Vous êtes sur la page 1sur 38

Petrogenesis of the Alkaline Igneous Rock Suites

of the Volcanic and Intrusive Centres of


Eastern Uganda
by B. c. KING

Department of Geology, Bedford College, University of London

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


ABSTRACT
The alkaline igneous centres of eastern Uganda comprise a younger Tertiary group re-
presented by dissected volcanoes and an older Cretaceous (?) group composed entirely of
equivalent intrusive formations.
Both the extrusive and the intrusive suites are characterized by extreme silica under-
saturation and richness in alkalis, especially soda; they largely conform to a nephelinite series
and an ijolite-carbonatite series, respectively.
From field, petrographic, and chemical data it is argued that the immediately parental
magma was of nephelinite/melteigite composition. The courses of evolution of the volcanic
and plutonic rocks followed, divergent trends, owing to differences in their physico-chemical
environments. The volcanic series evolved towards a phonolitic or trachytic residium, whereas
the ijolite series maintained silica undersaturation and tended towards final enrichmenj-in
lime, alkalis, and volatiles, including carbon dioxide. The final residuum from this course of
evolution was a carbonatite or alkali carbonatite.
The evidence suggests that the felspar-bearing rocks (trachytes, phonolites, syenites, and
nepheline syenites) have had a number of different modes of origin. Apart from evolution in
the volcanic series of crystallization differentiation, they appear to have formed by fenitization
of Basement rocks and by felspathization of felspathoidal rocks, such as ijolites. Extrusive
equivalents may then have arisen by mobilization.

OUTLINE GEOLOGY OF THE ALKALINE CENTRES


T H E alkaline igneous centres of eastern Uganda form a particularly favourable
field for genetic study since, although all show strikingly similar suite character-
istics, different aspects tend to be emphasized, or to be more readily examined,
at each individual centre. Moreover, since the centres relate to two periods of
igneous activity, with consequent variations in extent of erosion, all gradations
are represented from eroded volcanic cones to wholly intrusive complexes corre-
sponding to volcanic substructures, but from which all trace of the former
extrusive formations have been removed.
The country rocks over the entire region have been assigned to the ' Basement
Complex'. The predominant rocks are granitic and include both acid gneisses
and homogeneous granites. In south-eastern Uganda, Davies (1956) also records
diorites, tonalites, syenites, and granodiorites, while in southern Karamoja
limited areas of quartzo-felspathic and biotite granulites, amphibolites, and
pyroxene gneisses and granulites have been described by Macdonald (1961) and
Trendall (1961).
(Journal of Petrology, Vol. 6, Part 1, pp. 67-100, 1965]
68 B. C. KING—PETROGENESIS OF
The older complexes occur along a line trending NNE.-SSW., about 40 miles
in length, in the extreme south-east of Uganda (Fig. 2). They are represented
only by intrusive formations which are truncated by, and therefore older than,
the surface upon which the Tertiary volcanoes are built and are believed also to

3J-0E
TOROR•

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


2°O
N

TORORO,
SUKULU* , . ' '
Miles
0 10 20 30 40

FIG. 1. Location map of the complexes of eastern Uganda.

be younger than Karroo, chiefly by analogy with the Chilwa Series of Nyasaland.
Detailed descriptions of the centres are given by Davies (1956) and King &
Sutherland (1964).
Sukulu forms a circular group of hills, some 2\ miles across, and consists
almost entirely of carbonatites. The carbonatites show a complex pattern of con-
centric structures, related to a number of centres, extensively traversed by faults
and dykes, many of which are essentially radial in arrangement. Sovite is the
chief type of carbonatite. Magnetite, apatite, micas, tremolite, zircon, pyro-
chlore, and barytes are the commonest minerals other than carbonates. The
presence of a very narrow ring of alkali silicate rocks (including fenites) has been
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 69

MBALE

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


0"S0 N

E££3 Syenite, etc.


I F I Feniiized rocks

I | Basement granites.etc.
Miles
0 1 2 3 4
.-•'34°IOE

FIG. 2. The older alkaline complexes of south-east Uganda (after Davies, 1956).

established from pits and boreholes. Traversing the carbonatites are occasional
dykes of tinguaite and zones or veins of breccia and agglomerate.
Bukusu is about 4 miles in diameter. So far as can be judged from the evidence
of scant exposures, supplemented by numerous pits and boreholes, the complex
consists of a central mass of carbonatite which is partly separated from an
encircling ring of carbonatite by alkaline silicate rocks. Similar silicate rocks
enclose the outer carbonatites, and an extensive zone of fenitization, which is
comparatively well exposed, is developed in the invaded granites. The fenites
70 B. C. KING—PETROGENESIS OF
are characterized by the formation of a blue-green amphibole, reddish-brown
biotite and alkali felspar, and are evidently associated with the development
of arcuate shear zones in the granite. Syenitic rocks, composed almost entirely
of potash felspar, occur intermittently in the peripheral parts of the complex,
while inwards the silicate rocks are extremely variable and include pyroxenites,
biotite-pyroxenite, glimmerite, hornblendite, and ijolite. The occurrences of
peridotite and dunite are of interest as representing the only olivine-bearing
plutonic rocks recorded from any of the alkaline centres. Felspar-bearing rocks
range from nepheline syenites to malignites.
The carbonatite is a sovite in which magnetite, phlogopite, and apatite are

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


ubiquitous. Each of these minerals occurs locally in very high concentrations.
Knopite, tremolite, diopside, wollastonite, pyrochlore, baddeleyite, and fluorite
are also widely distributed.
At Sekululu, fenitized granites appear beneath an outlier of volcanics of the
Elgon series. Arcuate shear structures occur and the fenites, in consequence,
often show a characteristic foliation.
Budeda is the smallest of the complexes, being less than half a mile across, but
since it has been revealed from beneath the Elgon volcanics by the active down-
cutting of the Siroko River and its tributaries, numerous fresh outcrops are
available for study.
The central part of the complex is formed by an arcuate mass ranging in com-
position from pyroxenite or melteigite to ijolite and urtite, and enclosing two
very small areas of carbonatite. This mass, together with a number of smaller
masses of ijolite, lies within a wide area of syentic fenites developed from the
granitic Basement rocks. Cancrinite-bearing syenites occur intermittently around
the margins of the ijolite masses. A small number of dykes is found, consisting
variously of cancrinite phonolite, cancrinite syenite, and nepheline syenite.
The stages of fenitization may be closely followed. Initially, the granite
becomes recrystallized along grain boundaries and channels to a fine, granular
texture. Later, metasomatism occurs marked by the development of aegirine,
blue-green amphibole, and alkali felspar. With the disappearance of quartz
a syenite is produced, while locally nepheline is formed. Nevertheless, the mode
of origin of these rocks is readily apparent from the retention, to a late stage,
of relics of the granite felspars and the inheritance of an early shear or crush
structure as a foliation in the syenites and nepheline syenites.
Although the bodies of melteigite-ijolite were evidently intrusive and. indeed,
contain rotated xenoliths of fenite, the leucocratic rock types appear to have
been formed by nephelinization of original melteigite. Nepheline often develops
as porphyroblasts or pods, while ijolite and urtite are commonly found as
diffuse veins or patches in melteigite. The ijolitic rocks, like those of Napak
(p. 72) are extremely varied in composition and texture, but unlike those of
Napak consist almost entirely of pyroxene (ranging from about Di40He40Ac2o
to Di20He30Ac50) and nepheline.
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 71
Particularly around the margins of the masses of melteigite-ijolite, extensive
replacement by potash felspar has occurred, often in large poikilitic crystals or
prominent laths, with the formation of nepheline syenites. The adjacent fenites
are also affected and late porphyroblasts, veins and pods of alkali felspar are
widely developed. The felspathization of the ijolites is often accompanied by the
replacement of nepheline by cancrinite, with the development of cancrinite
syenites. In some rocks, cancrinite occurs in elongate prisms, quite unlike the
nepheline in habit but with the appearance of a primary mineral.
The small masses of carbonatite show transitions from the surrounding
melteigite-ijolite and enclose numerous remnants of these rocks, so that it seems

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


most probable that they have formed by replacement.
Tororo consists essentially of two masses of carbonatite surrounded by a wide
envelope of fenitized Basement granite. The larger, northern mass is about three-
quarters of a mile and the smaller one to the south about a quarter of a mile in
length. The outermost fenites are granitic, but inwards, with the elimination of
quartz, they become syenitic in composition. The new minerals that develop are
chiefly aegirine, blue amphibole, and alkali felspar—usually a streaky perthite.
Again, much of the recrystallization and new mineralization takes place initially
in veins and patches, along crush channels.
Around parts of the southern mass of carbonatite, nepheline syenite and
ijolite-melteigite occur in a belt of varying width. Much of this belt consists of
these rocks, together with syenitic fenite, as blocks in a coarse breccia. The ijo-
lites are, typically, very variable in texture and composition. The aegirine-
augite is often strongly zoned, while wollastonite and melanite are not un-
commonly found.
Closely associated with the carbonatite, but evidently preceding its emplace-
ment, the blocks in the breccia were partly felspathized, so that they consist of
rims of potash-felspar, surrounding and merging into cores of the original sili-
cate rock. By this process, ijolites have in places been converted to nepheline
syenites. The 'early' nepheline syenites may be distinguished from those formed
by felspathization of ijolite by the occurrence of an acicular, aegirine-rich
pyroxene rather than aegirine-augite in the former, and by the coarse texture
of the potash-felspar in the latter.
Banding of the carbonatite of the northern mass suggests a vertical ring
structure together, peripherally, with a tangential or radial dyke pattern
(Williams, 1952). The banding consists partly of streaks rich in magnetite. In
the southern mass, an aegirine-rich pyroxene is abundant and is similarly con-
centrated in bands. In the breccias, partial carbonation of the blocks has often
occurred, especially where they have been invaded by, or incorporated in, the
carbonatite.
Agglomerate dykes with ijolitic and ultrabasic fragments are commonly found
cutting the carbonatites, while narrow dykes of potash-rich trachyte cut the
fenitized Basement in places.
72 B. C KING—PETROGENESIS OF
The younger complexes consist of the great dissected volcanoes which are built
on an erosion surface of mid-Tertiary age and extend over a distance, from north
to south, of about 150 miles.
Napak is unique among the alkaline complexes of eastern Uganda in that
both extrusive rocks, representing remnants of the volcanic cone, and intrusive
rocks, occupying the site of the central vent, are represented (King, 1949)
(Fig. 3).
The great bulk of the volcanics is composed of agglomerates, the fragments
consisting predominantly of lavas, but also of Basement gneisses and, signi-
ficantly, of fenites and members of the ijolite series. The lavas are almost entirely

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


nephelinites and melanephelinites, formed essentially of nepheline and a pale
diopsidic pyroxene. Olivine and/or melilite occur in some varieties, black ore
is ubiquitous, while perovskite and apatite are common accessories.
The central intrusive complex, which is about 1 \ miles in diameter, is formed
largely of ijolites with a small central boss of carbonatite, and is surrounded by
a zone in which the Basement gneisses are fenitized. The ijolites are structurally,
texturally, and compositionally highly varied and complicated. A broad twofold
division was established in the field into an earlier 'ijolite complex' of melano-
cratic and mesocratic rocks, but also including an important urtite phase, and
a later, coarse-grained 'urtite complex', largely forming well-defined dyke-like
masses. Variations in composition chiefly depend on the relative proportions of
the two principal minerals, pyroxene and nepheline. Melanite, wollastonite,
sphene, and apatite are widespread, magnetite-ilmenite, perovskite and biotite
occur in some rocks, while cancrinite, pectolite, calcite, and zeolites are found
mostly as replacements.
Invariably, the earliest member of the ijolite complex is pyroxenite, which
occurs as more or less distinct, xenolithic masses. It consists largely of a nearly
colourless diopside, together with magnetite-ilmenite, perovskite, apatite, and
a very pale sphene. A bright, orange-brown biotite may occur. Melteigite de-
velops as a modification of the pyroxenite, and is associated with the intro-
duction of nepheline, although this mineral may occur only in small amounts;
simultaneously, the pyroxene becomes green coloured and more aegirine-rich
while magnetite-ilmenite and perovskite are replaced by melanite and a more
deeply coloured sphene. Surrounding and invading the xenolithic pyroxenite-
melteigite, there occurs a highly variable, mesocratic to leucocratic ijolite com-
posed of nepheline and aegirine-augite, together with melanite in widely varying
amounts but always showing a late, replacive habit. Ijolite pegmatites, of many
textural and compositional types, traverse the ijolite complex; they may be
diffuse or sharply defined, and irregular or parallel-sided. They always contain
a pyroxene which is more aegirine-rich than that of the host.
These rocks are invaded nearly everywhere by urtite-n; heline rock. In places,
the ijolite complex is elaborately net-veined by the nepheline-rock, while in
others a more diffuse invasion produces many types of leucocratic ijolite from
Z
m

O
z
m
O
OKATHUM c
KOPOPWA
1/1
on
C
H
m
(A
O
Volcanks ^ Carbonatite ^ Ijolites
fn
FIG. 3. Geological section across the Napak volcano (after King, 1949, with modifications). >
CO
ded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014

H
m
7)

C
o
>
z
o
74 B. C. KING—PETROGENESIS OF
the melanocratic and mesocratic rocks of the earlier complex. A later phase of
ijolite pegmatites then occurred.
Dyke-like bands of coarse-grained urtite transect all other members of the
ijolitic sequence. A pegmatitic wollastonite-urtite is the commonest rock type,
but melanite-wollastonite-urtites and wollastonite-pyroxene-urtites also occur.
Late stage alterations, often associated with shearing, are common; nepheline
is replaced by cancrinite, natrolite, and other zeolites, while wollastonite is con-
verted to pectolite and calcite forms indiscriminately. The place of melanite is
taken by a colourless andradite, and aegirine-augite often breaks down to
turbid alteration products which include haematite and calcite.

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


Locally, felspar-bearing rocks occur. Often they have the habit of complex
replacive veins, whereby the ijolites have been converted to nepheline syenites
or even syenite, with large plates of potash felspar and an acicular aegirine-rich
pyroxene. Cancrinite is common, often in slender prisms.
Turjaites invade the ijolites at one locality; the principal minerals are a very
pale diopsidic pyroxene, nepheline, and melilite, but each of these may be locally
dominant and each may occur as phenocrysts or as small grains in the ground-
mass; biotite is abundant in places.
The central carbonatite is largely sovite, but up to 15 per cent of magnesia
has been shown in some analyses, while siderite occurs sporadically. Magnetite
is often abundant and is the mineral by which a steep concentric flow structure
is chiefly marked. Pyrochlore and baddeleyite occur as accessories.
The Basement rocks involved in the fenitized aureole are quartz-oligoclase-
potash felspar-gneisses and granulites, in which variable amounts of biotite,
muscovite, and garnet occur. In the fenites, a characteristic shearing or fissuring
and an irregular small-scale fracturing are seen, along which a deep-green
aegirine ranging to a pale acmite, and a blue-green amphibole form feathery
rosettes and are often accompanied by a grey-green chlorite and a distinctive,
reddish biotite. As well as occupying the fissures these minerals also tend to
replace the original minerals of the gneiss, especially the quartz. Reconstitution
of the felspars and quartz also occurs, initially by granulation along channels,
but ultimately by the development of new alkali felspars throughout the rock,
with the complete elimination of quartz.
Elgon, Kadam, and Moroto are all closely similar in constitution and in none
has erosion advanced sufficiently to reveal the central plug. So closely, however,
do the volcanics of these mountains resemble those of Napak that it cannot be
doubted that central plugs of ijolite-carbonatite, comparable with that of Napak,
underlie all of them. All three volcanoes are constituted, very largely, of
agglomerates and tuffs dipping outwards at low angles, although lavas are evi-
dently more abundant in Moroto than in the other volcanoes; the fragments are
chiefly of lavas resembling those of the flows, but ijolite-bearing agglomerates
are also recorded. Among the lavas, nephelinites and melanephelinites
predominate, often bearing olivine and/or melilite. On Elgon, trachytes,
SiO PYROXENITE
TRANSITION: PYROXENITE- IJOLITE
IJOLITE
IJOLITE PEGMATITE
URTITE
NEPHELINE SYENITE
TURJAITE

AUO
2^3 Na2O
El

FeO MqO

Fe,O, CaO

It

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


PoO. TiO,
f
i
y I
i I I
25 2O IS IO 5 15 5 IO
2O WT.
FIG. 5. Diagrammatic representation of the chemical compositions of the main types of plutonic rocks from Lokupoi (Napak). The values are based on chemical analyses. The constituents have been arranged to illustrate sympathetic variations, as between
A12O3 and NasO, NasO and K2O, FeO and MgO.
Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014
SiO ANKARATRITE
MELILITE ANKARATRITE
MELANEPHELINITE
NEPHELINITE
PHONOLITE

AI2O3 Na 2 O

1 — 1
• : - • • • • • • : •-.* , : ;
• • • • . • • • . • .

K2O

FeO
MqO

FeoO, CaO

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


TiO,
P,O.

i i 1 1 1
25 2O 15 IO is 2O WT. 25

Fig. 4. Diagrammatic representation of the chemical compositions of the lavas from Mt. Elgon and Napak. The values are averages based on chemical analyses.
Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 75
trachyandesites, and phonolites are extensively represented (Davies, 1952), while
phonolites also occur at Kadam (Trendall, 1961) and phonolites, trachytes, and
trachybasalts at Moroto (Hytonen, 1959, p. 110; Macdonald, 1961). The largest
volcano, Elgon, must originally have had a basal diameter of at least 60 miles.
The volcanics are often cut by plugs and dykes of nephelinite and phonolite.
Toror is the most northerly of the group of dissected volcanoes and the vol-
canics have been entirely removed by erosion (DuBois, 1959). It consists of a
central mass of phonolitic agglomerate, intruding carbonatite which is now in
the form of a partial ring, about two miles in diameter. The carbonatite intrudes
granitic gneiss which is fenitized and brecciated towards the contact. Trachyte

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


forms a complex of dykes within the gneiss around the contact and is often also
involved in the brecciation. A later phase of trachyte forms dykes which cut
the carbonatite. The trachytes are very strongly potassic. Dykes, sheets, and
plugs of phonolite cut all formations, the dykes having a distinctly radial arrange-
ment (Hytonen, 1959, p. 82). The fenitization is slight compared with that of
the other complexes and consists chiefly in the development of biotite, amphi-
bole, and aegirine along cracks and fissures. Granulation of the quartz and
felspar occurs along narrow channels; of particular interest is the development
along some of the channels of potash-felspar which often assumes a parallel,
lath-like habit.

CHEMICAL CHARACTERISTICS OF THE ALKALINE SUITES


The associations among the alkaline igneous rocks may be summarized as:
Volcanic:1
Melanephelinite-nephelinite.
Phonolite-trachyte (including some trachybasalts).
Plutonic:
Pyroxenite-melteigite-ijolite-urtite (ijolite series)-carbonatite.
Nepheline syenite-syenite (including fenites and felspathized ijolites).
Chemical compositions of typical assemblages are represented graphically in
Figs. 4 and 5, which are based on analyses of twenty-nine lavas from Napak
and Mt. Elgon and of forty-one plutonic rocks from Lokupoi (Napak). Selected
chemical analyses and normative compositions of rocks from the complexes of
eastern Uganda are given in Tables 1 and 2.
By comparison with members of calc-alkaline suites, all assemblages in the
complexes of eastern Uganda are characterized by high values for alkalis
(especially soda) relative to silica, and high alkalis relative to alumina. Excluding
the less-abundant phonolites, trachytes, nepheline syenites, and syenites, the pre-
dominant rocks show relatively low values for silica despite wide variations in
the proportions of other constituents, and relatively high values for lime.
1
Minor intrusions may be classed with either the volcanic or plutonic in terms of petrography
and chemistry.
76 B. C. KING—PETROGENESIS OF

TABLE 1
Budeda, Toror, and Elgon

1 2 3 4 5 6 7 8 9

Syenitic fenite

Ankaraitrite
Si
Si ft) t^
&i

Phonolii

Trachyti

ankarat
Nephelii
Melteigi

Melilite
syenitic
Ijolite
B27 B7 B279 B223 T174 TI34 (3) (4) (ii)

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


Chemical Compositions
SiO 2 . 45-68 45-49 56-71 64 79 5117 54 69 4038 41 44 44 08
A12O3 3 33 11 99 16 30 17-78 1948 21 02 4-84 701 1007
Fe 2 O 3 671 708 2-36 205 1-86 6-38 8-93 3 16 885
FeO . 9-50 646 3 03 088 3-62 0 10 6-71 7-89 3-37
MgO 689 3-82 1 64 040 0-46 000 1089 26-48 796
CaO . 20-28 12-32 5-38 2 21 318 201 17 99 8-23 1431
Na 2 O 2-86 5-45 7-71 8-42 10-77 062 1 69 1 83 317
K2O 0 10 1-35 405 3 19 301 13-34 1 54 094 210
H2O+ 046 294 100 033 4-71
1 60 268 1 57 2-80
H2O- 019 061 035 0-99
CO S . Oil 1 01 030 n.d. 0 16 — — Nil 003
TiO 2 . 0-64 041 068 043 061 0-30 330 072 1 49
P2O6 2 10 016 0-54 019 005 028 075 0-26 096
MnO '. 041 035 019 009 0-22 013 036 0 18 036
BaO . — — — — 015 — — — 004
Total 99 07 98-83 99-89 10076 99 45 100-66 10067 100 06 10058

Normative Compositions
Quartz — — — 0-7 — — — — —
Orthoclase . 06 81 24-2 18-8 17 9 790 — — 12 2
Albite 60 189 35 8 709 25-8 — — — 5-8
Anorthite . 4-3 1-5 8-3 II 81
Leucite — — — — — — 70 4.4 7-2
Nepheline . 5-6 147 13 7 — 31 1 28 7-7 8-5 11-4
Na 2 SiO 3 . — — — — 04 — — — —
Acmite 66 36 — 5-4 — — — —
Diopside 62-4 31-9 161 2-2 110 — 56-2 22-8 43 0
Wollastonite 48 4-4 08 2-4 09
2CaO.SiO 2 — — — — 69 1-2
Olivine — — — 08 — 14 46-5 —
Magnetite . 64 102 16 19 — — 13 0 4.4 7-7
Haematite . 08 6-4 — 3-5
llmenite 1-3 0-8 14 0-8 12 0-5 6-2 14 29
Apatite 50 03 1-3 03 01 07 1-7 0-7 2-4
Calcite 02 2-3 07 0-4
Water — — — — — 16 — 19 38
Total 989 95 9 99-2 100-3 94 1 99-3 101-2 99-9 100-8

Analysts: 1-5, D . S. Sutherland; 6, J. M. Bartle; 7, N. Sahlbom; 8, W. H. Herdsman; 9, B. C. King.


ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 77

TABLE 2
Napak

1 2 3 4 5 6 7 8 9

WoUastonite
Meso-ijolite

nephelinite
Pyroxenite

Melteigite

Nepheline

Melilite
!

syenite
Turjite
.a

Ijolite

unite
N35C N520 N165 N114 N32 N170 N106 K41 K147

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


Chemical Compositions
SiO 2 . 41-39 3992 3959 37-59 45-59 3566 54-38 41-51 3835
A12O3 4-91 8-53 1363 1636 18 20 23-54 17 02 1266 13-29
Fe 2 O 3 1037 9-59 864 417 1 81 117 4-67 7-72 7-26
FeO . 9-77 6-75 389 395 1 98 517 2-57 632 6-22
MgO. 1006 993 3-57 291 050 501 109 649 5 99
CaO . 1802 16 42 19 94 21-95 1803 17-84 3-79 11-91 15-34
Na 2 O 1 52 3-68 466 580 965 5 65 5-55 463 3-72
K2O . 1 40 1 48 380 2-57 4-29 1 70 9-54 165 321
H2O+ 3 51 3 31
— 021 — 020 — 0-38 —
H 2 O" 041 0-34
CO 2 . — — 025 — — — — 010 008
TiO2 . 2-55 3 21 1 06 2-43 016 3-36 058 2-25 218
P2O5 . 002 095 0 21 213 003 023 035 105 082
MnO 021 0 10 010 012 0 18 015 012 018 017
BaO . — — — — — — — 004 003
Total 10022 100 69 99-34 10018 10042 99-86 99-66 10043 10031

Normative Compositions
Orthoclase . — — 39-5 9.4 —
Albite — — — — — — 5-2 —
Anorthite . 2-5 2-5 50 108 92 100
Leucite — — — — — — 13 1 — 148
Kaliophilite 4-7t 5-U 126$ 8-5* 14-21: 7-8J — — 17-0J
Nepheline . 68 167 213 26-7 37-5 25-8 18 7 18 5* —
Na2SiO3 . — 16 — — —
Acmite — — — — 5 1 — Ill — —
Diopside . 64-2 536 19-2 160 95 14 8 123 34 1 261
WoUastonite — — 22-4 10 1 305 — 1-7 — —
2CaO.SiO 2 09 10 40 12 2 2-4 — — 8-4
Hypersthene — — — — — 90 — 2-2 —
Olivine . — — 3-5
Magnetite . 15 1 12 8 9-5 60 16 12 Ill 104
Haematite . — 0-8 21 — — — —
Ilmenite 49 61 21 4-6 03 64 11 4-3 41
Apatite — 2-4 03 50 — 07 14 2-4 20
Calcite 06 — 02 02
Water . 39 36
33-4§
Total 99 1 1010 99 1 99-9 101 1 995 100 1 1005 1001

% Kaliophilite made in preference to calcium orthosilicate. § 2CaO.Al,O1.SiO,(Gehlenite). Nepheline made


in preference to olivine.
Analysts: 1-7, R. C. Tyler; 8-9, B. C. King.
78 B. C. KING—PETROGENESIS OF
Although the alumina content of the alkaline rocks is within the same range
as that of the calc-alkaline rocks, their content of alkalis for a given percentage
of alumina is considerably higher. Average values in the alkaline rocks conform
fairly closely to a 1:1 molar ratio of alkalis to alumina, which corresponds to the
ratio of these constituents in alkali felspars and felspathoids and reflects poverty
in, or absence of, the anorthite molecule.
A considerable scatter is shown, however, by the plots of alkalis and alumina
for individual rocks (Fig. 6). Those rocks in which the ratio is greater than unity
reflect the tendency for excess alkalis to combine with ferric iron to form the
acmite molecule and, among the leucocratic rocks, the development of such

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


late-stage minerals as cancrinite and pectolite. Where the ratio is less than unity,
the excess alumina to some extent substitutes for silica in pyroxenes, but more
generally it is accounted for by the presence of melilite (gehlenite).
The diagram also shows that whereas, collectively, the plutonic rocks conform
to a molar ratio of 1:1 for alkalis and alumina, the volcanics (on average) show
a relative deficiency in alkalis. Moreover, higher total values for alkalis plus
alumina are represented among the plutonic rocks than in any of the volcanic
rocks.
In calc-alkaline rocks, such as basalt and andesite, lime is partitioned between
the two predominant minerals, pyroxene and plagioclase. In the typical alkaline
rocks two minerals also predominate, namely pyroxene and nepheline, but the
pyroxene is the only mineral which contains significant amounts of lime.
Understandably, therefore, the pyroxene of the alkaline rocks is normally
diopsidic in composition and the only other varieties are those enriched in the
hedenbergite and acmite molecules, in contrast to the presence of calcium-poor
pyroxenes in the calc-alkaline rocks.
The non-appearance of plagioclase in the alkaline series, despite the abund-
ance of lime, is most clearly understood by a consideration of the form of the
boundary surface of plagioclase in the system CaO-Al 2 O 3 -SiO 2 -Na 2 O (Fig. 7).
The field of all but the most sodic plagioclase lies close to the CaO-AI2O3-SiO2
face of the tetrahedron. The presence of a comparatively limited amount of soda
in the system suffices to cause initial crystallization of nepheline in place of an
anorthite-rich plagioclase. In the same system (Fig. 11), the extensive field of
wollastonite and, correspondingly, the great extent of the wollastonite-nepheline
boundary surface illustrate the wide range of temperatures over which, in natural
magmas, pyroxenes (and wollastonite itself) may be expected to crystallize
simultaneously with nepheline.

MINERAL ASSEMBLAGES IN THE VOLCANIC AND PLUTONIC SUITES


Typically, the nephelinites consist of olivine (Mg-rich), diopsidic pyroxene,
nepheline, melilite, and iron ore (ilmenite-magnetite). Of these, olivine and
melilite are not always present. Perovskite is often an abundant accessory.
A green, diopside-rich, aegirine-augite is found in many lavas, but nearly pure
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 79

Pyroxenite. Melteigite
Ijolite
Urtite
Syenite. Nepheline Syenite. Fenite
Turjaite
Nephelinite. Melanephelinite
Phonolite. Trachyte

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


PLUTONIC FIELD

VOLCANIC FIELD

AVCRACC
OLIVINe BASALT

IO 12 4 16 IS 2O 22
A L U M I N A MOLAR EQUIVALENT OF WEIGHT °h

FIG. 6. Relations between alkalis and alumina in volcanic and plutonic rocks from Elgon, Napak,
and Budeda. With the exception of phonolites and trachytes, all the volcanics plot below the dotted
line across the volcanic field. Thus, as well as there being a general tendency for the ratio of alkalis
to alumina to be higher in the plutonic rocks, a large proportion of these rocks show higher values for
alkalis plus alumina than are shown by the majority of the volcanics. The plutonic and volcanic fields
are delimited by long and short dashes, respectively.

diopside is usual in those that contain olivine or melilite (cf. King, 1949, p. 31).
Although the same minerals, with the exception of olivine, are also recognizable
in the groundmass of the lavas, the matrix of the groundmass is not certainly
determinable, beyond the fact that it is light coloured, of low refractive index,
and very weakly birefringent or isotropic. However, judging from the analyses,
it is necessary to assume in many cases the presence of a mineral (or minerals)
80 B. C. KING—PETROGENESIS OF
with a higher ratio of silica to alkalis and alumina than in nepheline; analcite
seems the most likely, but natrolite is a common secondary mineral and is
abundantly represented in the amygdales. The nature of the matrix, which is
very difficult to separate, has not been systematically investigated. Hytonen

CaO

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


AI 2 O 3 Na2O

FIG. 7. Thefield(volume) of plagioclase in the system Na2O-CaO-AI2O3-SiO2. The diagram is based


on a tetrahedron constructed from the relevant, investigated, partial systems. The volume of plagio-
clase, the form of which is of necessity somewhat generalized, extends from the field of anorthite on
the Ca0-Al 2 0 3 -SiO 2 face to that of albite on the NajO-AljOa-SiO,, face. The upper surface of the
volume, which is shaded, is largely bounded by the volumes of nepheline and wollastonite. Actual
relations are complicated by such phenomena as partial solid solution between nepheline and
anorthite.

(1959) has recorded sodalite from the groundmass of a melanephehnite from


Napak, but analcite and natrolite are common in the groundmass of Toror
phonolites.
Of the felspar-bearing lavas the phonolites, by varying proportions of alkali
felspar and nepheline, show transitions from the nephelinites (Davies, 1952).
The phonolites are characterized by an aegirine-rich pyroxene in addition to the
nepheline and alkali felspar.
Members of the ijolite series consist typically of aegirine-augite and nepheline,
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 81
1
together with more or less abundant melanite, while wollastonite is prominent
in the leucocratic rocks. Cancrinite after nepheline, and pectolite after wollasto-
nite, are the usual secondary minerals. The mineralogy of the pyroxenites shows
a closer affinity with the lavas, namely, in the presence of diopsidic pyroxene,

SiO,
Fe,O: Fe,O,

Fe.O,

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


i2O_K2O
AI2O3 FeO
FIG. 8. Mineralogical phase diagram of the volcanic rocks. Representative analyses of lavas from Mt.
Elgon and Napak are plotted. The radii of the circles are proportional to the content of Fe2O3; the
largest circle represents about 45 per cent.

nepheline, iron-titanium oxides, and perovskite. The turjaites, with diopside,


nepheljne, melilite, and iron ore, are also comparable with the lavas.
The felspar-bearing plutonic rocks are characterized by nepheline (and can-
crinite in many examples), potash-rich alkali felspar, and an aegirine-rich
pyroxene.
Thus the volcanic rocks conform largely to the following mineral assemblages
(Fig. 8):
Olivine-diopside-nepheline-iron ore.2 (Olivine-nephelinites.)
Melilite-diopside-nepheline-iron ore.2 (Melilite-nephelinites.)'
Diopside-nepheline-(analcite). (Nephelinites.)
Absent from the ijolites of Budeda. Iron-titanium oxides.
6233.1 G
82 B. C. KING—PETROGENESIS OF
The ijolite series of the plutonic rocks consists essentially of the following
assemblages (Fig. 9):
Diopside-nepheline-iron ore.1 (Pyroxenites.)
Aegirine-augite-nepheline2-melanite. (Melteigite-ijolite.)
Nepheline2-wollastonite-melanite-aegirine-augite. (Urtites.)

SiO

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


Fe,O:

AI2O3 FeO
FIG. 9. Mineralogical phase diagram of the plutonic rocks. Representative analyses from Lokupoi
(Napak) and Budeda are plotted. The radii of the circles are proportional to the content of Fe 2 O 3 :
the largest circle represents about 5 per cent.

The mineral assemblages in the volcanic rocks are sufficiently close to nor-
mative minerals for direct correlation between the two to be possible. In the
case of the plutonic rocks, this is not so and the normative minerals differ con-
siderably from those of the mode. Nevertheless, the high degree of under-
saturation is clearly indicated and, indeed, in many cases calcium orthosilicate
appears in the norm. In the modal minerals, however, the silica deficiency is
accounted for largely by such relationships among normative minerals as the
following:
1
Iron-titanium oxides.
1
Determinations of the alkali ratios in fourteen samples of nepheline from ijolitic rocks gave values
for the kaliophilite molecule ranging from 17-5 to 24 per cent.
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 83
1
(i) The presence of Tschermak's molecule in the pyroxene in place of nor-
mative anorthite:
CaAl2Si2O8 = CaAI2SiO6+SiO2.
anorthite Ca-Tschermakite
In turjaite, as in many of the volcanic rocks, melilite (gehlenite) is essentially the
silica-deficient equivalent of normative anorthite:
2CaAl2Si2O8 = 3SiO2+Ca2Al2SiO7+Al2O3.
anorthite gehlenite (in pyroxene)
(ii) The occurrence of melanite garnet as an approximate equivalent of acmite

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


and wollastonite:
2NaFeSi2O6+3CaSiO? = Ca3Fe2Si3O12+Na2SiO3+4SiO2.
acmite wollastonite garnet
The acmite: diopside ratio in the norm is often greater than that shown by the
modal pyroxene, while wollastonite, although a common mineral in the urtites,
especially, often appears in the norm where none occurs in the mode. The
following relationship, however, is also relevant in the case of normative
wollastonite:
3CaSiO3+Fe2O3 = Ca3Fe2Si3012.
The Fe2O3 is represented normatively by haematite or in magnetite.
(iii) The substitution of titania for silica in melanite. Calculations based on
chemical analyses of melanites show that titanium must largely be accommodated
in the silicon structural sites.
Minerals such as pectolite account for the occasional occurrence of sodium
silicate in the norm or its appearance in an equation such as the first of (i)
a ove
- Na2SiO3+2CaSiO3 = Na2Ca2Si309.
(cf. pectolite)
On the diagrams showing the mineral assemblages (Figs. 8 and 9), representa-
tive analyses have also been plotted; the plots show a close accordance with the
simplified mineralogical fields appropriate to the volcanic and plutonic rocks
respectively, although some disparity is to be expected owing to the ambiguity
in the status of alumina, the entry of potash into biotite (which is abundant in
a few rock types), and the omission of titania which plays a number of different
roles.
Of particular interest and significance are the divergent trends shown by the
predominant members of the volcanic and plutonic suites; there are, thus, no
volcanic equivalents of much of the ijolite series. The tendency towards a linear
form in each of the trends suggests the predominance of a single process in the
evolution of each of the two suites, while the relations between the trends and
the fields of mineral phases strongly suggest that evolution has in each case been
dependent on sequences of crystallization of the minerals.
1
An account of the pyroxenes is in course of preparation. In general, chemical compositions show
that only limited amounts of the Ca-Tsch molecule are present.
84 B. C. KING—PETROGENESIS OF
A near coincidence of composition is shown, however, between melane-
phelinite and pyroxenite-melteigite. This is considered to be of fundamental
significance, for it is evidently in this compositional range that the immediately
parental magma is to be sought.
CRYSTALLIZATION OF THE NEPHELINITE SERIES
The order of crystallization among the phenocrysts was: olivine, diopside,
nepheline, melilite, but considerable overlap is represented by the last three.
Diopside, changing in composition towards aegirine-augite, and nepheline are
also prominent in the groundmass and it is evident that pyroxene and nepheline

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


crystallized simultaneously during much of the consolidation of the rock. The
composition of the final residue is represented approximately by the analcite (or
similar mineral) which constitutes the matrix.
In experimentally investigated systems containing quartz or felspars, there is
no reaction relation between forsterite and diopside so that separation of for-
sterite in excess of its stoichiometric proportion, with consequent enrichment
of the melt in silica, is not to be expected. The liquidus diagram for the diopside-
nepheline-silica system, however, shows a field of forsterite, which may be
regarded as the result of the partial desilication of diopside by nepheline. In the
diopside-nepheline-anorthite system, both forsterite and gehlenite have fields of
primary crystallization. These partial systems are thus not truly ternary. In both
of these systems early separation of olivine and melilite (or either of these
minerals) would lead to relative enrichment of the melt in silica.
Investigation of the 'binary' system diopside-nepheline (Bowen, 1928, p. 261)
has shown that, depending on composition and temperature, either diopside or
nepheline, together with olivine or melilite, or olivine and melilite, may be in
equilibrium with liquid. Such liquids must be relatively enriched in silica.1 Thus,
in a crystallizing melt having a composition expressible simply in terms of diop-
side and nepheline, the following types of reactions2 may be envisaged:
2CaMgSi2O6+2NaAlSiO4 -> Mg2SiO4+Ca2Al2SiO7+Na2SiO3+3SiO2, (1)
diopside nepheline forsterite gehlenite
4CaMgSi 2 O 6 +5NaAlSiO 4 -> 2Mg 2 SiO 4 +4NaCaAlSi 2 O 7 +NaAlSi 3 O 8 ,. (2)
diopside nepheline forsterite Na-me!ilite albite
8CaMgSi 2 O 6 +3NaAlSiO 4 -» Mg 2 SiO 4 +4Ca,MgSi 2 O 7 +3NaAlSi 3 O 8 . (3)
diopside nepheline forsterite akermanite albite
During the closing stages of crystallization of the lavas, as represented by the
groundmass, the accumulated products of (1) might react as follows:
Na2SiO3+3SiO2+Fe2O3 -> 2NaFeSi2O6, (4)-
acmite
NaAlSiO 4 +SiO 2 +H 2 O -> NaAlSi 2 O 6 . H2O. (5)
analcite
1
In relation to the CaO-MgO-SiO2 system, Bowen showed-that the addition of nepheline has the
effect of shifting the boundaries between pyroxenes and forsterite or akermanite towards the silica
corner (cf., also, Yoder & Tilley, 1962).
2
Comparable reactions are inferred by Yoder & Tilley (1962, p.-356).
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 85
A possible course towards residual magmas of phonolitic composition is sug-
gested by the three primary reactions, as well as a tendency for such magmas to
become undersaturated in alumina. It may also be noted that the crystallization
of melilite leads to impoverishment of the melt in lime more rapidly than by
separation of a diopsidic pyroxene.

The role of Melilite


Data on the compositions of the melilites are scant, but it is evident that con-
siderable variations occur. Optical properties for melilites of Napak lavas agree

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


with a composition Ak 30 Ge 70 (King, 1949, p. 33), but the much lower refractive
indices quoted by Davies (1952, p. 35) for Elgon imply a considerable content
of sodium melilite, although it is stated that the birefringence varies greatly.
On the other hand, the relatively high birefringence of melilite in a nephelinite
at Toror suggests a high content of iron akermanite.
Two chemical analyses of melilites are available; one from an Elgon lava
(Smith, 1953) has approximately the composition Ak65Ge14Fe-Ge6Na-Mel15;
the other, from a turjaite from Napak,1 corresponds to Ak47Ge22Fe-.Ge10Na-
Mel21. The absence of zoning in the large crystals of melilite in the turjaite
suggests that the composition may correspond to the maximum content of the
soda molecule (and lowest melting temperature) that is compatible with the
stability under the conditions of crystallization (cf. Deer et a/., 1962, fig. 61).
Experimental evidence shows that the stability of both akermanite and sodium
melilites is limited to high magmatic temperatures, which may explain the fact
that melilite is largely confined to the volcanic (or hypabyssal) rocks. It also
seems likely that retention of volatiles (and, possibly, increase of pressure) under
plutonic conditions inhibits the formation of melilite, although there are indica-
tions that sodium-melilite has a greater range of stability than akermanite or
gehlenite. The composition of melilite, having a structure in which Si, Al, and
Mg enter into similar tetrahedral groups, is evidently highly sensitive to the
composition of the liquid from which it is precipitated.
A reaction relation between melilite and pyroxene is suggested by the develop-
ment of symplectic intergrowths between the two minerals, as shown by the
turjaites, while the invariable enclosure of pyroxene by melilite shows that
the latter formed with declining temperature, a conclusion in agreement with
the relations between diopside and nepheline as determined by Bowen in 1922.
The findings of Chinner & Schairer (1962, p. 630), in their investigations on
parts of the system CaO-MgO-Al 2 O 3 -SiO 2 , are relevant to the problem of the
relations between pyroxene and melilite. The composition of pyroxene in equi-
librium with liquid will increase in aluminium content to the point where
pyroxene reacts with liquid to form an akermanite-rich melilite. The effect of
1
A melilite nephelinite recently collected from Napak is much richer in melilite than any previously
found and the mineral forms large phenocrysts. Its optical properties resemble those of the melilite
from the turjaite.
86 B. C. KING—PETROGENESIS OF
^Q

• Pyroxenne
O Melleigite
O l|ohte
O Urine
60 ® Felspalhpzed Ijolite
X Nepheline & Cancrinite
+ Fenite Syenite
° Turjaite

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


• Nephehmte & Melanephelmite
' Phonotile. Trachyte, etc.

180

FIG. 10. Modified von Wolff diagram of rocks from Elgon, Napak, and Budeda. The volcanic field
and main plutonic field are separately outlined. Normative anorthite has been subtracted from L to
give Fp (alkali felspar) and values for Fp and M recalculated accordingly. If An is plotted as a vertical
component on a three-dimensional graph, much of thefieldof nephelinite-melanephelinite (containing
appreciable An) is separated from the ijolite field (containing little or no An). This reflects the ten-
dency for alkalis to be lower in the volcanics than in the plutonics, so that in calculation, the excess
alumina is allocated to lime in An, which in the rocks themselves is expressed as melilite.

removal of early formed diopside would be to enrich the liquid in alumina and
lime, enhancing its ability to crystallize melilite.

EVOLUTION OF THE VOLCANIC ROCKS


On both the modified von Wolff diagram (Fig. 10) and the mineral phase
diagram (Fig. 8), plots of the compositions of the lavas lie within the field of
silica undersaturation, but form a belt roughly parallel to the saturation boun-
dary (pyroxene-alkali felspar). On Fig. 8 it is also seen that they fall very nearly
astride a join from forsterite (and melilite) to alkali felspar. These observations
suggest that the series could have been produced by early separation of forsterite
and melilite, together with pyroxene, from a magma of melanephelinite com-
position. The highly melanocratic rocks, which show a considerable scatter of
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 87
plots, probably represent the effects of enrichment by accumulation of early
formed crystals. It is possible, in view of the comparative rarity of lavas rich
in olivine, that accumulates of this mineral exist in depth, but have not been
erupted.
It is inferred that early crystallization of olivine, pyroxene, and melilite pro-
duced derivative magmas in which the silica content, relative to alkalis and
alumina, increased progressively from that of nepheline towards that of alkali
felspar. The separation of melilite also regulated the lime content of the residual
melts, so that they were progressively impoverished in this constituent. Bowen
(1922,1945) showed that the course of liquids in the Ge-An-Wo-Ne-Ab system

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


could pursue alternative courses, of which one involves separation of melilite,
together with wollastonite (representing pyroxene) and nepheline, leading ulti-
mately to enrichment in albite (Fig. II). 1 During the earlier stages of crystalliza-
tion, nepheline is the only phase which contains appreciable potash, but the
amount of this constituent is limited to some 25 per cent of the kaliophilite
molecule, so that potash also tends to accumulate relative to soda in residual
melts. This would explain the relative abundance of potash shown by many of
the phonolites and trachytes, and the occurrence of a potash-rich alkali felspar
in these rocks. The non-appearance of leucite in the series is to be ascribed to
the relative rates of enrichment of silica and potash; at the stage when the
potash-soda ratio was sufficiently high for the formation of leucite, the accumu-
lation of silica was enough to inhibit it.

CRYSTALLIZATION OF THE IJOLITE SERIES


The absence of olivine is a notable feature of the ijolites of Napak as com-
pared with the lavas, even in cases where the chemical compositions are closely
similar. Olivine is never found in the plutonic rocks of Budeda or Tororo, but
it does occur in peridotites at Bukusu. Peridotites or dunites are, indeed, some-
times represented in the alkaline complexes of Africa (e.g. Shawa, S. Rhodesia;
Johnson, 1959), but they are rare and cannot be regarded as typical. It seems
reasonable to infer that under the physico-chemical conditions of formation of
the ijolites, the crystallization of olivine is inhibited and it is not formed by
reaction, even for compositions that are poorer in silica than are represented by
points on the nepheline-diopside join. It is possible that lower temperatures of
crystallization, consequent upon the retention of volatiles, is an important fac-
tor, but the presence of the volatiles themselves, presumably chiefly water and
carbon dioxide, may have a similar effect. These could also be major factors in
the non-appearance of melilite and, indeed, it is possible that crystallization
occurred at temperatures below the field of stability of the melilites.
In general, the sequence in the ijolite series is from melanocratic to leucocratic
(pyroxenite-melteigite-ijolite-urtite), together with pegmatitic phases and
1
Further understanding of the relevant systems is required to account, more adequately, for the
absence of plagioclases of intermediate composition.
B. C. KING—PETROGENESrS OF

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


Na2Si03 SiO-
FIG. 11. Quaternary system Na2SiO3-CaSiO3-Al2SiO6-SiO2 showing the relationships of the fields
of nepheline and plagioclase. Phase boundaries have been inferred from a tetrahedral construction
based on investigations of the various partial systems in the system Na2O-CaO-Al2O3-SiO2. The
unnamed corner of the field represents Al2SiOs. The phase boundary between corundum (cor.) and
mullite has been omitted.
The field of wollastonite overlies the entire upper (front) surface of the plagioclase field and the
right-hand sector of the upper surface of the nepheline field. In more complex systems represented
by actual rocks, diopside would take the place of wollastonite over the upper part of the field.
The thermal divide, commencing at X on the wollastonite-nepheline boundary plane, separates com-
positions which trend towards the precipitation of meliliteand, ultimately, of alkali felspar from those
trending towards residues rich in alkali silicates (trends shown by arrows).

late-stage alterations. A generalized succession of the principal minerals is as


follows:
Early Late
Diopside Aegirine-augite Aegirine
Nepheline Cancrinite
Wollastonite Pectolite
Perovskite Sphene Melanite Andradite
Iron-titanium oxides
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 89
It can be inferred, too, from field and petrographic evidence that many of the
phases have formed by metasomatic replacement and, for example, that meltei-
gites represent modified pyroxenites, while urtites of certain types developed by
nephelinization of melanocratic rocks. Textural evidence suggests that melanite,
in all cases, has formed by replacement, chiefly at the expense of perovskite and
pyroxene.
The early crystallization of diopside and nepheline, although it involves ex-
traction of lime from the beginning, restricts the rate of extraction to molar parity
with magnesia, so that lime tends to accumulate. In the lavas, such a tendency is
dissipated by the crystallization of melilite. In the lavas, too, both ferrous and

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


ferric iron oxides, and titania, are accommodated largely in the iron-titanium
oxides which have an extended period of separation. Apart from the pyroxenites,
where these constituents are represented in iron-titanium oxides and perovskite,
the sequence of crystallization in the melteigites and ijolites leads to concentra-
tion of iron and titania.
The early change in composition of the pyroxene, from diopsidic to heden-
bergite-rich, is followed by a trend towards aegirine; this reflects not only the
build-up of ferric relative to ferrous iron, but also impoverishment of the magma
in magnesia. At the same time, the substitution of soda for lime in the pyroxene
enhances the tendency for relative enrichment in lime in later liquids, and may
determine the appearance of new lime-bearing minerals, notably melanite and
wollastonite.
Petrographic evidence suggests that enrichment of the pyroxene in acmite
depends partly on a reaction between nepheline and diopside with declining
temperature, but other factors must operate simultaneously. For example,
equations of the following types may be envisaged:
3CaMg(SiO3)2+6NaAlSiO4+2Fe2O3 ->-
diopside -nepheline
-+ 2NaFeSi 2 O 6 +Ca 3 Fe 2 Si 3 O 12 +MgAl 2 SiO 6 +2Na 2 SiO 3 , (1)
acmite andradite Mg-Tsch.
(melanite)
or
3CaMg(SiO3)2-}-6NaAlSiO4+Fe2O3 ->
diopside nepheline
-+ 2NaFeSi2O6+3CaSiO3+MgAl2SiO6+2Na2SiO3. (la)
acmite wollastonite Mg-Tsch..

The Tschermak's molecule is probably to be regarded rather as a deficiency


than as a product of the reaction; that is, its constituents should be subtracted
from the left of the equation rather than added to the right. Analyses of the
pyroxenes show that substitution of silicon by aluminium, which is never ex-
tensive, is greater in the most diopsidic pyroxenes; alumina increases again in
the acmitic pyroxenes, but here aluminium is evidently substituting in the Y
position.
90 B. C. KING—PETROGENESIS OF
Melanite (andradite) may be regarded as forming by a simple reaction between
wollastonite and ferric oxide, although there is no textural evidence that this
actually occurs: ,-,
J
3CaSiO3+Fe2O3->Ca3Fe2Si3O12. (2)
wollastonite andradite
Another possibility that has been envisaged (cf. Yoder & Tilley, 1962, p. 426)
relates to an increasing oxidation ratio as crystallization proceeds:
6CaFe"Si2O8+30 -»- 2Ca3Fe2"Si3O12+Fe2O3+6SiO2. (3)
hedenbergite andradite

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


The abrupt change in the compositional trend of the pyroxenes from enrich-
ment in hedenbergite towards aegirine may be due to the instability of a
hedenbergite-rich pyroxene at higher oxidation ratios. The appearance,
variously, of sodium silicate, silica, and ferric iron among the reaction pro-
ducts of (1) and (3) may contribute to the formation of aegirine:
Na2SiO3+Fe2O3+3SiO2 -> 2NaFeSi2O6.
The compositional relations between wollastonite and melilite molecules
illustrate the way in which wollastonite may be regarded, in some degree, as the
chemical substitute in the plutonic series for melilite in the volcanic series:
Ca2Al2Si07 = CaSiO3+CaO+Al2O3,
gehlenite wollastonite
Ca2MgSi2O7 = 2CaSiO 3 +MgO,
akermanite wollastonite
CaNaAlSi 2 O 7 = CaSiO 3 +NaAlSiO 4 .
sodium melilite wollastonite nepheline

The first two equations emphasize the extent of silica undersaturation in meli-
lites; the third equation represents a reaction which has been shown to occur in
artificial systems, wollastonite and nepheline being stable at lower temperatures
(Deer et ah, 1962, p. 241). Wollastonite is a mineral that is typical of the urtites
and ijolite pegmatites; its formation evidently depends on the molar concentra-
tion of lime, in derivative liquids, exceeding that of magnesia and ferrous iron
combined, and on it being in excess of the ferric iron with which it would com-
bine to form andradite (melanite). In the CaO-MgO-SiO 2 system, diopside and
wollastonite are in equilibrium with liquid over a significant range of tempera-
ture, a relationship that is unaffected by considerable amounts of alumina. The
variable order of crystallization between pyroxene and wollastonite may reflect
this relation, for although wollastonite is usually later than pyroxene, examples
of the investment of wollastonite by diopside are not uncommon.
The considerable volume of the stability field of primary wollastonite in the
system Na 2 O-CaO-Al 2 O 3 -SiO 2 is noteworthy, and particularly its extension
over compositions that are relatively poor in lime but contain much soda (Fig.
11). Indeed, the composition point for Na 2 0. 3CaO. 6SiO2 lies well within the
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 91
wollastonite field. The low-temperature invariant point that may be inferred,
involving wollastonite, nepheline, and the soda-lime silicates, Na2Ca3Si6O16 and
Na2Ca2Si3O9 (together with the divariant planes and the unvariant lines involv-
ing two or three of these phases), is especially relevant to the petrochemistry of
the later phases of the ijolites. Pectolite, an almost ubiquitous replacement pro-
duct of wollastonite, may be regarded as related to the synthetic Na 2 Ca 2 Si 3 0 9
by substitution of Na 2 O by H 2 O to give Na 2 Ca 4 Si 6 O 17 .H 2 O. In natural systems,
enrichment in volatiles, notably water and carbon dioxide, leads to the formation
not only of pectolite but also of cancrinite, largely at the expense of nepheline.
Cancrinite (3NaAlSiO4. (Na,Ca)CO3) can be considered to form in the reciprocal

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


system (Na2,Ca)-(Co3,Al2Si2O8) where the concentration of soda suffices to
preclude the appearance of the anorthite molecule (or corresponding scapolite)
(cf. Eitel, 1954, p. 896). The occurrence of zeolites as secondary minerals and in
late-stage veins is of similar significance.
In certain cancrinite syenites from Napak and Budeda, the cancrinite has a
distinctive form, unlike that representing pseudomorphs after nepheline (pp.
71 and 72). Experimental evidence (Eitel, 1954, p. 895) shows that at an appro-
priate concentration of calcite and at elevated pressure, cancrinite is the initial
phase to crystallize in the binary system calcite-nepheline. Similarly, pectolite
also occurs independently of pseudomorphs after wollastonite, suggesting that
at an appropriate water-vapour pressure and with adequate concentration of
soda, it, too, crystallizes directly.
The appreciable content of titania in the ijolite series is accommodated in
perovskite and iron-titanium oxides in the pyroxenites, and in sphene and
melanite in the melteigites and ijolites, the minerals forming reaction series in
that order. Whereas, however, in perovskite and sphene the titania content is
sensibly fixed, in melanite there is considerable variation. From a maximum of
8 or 9 per cent there is a progressive decline to 4 or 5 per cent, which is fairly
closely marked by a change in the melanite from almost opaque to a translucent
brown, in thin section, often seen in zoned crystals. The change can also be corre-
lated, in the rocks themselves, with the transition from melteigite to leucocratic
ijolite and urtite. In some of the urtites, and in late-stage veins, an almost colour-
less garnet, near andradite in composition and devoid, or nearly so, of titania,
reflects the ultimate impoverishment of solutions in this constituent. There is a
marked diminution in the magnesia content, and a gradual increase in the
alumina content, in the sequence from the melanites to andradite; a maximum
of about 14 per cent of the pyrope molecule and 15 per cent of grossularite
(molar) are represented by analyses of garnets from the Napak ijolites.

EVOLUTION OF THE IJOLITE SERIES


Plots of the compositions of members of the ijolite series lie parallel to the
join pyroxene-nepheline (in contrast to those of the volcanic series which
approximate to the forsterite—alkali felspar join). Both Shand (1933, p. 10) and
92 B. C. KING—PETROGENESIS OF
Barth (1936, p. 344) have regarded reactions involving, essentially, these two
pairs of minerals as important in the formation of felspathoidal rocks. Thus,
Shand's equation, plagioclase+olivine ^ nepheline+pyroxene, is given by
Barth in the form:
NaAlSi3O8-f-2Mg2SiO4 ^ NaAlSiO4+4MgSiO3.
albite forsterite nepheline enstatite

Among others, however, Yoder & Tilley (1962, p. 405) object to this because
there is no evidence that, under .normal conditions of a crystallizing magma,
nepheline and magnesian pyroxene are in equilibrium. The same principle, how-

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


ever, can be represented by an alternative equation in which diopside and
nepheline, minerals known to coexist, are employed:
4CaMgSi2O6+5NaAlSiO4 ^ 2Mg2SiO4+4NaCaAlSi2O7 + NaAISi3O8.
diopside nepheline forsterite Na-melilite albite

The mineral phases to the left and to the right are characteristic, respectively,
of the assemblages in the plutonic and volcanic series in the alkaline rocks.
Diopside and nepheline are additional phases in the volcanic series, and it is the
early separation of forsterite and melilite, leading to relative enrichment of silica
in the liquid, that is responsible for the trend towards an alkali felspar com-
position.
What determines these alternative courses of early crystallization, which pro-
foundly affect the compositions of derivative melts, cannot be decided with any
certainty, but it would form a fruitful subject for experimental inquiry. It has
been suggested (p. 87) that retention of volatiles, by lowering the temperature
of crystallization, might suppress the field of forsterite along the diopside-
nepheline join and render the melilites unstable. It seems possible, also, that the
presence of (CO3)" in the melt might, as an additional anion, favour the initial
crystallization of phases which have a higher ratio of silica to basic oxides; in
effect, ino- and tecto-silicates rather than neso- and sorosilicates. Experiments
by Yoder & Tilley (1962, p. 501) showed that an olivine nephelinite melt,
quenched at 1200° C under pressures of 31-4 and 40 kilobars, yielded a pyrox-
enite with spinel and phlogopite mica. Such pressures, which would correspond,
in terms of simple hydrostatic pressure, to a crustal depth of 100 km or more,
are far in excess of those relevant to the problem under discussion. Also, the
soda and alumina contents of the pyroxene produced, although not determined,
are almost certainly greater than those in the pyroxenes of the alkali pyroxenites.
Nevertheless, such experiments provide a pointer to the kind of factors that are
likely to be influential in modifying the mineral assemblage of the volcanic
series.
Bowen's analysis (1945) of the partial system Geh-An-Wo-Ne-Ab, and the
studies by Gummer (1943) and Spivak (1944) of that of Wo-sodium silicate-An-
Ne in the system Na 2 O-CaO-Al 2 O 3 -SiO 2 , also provide important clues to the
solution of this problem. On the boundary curve between wollastonite and
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 93
nepheline, the presence of a temperature maximum (thermal divide) would
determine alternative paths for crystallizing melts (Fig. 11). The one trends
towards melilite and ultimately, with fractionation, towards an albitic plagio-
clase while the other trends towards alkali-rich silicates. Following this latter
trend, there is no increase in silica in derivative liquids and lime tends to be
stored; especially in relation to magnesia and iron. The two trends thus corre-
spond, in simplified form, to the trends shown by the volcanic and plutonic
series, respectively.
The contrasting courses of evolution of the volcanic and plutonic series is

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


most effectively shown by plotting, against each other, the ratios of total iron
oxides: magnesia and lime:magnesia (Fig. 12). Enrichment in iron oxides rela-
tive to magnesia is characteristic of both series, but the ijolite series, alone, shows
a striking enrichment of lime relative to magnesia. The increase is progressive
from pyroxenites and melteigites to urtites and, indeed, many wollastonite
urtites and ijolite pegmatites show higher absolute values for lime than do those
of the main ijolite series.
Although a common parent magma is postulated, necessarily, for both the
volcanic and the plutonic series, slight differences in composition, achieved at
an early stage, might account for the divergence in trend. In particular, it is
evident that a slightly higher content of soda (or total alkalis) in the system
Na2O-CaO-Al2O3-SiO2 would favour migration towards an alkali silicate end-
point (Fig. 11). Comparison of chemical analyses suggests that even where the
compositions of volcanic and plutonic rocks are closely similar, the latter con-
tain relatively higher alkalis (Figs. 4 and 5). If an important factor in determining
the volcanic trend is loss of volatiles, it is likely that under the same conditions,
alkalis also tend to escape. Experimental evidence confirms this view, since it is
well established that alkalis volatilize readily from silicate melts of certain com-
positions, notably those that are high in lime (cf. Eitel, 1954, p. 1136). It is con-
cluded, from the evidence of the mineralogy and chemistry of the two series,
that the plutonics, in contrast to the volcanics, crystallized in a closed system
in which volatiles and alkalis were prevented from escaping.
Although the ijolite series generally trends towards rocks that are rich in
nepheline, the later stages of crystallization show some measure of diversity,
which may be regarded as reflecting in exaggerated form the more particular
chemical tendencies shown during the general course of evolution (Fig. 13).
Thus, the urtites proper show extreme alkali (especially soda) enrichment; the
wollastonite urtites involve additional enrichment in lime; and extreme lime
enrichment, with a lower concentration of alkalis, is shown by the ijolite peg-
matites. In any of these rocks, and particularly in the ijolite pegmatites, calcite
often occurs with the habit of a primary, although late-forming mineral. It is this
trend, towards absolute enrichment in lime and the separation of calcite as a
final phase of crystallization, that is regarded as providing the key to the deriva-
tion of the carbonatites.
CO

z
o
o I
t- m
o" H
7a
ded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014

O
O
m
z
m
in

O
Tl
& Vokon.ci tlqon6NopoV

,-'\ V'' • • Pyroien.tc6Melleiqite Pjioienet:


o • ljol<tc (mcl pcqmaiitei) • Budeda
o D U'hir . Napak
O • Nepheline t ConCrinilC'Sycnile • Tororo

lu

loq CaO/MqO

FIG. 12. Diagram illustrating the relations between the ratios Fe 2 O a +FeO/MgO and CaO/MgO in the alkaline rocks of
Mt. Elgon, Napak, and Budeda. The fields of the volcanics and of the ijolite scries are indicated. Tie-lines join the composi-
tions of pyroxenes to the plots of the containing rocks.
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 95
\CaO

90,

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


• Pyroxenite & Melteigite
o Ijolite •
o Urtite
© Ijolite pegmatite
e Felspathized Ijolite
+ Nepheline & Cancrinite'Syenite
X. Fenites

A Nephelinite & Melanephelinite


A Phonolite

FIG. 13. Triangular plot of the relative proportions of Na2O, CaO, and MgO in the ijolite series and
in the nephelinite-melanephelinites.

ORIGIN OF THE CARBONATITES


Since von Eckermann's re-interpretation of the Alno area in 1948, many
(e.g. Dawson, 1962) have followed him in supposing that the primary material
in alkaline complexes was a carbonatite magma, the various silicate rocks having
been produced by metasomatism of the country rocks by emanations from a
carbonate melt, accompanied, in many cases, by mobilization. This is a tempting
hypothesis, since carbonatite is often seen as the central member of a series
showing progressive reduction of silica; carbonatite is, itself, virtually devoid
of silica so that it is a simple matter to postulate the development, from pre-
existing silicate rocks, of any degree of desilication by reaction.
Moreover, in many complexes (e.g. Sukulu), carbonatite is the dominant
member. Nevertheless, it must be recognized that even the largest masses of
96 B. C. KING—PETROGENESIS OF
carbonatite are trivial by comparison with the volumes of nephelinite lavas and
pyroclastics represented by the volcanic superstructures. At Napak, where both
volcanics and plutonics are found, it is estimated that even if the central complex
extended downwards for 5 miles, its volume would amount to no more than
3 per cent of the nephelinites (lavas and pyroclastics) of the original cone, while
the carbonatite accounts for little more than 1 per cent of the mass of the central
complex.
Judging from the mineralogy of the carbonatites and the type of fenitization
produced where they are directly in contact with country rocks, the carbonatites
appear to have been generated and emplaced at relatively low temperatures.

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


Moreover, their appearance at late stages in the sequences, and their high content
of minor constituents, are only reconciled easily with their representing deriva-
tive magmas. This, indeed, is the conclusion reached from a study of their rela-
tions with the ijolite series, and of the evolutionary trends shown by the ijolites
themselves.
The non-carbonate minerals that occur in the carbonatites include many that
are found in the silicate rocks, but they are those minerals that characterize the
later members of the silicate rock series. These include aegirine (or near aegirine),
alkali felspar, micas, sphene, apatite, and magnetite; inclusions of silicate rocks
are not merely mechanically disintegrated in the carbonatite, but are converted
into phases that are in stable association with the carbonates; thus, aegirine-
augite is recrystallized and transformed to euhedral crystals of a more aegirine-
rich pyroxene. The continuity between the mineralogy of the carbonatites and
that of the silicate rocks supports the suggestion that they have been derived
from the latter.
The occurrence of potash felspar in stable association with calcite, rather than
with a mineral poorer in silica, is of particular significance in view of the evidence
of potash-rich felspathization associated with carbonatites. The probability is
that in the ijolite series, the rate of accumulation of potash only necessitates the
formation of an independent potash mineral contemporaneously with the de-
velopment of carbonatite. Since neither kalsilite nor leucite is stable in this
environment, potash felspar is formed:
CaMgSi2O6+2KAlSi2O6+2CO2 ^ 4(Ca,Mg)C03+2KAISi308.
The high content of alumina would, presumably, determine the combination of
silica with potash (or soda) in an alumino-silicate rather than with the cafemic
constituents.
Although silicate rocks with subordinate calcite, and carbonatites with silicate
minerals, occur, the two groups of rocks are, in the main, sharply defined from
each other in the field. It seems likely that with increasing concentration of car-
bon dioxide, accompanied by accumulating lime and alkalis, a stage is reached
at which a highly mobile fraction of alkali and calcium carbonates, of low melt-
ing temperature, separates from the rapidly crystallizing final silicate phases (cf.
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 97
1
Wyllie & Tuttle, 1960, 1962). It has been shown that in carbonate melts, silica
readily displaces the lime (or alkalis) from combination with carbon dioxide
(Eitel, 1954, pp. 879, 882), so that calcite will not appear until the silica concen-
tration has been considerably reduced.
The comparatively low temperature and the low pressure at which alkali
carbonatites can exist is shown by the observation of lavas of this composition
in the crater of Oldoinyo l'Engai, Tanganyika, by Dawson (1962) arid Guest
(1963). The mobility of intrusive carbonatites is strongly indicated by the way
in which they have broken through the envelope of silicate rocks, so that in
numerous examples of carbonatite complexes they are found in direct contact

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


with the 'basement' (James, 1958; King & Sutherland, 1960, p. 311).
The mode of origin of carbonatites inferred by the writer implies the likeli-
hood that initially, they normally contain alkali carbonates. The alkali meta-
somatism of the country rocks is largely to be ascribed to the presence of these
'fugitive' constituents. Only in so far as they are retained could the carbonatite
survive as a liquid at low temperatures and at atmospheric pressure. Thus,
alkali carbonatite, alone among the derivatives of the ijolite series, is capable of
becoming an extrusive rock.
Recent work on the isotopic composition of strontium in carbonatites shows
that the 87Sr/86Sr ratio is low, and comparable with that in basalts (Hamilton &
Deans, 1963). Nevertheless,.they contain an assemblage of trace elements similar
to that of granites, though in much greater abundance. For example, strontium
averages 4000 to 5000 p.p.m. in carbonatites as compared with less than 400
p.p.m. in granites (Hamilton, 1964). These two observations, considered in con-
junction, are difficult to explain except by supposing that carbonatites are de-
rived by extreme differentiation from an ultimate basaltic parent.

ORIGIN OF THE SYENITES AND NEPHELINE SYENITES, AND THEIR


VOLCANIC EQUIVALENTS
The evidence suggests that rocks of these compositions have originated in a
number of different ways:
(a) In the volcanics series, the subordinate phonolites and trachytes may have
developed by crystallization differentiation from the nephelinites. In this
connexion it is to be noted, however, not only that transitional rock types
are uncommon but also that, since phonolites and trachytes show similar
iron oxide: magnesia and lime: magnesia ratios (Fig. 12), they appear to
have evolved simultaneously along divergent trends.
{b) Rocks of the composition of syenites, or even of nepheline syenites
(Budeda, pp. 70-71), have formed demonstrably as fenites by metaso-
matic processes. It is, therefore, reasonable to infer that trachytes and
1
Recent experimental work by Koster van Groos & Wyllie (1963), which has revealed wide
immiscibility gaps in the systems Ab-Na 2 CO 8 -H ! O and 80Ab20An-Na2CO3-H2O (at 1 kb and around
750° C), appears to be highly relevant to explaining the separation of carbonate melts from silicates.
6233.1 H
98 B. C. KING—PETROGENESIS OF
phonolites of corresponding compositions could have developed as
mobilized fenites.
(c) Nepheline syenites and syenites have also resulted from felspathization of
ijolites, as at Napak and Tororo; a distinctive feature of such rocks is their
high content of potash felspar. This mode of origin is strongly indicated,
also, for the highly potassic felspar-rock and potash-trachyte found at
Toror and Tororo. Such potash-rich felspathization appears to be closely
associated with the formation and emplacement of carbonatites, in contra-
distinction to the earlier fenitization which appears to be dependent upon
invasion by ijolitic rocks (cf. King & Sutherland, 1960). It is appropriate

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


to observe that, although it might be assumed that the occurrence of such
potash-trachytes necessarily implies actual melting and mobilization of
metasomatized material, field and petrographic observations show that
disconnected streaks and channels of trachytic texture develop in the
fenitic aureole of Toror and Napak by undoubted replacement processes.

THE NATURE OF THE PRIMARY MATERIAL


The coincidence of the volcanic and plutonic compositional fields around the
composition of melteigite/melanephelinite is a strong indication that magna of
this composition was the immediate parent to the alkaline suites. There is a possi-
bility, in view of the occasional occurrence of peridotites in a few complexes
(e.g. Bukusu), that the original magma was appreciably richer in magnesia than
melteigite, and that accumulative olivine-rich rocks were removed at an early
stage.
The mode of derivation of the inferred melteigite/melanephelinite magma lies
necessarily in the realms of hypothesis. The fact, however, that although the
volcanics of eastern Uganda are devoid of normal basalts there are extensive
areas elsewhere in East Africa in which nephelinites and olivine-basalts (alkali
basalts) are found together, suggests that the ultimate parentage of the nepheli-
nites is connected with that of the much more abundant olivine-basalts. Indeed,
all gradations between the two exist, through such rock types as basanites and
tephrites. Nevertheless, the apparent impossibility of deriving a melteigite/
melanephelinite magma by any process or combination of processes affecting
a primary basalt magma suggests that this is a genetic problem, which depends
for its solution upon an understanding of the way in which parental, basaltic
magmas are themselves generated.
Yoder & Tilley (1962, pp. 506-7) conclude that the variability of basalt mag-
mas from oversaturated (tholeiitic) to undersaturated (alkaline and nepheline-
bearing) can be explained by postulating derivation from parental eclogite,
enriched either in garnet or omphacite components, respectively. This, however,
is by itself inadequate as an explanation for the genesis of the nephelinites, for
even assuming complete removal of the garnet, the silica content of the omphacite
is too high (around 50 per cent); indeed, this is apparent from the fact that
ALKALINE IGNEOUS SUITES OF EASTERN UGANDA 99
omphacite corresponds normatively to jadeite (equivalent to nepheline plus
albite) and diopside.
It appears necessary to postulate the presence or development in the primary
eclogite of an additional phase having a very low ratio of silica to basic oxides,
or even one devoid of silica, such as spinel. Reaction with spinel would simul-
taneously desilicate jadeite to nepheline, and form a molecule capable of entering
either into pyroxene or reacting to form garnet:
MgAl2O4+NaAlSi2O6 _» MgAl2SiO6+NaAlSiO6,
spinel jadeite Mg-Tsch. nepheline

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


MgAl 2 SiO 6 +CaMgSi 2 O 6 =^ CaMg 2 Al 2 Si 3 0 12 .
pyroxene garnet
Kushiro and Kuno (1963) recognize a similar difficulty in deriving the highly
undersaturated basaltic magmas (nephelinites) from a parental peridotite or its
mantle equivalent. Their explanation, however, depends upon the suggestion
that at sufficient depth, forsterite would melt incongruently to produce enstatite
and a liquid poorer in silica than forsterite itself. Such a derived melt would, at
higher levels, desilicate the jadeite molecule to form nepheline.
Both Yoder & Tilley (1962) and Kushiro & Kuno (1963) regard the influence
on phase relations by differences in pressure, and hence in depth, as a major
factor in determining variations in the compositions of basaltic magmas derived
by the melting or partial melting of parental materials. There is a measure of
agreement that the undersaturated (alkali) basalts originate at greater depths
than those of tholeiitic composition. An affinity between the nephelinite-
carbonatite suite and the kimberlites has often been noted, and the indications
that the source of kimberlites lies at a deeper level than that of basaltic magmas
provide significant support for assigning to nephelinites, also, an origin at greater
depth than that of normal basalts (cf. King & Sutherland, 1960, pp. 718-19).

REFERENCES
BARTH, T. F. W., 1936. The crystallization process of basalt. Amer. J. Sci. 31, 321-51.
BOWEN, N. L., 1922. Genetic features of Alnoitic rocks at Isle Cadieux, Quebec. Ibid. 3, 1-34.
1928. The evolution of the igneous rocks. Princeton Univ. Press.
1945. Phase equilibria bearing on the origin and differentiation of alkaline rocks. Amer. J. Sci.
243A, 75-89.
CHINNER, G. A., & SCHAIRER, J. F., 1962. The join Ca3AI2Si3012-Mg3AI2Si3012 and its bearing on the
system CaO-MgO-Al2O3-SiO2 at atmospheric pressure. Ibid. 260, 611-34.
DAVIES, K. A., 1952. The building of Mount Elgon, East Africa. Mem. Geol. Surv. Uganda, no. 7.
1956. The geology of south-east Uganda. Ibid. no. 8.
DAWSON, J. B., 1962. The geology of Oldoinyo Lengai. Bull. vole. 24, 249.
DEER, W. A., HOWIE, R. A., & ZUSSMAN, J., 1962. Rock-forming minerals. Vol. I. Longmans.
DuBois, C. G. B., 1959. The Toror Hills alkaline complex, central Karamoja, Uganda. Congr. Geol.
Intern., Mexico, 1956, Rept. 303.
ECKERMANN, H. VON, 1948. The alkaline district of Alno Island. Sver. geol. undersok. 36.
EITEL, W., 1954. The physical chemistry of the silicates. Univ. Chicago Press.
GUEST, N. J., 1963. Natro-carbonatite from Oldoinyo L'Engai, Tanganyika. Proc. Geol. Soc. no.
1606, 54.
GUMMER, W. K., 1943. The system CaSiO3-CaAI2Si2O?-NaAISiO4. / . Geol. 51, 503-30.
HAMILTON, E. I., 1964. Isotopic composition of strontium in carbonatite. Nature, 201, 599.
& DEANS, T., 1963. Isotopic composition of strontium in carbonatite. Ibid. 198, 776-7.
100 B. C. KING—ALKALINE IGNEOUS SUITES
HYTONEN, K., 1959. On the petrology and mineralogy of some alkaline volcanic rocks of Toror Hills,
Mt. Moroto and Morulinga in Karamoja, northeastern Uganda. Bull. Comm. ge'ol. Finlande,
184, 75-131.
JAMES, T. C , 1958. Summary of silicate rocks associated with carbonatite bodies in Tanganyika. Ceol.
Surv. Tanganyika (unpublished report).
JOHNSON, R. L., 1959. The geology of the Dorowa and Shawa carbonatites, Buhera District, Southern
Rhodesia. Ph.D. thesis, Univ. Leeds.
KING, B. C , 1949. The Napak area of Karamoja, Uganda. Mem. Geol. Surv. Uganda, no. 5.
& SUTHERLAND, D. S., 1960. Alkaline rocks of eastern and southern Africa. Sci. Progress, 48,
298, 504, 709 (3 parts).
1964. The carbonatite complexes of eastern Uganda, in The Carbonatites (editors TUTTLE,
O. F., & GITTINS, J.). Wiley and Sons.
KOSTER VAN GROOS, A. F., & WYLUE, P. J., 1963. Experimental data bearing on the role of liquid
immiscibility in the genesis of carbonatites. Nature, 199, 801-2.

Downloaded from http://petrology.oxfordjournals.org/ at University of York on August 14, 2014


KUSHIRO, I., & KUNO, H., 1963. Origin of primary basalt magmas and classification of basaltic rocks.
J. Petrol. 4, 75-89.
MACDONALD, R., 1961. Explanation of the geology of Sheet 36 (Nabilatuk). Geol. Surv. Uganda, Rep.
no. 5.
SHAND, S. J., 1933. The lavas of Mauritius. Quart. J. Geol. Soc. Lond. 89, 1-13.
SMITH, J. V., 1953. Re-examination of the crystal structure of melilite. Amer. Min. 38, 643-61.
SPIVAK, J., 1944. The system NaAlSiO4-CaSiO3—Na2SiO3. / . Geol. 52, 24-52.
TRENDALL, A. F., 1961. Explanation of the geology of Sheet 45 (Kadam). Geol. Surv. Uganda, Rep.
no. 6.
WILLIAMS, C. E., 1952. Carbonatite structure: Tororo Hills, eastern Uganda. Geol. Mag. Lond. 69,
286-92.
WVLLIE, P. J., & TUTTLE, O. F., 1960. Experimental verification for the magmatic origin of carbona-
tites. Rept. Intern. Geol. Congr., XXI Session, Norden, 13.
1962. Carbonatitic lavas. Nature, 194, 1269.
YODER, H; S., & TILLEY, C. E., 1962. Origin of basalt magmas: an experimental study of natural and
synthetic rock systems. J. Petrol. 3, 342-532.

Vous aimerez peut-être aussi