Vous êtes sur la page 1sur 41

HESPERIA

76

IDEA AND

(2OO7)

VISUALITY

Pages SSSS9S

IN HELLENISTIC
ARCHITECTURE
A Geometric
Temple
at

of

Analysis
of

the

Asklepieion

Kos

ABSTRACT
author uses analytic geometry
and AutoCAD
software to analyze
at Kos, revealing a circumscribed
the plan of Temple A of the Asklepieion
as the basis for the
and
Pythagorean triangle
plans design.This methodology
its results counter earlier doubts about the application of geometry to Doric

The

temple design and suggest the existence of an alternative to the grid-based


approach characteristic of Hellenistic
temples of the Ionic order. Appre
ciation of the geometric system underlying the plan of Temple A leads to a
consideration

of the role ofvisuality inHellenistic


architecture, characterized
inwhich abstract ideas shared by architects and scholars

here as the manner


conditioned

viewing

and influenced

the design process.

on the island of Kos was a


The Asklepieion
healing sanctuary and medi
some 4 km
cal school of great importance throughout
antiquity.1 It lies
southwest of the ancient polis of Kos, built on a terraced slope
commanding
sea.
views
of
the
In
its
the
state,
impressive
completed
complex consisted of
three separate terraces connected by stairways, each supporting structures
from various periods (Figs. 1,2).2
By the middle of the 3rd century b.c., the sanctuary's three terraces
were constructed.3 On the lower terrace, a
Doric stoa with ad
?-shaped
was built to enclose an
rooms
x 93 m
47
joining
approximately
space.4Major
architectural features on the middle terrace included an altar,
a
replaced by
more monumental
version in the following century, and temples dedicated
to thank Andrew

1.1 wish
for his
taking
which

constructive
an interest

criticisms
in my

are all the

versations.

stronger
I am indebted

Stewart
and for

arguments,
for our con
to Fikret

Yeg?l andDiane Favro for their de


voted

attention

inception

to

to this

completion.
to Erich Gruen

study from
I am also
and Craw

grateful
ford H.

Greenewalt

Jr. for their gen

American

School

The

of Classical

erous

encouragement
an initial
following

at an Art

my

arguments
Mediterranean
quium

of this
project
of
presentation
and
History
Collo

Archaeology
at the
of California,
University

Berkeley, inApril 2005. All drawings


and

are my

photographs

own.

2. For the history of the Koan


Asklepieion,
pp. 340-342,

Studies

see Sherwin-White
345-346.

at Athens

For

1978,
the devel

and
of the site,
opment
dating
Schazmann
and Herzog
1932,

see
p. 75;

Gruben 1986, pp. 401-410; 2001,


pp. 440-449.
3. Schazmann
pp. 72-75,

pis.

and Herzog

1932,

37,38.

4. A centrally placed propylon on its


north wing
served as the monumental
to the
entrance
Schazmann
sanctuary;
and Herzog

1932,

pp. 47-48.

556

R.

JOHN

SENSENEY

at Kos,

1.
Figure
Asklepieion
the middle
and lower
the upper
of the

Asklepios
restoration
the

terrace,

3rd-century

terraces

with

the

b.c.

of the altar

2nd-3rd-century

stoa

of

the

The
to

the

lower

terrace.6

first half of the 2nd century b.c. witnessed

upper

terrace

in a new

resulted

that

character

and additions

changes
for

the

as a

sanctuary

whole. To connect the upper terrace with the rest of the sanctuary below, a
new
a dominant central axis
a
grand staircase created
(Fig. 3).7 In addition,
new

marble

stoa

the

replaced

earlier

timber

structure.

In

the

center

of

the

50.4 x 81.5 m

approximately
temple of Asklepios
ple A

space enclosed by this stoa, a marble Doric


was
as
as 170 b.c.,
as
begun
early
today referred to Tem
to distinguish
it from the earlier temple of Asklepios

(Figs. 2-5)
on the terrace below.8
(Temple B)
placed before
Axially
the
middle
and
lower
terraces, Temple A became
overlooking
visual focus of the entire Asklepieion.
The
Doric

choice

of the Doric

stoas continued

through

order for Temple

to be common

the Hellenistic

fig. 74.
7. Schazmann

terrace, were
upper
this level. For these

pp. 22-24,
fig.
54.
45-48,

found

on

as well
features,
as the
monumental
altar,
2nd-century
see Schazmann
and Herzog
1932,

pp. 25-31,
pis. 12-14.
6. For
later marble
mann
figs.

34-39,

49-51,

60, 73,

is an archaism. While

and the nearby

5. Important
utilitarian
features,
as a
and wells
lo
springhouse
cated along the retaining wall for the
also

the dramatic

in all areas of the Greek world

period, Asia Minor

such

the staircase

down

islands reflect

62, 75,98,109,112,149,159,171,246,
and
18, pis.

8. Schazmann
pp. 3-13,
figs.
ple is oriented

and
3-14,

Herzog
10,11,

1932,
37-40,

1932,
Herzog
tem
1-6. The

pis.

25 degrees west
of north.
on a foundation
the
of limestone,
con
the
of
is
superstructure
temple
with
the
structed of marble
throughout

Built
the timber

and

portico

replacement,
and Herzog
1932,
15-17,
pi. 9; Coulton

its

see Schaz
pp.

14-21,

1976,

pp. 9,

of courses of poros
limestone
exception
blocks
in the interior walls
of the naos.

of

from

remains

Temple

(left), the 2nd-century

of

b.c.

and
(center),
a.d.
restoration

of the Temple of Apollo


to Asklepios
see
and Apollo
(Temples B and C, respectively;
Fig. I).5 On
the upper terrace, a TT-shaped stoa of timber construction
balanced the

view

(right)

IDEA

AND

IN

VISUALITY

HELLENISTIC

ARCHITECTURE

557

-C

2.

Figure

at Kos,

Asklepieion

view

:::?-%--S?:
;?:
?,',:R"?:?".

of the remains (in situ) of the upper


terrace

from

complex

the

southeast,

looking toward Temple A, with the


stoa

in the
foreground

Figure 3. Asklepieion
restored

plan

==J

at Kos,

of the upper

terrace

complex with Temple A

3W

for the Ionic order for temple architecture. As Vitruvius


predilection
bolstered this pref
such as Pytheos and Hermogenes
architects
indicates,
as its
erence with a theoretical
(Vitr. 4.3.1-2).9 Furthermore,
justification
was
measurements
traditional
in
its
omission
of
demonstrate,10 Temple A
to
and
Pytheos
Vitruvius
mentions

9. In addition

Hermogenes,
architect Arkesios,

to the 3rd century.

who

perhaps

For Arkesios,

the
dates
as well

as the
convincing
looked arguments
Vitruvian
the Doric

and

still much

over

the common

against
of a "decline"
conception
order in the 4th century

of
b.c.,

seeTomlinson
10. Schazmann
pp. 3-5,

pis. 2-5.

1963.
and Herzog

1932,

558

JOHN

R.

SENSENEY

??-?:-;
u ?;?:
i:
?:: :

?x??
::t:?i

r-?:?ii
:I
"'?ic??:r?
?-:

Figure
Temple

4. View
A

from

of the
the

remains

of

southwest

the kind of novel modifications

and "optical refinements" characteristic of


also
the
Foregoing
interesting and easily detectable schemes
of Ionic temples associated with Pytheos
and Hermogenes
(Fig. 6), the
seem to have been a
would
conventional
temple
strictly
reapplication of
the Parthenon.

order in the 2nd century b.c.


character of Temple A may represent only
the straightforward
a part of its story. As I argue below, a
measure
geometric
analysis of its
ments reveals the use of a compass in constructing
the interrelationships
of architectural
in plan according to circumferences.11 The di
elements
the Doric
Yet

ameters

share a simple arithmetical


relationship
a 3:4:5
of
proportions
Pythagorean
triangle,

of these circumferences

based on the whole-number


rather than amore

to irrational
strictly geometric
relationship pertaining
like v2 or v3, or their fractional approximations. The geometry
of the temple's plan is therefore very simple, and is not to be confused by
the analytic geometry
it.
required to substantiate

numbers

circumferences
concealed within
the
presence of theoretical
nature
raises
features
about
the
of
the
building's
interesting questions
Doric design process on aHellenistic
architect's drawing board. That such
an
is found in only a single (albeit prominent)
underpinning
example of
Greek temple architecture, as opposed to the more widespread
approach
of grid patterns, does not detract from its significance. As I will discuss,
the uniqueness
in a temple plan?as
of circumferential
op
relationships
to
the
kind of orthogonal
that temples of the Ionic
posed
relationships
The

to a dearth of specifically Doric temples during the


order permit?relates
inTemple A demon
Late Hellenistic
period. The interesting geometry
an important architectural
tenet that we might
strated here exemplifies
term "cryptomethodic,"
to
the
features
of the design
systematic
referring
process that cannot be appreciated
be recovered only through detailed

through
study.

casual observation,

but may

11. For
plans
berger

an excellent

in ancient
1997.

discussion

architecture,

of

see Hasel

IDEA

AND

IN

VISUALITY

HELLENISTIC

ARCHITECTURE

559

m
?-#-3.080
?^

-7.228
1

Joid
-

1.016*

CD-"

4.368

4.435-1?

&
9.272

k-+

i.ioa
O

T-"^.I
18.075

Figure

state

5. Measured

plan

of

Temple A according to material and


trace

remains,

exposed
foundations

shown

limestone

without

masonry

the
of the

-?-

."" 1"...'".""H

-??-~~-*

10 rn

56o

R.

JOHN

SENSENEY

IDEALISM AND HELLENISTIC VISUALITY


a
turning to technical discussion,
that an ancient architect should design

Before

Iwill first address the very premise


on geometry that
building based
the final product.12 It is important to
a

does not correlate experientially with


state from the outset that architects of the Hellenistic

world thought about


their buildings in terms different from those used by architects today.We
that Greek architects called their plan, elevation, and
know from Vitruvius
to the notion that
perspective drawings i??ou (Vitr. 1.2.1-2), corresponding
uses
to
in reference
the transcendent ideas (or forms) that
Platonic idealism
are
to be the ultimate
thought
reality underlying the perceptible objects of
the everyday world. As Lothar Haselberger
has admirably observed, the
is
between the philosophical
and architectural meanings
correspondence
not casual,13 and the full implications of this correlation have yet to be
appreciated in studies of ancient architecture.
rhetorical manner

The

inwhich

Plato

sometimes

this ideal

discusses

can seem

our own way of


as when he
ist vision
quite foreign to
thinking,
can
presents Socrates' argument that couches manufactured
only
by artisans
a realm
an
our
in
imitate
couch
existing
beyond
archetypical
imperfectly
senses

(Resp. 10.596e-597e).
to these isolated metaphors

to reduce Platos

it is unfair

Yet

conception
and parables, and the lack of any clearly stated
our attention instead
unifying theory of ideas in Plato's work should draw
as amodel for systematic
to the more
of
mathematics
general importance
universe

to the ultimate realities of the


of penetrating
idealism. Perhaps the most articulate expression of this

methods

and hierarchical

in Plato's

is the well-known
way of understanding
passage inwhich Socrates, after
an uneducated
a
slave
that
geometric proof, concludes
through
guiding
in the world
eternal truths lie beyond our embodied experiences
(Meno
to
is
it
theoretical
rather
than
the
Platonic
the
model,
82b-86c). According
the sensory that is privileged.
What

systems

geometric

to bear on the question of underlying


in architectural plans is what J. J. Pollitt terms the

this discussion

brings

of the Hellenistic

"scholarly mentality"

Roman

12. The

considerations
following
to the ancient world,
only
pertain
to how
but more
culturally
generally
of the world
based understandings

discussion

of this

in which

objects
For
constructed.

idea

Cartesian

are
a

in the contexts

perspectivalism,
and 19th-century
painting,
see
Jay 1988, esp. pp.
phy,

early modern
photogra
16-17. A

definition of visuality offered by Nor


man Bryson (1988, pp. 91-92) has
recently been evoked by Jas Eisner in
his new
context:
structs
between
screen

in a classical
study of visuality
con
of cultural
"the pattern
that stand
and social discourses
the retina
through

which

and

a
the world,
... Greek
and

had

people

in the

originating

no choice

but

to

look and through which they acquired

not

the way
anticipate
viewed
and visually

age.14 Perhaps

(at least in part) their sense of subjec


I focus
tivity" (Eisner 2007, p. xvii).
texts
and what
here not on subjectivity

of

images
in the classical

how
Temple
were

tell us about

can

and

world,

but

visuality
rather on

of
underpinning
to ways
of seeing that
and socially
conditioned.

the geometric
A relates

culturally
13. Haselberger
and primary
92-94,
sources

esp. pp. 77,

and

secondary
Arendt
(1958,

cited. Hannah

p. 90) offers
articulation
to notions

1997,

an excellent
of how
of models

philosophical
ideas relate
Platonic
and measures,

which may be useful for framing the


conceptual

connection

between

ideas

and

of the

drawn

based

geometrically
in architecture:

into measures,

ideas

models

"For the transformation


is

Plato

helped by analogy from practical life,


it appears
that all arts and crafts
are also
guided
by 'ideas,' that is, by
the 'shapes' of objects, visualized
by the
inner eye of the craftsman
then
who

where

them in reality through


reproduces
imitation. This
enables him
analogy
to understand
ter of the
he does

the transcendent

ideas

in the

the transcendent

charac

same manner
existence

as
of

the fab
lives beyond
it
and
therefore
process
guides
can
for
become
the standard
eventually
its success or failure."
the model,

which

rication

14. See Pollitt 1986, pp. 13-16.

IDEA

----m

AND

VISUALITY

IN

HELLENISTIC

?g.|

.?j

I??j?I??h?-1|?H?|

1,1.

Jlj
rrriT;r
Ifli

|
tia

11_lij_|

IB ?P

ARCHITECTURE

561

g?-^_||L?

ft

lE iB] f?fjf
loig,.?II_JiiL-JSIJ_lllll_?

[m

0
Figure 6. Restored plans showing the
grid systems of Pytheos's Temple of
Athena Polias, Priene (left), and
Hermogenes'
Magnesia
p. 70, fig. 23

Temple
After
(right).

of Artemis,
Coulton

1988,

10m

a taste for didactic


of the Library at Alexandria,
came to
displays of abstruse knowledge
strongly characterize Hellenistic
art and literature. A notable feature of works appears to have been the
deliberate potential for simultaneous appreciation from both common and
In architecture, in particular, this tendency is found in
erudite perspectives.
intellectual

ambience

as
Pytheos's Temple
examples such
inwhich the masses might marvel
knew

the building's proportions


of mathematical
precision.15

of Athena Polias at Priene (Fig. 6, left),


at its surface qualities, while those who
its plan as an expression
could understand

text depends
in part upon the writings
of earlier
Vitruvius, whose
insists
He
that
this
Hellenistic
architects, exemplifies
scholarly emphasis.
as
an architect's
in
and
like
geometry, music,
background
disciplines
tronomy is requisite (Vitr. 1.1.4, 8-10), a claim that he backs up at times
his eagerness to show
with pretentious displays of erudition. Sometimes
that he discusses, as
exceeds his command of the material
his knowledge
of the doubling of the square,
to the Pythagorean
which he follows immediately
theorem, without realizing that both of these theorems illustrate an identical
such limitations, he
(Vitr. 9.Praef.4-7).16 Despite
principle of proportion
when

15. Pollitt 1986, pp. 14-15.


16. See de Jong 1989, pp. 101
102.

he credits Plato with

the demonstration
with

an introduction

R.

JOHN

562

SENSENEY

*
5

,"i:.
?

, *

,,+

rt
.

~.: ",, !
I ,

,;

..~.~

,,
?,..

his scholarly quality in the context not only of general theory,


the plans of
but also of architectural design. His procedures for designing
both Latin and Greek theaters (Figs. 7,8) well illustrate this tendency, and
merit quoting at length. For the Latin theater, he writes (Vitr. 5.6.1-3):
demonstrates

Figure 7. Plan of a Latin theater


according

to Vitruvius's

description

to be constructed as follows.
plan of the (Latin) theater itself is
a line of circumfer
Having fixed upon the principal center, draw
ence
to be the perimeter at the bottom. In
equivalent to what is
it inscribe four equilateral triangles at equal distances apart and
The

as the
touching the boundary line of the circle just
astrologers do in
a
are making computa
figure of the twelve celestial signs when they
tions from the musical harmony of the stars. From these triangles,
side is closest to the scaena and in the spot
select the one whose
it cuts the curvature of the circle let the front of the stage be
located. Then draw through the center a parallel line set off from
that position to separate the platform of the stage from the space of

where

The wedges for spectators in the theater should


the orchestra....
run around the cir
so that the
angles of the triangles that

be divided

of the circle may provide the direction for each flight of


between
the sections up to the first curved cross-aisle. Above
steps
are to be laid out with aisles that alternate
the
this,
upper wedges
with those below. The angles at the bottom that produce the direc
tions of the flights of steps will be seven in number, and the remain
cumference

the arrangement of the scaena. In this


ing five angles will determine
center
in
the
the
way
ought to have the "palace doors" facing it
angle
and the angles to the right and left will designate the position of the
doors for "guest chambers." The two outermost angles will point to
the passages in the wings.17

17. Smith
S. Kellogg.

2003,

pp.

165-166,

trans.

IDEA

AND

IN

VISUALITY

I
I

HELLENISTIC

ARCHITECTURE

563

C
C

C
Ir
\ C?
r
-------?1---?--------??-,,,*,II,,,
(r
fr
C\CCI
*?
Y
r.f
L' '
1,
r
Ir \t'
I I`rr \
I,II
"
r
r, i
r
Ir\ II'
( II
"'?'~''"""~""~"""'~~""""""'
rtt ''
r
r
\
'r
?I
1
,r
I\ c
rr
t Z
r
r
rrr
4\
rr
r
?r
t
.
rt ?
\
1.
tr r rr ,
r
?
r t
rr
I ?r
f
r
I? r
c r
t, r
\ ?I
I

Figure 8. Plan of aGreek theater


according toVitruvius s description

'4?
??r \

for the Greek

And

t
?-Ir
?) C

theater

(Vitr. 5.7.1-2):

theaters some things are done differently. First, in the bot


tom circle, while the Latin theater has four triangles, the Greek has
three squares with their angles touching the line of circumference.

In Greek

The

is determined by the line of the side of


limit of the proscenium
the square that is nearest the scaena and cuts off a segment of the cir
cle. Parallel to this line and tangent to the outer circumference of the
segment, a line is drawn that delineates the front of the scaena. Draw
a line
through the center of the orchestra and parallel to the direction
of the proscenium. Centers are marked where it cuts the circumfer

ence to the
at the ends of the half-circle. Then, with
right and the left
the compass fixed at the right, an arc is described from the horizontal
distance at the left to the left-hand side of the proscenium. Again,
with the center at the left end, an arc is described from the horizontal
distance at the right-hand side of the proscenium_Let
the ascend
as
as
of
between
the
of
far
the first
seats,
steps
up
ing flights
wedges

18. Smith
S. Kellogg.
19. For
1989.

For

2003,

the Greek
a discussion

pp.

167-169,

material,

trans.

see Isler

of

questionable
to match Vitruvius's

attempts
scholarly
of Roman
theater design
description
to later Roman
see Sear 1990;
theaters,
2006,

pp. 27-29.

curved cross-aisle, be laid out on lines directly opposite the angles


of the squares. Above the cross-aisle, the other flights are laid out
between the first.At the top, as often as there is a new cross-aisle, the
number of flights of steps is always increased by the same amount.18
are not easy to follow, and it would be
to
prescriptions
tempting
them as indicating a fussy outlook on the part of Vitruvius
if not
for the fact that these geometric constructions were applied in surviving

These

dismiss
Greek

and Roman

to a basic geometry
prescriptions pertain
or squares, rather than considerations
triangles

theaters.19 The

of forms such as equilateral

564

JOHN

R.

SENSENEY

based on irrational numerical

case of both theater types,


relationships. In the
not contribute to
the cryptomethodic
described
would
patterns
arguably
nor
most
ancient
visitors have
would
any visible harmonic relationships,
been likely to perceive them. Certainly, there are less theoretically grounded
and
the locations of radial stairways, boundaries,
ways of determining
on
seem more sensible to
it
forms
based
and
would
doorways,
design
simple
intuition and functional criteria. On the other hand, our very concern with
issues may be a consequence
of an inherently modern prerequisite
that the design process directly correlate with sensory experience. For the
world who could read and understand
privileged few in the Hellenistic
these

such passages,

there was value of a different


to aworld

for material

Furthermore,
dence of underlying

kind: the value of discourse.

pertaining
ideas, our own privileging
in the final built form is arguably misplaced.

that validates

the indepen
of the tangible properties

Vitruvius's
reference to the drawings of astrologers reveals a signifi
cant interdisciplinary
issue at work in such architectural ideas. Given the
interests
of Hellenistic
architects, there is no reason to believe that
scholarly
in reference solely to the
the practice of architectural drawing developed
construction
Another
drawn
that Vitruvius describes
of
designs
buildings.
avaXrijiua, which was the graphic reference for solar
declensions
that served as the basis for sundials (Vitr. 9.1.1, 9.7.2-7). He
recon
an
algorithm for the drawing, which has allowed for its
provides
in detail

is the Greek

struction as amarkedly circumferential design (Fig. 9).20 In this curvilinear


the cir
quality, the analemma provides intriguing general comparisons with
cumferential geometry underlying the design of Temple A at Kos proposed
Interestingly, Berossos the Chaldean, whom Vitruvius credits with
to Kos and established
the invention of the semicircular sundial, moved
the Great s conquest of
there a school of astronomy following Alexander
In the course of the 3rd century, Berossos s
(Vitr. 9.2.1,9.8.1).
Mesopotamia
school amalgamated with elements of the Koan medical school to establish
below.

the discipline of medical astrology, concerned particularly with the moment


of conception as the basis for casting nativities (Vitr. 9.6.2).
I do not argue that there was any symbolic connection between Tem
and the analemmay let alone some sort of mystical value. The study
A
ple
is an inexact science, and we cannot lose
of architectural
iconography
we deal with a design that pertains solely to the
sight of the fact that here
architect s drawing board; it is only the circumferential
approach that is
of
of
shared
the
(and
ways
similar, underscoring
envisioning
possibility
?izz?ary drawing) forms among architects and those concerned with astral
as well as geometry. As I demonstrate
below, the curvilinear
phenomena
of Temple A pertains not to solar declen
element in the underpinning
to
sions, but rather to a Pythagorean
triangle. In this aspect, it is similar
the ways inwhich Greeks and Romans began their theaters with squares
an outlook characteristic of the ways in
or
triangles, and consistent with
the drawn
which educated men of the Hellenistic
period thought. While
an eternal and abstract form of the idea, the final built form
plan expresses
into presence inways that need not readily unveil its un
idea
that
brings
truth to the senses.
mathematical
derlying
then, visuality in architecture was consti
of the casual viewer, but also by the
solely by the perceptions

In the Hellenistic
tuted not

period,

20.
trations

Howes
e.g.,Thomas
in Howe
and Rowland

See,

pp. 288-289, figs. 114,115.

illus
1999,

IDEA

Figure

9. The

analemma

according

AND

VISUALITY

IN

HELLENISTIC

ARCHITECTURE

565

to

Vitruvius's description
spon
certainty of geometry. Framed by a monarchically
ac
agenda and the resulting practices of visualizing form
to geometry, modes of visual representation established the starting
cording
for
form in disembodied
abstractions
that were subject to math
point
epistemological
sored scholarly

rules or norms.

ematical

In this way, the squares underlying


the placement
were
of empirical features within
the curvilinear Greek theater
patently
mea
real. Similarly, the circles (and the Pythagorean
triangle that gives
sure to their
of
the
rectilinear
forms
underlying
experiential
proportions)
relate to a primary consideration:
the
that defines the visuality of the building by constructing
its
geometry
eternal idea. The square, circle, triangle, and other shapes are the ideas of
nature that engender the visible things in the world, be they a theater, a
or even a human
temple,
body (Vitr. 3.1.3). As in Vitruvius's discussion

Temple

at Kos discussed

below

of theaters, the drawing of a building may begin with geometry alone, and
only through the process of design arrive at the final form. In further sup
I argue that the design for the plan of Temple
port of these observations,
a
similarly began with the drawing of Pythagorean
triangle, from
the design and construction
which
evolved into a completed expression
that continues to reflect its origin, however imperfectly.
A at Kos

QUESTIONABLE METHODOLOGIES
very suggestion of a hidden system within an architectural
to touch a raw nerve among archaeologists
and architectural

The

plan tends
historians

alike. Far from striking an innovative note, such an approach falls squarely
a tradition that has so tried the
one
patience of readers that itmay

within

day risk outright exclusion from mainstream


scholarly research. Before
to
it
is
circumstance.
address
this
necessary
briefly
proceeding,

566

R.

JOHN

SENSENEY

In a penetrating essay on the Parthenon, Manolis Korres offers amark


assessment of efforts to present that celebrated building as an
edly negative
expression of ideal numerical relationships and harmonious proportions.21
such studies as "pseudo-science," Korres notes the disturb
that contradict the reality of the building.
ing tendency
error in the
to him,
According
approaches include impudent suggestions of
Characterizing

to argue theories

the reliance of proposed


temple's construction and published measurements;
on inaccurate, small-scale drawings rather than the
theories
geometric
found in the actual building; the inability to credibly
degrees of magnitude
the proposed geometric
shapes with analytic geometry; and the
or even
that may underlie such studies in
motivations
mystical
like these have articulated and, justifiably,
the first place.22 Observations

correlate
obsessive

even reinforced

general reservations about the rigor and value of


studies of architecture in various periods and
and
geometric
metrological
world.23
locations in the ancient Mediterranean
perhaps

to future stud
Although Korres's remarks may provide salutary caution
of geometric
ies, we may not entirely benefit from severe marginalization
one thing, the Parthenon
antedates
For
in
Greek
analysis
temple design.
the use of scale drawings in architectural planning.24 In buildings of the
were used
con
Hellenistic
(Fig. 6), questions
period, when such drawings
more
of
become
basis
the
considerably
applicable.25
cerning
geometric
plans
at the close of the Hellenistic
(1.2.1-2)
clearly
period, Vitruvius
Writing
describes Greek temple design process in terms of Tragic, or the creation
of a quantitative geometric
system, and SiaGeGi?, the placement of archi
Vitru
tectural elements according to that established geometry.26 While
vius's comments

statements

21. Korres's
pp. 79-80)
going back

(1994,

on Greek
scholarship
not be aware of the

might
attached
of esteem

degree
and his views,

to Korres

own

my
referenced

2005).
"genius" (Hurwit
the remarks of Korres

In

tically incisive,
that I invoke as a
and their implications
used
for
the
methodology
background
A at Kos. To
in my
of
analysis
Temple
be clear, I in no way draw any meaning

ful comparison between Temple A and


greater

of a different

era. The

environ

architectural

(1972), in

of the Asklepieion
analysis
both a proper
trigono
Lacking
and convincing
identi
analysis

metric

of salient

fication

structures

to relate

tended

features

architectural

to his
geometry,
proposed
that the sanctuary's
suggests
eras were
in
from various

lines

to one

established

another
at

through
related

angles
sight
to the
section." He does not
"golden
an
A. See
of Temple
attempt
analysis
Doxiadis
1972, esp. pp. 125-126,

fig. 77.

its markedly
and

sophistication
is an expression
the Parthenon
details,
of a completely
different mentality
from what we find in Temple
A, and
the product

Similar
pp. 79-80.
toward the
be directed

his

pertaining
Doxiadis

are characteris
study
and it is these remarks

in this

in execution

may
of Greek

cluding
at Kos.

to

With

1994,

to method

carry

ments by C. A. Doxiadis

view,

the Parthenon.

22. Korres

practice,

implications
of ancient
analysis
and not to the
temple architecture,
of the Parthenon
architecture
per se.

study

referred

Jeffrey Hurwit

ological
for any geometric

that

criticisms

that comes
something
set
across
in less formal
particularly
a recent
In an aside during
public
tings.
in Chicago,
Institute
lecture at the Art

for example,
as a
Korres

pertains

issues

(1923a, 1923b). Readers

architecture

discussion

lowing

similar misgivings
at least as far asWilliam

with

represent Hellenistic

comprehensively

reflect

Bell Dinsmoor
less familiar

cannot

fol

is

advocates

"a general

architects

exploited

rule

[that]

geometry

... but not

details

for the

unless
of whole
buildings
composition
or
con
were
concentric
they
partially
own
in
Wilson
centric
Jones's
plan."
of such complex
geometry
acceptance
and numerical
ings and other
in turn, elicited
to material

in circular
systems
structures
(2000b)
doubts

extending

of the Roman

2001.
e.g., Yeg?l
24. A particularly
in favor of detailed

period;

forceful
architectural

build
has,
even
see,

argument
draw

for at least one Classical-period

ings

Athenian building, the Propylaia, is


Dinsmoor
on

the

plans

Jr. 1985.
introduction

in Greek
1997,

berger
25. For

For
and

views
opposing
role of drawn

architecture,

p. 83.
the development

see Hasel

of scale

plans during theHellenistic period


and alternative
design

In support of his theory for


in the 5th
"facade-driven"
Doric
design
b.c., Mark Wilson
(2001,
Jones
century
23.

p. 678)
ancient

for resolving

in earlier

modes
periods,

pp. 51-67.
26. For
complexities

of architectural
see Coulton

1988,

Vitruvius's

use

in this passage,
p. 217.

of Greek

arising

from

terminology
see Fr?zouls
1985, esp.

IDEA

AND

VISUALITY

IN

HELLENISTIC

ARCHITECTURE

567

his stated reliance upon Greek architectural writers arguably merits con
into the geometric
of Hellenistic
tinued investigations
underpinnings
to
In
Vitruvius's
adherence
addition,
ap
apparent
grid-based
buildings.
proaches in Ionic temple design27 might elicit inquiry into procedures that
he does not elaborate upon: in the Doric order, where intercolumnar spatial
contractions do not lend themselves to an orthogonal grid, how might the
of taxis differ?28
geometric constructions
a

the Parthenon will continue


furthermore,
privileged monument,
to be a favored object of attention for numerous
lines of inquiry despite
of particular approaches. Dating
from the 19th century
condemnations
are
re
too
volumes
of
onward, however,
many
archaeological
published
we
ports with scientific measurements
pertaining to buildings about which
As

still know

relatively

electronic

little. To allow these to gather dust or occupy unused


space, instead of reaping what the laudable efforts of

storage
their excavators can tell us about ancient

design, will benefit


nor architectural historians. Furthermore,
accusa
neither archaeologists
tions of "intellectual totalitarianism"29 directed at proponents of geometric
architectural

serve
only to curtail productive discussion.
analysis could
as ra
Rather than framing various outlooks as scientific or mystical,
tional or obsessive, we might
instead see observations
such as those of
to reevaluate the methodologies
in
Korres as an opportunity
employed
and geometric
analyses. In addition, an inclusive view may
proportional
us
to methods
that allow for scientifically sound analyses that, in turn,
open
a
our
of Hellenistic
solidify
understanding
temples. Finally,
responsible,
and
mathematically
rigorous,
computer-based
approach to geometric
us build upon and refine the criticisms,
analysis will help
tations of similar studies.

rules, and expec

The present study uses analytic geometry and vector-based AutoCAD


of the design of
(or CAD) software to analyze the geometric underpinning
at Kos. In the course of this analysis, I also consider
Temple A
questions
the perceived
limitations of studies that attempt to unveil
surrounding
hidden numerical and geometric
systems. In order to avoid the inevitable
distortions of proportional
and geometric
that look correct
relationships
on
a plan drawn to reduced scale, my
only when overlaid
study instead
on the
measurements.
relies
In
other words,
directly
buildings published

27.
Howe

the comments

See

in Howe

pp. 5,14,149.
28. See Vitr.
expresses
tradition

his

of Thomas

and Rowland

1999,

where

4.3.1-8,

indebtedness

he

to the Ionic

of

by charac
as deficient,
interaxes
issue of columnar

Hermogenes
the Doric
order

terizing
leaves the

on elevations
and focuses
unexplained,
at the expense
of any discussion
of
of
Jones's related notion
plans. Wilson
"facade-driven"

Doric

design

(2000b,

pp. 64-65; 2001) is discussed below; see


also n. 23,

above.

29. Korres

1994,

p. 80.

the proposed geometric system is now mathematically


verifiable rather than
we
is
in
and
So
that
intuitive,
may furthermore
computation.
grounded
ensure both mathematical
accuracy and the relationship between the nu
systems and the concrete, graphic form of the revealed geometry,
the calculations have been verified through the use of AutoCAD.
CAD
is not a requisite for this study, but merely a convenient
tool that may

merical

allow

researchers

and

readers

simpler

recourse

to

the measurements

of

in an architectural

proposed relationships
tions themselves that demonstrate

form; it is ultimately the calcula


the geometry. This combined Cartesian

carries the potential of standardization


and computer-based
method
for
future studies, allowing for a truly scientific approach inwhich results may
be replicated to confirm their veracity. Provided that an analysis such as
this one relies upon previously published numbers rather than ones own
we may now set aside
measurements,
suspicions of personal agenda and
in the objectivity of the process.
have confidence

R.

JOHN

568

SENSENEY

ANALYSIS OF THE BUILDING


The

Metrology

Before

dimensions
revealed through analy
discussing the nonorthogonal
should consider the temple s general measurements.
From the 5th
was
a
common
rule of thumb that the width:length
century onward, it
ratio of a Doric
should match
temples plan (including the euthynteria)
sis, we

on the short and


long sides of its peristyle.30 With
rear and eleven along its flanks,
its
front
and
along
Temple A
no
appears to be
exception (Fig. 10). In plan, the temple's overall dimen
sions are 18.075 x 33.280 m,31 a differential of only 0.4% from a proper
the number

of columns

six columns

as a 0.143 m reduction of the overall


simple adjustment, such
or a 0.078 m increase in the width, would result in a
length
perfect whole

6:11 ratio. A

ratio.

number

in
Scholars usually account for such "errors" by citing constructional
as well as centuries of exposure to the ele
exactitude and adjustments,
ments.32 Other
the temple might
slight irregularities found throughout
the theoretical design and
support this notion of a difference between
the actual built form (Fig. 5). For example, there are slight variances in
the thickness of the eastern and western naos walls (1.028 and 1.016 m,
to the
respectively) and in the distances from the exterior of these walls
edges of the stylobate (3.313 and 3.380 m, respectively).33 As
naos is not centered on the
stylobate.34

a result, the

over
and deterioration
factors such as imperfect masonry
Although
for such disparities,
additional
consid
time are plausible explanations
erations deserve emphasis. If it is the architect's design to begin with a
6:11 plan, other features might
the maintenance
of
complicate
as the construction progresses.
in
the
final
built
form
perfect proportions
In the end, there will be a set of measurements
that are necessarily inter
as
of
the
and
the overall dimensions of
such
the
widths
related,
krepidoma
proper

30. See Coulton

Wilson

1974,

on

pp. 62-69;

Jones 2001, p. 694.

31. Schazmann

and Herzog

1932,

in courses

pi. 2.
32. For
ence

the architect

Wilson

and

the final

see

product,

Jones 2000b, pp. 11-14; Dwyer

p. 340.
33. Schazmann

2001,

pi.

and Herzog

1932,

34. The
reason
result

excavators

of Temple

is a
lack of symmetry
of earthquakes
that have shifted
that

the entire
an

and pronaos
eastward,
that I find unconvinc

explanation
see Schazmann
ing;

and Herzog
1932,
on
on-site
there
6.
Based
p.
my
analysis,
are differences
in the limestone
foun
dations
sides

on

the eastern

of the naos. While

and western
the masonry

that are roughly perpendic


In addition,
the

continue

tendencies

divergent
raised foundations
where

of the naos

the

separate
ing the

should

crews were
limestone

indicate

responsible
foundations

that
for

lay
on either

than
side of the temple. More
plausible
the eastward
shift of the entire celia is
that one
in

crew

establishing
or

stylobate,
resulting

committed
the eastern

possibly
in a distance

a minor
limit

error
of the

the euthynteria,
from the celia

is

edge of the euthyn


and southern
sides,
to 4.368 m on the eastern

the western

m was

the eastern
would

celia

axis of the

central

that

originally
side as well,

be a more

if the entire

difference

on

to the outer

opposed
side. If we maintain

4.435

itself,

that found

less than

side. This

as

into

interpretation
distances
supported
by the nearly equal
of 4.43 and 4.435 m from the naos

teria on

the two separate


approaches
at a line west
of the central axis

The

is 0.067

the western

walls

of the naos.

this

naos

that

on the western
side are tighter
joints
than those toward
the east. These

meet

2.

in courses

to these walls.

ular

the problem
of the differ
the abstract vision of
between

side runs

the eastern

of
that are parallel with
the long walls
the naos, that on the western
side runs

approximately
intended
for
the result

balanced
originally

than
design
on the
lay
in its present

temple
I therefore
favor

dimensions.

human

causes for
to natural
opposed
in the
the lack of symmetry
temple's
measurements.
Such errors can and do
error

occur

as

in the

laying of foundations,
of elements
the placement

affecting
the superstructure.

in

IDEA

AND

VISUALITY

IN

HELLENISTIC

ARCHITECTURE

569

14.666 M = 45 F = 24 T

4^

i-4

o
Uh

10m

Figure 10. Restored plan ofTemple A,


with
axes

measurements
(M

meters,

of the colonnade
F = Doric

T = triglyph width modules)

feet,

SENSENEY

R.

JOHN

57?

in turn relate to the sizes of the paving slabs and the


the stylobate, which
area where the relative propor
spacing of the columns they support. The
is the conversion from
tions of the various parts are subject to modification
the abstract units of the drawing board (such as 6 x 11) to actual metric
must be privileged
values. In determining
specifications, certain distances
while others must be adjusted to the space allotted them. Considerations
of the paving slabs, for example, may
such as the specific measurements
a
ultimately result in slight departure from the integral proportions of the
s
architect
original drawn plan.
One method of accounting for the overall and individual dimensions
of a building is a metrological
analysis. A recent study proposes that the
architect of Temple A first worked out the overall dimensions
according
were the usual corner
to a specific metrological
system.35 Only thereafter
to
of the Doric order worked out, resulting in adjustments
contractions
of the theoretical plan. This theory, however, relies upon
the dimensions
the
the identification of a 0.305 m "foot" as the common unit underlying
no such unit of
system. Simply put, there exists
temple's metrological
measurement
in the ancient Greek world, a fact that the theory's authors
in our understanding
of
contend with by advocating greater flexibility
Greek metrology.36
Instead of suggesting
issue of commensuration.

we may consider the


Jones
specifically, Wilson

new units of measurement,


For Doric

temples,
exam
system, at least for 5th-century
a
to this theory, the width of
standard triglyph expresses
ples.37 According
the module that establishes commensurability
throughout various elements
itself commonly corresponds to a
of the building.38 The triglyph module
a standard foot (e.g., 25 or 30), with a dactyl equal to
5-dactyl multiple of
standards.39
1/16 of a foot in accordance with Greek metrological

makes

a detailed

case for amodular

for the remains in situ are available for the


InTemple A, measurements
western half of the columns on the rear
the
and
central columnar interaxis
of the stylobate, aswell as four columns along the western lateral colonnade
m for the missing
eastern half
(Fig. 5). At the rear, the addition of 5.793
+
m
+
a
14.666
for
in
of
the entire
results
3.080
5.793
m)
(5.793
length
axis (Fig. 10). For the long sides, the temple's excavators posit columnar
of
interaxes of 3.05 m based on the remains in situ and a consideration
0.61 and 0.915 m, respec
the triglyphs and metopes, which measure
one
of 3.034 m (see Fig. 5)
interaxis
the
Thus
preserved
tively (Fig. II).40
would represent an unintended departure from the theoretical constant of
3.05 m, and we may thereby restore the theoretical lateral axes, excluding
the contracted corners, to a length of 24.4 m (8 x 3.05 m), as in Figure 10.
and the 14.666 m
the 24.4 m axes of the lateral colonnades
Therefore,
axes of the front/rear
respectively,

would

of a value equal to 0.61 m

35. Petit andDe Waele


36. Petit

colonnades

and De Waele

37.Wilson

1998.
1998,

esp.

an earlier essay, J. J. de
Jong
p. 62. In
to have
the measure
claims
analyzed
ments
of Temple
A, but offers no dis
or results
to his anal
cussion
pertaining
ysis;

see de
Jong

1989,

equal 40 and 24
(Fig. 10).41

esp. p. 104, fig. 3.

the possibility
have endured
author's
response

integral units,

Jones 2001. Regarding


that
into

comments
to Coulton

such

could

system
see the
later periods,
on p. 697, n. 107 (in

Jones 2001.

39. Wilson

Jones

2001,

frieze;
1932,

0.611
esp. p. 690.

measurements

are based

of the
surviving
fragments
see Schazmann
and Herzog

three

pp.

10-11.

41.24.4/40

1983).

38.Wilson

40. These
on

m.

= 0.610 m; 14.666/24 =

IDEA

AND

IN

VISUALITY

HELLENISTIC

ARCHITECTURE

571

.345fr
.960

1.5r

? .610 *

.803

\\\\ \ MI
<--1.056-5>j

IM I M

Hl

3.05 = 5 T

<?1.270-->

HS?1.515-3H<S-1.535?-3H
.358'
.355

5m

Figure 11. Restored elevation of


lateral colonnade of Temple A, with
measurements

in meters

width modules

(T)

and

also measure
0.61 m.42 A
Significantly, Temple As triglyph widths
shows that this value equals 30 dactyls of a 0.325 m
simple calculation

triglyph

"Doric"

shares a 2:3 relationship with


the
triglyph width
a 1:5 ratio with the average interaxial
of
the
spacing
and a 1:24 ratio with the axis of the facade colonnade?

foot.43 This

standard metope,
lateral colonnades,

all typical proportional


relationships
is
It
also
10,
ll).44
(Figs.
interesting
42.
pp.

and Herzog

Schazmann

combined

1932,

10-11.
43.0.610

from
m/30

= 0.02033

m.

Since

foot divides into 16 dactyls, 0.02033 m

x 16 = 0.325

m.

Varying

between

0.325

and 0.329 m, the Doric foot has been


known

since Wilhelm
study
D?rpfeld's
of
the late-5th-century
(1890)
inscrip
tion relating
in
the expenses
involved
the construction

of the Erechtheion,

and

with

measurements

various

buildings
Attica.
throughout

tions ofWilliam
who

coined

firmed

Wilson

Bell Dinsmoor
term Doric
of 0.326

m.

(1961),
con

foot,
See

also

Jones 2000a, p. 75; 2001,

p. 689. The
Salamis,
sent a

the

a value

taken

on the
Acropolis
The
investiga

metrological

previously
system based

relief

from

to repre
thought
on a 0.322 m foot

according toWilson
that these distances

Jones's study
of 40 and 24

to the measurements

according

of

Ifigenia Dekoulakou-Sideris
has now

Wilson
system

been

(1990),
shown

convincingly

by

Jones (2000a) to represent a


based

on a 0.3275

to 0.3280

Doric foot. For the divisibility of


the triglyph module into 20,25, 30,
etc., dactyls,
p. 690.

44.Wilson

seeWilson

Jones

Jones 2001.

2001,

JOHN

572

R.

SENSENEY

to precisely
75 and 45 Doric
feet.45 In
triglyph modules
correspond
to
of
the
actual
dimensions
the
drawn
the
building and
translating
plan
its features, then, it is reasonable to theorize that the architect may have
the colonnades
of the facade and rear, establishing
their axes
privileged
a
of 45 Doric feet. Through
this magnitude,
3:5 ratio finds the 75-foot
for the axes of lateral colonnades

measurement

(Fig. 10). This latter di


each of which
subdivides
into eight intercolumniations,
two
into two half triglyphs, one whole
and
metopes
triglyph,
(Fig. 11).
the distance separating the end columns of these axes es
Furthermore,
tablishes the measurements
for the contracted corners, and the remaining

mension

divides

of the facade and rear colonnades

three interaxial distances


to the criterion

according

the dimensions

of incremental widening

could be set

toward the center. In


slabs in accordance with

of the individual

paving
of the stylobate are es
spacing, the total dimensions
of
the
and
and
the
widths
stereobate
tablished,
euthynteria are set according
to the remaining distance necessary to maintain
the 6:11 ratio of the over

varying
this irregular column

all plan.
To insist upon this explanation, however, is to treat Temple A as we
one more
have the Canon of Polykleitos,
plausible theory
resulting in yet
that can never be proven. There are too many types of metrical units, too
and too many rationales for us to induce conclu
many ways of measuring,
a
sively
guiding metrological
is not so much a reasonable

system. What

is lacking in such approaches


to a pattern of numbers, such

correspondence
ratios, but rather something outside of the buildings
themselves that might verify the significance of those numbers, such as a
source or a basis in Euclidian geometry. While
Vitruvius validates
primary
as whole-number

the case for how this system relates


of the triglyph module,
to large-scale distances must remain provisional;
in this regard, we may
40
units
of
the lateral colonnades
for
the
wonder,
integral
example, why
as
corner
In
the
modular
interaxials.
exclude the
addition,
theory
applied
to Temple A cannot address a central aspect of design that is unrelated
to the trabeation: the placements of the walls of the naos and pronaos in
of
relation to the overall plan. We must therefore explore other methods
the relevance

analysis

in seeking

buildings

design.

The

Theoretical

to substantiate

Plan

a theory for the


underlying

and

Standards

for

logic of the

Accuracy

it is not enough to merely draw geometric


shapes
emphasizes,
to a scale of 1:100.46 Instead,
over the features of a
proposed
plan reduced
must be verified through analytic geometry.
In other
geometric
shapes
to
a
the elements they
words,
drawing should correspond
superimposed
As Korres

co
not
through Cartesian
overlap
only visually, but also mathematically
and
with
lines
ordinates with
expressed algebraically,
interrelationships
described in terms of slopes and curves with coefficient-based
formulas,
such a strict standard places a damper on continued
for example. Naturally,
to
ancient architectural plans, but the gains in
about
theorize
attempts
are arguably well worth the endeavor.
credibility

45.14.666 m is only 0.026 m


(or 0.18%)

in excess

of 14.640

m.

14.640/24 = 0.610 m; 14.640/45 =


0.325 m; 24.4/40 = 0.610 m; 24.4/75 =
0.325

m.

46. Korres

1994,

p. 80.

AND

IDEA

IN

VISUALITY

ARCHITECTURE

HELLENISTIC

573

a
point of emphasis has been that the degree of accuracy in
must
in
the
actual
the
tolerances
theoretical
geometry
approximate
plan's
the accuracy of the built form, however, elicits
construction.47 Determining
a bit of circular reasoning, since many of its elements must be measured
Another

same theoretical
to
plan that its author attempts
support.48
against the very
This need not be the case for every feature, however. In Temple A, for ex
columnar interaxes
ample, there exist slight variances in the noncontracted
such as 3.050 m and 3.034 m,49 that are likely to
of the lateral colonnades,
relate to a theoretical constant rather than an intentional irregularity.
On a larger scale, we may note that Temple As naos (9.272 m wide,
not in the exact center of the stylobate (15.965 m
including itswalls) lies
0.067 m off axis (Fig. 5).50 Given
the gen
wide), but an imperceptible
for symmetry even in conjunction with "optical refine
eral predilection
one
be hard-pressed to argue the plausibility of this feature as
would
ments,"
intentional. From the outer wall to the edge of the stylobate, the dis
sides of the naos measure

tances on the western

and southern

3.375 m,

and the diverging measurement

eastern

respectively,
side

an

represents

error

of

ca.

3.380

and

of 3.313 m on the

1.98%.51

Still, it may be inadvisable to isolate this error in the eastern pteron,


since the final built form is the product of multiple
interrelating compo
nents. The most conservative approach would be to calculate the percentage
of tolerance according to the entire width of the stylobate. This calculation
should pertain to the theoretical plan rather than the actual plan, with the
m from the eastern
only difference being the addition of the "missing" 0.067
side of the temple, resulting in awidth of 16.032 m for the stylobate and
an overall width of 18.142 m (see
the
1). In order to maintain
Appendix
strictest possible tolerance in my analysis of this theoretical plan, I will
cap the standard for accuracy at 0.42%
discussed here.52

in accordance with

the divergence

in the
It is important to emphasize
that this addition to the width
not
is
in
"stack
the
deck"
for
the
and
does
any way
slight,
plan
results of the analysis that follows. Instead of adding 0.067 m to the nar
rower side, we may be justified in adjusting for symmetry in the theoretical
the actual width and shifting the naos to the
plan either by maintaining
center (see Appendix
2),53 or by reducing the width of the naos by 0.067
m in order to balance the sides
3). As the calcula
evenly (see Appendix
2 and 3 demonstrate,
the results for each of
tions provided inAppendixes
theoretical

these alternative

47. Korres
48.
80),

theorists

methodological
to which
degree
(whether
whatever)

p. 79.
words
(1994, pp. 79
"refuse to be bound by the
that the
requirement
1994,

In Korres's

a theoretical

metrological,
approximates
should be no

definition

or
geometric,
to the actual

less than the de


building
the build
of accuracy with which
was constructed
in any
(which
ing itself

gree

case

such

theorists

ceiving)."
49. Schazmann
pi.

theoretical

are

plans

of con

incapable

surements

and Herzog

1932,

plane of the walls


51. 0.067/3.380

the strict tolerance

see

and others,

and Herzog
m
52. 0.067/16.032
53. This

and Herzog
1932,
this and all of the follow

Schazmann

p. 6, pi. 2. For
measurements
ing
naos, dimensions

under

Schazmann

2.
50.

remain well

of the naos

and pro
to the outside

relate
rather
m.

than

For

the

socle.

these mea

tent with
vators,

who

the result
the entire

solution
the views

Fig.
1932,

= 0.42%.

would

5, and
pi.

2.

be consis

of Temple
A's exca
as
the displacement

explain
of an
that shifted
earthquake
and
celia; see Schazmann

1932, p. 6. For the problems


Herzog
with
this theory, see n. 34, above.

574

R.

JOHN

SENSENEY

of 0.42%, and in fact produce results closer to 0% in the case of several


dimensions. The rationale for privileging the theoretical plan inAppendix
1,
not
to
most
to
is
the
rather
but
therefore,
convincing analysis,
adjust
provide
for symmetry in away that most thoroughly relates to the measurements
of
the actual plan; when 0.067 m is added to the eastern side of the stylobate,
the eastern and western
sides of the plan equal one another as well as the
side behind the southern wall of the naos.54
The theoretical plan of 18.142 x 33.280 m solves one problem but
leaves another unresolved. On the one hand, our expectation for integral
in the overall plan is satisfied, since the theoretical plan results
proportions
in a nearly perfect 6:11 form.55 On the other hand, the rationale for the
naos and pronaos remains unclear. While
the distance of
placement of the
an equal
the walls of the naos from the edge of the euthynteria maintains
1:1:1 ratio on the sides and rear, the space before the antae of the pronaos
shares no integral relationship with these distances.56 Nor
in the
discern any meaningful
proportional
relationship
dimensions
of the naos and pronaos.57 As I argue below,
a
servable correspondences
pertains to process of design
arithmetical

may we

readily

length-to-width
this lack of ob

grounded not in
but rather to a
relationships
orthogonal dimensions,
procedure executed with the rule and compass. This geometry
between

geometric
is quite simple, though it requires some detail and rigor to substantiate
it. In the following
section, I demonstrate how we may recover the plan's
specific design process

through

analytic geometry.

Analysis

Geometric

To properly analyze the plan, I rely on simple calculations based on the


of Temple A, with the only adjustment being a
published measurements
centered naos, flanked on either side by equal distances of 0.380 m from
of the naos to the edges of the stylobate.58 All relevant di
in the plan are mathematically
verified and expressed
agonal relationships
in the footnotes with reference to a single quadrant of a two-dimensional
1-3 with accompanying Fig
coordinate system. In addition, Appendixes
ure 23
the
and
tolerances
that demonstrate
equations,
provide magnitudes,
the outer walls

measurements
for all three theoretical plans
proposed geometry according to
described in the prior section. Whenever
relevant, the location of features
will be given as Cartesian coordinates, inwhich the southeastern corner of
the euthynteria's outer edge is at the origin 0,0, and the extreme northwest
54. The
m

4.430

the exterior

of the naos

wall

euthynteria,
to the 4.435

side measures

southern

from

to the outer

which

56. The

face of the
edge

of the

is

essentially
equal
measurement
of the west

ern side (see Fig. 5). Adding 0.067 m to


the narrower
bate,

eastern

therefore,

ratio

side of the

produces
for all three sides.

stylo
1:1:1
nearly

of only

0.02 m

33.280 m length.

naos

to the outer

teria

is 6.797

find

from

the plan's

distance

and

the pronaos
it preserved
be

Here,

the pro

whose

tolerance

would
from

integral

relationship
rear distances

ratio

I can

between

the

to the
euthyn

teria (4.435 m) and that of the front


m), with

(6.797
ance of 2.1%.

an

implausible

overall

57. The
naos

and pronaos
the closest

edge of the euthyn


m. Of these two measure

the closest
is a 2:3

were

facade,

m. The

ments,

from

edge,

5.742

lateral

55. (18.142/6) x 11 = 33.260, a dif


ference

distance

to the
stylobate
on the northern

toler

acceptable.
58. Thus,
the exterior
long
4.435

outer
m,

and 4.435
above.

integral
of 1.9%

the distances

m;

x 22.053
ratio
is again

and each

of the euthynteria
than the present

see

Fig.

m.

is 3:7,
un

between

of the naos

walls
edge
rather

of the

dimensions
are 9.272

5. See

equal
4.368

also n. 34,

AND

IDEA

IN

VISUALITY

Figure 12. Restored theoretical plan


of Temple A with geometric under

ARCHITECTURE

HELLENISTIC

575

M,M,M,UP

I II

pinning
corner at 18.142,33.280.
In addition to consulting the calculations provided
here, readers may replicate the proposed findings using CAD software. The
results of the following analysis were verified with AutoCAD.59
A significant result of this analysis emerges from the location of a
the outer corners of the antae and
theoretical central point from which
corners
are
the outer back
of the euthynteria
and thereby
equidistant,
The
of this cir
share a theoretical circumference
12).60
pertinence
(Fig.
cumference to the design process is supported by the rational relationships
it shares with other features. The overall width of the temple shares a
3:5 ratio with the diameter of the theoretical circumfer
whole-number
ence, with a tolerance of less than 0.1%.61 If caution advises us to consider
this ratio a possibly
59. For

all magnitudes
consistency,
to the millimeter.

are rounded
60. From
plan's

long

located on the
point
central axis at [9.071,

12.101], a theoretical line to [0, 0]


measures

is the square
root of the sum of the squares of 9.071

and

12.101.

15.123

From

m, which

the same

coordinates

on

the central

corner

sures

m.

nal

15.124

corner

a theoretical

axis,

the external

result, there is an additional whole-number

fortuitous

of either

line

to

anta mea

the exter
Specifically,
anta is at
of the western

[13.749,26.483]. Through simple sub


traction,
distances
from

we

find

of 4.678

[9.071,12.101].

these
m

at

coordinates
and
The

14.382

sum of these

+ 15.124 m) finds a
figures (15.123 m
theoretical

diameter

of 30.247

see

m;

Appendix 1.
61. (30.247 m/5) x 3 = 18.148 m,
a difference
overall

plans
therefore

of only
width

a tolerance

seeAppendix

1.

6 mm

from

of 18.142

m,

of less than

the
and
0.1%;

576

JOHN

R.

X/

SENSENEY

_'

I-'

....
__

0__

<'^)

_--_

~~~10m 0

that should give pause to our skepticism: the distance from


proportion
to the plans southern edge and the overall
the theoretical circumcenter
width of the temple share a 2:3 ratio, again with a tolerance of less than
with a baseline x-x
of
0.1%.62 We may illustrate this correspondence
3 units drawn across the entire width of plan at the ordinate correspond
a line
center point of the circumference,
along with
ing to the theoretical
y-y

of 2 units drawn from the circumcenter

to the edge of the euthynteria

(Fig. 12).

Figure 13. Restored theoretical plan


of Temple A with geometric under
pinning
62. (18.142 m/3) x 2 = 12.095 m,
a difference
ordinate
tolerance

of only 6 mm
from the
at 12.101 m, and therefore

63. The

we may express this


Geometrically,
relationship through the algorithm
of two circumferences with a radius of 2 units, each centered on either
terminus

of baseline

centered

at the middle

x-x

is
(Fig. 13). The larger circumference, which
of x-x\ intersects with the smaller circumferences

outer corners of the euthynteria. Both the


exactly at the points of the
are
for
and
mathematical
significance of these intersecting points
proof
ratio
of
of
the diameters
the smaller and
revealed by the whole-number
a tolerance of less than 0.1%.63 When
larger circles, equaling 4:5 with
in relation to the overall width of the temple (the 3 units of
conceived
this final dimension
x-x),
brings the geometric
principle underlying
the architect's

system

into striking

clarity:

the 3:4:5

dimensions

of a

of less than 0.1%.


diameter

theoretical

circumference

larger
(see n. 60, above),

equals
which
has

of the

30.247

a ratio

of 5:4 with 24.202 m (the diameter cor


to the radius of 12.101 m in
responding
x-x
from the baseline
the y dimension
to either back corner of the
euthynteria

at [0, 0] and [18.142, 0]), with an error


of less than

0.1%

calculated

by the

difference divided by the magnitude:


=
(30.247 m/5) x 4 24.198 m, a dif
ference

of 4 mm

from

24.202

m.

=
(24.202 m/4) x 5 30.253 m, a differ
ence

of 6 mm

from

30.247

m.

IDEA

AND

IN

VISUALITY

HELLENISTIC

ARCHITECTURE

577

X
0

-0-__

Os__

Figure 14. Restored theoretical plan


of Temple A with geometric under
pinning

1~~~~~010m

K^)0~~~
circumscribed

Pythagorean
triangle.64 In effect, this geometric form ABC
lies at the heart of the design, with the compass centered midway
along
its hypotenuse
and the circumference
coinciding with its angles and lines

(Fig. 14).
We

should understand
as

based construction

this geometric underpinning


interdependent. Even in the Roman

and its compass


period, architects

a square, let alone aT square. Instead, the method


of
a
lines
with
the
rule
producing perpendicular
highest precision employed
and compass, with straight lines drawn through circumferential
intersec
tions in the same manner
that is revealed through this analysis of Tem
ple A.65 It has already been observed that Roman buildings such as amphi
theaters would commonly begin with a Pythagorean
triangle, and arrive at
did not work with

the final design using the compass through various stages.66 This Roman
use of the
similar manner of
Pythagorean
triangle recalls a conceptually
64. The

theoretical

circumference

diameter

of the

30.247

larger
equals
of
(see n. 60, above). The magnitudes
or
18.142 m (the plan's overall width,
baseline
24.202 m (the diameter
x-x),

to the radius of 12.101


corresponding
in the y dimension
from the baseline
x-x

to either

back

corner

of the euthyn

teria at [0,0] and [18.142,0]), and


30.247

3:4:5, with

a maximum

error

in any dimension

of less than 0.1%.

65. See Roth Conges 1996, pp. 370


372; Taylor 2003, p. 38.
66.Wilson Jones 1993, pp. 401
406,

figs.

13,15,16.

57?

JOHN

-- ------ -- ----- ------

R.

SENSENEY

--l ------ --------

Figure 15. Proposed general con


struction of a 3:4:5 Pythagorean
to circumferential

triangle

according
with
intersections,

dashes

indicating

the baseline
Ionic column bases that was familiar to Greek architects as
constructing
as the Archaic
early
period.67 The transparency of the plan of Temple A
us
to
allow
understand
how aHellenistic
architect might construct the
may
a
Pythagorean
triangle itself. The formula appears to consist of baseline of
6 units, upon the center of which a compass with a radius of 5 units is set,
and on the ends of which are set compasses with radii of 4 units. By these
could be joined to form the perpendicular
lines
means, the intersections
of the triangle's sides as well as the diagonal of its hypotenuse
(Fig. 15).
In the case of Temple A, it appears that the larger circumference
of this
to
construct
extent
in
remained
define
the
of
the
pronaos
geometric
place
at the antae (Fig. 14).
To dismiss these results would

now
us to
a confluence
require
posit
of three separate coincidences
of whole-number
(3:4:5) with
proportions
a maximum
error that is consistent with the strictest possible
standard
in the actual building, along with a fourth (and
an inte
more
that these proportions
coincidence
engender
conspicuous)
to
Greek
of
central
mathematics.
More
form
gral geometric
significance
a
the
of
over,
circumscription
triangle graphically expresses
Pythagorean
bisectors meet at a circumcenter
Tha?es' theorem: three perpendicular
of tolerance

observable

located on the hypotenuse, which runs the length of the circle's diameter
that the Pythagorean
triangle
(Fig. 16).68 In turn, the basic proportions
center point and the di
of
establish
location
the
theoretical
the
yields
ameters

(Fig. 15). In the face of these internal


to Euclidian geometry, the balance
their
and
pertinence
correspondences
on the side of in
form
this
concerning
obviously falls heavily
resulting
of the circumferences

design rather than chance.


There is yet another integral proportion that completes the geometric
of the temple's plan. The diagonal across the naos from
underpinning

67. For
and

column

pp.

126-129.

a
68. By definition,
Pythagorean
see Eue.
is a
triangle
right triangle;
Elem.
3.31.

69.Width

tentional

a 1:1
correspondence
including its external walls shares
a
difference of only 0.1%.69 From
with the total width of the temple, with
a theoretical central point located on the cross-axes of the naos, therefore,
corner to corner

the distance to either edge of the temple's width and each of the external
a circum
corners of the naos is essentially
equal. This congruency suggests
to the design of the naos, whose diameter shares a
ferential underpinning

the Pythagorean
triangle
see Gruben
bases,
1963,

including
15.572 m,

and length of naos

its walls

equal

respectively.

9.272

According

and
to

theorem,
then, we
Pythagorean
and
find
the
each
square
square root

the

of their

sum,

thus finding

18.123

m.

Ifwe take the 0.019 m difference


between 18.123 and 18.142 m (the
total adjustedwidth of the temple) and
divide by either 18.123 or 18.142 m, we
find

a difference

of 0.1%.

IDEA

AND

IN

VISUALITY

HELLENISTIC

010

ARCHITECTURE

579

---=- --S---- ----

'-0^~~

:-

-~

-c
=- L d

- m

"S^~~~loZZ
Figure 16. Restored theoretical plan
of Temple A with geometric under

10~~~~~~~~~~~~~~~~~~~1

0 0t

pinning
the diameter of the large circle, with a toler
we
as a
If
accept these circumferences
(Fig. 17).70
guiding
method for the placement of features within the plan, their ratio would call
tomind Vitruvius s formula of 3:5 circumferences for the main proportions

whole-number

3:5 ratio with

ance of 0.1%

of plans in peripteral round temples (Vitr. 4.8.2).


One indication that the circumferences
here
geometric

underpinning
the ordinates

suggest

an intentional

is their planar interrelationship.


The distance
of their theoretical center points is 0.115 m.71 In

separating
a scale
are
plan, special markings
required to make this separation percep
tible (Fig. 18). Before considering why the architect might have centered
his compass at different points a hair swidth apart in his design, we might
consider

this separation in relation to the theoretical proportions


and ac
to which
it corresponds: the separation of these ordinates

tual dimensions
70. (30.247 m/5) x 3 = 18.148 m,
a difference of 0.1% from 18.123 m;
=
(18.123 m/3) x 5 30.205 m, a differ
ence

of 0.1%
71. The

cal center

from

30.247

coordinates
point

of the

m.
of the theoreti

larger

circle

are

[9.071,12.101]
the

smaller

(see n. 60,

above).

For

of the
circle, the coordinates
are [9.071,12.216].
center
The
point
sum
ordinate
here is determined
by the
of the center point of the naos and the
distance
of the naos from the south

outer

(15.572
edge of the euthynteria:
+ 4.43 m. For the distance
sepa
center
of
rating the theoretical
points
the larger and smaller circles: 12.216
= 0.115 m.
12.101

m/2)

580

R.

JOHN

--

SENSENEY

--

-0

--------\
---- - ---_

^=

Y-.
BiaS~~iiiiff~~ia1Q

lo

Figure 17. Restored theoretical plan


of Temple A with geometric under
pinning

represents a 0.38% difference, sowe remain within the strictest standard for
theoretical tolerances of 0.42% calculated according to the constructional
inexactitude found in the actual building.72
On

the other hand, the applicability of this standard here is dubious.


we cannot
the precise metrological
sys
Although
conclusively determine
tem
the
remains
fact
that
the architect or builders
underlying Temple A,
would have needed to convert any conceptual circumferential
geometry
to actual measurements
for orthogonal distances. After all, we cannot ex
pect masons to have laid out the building according to invisible circles with
an eye to
a shared theoretical center
maintaining
point. Due to such nec
in
the
essary adjustments
planning and building process, it is natural that
from original design elements are bound to occur. Since we
lack secure access to this intermediary
the
stage of metric
specification,
relevance of a precise calculation for the percentage of error in a common

deviations

center point (such as 0.38%) may be limited. Instead, we may conceive of


the divergence inmore experiential terms: in a building over 33 m
we
long,
find the two theoretical circumcenters
of the integrally proportioned
cir
cumferences
child's hand

at points only 0.115 m apart, or less than the


a
length of small
in relation to the distance from floor to vaults in the cathedral

72. As
relating
the widths
tion
bate),

of error

in the calculation
to the difference

of the ptera
to the entire width
the difference

of 0.67 m
(0.42%
of the

is here

in

in rela

stylo
calculated

to the
geometry,
according
complete
as
of
represented
by the diameter
the larger theoretical
circumference:
m = 0.38%.
0.115/30.247

IDEA

AND

IN

VISUALITY

?-DII

ARCHITECTURE

HELLENISTIC

zzzzzz

~-Q-)

--Q" /

EEEEz

\>QZ

581

oizzzzzzziq-;
Figure 18. Restored theoretical plan
of Temple A with geometric under
and
pinning
circumcenters

indicators

marking

iii

the

fc\7)

i^~^ i

10m
0

in Paris. In a structure where even the width of the stylo


of Notre Dame
bate is off by 6.7 cm, an additional inexactitude of 4.8 cm for an invisible
feature is insignificant, particularly when that feature was no longer relevant
during the actual building process.

The

Schematic

Plan

affords, I will
justification
Leaving aside the security that mathematical
now suggest
to
in
the
Hellenistic
which
this
relates
geometry
possible ways
the theoretical demon
architect s process of designing Temple A. Unlike
stration above, the following analysis takes into account the proportions
intention here is to explore further questions
of the actual building. My
to
the
design process, integrating what I hope is well-grounded
relating
the results of the above geometric analysis.
with
speculation
In designing Temple A, the architect would have needed to harmo
nize the 3:4:5 triangle underlying
the placement of the naos and pronaos
inmind how a compass is
with the 6:11 ratio of the overall plan. Keeping
that the simplest way of working with
centered, it isworth emphasizing
the tool is to conceptualize

circumferences

in terms of radii rather than

582

JOHN
'

R.

'
'
?'i K')l I?I tV
i

i ;

i ;

i ; r

SENSENEY

i ; i

i,

I?I
?

In this way, one need not resort to half-number


divisors, such
as 2.5, in order to create a whole-number
diameter such as 5 units. In
radii of 3, 4, and 5 on a baseline of 6, therefore, the architect
producing
would have created a 6:8:10-unit
this same divisor
triangle. Extending
to the overall plan, an additional 3 units in the y dimension
produces the
final 6:11 ratio of the temples plan, which repeats the 6 x 11 number of
columns for the intended colonnades and simplifies the process of
drawing
diameters.

by maintaining
integers.
This demonstration

of the architect

to circumferences

Figure 19. Restored theoretical plan


of Temple A overlaid with intersec
tions

of circumferences

with

3 units

s method

of locating the wall


still does not explain the rationale

termini according
behind where they were placed along those circumferences.
It is tempting
to suggest a
circumference-based
the architect
simple
algorithm whereby
out
x-x
have
worked
these placements. On the baseline
of 6 units,
might
center the compass on the termini and center,
drawing
radius. Repeat this procedure three times, each with

three circles of equal


radii of 3, 4, and 5
according to the cir

units, finding the location of the walls and corners


cumferential intersections (Figs. 19-21). Despite
the appeal of the resulting
to
it
would
be
inadvisable
this
plans, however,
procedure. As Korres
adopt
we cannot draw conclusions
on the basis
recognizes,
concerning geometry
of how that geometry appears to coincide with features when overlaid on a
scale plan.73 Rather, we must replicate such results mathematically.
Unlike

73. Korres

1994,

pp. 79-80.

radii

of

IDEA

AND

-NIo--

C-:o

/
/111.~-^

VISUALITY

IN

HELLENISTIC

_1_1...

~ I Ii I

..'. ..

I. . I

I
?

0^J)~~

tions

of circumferences

with

radii

of

4 units

583

\s

0 \

Figure 20. Restored theoretical plan


of Temple A overlaid with intersec

ARCHITECTURE

~lo10m

the case of the underlying geometry demonstrated


above, calculations do
away that satisfies the strictest
not verify the hypothesis
in
here
suggested
possible tolerance of 0.42% in the actual building.74
Still, even in cases where proposals hold up to such scrutiny, one
deserves recognition. There is, of course, a gap between our
of verifying the plan through analytic geometry and the ancient
of converting
the location of its features into magnitudes
for

consideration
method
method
74.

In the case of the 3-unit

radii

that

the naos

the cross-axis

from

are set 9.062

at [9.071,12.216].

we

corner

at [4.435,4.430]
dimensions
of 4.435

andjy
to calculate

of

diagonal
a difference
of 1.1%
showing
from the
9.062 m.
expected
In the case of the 4-unit
radii
8.961

m,

we may
reference
the line of
(Fig. 20),
either
wall.
That
of the
long pronaos
western
intersects with
wall
the central

of 4.636

erance

of the plan to the external


and the 12.101 m ra
a distance

in
resulting
the baseline

of 11.178

to the

intersection
in
= 4.6362
+
dimension
(12.1012
they
y2).
intersection
The western
wall's
with
the
lateral

as

23.360],
of 4.435

occurs

circumference

given
from

by
the outer

and

the 12.101

euthynteria
resulting
the baseline

at [13.707,

the wall's

in a distance
to the

distance

intersection

in the y dimension

is a tol

of 0.7%.

In the case of the 5-unit

radii

of the central
(Fig. 21), the intersection
anta
circle and the western
exterior
corner

is at [13.749,26.483],
of 15.124 m from

in a radius

(see n. 60, above)

12.101]

resulting

[9.071,
and a dis

tance of 14.382 m from [13.749,


If the circle with

12.101].
m

a radius

of

15.124

12.101], itwill intersectwith the line of

from

in the

theoretical

the anta

is centered

at the end of the

of the

edge
m radius,

of 11.259

of 0.081

from

of the naos

from

find

and 7.786

distance

given by
the midpoint

dius,

corner

sidered here: from [0,12.216] to the


naos

the distance

wall

the plan is symmetrical,


only
to be con
of the naos needs

Because
one

corners

as

at [13.707,23.279],

circumference

(Fig. 19), it has already been established

baseline

x-x

at [13.749,26.573],

at [18.142,
a differ

=
y dimension (12.1012 4.4352 + y2).

ence of 0.09 m from [13.749,26.483],

The

or an error of 0.6%.

difference

of these

intersections

584

JOHN

R.

SENSENEY

--r

rI

10om

the actual building. We might,


scale
therefore, ask how an architectural
in
would
have
been
created
the
Hellenistic
This
drawing
question
period.
is especially relevant to the planning of Doric
temples, where interstitial
contraction precluded convenient
slabs that ensure conformity to a grid-based
A reasonable answer in the case of Temple
intuitive process that begins with the initial

columnar

repetition

of uniform

paving

plan.
A, I suggest, lies in a simple
schematic sketch before the

the smaller
of the detailed drawing (see Fig. 22): (1) within
completion
rear
naos
at
set
exterior
of
the
the
lines
of
the
walls
the
and sides
circle,
with approximately equal distances to the outside edges of the overall plan
in accordance with the principle of symmetry; (2) where the lateral lines
same circle, set the spur walls
again intersect with the circumference of this
naos
and pronaos; (3) in conjunction with these same lateral
separating the
lines, set the antae at the intersection with the circumference of the larger
circle. In the drawing process itself, this result is most easily achieved in
away that is similar to what I describe above: first set the locations of the
s
corners and the antae
by establishing equal distances from the plan edges,
and then mark these points with the compass set on the termini and center
of the baseline x-x. In these ways, the logic of the overall design maintains
that are circumferential, which is in keep
symmetry with interrelationships
a process of
relies upon the rule and compass.
with
that
ing
drawing

Figure 21. Restored theoretical plan


of Temple A overlaid with intersec
tions

of circumferences

5 units

with

radii

of

IDEA

AND

IN

VISUALITY

HELLENISTIC

I- - -

I \

ARCHITECTURE

--1

585

--

Figure 22. Proposed geometric


underpinning of Temple A
If we

of a modular-based
again consider the hypothesis
metrology,
we can
on one manner
s
in
which
the.
speculate
plan designer might have
established
scale. Since the placement of features depends upon circum
considerations, while the production of elements such as paving
slabs must be related to orthogonal dimensions,
itwould appear that the
a fixed
in
the
the
scaling
drawing precedes
following way. In privileging
as 45 Doric feet for the colonnade axes of the front and
such
magnitude
ferential

the remaining elements in the drawing


rear, the architect could measure
and placements
of the walls and the varying
(such as the dimensions
of the individual slabs that make up the stylobate and steps)
dimensions
against these established distances and fix their sizes according to scale.
By its nature, this procedure would be inexact for two reasons. In the first
place, the expectation of symmetry in the final built form would dictate
naos walls to the
equal values for the distances from the exterior
edge of the
at
in
both
the
sides
when
and
fact
the
of
the drawn
rear,
geometry
stylobate
form would

show a very slight discrepancy between the lateral and rear


indeed, the separation of 0.38% in the centers of the theoretical

distances;
circumferences
75. Two
tect could

in which

ways
resolve

this

the archi

issue would

be

for the
(1) to center the compass
at a
smaller circumference
slightly dif
ferent location
discussion
(see
above),

a result of this very consideration.75


(Fig. 18) is likely to be
to measure
the
would
need
the features on the
Secondly,
plans designer
the
drawing surface by hand and convert them to varying values. Unlike
case with Ionic
a
the
varied
in
of
columns
Doric
temples,
spacing
temple
such as that at Kos dictated
that individual slabs could not repeat an
therefore, would need to be subdivided
prototype. Distances,
into varying units for the paving slabs in accordance with
the spatial

or (2) to
a
verbally
designate
larger
numerical
distance
for the area behind

established

the naos,

contractions.

and

subtract

this distance

the length of the walls


latter solution
pronaos. The
from

more

practical

especially
of verbal

of the
seems

both

and more

considering

probable,
Greek
traditions
in

specification

"incomplete
as discussed
preliminary
planning,"
by
Coulton
In either case the ad
(1985).
is very small, both
in relation
justment
to the
in
of 0.42%
tolerance
expected
the final

built

form

cal distance

it would

the original

scale

and

in the theoreti

correspond

drawing.

to in

dress

In the end, therefore, the measurements


would have needed to ad
the individual paving slabs in addition to the overall size of the

or
steps in this process,
stylobate
euthynteria. Because of the multiple
and the slight modifications
bound to occur in each of these steps, it is
not reasonable for us to theorize intended values for each element and
of the plan, given asmeasurements
down to the dactyl. Instead,
the significant result of this study remains the revealed correspondence
of the overall form to a rational, theoretical geometry
in which
the per
error
of
remain
within
the
strictest
centages
possible tolerance found in

dimension

the

actual

construction.

586

JOHN

R.

SENSENEY

CONCLUSIONS
at Kos suggests
A metrological
analysis of Temple A in the Asklepieion
that the triglyph module theory proposed byWilson
Jones for 5th-century
Doric temples may be applicable to this Hellenistic
example. This theory
cannot, however, account for the locations of features not associated with
the temples trabeation, such as the walls of the naos and pronaos. Since
was created in an era when the kind of drawn
plan described
Temple A
in Ionic temples,
is likely to have been already commonplace
by Vitruvius
we are
in
its
how
address
the
considerations
of
asking
justified
plan might
design particular to the Doric order, where transparent orthogonal relation
a
were not
grid
possible. A geometric analysis that
ships established with
to the methodological
issues
addressed by Korres demonstrates
responds
a
that
circumscribed Pythagorean
triangle forms the basis of Temple As
the placements
of the plans
of
the
problem
verifying
principal
rests
for
this
modular
the
evidence
system
geometric
solely upon
theory,
the internal, measurable correspondences
that conform to Euclidian norms.
Furthermore, we can replicate these results both by calculation and with
design,

in which

determine

circumferences

features. Unlike

the more

difficult

with the modular theory, we can speculate


axes
scale in the
that the colonnade
may have played a role in establishing
into
drawn plan, by allowing for the conversion of relative dimensions

CAD

software.

In combination

actual values for the building. The full implications of the results of this
a few observations
analysis cannot be explored in the present study, but
merit brief comment.76
the design process proposed here runs counter to the
cur
as well as
differing from
simple grid approach used in Ionic temples,
rent ideas about the way in which Doric temples were designed. Wilson
In its details,

Jones insists on the principle of "facade-driven" design for Doric temples,


in contrast to the "plan-driven" design for Ionic temples.77 In other words,
architects designed Doric temples strictly according to the commensuration
of elements in the facade, as opposed to the creation of a guiding plan that
the layouts of Ionic temples. Yet given the mixing of the archi
determined
tectural

orders

as

early

as

the

5th

century

b.c.?most

famously

witnessed

might question such categorical notions of mutual


exclusiveness, particularly in buildings as late as the Hellenistic
period. As
discussed above, it appears that the triglyph module may very well have
in the Parthenon?we

a
significant role in the design of Temple As facade. One might
played
wonder, however, why ancient architects who are likely to have been trained
in the details of both orders should necessarily have repressed planning
of a particular module. After all,
tendencies solely due to the employment
the very notion of mutu
demonstrated
has
Rowland
convincingly
Ingrid
an
to
of Vitruvius's
modern
transformation
be
"orders"
early
ally exclusive
notable
accommodate
like ancient buildings
themselves,
genera, which,
a discussion
omit
should
Vitruvius
That
of
interchangeability.78
degrees
of taxis in relation
traditions

to Doric

temples probably reflects his bias toward the


that formed the core of his architectural train

of Ionic design
s
ignorance

ing.79 If Vitruvius

of Doric

taxis stemmed

from

this limited

76.

In an article

in prog
currently
the results of the present
other considerations
analysis
along with
of ancient Greek
in the larger context
I assess

ress,

architectural

77.Wilson
2001.
78. The
rigidly

Jones 2000b, pp. 64-65;

notion

of the "orders"

defined

begin with
the milieu
continuing

tools,

drawing, masonry
of planning.

and methods

categories
Renaissance

of Raphael
later with

appears
thinkers

as
to
in

and Bramante,
Serlio,

Palladio,

and Vignola; see Rowland 1994; Howe


and Rowland

1999,

79. See Tomlinson

p. 15.
1963.

IDEA

AND

background,
extending
With

IN

VISUALITY

HELLENISTIC

ARCHITECTURE

587

there is no reason why we should perpetuate his ignorance by


to Hellenistic
it retrospectively
architects and their buildings.
support from the results of this analysis, it is even worth specu

a
on the
higher degree of
special potential of the Doric order for
in
in the case of
At
the
least
design process.
drawing-board
sophistication
a
in
in
columnar placements
Doric temple might
Temple A, the variations
lating

an alternative
approach to the location of the internal
the plan.80What
appears to have resulted was a system
more interesting than the
characteristic
arithmetical
simple
relationships
one
was
too
innovative
that
for reuse and
of the grid plan, but also
perhaps
have motivated

features within

partly for this reason, and partly because


of Doric temples altogether, the possibly
in Temple A may have disappeared
from

continued development.
Perhaps
of the "decline" in the production
Doric-related

method

found

common
practice well before Vitruvius picked up his pen. Yet Temple A
was not the final instance of this
approach, which appears to have extended
even
and
into aHellenistic-Roman
the
Doric
order
context, where
beyond
to
demonstrate
the application of the Pythagorean
temple plans continued
as their
guiding geometry.81 Ultimately,
triangle and 3:5 circumferences
of
form
of Temple A might have its
the
characteristic
however,
geometry
not
most recognizable
in
the
taxis of the architect's
legacy
cryptomethodic
in
the
caementicium
but
that
Roman
drawing board,
shapes
opus
finally al
lowed for permanent expression in three dimensions. Framed in this way,
the fully experiential
rise to a new aesthetic

80. In Temple
interaxes

A, the lateral corner


ca. 2.7 m,
measure

as
to the other average
opposed
axes of ca. 3.05 m. In the facade
rear colonnades,
interaxes measure

the corner

and central

sure 3.065

and 3.080

pi.

2.

and

column

ca. 2.7 m, while

the second

see Schazmann

inter

interaxes
m,

and Herzog

81. These
found

geometric
approaches
in two of the earliest
hellenizing

respectively;
1932,

have been unimaginable

are

temples in Italy during the Republican


period: theTemple of Juno atGabii of
ca. 160

b.c.

ca. 120-100
mea

that would

and

the round

b.c.

in Rome's

temple
Forum

of
Boa

rium. For
of Juno

of the
the geometry
Temple
see
at Gabii,
Almagro-Gorbea

1982; Jim?nez 1982, esp. pp. 63-74;


Almagro-Gorbea

give
in the Hellenistic

theory of Vitruvius.

architectural

column

of the idea and its reflection would

intersection

and Jim?nez

1982;

Coarelli

1987, pp. 11-21. For the round


see Rakob
1973.
and Heilmeyer
temple,
An
elaborated
of
such
geom
analysis
etry, its significance,
tions between
these

and

the connec

and the
examples
at Kos discussed
in the present
work
are themes
in a
that I explore
study
article
focused
(in
follow-up
progress)
on Roman

architecture.

APPENDIX

THEORETICAL

PLAN A

Select

and equations for theoretical


locations, coordinates, magnitudes,
are given below. The coordinates
A
in
correspond to measurements
plan
meters taken by Schazmann
to
and Herzog
converted
here
(see Fig. 5),82
an 18.142 x 33.280
0 and limit 18.142, 33.280 at
with
0,
origin
quadrant
the southeastern
and northwestern
extremes, respectively
(Fig. 23). For
additional equations, see text and notes above.
Location

Relevant

Coordinates

Circumference
1 A
1
B
1
C

0,0
18.142,0
24.202
D 2 4.435,4.430
E 2 13.707,4.430
F 2 13.707,20.002
G 2 4.435,20.002
H 1 13.749,26.483
I 1 4.393,26.483

Circumference
(Circumcenter

1
9.071,12.101)

Definitions
= Radius 1 + Radius 2
AC = Diameter
=
to A
Radius 1 distance from circumcenter
to H
Radius 2 = distance from circumcenter

Magnitudes
X, Y distances from circumcenter
= 9.071
XI: 9.071-0
= 12.101
Yl: 12.101-0
X, Y distances from circumcenter
9.071 = 4.678
X2:13.749
= 14.382
Y2:26.483-12.101

(or B)
(or I)

to A

to H
82. Schazmann
pi. 2.

and Herzog

1932,

AND

IDEA

IN

VISUALITY

HELLENISTIC

ARCHITECTURE

/ ^ID

EIL

F_

,-_

- 1^
L

__

_~

O-

Figure 23. Restored theoretical plan


of Temple A with geometric under
pinning
responding

and

indicated
to Cartesian

locations

cor
~10m
lomo

0WJ~

coordinates

Equations
Radiusl2
Radius

=Xl2

inMagnitudes
Differences
= 0.042
30.247-30.205
= 0.025
18.148-18.123

+Yl2

Radius

12=9.0712+
1 = 15.123

Radius

22 =X22 +Y22

0.042

/ 30.247

Radius

22=4.6782+14.3822
2 = 15.124

0.025

/ 18.123

12.1012

Tolerances

Radius
AC

= 15.123

+ 15.124

= 30.247

Pythagorean
Equations

Circumference
(Circumcenter

hypotenuse2
hypotenuse2

9.071,12.216)

hypotenuse

Diameter2 = (DE)2 + (EF)2


Diameter2
Diameter
6:10 Ratio
Equations

= 9.2722+
= 18.123

=
(18.123 / 6) x 10 30.205
=
(30.247 / 10) x 6 18.148

Triangle
=AB2 +
(Yl x 2)2
= 18.1422 + 24.2022
= 30.247

inMagnitude
Difference
AC
hypotenuse
30.247
30.247 = 0

15.5722

of Circumferences

= 0.1%
= 0.1%

1 and 2
Tolerance

0 / 30.247 = 0%

589

APPENDIX

THEORETICAL

PLAN B

Select

and equations for theoretical


locations, coordinates, magnitudes,
are
B
below.
The
in
coordinates
given
plan
correspond to measurements
meters taken by Schazmann and Herzog
(see Fig. 5),83 converted here to an
at the
18.075 x 33.280 quadrant with origin 0,0 and limit 18.075,33.280
a
southeastern and northwestern
and
extremes, respectively,
symmetrically
and
centered naos (see Fig. 23, scaled for the slightly differing dimensions
coordinates of Appendix
1).
Location

Relevant

Coordinates

Circumference
1 A
1
B

0,0
18.075,0
C 1 18.075,24.224
D 2 4.402,4.430
E 2 13.674,4.430
F 2 13.674,20.002
G 2 4.402,20.002
H 1 13.716,26.483
I 1 4.360,26.483

Circumference 1
9.038,12.112)
(Circumcenter
Definitions
= Radius 1 + Radius 2
AC = Diameter
=
to A
Radius 1 distance from circumcenter
to H
Radius 2 = distance from circumcenter

Magnitudes
X, Y distances from circumcenter
0 = 9.038
XI: 9.038
= 12.112
Yl: 12.112-0
X, Y distances from circumcenter
9.038 = 4.678
X2:13.716
Y2: 26.483
12.112 = 14.371

(or B)
(or I)

to A

to H
83. Schazmann
pi.

2.

and Herzog

1932,

IDEA

AND

Equations
Radiusl2
Radius

IN

VISUALITY

=Xl2

12=9.0382+12.1122
1 = 15.112

Radius

22 =X22 +Y22

Radius

22=4.6782+14.3712
2 = 15.113

AC

= 15.112

+ 15.113

ARCHITECTURE

inMagnitudes
Differences
30.225
30.205 = 0.020
= 0.012
18.135 -18.123

+Yl2

Radius

Radius

HELLENISTIC

Tolerances

= 30.225

0.020

/ 30.225

< 0.1%

0.012

/ 18.123

< 0.1%

Pythagorean
Equations

Circumference
(Circumcenter

hypotenuse2
hypotenuse2

9.038,12.216)

hypotenuse

Diameter2 = (DE)2+ (EF)2


Diameter2
Diameter
6:10 Ratio
Equations

= 9.2722+
= 18.123

15.5722

of Circumferences

=
(18.123 / 6) x 10 30.205
(30.225 / 10) x 6 = 18.135

Triangle
= AB2 +
(Yl x 2)2
= 18.0752+ 24.2242
= 30.224

inMagnitude
Difference
AC
hypotenuse
30.225
30.224 = 0.001
1 and 2
Tolerance

0.001

/ 30.224

< 0.1%

59I

APPENDIX

THEORETICAL

PLAN C

Select

and equations for theoretical


locations, coordinates, magnitudes,
are
in
C
The
coordinates
below.
plan
given
correspond to measurements
meters taken by Schazmann
to
here
and Herzog
converted
(see Fig. 5),84
an 18.075 x 33.280
0 and limit 18.075, 33.280 at
with
0,
origin
quadrant
the southeastern and northwestern
extremes, respectively, with the width of
the naos reduced

.067 m in order to provide symmetry (see Fig. 23, scaled


dimensions
and coordinates of Appendix
1).

for the slightly differing


Location

Relevant

Coordinates

Circumference
1 A
1
B
C
D
E
F

0,0
18.075,0
1 18.075,24.224
2 4.435,4.430
2 13.640,4.430
2 13.640,20.002

G 2 4.435,20.002
H 1 13.682,26.483
I 1 4.393,26.483
Circumference 1
(Circumcenter
9.038,12.107)
Definitions
= Radius 1 + Radius 2
AC = Diameter
=
to A
Radius 1 distance from circumcenter
to H
Radius 2 = distance from circumcenter

Magnitudes
X, Y distances from circumcenter
= 9.038
XI: 9.038-0
= 12.107
Yl: 12.107-0
X, Y distances from circumcenter
9.038 = 4.644
X2:13.682
12.107 = 14.376
Y2: 26.483

(or B)
(or I)

to A

to H
84. Schazmann
pi. 2.

and Herzog

1932,

IDEA

AND

Equations
Radiusl2=Xl2
Radius

IN

VISUALITY

12=9.0382+12.1072
1 = 15.108

Radius

22 =X22 +Y22

Radius

22=4.6442+14.3762
2 = 15.108

AC

= 15.108

+ 15.108

ARCHITECTURE

Differences inMagnitudes
30.216
30.148 = 0.068
= 0.041
18.130-18.089

+Yl2

Radius

Radius

HELLENISTIC

Tolerances

= 30.216

0.068

/ 30.216

0.041

/ 18.089

Pythagorean
Equations

Circumference
(Circumcenter

hypotenuse2
hypotenuse2

9.038,12.216)

hypotenuse

Diameter2 = (DE)2+ (EF)2


Diameter2
Diameter
6:10 Ratio
Equations

= 9.2052+
= 18.089

15.5722

of Circumferences

(18.089 / 6) x 10 = 30.148
(30.216 / 10) x 6 = 18.130

= 0.2%
= 0.2%

Triangle
= AB2 +
(Yl x 2)2
= 18.0752+ 24.2142
= 30.216

inMagnitude
Difference
AC
hypotenuse
30.216
30.216 = 0
1 and 2
Tolerance

0/30.216

= 0%

593

594

SENSENEY

R.

JOHN

REFERENCES
1982. "Levanta
M.
Almagro-Gorbea,
miento
del templo,"
fotogram?trico
in El Santuario
de Juno en Gabii

(BABesch
man
and

(Biblioteca It?lica 17), ed.M. Al


magro-Gorbea,
Almagro-Gorbea,
nez. 1982.
ci?n,

pp.

pp. 33-38.
and
M.,
J. L.Jim?
modula
"Metrologia,

pp.

Parthenon

Arendt,

H.

1958.

les soci?t?s

du

Bryson,

Culture

Seattle,
F

2),

ed. H.

et

in

Under

opment of the Greek Stoa, Oxford.


-. 1983. "Greek Architects
and
in
of Design,"
et soci?t?: De l'archa?sme

the Transmission

romaine.
grec ? lafin de la R?publique
Actes du
Rome,
international,
colloque

2-4 d?cembre1980 (C?FR 66),


Rome,
pp. 453-468.
-. 1985.
Preliminary
"Incomplete
in Greek Architecture:
Planning
Evidence,"

in Bomme

laer 1985, pp. 103-121.


-. 1988. Ancient
Greek Architects
at Work,

2nd

ed.,

Vitruvius'

De Architectural

in

non

ingratum. Proceedings
on
the International
Symposium
of
and the
"De architecture'
Vitruvius'

Hellenistic

and

Republican

di
archeologia
pp. 355-368.
B., Jr. 1985. "Preliminary

in Bommelaer

1985,

135

pp.

AM

Mass-System,"
167-187.

1972. Architectural

C. A.

Space
Mass.

15,

inAncient

Greece,

Cambridge,

Dwyer, E. 2001. "TheUnified Plan of


of the Faun," JSAH

the House
pp. 328-343.
Eisner,
J. 2007.

Roman

Eyes: Visuality
inArt and Text,

and Subjectivity
Princeton.
Fr?zouls,

E.

in Bommelaer

pp. 213-229.
G. 1963.
Gruben,

"Das

1985,

archaische

78, pp. 78-177.


der Griechen,
Tempel

Didymaion,"/?tf
-. 1986. Die

ed., Munich.
-. 2001. Griechische

und

Tempel

Munich.

Heiligt?mer,
Haselberger,
Likenesses:

et le dessin

"Vitruve

1985.

d'architecture,"

4th

60,

L.

1997.

"Architectural
and Plans

Models

in Classical

Architecture

of

Antiquity,"

JRA 10, pp. 77-94.

Ithaca.

de Jong, J. J. 1989. "Greek Mathemat


and
ics, Hellenistic
Architecture,

Munus

del settimo

attische
pp.

Rome.
"Towards

in Atti
Italy,"
internazionale

zur anti
W.
1890. "Beitr?ge
D?rpfeld,
ken Metrologie
5: Das
?gin?isch

Doxiadis,
del Lazio

Intercolumniations,"
Stylobate
BSA 69, pp. 61-86.
-. 1976. The Architectural
Devel

New

pp. 177-180.
the Parthenon

147.

and

and

Some

the

Australian

Planning of the Propylaia byMnesi


cles,"

standing Greek Temple Design: The

Architecture

"How

W.

Dinsmoor,

du

Foster,

pp. 87-114.
1987.1
santuari

repubblicana,
Coulton,
J. J. 1974.

"How

Planned,"

'dassica, Rome,

in Contem

(Discussions

Visuality

Coarelli,
et ?

in Vision

Field,"

Expanded

porary

congresso

8), Strasbourg.
Antiques
in the
1988. "The Gaze

N.

1923a.

47:6,

1923b.

Greece,

antiques.
26-28

de
Strasbourg,
colloque
1984
du Centre
(Travaux
janvier
sur le Proche-Orient
Recherche
la Gr?ce

B.

Temple Design: Asia Minor,

Friedrich,
Cambridge,
pp. 81-112.
ed. 1985- Le dessin
Bommelaer,J.-F,
Mass.,

Actes

Salamis,"

Australian
Architec
Planned,"
ture 48:1, pp. 241-244.
-. 1961. "The Basis of Greek

Au

thority?"m Authority (Nomos 1),

dans

"AMet

Was

"What Was

ed. C.J.

d'architecture

Leiden,

from

Was

Architecture
-.

Rome,

Almagro-Gorbea,
87-124.

Geert

1.1990.

Relief

W.

Dinsmoor,

enGabii (Biblioteca It?lica 17),


ed. M.

Suppl.
J. J. de Jong,
100-113.

rological

1987

January
ed. H.

AJA 94, pp. 445-451.

del
y reconstrucci?n
de Juno
Santuario

in El

2),

Dekoulakou-Sideris,

Rome,

trazado,

templo,"

20-23

ture, Leiden,

Architec

Howe,T.

N.,

Vitruvius:

and

I. D.

Rowland.

The Ten Books

ture, trans, with

1999.

on Architec

commentary,

New

York.

Hurwit, J. 2005. "TheAcropolis in the


Age of Pericles" (public lecture,Art
Institute,

Chicago,

November

2005).

IDEA

P. 1989.

Isler, H.

"Vitruvs

erhaltenen

die

in

Age,

non
ingratum. Proceedings
on
International
the
of
Symposium
Vitruvius'
"De architectural'
and the
inMunus

Hellenistic

Rakob,
Der

and

Roth

pp.

in

"Arquitectura,"

It?lica

Gorbea,

1994.

inModern

Impact

ed. P. Tournikiotis,

Almagro

F,

-.

pp. 54

and J. A. K. E. De Waele.

1998.

"Le dessin du temple A dAskl?pios


? Cos,"

John R.

Pharos

6, pp. 59-68.

of
of

117
611

temple
taft

Architectural

architecture
hoyne

drive,

champaign,

senseney@uiuc.edu

buell
mc

illinois

-.
der

Sherwin-White,
An Historical
Settlement

Study from
to the
Imperial

hall

621
6182o

Urbana-Champaign

Period

51), G?ttingen.

Jones, M.

2000a.

83, pp.

and
133

1993.

"Designing

RM

100, pp. 391

"Doric Measure

and

1:The

Design
of the Relief
from

dence

Evi

Salamis,"

AJA 104, pp. 73-93.


-.

2000b.
Architecture,

Principles
of
New Haven.

2001.

"Doric Measure

Roman

and

Architectural Design 2:A Modular

An

the Dorian

"The Doric

Critics

Architectural

94, pp. 249

Study, Oxford.
1978. Ancient
S.M.

JHS

Amphitheatres,"
442.

1932.
und

Theaters:

Process,

145.
Wilson

76,

and Roman

AJA

Builders:

1963.

Hellenistic

Criticism,"

-.
Roman

(Hypomnemata

at

Illinois

"Vitruvius

Design,"
2006.

Senseney

University
school

Theater

Roman

inArchitectural

R. A.

Order:

Herzog.

Baubeschreibung

R. 2003.
Study

Tomlinson,

aspects,"

ArtB

H. Morgan,

Cambridge.

258.

97.
Petit,

and R.

R,

Sear, F. 1990.
of

Times,

Athens,

ca

des

(Kos: Ergebnisse
Baugeschichte
und
Deutschen
Ausgrabungen
I), Berlin.
Forschungen

in The Parthenon

the Parthenon,"
and Its

ed. M.

pp. 39-86.
"The Architecture

Rome,

M.

Korres,

17),

Taylor,
A

"Modalit?s

81-104.

Schazmann,

El Santuario defu?o enGabii (Bib


lioteca

Orders,"

Asklepieion:
1982.

1996.

on Ar

Vitruvius

trans. M.

York.

"Raphael, Angelo
and the Genesis
of the

Colocci,

2003.

T. G.

595

emended by S. Kellogg, New

1973.
Heilmeyer.
am Tiber
in Rom,

1994.

Architectural

pp. 3-23.
J. L.

R.

ARCHITECTURE

Smith,

pp. 299-422.

I. D.

Rowland,

in
Contemporary
ity (Discussions
Culture
2), ed. H. Foster,
Seattle,

Jim?nez,

A.

Cong?s,

M?FRA108,

of
"Scopic Regimes
in Vision and Visual

Modernity,"

HELLENISTIC

chitecture,

pratiques
d'implantation
romains: Quelques
dastres

141-153.
1988.

Cambridge.
F., and W.-D.

Rundtempel
Mainz.

Architec
Republican
1987
ture, Leiden, 20-23 fanuary
(BABesch Suppl. 2), ed. H. Geert
man
and J. J. de Jong, Leiden,

pp.
Jay, M.

IN

VISUALITY

Pollitt, J. J. 1986.Art in theHellenistic

und

Regeln

Theaterbauten,"

AND

Reading
Cos:

of the Classical

Temple,"

AJA 105, pp. 675-713.


Yeg?l, F 2001. Rev. ofWilson Jones
2000b, JSAH 60, pp. 500-504.

Vous aimerez peut-être aussi