Vous êtes sur la page 1sur 552

Fédération Internationale des Instituts d’Études Médiévales



GENEVA, 12-16 JUNE 2012

Edited by

Laurent CESALLI, Frédéric GOUBIER, Alain DE LIBERA

With the collaboration of Manuel Gustavo ISAAC


Présidents honoraires :
L.E. BOYLE (†) (Biblioteca Apostolica Vaticana et Commissio Leonina,
L. HOLTZ (Institut de Recherche et d’Histoire des Textes, Paris, 1999-)

Président :
J. HAMESSE (Université Catholique de Louvain, Louvain-la-Neuve)

Vice-Président :
G. DINKOVA BRUUN (Pontifical Institute of Mediaeval Studies, Toronto)

Membres du Comité :
A. BAUMGARTEN (Universitatea Babeş-Bolyai, Cluj-Napoca)
P. CAÑIZARES FERRIZ (Universidad Complutense de Madrid)
M. HOENEN (Universität Basel)
M.J. MUÑOZ JIMÉNEZ (Universidad Complutense de Madrid)
R.H. PICH (Pontificia Universidade Católica do Río Grande do Sul, Porto
C. VIRCILLO-FRANKLIN (Columbia University, New York)

Secrétaire :
M. PAVÓN RAMÍREZ (Centro Español de Estudios Eclesiásticos, Roma)

Éditeur responsable :
A. GÓMEZ RABAL (Institución Milá y Fontanals, CSIC, Barcelona)

Coordinateur du Diplôme Européen d’Études Médiévales :

G. SPINOSA (Università degli Studi di Cassino)
Fédération Internationale des Instituts d’Études Médiévales


GENEVA, 12-16 JUNE 2012

Edited by

Laurent CESALLI, Frédéric GOUBIER, Alain DE LIBERA

With the collaboration of Manuel Gustavo ISAAC

Barcelona - Roma
ISBN: 978-2-503-56735-8

All rights reserved. No part of this publication may be reproduced,

stored in a retrieval system or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording or otherwise, without the
prior permission of the publisher.

© 2016 Fédération Internationale des Instituts d’Études Médiévales.

Largo Giorgio Manganelli, 3
00142 Roma (Italia)

Introduction VII

I. Formal Logic: Hylomorphism and Formal Validity

IWAKUMA Yukio, On Medium in the Early Twelfth Century 3

John MACFARLANE, Abelard’s Argument for Formality 41
Christopher John MARTIN, Abaelard on Logical Truth 59
Giulia LOMBARDI, Le caractère formel de la logique en tant qu’ars 77
Julie BRUMBERG-CHAUMONT, La forme syllogistique et le problème
des syllogismes sophistiques selon Robert Kilwardby 93
Catarina DUTILH NOVAES, The Form of a Syllogism: Mood or Figure? 117
Paul THOM, Analysing Arguments in the Thirteenth and Fourteenth
Centuries 133
Joke SPRUYT, John Wyclif on the Formal Nature of Inference 149
Riccardo STROBINO, What is Form All About? A 14th-Century Discussion
of Logical Consequence 173

II. Formal Semantics: Issues and Strategies

Sten EBBESEN, Habitudines locales 197

Ana María MORA-MÁRQUEZ, Aristotle’s Fallacy of Equivocation and
its 13th Century Reception 217
Leone GAZZIERO, «Utrum figura dictionis sit fallacia in dictione. Et
quod non videtur». A Taxonomic Puzzle or How Medieval
Logicians Came to Account for an Odd Question by an Impossible
Answer 239
Simo KNUUTTILA, Scotus’s Formal Semantics of Modal Notions 269
Harald BERGER, «Sortes differt ab omni homine». A Tension in Albert
of Saxony’s Concept of Merely Confused Supposition 283
Luca SBORDONE, Semantics and Pragmatics of Reference. Elements
of a Contemporary Theory of Supposition 303

Paloma PÉREZ-I LZARBE, Jerónimo Pardo on the Formality of the

Expository Syllogism 325

III. Natural or Ideal Language?

Allan BÄCK, Aristotelian Protocol Languages 343

Christoph KANN , Raina KIRCHHOFF , Formal Elements in Natural
Language. Sherwood’s Syncategoremata Revisited 373
Claude PANACCIO, Ockham on Nominal Definitions, Synonymy and
Mental Language 393
Ernesto PERINI-SANTOS, The Underdetermination of Mental Language
in William of Ockham and John Buridan 417
Joël BIARD, Jean Buridan : une philosophie du langage ordinaire ? 435
Stephen READ, Non-normal Propositions in Buridan’s Logic 453
Sara L. UCKELMAN, Beyond Formality: The Role of the Dialectical
Context in Medieval Logic 469
E. Jennifer ASHWORTH, How Natural is Natural Language? Some Post-
medieval discussions 485
Paolo NATALI, Appendix to an impossible cover 501

Index auctorum antiquorum et mediaevalium 511

Index auctorum recentiorum 515
Index rerum 521
Index codicum 537

Since their birth in 1973, when three eminent scholars were having
some drinks together in Warsaw, the European Symposia for Medieval
Logic and Semantics have played an important role in coordinating a
movement of rediscovery of late medieval philosophy of language, and
more specifically of its strong logical-semantic component. Besides the
acknowledgement of the sophistication and refinement of medieval logical
theories, as well as of the wide scope of the medievals’ inquiries, what has
particularly fuelled this enthusiasm is the awareness of striking affinities
with some of the main semantic approaches developed in the wake of the
late 19th-early 20th century revolution in logic; indeed, the two periods
seem to share formal ambitions. As far as logic is concerned, this was
shown as early as 1956 by Bocheński, who pointed to some late medieval
distinctions between formal and material consequences. In semantics, the
affinity was found in the elaborate late medieval theories of quantification,
reference, modalities, etc., and their role in attempts at formally determining
the truth conditions of propositions. In both periods logic has maintained
complex, and at time uneasy, relations with natural language, and both
have seen pragmatic, ‘ordinary language’ endeavours to overcome some
of the difficulties. Soon, however, modern scholars became careful not too
overlook the specificities of theories developed more than five hundred
years apart, in particular with respect to their ‘formal’ character.
Tackling the question whether medieval logic can be considered as
a formal logic, Alfonso Maierù noted already in 1972 that the efforts of
medieval logicians over the centuries could be seen as tending towards
the identification of logical «structures» in language and its use which
are formal enough to become objects of scientific consideration1. Maierù
pointed at the medievals’ «quest for a formal logic whose validity extends
to any field of knowledge and does not depend on the peculiarities of
different subject matters… but obeys only its own principles». But
Maierù also stressed that «the language investigated is a historical one,
Latin. One might thus wonder to which extent… one is allowed to speak
of ‘formal logic’ in the middle ages». In other words, medieval logic in

A. MAIERÙ, Terminologia logica della tarde scolastica, Edizioni dell’Ateneo,
Roma 1972, pp. 41-43.

all its complexity –one could just as well talk about medieval philosophy
of language– is characterized by a certain tension between the aims of
‘formalist ambitions’ and the constraints proper to natural language. Such
a configuration has made of the Middle Ages an exceptionally promising
field of investigation for historians interested not only in logic and the
development of formal approaches, but also, and more fundamentally, in
the question of what it means, for logic, to be formal.
Some forty-three years and twenty symposia after the Warsaw meeting
–and forty-four years after Maierù’s remarks–, our knowledge of the field
has considerably expanded, allowing us to reassess the question of the
formal character and formalising ambitions of medieval logic, as well as
that of the natural character of the language in (and on) which it operated:
in other words, the question of the nature, object and purpose of medieval
logic. The 19th European Symposium for Medieval Logic and Semantics,
which took place in Geneva from the 12th to the 16th of June 2012, intended
to contribute to such a reassessment. So do the 25 papers published in
these proceedings, which tackle the issue from three different angles. The
first part focuses on the notion of formal logic, and more specifically on
questions pertaining to hylomorphism and formal validity; the second
considers different ways the medieval philosophers had of conceiving (and
doing) logical semantics; and the third gathers investigations regarding
the extent to which the object language of medieval logic is natural –or
The conference at the origin of this volume was part of the Swiss
National Science Foundation Project nº 129877, «Formal Semantics
and Natural Language in 13th Century. Theoretical Devices and Applied
Perspectives»; it benefited from the financial support of the Marie Gretler
Stiftung, the Société académique de Genève as well as the Commission
administrative of the University of Geneva. We are very grateful to all these
institutions, as well as to Ana Gómez Rabal and the people from TEMA
for their superb editing work and their immense patience. Our gratefulness
also goes to the team that provided a crucial help with the indices: Parwana
Emamzadah, Alexander Eniline, Markus Erne, Charles Girard, Magali
Roques. And finally, we would like to thank all the contributors of this
I. Formal Logic: Hylomorphism and Formal Validity


The two Introductiones, Vienna and Escorial, which I edited as works

of William of Champeaux, contain treatises on media as appendix1. The
term medium means a special type of hypothetical sentence, which first
appeared in the two Introductiones around 1100, provoked controversy in
the first few decades of the 12th century, and declined in the later period.
In the present paper, I shall first discuss (1) what a medium is, then (2)
the controversy on media in the early 12th century, and lastly (3) the latest
treatise on media known so far.

1 What is a medium?

1.1. Media in William of Champeaux’s Introductiones

The earliest known texts on media, the Vienna and Escorial ones,
discuss several media, giving some of them special names such as

medium a partibus: Vienna 1.1, Escorial 2

si omne animal est asinus, omnis homo est lapis.
medium a subiecto: Vienna 1.2, Escorial 4
si omnis homo est lapis, tunc, si omnis lapis est homo, omnis
homo est asinus.
medium a praedicato: Vienna 1.3, (Escorial 3)
si omnis homo est lapis, tunc, si omnis lapis est homo, omnis
lapis est asinus.
medium parvum: Vienna 1.5, Escorial 1
si omnis homo est lapis, quidam lapis est lapis.

Center for Arts and Sciences, Fukui Prefectural University. Email: kumayuk@
Introductiones dialecticae <secundum Wilgelmum> IV (Ms Wien, VPL 2499,
ff. 35v-42v) and Introductiones dialecticae artis secundum magistrum G. Paganellum
III (Ms El Escorial, e IV 24, ff. 52v-53r), ed. in IWAKUMA Y., «The Introductiones
dialecticae secundum Wilgelmum and secundum magistrum G. Paganellum», Cahiers
de l’Institut du Moyen-Âge Grec et Latin (CIMAGL), 63 (1993) 75-78 and 112-114.

At first sight, these media appear to be sophistic, their antecedents

and/ or consequents being false. Our texts, however, do not treat them as
sophisms; instead, they give a proof for each of them. For example, the
Vienna treatise gives the following proof for the medium parvum:

Text 1*: The Vienna treatise 1.5, Ms Wien, VPL 2499, f. 38r, p. 782

Media. Nota hanc naturam in parvo medio: quotiens aliquid1

de aliquo universaliter, praedicatum praedicatur de se particulariter,
ut hic: si (A1) omnis homo est lapis, (A2) quidam lapis est lapis.

Quod sic probatur. Si (A1) omnis homo est lapis, tunc vera est
ipsa, scil. (A1) ‘omnis homo est lapis’, et iterum ista (A2) ‘quidam
lapis est homo’. Et, si verae sunt istae duae propositiones (A1)
‘omnis homo est lapis’ et (A2) ‘quidam lapis2 est homo’, tunc (A3)
quidam lapis est lapis (a subiecto, regula: si aliquid subicitur alicui
universaliter, et aliquid subicitur subiecto particulariter, primum
praedicatum praedicatur3 de secundo subiecto particulariter).
Collige terminos positos.

1 aliquid] aliq(uo)d W – 2 lapis est homo] homo e(st) lap(is) W – 3

praedicatur] particularit(er) W et Iwakuma 1993

The proof in the Escorial treatise runs perfectly parallel (ed. Iwakuma,
p. 112). It is certain, then, that the author, William, accepts media as true
The term medium is never explicitly defined in any of the relevant
works known so far, which I shall all discuss in detail below. But we have
two clues as to what the term medium means: the two expressions most
frequently used, namely media propositio and per medium.

1.2. Medium / medius / media

First, we should note that the object of our concern is medium in neuter,
not medius in masculine nor media in feminine. Boethius’ translation of
Aristotle’s Prior Analytics uses the term medium in neuter as well as medius

The notes numbers within the quotations refer to the critical apparatus at the end
of each quotation (eds.).
Ed. IWAKUMA, «The Introductiones dialecticae», pp. 45-114.

in masculine in the sense of a middle term of a categorical syllogism.

Therefore, after the mid-12th century, when the work of Aristotle’s began
to be widely circulated, medium in neuter came to be commonly used in the
sense of a middle term. Thus John of Salisbury in his Metalogicon, written
in 1159, reports that it was William of Champeaux who took up medium,
but John confuses William’s medium in neuter with medius (terminus) in
masculine3. In the first half of the century, however, ‘middle term’ is always
expressed by medius in masculine, apart from a very few exceptions;
therefore medium in neuter must have had then a different meaning.
The term media (propositio) in feminine is used in all the extant 12th
century commentaries on Boethius’ De syllogismis hypotheticis to refer to
a special type of hypothetical sentences of the form

si est A, est B; si est B, est C,

from which syllogisms are composed with conclusions of the form

ergo, si est A, est C.

Boethius does discuss such syllogisms in his De syllogismis

hypotheticis I (855C-866A10), but he never calls them by the name media
propositio, an expression used only by medieval logicians.

1.3. Per medium arguments

Another expression, per medium, is used throughout the first half of

the 12th century, even in the periods before and after medium itself was
discussed. It is usually used to mean a special type of proofs, namely those
of the form

John of Salisbury, Metalogicon III.9.43-53 (CCCM 98, p. 129): «Versatur in
his inventionis materia, quam hilaris memoriae Willelmus de Campellis, postmodum
Catalanensis episcopus, diffinivit, etsi non perfecte, esse scientiam reperiendi medium
termium et inde eliciendi argumentum. Cum enim de inhaerentia dubitatur, necessarium
est aliquod inquiri medium cuius interventu copulentur extrema. Qua speculatione an
aliqua subtilior vel ad rem efficacior fuerit, non facile dixerim. Medium vero necessarium
est ubi vis inferentiae in terminis vertitur. Si enim inter totas propositiones sit, ut potius
sit obnoxia complexioni partium quam partibus complexis, medii nexus cessat.»

if P, then Q; if Q, then R; therefore, if P, then R

or a longer chain

if P1, then P2; if P2, then P3; ...; if Pn-1, then Pn; therefore, if P1, then

The earliest usage of the expression per medium is found in the

following proto-vocalist text4 from the late 11th century.

Text 2: Ms Paris, lat. 544, f. 102

Quod autem falsae particulares et indefinitae ponant falsas

universales, probatur hoc modo per medium. Nam, si (P) falsa est
‘quidam homo [non] est albus’, (Q) vera est ‘nullus homo est albus’;
et, si (Q) vera est ‘nullus homo est albus’, (R) falsa est ‘omnis homo
est albus’. Et item, si (P’) falsa est ‘homo est albus’, (Q’) vera est
‘nullus homo est albus’; et, si (Q’) vera est ‘nullus’, (R’) falsa est
‘omnis homo est albus’. Rursus, si (P1) falsa est ‘quidam homo
non est albus’ vel ‘homo non est albus’, (P2) vera est ‘omnis homo
est albus’; et, si (P2) vera est ‘omnis homo est albus’, (P3) falsa est
‘quidam1 homo non est albus’ <vel ‘homo non est albus’>; quodsi
(P3) falsa est ‘quidam homo non est albus’ vel ‘homo non est albus’,
(P4) falsa est ‘nullus homo est albus’.

1 quidam homo non est albus] nullus homo est albus P

The expression ‘secundum medium’ still frequently appears with the

same sense in the Introductiones Montanae minores and maiores from the
mid-12th century5.
We should note that the per medium argument is never discussed as
such by Boethius, but is a product of medieval logicians. A per medium
argument is not the same as a media propositio syllogism. In media
propositio syllogisms, the constituent propositions always have the same

For the ‘proto-vocalism’, see Y. IWAKUMA, «Vocales Revisited», in T. SHIMIZU
– Ch. BURNETT (eds.), The Word in the Medieval Logic, Theology, and Psychology,
Brepols, Turnhout 2009, pp. 86-89.
See, for example, the Intr. Mon. Minores, pp. 27.14 and 26, 28.20 and 22, 29.3,
and 34.27, Ed. by L. M. DE RIJK, Logica Modernorum II-2, Van Gorcum, Assen 1967.

subject term; but in per medium arguments, P, Q, and R can well be

propositions with different subject terms. Presumably, paying little notice
to this difference, medieval logicians took the name media propositio from
the expression per medium, which had already been used before Boethius’
De syllogismis hypotheticis came to be known.

1.4. Conclusion

Now, it should be pointed out that the media were always proved by
a per medium argument. For example, the proof of parvum medium cited
before (Text 1) runs as follows:

A1 ¨ A1 ∧ A2, A1 ∧ A2 ¨ A3 ⊢ A1 ¨ A3.

It is also the case with all the arguments concerning media which I
shall discuss below.
I conclude, then, that a medium is a sentence with the conjunction
«if» which can be proved by a per medium argument. Special nominal
expressions such as medium parvum, etc. are given to some strange kinds
of such arguments, which are true in spite of the fact that their antecedent
and/or consequent are false.

2. Controversies over media in the early 12th Century

There began a controversy over proofs of media in the first decade of

the 12th century.

2.1. The dissimilitudo theory

The controversy at its earliest stage was reported in Vienna treatises

on media.

Text 3: the Vienna treatise 1.4, Ms Wien, VPL 2499, ff. 37v-38r, ed.
Iwakuma pp. 77-78

[1.1] Haec supradicta media, scil. medium a partibus et de praedicato

et de subiecto, hoc modo determinantur. †Vel magis† quando ex

aliquo antecedenti duo consequentia sequuntur, unumquodque per se,

ex quibus postea simul acceptis sequitur aliud, non licet iungi extrema,
quia medius terminus dissimiliter enuntiatus est, ut in argumentatione
medii a partibus ex1 ‘omne animal est asinus’ sequitur ‘omne non
asinus est non homo’ et ‘omnis asinus est non homo’ unumquodque
per se; et ex illis postea simul acceptis trahitur ‘omnis homo est bos’
vel aliquid tale; medius itaque terminus dissimiliter enuntiatus est,
scil. illae duae propositiones dissimiliter enuntiantur; quare non
probatur consequentia, quia non ostenditur sequi ex aliquo vero.
[1.2] Et hoc habet auctoritas Boethii2 dicentis quod, quando medius
terminus dissimiliter enuntiatur, non licet extrema concludere, si fit
talis dissimilatio quae impediat conclusionem.

[2.1] Sed contra respondetur quod, cum haec Boethius dixit, agebat
de modalibus |38r| propositionibus3; sed praemissae propositiones
non sunt modales; quare haec ratio non videtur cogens ut3 concedatur.
In nulla enim argumentatione †universaliter iterum dicitur† extrema
cohaerent, quia ibi semper medius terminus dissimiliter enuntiatur.
[2.2] Ratio etiam praedicta in medio a partibus non videtur valere,
†quod dicimus, quando aliqua propositio falsa concedatur pro vera,
quod debeat habere illud quod vera,† scil. quod praedicatur5 sit
maius vel par subiecto.

1 ex] Et W – 2 Boethii] ul’ W – 3 propositionibus] p(ro) omnib(us)

W – 4 ut concedatur] inconcedat(ur) W, Sed concedatur legit De
Rijk – 5 praedicatur sic W, sed praedicatum legit De Rijk; fortasse
sic legendum?

Unfortunately, some parts are corrupt, but we can at least see that there
were some people who did not accept proofs of media ([1]); however, the
author himself, William of Champeaux, does not agree with them ([2]).
Those who did not accept proofs of media claimed that they have a flaw
because the middle terms are accepted differently (dissimilliter) in some
steps of the proof ([1.1]), appealing to the authority of Boethius ([1.2]). We
shall call this the dissimilitudo theory.
William rejects the dissimilitudo theory, saying ([2.1]) that Boethius
only speaks of modal propositions, but no modal propositions appear in
the proofs of media. Boethius actually asserts in the passage alluded to (De
syl. hyp. I, 842C3-D5) that in order to construe syllogisms from mediae
propositiones, the middle term should not be enunciated dissimiliter, as for
example in this manner:

si est A, est B; et, si necesse est esse B, est vel non est C,

touching only modal propositions.

This is the beginning of the controversy.

2.2. Peter Abelard in his Dialectica

Abelard discusses relevant issues in Book III.1 of his Dialectica.

2.2.1. Before we discuss our issue, however, it is necessary to see that

Book III has a very complicated structure, which is not always clear from
De Rijk’s edition6.
Book III intends to discuss topical arguments. It begins with the
definition of loci (pp. 253.16-263.3) and its division (pp. 263.4-271.21),
but in the course of the latter, a long interpolation intervenes (pp. 271.21-
309.24). After the interpolation, a discussion of maxima propositio follows
(pp. 309-25-331.4), although they had already been discussed before (pp.
264.5-268.35), and then we have discussions of each locus (pp. 331.5 to
the end).

Table of contents of the <Tractatus> III <Liber> I (pages in De Rijk’s edition) :

<Definitio loci> 253
¶ Divisio locorum 263
<De differentia maximarum propositionum> 264
<De loco differentia> 268
======== interporation =======
<Utrum inferentia consequentiarum veritatem custodiat> 271
<I: Quod veritas consecutionis in necessitate constat> 271
<Quod necessarium est sempiternum> 279
<De differentia categoricae et hypotheticae enuntiationis> 282
<II: De consecutionis necessitate> 283
<De necessitate largiore et strictiore> 283
<De regulis antecedentis et consequentis> 286
<De significatione antecedentis et consequentis> 287
<Octo regulae> 288

Peter Abelard, Dialectica, Ed. by L. M. DE RIJK, Van Gorcum, Assen 1970.

<De regulis I-V> 288

<Notanda de regulis I-V> 292
<De regulis VI, VII, et VIII> 305
¶ De maximis propositionibus 309
¶ De loco a substantia 331
De locis a consequenti substantiam 339
¶ De <locis> extrinsecis 369
De locis mediis 406

2.2.2. The interpolation begins as follows, proposing two issues to be


Text 4: Abelard, Dialectica III, Ms Paris, BnF, lat. 14614, f. 152r,

pp. 271.22-26 ed. De Rijk

Sed nunc quidem, utrum omnium inferentia suprapositarum

consequentiarum veritatem custodia[n]t1, consideremus. Quod
facilius hoc modo fecerimus, si2 prius, (I) in quo hypotheticae
propositionis veritas consistat, deliberemus; ac (II) deinde3, quid4
sit harum consequentiarum <necessitas>5, investigemus.

1 custodia[n]t correxit De Rijk, c(us)todia(n)t V – 2 si correxit De

Rijk, at (non ac ut reportavit De Rijk) – 3 d(e)in(de) solvi, demum
solvit De Rijk – 4 quid correxi, q(ui) V et De Rijk – 5 <ncessitas>
addidi, <sensus> addiderunt Cousin et De Rijk

As for Issue I, Abelard asserts (pp. 271.26-279.13) that the truth

of hypothetical sentences consists in the necessity of logical entailment
(necessitas consecutionis). Two corollaries follow. The truth or necessity
of an entailment is eternal, that is to say, it is not affected by changes in
the world (pp. 279.13-282.29). Therefore it is different from the truth
of categorical sentences, which is affected by changes in the world (pp.
282.30-283.35). Joscelin of Soissons takes a similar position (cf. ¶ 2.3.2

2.2.3. As for Issue II (pp. 283.37-309.24), Abelard first distinguishes

two meanings of necessity (pp. 283.37-286.30). In a broader sense, it
means that if what the consequents say is not the case, what the antecedents
say cannot be the case. In a stricter sense, it means that antecedents in

themselves require consequents, in addition to the fact that antecedents

cannot be true without the consequents being such. The necessity of
entailment should be accepted in the latter, stricter, sense.

Text 5: Abelard, Dialectica III, Paris, BnF, lat. 14614 f. 153v, pp.
283.37-284.3 ed. De Rijk

Videntur autem duae consecutionis necessitates. Una quidem

largior, cum videlicet id quod dicit antecedens non potest esse
absque eo quod dicit consequens. Altera vero strictior, cum scilicet
|284| non solum antecedens absque consequenti non potest esse
verum, <sed1 etiam> ex se ipsum exigit. Quae quidem necessitas
in propria consecutionis sententia consistit et veritatem tenet
incommutabilem, ...

1 <sed etiam> addidit De Rijk

After that, Abelard discusses the following eight general rules for
an antecedens and a consequens in a consequentia (pp. 286.31-309.24),
asserting that only the first two cases are valid and that all the remaining
ones are invalid.

I {antecedens}⊢{consequens} II =I’ ¬{consequens}⊢ ¬{antecedens}
*III {antecedens}⊢ ¬{consequens} *VIII=III’ {consequens}⊢ ¬{antecedens}
*IV ¬{antecedens}⊢ ¬{consequens} *VII=IV’ {consequens}⊢ {antecedens}
*V ¬{antecedens}⊢{consequens} *VI=V’ ¬{consequens}⊢{antecedens}

In between the discussions of rules I-V (pp. 286.31-292.33) and VI-

VIII (pp. 305.8-309.24), Abelard makes a long and complicated digression
(pp. 292.34-305.7). This digression deserves a more detailed study, since it
is directly relevant to our present concern, medium.

2.2.4. In the digression, Abelard first points out (pp. 292.34-293.4)

that there are some people who assert that Rules III and IV are valid in
some cases, viz. in the topical arguments on locus ab oppositis and ab
immediatis. For example,

si est sanum, non est aegrum (against the invalid rule III)
si non est sanum, est aegrum (against the invalid rule IV).

But Abelard refuses, as is well known, the validity of locus ab oppositis

and ab immediatis. Against those who accept the two loci, Abelard argues
as follows (p. 293.5-25). The sentences above (‘si est sanum, ...’ etc.)
might be true according to the relation of things (secundum habitudinem
rerum), but the aforementioned rules (III and IV) should not be understood
according to the relation of things, but according to the whole meaning of
the antecedent (secundum totarum antecedentis enuntiationum sententiam).
There are many if-sentences (consequentiae) that are true according to the
former, but false according to the latter.
Abelard, then, gives some examples of such if-sentences7, viz.

(a) si omnis homo est lapis, quidam lapis est lapis

(b) si Socrates est Brunellus, Socrates est homo,
giving proofs for (a) and (b), as follows.

Text 6: Abelard, Dialectica III, Ms Paris, BnF, lat. 14614 f. 155r, pp.
294.6-17 ed. De Rijk

Sic tamen <verae>1 ostendi poterunt suprapositae consequentiae,

hoc modo. Si (a1) omnis homo est lapis
vera est, simul et (a2) quidam lapis est homo.
Quodsi verae sunt (a1) omnis homo est lapis
et (a2) quidam lapis est homo,
contingit necessario veram2 <esse>
(a3) quidam lapis est lapis.
Haec est enim forma inferentiae tertii modi primae figurae. Unde et
sic per extremorum coniunctionem infertur
(a) si omnis homo est lapis, quidam lapis est

Abelard, Dialectica III, ed. DE RIJK p. 293.31-32. He gives another example (p.
293.34) (c) «si Socrates (] id(em) Ms et De Rijk) est animal, <Socrates> est homo»,
and discusses it separately from (a) and (b) (pp. 302.18-303.28). This is irrelevant to
our present concern.

Sic quoque secunda (b) convincitur vera.

Si (b1) Socrates est Brunellus
vera est, simul et (b2) Brunellus est Socrates.
Ex his autem duabus necessario consequitur
(b3) Socrates est Socrates.
Ex qua etiam infertur
(b4) Socrates est homo.
Coniunctis itaque extremitatibus concluditur
(b) si Socrates est Brunellus, Socrates est

1 <verae> addidi – 2 veram <esse> lexit et addidit De Rijk, u(er)

a(m) V, <veram> esse lexit Cousin

Now, Sentence (a) is exactly the same as parvum medium although

Abelard does not explicitly call it by the name ‘medium’. Moreover, the
proof given for (a) runs perfectly parallel to that given in the treatises on
media (see Text 1 above). Abelard thus accepts parvum medium (a) as a
true sentence as well as (b) together with their proofs.

2.2.5. Abelard then discusses the dissimilitudo theory (pp. 294.18-

295.34 ed. De Rijk).

Text 7: Abelard, Dialectica III, Ms Paris, BnF, lat. 14614, f. 155r, p.

294.18-24 ed. De Rijk

Huic tamen extremorum coniunctioni <non>nulli1 resistere conantur

ex medii termini dissimilitudine. Ait enim in Hypotheticis suis
Boethius, cum numerum mediarum hypotheticarum inquireret
eumque numero earum quae ex tribus terminis iunguntur
compar<ar>et2, ut sunt illae, quae vel ex categorica et hypothetica vel
e converso connectuntur; nec unam vocari hypotheticam3 cui medius
terminus dissimiliter accipitur, nec extrema per eum connecti.

1 <non>nulli correxit De Rijk, nulli V, nonnulli lexit Cousin – 2

compar<ar>et correxit De Rijk, c(om)p(ar)& V, compararet lexit
Cousin – 3 ip(otheti)ca(m)] ip(otheti)cu(m) a.c. V

The proof of (b) above is criticized from the point of view of the
dissimilitudo theory, as follows.

Text 8: Abelard, Dialectica III, Ms Paris, lat. 14614, f. 155v, p.

295.14-24 ed. De Rijk

Sic quoque et ex dissimilitudine medii termini illi quoque superiori

argumentationi, quae ostendebat

(b) si Socrates est Brunellus, Socrates est homo,

per denegationem ultimae consequentiae, quae ait
si (b3) Socrates est Socrates, (b4) Socrates est homo,
resistere laborant. Aiunt enim istam consequentiam non
aliunde veram videri nisi ex eo quod Socrates hominis
inferius fuerit. Illum autem hominis inferius esse non iam
recipiunt, ubi in praemissa enuntiatione (b1) Brunellus (qui
hominem expellit) coniunctus Socrati fuerat, immo asinus1
per Brunellum (qui ei copulatur). Neque enim iam ipsum
aliud a Brunello esse nullo modo aiunt.
1 asinus correxerunt De Rijk et Cousin, asini V

2.2.6. Abelard next reports that many people reject the dissimilitudo
theory and accept the proofs of (a) and (b) as valid ones.

Text 9: Abelard, Dialectica III, Ms Paris, BnF, lat. 14614, f. 155v,

pp. 295.35-297.24 ed. De Rijk

Sed sunt plerique qui primae argumentationis consequentiae (a, b)

obviandum non censeant propter illam terminorum enuntiationem.
Neque enim resistendum in huiusmodi argumentationibus dicunt nisi
ex dissimilitudine medii termini. Medius autem <terminus>1 nondum
existit, una |296| tantum consequentia proposita. Ut enim medius
intercedat terminus, oportet ipsum et in prima consequentia subsequi
et in secunda antecedere. Priori vero con sequentiae contradicere non
audent eo quod compellantur confiteri falsas eas <esse>2 quoque, quae
ex falsis propositionibus vere texuntur, veluti ista
(f) si omnis homo est margarita, omnis homo est lapis.

Neque enim iam margarita (hominem prorsus continens) species

lapidis remanet.

Sed, <si>3 priores consequentias (a, b) ex ulla terminorum

enuntiatione non denegent, quomodo4 illam argumentationem
absolvunt quae Aristotelicam auctoritatem oppugnat, ubi in tractatu
Oppositorum: “Sanis, inquit (Cat. p. 75.5), omnibus, sanitas quidem
erit, languor vero non erit”?5 Est autem talis argumentatio quae
potius ostendat (g) si omne animal sit sanum, languorem esse; quod
plane Aristoteli contradicit, qui (g*) si omne animal sanum sit,
languorem non esse confirmat. Est autem huiusmodi argumentatio

si (g1) omne animal est sanum, (g2) omne languidum est

a toto. Animal namque etiam languentia continet. Quod
si (g2) omne languidum est sanum,
(g3) omne languidum est animal.
Unde et (g4) quoddam animal est languidum.
Unde et (g5) languor in quodam animali consistit,
et ita (g6) languor est.

Itaque per medium ostensum est quod

(g) si (g1) omne animal est sanum, (g6) languor est.

Quae quidem argumentatio cassari non dicitur ut6 in priori

consequentia (a, b).

Sed primam quidem consequentiam (g1 ¨ g2) pro eo contradicunt

quod prima eius positio (g1) quasi concessa praemittatur. Cuius
veritatis concessio (g1) iam omnino perimit continentiam languidi
in animali (g2). Ubi enim animal sano prorsus supponi conceditur
(g1), nullum continere languidum annuitur. Tunc itaque terminorum
enuntiationem (g1) unius quoque per se positae consequentiae
inferentiam (g1 ¨ g2) perimere dicunt, cuius habitudinis vim
enuntiationis veritas aufert, cum subcontinua<tiva>7 conditio
apponitur, id est cum id quod iam concessum fuerat coniungit.
Cum autem non <subcontinuativa8 est sed> solum continuativa nec
quicquam concessionis requiritur sed sola conditionis copulatio
attenditur, id, quod nec verum est nec concessum, consecutionem
impedire non dicunt.

Sed unde maiorem vim adversus enuntiationem secundae (g2 ¨ g3)

vel tertiae9 (g3 ¨ g4) consequentiae quam adversus enuntiationem

primae (g1 ¨ g2) habeat primae partis enuntiatio (g1), cum

videlicet nec vera sit nec concessa.10 Sicut in superioribus
argumentationibus, ubi vel ex
(d1) omne animal est homo
(d5) quoddam animal non est homo
vel ex (b1) Socrates est Brunellus
(b4) Socrates est homo

extrahebatur, non ex |297| primis quidem propositionibus (d1,

b1) ut ex concessis argumentari incipimus, nec ullam earum
concessionem requirimus, sed sub conditione solum si hoc est,
illud esse proponimus. Quae quidem conditionis copulatio aeque
et in veris et in falsis enuntiationibus consistit. Nulla itaque
exhibet<ur>11 ratio ut prioris enuntiatio partis magis sequentes
quam priores consequentias oppugnare valeat, sed12 aequaliter
eius veritas adversa utrisque13 fuerit. Sed, nec pro enuntiatione
quacumque impediri potest quaecumque consequentia vera
consistit. Quod enim sempiternum [non]14 est ac necessarium,
nullo potest impediri casu, nec propter enuntiare, ut Aristoteles
meminit (De int. 9 p. 16.3-4), erit aliquid aut non erit. Enuntiatione
ergo resistendum non est, sed15 veritate enuntiationis. Alioquin per
oppositam enuntiationem quamlibet possemus impedire veritatem,
qui sic16 quidlibet vel esse vel non esse enuntiare<mus>17. Si ergo
sola enuntiationis veritas impediat, nulla autem vera enuntiatio
sic18 praemissa consecutione<m>19 posteriorum hypotheticarum
impediat, profecto inique20 calumniantur ex veritate praemissae
enuntiationis consequentias posteriores. Nam consequentiae21, quae
praemittuntur, verae quidem conceduntur ab ipsis, veluti illa quae
si (d1) omne animal est homo, (d2) omnis asinus est
et si (b1) Socrates est Brunellus, (b2) Brunellus est Socrates.

Sed earum veritas aliarum veritatem nullo modo impedit. Neque

enim verum vero potest esse impedimento. Non itaque ex veritate
illius praemissae enuntiationis calumniari oportuit posteriores

1 <terminus> addidi – 2 <esse> addidi – 3 <si> addidi – 4 quomodo

correxi, q(uae) m(odo) V et De Rijk et Cousin – 5 ? (signum
quaestionis) addidi, . (punctum) V et De Rijk – 6 ut correxi, n(isi)

ut videtur V et De Rijk et Cousin – 7 subcontinua<tiva> correxit

De Rijk, subc(on)tinua V et Cousin – 8 <subcontinuativa est sed>
addidi – 9 tertiae correxerunt De Rijk et Cousin, tote V – 10 .
(punctum) correxi, ? (signum quaestionis) V et De Rijk; Cousin –
11 exhibet<ur> correxerunt De Rijk et Cousin, exibet V – 12 S(ed)
p.c. et a.c. Si V, si lexerunt De Rijk et Cousin – 13 utrisq(ue) p.c. et
a.c. utri(us)q(ue) V, utriusque lexerunt De Rijk et Cousin – 14 n(on)
V, expunxerunt De Rijk et Cousin – 15 sed veritate enuntiationis
correxi, s(ed) si forte en(un)tiatio(n)is V, sed si forte enuntiation<e,
sola veritate enuntiation>is correxit De Rijk, sed si forte enuntiationis
lexit Cousin et dixit “Locus perturbatus. Aut hic aliquid deest, quod
nobis non sccurrit, aut sed si forte enuntiationis delendum est” –
16 sic correxerunt De Rijk et Cousin, sit V – 17 enuntiare<mus>
correxit De Rijk, en(un)tiare V et Cousin – 18 sic correxerunt De
Rijk et Cousin, sit V – 19 consecutione<m> correxerunt De Rijk et
Cousin, c(on)secutio(n)e V – 20 inique correxit De Rijk, iniq(ui)ue
V, inique lexit Cousin – 21 consequentiae quae praemittuntur] q(uae)
p(rae)mittu(n)t(ur) c(on)seq(ue)ntie et transportanda indicavit V,
quae praemittuntur consequentiae lexerunt De Rijk et Cousin

This position is William of Champeaux’s. The ‘quasi concessa’ in

the quotation above is reminiscent of the following passage of William’s
Vienna treatise on media.

Text 10: The Vienna treatise 1.1, ms Wien VPL 2499 f. 36r, p. 75
ed. Iwakuma

Quaeritur hic a probante utrum ipse, <qui> argumentetur de ‘omne1

animal est asinus’, <argumentetur> de vera vel de falsa.
Respondetur2: de neutra, sed de concessa pro vera. Quo concesso,
oportet quod asinus sit maius animali subiecto secundum regularem
praedicationem, cum accidens <de subiecto> vel idem de se non
praedicetur ibi.

1 omne] omni W – 2 Respondetur] Regula W

Abelard agrees with William that (a) (= parvum medium) as well as

(b) is true and that their proofs are valid. Still, Abelard argues against
William’s position, since there is a difference between the two masters in
the acceptance of the proofs.

A work from William’s school, a version of C88, says as follows.

Text 11: C8 VP-version, Vat. Reg. lat. 230, f. 69va-b, Paris lat.
13368, f. 213va; the parts in parentheses {} are V-additions
SANIS1 NAMQUE (75.5). Hic2 probat a partibus contrariorum quod non
est necesse, si unum contrarium sit, et reliquum esse, sic3: |V69VB|
Quia necessarium `non´ est, si sanitas est, aegritudinem esse; et, si
albedo est, nigredinem esse. Quod ostendit his verbis: SANIS quidem
OMNIBUS existentibus, {id est si omne4 animal est sanum,} sanitas
erit, languor5 quidem non erit, id est OMNIBUS animalibus existentibus
per simplex esse {manente prima concessione, scilicet quod omne
animal sit sanum.}

Hic videtur posse probari quod liber denegat, scilicet quod

(g’) omnibus existentibus sanis, languor [non8] erit,
sic. Si (g’1) omnia9 animalia sunt sana, (g’2) omne animal est
<et, si (g’2) hoc est, (g’3) quoddam sanum est animal>;
et, si (g’3) hoc est, tunc (g’4) quoddam sanum10 est
et, si (g’4) hoc est, (g’5) quoddam animal est aegrum;
et, si (g’5) quoddam animal est aegrum, tunc (g’6)
aegritudo est.
Sed sic determinandum est, ubi dicitur
si (g’2) omne animal est sanum,
(g’3) quoddam aegrum11 est sanum,
quod, quando dicimus
(g’2) omne `animal´ est sanum,
ibi removemus aegritudinem ab omni animali; et ita non est vera
illa12 consequentia quae dicit
si (g’2) omne animal est sanum, (g’3) quoddam aegrum
est sanum.
{... ... ....

C8 is a commentary on Aristotle’s Categories, which was presumably originally
written by William of Champeaux, and preserved in various revisions by William’s
students. See IWAKUMA, «Vocales Revisited», pp. 89-91.

Sed13 videtur quod bene sequitur in veritate:

si (g’2) omne animal est sanum, (g’3) quoddam aegrum

est sanum.
Quod sic patet. Si (g’3) quoddam animal <est> aegrum, tunc,
si (g’2) omne animal est sanum, (g’3) quoddam aegrum
est sanum.
Sed, quando Aristoteles dicit quod
si omne animal sit sanum, languor non erit,

voluit pro certo verum esse omne animal esse sanum; et ideo,
manente illa concessione, non remanet languor. Quae consideratio
non potest esse in huiusmodi14 consequentiis
si (d1) omne animal est homo, (d2) omnis asinus est
Si enim concederemus ita esse confirmando quod (d1) omne animal
sit homo, non sequitur postea (d2) omnis asinus est homo, quia,
ubi confirmavimus omne animal esse hominem, ibi concessimus
perire15 omnes res sub animali praeter hominem. Unde dubitando,
non confirmando, accipitur prior propositio. Ad quod utile est
scire quod ‘si’ haec coniunctio, quando dicitur ativam, `tunc´ facit
dubitationem, quasi dicat
si hoc est quod dubium est, tunc est illud;
quando autem est subcontinuativa, tunc facit certitudinem, et valet
idem quod ‘quia’, quasi dicat
si hoc est quod verum est, (id16 est quasi ‘quia hoc est’),
est illud.
Et in his continuativa
si (d1) omne animal est homo, (d2) omnis asinus est
quae17 consequentia vera est, et omnes huiusmodi per se factae.}
Similiter, cum omnes sint albi, albedo quidem erit, sed18 nigredo
suum contrarium non erit.

1 Sanis P, Sanu(m) V – 2 h(ic) V, om. P – 3 sic P, om. V – 4 post

omne add. a V 5 la(n)gor(!) q(uidem) n(on) erit P, om. V – 6 langor

P, la(n)guor(um) V 7 u(ero) V, q(uidem) P – 8 n(on) PV, expunxi

– 9 o(mn)ia a(n)i(m)alia s(unt) sana P, o(mn)ib(us) a(n)i(m)alib(us)
exist(e)ntib(us) sanis V – 10 san(um) e(st) egr(um) V, egr(um) ÷
(= est) san(um) P – 11 egr(um) ÷ (= est) san(um) P, a(nima)l e(st)
egr(um) V – 12 illa V, om. P – 13 Sed videtur] Si uid(e)am(us) V
– 14 huiusmodi] h(uius) mo(d)is V – 15 perire + oes V – 16 id est]
q(uod) u(er)u(m) e(st) V – 17 quae] q(ua) V – 18 s(ed) V, om. P

According to these texts (Text 9 and 11), William considers that (g)
is false and (d) is true, since in (g) and (d) the conjunction ‘si’ should be
accepted differently, for in (g) it is subcontinuativa and equivalent to ‘quia’
while in (d) it is continuativa and simply equivalent to ‘if’. Abelard argues
against them that there is no reason at all to accept ‘si’ differently in (g)
and in (d).

2.2.7. Abelard reports, then, two more positions in this controversy,

together with counter-arguments against each position.
One is another kind of the dissimilitudo theory (pp. 297.25-301.30).

Text 12: Abelard, Dialectica III, Ms Paris, BnF, lat. 14614,

ff. 155v-156r, pp. 297.25-298.9 ed. De Rijk

Sunt autem et qui singulas recipiunt consequentias, sed extremorum

coniunctionem non admittunt propter dissimilitudinem
enuntiationis medii termini; cumque talis inducitur regula
si aliquid infert aliud, quod inferat aliud,
primum inferens inferre ul timum,
id subintelligendum et determinandum esse diiudicant:
terminis eodem modo acceptis,
non scilicet dissimiliter enuntiatis.
......a counter-argument ...... |298| ......
Nec, si Boethii dicta pensemus, qui dissimilem enuntiationem
medii termini calumniatur ipsamque extremorum coniunctionem
intercipere concedit, ut “Si, inquit (De syl. hyp. I, 842C13-15), ita
si est A, est B,
si necesse est esse B, est vel non est C”,

nihil illam dissimilitudinem huic pertinere videmus. Haec enim

dissimilitudo, quam appellant, non augmenta[n]t1 numerum
hypotheticarum mediarum, sicut Boethius voluit. Quippe eaedem
mediae propositiones eorumque termini manent.
...... counter-arguments ......
1 augmenta[n]t correxit De Rijk, augm(en)ta(n)t V

This is the same type of the dissimilitudo theory as that reported in the
Vienna treatise (see Text 3 above).
The second position reported and rejected by Abelard (pp. 301.30-
303.28) is as follows.

Text 13: Abelard, Dialectica III, Ms Paris, BnF, lat. 14614, f. 156v,
pp. 301.30-302.1 ed. De Rijk

Sunt autem qui et consequentias singulas in suprapositis

argumentationibus recipiant unamque extremorum coniunctionem;
sed propositum1 inconveniens, quod2 intellexera<n>t3, probatum
esse denega<n>t. Veluti, cum nos superius talem proposuimus
consequentiam probandam
(b) si Socrates est Brunellus, Socrates est homo,
quae <consequentia>4 ei inconveniens videbatur qui Brunellum et
hominem ut opposita |302| accipiebat.
1 propositum correxi, p(ro)po(n)it(ur) V et De Rijk – 2 quod +
<quod> addidit De Rijk – 3 intellexera<n>t ... denega<n>t correxi,
i(n)tellex(er)at ... denegat V et De Rijk – 4 <consequentia> addidi

2.3. Joscelin of Soissons

A somewhat later stage of the controversy is recorded in two sources

from the side of William’s students, Joscelin of Soissons (§ 2.3) and Walter
of Mortagne (§ 2.4 below).

2.3.1. SH4

Ms Orléans 266 contains several commentaries on the De syllogismis

hypotheticis. One of them, to which I give the number SH4, is a product

of the school of Joscelin of Soissons, since it mentions (f. 267a) ‘noster

m(agister) G.’ and again a little later to ‘m. Gosl.’’.

SH4 reports the controversy as follows.

Text 14: SH4, Ms Orléans 277, p. 271


[0] Unde videtur in`n´uere quod, si in media propositione medius

terminus eodem modo accipiatur, nullum sequitur inconve(niens).
[1.1] Quod videtur esse fal(sum), quia fal`sae´ videntur
(1) si omnis homo est lapis, quidam lapis est lapis, (= medium
(2) si Socrates est Plato, Socrates est Socrates,
(3) si omne ens est domus, nullum ens est domus,
(4) si omnis homo est Socrates, quidam homo non est Socrates1,
(5) si albedo est, nigre(do) est,
(6) si est lapis, aeger `est´,
(7) si nulla res est, dies est; et ita aliqua res est,
(8) si omne animal est homo, quoddam animal est homo.

[1.2] Non. Quae consequuntur ex mediis propositionibus, mediis

ter(minis) earum non dissimiliter acceptis, etsi quidam quibusdam
nugis et supradictas2 conclu`si´ones non sequi ex suis argu(mentis)
putant bene ostendere. Quorum rationes alias posu`i´mus, et eas
nihil valere ostendimus3.

[2.1] Nos vero, qui fere omnes consequentias <recipimus> ex

quibus supradica inconve(nientia) sequi videntur, qualiter illis
resistamus, b(revite)r ostendamus. [2.2] Dicendum est igitur non
esse medias propositiones, ex quibus supradictae4 sequi videntur,
ut hic contextus
(α) si omnis homo est lapis,
omnis homo est lapis et quidam lapis est homo;
(β) si omnis homo est lapis et quidam lapis est homo,
quidam lapis est lapis,
ex qua sequi videtur

(1) si omnis homo est lapis, quidam lapis est lapis.

[2.3] Et, quod non sit propositio me(dia) ille5 contextus, inde
apparet, quia non habet medium terminum, quia neque illas duas
propositio(nes)α neque ‘homo’ vel ‘lapis’. Illae duae propositiones
non sunt termini, quia hoc prohibuit Boethius, ubi dixit (De syl.
hyp. I 842D6-7) medi(um) t(erminum) +non5 dissi(militer) accipi;
quodsi propositiones medios diceret terminos, dissi(militer) accipi
medi(um) t(erminum)+ in media propositio<ne> perciperet, ut7 dicit
(De syl. hyp. II 859B13-167C7) inaequimoda8 propositio debere
poni in secunda et tertia9 figura affirmativa et ne(gativa), nec est
aliqua dissi(militudo) quam negare quod altera10 affirmat. Nec
‘homo’ vel ‘la(pis)’ est medius termi(nus), quia ‘homo’ vel ‘lapis’
non est talis termi(nus) qui sit prin(cipalis) pars talis cate(goricae)
quae sit consequens tantumβ, id est ita quod non <sit> alia
cat(egorica) propositio cum ea, <sed> consequens, inquam, primae
hypotheticae et antecedens11 secundae hypo(theticae)β. Nec auctor
de alia media propositione egit, ut apparet in exemplis primae
figurae, ut12 de tali, in consequenti cuius primae hypo(theticae)
talis vox esset praedi(catum) in una tantum cate(gorica), quae esset
praedi(catum) an(tecedent)is secundae hypo(theticae), et ostenderet
ex sua inventione antecedens iungi consequenti ad interpretandum
unum intell(ectum), ut hic
si est homo, est animal,
si est animal, est animatum <et> est cor(pus),
si est cor(pus), est substantia13

‘animal’ est medius terminus, quia ostendit ‘Socrates est homo’

iungi ‘Socrates est substantia14’, sicut superius dictum est, et15
multae sunt conse(quentiae).

1 Socrates] h(om)o O – 2 sup(ra)dictas] sup(ra)dictis a.c. O – 3

ost(e)ndim(us)] ost(e)ndem(us) a.c. O – 4 supradictae] sup(ra)dictis
i.t. et supra -is add. -e O – 5 ille contextus inde transportavi, in(de)
ille 9text(us) O – 6 +non - t(erminus)+ i.m. sinistr. add. O – 7 ut]
u(bi) ut videtur O – 8 inaequimoda] i(n)eq(ui)ssimo(!) O – 9 tertia +
[(et)] O – 10 altera] alt(er)a[m] O – 11 antecedens] o(mne)s O – 12
ut] n(isi) ut videtur O 13 substantia] sus. O – 14 substantia] sus.
O – 15 et] (et)(iam) O

α illae duae propositiones, scil. ‘omnis homo est lapis’ et

‘quidam lapis est lapis’.
β In media propositione, ‘si est a, est b; si est b, est c’
medius terminus ‘b’ debet esse principalis terminus, scil.
praedicatus, primae hypotheticae, scil. ‘si est a, est b’, et
antecedens secundae hypotheticae, sil. ‘si est b, est c’.

[1.1] mentions the eight if-sentences (1)-(8). The first is medium

parvum, and its proof mentioned in [2.2] is the same as the one
found in the Vienna and Escorial treatises (Text 1 above) and in
Abelard’s Dialectica (Text 6 above).

[1.2] alludes to the attack on the media from the dissimilitudo

theory. The author ([2.1]) rejects the attack and says that he accepts
almost all of the eight if-sentences (1)-(8). But the reason of his
acceptance ([2.2-2.3]), this time, is different from those given in the
Vienna treatise (Text 1 above) and in Abelard’s Dialectica (cf. pp.
297.31-301.30 ed. De Rijk). According to SH4, the dissimilitudo
theory is false in that it falsely presupposes that the first step of the
proof, namely ‘α and β’, is a media propositio.

2.3.2. SH4 continues as follows

Text 15: SH4, Ms Orléans 266, pp. 271-272 (cont. to Text 14)

[3.0] Et ostendamus quod1 <consequentiae supradictae> sint


[3.1] Quippe est alio modo iudicandum est de veritate cate(goricae)

quam hypo(theticae). Veritas enim cate(goricae) statum rei
comitatur; hypo(theticae) veritas non, immo, quae vera est
aliquando, sive res sint sive non, semper vera est, et ita semper est
omnis vera ne(cessaria). Quod auctor testatur. Dicit enim Boethius
in Hypotherticis, qui dicit (De syl. hyp. I, 833A1-5)): «‘si peperit,
cum viro concubuit’ non dicit ipsum partum esse <id esse> cum
viro concumbere, sed non potuit partus esse sine viri concubitu”.
Et alibi in eodem (De syl. hyp. I, 843C6-10): Qui inpugnare vult
talem consequentiam ‘si est A, est B’, non debet dicere ‘non est
A, non est B’, sed A posse esse sine B. Et alibi in eodem (De syl.
hyp. I, 843A14-B4): “Necessitas cate(goricae) terminis2 applicatur,
neces(sitas) vero hypo(theticae) quaerit conse(quentiam)”, id est
ostendit conse(quentiam) sequi ex an(tecedent)i necessario.

[3.2] Sunt tamen quidam qui dicunt necessi(tatem) hypo(theticae)

†nos(!?), s(cilicet) enuntiatum3† intel(lectui) comitari. Nec caremus
auctore. Dicit enim Por(phryius) (Ubinam?) †proprium speciem
accidens suum funda()†; et Boethius in To(picis) (De dif. top.,
1176D12-1177B12) hoc testatur, agens de hypoth(etica) quaestione
postquam egit de cate(gorica) quaestione.

[3.3] Alii, qui hoc non recipiunt, dicunt utrumque auctorem

accipere sequi pro praedicari. Sed hoc falsum est, quia, <cum>
de ca(tegorica) quaestione4 egerat communiter, de hypo(thetica)
quaesti(one) residuum erat tractare; et ideo de eis tractat in
hypo(theticarum) intel(lectibus). Alioquin superflueret ille tractatus.
‘In hypo(theticarum) intel(lectibus)’ dico, quia dicunt †in hypo()
illas, de quibus exemplificat, utrumque auctorem intel() si() ge()
cate()†, ut facit illas, scilicet (Cat. 8, p. 67.9-10) “unum de contrariis
fuerit5 quale, etc”.

[3.3.1] Et, quod sequi possit concipi pro praedicari, confirmant,

et Ar(istotelis) aucto(ritate) et B(oethii). Dicit enim Aristoteles
(Anal. pr. I.4, p. 11.10-11): “Cui non inest medium, sed omne vel
nullum sequitur primum”; Boethius in commento (Ubinam?): sequi
dico quotiens aliqua re nominata contigit alteram rem nominari, ut
‘omnis homo est rationalis’.

[3.3.2] Vera itaque vel necessaria est illa conse(quentia), secundum

alios quosdam, cuius antecedens non potest esse sine conse(quente).
Et vocant antecedens, non vel propo(sitionem) vel intel(lectum)
illius vel aliquam rem de qua |272| agatur per aliquam partem illius
propositionis, sed rem propositionis, quae nihil est, vel, quicquid6
dicant quid sit, non est vel substantia7 vel accidens. Quae sententia
satis apparet per se nihil valere.

[3.3.3] Alii plus addunt, scilicet et antecedens ex se exigit

conse(quens). Qui dicunt antecedens <non> vel propo(sitionem)
vel intel(lectum), <sed> res praedi(cativi). Sed, inquisiti quid sit
exigere antecedens, cadunt vel nugas aliquas fingunt.

[3.3.4] Alii dicunt conse(quentiam) illam ne(cessariam), sicut ma.

Pe., si consequentis8 sententia clauditur in an(tecedent)i, et vocat `et´
senten(tiam) conse(quentis) et an(tecedent)is earum propositionum
intel(lectus), ut intel(lectus) ‘Socrates est homo’ componitur ex his
intel(lectibus) ‘Socrates est animal’ ‘Socrates est corpus’; et sic de

ceteris. Cui potest opponi quod t(unc) sit multi. Quod tamen non
concedunt, <dicens quod> significat illos plures ut unum. A quibus
q(uaeritu)r, †non deest quid sit significare ut unum in h(aec)†, sic
non est necessaria ‘si non est animal, non est homo’; et haec est
necessaria ‘si non est homo, non est animal’, quae falsa est.

[4] Mag. W‘alterus’ fere idem sentit de ne(cessitate) conse(quentiae),

qui non attribuit tantum veris sed9 etiam fal(sas) dicit necessarias,
sed in alio modo.

[5] Qualiter ‘vero’ definiamus necessariam conse(quentiam) vel

Topicorum10 vel Categoriarum notulis11 inspicias.

1 quod] q(uae) O – 2 terminis] t(er)min(us)is(!) O – 3 en(un)tiat()m

i(n)tel(lectui)] eun()t() ter a.c. O – 4 quaestiones] q()s O – 5 fuerit]
fiunt i.t. et supra -unt add. -erit O – 6 quicquid] q(ua)n(do)‘q(uod)’
O – 7 substantia] sus. O – 8 consequentis] 9seq(ue)ntia O – 9 sed]
q(uam) O – 10 Topicorum vel Categoriarum] topi‘car(um)’ (ve)l
cate`goriar(um)´ O – 11 notulis + [i(n)spities p.c. et i(n)---nes a.c.] O

Unfortunately, this part is heavily corrupt. However, we can at least

see the following facts.
Joscelin asserts [3.1] that truth-conditions are different in categorical
and hypothetical sentences. The truth of categorical sentences is affected
by what happens in the world, while it is not the case with hypothetical
sentences. Joscelin’s position is so far similar to Abelard’s (cf. § 2.2.2
above), but he turns it into the thesis that all true hypothetical sentences
are always necessary (semper est omnis vera necessaria), while Abelard
says that all true hypothetical sentences are true from eternity (omnes verae
consequentiae ab aeterno sunt verae, Dialectica III p. 279.18 ed. De Rijk).
Joscelin reports three theories with respect to the question of what
are true and necessary hypothetical sentences ([3.3.2]-[3.3.4]). According
to the first theory ([3.3.2]), true and necessary hypothetical sentences are
those the antecedents of which cannot be true unless the consequents are
true; according to the second ([3.3.3]), we should add the further condition
that antecedens in themselves require consequens. The first corresponds to
what Abelard reports as necessity in a wider sense, the second to what he
reports as necessity in the strict sense (cf. § 2.2.3 above). The first is the
theory of Ulger of Angers, as I shall show below (§ 2.4.2). Abelard holds
the second theory in his Dialecitca (cf. § 2.2.3 above), but SH4 [3.3.4]

attributes to Abelard (“ma. Pe.”) the third theory, namely that understanding
(intellectus) of antecedents contains understandings of consequents. This
must have been the theory at which Abelard arrived after his Dialectica.
SH4 [4] says that Walter of Mortagne holds almost the same theory as
Joscelin’s. Walter’s theory is recorded in more detail in another text (see
§ 2.4 below).

2.4. Walter of Mortagne and the De dissimilitudine terminorum

2.4.1. The same ms, Orléans 266, contains on pp. 291a-293b another
treatise relevant to our issue, medium. It begins as follows.

Text 16: De dissimilitudine terminorum, Ms Orléans 266, p. 291a

Boethius dicit in Hypotheticis syllogismis (I, 842D9-14) quod, si

medius terminus dissimiliter enuntietur in utraque propositione,
non possunt concludi extrema nec ullus potest fieri syllogismus
nec propositio una. Triplex1 ergo `de´negat, in quo, ut videtur
verum dicere, [cum] haec triaα a media propositione, in qua medius
terminus diverso modo accipitur, removeat. Nam puto quod diverso
modo medius terminus pronuntiari intelligit aliquo de illis sex modis
quos Boethius (2 In Perherm., pp. 132.21-134.7) contra sophisticas
importunitates determinari praecipit.

1 Triplex lextio incerta, Rix i.t. et s.l. add. v- ut videtur O

α haec tria, scil. non posse concludi extrema, nec ullum
syllogismum fieri, et non unam propositionem fieri.

I shall call this text De dissimilitudine terminorum. The author explicitl

says that he follows Walter of Mortagne (see Text 17 [4] below).
After the incipit quoted above (Text 16), there follows:
I (pp. 291a-292a): some preliminary discussions.
II (pp. 292a-293a): discussions of the following seven sophisms.

(i) si nulla res est, aliqua res est. (= SH4 (7))

(ii) si omne animal est homo, quoddam animal non est homo. (= SH4 (8))
(iii) si omnis homo est lapis, quidam lapis est lapis. (= SH4 (1))
(iv) si Socrates est Plato, Socrates non est Plato. (= SH4 (2))
(v) si omnis homo est homo, quidam homo non est Socrates (= SH4 (4))

(vi) si albedo est, nigredo est. (= SH4 (5))

(vii) si omne ens est domus, nullum ens est domus. (= SH4 (3))

For each sophism, the author first mentions an accustomed proof

given to the sophism; secondly he shows how the proof was attacked
by those who held the dissimititudo theory; and finally the author gives
his own attack against the counterargument from the view-point of the
dissimilitudo theory.
III (pp. 293a-b): The author’s own solution to show that the proof of
each sophism does not hold, having a flaw different from that shown by the
dissimilitudo theory.

2.4.2. A passage from the preliminary discussions (Part I) mentions

many masters by name.

Text 17: De dissimilitudine terminorum, Ms Orléans 266, p. 291a-b

[1.1] Sed, quia de dissimilitudine loqui ex occasione huius loci

iam cepimus, dissimilitudin(um) diversitates, quae secundum
quorundam reputationes extremorum conclusionem impedire
videntur, exquiramus.

[1.2] Et, quia necessitas consequentiae praepediri dissimilitudine1

videtur, prius de necessitate consequentiarum loquamur; et, an omnis
vera <consequentia> sit necessaria, et e converso, sicut m. Gosl’ dicit
et quidam alii, seu non omnis vera sit necessaria, rimemur.

[2] M. itaque G., qui omnem, ut praemissum est, veram

consequentiam necessariam dicit, necessariam in absolutam
necessariam et determinatam necessariam dividit. Necessariam
autem absolutam illam aestimat cuius veritatem, postquam semel
est inventa, nec varietas temporum nec permutatio rerum alterat, ut
Socrates est homo.
Necessariam vero determinatam dicit quam non rerum natura
semper veram, immo veram aliquando <quando> aliquis rerum
eventus efficit, ut haec
si Socrates est Socrates, tunc Socrates sedet.
Non enim ista semper est vera, nisi tunc tantum quando Socrates

[3] M. autem Petrus non omnem veram reputat2 esse necessariam,

quia illam tantum necessariam arbitratur cuius antecedentis sententia
consequentis claudit sententiam (sententiam vero idem esse quod
intellectum dicit), ut
si est homo, est animal.
Hanc enim propositionem ‘Socrates est homo’ tres intellectus putat
habere, id est ‘Socrates est animal’ ‘Socrates est rationalis’ ‘Socrates
est mortalis’; et sic istius propositionis intellectus, quae est ‘Socrates
est homo’, huius alterius, quae est ‘Socrates est animal’, intellectum
sicut et reliquarum duarum quas continet, dicit continere.

[4] Sed m. Galt(er)ius, ut dictum est, relictis quibus in hac uti

solebat solutione circumlocutionibus, ad meliorem se contulit
sent(ent)ia<m>, quam nos tenemus. Nos autem |291b| illam veram
consequentiam sive necessariam esse dicimus, cuius dividentis
verus esse nequit intellectus.

[5] M. Vlg. et ceteri Asinarii illam necessariam consequentiam

solent dicere, cuius antecedens non potest esse sine consequente. ...
(counter-arguments against master Vlg.’s theory) ...

1 dissimilitudine] dissimilimilitudine(!) O – 2 reputat] iteputat(!) O

This passage gives a summary of what is already known to us, namely

[1] that the controversy began with the attack from the point of view of
the disimilitudo theory, [2] the theory of Joscelin of Soissons, and [3]
Abelard’s later theory. In addition, [4] it gives information on the theory of
Walter of Mortagne, namely that true and necessary hypothetical sentences
are those whose consequents’ negation cannot be true, and [5] tells us that
the theory according to which necessary hypothetical sentences are those
whose antecedents cannot be true unless their consequents are true is from
Ulger of Angers.

2.4.3. All of the seven sophisms ((i)-(vii)) discussed in Part II are

exactly the same as those enumerated in SH4 (Text 14 [1.1] above). But
the author explicitly calls them sophismata. On that point the position the
author takes is contrary to SH4, which accepts them as true sentences.
Now, (iii) in the De dissimilitudine terminorum is the same as medium
parvum and is called so by the author. Moreover its proof (cf. Text 18 [1])
runs parallel to that which is now familiar to us.

Text 18: De dissimilitudine terminorum, Ms Orléans 266, p. 292a-b

Huic etiam argumentationi parvi medii dissimilitudinem inferunt,

per quam conclusionem impediri dicunt. Cuius medii terminorum
notitia talis est:
(iii) si (A1) omnis homo est lapis, (A3) quidam lapis est
[1] Quod solet probari, sic. Si (A1) omnis homo est lapis, (A1')
omne non lapis est non homo; et sic (A1) omnis homo est lapis.
Item si (A1) omnis homo est lapis, (A2) quidam lapis est homo.
Et, si verae sunt istae duae (A2) ‘quidam lapis est homo’ et (A1)
‘omnis homo est lapis’, vera est (A3) ‘quidam lapis est lapis’. A

[2] Hanc igitur conclusionem tali dicunt infringi dissimilitudine,

qua dictitant in priori argumentatione illas duas propositiones, id
est (A2) ‘quidam lapis est homo’ et (A1) ‘omnis homo est lapis’, ad
illam quae est (A3) ‘quidam1 lapis est lapis’, ut duas consequi, et illas
easdem confirmant in sequenti consequentia ut unum antecedere.

[3] Quod qualiter unum sit, videat qui potest, quia ego nequeo
videre. Non enim, puto, eos velle quod coniunctio interposita de
duabus unam faciat, cum Boethius (2 In Perherm. II.5, p. 109.22-28)
istam ‘Iupiter tonat’ et ‘Iuno ninguit’ multiplicem, id est non unam
sed multas propositiones, esse dicat. Per quod plane velle videtur
quod per interpositam coniunctionem duae propositiones in unam
non redigantur. Si enim idcirco +illas+ ibi dicunt ut duo consequi,
hic autem velut unum antecedere, quia ibi cum duabus, hic autem
una enuntiantur prolatione, nos e converso hic cum duabus quidem
prolationibus, ibi vero una eas prolatione proferamus, ut propter
eandem causam, hic tanquam unum an<te>cedere, ibi quasi duo
consequi convincamus.

1 quidam lapis] o(mn)os h(omo) O

2.4.4. Text 18 [2] above gives a counter-argument against the proof

of (iii), and [3] rejects the counter-argument. The author of the De
dissimilitudine terminorum thus appears to agree with Joscelin of Soissons
and Abelard in that he accepts the truth of (iii) as well as the validity of its

proof. However, the author gives in Part III his own answer to the sophism

Text 19: De dissimilitudine terminorum, Ms Orléans 266, p. 293b

Hanc etiam argumentationem, per quam

(iii) si omnis homo est lapis, quidam lapis est lapis
<probatur>, sic possumus denegare, ut dicamus ex his duabus
propositionibus (A2) ‘quidam lapis est homo’ et (A1) ‘omnis
homo est lapis’ haec consequens (A3) ‘quidam lapis est lapis’ non

Quodsi quis ad hoc probandum illam <regulam> quam

praemisimus a praedic(a)to velit afferre, r(espondemus) neque
per r(egulam) aliquam consequentiam verificari concedimus,
neque ex consequentia regulam. Si enim quia vera est r(egula)
vera est consequentia, et quia non est vera consequentia nec vera
est r(egula). Quod falsum est. Potest namque vera esse regula, nec
tamen vera erit consequentia, quia nec erit consequentia nisi sit qui
eam proferat. Similiter quoque vera potest esse consequentia, nec
tamen ideo vera erit regula, cunctis tacentibus, quia nec etiam erit.
Regula enim et consequentia diversi sunt aeres nulla habitudine se

According to the author, the rule a praedicato, which was the basis
of the proof, does not confirm (iii), because the rule a praedicato and the
sentence (iii) are different airs (i.e. distinct physical phenomena), and so,
if nobody utters the rule or the sentence, the rule a praedicato can be true
while the sentence (iii) is not, and vice versa, since what is not uttered is
not true (since it is not existent). Strange assertion! By the same token, one
could deny any inferences by any rules! However, when the author says
that the rule and the sentence are different airs, he presupposes the vox-in-
substantia theory, which was the issue at stake in the early 12th century9.
Now, SH4 reports that Walter of Mortagne held almost the same
theory as Joscelin of Soissons, but that Walter attributed necessity not only
to true sentences but also to false ones (Text 15 [4] above). What it means

With respect to the vox-in-substantia theory and the controversy over it, see
IWAKUMA, «Vocales Revisited», pp. 81-171, at §§ 5.3.1-5.3.4, and pp. 113-116.

seems to be that Walter accepts (iii) as necessary, and that (iii) can at the
same time be a non-true sentence or deprived of any truth-value (as well as
of any other properties), viz. when it is not uttered.

3. The Parisian treatise De mediis

We have one more text dealing with media in ms Paris, BnF, lat.
13368, f. 194va-vb, which begins:

Text 20: Tractatus Parisiensis de mediis, Ms Paris, BnF, lat. 13368,

f. 194va

[1] R(egula). Haec est natura cuiusdam medii:


[2] ut hic: Si (¬A) nullus homo est asinus,

tunc, si (B) omnis albedo est albedo,
(C) quaedam quant(itas) non est linea.

[3] Quod probatur. ...... proof ......

There aren’t any clues as to the identity of the author, but it is highly
probable that this text was written at Paris in the first decades of the 12th
century, just as is the case with the other texts copied in the same ms ff.
This treatise deals with the following eight hypothetical sentences:

(1) *si nullus homo est asinus,

tunc, si omnis albedo est albedo, quaedam quantitas non est
(2) si (X) omnis homo est lapis, tunc,

For the most recent description of ff. 128-231 of this ms, see Y. IWAKUMA,
«Pseudo-Rabanus super Porphyrium (P3)», Archives d’Histoire Doctrinale et Littéraire
du Moyen Âge, 75 (2008) 43-196, at pp. 45-47.

si (Y) omnis homo est homo, (X) omnis homo est lapis.
(3) *si omnis homo est lapis, tunc, si quidam homo est homo,
quidam lapis est lapis.
(4) si, quia (X) omnis homo est animal, sequitur
‘si (Y) Sorates est homo, (Z) Socrates est animal,
tunc, si quia (Z) Socrates est animal, (W) omnis lapis est
et, quia <(V) omnis homo est lapis,
(X) omnis homo est animal,
... lacuna ....
(X ¨ (Y¨Z)) ¨ (((Y¨W)&(V¨X)) ¨ (V ¨ (Y¨W))11
(5) si, quia (X) omnis homo est lapis, inde sequitur
‘si (Y) omnis lapis est ainus, (Z) omnis homo est asinus’,
tunc, si quia (W) omnis equus est equus, (Y) omnis lapis est
et, quia (X) omnis homo est lapis, sequitur
‘si (W) omnis equus est equus, (Z) omnis homo est
(X ¨ (Y¨Z)) ¨ ((W¨Y) ¨ (X ¨ (W¨Y))
(6) si, quia (X) omnis homo est animal, sequitur
‘si (Y) Socrates est homo, (Z) Socrates est animal’,
tunc, si, quia (Z) Socrates est animal, (W) omnis lapis est
et, quia (X) omnis homo est animal, sequitur
‘si (Y) Socrates est homo, (W) omnis lapis est lapis’.
(X¨( Y¨Z)) ¨ ((Z¨W)¨(X¨(Y¨W))
(7) si, quia (X) omnis homo est animal, sequitur
‘si (Y) Socrates est homo, (Z) Socrates est animal,
tunc, si, quia (W) omnis asinus est asinus, (Y) Socrates est
et, quia (X) omnis homo est animal, sequitur
‘si (W) omnis asinus est asinus,
(V) omnis equus est equus’.
(X¨(Y¨Z)) ¨ ((W¨Y) & (Z¨V))¨(X¨(W¨V)))
(8) si, quia (X) omnis homo est animal, sequitur
‘si (Y) Socrates est homo, (Z) Socrates est animal’,
et, quia (Z) Socrates est animal,

(V ¨ (Y¨W)) is my addition, supplying the lacuna with ‘sequitur: quia (V)
omnis lapis est lapis, si (Y) Socrates est homo, (W) omnis lapis est lapis’.

‘si (Y) Socrates est homo, (X) omnis homo est

(X ¨ (Y¨Z)) ¨ ((Z¨X) ¨ (Z ¨ (Y¨X));

giving each sentence Regula [1] and a proof [3] to each sentence [2], just
as is the case with (1) (Text 20 above).

3.1. Pecuriality of the Parisian Media

This treatise is of a different character from those so far discussed in

this paper. It does not discuss any media common to the others. Among
the eight sentences discussed here, only the first three ((1)-(3)) bear some
similarity with the media known so far, in that they have a form X¨(Y¨Z),
and that their categorical constituents are nonsense such as ‘omnis homo
est lapis’ or ‘omnis homo est homo’. The last four ((5)-(8)), on the other
hand, have much more complicated forms, and all are, together with (2),
tautologies from the modern point of view, albeit some of categorical
constituents are again nonsense. That might also be the case with (4);
indeed, although its discussion is incomplete (it ends with a lacuna), it can
very well be a tautology if we fill the gaps as suggested in the list above.

3.2. William of Champeaux’s discussion of syllogisms

In order to understand this Parisian treatise on mediis, we first have

to note that the discussion of syllogisms in the Vienna and Escorial
Introductiones is unusual. Both the Introductiones deal with syllogisms as
part of discussions of hypothetical propositions. That is to say, categorical
syllogisms are discussed in such a form as

si omne S est M, tunc, si omne M est P, omne S est P,

while Boethius discusses them in the form

omne S est M, omne M est P, ergo omne S est P.

The same applies to hypothetical syllogisms.


A closer examination shows further intriguing features of the

discussion of categorical syllogisms in the Introductiones.
There aren’t any references to the Aristotelian-Boethian distinctions
of modes and figures. Instead, the two Introductiones divide categorical
syllogisms according to the distinction of whether the inferential force (vis
inferentiae) is in the subject term or in the predicate. Accordingly, some
valid syllogisms are discussed twice, once as those whose inferential force
lies in the subject term, and once more as those whose inferential force lies
in the predicate12.
The two Introductiones do not touch upon some of the valid syllogisms
discussed in Boethius’ De syllogismis categoricis13. At the same time, the
Escorial Introductiones discuss Camestrop and Camenop14, which are not
mentioned in Boethius’ De categoricis syllogismis, nor in Aristotle’s Prior
Analytics, nor in any other medieval discussion of syllogisms. Moreover,
some invalid syllogisms are affirmatively mentioned in the Escorial
Among syllogisms, both the Introductiones discuss strange non-
syllogistic sentences of such forms as

i) si omne S est P, tunc si omne P est S, omne S est P. (Vienna 5.1

(2), Escorial 8.2)
ii) si omne S est P, tunc, si quoddam P non est S, quoddam S non est
P. (Escorial 8.7)
iii) si omne S est P, tunc, si nullum P est P, nullum S est S. (Vienna
5.1 (5), Escorial 8.3)
iv) si omne S est P, tunc, si quoddam S est S, quoddam P est P.
(Vienna 5.2 (3), Escorial 8.8)

We should note here that i) has the same form as medium (2) of the
Parisian treatise, although the term for P is different.

The first mode of the first figure is repeated in the Vienna Introductiones 5.1(1)
and 5.2(1), in the Escorial Introductiones 8.1 and 8.10. The second mode of the first
figure in the Escorial 8.4 and 8.23. The forth mode of the first figure in the Escorial
8.14 and 8.23. And the forth mode of the first figure in the Escorial 8.5 and 8.26.
The Vienna Introductiones discuss only syllogisms I.1-3, II.2 and 4, III.1 and 3.
The Escorial treatise lacks the discussions of I.5 and 6, and III.1.
For Camestrop, see the Escorial Introductiones 8.21; and for Camenop, see the
Escorial Introductiones 8.25.
It is the case with Escorial 8.15, 8.16, and 8.27; cf. also 8.19.

For hypothetical syllogisms too, the treatment of the two Introductiones

is completely different from that in Boethius’ De syllogismis hypotheticis.
Boethius first distinguishes different types of hypothetical sentences and
then makes various syllogisms with the modus ponens and modus tollens
for each type of hypothetical sentences. For example, he construes twofold
syllogisms from hypothetical propositions of the form ‘si est A, est B’, as

si est A, est B; sed est A; ergo est B,

or si est A, est B; sed non est B; ergo non est A.

The two Introductiones, on the other hand, give various rules without
proof, such as

quicquid sequitur ad consequens, sequitur ad antecedens.

A ¨ C, C ¨ X ⊢ A ¨ X.

This rule, which they call ab antecedenti, is valid from our modern
point of view, but it is never found in Boethius’ works, nor is it a hypothetical
syllogism in the Boethian sense at all. In such a rule the antecedent and the
consequent can well have different subjects, but Boethius deals only with
syllogisms whose categorical constituents have the same subjects.
As I have shown elsewhere16, the two Introductiones were so influential
that many texts were later written on their model; but in all of later works,
syllogisms, both categorical and hypothetical, are discussed faithfully
following Boethius. This shows that the strange discussion in the two
Introductiones came soon to be forgotten and replaced by the Boethian one.
These facts strongly suggest that the doctrine of syllogisms contained
in the two Introductiones was developed under circumstances in which

Y. IWAKUMA, «William of Champeaux and the Introductiones», in H. A. G. – C.
H. KNEEPKENS (eds.), Aristotle’s Peri Hermeneias in the Latin Middle Ages: Essays on
the Commentary Tradition, Ingenium Publishers, Groningen-Haren 2003, pp. 1-30, at
§ 2.1, p. 3 (cf. also n. 10, p. 5). I can add the following to those five mentioned there: 6)
Introductiones Lemovicensis, Paris lat. 544, ff. 110r-114r; 7) Introductiones Duacenses,
Douai 749, ff. 101va-13vb (ed. Y. IWAKUMA, «Introductiones Duacenses in dialecticam:
An Edition», Veritas 15 (1996), pp. 21-35); 8) Introductiones Parvipontanae, Berlin,
lat. oct. 262, ff. 1ra-7rb; Introductiones Emmeranae, München, clm 14735, ff. 21r-22v;
and Introductiones Parisienses, Paris, nouv. acq. lat. 3239, ff. 1ra-19vb.

there was little knowlegde of Boethius’ monographs on syllogisms. It is

true that the De syllogismis hypotheticis is referred to in the Vienna treatise
(see Text 3 above), and that, by way of consequence, the De syllogismis
hypotheticis was already known when the Vienna treatise was written; but
the doctrine itself was obviously formed without knowledge of Boethius’
It might perhaps be the case that Peter Abelard brought the knowledge
of Boethius’ monographs on syllogisms to Paris. The monographs were
known well within the proto-vocalists sect17. A treatise written in the syle of
the Introductiones preserved in the ms Paris lat. 544 ff. 110r-114r18 contains
a genuinely Boethian discussion of syllogisms, and it was certainly written
by a proto-vocalist in the late 11th century.

3.3. Parisian Media and William’s discussion of syllogisms

Going back to the Parisian treatise on mediis, many media discussed

there present a remarkable similarity with hypothetical sentences discussed
in the two Introductiones, as is shown in the following table.

Wien II Escorial I Paris

- 8.2 =2 X ¨ (Y¨X)
3.1 (1) 9.1 A¨C ⊢ (X¨A) ¨ (X¨C)
(2) 9.4 (A¨C) ¨ ((X¨A) ¨ (X¨C))
3.2 (1) 9.2 A¨C ⊢ (C¨X) ¨ (A¨X)
(2) 9.5 (A¨C) ¨ ( (C¨X) ¨ (A¨X) )
(3) 9.6 A¨C ⊢ ¬C¨¬A
3.3 (1) 9.3 X¨A, C¨Y ⊢ (A¨C) ¨ (X¨Y)
(2) 9.8 ((A¨C)∧(X¨A)∧(C¨Y)) ¨ (X¨Y) *
(3) - (X¨Y) ¨ ((Y¨Z) ¨ (X¨Z)) *
- 9.7 (A¨C) ¨ ((C¨A) ¨ (A¨C)) *
4 (1) - ((X¨Y)∧(Z¨W)) ¨ ((Y¨Z) ¨ (X¨W))
(2) 9.9 =5 (X ¨ (Y¨Z)) ¨ ((W¨Y) ¨ (X ¨ (W¨Y))
- - (8) (X ¨ (Y¨Z)) ¨ ((Z¨X) ¨ (Z ¨ (Y¨X))
(3) 9.10 =6 (X ¨ (Y¨Z)) ¨ ((Z¨W) ¨ (X ¨ (Y¨W)))
(4) - (X ¨ (Y¨Z)) ¨ ( ((W¨X)∧(V¨Y)) ¨ ((W ¨ (V¨Z)) )

Instead, the proto-vocalist sect had likely little knowledge of Boethius’
commentary on the Categories. Cf. IWAKUMA, «Vocales Revisited», p. 100.
For this ms, see IWAKUMA, «Vocales Revisited», pp. 86-89.

- - (7) (X ¨ (Y¨Z)) ¨ ( ((W¨Y)∧(Z¨V)) ¨ ((X ¨ (W¨V)) )

(5) - (X ¨ (Y¨Z)) ¨ ( ((W¨X)∧(Z¨V)) ¨ ((W ¨ (Y¨V)) )
(6) - (X ¨ (Y¨Z)) ¨
( ((W¨X)∧(V¨Y)∧(Y¨U)) ¨ (W¨(V¨U)) )
- - (4) (X ¨ (Y¨Z)) ¨ ( ((W¨Y)∧(Z¨V)) ¨ ((X ¨ (W¨V)) )

It is remarkable that all the sentences listed here are tautologies in the
sense of our modern classical propositional logic.
The two Introductiones did not give proofs for any of them, but the
Parisian treatise gives thourough proofs for the last four media (for medium
(4) there is no proof, but a lacuna in the ms). For example, the proof of (5)
runs as follows.

Text 21: Tractatus Parisiensis de mediis, Ms Paris, BnF, lat. 13368,

f. 194va

[1] ‘R(egula)’:

[2] ut hic si, quia (X) omnis homo est lapis,

inde sequitur:
si (Y) omnis lapis est asinus,
(Z) omnis homo est asinus,
tunc, si quia (W) omnis equus est equus, (Y) omnis lapis est
et, quia (X) omnis homo est lapis,
‘si (W) omnis equus est equus,
(Z) omnis homo est asinus’.
[3] Quod sic probatur
[3.1] si quia (X) omnis homo est lapis,
inde sequitur:
si (Y) omnis1 lapis est asinus,
(Z) omnis homo est asinus,
tunc, si quia si (Y) omnis lapis est asinus,
(Z) omnis homo est asinus,
inde sequitur:
si (W) omnis equus est equus,

(Z) omnis homo est asinus.

et quia (X) omnis homo est lapis,
si (W) omnis equus est equus,
(Z) omnis homo est asinus.
A conseq(uenti), ubi nihil restat.
[3.2] Et <ideo>
si quia [si quia] hoc est, illud est,
et quia si (W) omnis equus est equus,
(Y) omnis homo est asinus,
sequitur illud idem consequens2.
[3.3] Probandum est quod illa conse(quentia)
si (W) omnis equus est equus, (B) omnis lapis est asinus
sit antecedens ad istam
si quia si (Y) omnis lapis est asinus,
(Z) omnis homo est asinus3,
et quia (W) omnis equus est equus,
(Z) omnis homo est asinus.
Quod probatur per locum a consequenti, ubi nihil restat. Et sic
probata est.

1 omnis + [h(om)o] P – 2 consequens] 9seq(ue)nti P – 3 asinus

repet. i.m. P

We can formalize [2] and [3] as follows:

[2] (X¨(Y¨Z)) ¨ ( (W¨Y) ¨ (X ¨ (W¨Z)) )

[3.1] (X¨(Y¨Z)) ¨ ( ((Y¨Z) ¨ (W¨Z)) ¨ (X¨(W¨Z)) ) a cons.
[3.2]=[2] ∴ (X¨(Y¨Z)) ¨ ( (W¨Y) ¨ ((X¨(W¨Z)) )
[3.3] ∵ (W¨Y) ¨ ( (Y¨Z) ¨ (W¨Z) ) a cons.

When we presuppose the deduction theory, namely

P ⊢ Q ⇒ ⊢ P¨Q,
[3.1] and [3.3] are themselves valid by the rule a consequenti, viz.
A¨C, C¨P ⊢ A¨P,

which is shown by the words ubi nihil est (‘there is nothing to prove
further’). Replace A with X, C with Y→Z and P with W→Y in [3.1]; and
A with W, C with Y and P with Z in [3.3]. The rule a consequenti validates,
too, that [2]=[3.2] follows from [3.3] and [3.1]. Replace A with X¨(Y¨Z),
C with W¨Y and P with X¨(W¨Z).

All the proofs of (6)-(8) proceed likewise, only the following three
rules being unproved:
Ab antecedenti A¨C, P¨A ⊢ P¨C
A consequenti A¨C, C¨P ⊢ A¨P
Ab antecedenti et consequenti
A¨C, P¨A, C¨Q ⊢ P¨Q;

(the deduction theory is always presupposed tacitly). It might also be the

case with (4), the discussion of which, ending on a lacuna, is incomplete.
In this way, we may reckon the parts (4)-(8) as a kind of formalization of
propositional logic without axioms and with three deductive rules. In fact,
employing the same method, it would be possible in principle to produce a
group of infinite tautological propositions.
It is, however, impossible to apply this view to the parts (1)-(3). Be
that as it may, historically speaking, there wasn’t any further significant
development in the direction of such a formalization. Attempts of this kind,
provided there were any, were soon forgotten and replaced by Boethius’
theory of hypothetical syllogisms, which is essentially different from our
modern propositional logic.


Without knowing Boethius’ monographs on syllogisms, William of

Champeaux developed a discussion of media or true if-sentences proved
by per medium arguments, giving special names such as medium parvum
etc. to those of them which are true in spite of the fact that they have
false antecedents and/or false consequents. Peter Abelard arrived at Paris
around the turn of the century, perhaps with the knowledge of Boethius’
monographs on syllogisms. William’s discussion of media came to be
attacked by the dissimilitudo theory, which was supported by Boethius’
De syllogismis hypotheticis. This attack provoked controversies, involving
many younger masters like Peter Abelard, Joscelin of Soissons, Walter
of Mortagne, Ulger of Angers, and others; and through the controversy
they each deepened their thought on the nature of logical entailment. This
polemical situation offered a possibility of developing something like the
modern formalization of propositional logic. But the possibility declined
soon, with the acceptance of Boethius’ theory of syllogisms.



Buridan defines a formal consequence as one in which no categorematic

terms occur essentially –one that remains valid no matter what the matter,
provided we keep the form the same1. A material consequence, by contrast, is
one that fails to hold «in all terms, keeping the form alike». So, for example,

(1) A man runs. Therefore an animal runs.

is a material, but not a formal consequence, because the consequence can

be destroyed by substituting «horse» for «man» and «wood» for «animal».
Similar definitions of formal and material consequence can be found in
Pseudo-Scotus and Albert of Saxony2.
Two things are striking about this medieval definition. The first is
that it is very close to the modern conception of formal consequence one
finds in Bolzano and Tarski3. The second is that although Buridan and the
other fourteenth-century logicians state these distinctions with a great deal
of precision, they say almost nothing about the point of the distinctions.
Why are these distinctions drawn in the way they are? What philosophical
purpose do they serve?
Although the resemblance of the medieval distinction to the modern
one has been widely noted, the lack of explicit motivation has not. I suspect
that is because we find the distinction so familiar and natural that we do
not not pause to think about what motivates it. But the question should be a

Department of Philosophy, University of California, Berkeley, 314 Moses Hall,
Berkeley, CA 94720-2390, USA. Email: jgm@berkeley.edu.
Tractatus de Consequentiis, Ed. by H. HUBIEN, Publications Universitaires,
Louvain 1976, I.4.
I. M. BOCHENSKI, A History of Formal Logic, transl. by I. THOMAS, Notre Dame
University Press, Notre Dame 1961.
B. BOLZANO, Wissenschaftslehre, Felix Meiner, Leipzig 1929, 2nd ed.; A. TARSKI,
«The Concept of Truth in Formalized Languages», in J. CORCORAN (ed.), Logic,
Semantics, Metamathematics, Hackett, Indianapolis 1983, pp. 152-278.

live one for us, as well. We can sort good inferences into those that can be
turned into bad ones by uniform substitution of nonlogical vocabulary, and
those that cannot. But what is so special about the inferences in the latter
class, the ones we call «formally valid»? Consider this pair:

(2) Snow is white, everything white is colored; therefore, snow is

(3) Snow is white. Therefore, snow is colored.

Is my knowledge that (2) is a good inference any more certain or apriori than
my knowledge that (3) is a good inference? Presumably not. Why, then, do we
care about the distinction between formally and materially valid inferences?
Is it just that the formally valid inferences are more amenable to systematic
treatment? But then, thinking of them as valid in a special way –formally
valid– would be akin to thinking of stars that can be studied using terrestrial
telescopes as a special kind of stars –telescopically accessible stars4.
Those who want to avoid this deflationary conception might be
tempted to something like the following line of thought. An inference like
(3) owes its validity to a fact about the world –the fact that whatever is
white is colored. Not a very exciting fact, perhaps, and one that can be
known apriori –but a fact nonetheless. In contrast, the validity of (2) does
not depend on any fact about white things, snow, or anything else. (2) is
valid entirely in virtue of its form or construction; its validity does not
depend on any fact about the world, however general. Generalizing, we
might say that formally valid inferences are inferences whose validity is
not grounded in any fact about the world. Of course, the claim that there
are such inferences is a substantive one. Let us call it the

Formality Thesis: There are inferences whose validity is entirely

grounded in their forms, and does not depend on any fact about the

For worries of this kind, see BOLZANO, Wissenschaftslehre, § 186; J. A. COFFA,
«Machian Logic», Communication and Cognition, 8 (1975) 103-129; J. BARWISE –
S. FEFERMAN (eds.), Model-Theoretic Logics, Springer-Verlag, New York 1985, p. 6;
J. ETCHEMENDY, «The Doctrine of Logic as Form», Linguistics and Philosophy, 6 (1983)
319-334; S. READ, «Formal and Material Consequence», Journal of Philosophical
Logic, 23 (1994) 247-265.
As stated, the Formality Thesis is ambiguous between a metaphysical thesis
(there are inferences that are valid solely in virtue of their forms, and not at all in

The Formality Thesis goes with a view of logic that seems quite natural.
On this view, the role of logic is to help us make explicit everything on
which an inference depends. When we have teased out hidden assumptions
to the point where we have a formally valid argument, then we know that
the process of explicitation has come to an end; we have made all of the
assumptions on which the inference depends explicit. We could not think
of logic this way if the Formality Thesis did not hold, since there would
always be further facts on which the validity of an inference depends,
which could in principle be made explicit as further premises.
As natural as it may seem, though, the Formality Thesis is a substantive
and controversial thesis. It does not follow immediately from the fact that
inference (2) holds «in all matter» –that is, on all uniform substitutions of
nonlogical terms– that it is good solely in virtue of its form, independently
of any facts about the world. Although the schema

(4) X is prime. X > 2. Therefore, X is odd.

yields a good inference for every substitution for X, the goodness of these
inferences is grounded in a fact about primes –namely, that every prime
greater than 2 is odd. So clearly the inference from «every instance of the
schema is a good inference» to «the instances are good solely in virtue
of being instances of the schema» is not a good one in general. Maybe it
is cogent when we restrict ourselves to schemata of a certain kind –those
in which the only fixed terms are «logical constants». But if so, that is
something that requires showing.
Which brings me, at last, to Abelard. Abelard interests me because,
unlike later medieval logicians, he offers an argument for the Formality
Thesis –a thesis his predecessors seem to have rejected. This argument
motivates his distinction between formal and material consequence –or,
in his terminology, perfect and imperfect entailment (inferentia6). He does
not slide from «good in all matter» to «good in virtue of form»; indeed,
he recognizes inferences that are good in all matter but depend for their

virtue of any facts about the world) and an epistemological thesis (there are inferences
that we can know to be valid solely on the basis of knowledge of their forms, and
independently of any knowledge of facts about the world).
In translating inferentia as «entailment», I follow C. MARTIN, «Logic», in
J. E. BROWER – K. GUILFOY (eds.), The Cambridge Companion to Abelard, Cambridge
University Press, Cambridge 2004, pp. 158-199.

validity on facts about the world. He gives a much subtler argument –and
it is a good argument, in its philosophical context. But it is not an argument
that can give us a good reason to accept the Formality Thesis.

Perfect and imperfect inferentia

An entailment is perfect, Abelard says, when

[…] from the structure (complexio) of the antecedent itself, the truth
of the consequent is manifest, and the construction (constructio) of
the antececent is so disposed that it contains also the construction
of the consequent in itself, just as in syllogisms or in conditionals
which have the form of syllogisms. (253.31-254.1)7

For example,

(5) If every man is an animal and every animal is alive, every man
is alive. (254.35)8

An entailment is imperfect, by contrast, when the connection between

antecedent and consequent takes its necessity «from the nature of things»
(ex rerum natura, 255.7-8), not from the construction of the antecedent and
consequent, as in

(6) If every man is an animal, every man is alive. (255.3)

Both perfect and imperfect entailments require a necessary connection

between antecedent and consequent –indeed, the sense of the consequent must
be contained in the sense of the antecedent (283.37-284.8). The difference is
not in the strength of the modal connection (255.12-13), but in its ground.
Given these definitions, the thesis that some entailments are perfect
amounts to the Formality Thesis9.

All parenthetical references are to Abelard’s Dialectica, Ed. by L. M. DE RIJK,
Van Gorcum, Assen 1956. Unless noted, translations from Abelard are my own.
Although Abelard is aware of the difference between arguments and conditionals,
he applies the concepts inferentia and consequentia, as well as the perfect/imperfect
distinction, to both (giving examples in both forms).
Probably in its epistemological variant. Abelard holds that the truth of «if it is
man, it is animal» does not depend on the existence of either man or animal: like all true

The dialectical background

Abelard’s discussion of perfect and imperfect entailments takes place in

the section of the Dialectica entitled «Of Topics». That may seem surprising,
since in Aristotle the topics are means for the discovery of valid syllogisms,
rather than grounds for their validity10. However, by late antiquity it had
become standard to conceive of the Topical maxims as axioms on which the
validity of arguments might rest. In De Topicis Differentiis Boethius defines
a maximal proposition as a «maximal, universal, principal, indemonstrable,
and known per se proposition, which in argumentation gives force to
arguments and to propositions»11. For example, the maxim

(7) Whatever is present to the genus is present to the species

«supplies force to» the argument below and «makes [it] complete from
without» (1188B-C):

consequences, it is an eternal truth (279.18). But if man and animal did not exist, then
(as will be explained later in this essay) there would be no locus differentia and hence
no Topical grounding. So the fact that man is species of animal cannot be the cause of
the entailment (consecutio) but only its proof (probatio) (265.10-12). This suggests
that what distinguishes perfect inferences from imperfect ones is a special epistemic
character: their validity can be known independently of all knowledge about the world.
For this view of Aristotle’s Topics, see J. ALLEN, «The Development of
Aristotle’s Logic: Part of an Account in Outline», in J. J. CLEARY – W. C. WIANS (eds.),
Proceedings of the Boston Area Colloquium in Ancient Philosophy, 11 (1995), p. 189;
E. STUMP, Boethius’s De Topicis Differentiis, Cornell University Press, Ithaca 1978,
pp. 168-177; N. J. GREEN-PEDERSEN, The Tradition of the Topics in the Middle Ages:
the Commentaries on Aristotle’s and Boethius’ «Topics», Philosophia Verlag, Munich –
Vienna 1984, p. 23. A contrasting view understands Aristotelian topical maxims as
quasi-logical laws that ground the validity of the inferences to which they are applied:
J. BRUNSCHWIG, Aristote: Topiques, vol. I, Société d’Édition «Les Belles Lettres», Paris
1967, pp. XL-XLI; W. A. DE PATER, «La fonction du lieu et de l’instrument dans les
Topiques», in Aristotle on Dialectic: The Topics (Proceedings of the Third Symposium
Aristotelicum), Oxford University Press, Oxford 1968, pp. 166, 174. But Aristotle does
not seem to think that the acceptability of the particular syllogisms that fall under a
topic depends on the acceptability of the general rules he gives; if he did, it would be
difficult to make sense of the fact that he often acknowledges counterexamples to the
maxims (e.g. at Top. 115b14, 117a18, 117b14, 121b30, 123b17, 124b19, 128b6).
STUMP, Boethius’s De Topicis Differentiis 1185B. This approach to the Topics
goes back at least to Themistius. S. EBBESEN, Commentators and Commentaries on
Aristotle’s Sophistici Elenchi, vol. 1, E. J. Brill, Leiden 1981, p. 118.

(8) Every virtue is advantageous. Justice is a virtue. Therefore,

justice is advantageous.

Commentators have wondered how the maxim here can play the role
Boethius assigns it, of supplying force to the argument and completing it
from without, when (8) has the form of a valid categorical syllogism12. But
the problem only arises if we assume that valid categorical syllogisms are
distinguished from other forms of arguments by the fact that they require
no external validation. Boethius shows no sign of accepting this view.
Sten Ebbesen claims that for Boethius as for Galen, «every inference
owes its cogency to an axiom»:

The implication of the Boethian theory would seem to be that

all proof proceeds, implicitly or explicitly, by instantiation and
detachment and, as some medievals saw, that a categorical syllogism
is not anything sui generis, as it depends on a law of inference of
the same type as the ones that licence inferences involving other
relations than plain predication13.

It is not clear that Boethius himself accepts all these implications. But
eleventh and twelfth-century logicians, whose main sources for syllogistic
theory were the works of Boethius, embrace them explicitly. In his
Dialectica, Garlandus Compotista says that the theory of Topics is prior
to the theory of categorical and hypothetical syllogistic, and all syllogisms
are ratified by topical maxims (per maximam propositionem sillogismus
approbatur)14. For instance, the syllogism

(9) Every animal is a substance. Every man is an animal. Therefore,

every man is a substance.

depends on the maximal proposition

GREEN-PEDERSEN, The Tradition of the Topics, pp. 68-69; cf. STUMP, Boethius’s
De Topicis Differentiis, pp. 183-184; Abelard, Dialectica, 257.34-258.9.
S. EBBESEN, «Ancient Scholastic Logic as the Source of Medieval Scholastic
Logic», in N. KRETZMANN – A. KENNY – J. PINBORG (eds.), The Cambridge History of
Later Medieval Philosophy, Cambridge University Press, Cambridge 1982, p. 112.
Garlandus Compotista, Dialectica, Ed. by L. M. DE RIJK, Van Gorcum, Assen
1959, 86.13.

(10) that which is universally attributed to the whole is [also

universally attributed] to the part (quod universaliter attribuitur
toti, et parti)15.

More generally, «categorical syllogisms are aided by the Topics from the
whole and from the part and from an equal» (114.18)16.
The early twelfth-century works on the Topics collected in Logica
Modernorum17 seem to follow Garlandus in taking all syllogisms to be
validated by topical maxims. The Introductiones dialectice Berolinenses,
for instance, take all syllogisms in the mood Barbara to be licensed by the
following topical maxim:

(11) If something is predicated universally of something, then if

something else is predicated universally of the predicate, that
same thing is predicated universally of the subject18.

Abelard’s picture of the Topics is largely taken over from this

Boethian tradition. The function of a Topic, according to Abelard, is to
confer inferential force on an entailment by grounding it in a real relation
among the things to which its terms refer (256.35-257.1). For example,
the conditional «if it is a man, it is an animal» is justified by the Topic
from species, since man is a species of animal, and we know that genus
necessarily applies to species (257.4-5). Following Boethius, Abelard
takes a Topic to have two components: a locus differentia and a maxima
propositio. The locus differentia (henceforth Differentia) is «that thing in
the relation of which to something else the soundness of the entailment
consists» (ea res in cuius habitudine ad aliam firmitas consecutionis

Ibid., 92.29.
Garlandus was anticipated in this view by Abbo of Fleury (945-1004) and other
early commentators on the Boethian Topics (see N. J. GREEN-PEDERSEN, The Tradition
of the Topics, pp. 144, 152). Green-Pedersen summarizes the pre-1100 works by
saying that they take the Topics to be an «[…] ‘underlying logic’ which shows or
explains why the arguments are valid» (p. 160). See also E. STUMP, Dialectic and Its
Place in the Development of Medieval Logic, Cornell University Press, Ithaca 1989,
p. 87; E. STUMP, «Topics: Their Development and Absorption into Consequences»,
in N. KRETZMANN – A. KENNY – J. PINBORG (eds.), The Cambridge History of Later
Medieval Philosophy, op. cit., p. 277.
L. M. DE RIJK, Logica Modernorum, Van Gorcum, Assen 1962-1967.
STUMP, Dialectic and Its Place, p. 116.

consistit, 263.7-8). Although the Differentiae are things, not relations,

they count as Topical Differentiae only insofar as they stand in relations to
other things19. In the example, the Differentia is man, which stands in the
species relation to animal. The maxima propositio (henceforth maxim) is a
general proposition justifying an inference from an antecedent proposition
containing a term for the Differentia to a consequent proposition containing
a term for the thing to which it is related. In the example, the maxim is
«of whatever the species is predicated, so is the genus» (de quocumque
praedicatur species, et genus, 263.18).
Abelard’s strikingly original move is to insist that some entailments do
not stand in need of topical grounding at all ([…] quia ita in se perfectae sunt
huiusmodi inferentiae ut nulla habitudinis natura indigeant, nullam ex loco
firmitatem habent, 256.34-5). Perfect entailments, he says, do not «take their
truth […] from the nature of things» (256.21-2). A sign of this independence
from things, Abelard claims, is that perfect consequences remain true
in «whatever terms you substitute» (255.32-3), whereas an imperfect
consequence «depends on the nature of things» and does not «remain true in
any terms whatsoever, but only in those which preserve the nature of the entail-
ment» (356.8-10). For example, the entailment in «if it is man, it is animal»
can be destroyed by replacing «man» or «animal» with «stone» (356.15-19).

Therefore those consequences are correctly said to be true from the

nature of things of which the truth varies together with the nature of
things. But those [consequences] of which the construction preserves
its necessity equally in any things at all, no matter what relations
they have, take their truth from the construction (complexione), not
from the nature of things… (256.20-23)

This is all that later medieval writers typically say about the distinction
between formal and material consequence: formal consequences hold «in
all terms». But Abelard cannot stop here, for as we have seen, the dominant
view at the time he is writing –and a view he explicitly attributes to Boethius
and Porphyry (257.32-258.13)– is that categorical syllogisms and other
perfect entailments are grounded in Topics. A proponent of such a view
could grant that syllogisms preserve validity in all substitution instances,
and maintain either that

GREEN-PEDERSEN, The Tradition of the Topics, p. 167.

1. for each instance, there is a Topic grounding the entailment in

some specific relation that holds between the things it concerns, or
2. there is a single Topic that grounds all of the instances in some
very general relations that hold between things.

Abelard offers arguments against both approaches (258-262 in his

treatment of inferences, 352-365 in his treatment of conditionals). It is a
measure of the success of these arguments, I think, that they do not get
repeated: it becomes customary in later medieval manuals to infer from
an inference’s being good «in all terms» to its being good «in virtue of its
construction» and not in virtue of the nature of things. But as we have seen,
Abelard cannot take this inference for granted. Indeed, he does not even
think that it is unrestrictedly valid. He claims that the consequence

(12) If it is alive, it is alive,

which certainly holds in all substitution instances, is not perfect in its

construction (ad inferentis constructionem): one would have to add the premise
«[…] and everything that is alive is alive» (255.19-27)20. Evidently, then, there
is more to perfection than mere preservation of validity «in all terms».
Abelard makes this point explicitly in his discussion of the hypothetical
syllogisms. Boethius had taken certain instances of what we now call
«affirming the consequent» to be valid by virtue of «the nature of the
things, in which alone these propositions can be asserted»21. For example,
in the inference

(13) If it is not a, it is b; but it is a; thus it is not b,

Boethius claims, the major premise can only be true when the terms a and
b are contraries, like «day» and «night». But when a and b are contraries,

This means that the one-premise conversion inferences necessary for the
reduction of second- and third-figure syllogisms to the first figure cannot count as
perfect. It seems odd that the validation of second- and third-figure syllogisms, which
are perfect in Abelard’s sense, should require the use of an imperfect inference. Does
Abelard ever discuss this issue?
L. OBERTELLO, De hypotheticis syllogismus di A.M. Severino Boezio: Testo,
traduzione, Logicalia; testi classici di logica, Paideia, Brescia 1969, II.ii.4-5.

and it is a, then it follows that it is not b. Hence the inference is valid for
all substitution instances in which the premises are true. In fact, Abelard
thinks Boethius’ claim that «if it is not a, it is b» can only be true when a
and b are contraries is simply wrong: «if it is not a, it is b» can be true, he
notes, when a is «animal» and b is «non-man» (499). But he goes on to
say that, even if Boethius were right that a and b had to be contraries, and
thus that no formal counterexample to the inference could be given, this
fact would not show that such inferences are syllogisms (and hence perfect

Even if it were possible, whenever the consequent were affirmed,

necessarily to affirm the antececent from any property whatever
–nevertheless there would be no form of syllogism in which, the
consequent having been affirmed in this way, one could affirm the
antecedent, or the antecedent having been denied, one could deny
the consequent, since the entailment of a syllogism is supposed to
be so perfect that no relation of things pertains to it. (502.19-25).

To say that an entailment is «perfect» is to say that our knowledge of its

validity is completely independent of our knowledge of «the nature of
things». Even if Boethius were right that (13) held in all terms for which
the premise could be true, that would not be something we could know
without knowing something about «the nature of things» –the relations
of contraiety between a and b. An entailment that holds in all terms, then,
need not be good in virtue of its construction.

Abelard’s arguments that syllogisms are perfect entailments

Let us now consider Abelard’s arguments for the claim that syllogisms
do not have Topical grounding. Recall that there are two ways in which
one might oppose Abelard’s claim. First, one might argue that the
validity of each individual syllogism is grounded in a particular relation
between things (the local strategy). Second, one might argue that there is
a single, very general relation between things that grounds the validity of
all syllogisms in a particular mood (the global strategy). Abelard shows
that neither approach will work. In my discussion, I will consider only
categorical syllogisms, though Abelard brings similar considerations to
bear on hypothetical ones.

The local strategy

Given a particular categorical syllogism, the obvious place to look for

a Topical Differentia is in the middle term. For example, in the syllogism

(14) All animals are alive. All men are animals. Therefore, all men
are alive.

one might naturally take «animal» to be the Differentia and apply the Topic
«from the genus», with the maxim «whatever is predicated of the genus is
also predicated of the species». But as Abelard points out, this Topic would
only explain the entailment from the second premise to the conclusion, not
the entailment from both premises together (258.14-17; cf. 356.4-11).
Even this kind of Topical grounding will be impossible when
syllogisms have false or accidentally true premises, for example:

(15) Every body is colored. But everything sitting is a body.

Therefore, everything sitting is colored (260.18-27).

In such a syllogism, «none of the propositions by themselves necessarily

imply the conclusion» (260.19-20). For there is no real relation in the nature
of things that could license the transition from either of these premises by
itself to the conclusion22. Body, for instance, is not the genus of sitting
thing, nor is colored thing the genus of body23. The only relation between
terms to which we might appeal here is the relation of predication: colored
is universally predicated of body, and body of sitting thing (cf. 259.1-9).
But «A is universally predicated of B» might taken to express either

(a) that A is asserted of all B (secundum vocum enuntiationem), or

(b) that in the order of things, A is true of all B (secondum rerum
cohaerentiam) (353.10-12; cf. 329.19-35).

Similar considerations lead Abelard to claim that «if man is a species of stone,
then if [something] is a man, it is a stone» is good in virtue of its construction (312). It
could not take its necessity from «the nature of things», because in the nature of things
man is not a species of stone. (312-3).
Abelard says at 285.20-29 that «if it is body, it is colored» is only accidentally

If it means merely (a) that A is asserted of all B, then it clearly cannot

ground a necessary entailment from «every C is B» to «every C is A»:

For who would concede that if «stone» were asserted universally

of «man» in some assertion, whether true or false, the consequence
which follows [i.e., «if every stone is an ass, then every man is
an ass», 353.5] would be true? This is why we can assert «stone»
(or anything else we like) of «man», but our assertion, which is
manifestly false, confers no truth on the consequence. (353.15-19)

If, on the other hand, the relation «A is universally predicated of B» means

that A is true of all B, then it is of no use in syllogisms with false premises,
such as

(16) All men are stones. All stones are asses. Therefore, all men are
asses (353.5).

Nor is it of any use when it is merely accidental that A holds of all B, since
entailment must be necessary (cf. 362.30-1). There are some categorical
syllogisms, then, for which no local topical maxim can be found. And once
we accept that one syllogism in Barbara holds in virtue of its construction,
we might as well accept that all do (since all have the same construction).

The global strategy

If the validity of categorical syllogisms depends on a Topical maxim,

then, it must be a maxim that captures the dependence of the conclusion
on both premises. Syllogisms in Barbara, for instance, might be thought to
depend on the rule:

(17) If B is predicated of A universally and C is predicated of B

universally, then C is predicated also of A universally24,

I have used schematic letters to make the principle clearer. Abelard uses
pronouns: «si aliquid praedicatur de alio universaliter et aliud praedicatur de praedicato
universaliter, illud idem praedicatur et de subiecto universaliter» (261.14-16). There
is a corresponding principle for hypothetical syllogisms: «si aliquid infert aliud et id
quod inferat existat, id quoque quod infertur necesse est existere» (261.25-6).

where «predicated of» is taken secundum rerum cohaerentiam. Might (17)

be a Topical maxim that gives syllogisms in Barbara their inferential force?
Abelard’s strategy here is to argue that (17), while perhaps a true
rule (regula), is not a Topical maxim, because it lacks a corresponding
Differentia (261.34-5, 265.25-266.2)25. The argument that (17) lacks a
Differentia is basically the same as the argument (rehearsed above) that
particular syllogisms lack a Differentia. The Differentia would have to
be some thing (res) that is predicated universally of some term in the
conclusion. The only obvious candidate is the middle term (B). But the fact
that B is predicated of all A could at best explain the entailment from one
premise of the syllogism to the conclusion (from «every B is C» to «every
A is C»), not the entailment from both premises to the conclusion. And it
explains this only if B is predicated of all A truly and necessarily: that is,
only if A and B stand in some beefier relation than mere predication –say,
genus and species (362.26-31). This will not be the case for all syllogisms
in Barbara.
Why should it matter whether or not (17) has a corresponding
Differentia and is thus a genuine maxim? Here Abelard is not as explicit
as he might have been, but I think we can reconstruct his reasoning. He is
trying to show that syllogisms are grounded in their construction alone,
not in «the nature of things». Apparently, he takes the fact that syllogisms
do not depend on any genuine maxims to be sufficient grounds for this
claim. Thus, although he does not deny that (17) is true if and only if the

(18) All A are B. All B are C. Therefore, all A are C.

is valid26, he denies that this equivalence shows that our knowledge of the
syllogism’s validity depends on how things are in the world. In order to

STUMP, Dialectic and Its Place, p. 96; GREEN-PEDERSEN, The Tradition of the
Topics, p. 197.
In this respect, (17) fares better than an alternative regula, (CS*): «If B is
predicated of A universally, then if C is predicated of B universally, then C is predicated
also of A universally» (si aliquid praedicatur de aliquo universaliter, tunc si aliud
praedicatur de praedicato universaliter, et de subiecto, 352.31-3). (17) and (CS*) are
not equivalent, because the law of exportation fails in Abelard’s logic. In fact, Abelard
argues, (CS*) and the corresponding regulae for other syllogistic moods have many
false instances (358.34-362.17).

understand Abelard’s reasoning here, we need to understand why he thinks

that only a genuine Topical maxim –one with a Differentia– can ground the
entailment in «the nature of things».
I propose that Abelard is thinking along the following lines. A Topical
maxim gives a rule for inference that is based on its Differentia: that is,
on some thing (res) in the world27. The inferential force (vis inferentiae)
which a maxim brings to an imperfect inference comes from the relation in
which the Differentia stands to a term in the conclusion of the inference (ex
habitudine quam habet ad terminum illatum, 256.36-7). For example, in the
valid consequence «if it is man, it is animal», the inferential force comes
from the relation (species) in which the Differentia (man) stands to animal.
The Differentia, then, is the thing (res) in the nature of which the validity
of imperfect inferences is grounded28. A regula without a Differentia, then,
although it might still be thought to ground the validity of inferences, could
not ground it in «the nature of things», as a maxim does.
To modern eyes, this reasoning appears to make an unwarranted
assumption: that the totality of facts about «the nature of things» is
exhausted by facts of the form

(19) A is F, or
(20) A stands in the relation R to B.

Given this assumption, it follows from (17)’s lack of a Differentia that (17)
is not a fact about «the nature of things» and must therefore depend for its
truth on something else: the construction or form of the syllogism, the way

GREEN-PEDERSEN, The Tradition of the Topics, p. 167. In the consequence «if it
is man, it is animal», the locus differentia is man; when Abelard calls the Topic «from
species», giving the relation in which the Differentia stands to something else, he
is saying «from where the locus comes» (unde sit locus, 264.5-34). Green-Pedersen
conjectures, plausibly, that Abelard insists that the Differentia be a thing and not
the relation itself because the latter approach would make the relations (e.g., genus,
species) into «independent realit[ies]» and contradict his nominalism (p. 168).
Cf. 255.7-9, on the consequence «if every man is animal, every man is alive»:
«These inferences, although they are imperfect in the construction of the antecedent,
nonetheless most often take their necessity from the nature of things, just as with [the
consequence] which we put down earlier from animal to alive, since the nature of
animal, in which as a substantial form alive inheres, never allows animal itself to exist
without life».

it is put together in thought and language. But if we relax the assumption

and count as facts about «the nature of things» facts with more logical
complexity, such as

(21) A, B, and C stand in the relation Q, or

(22) not both:{ (all A are B and all B are C) and not (all A are C) },
or even
(23) for all A, B, and C: A, B, and C stand in the relation Q,

then there is no longer any reason to think that (17) is not a fact about «the
nature of things», and consequently no reason to think that syllogisms in
Barbara do not depend on facts about the world: more general facts, to be
sure, than most Topically grounded inferences, but no less facts about «the
nature of things». Granted, the entailment in a categorical syllogism cannot
depend on the real relation of one thing to another; but might it not depend
on some more complex feature of the world?
This question would become acute for Kant –for whom «the nature of
things» consists of just the kind of complex, generalized relational facts
Abelard does not consider (e.g., the laws of Newtonian science)– and even
more pressing for Frege, Russell, and Wittgenstein, whose new logical
notation allowed the question to be raised in a more explicit way. But
Abelard doesn’t answer it. He is not even in a position to ask it. In order
to do so, he would have to reject the broadly Aristotelian assumption he
inherits from his sources and shares with all of his contemporaries, that
all facts about the world can be described by predicating «something of
something» (ti kata tinos)29. Given that assumption, Abelard is right to
deny that syllogisms depend for their validity on facts about the world.
Indeed, the same reasoning that leads Abelard to this conclusion
should lead him to accept the inference

(24) A is east of B. B is east of C. Therefore A is east of C.

«According to Abelard, if a statement of the form xRy is true, then what makes
it true is nothing but individual subjects and their monadic properties.» J. E. BROWER,
«Abelard’s Theory of Relations: Reductionism and the Aristotelian Tradition», Review
of Metaphysics, 51 (1998) 623.

as valid in virtue of its construction30. For suppose the premises were false.
What would be the Differentia? Since the inference is not valid in virtue of
B’s relation to something else, Abelard would reason, it must not be valid
in virtue of «the nature of things».
This point reveals the extent to which Abelard’s arguments for the
formality of syllogisms are unavailable to us today. Abelard would have to
concede that (24) is valid in virtue of its construction, while

(25) A is a donkey. Therefore, A is an animal.

is valid in virtue of the nature of donkeys. No modern advocate of the

Formality Thesis, I take it, would make a principled distinction between
these two cases. Similarly, as we have seen, Abelard takes syllogisms in
Barbara to be valid in virtue of their construction, while denying the same
status to

(26) If A then B. If not B, then not A.


(27) A is alive. A is alive.

Again, his views about the basis for the Formality Thesis –views we do not
share– would make a distinction of principle where we see none.


Unlike later medieval logicians who make a distinction between

formal and material consequence, Abelard explains why it is important to
distinguish between perfect and imperfect entailments. He argues that the
dominant view, on which all inferences are grounded in topical maxims,
cannot be sustained, and that we must recognize some inferences as not
needing external grounding. This is the first argument for the Formality
Thesis of which I am aware.
Abelard’s arguments seem to have been persuasive: the majority of
Abelard’s twelfth-century successors distinguish between «arguments

I am not aware of any passages in which Abelard discusses such inferences.

which rest upon loci [Topics] (locales) and those that are valid by their
form (complexionales)»31. The distinction persists in the thirteenth century
and is a likely ancestor of the fourteenth-century (continental) distinction
between formal and material consequence32. But we no longer find
arguments for the Formality Thesis that would support the distinction. The
reason, perhaps, is that there is no longer a concerted opposition. After
Abelard, it is taken for granted that valid inferences divide into those
whose validity can be attributed to their structure and those whose validity
depends on their terms and the nature of the things to which they refer.
It is tempting for contemporary advocates of the Formality Thesis to
point to fourteenth-century logicians as predecessors. But if I am right about
the philosophical basis of the medieval distinction, they should not do so.
We cannot accept the premises of Abelard’s argument for the Formality
Thesis, so if we are going to accept some version of the Formality Thesis
ourselves, it will have to be on other grounds33.

GREEN-PEDERSEN, The Tradition of the Topics, p. 200.
W. KNEALE – M. KNEALE, The Development of Logic, Oxford University Press,
Oxford 1962, pp. 274-275, p. 279; STUMP, Dialectic and Its Place, p. 127; GREEN-
PEDERSEN, The Tradition of the Topics, p. 198.
This essay is partially derived from the appendix to my dissertation:
J. MACFARLANE, «What Does It Mean to Say that Logic Is Formal?», University of
Pittsburgh, 2000. It was delivered at ESMLS XIX in Geneva in June 2012.


Boethius’ Logics

Boethius left twelfth century philosophers with three different logics, a

procedure for finding an argument to answer any given question, a puzzle,
and a problem. The logics are those for categorical and hypothetical
syllogisms and for his reworking of Cicero’s account of the Stoic
indemonstrables. The procedure is given in his account of the topics and
the puzzle is to reconcile this account with what he says in his treatments of
categorical and hypothetical syllogisms. The problem is to find a place for
certain familiar propositional inferences given that such inferences have no
place in any of Boethius’ logics.
Boethius sets out the logic of categorical syllogisms in De Syllogismo
Categorico1 (DSC). It corresponds to Aristotle’s exposition of non-modal
syllogisms in Prior Analytics 1-7, supplemented with the account of the
indirect figures developed, according to Boethius, by Theophrastus and
Eudemus. He defines a syllogism in DSC as2:

[…] oratio, in qua positis quibusdam et concessis aliud quoddam,

quam sunt ea, quae posita et concessa sunt, necessario contingit per
ipsa, quae concessa sunt.

This is not quite the definition given in the Prior Analytics which with
the recovery of Boethius’ translation of the work in the first half of the
twelfth century became standard in mediaeval logic. Boethius’ definition
adds to Aristotle’s the requirement of concession and while he himself
probably intended nothing much by it3, the addition allowed Abaelard to

University of Auckland (New Zealand), Faculty of Arts, 18 Symonds St
Auckland 1142. Email: cj.martin@auckland.ac.nz.
Anicii Manlii Severini Boethii De syllogismo categorico, Ed. by Ch. THOMSEN
THÖRNQVIST, Acta Universitatis Gothoburgensis, Göteborg 2008.
Boethius, De Syllogismo Categorico (DSC) II, 69.
See C. J. MARTIN, «“They had added not a single tiny proposition”: The Reception
of the Prior Analytics in the First Half of the Twelfth Century», Vivarium, 48 (2010)
157-190, p. 161, n. 13.

make a distinction crucial for him between arguments and conditionals.

Unlike Aristotle, although his examples are always given in the form of
arguments, Boethius formulates the moods of the categorical syllogism as
conditional propositions. For example barbara4:

Si A in omni B fuerit, et B in omni C fuerit, A terminus de omni C

praedicabitur, id est: Omne iustum bonum est. Omnis uirtus iustum
est. Omnis igitur uirtus bonum est.

Like Aristotle Boethius employs letters to formulate the moods of the

syllogism and replaces them uniformly with general terms in his examples.
Unlike Aristotle he offers a justification for the practice, telling us that it is
required in order to achieve generality and avoid the possibility of error5.
In DSC Boethius’ calls these schemata for syllogisms complexiones6 and
characterises them as having a form7 which in his commentary on Cicero’s
Topics he explicitly contrast with their component propositions as the matter
of syllogisms drawing an analogy in both works to the relationship between
the structure of a building and the materials from which it is constructed8.
Categorical syllogisms in all three figures are alike in that nothing in
addition to their premisses is required to prove the conclusion but the moods
of the secondary figures are imperfect in contrast to the perfect direct moods
of the first figure, which Boethius tells are called «indemonstrables»9,
because of the way in which their premisses are presented10:

[…] perfectus syllogismus est, cui ad integram probatamque

conclusionem ex superius sumptis et propositis nihil deest […];
imperfectus syllogismus est, cui nihil aeque ad perfectionem deest,
verumtamen in his quae in propositionibus sumpta sunt, aliqua desunt.

What is required to show that the conclusions of the secondary figures

follow from their premisses is provided entirely by the first figure, the
principles of conversion, and reduction to impossibility11.

Boethius, DSC II, 52-53 (89).
DSC II, 45.
DSC II, passim.
DSC II, 57, 101.
Boethius, In Topica Ciceronis (ITC), PL 64, 1046D-1047A; DSC II, 43-44.
DSC II, 75.
DSC II, 50.
DSC II, 60.

The second logic left by Boethius to the middle ages is the account
of the hypothetical syllogisms which he sets out in De Hypotheticis
Syllogismis12 (DHS). So far no ancient parallel for most of this material has
been discovered13 but Boethius’ remarks on it suggest that it is an attempt
to formulate an account of the logic of the conditional and disjunction in
Peripatetic terms14. Thus, he claims, though without giving any hint in
DHS of how to do this, where there is any doubt as to the truth of the
conditional premiss of a hypothetical syllogism it may be proved with a
categorical syllogism. The various forms of hypothetical syllogism do
not themselves, however, require anything further to guarantee that their
conclusion follows from their premisses, though a proof by reductio and
the affirmation of the antecedent is required to show that an argument by
denying the consequent holds15:
Just as in his account of the categorical syllogism Boethius formulates
hypothetical syllogisms schematically using letters for general terms since,
as I have argued extensively elsewhere, he has no notion of a propositional
content which might be substituted into the antecedent or consequent of a
simple conditional to yield a more complex conditional16. Rather he limits
himself to five classes of conditional, one simple, three compound, and a
mediate form consisting of pairs of simple conditionals, distinguishing for
each form in each class two different arguments –affirming the antecedent
and denying the consequent– where there is a syllogism in virtue the
structure of the propositions (complexio propositionum)17.
Thus all the instances of the schema «Si est A, est B, est A; ergo est
B» are valid hypothetical syllogisms as are all instances of «Si est A, non
est B, est A, ergo non est B» but these two argument schemata are not

Boethius, De Hypotheticis Syllogismis (DHS), Ed. by L. OBERTELLO, Paideia,
Brescia 1969.
For the very little that has been found, a treatment of only simple conditionals
and disjunction, see S. BOBZIEN, «A Greek Parallel to Boethius De Hypotheticis
Syllogismis», Mnemosyne, 54 (2002) 285-300.
Boethius, DHS 1.1.3.
DHS 2.1.6 «Non igitur syllogismus probatione, in eo quod syllogismus est,
indigebit […]», DHS 2.3. «[…] eaque necessitas tali ratione probabitur».
See C. J. MARTIN, «The Logic of Negation in Boethius», Phronesis, 36 (1991)
DHS 2.2-3. See C. J. MARTIN, «The logical textbooks and their influence», in
J. MARENBON (ed.), The Cambridge Companion to Boethius, Cambridge University
Press, Cambridge 2009, pp. 56-84.

presented by Boethius in DHS as themselves instances of a more general

propositional schema for affirming the antecedent.
Boethius also claims that conditionals of the form «Si non est A, est
B» are true only when A and B are both exclusive and exhaustive and so
support affirming the consequent and denying the antecedent, not, however,
in virtue of the structure of the propositions but rather, he says, in virtue of
the terms, or the nature of things18.

[…] hic quoque […] dicendum est secundum quidem ipsius

complexionis figuram nullum fieri syllogismum; secundum terminos
uero in quibus solis dici potest, necesse esse, si A fuerit, B non esse.

In DHS Boethius also provides a logic for disjunction, but only simple
disjunction, presented in exactly the same way as his logic for conditionals
with letters standing for general terms. A given form of simple disjunction
is, according to Boethius, equivalent to the conditional whose antecedent is
the opposite of the first disjunct and whose consequent is the second. Simple
disjunctions with both disjuncts affirmative thus support four inferences,
two in virtue of the complexio and two in virtue of the properties of the
terms disjoined, whereas the three other forms of disjunction support only
the two forms of disjunctive syllogism.
Boethius provided the middle ages with a method for discovering
arguments with his account of topical inference. In his commentary on
Cicero’s Topica he introduces the theory which he later summarised in De
Differentiis Topics (DT) where he follows the classification of the topics
given by Themistius as well as that provided by Cicero and proposes a
reconciliation of them. Boethius’ commentary was, it seems, little used in
the middle ages and the puzzle which I mentioned above is discussed rather
in connection with DT. It is, however, already clear in the commentary and
perhaps even more puzzling as it appears there. Cicero offers in his Topics
a recipe for answering questions by finding an argumentum, that is to say
a reason bringing conviction where something was in doubt, and tells us
that a locus, or topic, is where such a reason is to be sought, it is a sedes
argumenti. Boethius comments that every argumentum is presented in an
argumentatio, that is to say in a syllogism or an enthymeme, and so as
propositions. In DT he gives a more general and technical definition19:

DHS 2.3.6, for «rerum natura» see II.2.4.
Boethius, De Differentiis Topicis (DT), PL 64, 1174C.

[…] uis sententiae ratioque ea quae clauditur oratione cum aliquid

probatur ambiguum, argumentum uocatur.

The role of the locus, he tells us in both works, is to provide a general

principle, a so called maximal proposition, which in some way contains the
propositions which appear in the syllogism solving the question that we are
interested in20:

Maximas igitur, id est uniuersales ac notissimas propositiones, ex

quibus syllogismorum conclusio descendit.

In DT we are again given a rather more technical definition21:

Ac sicut locus in se corporis continet quantitatem, ita hae

propositiones quae sunt maximae, intra se omnem uim posteriorum
atque ipsius conclusionis consequentiam tenent, et uno quidem
modo locus, id est argumenti sedes, dicitur maxima principalisque
propositio, fidem caeteris subministrans.

A maximal proposition, Boethius notes, may be included in a syllogism

or, it will turn out much more importantly, may in some way provide the
argumentation with its strength from the outside22:

[…] aliae uero in ipsis quidem argumentationibus minime

continentur, uim tamen argumentationibus subministrant.

Just what Boethius intends by this is not at all clear. His characterisation
of maximal propositions and his examples suggests that an externally
posited maximal proposition somehow provides the premisses for the
argument. The maximal proposition «Of what the definitions are different
the substances are different» thus offers us a way of proving that two items
are distinct by showing that they have different definitions. Boethius’
example of an argumentum descending from this maximal proposition is23:

Boethius, ITC, 1051D.
Boethius, DT, 1186A.
Boethius, ITC, 1051D.
Boethius, DT, 1185C.

A jealous man is one who laments the happiness of others,

a wise man does not lament the happiness of others;

therefore a jealous man is not wise.

That is to say an argument which we could if we wished recast as

an instance of camestres. The puzzle is to understand exactly what role
Boethius supposes the maximal proposition to play in the argument given
that he has insisted, as we have seen, that nothing is needed apart from the
truth of the premisses to guarantee the truth of the conclusion in this or any
other mood or form of categorical or hypothetical syllogism.
In both ITC and DT on the other hand, Boethius gives the principles
of affirming the antecedent (posito antecedenti comitari quod subsequitur)
and denying the consequent (perempto consequenti perimi quod antecedit)
as maximal propositions for the loci ab antecedenti and a consequenti
from which are drawn the argumenta which prove the conclusion in
instances of these argument forms. In his detailed discussion of Cicero’s
remarks on these loci, with the locus a repugnantibus, as those which
are most properly used by dialecticians, however, there is no mention of
maximal propositions24. This is perhaps not surprising since the argument
forms in question are the Stoic indemonstrables of which Cicero gives the
extended list of seven. It is Boethius’ reworking of these which is the third
logic he bequeathed to the middle ages. Cicero formulates the first three
types of indemonstrable in terms of their general definitions which may
be interpreted as characterising a relation between predicates, as in the
account given in DHS, just as easily as a relation between propositional
contents. Boethius does indeed interpret them in this way and gives, for
example, the conditional premiss of the first type of indemonstrable in the
form «si hoc est, illud est» rather than «si hoc, illud». He goes on to add to
the Stoic meteorological impersonal «si dies est, lux est», his own favourite
«si homo est, animal est», the paradigmatic simple conditional of DHS.
As I have shown elsewhere, the absence of propositional negation in
Boethius’ logic is illustrated by his rewriting of a version the third Stoic
indemonstrable given by Cicero25. The argument which according to

Boethius, ITC V, 1129D-1145B.
MARTIN, «The Logic of Negation in Boethius», pp. 291-294.

Cicero is drawn from the locus a repugnantibus has for its first premiss
the negation of a copulative conjunction with first conjunct affirmative and
second negative, the second premiss is the first conjunct, and the conclusion
the negation of the second. Boethius rewrites the first premiss by replacing
the copulative conjunction «et» with the conditional conjunction «si». In
this rewritten form he construes both the preposed negation and that of the
consequent as predicative. The preposed negation acts on the consequent
negation to produce, in effect, an affirmation, and a true conditional.
The premisses of the fourth to seventh modes of indemonstrable are
given by Cicero with demonstrative pronouns apparently functioning
as propositional variables but while Boethius repeats them in this form
he shows no understanding of the propositionality of the connectives26.
He explains that the negated copulative of modes 6 and 7 hold only of
exclusive and exhaustive opposites, as do the disjunctions of modes 4 and
5, but treats it as special form of connection which despite its appearance
is not produced by applying propositional negation to a copulative
conjunction. There is thus nothing here to contradict his rejection in his
longer commentary on De Interpretatione of propositional negation and of
the copulative conjunction as proposition forming27.

The Discovery of Propositionality

The logicians of the twelfth century seem to have ignored Boethius’

reinterpretation of the Stoic indemonstrables, and their great achievement
in logic was to understand the connectives that appear in them, as the Stoics
did, propositionally. This achievement brings with it, however, the problem
of properly characterising the logic of the propositional connectives in
propositional terms.
Propositional negation appears for the first time in the middle ages, as
far as I know, in the Dialectica of Garland written at the end of the eleventh
or beginning of the twelfth century. In this work Garland, writing in the
millieu of Abaelard’s master Roscelin of Compiegne, presents a strenuously
anti-realistic account of logic adding, he tells us, to the rules given by
Boethius and Aristotle the results of his masters’ and his own research.

His presentation is entirely syntactical, see ITC, 1136D-1137A.
See MARTIN, «The Logic of Negation in Boethius», p. 284.

In his account of the negation of simple and quantified categorical

propositions Garland follows Boethius in maintaining that the negative particle
is applied to the predicate to signify its separation from the subject with no
hint that there is an alternative, propositional, account. The same is true of
his exposition of the logic of composite conditionals where he again follows
Boethius and gives, for instance28: «Si est homo, cum sit animatum, est animal,
sed cum sit animatum non est animal; ergo non est homo» as an example of
denying the consequent. In his treatment of conditionals, however, in setting
out one of the sophisms which are the peculiarly characteristic feature of his
work Garland argues that the conditional «Si est homo, est albus» is true since29:

Si est falsa sua negativa: non si est homo, est albus, vera est si est
homo est albus - nam talis negatio, scilicet ubi preponitur non,
dividit verum et falsum cum sua affirmatione.

Referring to this move a little later Garland briefly discusses the

claim that a proposition formed by preposing the negative particle in this
way to a conditional rather than that obtained by negating the consequent
is properly the «negative» of the original «affirmative» conditional. He
suggests that the issue may be resolved by construing «Si est homo, non est
albus» as meaning «if human holds of something, it does not follow for this
reason that it is white» and so as a conditional proposition dividing truth
and falsity with the original conditional and equivalent to the categorical
proposition «non: si est homo, est albus»30. He does not notice that such a
construal invalidates instances of modus ponens with conditionals of this
form as their hypothetical premiss.
Garland does, however, clearly recognise negation as propositional
operation and it appears once more in his Dialectica in a delightfully
complicated sophism where he argues31:

Homo est asinus. Utrum? Aut non est rudibilis, aut est asinus; sed
est rudibilis. Utrum? Aut non est non falsa ista propositio omnis
homo est rudibilis, aut vera est ista que dicit: homo est rudibilis;
sed ista propositio est non falsa: omnis homo est rudibilis. Utrum?

Garlandus, Dialectica, Ed. by L. M. DE RIJK, Van Gorcum, Assen 1959, p. 151.
Garlandus, Dialectica, pp. 133-134.
Ibid., pp. 135-136.
Ibid., p. 161. «Utrum?» in Garland’s sophisms means «Why is this?».

Aut non est non vera ista propositio non aut non est homo aut est
rudibile aut est non falsa: omnis homo est rudibilis […]

Garland follows Boethius closely in his treatment of the topics but

locates it, as the scientia inveniendi, before his accounts of both the
categorical and hypothetical syllogism, which together for him constitute
the scientia iudicandi. He notes furthermore that we may say that a maximal
propositions grounds the resolution of a question since a syllogism is proved
with a maximal proposition32. He does not, however, mention maximal
propositions in his treatment of categorical and hypothetical syllogism.
In other works from the beginning of the twelfth century topics are
associated with the moods of the categorical and the various forms of the
hypothetical syllogism. In the texts published by Yukio Iwakuma as the
Introductiones Dialecticae artis secundum magistrum G. Paganellum and
the Introductiones dialecticae <secundum Wilgelmum>33, which Iwakuma
associates with Abaelard’s other master, William of Champeaux, we find
for the first time the short list of topics which, as I have argued elsewhere,
seems to be derived from Boethius’ account of the relative extensions of
general terms in his Introductio ad Syllogismos Categoricos34. One of these
twelfth century Introductiones, Secundum Wilgelmum, defines a locus as a vis
argumenti35 and in both there are new topics corresponding to various logical
operations employed by Boethius in developing his account of categorical and
hypothetical syllogisms. For example the locus a pari per contrapositionem36:
Maximam propositionem: si vera est universalis affirmativa, vera est
sua conversa per contrapositionem. Assume: sed vera est ista omnis
homo est animal. Quare vera est sua conversa per contrapositionem
omne non animal est non homo.

Such principles, which are almost always in these texts called rules
(regulae) rather than maximal propositions, are given for each of the various
forms of conversion. In addition the locus a subiecto and a praedicato are
cited as the sources of the rules associated with the conditionalised forms

Garlandus, Dialectica, p. 86.
Y. IWAKUMA, «The Introductiones dialecticae secundum Wilgelmum and
secundum magistrum G. Paganellum», Cahiers de l’Institut du Moyen-Âge Grec et
Latin, 63 (1993) 45-114.
See MARTIN, «“They had added not a single tiny proposition”».
IWAKUMA, «The Introductiones dialecticae», I.6.3, p. 67.
Ibid., I.6.1, p. 94.

of the moods of the categorical syllogism. Thus for barbara in the form of
a conditional with a categorical antecedent and conditional consequent37:

Agendum est de illa quae constat ex categorica et hypothetica. In illa

categorica probat hypotheticam. […] Probat enim illam aliquando
ex vi praedicati, et tunc dantur hae regulae. Si aliquid praedicatur
de alio universaliter, tunc si aliud praedicatur de praedicato
universaliter, illud idem praedicatur de subiecto universaliter.

We are also provided with various rules drawn from the locus ab
antecedenti and a consequenti to prove compound conditionals, for
example contraposition, versions of transitivity, and more complex forms
such as «if something entails a hypothetical, then if something else follows
from the consequent of that hypothetical, it may be substituted for it in the
compound hypothetical»38. The examples presented in these texts like the
sophisms in Garland’s Dialectica show a considerable logical sophistication
and skill in manipulating complex propositions. The Introductiones do not,
however, have anything to say about the nature of inference in general and
for that we must turn to Peter Abaelard’s Dialectica the work in which he
sets out the logicial novelties for which he became notorious in the first
decade of the twelfth century in Paris.

Abaelard’s Logical Principles

Abaelard insists in his Dialectica and other works that the dividing
opposite of any given proposition is formed by negating it as a whole and
that this truth-functional opposite should be represented by preposing
the negative particle to the proposition. He distinguishes the operation of
propositional negation, iterable without limit, which he calls destructive
negation, from the application of the negative particle to the predicate of
an affirmation, forming what he calls its separative negation, to signify
the separation of the predicate from the subject, a proposition which is a
contrary of the original rather than its contradictory. Separative negation
is not iterable and if their subject term is empty both an affirmation and its
separative negation are false. Abaelard also insists, explicitly contradicting

Ibid., II.5, p. 70.
Ibid., I.9.10, p. 108.

Boethius, that the copulative conjunction «et» combines two propostions

to form a single proposition which may be negated with destructive
Abaelard’s understanding of these two propositional operations
is clear in his treatment in his Dialectica of Boethius’ catalogue of
hypothetical syllogisms and in his acceptance of a special case of what we
now call de Morgan’s rule. He investigates the possibility of interpreting
the «hypothetical» components of Boethius’ compound conditionals as
temporal propositions, true just in case both components are true40, and
canvases various suggestions as to how the negation of such propositions
might be represented with the negative particle applied to only the second
component. In the end, however, he despairs of a solution41:

Quem <sc. Boetium> tamen fortasse de destructione earum,

si attentius inspiciamus, nullo modo in quibusdam defendere
possumus, sed emendare.

The example given by Boethius and repeated by Garland thus becomes

in Abaelard’s presentation of the hypothetical syllogism: «Si est homo, cum
est animatum est animal; sed non cum est animatum est animal; ergo non
est homo»42. Where the scope of the preposed «non» is the whole of the
following compound temporal proposition.
In his account of hypothetical propositions Abaelard introduces
propositional disjunctions as equivalent to the conditionals whose
antecedents are the propositional opposites of the first disjunct and whose
consequents are the second disjunct. Abaelard’s account of the conditional
is intensional and so the same holds for disjunction with the consequence
that a necessary condition for the truth of an Abaelardian propositional
disjunction is that the disjuncts are exhaustive. His account of copulative
conjunction is, on the other hand, truth-functional and so de Morgan’s
Rule does not in general hold. It does hold, however, as Abaelard notes,
where the conjoined propositions exhaust all possibilities for a given kind
of subject. In particular we may argue by contraposition from the truth of

See C. J. MARTIN, «Logic», in J. BROWER – K. GUILFOY (eds.), The Cambridge
Companion to Abaelard, Cambridge University Press, Cambridge 2004, pp. 158-199.
Abaelard, Dialectica, Van Gorcum, Assen 1970, 2nd ed., pp. 481-488.
Ibid., p. 488.
Ibid., p. 306.

«Si Socrates neque est rationalis neque irrationalis, non est animal» to
that of «Si Socrates est animal, Socrates est uel rationalis uel irrationalis».

Si uero inter totas etiam propositiones quae antecedunt ad tertiam uis

inferentiae recipiatur, ac <si> uidelicet ita diceretur: si ita est in re ut
utraque istarum dicit propositionum, ita est in re ut tertia proponit,
nec tunc quoque destructionum inferentia fallit, ut uidelicet ita
dicamus quod si non est in re quod dicit ultima, non est in re totum
quod primae proponunt. Sicut enim simul antecedunt, ita et simul
auferri debent. Neque enim fallere potest ut si quid aliud necessario
exigit, si id quod exigitur non erit, nec quod exigit necesse est non

Abaelard treats Boethius’ compound conditionals as propositions in

which the principle connective is conditional but whose components are
temporal conjunctions because his theory of the conditional is intensional.
Indeed it is hyper-intesional, a relevantistic theory of the sort which we
now classify as connexive and like other relevantistic theories does not in
general allow a conditional with a categorical antecedent and conditional
consequent to be true. As we will see, Abaelard holds that such conditionals
are true in certain special cases but the examples given by Boethius are not
of this kind.
Abaelard’s theory of the conditional is found in his Dialectica in the
treatise on the topics which unlike Garland he places after his discussion
of the categorical syllogism but before that of the hypothetical syllogism
since he holds that the topics are the source of one kind of true conditional
but not of syllogisms. Abaelard and his followers the Nominales were
notorious, indeed, in the twelfth century for solving our puzzle by rejecting
it, maintaining that syllogisms, whether categorical or hypothetical have
no need of support from the topics44. Abaelard’s reasons for insisting
on this also provide him with a solution to the problem of the proper
characterisation of propositional inferences but as we will see that this
solution itself is somewhat problematic.

Ibid., p. 306.
Abaelard, Dialectica, pp. 256-263, see Y. IWAKUMA – S. EBBESEN, «Logico-
Theological Schools from the Second Half the 12th Century: A List of Sources»,
Vivarium, 30 (1992) 173-215, Extract n. 20, p. 180.

Abaelard’s general discussion of inference and conditional propositions

is striking in that he is at pains to distinguish between the necessity which
must connect the premisses and conclusion of a valid argument from the
necessity which is required to connect the antecedent and consequent of a
true conditional. For an argument to be valid it must, Abaelard agrees, be
necessarily truth preserving but for this all that is required is that it is not
possible for the premisses all to be true and the conclusion false at the same
time. For a conditional to be true on the other hand, according to Abaelard, it
must be such that in addition to it’s not being possible for the the consequent
to be false when antecedent is true, the sense of the antecedent must also
contain that of the consequent45. A true conditional is thus what we would
now characterise as an analytic truth and so if we identify logical with
analytic truth, Abaelard’s position is that logical truth requires more than is
required for valid inference. He argues for this distinction by invoking the
principles of connexive logic to provide him with rejection rules to which he
appeals to show that certain connections between the terms of propositions,
or between whole propostions, which certainly guarantee the necessity
required for valid argument, cannot guarantee the truth of the corresponding
conditional. There can thus be no deduction theorem for Abaelardian logic46.
Abaelard characterises the connection which holds between the
antecedent of a true conditional and its consequent as inferentia, which
we may translate as «entailment». The weaker sort of necessity required
for validity produces conditionals which are not true but which, according
to Abaelard, nevertheless have «maximal probability»47. His project in the
treatise on the topics in his Dialectica is to determine just which topics
provide us with entailments. The list of candidates is drawn from that
provided by Boethius rather than from the new short list but Abaelard adds
to it the loci a praedicato, a subiecto, and several others including those
which correspond to relations between whole propositions rather than
between their terms48.
A conditional expresses an entailment according to Abaelard just in
case the truth of the antecedent requires (exigit) that of the consequent

See e.g. Abaelard, «Super Topica Glossa», Ed. by M. DAL PRA, Pietro
Abaelardo: Scritti Logici, La Nuova Italia, Florence 1969, 2nd ed., pp. 205-330, 309.
See MARTIN, «Logic».
Abaelard, Dialectica, p. 285.
Ibid., pp. 352-369.

and this is so just in case the sense of the antecedent contains that of the
consequent. This condition may be met simply by the construction or form
of the antecedent guaranteeing that if it is true then so is the consequent,
in which case the entailment is, he says, perfect49. Alternatively it may be
guaranteed to hold by a topical principle, a maximal proposition which
provides the vis inferentiae, Abaelard’s new definition of a locus50, the
warrant, that is, that the character of the antecedent and consequent is such
that if the former is true then so is the latter, in which case the entailment
is imperfect51.
Abaelard explains this distinction in terms of substitution52. There is
absolutely no logical difference for him between perfect and imperfect
entailments. They are equally necessary but their necessity is in the one
case guaranteed by their complexio, and in the other by what we would
probably call a conceptual, or perhaps a metaphysical, truth. Perfect
entailments are such that they hold for all uniform substitution of general
terms while imperfect entailments hold only for the subsitution of general
terms or of propositions related in a particular way. Neither Abaelard, nor
anyone else that I know of in the middle ages, characterises complexional,
or formal truth, in terms of the uniform substitutablity of propositional
variables. Thus, according to Abaelard53:

[…] si ita proponatur: si omne animal est animatum et omne animatum

est animatum, omne animal est animatum perfecta quoque est
secundum complexionem inferentia. Quod quidem inde patet quia
ex se tantum, non ex natura terminorum, haec inferentia perfecta
est, quod, qualescumque terminos apponas, siue cohaerentes siue
remotos, nullo modo cassari ualet consecutio. […] Caeterae quoque
uerae consequentiae, quarum inferentia ex rerum natura pendet, non
in quorumlibet terminorum rebus uerae consistunt, sed in his tantum
quae naturam eius consecutionis seruant.

True conditionals such as «si Sortes est homo, est animal» thus rely on
fact that human is species of animal for the support of the locus a specie

Ibid., p. 253.
Ibid., p. 253.
Ibid., p. 253.
Ibid., p. 256.
Ibid., pp. 255-256.

with its maximal proposition that of whatever a species is predicated so is

its genus.
Despite, however, characterising perfect entailments in his introductory
remarks as those which hold for all uniform substitution of general terms and
repeating this claim later without suggesting that there are any exceptions
Abaelard goes on to remark that the conditional «si <omne animal> est
animatum, <omne animal> est animatum» is not perfect since54:

[…] quamuis certum sit idem sine se ipso non posse consistere
non est tamen inferentia complexionis perfecta, ubi idem ad
sui positionem simpliciter sequitur, nisi et ipsum in se contineri
demonstretur per adiunctam antecedenti propositionem quae est:
omne animatum est animatum.

Abaelard’s discussion the loci drawn from Boethius’ lists which he

considers worthy of investigation as of particular concern to logicians
produces the result that of those loci which provide general principles
governing the relationship, or as he says habitudo, between the kinds of
things signified by the subject and predicate terms of the antecedent of a
simple conditional and those signified by the subject and predicate terms
of the consequent only those which have to do solely with the nature of
the thing in which we are interested can provide a warrant for an imperfect
entailment. Such conditionals express, he says, a law of nature (lex naturae)
and a natural consecution (consecutio naturalis)55.
An imperfect entailment thus holds where the habitudes which relate
the terms of the antecedent and consequent have to do appropriately with
the definition or definitional components of a natural kind. «If Socrates
is human being, then he is an animal» is an imperfect entailment but «If
Socrates is a human being, then he is not an ass» is not, nor, surprisingly,
according to Abaelard, is «If Socrates is a mortal rational animal, then he
is a human being»56.
As I remarked above Abaelard insists that no locus is needed to
guarantee that the conditionalisation of a categorical or hypothetical
syllogism is true. It is, however, crucial for him that the conditionalisation
is formulated in the appropriate way. The versions we have seen given in

Ibid., p. 255.
Ibid., p. 280.
Ibid., pp. 331-332.

the Introductiones are compound conditionals with categorical antecedents

and conditional consequents but as I noted Abaelard holds that in general
such conditionals are false. The only propositions of this form which are
true, he insists, are those which have as their categorical antecedent a
proposition stating that a topical relationship holds which would warrant
the truth of the consequent in a topical proof of it. So for example «Si
animal est genus hominis, si est homo, est animal» is true57 but all the
conditionals given as instances of the moods of the categorical syllogism
in the Introductiones are false58.
As well as insisting that such conditionals are false Abaelard argues
that the attempt by their proponents to assign them a locus fails because
the locus needed and the maximal proposition drawn from it would have
to warrant the whole compound conditional59. If that locus were a thing,
its assignment would thus locate a feature of the subject and / or predicate
of the antecedent categorical which would warrant the truth of the whole
compound of which it is the antecedent. In general, however, Abaelard
argues, the only loci available to do this are those a praedicato and a subiecto
but these cannot do the job since the truth of a simple categorical, save in
the special circumstances mentioned, cannot entail that of a conditional.
The conditional formulation of the syllogism which Abaelard does
accept uses the copulative conjunction «et» to combine the premisses of
the syllogism to form the antecedent of a conditional with the conclusion
as its consequent60. He thus cannot accept the principle of exportation
which would allow him to export one of the antecedent conjuncts to form
an equivalent conditional with categorical antecedent and conditional
consequent61. Since the antecedent of Abaelard’s conditionalised syllogisms
is a compound of two propositions it does not have a single subject or
predicate which might provide a thing as locus to guarantee the truth of
the conditional, and indeed, as we have just seen Abaelard holds that such
conditionals do not require the aid of a locus of any sort.
In addressing the question of why Boethius claims that loci do support
syllogisms and offers examples which purport to show this Abaelard

Ibid., p. 282.
Ibid., p. 352.
Ibid., pp. 354-364.
Ibid., pp. 253-254.
Ibid., p. 359.

suggests that what he had in mind was perhaps that the loci concerned
warrant an enthymematic inference which could then used to support the
claim that a given syllogism is valid. To justify this suggestion he introduces
a crucial principle for his logic in both its general and particular form62:

Nam ubi simplex ualet inferentia, qui<s> non magis compositam

recipiat? Potest quoque ex simplici necessario probari composita.
Si enim uerum est: si omnis homo est animal, omnis homo est
animatus et ueram necesse est esse syllogismi inferentiam, hanc
scilicet: si omnis homo est animal et omne animal est animatum,
omnis homo est animatus Quicquid enim sequitur ad consequens,
et ad antecedens. Duae uero illae simul propositiones ad quamlibet
ipsarum antecedunt.

Here we have, stated for the first time, I think, in European logic both
the general principle of strengthening the antecedent, or monotonicity, and
the particular case of conditionalised simplification, as well as an appeal
to transitivity. Abaelard does not indicate what status these principles have
but we must assume that he regards them as logical truths. Just what kind
of logical truths, however, is not immediately obvious.
As I noted above, Abaelard holds that in addition to the loci which
are res, and the habitudes which they support, there are also loci and
maximal propositions which warrant entailments holding between whole
propositions. The locus ab antecedenti and a consequenti, Abaelard insists,
must indeed be understood in this way if we are to be able to understand the
logic of conditional proposition. The locus ab antecedenti is, however, he
notes, entirely empty, since someone who does not accept that «p entails q»
will not be convinced by its being pointed out to him that «p» is antecedent
to «q» and an appeal to the maximal proposition that where the antecedent
is posited so is the consequent63. Also drawn from the loci ab antecedenti
and a consequenti, however, are the principles of contraposition and
versions of transitivity64.
In addition to these Abaelard holds that equivalence of propositions is
a topical relationship65:

Ibid., pp. 260-261.
Ibid., p. 365.
Ibid., pp. 364-369.
Ibid., p. 270.

Secundum autem totas antecedentes propositiones locus consistit

cum dicitur: si omnis homo est animal, omne non-animal est non-
homo; si nullus homo est lapis, nullus lapis est homo; si (quia
Socrates est homo Socrates est animal), et (quia non est animal,
non est homo). Tota enim praecedens propositio toti consequenti,
secundum hoc quod ei aequipollet, comparatur.

Note that like repetition all of these conditionals hold for all subsitutions
of terms and Abaelard could certainly associate repetition with the locus a
pari if he wished to do so. Unfortunately he does not say any more about
the conversion of quantified categorical propositions or indeed provide
any reason for characterising syllogisms as perfect and complexional
while conversion is imperfect and local. The reason is perhaps that as far
is logic is concerned it does not matter. The logical status of an imperfect
entailment is precisely the same as that of a perfect entailment. Both are
necessary in precisely the same way.
Nowhere, as far as I know, does Abaelard classify the principles of
conditional simplification. Given what he says about repetition and his
remarks on propositional loci, it is hard, however, to suppose that he
would not classify it too as imperfect and local even though holds for all
substitution of terms. But again it doesn’t really matter since he clearly
supposes that it is logically true in just the same way as «si est homo,
est animal’ is logically true as are all instances of contraposition and
transitivity. Because he thinks this, however, he has left himself with no
room for escape from the arguments which would be directed at his logic
during his last decade in Paris66.

See MARTIN, «Logic».


De nos jours, il y a souvent des résistances à qualifier toute la logique

de formelle. Jonathan Barnes, dans Truth etc.1, relève que l’une des raisons
de ces résistances réside dans la confusion des termes ‘symbolique’
et ‘formelle’ utilisés pour décrire les systèmes de logique développés à
partir du XXe siècle. Je crois qu’il s’agit d’une confusion terminologique
commune, non seulement chez les non-logiciens, mais aussi chez certains
logiciens, comme on peut d’ailleurs le constater en lisant la plupart des
manuels d’introduction aux systèmes de logique symbolique.
Il est possible de voir une conséquence, parmi d’autres, de l’attitude qui
consiste à identifier logique formelle et symbolique dans le choix, répandu
depuis quelques années, d’attribuer l’appellation de logique informelle à
l’analyse de l’argumentation, dans l’idée que «les encadrements théoriques
de la logique axiomatique»2 ne sont pas adéquats pour servir de « fondement
à une analyse de la pratique argumentative »3. Ceci, à mon avis, a généré
davantage de malentendus terminologiques, et comporte le risque de
conduire à des contradictions ; car il est évident que, si « l’analyse des
sophismes »4 appartient au champ recherche de la logique informelle, avec
pour but la détection d’erreurs dans des cas concrets d’argumentation, il
faut nécessairement faire référence à une certaine forme (qui peut toutefois
être appelée ‘structure’, ‘modèle’), choisie comme critère de distinction
entre un bon et un mauvais argument.
Aucune analyse qui se dit logique ne peut faire abstraction de son
caractère formel, c’est-à-dire d’un caractère universel et organisé selon
un certain ordre donné à une certaine matière. Nous pouvons dire, donc,
une fois distingué entre ‘formel’ – qui spécifie que la logique s’occupe

Pontificia Università San Tommaso d’Aquino - Angelicum et Pontificia
Università Urbaniana, Roma, Italia. Email: g.lombardi@urbaniana.edu
J. BARNES, Truth etc. Six Lectures on Ancient Logic, Clarendon Press, Oxford
2007, pp. 274-276.
C. SALAVASTRU, Logique, argumentation, interprétation, L’Harmattan, Paris
2007, p. 149.
Ibid., p. 150.

des formes d’inférence – et ‘symbolique’ – qui spécifie qu’à l’intérieur

de l’étude des formes d’inférence, il est fait recours à des symboles dotés
d’une et d’une seule signification, afin d’éviter les ambiguïtés des langages
naturels –, que toute la logique, de l’Antiquité à nos jours, est formelle.
On qualifie donc de formelle la logique d’Aristote ainsi que celle de
Thomas d’Aquin, alors qu’elles ne sont pas construites sur la base d’un
langage symbolique, mais sur des langages naturels, d’une part le grec,
de l’autre le latin. Je ne rentre pas dans la question épineuse de savoir
si toute inférence valide l’est en vertu d’une forme logique ; j’aimerais
rester à un niveau à la fois plus général et plus fondamental : d’où vient
ce caractère formel de la logique ? Et est-ce que cela implique qu’il existe
une logique formelle universelle? La réponse négative à cette deuxième
question ressortira au fur et à mesure de l’argumentation proposée en
soutien à la réponse à la première question, qui, d’ailleurs, figure dans le
titre de cet exposé.
Mon propos est de montrer, par le biais de références à quelques
Commentaires de saint Thomas aux ouvrages d’Aristote, que ce caractère
formel de la logique provient de sa constitution en tant qu’ars – là où
‘ars’ est la traduction latine du mot grec ‘techne’, qui possède les traits
d’universalité et d’ordre. Le texte de référence pour le sens du mot ‘techne’
est le premier chapitre du Livre A de la Métaphysique d’Aristote; pour le
lien explicite entre forme et techne, je signale le passage du livre Λ du même
traité5, où Aristote précise que, normalement, ce qui est déterminé par une
forme ne peut subsister en dehors de la substance composée, « comme par
exemple la forme de la maison, à moins que par forme on entende techne ».
Le philosophe grec insiste même dans la Rhétorique sur le fait que toute
techne a affaire à l’universel et non au particulier6 : « aucune techne ne se
borne au particulier : ainsi, la médecine ne cherche point ce qui est propre
à la santé de Socrate ou de Callias, mais ce qui convient à tel ou tel en
général, et à telle ou telle maladie7. Car c’est là ce qui constitue la techne,
attendu que le nombre des matières particulières est infini, il est impossible
Aristote, Met Λ 3, 1070a13-15. Saint Thomas connaît et commente ce texte en
in Arist Met Λ, l. 3 n. 7-8 (In duodecim libros Metaphysicorum Aristotelis expositio,
éd. M. R. CATHALA – R. M. SPIAZZI, Marietti, Torino – Roma 1971, 2ª ed).
Aristote, Rhet I 2, 1356b30-35.
On remarque une différence entre ce texte et celui de la Métaphysique en ce qui
concerne l’exemple de la techne médicale : ici, Aristote, voulant souligner le caractère
universel de n’importe quelle techne et donc aussi de la rhétorique, considérée
relativement aux cas particuliers, ne fait aucune référence au but du technikos médecin

de les renfermer dans les limites d’une science. De même, la rhétorique

ne s’arrête point à ce qui est probable pour un particulier, pour Socrate ou
pour Hippias, par exemple. Elle ne s’occupe que de ce qui paraît tel à tous
les hommes en général, et par là elle ressemble encore à la dialectique ».
Le texte central pour ma recherche sera le Prologue du Commentaire de
Thomas d’Aquin aux Seconds Analytiques d’Aristote8, car nous y trouvons
une introduction à ce qu’est la logique, qui la caractérise justement comme
ars. Les premières lignes de ce Prologue vont nous offrir du matériel
utile pour tirer au clair dès le début ce que veut dire ars; à son tour, le
début du Commentaire à l’Ethique à Nicomaque9 nous aidera à préciser
sa nature de «certa ordinatio rationis». Nous pourrons de la sorte revenir
au commentaire aux Seconds Analytiques, afin de définir la nature de la
logique en tant qu’«ars artium», et son caractère formel.

1. Ars : certa ordinatio rationis

Thomas d’Aquin ouvre son commentaire aux Seconds Analytiques

par les mots suivants : « Sicut dicit Aristoteles in principio Metaphysicae,
hominum genus arte et rationibus vivit ».
Le renvoi explicite au début de la Métaphysique rend clair le fait que
la notion d’ars évoquée ici correspond à celle de techne décrite par Aristote
au sein du processus de connaissance comme le passage de l’individuel

de se servir des règles générales ainsi établies pour les appliquer au cas particulier ; ce
qui est explicite dans la Métaphysique (Met A 1, 981a18-20) quand est exprimée une
objection possible à l’efficacité de la techne par rapport à l’expérience : « Car ce n’est
pas l’homme que guérit le médecin, sinon accidentellement, mais Callias ou Socrate,
ou quelque autre individu qui se trouve appartenir au genre humain ». Il considère
important de le rendre explicite afin de ne pas donner l’impression d’être intéressé à
l’universel en lui-même, comme son maître. On peut ainsi dire que la techne est pensée
pour l’universel, mais que le technikos qui l’apprend doit exercer sa phronesis afin
d’appliquer la techne aux cas singuliers.
J’utiliserai l’abréviation ‘in Arist APo’ (Opera omnia, Editio Leonina, t. 1*/2:
Expositio libri Posteriorum, Commissio Leonina – J. Vrin, Roma – Paris 1989, 2ª ed.).
Pour une étude approfondie de la conception thomasienne de la logique, cf. l’ouvrage
fondamental de R. V. SCHMIDT, The Domain of Logic According to Saint Thomas
Aquinas, Martinus Nijhoff, The Hague 1966.
J’utiliserai l’abréviation ‘in Arist EN’ (Opera omnia, Editio Leonina t. 47/1 :
Praefatio. Sententia libri Ethicorum. Libri I-III, Ad Sanctae Sabinae, Roma 1969).

à l’universel saisi par de multiples expériences du même type10. Nous

relevons donc d’emblée le trait d’universalité qui doit caractériser la
logique; le trait relatif à l’ordre est quant à lui présent dans la formulation de
la définition d’ars que saint Thomas propose dans son Prologue, à savoir :
« nihil enim aliud ars esse videtur, quam certa ordinatio rationis, quomodo
per determinata media ad debitum finem actus humani perveniatur ».
En fait, cette définition ne dérive pas de la Métaphysique : Aristote
n’y donne pas de définition de techne, mais se concentre sur son origine
(γίγνεται), c’est-à-dire sur l’expérience, et sur le processus par lequel il
est possible d’atteindre une telle connaissance universelle. Saint Thomas,
pour sa part, conformément aux canons rhétoriques et pédagogiques d’une
introduction, cherche à susciter la docilitas en celui qui écoute11 : il offre par
conséquent une description-définition d’ars qui semble avoir des liens avec
la définition de techne comme systema que plusieurs témoins, tels Philon
d’Alexandrie, Galien et Sextus Empiricus, mais aussi l’un des scholiastes
byzantins du grammairien Denys le Thrace, font remonter aux Stoïciens.
Comme exemple, on peut prendre la double formulation du pseudo-
Galien : « la techne est un système – une mise ensemble (σύστημα) – de
compréhensions exercées ensemble (συγγεγυμνασμένων) vers un but
utile parmi ceux qui se trouvent dans la vie ; ou bien ainsi : la techne
est un système – une mise ensemble (σύστημα) – exercé ensemble
(συγγεγυμνασμένον) de compréhensions qui ont une relation vers un
unique but»12. Le rapport entre l’occurrence du mot latin ordinatio dans la
définition de saint Thomas et celle du mot grec ‘σύστημα’ dans le passage
de Galien est intensifié par la présence du participe ‘συγγεγυμνασμένον’ ;
car qu’il se réfère aux compréhensions – comme dans la première
version – ou directement au σύστημα, puisqu’il contient la préposition
‘syn’ – ‘avec’ – comme le mot ‘σύστημα’ lui-même, cela met l’accent sur
l’idée d’ensemble de parties organisées, fruit d’un exercice (grâce au verbe
‘γυμνάζομαι’) et, pourrait-on dire, d’une régularité.
Il est d’ailleurs possible non seulement de rapprocher le mot latin
‘ordinatio’ du mot grec ‘σύστημα’, mais aussi de reconnaître dans certaines

Aristote, Met Α 1, 981a5-7.
Thomas d’Aquin, in Arist De Anima.
Pseudo-Galien, Definitiones medicae 7 (Vol. XIX K p. 350): « τέχνη ἐστὶ
σύστημα ἐκ καταλήψεων συγγεγυμνασμένων πρός τι τέλος εὔχρηστον τῶν ἐν
τῷ βίῳ. ἢ οὕτως· τέχνη ἐστὶ σύστημα ἐκ καταλήψεων συγγεγυμνασμένον ἐφ’ ἓν
τέλος τὴν ἀναφορὰν ἐχόντων ».

formulations une même référence à une certaine fin («πρός τι τέλος»),

vers laquelle l’organisation est orientée. Le philosophe péripatéticien
Aspasius, au début de son commentaire à l’Ethique à Nicomaque13, fait
des remarques intéressantes à propos du mot ‘techne’, en relevant sa
plurivocité (« λέγεται δὲ τέχνη παρ’ αὐτοῖς τριχῶς », « ‘techne’ est dite
par eux de trois manières » ) ; ce qui explique que le mot ‘techne’ est
employé dans un sens générique (« ‘techne’ est dite le genre de toutes les
technai »14, c’est-à-dire des technai théorétiques, pratiques et poïétiques)
au commencement de l’Ethique à Nicomaque, alors que dans le livre VI
de ce même traité il est employé stricto sensu (« ἰδίως ») pour décrire
uniquement ce qui a trait à la production. Par rapport au sens générique
du terme, Aspasius admet que « quelqu’un pourrait définir ce que l’on
appelle techne comme un ensemble organisé de pensées qui amènent
à un seul but » (« ὁρίσαιτο δ’ ἄν τις τὴν οὕτω λεγομένην τέχνην
σύστημα ἐκ θεωρημάτων εἰς ἓν τέλος φερόντων »). Il est évident
qu’avec ce «quelqu’un» il veut indiquer les Stoïciens, même s’il change
l’expression typiquement stoïcienne ‘ἐκ καταλήψεων ’ en quelque chose
de plus péripatéticien comme ‘ἐκ θεωρημάτων’. Le mot ‘σύστημα’ est
toutefois encore là, avec sa valeur de rassemblement organisé, auquel a
été explicitement donné une orientation vers un but unique. En retournant
au texte de saint Thomas, on remarque quelque chose de semblable.
J’avais déjà suggéré la possibilité d’un rapprochement entre ‘σύστημα’ et
‘ordinatio’, terme utilisé par saint Thomas dans sa définition à l’intérieur
de l’expression ‘certa ordinatio rationis’ ; mais si l’on passe de son in Arist
APo à in Arist EN, on a confirmation du lien entre ces deux formulations
prises dans leur totalité. Souvenons-nous que celle de saint Thomas se
poursuit de la manière suivante : « quomodo per determinata media ad
debitum finem actus humani perveniatur ».
« Sicut Philosophus dicit in principio Metaphysicae sapientis est
ordinare » : ainsi le Docteur Angélique ouvre-t-il son commentaire, en
expliquant ces mots par le fait qu’il est propre au sage d’ordonner15 et en
ajoutant des réflexions à propos de la capacité de la raison d’avoir affaire à

Aspasius, Commentaire à l’Ethique à Nicomaque d’Aristote 2.15-3.3 (CAG,
vol. XIX pt. 1, éd. G. HEYLBUT, Reimer, Berlin 1889).
Aspasius, Commentaire à l’Ethique à Nicomaque d’Aristote 2.16-17 : « τὸ
γένος τῶν τεχνῶν ἁπασῶν τέχνη λέγεται ».
Saint Thomas ouvre sa Summa contra Gentiles avec cette même remarque.

l’ordre, en particulier de connaître l’ordre qui subsiste entre les choses, alors
que les facultés sensibles n’ont que la capacité de connaître quelque chose
« absolute », c’est-à-dire chaque chose de manière séparée, et non dans son
lien avec les autres choses. Avant de considérer les différentes manières
dont la raison entre en relation avec l’ordre, saint Thomas distingue deux
types d’ordre qui peuvent être relevés dans les choses : l’un est celui des
parties par rapport au tout, l’autre est celui des choses par rapport à un
but ; or c’est ce dernier qui est jugé le plus important (« principalior »), à
savoir celui des choses par rapport à un but (« ordo rerum in finem »). Nous
trouvons donc dans ces lignes d’ouverture d’in Arist EN la raison pour
laquelle dans le Prologue d’in Arist APo saint Thomas a tenu à préciser la
nature finale de cette certa ordinatio que l’ars dénote16. Mais dans la suite
d’in Arist EN, il nous aide aussi à comprendre pourquoi dans in Arist APost
il a voulu indiquer qu’il s’agit d’une « certa ordinatio rationis », sous-
entendant ‘humanae’17 ; ici, il me semble, nous avons une contribution
importante et originale de la part du Docteur Angélique. « Ordo autem
– distingue saint Thomas – quadrupliciter ad rationem comparatur » ;
toutefois, avec la première distinction il met au clair que la raison n’est
pas responsable de n’importe quel type d’ordre ; par exemple, l’ordre des
choses naturelles (ordo rerum naturalium) est quelque chose que « ratio
non facit, sed solum considerat ». De cette manière il fixe des limites
au pouvoir de la raison humaine, tout en lui reconnaissant une plasticité
particulière, qu’il manifeste à travers les trois distinctions suivantes :
« alius autem est ordo, quem ratio considerando facit in proprio actu, […].
Tertius autem est ordo, quem ratio considerando facit in operationibus
voluntatis. Quartus autem est ordo, quem ratio considerando facit in
exterioribus rebus, quarum ipsa est causa, […] »18. La capacité de la raison
humaine à connaître un ordre dont elle n’est pas responsable, lorsqu’elle

Cf. Summa Theologiae I, q. 103, a. 2 ad 2: « Ad secundum dicendum quod
philosophus loquitur de finibus artium, quarum quaedam habent pro finibus operationes
ipsas, sicut citharistae finis est citharizare; quaedam vero habent pro fine quoddam
operatum, sicut aedificatoris finis non est aedificare, sed domus. Contingit autem
aliquid extrinsecum esse finem non solum sicut operatum, sed etiam sicut possessum
seu habitum, vel etiam sicut repraesentatum, sicut si dicamus quod Hercules est finis
imaginis, quae fit ad eum repraesentandum ».
Il y a dans d’autres textes l’expression ‘divina ordinatio’.
Ce qui dans le commentaire aux Analytiques Seconds est considéré dériver de
actes de la main (manus actus).

devient active, « considerando », devient aussi productive (‘facit’) d’un

ordre qui peut concerner soit ses propres actes, c’est-à-dire le domaine
logique, sur lequel nous allons nous arrêter dans le prochain paragraphe,
soit les opérations de la volonté, c’est-à-dire le domaine de l’agir, de la
praxis, soit enfin ses opérations productives, c’est-à-dire le domaine de la
poiesis. Ne s’agit-il pas là du développement de l’idée de techne présentée
par Aristote au début de Métaphysique A1 et éclaircie par Aspasius dans
le texte cité ? Ce passage de « non facit, sed solum considerat » au triple
« considerando, facit » ne veut-il pas exprimer l’idée d’Aristote selon
laquelle « la techne apparaît (γίγνεται) lorsque d’une multitude de notions
issues de l’expérience se dégage un seul jugement universel applicable
à tous les cas semblables »19 ? Après tout, Aristote voulait souligner que
le processus de connaissance de l’homme ne peut atteindre un certain
degré d’abstraction qu’après avoir reconnu un certain type de relation de
ressemblance entre les choses, qui avaient été distinguées par la vue. La
limite ou l’enracinement au niveau du réalisme que saint Thomas veut fixer
avec le « non facit, sed solum considerat », laisse tout de même à la raison
la liberté de s’exprimer avec une grande plasticité pour donner sa propre
organisation aux connaissances qui dérivent de ses propres actes, ou bien
des actes de sa volonté, ou encore de sa capacité productive, de manière
à produire différentes sciences ou artes20, à savoir celles de la logique
(autrement dite rationalis philosophia), de la moralis philosophia ou des
artes mechanicae21.

Aristote, Met Α 1, 981a5-7, traduction TRICOT modifiée, dans Aristote,
Métaphysique, tome 1, livres A-Z, traduction et notes par J. TRICOT, Vrin, Paris 2000.
C’est saint Thomas lui-même qui utilise le mot ‘scientia’ au lieu d’ars : « Et
quia consideratio rationis per habitum scientiae perficitur, secundum hos diversos
ordines quos proprie ratio considerat sunt diversae scientiae ». Aristote, toujours dans
la Métaphysique A 1, souligne d’ailleurs explicitement qu’au niveau du processus de
connaissance, à cause de leur caractère universel, il n’y pas de distinction entre techne
et episteme. Leur différence est par contre relevée dans le livre VI de l’Ethique à
Nicomaque et concerne la nature de leurs objets, contingente pour l’une, nécessaire
pour l’autre.
Il faut préciser que saint Thomas mentionne aussi les sciences qui dérivent du
premier type d’ordre, celui que la raison « non facit, sed considerat » : il s’agit des
sciences théorétiques. Cela soulève une question quant à une sorte de chevauchement
de différents ordres : même si l’objet de connaissance a en lui-même un ordre qui n’est
pas le résultat de la raison, la constitution d’une connaissance en science demande le
rôle actif (facit) de la raison elle-même. Peut-être que c’est pour cela que la logique est

Revenons donc à notre texte d’in Arist APost, notamment à la première

partie de la formule définitoire « certa ordinatio rationis », pour arriver
enfin à préciser la nature d’ars de la logique. Parmi les trois mots, prenons
d’abord ‘ordinatio’ et remarquons la différence avec le mot ‘ordo’ utilisé
en in Arist EN : une explication possible peut être trouvée à partir de la
présence du suffixe ‘-tio’ qui en général donne aux substantifs de dérivation
verbale la référence à une action dans son résultat22 ; ainsi, l’ordinatio est la
structure ordonnée qui ressort d’un certain type d’ordre. En outre, d’après
saint Thomas23, l’ordinatio est l’ordo pris abstracte (en opposition avec
concretive); or ceci rappelle la manière dont Aristote fonde la techne ; ainsi
donc, si l’on associe abstracte à universaliter ou formaliter, on pourra trouver
confirmation du caractère universel, voire formel, de n’importe quelle ars.
Passons au mot ‘rationis’. Grâce à la distinction en types d’ordre
avec lesquels la raison est en relation, il est maintenant clair que l’ajout
de « rationis », qui n’a pas de correspondant dans la formule grecque, sert
à limiter la classe des résultats d’un ordre à ces trois types que la raison
«considerando facit» (parmi lesquels il y a l’ordre d’où provient la logique),
parce que la totalité de la formule définitoire explique (« enim ») ce qui
précède, à savoir « ad actus humanos faciliter et ordinate perficiendos
diversae artes deserviunt ». La diversité des actes humains ainsi que la
diversité correspondante des artes24, c’est justement ce que nous venons

placée par saint Thomas – à la suite de Boèce – à l’intérieur des sciences théorétiques,
avec une fonction précise, à savoir « adminiculum quoddam ad alias scientias […]
prout ministrat speculationi sua instrumenta, scilicet syllogismos et diffinitiones et alia
huiusmodi, quibus in scientiis speculativis indigemus » (Super De Trinitate, pars 3,
q. 5, a. 1 ad 2) et « ipsa docet modum procedendi in omnibus scientiis » (Super De
Trinitate, pars 3, q. 6, a. 1 ad 13).
Cet élément renforcerait d’ailleurs le parallèle avec le mot grec systema,
puisque ce dernier contient lui-même un suffixe, -ma, qui porte aussi la valeur de
résultat d’une action, en l’occurrence de la mise ensemble ; voir, par exemple, pragma,
comme résultat d’une praxis, d’une action.
Voir, par exemple, Thomas d’Aquin, II Sent., d. 9, q. 1, a. 1 ad 2 : « Ordo potest
sumi dupliciter : vel secundum quod nominat unum gradu tantum, sicut qui sunt unius
gradus, dicuntur unius ordinis ; et sic ordo est pars hierarchiae ; vel secundum quod
nominat relationem quae est inter diversos gradus, ut ordo dicatur ipsa ordinatio ; et
sic sumitur quasi abstracte, et sic ponitur in definitione hierarchiae ; primo autem modo
sumitur concretive, ut concretive, ut dicatur ordo unus gradus ordinatus ».
A propos de la pluralité des artes, voire technai, que l’être humain, à la
différence des autres animaux, est capable d’apprendre et d’exercer, voir Galien,

de relever dans le passage d’in Arist EN. Je voudrais faire aussi quelques
remarques à propos des adverbes ‘faciliter’ et ‘ordinate’, qui se répètent
deux fois dans la suite du texte, afin de souligner le type de contribution que
toute ars peut apporter à l’accomplissement de n’importe quel acte humain.
Nous avons suffisamment vu ce qu’il en est d’ordinate : c’est justement
parce que l’ars « n’est rien d’autre que certa ordinatio rationis » qu’elle
contribue à accomplir ordinate certains actes humains ; l’adverbe faciliter
évoque l’idée d’un habitus, étant donné que l’ars, voire la techne, a déjà
été considérée par Aristote comme l’une des vertus intellectuelles. Il s’agit
d’une facilité acquise, après l’exercice répété d’actes qui présupposent
un effort, et donc une certaine difficulté, comme saint Thomas le relève
dans les termes suivants25 : « ex consuetudine efficitur aliquid facile et
delectabile quod prius erat difficile : et hoc est signum habitus generati,
scilicet delectatio operis ». Rappelons que la facilité recherchée ne l’est
pas pour elle-même, mais justement pour concentrer davantage les efforts
sur le but qu’on veut atteindre.
Il nous reste encore à analyser l’adjectif ʻcertaʼ ; il y a en premier
lieu la question de la qualification qu’il est censé donner. La concordance
faite dans l’apparat critique de la première édition de la Léonine avec un
passage de la Summa Theologiae26, où saint Thomas définit la lex comme

Adhortatio ad artes addiscendas 1, 1-9 (Galien, tome II, Exhortation à l’étude de la

médecine. Art médical, texte établi et traduit par V. BOUDON, Les Belles Lettres, Paris
2002). J’ai consacré quelques notes à ce sujet dans « Logos e techne. Claudio Galeno e
Clemente Alessandrino, sulle basi formative necessarie alla perfezione dell’uomo » in
F. CARDERI – M. MANTOVANI – G. PERILLO (a cura di), Momenti del logos. Ricerche del
“Progetto LERS” (logos, episteme, ratio, scientia) in memoria di Marilena Amerise e
Marco Arosio, Edizioni Nuova Cultura, Roma 2012, pp. 123-158.
Voir, par exemple, le passage suivant, Thomas, in III Sent., d. 33, q. 1, a. 2 ad
2 : « ex consuetudine efficitur aliquid facile et delectabile quod pius erat difficile : et
hoc est signum habitus generati, scilicet delectatio operis » (« par l’habitude une action
qui était difficile à réaliser devient facile et délectable, le signe de l’acquisition d’un
habitus étant en effet le plaisir à accomplir l’activité», trad. dans J.-B. ÉCHIVARD, Une
introduction à la philosophie. Les proèmes des lectures de saint Thomas d’Aquin aux
œuvres principales d’Aristote, vol. 2, Science rationnelle et philosophie de la nature,
François-Xavier de Guibert, Paris 2005, p. 55).
Thomas d’Aquin, Summa Theologiae I-II, q. 90, a. 4 co: « Unde promulgatio
necessaria est ad hoc quod lex habeat suam virtutem. Et sic ex quatuor praedictis
potest colligi definitio legis, quae nihil est aliud quam quaedam rationis ordinatio ad
bonum commune, ab eo qui curam communitatis habet, promulgata ». Je voudrais
aussi signaler Summa Theologiae II-II, q. 88, a. 5 co : « Et ideo ipsa ordinatio actuum

« quaedam ordinatio… » pourrait nous amener à prendre ‘certa’, ici,

comme un synonyme du ‘quaedam’ que l’on trouve dans la Summa : de
cette manière, l’idée que l’ars serait alors «une certaine mise en ordre»
laisserait ouvert le choix entre une variété d’ordinationes rationis, et
notamment, d’après in Arist EN, les trois types que nous avons abordés,
à savoir « est ordo, quem ratio considerando facit in proprio actu », « est
ordo, quem ratio considerando facit in operationibus voluntatis», «est
ordo, quem ratio considerando facit in exterioribus rebus, quarum ipsa
est causa, […] ». Toutefois, si l’idée d’un sens de ‘certa’ comparable à
l’adjectif indéfini ‘quaedam’ est très séduisante, elle ne correspond pas
à l’usage que saint Thomas en fait normalement, qui est plutôt celui de
qualifier quelque chose comme étant certain, sûr, déterminé, clair. Mais
qu’est-ce que cela veut dire ? Que l’ars est une mise en ordre précise en
ce sens qu’il n’y en a qu’un type ? Cela risque de n’être pas cohérent avec
l’idée de saint Thomas que l’être humain est capable d’acquérir une variété
d’artes, et que cette variété, comme nous venons de le voir, est basée sur
la variété d’ordres que la raison est capable de donner à différents actes
humains. Il y a cependant aussi la possibilité que ce ‘certa’ ne serve pas à
préciser en quantité l’ordinatio, mais à en définir le caractère bien établi
à partir d’une distinction bien établie des règles27. L’adjectif est utilisé
avec ce sens-ci dans d’autres textes de saint Thomas28 pour marquer la

cuiuscumque virtutis in servitium Dei est proprius actus latriae. Manifestum est
autem ex praedictis quod votum est quaedam promiss/io Deo facta, et quod promissio
nihil est aliud quam ordinatio quaedam eius quod promittitur in eum cui promittitur.
Unde votum est ordinatio quaedam eorum quae quis vovet in divinum cultum seu
obsequium ».
Voir, par exemple, Thomas d’Aquin, Summa Theologiae I-II, q. 104, a. 4 co:
« Respondeo dicendum quod, cum lex sit quasi quaedam ars humanae vitae instituendae
vel ordinandae, sicut in unaquaque arte est certa distinctio regularum artis, ita oportet
in qualibet lege esse certam distinctionem praeceptorum, aliter enim ipsa confusio
utilitatem legis auferret. Et ideo dicendum est quod praecepta iudicialia veteris
legis, per quae homines ad invicem ordinabantur, distinctionem habent secundum
distinctionem ordinationis humanae ».
Thomas d’Aquin, Summa Theologiae II-II, q. 49, a. 5 ad 2 : « Particularia
autem operabilia, in quibus prudentia dirigit, recedunt praecipue ab intelligibilium
conditione, et tanto magis quanto minus sunt certa seu determinata. Ea enim quae
sunt artis, licet sint singularia, tamen sunt magis determinata et certa, unde in pluribus
eorum non est consilium, propter certitudinem, ut dicitur in III Ethic. Et ideo quamvis
in quibusdam aliis virtutibus intellectualibus sit certior ratio quam prudentia, tamen

différence entre l’ars, qui porte sur l’universel, et la prudentia, qui porte
sur le particulier : toutes les deux correspondent à une mise en ordre, mais
dans un cas, puisque les choses-mêmes sont « determinata et certa »,
elle aussi est « certa » ; alors que dans l’autre, les choses étant « varia et
incerta », la mise en ordre ne peut pas trouver en elle-même une régularité ;
c’est pour cela que l’homme « bene ratiocinativus » est capable de mettre
en lien l’ars acquise et la prudentia, de manière à appliquer les principes
universaux aux cas particuliers.
Ces remarques à propos de la définition d’ars et de ces deux adverbes,
faciliter et ordinate, doivent nous permettre de montrer comment tout ce
discours préalable servait à saint Thomas pour caractériser la logique.

1. La logique, ars artium

Saint Thomas ouvre son Prologue d’in Arist APo avec une référence à
l’ars et à sa valeur pour l’identité de l’homme par rapport aux autres
animaux, dont il se sert pour proposer une introduction à la logique ;29 la
raison, qui a un rôle de direction dans n’importe quelle ars, est capable de
donner un ordre à ses propres actes, ce dont résulte l’ars logique. Car la
raison, dans sa partie intellective (intellectiva pars), est en mesure de
réfléchir sur elle-même : « et haec ars est logica, id est rationalis scientia »
– dira saint Thomas à la fin du raisonnement – « quae non solum rationalis
est ex hoc, quod est secundum rationem, quod est omnibus artibus commune,
sed ex hoc, quod est circa ipsum actum rationis sicut circa propriam
materiam ». L’explication de l’appellation «rationalis», qui, effectivement,
semble de prime abord redondante, est intéressante, car le caractère rationnel
de la logique est déjà mis au clair par le fait même qu’elle est une ars et, par
conséquent, une « certa ordinatio rationis ». Mais la logique a quelque
chose de différent des autres artes : la raison n’y est pas seulement engagée
dans le rôle ordinateur – d’où provient ordinatio rationis – mais aussi, en
ses actes, dans le rôle de matière qui doit être ordonnée, ce qui explique

ad prudentiam maxime requiritur quod sit homo bene ratiocinativus, ut possit bene
applicare universalia principia ad particularia, quae sunt varia et incerta ».
Il vaut la peine de rappeler que les commentateurs néoplatoniciens fournissaient
une introduction à la logique d’Aristote avant de commenter les Catégories, voir par
exemple l’Isagoge de Porphyre.

justement l’appellation « rationalis scientia »30. Mais revenons en arrière,

pour suivre de près l’argument qui amène à reconnaître la logique comme
ars de la raison : « Si igitur ex hoc, quod ratio de actu manus ratiocinatur
adiventa est ars aedificativa vel fabrilis, per quas homo faciliter et ordinate
huiusmodi actus exercere potest, eadem ratione ars quaedam necessaria
est, quae sit directiva ipsius actus rationis, per quam scilicet homo in ipso
actu rationis ordinate, faciliter et sine errore procedat. Et haec ars est
logica, idest rationalis scientia ». La raison, répétons-le, est capable de
réfléchir sur ses propres actes. Or, par la réflexion sur les actes de la main,
la raison est déjà capable de produire une ordinatio, c’est-à-dire une ars : il
s’agit de l’une des artes produites par un ordre du troisième type selon le
schéma d’in Arist EN, comme l’architecture, grâce à laquelle l’homme peut
procéder avec ordre et facilité – on a donc ici la deuxième occurrence du
couple d’adverbes ‘ordinate’ et ‘faciliter’. Par conséquent, et d’autant plus
qu’elle part de la réflexion sur elle-même, la raison sera obligée – saint
Thomas parle en termes de nécessité – de produire une ars qui donne un
ordre aux actes de l’homme, c’est-à-dire – et voici la troisième occurrence
du couple d’adverbes –, qui en rend l’exécution facile et ordonnée, mais
aussi – voici l’élément nouveau – « sine errore ». Cette ars, c’est justement
la logique, de la même façon que dans in Arist EN il a été indiqué à propos
du premier type d’ordre que la ratio « considerando facit ». Il y a sans doute
un ordre naturel de la raison que la raison même peut arriver à connaître
(« considerando ») ; mais la mise en ordre, l’ordinatio, qui correspond à la
logique, c’est le résultat de l’ordre possible que la raison trouve afin de
rendre l’exécution de ses actes facile et sans erreur. La question que nous
pouvons poser concerne justement l’ajout d’une troisième locution
adverbiale, « sine errore », à « ordinate » et « facile » : pourquoi, lorsque
saint Thomas se réfère soit à n’importe quelle ars, soit au cas particulier des

Il est ici évident que ars et scientia sont utilisés de manière synonyme ; mais il
y a aussi des passages où saint Thomas veut marquer une distinction, comme le suivant,
Summa Theologiae I-II, q. 57, a. 3 ad 3 : « Ad tertium dicendum quod etiam in ipsis
speculabilibus est aliquid per modum cuiusdam operis, puta constructio syllogismi
aut orationis congruae aut opus numerandi vel mensurandi. Et ideo quicumque
ad huiusmodi opera rationis habitus speculativi ordinantur, dicuntur per quandam
similitudinem artes, sed liberales; ad differentiam illarum artium quae ordinantur ad
opera per corpus exercita, quae sunt quodammodo serviles, inquantum corpus serviliter
subditur animae, et homo secundum animam est liber. Illae vero scientiae quae ad
nullum huiusmodi opus ordinantur, simpliciter scientiae dicuntur, non autem artes ».

artes productrices, n’utilise-t-il que le couple adverbial « ordinate » et

« faciliter », tandis que dans le cas de l’ars de la logique, il ajoute aussi
«sine errore» ? Cela veut-il dire que le fait de procéder selon la vérité ne
concerne que la logique, alors que les autres artes peuvent procéder dans
l’erreur ? Assurément non, puisque sans la vérité parmi les buts, il ne peut
s’agir d’une ars ! Je crois que pour comprendre le sens de ce « sine errore »
nous pouvons demander de l’aide à l’un des pères de la logique
contemporaine, Gottlob Frege, notamment lorsqu’il relève que la tâche
propre de la logique est « de découvrir les lois de l’être vrai »31. Cette
remarque semble bien en ligne avec le rôle que saint Thomas, dans son
Commentaire au De Trinitate de Boèce, attribue à la logique par rapport aux
sciences spéculatives dont le but est justement le vrai32 : « Res autem, de
quibus est logica, non quaeruntur ad cognoscendum propter se ipsas, sed ut
adminiculum quoddam ad alias scientias ». Dans un passage suivant du
même commentaire, là où saint Thomas aborde la question de l’ordre à
suivre pour l’apprentissage des disciplines (sciences ou artes) et donne la
raison pour laquelle il faut commencer par l’étude de la logique, il dit « quia
aliae scientiae ab ipsa dependent, in quantum ipsa docet modum procedendi
in omnibus scientiis »33 ; certes, pour que la logique puisse enseigner le
modus procedendi dans toutes les autres sciences, elle doit être construite de
manière à permettre d’apprendre la façon de procéder « sine errore ». Mais
qu’est-ce que cela veut dire ? Je crois que cela a trait au caractère, si l’on
peut dire, spécialement universel de la logique, qui en fait une ars artium.
Mais il est important de préciser d’emblée que cette universalité ne fait pas
que la logique contient les autres artes : dire que la logique est ars artium
ne signifie pas qu’elle est l’ars architectonique34 relativement aux autres.

G. FREGE, « The thought: A Logical Inquiry », Mind, LXV-259 (1956) 289-311:
« The word “true” indicates the aim of logic as does “beautiful” that of aesthetics or
“good” that of ethics. All sciences have truth as their goal; but logic is also concerned
with it in a quite different way from this. It has much the same relation to truth as
physics has to weight or heat. To discover truths is the task of all sciences; it falls to
logic to discern the laws of truth. […] Rules for asserting, thinking, judging, inferring,
follow from the laws of truth ».
Thomas d’Aquin, in Boe De Trin, pars 3, q. 5, a. 1 ad 2.
Thomas d’Aquin, in Boe De Trin, pars 3, q. 6, a. 1 ad 13.
Voir, par exemple, Thomas d’Aquin, in Arist EN, lib. 1, l. 1, n. 17 : « Et
dicit quod in omnibus artibus vel virtutibus hoc communiter est verum, quod fines
architectonicarum sunt simpliciter quoad omnes magis desiderabiles, quam fines
artium vel virtutum, quae sunt sub principalibus. Quod probat per hoc, quod homines

Pour le formuler différemment : nous avons vu que toute ars se constitue

par des règles générales à partir d’expériences similaires. Cette universalité
est nécessaire pour saisir le pourquoi de l’objet que l’on veut connaître.
Mais celui qui a appris et exerce une ars n’est pas reconnu comme artifex
par l’application conséquente et indiscriminée de ces règles ; celle-ci peut
même causer de graves erreurs. Pensons, par exemple, à la médecine : il faut
exercer la prudentia (phronesis en grec) pour juger les cas particuliers. Par
contre, dans le cas de l’ars qu’est la logique, une fois que l’artifex a compris
le pourquoi et a établi les « determinata media » pour arriver « ad debitum
finem », il peut procéder en suivant les règles sans risquer de tomber dans la
faute, sine errore35. C’est pour cette raison que j’ai attribué à la logique un
caractère spécialement universel, puisque, comme les autres artes, elle se
constitue de manière universelle, mais à la différence des autres, elle peut
aussi procéder de manière universelle. Pour cela, elle peut être considérée
modus operandi, ou comme saint Thomas l’indique in in Arist APo, « ars
artium » ; dans cette tâche, il faut que la logique ait davantage un caractère
universel, voire formel. Et le fait de lui reconnaître cette tâche ne me semble
pas contraposé au constat que fait J.-B. Échivard quand il commente
justement ce passage d’in Arist Apo et la formule « ars des arts» : «Thomas
ne parle pas de “forme” universelle de la raison »36. C’est exact, il n’en
parle pas, parce que, justement, il n’existe pas de forme universelle de la
raison ; s’il est vrai que l’ordre des choses naturelles n’est pas propre à la
raison, celle-ci est toutefois capable d’organiser ses actes en leur donnant
une certaine forme qui de par sa nature est universelle et adéquate à une
certaine matière, afin que sa réalisation procède de manière facile et
ordonnée. La raison est capable de réfléchir sur elle-même, mais pas de
manière absolue et totale, sinon il ne s’agirait pas de la raison humaine et la
logique s’identifierait avec la sophia. Or nous savons que cela n’est pas le
cas pour saint Thomas, ni d’ailleurs pour Aristote, qui se distingue de son
maître en niant justement l’identification de la dialectique avec la

persequuntur, id est quaerunt, illa, id est fines inferiorum artium vel virtutum gratia
horum, idest propter fines superiorum. Litera autem suspensiva est, et sic legenda:
quaecumque sunt talium sub una quadam virtute... in omnibus utique architectonicarum
fines ».
La méthode actuelle des tables de vérité peut constituer un bon exemple: en sa
qualité de mécanique, elle résume les trois attributs – exprimés chez saint Thomas de
manière adverbiale – à savoir le fait d’être facile, ordonnée et sans erreur.
ÉCHIVARD, Une introduction à la philosophie, vol. 2, p. 57.

philosophie. Par contre, nous pouvons nous appuyer sur des textes du
Docteur Angélique pour lui attribuer l’idée qu’une variation de la forme est
possible à l’intérieur d’une ars et, donc, à l’intérieur de la logique ; nous
trouverons de la sorte la réponse à la deuxième question posée au début de
ce travail. Ainsi, dans le commentaire à la Physique d’Aristote37, afin de
rendre clair que les animaux, à la différence des humains, procèdent dans
leurs opérations selon un principe naturel et non selon l’intellect (per
intellectum), saint Thomas montre qu’il leur manque la possibilité de varier,
qui caractérise ce qui provient de l’intellect et, ajoute-t-il, de l’ars : « omnis
enim hirundo similiter facit nidum, et omnis araneus similiter facit telam,
quod non esset si ab intellectu et arte operarentur » ; pour expliquer en quoi
consiste cette variation, il fait référence à la capacité de l’artifex de varier la
forme qu’il juge en tant qu’objet de son ars: « omnis enim hirundo similiter
facit nidum, et omnis araneus similiter facit telam, quod non esset si ab
intellectu et arte operarentur: non enim omnis aedificator similiter facit
domum, quia artifex habet iudicare de forma artificiati, et potest eam
variare »38. De nouveau, si cela vaut pour l’ars en général, cela vaut a
fortiori pour l’ars artium, la logique. Nous pouvons ainsi voir même les
différents systèmes de la logique contemporaine comme une variation de la
forme sur la base de la matière choisie, avec en commun un même but.
Mais, comme je l’ai indiqué dans l’introduction, je ne veux pas rentrer dans
la question de la forme logique et, par voie de conséquence, de la matière
logique. Dans l’étude proposée ici, j’ai voulu me concentrer sur un sens plus
large de ‘forme’, qui correspond à celui d’ars, ainsi que nous pouvons le
relever dans le livre Λ de la Métaphysique d’Aristote. Dans les textes de
saint Thomas que j’ai cités tout au long de ce travail, le rapprochement entre
forme et ars reste implicite, mais ses attributs, notamment, celui de
l’universalité et de l’ordre, sont bel et bien présents. Ces textes ont, de fait,
montré l’aspect universel et ordonné de n’importe quelle ars, et plus
particulièrement de l’ars logique.

Thomas d’Aquin, in Arist Phys, lib. 2, l. 13, n. 5.
Cf. aussi Thomas d’Aquin, De potentia, q. 6, a. 1 ad 12: « Ad decimumse-
cundum dicendum, quod ars divina non totam seipsam explicat in creaturarum
productione; et ideo secundum artem suam potest alio modo aliquid operari quam
habeat cursus naturae; unde non sequitur quod si potest facere contra cursum naturae,
possit facere contra suam artem: nam et homo artifex potest aliud artificiatum facere
per suam artem contrario modo quam prius fecit ».




L’étude des difficultés que pose à Robert Kilwardby l’analyse

logique des syllogismes sophistiques permet de suivre l’émergence d’une
interrogation théorique sur la forme et la matière syllogistiques. On y
voit comment les discussions sur la démarcation entre les arguments
matériellement et formellement sophistiques dans les Réfutations
sophistiques, déjà très élaborées au moment où démarrent les premiers
commentaires aux Premiers Analytiques, y jouent un rôle décisif1.
De nombreux spécialistes de la logique aristotélicienne, héritiers de
Jaeger et naturellement enclins à une approche formelle de la logique,
distinguent deux sens de « syllogisme ». Il s’agit de rendre compte de
l’existence d’inférences nécessaires, parfois appelées par Aristote
syllogismoi, qui ne suivent manifestement pas les modes et les figures
syllogistiques des Premiers analytiques : on les trouve dans les Topiques,
les Réfutations sophistiques, et dans certains passages des Premiers
analytiques eux-mêmes, pour lesquelles on parle volontiers de «syllogisme
au sens large». On traduit alors parfois syllogismos par «déduction»2, par
opposition au «syllogisme au sens strict» ou « syllogisme » tout court.
Cette distinction, sinon cette double traduction, n’est pas sans fondement,
puisqu’Aristote appelle « syllogismes » des arguments qui ne sont pas en

Laboratoire d’Études sur les Monothéismes (CNRS), 7 rue Guy Môquet, 94801
Villejuif. Email: brumberg@vjf.cnrs.fr
Comme l’avait souligné S. EBBESEN, « The Way Fallacies were treated in
Scolastic Logic », Cahiers de l’Institut du Moyen-Âge Grec et Latin (CIMAGL), 55
(2001) 107-134, p. 125.
Voir par exemple la traduction du terme syllogismos par J. BRUNSCHWIG
au début des Topiques (100a25-27): « un raisonnement déductif est une formule
d’argumentation dans laquelle, certaines choses étant posées, une chose distincte de
celles qui ont été posées s’ensuit nécessairement », Les Belles Lettres, Paris 2002,
p. 1 ; voir également G. STRIKER, Aristotle’s Prior Analytics, Book 1, Oxford 2009, qui
choisit « deduction » mais utilise ponctuellement « syllogism ».

modes et figures, et parfois, au contraire, les contraste avec les syllogismes3.

Elle se fait néanmoins au mépris d’une définition quasi identique du
« syllogisme » dans tous les textes précités, ainsi que dans la Rhétorique4.
Pour les commentateurs médiévaux, comme pour les exégètes tardo-
antiques, l’Organon a un auteur, Aristote, il a une cohérence interne
de principe, et la critique génétique n’est pas encore à l’ordre du jour.
Aussi tous les syllogismi sont-ils des syllogismes, l’absence de mise en
forme syllogistique effective dans le texte aristotélicien étant considérée
comme un non problème. C’est le cas dans les ouvrages de philosophie
naturelle dont l’étude systématique commence de manière concomitante
à celle des Premiers analytiques et des Topiques, où l’on reconstitue
systématiquement les syllogismes sous-entendus par Aristote : la théorie
du syllogisme démonstratif est spontanément perçue (et, au besoin, ré-
instaurée) comme la logique sous-jacente de la science. Il en est de
même dans les ouvrages logiques : on reconstitue des raisonnements en
« Barbara » et autres « Darii » dans les Topiques, et, ce qui nous intéresse
ici au premier chef, dans les Réfutations sophistiques. Dans le même
temps, le modèle alexandrin de divisions des syllogismes à l’aide du
couple matière-forme transmis par « Alexandre »5 commence à saturer

Voir J. BARNES, (« Proof and the Syllogism », in E. BERTI (ed.), Aristotle on
Science: The Posterior Analytics, Padua 1981, pp. 17-60) qui explique avec raison que
sans cette distinction le programme de réduction de tous les arguments valides aux
syllogismes n’aurait pas de sens.
Pour les Topiques, voir note 2 supra. Premiers analytiques 24b17-22 : « le
syllogisme est un discours, dans lequel certaines [choses] étant posées, quelque chose
d’autre que celles-ci en résulte nécessairement par le seul fait de celles-ci. Par le seul
fait de celles-ci : je veux dire que c’est par elles que le résultat est obtenu ; à son
tour, l’expression “c’est par elle que la conséquence est obtenue” signifie qu’aucun
terme étranger n’est en sus requis pour produire la conclusion nécessaire», trad. J.
TRICOT, Vrin, Paris 2001, pp. 4-5 (légèrement modifiée) ; Rhétorique 1356b15-18 :
« de l’existence de certaines choses, il résulte – à cause d’elles – une chose différente et
distincte d’elles, du seul fait que ces choses-là existent soit de manière universelle, soit
dans la plupart des cas, c’est ce qu’on appelle là un syllogisme et ici un enthymème »,
trad. P. CHIRON, Flammarion, Paris 2007, p. 129 ; Réfutations sophistiques 168a25:
« La déduction (= syllogismos) s’effectue à partir de choses posées de telle façon
qu’elles entraînent nécessairement l’assertion d’une chose différente d’elles, mais
qui résulte d’elles », trad. L. A. DORION, Aristote, les Réfutations sophistiques, Vrin,
Paris – Laval 1995, p. 119.
Sur « Alexandre », auteur anonyme de commentaires aujourd’hui perdus (mais
lus au Moyen Âge) aux Réfutations sophistiques, aux Premiers analytiques et aux

le champ de la logique, d’abord pour la classification des arguments

déficients dans les Réfutations sophistiques, puis pour toute enquête
s’intéressant à la définition du syllogisme et de la forme syllogistique
après la « redécouverte »6 des Premiers analytiques.
C’est ainsi que naît le problème spécifique que nous souhaitons
aborder ici : après le commentaire de l’Anonymus Cantabrigiensis au
Réfutations sophistiques, situé au tournant des XIIe et XIIIe siècles7, un

Seconds analytiques inspirés des commentateurs tardo-antiques et particulièrement de

Philopon, voir l’étude de S. EBBESEN, Commentators and Commentaries on Aristotle’s
Sophistici Elenchi : a Study of Post-Aristotelian Ancient and Medieval Writings on
Fallacies, Leiden 1981, vol. II, pp. 233-530, en particulier pp. 346-347 et vol. III,
pp. 4-7), et une série de publications du même auteur, notamment, « Analysing
Syllogisms or Anonymous Aureliensis III, the (presumbably) Earliest Extant Latin
Commentary on the Prior Analytics and its Greek Model », CIMAGL, 37 (1981) 1-20,
version actualisée dans ID., Greek-Latin Philosophical Interaction, vol. 1, Ashgate,
Aldershot – Burlington 2008, pp. 171-186, « Fragments of ‘Alexander’’s commentaries
on Analytica posteriora and Sophistici elenchi » (CIMAGL, 60 (1990) 113-120), parue
en 2008 dans EBBESEN, Greek-Latin Philosophical Interaction, vol. 1, op. cit., pp. 187-
202 et « Anonymi Parisiensis Compendium Sophisticorum Elenchorum: The Uppsala
Version », CIMAGL, 66 (1996) 253-312.
On sait que la syllogistique assertorique a été continûment transmise au Moyen
Âge par les opuscules de Boèce et que les Premiers analytiques étaient déjà connus
à l’époque d’Abélard. Ils ont eu une réception très intéressante au XIIe siècle, pour
laquelle Ch. Martin a récemment fait des hypothèses d’une grande pertinence (« “They
had added not a single tiny proposition” : The Reception of the Prior Analytics in the
First Half of the Twelfth Century », Vivarium, 48 (2010) 159-192). Mais cela ne veut
pas dire qu’il faille considérer l’apparition d’une exégèse complète et systématique des
Premiers analytiques au tournant des XIIe et XIIIe siècles comme un non-événement.
On a tendance à relativiser l’importance des débuts d’une lecture systématique du texte
des Premiers analytiques dans l’histoire de la logique, en insistant sur l’importance
des opuscules boéciens, lorsqu’on considère que l’essentiel des Premiers analytiques
consiste dans la syllogistique assertorique (livre I, chap. 1 à 7) et, parfois, modale (I,
chap. 8-26), à laquelle on reconnaît tout de même une réception spécifique au XIIIe
siècle. C’est précisément cette conception qui est écartée ici. Il s’agit au contraire de
montrer l’importance d’une lecture globale des Premiers analytiques (y compris la
fin du livre I et tout le livre II) comme porteurs d’une théorie du syllogisme à laquelle
contribuent de manière notoire les Réfutations sophistiques.
Éd. S. EBBESEN, à paraître. Voir J. BRUMBERG-CHAUMONT, « Form and Matter in
the Anonymus Cantabrigiensis », in B. BYDÉN – Ch. THOMSEN THÖRNQVIST (éds.), The
Aristotelian Tradition: The Reception of Aristotle’s Works on Logic and Metaphysics in the
Middle Ages, à paraître en 2017. Ce « Barbara sophistique » apparaît pour la première fois,
à notre connaissance, dans ce commentaire, et il est repris dans toute la tradition ultérieure.

certain nombre de paralogismes dans les commentaires aux Réfutations

sophistiques sont reformulés dans des arguments qui suivent les modes et les
figures syllogistiques, tout en étant parfois décrits comme « formellement
déficients ». L’exemple typique est, pour le paralogisme de l’accident, le
« Barbara sophistique » : « omne aes est naturale ; omnis statua est aes ;
ergo omnis statua est naturalis », présent dans l’Anonymus Cantabrigiensis.
Nous avons choisi de nous concentrer ici sur les ouvrages logiques
de Robert Kilwardby dont les thèses servent de référence aux auteurs de
sa génération, tels Albert le Grand, et de matrices pour la production de
nouveaux problèmes dans la génération suivante8. L’enquête s’appuie sur
des textes dont l’authenticité n’est pas mise en question, tels le commentaire
aux Premiers analytiques, aux Topiques, et le De Ortu scientiarum,
auxquels s’ajoute le commentaire aux Réfutations sophistiques, qui n’a pas
été attribué de manière définitive.

1. Arguments formellement et matériellement sophistiques

Au début des Topiques (I, 1), Aristote distingue parmi les raisonne-
ments éristiques 1) les raisonnements qui sont des syllogismes mais partent
de prémisses apparemment admises, 2) les raisonnements qui ne sont que
des apparences de syllogismes mais partent de prémisses véritablement
admises, 3) les raisonnements qui sont des apparences de syllogismes et
partent de prémisses apparemment admises, et 4) les paralogismes en un
sens qui est propre aux Topiques, qu’on retrouve aussi dans les Seconds
analytiques: ce sont les raisonnements syllogistiques qui partent de
prémisses apparemment propres à une science donnée, mais qui lui sont
en fait extrinsèques. Nous laissons ce sens de « paralogisme » de côté ici.
Cette classification a été universellement interprétée par tous les
commentateurs anciens, médiévaux et contemporains à partir de la
distinction entre la matière et la forme : les raisonnements de type 1 sont
matériellement éristiques, tandis que ceux de type 2 et 3 sont formellement

Voir notamment les commentaires aux Réfutations sophistiques anonymes
(Incerti Auctores) édités par EBBESEN en 1977, que nous évoquons plus loin en
conclusion, le commentaire de Simon de Faversham aux Réfutations sophistiques, le
commentaire de Boèce de Dacie aux Topiques.

La façon dont on comprend la typologie des arguments éristiques

des Topiques instrumente à son tour la manière dont le classement des
syllogismes sophistiques est interprété dans les Réfutations sophistiques.
Celle-ci dépend également de la manière dont la première phrase du traité
est lue, selon l’interprétation que l’on donnera du kai/et :

De sophisticis autem elenchis et (kai) de his qui videntur quidem

elenchi, sunt autem paralogismi sed non elenchi dicemus (Sophistici
Elenchi, AL VI/1-3, p. 5).

Ce qui peut se traduire selon que le kai / et est compris comme un kai
épéxégétique ou comme une conjonction (« numeraliter » dans l’exégèse
médiévale) :

Traitons maintenant a) des réfutations sophistiques et b) des

[arguments] qui ont l’apparence de réfutations mais sont des
paralogismes et non des réfutations.

Ou bien par :

Traitons maintenant des réfutations sophistiques, c’est-à-dire des

arguments qui se présentent comme des réfutations, mais qui sont
en fait des paralogismes et non des réfutations (Les Réfutations
sophistiques, trad. L.-A. Dorion, p. 119).

L’interprétation «numeraliter», comme une conjonction, du « kai

/ et », couplée avec l’introduction du couple matière-forme, induit la
classification suivante : a) les syllogismes matériellement sophistiques,
mais syllogistiques tout de même (les arguments de type n° 1 des
Topiques), et b) les raisonnements formellement éristiques qui sont des
apparences de syllogismes (les arguments nº 2 et 3 des Topiques), appelés
« paralogismes » dans les Réfutations sophistiques en un sens distinct de
celui des Topiques I, 1.
Il s’agit d’une interprétation spontanée, qu’on retrouve, par exemple,
dans l’introduction de Jacques Brunschwig à sa traduction des Topiques9
ou dans le commentaire de L.-A. Dorion aux Réfutations sophistiques10.
Elle est typique d’une conception moderne de la notion de forme logique

Op. cit., p. XXXVI.
Op. cit., p. 280.

en ce qu’elle présuppose naturellement qu’un défaut de matière n’entraîne

pas nécessairement la corruption de l’être syllogistique du raisonnement,
la forme étant «indifférente» au contenu avec lequel on la remplit, tandis
qu’un défaut de forme implique nécessairement que le raisonnement ne
peut plus être un syllogisme.
Ce présupposé pose un problème à nos maîtres du XIIIe siècle, pétris
d’hylémorphisme aristotélicien, où l’union de la forme et de la matière
est une « composition essentielle » dans laquelle l’absence d’un des deux
composants, aussi bien la matière que la forme, entraîne la non-existence
du composé. Ils sont en outre fortement influencés par l’interprétation
d’Avicenne, qui fait entrer la matière dans la définition des composés
hylémorphiques dont la forme ne peut être séparées de la matière, fût-
elle une matière « générale ». Ces notions s’appliquent au composé
hylémorphique qu’est le syllogisme, en dépit du fait qu’il est considéré
comme un artéfact. Non seulement un composé doit avoir une matière et
ne peut persister si sa matière est « corrompue », s’il n’a plus de matière du
tout, mais il doit avoir en outre une matière qui soit un substrat adapté à sa
forme, si ce n’est son substrat propre, comme dans le cas des êtres naturels.
D’un autre côté, il existe une équivalence spontanée, héritée des
commentateurs tardo-antiques, entre les notions de prémisses fausses
et celle de prémisses sophistiques. La notion de fausseté est elle-même
rattachée, quoique de manière implicite, à la notion de déficience : une
prémisse sophistique est une prémisse fausse, et, par conséquent une matière
déficiente pour la forme syllogistique. C’est notamment pour répondre à
ce problème que Robert Kilwardby va donner au syllogisme simpliciter,
une matière, les « lettres » elles-mêmes, les « termes transcendants »
selon la terminologie médiévale. Reste à savoir si ce dispositif permet de
répondre à la difficulté précise posée par le respect des modes et des figures
syllogistiques par des arguments formellement déficients : sont-ils tout de
même des syllogismes simpliciter ?
Quelques clarifications sur la notion de « matière du syllogisme »
s’imposent d’abord.

2. Matières du syllogisme

On peut dégager au moins deux acceptions distinctes de la notion de

« matière du syllogisme » :

[Sens 1] Une notion, dont témoigne Ammonius, est celle de « matière

des propositions ». Il s’agit d’un sens technique de « modalité matérielle »,
qui caractérise les relations entre les choses signifiés par les termes de
la proposition (relations nécessaires, impossibles, contingentes). Se sont
agrégées à ce sens des modalités aléthiques et épistémiques. Dans les
discussions sur la classification des arguments déficients, c’est le critère
aléthique qui a été privilégié : les arguments matériellement déficients sont
les arguments partant de prémisses fausses11 ; ils sont en outre éristiques
puisqu’ils se donnent l’apparence du vrai. C’est parfois ce sens que prend
la « matière du syllogisme ». Un exemple typique d’un syllogisme qui
pèche par la matière est alors le syllogisme « quoniam » des Analytiques
II, 2 et 4, ou bien le syllogisme « seulement-matériellement-déficient » des
Topiques et des Réfutations sophistiques.
[Sens 2] Parfois la matière du syllogisme n’est pas rabattue sur la
matière de la proposition mais elle est considérée comme la proposition
elle-même, comme proposition, indépendamment de sa matière. C’est ce
deuxième sens de matière qui est en jeu quand Robert Kilwardby attribue
au syllogisme simpliciter une matière qui lui est propre et qui ne peut faillir.
Parfois encore la matière du syllogisme est une notion plus vague
et correspond au contenu concret dans la forme syllogistique du fait du
remplacement des « lettres » par des termes concrets. Ce dernier sens de
matière est spécifiquement lié à l’idée que l’usage des lettres dans les Premiers
analytiques permet d’étudier la forme indépendamment de toute matière.

3. Le syllogisme simpliciter comme un composé de forme et de matière

dans le commentaire aux Premiers analytiques

La position de Robert Kilwardby sur la forme et la matière du syllogisme

est très claire dans le prologue de son commentaire aux Premiers analytiques
(ca. 1240). Le logicien anglais s’y interroge sur la nature du syllogisme

Sur cet amalgame du sophistique et du faux, voir la classification des arguments
en fonction des modalités aléthiques dans la tradition tardo-antiques de l’Organon long,
qui sera repris dans la tradition arabe, où les prémisses des argument dialectiques sont
à moitié vraies ou à moitié fausses, mais plus vraies que fausse, les rhétoriques plus
fausses que vraies, les sophistiques fausses mais apparemment vraies et les poétiques
tellement fausses qu’elles n’ont pas même l’air vraies. Voir le tableau proposé par
EBBESEN, Commentators and Commentaries, vol. I, p. 91.

pur et simple (simpliciter) et sur sa capacité à être le sujet d’une science,

et plus précisément celle des Premiers analytiques. Une fois établi que le
syllogisme pur et simple est une réalité suffisamment constituée en soi, dotée
de sa bonitas propre, pour être objet d’une science, la question est de savoir
si le syllogisme pur et simple est l’objet des Premiers analytiques. Plusieurs
objections sont avancées. Une objection nous dit qu’un genre n’a pas d’être
séparément de ses espèces, que le syllogisme pur et simple est le genre
des syllogismes démonstratifs et dialectiques, donc que le syllogisme pur et
simple n’a pas d’existence séparée. Il ne peut ainsi être l’objet des Premiers
analytiques, qui portent sur un syllogisme qui possède un être distinct du
syllogisme dialectique et démonstratif, c’est-à-dire un syllogisme dans une
matière transcendante, comme dans ces termes « A » et « B » et non dans
une matière probable ou une matière nécessaire12. Une autre objection nous
dit que puisque les autres livres portent sur le syllogisme contracté dans une
matière donnée, il apparaît que les Premiers analytiques ne portent pas sur
le syllogisme simpliciter mais sur la forme du syllogisme.
La réponse de Kilwardby se concentre d’abord sur ce dernier point :

Contre ceux qui disent qu’Aristote traite seulement de la forme du

syllogisme13 : il définit en effet au début [des Premiers analytiques] la
proposition et le terme en vue du syllogisme et ceux-ci sont ensemble
les [composants] matériels pour le syllogisme. De même il définit
le syllogisme pur et simple, mais celui-ci ne revient pas seulement
à la forme du syllogisme. Bien que le syllogisme pur et simple dont
s’occupent les Premiers analytiques soit donc une forme relativement
au syllogisme dialectique et au syllogisme démonstratif, il n’apparaît
pourtant pas qu’il faille dire que le livre des Premiers analytiques
porte uniquement sur la forme, mais plutôt qu’il considère aussi bien
la forme que la matière du syllogisme en général14.

« Sillogismus in materia transcendenti, ut in hiis terminis A et B, nec in materia
probabili neque in materia necessaria », Expositio Egidii Romani super libros priorum
Analeticorum Aristotelis cum textu euisdem, Venetiis 1499, f. 2rb.
C’est la position de l’anonymous Aurelianensis III, un commentaire incomplet
aux Premiers Analytiques de la fin du XIIe siècle, le premier connu à ce jour. Voir le
prologue dans l’édition partielle de EBBESEN, « Analysing Syllogisms or Anonymus
Aurelianensis III », Repr. in Greek-Latin philosophical interactions, vol. 1, op. cit., pp.
171-187) et l’édition à paraître de Ch. THÖRNVISQ-THOMSEN.
« Adhuc contra eos qui dicunt Aristotelem solum determinare de forma
syllogismi. Deffinit enim in principio propositionem et terminum propter syllogismum

La réponse à l’objection précédente est la suivante :

Il faut dire que le syllogisme pur et simple n’est pas véritablement

un genre pour [les syllogismes] dialectique et démonstratif, puisque
le genre se divise véritablement par les différences formelles
qui, lui advenant, constituent véritablement les espèces. Mais les
syllogismes dialectique et démonstratif ne sont pas constitués
par des différences véritablement formelles qui adviendraient au
syllogisme pur et simple, mais plutôt par des différences matérielles.
Le syllogisme pur et simple est donc quelque chose de constitué
en acte formellement en amont [des syllogismes] dialectique et
démonstratif. Il fait abstraction de ces derniers selon son être en
ce sens qu’il est possible d’en avoir une connaissance certaine et
d’en donner un exemple dans des termes communs en lesquels ne
sont ni le syllogisme dialectique, ni le syllogisme démonstratif. S’il
était véritablement un genre il ne serait pas véritablement quelque
chose de constitué dans l’être en amont de ceux-ci et il ne serait pas
possible d’avoir une connaissance certaine de quoi que ce soit à son
propos en dehors de son instanciation dans un syllogisme spécial,
dialectique ou démonstratif15.

Le De Ortu scientiarum (ca. 1250) précise que la matière du syllogisme

simpliciter est comme la matière intelligible des objets mathématiques,

et omnia hec sunt materialia syllogismo. Item definit syllogismum simpliciter sed
syllogismus simpliciter non dicit solum formam […] quamvis ergo syllogismus
simpliciter de quo agitur in Prioribus sit forma ad syllogismum dialectium et
demonstrativum, tamen non videtur esse dicendum quod liber Prioribus sit tantummodo
de forma sed magis considerat tam formam quam materiam syllogismi in genere »,
Expositio Egidii Romani super libros priorum Analeticorum, f. 2rb.
« Dicendum quod syllogismus simpliciter non est vere genus ad dialecticum
et demonstrativum quia vere genus dividitur per formales differentias per quas ei
advenientes constituuntur vere species. Sed dialecticus syllogismus et demonstrativus
non sunt constituti per differentias vere formales syllogismo simpliciter adveninentes
sed magis per differentias materiales magis. Et ideo syllogimus simpliciter est actu et
formaliter constitutum in esse ante syllogismum dialecticum et demonstrativum. Et ideo
abstrahit eis secundum esse in tantum quod potest aliquid certificari et exemplificari
de ipso in terminis communibus in quibus neque fit dialecticus syllogismus neque
demonstrativus. Si autem (éd.: ante) esset vere genus tunc non esset aliquid vere
(éd.: consideratum) constitutum in esse ante ipsos neque posset tunc certificari de
syllogismo simpliciter aliquid nisi manifestato syllogismo speciali ut in dialectico vel
demonstrativo », Expositio Egidii Romani super libros priorum Analeticorum, f. 2rb.

différente de la matière des différents cercles matériels qui peuvent être

réalisés sur son modèle16.
Toute la discussion de Kilwardby présuppose un « principe
hylémorphique » clairement formulé dans le commentaire aux Topiques
qui lui est attribué.

4. Le principe « hylémorphique » dans le commentaire aux Topiques

À propos des syllogismes seulement matériellement éristiques, qui

sont des syllogismes, Robert Kilwardby envisage l’objection consistant à
dire que si la matière est fautive la forme ne peut subsister, pas plus que
le composé : les syllogismes matériellement éristiques ne pourraient donc
pas être des syllogismes. C’est exactement la thèse qui sera condamnée
à Oxford en 1277 sous l’égide de Robert Kilwardby: « le syllogisme qui
pèche par la matière n’est pas un syllogisme »17.

« Quia istae differentiae ratiocinationum ex propriis [i. e. pour le syllogisme
démonstratif] et ex communibus [i. e. pour le syllogisme dialectique] summuntur ex
parte materiae […] et forma ratiocinationis eadem est in omni materia, ideo oportuit
logicam tradere tractatum de modo ratiocinandi in genere prout abstrahit ab omni
materia propria vel communi, ad quam videlicet respiceret tam demonstrator quam
dialecticus ratiocinari volens. Eadem enim est forma syllogistica in utraque materia,
scilicet in necessaria et probabili. Nota tamen quod non ita abstrahit haec forma quod
determinetur de ipsa sine omnimoda materia, quia hoc esse non posset. Sed sicut
mathematica abstrahuntur a materia physica tantum et nihilominus ipsa habent suam
materiam intelligibilem […] sic forma syllogistica et omnino ratiocinativa abstrahitur
a forma communi et propria, id est probabili et necessaria. Habet tamen secum
quandam materiam simpliciorm quae est intra utramque dictam, scilicet tres terminos,
duo extrema et medium unum ex quibus connectuntur duae propositiones. De forma
autem syllogistica et omnino ratiocinativa in libro Priorum Aristotelos determinetur,
ubi ponuntur semper tres termines constituentes duas propositiones, sed tales qui
abstrahant a materia probabili et necessaria », De Ortu scientiarum, éd. A. G. JUDY,
Oxford 1976, p. 170.
« Martii 18, Oxoniae […] 2. Item quod sillogismus peccans in materia non est
sillogismus », Chartularium universitatis parisiensis 1, 1200-1286, éd. H. DENIFLE –
A. CHATELAIN, Paris 1889, n. 474, p. 558. Une thèse que nous avons identifiée comme
soutenue par Albert le Grand dans ses paraphrases aux Topiques et aux Réfutations
sophistiques (voir conclusion et J. BRUMBERG-CHAUMONT, « Les divisions de la logique
selon Albert le Grand », in EAD. (éd.), ‘Ad notitiam ignoti’. L’organon dans la Translatio
studiorum à l’époque d’Albert Le Grand, Brepols, Turnhout 2013, pp. 335-416.

La réponse ne consiste pas à remettre en question le principe

hylémorphique évoqué, auquel Robert Kilwardby souscrit pleinement, mais
à dire que le syllogisme simpliciter a une matière essentielle qui ne peut
faillir. Il va plus loin en disant que les matières des syllogismes dialectiques
et démonstratifs leur sont accidentelles en tant que syllogismes :

On s’interroge, en septième lieu, sur le fait qu’[Aristote] indique

que [le syllogisme éristique] est un syllogisme même s’il pèche par
la matière. Cela semble faux puisque tout composé de forme et de
matière est tel que si la matière ou la forme fait défaut le composé fait
défaut, et ceci parce que chaque forme n’appelle pas n’importe quelle
matière mais celle qui lui est propre : c’est ainsi que le syllogisme qui
se fait à partir d’une [prémisse] fausse, ou de deux fausses, n’est pas
un syllogisme parce qu’en ce cas la matière fait défaut.

Il faut répondre que la matière essentielle dans le syllogisme consiste

dans les trois termes et les deux propositions, et que si cette matière
fait défaut, le syllogisme ne subsiste pas.
Il y a, selon un autre mode, une matière accidentelle du syllogisme
c’est-à-dire les trois termes et les deux propositions considérées
selon la disposition que les propositions sont admises, vraies et
nécessaires, et si cette matière fait défaut, il peut bien il avoir alors
un syllogisme, et cela parce que cette matière est accidentelle18.

Ce texte est remarquable car le « principe hylémorphique » y est

clairement exprimé : une forme ne peut être réalisée dans n’importe quelle
matière, mais seulement dans la matière qui lui est appropriée, de sorte

« Septimo queritur super hoc quod innuit sillogismus esse [i. e. le syllogisme
éristique] etsi sit pecatum in materia. Hoc enim videtur falsum quia unumquodque
compositum ex materia et forma sic se habet quod si fiat defectus vel in materia vel in
forma, erit defectus in composito, et hoc quia quelibet forma non appetet quamlibet
materiam, set sibi propriam […] tunc quod sillogismus ex falsis vel ex falso non
sit sillogismus et hoc quia ibi est peccatum in materia. Et dicendum quod materia
essentialis in sillogismo sunt tres termines et due propositiones ; et si defectus in
ipsa materia non manet sillogismus. Alio modo <materia> sillogismi accidentalis
scilicet tres termini et due propositiones sub istis dispositionibus quod propositionibus
sint probabiles, vere et necessarie, et si sit defectus in ista materia bene tunc potest
sillogismus esse, et hoc quia hec materia accidentalis », In Librum Topicorum, éd. O.
WEIJERS, « Le commentaire sur les Topiques attribué à Robert Kilwardby », Documenti
e Studi sulla tradizione filosofica medievale, VI (1995) 107-143, pp. 132-133.

que si cette matière fait défaut, la forme fait défaut, ainsi que le composé.
Plus précisément, la matière du syllogisme au sens 2 est essentielle, tandis
que la matière du syllogisme aux sens 1 est accidentelle. La thèse n’est
pas que la matière au sens 1 est accidentelle à l’égard du syllogisme
simpliciter, ce qui serait banal, mais qu’elle est accidentelle à l’égard d’un
syllogisme donné dont elle est la matière. Lorsque cette matière fait défaut,
le syllogisme reste un syllogisme en vertu de la matière au sens 2, qui lui
confère l’être-syllogisme. La matière du syllogisme simpliciter est ainsi
comme « sous-jacente » au syllogisme concret et elle se substitue à la
matière concrète si celle-ci fait défaut.

5. Le « Barbara sophistique » dans les Premiers analytiques

Revenons à présent à l’analyse des arguments sophistiques qui suivent

une combinaison syllogistique utile dans le commentaire aux Premiers
Analytiques. Robert Kilwardby envisage l’argument qu’ils pourraient être
considérés comme un contre-exemple au premier mode de la première
figure au prétexte que le Barbara sophistique conclurait du dici de nullo à
partir du dici de omni. On sait en effet que dans un syllogisme le faux ne
peut suivre du vrai. Si on peut produire un cas où une paire de propositions
vraies donne une proposition fausse, comme Aristote lui-même le fait dans
sa discussion sur la syllogistique modale, on doit exclure la combinaison
en question. On sait également qu’une combinaison est non conclusive si
aussi bien une proposition que son contraire peut suivre, comme dans le cas
de la combinaison d’une majeure universelle affirmative et d’une mineure
universelle négative dans la première figure (Anal pr I, 4). La réponse de
Robert Kilwardby consiste à dire que le dici de omni suit toujours, du moment
qu’une conception suffisamment abstraite de la relation de prédication est
adoptée, et que le Barbara sophistique n’est pas un contre-exemple puisque
le vrai suit bien du vrai. Le fait que la conclusion ne puisse être vraie que si
le prédicat est considéré comme se prédiquant par accident du sujet19 n’est
pas un problème, étant admis contrefactuellement que toutes les statues sont

C’est par accident – en vertu de sa composition matérielle hic et nunc, et non
de son essence – que la statue est « naturelle », parce qu’il se trouve qu’elle est faite en
cuivre. Elle ne le serait pas si elle était faite en plastique, pour peu que le plastique soit
considéré comme une matière « artificielle » ou si elle était elle-même faite d’artéfacts,
comme dans une accumulation d’Arman. L’important ici est que ce n’est pas en tant
que statue qu’elle est naturelle.

faites en métal, car la prédication impliquée dans le dici du dici de omni

fait abstraction de la nature de la prédication, naturelle ou accidentelle20. Le
Barbara sophistique est bien un syllogisme simpliciter en Barbara :

Un doute peut être soulevé à propos du premier mode [de la

première figure]. Il semble en effet que cette combinaison est inutile
puisqu’on trouve des termes qui donnent l’inhérence « dans tout »
ou « dans aucun » […] comme dans le cas « Tout bronze est naturel,
la statue est en bronze, la statue est naturelle ». Il faut dire que ce
cas ne vaut rien, puisqu’il s’agit ici de la forme syllogistique dans
une matière commune, qui est dégagée des matières probables,
nécessaires et apparentes, de sorte que la forme dont il s’agit ici
peut non seulement se trouver dans les syllogismes dialectiques et
démonstratifs, mais aussi dans les sophistiques. C’est pourquoi il
faut dire que la conclusion est bien obtenue, c’est-à-dire « toute
statue est naturelle », selon l’artisan de ce livre, ou bien, si ce
n’était pas le cas, il lui faudrait refuser la première figure. La
forme [de cet argument] est bonne selon lui et elle n’est pas exclue
de la forme syllogistique telle que traitée ici. Pour le prouver, il
faut savoir qu’il y a un double syllogisme, c’est-à-dire celui dont
la nécessité est locale, dans lequel à partir de la majeure ou de la
mineure la conclusion suit nécessairement, et il s’agit du syllogisme
dialectique ou démonstratif ; l’autre syllogisme est celui dont la
nécessité est celle de la combinaison uniquement, qui est causée

Ici Robert Kilwardby prend manifestement position dans un débat sur la
relation entre dici de omni et transitivité de la prédication, qu’elle soit essentielle ou
non, dont l’Anonymous Cantabrigiensis se faisait déjà l’écho au début du XIIIe siècle
à propos du « Barbara fallacieux » : « De huiusmodi tamen habetur controversia.
Quidam enim non dicunt has maximas esse ‘Quicquid de praedicato praedicatur, et de
subiecto’ ‘Quicquid a praedicato removetur, et a subiecto removetur’, immo falsas has
dicunt, sed per adiectionem, posse fieri veras, sic sc.: ‘Quando aliquid de aliquo ut de
subiecto <praedicatur>, quicquid de praedicato praedicatur ut de subiecto, de subiecto
praedicabitur ut de subiecto’, ‘Quando alterum praedicatur de altero ut de subiecto,
quicquid removetur a praedicato ut ab extraneo removetur a subiecto ut ab extraneo’.
Alii dicunt has veras ‘Quicquid praedicatur de praedicato, praedicatur de subiecto’
‘Quicquid removetur a praedicato, removetur a subiecto’, nec in aliquibus terminis –
sive extranee sive non extranee sumptis – posse inveniri instantiam. Nam cum naturale
praedicetur de aere et aes de statua, naturale etiam praedicatur de statua, sic ‘Omnis
statua est naturale’ vel sic ‘Quicquid est statua est naturale’, non sic autem ‘Omnis
statua est naturalis’, et cum artificiale removeatur ab aere, removebitur etiam a statua,
sic ‘Nulla statua est artificialis’ », éd. S. EBBESEN, à paraître 2016.

par la disposition correcte des termes et des propositions les uns

par rapport aux autres, et il est commun au syllogisme dialectique,
démonstratif et sophistique ; c’est de ce syllogisme et de cette forme
que nous traitons ici. L’artisan des Premiers analytiques traite
abstraitement le syllogisme et, de même, l’être-prédiqué, dans les
propositions syllogistiques de sorte qu’il fait abstraction de l’être
par soi ou de l’être par accident, raison pour laquelle il concède
les prédications accidentelles. Quand on argumente ainsi en disant :
« Tout bronze est naturel, la statue est en bronze, etc. », on dira
que la conclusion est seulement vraie par accident, sans aucunement
invalider la forme de l’argument21.

La solution de Kilwardby consiste ainsi à faire de tout argument

une réalité gigogne : toit syllogisme est double puisque qu'il y a un
syllogisme simpliciter sous-jacent à tout syllogisme concret. Ce dernier
demeure un syllogisme dont la conclusion est obtenue formellement, en
vertu de la combinaison seule, et ainsi en vertu du principe du dici de

« Forte dubitatur de primo modo. Videtur enim quod sit inutilis coniugatio quia
est reperire terminos omni et nulli inesse […] quod autem nulli inesse patet hoc : “ Omne
aes est naturale, statua est aes, statua est naturalis ”. Et dicendum quod instantia nulla
est, hic enim determinetur forma syllogistica in communissima materia, quae abstrahit
a materia probabili et necessaria et apparenti, unde forma quae hic determinatur non
tantum inveniri potest in dialecticis et demonstrativis, sed etiam in sophisticis. Unde
dicendum quod sequitur conclusio, scilicet “ omnis statua est naturalis ” secundum
artificem huius libri, vel si non, neganda est prima secundum ipsum ; forma enim bona
est secundum ipsum, et non excluditur a forma syllogistica hic determinata. Ad huius
evidentiam sciendum quod duplex est syllogismus, scilicet ille cuius necessitas est
localis, ubi ex maiori vel minori necessario sequitur conclusio, et talis est syllogismus
dialecticus vel demonstrativus ; et alius est cuius necessitas est complexione tantum,
hoc causata ex debita complexione terminorum ad invicem et propositionum, et talis
est communis syllogismo dialectico, demonstrativo et sophisticos, et talis syllogismi
necessitas et forma hic determinatur. Artifex igitur libri Priorum abstrahit syllogismum
et similiter esse predicatum in propositionibus syllogisticis unde abstrahit illud
esse ad esse per se et esse per accidens et ita concedit praedicationes accidentales.
Unde cum sic arguatur : “ Omne aes est naturale, omnis statua est aes, ergo etc. ”
diceret quod conclusio non est vera quamvis per accidens, numquam autem formam
arguendi negaret », Robert Kilwardby, Expositio Egidii Romani super libros priorum
Analeticorum Aristotelis cum textu euisdem, éd. N. J. GREEN-PEDERSEN, « Discussions
about the Status of the Loci Dialectici in Works from the Middle or the 13th Century »,
CIMAGL, 20 (1977) 38-78, p. 75. Même objection et même réponse chez Albert le
Grand, Libri Priorum Analyticorum, pp. 490B-491A.

omni, même quand sa matière est sophistique. Par conséquent les Barbara
sophistiques sont parfaitement syllogistiques et non formellement
Cette analyse du Barbara fallacieux n’est pas maintenue pas Robert
Kilwardby dans le De Ortu scientiarum, et elle n’est généralement pas
celle adoptée dans les décennies suivantes.

6. L’analyse des arguments matériellement et formellement éristiques

dans le De Ortu scientiarum.

Le De Ortu scientiarum offre d’abord une analyse de la typologie des

arguments éristiques en fonction de la forme et de la matière qui ne permet
pas vraiment de rendre compte du cas du «Barbara sophistique», pourtant
mentionné. Comme dans la tradition antérieure des commentateurs tardo-
antiques, Robert Kilwardby prend comme exemple d’un raisonnement
éristique qui pèche par la forme un raisonnement qui suit une combinaison
inutile – il est en effet dans la première figure avec une prémisse en A et
l’autre en E, avec une conclusion en E, qui part de prémisses vraies pour
aboutir à une conclusion fausse. Il prend comme exemple de syllogisme qui
pèche par la matière un argument qui suit une combinaison syllogistique
dont la majeure est apparemment admise. Comme exemple d’argument
à la fois matériellement et formellement éristique, il choisit un argument
qui suit la même combinaison inutile que la première fois mais dont une
prémisse est fausse22.

« Et iste [i. e. le syllogisme éristique] revera multis modis est. Aut enim
non facit quod deberet. Ille autem facit quod non deberet introducit propositionem
vel conclusionem falsam ad faciendum habitum falsum, quod non deberet facere
ratiocinatio. Et iste est tribus modis, secundum Aristotelem in I Topicorum, quia aut
peccat in forma tantum, scilicet quando consequentia non tenet, ut : omnis homo est
animal, asinus non est homo, ergo non est animal ; aut in materia tantum, scilicet
quando consequantia bona est sed aliqua praemissarum est falsa et improbabilis, quae
tamen apparet probabilis, ut : omnis statua est naturalis, figura Herculis est statua,
ergo est naturalis – prima quae est falsa et improbabilis potest apparere per fallaciam
accidentis sic : omnis statua est aes, et omne aes est naturale, ergo omnis statua est
naturalis – aut peccat in materia et in forma, ut : omnis homo est animal, risibile non
est homo, ergo non est animal. Minor est falsa et improbabilis. Potest tamen videri
alicui probabilis per fallaciam accidentis sic : nullum proprium est homo, risibile est
proprium, ergo non est homo […] », De Ortu scientiarum, p. 174.

Robert Kilwardby nous dit ensuite que les treize fallacies, qu’elles
soient in dictione ou extra dictionem, pèchent par la forme23, ce qui
n’est guère cohérent avec les propos précédents, puis conclut de manière
exagérément optimiste en affirmant qu’il s’est suffisamment expliqué sur
les arguments qui pèchent par la forme ou la matière.
Il ne nous dit donc pas comment analyser le « Barbara sophistique »
du point de vue du couple matière forme. Or c’est un point crucial pour un
auteur qui considère que la forme du syllogisme simpliciter réside dans la
combinaison utile, tandis que sa matière est constituée par les propositions
formées de lettres.
La réponse arrive beaucoup plus loin. Il s’agit alors d’expliquer
pourquoi, même si le syllogisme dialectique et le syllogisme sophistique
relèvent de la même science et du même traité, la dialectique et les Topiques,
il a fallu traiter spécifiquement du syllogisme éristique. Une raison est
que le syllogisme éristique se produit de manières plus nombreuses que
le raisonnement qui suit une combinaison inutile, qui pèche par la forme,
ou que le pseudépigraphème, le paralogisme scientifique qui pèche par
la matière. C’est que le syllogisme éristique pèche à la fois par la forme
et par la matière et qu’il pèche par la forme de deux façons, soit qu’il
est en conformité (modificatio) seulement vocale et non réelle avec le
syllogisme, comme c’est le cas de la fallacie de l’accident, soit qu’il suit
une combinaison inutile24. L’exemple n’est plus le Barbara fallacieux à
propos de la statue, mais un « Darii fallacieux » à propos des bains :

Le paralogisme scientifique ne pèche que par la matière et la

combinaison inutile que par la forme. Mais l’argument éristique est
dit tel parfois parce qu’il ne fait pas ce qu’il est censé faire, parfois
parce qu’il fait ce qu’il n’est pas censé faire, et il pèche soit par la
forme, soit par la matière, soit par les deux. Parmi ceux qui pèchent par

« De his modis in Elenchi agitur, quia de eo qui peccat in forma in illa parte ubi
agitur de locis sophisticis in dictione et extra dictionem, ibi enim agitur de apparenti
syllogismo, et hic est qui peccat in forma […] », De Ortu scientiarum, p. 174.
La notion de modificatio désigne ici la disposition syllogistique en général ;
c’est une notion technique propre à la tradition latine des Réfutations sophistiques, où
asyllogizatoi est traduit une fois par immodificati (voir SE 168a21, Aristoteles Latinus
6:1-3, 15). Il s’agit d’un passage, au début du chapitre 6 à propos de la réduction de tous
les paralogismes à l’ignorance de la réfuation, où Aristote enjoint de d’abord vérifier
si les arguments ne sont pas assyllogizatoi, c’est-à-dire, non concluant nécessairement,
puis de vérifier s’ils respectent la définition du syllogisme.

la forme, certains sont dépourvus de conformité syllogistique réelle,

mais non vocale, d’autres de conformité à la fois réelle et vocale. Dans
l’exemple : « Toute eau est naturelle, quelque bain est d’eau, donc
quelque bain est naturel », il n’y a pas de défaut dans le mode vocal,
mais seulement dans le mode réel. Mais si l’on dit : « Toute eau est
naturelle, aucune sécheresse n’est de l’eau, donc aucune sécheresse
n’est naturelle », l’argument pèche également quant à la conformité
vocale puisque la mineure est négative dans la première figure25.

Le Darii fallacieux appartient aux arguments formellement éristiques,

comme c’est le cas, nous l’avons vu, de toutes les fallacies, qu’elles
soient in dictione ou extra dictionem. Mais il possède tout de même une
conformité (modificatio) syllogistique sur un plan vocal. Il y aurait donc
une clause supplémentaire sur le mode réel, et pas seulement vocal, pour
la démarcation des syllogismes, clause qui est absolument nécessaire pour
pouvoir rendre compte des fallacies qui suivent une combinaison utile
tout en péchant par la forme. Mais comment est-elle compatible avec ce
qui a été dit précédemment de la forme et de la matière du syllogisme
simpliciter, où la signification des termes, pourtant nécessaire pour pouvoir
juger de l’existence d’une disposition réelle et pas seulement vocale, ne
doit pas être prise en compte ? Une doctrine comparable, quoiqu’exprimée
de manière plus tranchée, apparaît dans le commentaire aux Réfutations
sophistiques attribué à Robert Kilwardby.

7. Le commentaire aux Réfutations sophistiques attribué à Robert


Ce commentaire utilise la même solution que celle mise en œuvre dans

le De Ortu scientiarum, c’est-à-dire la distinction entre la conformité réelle

« Paralogismum enim disciplinae tantum in materia peccat ; inutilis coniugatio
tantum in forma ; sed litigiosus aliquando dicitur quia non facit quod deberet, aliquando
quando facit quod non deberet, et iste peccat tum in materia tum in forma tum utraque.
Et qui peccat in forma aliquando caret modo syllogistico reali tantum, aliquando
vocali et reali, verbi gratia : omnis aqua est naturalis, sed aliquod balneum est aqua,
ergo est naturale. Hic nullus est defectus in modificatione vocali sed reali tantum. Sed
si sic dicatur : omnis aqua est naturalis, nulla siccitas est aqua, ergo nulla siccitas est
naturalis, peccatum est etiam in modo vocali, quia minor est negativa in prima figura »,
De Ortu scientiarum, pp. 189-190.

et vocale. L’auteur26 considère en effet que toutes les fallacies pèchent contre
la forme du syllogisme. Il prend comme exemple la fallacie de l’accident,
laquelle suit une combinaison utile selon les Premiers analytiques. Il faut
donc trouver une solution. Celle-ci apparaît au cours de la réponse à un
certains nombres de dubitationes dans le prologue.
L’auteur s’interroge sur la possibilité d’une science du syllogisme
sophistique et s’il s’agit de la logique. Un argument contra consiste à dire
que la logique s’occupe du syllogisme et de ses parties et que le syllogisme
sophistique n’est ni une partie de syllogisme ni un syllogisme. L’auteur
y répond et défend un sens dans lequel le syllogisme sophistique est un
syllogisme, de sorte qu’il entre dans l’objet de la logique, et un sens en
lequel il pèche par la forme et n’est pas un syllogisme. Le premier sens se
rapproche de la description du syllogisme simpliciter du commentaire aux
Premiers analytiques, puisque l’argument formellement éristique suivant
une combinaison syllogistique possède les propositions et les termes en
tant que propositions et termes à titre de matière, et le premier mode de la
première figure à titre de forme. En revanche, il n’est pas un syllogisme si
le mode réel est pris en compte :

Il faut dire que le syllogisme sophistique est un syllogisme, ce qui

peut se comprendre de la façon suivante : parmi les syllogismes,
certains pèchent par la matière, d’autres par la forme, et d’autres

Pour l’attribution du commentaire contenu dans le ms. Paris, BnF, lat. 16619 à
Robert Kilwardby voir O. LEWRY (« Robertus Anglicus and the Italian Kilwardby », in
A. MAIERÙ (éd.), English Logic in Italy in the 14th and the 15th Centuries, Bibliopolis,
Napoli 1982, pp. 33-52, p. 38), qui utilise des arguments philosophiques, stylistiques
et doctrinaux. Nous ne pouvons discuter ici dans les détails la question de l’attribution,
mais nous sommes réservés sur les arguments doctrinaux dans la mesure où le
prologue contient des éléments de doctrines dissonant par rapport à ceux contenus
dans le commentaire aux Premiers analytiques et au De Ortu scientiarum. Il s’agit
en particulier de l’idée que les syllogismes dialectiques scientifiques et sophistiques
ne sont pas des variations matérielles du syllogisme simpliciter et de la thèse selon
laquelle ils en sont des espèces (différant formellement) destinées à être prise en
charge dans des traités distincts (ce qui va exactement à l’encontre de ce que dit le
De Ortu scientiarum). Il y a d’autres dissonances. Une hypothèse toujours ouverte est
qu’il pourrait s’agir d’un commentaire « patchwork », comme c’est le cas d’un autre
manuscrit contenant des commentaires de Robert Kilwarby, le ms. Oxford, Bodleian
Canonici misc. 403, et que le prologue viendrait d’un autre texte. L’auteur en serait un
auteur fortement influencé par Robert Kilwardby, avec des positions divergentes sur
certains points.

encore par les deux. Le syllogisme qui pèche par la matière est un
syllogisme, comme l’affirme clairement Aristote dans plusieurs
passages. Mais le syllogisme qui pèche par la forme comme «Toute
eau est naturelle, le bain est d’eau, donc le bain est naturel», est bien
un syllogisme car on retrouve d’une certaine manière en lui toute la
nature du syllogisme. Il possède en effet la matière requise pour un
syllogisme, trois termes et deux propositions et il possède aussi la
forme : il est en effet dans la première figure. En lui les trois termes
sont en effet disposés de telle sorte que le premier inhère dans tout le
moyen terme et le moyen terme dans tout le troisième, et quand il en
est ainsi, on est dans la première figure comme le dit Aristote dans
les Premiers analytiques. On est en outre dans le premier mode :
les prémisses sont en effet affirmatives universelles et la conclusion
universelle affirmative. Il n’en demeure pas moins que la conformité
syllogistique est double, vocale ou réelle. La conformité vocale
s’observe ici, de sorte qu’il y a bien une combinaison syllogistique
en ce qui concerne le son vocal, je veux dire dans le syllogisme qui
pèche par la forme. Il appert donc que le syllogisme sophistique est
une espèce du syllogisme, et qu’ainsi l’objection précédente n’a plus
d’objet. Quand Aristote dit ensuite que le syllogisme sophistique
n’est pas un syllogisme, il veut dire que ce n’est pas un syllogisme
dont la combinaison existe sur un mode réel27.

La même solution apparaît plus loin, à propos de la distinction des

genres d’argumentations dans les Réfutations sophistiques, laquelle

« Dicendum quod syllogismus sophisticum sillogismus est et hoc habet sic intelligi :
Sillogismis sophisticis quidam peccat in materia, quidam in forma, quidam in utrumque.
Sillogismus peccans in materia est sillogismus, sicut plane vult Aristoteles in pluribus locis.
Sillogismus autem peccans in forma ut iste “omis aqua est naturalis, balneum est aqua,
ergo balneum est naturale” sillogismus est, quia in eo tota natura sillogismi quodammodo
reperitur. Habet enim materiam sillogismo debitam ut tres terminos et duas propositiones,
similiter autem et formam. Habet enim primam figuram. Hic enim disponuntur tres terminos
ita quod primum inest omni medio et medium omni <tertio>, et quando ita est, tunc est
prima figura, ut dicit Aristoteles in libro Priorum. Est etiam ibi modus prime figure. Sunt
enim premisse universales affirmative et conclusio universalis affirmativa.Verumptamen
modus in sillogismo duplex est, scilicet vocalis et realis. Vocalis autem hic reperitur
unde tota complexio sillogismi quantum est ex parte vocis – dico <tum> in sillogismo
peccante in forma. Patet ergo quod sillogismus sophisticus species est sillogismi et ita perit
oppositio prius facta. Cum enim dicit Aristoteles inferius quod sillogismus sophisticus non
est sillogismus ipse intendit quod non est sillogismus cuius complexio est modus realis»,
Liber Elenchorum, ms. Paris, BnF, lat. 16619, f. 1vb19-35.

inclut les syllogismes éristiques. L’auteur précise qu’il semble y avoir

une contradiction puisque, d’un côté, le syllogisme éristique n’est pas
un syllogisme, étant un syllogisme apparent, mais, d’un autre côté, il
appartient aux argumentations, et doit être un syllogisme, puisqu’il n’est
ni une induction, ni un enthymème, ni un exemple. La réponse consiste
à dire que le syllogisme est double et que le syllogisme éristique est un
syllogisme du point de vue de la réalisation vocale, mais pas du point de
vue de la réalisation réelle28.
L’auteur du commentaire maintient les fallacies qui suivent une
combinaison utile au sein des syllogismes en s’appuyant sur une conception
syntactique de la forme syllogistique dans laquelle les parties matérielles sont
les termes concrets eux-mêmes pris en tant que termes, indépendamment
de leur signification, comme de simples séquence sonores qui n’exigent
que le respect de la substitution uniforme vocale. Le double syllogisme
n’est plus le syllogisme simpliciter et probans, mais le syllogisme réalisé
vocalement ou réellement. Il donne ainsi un sens très affaibli à l’affirmation
d’Aristote selon laquelle les arguments formellement sophistiques sont
des syllogismes apparents : ce n’est pas qu’ils sont apparemment des
syllogismes sans en être, mais ils sont réellement des syllogismes dans
la mesure seulement où ils en ont l’apparence sonore ou visuelle. Il ne
nous dit pas en outre clairement ce qu’est une conformité réelle : est-
il exigé, par exemple, dans un syllogisme en Barbara, que le prédicat
se dise de tout le sujet avec vérité, ou que la prédication soit seulement
uniforme des prémisses à la conclusion, quelle que soit la valeur de vérité
des propositions ? Enfin la question demeure de savoir si les arguments
formellement éristiques possédant une combinaison syllogistique sur un
plan purement vocal sont des syllogismes simpliciter ou non.
L’auteur ne se prononce pas directement sur cette dernière question.
Il semble entretenir une conception du syllogisme simpliciter assez
profondément différente de celle qu’on a pu lire dans le commentaire aux
Premiers analytiques de Robert Kilwardby. Il dit en effet explicitement que
les différences qui s’ajoutent au syllogisme simpliciter pour constituer les

« Sillogismus enim duplex est secundum duplicem eius perfectionem. Habet
enim perfectionem in modo duplicem scilicet vocalem et realem. Unde sillogismus
apparens et non existens non est sillogismus quoad ad perfectionem realem et est
existens sillogismus quantum ad perfectionem vocalem », Liber Elenchorum, ms.
Paris, BnF, lat. 16619, f. 3va12-16.

syllogismes dialectiques, scientifiques et sophistiques sont des différences

diviseuses formelles à l’égard de celui-ci, et des différences constitutives
matérielles des trois syllogismes, qui les différencient entre eux. Elles
ne sont donc pas des différences matérielles du syllogisme simpliciter,
thèse qui est considérées comme la «marque de fabrique» d’une théorie
proprement kilwarbienne29.
Cet usage assez peu orthodoxe des notions de différences constitutives
et diviseuses pourrait peut-être s’expliquer de la façon suivante : en se
voyant adjoindre les déterminations ‘sophistiques’, ‘dialectique’ et
‘scientifique’ le syllogisme simpliciter devient autre substantiellement,
i.e. il cesse d’être, précisément, simpliciter. En revanche les différences
‘sophistiques’, ‘dialectique’ et ‘scientifique’ ne différencient les trois
syllogismes que matériellement puisqu’ils ont une forme commune. Ceci
ne peut avoir de sens que si le syllogisme simpliciter n’est considéré que
comme la forme commune que les syllogismes concrets partagent, à l’instar
du genre qui ne subsiste pas en dehors de ses espèces, et non comme une
réalité logique auto-subsistante à part des syllogismes concrets eux-mêmes,
tel qu’observé dans le commentaire de Robert Kilwardby aux Premiers
analytiques. Il n’est d’ailleurs plus question dans notre commentaire des
« lettres » comme matière transcendante du syllogisme simpliciter.

« Sed tunc queritur cuiusmodi differentie sunt differentie predicte, scilicet
necessarium, probabile et sophisticum. Et videtur quod sint differentie materiales
silligismi. Per hec enim dividitur sillogismus penes materiam propositionum que sunt
eius principia materialia. Quod autem sint formales, videtur sic. Illa que diversificant
speciem in tractatibus separatis habent determinari ; que autem non diversificant
speciem sed solum materiam debent determinari in eodem tractatu. Unde si ista essent
materialia respectu sillogismi, in eodem tractatu determinasset Aristoteles de hiis. Nunc
autem de hiis in diversis tractatibus determinat, et ideo videtur quod sint differentie
formales. Et dicendum quod possumus istas differentias considerare dupliciter, aut in
se, et sic sunt quedam forme, vel in comparatione ad sillogismum, et sic dupliciter : aut
enim in comparatione ad sillogismum simpliciter, et sic sunt differentie formales. Cum
enim adveniunt ad sillogismum simpliciter, ipsum informant. Vel possunt comparari
ad silllogismos per eas constitutos, et sic sunt materiales. Faciunt enim diversitatem
materialem diversitatem <esse> istorum trium sillogismorum ad invicem : probabilis
scilicet, necessarii et apparentis », Liber Elenchorum, ms. Paris, BnF, lat. 16619, f.


Les difficultés posées par l’héritage kilwardbien sont diversement

reçues. Une option, explorée dans la génération suivante, a consisté à
réviser la description initiale de la matière et de la forme du syllogisme
simpliciter en y ajoutant des clauses supplémentaires portant sur la manière
dont les termes concrets doivent remplacer les lettres des syllogismes
simpliciter. C’est ce qu’on observe dans le premier commentaire aux
Réfutations sophistiques édité par Sten Ebbesen en 1977, daté des années
127030. Les clauses sont, pour les termes, l’unité de significations, et pour
les propositions dans leur ensemble, l’uniformité de prédication, de sorte
que les arguments formellement déficients qui suivent vocalement une
combinaison utile ne sont pas des syllogismes simpliciter. La discussion
menée dans ce commentaire met particulièrement bien en relief la difficile
question du statut théorique des «lettres» comme matière, question qui
n’avait pas été abordée par Robert Kilwardby. N’étant pas des variables
mises pour des items, mais des «lettres d’attente» (dummy letters), c’est-
à-dire des termes, dotés d’un contenu sémantique, quoique général, elles
sont à l’interface de la forme et de la matière. L’anonyme précise en effet
qu’elles appartiennent à la forme du syllogisme simpliciter, pour laquelle
le seul respect du mode et de la figure ne saurait suffire, tout en relevant
de la dimension matérielle de celle-ci. On peut considérer les clauses sur
l’unité et la conformité venues enrichir la conception kilwardbienne du
syllogisme simpliciter comme une tentative, précisément, pour délimiter
le contenu sémantique des termes transcendants. Il reste bien général,
puisqu’indifférent à telle ou telle signification individuelle. Mais il va au-
delà d’une simple stipulation préalable sur la nature des termes concrets
substituables – par exemple : « pas de termes équivoques » – qui s’en
tiendrait au plan d’une sémantique formelle, sans pouvoir prendre en
charge l’éventail des ambiguïtés et des variations sémantiques intra-
argumentatives préjudiciables à la validité de l’inférence syllogistique,
telle que, précisément, la fallacie de l’accident31.

Incerti Auctores, Quaestiones super Sophisticos Elenchos, éd. S. EBBESEN,
DSL – Gad, Copenhagen 1977, pp. 25-27 (Corpus Philosophorum Danicorum Medii
Aevi VII).
Ockham emploie cette méthode qui consiste à exclure certains types de termes
dans la définition du syllogisme, méthode qui ne peut fonctionner pour des cas très
complexes comme celui du Barbara sophistique dans le paralogisme de l’accident.

Il s’est sans doute trouvé à l’époque des commentaires logiques

de Robert Kilwardby des maîtres qui, fidèles à la fois à la conception
kilwardbienne du syllogisme simpliciter et à la distinction entre conformité
et réalisation vocale et réelle, ont dû définir le syllogisme simpliciter
uniquement sur un plan vocal, pour pouvoir soutenir que les fallacies
formellement déficientes qui suivaient une combinaison utiles étaient tout
de même des syllogismes en tant que syllogisme simpliciter. C’est en effet
une position qui est attestée dans les commentaires aux Topiques du groupe
« parisien » (auquel aurait appartenu Albert le Grand) de l’époque, édités
partiellement par Green-Pedersen, position qui y est mentionnée pour
y être critiquée. C’est aussi le point de départ de la discussion d’Albert
le Grand dans son commentaire aux Réfutations sophistiques, dont la
position renverse la proposition kilwardbienne. Elle fait en effet clairement
l’équivalence entre matière vocale et matière apparente. Ce n’est pas que
les arguments sophistiques ont réellement l’apparence de syllogismes :
ce n’est que’apparemment qu’ils sont des syllogismes, en vertu d’une
disposition syllogistique apparente, vocale. Il en résulte que toutes les
fallacies formellement déficientes, même quand elles ont l’air d’être

Mais ce n’est pas un problème pour Ockham puisqu’il considère, contre certains
logiciens (en fait tous ceux des générations précédentes) que ce types d’arguments
sont des syllogismes du premier mode de la première figure parfaitement acceptables :
« quattuor modi primae figurae tenent in omnibus terminis, nec refert utrum sumatur
sub terminus substantialis vel accidentalis, quale quid vel hoc aliquid, sive quantum,
sive aliquid […] dummodo per primam propositionem denotetur praedicatum vere dici
vel removeri ab omni illo de quo dicitur subiectum et per secundam propositionem
denotetur illud quod prius fuit subiectum non aequivoce acceptum dici de assumpto,
et postea in conclusione illud qui fuit praedicatum in prima dici vel removeri de illo
praecise quod fuit subiectum in secunda propositione. Ex isto sequitur quod tales
syllogismi sunt boni ‘omne coloratum est, omne album est coloratum, ergo omne
album est’; ‘omne animal est homo, omnis asinus est animal, omnis asinus est homo”;
Et ideo errat Magister Abstractionum assignando praedictis syllogismis fallaciam
accidentis cum isti syllogismi regulentur per dici de omni et sint evidentes. Similiter
tales syllogismi sunt boni nisi aequivocatio impediat ‘omnis homo continetur in genere
substantiae, album est homo, igitur album continetur in genere substantiae’ […] et
ceteri huiusmodi, in quibus volunt aliqui moderni assignare fallaciam accidentis,
nescientes naturam syllogismi nec fallaciam accidentis. Advertendum est tamen quod
numquam syllogismus est regulatus per dici de omni vel per dici de nullo quando
aliquis terminus aequivocus sumitur », Somme de Logique III/1 (traité du syllogisme),
trad. française J. BIARD – C. GRELLARD – K.S. ONG-VAN-CUNG, Franciscan Institute
1974, pp. 12-14.

disposées syllogistiquement d’un point de vue vocal, sont non-conformes

(immodificati), non disposées en figures syllogistiques (infigurati) et
sont des syllogismes apparents32. Quant au mode réel, il n’inclut pas
seulement la stabilité sémantique des termes et des prédications, mais la
disposition réelle des choses telles que prédiquées, autrement dit la vérité
des prémisses33.

« Sciendum quod quia immodificati sunt, per consequens etiam sunt infigurati:
quamvis enim figuram vocalem habeant, non tamen habent figuram realem : quia vel
medium est multiplex et non debite se habet ad extrema, sicut in fallaciis in dictione :
vel secundum rem indebite accipitur, ut in fallaciis extra dictionem : et sic non habet
figuram realem », Liber Elenchorum, p. 603B ; « De his qui videntur elenchi, quamvis
in veritate non sint elenchi, sed sunt paralogismi a para quod est juxta, et logos quod
est ratio sive ratiocinatio, secundum quod ratio cadit in diffinitione argumenti, quando
dicitur quod argumentum est ratio rei dubie faciens fidem : sic enim ratiocinatione
paralogismi videntur elenchi, sed non sunt, qui peccant in forma syllogismi, et nulla
res illius habet nomen, cujus non habet formam, cum omnis denominatio sit a forma :
sed propter similitudinem quam habet ad elenchum, dicitur paralogismus, hoc est,
conjugatio propositionum juxta syllogismum per similitudinem facta, cum tamen
consequentiam syllogismi non habet ex vera medii et extremorum positione in modo et
figura naturali et reali », Liber Elenchorum, p. 527A-B.
« Quod autem diximus sophisticum elenchum peccare in materia, intelligendum
est quod dictum est de materia secundum vocem scilicet et rei naturam et conjugationem
naturalem rerum significatarum in oratione : et materiam quidem secundum vocem
habet paralogismus peccans in materia : et sic syllogismus est verus proximam et
essentialem materiam habens syllogismi : realiter autem secundum quod ea quae sunt
in voce, sunt ad rem significandam, non habet materiam : et ideo non est syllogismus,
cum materiam non habens, per consequens et forma reali sit necesse carere : vocaliter
igitur syllogismus est, et non realiter : significatio enim sermonis adjacet falso, sicut
figura hominis adjacet mortuo : nam <non> statim cadunt vel corrumpuntur a figura
quae cadunt ab esse: quamvis figura comitetur et consequens sit esse, sicut dicit
Aristoteles in quarto Meteororum », Liber Elenchorum, p. 528. Sur ce texte et sur la
justification de l’ajout du <non> voir BRUMBERG-CHAUMONT, ‘Ad notitiam ignoti’.


1. Introduction

According to what can be described as the classical, ‘textbook’ account

of logic, the quintessential objects of logical investigation are the forms of
arguments, which are typically understood as corresponding to argument
schemata. Here is a passage which nicely illustrates the general idea:

We say that (1), (7) and (8) [examples of arguments previously

given] have a particular form in common, and that it is this form
which is responsible for their validity. This common form may be
represented schematically like this:

A or B
Not A

These schematic representations are called argument schemata.

The letters A and B stand for arbitrary sentences. Filling in actual
sentences for them, we obtain an actual argument. Any such
substitution into schema (11) results in a valid argument, which is
why (11) is said to be a valid argument schema1.

A little further:

Logic, as the science of reasoning, investigates the validity of

arguments by investigating the validity of schemata. For argument
schemata are abstractions which remove all those elements of
concrete arguments which have no bearing on their validity2.

Faculty of Philosophy, Oude Boteringestraat 52, 9712 GL Groningen, The
Netherlands. Email: c.dutilh.novaes@rug.nl
L. T. F. GAMUT, Logic, Language, and Meaning, vol. 1, University of Chicago
Press, Chicago 1991, p. 3.
GAMUT, Logic, p. 4.

(In this passage, schematic letters take the place of sentences, but they
can of course also take the place of terms, such as in «Every A is B, and
every B is C. So Every A is C.») Elsewhere3, I have referred to this view
as the ideology of ‘logical hylomorphism as we know it’ (LHAWKI), and
argued that it is a dogmatic and essentially misguided conception of the
nature of logic. One of my strategies to criticize LHAWKI was to engage
in a project of ‘conceptual genealogy’4 so as to unveil the substantive and
often contentious assumptions made along the way in the history of logic,
leading to the establishment of LHAWKI as the received view about logic.
Clearly, against this background, investigating the historical paths leading
to the association of the form of an argument to a schema is of the utmost
relevance, and this will be the topic of this paper.
In the first half of the paper, I review a specific chapter in these
developments: the transformation from figure to mood as that which is
viewed as corresponding to the form of a syllogism. I start with the absence
of hylomorphic concepts in Aristotle’s logical texts, and then discuss the first
applications of hylomorphism to arguments with the ancient commentators,
the equivocal understanding of the form of a syllogism (between figure and
mood) in Latin medieval authors of the 12th and 13th centuries, and the
establishment of a schematic understanding of the form of arguments in
general, as exemplified by Buridan’s treatise on consequence, in the 14th
century5. However, in the second half of the paper, I also show that, in the
same treatise, Buridan himself makes extensive use of syllogistic figures
for logical theorizing, in particular in the fourth book dedicated to modal
syllogistic. Hence, contrary to what the textbook passage above seems to
suggest, investigating schemata is certainly not the only way to study the
validity of arguments in a systematic way: in the particular case of modal
syllogistic, figures offer a very convenient vantage point, for reasons to be
explained in due course.
C. DUTILH NOVAES, «Reassessing logical hylomorphism and the demarcation of
logical constants», Synthese, 185 (2012) 387-410.
On this concept, see C. DUTILH NOVAES, «Conceptual genealogy for analytic
philosophy», in J. BELL – A. CUTROFELLO – P. M. LIVINGSTON (eds.), Beyond the Analytic-
Continental Divide - Pluralist Philosophy in the Twenty-First Century, Routledge,
London 2015 , pp. 75-108.
These developments are discussed in more detail in DUTILH NOVAES, «Reassessing
logical hylomorphism», and C. DUTILH NOVAES, «Form and matter in later Latin
medieval logic: the cases of suppositio and consequentia», Journal of the History of
Philosophy, 50 (2012) 339-364.

The upshot is thus that the LHAWKI ideology is mistaken in

circumscribing logic to the study of schemata. In turn, this suggests that the
widely endorsed, quasi-metaphysical view that the form of an argument
simply is the underlying schema is not nearly as straightforward as it
might be thought; in certain cases, other aspects of arguments seem to be
equally qualified to play the role of ‘forms’ and be the object of logical
investigation – syllogistic figures in particular6.

2. From figure to mood

As is well known, the father of logic and the father of hylomorphism

is not the father of logical hylomorphism. Nowhere in his logical writings
does Aristotle apply the form-matter distinction to logical objects such as
arguments. True enough, the theory of syllogistic presented in the Prior
Analytics presupposes something like a schematic understanding of
arguments, in particular for Aristotle’s method of proving the invalidity
of a syllogistic mood by providing a counterexample: a ‘model’ in which
the premises are true and so is the contradictory of the putative conclusion.
However, the only reference to something like the form or matter of an
argument, and an oblique one at that, is to be found in the Physics and in
an almost identical passage in the Metaphysics, where he seems to suggest
that the premises are the material cause of the conclusion.
Some centuries later, however, the ancient commentators such as
Alexander of Aphrodisias, Ammonius, Simplicius and others, made
extensive use of the hylomorphic framework in their commentaries on the
Prior Analytics7. The earliest such applications that we are aware of are
those in Alexander’s commentary on the Prior Analytics:

Following Corcoran («Schema», in E. ZALTA (ed.), Stanford Encyclopedia
of Philosophy, http://plato.stanford.edu/entries/schema/, 2012), on the present
understanding of ‘schema’, a syllogistic mood counts as a schema, but not a syllogistic
figure. In personal communication, John MacFarlane points out that, on a broader
understanding of schemata, syllogistic figures may also count as schemata.
Barnes offers an extensive overview of applications of hylomorphism to logic
in the tradition of the ancient commentators, see J. BARNES, «Logical Form and
Logical Matter», in A. ALBERTI (ed.), Logica, Mente e Persona, Olschki, Firenze 1990,
pp. 1-119; J. BARNES, Truth etc., Oxford University Press, Oxford 2007, chap. 4.

The [syllogistic] figures are like a sort of common matrix: by fitting

matter into them, it is possible to mould the same form in different
sorts of matters. For just as things fitted into one and the same
matrix differ not in form and figure but in matter, so it is with the
syllogistic figures8.

For our purposes, it is crucial to notice that Alexander seems to view

the form of a syllogism as corresponding to its figure, not its mood. In
the same text, he seems to suggest that the form of a valid syllogism is
somehow related to its validity and reliability:

Combinations are called syllogistic and reliable if they do not alter

together with differences in the matter –i.e. if they do not deduce
and prove different things at different times, but always and in every
material instance preserve one and the same form in the conclusion.
Combinations which change and alter configuration together with
the matter and acquire different and conflicting conclusions at
different times, are non-syllogistic and unreliable9.

The last remark can be read as suggesting that something like the mood
of a syllogism is what corresponds to its form, as within each figure there
are valid as well as invalid moods. But generally speaking, the ancient
commentators tended to associate the form of a syllogism to its figure, at
least in their explicit statements on the matter.
In early Latin medieval texts, references to the form and matter of
syllogisms or arguments all but disappear completely, arguably due to the
fact that Boethius does not make systematic use the hylomorphic framework
in his writings on syllogistic10. Nevertheless, some of the gist of the logical
hylomorphism of the ancient commentators is present in Boethius, for
example in the distinction between propositionum complexio and rerum
natura (e.g. HS II ii 5)11. Picking up from Boethius, Abelard’s distinction

Alexander of Aphrodisias, On Aristotle’s Prior Analytics 1.1–7, Translated by
J. BARNES – S. BOBZIEN – K. FLANNERY – K. IERODIAKONOU, Duckworth, London 1991, p. 48.
Alexander of Aphrodisias, On Aristotle’s Prior Analytics, op. cit., p. 114.
One exception: Boethius, De syllogismo categorico, Ed. by Ch. THOMSEN
THÖRNQVIST, University of Gothenburg, Gotheburg 2008, p. 830D (Studia Graeca et
Latina Gothoburgensia 68). (I owe this reference to Chris Martin).
E.g. Boethius, De hypotheticis syllogismis, Ed. by L. OBERTELLO, with Italian
translation, Paideia, Brescia 1969, II ii 5 (Istituto di Filosofia dell’Università di Parma,
Logicalia 1).

between perfect and imperfect inferences in the Dialectica is very much

in the spirit of the logical hylomorphism of the ancient commentators, but
with no use of specific hylomorphic terminology.
In contrast, in the only currently known commentary on the Prior
Analytics from the 12th century, the so-called Anonymus Aurelianensis III
(henceforth, AA III), the form-matter distinction is extensively applied to
syllogisms, apparently under direct Greek influence (as argued by Ebbesen
and Thomsen Thörnqvist)12. But AA III adds a crucial element, which
apparently is not to be found in the ancient commentaries: an equivocal
(‘duplex’) understanding of the form and matter of syllogisms.

The form of a syllogism is understood equivocally, i.e. as the

disposition of terms which is called its figure, and the disposition
of sentences, called its mood, and in both cases it is unique to each
syllogistic type13.

Just as the form of a syllogism is equivocal, so is its matter, i.e.

terms and sentences14.

Similar suggestions are found in the early 13th century anonymous

treatise known as Dialectica Monacensis, and in Robert Kilwardby’s
commentary on the Prior Analytics, dated circa 1230:

It must be noted that every totality is constituted of form and

matter. As a syllogism is such a totality, it too must be constituted
of form and matter. The matter of a syllogism is equivocal, namely
proximate or remote. The remote matter are the three terms: such a
trio of terms constitutes every syllogism, and it is impossible that it

S. EBBESEN, «Analysing Syllogisms or Anonymus Aurelianensis III, The
(Presumably) Earliest Extant Latin Commentary on the Prior Analytics, and its
Greek Model», Cahiers de l’Institut du Moyen-Âge Grec et Latin, 37 (1981) 1-20;
Ch. THOMSEN THÖRNQVIST, «The ‘Anonymus Aurelianensis III’ and the Reception of
Aristotle’s Prior Analytics in the Latin West», Cahiers de l’Institut du Moyen-Âge Grec
et Latin, 79 (2010) 25-41.
«Forma autem syllogismi duplex est, sc. dispositio terminorum quae figura
vocatur, et dispositio propositionum, quae dicitur modus, et utraque unica est ad omnia
genera syllogismorum.» (EBBESEN, «Analysing Syllogisms», p. 14) The translations
from Latin are my own, unless otherwise stated.
«Sicut autem forma syllogismorum est duplex, ita est materia duplex, scilicet
termini et propositiones.» (EBBESEN, «Analysing Syllogisms», p. 15).

be composed of more or fewer [terms]. The proximate matter are the

three sentences. And just as matter is equivocal, so is the form [of a
syllogism]. The terms must have a figure, and it is the arrangement
of the trio of terms which is produced by subjecting and predicating.
And since three terms can only receive three arrangements in
subjecting and predicating, and there are two premises, we can only
have three figures. Similarly, from the sentences comes the mood,
namely quality and quantity, and this makes a syllogism. The mood
is thus the disposition of the trio of sentences which pertains to
quality and quantity15.

And thus we see in a syllogism an order in matter and form. For

matter, as the term is its remote and unarranged matter, the sentence
is truly its proximate and arranged matter. For form, as the figure is
the incomplete form in potentiality towards future forms, the mood
is thus the ultimate, completing form of a syllogism16.

In other words, according to these texts, there are two ways in which
the form of a syllogism can be understood, just as there are two ways in
which the matter of a syllogism can be understood: the form of a syllogism
can be either its figure or its mood, and correspondingly its matter can be

«Notandum est quod omne totum constat ex materia et forma. Cum autem
sillogismus sit quoddam totum, necesse est ipsum constare ex materia et forma.
Materia autem sillogismi duplex est, scilicet propinqua et remota. Remota materia
sunt tres termini: ex tribus enim terminis constat omnis sillogismus et impossibile
est quod ex pluribus vel paucioribus fiat. Propinqua vero materia tres propositiones
sunt. Duplici autem materie debetur duplex forma. Terminis enim figura debetur. Et
est figura ordinatio trium terminorum quae attenditur in subiciendo et predicando.
Et quoniam tres termini non possunt recipere nisi triplicem ordinationem in
subiciendo et predicando et hoc quantum ad duas premissas propositiones, propter
hoc, inquam, non habemus nisi tres figuras. Item, propositionibus debetur modus,
scilicet qualitas et quantitas, ut fiat sillogismus. Modus enim est dispositio trium
propositionum que attenditur in qualitate et quantitate.» (quoted in L. M. DE RIJK,
Logica Modernorum, vol. II-2, Van Gorcum, Assen 1967, p. 491).
«Et sic invenimus in syllogismo ordinem in materia.et in formis. In materiis
quia terminus est materia eius remota et indisposita, propositio vero est materia
propinqua et disposita; in formis etiam quia figura est forma incompleta et in potentia
ad ulteriorem formam. Modus autem est forma ultima syllogismi completiva.» (quoted
in P. THOM, Logic and Ontology in the Syllogistic of Robert Kilwardby, Brill, Leiden
2007, p. 57).

either its terms or its sentences. According to the Dialectica Monacensis

and Kilwardby, but not AA III, this follows from the fact that matter can be
understood in two specific ways, namely either as remote or as proximate.
Kilwardby adds that the mood of a syllogism is its ultimate, completing
form, while its figure is its form in potentiality; this observation clearly
anticipates the later prevalence of the schematic understanding of the form
of arguments (schematic in the sense of viewing the form of an argument
as corresponding to a schema).
It is not clear where this equivocal understanding of the form and
matter of syllogisms is coming from. As we have seen, there is some
ambiguity in how the ancient commentators talk about the form of a
syllogism: they usually say explicitly that it corresponds to its figure,
but some of their observations also suggest that the mood of a syllogism
might also be viewed as its form. Moreover, it all seems to indicate that
the equivocal understanding is not to be traced back to Greek sources, thus
it might well be an independent innovation. Indeed, to my knowledge, the
distinction between materia propinqua and materia remota is common in
Latin texts but is not to be found in Greek texts. At any rate, the equivocal
understanding of the form and matter of syllogisms seems to have been a
fairly widespread doctrine by the first half of the 13th century already, as
attested by its presence in these texts.
Regrettably, though, as long as we do not have a better grasp of
the historical origins of the equivocal understanding of the form of a
syllogism, we are missing an important piece of the puzzle. From the point
of view of the project of ‘conceptual genealogy’ previously mentioned,
the equivocal understanding of the form (and matter) of a syllogism is
absolutely crucial: it represents the transition from the early non-schematic
logical hylomorphism of the ancient commentators to the later schematic
understanding of the form of arguments, syllogisms in particular.
By the end of the 13th century, the concepts of form and matter were
being applied to arguments/consequences more generally, not only to
syllogistic arguments, as can be seen in an illustrative passage by Simon
of Faversham:

When it is said that «an animal is a substance; therefore a man is

a substance is a good consequence» I reply that this consequence
does not hold in virtue of form (ratione formae), but rather in virtue
of matter. Because according to the Commentator on the first book
of the Physics, an argument which is valid (concludens) in virtue of

form must hold in all matter. This consequence, however, holds only
for features which are essential […] and so this consequence is not
formal (formalis)17.

It may well be that the gradual disappearance of the idea that the form
of a syllogism corresponds to its figure is related to the fact that, once the
notions of form and matter are generalized to other kinds of arguments/
consequences, there is no obvious counterpart to figures in non-syllogistic
arguments. A syllogistic mood, by contrast, is simply a schema, a notion
which can be generalized to arguments of all kinds18. In this sense, the
gradual transition in the 13th century from the primacy of syllogistic
arguments towards a more general conception of arguments, culminating
in 14th century theories of consequence, may have been an important factor
in the establishment of a schematic conception of the formal.
It is in Buridan’s treatise on consequence that one finds the clearest
medieval formulation of the schematic understanding of the form of an
argument, which provides the basis for Buridan’s substitutional definition
of formal consequences:

A consequence is called formal if it is valid in all terms retaining a

similar form. Or if you want to put it explicitly, a formal consequence
is one where every proposition similar in form that might be formed
would be a good consequence, e.g., «That which is A is B, so that
which is B is A»19. (my emphasis)

«Et cum dicitur “Hic est bona consequentia: ‘animal est substantia, ergo homo
est substantia’”, dico quod ista consequentia non tenet ratione formae, sed ratione
materiae. Non tenet ratione formae quia secundum Commentatorem I Physicorum
sermo concludens virtute formae debet tenere in omni materia; ista autem consequentia
tantum tenet in essentialibus, et hoc propter identitatem naturae importatam in talibus
per antecedens et consequens; et propter hoc consequentia ista non est formalis.»
(Simon of Faversham, Quaestiones Super Libro Elenchorum, Ed. by S. EBBESEN et al.,
Pontifical Institute of Medieval Studies, Toronto 1984, here a. 36, p. 200, translation
from C. MARTIN, «Formal Consequence in Scotus and Ockham: Towards an Account of
Scotus’ Logic», in O. BOULNOIS – E. KARGER – J.-L. SOLÈRE – G. SONDAG (eds.), 1302:
Duns Scot à Paris 1302–2002, Brepols, Turnhout 2005, pp. 117-150, here p. 135).
Recall that, following Corcoran «Schema», for the present purposes a syllogistic
mood counts as a schema, but not a syllogistic figure.
«Consequentia ‘formalis’ uocatur quae in omnibus terminis ualet retenta forma
consimili. Vel si uis expresse loqui de ui sermonis, consequentia formalis est cui omnis
propositio similis in forma quae formaretur esset bona consequentia, ut “quod est A est B;

I say that when we speak of matter and form, by the matter of a

proposition or consequence we mean the purely categorematic
terms, namely, the subject and predicate, setting aside the
syncategoremes atteched to them by which they are conjoined or
denied or distributed or given a certain kind of supposition : we say
all the rest pertains to the form20.

Thus, according to Buridan, the form of an argument is (at least

partially) defined by the syncategorematic terms occurring in it, while
its matter corresponds to its categorematic terms. At first sight, this may
appear to be an almost exact formulation of LHAWKI, as exemplified
by the textbook passage quoted above. There is however a fundamental
difference: nowhere does Buridan say that the form of an argument
(consequence) is that in virtue of which it is valid. He does say that formal
consequences make their validity evident to us, but this has epistemic
rather than logical import. Nevertheless, with Buridan we clearly no longer
have an equivocal understanding of the form of an argument: it uniquely
corresponds to what we now refer to as an argument schema, which in the
case of syllogisms corresponds to syllogistic moods.

3. Buridan on the figures of modal syllogisms

Does this mean that Buridan does not attribute any usefulness to the
concept of syllogistic figure for logical theorizing? Here again the answer is
negative. Unlike the authors of modern logical textbooks, in practice Buridan
does not endorse the view that logic «investigates the validity of arguments
by investigating the validity of schemata», at least not exclusively. In fact,
the concept of syllogistic figure plays a prominent role in particular in his
analysis of modal syllogisms, as I will show now. The main technical reason

ergo quod est B est A”.» (John Buridan, Tractatus de Consequentiis, Ed. by H. HUBIEN,
Publications Universitaires, Louvain 1976, pp. 22-23, translation in John Buridan, Treatise
on Consequences, transl. by S. READ, Fordham University Press, New York 2015, p. 68).
«Et dico quod in proposito, prout de materia et forma hic loquimur, per ‘materiam’
propositionis aut consequentiae intelligimus terminos pure categorematicos, scilicet
subiecta et praedicata, circumscriptis syncategorematicis sibi appositis, per quae
ipsa coniunguntur aut negantur aut distribuuntur uel ad certum modum suppositionis
trahuntur; sed ad formam pertinere dicimus totum residuum.» (Buridan, Tractatus de
Consequentiis, p. 30, transl. p. 74.).

for this is that focus on figures allows the logician to keep track of the
property of ampliation, which is a crucial property of modal sentences.
But before turning to modal syllogistic, the topic of the fourth book
of Buridan’s treatise, we must attend to his definition of the semantics
of modal sentences as presented at the beginning of the second book, on
consequences between modal sentences. Buridan’s understanding of the
semantics of divided modal sentences differs from that of e.g. Ockham,
and his results on modal syllogistic depend crucially on the exact details of
the semantics he attributes to modal sentences. Here is what he says:

It should be realized that a divided proposition of possibility has a

subject ampliated by the mode following it to supposit not only for
things that exist but also for what can exist even if they do not. […]
So the proposition ‘B can be A’ is equivalent to ‘That which is or
can be B can be A’21.

In practice, the subject of a divided modal sentence functions for

Buridan as a disjunctive term: for example, ‘A is necessarily B’ is equivalent
to ‘That which is or can be A is necessarily B’.
Turning now to modal syllogisms, in the fourth book Buridan proves
a number of general conclusions for modal syllogisms, usually resorting
to figures as the concept affording the suitable level of generality. True
enough, the concept of validity remains tied to syllogistic moods, just as in
the case of assertoric syllogisms, as Buridan himself points out:

When in the preceding conclusion I said or when I will say later,

«is valid,», I always understand by «valid» not that it is valid in all
combinations, but that it is valid in all moods which were claimed
to be valid for assertorics22.

«[…] supponendum est quod propositio diuisa de possibili habet subiectum
ampliatum per modum sequentem ipsum ad supponendum non solum pro his quae
sunt sed etiam pro his quae possunt esse quamuis non sint. Unde sic est uerum quod
aer potest fieri ex aqua, licet hoc non sit uerum de aliquo aere qui est. Et ideo haec
propositio ‘B potest esse A’ aequiualet isti: ‘Quod est uel potest esse B potest esse A’.»
(John Buridan, Tractatus de Consequentiis, p. 58, transl. p. 97).
«Et cum in dicta conclusione dixi uel dicam post ‘ualet’, semper ego per ‘ualere’
intelligo non quod ualeat in omnibus combinationibus sed quod ualeat in omnibus
modis qui positi fuerunt ualere de inesse.» (John Buridan, Tractatus de Consequentiis,
p. 115, transl. p. 143).

He will then investigate, for each of the syllogisms that are valid in
the case of assertoric sentences, whether they also correspond to valid
modal syllogisms, and of which kind. In other words: within the valid
moods for assertoric syllogisms, which combinations of the modalities of
necessity and possibility will also produce valid modal syllogisms? This is
the general question. Now, rather than proceeding by an analysis of modal
syllogistic moods, Buridan focuses on syllogistic figures. As we shall see,
this is a clever move, since what determines the properties of a modal
syllogism besides its mood is the ampliation of its terms, which can be kept
track of through its figure, as some examples will illustrate.
Let us start with his fourth conclusion:

Fourth Conclusion: In the first figure with both [premises] of

necessity or of possibility or one of necessity and the other of
possibility there is always a valid syllogism to a conclusion of the
same kind as the major [premise]23.

These are evident by the dictum de omni et nullo. They are all perfect
syllogisms or nearly perfect. For if the major [extreme] is explicitly
expressed by a disjunction of the verb ‘is’ with the verb ‘can’, then
if the minor [premise] is of possibility it will be clearly subsumed
under the distribution of the major [extreme]; while if the minor is
of necessity, the same is true, since that of possibility follows from
that of necessity24.

In order to analyze this claim, let us introduce some notation to capture

the ampliation/supposition of terms.

[ ]A : ‘A’ supposits for that which is necessarily A.

<>A : ‘A’ supposits for that which can be A.

«In prima figura ualet semper syllogismus ex ambabus de necessario uel de
possibili aut ex una de necessario et alii de possibili ad conclusionem talis modi qualis
est maior.» (John Buridan, Tractatus de Consequentiis, p. 115, transl. p. 143).
«Haec sunt manifesta per dici de omni uel de nullo. Et sunt omnes syllogismi
perfecti uel quasi perfecti. Si enim maior explicite exprimatur per disiunctionem
huius uerbi ‘est’ ad hoc uerbum ‘potest’, tunc si minor fiat de possibili erit manifesta
sumptio sub distributione maioris; si autem minor fiat de necessario, adhuc idem redit,
quia ad illam de necessario sequitur illa de possibili.» (John Buridan, Tractatus de
Consequentiis, p. 115, transl. p. 143).

*A : ‘A’ supposits for that which is A.

*<>A : ‘A’ supposits for that which is or can be A.
/ : any copula joining subject to predicate.

I will use ‘/’ to denote the copula connection, of any kind. Precisely
because, at this point, the idea is to focus exclusively on figures rather than
on moods, no attention will be paid to the two kinds of copulas (affirmative
or negative), nor to the quantity of the sentences, as these are two properties
pertaining to mood. In other words, at this point we abstract away from
quality and quantity, focusing exclusively on the mutual disposition of the
terms which is what defines the figure of a syllogism.
Thus, a generic modal categorical sentence of necessity is represented

*<> A/[ ]B

A modal sentence of possibility is represented by:

*<> A/<>B

A first-figure syllogism with two premises of necessity becomes:

*<>A/[ ]B
*<>B/[ ]C
*<>A/[ ]C

In the conclusion, A and C simply maintain the ampliation they have

in the premises, which can be immediately (almost graphically) perceived.
As for the middle term B, it has different suppositions in the two premises,
but not in a way that would block the conclusion, because the ampliation
of B in the minor premise is subsumed under the supposition of B in the
major premise: everything which is necessarily B is or can be B as well.
A first-figure syllogism with two premises of possibility becomes:


Here, a similar reasoning applies: the ampliation of B in the minor

premise is subsumed under the supposition of B in the major premise, as
what can be B a fortiori satisfies the disjunctive clause of what is or can
be B.
The mixed case follows by a similar reasoning. Let us use # and ^ to
denote either one of the modalities; we then obtain:


As we have seen, both in the case of a minor premise of necessity and

in the case of a minor premise of possibility, the ampliation of the minor
middle term is subsumed under the supposition of the major middle term.
So in the conclusion, C simply carries over the ampliation that it has in
the major premise, which is thus determined by the modality of the major
Let us now turn to the third figure. Buridan states the following

Sixth Conclusion: In the third figure a conclusion of possibility

always follows from two premises of possibility, and a conclusion of
necessity from two [premises] of necessity, and a conclusion of the
same kind as the major [premise] from one [premise] of necessity
and the other of possibility25.

With two premises of possibility we have:

*<>A/ <>B
*<>A/ <>C
<>B/ <>C

«In tertia figura semper ex ambabus de possibili sequitur conclusio de possibili,
et ex ambabus de necessario sequitur conclusio de necessario, et ex una de necessario
et altera de possibili sequitur conclusio modi similis modo maioris.» (John Buridan,
Tractatus de Consequentiis, p. 116, transl. p. 144).

Now, since clearly <>B/ <>C implies *<>B/ <>C (because the subject
is read as a disjunctive term), we obtain:

*<>A/ <>B
*<>A/ <>C
*<>B/ <>C

For two premises of necessity, the conclusion is in first instance

[ ]B/[ ]C, but this implies <>B/[ ]C, which in turn implies *<>B/[ ]C.
With mixed premises, we obtain:


Now, if ^ is either <> or [ ], as we have seen, in both cases it implies

*<>B/#C, thus the conclusion is of the same modality as the major premise.
For the second figure, the reasoning is slightly more complicated, as it
involves indirect proofs, but the general strategy is similar.
The general idea is thus that the two terms in the conclusion carry
over the ampliation they have in the premises to the conclusion. As for
the middle term, some caution is required to ensure that, in the case of
the first figure, if it has different ampliations/suppositions in each of the
premises, the supposition of the minor middle term is subsumed under the
supposition of the major middle term. For the third figure, the intermediary
conclusions having <> or [ ] as the modality ranging over the subject
then entail the more general conclusion with *<> as the subject’s modality.
This reconstruction follows closely Buridan’s own reasoning to show the
correctness of his general claims.
The main methodological point emerging from these considerations
is that focus on figures allows for a geometrical, quasi-algebraic approach
to modal syllogisms, based on the position of the terms in the syllogism –
which is precisely what a syllogistic figure is about– and their ampliations.
And yet, syllogistic figures are not schemata, and no reference is made
to the so-called ‘logical terminology’, besides the ampliative effect of
modal terms. Clearly, investigating the validity of schemata is not the
only viable approach to investigate the validity of arguments, and in this

case the focus on figures allows for a greater level of generality in the
True enough, and as remarked before, the case of syllogistic is rather
special in that a syllogism has two distinctive properties, a mood and a
figure, whereas it is not immediately obvious what the counterpart of a
syllogistic figure would be for arguments in general. So the claim is not
that the ‘figure perspective’ should be adopted across the board for the
analysis of the validity of arguments, but rather that, at least in the case of
modal syllogisms, and coupled with the concept of ampliation, it offers a
particularly fruitful vantage point for logical analysis.

4. Conclusion

Elsewhere26 I argued that schematic logical hylomorphism is a

misguided doctrine, at least if it is meant to offer an account of that in
virtue of which arguments are valid, entailing the view that validity is
exclusively a matter of form. I also argued that this version of logical
hylomorphism is based on a rather shaky conception of the metaphysics
of arguments: it is a mereological hylomorphism which requires a unique,
principled and sharp separation of the form of an argument from its matter
–something that might simply be impossible to achieve, as the vexing and
unresolved issue of the demarcation of logical constants suggests. This
does not mean that focusing on schemata cannot be a fruitful approach in
logical inquiry, and much progress in the history of logic has been made
on the basis of the notion of schemata. But the schematic approach is best
seen as an instrumental, pragmatically fruitful stance, not as providing
answers to hard philosophical questions such as what warrants the validity
of arguments.
In this paper, I began by reviewing the developments leading from
the association of the form of a syllogism to its figure with the ancient
commentators to the association of the form of arguments in general to
schemata, including syllogistic moods, in the 14th century with Buridan.
But while Buridan relies on schemata to introduce the concept of formal
consequence, in the same treatise he makes extensive use of syllogistic
figures for logical analysis, as discussed in the second half of the paper.

DUTILH NOVAES, «Reassessing logical hylomorphism».

Buridan thus illustrates the general attitude in logic that I would like
to recommend: one should adopt the perspective that is most suitable
for the particular subject matter under investigation –in other words,
a methodologically liberal stance. Schemata are not the only kinds of
constructions relevant for logical analysis. For modal syllogisms, Buridan
has shown that adopting the figure-perspective is a particularly fruitful



Aristotle tells us that his investigations into the syllogism in the Prior
Analytics fall into three parts, dealing respectively with the theory of
syllogistic deduction, how to find a middle term so as to make a syllogism,
and how to analyse a piece of discourse into syllogistic form1. I want to
talk about this third part, the part that deals with analysis. I will argue
that anyone who takes this part of the Analytics to be part of logic, must
have a conception of logic that is partly non-formal, that Robert Kilwardby
conceived of logic in this way, that he was aware of the difference in kind
between analysis and deduction, that some of his successors weakened his
concept of syllogistic form, that they conceived of some matters which
Kilwardby dealt with under the rubric of analysis as better dealt with under
the rubric of deduction, and that in extending the reach of formal logic in
this way they no longer treated analysis into logical form as a significant
part of logic.
Kilwardby devotes close and extended attention to difficulties that
arise in putting informal reasoning into syllogistic form. He poses and
answers more than 30 detailed questions to these difficulties. Many later
question-commentaries displayed less interest in these chapters. Two of the
thirteenth-century commentaries listed by Sten Ebbesen have no questions
at all on the subject2. Simon of Faversham has only 3 questions –two about
oblique syllogisms and one about reduplicated propositions3. Radulphus
Brito shows greater interest in these matters, having 16 questions covering
most of the Aristotelian material4.

School of Philosophical and Historical Inquiry, The University of Sydney, NSW
2006. Email: paul.thom@sydney.edu.au
Aristotle, Prior Analytics A32, 47a1.
Anonymus Wignorniensis and the anonymous Quaestiones primae in ms
Cambridge, Gonville & Caius 611/341: 25ra-47vb. See S. EBBESEN, «The Prior
Analytics in the Latin West: 12th-13th centuries», Vivarium, 48 (2010) 96-133, pp. 123-
EBBESEN, «The Prior Analytics in the Latin West», pp. 118-119.
Ibid., p. 122, pp. 132-133.

My focus will be on Kilwardby’s procedure in reducing arguments

to syllogistic form, with special attention to his treatment of arguments
in which one or other of the terms is infinitated, i.e. is subjected to an
unqualified term-negation. I will also briefly look at Kilwardby’s treatment
of reduplicated propositions. Finally, I will try to draw out some general
features of the development of logic in second half of the thirteenth and the
early fourteenth centuries.

Kilwardby’ account of reduction to syllogistic form

Kilwardby’s treatment of this material is based on a strong conception

of syllogistic form, and it maintains a clear demarcation between what is
yet-to-be-syllogized and what exhibits syllogistic form.


The process whereby what is yet-to-be-syllogized gets expressed as a

syllogism is, in Aristotle’s word, a process of analysis; Boethius’s word is
reductio. Kilwardby describes the process in the following way:

a certain reduction of an utterance that is to be syllogized is made

when it is syllogized in actuality, and this is when an utterance that
is arranged in a confused and disorderly way has to be reduced
subsequently to syllogistic form5.

The concept of ordering occupies an important place in Kilwardby’s

view of the content of the Prior Analytics. He says that the first section
of Book I, i.e. Chapters 1-26, has as its main concern the ordering of the

Robert Kilwardby, Notule libri Priorum [henceforth, NLPri] Ms Firenze,
Biblioteca Nazionale Centrale [henceforth, BNCF] J.10.48, f. 49vb: «Et dicendum quod
reductio sillogismorum est multiplex. Aut enim est oratio sillogistica actu sillogizata
uel potentia tantum. Si potentia, sic fit quedam reductio orationis sillogizande in actu
sillogizata, et hoc est quando oratio confuse et inordinate disposita est, et postea reduci
debet in formam sillogisticam (talis autem est reductio que determinatur hic)» (A32-
33, dub. 1).

middle term in relation to the extremes6. Thus he sees the theory of the
syllogism (which is what is presented in those chapters) as being principally
concerned with a type of ordering. It is not surprising, therefore, that he
views the section of the book dealing with the application of that theory
as showing us how to uncover or impose order in materials that are to-be-
syllogized. For, to understand something as a syllogism is to understand it
as possessing a certain type of order.
Kilwardby characterises the relationship between the yet-to-be-
syllogized and the syllogized in two different ways. What is yet-to-be-
syllogized is ‘potentially a syllogism’7. At the same time, in actualizing
this potential, ‘we must … correct the way of taking <the terms> if they
are taken in an inappropriate way’8. These seem to be antithetical activities:
to actualise the potential that exists in something is to uncover something
that is already latent in the thing’s nature, whereas to correct something
is to replace something in it by something that initially was not in it. Its
potentialities are internal to the thing, whereas corrections assume a standard
that is external to it. The antithesis is real, but it is also inevitable given that
to express something in syllogistic form is to interpret it. An interpreter
certainly aims to find something within the object of interpretation that
will make of it the kind of sense that is sought; but sometimes sense can
be made of the object only by bringing it into alignment with an external
standard. This double polarity is a general feature of the activity of
interpretation, and therefore applies to the particular case where inferential
material is interpreted syllogistically. Kilwardby shows an awareness of
the interpretive nature of reduction to syllogistic form when he says that

Kilwardby, f. 28ra: «In prima determinat sillogismorum generationem, in
secunda, (cum dicit ‘Quomodo autem reducemus’) eorumdem reductionem.
Ad generationem autem sillogismorum duo pertinent, scilicet ordinacio medii
respectu extremorum et eiusdem inuencio. Ideo prima in duas. In prima determinat
ordinacionem medii respectu extremorum, in secunda, (cum dicit ‘Quomodo autem
idonei’), determinat eiusdem inuencionem.»
Kilwardby, f. 49vb: «Et dicendum ad primum quod si oratio que solum
potentia sillogizata est debeat in formam sillogisticam reduci, necesse est illam
formam sillogisticam prius cognosci. Sed illa non cognoscitur nisi per sillogismorum
generationem» (A32-33, dub. 2).
Kilwardby, f. 51ra: «quasi diceret, secundum predicta documenta considerandum
est terminos in reductione et corrigendum est eorum acceptionem si inconuenienter
accipiantur» (A37, 49a10).

the syllogism is ‘understood’ in the material to-be-syllogized9, adding that

in the process of reduction we must not be misled into ‘accidental ways of
understanding’ but must articulate the ‘essential comparison of the middle
to the extremes’10.
Some modern writers take the process of putting something into
logical form to be one of translation. Alex Oliver holds this view. When
we put somehing into logical form, according to Oliver, we are simply
translating pre-formal material into sentences of a given logic, with the
aim of preserving some at least of the original sentences’ inferential
properties11. By contrast, the process of reduction, as Kilwardby conceives
it, sometimes requires that we re-order things and correct errors. Moreover,
as we shall see, it sometimes requires that we suppress existing material
and introduce new material. But, to change the order, to make corrections,
deletions and additions –these are not the activities of a good translator.
So Kilwardby presumably would not accept an Oliver-style account of the
activity of reduction to syllogistic form.
Like all interpretation, reduction to syllogistic form aims to make
a specific kind of sense of its object, and it presupposes an interpreting
scheme that is capable of endowing that kind of sense on an object of
interpretation. In the present case the aim is to make rational sense of the
object. The interpreting scheme, in general terms, is a theory of logic,
and the reduction is a rational reconstruction. Specifically, the logical
theory that Kilwardby presupposes in commenting on these chapters is
the Aristotelian theory of the syllogism, and it is assumed that rational
discourse can be reduced to syllogistic form.
The process of reducing an inference to syllogistic form cannot itself
be seen as a formal inference. It is therefore appropriate that the expressions
Kilwardby uses in describing this process (the expressions ‘introduction
of order’, ‘essential understanding’, ‘actualisation of potential’, and

Kilwardby, f. 50ra: «Sed minor huius sillogismi intelligitur in secunda
propositione orationis dicte universaliter intellecte» (A32-33, dub. 5).
Kilwardby, f. 51rb: «Et dicendum ad primum quod talis diuersitas medii que
est secundum obliquum et rectum non est nisi secundum accidentales modos
intelligendi, et hec diuersitas non creat diuersitatem in comparatione essentiali medii
ad extrema, et ideo talis diuersitas medii non impedit sillogismum» (A36-37, dub. 3).
Compare A. OLIVER, «The matter of form: logic’s beginnings», in J. LEAR – A.
OLIVER (eds.), The Force of Argument: essays in honor of Timothy Smiley, Routledge,
London 2010, pp. 165-185, p. 180.

‘correction’) are all distinct from the technical terms used by logicians to
describe the relation between premises and conclusion in a valid inference
(‘entailment’, ‘deducibility’, ‘syllogize’). The distinction demonstrates
Kilwardby’s firm resolve to maintain a conceptual space and a distinctive
vocabulary for the yet-to-be-syllogized. He thought that the study of the
syllogism formed the core of logic, but he knew that one essential part of
that study was non-formal.

Syllogistic form

The products of this process are inferences expressed in syllogistic

form. In order to possess syllogistic form, an inference must contain
nothing superfluous and must omit nothing that is needed, so that there are
two propositions, which are related in the due manner12 i.e. with one as a
whole and the other a part13, with the whole stated before the part14, and
with the propositions transcribed into predicative form15.
Kilwardby’s conception of syllogistic form relies on a traditional
exposition of Aristotle’s definition of the syllogism –in particular, he
understands the Aristotelian requirement that in the premises some things
are stated [positis in Boethius’s translation] to mean that the premises must
be stated in figure and mood, and that figure and mood constitute the form
of the syllogism16. But as his commentary continues, this conception is

Kilwardby, f. 49va: «Consequenter determinat difficultatem siue errorem
accidentem circa habitam reductionem. Et primo manifestat in terminis qualiter
fit deceptio ex superfluitate uel diminuitione» (A32, 47a22). Ibid.: «Non statim
temptandum est tales orationes reducere antequam fiat acceptio duarum propositionum
debito modo se habencium» (A32, 47a28).
Kilwardby, f. 49va: «docet quomodo se debent habere propositiones, dicens
quod altera ut totum, reliqua ut pars» (A32, 47a12).
Kilwardby, f. 50ra: «hic non est debita positio propositionum, illa enim que ut
pars est primo ponitur, que autem est ut totum secundo, tamquam ipsa sub sua parte
accipiatur» (A32-33, dub. 7).
Kilwardby, f. 50ra: «Si autem transferantur propositiones in predicatiuas sic,
‘Omnis homo est animal et omne animal est substantia’, iam non est nisi peccatum
unum contra sillogismum, scilicet indebita potitio propositionum» (A32-33, dub. 7).
Kilwardby, f. 28vb: «Et dicendum quod sicut ordo est in materiis – quedam
enim est remota et indisposita, quedam autem propinqua et disposita – sic et in
formis. Quedam enim forma materialis est et in potentia ad formam ulteriorem,

strengthened. In order for an inference to possess syllogistic form it is

not enough that it be in some figure and have some mood; it must also
be in accord with the principles governing individual syllogistic figures
and moods17. For example, an inference in the second figure must have
a negative premise; and, since Aristotle tells us that all contingency-
propositions are affirmative in form, this means that no second figure
inference having a contingency-premise and an affirmative necessity-
premise possesses syllogistic form18.
Further requirements emerge elsewhere in Robert’s commentary. In
his discussion of A28 he argues that third figure inferences with a singular
term as middle are not properly speaking syllogisms, because while they
are in a figure they are not in any mood19. Syllogistic mood, it seems,
requires either universal or particular premises, not singulars. In a note
on A9 he also excludes singular propositions from appearing in mixed
necessity/assertoric first figure syllogisms20.

quedam autem est ultima et completiua. Et sic inuenimus in sillogismo ordinem

esse in materiis et in formis – in materiis quia terminus est materia eius remota et
indisposita, propositio uero materia propinqua et disposita; in formis quia figura est
forma incompleta ens in potentia ad ulteriorem formam, modus autem est forma
ultima sillogismi completiua. Et respondet incompletum in formis incompleto
in materiis, scilicet figura termino, et completum in formis completo in materiis,
scilicet modus propositioni» (A4, dub. 3).
These principles are listed in P. THOM, Logic and Ontology in the Syllogistic of
Robert Kilwardby, Brill, Leiden 2007, pp. 176-177, 238.
See THOM, Logic and Ontology, pp. 231-234.
Kilwardby, f. 48ra: «Consequenter queritur cum duplex sit dispositio formalis
in sillogismo, scilicet modus et figura, et sunt inutiles inspectiones peccantes contra
modum, quare non sunt alique inutiles peccantes contra figuram.» […] «Et dicendum
quod hoc est quia due propositiones facte in tribus terminis per situm terminorum
ex necessitate determinant figuram, sed non de necessitate determinant modum. Et si
abesset figura, nulla esset dispositio ad sillogismum, et ideo quantum ad figuram nulla
deficit inspectio» (A28.1, dub. 2). See THOM, Logic and Ontology, pp. 114-115.
Kilwardby, f. 33rb: «Et nota quod in minore accipienda est minor extremitas
‘sub’ medio, non quod sit sub eo accidentaliter, neque quod sit sub eo ut nunc, sed
quod sit sub eo indifferens ei; uerba gratia, necesse est omnem hominem esse animal.
Deinde non est accipiendum sub homine aliquid quod est sub eo ut nunc (cuiusmodi
est ‘Socrates’), neque quod secundum accidens (cuiusmodi est ‘album’), sed aliquid ei
indifferens (cuiusmodi est ‘aliquis homo’)» (A9, nota 2).

Infinite terms

In Chapter 32 Aristotle discusses ways in which we can be misled

into thinking that an inference is syllogistic when it isn’t so. For example,
we may take an inference to be syllogistic because its conclusion follows
necessarily from its premises21. But it may not be so; and if not we will
have to add any unstated premise that is needed, and delete any stated
premise that is superfluous, for a syllogism to come about22. As an example
Aristotle cites the inference:

When a non-substance is destroyed, a substance is not destroyed;

when a thing’s parts are destroyed, the thing is destroyed; so
substance-parts are substances.

This might seem to satisfy the Aristotelian definition of a syllogism,

because its conclusion is different from its two premises and follows from them
necessarily. But as against that, it cannot be a syllogism, because it derives an
affirmative conclusion from premises one of which is negative, and because
the conclusion contains a term (‘substance-part’) that is not in the premises23.
Kilwardby notes that there is a syllogism leading to the desired
conclusion, namely:

Kilwardby, f. 49vb: «multociens fit deceptio eo quod putatur esse sillogismus
qui non est, et hoc aliquando in orationibus necessariis quia enim necessitatem habent
creditur quod sint sillogismi» (A33, 47b15).
Kilwardby, f. 49va: «Ideo oportet si sui necessariorum omissum sit apponere,
si quid autem superfluum interrogatum auferre» (A32, 47a14).
Kilwardby, f. 50ra: «Consequenter dubitatur de suo exemplo ubi dicit
conclusionem sequi sed propositiones deficere ut hic: Non substantia interempta
non interemitur substantia, sed partibus rei interemptis interimitur res, ergo partes
substantie sunt substantie. Uidetur enim quod sit sillogismus, cum sit oratio in qua
quibusdam positis etc.
Oppositum autem uidetur multipliciter. Et primo per hoc quod altera premissarum
negatiua est et conclusio affirmatiua, quod non contingit in sillogismo. Adhuc
quicumque terminus est in conclusione oportet quod sit in premissis, sed in dicta
oratione est aliquis terminus in conclusione qui non est in premissis. Adhuc omnes
termini premissarum preter medium inueniri debent in conclusione, sed sic non est in
dicta oratione, quare non est sillogismus. Et hoc uerum est.
Ad obiectum dicendum quod non est ibi recta dispositio terminorum et
propositionum. Et ideo non conuenit ei diffinitio silogismi licet concludat ex
necessitate» (A32-33, dub. 4).

Things whose destruction leads to the destruction of a substance

are substances; but the destruction of substance-parts leads to the
destruction of a substance; so substance-parts are substances.

But the premises of this syllogism are not those of the given syllogism.
The major premise here is an affirmation about things whose destruction
leads to the destruction of a substance, whereas the original major was a
denial about non-substances. The new minor premise, on the other hand,
can be understood in the original minor, because whereas the original
minor made a statement about the parts of things in general the new minor
is about substance-parts in particular.
Kilwardby proposes to turn the inference into a syllogistic argument
leading from the original premises to the original conclusion in two stages.
In the first stage, given that the desired conclusion is about substance-parts,
he replaces the original minor premise (which is about things in general)
with a premise whose subject is restricted to substance-parts. One might
expect that he would also reformulate the negative major premise with
an affirmative, given that the desired conclusion is affirmative. However,
Kilwardby cannot do this, given that affirmatives are in general supposed
not to follow from negatives. So he retains the major premise in its negative
form. These two premises will lead by a second-figure syllogism to the
conclusion that no substance-part is a non-substance:
When a non-substance is destroyed, a substance is not destroyed;
but when substance-parts are destroyed, a substance is destroyed;
so substance-parts are not non-substances24.

Kilwardby, f. 50ra: «Queritur igitur quomodo desunt ibi propositiones.
Et dicendum quod proprie premisse huius conclusionis ‘Partes substantie sunt
substantie’ sunt due: ‘Quibus interemptis interimitur substantia, substantie sunt; sed partibus
substantie interemptis interimitur substantia; ergo partes substantie sunt substantie’. Iste
propositiones desunt in predicta oratione ad hoc quod immediate sillogizetur conclusio.
Sed minor huius sillogismi intelligitur in secunda propositione orationis dicte uniuersaliter
intellecte. Per prosillogismum autem sillogizatur sic, ex prima propositione posita in litera
et minore subintellecta, in secunda propositione per secundam figuram: ‘Non substantia
interempta, non interimitur substantia; partibus substantie interemptis interimitur
substantia; ergo partes substantie non sunt non substantie’. Ex hoc autem et quodam uero
coassumpto sillogizatur sic: ‘Que sunt et non sunt non substantie sunt substantie; sed partes
substantie sunt et non sunt non substantie; ergo partes substantie sunt substantie’. Sic igitur
patet quomodo ex prosillogismo sillogizatur et qualiter desunt propositiones et que, et quod
oratio quam ponit quamuis sit necessaria non tamen est sillogismus» (A32-33, dub. 5).

In the second stage, Kilwardby’s aim is to get from the universal

denial that no substance-part is a non-substance to the desired universal
affirmation that every substance-part is a substance. To this end, he brings
in what he calls the ‘true assumption’ that any being that is not a non-
substance is a substance. As it stands, the intermediate conclusion, that
substance-parts are not non-substances, doesn’t mesh syllogistically with
this assumption; so he reformulates that conclusion as ‘Substance-parts
are beings that are not non-substances’. He is then able to deduce by a first
figure syllogism that substance-parts are substances.
The whole analysis can be represented in the following way (where D
stands for ‘thing such that a substance is destroyed when it is destroyed’, P
for ‘substance-part’ and S for ‘substance’):

When a non-S is destroyed an S is not destroyed

¨ (1) No non-S is a D
When a thing’s part is destroyed the thing is destroyed
¨ (2) Every P is a D
No P is a non-S ¨ Every P is a being that is a non-S
(3) Every being that is not a non-S is an S
Every P is an S

Figure 1. Reduction of ‘Every substance-part is a substance’ (Kilwardby)

The original premises (‘When a non-substance is destroyed, a

substance is not destrotyed’ and ‘When a thing’s parts are destroyed the
thing is destroyed’) are first reformulated as propositions (1) and (2).
(The arrows indicate this process of reformulation.) (1) and (2) lead by a
prosyllogism in Cesare to the intermediate conclusion ‘No P is a non-S’.
This is reformulated as ‘Every P is a being that is not a non-S’. And this,
together with Kilwardby’s extra assumption (3) leads by a syllogism in
Barbara to the desired conclusion ‘Every P is an S’.
This process involves two types of transformation, indicated
respectively by arrows connecting statements, and by horizontal lines under
premise-pairs. Transformations of the first type are strategic reformulations
that are aimed at getting the inference into a format that can be handled
by syllogistic theory. Those of the second type are formal transformations
within syllogistic theory.

Interesting here is the transformation of ‘No P is a non-S’ into ‘Every P

is a being that is not a non-S’. The formal terms in which I have expressed
this transformation might suggest that I am attributing to Kilwardby a
formal logical rule to this effect. But this impression would be misleading.
There doesn’t seem to be any body of logical theory in Kilwardby to which
such a formal transformation would belong. I think he simply takes these
two expressions as different formulations of the same proposition; thus
this transformation is of the same general type as the initial interpretive
reformulations of the original premises.
One further feature of Kilwardby’s analysis must be mentioned.
He presents the proposition (3) as a true assumption; but it looks like
something more significant than that. It looks as if it ought to be part of a
formal logical theory of negative terms. I don’t think it has this character
in Kilwardby’s thought; but we will see that it, or something rather like it,
does take on this character in the work of a later logician.

Reduplicative propositions

In commentaing on Aristotle’s Chapters 32-46 Kilwardby considers

a number of difficult cases for the application of syllogistic theory to
ordinary inferences. I will take his treatment of reduplicative propositions
as representative.
In Chapter 38, Aristotle considers the reduction to syllogistic form of
inferences containing reduplicated propositions (i.e. propositions in which
A is said to be B insofar as it is C). Kilwardby notes that reduplication
appears to be a property of the proposition’s subject:

It appears as if reduplication is a property of the subject, because it

bespeaks the cause why the predicate inheres in the subject25.

But he finds reason to support Aristotle’s view that reduplication

attaches to the proposition’s predicate:

Reduplication indicates the cause of the predicate’s inhering in

the subject, but the relation of inherence begins from the predicate

Kilwardby, f. 51vb: «Et uidetur subjecti, quia reduplicatio dicit causam entem
in subiecto quare predicatum ei inest» (A38-40, dub. 1).

and ends in the subject, so reduplication is properly added to the


He notes all of Aristotle’s examples of syllogistic reasoning with

reduplicated terms. He also identifies some classes of such reasoning not
mentioned by Aristotle, pointing out that there are such inferences with
negative premises27, and that there can be syllogisms in all the figures with a
reduplication of the middle and the minor28. What he does not do is to provide
any comprehensive account of the semantics of reduplicative propositions or
of the logic of inferences containing them. He says that Aristotle doesn’t
mean to teach the modes of syllogizing with reduplication, but only to show
how reduction to terms is be done when reduplication is present29.

Later formal developments

The century following Kilwardby’s work saw huge changes in the

way logic was studied in the Latin West. Three sorts of development are
especially striking in the present context. First, the concept of syllogistic
form as propounded by Kilwardby was loosened by dropping some of
Kilwardby’s requirements. Second, greater interest was devoted to non-
syllogistic forms of inference. And third, the Aristotelian art of reducing
pre-theoretic materials to syllogistic form played a decreasing role in the
writings of logicians.

Kilwardby, f. 51vb: «Reduplicatio autem […] signat causam inherencie, que
quidem inherencia incipit a predicato et terminatur ad subiectum, et ideo additur
predicato proprie» (A38-40, dub. 1).
Kilwardby, f. 51vb: «Et quod contingat palam est in exemplo sic: ‘Nullum
eligendum est malum in quantum eligendum; omne bonum est eligendum; ergo nullum
bonum est malum in quantum eligendum’. Hic iam facta est reduplicatio medii. Fit
etiam reduplicatio minoris sic: ‘Nullum corpus in quantum egrum est sanum in
quantum egrum; omne egrum est corpus in quantum egrum; ergo nullum egrum est
sanum in quantum egrum’» (A38-40, dub. 3).
Kilwardby, f. 52ra: «Et dicendum quod in secunda figura fieri potest minoris
reduplicatio in minori propositione, sed non potest fieri reduplicatio medii. In tercia
autem figura econuerso fit reduplicatio medii sed non minoris» (A38-40, dub. 4).
Kilwardby, f. 51vb-52ra: «Et dicendum quod Aristoteles non intendit hic docere
modos sillogizandi cum reduplicatione, sed tantum intendit quomodo in terminos
facienda est reductio duarum propositionum ubi sumitur reduplicatio» (A38-40 dub.3).

With regard to the notion of syllogistic form, Ockham continued to

maintain the traditional exposition of the Aristotelian definition of the
syllogism, according to which a syllogism must be in a figure, but he
allowed that the premises could be singular propositions30. Buridan widened
the application of the concept of a syllogism even beyond the limits set by
Ockham. He rejected the traditional gloss on Aristotle’s definition of the
syllogism, and with it the requirement that every syllogism be in one of the
three figures. He argued that, since we use the definition of the syllogism
in assessing which moods are syllogistic and which are not, the definition
cannot presuppose which moods are syllogistic; instead, we infer that
a mood is syllogistic from the fact that its conclusion follows from its
premises by virtue of a middle that is distinct from its extremes31. What
makes a syllogistic inference valid is that a subsumption is made under a
distributed middle term, whether that term occurs as a subject or predicate
or whether it occurs as a part of a subject or predicate. Given his revised
concept of a syllogistic inference and this test of validity, Buridan has a
theoretical basis on which he can determine the validity or invalidity of
syllogistic inferences containing oblique terms32. By contrast, Kilwardby’s

William Ockham, Summa logicae = Opera philosophica vol. 1, Ed. by P.
BOEHNER – G. GÁL – S. BROWN, Editiones Instituti Franciscani Universitatis St.
Bonaventurae, St. Bonaventure NY 1974, III.i.1: «Sciendum est tamen, quod definitio
communis omnibus praedictis est ista: Syllogismus est oratio in qua ex duabus
praemissis dispositis in modo et figura de necessitate sequitur conclusio. III.i.8: Et
regulae, quae dictae sunt prius, quando maior est universalis, sunt etiam servandae,
quando maior est singularis.»
John Buridan, Quaestiones in Analytica Priora, Ed. by H. HUBIEN, unpublished
typescript [http://www.logicmuseum.com/wiki/Authors/Buridan/ Quaestiones_
in_analytica_priora] I q.42: «Ideo non est necessarium quod omnis syllogismus sit
proprie et simpliciter in aliqua illarum trium figurarum. Nec illa expositio ‘syllogismi’
est bona, seu conveniens. Quia per definitionem ‘syllogismi’ nos examinamus et
concludimus qui modi sint debiti et qui modi non sint debiti. Ideo definitio ‘syllogismi’
non praesupponit qui modi sint debiti, sed hoc magis infertur ex eo quod videmus
conclusionem sequi ad praemissas per medium alienum ab extremitatibus. Quia
conversiones e aequipollentia, licet sint consequentiae necessariae, tamen non merentur
dici syllogismi, quia non concludunt per aliquod medium alium ab extremetitatibus.
Et tamen ista alietas medii tangebatur in definitione ‘syllogismi’ per istam clausulam
“necesse est aliud sequi”.»
Buridan, Quaestiones I, q. 42: «Concludo etiam finaliter. Dico quod omnis
terminus distributus in maiori propositione sub quo accipitur alius terminus in minori
propositione potest et debet dici ‘medium syllogisticum’, sive ille terminus distributus

account of oblique terms requires him to reduce inferences containing

them to one the Aristotelian syllogistic figures.
With regard to non-syllogistic inferences, Ockham made a notable
advance through his theory of exponible propositions. Ockham’s usage
of the expression ‘expositio’ in this context is not the traditional one that
denotes a kind of interpretation; rather, in his usage ‘exposition’ denotes
a kind of logical consequence. An exponible proposition is one which, by
virtue of its form, entails each of several exponents, whose conjunction
in turn entails that proposition33. The exponents follow logically from the
exponible proposition and it follows from them. In this respect the relation
between the exponents and the exponible proposition differs from that
between the premises and conclusion of a syllogism. For no syllogism,
stated in its most general form, is such that the truth of its premises follows
from the truth of its conclusion34.
Ockham states a rule for the validity of inferences containing exponible
propositions. An exponible conclusion follows from two premises, one at
least of which is exponible, if and only if every exponent of the conclusion
(or the conclusion itself) follows from some exponents of the premises (or
from an exponent of one premise together with the other premise)35. This
points to another difference between these inferences and syllogisms. An
inference such as Ockham describes may very well be from premises some
of which are redundant –something that never happens in a syllogism stated

sit totale praedicatum vel pars praedicati, vel totale subiectum maioris, vel pars
subiecti.» See John Buridan, Summulae de Dialectica: an annotated translation, with
a philosophical introduction by G. KLIMA (Yale University Press, New Haven 2001),
§§ 4.3.6 and 5.8.1-5.8.4.
Ockham, Summa logicae II.11: «Et est sciendum, quod quaelibet categorica,
ex qua sequuntur plures propositiones categoricae tamquam exponentes, hoc est
exprimentes quid ista propositio ex forma sua importat, potest dici propositio
aequivalens propositioni hypotheticae.»
P. THOM, The Syllogism, Philosophia, Munich 1981, p. 224, Corollary 1. (But
‘some members of A are true’ should read ‘some members of Q are false’.)
Ockham, Summa logicae III.i.65: «Pro quibus utendum est ista regula generali,
quod quandocumque quaelibet exponens conclusionis vel ipsa conclusio sequitur ex
aliquibus exponentibus praemissarum vel ex aliqua exponente unius praemissae et alia
praemissa, semper est syllogismus bonus, et aliter non. Et ideo ad videndum, an talis
discursus sit bonus vel non, oportet diligenter videre, quae sunt exponentes talium
praemissarum et conclusionis et quomodo se habent secundum consequentiam ad

in its most general form36. This feature of inferences containing exponible

propositions can be illustrated as follows. Suppose an inference’s conclusion
has two exponents, and that both of them follow from the first premise by
itself no matter what the second premise is. Ockham’s rule is satisfied, but
the inference has an irreducibly redundant premise. Surprisingly, however,
Ockham calls inferences that satify his rule syllogisms (‘the syllogism
is always good’). Perhaps he is using’syllogism’ here in a broad sense to
cover all valid inferences from more than one premise.
A proposition like ‘An ass is a non-man’ has as its exponents ‘An
ass is something’ and ‘An ass is not a man’37. Thus, ‘What is something,
but is not an X, is a non-X’ is true. This principle is reminiscent of
Kilwardby’s principle that what is, but is not an S, is a non-S. But, unlike
his predecessor, Ockham is able to put his principle to work in a systematic
way, thanks to his rule for inferences containing exponible propositions.
Thus, the inference ‘All men are reasoners, Brownie is a non-reasoner, so
Brownie is a non-man’ is valid because the conclusion has two exponents
(‘Brownie is something’ and ‘Brownie is not a man’), and the first of these
is an exponent of the minor, while the second follows syllogistically from
the major premise together with an exponent of the minor (‘Brownie is not
a reasoner’).
There are many types of exponible proposition in Ockham’s logic.
Reduplicative propositions are among them. A proposition ‘A insofar as
it is B is C’ has the exponents ‘A is C’, ‘A is B’, ‘Every B is C’ and ‘If
something is B it is C’ where this last proposition is understood to be the
sort of conditional that links a lower to a higher term in the same category or
two convertible terms38. Such propositions therefore, like those containing

THOM, The Syllogism, p. 193, Corollary.
Ockham, Summa logicae II.12: «[…] propositiones, in quibus ponuntur termini
negativi, privativi et infiniti, sunt aequivalentes propositionibus hypotheticis […] Unde
quaelibet propositio, in qua ponitur terminus infinitus, habet duas exponentes, unam
affirmativam, in qua iste terminus ‘aliquid’ in singulari vel in plurali vel aliquis alius
terminus aequipollens tali subiicitur vel praedicatur. Unde ista ‘Asinus est non-homo’
aequivalet ista ‘Asinus est aliquid, et asinus non est homo’.»
Ockham, Summa logicae II.16: «Si fiat reduplicato gratia concomitantiae tunc
ad veritatem ipsius requiruntur quatuor propositiones tamquam exponentes eam:
una in qua praedicatum principale vere praedicetur de subiecto principale, alia in
qua illud super quod cadit reduplicatio praedicetur de subiecto principale, tertio in
qua praedicatum principale praedicetur de illo super cadit reduplicatio universaliter,
quarta erit una conditionalis vera ab illo super quod cadit reduplicatio ad praedicatum

infinite terms, are subject to Ockham’s general rule for inferences involving
exponible propositions. For example the inference ‘Every A insofar as it is
B is C, some D is an A, so some D insofar as it is a B is a C’ can be shown
to be valid by Ockham’s analysis into exponents. The analysis will show
that every exponent of the conclusion either is an exponent of the major
or else follows from the minor together with an exponent of the major.
The exponents of the conclusion are (1) ‘Some D is a C’, (2) ‘Some D is
a B’, (3) ‘Every B is a C’, and (in a suitably strong sense) ‘If something
is a B it is a C’. Now, (1) follows syllogistically from the minor (‘Some
D is an A’) together with an exponent of the major (‘Every A is a C’). (2)
follows syllogistically from the minor together with an exponent of the
major (‘Every A is a B’). (3) and (4) are exponents of the major39. So here
again Ockham has a way of reducing a class of inferences to formally
valid inferences some of which are not syllogistic since they are inferences
between an exponible proposition and its exponents.


In the territory we have surveyed, Kilwardby’s formalizing ambitions

were considerable –even heroic. He shared Aristotle’s faith in the possibility
of reducing inferences with infinite or oblique terms, and with reduplicative
propositions, to the standard syllogistic forms without developing new
logics to deal specifically with those types of term and proposition.
The reductions he envisaged fall in the area of the application of a
formal logical theory. In this non-formal part of logic, the general nature

principale illo modo quo ab inferiori ad superius dicitur esse bona consequentia et
quomodo dicitur quod ex uno convertibilium sequitur reliquum.»
Ockham, Summa logicae III.i.65: «Sicut ad videndum an iste discursus sit
bonus ‘Omnis homo inquantum homo est risibilis, animal est homo, igitur animal
inquantum hom est risibilis’ oportet videre exponentes maioris et conclusionis. Unde
exponentes maioris sunt istae ‘Omnis homo est homo’, ‘Omnis homo est risibilis’,
‘Si aliquid est homo ipsum est risibile’. Exponentes conclusionis sunt ipsae ‘Animal
est homo’, ‘Animal est risibile’, ‘Omnis homo est est risibilis’, ‘Si aliquid est homo
ipsum est risibile’. Nunc autem sequitur ‘Omnis homo est homo igitur animal est
homo’. Similiter sequitur ‘Omnis homo est risibilis igitur animal est risibile’. Aliae
duae exponentes conclusionis sunt exponentes maioris. Ex quo patet quod omnes
exponentes conclusionis sequuntur ex antecedente et ita discursus est bonus.»

of the problem to be solved is this: how can certain pre-theoretic terms,

propositions and inferences be adequately represented in a given logical
theory. Solutions are arrived at through an art of interpretation not through
techniques of deduction.
Kilwardby’s successors no longer wanted to reduce inferences with
infinite or oblique terms to Aristotelian syllogisms, but they succeeded in
absorbing into the domain of formal logic some of what Kilwardby treated
as a matter of interpretation. That was a gain. At the same time, the art of
uncovering the formal deductions that can be found in informal discourse
no longer played a major role in their presentations of logic. And this was a
loss, because this is an art that every student of logic has to learn.



The notion of ‘forma’ plays a variety of roles in medieval logic. An

example of the use of ‘forma’ is the so called forma-materia device, as
a way to analyse and establish the truth value of sophisma-sentences1.
Furthermore, the concept of ‘forma’ plays an important part in medieval
discussions on inference: an inference is often said to be valid in virtue of
its ‘forma’.
Besides having a use in logic, the concept of form of course also features
prominently in the philosophical outlook of the Formalist movement2.
One of its representatives, the Spanish Scotist, Petrus Thomae3 devoted
a considerable effort on a thorough analysis of Scotistic terminology, and
especially the concept of ‘distinctio formalis’. He wrote a short treatise
called De unitate minori4, in which he analyses the concept of unity which
is less than numerical. In this short tract the author discusses numerous
arguments brought forward by others on specific philosophical issues, and
in his own replies to their accounts, he frequently refers to the ‘forma’ of
an argument.
In this contribution I will delve into the idea of ‘formalness’ as found
in the work of the medieval logician John Wyclif. It is quite remarkable
that the ‘realist’ Wyclif wants nothing to do with the so-called consequentia

Associate Professor, Faculty of Arts and Social Sciences, Maastricht University,
P.O. Box 616, 6200 MD Maastricht NL. Email: joke.spruyt@maastrichtuniversity.nl
J. SPRUYT, «The Forma-Materia Device in Thirteenth-Century Logic and
Semantics», Vivarium, 41 (2003) 1-46.
I.e., the followers of Scotus, who adhered to the formal distinction.
Petrus Thomae was probably born in Catalonia, Spain (ca. 1280?), and it is likely
that he entered the Franciscan Order in the province of Aragon in Catalonia about
1300. For biographical details of Petrus Thomae, see G. G. BRIDGES, O.F.M., Identity
and Distinction in Petrus Thomae, O.F.M., St. Bonaventure − Louvain − Paderborn
1959, pp. 1-2 (The Franciscan Institute Publications, Philosophy Series, no. 14).
E. P. BOS, The Tract De unitate minori of Petrus Thome, Peeters, Louvain 2002
(Recherches de théologie et philosophie médiévales, Biblioteca, 5).

materialis. Considering his adamant rejection of this particular, quite

commonly accepted type of consequentia as a way to handle awkward
kinds of inferences, we need to understand his conception of formality. It
is interesting to briefly consider the treatise by Petrus Thomae as a prelude
to Wyclif, because the former’s perspective on the forma of an argument
resonates in Wyclif’s account.
As I hope to show, Wyclif’s notion of inference hinges on two
separate foundations of logic. While his accounts of conditional and
inferential expressions is expressed in terms of the logical truth-conditions
of an inference, ultimately his analysis of the formalness or validity of
inferences is based upon an ontological starting point, i.e., the ontology of
the propositions. I shall first discuss his conception of inference featuring
in his account of consequentia and that of conditional expressions. The
most significant aspect of the latter is Wyclif’s critique of the distinction
between consequentia formalis and consequentia materialis. In conclusion
I shall briefly turn to the Logica Morelli, a work that was written later
than John Wyclif’s, and which seems to return to an earlier conception of
For starters, let us first review the ways in which formality had come
up in Syncategoremata treatises of the thirteenth century.

1. ‘Matter’ and ‘form’ in the context of inference (consequentia)

1.1. The thirteenth century

For some time now, research has been done in the development of the
distinction between formal and material consequences in logic5. While in

See, among others, F. SCHUPP, Logical Problems of the Medieval Theory of
Consequences: with the edition of the Liber Consequentiarum, Bibliopolis, Napoli
1988 (History of Logic, 6); C. DUTILH NOVAES, «Form and Matter in Later Latin
Medieval Logic: the cases of suppositio and consequentia», in Journal of the History
of Philosophy, 50 (2012) 339-364; S. READ, «Formal and Material Consequence,
Disjunctive Syllogism and Gamma», in K. JACOBI (ed.) Argumentationstheorie.
Scholastische Forschungen zu den logischen und semantischen Regeln korrekten
Folgerns, E. J. Brill, Leiden – Köln – New York 1993, pp. 233-262 (Studien und
Texte zur Geistesgeschichte des Mittelalters, 38); S. READ, «Formal and Material
Consequence», Journal of Philosophical Logic, 23 (1994) 247-265.

the thirteenth century we already see indications of this type of distinction,

the expressions consequentia formalis and consequentia materialis only
became common lore in the fourteenth century.
In the thirteenth century, the nature of inference came up in Sophistaria
treatises and Syncategoremata. The syncategorematic term ‘si’ was taken
as a starting point for discussions about the nature of inferentiality. Most
authors explained the meaning (consignificatio) of this expression in
terms of a kind of causality. However, the nature of the causality involved
was interpreted in different ways. Sometimes the authors highlighted the
inferential function of ‘si’, whereas others were more inclined to include
some kind of ontological causality in it as well.
One of the most problematic issues that worried the Syncategoremata
authors was how they should deal with the standard description of a good
inference, which implies that an inference is in order even if its antecedent
is false, no matter what the consequent. And this in turn suggests that from
an impossible antecedent anything at all follows (ex impossibili sequitur
quidlibet). Thirteenth-century authors had two ways to cope with this
outcome: one way was to qualify the notion of impossibility in such a way
that in order to function as antecedent of an inference at all, the impossible
in question must have some bearing on what is expressed in the consequent
–in which case the rule was rejected–; others accepted the principle that
from an impossible anything follows, but that it does not apply in every
kind of inference6.
The former type of solution was suggested by authors like Peter of
Spain, who rejected the ‘ex impossibili’ rule. There were others, such as
Nicholas of Paris, who accepted it, but only in what were called ‘non-
natural’ consequences. In general the difficulty of how to deal with ‘ex
impossibili’ consequences was brought on by the authors’ commitment (to
a greater or lesser degree, depending on the authors in question) to a system
of logic that was mainly based upon the topical relationships between the
terms featuring in antecedent and consequent of an conditional expression
or inference.
The understanding that there could be such a thing as arguing from
anything at all from an impossible antecedent, and, conversely, that a

For a detailed discussion, see J. SPRUYT, «Thirteenth-Century Positions on the
Rule ‘Ex impossibili sequitur quidlibet’», in K. JACOBI (ed.), Argumentationstheorie,
op. cit., pp. 161-193.

necessary proposition could be argued for starting from any antecedent

whatsoever, later on led logicians to identify an exceptional class of
consequences, which became labelled as consequentia materialis. This
strategy was adopted by people like Walter Burleigh and William of
Ockham: the consequentia formalis was identified as a valid inference,
the validity of which was based upon the semantic relationships between
the terms featuring in is antecedent and consequent, whereas in material
consequences such a relationship is absent. Thus they were able to
accommodate the paradoxical cases of inference, i.e. inferences from an
impossible antecedent and inferences to a necessary consequent, which
were presumed to be valid as well, but not in the same way as consequentie
Before we proceed with Wyclif’s position on valid inference, we shall
quickly consider some of Petrus Thomae’s remarks on ‘forma’.

1.2. Petrus Thomae on ‘forma’

The expression consequentia formalis was not the only one used
in connection with the inferential relationships. As mentioned above,
Petrus Thomae speaks about the forma to highlight a certain feature
of an argument. In De unitate minori, the author puts forward a list of
questions, which he deals with in the usual way by presenting arguments
pro and con a specific position. (Below I shall present two examples
of how Petrus presents his case.) In order to recognise the significance
of Petrus’s accounts in this connection (and Wyclif’s too, as we shall
see later on), it is important that we first grasp what the notion of
inferentiality involves. From what the author says we can gather that
he takes inference to be a kind of true production of a conclusion from
a (some) premise(s). So the notion of ‘true’ is a kernel ingredient, and
requires serious attention.
Let us now turn to some of his examples. (1) In qu. 5, art. 3, the following
syllogism is brought forward: whatever is truly real, is truly something

For a general overview of the developments, see C. DUTILH NOVAES, «Medieval
Theories of Consequence», in E. N. ZALTA (ed.), The Stanford Encyclopedia of
Philosophy (Summer 2012 Edition), URL = <http://plato.stanford.edu/archives/

positive; now a privation and a negation are truly something real; therefore
they are truly something positive8. After having presented this argument,
Petrus responds to it by making a distinction between two ways of taking
‘truly real’ (vere reale): in the first sense it is something positive, and in
the second sense it is something that follows from something real, whether
or not this thing itself is positive9. Petrus’s response to this argument is
preceded by the phrase ‘ad formam’: as to the form, Petrus explains, it
should be noticed that the major is true with regard to ‘truly real’ as taken
in the first sense, but in this case the minor is not assumed under the major
premise10. Hence he disagrees with the conclusion.
In the example just presented, the phrase ‘following from something
real’ appears to be precisely the way to describe the relationship of
inference. This particular conception in terms of ‘truly following’ is
further explained by means of the generic concept of true production, as
we can see in the second example. (2) In qu. 8, art. 4, we find the following
argument: according to the Philosopher, Met. VII, c. 13, all that comes
about, comes about by something similar, to which the Commentator notes,
«Everything that is generated is generated by something similar in name
and definition»; therefore everything that is generated is generated by a
univocal generation, and consequently no production is equivocal11. Later
on Petrus replies to this syllogism: ‘ad formam’, he contends, the sayings
of both Aristotle and the Commentator, should be understood as applying

Petrus Thomae, De Unitate Minori, qu. 5, art. 3, ed. BOS, p. 42.985-986: «Quod
vere est reale, vere est positivum; sed privatio et negatio sunt vere aliquid reale; ergo
sunt vere aliquid positivum.»
Ibid., p. 42.988-990: «Respondeo: ‘vere reale’ aliquid potest dici, vel quia
rem positivam dicit, et sic solum ens positivum est vere reale; vel quia veram rem
sequitur, et sic quicquid sequitur veram rem aliquam potest dici vere reale, sive istud
sit positivum, sive non.» For something non-positive, see below, p. 162.
Ibid., p. 43.995-996: «Ad formam: maior est vera de vere reali primo modo
accepto; sed tunc minor non sumitur sub maiori.»
Ibid., qu. 8, art. 4, ed. BOS, p. 69.1876-1881: «Secundum Philosophum*,
septimo Metaphysicae, capitulo <decimo> tertio, omne quod fit, fit a conveniente,
ubi Commentator, commento trecesimo, ‘omne quod generatur, generatur convenienti
nomine et ratione’; ergo omne quod generatur, generatur generatione univoca, et per
consequens nulla productio est equivoca.» *Cf. Metaphysica VII, 13, 1038b23-27:
«Amplius autem impossibile et inconveniens hoc esse substantiam, si est ex aliquibus,
non ex substantiis esse, nec ex eo quod hoc aliquid sed ex quali. Prius enim erit non
substantia et quale substantia et hoc, quod est impossibile.»

to natural generation, which proceeds from something that is specifically

the same to something that is specifically the same12.
In both cases just mentioned (and many other examples can be found
in this tract), the use of the phrase ‘ad formam’ applies, it would seem to
me, to the shape or schematic form of the argument, i.e. the structure of the
syllogism. We shall now consider Wyclif’s account.

2. John Wyclif (1320-1384)13 on inference and conditionals

In Wyclif’s logic, the complicated issues surrounding the notion of

‘formalness’ in medieval conceptions of logic takes an interesting turn.
Wyclif is known for his tendency to reify all kinds of objects of thought.
The ontological perspective that is so predominant in different parts of his
logic also shines through in his discussion of inference.
Before we turn to the texts, a few preliminary remarks should be made.
The premises of arguments of course are propositions, and the domain
of propositions featuring in arguments is tied up with different kinds of
certainty. In Wyclif’s account, premises are understood as expressing
some state of affairs14. When it comes to the relationship of inference
between propositions, the ontological status of what is expressed by these
propositions acting as premises is decisive.
Wyclif’s conception of inference is brought forward inter alia two
specific sections of his Logic, viz. in the chapter on consequentiae and the
one on conditionals.

Petrus Thomae, De unitate minori, qu. 8, art. 4, ed. BOS, p. 71.1951-1953: «Ad
formam: dictum Philosophi et Commentatoris debet intelligi de generatione naturali et
per se, et que procedit ab eodem specie et ad idem specie.»
Iohannes Wyclif, Tractatus de logica, Ed. by M. H. DZIEWICKI (3 vols.), Trübner
for the Wyclif Society, London 1893-1899 (Repr. Johnson Reprint Corporation, New
York – London; Minerva, Frankfurt am Main; hereafter cited as Iohannis Wyclif,
L. CESALLI, «Le ‘pan-propositionalisme’ de Jean Wyclif», Vivarium, 48 (2005)

2.1. Consequentiae

In the thirteenth chapter of his Logic, Wyclif brings speaks about

the good and formal consequence (consequentia bona et formalis)15.
He gives one standard definition only of the expression ‘consequentia’;
a consequentia is a kind of relationship between an antecedent and a
consequent, combined with an inferential expression, or an aggregate of
an antecedent and a consequent, combined with an inferential expression16.
Inferential expressions include words such as ‘ergo’, ‘igitur’, ‘ideo’, and
‘quare’17. Unlike the author of Logica Morelli (to which we shall turn
later), Wyclif does not give a separate definition of a good and formal
consequentia, but immediately presents twenty-two rules to identify them.
On the face of it these rules seem to run parallel with the standard set
presented by Ralph Strode, the anonymous author of Logica Morelli and
others. It is useful to take a closer look at some of these rules, in order to
find out what role the notion of formality plays in them.
The first rule Wyclif presents seems straightforward enough: a
consequence is good and formal in which the consequent is formally
understood in the antecedent. On the other hand, this description is somewhat
circular, because of the occurrence of ‘formalness’ in definiendum and
definiens. By way of explanation, Wyclif simply gives a few examples,
one of which runs, ‘Peter is charitable; therefore, he is virtuous.’ The
reason why this consequence fits the bill is because the consequent ‘he is
virtuous’ is understood in the antecedent18. Here it seems obvious to take
the expression ‘formal’ as applying to a semantic relationship between the
significata of the expressions used, such that being charitable implies being
virtuous because being virtuous is part of what it means to be charitable.

Iohannes Wyclif, Tractatus de logica I, ed. DZIEWICKI, pp. 42ff.
Ibid., cap. 13, p. 43.2-5: «Consequencia est quedam habitudo inter antecedens
et consequens, cum nota consequencie* [*ed. consequente]. Vel: consequencia est
quoddam aggregatum ex antecedente et consequente cum nota consequencie.»
Ibid., p. 43.12-13: «Et sunt note consequencie ergo, igitur* [*ed. etc.], ideo et
quare* [*ed. quia].»
Ibid., p. 43.6-12: «Ad cognoscendum que sunt consequencie bone et formales
dantur 22 regule. Quarum prima est hec: Quelibet consequencia est bona et formalis in
qua consequens formaliter intelligitur in antecedente; ut sic argumentando: Petrus est
caritativus; ergo ipse est virtuosus, quia hoc consequens, ipse est virtuosus, intelligitur
in hoc antecedente, Petrus est caritativus.»

In the second and third rules, Wyclif’s take on the formal nature of an
inference is slightly different. The second rule runs: any consequence is
good and formal when the contradictory of the antecedent follows from the
contradictory of the consequent. The following example is given: ‘A man
is running; therefore an animal is running’, which, the author explains, is
a good and formal consequence because it formally follows, ‘No animal is
running; therefore no one is running’19. The corollary of this rule runs: any
consequence is good and formal when the contradictory of the consequent
formally conflicts with the antecedent. Hence the following is in order,
‘He understands everything; therefore he understands something’, because
the proposition ‘He understands nothing’, i.e. the contradictory of the
antecedent, conflicts with ‘He understands everything’20. In these examples
another condition for the formal validity of an inference is introduced, i.e.,
the logical relationship between the signa quantitatis contained in the
propositions at issue.
In the fourth rule Wyclif returns to the semantic relationships between
terms featuring in the propositions involved to explain the validity of a
consequence: Whatever is antecedent to the antecedent, is antecedent to
the consequent of the same antecedent. This rule vouches for the validity of
an inference like, ‘Some man is sensing; therefore some body is sensing’,
given the validity of the inference, ‘Some animal is sensing; therefore some
body is sensing’. This argument is acceptable because the proposition,
‘Some man is sensing’ can be antecedent to the first consequent, viz. ‘Some
animal is sensing’, and therefore it can be antecedent to its consequent21.

Ibid., p. 43.14-18: «Secunda regula est ista: Quelibet consequencia est bona et
formalis quando ex contradictorio consequentis sequitur contradictorium antecedentis,
ut: homo currit; ergo animal currit, quia sequitur formaliter: nullum animal currit;
ergo nemo currit.»
Ibid., p. 43.19-23: «Tercia regula est ista: Quelibet consequencia est bona
et formalis ubi contradictorium consequentis formaliter repugnat antecedenti, ut:
omnia intelligit, igitur aliquid intelligit, quia ista repugnant: nihil intelligit et omnia
Ibid., p. 43.24-33: «Quarta regula est ista: Quicquid antecedit ad antecedens
antecedit ad consequens. Hoc est sic intelligendum quod quelibet proposicio que
antecedit ad antecedens potest antecedere ad consequens illius antecedentis, ut hic:
aliquod animal sentit; ergo aliquod corpus sentit; et sic sequitur: aliquis homo sentit;
ergo aliquod corpus sentit, quia iste propositio, aliquis homo sentit, potest antecedere ad
hoc consequens prioris consequencie, scilicet aliquod animal sentit, et per consequens
potest antecedere ad hoc consequens, aliquod corpus sentit.»

The antecedence in question here is of course tied up with the semantic

connection between the terms ‘man’, ‘animal’ and ‘body’.
The fifth rule Wyclif mentions is an account of the good formal
inference a primo ad ultimum. This rule is to the effect that if you argue in
a series of consequences from the first to the final proposition, and if in that
sequence of inferences from the first to the final proposition the inferences
are good and formal and not varied, then the entire inference from the first
to the final proposition is good as well. The author also mentions a possible
objection to the rule, by adducing the sophisma-sentence ‘No time is;
therefore some time is’, which is argued for on the basis of the inferences
in between, ‘No time is; therefore it is not day’, ‘It is not day; therefore
some time is’; he counters this objection by explaining that this is not a
formally good example of an argument a primo ad ultimum, because the
inferences in between are varied22. He does not in this particular example
explicitly pass any judgement on the formality and goodness of the
inference. Instead he seems to be taking the absence of varied inferences
in between as a condition that is separate from the formality and goodness
of the inferences at issue.
In his sixth and seventh rules, Wyclif, explicitly refers to a ‘formal
and good’ inference. The sixth rule simply states that from a universal
to its subaltern particular an inference obtains (consequentia tenet), both
affirmatively and negatively, like in, ‘Every virtue is good; therefore some
virtue is good’, and ‘No charity is a vice; therefore some charity is not
a vice’. The seventh rule mentions one condition for producing a ‘good’
consequence: to argue from a particular to its indefinite is good, both
affirmatively and negatively, like in the example, ‘Some charity is a virtue;

Ibid., p. 44.10-17: «Alia regula est ista: Quando argumentatur a primo ad
ultimum, ubi omnes conseqencie intermedie sunt bone et formales et non variate,
est consequencia bona. Et argumentatur a primo ad ultimum quando consequens
prioris consequencie est antecedens posterioris consequencie, ut sic argumentando:
Ordinata dileccio est, ergo caritas est; caritas est, ergo virtus est; virtus est; ergo
bonitas est. Sed argumentatur contra istam regulam: nullum tempus est, ergo dies
non est; dies non est, ergo* [*ed. et] aliquod tempus est. A primo ad ultimum:
nullum tempus est; ergo aliquod tempus est. Ista consequencia non valet, et tamen
argumentatur per regulam predictam. Ergo regula illa est falsa. Dicendum est quod
consequencie intermedie sunt variate, quia plus ponitur in antecedente secunde
consequencie quam fuit consequens prime consequencie. Ideo non argumentatur per

therefore charity is a virtue’, and in the negative, ‘Some charity is not a

vice; therefore charity is not a vice’23.
A final example in Wyclif’s exposé of rules is interesting here to take a
look at, viz.: from the false follows a truth, but from a truth never formally
follows something false. This rule is tied up with the famous principle
‘ex impossibili sequitur quidlibet’, which Wyclif presents to us with the
little verse: ‘From falsehoods a truth, from a truth nothing but a truth’. To
illustrate this rule the author gives a few examples, such as, ‘Man is an
ass; therefore God is’, and ‘No god is; therefore no world is’, or from the
premise ‘No god is’, you could even conclude ‘A world is’, or anything
you like. These inferences are all valid in virtue of the ‘ex impossibili’ rule,
he says24. It is noteworthy that this rule is explicitly explained in terms of
formaliter sequi, but as yet it only features in a negative way: arguing from
something true to something false is never formally valid.
What we has just been brought forward is not all Wyclif has to say
on the issue of reasoning from impossible assumptions. In Logica II
arguments ex impossibili implicitly come up for discussion again, this time
in connection with an exposition of conditional sentences.

2.2. Conditional sentences and their truth conditions

Before we discuss his views on validity in this connection, let us first

consider what Wyclif has to say about conditional sentences in general.

Ibid., p. 44.18-27: «Alia regula est ista: ab universali ad suam particularem
subalternam, tam affirmative quam negative, tenet consequencia. Affirmative, ut hic:
quelibet virtus est bona; ergo aliqua virtus est bona; negative, ut hic: nulla caritas
est vicium; ergo aliqua caritas non est vicium. Alia regula est: A particulari ad suam
infinitam, tam affirmative quam negative, est consequencia bona. Ut: quedam caritas
est virtus; ergo caritas est virtus. Negative, ut sic: quedam cartitas non est vicium;
ergo aliqua caritas non est vicium.»
Ibid., p. 45.4-14: «Alia regula est ista: Ex falso sequitur verum, sed numquam
ex vero sequitur falsum formaliter, versu: Ex falsis verum, ex vero nil nisi verum. Ut
bene sequitur: homo est asinus; ergo Deus est. […] Sed ex falso sequitur verum. Nam
ista regula est quod ex impossibili sequitur quodlibet (vel sequi potest); ut sequitur:
nullus deus est; ergo nullus mundus est, et similiter quod mundus est, vel quicquid
volueris concludere.»

2.2.1. General remarks about conditionals

Conditional sentences are comparable to syllogisms. (Incidentally, for

Aristotle the syllogism is by definition an implication.25) Yet they do have a
few defining characteristics of their own. In his words, «Every hypothetical
proposition subordinated under a conditioned act26, is a conditional, like
the following: ‘If you are a man, you are an animal’, as are the others,
which are commonly called consequences»27.
It is interesting how Wyclif distinguishes conditionals from
consequences. A consequence he defines as «a relationship of truth between
the consequent and the antecedent, which is included in the significate of
every true conditional». Hence the expressions ‘ergo’, ‘ideo’, and ‘igitur’,
and the like, «by connoting causation, posit a consequence that is erroneous
in form nor matter»28.
Here we see again that inference is defined in terms of ‘causation’.
The thirteenth-century author of a Syncategoremata treatise, Johannes

For Aristotle’s view of the syllogism, see L. M. DE RIJK, Aristotle. Semantics and
Ontology. Vol. I, General Introduction. The Works on Logic; Vol II: The Metaphysics.
Semantics in Aristotle’s Strategy of Argument, E. J. Brill, Leiden – Köln – New York
2002; Vol. I, pp. 568-572 (Philosophia Antiqua, 91).
The expression «sub actu condictionato» should be understood in an analogous
way as in the pair «actus exercitus/actus significatus». This distinction is often used
to explain the difference between an expression that signifies something, like the
noun ‘negatio’, and an expression that has a specific function, such as ‘non’, which
is not exactly said to signify, but instead to carry out a negation. In this case then the
expression «actus condicionatus» refers to the function of a consecutive expression,
such as ‘si’. See G. NUCHELMANS, «The Distinction ‘Actus Exercitus / Actus
Significatus’ in Medieval Semantics», in N. KRETZMANN (ed.), Meaning and Inference
in Medieval Philosophy: Studies in Memory of Jan Pinborg, Kluwer Academic
Publishers, Dordrecht 1988.
Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 182.3-7: «Et primo
supponatur omnem ypotheticam subordinatam actu condicionato, esse condicionalem; ut
est talis: Si tu es homo, tu es animal, et cetere que vulgariter vocantur consequencie […].»
Ibid., p. 182.6-15: «[…] quamvis secundum vim vocis habitudo veritatis
posterioris ad priorem sit consequencia (qualis est in significato cuiuscumque
condicionalis vere) inter antecedens naturalis prius et suum naturaliter consequens.
Et hinc patet quod differunt iste note consequenciarum, si, ergo, igitur, et forte omnia
sincategorematica* [*cor. ex ed. synkategorica]; que nos ponimus synonima. Nam iste
coniuncciones ergo, ideo, igitur, et si que consimiles, connotando causacionen, ponunt
consequenciam nec in materia nec in forma peccantem.»

Pagus clearly explains the meaning of the expression ‘cause’ in the

description of inference: it pertains to the relationship of entailment
between antecedent and consequent, and not to being29. Wyclif too hints
at what kind of entailment he has in mind here. He says that conditionals
do not posit a consequence that is erroneous in form nor matter (unlike
consequences, that is), because the following is necessary: ‘If you
are an ass, you can bray’30. In this particular example we can identify
the relationship between antecedent and consequent with the kind of
causation we have spoken of above, when we looked at the expression
‘truly real’ as explained by Petrus Thomae. From the nature of one thing,
i.e., something’s being an ass, truly follows its having the ability to bray.
What follows from one thing, your ability to bray, is itself something
positive. An example of something negative following from the same
antecedent could be what is expressed in a consequent like, ‘Therefore,
you are irrational.’

2.2.2. On different kinds of entailment

In the previous paragraph, Wyclif only indirectly shows us how he

conceives of inferentiality. In what comes next, he is a bit more explicit.
The expression si, the author tells us, sometimes stands for ‘because’,
and sometimes expresses an absolutely necessary conditional truth. In the
latter case he distinguishes between a conditionalised truth ut nunc, and a
conditionalised truth obtaining for all eternal time31.

Cf. J. LE PAGE, in H. A. G. BRAAKHUIS, De 13de Eeuwse Tractaten over
Syncategorematische Termen. I: Inleidende Studie. II: Uitgave van Nicolaas van Parijs’
Sincategoreumata (diss.), Nijmegen 1979; for a commentary see SPRUYT, «Thirteenth-
Century Positions».
Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 182.15-18:
«Sed non sic condicionalis, cum hoc sit necessarium si tu es asinus, tu es rudibilis. Et
sic differunt note consequenciarum secundum genera causandi.»
Ibid., p. 182.22-30: «Sed [...] notandum quod si quandocumque* [*ed.
quantocunque] ponitur pro quia, quandocumue simpliciter dicit necessariam veritatem
condicionatam. Et hoc dupliciter variatur: Vel sic quod sit veritas condicionata* [*ed.
condicionis] tenens ut nunc, ut hic: si ego sum Rome, falsum est verum. Nam veritati
eterne repugnat quod nunc sim Rome, nisi quodlibet sequatur. [...] Vel secundo quod sit
veritas condicionata tenens pro omni tempore eterno, ut si deus est, ipse vult mundum
esse. Vel pro tempore eterno a parte post, ut si ego non sum, nichil fuit.»

This distinction between different kinds of conditionals features in the

logic of the thirteenth and fourteenth centuries too. In his discussion of
the sophisma-sentence ‘Nichil est verum nisi in hoc instanti’, Matthew of
Orléans, for example, distinguishes the consequentia ut nunc, which he
explains as applying to a certain time, as opposed to something’s obtaining
simpliciter, that is to say applying for any time whatsoever. Taken in the
first way, the sophisma-sentence is false, because in that case it denotes
the consequence for any time whatsoever32.
Wyclif’s deals with the conditional simpliciter in a similar manner as
Matthew of Orléans does, but has his own way of describing the nature
of these different kinds of conditionals. As an example of the conditional
ut nunc, Wyclif presents the proposition, ‘If I am in Rome, the false is
true’; this conditional is true. Its truth is explained in a remarkable way
by Wyclif: it is a truth of a condition obtaining ut nunc, because, as he
says, that I should now be in Rome is in conflict with eternal truth, unless
anything follows33. The reason for this being the case is that my now not
being in Rome is a true, negative state of affairs.
Although as such Wyclif’s reference to eternal truth is not surprising,
given his views on proposition, it seems a bit out of place in this connection.
What kind of truth is he referring to here? Or is he just talking about the
truth-condition itself as an eternal truth? The label ut nunc suggests that we
are dealing with contingency. Yet at the same time necessity is involved:
my now not being in Rome is a true negative state of affairs, which is
undeniable, and cannot be explained away.
With regard to the example just mentioned, then, two claims are made:
first there is such a thing as a truth depending upon a condition obtaining

Matthew of Orléans, Sophistaria sive distinctiones sophismatum, Edited with
an Introduction and Notes by J. SPRUYT, E. J. Brill, Leiden − Boston − Köln 2001, III,
182-183, pp. 282-283 (Studien und texte zur Geisteschichte des Mittelalters, 74): «Si
autem teneatur [sc.‘nisi’] consecutive, adhuc est duplex, quia hec dictio ‘nisi’ potest
denotare consequentiam ut nunc vel consequentiam simpliciter. Si consequentiam
simpliciter, tunc debeat quod falsa est, quia tunc denotat consequentiam pro quolibet
tempore. Item. In consequentia simpliciter exigitur localis habitudo et illa duo non
reperiuntur in hac ‘quicquid est verum est verum in hoc instanti’. Si autem denotat
consequentiam ut nunc, tunc est vera, et est sensus ‘nichil est verum etc.’, idest si
aliquid est verum pro tempore quod est nunc, est verum in hoc instanti. Et sic non
denotatur consequentia respectu cuiuslibet temporis, sed respectu temporis quod est
The text is quoted in full, see above, n. 31.

now, and second, from something which is impossible, and therefore false,
viz. that I should now be in Rome, anything follows, among other things
that the false is true. (Incidentally, in this particular example it is not any
kind of impossibility he is talking about, but simply the impossibility of
something’s being the case at the very moment it is not the case.)
The second type of conditional he talks about in this connection is a kind
of conditionalised truth obtaining for all eternal time, e.g. ‘If God is, he wills
the world to be’, or, as he explains, for an eternal time a parte post, e.g. ‘If I am
not, nothing has been’34. These two are labelled as consquentiae per accidens.
After making yet another distinction between different kinds of
consequentiae per accidens and giving some examples of conditionals
that include a condition under a specific form of the verb (subjunctive,
imperative) and of propositions that imply a consequence35, Wyclif presents
a division between three main types of truth of a conditional.

2.2.3. On the truth of conditionals

A conditional can be true in three different ways, namely simply and

absolutely (simpliciter et absolute), per accidens, or ut nunc. The three
kinds of true conditional have in common not only that they primarily
signify truth36, but in general a true conditional is such that it is impossible
that the truth is conditionally assumed unless its truth is conditionally
deduced, and the other way round. And this is what is usually said that
a consequence is true when it is impossible for the antecedent of this
expression to thus primarily signify it is true unless its consequent is true37.

See for the text, above, n. 31.
Cf. Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, pp. 182.30-
For an analysis of Wyclif’s use of primarie significare, see CESALLI «Le ‘pan-
propositionalisme’», pp. 147-150.
Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 183.21-30:
«Ex istis patet quod tripliciter condicionalis est vera: primo simpliciter et absolute,
secundo per accidens, et tertio ut nunc. Et omnis condicionalis vera convenit in hoc
cum qualibet, non solum quod primarie significat veritatem, sed quod impossibile est
veritatem condicionaliter assumptam esse nisi sit veritas eius condicionaliter deducta,
et econverso* [*ed. econtra]. Et hoc est quod principaliter solet dici quod signanter
tunc est consequencia bona quando impossibile est antecedens illius sic primarie
significantis esse verum nisi suum consequens sit verum.»

The truth conditions of a conditional expression are connected with

the notions of impossible and necessary. The latter is said in three ways,
namely simpliciter, per accidens and secundum quid: a simply necessary
truth is such that in absolutely no way whatsoever it cannot not be the
case, a truth that is necessary per accidens is such that it conflicts with the
eternal order that for some given part of time what is expressed by it should
not be the case, and while a truth that is necessary secundum quid is only
the case temporarily, nevertheless it conflicts with the eternal order that it
should not be the case38. Correspondingly impossible is distinguished into
per accidens and secundum quid: everything that is necessary per accidens
or secundum quid can not-be, and likewise everything impossible per
accidens or secundum quid can be39.

2.2.4. On the formality issue as such

In the next part of the chapter Wyclif identifies a number of rules that
follow from the considerations he has just presented. What is remarkable
in his account in general here is how the metaphysics of necessity and
impossibility enter the scene, as we shall see below. The outcome of his
account is also worthwhile to look at. Firstly, it clearly turns out that
with regard to the formality issue, Wyclif disagrees with the well-known
distinction between material and formal consequence. Secondly, he alludes
to the failure of relevance issue, that was to become a domain of interest
later on in the history of logic. And finally, some light is shed on precisely
what he takes a conditional sentence to be, that is, in terms of what exactly
it is supposed to posit. Let us first look at the rules he presents.

This account of necessity is in accordance with Aristotle’s famous rule of De
interpret. 9, 19a23-24: «That what is, is, when it is, and what is not, is not, when it is
not, is necessary.» For a detailed analysis, see DE RIJK, Aristotle, Vol. II, pp. 286-288.
Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 183.31-41:
«Et ita tripliciter dicitur impossibile et necessarium, scilicet simpliciter, per accidens
et secundum quid. Simpliciter necessarium est quod de nulla potencia potest non esse.
Necessarium per accidens est veritas quam ordinacioni eterne repugnat pro aliqua parte
dati temporis eterni non esse. Et necessarium secundum quid quidem solum temporaliter
est, sed legi eterne repugnat ipsum non esse. Et correspondenter de impossibile per
accidens et secundum quid. Unde sicut omne necessarium per accidens aut secundum
quid potest non esse, omne impossibile per accidens aut secundum quid potest esse.»

The first rule regarding true conditionals is the following:

proportionally, just as a conditional is true, so it is necessary, and just as
it is false, it is impossible, and the other way round40. To this the author
adds that logicians only accept that a conditional is true if it is absolutely
necessary. So it is not surprising, he continues, that this rule stays well away
from any extension of the ‘necessity’ involved in the evaluations of true
conditionals. By way of explanation, Wyclif lists a number of propositions
that follow from a specific interpretation of ‘necessary’ (incidentally, the
implications he mentions seem to have to do with the kind of necessity
attached to the antecedent or the consequent of the conditionals at issue):
that from something absolutely necessary something necessary per
accidens follows: e.g. from God is it follows that he wishes the world to
be, and from this it follows that the world is, and in brief anything else that
is contingently true. And from this it follows that everything in the future
will necessarily come about. It also follows that a consequence can be
posited about impossibles as well, because such an impossible can be true
per accidens41. After coming up with some other confusions resulting from
an incorrect take on necessity, he says that therefore we restrict the truth of
a conditional to the absolutely necessary42.
The next rule he mentions, is that from anything that is false (whether
insofar as it is signified, or insofar as it is a sign)43, a truth follows, but
never the other way round. The second part of the rule is expressed in
a peculiar way: Wyclif says that «Never from something true something

Ibid., p. 184.1-3: «Ex istis patent quedam regule. Prima quod proporcionaliter
sicut condicionalis est vera, sic necessaria, et sicut ipsa est falsa, sic impossibilis, et
econverso* [*ed. econtra].»
Ibid., p. 184.4-14: «Communitas tamen sophistarum non admittit condicio-
nalem esse bonam nisi fuerit absolute necessaria. Ideo non mirum si regule eorum
dissonant ab ampliantibus necessarium et impossibile. Nam ex absoluto necessario
sequitur necessarium per accidens, ut ex deum esse sequitur ipsum velle mundum
esse, et per consequens mundum esse; et breviter omne aliud verum, quantumlibet
contingens. Ex quo sequitur quod omne futurum necessario eveniet, non quidem
necessitate absolutam sed necessitate ex supposicione. Sequitur eciam quod
consequencia sit ponibilis et de impossibilibus, cum talis impossibilis per accidens
potest esse vera.»
Ibid., p. 185.18-20: «Dimissis ergo istis usque ad ultimum casum, fiat restriccio
veritatis condicionalis ad absolute necessariam.»
As he explains later on, a sign is true when its primary significate is the case at
some time; see below, note 48.

false follows on the part of reality (ex parte rei), to the extent that it is false,
because to any such extent, it is simply not the case»44.

2.2.5. On the expressions ‘true’ and ‘false’

Wyclif then enters into a quite detailed description of how to take

this particular rule, by looking at what precisely is involved in the use
of the expressions ‘true’ and ‘false’. First of all, as a corollary of the rule
he just mentioned, he puts forward that a contingent falsehood (falsum
contingens) can be true, and consequently it can be necessary in a certain
way. So in truths per accidens, not only from something false something
true follows, but something necessary will be false and something true will
be impossible45. From his examples we can gather that he means to say that
truth per accidens is such that it can change, owing to the change of the
conditions which happen to be true per accidens. For it can be necessary
that a stone has killed Socrates, but this can be impossible later on, after
the stone has been destroyed (because in that case the stone is not there
any more). And a similar account applies in ‘Something false will be true’,
and ‘That which at one time is not, will at another time be’, for if this
is now, then ‘this is’ is false, and if that is, then ‘that is’ is true. Nor is it
deceitful that if this will then be false or impossible, then that will be,
because exactly the opposite follows46.
All these cases of change from truth to falsehood and vice versa,
lead Wyclif to talk about the concepts of ‘nature’ and ‘false’ themselves.
What exactly is going on when something true changes into something

Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 184,16-19:
«Secundo patet ex falso, tam signo quam signato, sequi verum, sed numquam ex vero
sequi falsum ex parte rei pro mensura pro qua est falsum. Pro omni enim tali mensura,
ipsum non est.» Once again we are confronted with the idea of ‘truly following’, as
outlined above, p. 155.
Ibid., p. 184.19-23: «Potest tamen falsum contingens esse verum, et per
consequens necessarium aliquo modo. Unde in talibus que sunt per accidens, nedum
ex falso fiet verum, sed necessarium erit falsum et verum erit impossibile.»
Ibid., p. 184.23-20: «Hoc enim potest esse necessarium iste lapis occidit
Sortem, et hoc potest esse impossibile post corrupcionem lapidis. Idem ergo in re est,
falsum fore verum et illud quod aliquando non est aliquando fore, quia si hoc nunc non
est, tunc hoc esse est falsum, et si illud est, tunc illud esse est verum. Nec est color, si
hoc tunc erit falsum vel impossibile, quod tunc erit, cum oppositum sequitur.»

false? What does it tell us about the ontological status of a falsehood and
falsehood? Wyclif sides with a point of view brought forward by others,
who say that what is true is false because what is at one time the case,
at another time is not the case. Hence it does not follow: this is false or
impossible, therefore it is not the case; but it does follow that for some
period in time it is conflicting that that should necessarily be the case.
However, something is not true and false together at the same time. And in
this way something false follows from something true, as from, God wills
the day of judgement to be, follows that it [i.e., the day of judgement] is,
which is false until now, but true at the time in which it is the case47.
From this explanation it would appear that for Wyclif, the truth of
what is expressed in a premise must be an ens possibile, even though it
might not be the case at a given point in time. The only kind of premise
which cannot be true in any way at all is an impossible simpliciter. In the
case under consideration, Wyclif takes the antecedence and consequence to
apply to states of affairs that are or are not the case at a given point in time.
This also ties in with his remark that a sign is true if its primary significate
is the case at some time48.
The next two rules Wyclif deals with are the two problematic cases,
viz. of arguing from something impossible, and of arguing to something
necessary. It is worthwhile to consider what he has to say carefully, because
his explanation and examples clearly reveal the author’s ontological
perspective on inference.

Ibid., p. 184.31-41: «Unde aliqui dicunt quod <si> verum converteretur in
falsum et impossibile, et ipsum falsum habebit esse possibile vel impossibile. Alii
autem dicunt quod verum est falsum, quia quod pro uno tempore est, pro alio non
est. Unde non sequitur hoc est falsum vel impossibile; ergo non est; sed bene sequitur
quod pro aliqua mensura repugnat necessario illud esse. Non tamen simul et semel est
verum et falsum. Et sic ex vero sequitur falsum; ut ex deum velle diem iudicii esse,
sequitur illum esse, quod adhuc est falsum sed in tempore suo verum. Et illud reputo
esse probabile.»
Ibid., p. 185.5-6: «Ego autem voco signum verum si suum significatum
primarium est aliquando.» The primary significate of a proposition is identified with
the state of affairs expressed in that proposition, e.g. the primary significate of ‘Sortes
est homo’ is Sortem esse hominem. The primary significate is distinguished from the
secondary significate of a proposition, which in the case just mentioned would be
Sortem esse animal.

2.2.6. The two paradoxical principles

The sixth rule concerning conditional sentences Wyclif mentions here

is the ‘ex impossibili’ rule: «From anything simpliciter impossible follows
any conclusion that can be drawn, just as from any antecedent follows
whatever can be assigned absolutely necessary.» He gives an example of
the first rule: if it is impossible that you are an ass, you could only be an
ass if God did not exist, because if the former could be the case together
with the fact that God exists, then it can be this way, and so it would not
be impossible for the consequent to be the case49. Via this modus tollens, a
reductio ad absurdum, it is claimed that it cannot be true at the same time
that God exists and that you are an ass.
In a final paragraph on the rules he has just presented, Wyclif concludes
with the corollary of ‘ex impossibili’ rule, viz. that every good inference
posits that God is. The necessities are chained to each other in such a way
that if one is posited, it is impossible that the other should be destroyed,
just as if one impossible is posited, anything else that can be concluded
formally follows from it50. The ultimate standard for proving these two
rules is that they follow from the description of a good consequentia
he had given earlier51, which is tied up with his idea of truly following.
What Wyclif’s analysis and examples of these two problematic principles
demonstrate, in my opinion, is first, that by ‘impossibility’ he means a state
of affairs that simply cannot be the case, that is, something that totally
clashes with the eternal truth that God exists. If that truth should not be the
case, then anything, i.e., any state of affairs could be the case too. The latter

Ibid., pp. 185.36-186.2: «Sexto patet quod ex omni simpliciter impossibili
sequitur quodlibet concludendum, sicut ex omni antecedente sequitur quodlibet
absolute necessarium assignandum. Si enim possibile est te esse asinum, non potest
esse quod tu es asinus nisi deus non sit, quia si posset esse cum hoc quod deus sit, tunc
potest sic esse, et per consequens non est impossibile sic esse; quod tamen datum est.
Et eodem modo probatur secunda pars regule.»
Ibid., cap. 8, p. 186.3-11: «Septimo patet quod omnis talis consequencia ponit
fomaliter deum esse, et econverso* [*ed. econtra]. Et per consequens necessitates
sunt sic concathenathe quod posita una, impossibile est quod reliqua destruatur, sicut,
posito uno impossibili, formaliter sequitur quidlibet concludendum. Si enim quelibet
talis ponit aliqualiter esse, utputa veritatem significatam primarie per eandem, sequitur
quod quelibet talis ponit primam veritatem esse, ex qua posita, sequuntur due partes
See above, p. 157.

is a necessary consequence of such an impossibility. Conversely, whatever

state of affairs might be the case, it cannot possibly lead to the destruction
of the truth that God exists. So the formal validity of an inference is not just
connected with the kind of disputational necessity as found in Aristotle52,
but ultimately resides in a necessary order of the world as willed by God.
In the previous section, we considered Wyclif’s take on the two
paradoxes of implication. However, he also sets out to counter the
common objections against the two controversial rules. These objections
are interesting to consider, because Wyclif’s response to them indicates yet
again what his perspective on the nature of inferentiality involves. Let us
consider these objections and Wyclif’s responses to them one by one.

2.3. Wyclif’s defense of the principles

His first criticism is that according to the «old» rules of logic, no

conditional posits anything, so from none of them would follow that God
is. And by the same token, something impertinent would only follow from
an impossible materialiter; this is what the old rules of logic claim, adding
that from something impertinent never follows something necessary,
except materially. Thirdly, many necessities are such that they can not-be;
because otherwise there would not be an order between them; but if one
truth is posited, whatever possible truth will be posited53.
In all three of these objections, the most problematic feature of
the definition of a good consequence appears to be that it can include
propositions that do not seem to have anything to do with each other. In
other words, the arguments all relate to the idea that entailment only applies
to propositions that are relevant to each other on the basis of semantic

For a discussion of Aristotle’s account of semantic or dispositional necessity
in connection with the ascription of truth-alues to statements about future contingents,
see DE RIJK, Aristotle, Vol. I, pp. 268-282.
Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 186.14-
21: «Sed contra illud argumentatur primo per hoc quod iuxta antiquas regulas, nulla
condicionalis quicquid ponit; ergo non ex qualibet sequitur Deum esse. Secundo sic:
nunquam, nisi materialiter, sequitur impertinens ex impossibili, sicut antique regule
sumant, addentes quod nunquam sequitur necessarium nisi materialiter ex impertinenti.
Tercio: multe necessitates possunt non esse, quia aliter non esset ordo inter illas; sed,
posita una veritate, poneretur quelibet veritas possibilis.»

relationships between the terms involved. In the logic of someone like

Peter of Spain, there is no room for irrelevance between propositions of an
inference, because he takes the definition of consecutio in terms of causa
consequendi to express a topical connection between the different terms
of the proposition. That is why Peter only accepts one kind of case of
something following from an impossible, viz. if what is said to follow is an
outcome of the semantic relationships between the terms in the antecedent
and the consequent54.
Contrariwise, Wyclif thinks there is no reason whatsoever to reject the
rules as they stand, nor to modify them in such a way as to make room for
what would qualify as exceptional cases of good inferences. To be sure,
he partially agrees with the old qualification of a conditional, i.e., that it
does not posit anything, but assumes that this statement means something
different than what the opponents of the two ‘problematic’ rules seem
to think: on Wyclif’s own account, no conditional containing contingent
extremes (‘extremes’ in this case applies to the propositions involved in the
inference) posits the other extreme. For example, it is not necessary that
on the assumption that if you are running you are moving, that either you
are running or you are moving. However, such a conditional does posit its
primary significatum and all of its causes55, i.e., its (ontologically based)
As to the second objection to the rule ‘ex impossibili’ rule, arguing
that from an impossible something impertinent can only follow something
materialiter (as opposed to formaliter, that is), Wyclif explains what it
means to be ‘impertinens’; in this description Wyclif’s conceptualisation
of inferentiality is highlighted again. Our author’s response to the objection
at issue is that nothing is simpliciter impertinent to something necessary
or impossible, because it conflicts with everything in such a way that it is
impossible to assign anything whatsoever to it, and everything that is in
such a way necessary follows from anything. These are the two ways in

For an analysis of Peter of Spain’s views concerning this matter, see SPRUYT,
«Thirteenth-Century Positions».
Iohannes Wyclif, Tractatus de logica II, cap. 8, ed. DZIEWICKI, p. 186.23-30:
«Ad primum dico quod assumptum non est regula, cum obliquat a veritate. Verumtamen
cum antiqua sentencia communis non sit in toto falsa, intelligitur isto modo: nulla
condicionalis de contingentibus extremis ponit alterum illorum; ut non oportet si tu
movearis te currente quod vel curras vel movearis, quelibet tamen talis condicionalis
ponit suum primarium significatum et omnem eius cuasam.»

which a proposition is called ‘pertinent’ to another, one either because it is

antecedent to it, or because it conflicts with it. In a word, if A is pertinent to
B because it follows from B, by the same token B is pertinent to A because it
is antecedent to it. Pertinence then is a relation of sameness (equiparantia)
that can be equally founded in antecedence as it can in consequence56.
In his response to the final argument, Wyclif explicitly addresses the
nature of formalness involved in the relationship of entailment, and why it
will not do to talk about material consequences. He explains that it is just as
incoherent to have a good consequence that is not good in form, as it would
be to have matter without form; this is to say that inference is inference
owing to form, and even matter is what it is owing to form. The standard of
truth and the form by which any other conditional obtains, he says, is the
following: if of two truths one is impossible to be without the other, then if
one is, the other is too. Of course one can still make a distinction between
different forms of argument, viz. substantial, syllogistic, enthymematic
inferences and inductions, because some inferences obtain that are called
formal, but they are not called material57. Clearly, the ‘form’ he is talking
about here is the shape, or the schema of the argument. But no matter what,
it always comes down to the principle just mentioned: a consequence can
only be good if it at least obtains according to that form. And this means
that in order to assess the formal goodness of any inference of logic, it is
enough that it is an inference and that, from what the primary significates
of its antecedent and its consequent have demonstrated, it is impossible for
the one to be without the other58.

Ibid., p. 186.31-40: «Ad secundum dicitur quod nichil est impertinens simpliciter
necessario vel impossibili, cum omni tali impossibili repugnat quodlibet assignandum, et
omne sic necessarium sequitur ad quidlibet. Et ex istis duobus modis dicitur proposicio
pertinens alteri: vel quia antecedit ad illam, vel quia repugnat illi. Si enim A est pertinens
B quia sequitur ad B, per idem B est pertinens A quia antecedit ad A, cum pertinencia sit
relacio equiparancie eque fundabilis in antecedentia sicut in consequencia.»
Ibid., pp. 186.40-187.8: «Ulterius dicitur quod tam repugnat consequenciam
esse bonam et non de forma quam repugnat materiam esse informem. Nam ista est
veritas et forma exemplaris per quam qualibet alia condicionalis tenet: si duarum
veritatum unam impossibile est esse cum hoc quod non sit reliqua, tunc, si illa
est, reliqua est. Verumtamen est dare aliquas formas, substanciales, syllogistica,
entimematica, et inducciones, quia tenent alique consequencie que vocantur formales;
sed non iste vocate materiales.»
Ibid., p. 187.8-14: «Non tamen est possibile quod quod aliqua consequencia
teneat nisi teneat ad minimum per hanc formam. Sufficit ergo ad investigandum

It now has become quite clear that for Wyclif, unlike the authors of the
thirteenth century, some logicians of the fourteenth (inter alios John Buridan
and William of Ockham)59, the expression consequentia formalis indeed
refers to a logically valid inference, the validity of which is explained in
terms of the truth condition of the inferential sign ‘si’ only: a consequentia
formalis is such that the antecedent cannot be true without the consequent,
and that is it. Wyclif does not need an extra type of inference, i.e., the
consequentia materialis, because for him, entailment is tied up with being
in some way or another, and any kind of being, even matter, is what it is
due to a form (as in forma dat esse). Only impossibilia simpliciter are such
that they clash with any kind of being whatsoever, and as such they cannot
possibly be the case. And ultimately, and quite in line with this view, the
‘ex impossibili’ rule seems to account, in negative terms, for the causal
connection between antecedent and consequent in entailment.

3. Concluding remarks: rehabilitating the consequentia materialis

If we compare John Wyclif’s account of the formalness of inference

with what was said about the consequentia formalis earlier on, we can
see a change in the meaning of the word ‘formal’: from an expression
more specifically tied up with the semantic relationships between linguistic
expressions (a way in which it was indeed also used by Wyclif, in the
chapter on consequences, see above p. 157), it has now acquired a use for
the relationship between propositions as such, viz. the form by which an
inference is an inference.
However, that is not to say that the original distinction between
consequentia formalis and consequentia materialis disappeared altogether.
The paradoxes of implication continued to put logicians on the alert, like
the author of the Logica Morelli. This fifteenth-century treatise60, written in

bonitatem formalem cuiuscumque consequencie logice quod sit consequencia;

et, demonstratis primarie significatis per suum antecedens et suum consequens, sit
impossibile hoc esse nisi hoc sit.»
SCHUPP, Logical Problems, p. 42.
Logica Morelli, Ed. (from the Manuscripts, With an Introduction, Notes and
Indices) by J. SPRUYT, Brepols, Turnhout 2004 (Studia Artistarum, 12); also see
J. SPRUYT, «A Fifteenth-Century Treatise on Consequences», Vivarium, 26 (1999) 187-

the tradition of John Buridan and Ralph Strode, once again recognises (or
continues to recognise) a difference between these two kinds of inference.
It is the consequentia formalis which is the default, so to speak: this one is
described as a good consequence to which the expression ‘formal’ applies,
because what is expressed in the consequent is formally a part of the
understanding of what is expressed in the antecedent. The consequentia
materialis is the odd one out: it is only there to explain the ‘ex impossibili’
rule, and the rule ‘ad quodlibet necessarium’.




This paper is concerned with problems of definition and classification

of types of valid consequences in the late 14th century1. I will examine the
discussion of consequences in Peter of Mantua’s Logica (early 1390s) and
focus on his original way of understanding certain familiar distinctions (ut
nunc and simpliciter validity; material and formal validity). His theory
of consequences is an original account that combines elements from the
English tradition and from the Continental tradition in one eclectic model2. I

Tufts University, Department of Classics, 328 Eaton Hall, 5 the Green, Medford,
MA 02155, United States. Email: riccardo.strobino@tufts.edu. I wish to thank Chris
Martin for his willingness to discuss at length inseparability and logical consequence
in medieval logic, and for many insightful comments on the main ideas of this paper.
All shortcomings are my own.
I shall not deal with parallel issues arising in earlier phases of medieval logic
that are the object of original discussions, notably in the 12th century by Peter Abelard,
and in the 13th and early 14th century by Robert Kilwardby and John Duns Scotus.
The only existing study of Peter of Mantua’s theory of consequences, to date, is an
excellent article by M. BERTAGNA, «La dottrina delle conseguenze nella Logica di Pietro
da Mantova», Documenti e studi sulla tradizione filosofica medievale, 11 (2000) 459-495.
The article discusses at length some of the distinctions I examine in this paper, leaving
untouched the issue of latitudo which is what my analysis primarily aims at. A crucial
point where I also depart from Bertagna’s work is the choice of textual basis. His paper is
based on one of the two early printed editions of Peter’s Logica that came out in Venice
in 1492, and does not take into account the manuscript tradition altogether (see f. 3 below
for bibliographical references). The texts I discuss here, by contrast, are derived from a
systematic investigation of the manuscript tradition. Although the resulting text is not yet
in the form of a critical edition, some conclusions can already be firmly established: among
them the fact that a long section at the beginning of the treatise of consequences in all six
manuscripts of the Logica is entirely missing from the early printed editions. That section
contains better versions of the definitions discussed by Bertagna. The latter might be (i)
interpolations, (ii) belong to an alternative version of the text absorbed at some stage of the
transmission by the main version, or (iii) the outcome of an effort of correction in a text that
contained obvious repetitions and alternative formulations of the same arguments.

shall discuss some fundamental ideas about validity and form, and propose
an explanation for Peter’s contention that formal validity comes in degrees3.
The structure of Peter of Mantua’s treatise on consequences has two
clearly identifiable parts. The first part contains a set of definitions (valid
consequence in general; ut nunc valid consequence; materially valid
consequence; formally valid consequence) and classificatory distinctions
(ut nunc/ut semper, material/formal) along with a number of objections and
clarifications. The second part contains a list of general rules and special rules,
some of which are discussed at great length, and occupies the by-far largest
portion of the text. I shall be concerned in this paper with the first part only.
At the opening of the treatise Peter gives his definition of a valid
consequence (consequentia bona):

[Validity] A <valid> consequence is a rational or conditional propo-

sition between whose principal parts [i.e. antecedent and consequent]
there is a relation such that it is impossible for the contradictory of
the consequent to be the case along with the antecedent, without
introducing a new signification for one or more terms4.

On Peter of Mantua, see T. E. JAMES, «Peter Alboini of Mantua: Philosopher–
Humanist», Journal of the History of Philosophy, 12-2 (1974) 161-170. Cf. also for
an updated bibliography R. STROBINO, ‘Concedere’, ‘Negare’, ‘Dubitare’, Peter of
Mantua’s Treatise on Obligations, Ph.D. Dissertation Scuola Normale Superiore, Pisa
2009. For two recent studies on chapters of the Logica, see R. STROBINO, «Contexts
of Utterance and Evaluation in Peter of Mantua’s Obligationes», Vivarium, 49 (2011)
275-299, and R. STROBINO, «Truth and Paradox in Late XIVth-Century Logic: Peter
of Mantua’s Treatise on Insoluble Propositions», Documenti e studi sulla tradizione
filosofica medievale, 23 (2012) 475-519.
The manuscripts that contain the treatise on consequences are: (1) [O]xford,
Bodleian Library, Canon. misc. 219, ff. 85va-95va; (2) [B]erlin, Staatsbibliothek
Preussischer Kulturbesitz, Hamilton 525, ff. 82vb-88vb; (3) [M]antova, Biblioteca
Comunale, Ms. 76 (A III 12), ff. 61rb-72ra; (4) Venezia, Archivio dei [P]adri
Redentoristi di Santa Maria della Fava, Ms 457, ff. 47va-57ra; (5) [V]enezia, Biblioteca
Nazionale Marciana, L.VI.128 (2559), ff. 52vb-62rb; (6) Città del Vaticano, Biblioteca
Apostolica Vaticana, Vat. [L]at. 2135, ff. 44vb-54vb. The early printed editions are:
(7) [E1]: [Johannes Herbort, Padua 1477], sig. II4+1ra-IIII3+3vb; (8) [E2]: [Antonius
Carcanus for] Hieronymus De Durantibus, Papie 1483, sig. H3+3va-K2va; (9) [E3]:
Bonetus Locatellus, Venetiis 14921, sig. E4+2va-F4+1rb (used by Bertagna in his
article); (10) [E4]: Simon Bevilacqua, Venetiis 14922, sig. Hva-I3vb.
Peter of Mantua, Consequentie, Ms O, f. 85va: «Consequentia <bona> est
propositio rationalis vel conditionalis inter cuius partes principales talis habitudo

Validity in general is understood in terms of the impossibility of

the conjunction of the antecedent of a consequence and the negation of
its consequent, under the implicit assumption that categorematic terms,
i.e. all terms that are not logical constants, signify as they ordinarily do
(if they do not signify as they ordinarily do, because a new imposition
has been made that changes their ordinary signification, then the
consequence may no longer be valid: in such a case, one would need
to look at the new signification of the term(s) and establish whether
the conjunction of the antecedent and the negation of the consequent is
impossible or not).
Thus, Peter’s definition assumes that the following holds for any
consequence Φ of the form ‘p; therefore q’ (rationalis) or ‘if p, then q’

Ф is valid iff ¬◇(p ⋀ ¬q) (Val)

Ф is a valid consequence if and only if the conjunction of its antecedent

and the negation of its consequent is impossible. Now, there are three
fundamental reasons why such a conjunction of two propositions may
turn out to be impossible. Either (i) the first conjunct (the antecedent) is
impossible, or (ii) the second conjunct (the negation of the consequent)
is impossible, or (iii) they are both possible, if taken separately, but
incompatible with each other:

(i) ¬◇p iff ¨¬p (ex impossibili quodlibet)

(ii) ¬◇¬q iff ¨q (necessarium a quolibet)
(iii) ◇p, ◇¬q, but ¬◇(p ⋀ ¬q)

In the rest of the paper, I will refer to cases (i)-(ii) as cases of trivial
validity, and to case (iii) as a case of genuine validity5.

est quod non potest contradictorium consequentis sine nova termini vel terminorum
significatione stare cum antecedente eiusdem».
All texts discussed in the paper are the result of the author’s preparatory work for
a critical edition.
Similarly, in the next section, cases (iv)-(v) will count as trivial, and case (vi)
as genuine. The distinction between trivial and genuine validity is the only case in
which I am going to use conceptual vocabulary that Peter himself does not makes

First distinction: ut nunc and simpliciter

What does ‘impossible’ exactly mean in this context? The answer

to this question marks the first division between types of valid
consequences according to Peter of Mantua6. One way is to read
impossible as impossible per accidens, the other is to read impossible
as impossible simpliciter. Impossible per accidens propositions
are propositions that are impossible after a certain moment in time,
but were possible before that moment, like ‘Adam did not exist’
(assuming that Adam actually came into existence at some point).
Impossible simpliciter propositions are, by contrast, propositions that
are impossible at all times, like ‘A human being is a donkey’ or ‘God
does not exist’7. The same applies, mutatis mutandis, to the notion of
necessity which can accordingly stand either for necessity per accidens
or for necessity simpliciter.
Thus, the impossibility of the conjunction of the antecedent and
the negation of the consequent of a valid consequence can be either
unconditional, as in cases (i)-(iii) above, or relativized to some time.
In the latter case, the antecedent and the negation of the consequent
are neither impossible in their own right nor in principle incompatible
with each other, but rather one of the following obtains: either (iv)
the antecedent is impossible per accidens or (v) the consequent is
necessary per accidens (or the negation of the consequent is impossible
per accidens, which is the same) or (vi) they are both possible, if taken
separately, but incompatible with each other from a given time on (i.e.
not unconditionally):

use of. I introduce the distinction beetween trivial and genuine validity to refine the
criterion of justification for a ranking of degrees of (formal) validity, which will be
introduced below. Besides, I believe the distinction is useful in its own right to avoid
the conceptual confusion that sometimes arises in connection with the distinction
between material validity and formal validity which one might be easily tempted to
take as a surrogate of the former.
Peter of Mantua, Consequentie, Ms O, f. 85va: «Consequentiarum alia bona
<ut> nunc, alia bona ut semper».
These are standard examples of propositions that are usually taken to be
metaphysically impossible.

(iv) ¬◇i p iff ¨i ¬p (ex impossibili per accidens

quodlibet ut nunc)
(v) ¬◇i¬q iff ¨i q (necessarium per accidens a
quolibet ut nunc)
(vi) ◇ p, ◇¬q, but ¬◇i(p ⋀ ¬q)

According to whether the impossibility in question is of one kind or of

the other, the first distinction that Peter introduces is a distinction between
ut nunc valid consequences, as in (iv)-(vi), and ut semper or simpliciter
valid consequences, as in (i)-(iii). The first class is explicitly defined in
the treatise, while the second class is not, although its definition de facto
coincides with the definition of a valid consequence, if we implicitly take
the relation of incompatibility between the antecedent and the negation
of the consequent to hold at all times. Peter defines ut nunc validity as

[Ut nunc validity] An ut nunc valid consequence is a rational

or conditional consequence such that it is no longer possible for
the contradictory of its consequent to be the case along with its
antecedent without introducing a new signification for the terms,
although at some time it was possible for the opposite of the
consequent to be the case along with its antecedent8.

A theory that accounts for ut nunc valid consequences relies on

the fundamental assumptions that the past cannot be modified (nulla
potentia est ad preteritum), and that a false claim about the past cannot
possibly be turned into a true claim (per nullam potentiam potest
The table below summarizes the first division with the six cases in
which a consequence is said to be valid.

Peter of Mantua, Logica, Ms O, f. 85va: «Consequentia bona ut nunc est
consequentia rationalis vel conditionalis cuius iam contradictorium consequentis non
potest stare cum antecedente eiusdem sine nova impositione terminorum quamvis
aliquando potuit consequentis oppositum cum antecedente eius stare».

Simpliciter valid Ut nunc valid

(Simpliciter (Per accidens
impossibility) impossibility)

(i) ¬◇ p iff ¨¬p (iv) ¬◇i p iff ¨i ¬p Trivial

(ii) ¬◇¬q iff ¨q (v) ¬◇i¬q iff ¨i q Trivial

(iii) ◇ p, ◇¬q (vi) ◇ p, ◇¬q

but but Genuine
¬◇(p ⋀ ¬q) ¬◇i(p ⋀ ¬q)

Second distinction: material and formal

The second distinction that Peter introduces is only internal to the

class of consequences that are valid ut semper (or simpliciter)9. Those
are always either materially valid or formally valid. Now, what is it for a
consequence to be materially valid or formally valid? It is for it to be valid
and to satisfy an additional condition. Such a condition has to do with how
consequences having the same form as the original behave with respect to
validity, that is to say whether they, too, are valid or not.
It is in fleshing out this distinction that Peter gives his original
contribution to the theory of consequences in the late 14th century. His
account combines elements both from the English and from the Continental
tradition10. Before getting to it, however, let us look at the definitions.

Peter of Mantua, Consequentie, Ms O, f. 85vb: «Sed consequentiarum simpliciter
seu ut semper quedam est formalis quedam est materialis».
For a comprehensive list see BERTAGNA, «La dottrina delle conseguenze»,
pp. 489-490, ff. 52-53. Suffice it to note that by English tradition one should understand
here a variegated and fairly large group of authors and texts from the early 14th century
to the second half of the 14th century (including Walter Burley, the Consequentie
secundum modum Oxonie, Richard Ferrybridge, Ralph Strode only to mention a few
names). By Continental tradition one should understand a narrower group of auhtors
mainly active in Paris: most notably John Buridan, who seems to be the originator of
that tradition and whose influence later in the century came to be remarkable in the rest
of Continental Europe, especially through the works of other “Parisians” like Albert of
Saxony and Marsilius of Inghen.

[Material validity] A simpliciter material consequence is a

rational or conditional proposition (i) between whose principal
parts there is a relation such that it was, is and will be impossible,
without introducing a new signification for one or more terms, for
the contradictory of the consequent to be the case along with the
antecedent (or one that can be converted with one such proposition
without introducing a new signification), and (ii) such that it is not
the case that any proposition having the same form is or may be
valid without introducing a new signification11.

Material validity is understood as validity plus the condition that at

least one consequence having the same form as the consequence under
evaluation be invalid.

Φ is materially valid iff

(i) Φ is valid (Val)


(ii) there is a Ψ such that Ψ has the same form as Φ, and Ψ is not
valid ¬(SF)

If we call (SF) the requirement of validity under sameness of form with

respect to a given consequence Φ (i.e. the condition that all consequences
having the same form as Φ be valid), then the definition of material
validity is obtained by simply adding to the general definition of validity
the condition that (SF) be not satisfied, which I express as ¬(SF).
The definition of formal validity, by contrast, is obtained precisely by
adding (SF) itself to the general definition of validity:

[Formal validity] A simpliciter formal consequence is a rational

or conditional proposition (i) between whose principal parts there
is a relation such that it was, is and will be impossible, without
introducing a new signification for one or more terms, for the

Peter of Mantua, Consequentie, Ms O, f. 85vb: «Consequentia materialis
<simpliciter> est propositio rationalis vel conditionalis (i) inter cuius partes principales
est talis habitudo quod non potest contradictorium consequentis sine nova termini vel
terminorum significatione stare cum antecedente eiusdem nec potuit nec poterit, vel
que potest esse convertibilis sine nova impositione cum una tali, (ii) cui non valet vel
valere potest <quelibet sibi> similis in forma sine nova impositione».

contradictory of the consequent to be the case along with the

antecedent (or one that can be converted with one such proposition
without introducing a new signification), and (ii) such that any
proposition having the same form is or may be valid without
introducing a new signification12.

In other words,
Φ is formally valid iff

(i) Φ is valid (Val)


(ii) for all Ψ, if Ψ has the same form as Φ, then is Ψ valid (SF)

In the definition of formal validity, Peter’s wording in Latin is somehow

loose, because instead of requiring that all consequences with the same form
as the original be valid tout court (as is shown in the above reconstruction),
the second clause requires that they be or may be valid, which seems to allow
for a scenario in which they are not. If we were to take this seriously, the
theory itself would lose much of its appeal and interest, as the criterion would
not give a univocal way to identify the class of formally valid consequences
(in fact those putative formal consequences would just turn out to be material
consequences, because whenever (SF) is not satisfied, then if a consequence
is valid, it is so only materially). I believe the text should be taken to mean
that any potential consequence having the same form as the original is valid.
The potest in the definition of formal validity is not there to mean that a
consequence with the same form as the original may or may not be valid. The
inaccurate formulation is most likely an effect of deriving the definition of
formal validity from the definition of material validity by removing the sign
of negation from the second clause cui non valet vel valere potest quelibet
sibi similis in forma to obtain the rather unsettling cui valet vel valere potest
quelibet sibi similis in forma. How the passage should be read, if the theory
must make sense at all, seems to be straightforward.

Peter of Mantua, Consequentie, Ms O, f. 85vb: «Consequentia formalis
<simpliciter> est propositio rationalis sive conditionalis (i) inter cuius partes principales
est talis habitudo quod non potest contradictorium consequentis sine nova impositione
termini vel terminorum stare cum antecedente eiusdem <nec potuit nec poterit> vel
que potest esse sine nova impositione convertibilis cum una tali, (ii) cui valet vel
valere potest quelibet <sibi> similis in forma sine nova terminorum significatione».

Third distinction: trivial and genuine

A third distinction, which does not occur in the text but is nonetheless
useful to understand the picture at a higher level of generalization, is the one
between what I have called trivial consequences and genuine consequences.
The distinction cuts across both ut nunc and simpliciter consequences, and
within the latter class, across both material and formal consequences. In
other words, there are both trivial and genuine ut nunc consequences; there
are both trivial and genuine material consequences; and there are both trivial
and genuine formal consequences. Trivially valid consequences are valid
consequences that are such either because their antecedent is impossible
(ut nunc or simpliciter, and in the latter case, either because it expresses a
metaphysical impossibility or because it expresses a formal contradiction)
or because their consequent is necessary (ut nunc or simpliciter, and in the
latter case either because it expresses a metaphysical necessity or because
it expresses a formal tautology)13. Genuinely valid consequences are valid
consequences that are such because the conjunction of their antecedent and
the negation of their consequent is impossible (ut nunc or simpliciter) but
neither is their antecedent impossible nor their consequent necessary (in
none of the three aforementioned senses, i.e. ut nunc, metaphysically or
logically). The impossibility of the conjunction of the antecedent and the
negation of the consequent depends in this case on some kind of genuine
relation between the terms involved in the two propositions that makes the
consequent inseparable from the antecedent.
Now, when there is a genuine relation between the terms occurring in
the antecedent and in the consequent such that having the former without
the latter is impossible, the relation itself may be of different sorts. As we
shall see shortly, the claim that (formal) validity comes in degrees can be
satisfactorily explained if it is connected with the idea of classifying the
types of relations that may hold between the terms in the antecedent and
in the consequent and rank them according to some plausible criterion of
increasing (or decreasing) strength.

Standard examples are identity statements and their negations. For instance
‘Tu differs a te’ (‘You differ from yourself’) counts as a paradigmatic case of
formal contradiction, because it denies a true identity statement, and not merely as a
metaphysically impossible proposition like ‘A human being is a donkey’ or ‘God does
not exist’.

The reasons why the distinction between trivial and genuine validity
applies to both ut nunc and simpliciter valid consequences are obvious once
we define trivial cases as those in which the impossibility of the conjunction of
the antecedent and the negation of the consequent depends on the impossibility
of one of the two conjuncts (i.e. either the antecedent is impossible, or the
consequent is necessary and hence its negation is impossible), and genuine cases
as those in which the impossibility depends on the intrinsic incompatibility
between the antecedent and the negation of the consequent, or equivalently on
the inseparability of the consequent from the antecedent.
The table below brings together the distinction from the first section (ut
nunc and simpliciter) and the distinction from the second section (material
and formal), and lists all types of valid consequences, while making the
reason of their type of validity explicit:

Simpliciter Ut nunc

Material Formal

(1) (2) (3)

Impossible Impossible Impossible
antecedent antecedent antecedent
(contradiction) (per accidens) Trivial

(4) (5) (6)

Necessary Necessary Necessary
consequent consequent consequent
(tautology) (per accidens)

(7) (8) (9)

Impossible Impossible Impossible
conjunction only conjunction only conjunction only Genuine
(not in all terms14) (in all terms) (per accidens)

¬(SF) (SF)

I.e. the consequence does not hold for every substitution of categorematic

There are six types of trivial consequences. They are such either by
virtue of their antecedent being impossible, or by virtue of their consequent
being necessary. If this is the case only from a given time on, before which
the conjunction of the antecedent and the negation of the consequent was
not impossible, then we are talking about ut nunc valid consequences.
If, by contrast, the conjunction of the antecedent and the negation of the
consequent is impossible at all times, then we are talking about simpliciter
valid consequences. In the latter case, according to whether the requirement
expressed by (SF) –i.e. the validity of all consequences having the same form–
is satisfied or not, we will have formal or material consequences, respectively.
Below is a list of paradigmatic examples for each category, taken from
Peter’s discussion:

(1) God does not exist; therefore, you are disputing

Nullus Deus est; ergo, tu disputas
(2) You differ from yourself; therefore, a stick stands in the corner
Tu differs a te; ergo, baculus stat in angulo
A human being is not a human being; therefore, a goat is
Homo est non homo; ergo, capra disputat
(3) Adam did not exist; therefore, a chimera exists
Adam non fuit; ergo, chimera est
(4) You are running; therefore, God exists
Tu curris; ergo, Deus est
(5) You are running; therefore, the king is sitting or he is not sitting
Tu curris; ergo rex sedet vel non sedet
(6) God exists; therefore, Adam existed
Deus est; igitur Adam fuit
(7) Socrates believes that every human being is being deceived;
therefore Socrates is being deceived
Sor credit quod omnis homo decipitur; ergo, Sor decipitur
(8) A human being is running; therefore, an animal is running
Homo currit; ergo, animal currit
A human being is running; therefore, something capable of
laughter is running
Homo currit; ergo, risibile currit
Only a human being is running; therefore, a chimera is not running
Tantum homo currit; ergo, chimera non currit

(9) This instant exists; therefore, you are disputing (indicating the
present instant and assuming that you are actually disputing)
Hoc instans est; ergo, tu disputas (demonstrato instanti presenti
et te disputante).

Form under scrutiny

One of the features of the ‘dialectical context’ in the first part of Peter
of Mantua’s treatise is the presence of several objections and remarks
concerning sameness of form and validity. It is in this context that his
reception of elements from both the English and the Continental 14th-
century tradition on consequences becomes evident in its originality. The
English tradition tends to privilege a characterization of formal validity
that appeals to the requirement that the meaning of the consequent be
understood or contained in that of the antecedent (this being the reason
of the former’s inseparability from the latter). Continental authors, by
contrast, have a notion of formal validity that focuses exclusively on the
syntactical structure of arguments and conditional statements (formally
valid consequences are those that are valid in all terms). As a result, a
paradigmatic case on whose evaluation the two traditions will strongly
disagree is that of inferences like

(10) A human being is running; therefore, an animal is running.

Homo currit; ergo animal currit

Both traditions would agree that this is a valid consequence because

the conjunction of the antecedent and the negation of the consequent is
impossible. But they would evaluate differently the sense in which it is said
to be valid. English authors would claim that it is formally valid because
the consequent is understood in the antecedent. Continental authors would
claim that it is materially valid because an invalid consequence of the
same form can be exhibited, which implies that the above consequence is
not valid in all terms (it can be turned into a formally valid consequence,
however, by adding to the antecedent the premise ‘All human beings are
Peter of Mantua’s move consists in bringing together these two
traditions and combining essential elements of both in a new original

account15. The spirit of his model can best be captured by the following
description: fundamentally Continental with an English notion of form. It is
fundamentally Continental because the distinction between formal validity
and material validity is grounded on whether a consequence is valid in all
terms or not. Formally valid consequences are, on Peter’s account, those that
remain valid for all uniform substitutions of categorematic terms, under the
assumption that form is preserved; materially valid consequences are those
that remain valid only for some uniform substitution(s) of categorematic
terms (i.e. not for all), under the assumption that form is preserved. The
difference with the Continental approach lies in what should count as
form. And it is with respect to this central notion that Peter brings in a
decisive element from the English tradition. In the above example, form
for Continentals exclusively depends on the syntactical features of the
proposition. This is why

(11) A human being is running, therefore a piece of wood is running

Homo currit; ergo lignum currit

can qualify as a counterexample against the formal validity of (10), and is

supposed to show us, in Continental terms, that (10) is just materially valid
(as is (11), for that matter).
Peter, however, explicitly rejects this view. According to him, the form
of a consequence is not merely determined by syntactical features, like
the type and position of logical constants that occur in it. His notion of
form is what I shall call a thick notion of form. In addition to syntactical
features, which remain necessary to determine the form of a consequence,
there is another factor that needs to be taken into account, namely the type
of relation holding between terms in the antecedent and in the consequent
(notably, between the two subject terms and between the two predicate
terms)16. Between such pairs of terms there may or may not be a connection.

The point is extensively discussed and convincingly argued for in the
conclusion of BERTAGNA, «La dottrina delle conseguenze», pp. 488-495.
We shall assume, for the sake of simplicity, that a consequence of the form ‘if p,
then q’ or ‘p; therefore, q’ can be analyzed in the most basic case as ‘if A is B, then C is
D’ or ‘A is B; therefore C is D’, where A, B, C, and D are terms, and that the relations
we are primarily interested in are those between pairs A-C and B-D, respectively. The
analysis varies in function of the quality and quantity of the propositions in question
and of the presence of certain logical constants (such as exceptives, exclusives and the

If there is no connection, the terms are said to be irrelevant (impertinentes);

if there is a connection, the terms are said to be relevant (pertinentes), and
there is also a connection (consecutio) between antecedent and consequent.
The thick notion of form used by Peter instead of the merely syntactical
notion of form used by Continentals is the result of supplementing the
latter with considerations about the semantics of the terms involved (to
put it otherwise, one might think of it as a the result of systematically
importing topical relations in the process that leads to determining the
form of a consequence). And such considerations are precisely what lies
at the core of the English notion of formal validity. Leaving aside the
case of syllogistic inferences, which does not tell us much given that they
are regarded by both traditions as paradigmatic cases of formally valid
consequences17, the watershed between the two is represented by the way
in which we are supposed to evaluate inferences like (10) and (11).
In addition to providing a justification for the claim that affirmative
conditional statements involving, for instance, hierarchically subordinated
and superordinated terms are formally valid (not just materially valid
as Continentals would claim), Peter’s account has another effect: by
definition it makes consequences that have a formal contradiction as their
antecedent or a tautology as their consequent formally valid (not just
materially valid as English authors would claim). Thus, the modification
of the notion of form operated by Peter results in a partial rearrangement
of what counts as materially valid and what counts as formally valid for
the two traditions. Just as some consequences that are materially valid for
Continental authors turn out to be formally valid for Peter (in agreement
with the English tradition), so some consequences that are materially valid
for English authors turn out to be formally valid for Peter (in agreement
with the Continental tradition). An example of the first type are inferences
like (10); an example of the second type are inferences like (2) and (5).

like). We shall not go into details here for lack of space, but it may be useful to keep
in mind that the conditions for each type of inference are ordinarily laid down and
discussed in the contexts where general rules and special rules of consequences are
Syllogistic inferences are not a good test because they satisfy two distinct
criteria. As a result, even if they are regarded by both traditions as formally valid, it is
because of different reasons. They are valid in all terms (under uniform substitution of
categorematic terms) and the conclusion is contained or understood in the premises.

Degrees of formality (latitudo)

There is yet another aspect, however, that makes Peter’s views on

consequences interesting. His redefinition of the notion of form, achieved
by supplementing the syntactic criteria with considerations about the
relations between pairs of corresponding terms in the antecedent and in
the consequent, results on the one hand in making certain consequences
formally valid that would otherwise have turned out to be just materially
valid on the Continental account. But it also makes formally valid certain
consequences that are trivially valid and whose formal features make all
consequences having the same form valid, as in the case of consequences
with a contradictory antecedent or a tautological consequent. Therefore,
the class of consequences that count as formally valid according to Peter’s
revised notion of form will contain members that are quite heterogeneous
in nature: conditional claims about genuinely inseparable terms as well as
conditional claims that are trivially valid without exhibiting any connection
whatsoever between antecedent and consequent18.
A series of objections and remarks that are put forward in the text
after the definition of formal validity seems precisely to address this sort
of concern. Two of them prove to be of particular interest. First, against his
own proposed definition of formal validity, Peter considers the objection
that such a definition would vouch for inferences with no connection
between premises and conclusion (a point the English tradition would have
been sensitive to):

This consequence is valid ‘A human being is not a human being;

therefore a goat is disputing’, and yet there is no relation (habitudo)
between antecedent and consequent, because every relation which
is a consequence is a connection; but there is no connection in that
proposition [sc. ‘A goat is disputing’], for it is about terms that are
irrelevant to [the terms occurring in] ‘A human being is not a human
being’; therefore, from the latter to the former there is no connection
(consecutio) nor relevance (pertinentia)19.

This is just as true of material consequences, too: some are trivial and some are
genuine. But Peter does not seem to attend to the fact that this sort of heterogeneity is
not an exclusive feature of the class of formally valid consequences.
Peter of Mantua, Consequentie, Ms O, f. 86va: «Item hec consequentia est bona
‘Homo est non homo; ergo, capra disputat’ et tamen nulla est habitudo inter antecedens

An objection of this sort only makes sense from the perspective of a

theorist who believes that it is wrong to associate the notion of a formally
valid consequence with cases in which there is no relation or connection
whatsoever between antecedent and consequent. Second, in the context of
another objection raised primarily as a pretext for examining more closely
the claim that form is determined exclusively in function of the syntactical
features of a consequence (a claim which will be later categorically
rejected), Peter gives the standard example of consequences that have
the same form according to Continentals (but not according to his own
criterion), and makes the following point:

This consequence is valid ‘Terms are individually arranged in the

same way in one [consequence] as in the other, the quantity, quality,
and supposition are the same, and any property or mode of signifying
a term has in one consequence, [the corresponding term in] the other
has the same; therefore, they have the same form’20. The above
consequence holds according to those who assume that the form of a
consequence is determined only with respect to the aforementioned
[criteria]. They claim that the following consequences have the
same form: ‘A human being is running; therefore, an animal is
running’ and ‘A human being is running; therefore, a piece of wood
is running’21.

How can Peter effectively deal, at one and the same time, with the
concerns of Continentals about sameness of form, and those of the English
tradition about the absence of a genuine connection in certain kinds of
formal consequences (the trivial ones)? As for the Continental objection,
he simply discards it by putting forward his own revised definition of

et consequens, quia omnis habitudo que est consequentia est consecutio; sed nulla est
consecutio illius propositionis, quia est de <terminis> impertinentibus ad illam aliam
‘Homo est non homo’; ergo illius ad aliam nulla est consecutio nec pertinentia».
Peter of Mantua, Consequentie, Ms O, f. 86va: «Est similis terminorum
discretio in una sicut in alia, similis qualitas, quantitas, suppositio, et qualiscumque
proprietas alicuius termini est in una seu modus significandi talis est in alia; igitur, ille
sunt similium formarum».
Peter of Mantua, Consequentie, Ms O, f. 86va: «Tenet consequentia apud eos
qui ponunt formam consequentie attendi solum penes illa iam nominata, qui dicunt
has consequentias esse similium formarum: ‘Homo currit; igitur animal currit’, ‘Homo
currit; igitur, lignum currit’».

form, which enables him to argue that Continental counterexamples to

consequences he regards as formally valid are merely putative, because
in fact they do not satisfy the requirement of sameness of form once this
notion is understood according to the new definition:

To solve this difficulty, we shall assume first that the formality of a

consequence is determined with respect to (i) the mutual relevance
of the terms occurring in the consequent and in the antecedent, (ii)
their properties and (iii) order. Hence it is clear that the following are
not consequences with the same form ‘A human being is running;
therefore, something capable of laughter is running’, ‘A human
being is running; therefore, a piece of wood is running’, because the
relevance of the terms is not the same22.

In this passage Peter makes it clear that it is not merely in terms of

syntactic features that the form of a consequence is determined, but also
by taking into account the relations holding between the terms of the
antecedent and those of the consequent. In particular one must look at
the relation between the two subject terms and at that between the two
predicate terms in the antecedent and in the consequent. The relations
holding between the terms of those two pairs are an essential part of the
form of a consequence. This becomes especially relevant when it comes
to evaluating sameness of form, for it is a necessary condition for any
consequence Ψ having the same form as Φ that the relations between the
terms in Φ be preserved by the terms in Ψ. The above point is therefore
a piece of criticism leveled against an insufficient characterization of the
notion of form which Peter intends to replace with his own thick notion, by
requiring that all bits of information concerning the relations between pairs
of terms be taken into account.
As for the objection against the heterogeneity of the class of formally
valid consequences, and in particular against the idea of including in that
class consequences that exhibit no relevance or connection between the
terms in the antecedent and those in the consequent, Peter responds by

Peter of Mantua, Consequentie, Ms O, f. 86va: «Sed pro istorum solutione
accipitur primo quod formalitas consequentie attenditur penes terminorum pertinentiam
consequentis et antecedentis ad invicem, et proprietatem <et ordinem>. Ex quo patet
quod iste non sunt consequentie similes ‘Homo currit; igitur risibile currit’, ‘Homo
currit; igitur lignum currit’ quia non est similis terminorum pertinentia».

putting forward an elusive claim about the fact that formal validity admits
of a range of degrees (habet latitudinem).

Secondly, we shall assume that the formality of a consequence comes

in degrees and that one consequence is more formal than another23.

No explanation is given to substantiate this contention. But if we try to

spell out the claim that formality comes in degrees, a plausible explanation
could be given in the following way. All formally valid consequences are
valid, and formally so, that is to say the conjunction of their antecedents and
the negations of their consequents is impossible, and any consequence having
the same form is such that the conjunction of its antecedent and the negation
of its consequent is impossible. Yet, one might be inclined to think that some
are valid in a stronger sense than others, despite their being all formally
valid (i.e. despite the fact that they both satisfy the two conditions for formal
validity). For instance, a consequence having an explicit contradiction as its
antecedent or a tautology as its consequent is certainly valid (trivially so), and
no consequence having the same form will fail to be valid (any consequence
having the same form will have by definition an explicit contradiction as its
antecedent or a tautology as its consequent). But the pairs of terms involved
would, most likely24, have no relevance to or connection with one another.
Consider, by contrast, a consequence that is formally valid because of the
genuine inseparability of the terms involved. That consequence will be valid
and any consequence having the same form will also be valid, just as in the
previous case, but there will be in this case an additional element, namely a
connection between the terms. And from here one could take the argument
even further. Not only the existence of a relation of consecutio between
terms (or, better, a consecutio between pairs of corresponding terms: subject-
subject or predicate-predicate) may justify a higher position in the ranking for
a consequence of this sort as opposed to consequences exhibiting no relation
whatsoever, but even within the category of consequences exhibiting such

Peter of Mantua, Consequentie, Ms O, f. 86va: «Accipitur secundo quod
formalitas consequentie habet latitudinem et una consequentia est magis <formalis>
[reading formalis with BLM instead of similis in O] alia».
Consider cases like ‘Homo est non homo; ergo animal est non animal’ or
‘Homo est asinus; ergo homo est animal’. Peter does not discuss such examples, but
it seems that he would nonetheless be committed to regarding them both as cases of
formally valid consequences.

relations, one might be inclined to distinguish between stronger and weaker

degrees of formal validity according to the type of relation involved. For
example one might want to take a consequence whereby two terms stand in
the species/genus relation to be stronger than one whereby two terms stand
in the subject/proprium relation25.
In the case of (trivially) formally valid consequences having
either explicit contradictions as their antecedents or tautologies as their
consequents, the validity of consequences of the same form is guaranteed
by the fact that the latter will have by definition antecedents or consequents
with the same logical structure, if they are to preserve the form of the
original. Likewise in the case of (genuinely) formally valid consequences.
For what is it after all for a consequence to have the same form as a valid
consequence exhibiting a particular kind of relation between suitable pairs
of terms? It is for that consequence to exhibit in turn the same kind of
relation between the corresponding pair of terms occurring in its own
antecedent and consequent. If the original consequence was valid in virtue
of a relation holding between its terms, then by definition any consequence
with the same form will be valid in virtue of the same relation.
The relations between pairs of terms preserved by consequences
sharing the same form may be of different types. At least two are explicitly
identified in the context of Peter’s discussion, namely relations between
terms and their constituents (i.e. terms that belong to the same predicamental
line26) and relations between terms and non-constituent inseparable terms.
In the case of a constituent relation, the inseparability of two terms is a
result of the genuine inconceivability of one without the other, whereas in
the case of a non-constituent relation27, the inseparability of the two terms
is only a form of strict inseparability: the two terms are such that one can in
fact never be taken apart from the other, but it is not the case that one cannot
be conceived without the other. In the second case there is no containment
of sense; in the first case there is. Moreover, containment of sense entails

The examples used here make sense against the backdrop of an Aristotelian
theory of predicables, which in some form or another is presupposed by all the
accounts I have been discussing. However, nothing prevents the same considerations
put forward here from applying to theories that allow for other privileged relations
between types of terms.
Expressing, in other words, definitional connections.
Or any other inseparable accident, to keep the Aristotelian (broadly construed)
leitmotif going.

strict inseparability but not the other way around. For this reason, the class
of formally valid consequences associated with the first notion and the
class of those associated with the second notion may plausibly be taken to
represent degrees of formal validity of different strength.
In sum, there are at least three senses in which a consequence may be
formally valid. All senses satisfy the definition of formal validity, but they
do so in three different ways, or in virtue of three different reasons. The
reason may be either (i) a merely syntactic feature of their antecedent or
(ii) a merely syntactic feature of their consequent, without there being any
relation between the terms, or (iii) a genuine relation holding between the
terms. In the latter case, at least two types of fundamental relations seem to
stand out as natural candidates: containment of sense and strict inseparability
(associated in this context with an underlying theory of predicables).
In Peter’s work there is no explicit attempt to draw distinctions
between degrees of material validity, but the above line of reasoning might
apply to that class of consequences at least in one significant respect. The
distinction between genuine and trivial validity offers a way to rank the
cases in which there is a connection between antecedent and consequent
above the cases in which no such connection exists (material consequences
that are trivially valid), although it should be kept in mind that even when
such a connection exists (material consequences that are genuinely valid)
this is not the case unconditionally but rather only for some substitution(s)
of terms.
Thus, the claim that formal validity comes in degrees can best be
clarified if we look at the variety of relations that may hold between terms
in the antecedent and in the consequent. It is a classification that relies
first on the distinction between the existence or non-existence of a relation
between the terms, which sets trivial consequences apart from genuine
ones (the latter being stronger than the former), and secondly, when such
a relation exists, on its strength. The second parameter applies only to
the class of formal consequences that are genuinely valid. The stronger
the relation, the higher the degree of formal validity. The following table
shows how different notions of «following from» can be classified in order
to achieve an informal characterization of the latitudo (range of degrees) of
validity for a logical consequence, based on the idea that different notions
of inseparability may govern the connection between terms occurring in
the premises and in the conclusions of a valid inference.

1. Conceptual inseparability

2. Strict inseparability (in all Genuine


3. Contradictory antecedent
or Trivial
Tautological consequent

4. Strict Inseparability (not in all


5. Impossible antecedent
Necessary consequent


The account of logical consequence we can extract from Peter of

Mantua’s theory of consequentie brings together with originality the two
main trends developed in 14th-century discussions. It combines elements
of the English tradition, notably the idea that formal validity is grounded
on the intuition that the sense of the consequent should be understood
or contained in that of the antecedent, with elements of the Continental
tradition, notably the idea that formal validity means validity in all terms
under sameness of form. As one might expect, given that the definitions
differ in the two traditions, the class of materially valid consequences
and the class of formally valid consequences will not coincide but merely
overlap. Peter’s account offers an original way to look at the distinction
by supplementing the Continental account with a revised notion of form.
His approach changes quite significantly the spirit of the original project

by appropriating the idea of relevance (pertinentia) between the terms

featuring in the antecedent and in the consequent. By looking at the types
of relations between the terms, we can make sense of Peter’s claim that
formal validity admits of a range of degrees, and that some consequences
are more formal (or formally valid) than others.
I have argued that at the lowest level in the ranking are formal
consequences that are trivially valid, i.e. those consequences that are
valid either because their antecedents are contradictions or because
their consequents are tautologies. Above those are formal consequences
that are genuinely valid. The characteristic of such consequences is that
they are formally valid because of a relation holding between the terms
in their antecedents and their consequents. According to the type of
relation in question, we will either have, at an intermediate level, genuine
formal consequences in which the consequent is strictly inseparable
from the antecedent, or we will have, at the highest level, genuine formal
consequences in which the consequent is conceptually inseparable from
the antecedent. The reason of the internal division is that conceptual
inseparability (or containment of sense) entails strict inseparability but
not the other way around. Consequences that satisfy the requirement of
conceptual inseparability are therefore those that deserve the crown.
II. Formal Semantics: Issues and Strategies


13th-Century logic was never «formal» quite in the sense in which

modern logicians talk of «formal logic». But there was an awareness that
one characteristic of logic was that one ought to distinguish between a
theoretic variant (logica docens) and an applied one (logica utens)2, and
that the entities dealt with in pure logic are not real in the same sense as the
objects of the natural sciences. In short, that pure logic is about the structure
rather than about the matter of the arguments of applied logic. About the
middle of the century it became common to describe logical objects with
the Avicennian formula intentiones secundae (adiunctae primis) or simply
as intentiones, though any number of qualifications could be added to the
The same conceptual framework was used to describe the objects of
the sub-disciplines of logic. Thus there would be a dialectica docens and
a dialectica utens3, and the core elements of dialectic would be a certain
class of intentiones. Not everybody was equally good at keeping theoretic
and applied logic apart, and not everybody had a developed theory of the

Centre for the Aristotelian Tradition, Saxo Institute, University of Karen
Blixensvej 4 DK-2300 Copenhagen, Denmark.. Email: se@hum.ku.dk.
I would like to express in this place my gratitude to Dr. N. J. GREEN-PEDERSEN,
who has liberally allowed me to use transcriptions of Topics commentaries that he made
in the 1970s and early 1980s in preparation for his landmark book The Tradition of the
Topics in the Middle Ages. The Commentaries on Aristotle’s and Boethius’ ‘Topics’,
Analytica, Philosophia Verlag, München –Wien 1984. More information about the key
concepts of the present essay, intentio and habitudo (localis), in the context of the
Topics may be found in Green-Pedersen’s book via the entries ‘intention’ in the index
p. 450, and ‘locus: and relations (habitudines)’ on p. 452. Green-Pedersen kindly read
an almost-finished version of this article and helped me improve it.
As is my wont, I often tacitly change the orthography and punctuation of editions
that I quote.
For more about this distinction, see S. EBBESEN, «Logica docens / utens»,
Historisches Wörterbuch der Philosophie, vol. 5, Schwabe, Basel 1980, cols. 353-355.
For dialectica docens / utens, see also N. J. GREEN-PEDERSEN, «On the
Interpretation of Aristotle’s Topics in the Thirteenth Century», Cahiers de l’Institut du
Moyen-Âge Grec et Latin, 9 (1973) 1-46, at pp. 14-15.

intentiones of logic. One who was good at keeping things apart and did
have a developed theory was Boethius of Dacia (floruit ca. 1270-1275).
Unfortunately, we only have one of his question commentaries on the
Organon, but fortunately the one that has survived, his Quaestiones super
librum Topicorum, is a very mature work –actually, it seems to be a second,
revised edition4.
In this article I will present some striking features of Boethius of
Dacia’s theory of the nature of dialectic5. This must function as a stand-in
for his general theory of logic, which must have shared central features with
his theory of this particular sub-discipline. True, there are a few statements
about the subject of logic in the extant œuvre, but the information that «the
way of knowing and its variants by means of which the various entities are
known or grasped form the subject of logic»6 by no means suffices for a
reconstruction of the details of his views about logic in general.
A central term in Boethius of Dacia’s theory of dialectic is habitudo
localis. In the course of this article I shall try to convey an impression of
what he thought such a «local habitude» is, and at the end of the paper
I shall investigate the origin of the term. I render localis as ‘local’, but
‘topical’ would also have done. The adjective indicates that the habitude is

Boethius Dacus, Quaestiones super librum Topicorum, Ed. by. N. J. GREEN-
PEDERSEN – J. PINBORG, DSL – Gad, Copenhagen 1976 (Corpus Philosophorum
Danicorum Medii Aevi, VI.1). The case for the preserved text being a revised edition
is stated on p. XXV of the edition.
Boethius of Dacia’s theory of dialectic has already been examined in GREEN-
PEDERSEN’s The Tradition of the Topics, pp. 228-230. While I do not disagree with his
analysis, the focus of mine is different from his. For general descriptions of Boethius
of Dacia’s logic and theory of science, see (1) J. PINBORG, «Die Logik der Modistae»,
Studia Mediewistyczne, 16 (1975) 39-97 (Repr. in J. PINBORG, Medieval Semantics.
Selected Studies on Medieval Logic and Grammar, Variorum, London 1984). (2) S.
EBBESEN, «Boethius of Dacia: Science is a Serious Game», Theoria, 66 (2000) 145-158
(Repr. in S. EBBESEN, Topics in Latin Philosophy from the 12th-14th Centuries, Collected
Essays of Sten Ebbesen, vol. 2, Ashgate, Farnham – Burlington 2009, pp. 153-162).
On Boethius of Dacia’s philosophy as a whole, see (a) J. PINBORG, «Zur Philosophie
des Boethius de Dacia. Ein Überblick», Studia Mediewiztyczne, 15 (1974) 165-185
(Repr. in ID., Medieval Semantics). (b) S. EBBESEN, Den danske filosofis historie i
middelalderen, ca. 1170-1536, Gyldendal, København 2002.
Boethius Dacus, Modi Significandi, qu. 7.40-42, Ed. by J. PINBORG, DSL – Gad,
Copenhagen 1969 (Corpus Philosophorum Danicorum Medii Aevi, IV), p. 29: «modus
sciendi et differentiae eius, quibus sciri habent vel cognosci diversa entia, subiectum
est in logica.»

one connected to the loci/τὀποι of dialectic. A habitude is roughly the same

as a relation, but only roughly. According to the Oxford English Dictionary
(2nd ed.), this sense of the word is obsolete; the last occurrence quoted in
the dictionary is from Berkeley and dated 1732. However, I propose to
revive the use of ‘habitude’ to render the medieval technical term habitudo.
Unsurprisingly, the very first question in Boethius of Dacia’s
commentary on Aristotle’s Topics is «Whether dialectic is a science»
(Utrum dialectica sit scientia). Here is his determination7:

The correct answer is that dialectic is a science –a science that provides

knowledge about how to argue constructively or destructively for or
against the attribution of any predicate whatsoever in any matter
whatsoever on the basis of probable signs which habilitate the
subject to participate in the predicate without necessitating this (as
will appear below). So, in spite of the fact that a conclusion reached
by means of a dialectical syllogism is not as such known but only
believed in and not necessary, yet the art through which one knows
how to syllogize dialectically on the basis of probable signs is a
science and both certain and necessary.

In the answer to the first ratio in contrarium, he makes clear that it is

dialectica docens that is a genuine science, genuine knowledge8:

Dialectica docens proceeds from necessary principles, but dialectica

utens (i.e. dialectic applied to do a job) proceeds from probable
<assumptions>, for in so far as it is applied to do a job, it is applied
to some special matter and syllogizes on the basis of probable signs
belonging to that matter. This is why dialectic produces opinion and
not certain knowledge about specific things.

Boethius Dacus, Quaest. Top. I.1.25-32, ed. GREEN-PEDERSEN – PINBORG,
p. 12: «Solutio: Dicendum est quod dialectica est scientia qua scitur quomodo
quodlibet praedicatum est construendum et destruendum in qualibet materia ex
signis probabilibus, quae habilitant subiectum ad participationem praedicati et non
necessitant, sicut patebit inferius. Unde, licet conclusio per dialecticum syllogismum
conclusa in quantum talis non sit scita sed opinata et non necessaria, ars tamen per quam
scitur dialectice syllogizare ex signis probabilibus est scientia certa et necessaria.»
Ibid. I.1.38-44, p. 13: «dialectica docens procedit ex principiis necessariis, sed
dialectica utens, seu secundum quod applicatur ad opus, tunc applicatur ad aliquam
materiam specialem, quam syllogizat ex signis probabilibus illius materiae. Et ideo
dialectica de rebus specialibus facit opinionem et non certam scientiam.»

To the objection that dialectic cannot be a science because it is about

none of the standard sorts of entity (natural, mathematical, divine, moral),
Boethius replies9:

Dialectic is about a sort of entity, viz. the way in which a thing can
be known by means of probable <assumptions> and signs of the
thing in question, and this way of knowing is a sort of entity in an
extended sense of the word ‘entity’, even though it is none of the
entities on which you base your argument.

The promise to elucidate the somewhat obscure talk about «probable

signs» in the first quotation, above, is redeemed in question 210:

It should be understood that it belongs to the dialectician as such

to consider the structural traits (rationes) that accrue to things,
i.e. those common intentions that are the foundations of the local
habitudes (habitudines locales) by means of which the dialectician
corroborates his arguments, i.e. the intentions of genus and species,
cause and effect, relatives, contraries, etc. And since, obviously,
those intentions are found in all things, […] the result is that
dialectic is a common science and that the dialectician can argue
in every discipline and every matter. However, as those common
intentions from which the dialectician derives his arguments are
not the cause of the conclusions which he concludes in specific
sciences, but <mere> signs, the dialectician cannot have certain
knowledge about such conclusions, but <just> opinion. For without

Ibid. I.1.52-56, p. 13: «dialectica est de aliquo ente, scilicet de modo quo
cognosci potest res per probabilia et signa illius rei, et iste modus cognoscendi est
aliquod ens extenso nomine entis, licet sit nullum entium ex quibus tu arguis.»
Ibid. I.2.25-42, pp. 14-15: «intellegendum est, quod dialecticus per se habet
considerare rationes, quae rebus accidunt, videlicet istas communes intentiones,
in quibus fundantur habitudines locales, per quas dialecticus suas argumentationes
confirmat, scilicet intentionem generis et speciei, causae et effectus, relativorum,
contrariorum et ceteras tales intentiones. Et quia istae intentiones in omnibus rebus
sunt, ut de se patet […] ex hoc contingit quod dialectica est scientia communis et quod
dialecticus potest arguere in omni arte et in omni materia. Quia tamen istae communes
intentiones ex quibus dialecticus argumenta sua sumit non sunt causa conclusionum
quas concludit in scientiis specialibus, sed signa, ideo dialecticus de his conclusionibus
certam scientiam habere non potest, sed opinionem. Sine causis enim non est scire, sed
sine eis bene contingit opinari.»

causes one cannot know, but it is perfectly possible to entertain an

opinion without them.

The intentions are not causes of conclusions because intentions are

not constituents of real things. There is an order: (1) things (res) with their
ways of being (modi essendi), (2) common intentions, (3) local habitudes11:
First comes the thing itself, second the common intention founded on
things, third the local habitude. And because of various ways of being (or
various natures) of the thing, the thing contains various common intentions
and local habitudes, which are signs of various dialectical consequences12
from this thing to those things and from other things to this thing.
At first glance, this passage might seem to make the intentions internal
to things in the same way that the modi essendi are, but this is not what
Boethius wants to say. In question I.5 he discusses what is the cause of
entailment in (1) Socrates is a man, therefore he is an animal and (2)
Socrates is white, therefore he is not black. The cause of entailment in (1)
is not the habitude of species to genus and genus to species, but the fact
that man is inseparable from animal, and similarly with (2) the cause is not
the common intention of contrariety and the local habitude of contrary to
contrary, but the fact that white and black are incompossible. De facto, the
relevant intentions and habitudes always accompany their things, but they
are separable in thought13:

Ibid. I.7.17-22, p. 28: «primo est res ipsa, secundo est ipsa communis intentio
fundata in rebus; tertio est habitudo localis. Et propter diversum modum essendi
(sive propter naturas diversas rei) habet res in se diversas communes intentiones et
habitudines locales, quae sunt signa diversarum dialecticarum consequentiarum huius
rei ad illas res et aliarum rerum ad hanc rem.»
‘Consequences’ here means «relations of following», which, however, in
medieval terms is a habitudo rather than a relatio. Hence my choice of the clumsy
Boethius Dacus, Quaest. Top. I.5.67-78, ed. GREEN-PEDERSEN – PINBORG, pp. 22-
23: «licet enim res significatae per hoc quod est ‘album’ et ‘nigrum’ numquam sint sine
communi intentione contrarietatis et sine habitudine locali contrarii ad contrarium,
ipsa tamen natura rerum et ipsa habitudo localis non sunt idem, sed sunt aliud et
aliud. Circumscripta per intellectum communi intentione et habitudine locali contrarii
ad contrarium a rebus significatis per hos terminos ‘album’ et ‘nigrum’, adhuc esset
consequentia bona ‘hoc est album, ergo non est nigrum’, eo quod naturae istarum
rerum incompossibiles sunt. Et ex hoc apparet, quod communis intentio et habitudo
localis contrarii ad contrarium non sunt causa huius consequentiae, quia illud non est
causa, quo ablato adhuc remanet effectus.»

Although the things signified by ‘white’ and ‘black’ never are

without the common intention of contrariety or without the local
habitude of contrary to contrary, the nature of the things and the local
habitude are not identical, but distinct. If, as a thought experiment,
we were to pretend that no common intention or local habitude of
contrary to contrary were connected to the things signified by the
terms ‘white’ and ‘black’, ‘This is white, therefore it is not black’
would still be a good consequence, because the natures of those
things are incompossible. This shows that the common intention
and the local habitude of contrary to contrary are not the cause
of this consequence, for what can be removed with the effect still
remaining is not a cause.

Why is it permissible to counterfactually disregard the intentions and

habitudes? Obviously because they are entia rationis, and not intrinsecally
linked to things the way the things’ modi essendi are. So obviously that
Boethius of Dacia did not feel he had to make a point of the fact.
Now, the «ways of being» or «properties» which are the real
counterparts of the intellect’s intentiones, play a central role not only in
Boethius of Dacia’ attempt to anchor dialectic in reality, but also in his
attempt to do the same to grammar, and he was perfectly aware of this.
In his Questions on Priscian Major, more commonly known as Modi
Significandi, he repeatedly points out the parallelism between the way
the building blocks of grammar, the «ways/modes of signifying» (modi
significandi) and the dialectician’s local habitudes are derived from real
properties14. In qu. 19 he directly asks «whether the dialectician derives his
local habitudes and the grammarian his ways of signifying from the same
real properties» (utrum eaedem in re sint proprietates a quibus dialecticus
accipit habitudines locales et grammaticus modos significandi). The
answer is a qualified «Yes»: some intentions and habitudes and some ways
of signifying share a foundation in the same property, as is the case with
the dialectician’s universal and the grammarian’s mode of signifying of an
appellative noun. Other items in the tool-boxes of the dialectician and the
grammarian cannot be matched the same way.
The parallelism between grammar and dialectic may then be depicted
as follows:

See in particular Boethius Dacus, Modi Significandi, qu. 17-19, ed. PINBORG,
pp. 62-71.

(1) proprietates in re ¨ (2) modi intelligendi ¨ (3) modi significandi

(1) proprietates in re ¨ (2) intentiones communes ¨ (3) habitudines

The transition from (1) to (2) is where the flexibility of the human
mind shows itself, as it may interpret the same property correctly in
more than one way and apply the resulting understandings for different
purposes. The items labelled (3) are the fundamental primitive terms of
the respective disciplines. The grammarian «teaches how to express an
intended conceptual content of the mind in congruous speech, irrespective
of subject matter»15, and he does so by formulating rules that spell out
which combinations of ways of signifying will produce congruity.
The dialectician «teaches how to argue dialectically, i.e. on the basis of
probable <assumptions>» and as such it is a special science, «but as this
dialectical way of arguing is common to just any matter and all things […]
in this respect dialectic is a common science»16. Boethius does not say
much about maxims except that they corroborate (confirmant) arguments17,
but he would probably have agreed that they are the dialectician’s rules,
which spell out how a good dialectical argument may be structured. As
Radulphus Brito was to say in the 1290s18:

Ibid., qu. 7.106-108, p. 32: «Docet ergo grammatica modum exprimendi
conceptum mentis intentum per sermonem congruum in omni materia».
Boethius Dacus, Quaest. Top. I.2.18-24, ed. GREEN-PEDERSEN – PINBORG, p.
14: «dialectica quantum ad scibile, quod ipsa docet, est scientia specialis; docet enim
aliquod scibile speciale sicut modum arguendi dialectice sive ex probabilibus. Quia
tamen iste modus arguendi dialectice communis est omni materia et omnibus rebus,
[…] secundum istum modum dialectica est communis scientia.»
Thus Boethius Dacus, Quaest. Top. II.6.13-20, ed. GREEN-PEDERSEN –PINBORG,
p. 120: «idem in re est dignitas, communis animi conceptio et maxima propositio; sed
dicitur dignitas propter suam evidentiam […] Communis animi conceptio dicitur quia
communiter quilibet intellectus, nisi fuerit febre ignorantiae infirmus, sibi consentit.
Dicitur autem maxima propositio propter suam maximam potestatem quam habet in
omnibus aliis confirmandis.»
Radulphus Brito, Quaestiones super libro Topicorum Boethii, qu. II.2, Ed. by
N. J. GREEN-PEDERSEN, Cahiers de l’Institut du Moyen-Âge Grec et Latin, 26 (1978)
1-92, pp. 26-27: «Et dico quod locus maxima et differentia maximae differunt, sicut
satis visum est, quia maxima est propositio explicans habitudinem termini ad alterum,
ut visum est. Sed differentia maximae est habitudo termini ad terminum ut confirmans
argumentum. Modo ista differunt sicut explicans et explicatum.»

A maxim is a proposition that explicates the habitude of one term to

another […], but a difference of a maxim is the habitude of one term
to another in so far as it corroborates an argument. Now these two
differ in that one is explicating, the other what is explicated.

A key feature of the dialectician’s habitudes is their relational


As those common intentions are relational (respectivae), because

one belongs to a thing as compared to another thing and vice versa
–for the intention of genus does not belong in an absolute way to
some thing, but in so far as it is compared to another thing, of which
it is the genus, and similarly with all other intentions– therefore the
local habitudes, which are founded on those common intentions,
are relative (respectivae), for a local habitude is never founded
on something in an absolute way. And because a habitude is just
that of one thing to another, therefore consequence is founded on a

Notice that Boethius avoids the use of the word relatio. He clearly did
not want the habitudes to be burdened with all the problems that attach to
the relations of Aristotle’s Categories.
But if habitudes and intentions are derived from real properties,
how can the dialectician avoid being involved in metaphysical
considerations, and how can he, without straying into the domain
of metaphysics or natural science use his understanding about the
habitude of contrary to contrary to form an argument of the type: ‘This
is hot, therefore this is not cold’? And he certainly must not trespass
on the grounds of other sciences. Boethius of Dacia understood the
Aristotelian theory of science to imply that there must be watertight
bulkheads between any two sciences, unless one be a subordinate of

Boethius Dacus, Quaest. Top., Prooemium, ed. GREEN-PEDERSEN – PINBORG,
pp. 202-210, pp. 10-11: «Et quia illae communes intentiones sunt respectivae, quia
una debetur rei in comparatione ad aliam et e converso – intentio enim generis non
competit alicui rei absolute, sed in comparatione ad alteram rem, cuius est genus, et
eodem modo dicendum est de omnibus aliis – ideo habitudines locales, quae fundantur
in illis communibus intentionibus, respectivae sunt; numquam enim habitudo localis
in aliquo absolute fundatur. Et quia habitudo non est nisi unius ad alterum, ideo in
habitudine fundatur consequentia.»

the other, and similarly between science/knowledge and belief. This

fundamental principle is neatly summed up in a sentence that he repeats
on several occasions20:

Nullus artifex potest concedere aliquid vel negare nisi ex principiis

suae scientiae.

No specialist can concede or deny anything except on the ground of

the principles of his own science.

Boethius of Dacia’s solution of apparent violations of the principle is

that several sciences may well be mastered by one and the same person,
but he just should not mix up his role as a metaphysician with his role as a
dialectician, for instance. Qua dialectician he may take over some insights
that he has reached as a metaphysician, but he could not possibly have
reached those insights by doing dialectic. The theme is a recurrent one in
Boethius’ works.
Thus in question 27 of book I on the Topics he asks whether it is
the dialectician’s task to establish the division into ten categories. Since
Aristotle introduces the tenfold division in Topics I.9, this might seem to
be the case. But no21:

Qua dialectician, the dialectician only considers categorial things

per accidens and distinguishes between them, in order that based
on the things and their properties he may grasp the local habitudes
and the intentions themselves, the maxims, axioms and common
conceptions of the mind, all of which he considers per se, but the
things themselves only per accidens.

Just in case my readers start to wonder whether Boethius is not here

introducing a dangerous multiplicity of objects for the dialectician to

Here quoted from Boethius Dacus, Quaestiones De generatione et corruptione,
qu. 2, Ed. by G. SAJÓ, DSL – Bagge, Copenhagen 1972 (Corpus Philosophorum
Danicorum Medii Aevi VI.1), p. 8.
Boethius Dacus, Quaest. Top. I.27.26-31, ed. GREEN-PEDERSEN –PINBORG, p. 72:
«dialecticus in quantum dialecticus res praedicamentales per accidens considerat et eas
distinguit, ut ex rebus et earum proprietatibus cognoscat habitudines locales et ipsas
intentiones, maximas propositiones, dignitates et communes animi conceptiones; et
haec omnia per se considerat, res autem ipsas per accidens.»

study, he elsewhere makes clear that he considers ‘maxims’, ‘axioms’ and

‘common conceptions of the mind’ to be referentially equivalent22.
The importance of the nullus artifex principle is spelled out even
more clearly in question I.4 on the Topics, where it is asked whether a
dialectician can study any specific thing (utrum dialecticus consideret rem
specialem). The answer, of course, is a «No!», though with qualifications23:

It is not the dialectician’s job qua dialectician to contemplate specific

things, for any specific thing, be it mathematical, natural or divine,
pertains to some specialist together with everything whatever that is
in any way an attribute of that thing. As such, it is the dialectician’s
proper job to contemplate those common intentions and local
habitudes by means of which he corroborates his arguments, but
because those common intentions and local habitudes do not
have their foundation anywhere but in things, for this reason the
dialectician can in some way and per accidens contemplate the
natures of specific things. However, he does not do this in his capacity
of dialectician but in so far as he is simultaneously a philosopher.
It thus results that when actually presenting an argument the

See ibid. II.6.13-20, p. 120, quoted in note 17, above.
Ibid. I.4.15-42, pp. 17-18: «Dicendum quod dialecticus secundum quod
dialecticus per se res speciales considerare non habet; omnis enim res specialis sive
sit mathematica sive naturalis sive divina pertinet ad aliquem artificem specialem, et
omnia, quaecumque attribuuntur illi rei, quicumque fuerit ille modus attributionis.
Dialecticus autem per se habet considerare illas communes intentiones et habitudines
locales, per quas argumenta sua confirmat. Et quia istae communes intentiones et
habitudines locales non fundantur nisi in rebus, ideo dialecticus aliquo modo habet
considerare naturas rerum specialium per accidens. Hoc autem non facit in quantum
dialecticus, sed in quantum simul cum hoc est philosophus. Ex hoc apparet, quod
dialecticus arguens debet esse dialecticus et philosophus. Debet enim esse dialecticus,
ut possit considerare communes intentiones et habitudines locales, per quas suum
argumentum confirmat, et probabilitatem suarum propositionum, quas adducit. Sed
philosophus debet esse, ut sciat naturas rerum, quae significantur per terminos, utrum
scilicet sibi debeatur talis communis intentio vel alia. Si enim ignoraret naturam huius
rei, quae per hoc quod est ‘homo’ significatur, et naturam huius rei, quae per hoc quod
est ‘animal’ significatur, non posset scire, quod ei, quod est homo, debetur intentio
speciei et non generis, et ei, quod significatur per hoc quod est ‘animal’, intentio
generis et non speciei. Unde cum dialecticus sic arguit ‘Socrates est albus, ergo non est
niger’, nullo modo posset scire hoc esse bonum argumentum, nisi sciret naturas rerum,
quae significantur per hos terminos, esse incompossibiles et esse tales, quod positio
unius in aliquo subiecto est remotio alterius ab eodem».

dialectician must be both a dialectician and a philosopher. He must

be a dialectician in order to be able to contemplate the common
intentions and local habitudes by means of which he corroborates
his argument, and the probability of the premisses that he adduces,
but he must be a philosopher in order to know the natures of the
things signified by the terms <in his argument>, whether, that is,
such or such an intention matches them. For if he did not know the
nature of the thing that is signified by ‘homo’, and the nature of the
thing that is signified by ‘animal’, he could not know that it is the
intention of species, and not of genus, that matches man, or that it
is the intention of genus, and not of species, that matches what is
signified by ‘animal’. So when the dialectician argues ‘Socrates is
white, therefore he is not black’, he would have no way to know that
this is a good argument unless he knew that the natures of the things
signified by those terms are incompossible and such that positing
one of them in some subject entails the removal of the other from it.

In spite of Boethius of Dacia’s constant moving forth and back between

the pure dialecticus docens and dialecticus utens, the main point is clear,
I submit. Pure dialectic is a science about such common intentions as
contrariety and such and local habitudes as that of contrary to contrary.
Apart from syncategoremes, names for these relational entia rationis and
some other logical words, like ‘is predicated’, the axioms and theorems
of a dialectician’s science may not contain anything but placeholders like
‘whatever’. As soon as you replace a ‘whatever’ with a word with genuine
content, you are leaving the sphere of dialectic and only get the right result if
you know about the thing signified by the word you use in your instantiation.
Boethius of Dacia does, however, leave an inconsistency which he
could easily have weeded out. In the passage just quoted, he assigns to the
dialectician as such the ability to judge the probability of the premisses of
a dialectical argument. But that presupposes the sort of knowledge about
the terms of an actual argument that the dialectician is not supposed to have
qua dialectician. This slip was to have unfortunate consequences.
A generation after Boethius of Dacia, Radulphus Brito in question
I.19 of his Questions on Aristotle’s Topics asks «Whether dialecticus utens
can judge the truth of propositions framed in specific terms»24. In the

Radulphus Brito, Quaestiones super Topica Aristotelis, qu. II.19, Ms Paris,
BnF, lat. 11132, f. 18rA: «Utrum dialecticus utens possit iudicare veritatem <in>
propositionibus factis in terminis specialibus.»

determination he mentions two views, both of which he rejects: (1) that

there is no way a dialectician can judge the truth of such propositions, (2)
that an artifex mixtus can do so25:

Others say that someone who is a mixture of a dialectician and a

specialist can judge the truth of propositions framed in specific
terms, for in his capacity of dialectician he knows about second
intentions, and in his capacity of specialist he knows about the
terms, and thus someone who is a mixture of a dialectician and a
specialist can know about and judge the truth of propositions framed
in specific terms.

View (2) is Boethius of Dacia’s and leaves pure dialectic free to say
nothing about any actual thing. Radulphus could have adopted view (2)
and rectified Boethius of Dacia’s slip by denying (3) that a dialectician
as such can judge the probability of premisses containing categorematic
terms. Instead he chose to reject (2) and support (3), although in a rather
woolly fashion26:

The dialectician cannot properly judge the truth of propositions

framed in specific terms, but he can estimate their truth

Admittedly, Radulphus links the dialectician’s ability to estimate the

truth-value of such propositions to his being the sort of dialectician that,
Radulphus thinks, Aristotle has in mind in his Topics, namely one «who
deals with the dialectical syllogism with a view to its use and application»27.

Radulphus Brito, Quaest. Topica Arist., qu. II.19, Ms Paris, BnF, lat. 11132,
f. 18rB: «Item alii dicunt quod artifex mixtus ex dialectico et artifice speciali potest
iudicare veritatem in propositionibus factis in terminis specialibus, quia in quantum
est dialecticus cognoscit intentiones secundas, sed in quantum est artifex specialis
cognoscit terminos; et sic artifex mixtus ex dialectico et alio artifice speciali potest
cognoscere et iudicare veritatem in propositionibus specialibus.»
Radulphus Brito, Quaest. Topica Arist., qu. II.19, Ms Paris, BnF, lat. 11132,
f. 18rB: «dico breviter ad praesens quod dialecticus non habet iudicare veritatem
proprie in propositionibus factis in terminis specialibus, tamen veritatem habet in
propositionibus factis in terminis specialibus aestimare.»
Radulphus Brito, Quaest. Topica Arist., qu. II.19, Ms Paris, BnF, lat. 11132,
f. 18rB: «Sed dialecticus qui hic instruitur determinat de syllogismo dialectico secundum
eius usum et applicationem. Ergo habet applicare syllogismum dialecticum ubicumque
ipso utimur.» Cf. Radulphus Brito, Quaest. Top. Boethii, qu. I.1, ed. GREEN-PEDERSEN,

This still leaves the way open for him to operate with a pure dialectic
and a pure dialectician, but he seems not to have been really interested
in Boethius of Dacia’s quest to reserve a place for disciplines that do not
care about content. Part of the explanation may be that his own theory of
intentions links second intentions rather strongly to real things –but that is
not a matter I can discuss here.
I know of no later author who either follows in Boethius of Dacia’s
footsteps or even polemicizes against him. It looks as if he was a lone wolf.

Appendix. The origin of the term habitudo localis.

The word habitudo is fairly common in 13th-century logic, and often

it is little but a stylistic variant of relatio28, but it is particularly common
in contexts in which the author has reasons to avoid importing all the
problematic properties of Aristotelian relatives. Thus people will speak
about the habitudo of accidents to substances and the habitudo that links
each sense of an analogous term to the common core29. Similarly, in a

p. 2: «Ulterius intellegendum est quod differenter determinatur hic de loco et in Topicis

Aristotelis, quia hic determinatur de loco secundum suam essentiam, sed in Topicis Aris-
totelis quantum ad sui usum et applicationem, ut applicatur ad terminandum problemata de
quattuor praedicatis. Et propter hoc apparet quare hic vocantur loci et in libro Aristotelis
considerationes, quia in Topicis Aristotelis determinatur de eis quantum ad suum usum et
applicationem, et ‘consideratio’ magis nominat illud quod est in usu et applicatione; sed hic
vocantur loci quia hic determinatur de eis quantum ad eorum essentiam.»
E.g., Petrus de Alvernia, Quaestiones super Metaphysicam, qu. V.27, Ms
Cambridge, Peterhouse 152, f. 181rB-vA, Oxford, Merton 292, f. 295rB: «Dicendum
quod relationem necesse est esse ens. Et hoc apparet primo sic: quando illud ex quo
causantur alia quaedam entia secundum quod huiusmodi oportet esse aliquod ens;
sed ex ipsa relatione causantur alia quaedam entia, puta actio et passio, situs, habitus,
quando et ubi. Omnia enim ista consistunt in relatione unius ad alterum, vel saltem
derelinquuntur {-itur codd.} ex tali relatione vel habitudine. Quare manifestum est
quod relatio est aliquod ens.»
Incerti Auctores, Quaestiones super Sophisticos Elenchos, qu. 809.58-63, Ed.
by S. EBBESEN, DSL – Gad, Copenhagen 1977 (Corpus Philosophorum Danicorum
Medii Aevi VII), pp. 270-271: «ubicumque aliqua duo analogice continentur sub
aliquo tertio posterius habet definiri per habitudinem ad prius, sicut accidens definitur
per habitudinem ad substantiam, et sanum in urina vel diaeta per habitudinem ad
sanitatem animalis. Syllogismus autem dialecticus non definitur per habitudinem ad
demonstrativum, nec e converso.»

conjunctive proposition, et may be said to express a habitudo that links the

conjuncts30. In a discussion of whether inherence is essential to accidents,
Radulphus Brito on one occasion carefully distinguishes habitudo from
relatio, making the former the genus of the latter31:

<Ratio principalis> Item, inhaerentia est relatio, quia est habitudo

quae dicitur ad alterum; sed relatio non est de essentia accidentis;
probatio quia octo sunt genera accidentis absoluta; ergo inhaerentia
non est de essentia accidentis.

<Ad rationem> Ad aliud, cum dicitur quod inhaerentia non est de

essentia accidentis, et etiam <quod> sit quaedam relatio et quaedam
habitudo, dico quod duplex est habitudo: quaedam est ad aliud
tamquam ad terminum, et talis est relatio; alia est habitudo ad aliud
sicut ad subiectum, et illa est de ratione essentiali accidentis. Ideo etc.

The special habitudines that are local habitudes have a history that
goes back at least to Abelard. Admittedly, he does not say habitudo localis,
but he does talk of the habitudo between man and animal which possesses
an inferential force and is the locus from a species32:

Locum ergo generaliter definientes vim inferentiae dicimus. Veluti

cum talis proponitur consequentia: ‘si est homo, est animal’,
‘homo’, cuius habitudo ad ‘animal’ vim inferentiae tenet, locus
dicitur, cumque ‘homo’ ad ‘animal’ utpote species ad genus suum
sese habeat, locus ipse a specie assignandus est.

The first appearance of the complex term habitudo localis that I know
of is in Abbreviatio Montana33, which is supposedly from the middle of the

Incerti Auctores, Quaest. SE, qu. 815.122-128, ed. EBBESEN, pp. 292-293: «Si
enim quaeratur utrum ʻdeus est et tu es asinus’ non dicendum quod illa vera est propter
hoc quod altera pars huius [est] assertionis vera est, scilicet deum esse, nam absolute
non asseritur deum esse, sed sub habitudine ad te esse asinum, quae quidem habitudo
significatur per notam copulationis, et quia deum esse sub tali habitudine falsum est,
ideo totam copulativam iudicamus esse falsam.»
Radulphus Brito, Quaestiones super Metaphysicam, qu. VII.2, Ms Firenze,
BNC, E.I.252, f. 292rB-vA.
Petrus Abaelardus, Dialectica, Ed. by L. M. DE RIJK, Van Gorcum, Assen 1956
(2 ed. 1970), p. 253.
See L. M. DE RIJK, Logica Modernorum, vol. II.2, Van Gorcum, Assen 1967,
pp. 85ff.

12th century, but it is a remarkable fact that none of the other texts printed
in De Rijk’s Logica Modernorum II.1 uses the term. Nor does it occur in
such works from the first half of the 13th century as Tractatus Monacenses34
or Summa ‘In omni doctrina’35. It resurfaces about the middle of the 13th
century in the Topics commentary of Nicholas of Paris36, and then became
a standard term in texts on dialectic.
Abelard’s use of habitudo in this context has nothing of the appearance
of a novelty –one gets the impression that he is using a well-established
term. Yet it is not found in the earliest published commentary on De
topicis differentiis, the late 11th- commentary work that goes under the
title of Primum oportet37, and the only occurrence of the word in Manlius
Boethius’ De topicis differentiis38 is not in a relevant context, while it is
quite absent from his commentary on Cicero’s Topics.
Old Boethius is, nevertheless, the original source of the medieval
logicians’ use of habitudo in dialectic. In his translation of Porphyry’s
Isagoge, σχέσις is rendered habitudo, and most importantly, this happens in
a passage dealing with the relation of an intermediary species to its superior
genus and its inferior species. Such a species has two habitudines, one to
the inferior and one to the superior item in the genus-species hierarchy,
whereas a highest genus or a lowest species has only one habitudo39:

Tractatus Monacenses is my name for a collection of treatises found in ms
München, BSB, clm 14763, ff. 121r-141r, of which I am preparing an edition.
The Summa ‘In omni doctrina’, Ed. by E. P. BOS, Éditions de l’Institut
Supérieur de Philosophie – Peeters, Louvain-La-Neuve – Louvain 2001 (Philosophes
Médiévaux, XLIII).
See text in N. J. GREEN-PEDERSEN, «Discussions about the Status of the Loci
Dialectici in Works from the Middle of the 13th Century», Cahiers de l’Institut du
Moyen-Âge Grec et Latin, 20 (1977) 1-37, at p. 66. The texts published in that article
are a treasure trove for anyone interested in the history of habitudines and intentiones
in dialectic.
Edition in H. HANSEN, «An Early Commentary on Boethius’ Topics», Cahiers
de l’Institut du Moyen-Âge Grec et Latin, 76 (2005) 45-130.
Boethius, De Topicis Differentiis II.8.9, Ed. by D. Z. NIKITAS, Academy of
Athens – Vrin – Ousia, Athens – Paris – Bruxelles 1990 (Philosophi Byzantini 8), p.
39 (= PL 64, 1191B): «Est enim proportio: “nam ut sese gubernator habet ad navim,
ita magistratus ad civitatem”. Hic autem locus distat ab eo qui a similibus ducitur: ibi
enim una res unicuilibet alii comparatur; in proportione vero non est similitudo rerum,
sed quaedam habitudinis comparatio.»
Porphyrius, Isagoge, translatio Boethii, in Aristoteles Latinus I.6-7, Ed. by
L. MINIO-PALUELLO, adiuvante B. G. DOD, Desclée de Brouwer, Bruges – Paris 1960,

Quae vero sunt in medio, eorum quidem quae supra ipsa sunt, erunt
species, eorum vero quae post ipsa sunt, genera. Quare haec quidem
duas habent habitudines: eam quae est ad superiora, secundum
quam species ipsorum esse dicuntur, et eam quae est ad posteriora,
secundum quam genera ipsorum esse dicuntur. Extrema vero unam
habent habitudinem; nam et generalissimum ad ea quidem quae
posteriora sunt habet habitudinem, cum genus sit omnium id quod
est supremum, eam vero quae est ad superiora non habet, cum sit
supremum et primum principium; specialissimum autem unam habet
habitudinem, eam quae est ad superiora quorum est species, eam vero
quae est ad posteriora non diversam habet, sed etiam individuorum
species dicitur (sed species quidem individuorum velut ea continens,
species autem superiorum velut quae ab eis continetur).

This is indubitably the main source of the habitudo localis, but the
adoption of the term may also have been facilitated by the fact that it
occurs a couple of times in Boethius’ commentary on the Categories. He
thus says that some think up and down do not belong under quantity, rather,
they think they are habitudines, or in Greek: σχέσεις, defined by a relation
to our heads and feet, respectively40:

Sed quidam volunt non esse quantitatis quod sursum dicitur et

deorsum sed potius habitudines, quas Graeci σχέσεις vocant:
quae enim pars ad caput nostrum est, hunc sursum vocamus; quae
pars pedibus subiacet, illa deorsum dicitur; quocirca secundum
habitudinem quandam quodammodo ad nos ipsos relata sursum
deorsumque praedicamus.

pp. 10.14-11.6. Original text in Porphyrii Isagoge, Ed. by A. BUSSE, Reimer, Berlin
1887 (Commentaria in Aristotelem Graeca, 4.1), p. 5.6-16: «τὰ δὲ μέσα τῶν μὲν πρὸ
αὐτῶν εἴη ἂν εἴδη, τῶν δὲ μετ’ αὐτὰ γένη. ὥστε ταῦτα μὲν ἔχει δύο σχέσεις,
τήν τε πρὸς τὰ πρὸ αὐτῶν, καθ’ ἣν εἴδη αὐτῶν εἶναι λέγεται, τήν τε πρὸς τὰ
μετ’ αὐτά, καθ’ ἣν γένη αὐτῶν εἶναι λέγεται· τὰ δὲ ἄκρα μίαν ἔχει σχέσιν· τό τε
γὰρ γενικώτατον τὴν μὲν ὡς πρὸς τὰ ὑφ’ ἑαυτὸ ἔχει σχέσιν, γένος ὂν πάντων
τὸ ἀνωτάτω, τὴν δὲ ὡς πρὸς τὰ πρὸ ἑαυτοῦ οὐκέτι ἔχει, ἀνωτάτω ὂν καὶ ὡς
πρώτη ἀρχὴ καί, ὡς ἔφαμεν, ὑπὲρ ὃ οὐκ ἂν εἴη ἄλλο ἐπαναβεβηκὸς γένος· καὶ
τὸ εἰδικώτατον δὲ μίαν ἔχει σχέσιν τὴν μὲν ὡς πρὸς τὰ πρὸ αὐτοῦ, ὧν ἐστιν
εἶδος, τὴν δὲ ὡς πρὸς τὰ μετ’ αὐτὸ οὐκ ἀλλοίαν ἔχει, ἀλλὰ καὶ τῶν ἀτόμων
εἶδος λέγεται ἀλλ’ εἶδος μὲν λέγεται τῶν ἀτόμων ὡς περιέχον αὐτά, εἶδος δὲ
πάλιν τῶν πρὸ αὐτοῦ ὡς περιεχόμενον ὑπ’ αὐτῶν.»
Boethius, In Categorias Aristotelis. Patrologia Latina 64, J.-P. MIGNE, Paris
1847, 212B; cf. 223C.

This seems to be in accordance with the standard use of the term

σχέσις among Boethius’ Greek contemporaries, who appear to use the
word about all sorts of external properties –Aristotelian relations, position
in time and space, etc.
Later, when dealing with relations, Boethius stresses that genuine
relatives owe their «convertible» predicability to some habitudo. By saying
«A child is the child of his parent, and a parent is the parent of his child»
we explicate a habitudo and a comparatio and a continentia utrorumque,
as it were. By contrast, if we say «When the sun is above the earth it is
day, and when it is day the sun is above the earth» we do not point to any
habitudo, but just to a consequentia, i.e. that day and sunlight invariably
follow each other41:

Aliam vero attulit causam prorsus gravem: ait enim proprium

esse hoc relativorum, non secundum suam nuncupationem sed
secundum aliquam habitudinem, eodem modo converti. Qui enim
dicit «cum sol est super terram, dies est, et cum dies est, sol est
super terram» nullam habitudinem monstrat sed tantummodo
consequentiam ostendit. Consequitur enim super terram solem
esse cum dies est, et cum sol super terram cursus agat, diem
esse; cum vero aliquis dicit «filius patris filius, pater filii pater»,
habitudinem et comparationem et quodammodo continentiam
utrorumque declarat. Atque hoc quoque in alia quavis relatione
spectare licet. Quocirca, quoniam omnia ad aliquid secundum
quandam ad se invicem habitudinem continentiamque dicuntur,
secundum continentiam quoque et habitudinem eorum conversio
facienda est, qua in re nos quoque graviter dicentis Iamblici
auctoritati concedimus.

This suggests that to Manlius Boethius a habitudo is something

only incomplex items can possess, not propositions, and this view about
σχέσεις may be one he has taken over from Iamblichus (see the end of
the quotation), but it does not seem to have been standard doctrine. John
Philoponus, a younger contemporary of Boethius’, uses σχέσεις both of
relational quasi-properties of substances and of the relations between the
constituent propositions of a syllogism42.

Boethius, In Categorias 225A-B.
Joannes Philoponus, In Aristotelis Categorias commentarium, Ed. by A.
BUSSE, Reimer, Berlin 1898 (CAG, 13.1), p. 47.12-14: «ἀλλὰ λέγομεν ὅτι τὰ πρός

Nor were the medievals to follow Boethius in denying that propositions

could be the bearers of habitudes. Thus, in a sophisma from the 1270s
or thereabouts, the author says that the conjunction ‘if’ exercises the
act of consecution and expresses the habitude of the antecedent to the
consequent43. About 1250/60 Robertus de Aucumpno had proposed to
see the habitudes involved in dialectical syllogisms as a formal trait of
such syllogisms (though primarily linked to the terms involved, one would
presume)44, and towards 1280 Simon of Faversham even says that the form
of a syllogism simpliciter is the necessary habitude between the premisses
and the conclusion together with the unity of the middle term with the

τι, ὡς αὐτὸς ὑποκατιὼν λέγει, σχέσεις τινές εἰσι μόνον καὶ οἰκείαν ὑπόστασιν
οὐκ ἔχουσιν, ἐν δὲ ταῖς ἄλλαις κατηγορίαις τὸ εἶναι ἔχουσιν.» Id., In Aristotelis
Analytica Priora commentaria, Ed. by M. WALLIES, Reimer, Berlin 1905 (CAG, 13.2),
pp. 259.29-260.2: «ἐπειδὴ οὖν, φησί, δέδεικται ὅτι αἱ ποιοῦσαι τὸν συλλογισμὸν
προτάσεις ἔχουσί τινας πρὸς ἀλλήλας σχέσεις, διελὼν εἰς δύο τμήματα ταύτας
τὰς προτάσεις περὶ ἑκατέρου τμήματος ἐρωτήσαιμι.»
Sophismata Wigornensia 11, Ms Worcester Cath., Q.13, f. 41rB: «argumentis
respondeo dicendo quod ʻsiʼ exercet actum consecutionis et dicit habitudinem
quandam antecedentis ad consequens, et ratione illius actus sic externi datur intelligi
actus significatus».
Robertus de Aucumpno, Commentarium Sophisticos Elenchos, Ms
Cambridge, Peterhouse 206, f. 134rB, and Paris, Mazarine 3489, f. 1rA-B (minor
variants not registered): «dicendum quod demonstratio eo quod contrahit ad rem
nullam formam argumenti addit super syllogismum, et ideo non est obliquitas alia
contra demonstrationem ratione formae quam contra syllogismum simpliciter, nulla
enim habitudo reperitur in re quae sit formalis respectu argumenti, quia non est in
re habitudo nisi huius causae ad hunc effectum vel huiusmodi, habitudo autem
ratione causae eadem est cum habitudine syllogismi simpliciter, particularitas autem
superaddita est potius ratione materiae quam formae, non sic autem est de syllogismo
dialectico, habitudo enim intentionum et intentiones formales sunt respectu rerum.»
Similar statements are found in several Topics commentaries from about the middle of
the 13th century, as pointed out by GREEN-PEDERSEN, The Tradition of the Topics, p. 255.
For a good example, see his «Text 8» (Lisboa-Robert), p. 360: «Est enim illatio duplex:
una quae causatur ex qualitate et quantitate propositione et ordine earundem, et haec
determinatur in libro Priorum; alia est quae causatur ex habitudine terminorum, quae
habitudo vocatur locus, et haec determinatur in libro Topicorum. Quare manifestum est
quod dialecticus syllogismus novam formam addit ultra syllogismum simpliciter. Ideo
doctrina de syllogismo dialectico et syllogismo simpliciter sunt doctrinae separatae.
Demonstrativa vero ultra principia syllogismi simpliciter solummodo addit principia
ex quibus fit illatio, scilicet materialia principia, non enim addit principia quae sunt
rationes inferendi.»

extremes. Syllogistic figure or mode are given the less important status of
condiciones materiae45.
However, in spite of such apparent deviations from Manlius Boethius’
usage, there is no doubt that he is the Latin world’s source for this
philosophical term.

Simon de Faverisham, Sententia Topicorum, Ms Leipzig, UB, 1359, f. 25rB,
ad 100b25sq.: «Ulterius notandum quod forma syllogismi est habitudo necessaria
praemissarum ad conclusionem cum unitate medii ad extrema medii.» Anonymus
SF (the author of the first set of questions in Incerti Auctores, Quaestiones super
Sophisticos Elenchos), who composed his work a few years before Simon, would
beg to differ (qu. 11.39-45, ed. EBBESEN, pp. 25-28): «Ex hiis accipio quod forma
syllogismi simpliciter sit consequentia syllogistica fundata supra uniformem
comparationem medii ad extrema sub debita quantitate et qualitate, unde forma
syllogismi non est habitudo terminorum, sed est aliquid in habitudine. Et ideo
intellege quod unitas termini, uniformis comparatio medii ad extrema, modus et
figura sunt dispositiones materiales ad formam syllogismi simpliciter.» Notice,
however, that Simon and the anonymous speak of different types of habitudo. The
latter’s consequentia syllogistica corresponds to the former’s habitudo necessaria
praemissarum ad conclusionem, but the two disagree about the status of the
relationship of the middle to the extremes.



In the first chapter of the Sophistical Refutations, Aristotle puts forth the
claim that human language is necessarily ambiguous ; for it is a usual fact of
human communication that linguistic expressions represent different things
in different communication-contexts. Given the unavoidable ambiguity
of human language, Aristotle’s approach to ambiguity in general –and to
equivocation in particular– aims at recognizing and avoiding or resolving
linguistic ambiguity. The particular linguistic situation that Aristotle has in
mind in the Sophistical Refutations, as well as in the Topics, is the dialectical
joust. In both treatises Aristotle gives the respondent a considerable number
of tricks/tests that can help him to unveil and resolve the questioner’s
ambiguous uses of language. This, of course, entails the respondent’s
capacity of recognizing two fundamental intentions in the questioner’s
mind –the intention of deceiving and the intention that lies behind his use
of words. There are thus two features that make the approach to ambiguity
in Aristotle’s dialectic fundamentally pragmatic1: first, the questioner’s
intention as determining the meaning that expressions have in different
acts of communication; and second, the respondent’s recognition of those
meanings and intentions as necessary for avoiding and resolving ambiguity.
After having claimed in the first chapter of the Sophistical Refutations
that the most common and natural cause of a sophistical refutation is
the unawareness of the unavoidable ambiguity of language2, Aristotle

Department of Philosophy, Linguistics and Theory of Science, University of
Gothenburg FLoV, Box 200, 40530 Gothenburg, Sweden. Email: ana.maria.mora.
By «pragmatic» I mean the Gricean sense of an intimate link between, on the
one hand, the interlocutors’ intentions of communicating and their recognition of each
others intentions, and on the other hand the determination of meaning and the efficacity
of human communication; see e.g. P. GRICE, «Utterer’s Meaning and Intentions» and
«Utterer’s Meaning, Sentence-Meaning, and Word-Meaning», in Studies in the Way
of Words, Harvard University Press, Cambridge MA 1989, pp. 86-116 and 117-138.
Aristotle claims that human language is necessarily ambiguous because there
is only a limited number of articulate utterances that can be used in order to represent

introduces in the fourth chapter six possible sophistical refutations that

depend somehow on this ambiguity –refutations depending on equivocation,
amphiboly, accent, composition, division and form of the expression3.
Aristotle illustrates equivocation with three sophistical arguments.
The first one depends on the equivocation of the verb μανθάνειν, which
represents both the use of knowledge and the acquisition of knowledge; the
second one depends on the equivocation of τὰ δέοντα, which represents both
what must be because it is good and what must be because it is inevitable;
and the third one depends on the equivocation of the participle ὁ κάμνων,
which represents both he who is ill now and he who was ill in the past4.

an unlimited number of objects: «For since it is impossible to argue by introducing

the actual things under discussion, but we use names as symbols in the place of the
things, we think that what happens in the case of names happens also in the case of
the things, just as people who are counting think in the case of their counters. But the
cases are not really similar; for names and a quantity of terms are finite, whereas things
are infinite in number; and so the same expression and a single name must necessarily
signify a number of things.» Aristotle, Sophistical Refutations (hereafter, SE) 1.165a3-
17 (transl. E. S. FORSTER, Harvard University Press, Cambridge MA 1955, p. 13).
For other discussions of this passage, see L.-A. DORION, in Aristote, Les réfutations
sophistiques, intr., trad. et comm. par L.-A. DORION, Vrin, Paris 1995, pp. 206-209; P.
FAIT, in Aristotele, Le confutazioni sofistiche. Organon VI, a cura di P. FAIT, Editori
Laterza, Roma – Bari 2007, pp. 102-104; and S. G. SCHREIBER, Aristotle on False
Reasoning. Language and the World in the Sophistical Refutations, State University of
New York Press, Albany 2003, pp. 11-18.
Cf. Aristotle, SE 4.165b25-28. Aristotle does not give an explicit deduction
showing the exhaustivity and mutual exclusivity of this list, but a deduction at least as
old as Galen was supplied in the commentary tradition of this passage, a deduction to
which I shall come back later. See also S. EBBESEN, Commentators and Commentaries
on the Sophistici Elenchi. A Study of Post-Aristotelian Ancient and Medieval Writings
on Fallacies. Volume I. The Greek Tradition, Brill, Leiden 1981, pp. 78-87.
«Arguments such as the following are based on equivocation: “Those who know,
learn; for it is those who know the use of the letters that learn what is indicated to them”.
Here “learn” is equivocal, meaning “understand by using knowledge” and “acquire
knowledge”. Or again, “Evils are good, for what must exist is good, and evils must
exist”. Here “must exist” is used in two senses; it means “what is necessary”, which is
often true of evils (for some evil is necessary), and we also say that good things “must
exist”. Or again, “the same man is seating and standing and is a sick man and restored
to health; for it is the man who stood up that is standing, and it is he who was recovering
his health that is restored to health, but it was the man who was seated that stood up and
the man who was sick that was recovering”. For that “the sick man” does such and such
a thing or has such and such thing done to him, has not one meaning only but at one

Some lines later he adds that equivocation and amphiboly can happen in
three ways. First, when the same expression –simple or composed– properly
(κυρίως) represents several things, as in the case of the nouns «eagle»
(ἀετός) and «dog» (κύων)5; then, when it is a regular practice to use an
expression in a way that differs from its most common use; finally, when
some words signify only one thing when they are taken by themselves, but
they produce an ambiguous sentence when they are taken together, as in the
case of ἐπίσταται γράμματα. However, it must be noted that contrary to
the medieval reading of this chapter Aristotle claims neither that these three
modes should be mapped onto the three examples given above, nor that they
amount only and exclusively to a typology of equivocation6.
As regards the resolution of equivocation, it should be stressed that for
Aristotle the burden of resolution lies on the respondent of the dialectical
exchange. Hence, in chapter 19 of the Sophistical Refutations, Aristotle
gives the respondent instructions on how to react before a questioner who
is exploiting equivocation: i) if the equivocation is in the conclusion, but
the conclusion is not the contradictory of what is to be refuted, then there
is no refutation and therefore there is no need to resolve the equivocation;
ii) but if the equivocation is in the conclusion, and there is refutation, then
the respondent should argue that the questioner did not deny the same thing
that was claimed in the premise; iii) now, if the equivocation is in one
of the questions, the respondent should not reply simply yes or no, but

time means “the man who is now sick”, and at another time “the man who was formerly
sick”.» Aristotle, SE 4.165b30-166a6 (transl. FORSTER, p. 19).
κύων, for instance, can represent a dog, a dogstar and a Cynic philosopher.
Cf. Aristotle, SE 4. 166a14-21. This is rather a typology of ambiguity covering
both equivocation (first two cases) and amphiboly (last case). The first two cases can be
related to a division of nominal expressions in the Rhetoric: «A word in its prevailing
and native meaning and metaphor are alone useful in the lexis of prose. A sign of this
is that these are the only kinds of words everybody uses; for all people carry on their
conversations with metaphors and words in their native and prevailing meanings.»
Aristotle, Rhet. 3.1404b31-37 (transl. G. A KENNEDY, in On Rhetoric: A Theory of Civic
Discourse, Oxford University Press, Oxford 2007 p. 199). Thus, the first two cases
seem to make reference to the «native» meaning of a word and to its metaphorical
meanings, while the third case seems to make reference to syntactic ambiguity. For a
discussion of the relation between this passage in the Sophistical Refutations and the
notion of «metaphor» in the Rhetoric, see SCHREIBER, Aristotle on False Reasoning, pp.
179-185. For a more general study of the Aristotelian notion of metaphor, see J. KIRBY,
«Aristotle on Metaphor», American Journal of Philology, 118 (1997) 517-554.

he should stress that the question has different meanings and ask for a
resolution of the equivocation; iv) nevertheless, if he fails to notice the
equivocation in the question, it is possible to wait until the conclusion in
order to ask for a resolution of the equivocation7. On the other hand, in
the Topics Aristotle gives the respondent a number of tests that he can
apply to expressions so as to unveil equivocation, for instance: i) to check
whether their different uses have different contraries; ii) to check whether
they have different contradictories; iii) to check whether they fall under
different categories of predication; iv) to check whether they belong to
different genera; and v) to check whether they have different definitions8.
Consequently, the resolution of equivocation depends on the respondent’s
capacity of recognizing that the questioner has shifted the expression’s
meaning, and it is recommended only when there is a risk of linguistic
misunderstanding9. It is completely natural that the same word has different
meanings; the problem arises when equivocal words produce ambiguous
premises and fallacious ways of convincing through arguments.
Notwithstanding, for most medieval commentators of the Sophistical
Refutations, the passage SE 4.166a14-21 introduces a typology of equivocation
that can be mapped onto the examples that Aristotle gives in the passage SE
4.165b30-166a6, and they remarkably focus on the equivocation of simple
words rather than on the equivocation of expressions irrespective of whether
they are simple or composed. But, along the lines of Aristotle’s concern with
sophistical arguments rather than with equivocal words, they also discussed
extensively the problem of what can be inferred from ambiguous statements.
They also devoted a great deal of attention to the problem of how to produce
statements with equivocal words so as to assure that their truth-value can be
univocally determined. Now, the solutions that they give to this last problem
are not explicitly found in Aristotle’s works but come from the late ancient
interpretative traditions of Aristotelian dialectic.
My aim in the pages to follow is to present the different ways in which
13 -century Arts masters deal with the problem of inference from ambiguous
statements. I shall first introduce the 13th-century definitions of fallacy and

Aristotle, SE 19.177a9-32.
Unveiling and resolving equivocation is the second of the four dialectical tools
that Aristotle introduces in Topics I.15, together with the recollection of premises, the
discovery of differences and the perception of likenesses.
See e.g. Top. I.18.

of equivocation; I shall then present the typology of equivocation put forth

by these authors; thereafter I shall present two cases that were used in their
discussions on inference from ambiguous statements, cases that are closely
connected to the question of resolution of equivocation in a disputational
context. I shall conclude by showing that towards the end of the 13th
century a group of masters presents us with a solution where the speaker’s
intentions and the listener’s recognition of these intentions play a more
important role than they did in their predecessors. Consequently, I shall
claim that their approach to equivocation –albeit considerably different in
important respects– is closer to Aristotle’s approach in that they put back
on the listener the burden of disambiguation.

Fallacies and equivocation in the 13th century

The aim of a sophistical disputation (disputatio) is to produce an

apparent knowledge by means of a sophistical locus10. There are thirteen
sophistical loci: six depending on the expression (in dictione) and seven
outside of the expression (extra dictionem). The medieval tradition is more
concerned with fallacies than with refutations, even if the notion of fallacy
does not prevail in the Aristotelian treatise itself. A fallacy is then both the
deceit caused by an apparent knowledge and the cause of this deceit –the
sophistical locus itself.
Any fallacy in the sense of sophistical locus involves both the cause
of the appearance (causa apparentiae) and the cause of the non-existence
or the cause of the non-validity of the argument (causa non existentiae vel
causa falsitatis).
The fallacies depending on the expression are produced because we
represent several things with the same word or sentence –because of the
multiplicity of signification of one and the same expression. Thus, the
unity of the expression is the cause of the appearance, and the multiplicity
of signification is the cause of the non-validity of the argument (causa
non existentiae). This multiplicity can be actual, potential, or apparent
(fantastica). Equivocation and amphiboly depend on the actual multiplicity
of the expression. Accent, composition and division depend on its potential

See also S. EBBESEN, «The Way Fallacies were Treated in Scholastic Logic»,
Cahiers de l’Institut du Moyen-Âge Grec et Latin (CIMAGL), 55 (1987) 107-134.

multiplicity. And form of the expression depends on its apparent multi-

plicity. The source of this division of multiplicity is explicitly attributed
by some authors to a certain Alexander, but a similar deduction of the six
fallacies in dictione, based on the actual, potential or apparent multiplicity
of the expression can be traced back at least to Galen (2nd century AD)11. In
his De captionibus, Galen supplies the deduction of the types of refutation
depending on the expression that Aristotle mentions, but does not make
explicit, in the passage 4.165b27-30 of the Sophistical Refutations. Galen
tells us that the multiplicity of the expression is the vice (κακία) that causes
a sophistical refutation depending on the expression. This vice can be
actual (ἐνεργείᾳ), potential (δυνάμει) or only apparent (φαντασίᾳ), and
it can befall a word or a sentence. Thereafter, the deduction goes exactly
as it goes in the 13th-century, except that for Galen the deduction concerns
types of sophistical refutation, and not types of fallacy.
Summing up, there is equivocation (aequivocatio) when the same word
(dictio) actually signifies two or more different things ; and the fallacy of
equivocation corresponds to the apparent knowledge that we acquire when
we are not able to make a distinction between the diverse significates of
one and the same term in an argument12.

Types of equivocation

According to the 13th-century interpretation of the passage SE

4.166a15-23, Aristotle gives there a division of equivocation into three

Galen, Capt., cc. 2-3, pp. 3-4; for a detailed account of Galen’s deduction, see
EBBESEN, Commentators, pp. 78-87.
See e.g. Peter of Spain, Tractatus. Called afterwards Summulae logicales, ed.
and intr. by L. M. DE RIJK, Van Gorcum, Assen 1972, pp. 98:5-99:20: «[…] alio autem
modo fallacia dicitur causa sive principium illius deceptionis. […] in qualibet fallacia
isto secundo modo sumpta duplex est principium sive duplex causa, scilicet […] causa
apparentiae […]; alia causa […] est […] causa non existentiae, quod idem est, sive
causa falsitatis. Causa apparentiae in qualibet fallacia est quod movet ad credendum
quod non est […] causa falsitatis est quod dacit creditum esse falsum. Et quia ista
duo principia sive istae duae causae sunt in qualibet fallacia, ideo oportet quod sint
in aequivocatione […] causa apparentiae […] est unitas dictionis eiusdem simpliciter
[…] causa falsitatis […] diversitas rationum vel rerum significatarum. Aequivocatio
est cum diversae rerum rationes in eodem simpliciter nomine uniuntur. […] Fallacia
aequivocationis est deceptio causata in nobis ex impotentia distinguendi diversas in
eodem nomine rationes simpliciter.»

different types that can be mapped onto the three examples of ambiguity
that he gives in the passage SE 4.165b30-166a6. Authors from the period
generate this typology in two different ways. In the first one, common in
British authors, equivocation results from one term that represents several
things; but a term can represent several things by itself or in conjunction
with another term; if by itself, this can be either properly or derivatively, so
that we have the first and second type; if in conjunction with another term,
then we have the third type13.
Continental authors reject this deduction, because it would entail that
in the third type the term is not actually, but only potentially, equivocal.
They provide instead another deduction of the same typology according
to which equivocation stems from a term actually signifying several
things; but this can be either because of its significates or because of its co-
significates (e.g. co-signification of time); if because of its significates, this
can be either properly or derivatively, so that we have the first and second
type; if because of its co-significates, we have the third type14.
The three types of equivocation are illustrated with the three sophistical
arguments taken from the passage SE 4.165b30-166a6 as follows:

See e.g. William of Sherwood, Introductiones in logicam. Einführung in
die Logik, Ed. by H. BRANDS – C. KANN, Felix Meiner, Hamburg 1995, at p. 170:
«Est ergo aequivocatio eiusdem dictionis diversa significatio. Et hoc potest esse
tripliciter, aut quod dictio significet de se plura aut ex coniunctione eius cum alio.
Et primum dupliciter scilicet quod proprie significet plura aut transumptive.» And
Robert Kilwardby, Commentarium in Sophisticos Elenchos, Ed. by S. EBBESEN, in
CIMAGL, 67 (1997) 152-169, at p. 152: «Pars prima habet tres secundum tres modos
aequivocationis, primo enim ponit orationem in qua est aequivocatio ex principali
significatione dictionis, secundo cum dicit et rursus ponit orationem in qua est
aequivocatio ex eo quod dictio significat plura non principaliter, sed unum principaliter
et aliud transumptive; tertio cum dicit amplius ponit orationem in qua est aequivocatio
eo quod dictio per se significat unum solum et coniuncta alii significat plura.»
See e.g. Peter of Spain, Tractatus (ed. DE RIJK, p. 105:14-20): «Ratio diversitatis
[…] est in hoc quod aequivocatio fit tripliciter. Quia quod dictio plura significet, aut
est ex significatione aut ex consignificatione, […] si est ex significatione, tunc ea quae
significantur aut de pari […] aut per prius et posterius; […] si ex consignificatione,
sic est tertius.» And Nicholas of Paris, Notulae super librum Elenchorum, Ed.
by S. EBBESEN, in CIMAGL, 67 (1997) 169-179, at p. 173: «[…] modus proprius
aequivocationis causatur aut a parte significati, sic sunt duo primi; aut consignificati,
et sic tertius. Si a parte significati, dupliciter: vel aequaliter significati, sic est primus
modus; vel per prius et posterius, et sic est secundus[…]».

a) When a term has several significations that are equally proper:

Example: Quicumque sunt grammatici discunt;
scientes sunt grammatici;
ergo scientes discunt.

Here «discunt» is equivocal for using a discipline and acquiring a

Some authors (e.g. Peter of Spain) also illustrate this type with the
noun «canis» and the following argument :
Example (a’): Omnis canis est latrabile;
sed quoddam marinum animal est canis;
ergo quoddam marinum animal est latrabile.
b) When a term represents several things, one primarily and the
others derivatively:
Example: Omne expediens est bonum;
malum est expediens;
ergo malum est bonum.

Here «expediens» is equivocal for the necessity that can be in bad

things and the one that must be in good things. Medieval interpreters
discuss this type of equivocation as a case of analogy15.
c) When a term represents only one thing by itself, but several things
when it is joined to another term:
Example: Quicumque sanabatur sanus est;
laborans sanabatur;
ergo laborans sanus est.

Here «laborans» is equivocal, because the present participle can

represent both what is the case now (with a predicate in the present tense,
such as «sanus est») and what was the case in the past (with a predicate in
the past tense, such as «sanabatur»).

For medieval analogy see E. J. ASHWORTH, «Signification and Modes of
Signifying in Thirteenth-Century Logic: A Preface to Aquinas on Analogy», Medieval
Philosophy and Theology, 1 (1991) 39-67; ID., «Analogy and Equivocation in
Thirteenth-Century Logic: Aquinas in Context», Mediaeval Studies, 54 (1992) 94-135;
ID., Les théories de l’analogie du XIIe au XVIe siècle, Vrin, Paris 2008 (see pp. 111-113
for a complete list of Ashworth’s studies on the medieval notion of analogy).

The validity of «Canis est, ergo latrabile est»

A common question raised in 13th-century commentaries on the

Sophistical Refutations is whether an equivocal term like «canis» represents
its significates conjunctively or disjunctively16. From Robert Kilwardby’s
introduction to the question, it seems likely that the question arises from
a confrontation with chapter 8 of Aristotle’s De interpretatione17. In this
chapter, Aristotle seems to establish an equivalence between «a cloak is
white», «a horse and a man is white» and «a horse is white and a man is
white», where «cloak» has been given as a name to both man and horse18.
This could be interpreted as entailing an equivalence between «cloak» and
«man and horse» (or in medieval terms between «canis» and «latrabile et
marinum et caeleste»). What is far from clear, though, is the logical import
of this equivalence. For Aristotle is not saying that the affirmation «a cloak
is white» entails that «a man is white» and «a horse is white» are both true at
the same time so that both of them can be inferred from «a cloak is white».

For other discussions of this problem, see S. EBBESEN, «Can Equivocation
be Eliminated ?», Studia Mediewistyczne, 18 (1977) 103-24; ID., «Is “canis currit”
Ungrammatical? Grammar in Elenchi Commentaries», Historiographia Linguistica, 7
(1980) 53-68; and C. MARMO, Semiotica e linguaggio nella scolastica, Istituto Storico
Italiano per il Medio Evo, Roma 1994, chapters 5.1-5.3.
Kilwardby’s commentary on the Sophistical Refutations is perhaps the first
(partially) edited extant witness of this question. The question he raises is: «Habito quod
una dictio possit significare plura, quaeritur an significet omnia illa sub disiunctione vel
sub copulatione.» And his second argument against significatio sub disiunctione refers
explicitly to Int. 8: «Item hoc patet per Aristotelem in libro Perihermeneias. Dicit enim
quod si tunica imponatur ad significandum hominem et equum, et dicatur “tunica est
alba”, qui sic dicit aut dicit quod homo est albus et equus est albus, aut nihil dicit; sed
non dicit nihil; ergo dicit haec duo coniuncta per copulationem.» Robert Kilwardby,
Comm. in SE (ed. EBBESEN, p. 156). See also Albert the Great, Liber I Elenchorum t. II
c. II (ed. P. JAMMY, Alberti magni expositio sophisticorum elenchorum, C. Prost, Lyon
1651, vol. 2, pp. 539-542).
Aristotle, Int. 8.18a18-27 (transl. J. L. ACKRILL, Clarendon Press, Oxford 1963,
pp. 49-50): «But if one name is given to two things which do not make up one thing,
there is not a single affirmation. Suppose, for example, that one gave the name “cloak”
to horse and man; a cloak is white would not be a single affirmation. For to say this is
no different from saying a horse and a man is white, and this is no different from saying
a horse is white and a man is white. So if this last signifies more than one thing and
is more than one affirmation, clearly the first also signifies either more than one thing
or nothing (because no man is a horse). Consequently it is not necessary, with these
statements either, for one contradictory to be true and the other false.»

On the contrary, he is claiming that «a cloak is white» is not susceptible of

truth-determination because it represents both «a man is white» and «a horse
is white», which is not the same as saying that it is equivalent to «a man is
white and a horse is white». «A cloak is white» is a multiple statement and as
such it should not be granted as true or false19. Consequently, nothing should
be inferred from it. It is therefore with good reason that medieval interpreters
wonder about the logical import of the equivalence between «canis» and
«latrabile et marinum et caeleste», and more importantly about the logical
import of the conjuction in «latrabile et marinum et caeleste».
Kilwardby explains the conjunction by saying that an equivocal term
represents its significates conjunctively, without accepting that representing
them conjunctively amounts to their actual conjunction. Accordingly, he
also rejects the validity of the inference «Canis est, ergo latrabile est». In
order to explain what it is for the equivocal term to represent its significates
conjunctively without this entailing their actual conjunction, he tells us that
the equivocal term is something between a universal whole and an integral
whole, in that it shares certain features with both of them20.
As regards its similarity with an integral whole, Kilwardby tells us that
the equivocal term represents actually all its significates, because when it
is uttered different people can take it as representing different things at the
same time21, just as different people can focus on different parts of a house,
for instance when they are watching it from different viewpoints.

Cf. Aristotle, SE 17.175b39-176a19 (transl. FORSTER, p. 95): «If one does not
make two questions into one, the fallacy which depends on equivocation and ambiguity
would not exist either, but either refutation or absence of refutation. For what is the
difference between asking whether Callias and Themistocles are musical and asking
the same question about two people with the same name? For if one indicates more
things than one, one has asked more questions than one. If therefore it is not correct to
demand simply to receive one answer to two questions, clearly it is not proper to give
a simple answer to any equivocal question, even though the term is true of all of the
subjects, as some people claim that one ought. […] If therefore one must not give one
answer to two questions, it is obvious that neither should one say «yes» or «no» where
equivocal terms are used; […]»
Robert Kilwardby, Comm. in SE (ed. EBBESEN, p. 157): «[…] dictio aequivoca
significat plura et sub copulatione. Verumtamen notandum est quod dictio aequivoca
uno modo participat naturam totius integralis, alio modo naturam totius universalis;
unde est quasi quoddam totum medium inter totum universale et totum integrale.»
See Robert Kilwardby, Comm. in SE (ed. EBBESEN, pp. 157-158): «Naturam
autem totius integralis habet in hoc quod sicut totum integrale simul actu continet in se

As regards the limits of its similarity with universal terms, Manlius

Boethius had already pointed out in his De divisione that the resolution
of an equivocal term should not be confused with the determination of
universal terms :

But since some things are equivocal, others univocal, and univocals
are precisely the things we assume for the cutting up of genera
whereas with equivocals there is division only of the signification,
it must first be determined what is univocal and what equivocal, lest
we be deceived into resolving an equivocal name into significant
spoken sounds as though into species22. (transl. MAGEE, p. 17)

Both the equivocal term and the generic term are more universal than
their parts, but the genus is more universal both as regards nature and
predication, while the equivocal term is so only as regards predication.
Thus, the equivocal term shares some logical features with universal terms,
without being more universal in nature than its parts. Hence, it cannot be
displayed into its parts as a genus is into its species (for instance, «animal»
is displayed into «man» by means of the specific difference «rational»)23.
But the equivocal term behaves as a universal term in that, just as the

omnes suas partes […] si ergo aliquis proferat “canis currit” simul et semel et in eodem
instanti percipiunt Socrates et Plato diversa significata […]»
Manlius Boethius, De divisione, critical edition, translation and commentary by
J. MAGEE, Brill, Leiden 1998, at pp. 16:24-18:1: «Sed quoniam alia sunt aequivoca, alia
univoca, et quae sunt univoca ipsa in generum suscipimus sectiones, quae vero sunt
aequivoca in his divisio sola significationis est, videndum prius est quid sit univocum
quid aequivocum ne, cum ista fefellerint, aequivocum nomen quasi in species ita in
significativas voces resolvamus.»
Manlius Boethius, De divisione (ed. MAGEE, pp. 10:32-12:15): «Secundum
se autem divisionis huiusmodi differentia est. Differt enim divisio generis a vocis
divisione quod vox quidem in proprias semper significationes separatur, genus non in
significationes sed in quadam a se quodammodo creatione disiungitur, et genus semper
speciei propriae totum est et universalius in natura, aequivocatio vero universalior
quidem significata re dicitur tantum voce, non etiam totum est in natura. Illo quoque
a vocis distributione dividitur, quod nihil habent commune praeter solum nomen quae
sub voce sunt, quae vero sub genere collocantur et nomen generis et definitionem
suscipiunt. […] Generis vero apud omnes eadem divisio distributioque permanet,
unde fit ut vocis quidem divisio ad positionem consuetudinemque pertineat, generis ad
naturam, nam quod apud omnes idem est natura est, consuetudinis vero est quod apud
aliquos permutatur.»

universal term, it does not allow downward entailment. Hence, just as

«Animal est, ergo homo est» is not valid, in the same way «Canis est, ergo
latrabile est» is not valid either24. And this is because both the universal
and the equivocal term are applicable to all their parts, but without positing
their actual conjunction.
In the second half of the century, the Incertus SF –the author of an
anonymous commentary on the Sophistical Refutations– answers the
question whether an equivocal term represents its significates conjunctively
or disjunctively, by saying that although the equivocal term represents
actually all its significates, it is neither equivalent to their conjunction
nor to their disjunction25. It is noteworthy that this does not amount to
rejecting Kilwardby’s solution; for strictly speaking they both agree on
the fact that the equivocal term is equivalent to neither the conjunction
nor the disjunction of its significates, at least insofar as inference from
an ambiguous statement is concerned. Accordingly, for the Incertus SF
the inference «Canis currit, ergo latrabile currit» is not valid, because the
antecedent has an equivocal subject, a statement with an equivocal subject
is not an unity, and therefore its truth or falsity cannot be determined
without a disambiguation of its subject26.

Robert Kilwardby, Comm. in SE (ed. EBBESEN, pp. 158-159): «Item, nomen
aequivocum habet in se naturam totius universalis in hoc quod sicut totum universale
secundum suam substantiam totam est in qualibet sui parte, ita vox aequivoca secundum
se totam est in quolibet significato per ipsam, […] In dictione autem aequivoca non est
ita, immo quodlibet significatorum significatur per totam vocem, et in talibus non est
praedicta propositio intelligenda, quia hoc modo participat dictio aequivoca naturam
totius universalis […] Sicut igitur a toto universali ad suam partem non tenet processus
affirmando, similiter nec tenebit a toto aequivoco ad aliquod suorum significatorum.»
Incertus SF, Quaest. sup. SE, q. 45 (ed. S. EBBESEN, in Auctores Incerti,
Quaestiones super Sophisticos Elenchos, ed. S. Ebbesen, GAD, Copenhagen 1977.
pp. 94:142-95:168): «Unde terminus aequivocus neutro istorum modorum sua
significata importat, quia nec copulationem nec disiunctionem in suo modo significandi
vel significato includit. Dico tamen quod terminus aequivocus habet sua significata
in actu, ut secundo acceptum est, ita quod sicut definitio habet suas partes in actu,
ita terminus aequivocus habet sua significata in actu, […] Et illud est quod dixerunt
antiqui quod terminus aequivocus non significat hoc et hoc et hoc, nec significat hoc
vel hoc vel hoc, sed significat hoc hoc hoc, […]»
Incertus SF, Quaest. sup. SE, q. 47 (ed. EBBESEN, p. 103:63-72): «[U]nitas
simpliciter <non> reperitur in hoc antecedente quod est “canis currit”, quia sicut
aliquid est ens ita est unum, […]; sed hoc antecedens “canis currit” entitatem habet a
suis terminis, terminus autem eius non est unus, non enim est ens unum nec in re nec in

At the time of Incertus SF’s commentary there is a development of

the discussion ; for commentaries of the period commonly raise the further
question whether it is possible to resolve an ambiguous statement at the
time of its utterance by means of the adjunction of a qualifying term to the
equivocal term27. In other words, since it is now generally accepted that
«canis est, ergo latrabile est» is not a valid inference, the question becomes
whether «canis quadrupes est, ergo latrabile est» is valid or not.
The question actually raised in the question-commentaries of the second
half of the century is whether an equivocal term can be disambiguated by
means of the adjunction of another term, so that a statement containing it
is susceptible of truth-value. This amounts to asking whether it is possible
that the speaker produces a univocal statement, even though it contains an
equivocal word. The adjunction of a disambiguating term, in turn, can be
either immediate –as in «canis quadrupes»– or mediate –as in «Canis est
quadrupes». Incertus SF’s position is that the equivocal term cannot be
disambiguated by a mediate adjunction –by a predicative adjunction; for
in «Canis est quadrupes», «est cuadrupes» is predicated of «canis», so that
it is only true if «canis» stands for a dog, but it would be falsely attributed
to the constellation and to the sea-animal. And in any case, in order to
determine the truth-value of the statement, it is necessary to determine
what the subject-term «canis» stands for28.
Nevertheless, Incertus SF accepts that the signification of «canis» can
be restricted to only dogs by the immediate adjunction of a qualification
–for instance, by the immediate adjunction of «quadrupes» as in «canis
quadrupes». Consequently, he claims that it is possible to determine the
force of the expression (virtus sermonis), irrespective of whether it is from

ratione, et sic illud antecedentem non habet unitatem, et ideo huiusmodi consequentia
est distinguenda et universaliter omnis consequentia ubi antecedens non est unum et
consequens est unum est distinguenda; sed ita est in proposito quare et cetera.»
See also EBBESEN, «Can Equivocation be Eliminated»; and ID., «The Dead Man
is Alive», Synthese, 40 (1979) 43-70.
Incertus SF, Quaest. sup. SE, q. 55 (ed. EBBESEN, p. 127:29-38): «Dico ad hoc
quod quicquid sit de determinatione immediata, dico quod determinatio mediata non
tollit aequivocationem […]; sed in talibus propositionibus in quibus est praedicatum
pertinens subiecto pro aliquo eius significato cadit distinctio, sicut patet hic “canis
currit”: li currit enim non competit nisi pro uno significato, tamen ista est distinguenda,
quare in talibus tales determinationes non terminant aequivocationem termini
aequivoci. Et hoc patet per rationem quia aliquid tale quod ei attribuitur mediate, non
attribuitur ei ut idem sed ut praedicatum habens rationem alterius extremi.»

the part of the speaker or from the part of the listener –«canis» can be
absolutely narrowed by the adjunction of «quadrupes», so as to produce
a univocal statement that contains «canis quadrupes» as a subject29.
However, he claims not to have a good logical reason to support his
position, except for an argument of authority that he takes from Manlius
Boethius. In De divisione, M. Boethius puts forth the statement «Canna
romanorum sanguine sorduit», where the term «canna» can represent both
a river and a reed. M. Boethius disambiguates (determinat) the statement
by adding a qualification to the term «canna» –the demonstrative pronoun
«hic», as in «Hic canna Romanorum sanguine sorduit»30. The adjunction
of «hic» makes that «canna» cannot be taken to represent a reed, for
«canna (reed)» is a feminine noun and «hic» a masculine demonstrative.
However, Incertus SF remains silent about the fact that M. Boethius also
accepts a qualification in the predicate as another way of disambiguating
the statement, as for instance in «Canna Romanorum sanguine plenus
At any rate, Incertus SF’s position is that the equivocal term according to
its force actually represents all the significates to which it has been imposed,
unless its signification is narrowed by the adjunction of an immediate

Incertus SF, Quaest. sup. SE, q. 55 (ed. EBBESEN, pp. 127:51-128:65): «[E]t
ideo sic in proposito loquendo de termino aequivoco secundum virtutem sermonis
dicendum est absolute quod determinatur et non solum secundum usum utentis. Quod
patet quoniam recipimus hanc veram “canis latrabilis currit”, sed si de virtute sermonis
non determinaretur li canis, tunc incompossibilia et contradictoria importaret, […] Et
ideo dicendum quod potest determinari.»
Manlius Boethius, De divisione (ed. MAGEE, p. 46:18-25): «Dividitur
qualibet adiectione quae terminet, vel generis vel casus vel alicuius articuli; ut cum
dico “Canna Romanorum sanguine sorduit” et calamum demonstrat et fluvium,
sed dividimus sic: articulo quidem, ut dicamus “hic Canna Romanorum sanguine
plenus fuit”; vel genere, ut “Canna Romanorum sanguine plenus fuit”; vel casu vel
numero, in illo enim singularis tantum est, in illo pluralis, et de aliis quidem eodem
modo.» Note that just some lines before M. Boethius also puts forth the possibility
of disambiguation through definition; cf. Id., De divisione (ed. MAGEE, p. 46:14-18):
«Dividuntur autem significationes aequivocarum secundum aequivocationem unius
particulae orationum definitione, ut cum dico “homo vivit” intellegitur et verus
intellegitur et pictus; dividitur autem hoc modo: “animal rationale mortale vivit”,
quod verum est, “animalis rationalis mortalis simulatio vivit”, quod falsum est.» Note
also that in both cases what is at stake is not the disambiguation of the equivocal
term per se, but rather the disambiguation of the whole statement so as to univocally
determine its truth-value.

qualification, just as the authorities show31. Hence, «Canis est, ergo latrabile
est» is not a valid inference, but «Canis quadrupes est, ergo latrabile est» is
valid, because the truth-value of the antecedent can be univocally determined
through the adjunction of «quadrupes». If we allowed the statement «Canis
est» to be a case of semantic opacity because of equivocation –a statement
where the truth-value of the statement cannot be determined because of the
equivocation of «canis»–, we could also claim that Incertus SF’s treatment of
the situation is along the lines of formal semantics, in that it makes the value
of «canis» depend on the value of «quadrupes», so as to make it possible to
interpret «Canis quadrupes est» as univocally true32. It is noteworthy, though,
that this is not strictly speaking an Aristotelian concern. Although Aristotle
accepts indeed the possibility for the respondent to unveil equivocation
through some tests such as substitution of the definition, he is not concerned
with the absolute determination of a statement’s truth-value. And this is
perhaps the reason why this question is not raised in earlier commentaries
–e.g. Kilwardby’s and Albert’s– which are more concerned with the faithful
interpretation of the Aristotelian doctrine than with exhausting the discussion
of a logical problem.
Incertus SF represents the second stage of a discussion that takes a
more pragmatic turn towards the end of the 13th century33. For there is a
group of commentators who reject Incertus SF’s treatment of the problem
of inference from ambiguous statements and who instead take a position
that is closer to Aristotle’s pragmatic approach to equivocation in the

Incertus SF, Quaest. sup. SE, q. 55 (ed. EBBESEN, p. 128:60-79): «Et ideo
dicendum quod potest determinari. […] sed cum ei additur aliquid quod natum est ei
convenire pro uno significato, tunc auctores utuntur eo pro uno significato, ut patet in
auctoribus iam adductis. Et ideo si ex alio non potest accipi probatio significationis nisi
ex usu auctorum, nec impositionis, tunc dicemus quod est ratio impositionis termini
aequivoci quod imponitur ad significandum omnia significata cum per se accipitur
sine determinatione […] cum autem sumitur cum aliqua determinatione tali, significat
tantum illud unum; nec est alia ratio quaerenda nisi quod sic utuntur auctores.»
Formal semantics, however, has been concerned with referential opacity, as it
is for instance the case in quantified statements or in statements that contain relative
pronouns. But we can see, I believe, the case of «Canis quadrupes est» as an extreme
case, where the referential opacity is primarily caused by the equivocation of the subject.
See also C. MARMO, «A Pragmatic Approach to Language in Modism», in
S. EBBESEN (ed.), Sprachtheorien in Spätantike und Mittelalter, Gunter Narr Verlag,
Tübingen 1995, pp. 169-183; MARMO, Semiotica e linguaggio, pp. 346 sqq.; and ID.,
La semiotica del XIII secolo, Bompiani, Milano 2012, pp. 126-132.

Sophistical Refutations. These authors are Radulphus Brito, Thomas of

Wyk and the Anonymus Pragensis.
To the question whether an equivocal term represents its significates
conjunctively or disjunctively, they answer, just as the Incertus SF, that it
does not represent them in either way, even though it actually represents all
of them at the same time because of its several impositions34. They also agree
that the question about the validity of the inference «Canis est, ergo latrabile
est» cannot be answered with either yes or no, because one ought not give
a simple answer to a multiple question35. The inference «Canis est, ergo
latrabile est» is ambiguous, because the antecedent has an equivocal subject,
and the validity of an inference depends on the truth of the antecedent, a truth
that cannot be determined precisely because the antecedent is ambiguous.
Hence, along the lines of Aristotle’s advice, they propose not to give a yes or
no reply, but to make explicit the antecedent’s ambiguity36. More interesting,

See e.g. Thomas of Wyk, Fallaciae, ed. S. EBBESEN in CIMAGL, 68 (1998)
139-143, at p. 141: «Verumtamen terminus aequivocus actu omnia sua significata
repraesentat sub propria ratione […]; una enim impositio aliam tollit. Et hoc innuit
Commentator cum dicit quod terminus aequivocus unum significat ac si aliud non
significaret, per hoc innuens quod nulla habitudine respicit sua significata, nec
copulative nec disiunctive.»
Cf. SE c. 17 and Int 8.18a13-27.
Anonymus Pragensis, Quaestiones super Aristotelis Sophisticos Elenchos, q.
14, ed. D. MURÈ, in CIMAGL, 68 (1998) 63-97, at pp. 81-82: «Ad hoc solvendum,
est notandum quid sit consequentia. Et est nihil aliud quam habitudo consequentis
ad antecedens, quae principaliter fundatur in antecedente. […] Secundo dico quod si
subiectum est univocum, et passio est univoca, et si subiectum aequivocum, et passio
aequivoca […] cum igitur subiectum in quo fundatur consequentia sit aequivocum,
passio haec, scilicet consequentia, est aequivoca. Et quia ad quaestionem plures non est
danda una responsio, ideo cum quaeritur utrum sequitur […] non debet dici simpliciter
quod sequitur nec […] quod non sequitur; sed debet distingui quod, si «canis» in
antecedente supponit pro latrabili, tunc est bona consequentia; si vero «canis» accipitur
pro significatis aliis, non est bona consequentia.» See also Radulphus Brito, Quaest.
sup. SE, q. I.15 (ed. S. EBBESEN, in CIMAGL, 68 (1998) 185-227, at p. 205 p. 205):
«Non sequitur: modo in ista consequentia antecedens potest esse verum et consequens
falsum; ideo etc. […] Ad istam quaestionem dico […] quod non est respondendum
unica responsione, immo ista est distinguenda, et debet dici quod pro uno significato
est vera et pro alio est falsa; […] omnis intentio fundata in aliquo obiecto habet
multiplicitatem illius obiecti; ergo cum consequentia sit quaedam intentio fundata in
aliquo obiecto complexo mediante habitudine terminorum illius complexi, secundum
multiplicitatem terminorum in tali complexo positorum multiplicatur consequentia.

though, is their position as regards the resolution of ambiguous statements;

for, contrary to Incertus SF, they reject the idea that the adjunction of a term
–whether mediate or immediate– can determine the significate of an equivo-
cal term so as to make the statement univocal.
It is their position that an equivocal term represents actually and at
the same time all the significates on which it was imposed every time
it is uttered. However, they reject the idea that there can be an absolute
disambiguation of the equivocal term –a disambiguation that does not result
from the determination of a specific significate from the part of the listener.
The significates of an equivocal term cannot be narrowed by adjunction
absolutely –simply through the adjunction of a qualifying term– because
the equivocal word, whenever it is uttered, represents by its very essence
and actually all of its significates37. Nonetheless, the speaker can make
clear her intended significate with a qualification of the equivocal term, so
that the listener, by virtue of his charitable understanding (i.e. because he
wants the act of communication to succeed), can take the term to represent
the significate that is coherent with the qualifying term38. Therefore, we

[…] modo omnis consequentia multiplex est distinguenda; ergo talis consequentia est
distinguenda, et debemos dicere quod pro uno significato est vera et pro alio falsa […]»
See e.g. Thomas of Wyk, Fallaciae (ed. EBBESEN, p. 141): «Intelligendum est
etiam quod terminus aequivocus non potest contrahi per mediatum nec immediatum
adiunctum, quia contractio proprie est communis potentia repraesentantis multa, et hoc
per aliud magis determinatur; sed in termino aequivoco nulla significatio est communis
quae contrahi possit, […] terminus aequivocus actualiter et non potentialiter respicit
sua <significata>, ideo in aequivocis non est proprie contractio. Verumtamen sic
dicendo «canis latrabilis» virtute huius expressi pertinentis ad alterum significatum
termini aequivoci inducitur intellectus audientis magis <ad> apprehendendum unum
significatum quam aliud. Verumtamen de virtute sermonis non coartatur ad unum
significatum tantum, sed omnia adhuc repraesentat.»
Anonymus Pragensis, Quaest. sup. SE, q. 16 (ed. MURÈ, p. 88): «Ad hoc dicendum:
cum quaeritur utrum aequivocatio possit determinari per adiunctum, potest intelligi
dupliciter: vel de virtute sermonis vel de bonitate intelligentis. Si quaestio quaerat utrum
aequivocatio determinetur per adiuncta virtute sermonis, dicendum quod non; et huius
ratio est, quia in secundo huius dicit Philosophus: si dicatur “Coriscus est musicus”
distinguenda est; et si dicatur “hic Coriscus est musicus”, adhuc est distinguenda; et
tamen hic est determinatio, scilicet “hic”. […] Sed quantum est de bonitate intelligentis,
dicendum quod determinatio distinguit terminum aequivocum; et huius ratio est,
quoniam illud est possibile de bonitate intelligentis, quo accepto, nulla repugnantia
sequitur. Sed nulla repugnantia sequitur quod intellectus, accepto uno significato, non
recipit reliquum in termino aequivoco. Ergo terminus aequivocus distinguitur quantum

could suppose that to the question whether «Canis quadrupes est, ergo
latrabile est» is a valid consequence, they would answer that, despite the
adjunction of «quadrupes», the question is still multiple, unless «cannis» in
the antecedent is disambiguated in a way that could go like this:

Speaker: Is «Canis quadrupes est, ergo latrabile est» a valid

Listener: That depends on what is the significate of «canis». It is
neither a constellation nor a sea-animal, right?
Speaker: Right!
Listener: Ok, then it is a valid inference.

Thus, the determination of the antecedent’s truth-value and the

determination of the inference’s validity cannot result from a formal
semantic analysis of the antecedent. They can only result from a
collaborative determination that depends partly on the listener’s recognition
of the speaker’s intended value. I also submit that their position is closer
to Aristotle’s general attitude towards equivocation in his dialectical
treatises, even though their analyses rely heavily on the notion of linguistic
imposition, a notion that is alien to the Aristotelian treatises themselves.

Is «Laborans sanus est» to be disambiguated?

I would like to finish with the discussion of another problem regarding

resolution of equivocation –the one related to the question of whether the
sentence «Laborans sanus est» ought to be disambiguation39. The problem

est de bonitate intelligentis.» See also Radulphus Brito, Quaest. sup. SE, q. I.17 (ed.
EBBESEN, pp. 213-214): «Dico tamen duo ad istam quaestionem: primo quod terminus
aequivocus de virtute sermonis non potest contrahi ad alterum significatum solum per
aliquam determinationem sibi adiunctam; secundo dico quod de bonitate intelligentis
potest contrahi vel determinari per aliquam determinationem sibi adiunctam. Primum
declaratur sic: quia omne illlud quod est determinabile per aliud est in potentia ad illud
per quod determinatur, sicut dicendo “homo albus” “homo” ibi determinatur per “albus”
quia est in potentia ad albedinem; modo terminus aequivocus non est in potentia ad
sua significata […] Secundo declaratur sic: quia aliquis potest intelligere terminum
aequivocum pro uno significato ita quod non pro alio, et ita potest ipsum intelligere sumi
pro illo significato ita quod non pro alio, et sic de bonitate intelligentis potest contrahi ad
alterum eius significatum per aliquam determinationem adiunctam sibi.»
For other discussions of «Laborans sanus est» see S. EBBESEN, «Les grecs
et l’ambigüité», in I. ROSIER (ed.) L’ambigüité: Cinq études historiques, Presses

lies, first of all, in the nature of the present participle «laborans»; for it is
not evident at all whether it is equivocal, and if so, why it is equivocal.
The problem arises from the description of the third type of equivocation
that we find in British authors such as Robert Kilwardby and William of
Sherwood, according to whom there is equivocation of the third type
when a term represents only one thing in itself, but many when it is joined
to another term. The paradigmatic example that is given is the present
participle «laborans», which can co-signify both the present and the past
when it is joined to a verb in the imperfect tense, but only the present when
it is taken by itself. But according to the standard definition, equivocation
takes place when a term has several significates actually and in itself, and
not when it has one actually and in itself and several potentially (and only
actually in the context of a statement). Kilwardby and Sherwood go about
this objection by stressing that the action denoted by the present participle
is primarily found in present things, but that because of its adjunction to a
verb in the imperfect tense, the present participle can derivatively denote an
action in the past. Therefore, the statement «Laborans sanus est» does not
need to be disambiguated, because the present participle with the present
tense of a verb retains its main co-significate –the present40.

Universitaires de Lille, Lille 1988, pp. 15-32; and S. EBBESEN – I. ROSIER, «Robertus
Anglicus on Peter of Spain», in P. PÉREZ-ILZARBE – I. ANGELELLI (eds.), Medieval and
Renaissance Logic in Spain: Acts of the 12th Symposium on Medieval Logic and
Semantics, Olms, Hildesheim 2000, pp. 61-95.
See e.g. William of Sherwood, Die Introductiones in logicam des Wilhelm von
Shyreswood, ed. M. GRABMANN, Verlag der Bayerischen Akademie der Wissenschaften,
München 1937, at pp. 87:36-88:34: «Exemplum tertii. Quisquis sanabatur, sanus est.
Laborans sanabatur. Ergo laborans sanus est. Iste enim terminus laborans nec de se nec
transumptive dat intelligere nisi praesens, sed ex coniunctione cum hoc verbo sanabatur
dat intelligere praeteritum. […] Contra ultimum modum sic. Dictio est prior unaquaque
oratione, retinet ergo esse dictionis antequam ingrediatur orationem. Hoc autem habet ex
sua significatione. Suam ergo significationem habet antequam ingrediatur orationem et non
ex ordinatione sui in oratione. Laborans ergo, cum ex se significet unum, non significabit
aliud ex ordinatione sui cum alio. […] Dicendum quod necesse est dictionem habere
significationem ante orationem et ab ea nulla potest habere. Verumtamen significatio, quam
habet ex adiuncto suo, poterit permutari et hoc est non in omni dictione, sed in illa, cuius
significatio vel consignificatio est una intentio participata a pluribus secundum prius et
posterius. Et tunc illa dictio significabit de se illud, quod primo participat illa intentionem;
ex adiunctione autem [aut Grab.] potest significare illud, quod posterius eam participat.
Sic est hic. Haec dictio enim laborans consignificat praesens, quod primo salvatur in
praesenti simpliciter, per posterius autem et diminute in praesenti de praeterito.»

Nicholas of Paris and Peter of Spain give another description of the third
type of equivocation, which eludes the problem in Kilwardby and Sherwood’s
description. According to Nicholas and Peter, equivocation of the third type
happens when a term has several co-significations actually and in itself, as it
is the case of the present participle, which actually denotes the present, the
past and the future41. This is more or less the description adopted by Incertus
SF42, the Anonymus Pragensis43, Thomas of Wyk44 and Radulphus Brito.

Nicholas of Paris, Not. sup. lib. Elench. (ed. EBBESEN, p. 173): «Ad aliud
dicimus quod sicut dicit Priscianus participia praesentis temporis confundunt omne
tempus, praesens sc. praeteritum et futurum. Unde secundum confusionem aequivoca
sunt et tempora diversa consignificant de se. Dicimus ergo, cum sic dicitur “laborans
sanabatur”, li laborans ab illo verbo “sanabatur” non accipit consignificationem
plurium temporum; aut si accipit, non accipit in quantum signum sive significans, ex
quo causatur aequivocatio, sed forsitan accipit modum supponendi, et ita ut supponens
et non ut significans habet ab illos consignificationem plurium temporum, a se vero
habet ut sit signum et non ut supponat.» Cf. Priscian, Institutiones XI.VII.25.
Incertus SF, Quaest. sup. SE, q. 53 (ed. EBBESEN, p. 119:200-220): «Et ideo
falsum dicunt quando ponunt quod in tertio modo aliquid significet unum de se et habet
aptitudinem ad aliud ex adiunctione […] et haec ratio potest haberi a Commentatore.
Et iterum cum dicunt quod participium desinens in -ans vel in -ens de se est praesentis
temporis et ex alio habet quod sit praeteriti, hoc falsum est, quia secundum Priscianum
tale participium significat praesens confusum, quod quid congregatum est ex praeterito
et futuro, quare pro omni potest accipi […] Et ideo dicendum quod haec “laborans
sanus est” distinguenda est sicut et alia, cum li laborans per se significet utrumque
tempus actu ex impositione, sicut ostendunt rationes primo adductae.»
Anonymus Pragensis, Quaest. sup. SE, q. 19 (ed. MURÈ, p. 96): «Ad istam
quaestionem dicunt aliqui quod ista propositio “laborans sanus est” non distinguitur,
et huius rationem assignant, quia li laborans praesens tempus designat <actu> et
praeteritum designat in potentia, et non tenetur pro ipso praeterito nisi cum sibi additur
determinatio praeteriti temporis, cum dicitur laborans sanabatur; sed cum dicitur
“laborans sanus est” tenetur pro praesenti, ideo non est distinguenda. Sed haec positio
non valet, quia <si> iste terminus importaret praesens actu et praeteritum in potentia,
iam non faceret multiplex actuale, quod est contra fallaciam aequivocationis […]
Propter quod dicerunt alii quod “laborans” importat utrumque tempus, tamen unum
per prius, sc. praesens, reliquum per posterius, sc. praeteritum. […] Haec opinio iterum
nulla est […] quia sic iste modus tertius non differret a secundo modo.»
Thomas de Wyk, Fallaciae (ed. EBBESEN, p. 143): «Circa tertium modum aequi-
vocationis primo assignatum arguitur: Si “laborans” tantum de se consignificat unum
tempus et a<liu>d haberet ex adiuncto, multiplicitas igitur in hac “laborans sanabatur”
non esset in hac dictione “laborans” sed in tota oratione primo, quod <est> contra rationem
aequivocationis, quia aequivocatio est in dictione et non in oratione primo. Ad hoc dicendum
quod ei quod est “laborans” non confertur modus significandi sive consignificatum ex

On the one hand, all these authors reject the description of the third type of
equivocation in terms of the adjunction of another term for two main reasons:
first, because it entails that the ambiguity of «Laborans sanabatur» stems
from the verb «sanabatur» and not from the equivocal term «laborans», since
what produces the multiplicity of signification is the imperfect tense; second,
because if «laborans» were not actually equivocal, but only potentially, it
would not be equivocal at all; for equivocation requires the actual multiplicity
of signification. On the other hand, to the question about the need of
disambiguating the statement «Laborans sanus est», they reply with the
affirmative: first, because irrespective of the tense of the verb, «laborans»,
because of its imposition, always denotes the past and the present; second
because this multiple co-signification, which belongs to the very essence of
«laborans», cannot be narrowed by the adjunction of an accidental predicate45.
Radulphus Brito stresses in addition to this that the resolution needed
here is not a resolution of ambiguity due to amphiboly, but a resolution
of ambiguity due to equivocation. He stresses this because of a problem
that was discussed at least from Nicholas of Paris onwards. If multiple
co-signification of time is a cause of equivocation, why the same is not the
case for the multiple co-signification of number, of case or of gender? The
British tradition considers a term like «episcopi» as equivocal, because
it can represent both the nominative plural and the genitive singular, so
that the statement «Isti asini sunt episcopi» is ambiguous because of
the equivocation of case and number of «episcopi». In the Continental
tradition, on the contrary, case and number are not causes of equivocation,
so it is wrong to consider «episcopi» as an equivocal term46. Nicholas and

adiuncto, sicut supponitur in ratione, sed “laborans” duo tempora consignificat, praesens
et praeteritum. Utrumque sibi conceditur ex impositione, diversimode tamen, quia praesens
tempus sibi conceditur ubicumque ponitur respectu cuiuscumque verbi, aliud sibi conceditur
respectu verbi convenientis. Unde non est imaginandum quod unum habeat ex se et aliud trahat
ex adiuncto, et secundum hanc imaginationem procedunt rationes ad hanc quaestionem.»
See e.g. Anonymus Pragensis, Quaest. sup. SE, q. 19 (ed. MURÈ, p. 96): «Ideo
dicendum quod haec propositio “laborans sanus est” distinguenda est: cuius ratio est,
quia posita causa per se, ponitus effectus per se; sed causa per se distinctionis est
quod “laborans” importat duo tempora, […] ; ergo quaecumque dictio sibi coniugatur,
semper haec duo tempora significat, nam accidentale non transmutat essentiale; sed
impositio est essentialis in termino et additio […] est accidentalis; ergo non potest tolli
significatum eius per illius termini additionem […]»
See EBBESEN – ROSIER, «Robertus Anglicus on Peter of Spain» for another
discussion of the case «episcopi».

Brito provide us with the following explanation: While co-signification of

number, case and gender are causes of the constructibility of a term with
another term so as to produce a grammatical sentence, co-signification
of time is not the cause of grammatical constructibility. The syntactic
multiplicity of terms with a multiple co-signification of number, gender,
etc., is the cause of fallacies due to amphiboly or to form of the speech
and not the cause of fallacies due to equivocation. The ambiguity of the
sentence «Laborans sanus est» does not stem from a syntactic multiplicity
of the sentence, but from the semantic multiplicity of «laborans» that comes
from its multiple co-signification of time. Hence, the sentence needs to be
disambiguated as a case of equivocation, and not as a case of amphiboly47.
And just as in the case of «Canis currit», this disambiguation cannot
be due to the adjunction of a predicate as «sanus est». Incertus SF would
probably agree that the sentence can be disambiguated by the immediate
adjunction of a term to the equivocal term, as in «Laborans nunc sanus est».
A solution that cannot be accepted by the Anonymus Pragensis, Thomas
of Wyk and Radulphus Brito, as it was shown in the case of the inference
«Canis est, ergo latrabile est».
In conclusion, the Anonymus Pragensis, Thomas of Wyk and
Radulphus Brito put forth an analysis of the fallacy of equivocation that
is remarkably Aristotelian –and pragmatic– in that the disambiguation
of a statement containing an equivocal term can only be the result of a
collaborative effort between the interlocutors ; effort that fundamentally
includes an inference by the listener about the intended meaning of the

See e.g. Radulphus Brito, Quaest. sup. SE, q. I.19 (ed. EBBESEN, pp. 224-225):
«Omnis oratio in qua est aliquod multiplex est distinguenda […] multiplicitas quae
est in oratione ex parte dictionis absolute consideratae est multiplicitas secundum
aequivocationem; modo dicendo “laborans sanus est” in ista oratione est multiplicitas
ex parte dictionis absolute consideratae et secundum se […] si enim aliqua dictio
habeat diversos modos significandi respectivos qui sint principium referendi ipsam ad
aliam dictionem, tunc habet esse multiplicitas amphibolica, […] ibi est multiplicitas ex
parte temporis; modo tempus non est modus significandi respectivus qui sit principium
uniendi unum constructibile cum alio constructibili; et ideo ista est distinguenda
secundum aequivocationem. Tertium declaratur, quia tertius modus aequivocationis est
quando aliqua dictio secundum se significat unum, coniunctum autem alteri significat
plura; […] et si ex coniunctione modi significandi cum significato <laborans> importat
plura; ergo etc. […] numerus et persona vel casus vel genus non propter hoc redditur
talis dictio aequivoca.»



Steno Ebbesen magistro sapientissimo carissimoque

1. Introduction

TAXONOMY MATTERS: ARISTOTLE. If the interest an author takes in a subject

can be measured at all, the educated guess would be that it is commensurate
with the attention he devotes to it. It thus seems a fair assessment to say
that –when handling sophistical topics– classificatory issues were a matter
of great concern to Aristotle. For one thing, he discussed at length and
eventually discarded at least one competing classification of fallacies
according to which these are to be differentiated depending on whether they
aim at the thought or at its verbal expression1. For another, he remarkably
engineered his own taxonomy. An all-embracing genus, the ignorance of
what a refutation is2, encompasses two comprehensive species –one dwelling

UMR 8163 «Savoirs, Textes, Langage» (STL), Université Lille 3 Rue du
Barreau, 59650 Villeneuve-d'Ascq. Email: leone.gazziero@univ-lille3.fr.
Aristotelis sophistici elenchi, Ed. by D. ROSS, Clarendon Press, Oxford 1958,
10, 170b 12-16: «οὐκ ἔστι δὲ διαϕορὰ τῶν λόγων ἣν λέγουσί τινες, τὸ εἶναι
τοὺς μὲν πρὸς τοὔνομα λόγους, ἑτέρους δὲ πρὸς τὴν διάνοιαν· ἄτοπον γὰρ τὸ
ὑπολαμβάνειν ἄλλους μὲν εἶναι πρὸς τοὔνομα λόγους, ἑτέρους δὲ πρὸς τὴν
διάνοιαν, ἁλλ’ οὐ τοὺς αὐτούς [contrary to what some say, there is no distinction
between arguments aiming at the word and arguments that aim at the thought. In fact, it
is absurd to think that arguments aim either at the word or at the thought and that they are
not the same]». The issue is brilliantly discussed in M. HECQUET-DEVIENNE, «La pensée
et le mot dans les Réfutations sophistiques», Revue philosophique, 2 (1993) 179-196.
Aristotelis sophistici elenchi 6, 169a 18-21: «πάντες οἱ τ<ρ>όποι πίπτουσιν εἰς
τὴν τοῦ ἐλέγχου ἄγνοιαν, οἱ μὲν οὖν παρὰ τὴν λέξιν, ὅτι ϕαινομένη ἀντίϕασις,
ὅπερ ἦν ἴδιον τοῦ ἐλέγχου, οἱ δ’ ἄλλοι παρὰ τὸν τοῦ συλλογισμοῦ ὅρον [all
fallacies fall under the ignorance of what a refutation is. Those depending on expression
because the contradiction –which is the distinctive feature of refutation– is apparent
only; the others because they violate the definition of the deduction]». Cf. Aristotelis

on language, the other not3– whose twelve subsets4 account for all kinds of
failures to meet either one of the two requirements a refutation ought to
meet in order to do its job, which is to establish a real contradiction by
means of a genuine deduction5. Aristotle even set himself to prove (both by
way of induction and deduction) that his sixfold division of fallacies which
have to do with expression is exhaustive: no fallacy involving linguistic
features has been neglected and none falls outside those he mentioned6.

sophistici elenchi 6, 168a 17-20: «ἢ δὴ οὕτως διαιρετέον τοὺς ϕαινομένους

συλλογισμοὺς καὶ ἐλέγχους, ἢ πάντας ἀνακτέον εἰς τὴν τοῦ ἐλέγχου ἄγνοιαν,
ἀρχὴν ταύτην ποιησαμένους· ἔστι γὰρ ἅπαντας ἀναλῦσαι τοὺς λεχθέντας
τρόπους εἰς τὸν τοῦ ἐλέγχου διορισμόν [apparent deductions and refutations must
be classified either the way we did or be reduced to the ignorance of what a refutation
is, which we acknowledge as their origin. In fact, it is possible to show that all the
aforesaid fallacies neglect one aspect <or another> of the definition of the deduction]».
Aristotelis sophistici elenchi 4, 165b 23-24: «τρόποι δ’ εἰσὶ τοῦ μὲν ἐλέγχειν
δύο· οἱ μὲν γάρ εἰσι παρὰ τὴν λέξιν, οἱ δ’ ἔξω τῆς λέξεως [there are two ways of
refuting: one has to do with expression, the other is independent of it]». As is well
known, Aristotle’s classification of fallacious reasoning rests upon the alternative
whether linguistic features play a role or not. Here λέξις means everything that has
to do with the way we talk about things: written or spoken words may be equivocal
(homonymy, figure of speech and accent exploit their ambiguous features), turns of
phrase may be equivocal too (amphiboly, composition and division –on the other
hand– take advantage of their syntactical arrangements).
Equally distributed within and outside verbal expression: «ὁμωνυμία,
ἀμϕιβολία, σύνθεσις, διαίρεσις, προσῳδία, σχῆμα λέξεως [homonymy, amphiboly,
composition, division, accent, form of expression]» (Aristotelis sophistici elenchi 4,
165b 24-27) and «παρὰ τὸ συμβεβηκός, […] τὸ ἁπλῶς ἢ μὴ ἁπλῶς ἀλλὰ πῂ ἢ
ποὺ ἢ ποτὲ ἢ πρός τι λέγεσθαι, […] τὸ παρὰ τὸ ἑπόμενον, […] τὸ παρὰ τὸ <τὸ>
ἐν ἀρχῇ λαμβάνειν, […] τὸ <τὸ> μὴ αἴτιον ὡς αἴτιον τιθέναι, […] τὸ τὰ πλείω
ἐρωτήματα ἓν ποιεῖν [the fallacy of accident; the fallacy in which an expression is
either said without qualification or not without qualification but with some qualification
related to manner, place, time or relation; the fallacy of assuming the point to prove;
the fallacy that states that something is cause without it being one; the fallacy that ask
multiple questions as one]» (Aristotelis sophistici elenchi 4, 166b 21-27) respectively.
Aristotelis sophistici elenchi 1, 164b 27-165a 3: «ὁ μὲν [165a] γὰρ συλλογισμὸς
ἐκ τινῶν ἐστι τεθέντων ὥστε λέγειν ἕτερον ἐξ ἀνάγκης τι τῶν κειμένων διὰ
τῶν κειμένων, ἔλεγχος δὲ συλλογισμὸς μετ’ ἀντιϕάσεως τοῦ συμπεράσματος
[deduction occurs when something new is necessarily involved by what has been
previously stated. The refutation is a deduction which contradicts what has been
presented as a conclusion]».
Aristotelis sophistici elenchi 4, 165b 28-29: «τούτου δὲ πίστις ἥ τε διὰ τῆς
ἐπαγωγῆς καὶ συλλογισμός [This can be proved by both induction and deduction]».


topic was popular enough for him to pride himself on having successfully
dealt with the rationale behind Aristotle’s claim for completeness whilst other
interpreters had failed to explain it7. Commentators will not forget Galen’s
lesson and the issue will be addressed as a matter of routine by Byzantine and
Latin scholars alike. Especially the latters placed considerable emphasis on
the issue of exhaustiveness and spared no effort in order to describe in detail
the architecture of Aristotle’s classification. Actually, from the very start,
Latin commentators stressed the fact that Aristotle’s taxonomy is a coherent,
principle-ruled, derivational system8. They also spent considerable time

Galeni de captionibus in dictione, Ed. by S. EBBESEN, Commentators and
Commentaries on Aristotle’s Sophistici elenchi, Brill, Leiden 1981, II, pp. 6.22-7.4:
«τῶν οὖν ἐξηγησαμένων [7] αὐτὸν οἱ μὲν οὐδ’ ἐπεχείρησαν ταῦτ’ ἀκριβῶσαι τὸν
προσήκοντα τρόπον, οἱ δ’ οὐκ ἔτυχον. ἡμεῖς δὲ πειραθῶμεν, οὐκ Ἀριστοτέλους
ἕνεκεν οὐδ' ὡς τῷ λόγῳ βοήθειάν τινα πορίζοντες, ἀλλ’ ἡμῶν αὐτῶν [some
commentators did not even try go give a precise account of Aristotle’s way, others did
not succeed. Let’s try, neither for Aristotle’s sake nor for the sake of the text, but for
ourselves]». Galen managed at best to keep only half of his promise. As far as I know,
no one – Valentina di Lascio excepted – has ever fulfilled the other half and delivered the
real McCoy, that is the key to understand Aristotle’s « proof through syllogism ». That
she did in a remarkable essay, namely «The Theoretical Rationale behind Aristotle’s
Classification of the Linguistic Fallacies in the Sophistical Refutations», Logical analysis
and history of philosophy, 15 (2013) 55-89.
Four texts from early Latin literature will illustrate the point. 1. «ignorantia
elenchi dicitur esse principium et origo omnium fallaciarum, […]. Ad quam quidem
ignorantiam omnes redargutionis species rediguntur [we say that the ignorance
of what a refutation is is the principle and the origin of all fallacies [...]. In fact, all
their species come down to the ignorance of what a refutation is]» (Anonymi summa
sophisticorum elenchorum, Ed. by L. M. DE RIJK, Logica Modernorum, vol. I, Van
Gorcum, Assen 1962, p. 416.16-18). 2. Anonymi Aurelianensis I commentarium in
Sophisticos elenchos, Ed. by S. EBBESEN, Cahiers de l’Institut du Moyen-Âge Grec et
Latin (CIMAGL), 34 (1979) 162.27-163.6: «convenienter facta est superius fallaciarum
divisio, ergo aut sic est dividendum apparentes syllogismos, ut supra divisimus,
aut reducendum omnes in ignorantiam elenchi, ut scilicet dicamus quod in omni
paralogismo est ignorantia elenchi et omnes fallacia sub illa specie est coartandum.
[163] Non autem dico quod illi qui primam tenant divisionem sic debeant dividere, sed
his, id est ab his, est reducendum omnes fallacias sub ignorantia elenchi. Vel his, id
est secundum hos, qui faciunt hanc, id est constituunt ignorantiam elenchi principium
omnium fallaciarum, id est quasi genus omnium [the division Aristotle introduced by

trying both to single out each acknowledged kind of fallacious reasoning and
to list its various moods or subsets. Accordingly, the question «in how many
varieties a given fallacy comes (quot modis varietur)» received at least as

REFUTATION IS”, that is: as if we said that every paralogism has to do with the ignorance
of what a refutation is and that one must subsume every fallacy under ignorance as
one of its species. I do not claim that those who accept the first classification have
to make a division according to the second classification, but “THOSE”, that is: they
must subsume every fallacy under the ignorance of what a refutation is; or “THOSE”,
that is: for those “WHO DO THAT”, that is: those who make the ignorance of what a
refutation is the principle of all fallacies altogether, as if it were the genus of them
all]». 3. Anonymi Aurelianensis II de paralogismis, Ed. by S. EBBESEN, CIMAGL, 16
(1976) 77.20-21: «omnia genera fallaciarum ad hanc tamquam ad unum genus reduci
possunt [all the kinds of fallacy may be reduced to the fallacy of the ignorance of what
a refutation is as if it were their only genus]». 4. Anonymi fallacie londinenses, Ed.
by L. M. DE RIJK, Logica modernorum, vol. II, Van Gorcum, Assen 1967, p. 672.4-
8: «fallacia secundum ignorantiam elenchi est deceptio proveniens ex obmissione
aliquorum quae observanda sunt in descriptione elenchi. Et secundum hoc non aliqua
tresdecim fallaciarum. Unde Aristoteles docet reducere omnes alias fallacias ad hanc
fallaciam [the fallacy of the ignorance of what a refutation is is a deception that arises
when one of the requirements to be satisfied according to the refutation’s definition
has been neglected. In this respect, the ignorance of what a fallacy is does not belong
to Aristotle’s thirteen kinds of fallacies. This is why Aristotle advises that all other
fallacies be reduced to the fallacy of the ignorance of what a refutation is]». That
being said, taxonomic expediency did not prevent Latin authors from raising problems
about the fact that ignorantia elenchi seems to be both inclusive of and included in
the distinction between fallacies that dwell on verbal expression and fallacies who do
not, in so far as Aristotle ranked it amongst the latters. Peter of Spain, for instance, felt
the need to address the issue: «fieri quidem solet duplex distinctio ignorantie elenchi,
secundum quod est una specialis de tredecim fallaciis, et secundum quod est generalis
ad quam omnes tredecim fallacie reducuntur [as far as ignorance of what a refutation
is is concerned, a double distinction is usually made, according to which ignorance of
what a refutation is is both particular and general. Particular in so far as it is one of the
thirteen fallacies and general in so far as it is the fallacy to which all others may be
reduced]» (Petri hispani portugalensis tractatus, Ed. by L. M. DE RIJK, Peter of Spain
(Petrus Hispanus Portugalensis). Tractatus called afterwards Summule logicales, Van
Gorcum, Assen 1972, VII, p. 180.5-7). As did the anonymous author of the Fallaciae
ad modum Oxoniae, Ed. by C. R. KOPP, Köln Universität, Köln 1985, p. 128, Albert
the Great (Alberti magni expositio sophisticorum elenchorum, Ed. by P. JAMMY, C.
Prost, Lyon 1651, p. 887a), Giles of Rome in his Expositio super libros elenchorum,
Venetiis per Bonetum Locatellum, 1496, 18vb 28-34 and Anonymus agdavensis in
his Quaestiones super Sophisticos elenchus, Ms Anger, Bibliothèque municipale, 418
(405), f. 180ra 24-25.

much attention as the question «what the fallacy itself is (quid sit)»9. Moreover,
Westerners resorted to specific devices in order to solve classificatory puzzles.
Besides the inherited opposition between Form and Matter which was much
used to tell apart sophisms whose mistake is to jeopardize either the identity
of their subject matter or the compelling form of their entailment10, the most
sophisticated tool –and the more innovative too– they developed along the
way was the distinction to be made between the way an argument goes
wrong and the way it fools us. On that ground, they differentiated between
what they called a causa apparentiae (what gives an argument a respectable
appearance) and a causa defectus or non existentiae (the reason why –despite
looking good– it is defective or fails to imply its conclusion)11. All of which

In fact, more often than not, the two questions went hand in hand, as is illustrated
time and again by the use of formulaic repetitions. Cf. e.g. Anonymi Cantabrigiensis
commentarium in Aristotelis sophisticos elenchos, Ms Cambridge, St John’s D.12, f.
85rb («dicatur quid sit aequivocatio, quid fallacia secundum aequivocationem, quot
modis habeat fieri [it ought to be explained what homonymy is, what the fallacy of
homonymy is and in how many varieties it comes]», f. 86rb («circa hanc fallaciam
considerandum quid sit amphibologia, quid fallacia secundum amphibologiam, et quot
modis habeat fieri [about this fallacy one must take into account what amphiboly is,
what the fallacy that dwells on amphiboly is and in how many varieties it comes]»), f.
87va («videndum est ergo quid sit compositio, quid divisio, quid fallacia compositionis,
quid fallacia divisionis, quot modis fiant istae fallaciae [we have to consider what
composition is, what division is, what the fallacy of composition is, what the fallacy
of division is and in how many varieties they come]»), f. 88vb («circa hanc fallaciam
considerantdum erit quid sit figura dictionis, quid fallacia secundum figuram dictionis,
et quot sunt huius fallaciae modi [about this fallacy one must take into account what
figure of speech is is, what the fallacy that dwells on figure of speech is and in how
many varieties it comes]»), f. 89va («videndum est quid sit accidens, quid fallacia
secundum accidens, quot modi paralogismorum fiant secundum hanc fallaciam [we
have to consider what accident is, what the fallacy of accident is and how many kinds
of paralogisms occur according to it]»).
The literature on «logical form» vs. «logical matter» is both extensive in
quantity and varied in quality. Besides J. BARNES, «Logical Form and Logical Matter»,
in A. ALBERTI (ed.), Logica, Mente e Persona, Olschki, Firenze 1990, pp. 16-39,
which is quoted at every turn, a sensible introduction to the problem is to be found
in S. EBBESEN, «The Way Fallacies were Treated in Scholastic Logic», CIMAGL, 55
(1987) 107-134.
The early Dialectica monacensis makes the distinction very clearly when
tackling the fallacy of figure of speech: «fallacia autem figurae <dictionis> est
deceptio proveniens ex similitudine dictionis cum dictione, vel etiam ex diversitate
significatorum vel consiginificatorum. Heae enim sunt causae ipsius moventis, scilicet

makes perfect sense, notably for two reasons: first of all, Aristotle makes it

causa apparentiae ex parte signi et causa falsitatis ex parte significati vel consignificati
[the fallacy of figure of speech arises because of the similarity between expressions
and the diversity between the things these signify or cosignify. Two are, in fact, the
causes that bring about fallacies of this kind, that is: a cause which accounts for their
deceptive appearance, which has to do with words, and a cause which accounts for their
falsehood, which has to do with the things these words signify or cosignify]» (Tractatus
de sophistica argumentatione (Dialectica monacensis), Ed. by DE RIJK, Logica
Modernorum, II, p. 579.9-13). Along the same lines, Peter of Spain established two
sets of synonymous expressions, which will become pretty standard in later literature:
«principium autem motiviun sive causa apparentiae in qualibet fallacia est quod movet
ad credendum quod non est. Principium vero defectus sive causa falsitatis est quod
facit creditum esse falsum [the cause or the principle which produces the deceptive
appearance in every fallacy is what leads someone to believe what is not the case.
The principle of the flaw or the cause which accounts for the fallacy’s falsehood is
what is actually responsible for the falsehood of what one is led to believe]» (Petri
hispani portugalensis tractatus VII, p. 98.13-16); further, in a digression, he shrewdly
remarks: «in aequivocatione principium motivum ab unitate dictionis sumitur et
principium defectus a parte rerum significatarum [in the case of fallacies of homonymy,
the principle that accounts for the deception has to do with single words which mean
more than one thing, whereas the principle of the flaw has to do with the things the
ambiguous word signifiy or cosignify]» (p. 122.5-7; cf. p. 128.13-15 for a similar point
concerning the fallacy of accent). The distinction itself will be successful enough to
serve as a structuring factor in commentators’ questioning, as is most clearly the case
in the Summa Lamberti, Ed. by F. ALESSIO, La Nuova Italia, Firenze 1971, VII, where
the very same questions are asked about any given family of fallacies, namely what
are its causa apparentiae, its causa defectus and its modi: cf. e.g. «sequitur de fallacia
accentus circa quam videndum est quid sit accentus, et quid fallacia accentus, quae
causa apparentiae, quae causa falsitatis in ipsa, et quot sunt modi paralogizandi in ipsa
[the fallacy of accent is discussed next and one must consider what accent is, what the
fallacy of accent is, what causes its deceptive appearance, what accounts for its falsity
and in how many ways it leads to draw a false inference]» (p. 166.1-4); «sequitur de
fallacia figurae dictionis circa quam videndum [169] est quid sit figura dictionis et quid
fallacia figurae dictionis, quae causa apparentiae, quae causa falsitatis et quot modi
paralogizandi sunt in illa fallacia [the fallacy of figure of speech is discussed next and
one must consider what figure of speech is, what the fallacy of figure of speech is,
what causes its deceptive appearance, what accounts for its falsity and in how many
ways it leads to draw a false inference]» (pp. 168.43-169.3); «in paralogismis qui fiunt
secundum accidens semper oportet tria reperiri: secundum rem, substantiam et accidens
assignatum vel attributum, ideo videamus prius quis terminus debeat dici generaliter res
subiecta, quis accidens et quis attributum et per hoc videbitur quid sit accidens; postea
videndum est quid sit fallacia accidentis, quae causa apparentiae, quae causa falsitatis
et quot modi paralogizandi sunt in ea [in fallacies of accident one always need to find

clear from the start12 that his main concern with sophistic arguments is not so
much that they are poor arguments but rather that they manage not to appear
so13. Secondly and foremost, their place in Aristotle’s classification depends
on the way this illusion works: for instance, «homonymy», «amphiboly» and

out three things: what is ascribed or attributed according to the thing, the substance and
the accident. Therefore, let’s consider first which term generally deserves to be called
a subject, which one deserves to be called an accident or an attribute. This way, we
will ascertain what accident means here. One has to consider next what the fallacy of
accident is, what causes its deceptive appearance, what accounts for its falsity and in
how many ways it leads to draw a false inference]» (p. 173.28-35); «sequitur de fallacia
consequentis, circa quam videndum est quid sit consequens, quid fallacia consequentis,
quae causa apparentiae, quae causa falsitatis et quot sunt modi paralogizandi in ea [the
fallacy of consequent is discussed next and one must consider what consequent is, what
the fallacy of consequent is, what causes its deceptive appearance, what accounts for its
falsity and in how many ways it leads to draw a false inference]» (p. 195.8-11).
Cf. Aristotelis sophistici elenchi 1, 164a 20-26: «περὶ δὲ τῶν σοϕιστικῶν
ἐλέγχων καὶ τῶν ϕαινομένων μὲν ἐλέγχων, ὄντων δὲ παραλογισμῶν ἀλλ’ οὐκ
ἐλέγχων, λέγωμεν ἀρξάμενοι κατὰ ϕύσιν ἀπὸ τῶν πρώτων. ὅτι μὲν οὖν οἱ μὲν
εἰσὶ συλλογισμοί, οἱ δ’ οὐκ ὄντες δοκοῦσι, ϕανερόν. ὥσπερ γὰρ καὶ ἐπὶ τῶν
ἄλλων τοῦτο γίνεται διά τινος ὁμοιότητος, καὶ ἐπὶ τῶν λόγων ὡσαύτως ἔχει
[we will discuss the sophistical refutations, that is, refutations which appear to be
so while they are not, being paralogisms instead. As befits the natural order, we will
begin with what come first. That some deductions are really what they seem to be,
while others only looks like deductions, is evident. As it happens in other matters,
this arises from similarity. And this is the case with arguments as well]».
Albert the Great will stress the fact that where there is no such appearance there’s
no fallacy involved either: «si causam apparentiae non haberet, non deciperet [where
nothing produces a deceptive appearance, there is no deception either]» (Alberti magni
expositio sophisticorum elenchorum, p. 850b 52-53). A similar point is made by Giles
of Rome: if a bad argument does not appear to be sound, it is not a fallacy (cf. Aegidii
romani expositio super libros elenchorum 58vb 47-48: «dubitaret forte aliquis, quia
videtur hic nullam esse fallaciam, nullam enim videtur habere apparentiam [one might
be perplexed, for no fallacy seems involved here, since there is no deception]»). Simon
of Faversham will do the same: «paralogismus enim secundum quamlibet fallacia
debet apparere bonus syllogismus, aliter non falleret; ergo oportet quod quaelibet
fallacia habeat aliquid quod faciat ipsam apparere esse bonum syllogismum [whatever
the fallacy involved, a paralogism has to look like a sound deduction, otherwise it
would not be deceitful. Therefore, any fallacy whatsoever needs something that makes
it looks like a sound deduction]» (Simonis de Faverisham quaestiones novae super
libro elenchorum, Ed. by S. EBBESEN – T. IZBICKI – J. LONGEWAY – F. DEL PUNTA –
E. SERENE – E. STUMP, Pontifical Institute of Mediaeval Studies, Toronto 1984, q. 10,
p. 128.122-124).

«figure of speech» all involve a double meaning14, but they are not deceitful
in the same way15. And this is indeed why they are distinct fallacies: they
may well share the same causa defectus, but –their causa apparentiae being
different– they are different. This is of course a pretty strong claim, but it is
not at all an unusual one in medieval literature16. A typical example is to be
found in the views of Anonymus salmaticensis-florentinus, who is strongly
committed to the idea that the reason why a fallacy does not look like one

This is openly stated in Aristotelis sophistici elenchi 6, 168a 23-25: «τῶν μὲν
γὰρ ἐν τῇ λέξει οἱ μέν εἰσι παρὰ τὸ διττόν, οἷον ἥ τε ὁμωνυμία καὶ ὁ λόγος καὶ ἡ
ὁμοιοσχημοσύνη [some fallacies that have to do with expression depend on a double
meaning, as –for instance– homonymy, amphiboly and figure of speech]».
This is precisely the way Anonymous cordubensis explains why the three
are different fallacies, that is, on account of their different principia apparentiae.
Since «fallaciae distinguuntur penes principia apparentiae, ipsarum fallaciarum
diversarum necesse est esse diversa principia [fallacies are told apart mostly by the
principles that account for their deceptive appearance, for different fallacies must
have different principles]» (Anonymi Cordubensis quaestiones super sophisticos
elenchos, Ed. by S. EBBESEN, Incertorum auctorum quaestiones super sophisticos
elenchos, G.E.C. Gad, Copenhague 1977, q. 820, p. 306.10-11), «ad hoc dicitur quod
principium apparentiae proprium aequivocationis per quod distinguitur a quolibet
loco sophistico est unitas vocis incomplexae secundum materiam et formam. Ex hoc
enim patet distinctio eius a fallacia amphiboliae quoniam in amphibolia est unitas
vocis complexae, hic autem vocis incomplexae […]. Distinguitur autem a figura
dictionis quoniam in figura dictionis non est unitas vocis incomplexae secundum
substantiam vocis, sed solum secundum similitudinem terminationum [The answer
to that is: the principle which both accounts for the deception peculiarly associated
with homonymy and allows to tell homonymy apart from any other fallacy is the
material and formal unity of the single word which happens to be ambiguous.
It is thereby evident why the fallacy of homonymy differs from the fallacy of
amphiboly, in so far as the unity involved is alternatively the unity of a single
expression (homonymy) or the unity of a complex expression (amphyboly) [...].
The fallacy of homonymy also differs from the fallacy of figure of speech, for in
the latter the unity involved is not the unity of a single expression according to its
substance, but rather the unity according to the similarity between words because
of their ending]» (Anonymi Cordubensis quaestiones super sophisticos elenchos, q.
820, p. 307.24-35).
Cf. e.g. Thomae de Aquino (?) de fallaciis, Ed. by H. F. DONDAINE, Editori di
San Tommaso, Roma 1976, p. 405a 16-48 and p. 405b 1-30. Anonymi G&C 611-II
quaestiones in sophisticos elenchos, Ed. by S. EBBESEN, «Texts on Equivocation. Part
II. Ca. 1250 - ca. 1310», CIMAGL, 68 (1998) 183.17-20. Radulphi britonis quaestiones
super sophisticos elenchos, Ed. by S. EBBESEN, CIMAGL, 53 (1986) 122.15-19 and

is both its most important and its most distinctive feature17. Another good
example is Simon of Faversham, for whom what produces the illusion that a
fallacy is a sound argument is the very thing that makes it the fallacy it is. In
Simon of Faversham’s words, the causa apparentiae is to perform a double
function, that is, on the one hand, it makes it possible to tell good arguments
from bad ones and, on the other hand, it sets any given fallacy apart from any
WHEN TAXONOMY FAILS: A CASE STUDY. Sure enough, Aristotle and
medieval commentators alike allowed for occasional overlaps between
families of fallacies19. After all, no stretch of imagination is required to

It is plain that, in the Anonymous’ eyes, sophistical appearances are not only
utterly important but that they also make all the difference in taxonomical matters:
Anonymi salmaticensis-florentini quaestiones super Sophisticos elenchos, Ed. by
S. EBBESEN, Incertorum auctorum quaestiones super Sophisticos elenchos, q. 82,
p. 189.25-33: «in fallacia nihil est prius causa apparentiae, cum ex ipsa ratio fallaciae
accipitur […]; a causa enim apparentiae sumitur ratio fallaciae in se et distinctio eius
ab omnibus aliis [in a fallacy nothing takes precedence over the cause that accounts
for its deceptive appearance, since it is the very thing that makes a given fallacy the
fallacy it is [...]; as a matter of fact, the cause that accounts for a fallacy’s deceptive
appearance accounts also for both its being the fallacy it is and the way it differs from
any other fallacy]».
Simonis de Faverisham quaestiones novae super libro elenchorum, q. 20,
p. 151.27-29: «illud est causa apparentiae in aliqua fallacia quod facit ipsam esse
fallaciam et quod facit ipsam esse distinctam ab omnibus aliis [in any fallacy the cause that
accounts for its deceptive appearance is the very thing that makes it a fallacy and makes it
differ from any other fallacy]». Furthermore, Simon of Faversham will identify the causa
apparentiae with the formal principle of the fallacy itself: «sicut entia distinguuntur
per suas formas ita distinguuntur fallaciae per suas causas apparentiae. Causa enim
apparentiae in qualibet fallacia est quid formale. Et ideo fallacia quae habet causam
apparentiae distinctam est fallacia distincta [just as things differ because of their forms,
fallacies too differ because of the causes that account for their deceptive appearance. In
fact, such causes are their formal element. Therefore, two fallacies that have not the same
cause that account for their deceptive appearance are different]» (Simonis de Faverisham
quaestiones novae super libro elenchorum, q. 33, p. 189.22-25).
Aristotelis sophistici elenchi 24, 179b 17: «οὐδὲν δὲ κωλύει τὸν αὐτὸν λόγον
πλείους μοχθηρίας ἔχειν [nothing prevents the same argument from having multiple
flaws]». Cf. Anonymi summa sophisticorum elenchorum, p. 417.22-24; Fallaciae
vindobonenses, Ed. by DE RIJK, Logica Modernorum, I, p. 525.26-27; Anonymi
monacensis commentarium in sophisticos elenchos, Ed. by L. GAZZIERO, «The Latin
“Third Man”. A Survey and Edition of Texts from the XIIIth Century», CIMAGL, 81
(2012) 42.12-15; Roberti <Kilwardby> commentarium in sophisticos elenchos, ibid.,

fancy nasty quibblers trying more than one trick at a time or clumsy people
rambling when reasoning in their heads. But such trespasses are so gross
that they are not much of a threat to the integrity of Aristotle’s division.
That being said, other entanglements are of a more serious nature and
may possibly lead to the conclusion that a structural revision of Aristotle’s
taxonomy is in order, but –then again– such crossings are vouched for by
Aristotle himself: the first examples that spring to mind are –of course– the
inclusion of the fallacy of consequent within the fallacy of accident and the
symmetry between the fallacies of composition and division20. Medieval
authors provided nice, even funny examples of fallacies open to more than
one interpretation21. Still, some crossovers are neither trivial nor supported
by the text. The one I wish to investigate will turn out to be both disruptive
and ill inspired.

p. 52.3-22; Nicholai parisiensis notulae super librum elenchorum, ibid., pp. 54.25-
55.8; Roberti codicis veneti commentarium in sophisticos elenchos, ibid., p. 60.6-14;
Roberti de Aucumpno commentarium in sophisticos elenchos, ibid., p. 78.1-14; Alberti
magni expositio sophisticorum elenchorum, p. 938b.
Aristotelis sophistici elenchi 6, 168b 27-28: «οἱ δὲ παρὰ τὸ ἐπόμενον μέρος
εἰσὶ τοῦ συμβεβηκότος [the fallacies of consequent are a subset of those of accident]»
(cf. 7, 169b 6-7 et 8, 170a 4-5); 23, 179a 13-15: «παρὰ σύνθεσιν ὁ λόγος, ἡ λύσις
διελόντι, εἰ δὲ παρὰ διαίρεσιν, συνθέντι [when arguments turn on composition, they
are to be solved by means of a division; when they turn on division, then they are to be
solved by means of a composition]» (cf. Aristotelis ars rhetorica, Ed. by R. KASSEL,
Walter de Gruyter, Berlin 1976, II, 24, 1401a 25-26 where Aristotle mentions both
as one fallacy rather than two: «ἄλλος τὸ <τὸ> διῃρημένον συντιθέντα λέγειν ἢ
τὸ συγκείμενον διαιροῦντα [another fallacy consists in asserting conjointly what is
separated and separately what is conjoined]»).
Medieval humor is not for all tastes, nor are medieval logicians’ examples (cf.,
e.g., those Abelard peppered his gloses on Aristotle’s Peri hermeneias with, edited by
Y. IWAKUMA, «Pierre Abélard et Guillaume de Champeaux dans les premières années du XIIe
siècle. Une étude préliminaire», in J. BIARD (ed.), Langage, sciences, philosophie au XIIe
siècle, Vrin, Paris 1999, p. 95). The argument they usually brought up in order to illustrate
fallacies open to more than one solution should be to everybody’s liking: «quicumque sunt
episcopi sunt sacerdotes; isti asini sunt episcopi; ergo isti asini sunt sacerdotes [all bishops
are priests; theses asses are bishops (these asses belong to the bishop); therefore, these
asses are priests]». Cf. Tractatus de sophistica argumentatione (Dialectica monacensis), p.
562.10-12; Petri hispani portugalensis tractatus VII, p. 108.19-20; Introductiones magistri
Guillelmi de Shyrewode in logicam, Ed. by C. KANN – H. BRANDS, Meiner, Hamburg 1995,
VI, p. 172.73-74; Summa Lamberti, VII, p. 152.7-28 and pp. 154.28-155.4; Thomae de
Aquino (?) de fallaciis IV, p. 407a 56-58; Rogeri Baconi Compendium studii theologiae,
Ed. by T. S. MALONEY, Brill, Leiden 1988, p. 139, 116.22-23.

2. Anomalies and random variables


ANSWER. When Medieval Latin commentators asked an odd question, this
is usually the sign that either they lacked a piece of information or they
were facing an anomaly in processing available data. When their answer
proves to be at odds with the basic tenets of a theory they otherwise seem
to grasp pretty well, one had better start looking for both. The question
I’ll be using as a Freudian slip of sorts is very odd indeed and it took the
form of a dilemma: «whether the fallacy of figure of speech is a linguistic
fallacy or not». The answer is even more telling, since no reader in his
right mind would admit, let alone accept that a sophism depending on the
shape of words may be independent from expression rather than related to
it. And yet this is precisely the claim whose grounds I am going to discuss.
For once, the deficit in information is not directly related to the status of
the Aristotelian corpus bequeathed to the Latins over the centuries. It has
more to do with the fact that no ancient scholium or excerpt had been
handed down –via Boethius, James of Venice or the Arabs– about the
most thought-provoking piece of argument Aristotle introduced in order to
illustrate what figure of speech is and how it works, namely the so called
«Third Man». The anomaly is an occasional (albeit very widespread)
mismatch between fallacies of accident and fallacies of figure of speech,
which easily qualifies as one of the most peculiar episodes in the history of
Aristotle’s Latin exegesis.
number of interpreters, both ancient and modern, have suggested22, the

Anonymi glosae in Aristotelis Sophisticos elenchos, Ed. by DE RIJK, Logica
Modernorum, I, p. 214.20-22: «notandum quod “accidens” dicitur hic predicatum, sive
de se tantum sive de alio predicetur, sive sit substantiale sive accidentale [it should
be noted that “accidens” means here predicate, whether it is predicated of itself or
of something else, whether it is essential or accidental]»; cf. p. 214.10: «secundum
accidens, idest secundum praedicatum [“secundum accidens”, that is to say predicate
related]». Anonymi summa sophisticorum elenchorum, p. 356.7-10: «“accidens”
enim ibi largo modo accipitur, scilicet pro quolibet predicato, sive accidentale sit
sive substantiale [“accident” is taken here in a broad sense and means any predicate,
whether accidental or essential]». Anonymi parisiensis compendium sophisticorum
elenchorum, Ed. by S. EBBESEN – Y. IWAKUMA, CIMAGL, 60 (1990) 88.19-22: «accidens
autem hic appellat Aristoteles praedicatum: cum enim subiectum et accidens relativa
sunt, et quod in propositione subicitur subiectum dicatur, non debet mirum videri si eius

fallacy of accident’s name does not imply that only accidental features
are involved. As it is the case elsewhere in Aristotle’s corpus23 and as it
is demonstrated by his own choice of examples24, «accidens» means here
much the same as «predicate» tout court, that is: without restriction. As its
definition goes25, the fallacy of accident leads to believe that what is said

praedicatum accidens appelletur [Aristotle calls here “accidens” the predicate. Since
the subject and the accident are relative and the subject is said to be underlying, it does
not come as a surprise that its predicate is called “accidens”]». Anonymi cantabrigiensis
commentarium in Aristotelis sophisticos elenchos, 89vb 17-18: «accidens in hac
iunctura locutionis “fallacia secundum accidens” dicitur praedicatum sive ipsum sit
accidentale praedicabile sive substantiale [“accidens” within the expression “fallacia
secundum accidens” means predicate, whether it is something predicated accidentally
or essentially]»; cf. 89vb 4-5: «dicitur accidens omne praedicabile sive accidentale
sive quod non <est> accidentale sive substantiale [we call “accidens” everything
that may be predicated, either accidental or non-accidental, that is to say essential]».
Anonymi fallacie londinenses, p. 669.4-5: «accidens prout hic accipitur idem est quod
praedicatum [the way “accidens” is understood here, it means the same as predicate]».
Guillelmi de Montibus (?) fallaciae, Ed. by Y. IWAKUMA, «The Fallaciae and Loci of
William de Montibus. An Edition», Journal of Fukui Prefectural University, 2 (1993)
15.17-18: «secundum accidens dicitur quasi secundum praedicatum [“secundum
accidens” means pretty much the same as “predicate related”]». Anonymi fallaciae
lemovicenses, Ed. by S. EBBESEN – Y. IWAKUMA, CIMAGL, 63 (1993) 30.29: «prout
hic dicitur accidens idem est quod praedicatus [what is called here “accidens” is the
same as “predicate”]». Modern scholars who hold the same view are, amongst others,
M. MIGNUCCI, «Puzzles about Identity. Aristotle and His Greek Commentators», in
J. WIESNER (ed.), Aristoteles. Werk und Wirkung, W. de Gruyter, Berlin 1985, I, p. 75,
D. Zaslawsky, «Le sophisme comme anomalie», in B. CASSIN (ed.), Le plaisir de
parler. Études de sophistique comparée, Editions de Minuit, Paris 1986, p. 192, and
L.-A. DORION, Aristote. Les réfutations sophistiques, Presses de l’Université Laval –
Vrin, Montréal – Paris 1995, p. 233, note 57.
συμβαίνω and κατηγορέω are synonyms in Aristotelis topica, Ed. by
J. BRUNSCHWIG, Les Belles Lettres, Paris 2007, VII, 1, 152a 33-37 and 152b 25-29,
as well as in Aristotelis sophistici elenchi 7, 169b 4-6. Likewise, συμβεβηκός and
κατηγορούμενον are synonyms in Aristotelis analytica posteriora, Ed. by W. D. ROSS,
Clarendon Press, Oxford 1949, I, 4, 73b 8-10.
There can be no doubt that being a man is an essential feature of the individual
man (cf. Aristotelis sophistici elenchi, 5, 166b 33-36) or that being a figure is an
essential feature of any given figure, a triangle for instance (cf. Aristotelis sophistici
elenchi 6, 168a 40 - 168b 4).
Aristotelis sophistici elenchi 5, 166b 28-32: «note 25: οἱ μὲν οὖν παρὰ τὸ
συμβεβηκὸς παραλογισμοί εἰσιν ὅταν ὁμοίως ὁτιοῦν ἀξιωθῇ τῷ πράγματι καὶ
τῷ συμβεβηκότι ὑπάρχειν. ἐπεὶ γὰρ τῷ αὐτῷ πολλὰ συμβέβηκεν, οὐκ ἀνάγκη

of a predicate may be also said of its subject and vice versa26. As a result,
fallacies of accident occur when one is unable to determine beforehand
what belongs to both (the subject and the predicate) and what belongs

πᾶσι τοῖς κατηγορουμένοις καὶ καθ’ οὗ κατηγορεῖται ταὐτὰ πάντα ὑπάρχειν

[fallacies of “accident” occur when a predicate whatsoever is believed to belong in the
same way to a thing and to something that is predicated of it. Since many attributes are
predicated of the same thing, it is not necessary that all the attributes belong both to the
thing and to all of its predicates]».
Medievals acknowledged that the fallacy of accident goes both ways, as
the following five examples make it pretty clear. Anonymi summa sophisticorum
elenchorum, p. 356.1-8: «secundum accidens ergo fiunt paralogismi, ut dicit Aristoteles,
quando quodlibet similiter fuerit assignatum inesse rei subiectae et accidenti, id est
praedicato, et e converso, id est quando idem assignatur convenire accidenti sive
praedicato, quod inest et rei subiectae [as Aristotle says, fallacies of accident occur
when something whatsoever is similarly said to belong both to the underlying thing
and to the accident, that is to say to the predicate, or – the other way round – when
it is said to fit the accident, that is to say the predicate, to which it belongs, and the
underlying thing]». Anonymi cantabrigiensis commentarium in Aristotelis sophisticos
elenchos, 89vb 19-21: «ut dicatur fallacia secundum accidens deceptio proveniens ex
omissione \habitudinis/ unius praedicabilis ad aliud sive praedicati ad subiectum sive
subiecti ad praedicatum [what we call the fallacy of accident is a deception which
arises from disregarding the relation of one predicate to the other, whether this relation
is the predicate’s relation to the subject or the subject’s relation to the predicate]».
Anonymi fallacie londinenses, p. 669.6-9: «fallacia secundum accidens est deceptio
proveniens ex obmissa habitudine praedicati ad subiectum, vel econverso, quia
haec fallacia provenit tam ex obmissa habitudine subiecti ad praedicatum quam ex
obmissa habitudine praedicati ad subiectum [the fallacy of accident is a deception
which arises from disregarding the relation either of the predicate to the subject or
of the subject to the predicate, for such a fallacy arises as much from a neglected
relation of the subject to the predicate than from a neglected relation of the predicate
to the subject]». Guillelmi de Montibus (?) fallaciae, p. 16.1-7: «incidit autem haec
fallacia in argumentatione quandoque aliquid assignatur subiecto quod non potest
assignari praedicato vel e converso. Est enim fallacia secundum accidens deceptio
proveniens ex omissa habitudine praedicati ad subiectum vel e converso [such a
fallacy occurs in arguments which ascribe to the subject what cannot be ascribed to
the predicate and vice versa. As a matter of fact, the fallacy of accident is a deception
which arises from disregarding the relation either of the predicate to the subject or
of the subject to the predicate]». Anonymi salmaticensis-florentini quaestiones super
Sophisticos elenchos, q. 84, p. 194.24-31: «accidens est aliquo modo idem subiecto de
quo dicitur et aliquo modo non, et sic sumitur accidens in fallacia accidentis, ut dicit
Commentator, et secundum hoc dicit modos accidentis: uno modo ex eo quod aliquid
quod inest praedicato denotatur inesse subiecto, ut “homo est animal, sed animal est

exclusively to either one (alternatively the subject or the predicate)27.

As far as such determination depends rather on states of affairs than on
names, one may safely assume that fallacies of accident have little –if
anything at all– to do with linguistic considerations. Moreover, even when
such considerations have been forced upon them, fallacies of accident
verged on homonymy rather than on figure of speech28. Which is one

genus, ergo homo est genus”; alio modo ex eo quod illud quod inest subiecto denotatur
inesse praedicato, ut “Socrates est homo, Socrates est individuum, ergo homo est
individuum”; et sic duo modi principales sunt [the “accident” is, in a way, the same as
the thing it is predicated of and, in another way, it is not. As the Commentator has it,
this is how “accident” has to be understood in the fallacy of the same name and how its
moods have to be assigned therein. One arises from the fact that what is attributed to
the predicate is meant to be predicated of the subject as well, as in: “man is an animal,
but animal is a genus, therefore man is a genus”. Another arises from the fact that what
is attributed to the subject is attributed to the predicate as well, as in: “Socrates is a
man, Socrates is an individual, therefore man is an individual”. This is why there are
two main moods of the fallacy of accident]».
Aristotelis sophistici elenchi 7, 169b 3-6: «<169a 22: ἡ δ’ ἀπάτη γίνεται>
τῶν δὲ παρὰ τὸ συμβεβηκὸς διὰ τὸ μὴ δύνασθαι διακρίνειν τὸ ταὐτὸν καὶ τὸ
ἕτερον, καὶ ἓν καὶ πολλά, μηδὲ τοῖς ποίοις τῶν κατηγορημάτων πάντα ταὐτὰ
καὶ τῷ πράγματι συμβέβηκεν [in fallacies of accident the deception arises from the
incapacity to distinguish what is the same and what is different, what is one and what
is many, as well as from the incapacity to tell which predicates have the same attributes
as their subjects]».
Although they should have known better –and some of them actually did,
as argued at length in a forthcoming paper in Acta philosophica: «Exempla docent.
How to Make Sense of Aristotle’s Examples of the Fallacy of Accident (Doxography
Matters)»– Latins brought the fallacy of accident and the fallacy of homonymy
together on the grounds of a variation in the supposition of the middle term observed
in tokens of both types. Cf. Anonymi summa sophisticorum elenchorum, p. 294.10-
16 and pp. 357.25-358.4; Anonymi fallaciae vindobonenses, p. 527.11-29; Anonymi
tractatus de sophistica argumentatione (dialectica monacensis), p. 585.23-34; Petri
hispani portugalensis tractatus VII, pp. 106, 148.19-293; Anonymi monacensis
commentarium in sophisticos elenchos, p. 42; Rogeri Baconi summulae dialectices,
Ed. by A. DE LIBERA, Archives d’Histoire Doctrinale et Littéraire du Moyen Âge, 54
(1987) 261, 627-629; Summa Lamberti VII, pp. 181.28-182.11; Thomae de Aquino
(?) de fallaciis IX, p. 411a 75-88; Anonymi salmaticensis-florentini quaestiones super
Sophisticos elenchos, q. 83, p. 192.20-35; Aegidii romani expositio super libros
elenchorum 17ra 40-45; Ioannis Duns Scoti quaestiones super librum elenchorum
Aristotelis, Ed. by R. ANDREWS – O. BYCHKOV – S. EBBESEN – G. ETZKORN – G. GAL
– R. GREEN – T. NOONE – R. PLEVANO – A. TRAVER, The Franciscan Institute St.
Bonaventure University, St. Bonaventure NY 2004, q. 44, p. 471.1-4; Ioannis Buridani

more reason Latin commentators should have kept them apart, since they
ordinarily took very seriously the difference between homonymy and form
of expression, whose «actual» and «imaginary» polysemy they opposed
in line with a tradition that –under «Alexander»’s patronage– goes back
to Galen through Michael of Ephesus and James of Venice29. To make a
(very) long story short, homonymy is tantamount to using one word with
multiple meanings30, while figure of speech occurs when using different

quaestiones elenchorum, Ed. by R. VAN DER LECQ – H. A. G. BRAAKHUIS, Ingenium,

Nijmegen 1994, q. 14, 73.138-143.
In this connection, Anonymus digbeianus, Anonymus salmaticensis-
florentinus, Simon of Faversham, Anonymus G&C 611-II, Duns Scotus and Radulphus
Brito especially deserve to be mentioned, for they emphatically underscored such a
difference. For the sake of brevity, I will only quote Simon of Faversham’s Quaestiones
veteres super libro elenchorum: «specialiter distinguitur <fallacia aequivocationis> a
figura dictionis, [80] quia in figura dictionis sub unitate vocis secundum substantiam
non latent plura significata, sed magis sub similitudine vocis, et quia ibidem non latent
plura significata secundum substantiam vocis, ideo dicimus quod ibi est phantastica
multiplicitas [the fallacy of homonymy especially differs from the fallacy of figure
of speech, for the multiple meanings are not dissimulated by a single word but by
a similarity between words. In so far as the multiple meanings involved in figure
of speech are not dissimulated by one word only, we call its multiplicity “phanta-
sised”]» (Simonis de Faverisham quaestiones veteres super libro elenchorum, Ed.
Pontifical Institute of Mediaeval Studies, Toronto 1984, q. 19, pp. 79.20-80.25;
cf. q. 10, p. 126.59-63. As for the others, cf. Anonymi digbeiani in sophisticos
elenchos, S. EBBESEN (ed.), CIMAGL, 53 (1986) 121.17-22; Anonymi salmaticensis-
florentini quaestiones super Sophisticos elenchos, qu. 80, 179.1-3 and 179.21-181.59;
Anonymi G&C 611-II quaestiones in sophisticos elenchos, p. 183.17-30; Ioannis Duns
Scoti quaestiones super librum elenchorum Aristotelis, q. 42, pp. 465.7-12 and 19-20;
Radulphi britonis quaestiones super sophisticos elenchos, p. 120.34-36, p. 121.33-34
and p. 125.13-17. The origin and history of the distinction between multiplex actuale,
potentiale et phantasticum has been meticulously reconstructed by S. EBBESEN, whose
«Philoponus, “Alexander” and the Origins of Medieval Logic», in R. SORABJI (ed.),
Aristotle Transformed. The Ancient Commentators and Their Influence, Duckworth,
London 1990, pp. 445-462 is the best place to start looking.
Cf. Anicii Manlii Severini Boethii de divisione liber, Ed. by J. MAGEE, Brill,
Leiden 1998, p. 8.16-30. Anonymi aurelianensis I commentarium in Sophisticos
elenchos, p. 85 and p. 95. Fallaciae vindobonenses, p. 499.3-4. Cf. Guillelmi de
Montibus (?) fallaciae, p. 6.1-3. Petri hispani portugalensis tractatus VII, p. 98.25-26.
Fallaciae ad modum Oxoniae, p. 20, particularly the Q version. Roberti Kilwardby (?)
commentarium in Sophisticos Elenchos, Ed. by S. EBBESEN, «An Inventory of Texts
about Equivocation», CIMAGL, 67 (1997) 161.28-30. Introductiones magistri

words whose morphological resemblance conveys the illusion that they

signify the same thing or the same kind of things31. The latter is indeed
the sort of quandaries Aristotle had in mind defining fallacies that depend
on the form of expression32: when things that are not the same are said in

Guillelmi de Shyrewode in logicam, VI, p. 170.64 and pp. 172.94-174.102. Anonymi

e Musaeo 133 commentarium in Aristotelis sophisticos elenchos, Ed. by S. EBBESEN,
«An Inventory», op. cit., p. 165.12-13 and pp. 165.30-166.3. Nicholai parisiensis
notulae super librum elenchorum, Ed. by S. EBBESEN, «An Inventory», op. cit.,
p. 170.9-11. Nicholai parisiensis de fallaciis (summae metenses), pp. 474.15-475.3.
Roberti codicis veneti commentarium in sophisticos elenchos, p. 66.13-14. Summa
Lamberti, VII, p. 148.28-30. Roberti de Aucumpno commentarium in Sophisticos
Elenchos, Ed. by S. EBBESEN, «An Inventory», op. cit., p. 184.14-16, p. 185.10-11 and
p. 188.17-19. Alberti magni expositio sophisticorum elenchorum, pp. 847b.56-848a.2
and p. 850b.56-59. Anonymi Basileensis quaestiones in Aristotelis Categorias, Ed. by
S. EBBESEN, «Texts on Equivocation. Part II. Ca. 1250 - ca. 1310», CIMAGL, 68 (1998)
113.5-11. Thomae de Aquino (?) de fallaciis, pp. 405b 32-406a 10. Anonymi digbeiani
in sophisticos elenchos, pp. 108.28-109.2. Aegidii romani expositio super libros
elenchorum, 10rb 3-6. Anonymi Cordubensis quaestiones super sophisticos elenchos,
q. 827, p. 329.43-46. Anonymi tres quaestiones de aequivocatione, Ed. by S. EBBESEN,
«Texts on Equivocation. Part II», p. 129.10-11 and p. 137.8. Anonymi pragensis
quaestiones super Aristotelis sophisticos elenchos, Ed. by D. MURÈ, «Anonymus
Pragensis on Equivocation», CIMAGL, 68 (1998) 74.2-21 and p. 92.11-13. Thomae
de Wyk fallaciae, Ed. by S. EBBESEN, «Texts on Equivocation», p. 139.9-14. Anonymi
G&C 611-II quaestiones in sophisticos elenchos, p. 144.18-22, p. 145.9-10 and 16-30.
Radulphi britonis quaestiones super sophisticos elenchos, pp. 192.23 - 193.15 and
Cf. Anonymi aurelianensis I Commentarium in Sophisticos elenchos,
p. 124.31-33; Anonymi Cantabrigiensis Commentarium in Aristotelis sophisticos
elenchos, 88vb; Fallaciae vindobonenses, p. 515.4-12; Fallaciae parvipontanae, Ed.
by DE RIJK, Logica Modernorum, I, p. 586.24-26; Guillelmi de Montibus (?) fallaciae,
p. 13.26-28; Anonymi tractatus de sophistica argumentatione (dialectica monacensis),
p. 578.34-35; Petri hispani portugalensis tractatus VII, p. 135.11-25; Fallaciae ad
modum Oxoniae, p. 97-98; Introductiones magistri Guillelmi de Shyrewode in logicam
VI, p. 188.288-290; Rogeri baconi summulae dialectices, p. 254.28-29 and p. 258.6-
259.2; Summa Lamberti VII, p. 169.5-22; Alberti magni expositio sophisticorum
elenchorum, p. 859.33-43; Thomae de Aquino (?) de fallaciis, p. 410b 2-17; Anonymi
digbeiani in sophisticos elenchos, p. 63.1-4; Anonymi Cordubensis quaestiones super
sophisticos elenchos, q. 838, pp. 365.100-366.127; Simonis de Faverisham quaestiones
novae super libro elenchorum, q. 10, p. 128.117-119; Radulphi britoni quaestiones
super sophisticos elenchos 121-125 (in particular 123.9-19).
Aristotelis sophistici elenchi 4, 166b 10-15: «οἱ δὲ παρὰ τὸ σχῆμα τῆς
λέξεως συμβαίνουσιν ὅταν τὸ μὴ ταὐτὸ ὡσαύτως ἑρμηνεύηται, οἷον τὸ

pretty much the same way, this is likely to confuse people about what those
things are and how they stand with respect to each other.
STATING THE PROGRAM. After addressing the problem of why the fallacies
of accident and of figure of speech do not belong together, we may turn
our attention to the rather fortuitous chain of events that brought the two
together. In other words, it is time to ask the question: if fallacies of figure of
speech depend on linguistic features to such an extent that there is no point
in assuming that they do not resort to language itself one way or another,
how come then medieval authors repeatedly asked themselves «utrum figura
dictionis sit fallacia in dictione», which I’ll translate without much concern
for literality: «whether or not the form of expression, as a source of fallacious
reasoning, depends on expression»? Anonymus salmaticensis-florentinus33,

ἄρρεν θῆλυ ἢ τὸ θῆλυ ἄρρεν ἢ τὸ μεταξὺ θάτερον τούτων, ἢ πάλιν τὸ ποιὸν

ποσὸν ἢ τὸ ποσὸν ποιόν, ἢ τὸ ποιοὐν πάσχον ἢ τὸ διακείμενον ποιοῦν, καὶ
τἆλλα δ’ ὡς διῄρηται πρότερον [fallacies of figure of speech occur when what
is not the same is said in the same way; for instance, when something masculine is
designated by means of an expression which is rather feminine, or when something
feminine is designated by means of an expression which is rather masculine, or
when something neuter is said by means of an expression which is alternatively
rather masculine or feminine; or –again– when a quality is said by means of an
expression which looks like a term for a quantity or when a quantity is said by means
of an expression which looks like a term for a quality; or –again– when an action is
said by means of an expression which looks like a term for an affection or when a
state is said by means of an expression which looks like a term for an action; and so
forth according to the division previously made]». Aristotelis sophistici elenchi 7,
169a 29-35: «τῶν δὲ παρὰ τὸ σχῆμα διὰ τὴν ὁμοιότητα τῆς λέξεως. χαλεπὸν
γὰρ διελεῖν ποῖα ὡσαύτως καὶ ποῖα ὡς ἑτέρως λέγεται (σχεδὸν γὰρ ὁ τοῦτο
δυνάμενος ποιεῖν ἐγγύς ἐστι τοῦ θεωρεῖν τἀληθές, μάλιστα δ’ ἐπίσταται
συνεπινεύειν), ὅτι πᾶν τὸ κατηγορούμενόν τινος ὑπολαμβάνομεν τόδε τι,
καὶ ὡς ἓν ὑπακούομεν [as far as fallacies of figure of speech are concerned, the
deception arises from the similarity among expressions. In fact, it is difficult to
tell apart things said in the same way and things said differently (he who is able to
do this is almost on the verge of discovering the truth, all the more so will he be
able to answer advisedly), for we trust everything predicated of something else to
be an individual thing and we understand it as being one]». This is a very strong
philosophical point in its own right and a very wise lesson at that: as a matter of
course, language by itself teaches us next to nothing about how the world is. Since
we talk about different realities as if they were just the same, we simply cannot
trust words to tell them apart.
Anonymi salmaticensis-florentini quaestiones super Sophisticos elenchos,
q. 75, pp. 170.1-172.27.

Anonymus pragensis34, Duns Scotus35, Radulphus Brito36, John Buridan37 and

Marsilius of Inghen38 even devoted entire questions of their commentaries to
the problem. How they came to think of the question as a sensible one to ask
is a puzzle worth solving. Sten Ebbesen39, Irène Rosier-Catach40 and Andrea
Tabarroni41 have gone a long way in helping us understand more about the

Anonymi pragensis, q. 28, pp. 64.8-66.14 asks the question in a slightly
different form: «consequenter quaeritur utrum fallacia figurae dictionis habeat
principium apparentiae ex parte vocis [the question arises next whether the origin of
the fallacy of figure of speech has to do with verbal expression]».
Ioannis Duns Scoti quaestiones super librum elenchorum Aristotelis, q. 37,
pp. 437.1-443.6: «utrum figura dictionis sit locus in dictione [whether the fallacy of
form of expression is a fallacy depending on expression]».
Not only Radulphus Brito asked the question, but he came as close as one
can get to give the right answer: «cum dicitur “illa fallacia non est in dictione cuius
causa apparentiae sumitur ex parte rei”, verum est. Et cum dicitur quod fallacia figurae
dictionis est huiusmodi, falsum est, immo sumitur ex similitudine dictionis cum
dictione, vel in modo appellandi vel in concretione vocum. Et cum dicitur quod ista
fit quando unum praedicamentum commutatur in aliud, verum est; sed hoc non est
per similitudinem sumptam ex parte rei, sed [119] per similitudinem dictonis cum
dictione, quae sumpta est ex parte vocis; et si commutatur unum praedicamentum in
aliud per similitudinem sumptam ex parte rei, tunc magis habet esse fallacia accidentis
[it is true to say that when the cause accounting for a fallacy’s deceptive appearance
has to do with how things are, then the fallacy at hand is not a fallacy depending on
expression. Nevertheless, if one says that such is the case of form of expression, he is
wrong. On the contrary, what causes its deception has to do with the similarity among
expressions, either through the way these expressions designate things or through their
verbal morphology. Besides, if one says that the fallacy of form of expression occurs
when a shift between categories occurs, he is right, but this does not happen because
of a similarity between things, rather because of a similarity between expressions. If
the shift between two categories is brought about by a similarity between things, then
such an argument is rather a fallacy of accident]» (Radulphi britonis quaestiones super
sophisticos elenchos, qu. 32, pp. 118.38-119.4).
Ioannis Buridani quaestiones elenchorum, q. 13, pp. 63-68.
Marsilii de Inghen quaestiones elenchorum, Wien, Österreichische
Nationalbibliothek, 5342, qu. 24, ff. 43ra-44vb; Venezia, Marciana, Latina VI. 146
(coll. 2658), qu. 23, f. 143r.
EBBESEN, Commentators and Commentaries, I, pp. 197-223.
I. ROSIER, «Évolution des notions d’equivocatio et univocatio au XIIe siècle»,
in I. ROSIER (ed.), L’ambiguïté, Presses Universitaires de Lille, Lille 1988, pp. 103-166.
A. TABARRONI, «Figure of Speech and Aristotle’s Division of Fallacies», in
C. CELLUCCI – M. C. DI MAIO – G. RONCAGLIA (eds.), Logica e filosofia della scienza,
ETS, Pisa 1994, pp. 15-24.

problem and related issues (namely, the shifting association of the Boethian
inherited fallacy of univocation and –in turn– homonymy, figure of speech
and accident). As for today, I would like to add a new element to the picture.
In order to do so I will follow a promising thread in an early –possibly the
earliest– collection of quaestiones on Sophistici elenchi, whose author –a
Parisian master of the 1270s– S. Ebbesen dubbed Anonymous cordubensis
after the cordovan library where survives the only manuscript of his work.

3. Anonymus C

SACRA PAGINA. Anonymi cordubensis quaestiones super sophisticos

elenchos, q. 834, 352.14-353.28: «Consequenter quaeritur utrum figura
dictionis sit fallacia in dictione. Et quod non videtur: omnis fallacia cuius
principium apparentiae est a parte [353] rei et non a parte vocis est fallacia
extra dictionem et non in dictione; figura dictionis est huiusmodi; quare et
cetera. Maior patet, quia secundum Alexandrum ex hoc dicuntur fallaciae
in dictione quia principium apparentiae habent a parte vocis, et fallaciae
extra dictionem quia principium apparentiae habent a parte rei. Minor
declaratur, nam ex eo quo ratio fallaciae sumitur ex sua causa apparentiae,
et in quolibet modo fallaciae manet ratio fallaciae, necesse est in quolibet
modo cuiuscumque fallaciae manere eandem causam apparentiae. Nunc
in tertio modo figurae dictionis non est principium apparentiae a parte
vocis, ut patet ibi: “Coriscus est alter ab homine, ergo est alter a se”,
“Coriscus” enim et “homo” nullam convenientiam habent. Quare ibi causa
apparentiae erit a parte rei [next the question arises whether the form of
expression, as a source of fallacious reasoning, depends on expression.
This appears not to be the case. Any fallacy whose appearance has factual
rather than verbal grounds is a fallacy independent of expression rather
than a fallacy depending on it[a]. Such is the case of <the fallacy of> form
of expression. This is why, etc. The major premise is obvious, since –
according to «Alexander»[b]– fallacies, which depend on expression, are
so called because of the verbal nature of what produces their illusion; on
the other hand, fallacies, which do not depend on expression, are so called
because of the factual nature of what produces their illusion. The minor
premise is thus to be explained: since a fallacy is what it is on account
of what produces its illusion and what makes it the fallacy it is remains
the same throughout its modes[c], it is necessary that what produces its
illusion remains the same in whichever variety a given fallacy comes.

Now, in the third mode of fallacies that depend on form of expression[d],

what produces the illusion does not depend on words, as it becomes clear
through this example[e]: “Coriscus is other than man; thus he is other than
himself”[f]. As a matter of fact, the expressions “Coriscus” and “man” have
no resemblance at all. This is why what produces here the illusion will be
factual in nature]».
IN DICTIONE». I.e. any fallacy whose illusion or (deceptive) appearance
depends on the things we say rather than on the way we talk about them
falls outside expression. Anonymus cordubensis holds fast to what was
at that time a commonplace. As a matter of fact, as early as the Anonymi
glosae in Aristotelis Sophisticos elenchos, p. 205.7-9 and p. 213.27-29
the association between in dictione and a parte vocis, on the one hand,
and extra dictionem and a parte rei, on the other hand, appears to be
taken for granted. As it will be afterwards: cf. Anonymi summa sophis-
ticorum elenchorum, p. 286.19-21 and p. 353.14-16. Anonymi pari-
siensis compendium sophisticorum, p. 73.2-5. Anonymi compendiosus
tractatus de fallaciis ex codice parisino latino 6674, Ed. by S. Ebbesen,
Cahiers de l’Institut du Moyen-Âge Grec et Latin, 34 (1979) 186.48-
187.1. Anonymi Aurelianensis I commentarium in Sophisticos elenchos,
pp. 132.29-133.2. Anonymi Aurelianensis II tractatus de paralogismis,
pp. 51.28-52.2 and p. 157.16-19. Fallaciae parvipontanae, p. 551.10-27,
p. 552.1-2 and p. 592.5-16. Anonymi fallacie londinenses, p. 647.6-8.
Guillelmi de Montibus (?) fallaciae, p. 15.21-23. Tractatus de sophistica
argumentatione (Dialectica monacensis), p. 558.25-27, p. 559.4-6 and
p. 584.22-25. Petri hispani portugalensis tractatus VII, p. 145.22-25.
Fallaciae ad modum Oxoniae, 16. Introductiones magistri Guillelmi de
Shyrewode in logicam VI, p. 168.38-44. Nicholai parisiensis notulae super
librum elenchorum, Praha Knihovna Metropolitni Kapituli, L.76 (1322),
p. 61vb. Nicholai parisiensis de fallaciis (summae metenses), p. 474.1-10.
Rogeri baconi summulae dialectices, pp. 239.26 - 240.9. Summa Lamberti
VII, p. 146.20-36. Alberti magni expositio sophisticorum elenchorum,
p. 846b 33-46. Thomae de Aquino (?) de fallaciis IV, pp. 405a 44 - 405b
6 and p. 411a 1 - 411b 11. Anonymi salmaticensis-florentini quaestiones
super sophisticos elenchos, q. 35, p. 65.9-10. Aegidii romani expositio
super libros elenchorum 27va 17-24 and 49rb 62 - 62va 15. Simonis de
Faverisham quaestiones novae super libro elenchorum, q. 1, p. 106.95-

99. Anonymi pragensis, q. 28, p. 64.13-15. Radulphi britonis quaestiones

super sophisticos elenchos, p. 118.7-10 and p. 122.3-4.
[b] «SECUNDUM ALEXANDRUM». As he already did in quaestio 820
(p. 306.4-9) on this very issue, Anonymus C appeals here to the authority of
what was believed to be Alexander of Aphrodisias’ exegesis of Aristotle’s
Sophistici elenchi. The reference made to «Alexander» is merely a way of
reproducing the traditional alternative, with a peculiar twist nonetheless,
for the Anonymous is about to misapply it in his attempt to show –against
Aristotle42 and, one might add, against exegetical common sense– that
«form of expression» is a sophistic trick which does not depend on
linguistic expression rather than one which exploits it to its advantage.
is quite understandable and reflects the fact that –as a general rule– fallacies
related to the same family are to be solved in one and the same way43. It is
nonetheless about to backfire and become a rather strong argument against
Anonymus cordubensis view. As a matter of fact, the same consideration
may be put forward to prove the opposite, as an interesting development in
a relatively close text is to show: «sed illud non videtur omnino esse verum,
quia ratio fallaciae sumitur a principio apparentiae. Si igitur sit fallacia una
in quolibet modo, oportet principium manere unum in quolibet modo; et
cum in figura dictionis principium apparentiae sit unitas vocis secundum
qualitatem sub qua latent plura, cum similis modus appellandi non
necessario habeat similitudinem vocis, non est hoc sufficiens ad principium
in hac fallacia [but this does not appear to be entirely true, since a fallacy is
what it is on account of what produces its illusion. Now, if a fallacy is the
same throughout its modes, what makes a fallacy what it is will necessarily
be the same in each mode. Further, given that what produces the illusion in
a fallacy of the form of expression is the qualitative unity of words which
happen to stand for multiple things, in so far as similar designations are
not necessarily expressed in the same way, this is not enough to cause that
particular fallacy]» (Anonymi salmaticensis-florentini quaestiones super
Sophisticos elenchos, q. 76, p 173.40-46 ; cf. q. 74, pp. 169.22 - 170.31).
[d] «IN TERTIO MODO FIGURAE DICTIONIS». Anonymus C refers to the
well-known confusion between a «this something» and «what qualifies it
as the something it is» («ex mutatione quale quid in hoc aliquid»). Since

Cf. Aristotelis sophistici elenchi 4, 165b 23-27 (quoted above, note 3).
Cf. Aristotelis sophistici elenchi 20, 177b 31-33 and 24, 179b 11-12.

A.J. Smith, «TODE TI in Aristotle», The Classical Review, 35, 1921,

p. 19 the issue has been widely studied. Two essential readings deserve
a special mention: J. Kung, «Aristotle on Thises, Suches and the Third
Man Argument», Phronesis, 26, 1981, pp. 207-247 and S. Ebbesen, «Hoc
aliquid - Quale quid and the Signification of Appellatives», Philosophia,
5-6, 1975-1976, pp. 370-392.
[e] «UT PATET IBI». What follows is very much the linchpin of Anonymus
C’s argument, which he regarded himself as the most substantial piece of
evidence in order to prove that this mode of figura dictionis may be set
apart from the others and be removed –so to speak– from within the sphere
of language.
here is a distinguished argument in an abridged form. The missing premise
–which is to be supplied from Aristotelis sophistici elenchi, 5, 166b 33–
being: «Coriscus est homo [Coriscus is a man]».

LECTURA. This is hardly the most memorable piece of Aristotelian

scholarship the Latin Middle Ages left us, but it is most certainly a text that
has the qualities of its faults. Three deserve to be pointed out in particular:
1. One of Anonymus C’s assumptions rests on an apocryphal source:
the Latin version of a commentary by Alexander of Aphrodisias
whose original greek version, if it ever existed, was long lost.
2. Anonymus C built his case on an example –«Coriscus est alter ab
homine, ergo est alter a se»– which is both as Aristotelian as it gets
and remarkably out of place.
3. Anonymus C hammered home his view by stating that «homo» and
«Coriscus» bear no discernible similarity.

PARERGA. Before taking up point two, which is by far the most

important, a few words will suffice to explain why one may leave aside
issues one and three:
1. It is true that Anonymus C trusted an authority which we know
today is not the one he thought it to be. However, it doesn’t really
matter whether he took so basic a principle of Aristotelian doctrine
as the distinction between paralogisms in dictione and extra
dictionem –directly– from Aristotle or –indirectly– from Pseudo-
Alexander. Furthermore, the enigma of the «Latin Alexander» has
been brilliantly solved by Sten Ebbesen, who has proved that the

alleged Latin fragments of a lost Greek commentary by Alexander

of Aphrodisias are, in fact, a set of scholia James of Venice drew
from Micheal of Ephesus’ commentary on Aristotle’s Sophistici
3. It is also true that Anonymus C’s morphological speculations are
completely beside the point. For one thing, he must have known
very little Greek in order to miss the fact that, while «homo» and
«Coriscus» do not bear any resemblance, Κορίσκος and ἄνθρωπος
do have the same form of expression. For another, he must have
had very peculiar linguistic notions in order to believe that one may
explain anything about what happens in his own language because
of what happens in another. However, since the argument is an
aristotelian example of the fallacy of accident45, it would reach its
conclusion whatever the morphology of the words involved.

TRANSLATION MATTERS. When it comes to the second issue we

encountered, it is an entirely different story. In fact, treating the «Coriscus
other than himself» argument as a fallacy of figure of speech provides us
with a lead as to how and why accidens ended up playing a key role in
a discussion whose focus is figura dictionis, a different type of fallacy
altogether. Since Anonymus cordubensis is far from being an isolated
case, there’s only one plausible explanation for the consensus Latins
reached over this particular issue, which is that the connection between the
«Coriscus other than himself» argument and the fallacy of figure of speech
was hinted at in Aristotle’s text itself. When Latin scholars rediscovered
Aristotle’s Sophistici elenchi –that is, about the time of Peter Abelard,
Thierry of Chartres and Adam of Balsham (mid-XIIth century)– it was by
and large through Boethius’ translation46. Not only did Boethius take a few

S. EBBESEN himself tells the story of «Alexander»’s recovery in «The Greek
under the Latin and the Latin under the Greek», Greek-Latin Philosophical Interaction.
Collected Essays of Sten Ebbesen. Volume 1, Ashgate, Aldershot – Burlington 2008,
pp. 1-7, a text philologists and philosophers alike should read each and every morning
before starting to work.
Aristotelis sophistici elenchi 5, 166b 32-33: «οἷον “εἰ ὁ Κορίσκος ἕτερον
ἀνθρώπου, αὐτὸς αὑτοῦ ἕτερος· ἔστι γὰρ ἄνθρωπος” [for instance, if Coriscus is
other than man, he is other than himself, for he is a man]».
Cf. L. MINIO-PALUELLO, «Boezio, Giacomo Veneto, Guglielmo di Moerbeke,
Jacques Lefèvre d’Étaples e gli “Elenchi sophistici”», Rivista di filosofia neo-scolas-

liberties with the text –mainly in the choice of examples47– but –what is
more– he changed for good the face of at least one argument related to the
form of expression.
much labour to fully discuss the «Third man» argument. Besides, as it will
soon become clear, its reconstruction is not –strictly speaking– required. A
minimal account will do48. As conveyed by its conclusion (ὅτι ἔστι κτλ.) –

tica, 44 (1952) 399-400; «Jacobus Veneticus Grecus. Canonist and Translator

of Aristotle», Traditio, 8 (1952) 265-304; «Gli “Elenchi sophistici”: redazioni
contaminate colla ignota versione di Giacomo Veneto (?); frammenti dello ignoto
commento d’Alessandro di Afrodisia tradotti in latino», Rivista di filosofia neo-
scolastica, 46 (1954) 222-231; «Giacomo Veneto e l’Aristotelismo Latino», in
A. PERTUSI (ed.), Venezia e l’Oriente fra tardo Medioevo e Rinascimento, Sansoni,
Firenze 1966, pp. 53-74. B. G. DOD, Praefatio, in Aristoteles latinus. De sophisticis
elenchis. Translatio Boethii, Fragmenta Translationis Iacobi et Recensio
Guillelmi de Moerbeke, Brill, Leiden 1975, pp. XII-XIV; «Aristoteles latinus», in
N. KRETZMANN – A. KENNY – J. PINBORG (eds.), The Cambridge History of Later
Medieval Philosophy. From the Rediscovery of Aristotle to the Disintegration
of Scholasticism (1100-1600), Cambridge University Press, Cambridge 1982,
pp. 45-79. S. EBBESEN, «Jacobus Veneticus on the Posterior Analytics and Some
Early 13th Century Oxford Masters on the Elenchi», CIMAGL, 21 (1977) 1-9;
«Anonymi Aurelianensis I commentarium in Sophisticos elenchos. Introduction:
Boethius, Jacobus Veneticus, Michael Ephesius and “Alexander”», CIMAGL,
34 (1979) p. XXXVII; «Review Article. Union Académique Internationale Corpus
Philosophorum Medii Aevi. Academiarum consociatarum auspiciis et consilio
editum. Aristoteles Latinus VI I-3 De Sophisticis Elenchis Translatio Boethii,
Fragmenta Translationis Iacobi, et Recensio Guillelmi de Moerbeke, Edidit
Bernardus G. DOD, Brill, Leiden & Desclee de Brouwer, Bruxelles 1975, pp. XLII
+ 152», Vivarium, 17 (1979) 69-80.
A few Westerners seem to have been aware of the fact that Boethius did not
translate but rather adapted (from Vergil and Horace) two of Aristotle’s examples.
Cf. Anonymi summa sophisticorum elenchorum, p. 326.1-8; Anonymi parisiensis
compendium sophisticorum, p. 84.23-28; Anonymi aurelianensis I commentarium
in Sophisticos elenchos, p. 123.3-4 and p. 123.26-33; Anonymi Cantabrigiensis
commentarium in Aristotelis sophisticos elenchos, 88Vb. Cf. L. MINIO-PALUELLO, «The
Text of Aristotle’s Topics and Sophistici elenchi. The Latin Tradition», The Classical
Quarterly, 5 (1955) 110.
Interested readers will find a more detailed story in L. GAZZIERO, «“Et
quoniam est quis tertius homo”. Argument, exégèse, contresens dans la littérature
latine apparentée aux Sophistici elenchi d’Aristote», Archives d’histoire doctrinale
et littéraire du Moyen Âge, 80 (2013) 7-48. Relevant sources have been edited in
L. GAZZIERO, «The Latin “Third Man”», op. cit., pp. 11-93.

which is, anyway, all Aristotelis sophistici elenchi, 22, 178b 36 - 179a 10 has
to offer– the agument aims at inferring the existence of a third man (a τρίτος
ἄνθρωπος precisely) besides man himself (παρ’ αὐτὸν) and individual
men (καὶ τοὺς καθ’ ἕκαστον). Even if Aristotle’s main concern is to
explain how to avoid the «Third man» rather than to recount the argument
itself, one can be reasonably sure that the decisive move is to bring about
the idea that the universal is on a par with the particulars whose universal
it is, for –as Aristotle says in 179a 4-5– «οὐτὸ ἐκτίθεσθαι δὲ ποιεῖ τὸν
τρίτον ἄνθρωπον, ἀλλὰ τὸ ὅπερ τόδε τι εἶναι συγχωρεῖν [setting apart
does not produce the “Third Man”, rather the admission that <what is set
apart> is an individual thing]». All of which makes excellent sense, given
that Aristotle thought that –no matter how dire the consequences– being
wrong about which is which between particulars and universals is an easy
mistake to make and, more to the point, a mistake we cannot easily avoid
because of what can only be described as a fact of language. In fact, not
only is it most natural to assume that everything we say refers to something
that exists49, but such delusion is all the more likely to occur when talking
about substances, as one can easily draw from a well-known passage of
Aristotle’s Categories50, where the way we name things is held responsible

Aristotelis sophistici elenchi, 6, 168a 25-26: «σύνηθες γὰρ τὸ πάντα ὡς τόδε
τι σημαίνειν [we usually speak of everything as though it were an individual thing]»;
cf. 7, 169a 33-34: «πᾶν τὸ κατηγορούμενόν τινος ὑπολαμβάνομεν τόδε τι [we
trust everything predicated of something else to be an individual thing]».
Aristotelis Categoriae, Ed. by R. BODÉÜS, Les Belles Lettres, Paris 2001, 5,
3b 10-18: «πᾶσα δὲ οὐσία δοκεῖ τόδε τι σημαίνειν. ἐπὶ μὲν οὖν τῶν πρώτων
οὐσιῶν ἀναμϕισβήτητον καὶ ἀληθές ἐστιν ὅτι τόδε τι σημαίνει· ἄτομον γὰρ
καὶ ἓν ἀριθμῷ τὸ δηλούμενόν ἐστιν. ἐπὶ δὲ τῶν δευτέρων οὐσιῶν ϕαίνεται μὲν
ὁμοίως τῷ σχήματι τῆς προσηγορίας τόδε τι σημαίνειν, ὅταν εἴπῃ ἄνθρωπον
ἢ ζῷον· οὐ μὴν ἀληθές γε, ἀλλὰ μᾶλλον ποιόν τι σημαίνει, - οὐ γὰρ ἕν ἐστι τὸ
ὑποκείμενον ὥσπερ ἡ πρώτη οὐσία, ἀλλὰ κατὰ πολλῶν ὁ ἄνθρωπος λέγεται
καὶ τὸ ζῷον. οὐχ ἁπλῶς δὲ ποιόν τι σημαίνει, ὥσπερ τὸ λευκόν· οὐδὲν γὰρ
ἄλλο σημαίνει τὸ λευκὸν ἀλλ’ ἢ ποιόν, τὸ δὲ εἶδος καὶ τὸ γένος περὶ οὐσίαν τὸ
ποιὸν ἀϕορίζει -, ποιὰν γάρ τινα οὐσίαν σημαίνει [it looks like every substance
refers to an individual thing. As far as primary substances are concerned it is true and
undisputable that each refers to an individual thing, for what we refer to is something
particular and one in number. As for the secondary substances, on the other hand,
they seem to refer to an individual thing, as when we say “man” or “animal”, on
account of the form of denomination (τῷ σχήματι τῆς προσηγορίας). Yet, this is
not true, for they rather signify something that is such and such. In fact, the subject
is not one, as in the case of primary substances; in fact, man and animal are said

for this confusion: the form of appellatives being roughly the same in all
cases, it is not by means of a linguistic analysis that one shall settle whether
a given name refers to a particular or to a universal thing. Be that as it may,
Boethius’ translation of «καὶ ὅτι ἔστι τις τρίτος ἄνθρωπος παρ’ αὐτὸν
καὶ τοὺς καθ’ ἕκαστον» was to change the game altogether, since it sounds
«et quoniam est quis tertius homo A SE ET AB UNOQUOQUE», which is as literal
as it gets, except for the fact that Boethius translated as if his Greek model
read αὐτὸν instead of αὑτὸν51. As a result, Latin commentators understood
the expression tertius a se as if it meant either diversus a se (different from
himself) or alter a se (other than himself), which –as far as I know– they did
without exception52. Such understanding proved fatal for the «Third man».

of many things. That being said, secondary substances do not mean a quality tout
court, as white does. As a matter of fact, white means just a quality, whereas species
and genus have more to do with a determination of the substance, they rather signify
what qualifies a given substance as the substance it is]». For all practical purposes, the
σχῆμα τῆς προσηγορίας is synonymous with the σχῆμα τῆς λέξεως of the fallacy
of the same name: προσηγορία is the denomination according to a certain name (the
term occurs in Aristotle’s definition of παρώνυμα in chapter one of the Categories:
«παρώνυμα δὲ λέγεται ὅσα ἀπό τινος διαϕέροντα τῇ πτώσει τὴν κατὰ τοὔνομα
προσηγορίαν ἔχει, οἷον ἀπὸ τῆς γραμματικ ῆ ς ὁ γραμματικὸς καὶ ἀπὸ τῆς
ἀνδρείας ὁ ἀνδρεῖος [we call paronym things that are named after something else
through a flexion of the name of the latter: for instance, grammarian is named after
grammar and brave after bravery]» (Aristotelis Categoriae, 1, 1a 12-15). Both notion
and their relation have been thoroughly dealt with by F. Ildéfonse, «Ta skhêmata tês
lexeôs», in M. S. CELENTANO – P. CHIRON – M.-P. NOËL (eds.), Skhèma/Figura. Formes
et figures chez les Anciens. Rhétorique, philosophie, littérature, Editions Rue d’Ulm/
ENS, Paris 2004, pp. 143-157.
Since it is more than likely that the oncial letters of his manuscript lacked
diacritical marks, it does not come as a surprise that Boethius got the breathing wrong.
Roberti Grosseteste quod fertur commentarium in Sophisticos elenchos, Ed. by
L. GAZZIERO, «The Latin “Third Man”», op. cit., pp. 30-32; Petri hispani portugalensis
tractatus VIII, pp. 141.31-143.19; Anonymi Monacensis Commentarium in
Sophisticos Elenchos, pp. 33-44; Roberti Kilwardby (?) commentarium in Sophisticos
Elenchos, pp. 45-53; Nicholai Parisiensis Notulae super librum elenchorum, 54-59;
Roberti <Kilwardby ?> commentarium in sophisticos elenchos, pp. 60-67; Roberti
de Aucumpno commentarium in Sophisticos Elenchos, pp. 68-82; Alberti magni
expositio sophisticorum elenchorum, pp. 938b-939a; Roberti anglici commentarium in
Sophisticos elenchos, Ed. by L. GAZZIERO, «The Latin “Third Man”», op. cit., pp. 83-
87; Thomae de Aquino (?) de fallaciis IX, p. 411.66-81; Anonymi salmaticensis-
florentini quaestiones super Sophisticos elenchos, q. 80, pp. 179-185; Aegidii Romani
expositio super libros Elenchorum, 54va-55rb; Anonymi bavarici lectura super librum

To make another long story short, this is how the argument was refined out
of existence: Latin commentators just lost sight of the «Third man» as an
argument and resorted to those they had at hand in order to make sense of
what they read. As it happens, they had the good fortune and the flair to
find a perfect match in Aristotle’s discussion of the fallacies of accident.
As true to the text as Boethius allowed them to be, they were brought to
take the «tertius a se» apart from the «ab unoquoque». Instead of thinking
of them as two clauses of one and the same conclusion («there is a “third
man” beside man and individual men»), Latin commentators considered
the «tertius a se» and the «ab unoquoque» as two different conclusions («a
given man is other than himself» and «he is other than any other man»).
Therefore, if there are two conclusions instead of one, there are also two
arguments rather than one. This split issue became the standard story and
we find it repeated time and again in the XIIIth and XIVth centuries53. If the
second line of reasoning, which stipulates that a man differs from any other
(«ab unoquoque»), is usually treated as a mere repetition of the first, whose
conclusion is that a man differs from himself («a se»)54, Latin commentators
displayed sometimes an uncanny ingenuity. The palm of sophistication goes
to Anonymus bavaricus who most of the time rates as an average, run-of-
the-mill commentator. On this occasion, nevertheless, he outdid himself

Elenchorum, Ed. by L. GAZZIERO, «The Latin “Third Man”», op. cit., pp. 88-91;
Anonymi Cordubensis quaestiones super sophisticos elenchos, q. 838, pp. 362-366;
Simonis de Faverisham quaestiones novae super libro elenchorum, q. 19, pp. 148-150;
Ioannis Duns Scoti quaestiones super librum elenchorum Aristotelis, q. 41, pp. 459-
463; Anonymi expositio super libros Elenchorum, Ed. by L. GAZZIERO, «The Latin
“Third Man”», op. cit., pp. 92-93.
Anonymus monacensis is as good an example as any and more explicit than
most: «“QUONIAM QUIS HOMO”, id est aliquis homo, “EST TERTIUS”, id est diversus a se
ipso. Et per hoc habetur conclusio primi paralogismi. “ET AB UNOQUOQUE”, id est aliquis
homo potest concludi esse diversus ab unoquoque alio. Et per hoc habetur conclusio
secundi paralogismi [“QUONIAM QUIS HOMO”, that is: a certain man; “EST TERTIUS”, that
is: is different from himself; and we have here the conclusion of the first sophism.
“ET AB UNOQUOQUE”, that is: it may be inferred that a certain man is different from any
other. And we have here the conclusion of the second sophism]» (Anonymi monacensis
commentarium in Sophisticos Elenchos, p. 35).
Anonymi monacensis commentarium in sophisticos elenchos, p. 35; Roberti
<Kilwardby> commentarium in sophisticos elenchos, p. 46; Roberti de Aucumpno
commentarium in sophisticos elenchos, p. 69; Alberti magni expositio sophisticorum
elenchorum, p. 938b 21-33; Roberti anglici commentarium in Sophisticos elenchos,
p. 83; Aegidii romani expositio super libros elenchorum 54va 3-7.

(and everybody else in the process). As a matter of fact, not only did he pull
two arguments out of his sleeve where there was only one to begin with,
but he also managed to match them perfectly with the first and the second
example Aristotle offered of the fallacy of accident55. If symmetry is the seal
of truth, seldom two wrongs came so close to make a right.
EPILEGOMENA. Whether in its more subtle formula or in its ordinary
capacity, the appeal of this solution was immensely strong. While nothing
in Aristotle’s words pointed in its direction, the association of the «Third
man» and the «Coriscus other than himself» sophism proved so successful
that more than one Latin reader believed he found it in Aristotle’s text
itself. Around 1280, Simon of Faversham –who was neither the first nor the
last to do so56– quoted Aristotle himself as the final authority on the issue:

Anonymi bavarici lectura super librum Elenchorum, p. 88: «Primus paralogismus
formatur sic: “ab homine Coriscus est alter; Coriscus est homo; ergo, alter a se”.
Conclusionem solum ponit cum dicit: “ET QUONIAM QUIS”: id est, aliquis est homo tertius
a se et paralogyzetur sicut dictum est. Tunc ponit alium cum dicit: “Coriscus est alter a
Platone et Plato est homo, ergo alter ab homine”. Conclusionem ponit cum dicit “ET AB
UNOQUOQUE” [the first paralogism goes like this: “Coriscus is other than man; therefore
he is other than himself”. Aristotle states the conclusion only: “ET QUONIAM QUIS”, that
is to say: some man is other than himself, and the paralogism is argued for as it has
been said. Then he brings about the other paralogism: “Coriscus is other than Socrates;
Socrates is a man; therefore Coriscus is other than man”. Aristotle states the conclusion
when he says: “ET AB UNOQUOQUE”]». NOTA BENE: «ab homine Coriscus est alter; Coriscus
est homo; ergo, alter a se» is none other than the first fallacy of accident Aristotle
discussed at the beginning of chapter 5 of Sophistici elenchi (quoted above, note 45);
while «Coriscus est alter a Platone et Plato est homo, ergo alter ab homine» –apart from
the fact that Plato has replaced Socrates– is pretty much the same as the second fallacy
of accident Aristotle discussed right after the first we just mentioned: «ἢ εἰ Σωκράτους
ἕτερος, ὁ δὲ Σωκράτης ἄνθρωπος, ἕτερον ἀθρώπου ϕασὶν ὡμολογηκέναι διὰ
τὸ συμβεβηκέναι οὗ ἔϕησεν ἕτερον εἶναι, τοῦτον εἶναι ἄνθρωπον [otherwise, if
Coriscus is other than Socrates, since Socrates is a man, they pretend that it has been
admitted that he is other than man because of the fact that man is predicated of Socrates
and Coriscus is said to be other than Socrates]» (Aristotelis sophistici elenchi 5, 166b
Cf. Petri hispani portugalensis tractatus VII, p. 142.4-8; Roberti <Kilwardby>
Commentarium in Sophisticos elenchos, p. 51; Nicholai parisiensis notulae super
librum elenchorum, p. 54; Alberti magni expositio sophisticorum elenchorum,
p. 939a; Anonymi salmaticensis-florentini quaestiones super sophisticos elenchos,
q. 80, p. 179.21-23 and q. 85, p. 196.16-21; Anonymi Cordubensis quaestiones super
sophisticos elenchos, q. 838, p. 363.32-33; Ioannis Duns Scoti quaestiones super
librum elenchorum Aristotelis, q. 41, p. 460.14-17.

«Philosophus dicit, secundo huius, quod hic est figura dictionis: “Coriscus
est alter ab homine; ergo, Coriscus est alter a se” [says the Philosopher, in
the second book <of the Sophistici elenchi>, that we have here a case of
fallacy of figure of speech: “Coriscus is other than man ; therefore he is
other than himself”]»57. It would not be historically accurate to say that the
collapse of the Latin «Third man» brought alone the fallacy of accident and
the fallacy of figure of speech together. That said, the exegetical anomaly
it turned out to be soon became the single strongest reason in favour of this
unlikely association. At any rate, it proved compelling enough to make
Latin commentators wonder wether some fallacies of form of expression
fall outside expression itself and to persuade some of them that, contrary to
all expectations, this may well be the case.

Simonis de Faverisham quaestiones novae super libro elenchorum, q. 19,
p. 148.17-18. As is well known, a distinctive feature of the Latin tradition of Aristotle’s
tract was its division in two books: the first ends at 16, 175 where the second begins.
This division is –in all probability– a Latin invention, for there is no trace of it in
Aristotle nor in the Greek and Byzantine tradition. That being said, it is far from
arbitrary. As a matter of fact, the first fifteen chapter of Aristotle’s work’s focus is upon
sophistic objectives (metae) and techniques (fallaciae). From chapter 16 on, Aristotle’s
attention turns to the ways we can counter or neutralize sophistic arguments.


Duns Scotus introduced the notion of logical potency or possibility,

which he discussed in various parts of his philosophical and theological
treatises. Even though he did not write any separate study of this concept,
it had a considerable influence on philosophy and logic in the fourteenth
century and later. In this paper I would like to pay attention to the logical
layer of Scotus’s modal theory. I am particularly interested in his view that
logical modalities are what they are independently of whether God exists
or not and how this idea is related to Scotus’s assumption that the realm
of intelligibility, which is not limited by anything but logical possibility, is
actually infinite and cannot be other than it is.

1. Modal Metaphysics

In his modal metaphysics, Scotus famously makes use of a model of

divine psychology in which various acts are separated by the instants of
nature; that is, they are distinguished from each other conceptually but
not temporally. This flexible tool is employed in a pragmatic rather than a
systematic way. Scotus says, for example, that in the first instant of nature
God understands itself as such and in the second instant everything that
might be created, giving this collection an objective being as the object
of understanding (esse intelligibile)1. The content of the actually infinite
aggregate of understandable objects is said to have esse possibile with
respect to divine power2. Scotus’s approach to intelligible possibilities
is propositional in the sense that he describes them as what is meant by

University of Helsinki, P.O. Box 33 (Yliopistonkatu 4) 00014 University of
Helsinki, Finland. Email: simo.knuuttila@helsinki.fi
Duns Scotus, Ordinatio I.35, n. 32; I.43, n. 14-16 (John Duns Scotus, Opera omnia,
Typis Polyglottis Vaticanis, Civitas Vaticana 1950-, VI, 258, 358-360). For the intentional
or objective being of things, see Ordinatio I.27.1-3, n. 54 (Ed. Vat. VI, 86); Ordinatio
IV.1.2, n. 3 (John Duns Scotus, Opera omnia, ed. by L. WADDING, Lyon 1639, 8, 56-57).
Ordinatio I.43, n. 14 (Ed. Vat. VI, 358-359). For the actual infinity, see note 5

complexes or propositions. In describing the logical potency which forms

the basic level of all possibilities, Scotus writes:

This potency is a modus of a composition of intellect, and it is caused

by the disposition of the terms of the composition, namely, that they
are not repugnant. And though there often is a corresponding real
potency in reality, it is not an essential part of the nature of this
potency. In this way it was possible that the world will exist before
its creation, if there was an intellect forming this composition ‘The
world will exist’3.

The possibility of what is expressed by the proposition ‘The world

will exist’ exemplifies Scotus’s notion of being which, in its most general
sense, refers to everything to which to be is not repugnant4. The possibilities
associated with a real activating power are called metaphysical or real
possibilities, and are formally based on the logical potency or compatibility
between the meanings of the terms used in the description of possibilities.
The coherence of a proposition requires, of course, that the terms signify
things which might be5.
Contrary to what some of Scotus’s followers assumed, he did not mean
that God’s thought is required for the truth of true possibility propositions.

Quaestiones super libros Metaphysicorum Aristotelis VI-IX, Iohannis Duns
Scoti Opera Philosophica 6, ed. by R. ANDREWS et al., The Franciscan Institute, St.
Bonaventure 1997, IX.1-2, n. 18 (514). This is meant to be an explanation of that notion
of possibility in Aristotle which is not based on a potency. Many thirteenth-century
authors refer to this conception in defining divine omnipotence as a power to actualize
what is not contradictory; see J. SCHMUTZ, «Qui a inventé les mondes possibles?», in
J.-C. BARDOUT – V. JULLIEN (eds.), Les mondes possibles, Presses Universitaires de
Caen, Caen 2006, pp. 9-45.
Scotus argues, for example, that when Aristotle says that in research into things
one should first ask whether it is before asking what it is, this ‘whether it is’ (si est)
should be not taken to refer to actual existence but to being as that to which existence is
not repugnant. See Ordinatio IV.1.2, n. 8; L. M. DE RIJK, «Einiges zu den Hintergründen
der scotischen Beweistheorie: die Schlüsselrolle des Sein-Könnens (esse possibile)»,
in A. ZIMMERMANN (ed.), Kölner Universität im Mittelalter, Miscellanea mediaevalia 7,
de Gruyter, Berlin 1989, p. 183.
For a detailed analysis of relevant texts, see L. HONNEFELDER, Scientia
transcendens. Die formale Bestimmung der Seiendheit in Realität in der Metaphysik
des Mittelalters und der Neuzeit (Duns Scotus – Suárez – Wolff – Kant – Peirce), Felix
Meiner, Hamburg 1990, pp. 3-56.

On the contrary, he says in many places that what is possible is possible

because of itself, not because of something else. (See below.) Logical
potencies or possibilities as compatibilities of meanings are not entities and
have no kind of existence, but are like norms for any complex formed in
understanding or realized by a power. Scotus’s assumption of the objective
being of all coherent combinations in divine intellect did not mean that
God created their logical possibility or is otherwise their source. This is
not required because they are formally coherent by themselves. However,
he believed that because the logical potency is not a power of existence,
there is no path from logical possibilities to actuality except through the
activating power of an existing being6.
Possibilities are structured on the basis of the relation of compossibility.
For each non-necessary possible state of affairs, its denial is also
possible with respect to the same time, though not compossible. Logical
impossibilities are parasites in the order of possibilities, being incompatible
combinations between possible ingredients. Such combinations lack logical
potency and consequently represent absolute nothingness7. The systematic
role of compossibility in Scotus’s modal considerations is shown by his
characterizing logical potency or possibility as non-incompatibility between
the terms8. In his identity view of predication, the compatibility between
significant units means their being simultaneously predicable of the same
subject, so that the result does not represent the nothingness of impossibility9.
Scotus explains his idea of possibility and truth by stating that at one
instant of nature propositions about logically contingent things are present
in the divine intellect without a truth value and at another instant the divine
choice makes a number of propositions true; namely, those which express
things included in divine providence pertaining to the actual world10. This

See Ordinatio I.43, n. 14 (Ed. Vat. VI, 358).
Ordinatio I.43, n. 16-18 (Ed. Vat. VI, 359-361); cf. Lectura I.39, n. 72 (Ed. Vat.
XVII, 504).
Ordinatio I.7.1, n. 27 (Ed. Vat. IV, 119).
For Scotus’s view of predication, see G. PINI, «Scotus on Assertion and the
Copula: A Comparison with Aquinas» in A. MAIERÙ – L. VALENTE (eds.), Medieval
Theories on Assertive and Non-Assertive Language. Acts of the 14th European
Symposium on Medieval Logic and Semantics, Olschki, Florence 2004, pp. 320-332
(Lessico Intellettuale Europeo, 97).
«But the divine intellect offers the terms of a future contingent complex to
his will as neutral concerning it. The terms of the future contingent do not include

corresponds to his general view of the psychology of intellection, according

to which propositions are understood as being neutral before they are
understood as being true or false. Understanding the truth is a separate act
about the relation between the proposition and what it signifies11. Antonie
Vos and some others following him have stressed that God chooses those
propositions to be true which express the states of affairs to be realized
in the actual world, without making other propositions true with respect
to possible worlds12. Scotus did not use the term ‘possible world’ and no
medieval thinker developed anything like a systematic theory of various
relations between possible worlds, which is typical of the possible worlds
semantics of the last century. In spite of this, there are some similarities.
According to possible worlds semantics, modality could be understood as
multiplicity of reference with respect to alternatives, and this was a central
idea in Scotus’s theory as well13.

knowledge of the contingent complex, because the terms are not the cause of such
truth because then it would be immediately true. Therefore the divine intellect has a
neutral cognition of such terms before an act of the will, that is, the choice of the actual
providence. […] And so contingents of this kind are true because their truth is first
caused by an act of the divine will. It is not the case that because these contingents
are true, the will wills them to be true, but rather the reverse is the case. And therefore
when the truth caused in a complex of such terms is determined by an act of will, then
the divine intellect first knows one part of a contradiction of contingents to be true.»
(Duns Scotus, Reportatio IA, 38. 1-2, n. 37, transl. by G. FROST in «John Duns Scotus
on Godʼs Knowledge of Sins: A Test-Case for God’s Knowledge of Contingents»,
Journal of the History of Philosophy, 48 (2010) 17-18).
“The intellect forms and apprehends many propositions in the second act, which
in fact are neutral to it, according to Book I of the Topics. Although there is a formal
truth or falsity in that act, depending on whether there is a conformity to external
things or not, it is not there objectively because this conformity is not apprehended”
(Quaestiones super libros Metaphysicorum Aristotelis VI.3, 37 (69)).
A. VOS, The Philosophy of John Duns Scotus, Edinburgh University Press,
Edinburgh 2006, pp. 271-273, 492.
Many late medieval thinkers, such as Scotus, Ockham, and Buridan, distanced
themselves from traditional interpretations of Aristotle’s modal syllogistic and other
received views of necessity and possibility. While the modal paradigms criticized, most
typically the frequency model for necessity and possibility, were derived from ancient
philosophy, the new concepts associated with the idea of modality as alternativeness were
for the most part medieval, first suggested by some twelfth-century figures and further
developed in late medieval times, after the less innovative thirteenth-century Aristotle
reception. Some elements of late medieval modal insights were embedded in Leibniz’s

Scotus’s remarks on neutral propositions were associated with his

theological view of why the actual world is as it is. One set of possibility
propositions is true because of God’s decision to create the actual world.
This choice could have been otherwise and correspondingly the contingent
truths could be other than they are. Scotus states about alternative
providential plans:

God contingently predestinates whom he predestinates and he

may not predestinate that person, not both simultaneously nor
successively, but both separately, in an instant of eternity. (Ord.
I.40, n. 5, Ed. Vat. VI, 310)

If ‘Socrates will be beatified’ is contingently true, ‘Socrates will not

be beatified’ would be true if the divine providential plan were different.
It would not be true in a Leibnizian counterpart world, but it could be true
in relation to an alternative of the actual world. It is also interesting that
both alternatives are about the same possible person, not about separate
counterparts as in Leibniz’s possible worlds or in Lewis’s parallel universes.
The basic structure of Scotus’s modal metaphysics can be summarized
as follows. The actual world is a state of affairs which obtains. There is a
complete set of propositions which are true about the contingent things in
the actual world and an infinite number of possibility propositions which
are not true about it. For each temporally definite proposition p, the terms
of which are not incompatible, there is a possible state of affairs S such that
if S had been actual, p would have been formally true. The same holds for
the conjunctions of propositions which are compossible. This is the kernel
of the formal semantics of modality in Scotus. The assertoric part of a
possibility proposition is formally true about the actual world or would be
true if the world were different in a relevant way.

philosophy of possible worlds and known through it by later writers, but otherwise these
developments were hardly mentioned in philosophical works until the middle of the last
century, when many scholars began to investigate medieval logic and semantics, including
modal logic and modal theories. This interest was stimulated by the lively discussion of
modalities in the 60s and 70s and also, but not only, by the observation that, as distinct
from ancient theories, there were some similarities between the philosophical assumptions
of medieval theories and Kripkean possible worlds semantics, most obviously the
association of the meaning of modal terms with the idea of simultaneous alternatives.
See S. KNUUTTILA, «Modality», in J. MARENBON (ed.), The Oxford Handbook of Medieval
Philosophy, Oxford University Press, New York 2012, pp. 312-341.

2. The Actually Infinite Content of Intelligibility

In arguing for the infinity of divine omniscience, Scotus explains that

when an omniscient intellect thinks about all thinkable things, the content
of this act consists of an actually infinite number of intelligible things. His
main interest is to shed light on the intensive infinity of divine knowledge,
and he thinks that this kind of infinity is required by the actual quantitative
infinity of the content of the faculty:

Are there not […] infinite intelligibles and are these not actually in
an intellect which actually understands everything? Therefore the
intellect which actually understands all things at once is infinite […]
There can be no end of things which are potentially infinite, taking
one after the other. If all these things are actually simultaneous, they
are actually infinite. But intelligibles are like this with respect to a
created intellect, as is clear, and in you [God] all things, which are
intelligible for a creature successively, are actually simultaneously
understood. Therefore there are there actually infinitely many things
understood. (De primo principio IV.9, n. 68)14

Beginning from the Aristotelian notion of potential infinity, which was

the standard concept in natural philosophy of his time, Scotus moves to the
actual quantitative infinity which he regards as not inconsistent in itself15.
The intentional correlate of the divine intellect, the realm of what can be
understood, is actually infinite and the same for any omniscient intellect,
since the intellect is a power of grasping what is intelligible, and this is
necessarily what it is for any intellect16.
The idea of the infinite totality of what can be understood was used by
Scotus in many places, one of these being his account of human psychology
in the hypostatic union. Aquinas and Scotus follow Peter Lombard in

John Duns Scotus, Tractatus de primo principio / Abhandlung über das
erste Prinzip, ed. with a translation and notes by W. KLUXEN, Wissenschaftliche
Buchgesellschaft, Darmstadt 1974. The translation is taken, with small modifications,
from R. CROSS, Duns Scotus on God, Ashgate Studies in the History of Philosophical
Theology, Ashgate, Aldershot 2005, pp. 91-92.
Quodlibet V, n. 2-3 (ed. WADDING XII, 118), CROSS, Duns Scotus on God,
pp. 93-99.
Lectura I.2.1.1-2, n. 57 (Ed. Vat. XVI, 131); Ordinatio I.3.1.4, 268 (Ed. Vat.
III, 163-164); I.38. n. 9 (Ed. Vat. VI, 306).

assuming that the cognitive capacities of Christ were as perfect as possible.

According to Aquinas, Jesus Christ had three sorts of knowledge: the beatific
vision of God’s essence as a grace-based perfection, the infused knowledge
of things in the world which in principle can be naturally known, and the
standard human knowledge through abstraction from the phantasms. Christ
saw the essence of God, as far as this was possible for a finite mind, and
all things in the created world through participation in the divine light in
the beatific vision. As for the infused knowledge, Aquinas states that the
soul of Christ knew everything in the Word, that is, in the second person of
the Trinity, which was the divine part of the hypostatic union. The infused
knowledge which the intellect received from the Word was partially the
same as the first knowledge, but Aquinas treated this as the perfection of
the human passive intellect, which was activated by immediate divine
causation. Christ’s habitual infused knowledge made him omniscient about
things in the created order. He was not fully omniscient, however, since he
did not have knowledge of God’s unrealized possibilities. Aquinas believed
that ontological possibilities were determined by the essence of God and
that God could have created things quite different from the actual ones.
These unrealized metaphysical possibilities are not known to created finite
minds, which have only partial knowledge of God’s infinite essence17.
Deviating from Aquinas, Scotus argued that, apart from the immediate
vision of the Godhead, the soul of Christ saw in the Word everything the
omniscient Word saw, i.e., all facts about the actual world and its history
as well as the actually infinite number of unrealized possibilities. This
epistemic optimism was based on the view that possibilities are expressed
by non-contradictory propositions which are similarly intelligible to
any intellect. The statement that Christ could actually see in the Word
everything the Word saw may be understood in two different ways:

One way of understanding this is that the soul of Christ has one
vision about the Word as the primary object and about all objects
which shine in the Word as secondary objects, these secondary

Thomas Aquinas, Summa theologiae III.9-12 (Opera Omnia, ed. Leonina
XI, Rome, 1888-1906, 138-170). This paragraph is derived from S. KNUUTTILA,
«The Psychology of Incarnation in John Duns Scotus», in K. EMERY JR. et al. (eds.),
Philosophy and Theology in the Long Middle Ages: A Tribute to Stephen F. Brown,
Studien und Texte in Geistesgeschichte des Mittelalters 105, Brill, Leiden 2011,
pp. 737-748.

objects being not separately attended to. It does not follow that
an infinite act is required to found these attentions, because they
are merely potential, and so no actual infinity is assumed since the
object is not actual. In a second way it can be taken to mean that
that there is a proper vision of each object and therefore an infinite
number of actual visions is simultaneously received from the Word
as a cause. This second way demands that something infinite is
assumed, which seems to contradict many authoritative opinions of
the philosopher and saints18.

Is the knowledge of infinite possibilities present in the soul in one act or

several acts? While not arguing for the actually infinite number of separate
acts in the human intellect of Christ, Scotus assumes that there might be an
actually infinite number of simultaneous acts in a finite intellect, if they are
directly caused by a divine power. He wonders why should there be any
limit, if there may be several simultaneous acts or simultaneous distinct
aspects in one act. A third possibility is that the Christ may see in the Word
everything he wants to see as in a mirror –while the whole content is not
simultaneously actual, he always has access to the infinite knowledge in the
Word. This is the view which Scotus seems to regard as most probable19.
Scotus’s somewhat surprising considerations of Christ’s knowledge
show his interest in the conception of actual infinity, which he mostly
treated as part of his idea of the intensive infinity of divine attributes. The
assumption of infinite things in esse intelligibile is part of the theological
model of God’s eternal choice between possible alternatives. When the
basic criterion of the ways things could be is their consistent description,
their number at this level is obviously infinite. As all possibilities are
knowable by divine omniscience, their being known is best guaranteed by
assuming that they form an actually infinite totality20. Such a totality could
be formed by one and the same proposition as infinitely multiplied, but
Scotus’s idea is, of course, that the propositions are different and express
the infinity of the various ways things could be.

Ordinatio III, dist. 14, q. 2, nn. 58 and 68 (Ed. Vat. IX, 449-452).
Ordinatio III.14.2, n. 58-76 (Ed. Vat. IX, 449-455).
Thomas Aquinas argued that an actual quantitative infinity is impossible
because it cannot form a definite quantity; see Summa theologiae I.7.4.

3. The Source of Possibility

Scotus says that when God thinks about something which could
become actual, it is possible formaliter ex se and principiative through
God. The latter qualification refers to its being produced in esse intelligibile
by divine omniscience and then becoming included in esse possible with
respect to divine omnipotence. These formulations are understood by
some recent commentators as follows. Fabrizio Mondadori argues that
the formal aspect of the possible is not dependent on God, but possibility
is predicated of something which is apparently a thought in God’s mind.
Tobias Hoffmann argues, like Mondadori, that Scotus denies that God is
responsible for giving things the possibility of being. While possibilities do
not depend on God in this sense, only God can initiate the eidetic natures of
creatures of which possibility is predicated. If God did not exist, there would
be no eidetic natures and thus no possibles to be considered. Richard Cross
follows this interpretation21. According to the alternative interpretation,
God is responsible for the objective being of esse intelligibile and esse
possibile but not what they are22.
Even though Scotus does not formulate his views as clearly as one
might hope, he clearly did not make logical possibilities dependent on
divine intellect in any way. Let us consider the example of the possibility
of the world without God which Scotus discusses in many places,
assuming that nothing exists and that there is then an intellect formulating
a proposition which states that a world is possible. This proposition is said

T. HOFFMANN, Creatura intellecta: Die Ideen und Possibilien bei Duns Scotus
mit Ausblick auf Franz von Mayronis, Poncius und Mastrius, Aschendorff, 2002
(Beiträge zur Geschichte der Philosophie und Theologie des Mittelalters, N.F., 60),
«Duns Scotus on the Origin of the Possibles in the Divine Intellect», in S.F. BROWN
et al. (eds.), Philosophical Debates at Paris in the Early Fourteenth Century,
Brill, Leiden 2009, pp. 359-379; F. MONDADORI, «The Independence of the Possible
According to Duns Scotus», in O. BOULNOIS et al. (eds.), Duns Scot à Paris: Actes du
colloque de Paris, 2-4 Septembre 2002, Brepols, Turnhout 2004, pp. 313-374 (Textes
et études du Moyen Age 26); R. CROSS, «Recent Work on the Philosophy of Duns
Scotus», Philosophy Compass, 5/8 (2010) 667-675.
HONNEFELDER, Scientia transcendens, p. 54, stresses that God’s intellect
knows all intelligible things by necessity; see also S. KNUUTTILA, «Duns Scotus and
the Foundations of Logical Modalities», in L. HONNEFELDER at al. (eds.), John Duns
Scotus: Metaphysics and Ethics, Brill, Leiden 1996, p. 137 (Studien und Texte zur
Geistesgeschichte des Mittelalters 53).

to be true; similarly, what is affirmed by the proposition ‘The world will

exist’ is said to be possible. No God is needed here23. In one variant of these
arguments, Scotus adds that if God will be later, this is irrelevant for the
logical possibility:

If, before the creation of the world, there was not only no world but
also, per incompossibile, no God (although he began by itself exist
and was then capable of creating the world), and if there had been
an intellect before the world forming this ‘The world will exist’, it
would have been possible because the terms are not incompatible,
not on the basis of a principle in a possible being or a corresponding
active principle, and in this case ‘The world will exist’ would not
have been possible because of God’s potency but, formally speaking,
because of the potency which was the non-incompatibility of these
terms, although this non-incompatibility was also accompanied by
an active potentiality with respect to this possibility. (Ordinatio
I.7.1 n. 27, Ed. Vat. IV, 118-119)

Explicating the view that what a proposition expresses is possible if

the terms are not incompatible, Scotus compares this formal possibility
with potency based possibilities. He argues that propositional formal
possibilities can be said to be possible by the potency derived from the
compatibility of terms, without attention to potencies as powers. Elsewhere
he calls this formal potency a logical potency.
In explaining the formal truth of a proposition, Scotus, as mentioned
above, says that when a proposition is formed by an intellect, there is a
formal truth or falsity in the act, depending on whether it conforms with
things or not, independently of whether the proposition is asserted or not.
Formal truth or falsity applies to any apprehensive act of combining or
dividing significative terms24. Objective truth or falsity belongs to these
complexes as asserted or denied. Formal truth as a relation of conformity
between a cognitive representation and reality pertains to a small part
of the representations expressed by compatible complexes which are
formally possible, the majority of these being expressed by formally false

Lectura I, d. 7, n. 32 (Ed. Vat. XVI, 484); Lectura I, d. 39, q. 1-5, n. 49 (Ed.
Vat. XVII, 494); Quaestiones super libros Metaphysicorum Aristotelis IX, q. 1-2, n.
18 (514).
See also PINI, «Scotus on Assertion», pp. 324-326.

propositions. Similarly, a conceptually necessary proposition is said to be

per se known independently of whether the subject understands that it is
per se known25. The same holds for modal propositions in general; the
formal modality qualifies a proposition on the basis of the logical potency
or impotency, independently of whether anyone attends to the compatibility
or incompatibility of the meanings of terms.
An omniscient intellect immutably knows everything that can be
signified by the combinations of compatible terms and provides them with
an ontological status as the objects of understanding. Having this image in
mind Scotus often speaks about possibilities as divine thoughts. Because
of God’s eternal act of thinking, the meanings of all consistent propositions
occur in intelligible being necessarily, because the divine intellect is not
free, and they form an actually infinite realm of possible representations
of how things could be. This part of Scotus’s modal theology had many
followers, such as Mastrius. It is also the view of Leibniz who distances
himself from other interpretations:

The late Jacob Thomasius […] made the apt observation […] that
one ought not to say, with some Scotists, that the eternal truths
would subsist, even if there were no intellect, not even that of God.
For it is, in my opinion, the divine intellect which makes the eternal
truths real, although his will has no part in it. Whatever is real must
be founded in something existent. It is true that an atheist may be
a geometer, but if there were no God, there would be no object of
geometry, and without God there would neither be anything existent
nor anything possible26.

Treating possibilities as divine thoughts played an important role in

Scotus’s theological creationism. While these reified possibilities had an
objective being as a consequence of the intensively infinite attributes of
the metaphysically simple and perfect divine entity, they were formally
possible by themselves and not by divine activity. All interpreters seem to
agree with this picture. Divided opinions concern the question of whether
Scotus assumed that the objects of God’s thought are formally possible
by themselves or vice versa. The abstract question of the first origin of

Ordinatio I.2.1.1-2, n. 22 (Ed. Vat. II, 136).
G.W. Leibniz, Essais de Théodicée 184, in Die philosophischen Schriften von
G.W. Leibniz, ed. C.L. GERHARDT, Berlin 1875-1890, vol. VI, p. 226.

modalities is not directly relevant for understanding Scotus’s modal

semantics, to be sure, but it is of some interest that Scotus would have
not accepted the view of Leibniz just quoted. Formal possibilities and
impossibilities receive some kind of objective existence when there is any
intellect whatsoever which forms a proposition, but the possibilities or
impossibilities which are expressed by the propositions are not constituted
by them. If these are the meanings of the propositions, should they be
actual in order to be called possible?
The terms of esse intelligibile and esse possibile are used in Ordinatio
I.43 as part of Scotus’s answer to Henry of Ghent’s theological arguments
about whether the reason for being impossible is derived from God or other
things. Scotus explains that while possibilities can be discussed as having esse
possibile with respect to divine power, this status is derived from their being
first understood and having a being in esse intelligibile in divine thought:

An object is not primarily possible because of the potency by

reason of which it [God] is understood to be omnipotent, but by the
divine intellect which first produces it in intelligible being, and the
intellect is not formally the active power by which God is said to be
omnipotent, and when a thing is produced in such a being, namely
intelligible, by the divine intellect in the first instant of nature, it has
by itself a possible being in the second instant of nature because it
is not formally repugnant for it to be and it is formally repugnant
for it to have a necessary being by itself. (Ordinatio I.43, n. 14, Ed.
Vat. VI, 358)

These possibilities are formaliter possibilities by themselves and

principiative by the intellect because their objective being is caused by the
divine act of production27.
Scotus assumes that the basic level of divine modal knowledge can be
described as an actually infinite and complete set of understood propositions
which express the possible ways things could be. When the divine
combinatory art fills the area which is fixed by these conditions, everything
that is formally possible receives a minor intentional existence as an object
of God’s mind. If such an infinite knowledge did not exist, even then one
could say that if any of these propositions is formulated, it would express a
possibility, and if it is a possibility proposition, it would be true.

Ordinatio I.43, n. 7 (Ed. Vat. VI, 354).

4. Formal Modality

When Scotus discussed the omniscient divine or human mind, he found

it convenient to deal with possibilities in a formal way instead of describing
what they are. This was also the part of his theory which proved to be
influential. In this approach, possibilities are primarily defined as logical
possibilities which are expressed by propositions the terms of which are
not mutually exclusive and to which being is not repugnant in this sense.
If the terms are incompatible, the propositions express impossibility, and if
their denials are incompatible, they express necessity. Because this structure
pertains similarly to divine and human intellect, possibilities are treated as if
in the eternal instant. Questions of time and power are discussed separately
and do not belong to the essence of the formal theory, except that singular
contingency propositions can be treated as temporally definite. The semantics
of contingency propositions is spelled out with respect to the actual order in
which they are true, but could be false, or are false, but could be true.
A special feature of Scotus’s theory is that he often operates with the
idea that there is an actually infinite number of ways in which things could
be. An infinite mind is needed for knowing them all simultaneously, but this
mind does not decide what belongs there and what does not. The reason
for the assumption of actual quantitative infinity in Scotus is the idea of
the completeness of the realm of intelligibility. It includes all the ways
things could be. Their being knowable through possibility propositions is
conceptually prior to their being known, and they are exhaustively known
through an actually infinite number of different possibility propositions.
I don’t comment on the philosophical problems associated with the
assumption of actual infinity of intelligibility. It is a historically interesting
part of an attempt to explain how the realm of logical possibilities is fixed
independently of divine omniscience. In Scotus’s metaphysical theism, the
eidetic content of significative units and their combinations are primarily
understood by divine intellect, but this takes place necessarily with respect
to the complete realm of what might be. This has existence only when it is
understood, but it is not constructed by understanding it28.

That the possibilities are what they are is called a brute metaphysical fact in
P. KING, «Duns Scotus on Possibilities, Powers, and the Possible», in T. BUCHHEIM
et al. (eds.), Potentialität und Possibilität: Modalaussagen in der Geschichte der
Metaphysik, Frommann-Holzboog, Stuttgart 2002, pp. 275-299.

Formal or logical modality is discussed here as a non-temporal matter.

The modalities relevant in discussing nature and action need a reference
to time and power, and this is the next step in Scotus’s theory, in which
the formal theory is regarded as basic. This was an influential model
in the fourteenth century. The theory of modal syllogistic and modal
consequences was developed at the level of logical modalities, and the
theories of power, time, and natural modality were treated separately,
using the formal theory as a normative background. There were some
areas in which logical modalities and natural or otherwise conditionalized
modalities were treated in interplay. In theology, the distinction between
God’s absolute and ordained power was understood as partially analogous
to that between logical and natural modalities, and this was even more
influential in natural philosophy and its interest in investigating the
principles of natural philosophy secundum imaginationem, by assuming
contrafactual and logically possible cases in order to see what followed29.

See KNUUTTILA, «Modality»; for God’s absolute and ordained power, see
H. GELBER, It Could Have Been Otherwise: Contingency and Necessity in Dominican
Theology at Oxford, 1300-1350, Brill, Leiden 2004, pp. 309-349 (Studien und Texte
zur Geistesgeschichte des Mittelalters, 81).



In his Logica, Albert of Saxony defines merely confused suppo-

sition thus, that a descent under the general term to the singular terms
can be done by way of a sentence with a disjunct extreme, but not
by way of a disjunctive sentence or a conjunctive sentence, that is in
scholastic terms, a descent can be done disiunctim, but not disiunctive
nor copulative:

Suppositio personalis confusa tantum est acceptio termini pro

quolibet, quod significat ex impositione vel naturaliter proprie, sub
quo virtute istius acceptionis licet fieri descensum ad sua singularia
per propositionem de disiuncto extremo et non per propositionem
disiunctivam nec copulativam. Sic supponit praedicatum in ista
propositione “Omnis homo est animal”, sub ipso enim licet fieri
descensum ad eius supposita per propositionem de disiuncto
extremo et non per propositionem disiunctivam nec copulativam.
Unde bene sequitur “Omnis homo est animal, ergo omnis homo est
illud vel illud animal”, ita quod hoc totum disiunctum “illud vel
illud animal” est verificabile de isto termino “homo” sumpto cum
distributione significative accepto1.

He then adds a longer passage in which he criticizes the view of

«some» who add to this definition «vel copulatim», that is, that the relev-
ant descent can be done by way of a sentence with a disjunct extreme

Universität Graz, Institut für Philosophie, Heinrichstraße 33, A-8010 Graz.
Email: harald.berger@uni-graz.at.
Albert von Sachsen, Logik, Lateinisch – Deutsch, übersetzt, mit einer Einleitung
und Anmerkungen hg. v. H. BERGER, Meiner, Hamburg 2010, ch. II.3, pp. 262-264
(Philosophische Bibliothek, 611). Cf. also rule 11 in ch. II.6, ibid., pp. 310-312. On this
regula antiqua ‘Quidquid mobilitat immobilitatum, hoc immobilitat mobilitatum’ see,
e.g., F. SCHUPP, Logical Problems of the Medieval Theory of Consequences. With the
Edition of the Liber consequentiarum, Bibliopolis, Napoli 1988, pp. 161-162 (History
of Logic, 6).

or with a conjunct extreme. This addition, however, is superfluous accor-

ding to Albert2.
This view criticized here by Albert is clearly that of Thomas Maulfelt
who defines merely confused supposition thus, that a descent can be done
by way of a sentence with a disjunct (and conditionalized) extreme or
with a conjunct extreme. (The clause «condicionatim = de condicionato
extremo» need not concern us here.) Thomas for his part states that the
definition which only comprises a descent by way of a sentence with a
disjunct extreme is insufficient, indeed false3.
The case for Thomas’s addition of the clause «vel copulatim» to the
definition of merely confused supposition is the analysis of the sentence
«Sortes differt ab omni homine» which does not allow a descent under
«homine» disiunctim, but only copulatim, since Socrates is this man or that
man or that man etc., but he is not this man and that man and that man etc.
This is faithfully reported by Albert4.
So we have Albert’s view that a definition such as that of Thomas
comprises more than needed, and Thomas’s view, that a definition
such as that of Albert is insufficient. As a matter of fact, the definition
of suppositio confusa tantum given by Albert in his Logica seems to
be the common one. It is to be found in this or that way in, e.g.,
William of Ockham 5, John Buridan 6, Marsilius of Inghen 7, John of

Albert von Sachsen, Logik, op. cit., pp. 264, 266 and 268. The rejection is on
pp. 266 and 268, «Breviter dico, quod non oportet addere […] propter hoc tamen non
sequitur, quod supponat confuse tantum».
Thomas Maulfelt, Suppositiones, Ms Erfurt, Universitätsbibliothek, Dep. Erf.,
Cod. Amplon. Q. 271, f. 45r-v. Cf., e.g., S. READ, «Thomas of Cleves and Collective
Supposition», Vivarium, 29 (1991) 50-84, at p. 71 and n. 66.
Thomas Maulfelt, Suppositiones, op. cit., f. 45v. Cf. Albert von Sachsen, Logik,
op. cit., pp. 264 and 266, «Nec valet, quod aliqui addunt ‘vel copulatim’ […] tunc esset
omnis homo, quod est falsum».
Guillelmi de Ockham, Opera Philosophica I: Summa Logicae, Ed. by
P. BOEHNER – G. GÁL – S. BROWN, The Franciscan Institute, St. Bonaventure NY 1974,
Ps. I, cap. 70, pp. 209-212, at p. 211.
Johannes Buridanus, Summulae de suppositionibus, Ed. by R. VAN DER LECQ,
Ingenium, Nijmegen 1998, pp. 50-51 (Artistarium, 10-4), at p. 50.
Marsilius of Inghen, Treatises on the Properties of Terms, Ed. by E. P. BOS,
Reidel, Dordrecht – Boston – Lancaster 1983, pp. 56 and 58 (Synthese Historical
Library, 22).

Holland 8. For chronological reasons Maulfelt’s target seems to be

Surprisingly enough, however, Albert in his Sophismata holds the
very view he criticizes in the Logica. In the introduction to the section on
terms like «differt» in the second part of the Sophismata, Albert says that
defining merely confused supposition in terms of a descent by way of a
sentence with a disjunct extreme is insufficient. Rather one needs to add
«or with a conjunct extreme»9. So here he adheres to the view of Maulfelt10.
The crucial passage runs:

Dico, quod communiter auctores non sufficienter definiunt

suppositionem confusam tantum definientes eam sic, quod
suppositio tantum est, virtute cuius sub termino sic supponente
valet descendere ad eius supposita per propositionem de disiuncto
extremo, sed oportet addere “vel de copulato extremo”, quo addito
suppositio confusa tantum est sufficienter definita. Modo, licet in illa
propositione “Omnis homo differt ab omni homine” non possit fieri
descensus sub isto termino “homine” ad eius supposita disiunctim,
tamen valde bene copulatim11.

If the Logica Alberti and the Sophismata Alberti would have been
handed down to us anonymously, we certainly would assume that they
were composed by two different authors. But there is not the slightest
doubt regarding the authenticity of these two works, indeed they are the
most important and most famous works of Albert’s in the field of logic. So
given the identity of the author, it follows that Albert has changed his mind
on these matters.

John of Holland, Four Tracts on Logic (Suppositiones, Fallacie, Obligationes,
Insolubilia), Ed. by E. P. BOS, Ingenium, Nijmegen 1985, pp. 16-18 (Artistarium, 5).
Albertus de Saxonia, Sophismata, Olms, Hildesheim-New York 1975 (repr. of
the print Paris 1502), ff. i2vb-i3rb. Mischa von Perger, who is preparing a critical
edition of Albert’s Sophismata, tells me that all (apart from the few he has not yet seen)
manuscripts and early prints have substantially the same text.
I am indebted to Michael J. Fitzgerald, who has first drawn my attention to
this Sophismata-passage some time ago. He has now addressed the problem for his
own, see M. J. FITZGERALD, «Unconfusing Merely Confused Supposition in Albert of
Saxony», Vivarium, 50 (2012) 161-189.
Draft of a critical edition by M. vON PERGER, cf. repr. of print Paris 1502,
f. i3ra-b.

I must confess, however, that I have not yet any convincing idea
regarding why, when and in what direction this change of mind of Albert’s
took place.
In the introduction to my edition of the Logica Alberti, I have argued
that there are two redactions of the Logica, the first dating from the first
half of the 1350s, the second dating from 1360, and that the Sophismata are
composed after the first and before the second redaction of the Logica12.
By the way, since then I even got inclined to assume a third redaction
of the Logica in between, so that we had a first redaction without any
treatise on dialectical consequences, that is on the Topics, a second one
with such a treatise at the very end of the whole work, that is in Treatise
VI after the Insolubles and Obligations, and finally a third redaction with
that treatise on the Topics incorporated into Treatise IV on Consequences.
The assumption of such a third redaction is also indirectly corroborated
by an extant anonymous Logica clearly dependent on Albert’s which is
incomplete and scattered in two manuscripts preserved in Prague13.
Regarding the Sophismata, Mischa von Perger and I have seen now
almost all known manuscripts and prints, and in all of these witnesses, the
pertinent text is quite the same, as was already noted.
The tradition of the Logica, however, has some surprising peculiarities
in this respect: The Erfurt manuscript, Codex Amplonianus Quarto 242,
which represents the first redaction, has «per propositionem de disiuncto
extremo vel de copulato extremo», that is, Maulfelt’s view, and has no trace
at all of the passage «Nec valet, quod aliqui addunt “vel copulatim”»14:

Suppositio personalis confusa tantum est acceptio termini pro

quolibet, quod significat ex impositione vel naturaliter proprie,
[pro] <sub> quo virtute illius suppositionis [licet] potest fieri
descensus ad sua singularia per propositionem de disiuncto extremo
vel de [copulativo] <copulato> extremo et non per propositionem
disiunctivam nec copulativam. Unde hac suppositione supponit hic

Albert von Sachsen, Logik, op. cit., pp. LXXVII and LXXIX.
Praha, Archiv Pražského hradu, fond Knihovna Metropolitní kapituly u sv.
Víta, M. 33, ff. 1ra-36vb, and O. 55, ff. 61r-65r and 51r-56v (in this order!). On f. 61r it
reads «Liber sextus et ultimus, qui continet tres tractatus. Primus erit de insolubilibus,
secundus de obligationibus, tertius de locis dyalecticis». Cf. H. BERGER, «Erträge einer
Bibliotheksreise nach Prag», Codices Manuscripti & Impressi, 89-90 (2013) 13-23.
Erfurt, UB, Dep. Erf., Cod. Amplon. Q. 242, f. 18r.

terminus “animal” in illa propositione “Omnis homo est animal”.

Nam bene sequitur “Omnis homo est animal, ergo omnis homo est
illud animal vel illud vel illud”. Similiter hac suppositione supponit
ly [homo] <“homine”> in illa propositione “Omnis homo differt ab
omni homine”, quia bene sequitur “Igitur omnis homo differt ab isto
homine [vel] <et> ab illo”, nec valet consequentia “Omnis homo
differt ab isto homine et ab isto, ergo omnis homo differt ab illo
homine”, quia arguitur ab [uno] <una> de [copulativo] <copulato>
extremo, ac si esset copulativa etc. – Suppositio confusa distributiva
est …

So here the Logica does agree with the Sophismata regarding merely
confused supposition. The Leipzig manuscript (= L), containing the third
redaction, and dated from 1381, has the definition «per propositionem de
disiuncto vel copulato extremo» together with the «Nec valet»-passage,
what yields of course an incoherent text15. Quite the same is true of the
print of 1522 (= X), but the editor Pietro Aurelio Sanudo apparently has
tried to save the passage, e.g. by omitting «vel copulato extremo» in the
definition and by adding «secundum quosdam», so that the incoherence
with the «Nec valet»-passage is not that manifest16. By the way, in the
margin, Sanudo refers to Peter of Mantua as a supporter of the «vel
copulatim»-view, Maulfelt apparently being unknown to him.

L25rb / X11va-b: Suppositio personalis confusa tantum est acceptio

termini pro [cuilibet L] <quolibet X>, quod significat ex impositione
vel naturaliter proprie, sub quo virtute illius suppositionis (licet L
potest X) fieri descensus ad sua singularia per propositionem de
disiuncto (vel copulato L om. X) extremo et non per propositionem
disiunctivam (vel L nec X) copulativam. Sic (enim X om. L) supponit
praedicatum in illa propositione “Omnis homo est animal”, sub ipso
enim (licet L debet X) fieri descensus ad (eius L sua X) supposita
per propositionem de disiuncto extremo et non per propositionem
disiunctivam nec copulativam. Unde (hac suppositione supponit
hic terminus “animal” in ista propositione “Omnis homo est

Leipzig, UB, Ms 1367, f. 25rb. Cf., e.g., also Ms Paris, BnF, lat. 6670 (dated
1417), f. 42r-v, for the same incoherence.
Albertus de Saxonia, Perutilis Logica, Olms, Hildesheim – New York 1974
(repr. of the print Venice 1522), f. 11va-b. The editor’s marginal note to the «Nec valet»-
passage in the main text runs «Huius opinionis fuit Pe. Ma. in tracta. suppositionum».
Cf. Petrus Mantuanus, Logica, e.g. Pavia 1483, f. a3ra.

animal”, nam X om. L) bene sequitur “Omnis homo est animal,

ergo (omnis homo est L om. X) illud vel illud animal”, ita quod
hoc totum disiunctum “illud vel illud (animal L om. X)” (est L
sit X) verificabile de illo termino “homo” significative accepto.
Similiter (secundum quosdam X om. L) hac suppositione supponit
ly “homine” in illa (propositione X om. L) “Omnis homo differt ab
omni homine”, quia bene sequitur “(Omnis homo differt ab omni
homine L om. X), igitur omnis homo differt ab illo [ab illo add. X]
et ab illo homine”, neque valet consequentia “Omnis homo differt
ab illo et ab illo homine, igitur (omnis homo X om. L) differt ab illo
homine”, quia arguitur ab una (L uno X) de copulato extremo, ac si
esset (X sit L) (una X om. L) copulativa. – Nec valet, quod aliqui
addunt “vel copulatim (L copulativi X)”. Unde dicunt …

The presumably oldest manuscripts of the Logica, however, viz. Paris,

BnF, lat. 14715, Pommersfelden, Schlossbibliothek 236, and Prague,
NK, IV. G. 417, unanimously have the definition with only «de disiuncto
extremo» and the «Nec valet»-passage. By the way, this very passage is
also edited from six manuscripts and the early print as an appendix to
Stephen Read’s paper of 1991 on Albert of Saxony vs. Thomas Maulfelt18.
So it seems that the Sophismata text is stable, whereas the Logica text
is not19. Perhaps this means that the Sophismata antedate the Logica, and
also, that in the course of the development of the Logica, the Maulfelt-
view has been abandoned by Albert. This does not fit well, however, with
the fact that in the Sophismata there are clear references to the Logica, but
that there are none in the opposite direction, apart from a single implicit

See Albert von Sachsen, Logik, op. cit., pp. LXX-LXXIV, nos. 21, 24, 27. Cf.,
e.g., also Paris, BnF, lat. 18430, f. 18va-b.
S. READ, «Descensus copulatim: Albert of Saxony vs. Thomas Maulfelt», in
J. BIARD (ed.), Itinéraires d’Albert de Saxe. Paris-Vienne au XIVe siècle, Vrin, Paris
1991, pp. 71-85 (Études de philosophie médiévale, 69), Appendix at pp. 83-85.
In his Quaestiones circa Logicam, qu. 17, ad rationem 5am, Albert only says that
in ‘Sortes differt ab omni homine’ the term ‘homine’ stands non distributive. If this
meant confuse tantum, there would be a further conflict with his determinate view. But
presumably non distributive means immobiliter (in the sense of not allowing a descent
to a conjunction of sentences) and covers both determinate and merely confused
supposition. See M. J. FITZGERALD (ed.), Albert of Saxony’s Twenty-Five Disputed
Questions on Logic, Brill, Leiden – Boston – Köln 2002, pp. 238-239, § 311, and
p. 244, § 321 (Studien und Texte zur Geistesgeschichte des Mittelalters, 79), with a
text not completely correct, however.

reference in only some of the witnesses (viz. at the end of ch. III.10 on
sentences with «incipit» and «desinit»)20.
Now, it is true, of course, that there is a logically relevant difference
between the passages at stake in the Logica and in the Sophismata, since in
the former the example used is the singular sentence «Sortes differt ab omni
homine», whereas in the latter it is the universal sentence «Omnis homo
differt ab omni homine». By the way, this very difference is sometimes
explicitly stated in the relevant tradition, e.g. in Marsilius of Inghen or
in the first Berlin-text (edited in the appendix below) under the name of
a certain Hoklem21. But with Albert, it is the proper definition of merely
confused supposition that is primarily called into question.
Regarding the singular sentence «Socrates differs from every man»,
there are the following views: For Maulfelt, the term «man» has merely
confused supposition in the sense of a descent to a conjunct extreme
(copulatim), because the following inference holds: «Socrates differs from
every man, therefore Socrates differs from this and this and this man» and
so on for every single man. The consequent is true as well as the antecedent,
because Socrates is not this man and this man and this man and so on, but
rather this man or that man or that man and so on, as was already stated
above (see also the second Berlin-text edited in the appendix below)22.
For Albert, as a Maulfelt-critic, however, this is not a case of merely
confused supposition, but rather of determinate supposition, what implies
a descent to a disjunctive sentence (disiunctive as opposed to disiunctim).
So according to Albert, it follows that Socrates differs from this man, or
Socrates differs from this man, and so on. Since all but one part of this
disjunction are true, the disjunction itself is true, the false part being the one
where «this man» refers to Socrates himself (for the logical details I refer to
the fine studies by Stephen Read, Christoph Kann, and Hartmut Brands)23.

Cf. on this Albert von Sachsen, Logik, op. cit., p. LXXVII. In the print Paris
1502 (repr. 1975, op. cit.), f. h1va, instead of ‘maximam’ it should be read ‘materiam’,
as Mischa von Perger tells me.
Marsilius of Inghen, Treatises on the Properties of Terms, op. cit., p. 70, lines
1-11. Cf. also Johannes Dorp in Johannes Buridanus, Compendium totius Logicae,
Venedig 1499, repr. Minerva, Frankfurt am Main 1965, f. i5rb-va. – Ms Berlin, SBPK,
lat. fol. 206, f. 295ra.
Ibid., f. 319rb-va.
READ, «Descensus copulatim»; ID., «Thomas of Cleves»; C. KANN, Die
Eigenschaften der Termini. Eine Untersuchung zur Perutilis logica Alberts von

Some light on this 14th century discussion, even though unfortunately

not on the problem found in Albert of Saxony24, is shed by an interesting
miscellaneous manuscript on logic dating from the first half of the fifteenth
century, viz. Berlin, Staatsbibliothek Preußischer Kulturbesitz, Ms. lat.
fol. 206, a paper manuscript of some 330 folios, partly dated from 1433
(f. 152ra), with references to Rostock (f. 165ra) and Lübeck (f. 290rb).
Our problematic sentence «Sortes differt ab omni homine» is treated as
a sophisma twice in it (see below). About the first half of this manuscript
is filled by John Dorp’s Commentary on Buridan’s Summulae25. Another
part, e.g., deals with some of Albert of Saxony’s Sophismata26. And several
parts are commentaries on Marsilius of Inghen’s parva logicalia27.
The two parts relevant for my purpose here are an anonymous
commentary on Marsilius of Inghen’s Suppositiones, ff. 287ra-300vb, and an
anonymous commentary on Marsilius’s Consequentiae, ff. 311ra-320vb. The
sophisma «Sortes differt ab omni homine» does occur in both of those works,
the first occurrence is on ff. 294va-295va, the second one on ff. 318vb-319va.

Sachsen, Brill, Leiden – New York – Köln 1994, pp. 96-105 (Studien und Texte zur
Geistesgeschichte des Mittelalters, 37); H. BRANDS, «Referenztheorie und freie Logik im
Spätmittelalter», Philosophisches Jahrbuch, 102 (1995) 33-60; H. BRANDS, «Descensus
copulatim und universelle Spezialisierung bei Thomas Manlevelt», in M. KINTZINGER
– S. LORENZ – M. WALTER (eds.), Schule und Schüler im Mittelalter. Beiträge zur
europäischen Bildungsgeschichte des 9. bis 15. Jahrhunderts, Böhlau, Köln – Weimar
– Wien 1996, pp. 165-185 (Beihefte zum Archiv für Kulturgeschichte, 42).
Text no. 5, f. 188ra-218va, is a commentary on Albert’s Sophismata, but breaks
off in the first part. The commentary begins, «Secundum venerabilem Boetium De
phylosophico consolatu malum enim non evitatur nisi cognitum […] Videnda sunt pro
praesenti aliqua sophismata Alberti stilo facili composita […] Igitur divino adiutorio
Albertus in suis Sophismatibus dicit, quod velit ponere sophismata varia habentia
difficultatem ratione diversorum synkategorematum».
Cf. E. P. BOS, «Die Rezeption der Suppositiones des Marsilius von Inghen bei
Johannes Dorp (Paris) und in einem anonymen Prager Sophistria-Traktat (um 1400)»,
in M. J. F. M. HOENEN – P. J. J. M. BAKKER (eds.), Philosophie und Theologie des
ausgehenden Mittelalters. Marsilius von Inghen und das Denken seiner Zeit, Brill,
Leiden – Boston – Köln 2000, pp. 213-238. See p. 217, n. 17, on mss Berlin and others,
pp. 217-223 on Marsilius rather than Buridan as commented on by Dorp in the 4th part
on suppositions.
Cf. above n. 24.
Cf. Marsilius of Inghen, Treatises on the Properties of Terms, op. cit., p. 27,
no. I (five commentaries on Marsilius’s Suppositiones, Ampliationes, Appellationes,
Restrictiones, Consequentiae on f. 287ra-320vb).

The very nature of a sophisma consists in there being good reasons for
considering the sentence at stake as being true, as well as for considering
it as being false. In our case, the first intuition, presumably, will be that
the sentence is false, because there is one man from whom Socrates does
not differ, namely Socrates himself. On the other hand, Socrates is not
identical with every man, so it is true that he differs from every man.
Exactly these points are nicely stated at the beginning of the first
sophisma (f. 294va-b). Disproof: Socrates differs from every man,
Socrates is a man, therefore Socrates differs from himself. The consequent
is impossible, the minor sentence is true, so the major sentence will be
false; but that is precisely the sophisma.
The proof is given by way of the interpreting sentences (exponentes):
Socrates exists, and every man exists, and Socrates is not every man;
therefore Socrates differs from every man.
The response, at the end, states that the sophisma is true, as is shown
by the interpreting sentences just quoted. The disproof is rejected, because
the term «man» in the sophisma is not distributed. And this is indeed the
crucial point here, viz. «every» as a part of the predicate «every man»
versus «every» as a quantifier of the predicate «man».
In between it is stated that the ablative following the verb «differt»
has confuse distributive supposition, as long as no other distributive sign
does prevent this, because in that case, the supposition would change to non
distributive. And here occurs a reference to a certain Hoklem who is reported
to say that when a further distributive sign is added, the supposition changes
from confuse distributive to determinate, if the sentence is a singular or an
indefinite one, or to merely confused, if the sentence is a universal one.
So in the sophisma «Sortes differt ab omni homine» the term «homine»
has determinate supposition according to this Hoklem, since it is within
the scope of two distributive signs in a singular sentence, viz. the negation
(semantically) included in «differt» and the universal quantifier «omni».
I take «Hoklem»28 to be a misspelling of «Ockham», and the reference
could be to his Summa Logicae, part I, chapter 7429. By far better, however,
the passage at the end of column 295ra is matched by Marsilius of Inghen’s
16th rule of suppositions30.

Further occurrences of this name are, e.g., on ff. 289va, 296ra, 296rb.
Ockham, Summa Logicae, op. cit., pp. 228-230, the 4th rule at pp. 229-230.
Marsilius of Inghen, Treatises on the Properties of Terms, op. cit., p. 70.

In the second sophisma there is one argument (at the end of col. 319rb)
saying that the resulting sentences from the descent, called descendentes
here, are false and that, as a consequence, the sophisma itself is false. In
support for this, it is claimed that the term «man» in the sophisma has merely
confused supposition, because it is within the scope of two distributive
signs, and that, as a consequence, the descent must be disiunctim. But since
Socrates is this or this or this man, he does not differ from this or this or this
man, and so the sophisma is false.
In responding to this argument the Anonymous refers to two views
(opiniones), the first ascribed to Magister Thomas Maulvelt31, the second
to Biridanus, Mercilius et alii moderni loyci.
The first opinion, attributed to Maulfelt, is not well stated, since
the crucial copulatim-point is missed. It is said that according to
Maulfelt the term «man» has merely confused supposition, but that the
descent must not be disiunctim, as the previous argument suggested,
but otherwise. The following descent, however, is not a copulatim
one, but such that the disjunct term is outside the scope of «differs»,
viz. preceding it rather than following it, as in the argument which is
rejected here.
The second view, attributed to Buridan, Marsilius, and further
modern logicians, is presented thus that the term «man» in the sophisma
does not stand in merely confused supposition, but rather in determinate
supposition. Accordingly they say that not always merely confused
supposition results, when two distributive signs affect the same term, but
only when this term is distributed, what, however, is not the case with
our sophisma.
So this source clearly distinguishes the Maulfelt-view on the one
hand, and the view of Buridan and Marsilius on the other. To the former,
we may add the Albert of the Sophismata Alberti and Peter of Mantua,
to the latter we may add Ockham, the Albert of the Logica Alberti,
and John of Holland32 who is closely following the Logica Alberti, as
is shown by his editor Egbert Bos33. It must be noted, however, that
Buridan explicitly says that the term «homine» in the sentence «A

The spelling here seems to be ‘Maulvelvelt’, perhaps (but rather unlikely)
‘Mauwelvelt’. Cf. also f. 290va.
See John of Holland, Four Tracts on Logic, op. cit., pp. 16-18, 21-22, 27-28.
Ibid., pp. 161-166.

differt ab omni homine» has merely confused supposition34, so that his

mention among the supporters of the determinate view seems not to be
Albert of Saxony is not even mentioned here by name but, as we have
shown, he exemplifies both ways of dealing with such sophismata. It would
be natural to assume that Albert of Saxony’s change of mind in this respect
is due to the intellectual milieu at Paris in the 1350s. But again, the details
are unfortunately completely unknown. And scarcely known are also the
details of times and places of Maulfelt’s teaching and writing as well as of
the reception of his ideas, although he seems to be the crucial figure in this
particular debate.

Appendix: Two texts from Ms Berlin, Staatsbibliothek Preußischer

Kulturbesitz, lat. fol. 206, ff. 294va-295va and 318vb-319va

On this manuscript see above and also the catalogue description by

Valentin Rose35.
The two pieces edited below are from the text no. 8, f. 287ra-300vb, a
commentary on Marsilius of Inghen’s Suppositiones, and from the text no.
12 (nos. 12 and 13 in Rose), ff. 311ra-320vb, a commentary on Marsilius’s

/294va/ … Sortes differt ab omni homine.

Sortes differt ab omni homine. Improbatur sic: Sequitur bene /294vb/
“Sortes differt ab omni homine, Sortes est homo, ergo Sortes differt a se
Sorte”. Consequens est impossibile, consequentia virtualiter tenet in Darii,
minor est vera, ergo maior erit falsa, quae est sophisma.
Probatur sic per eius exponentes: “Sortes est et omnis homo est et
Sortes non est omnis homo, igitur Sortes differt ab omni homine”.

Johannes Buridanus, Summulae de suppositionibus, op. cit., p. 63, lines 19-20.
V. ROSE, Verzeichniss der lateinischen Handschriften der Königlichen
Bibliothek zu Berlin, 2. Bd., 3. Abt., Asher, Berlin 1905, pp. 1227-1228, no. 973. Cf.
also B. MICHAEL, Johannes Buridan: Studien zu seinem Leben, seinen Werken und
zur Rezeption seiner Theorien im Europa des späten Mittelalters, Dissertation, Berlin
1985, vol. 2, pp. 542-