Download d ated 2010

Trans. geo!. Soc. S. Afr., 82 (1979), 109-131
PETROGENESIS OF CALC-ALKALINE METALAVAS IN THE MID-PROTEROZOIC
HAIB VOLCANIC SUBGROUP, LOWER ORANGE RIVER REGION
by
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
D. L. REID
ABSTRACT
Precambrian igneous rocks underlie the mountainous tract that straddles the lower Orange River near
Vioolsdrif. The oldest rocks are lavas and related fragmental rock types of the Haib Volcanic Subgroup.
Total rock isotopic ages reported elsewhere indicate that the volcanics were erupted about 2000 m.y.
ago. The earliest volcanics were predominantly andesitic-rhyolitic, being made up of two distinct components: (I) non-porphyritic rhyolites (?ignimbrites) and related pumice sheets and bedded tuffs, and (2)
a differentiated suite of porphyritic lavas ranging in composition from andesite to rhyolite, with andesite
dominant. Later extrusive activity was characterised by an increased proportion of basalticandesitic and andesitic material, in the form of porphyritic lavas, pyroclastic beds and volcanogenic
sediments.
The differentiated suite of porphyritic lavas follows a high-K calc-alkaline trend, ranging in composition from basaltic andesite, through andesite and dacite, to rhyolite. An attempt has been made to
explain the chemical variation in terms of a stepwise crystal fractionation model. Results for both major
and trace elements are encouraging and indicate that the more mafic lavas of the Haib Volcanic Subgroup (basaltic andesite to dacite) could have been produced by progressive removal of ortho- and
clinopyroxene (augite), plagioclase and titanomagnetite. The porphyritic rhyolites could have been
produced by removal of hornblende, biotite, plagioclase and titanomagnetite from a parental dacite.
The most basic lava (basaltic andesite) is probably not a primary magma and a more basic precursor is
preferred. The Haib basaltic andesites could have been derived from this primary basaltic precursor by
five to ten per cent olivine fractionation.
The early non-porphyritic rhyolites do not appear to be related to the porphyritic lava suite through
fractional crystallisation and may represent a separate acid magma. The nature and source of this
magma is not well-defined at present.
If the crystal fractionation model for the origin of the porphyritic lava suite is accepted, then at least
60 per cent of the Haib Volcanic Subgroup constitutes juvenile addition from the upper mantle. A major
mid-Proterozoic crust-producing event in the lower Orange River region is therefore indicated.
CONTENTS
I.
I I.
III.
IV.
V.
INTRODUCTION
GENERAL DESCRIPTION
METAMORPHISM
ALTERATION . .
GEOCHEMISTRY
A. Variation Trends
B. Trace Elements
VI. PETROGENESIS . . . .
A. Fractional Crystallisation
..
I. Choice of Phenocryst Phases
2. Least Squares Approximations . . . . .
3. Enrichment of Incompatible Trace Elements
4. Compatible Trace Element Behaviour . . .
5. Summary . . . . . . . . . . . . . . .
. .
B Origin of the Non-Porphyritic Rhyolites
C. Magnetite Fractionation and the Calc-Alkaline Series
D. Magma Genesis and Source Characteristics
I. Basaltic Andesite (BA) as a Primary Magma
2. Basaltic Andesite (BA) as a Derivative Magma
VII. DISCUSSION
ACKNOWLEDGMENTS
.......... .
REFERENCES . . . . . . . . . . . . . . . .
I. INTRODUCTION
Precambrian metavolcanic suites (commonly called
greenstones) have recently received much attention because of their alleged importance in recognising ancient
tectonic environments and in estimating the composition
of possible source regions. Formulation of such interpretative models depends critically on estimates of parental
and/or primary magma compositions. Modelling of major
and trace element variation trends may provide constraints on both the composition and number of parental
and primary magmas. Such a quantitative approach has
produced reasonably encouraging results from studies of
Archaean greenstones in southern Africa and Australia
Page
109
110
112
112
113
113
118
118
118
119
121
122
123
125
125
126
126
126
128
128
129
129
(e.g. Bickle et at., 1976; Condie, 1976; Sun and Nesbitt,
1977).
Younger greenstone sequences are also well-preserved
throughout southern Africa; one such Proterozoic (2,0 1,9 Ga, Reid, 1979a) suite of metavolcanics crops out in
the lower Orange River region (Fig. 1) and consists of at
least 8 (XX) metres of mixed mafic to felsic lavas, pyroclastics and volcanogenic sediments (Haib Volcanic Subgroup, Blignault, 1977; Kroner and Blignault, 1977). The
figure of 8 (XX) metres can only be regarded as a minimum
because the base of the volcanic succession is cut by
younger granitic plutons (Vioolsdrif Intrusive Suite, Reid,
t 979b) and the top is unconformably overlain by late Pre-
110
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
9
Nilma ,1ml Karoo sedlmp./Hs. overburden
o
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
-
20 40 60 80 100
km
Figure 1
Locality map of the lower Orange River region. Subdivision of the Precambrian is after Kroner and Blignault (1977). The present study
area is outlined.
cambrian (Nama) and Mesozoic (Karoo) sediments. The
present extent of the Haib Volcanic Subgroup covers an
area of about 2000 km2, but possible correlatives may
occur under the younger sedimentary cover to the' north
and west of the present study area (Fig. I).
Volcanics of the De Hoop Subgroup occurring in the
north-eastern Richtersveld may be correlatives; certainly
the structural trend of these latter volcanics is the same as
that displayed by the Haib Volcanic Subgroup, such that
one suite could be the lateral continuation of the other. [f
such a correlation is correct, then the stratigraphic scheme
proposed by Kroner and Blignault (1977) for the Proterozoic basement underlying the lower Orange River region,
may be simplified. The Orange River Group (as proposed
by these authors) may be considered to consist of a basal
volcanic pile (Haib Volcanic Subgroup = De Hoop Volcanic Subgroup), overlain conformably by sediments of
the Rosyntjieberg Formation. Results of petrographic
(Ritter, 1978) and geochemical investigations of the De
Hoop volcanics indicate that this more simple alternative
scheme is not ruled out on petrological grounds. At present there is no radiometric data on the De Hoop volcanics
that could give any indication of their true age. As a result,
there appears to be more evidence for a correlation between the De Hoop and Haib Subgroups than against, so
the simpler two-fold subdivision of the Orange River
Group is currently preferred.
The Orange River Group forms the country rock into
which the Vioolsdrif batholith was emplaced and taken together, the sedimentary-volcanic-plutonic assemblage
constitutes the lithology and displays the radiometric age
pattern that characterises the Richtersveld Province
(Kroner and Blignault, 1977; Reid, 1979a, b).
II. GENERAL DESCRIPTION
The exposed portion of the Haib Volcanic Subgroup in
the present study area (Fig. 2) consists of acid - intermediate volcanics (rhyolites, dacites, andesites) at the
base, overlain by a sequence of andesites, basaltic andesites and minor felsic volcanics. [n order to highlight this
rough compositional difference between the top and bottom of the volcanic succession, two formations have been
provisionally proposed by Blignault (1977). The felsic
lower part of the succession is included within the Tsams
Formation (r. . .5 000 metres thick), while the more basic
top is included within the Nous Formation (r. . .3 000 metres
thick). The contact between the two formations is arbitrarily taken at the point where basaltic andesitic volcanics
(lava flows or pyroclastic tuffs) first become abundant.
Field mapping of the De Hoop Subgroup in the northeastern Richtersveld by Ritter (1978) indicates that a
similar two-fold subdivision can also be applied to these
volcanics.
No systematic compositional variation (e.g. mafic to felsic cycles) appears to be present within the Haib Volcanic
Subgroup. In fact, rapid thickness changes, local lithologic
complexity and lateral inconsistency characterise the
entire succession. Such variation is common in modern
mixed volcanic piles (Carmichael et al., 1974). Despite
metamorphic recrystallisation and moderate deformation,
individual lava flows, pyroclastic beds, porphyritic textures,
amygdales, flow banding and other primary volcanic features are still preserved. Fragmental volcanic material is
very abundant and probably about 60 per cent of the
Tsams Formation and almost all the felsic material in the
Nous Formation is made up of fragmental rock types.
Clearly a large proportion of the Haib Volcanic Subgroup
METALAVAS IN HA1B VOLCANIC SUBGROUP
1[1
Younger Cover
Diorite -
Granit~ V I S
Basic rocks
(~t[~ift
Nous Fm
~r\?~~1
Tsams Fm
J
N
\1):
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
-\.
~WAATKOP
-.
•
10 KM
~ ~
~
%
Figure 2
General geological map of the present study area. Based on Von Backstrom and De Villiers (1972) and Blignault (1977). VIS = Vioolsdrif
Intrusive Suite, HVG = Haib Volcanic Subgroup.
was the product of explosive subaerial eruptions.
Fragmental rock types exhibit grain sizes from aphanitic
tuffs to very coarse breccias and agglomerates. Metamorphic recrystallisation makes it very difficult to estimate the
original grain size of many fragmental rock types. Weathered outcrops often have individual clasts etched out,
although none can be distinguished in freshly broken
surfaces or thin sections. Size gra.ding, thickness variations
and mantled bedding within the fragmental units serve to
determine the attitude of the volcanic succession. Such
diagnostic features can only be observed in weathered outcrops. Fine-grained tuffs and volcanogenic sediments are
almost indistinguishable from aphyric lavas. However,
careful examination of outcrops often results in the discovery of faint bedding in most fragmental deposits. Apart
from acid varieties, aphyric lavas seem to be rare in the
Haib Volcanic Subgroup.
The most mafic lavas are basaltic andesite, containing
dark pseudomorphed mafic phenocrysts and subordinate
plagioclase set in a very fine-grained groundmass. No fresh
mafic phenocrysts were ever encountered, although the
nature of the secondary products and crystal outline
suggest that clinopyroxene and an Fe-Ti oxide (probably
titanomagnetite) were probably present. Other possible
mafic silicate phenocrysts include orthopyroxene and/or
amphibole. Plagioclase phenocrysts are invariably turbid,
with reconstitution to fine-grained saussurite (epidote and
GfOl 8211 -
Ml
sericite) well advanced. The groundmass is now a granoblastic mosaic of fine-grained secondary minerals, including
quartz, epidote, albite, chlorite, biotite and actinolite.
Andesites differ from basaltic andesites in having plagioclase as the dominant phenocryst phase. Pseudomorphed
mafic silicate phenocrysts and titanomagnetite microphenocrysts are subordinate. There is complete gradation
from basaltic andesite to andesite, which involves the progressive change in relative proportions of mafic silicate
phenocrysts to plagioclase.
Dacites are lighter coloured than the andesites but also
contain plagioclase as the dominant phenocryst phase.
The most abundant secondary mineral replacing original
mafic silicates is biotite. Microphenocrysts of titanomagnetite, quartz and, less commonly, apatite also occur.
The appearance of phyric quartz marks the transition from
andesite to dacite. As in the case of the basaltic andesite,
there is a complete gradation in petrographic features
from andesite to dacite.
Porphyritic rhyolites with conspicuous phenocrysts of
plagioclase, quartz and altered biotite may be distinguished from more abundant non-porphyritic rhyolites. A
significant proportion of the non-porphyritic rhyolites are
probably ignimbrites, since they form extensive sheets that
can be followed for many kilometres. Porphyritic rhyolites
are laterally impersistent units, except .where they
coalesce to form large dome complexes.
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
112
III. METAMORPHISM
The original igneous mineralogy of the lava~ has be~n
replaced by a metamorphic asse~blage. Basaltic andesite
metalavas (the only lavas whose mIneralogy has been st~d­
ied in detail) contain the assemblage quartz-alblteepidote-chlorite-actinolite-biotite, which i~ diagn~stic. of
low-grade (greenschist facies) metamorphism (Miyashiro,
1973). Two interesting metamorphic reactions are worthy
of note: the decomposition of titanomagnetite phenocrysts
and marked compositional variation exhibited by metamorphic amphibole.
. .
Titanomagnetite microphenocrysts have been oXldlsed
to an aggregate of magnetite and ilmenite, the latter forming intersecting plates. Further reacti~n involves ~he
formation of sphene after ilmenite, but with the replacIng
mineral retaining the ~ross-hatched distribution. Pseudomorphs after titanomagnetite can often be recognised
from the unusual arrangement of secondary sphene.
Metamorphic amphibole is actually a complex intergrowth of actinolite and blue-green horn?lende. Figur~~ 3
and 4 summarise the wide range in amphibole composition
found in the metabasaltic andesites of the Haib Volcanic
Subgroup. The transition from actinolite to blue-gre~n
hornblende involves a deepening in colour, increases In
A1 20 3 , Ti0 2, Na 20 + K20 and FeO*/MgO (F~O* = ~otal Fe
as FeO). The range may be found in one thIn section and
does not reflect any systematic regional or bulk chemical
variation. EMP analyses of the various phases (light and
dark coloured) do not reveal any distinct compositional
populations that may be expected if there is a miscibility
gap (see Fig. 3). Grapes and Gr~ha~ (1978) have re~ently
reviewed the problem of actInolIte-hornblende Intergrowths in metabasites and conclude that the association
is a function of incomplete reaction in the passage from
low- to medium-grade metamorphism.
IV. ALTERATION
A potential problem that plagues geochemical studies of
Precambrian greenstones is the effect of chemical alteration, which may have accompanied metamorphism or
later intrusive activity. As a prelude to a discussion of the
geochemical data, it is pertinent to note that the Haib Volcanic Subgroup has been folded, metamorphosed, .locally
sheared, intruded and exposed to at least three perIods of
.4
.2
... 1.53
...
U,
FeOjlMgO
1.0
...
.6
...
..... ...
... ...
......
... ... ...
4
I
12
Figure 3
Variation in Ti0 2 and FeO*/MgO with AIP3 in metamorphic
amphiboles in the Haib basaltic andesites.
3
2.5
8
Richterite
Actinolite
Ca+Na+K
• t
ACTINOLITES
~
.......
...
...
Edenite
HORNBLENDES
Si
6 Tschermilkite
P.rgilsite
Figure 4
. , .
Variation in Si with Ca + Na + K in metamorphic amphiboles In
the Haib basaltic andesites. Atomic proportions calculated on the
basis of 23 oxygens. Boundary between actinolite and hornblende
is after Miyashiro (1973).
weathering. It is, therefore, difficult to avoid ~he concl~­
sion that a significant proportion of the Halb VolcaniC
Subgroup has had ample opportunity to be affected by a
variety of alteration processes. The aim of the present
study was more concerned with avoiding alt~red ro~ks,
rather than to monitor chemical changes dUrIng varIOUS
processes of alteration.
Apart from the effects of present-day sub-aerial weathering and the greenschist facies metamorphism, two types
of alteration can be recognised: (I) alteration associated
with sulphide mineralisation (confined to the Haib copper
prospect at Tsams) and (2) localised metasomatism within
shear zones. There is, however, abundant field evidence
for widespread alteration which is not obviously related to
sulphide mineralisation or shear zones. This alteration is in
the form of diffuse, pale-coloured zones immediately surrounding veins of silica, carbonate, albite, chlorite and
epidote. The veins have sometimes developed such a
closely spaced intersecting network that it is impossible to
secure "fresh" volcanic rock from some outcrops. Veins
containing the aforementioned minerals vary in width
from 10 cm down to sub-millimetre dimensions. Inspection of many "fresh" looking specimens revealed the presence of microscopic veinlets. Alteration associated with
the veins is commonly replacement of the country rock
by the minerals in the vein, resulting in epidote-quartz,
heavily carbonated, silicified and chloritised rocks.
Fragmental rock types are particularly susceptible to
alteration, especially epidotisation, and characteristically
occur as pale green, hard, brittle, fine-grained rocks showing relict bedding. Flow boundaries are also heavily
altered, probably reflecting greater permeability, since it
has been demonstrated by Smith (1969) and Jolly and
Smith (1972), that hydrous fl uids are the controlling agents
in most processes of alteration .
The main criterion used for selecting samples for analytical work was based on appearance in hand specimen
and under the microscope. Altered samples from near the
Haib copper prospect were easily avoided because of their
bleached appearance and the presence of disseminated
pyrite replacing the primary mafic minerals. Any rocks
displaying penetrative schistosity or foliation were
rejected, as were hard, unweathered, unfoliated rocks
containing conspicuous veins, pods or stringers of epidote,
113
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
METALAVAS IN HAIB VOLCANIC SUBGROUP
silica, carbonate or chlorite. Samples that passed through
all the visual and microscopic tests were hard, unweathered, unfoliated rocks containing very few or no
veins, but which were composed of metamorphic and relict
igneous minerals.
The majority of samples have less than two per cent H 2 Q+
(water lost above 110°C), which is lower than the arbitrary
cut-off value (2,5 per cent) used by -Canadian workers (e.g. Irvine and Baragar, 1971) for their studies of Canadian greenstones. Rocks containing H2 Q+ greater than 2,5 per cent were
considered to be more prone to alteration. Those samples with
less than 2,5 per cent H2 Q+ were subsequently recalculated to
100 per cent volatile free, apparently in the belief that at this
level, water had been merely added, rather than being a
replacement component. In view of the comparable H 2Q+
contents in the Haib Volcanic Subgroup lava samples
analysed, the visual and microscopic methods employed in
this study are satisfactory, at least for avoiding excessively
hydrated samples.
It was decided to adopt the method of Le Maitre (1976)
to estimate Fe 20/FeO. Predicted values for this ratio in all
rocks were found to be high (~ 0,6) but which were consistent with the ubiquitous presence of phyric magnetite.
Sr and Pb isotopic data reported in Reid (1979a) conform to geologically meaningful trends and indicate that
the Rb-Sr, Th-Pb and Pb-Pb systems in the rocks finally
analysed were not affected by later alteration. The U-Pb
system, however, shows evidence of recent disturbance
and probably involves the recent removal of U by surface
weathering and/or ground-water leaching.
To summarise, the Haib Volcanic Subgroup lavas have
not entirely escaped alteration but the effects associated
with weathering, hydrothermal activity, contact metasomatism and shearing, can be avoided by careful visual
and microscopic examination. Samples remaining after
these tests were only a small fraction of the total number
collected. The Fe 20/FeO ratio needed to be adjusted
because the ubiquitous presence of secondary epidote and
magnetite-ilmenite aggregates after phyric titanomagnetite, pointed to significant post-consolidation oxidation. Investigation of other components such as water content, Sr
and Pb isotopes, resulted in a few more samples being rejected, but highlighted the need to concentrate on average
trends rather than individual compositions.
is well illustrated in a plot of AI 20 3 versus OJ. (Fig. 7)
where the average trend for the Haib Volcanic Subgroup is
compared with other calc-alkaline trends. A well-defined
maximum at OJ. = 60 (andesite) contrasts with the progressive decrease exhibited by the average Cascade trend.
Maxima in the AI 2 0 3 versus 0.1. plot are not unknown and a
number of calc-alkaline igneous suites (e.g. Borrowdale and
Salina) showing an early rise and subsequent fall in Al 2 0 3 with
differentiation are also shown in Fig. 7.
Another important major element feature is the relatively high K 20 contents of the Haib Volcanic Subgroup
lavas. Figure 8 is a compilation of average K20 versus Si02
trends for many' volcanic provinces in the circum-Pacific
region (after Gill, 1970). Within the continuous spectrum
of average trends, the Haib Volcanic Subgroup lavas show
affinity with so-called high-K calc-alkaline or some shoshonitic suites. There is considerable disagreement as to
the distinction between high-K calc-alkaline and shoshonitic trends and a number of arbitrary subdivisions have
been suggested (Taylor, 1969; Taylor et al., 1969a; Jakes
and White, 1972a; Jakes and Gill, 1972; Mackenzie and
Chappell, 1972; Peccerillo and Taylor, 1976). The trend
shown by the Haib Volcanic Subgroup relative to the
others is more important than what name is finally used,
but for purposes of reference, the Haib Volcanic Subgroup
,
.'
'.
1.0
70
60
. ,,:1
•• ' fl'
.....
.S
"
'!
..
18
.. '
:
:
•~ e.
'
..
.~
H
......
CaO
'; .
.'
~
.
.' .".
'
-.', ::.
..... .1
K2 0
..
.:.
~
"
."".
Na20
.'
f •
~
.
2
."
.. :-.
MgO
~
..
.... ,
.'
GEOCHEMISTRY
Individual chemical analyses are listed in Reid (1977)
and all major oxide data have been recalculated to lOOper
cent volatile free with Fe 20/FeO adjusted, prior to interpretation. CIPW norms calculated from the recast analyses were used to determine Oifferentiation Index (0.1.,
Thornton and Tuttle, 1960).
GEOL 82/1
!'t
'
18
.
FeO·
"1203
V.
A. Variation Trends
Major element variation is illustrated in Fig. 5, where all
oxides are plotted against OJ. The close correlation between Si0 2 and 0.1. indicates that the major element variation trends resemble the conventional Harker diagrams.
The salient feature of Fig. 5 is that there is a regular variation of all major oxides with degree of differentiation.
Clearly the petrographic gradation from basaltic andesite
to dacite mentioned earlier, is accompanied by a serial
variation in chemical composition. Subsequent variation
diagrams display average trends rather than individual
data, in order to emphasise comparisons. Average compositions are listed in Table I and have been compiled by
subdividing the lava suite using the scheme of Peccerillo
and Taylor (1976).
The calc-alkaline nature of the lava suite is shown by the
lack of marked iron enrichment in the AFM diagram (Fig.
6) and an alkali-lime index of 61. However, the suite does
not exhibit the high Al 20 3 contents so characteristic of
modern calc-alkaline suites (Jakes and White, I972a). This
... .
.. .,::.-
."
.... '! ..
"
,
"
..
t
•0\
:
MnO
.'.'
..
:
....,.
.
:'
!
P20S
i
.3
.: .
DI
40
..
.'
:
J
D1
80
80
40
60
80
.i
Figure 5
Plot of all major oxides (volatile free) against OJ. (Differentiation
Index).
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
114
.
T~llI
Average Composition of Lavas Within the Haib Volcanic Subgroup. Major Oxides have bee!, Recalculated on a VolatIle
Free Basis, with Fe 2 0/FeO Adjusted According to the Scheme of Le MaItre (1976)
b
Rhyolite
Rhyolite a
Dacite
Andesite
Basaltic
andesite
Majors (wt.%)
Si02
Ti02
AI2 0 3
Fe 2 0 3
FeO
MnO
MgO
CaO
Nap
KP
P2 O,
Traces (ppm)
Ba
Rb
Zr
Nb
Th
Pb
Y
U
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
Sr
Ni
Co
Cr
V
Sc
Zn
Cu
c.I.P.W. norms (wt.%)
Qz
C
Or
Ab
An
DiWo
DiEn
DiFs
! Di
HyEn
HyFs
!Hy
II
Mt
Ap
D.1.
N - 5
N = 18
N = 18
N = 3
N=IO
54,95
0,73
14,94
3,40
5,49
0,16
8,09
8,29
1,87
1,86
0,23
59,10
0,74
16,10
3,11
4,04
0,12
4,90
6,35
2,59
2,74
0,23
65,68
0,59
15,73
2,32
2,40
0,09
2,26
4,16
2,81
3,82
0,16
70,49
0,41
14,83
1,40
1,16
0,04
0,97
2,60
3,34
4,82
0,09
74,66
0,35
13,46
1,01
0,77
0,04
0,39
0,96
3,33
5,03
0,10
739
61
116
4
7
18
17
0,8
471
110
42
344
184
28
82
74
107O
153
175
10
18
34
21
3,4
406
20
16
44
87
14
69
97
959
98
144
6
9
23
17
1,4
531
58
29
144
151
17
78
62
1468
227
305
15
21
32
39
3
128
985
218
188
13
23
30
20
5
246
9
3
18
37
7
33
19
9
39
23
7,60
12,75
22,31
25,88
10,97
15,84
26,90
5,38
3,75
1,19
10,32
16,40
5,19
21,59
1,38
4,93
0,50
34,4
16,20
21,92
24,20
2,49
1,74
0,54
4,77
10,46
3,30
13,76
1,41
4,52
0,50
50,9
22,60
23,76
19,03
0,29
0,21
0,05
Q,55
5,42
1,43
6,85
1,12
3,47
0,34
68,7
28,50
28,27
11,24
0,48
0,37
0,05
0,90
2,05
0,33
2,38
0,77
2,04
0,19
82,7
33,55
1,00
29,73
28,16
14,94
0,97
0,08
1,05
0,67
1,46
0,22
91,4
a: Porphyritic variety: b: Non-porphyritic variety
F
F
AL------.........JL.-----....I M
A
L - . -_ _ _ _----lt._ _ _ _ _......1
M
Figure 6
A (N~9 + ~C?)- F(FeO*)-:- M(M~O) plot. Typ~cal calc-alkaline trend is that for the Cascade.s (after qtr.m.ichael et at., 1974, open circles); typical
tholeIItIc ~rend IS that for. ThJngI?~h (after Car!1'lIchael, 1964,. open sq~ares). The average Halb trend GOlnlng closed triangles) closely follows the
calc-alkaline Cascades SUite. IndiVidual data pOints for the Halb Volcamc Subgroup have been plotted separately in the right hand diagram in order
to show that the average trend does indeed follow that defined by all the data.
115
METALAVAS IN HAIB VOLCANIC SUBGROUP
cal data available, Kuno (1968) and Brown (1967) defined
three fields in the total alkalis-silica diagram (Fig. 9) that
correspond to: tholeiitic, calc-alkaline and alkaline series.
Thus the calc-alkaline series is characterised by low iron
enrichment, moderate K,O contents and moderate total
alkali content. However, the intermediate field of the total
alkalis-silica plot is not exclusively occupied by calc-alkaline trends. For example, the Thingmuli lavas, shown previously as a typical tholeiitic suite (see Fig. 6), exhibit an
..•..
8
6
40
60
80
4
Figure 7
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
Plot of average AI,O, against D.1. for a number of volcanic suites.
Typical calc-alkaline = Cascades (open circles); typical tholeiite
= Thingmuli (open squares) and Red Hill dolerite-granophyre
(McDougall, 1%2, open triangles); other calc-alkaline suites =
Borrowdale (Fitton, 1972, closed squares) and Salina (Keller,
1974, closed circles). The Haib trend joins the closed triangles.
lavas are considered to display a high-K calc-alkaline
trend.
Kuno (1966) has demonstrated that a systematic increase in total alkalis at constant silica exists in the Cenozoic lavas of Japan as they are traced across the island arc
from the oceanic to the continental side. With the chemi-
2
50
60
Figure 9
Average (Na,O + K,O}-SiO, plot for the Haib lavas (closed
triangles), together with a typical calc-alkaline (Cascades, open
circles) and tholeiitic (Thingmuli, open squares) suites. The three
fields are those defined by Kuno (1968) and Brown (1967). A = alkaline series, CA = calc-alkaline series, TH = tholeiitic series.
Note that all three suites plot in the CA field and therefore cannot
be readily distinguished in this type of plot.
12
4
2
FigureS
Comparison between the Haib average K,O-SiO, trend (H) and other volcanic provinces. Trends 1-18 are for Cenozoic suites in the
circum-Pacific region (compilation in Gill, 1970). 19 = shoshonitic suite of Fiji (Gill, 1970); 20 = Barby shoshonitic suite (Watters, 1974);
21 = Salina (Keller, 1974); 22 = Borrowdale (Fitton, 1972).
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
116
80
Si0 2
70
60
~
50
·
,·
•
•·
·
··.. ··
·
· ·· ·
·
·· .
I
AI 2 0 3
16
~
14
~
I
.5
Fe 2 03
r5
4
2
~
·
· · · ·. ·
.·· ···.s'.:.
. ·· ·
.
. .·
Na 0
. ·
. · ·,. .
. . . .·.· .·· ·
MgO.
~10
~5
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
K 20
2
~
4 r
2
I
I
I
I
20
40
60
r6
-4
r2
I
80
01
20
:'
I
I
I
I
6
.
·· ·· ·
·.· ·.·.....
..
L
I
I
I
CaO
6
I
I
I
I
I
I
· .. · · .
·. ,.., .
· · ·.... ·
~10
1
8
I
T
I
I
·. ··. .·
·,
·
.
.
· ··
·
· · · ··. ·
12
·
.·. ··· ·'."'.
.
.
· • .· · ·
-1.0
I
..
I
18
Ti0 2
I
·
· ·. .
·
. · ·.,.
·· ··.:
· · · .
·
I
I
40
60
·
I
1
I
·
·
80
01
Figure 10
Major oxide composition of fragmental rock types (pyroclastics, volcanogenic sediments).
average trend that lies entirely within the calc-alkaline
field of Fig. 9. Tholeiitic and calc-alkaline volcanic provinces may not be readily distinguished on a total alkalissilica diagram. While the total alkalis-silica variation exhibited by the Haib Volcanic Subgroup lavas is consistent
with their calc-alkaline affinities, this feature may not be
invoked as independent evidence alone.
The uniformly low Ti0 2 contents (less than 0,8 per cent)
of all the Haib Volcanic Subgroup lavas (see Fig. 5) is also
consistent with their calc-alkaline affinity (Jakes and
White, I972a; Carmichael et al., 1974). As in the case of
total alkalis, Ti0 2 contents cannot be invoked as independent evidence, since Engel et al. (1965) report modern
oceanic basalts with low Ti0 2 (less than one per cent).
Fragmental rock types (pyroclastics, volcanogenic sediments) are essentially mechanical mixtures of the various
lavas described previously. The wide range in lava compositions preclude any simple mixing models, because of the
great variety of possible end members. In addition, the
permeable nature of the original volcanic debris has facilitated the migration of fluids, resulting in widespread alter-
ation. Chemical variation caused by alteration has, therefore,
been superimposed on that brought about by mechanical
mixing. The variation diagrams in Fig. 10 illustrate the
wide range in major element composition of the fragmental rock types. The range is of the same order as that exhibited by the lavas, but shows more scatter, which suggests
that alteration has probably played a prominent part in
producing the present compositions. One interesting feature of Fig. 10 is the lack of correlation between K20 and
0.1., despite the fact that 0.1. is strongly influenced by the
amount of normative orthoclase. Acidic fragmental rock
types (0.1. = 80) have similar K20 contents to some basaltic andesite lavas (0.1. = 40). In view of the absolute abundances involved (K 20 = 1,2 per cent), the high 0.1. fragmental rocks appear to have lost K20 by some leaching
process during alteration. The prevalence of high 0.1., low
K20 fragmental rock types in the Haib Volcanic Subgroup, suggest that these chemical peculiarities may be
use!ul .in identifying such rocks when metamorphic recrystalhsatton has rendered them indistinguishable from true
lavas.
Co
Ni
100 ••
.. .,
• e'.
A
,
...
.'
c7
'... •• f.'
~
'0
.,
'
.
E
0
~.
~.
~ . : ...
.. .
. ...
«)00
E
..
00
Rb
Ba
.. .. .,
",'
~.,
•
A
'.
• ,0: "
.:
)
0
e
•••
.:"
'0
• '0
•
,
...
,
•
E
•
000
~.'
•
'
,)
.."
.
".
'0'
0
'00
E
e
... . F ..
8
A
Cr
V
!
A
.'
•
'00
A
Zr
'If :-- ••
,00
<!
, i '.
i
00
...
Nb
",\Q...
:'~
.....
......
'0
~
100A'
'F
.
-•
o'c .;.:~=
.. . ....
~
A
..
~
0
~
.' , - e
.~
00
E
..
°0
.~
:.:
.~
. .,'
~
m
-l
»
I
e
B'
:
F
r
»
<
»
,,~ ..!. .. - ...f ,.,
V>
Z
..... f ...
J:
Pb
Se
B
·
00
e
-...
• " ..
Zn
'00
~
.
<0
,00
0
J
J
D1
.'
,
60
duced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
A
~~
•
Th
.. ,:. .... ,
,...
,: .
.. :' ,. -. ..
10
F
'
,"
"
~
e
,
'0
E
A
•
Sr
.. . •.. ·11.
.'
A
'.
'00 •
'.
,
80
o·
~
'"
.. •
<.
.
.'
80
..
e'
I
., .- .. e
...
'"
<
0
r
n
»
z
()
V>
c
•
'"
0
'c0"
A
"
Y
.Il ' •
~
\
'
-'
..
~ .' ~
FO":
'0
... :-: 'oa'•...•
..
."
'0: . ,
..
DI
D1
<0
.
. '. .... .
'. '.
'. ,.'
•• q
•
d
000
'
D1
60
'.
~
..
.'
.'. ,
'
•
A
Cu
.,.
0
e
..... 0..
Ii
.': .
•
·.
0
B
.
U
.....
;>:
80
80
<0
Figure II
Variation of trace element composition with D.l. in the Haib lavas. Average compositions are from Taylor (1969); A =- high alumina basalt. B = low Si andesite, C
andesite. D = high K andesite, E = dacite, F = rhyolite . NOie that the trace element abundances are plotted on a log scale. DL = Detection limit.
60
~
=
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
118
and Ringwood, 1968). Partial fusion alone is not c~~sid­
ered to be responsible for the range in lava composltlo~s
encountered in the Haib Volcanic Subgroup, because thiS
implies that every lava constitutes a separate primary. ~elt,
which is considered unlikely. If the presence of a limited
number of primary magmas is conceded, then the. entir~
spectrum of lavas may be produced by the superimpOSItion of the other two processes (fractionation and contamination) on the partial melting process. .,
.
Porphyritic lavas ranging in compoSItion from basaltiC
andesite to rhyolite have isotopic compositions that fall on
well-defined Rb-Sr, Th-Pb and Pb/Pb isochrons, all of
which yield the same age (Reid, 1979a). A common initial
Sr and Pb isotopic composition is, therefore, indicated.
The operation of processes such as magma mixing and/or
wall rock assimilation are, therefore, ruled out, except in
the unlikely case where all mixing end members (liquid or
solid) are isotopically indistinguishable. Much of the evidence for magma mixing is petrographic (e.g. incompatible phenocryst assemblages, intermingling of two or
more different glasses) and is, therefore, susceptible to
being obscured by metamorphism. Certain lavas within
the Haib Volcanic Subgroup may have been the result of
magma mixing, but the badly preserved state of all the
rocks preclude a positive identification of such a petrogenetic process and thus it will not be further discussed.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
B. Trace Elements
Figure II illustrates the variation of 15 trace elements
with 0.1. A log scale has been used in order to accommodate the wide range in abundances displayed by the trace
elements. Ba, Rb, Zr, Nb, Th, U, Pb and Y show positive
correlations, while Sr, Ni, Co, Cr, V and Zn show negative
correlations with O.l.
Ba, Rb, Pb, Th and U in the Haib Volcanic Subgroup
lavas are enriched relative to "average" calc-alkaline lavas
(as estimated by Taylor, 1969; lakes and White, I972a),
but are similar to high-K calc-alkaline suites (see Fig. II).
The suspected affinity of the Haib Volcanic Subgroup
lavas with high-K suites is, therefore, confirmed by the observed abundances of the above trace elements.
Compared with modern calc-alkaline igneous rocks, the
mafic to intermediate lavas of the Haib Volcanic Subgroup
contain higher Ni and Cr contents. Taylor (1969) has highlighted the relatively low Ni contents in calc-alkaline lavas
(commonly less than 30 ppm). Several exceptions have
been subsequently described (e.g. Hedge, 1971; Zielinski
and Lipman, 1976; Noble et al., 1975), so the Haib Volcanic Subgroup mafic to intermediate lavas cannot be
regarded as a unique high Ni calc-alkaline suite. Similar
high levels of Ni ('"'-1100 ppm) and Cr (200-400 ppm) have
been reported in calc-alkaline greenstones of Rhodesia
(Hawkesworth and O'Nions, 1977) and Western Australia
(Hallberg et al., 1976). Baragar and Goodwin (1969) report
high Ni contents in "andesite" from the numerous Precambrian greenstone belts of the Canadian shield. These
authors do state, however, that these intermediate lavas
are transitional to theoleiites and, therefore, may not be
strictly comparable.
A. Fractional Crystallisation
The reasonably coherent major and trace element variation trends, together with the results of the Sr and Pb
isotope study, suggest that closed system fractional
crystallisation may be a viable process to produce the lava
suite. A fundamental assumption upon which many
models involving fractional crystallisation are based, is
that the trends shown in Figs. 5 and II represent a liquid
line of descent. It is quite likely that many porphyritic
lavas in the Haib Volcanic Subgroup are phenocryst enriched, with the result that their bulk compositions fall off
the liquid line of descent. If a unique liquid line of descent
does in fact exist, the effects of phenocryst accumulation
may be overcome if a large number of samples are analysed, because the effects will probably appear as scatter
VI. PETROGENESIS
Eruption of lavas which vary from basaltic andesite to
rhyolite implies the existence of magmas exhibiting the
appropriate compositional range. Fundamental processes
responsible for variation in magma composition include
fractional crystallisation, partial melting and contamination. Separating these processes is probably artificial,
since magmas are initially generated by partial fusion and
possibly experience variable amounts of fractionation and
contamination during their ascent (O'Hara, 1968; Green
TABLE II
Estimated Liquid Compositions Used in the Petrogenetic Modelling Study
BA
BAI
A
0
R
NR
K2 0
P2 0 5
54,6
0,70
14,4
9,3
0,17
8,5
9,0
1,8
1,3
0,21
58,0
0,80
15,8
7,5
0,14
5,8
7,0
2,3
2,4
0,23
62,6
0,68
16,2
5,6
0,11
3,3
5,2
2,7
3,4
0,21
65,8
0,60
15,9
4,4
0,10
2,10
4,.Q
3,0
3,9
0,18
71,1
0,29
14,5
2,3
0,04
1,1
2,7
2,9
4,8
0,08
74,2
0,34
13,7
1,3
0,03
0,28
0,71
3,4
5,7
0,03
OJ.
29,7
40,0
60,7
70,0
81,4
93,2
Ba
Rb
Zr
Nb
Th
U
Pb
Y
Sr
Ni
Co
Cr
719
40
109
3,2
4,9
0,7
15,3
17,6
450
125
47
420
200
30
92
270
15,0
830
82
132
5,0
8,4
1,0
22
18,1
525
73
36
165
180
20
86
243
24,0
960
128
152
7,3
13,1
2,3
30
19,4
485
35
21
63
123
15
72
221
29,4
1 113
154
167
8,9
16,2
2,9
26
21
390
20
985
218
188
12,9
18,6
4,0
29
22
246
8,0
2,7
18
40
6
33
183
40,5
1468
227
305
15,4
21
2,9
32
39
128
Si02
Ti0 2
Al2 0 3
FeO
MnO
MgO
CaO
N~O
V
Sc
Zn
K/Rb
K/Ba
II
35
74
10
58
210
29,1
9
39
208
32,2
119
METALAVAS IN HAIB VOLCANIC SUBGROUP
about average trend lines. For the purposes of this study,
the average trends displayed in Figs. 5 and II have been
modelled rather than the composition of any particular
lava.
Average trend lines have been fitted through the major
and trace element plots and estimated liquid compositions
have been read off at predetermined intervals. The
average trends were first estimated visually and then
reproduced either by first order (linear) or second order
polynomial curves. Table II lists the estimated liquid compositions which correspond to the following magma types:
(i) BA
Basaltic andesite (most basic magma)
(ii) BAI
Basaltic andesite (maximum Ti0 2 content)
(iii) A
Andesite (maximum Al 20 3 content)
(iv) 0
Dacite (arbitrarily chosen at 0.1. = 70)
(v) R
Rhyolite (porphyritic variety, 0.1. = 80--93)
(vi) NR
Rhyolite (non-porphyritic variety with highest
0.1.> 90).
The model evaluated below involves the stepwise derivation of the magma types from parental basaltic andesite
(BA). The status of the parental magma (i.e. primary or
derivative) will be examined later. The fractional crystallisation model has been subdivided into a series of steps
(BA-BAI, BAl-A and so on), each of which have been
evaluated separately. There have been a variety of
approaches adopted towards petrogenetic modelling in
recent literature (e.g. only major elements, Erikson, 1977;
only trace elements, Allegre et al., 1977; major and trace
elements, Ewart et al., 1973). Ideally the major elements
I. Choice of Phenocryst Phases
Inspection of thin sections of basaltic andesite reveals
the presence of plagioclase and magnetite as subordinate
phenocryst phases. The dominant phenocrysts are of some
mafic silicate (or silicates) which are now replaced mainly
by actinolite and chlorite. The possible identity of this
phase (or phases) may be established by inspection of Fig.
12, which summarises the change in major element
composition in the passage from basaltic andesite (BA) to
andesite (A).
Common phenocryst phases in modern, fresh basaltic
andesites and andesites (in addition to plagioclase and
magnetite) include olivine, orthopyroxene, clinopyroxene
(augite) and sometimes hornblende. Olivine is probably
not present because of the fairly strong quartz normative
nature of the Haib basaltic andesites and the absence of distinctive alteration products (e.g. serpentine, bowlingite,
iddingsite, talc, Mg-Fe amphiboles, secondary magnetite).
Since differentiation is in the direction of increasing silica
content, it is unlikely that olivine would persist for any significant crystallisation interval without reacting with the
plag
AI203
18
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
should be modelled first and subsequently tested for consistency with trace elements. However, the metamorphosed nature of the lava suite has precluded this totally
objective approach and trace element data have had to be
used to obtain information on possible crystallising phases.
As a result, certain sections involve the discussion of both
major and trace elements together.
MgO
opx cpx hbl
OA
10
mt
BA
BA
14
5
opx
OA
plag
opx
CaO
10
FeO'>
10
mt
BA
BA
hbl
5
plag
opx
5
cpx
mt
hbl
cpx
OA
0A
50
60
70
plag ,opx
BA
OA
.5
mt
50
60
70
Figure 12
Plot of average major oxide composition ag~i~st Si0 2 f~r the !l10~e bas!c lavas in the .Haib Volcanic Subgroup (data listed in Table II). Vectors represent the path taken by the compositions of reSidual hqulds (mmeral control hnes) dunng the removal of the appropriate mineral.
120
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
TABLE III
.
Typical Compositions of the Common Low Pressure Phenocryst Phases in Calc-alkaline Lavas. Data Extracted from Ewart (1976a), Jakes and WhIte
(l972b) and Cawthorn (1976)
BlOT
USP
PLAG
MT
OPX
CPX
HBL
I
2
Si02
Ti02
AI 20 3
FeO*
MgO
CaO
53,0
0,2
1,0
17,0
26,8
2,0
N~O
-
53,2
0,2
1,0
19,1
24,4
2,1
-
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
K20
I
2
I
50,0
50,7
(0,2-1,0)
2,0
2,0
13,3
10,5
17,8
15,4
17,9
19,0
0,2
0,2
-
-
46,5
2,0
7,7
16,1
13,7
II,S
1,4
1,1
-
48,9
1,1
8,5
18,6
9,5
11,9
0,9
0,5
surrounding liquid to produce orthopyroxene. If average
compositions of the other common phenocryst phases are
adopted, it is possible to construct control lines which represent the behaviour of magmas affected by the removal of
any of these phases. Table III lists the adopted major
element compositions of the various phenocryst phases,
which have been based on data extracted from Ewart
(l976a) for pyroxenes and magnetite, Cawthorn (1976) and
Jakes and White (l972b) for amphiboles. The plagioclase
composition corresponds to the average obtained from a
limited number of microprobe analyses (range An 70-An'J'
average An6Q).
Comparison with mineral control lines indicates that
simple fractionation of anyone phenocryst phase could
not produce the observed trend (BA-BAI-A) in Fig. 12.
The only possible exception is hornblende. The wide compositional range exhibited by this mineral could probably
result in a fan of control lines that straddle the observed
trend. Suitable choice of AI 20 J, FeO* and MgO contents in
the predicted hornblende could produce control lines that
coincide with the observed trend. However, the CaO content of most volcanic hornblendes does not show similar
variability (Jakes and White, 1972b; Cawthorn, 1976) and
the control line in the Cao versus Si02 plot diverges significantly from the observed trend. It is concluded that simple
hornblende fractionation to produce the derivative magmas is unlikely.
Further inspection of Fig. 12 reveals that the most likely
combination of phenocryst phases that would produce the
observed trend is two pyroxenes and plagioclase. This
combination will satisfy the Al 20 J versus Si02 path and the
coprecipitation of orthopyroxene will offset the depletion
in CaO caused by clinopyroxene-plagioclase fractionation. Another important feature of the major element
variation trend is illustrated in Fig. 13. The calc-alkaline
affinity of the lava suite is clearly displayed by the negligible change in MgO/FeO* in the passage from basaltic andesite to andesite. The mineral control lines plotted in Fig.
MgO
mt-,'Pla
8
g
hbl~
/::' l
oPX,cpx
c:J.
A
4
FeO*
8
2
An60
AnlOO
Ab lOo
-
-
53
-
-
-
30
80
93
-
43,2
36,6
-
68,7
IS
-
-
-
2
10
Figure 13
MgO-FeO* plot of the more basic lavas in the Haib Volcanic
Subgroup, together with mineral controllines.
I
12,5
4,5
-
-
20,2
-
-
19,4
-
11,8
-
-
36
-
64
-
39,0
1,6
16,5
20,5
12,7
-
0,06
9,6
13 indicate that two pyroxen e-plagioclase fractionation
will result in derivative magmas being enriched in FeO*
relative to MgO. In order to produce the observed trend, a
mineral with low MgO/FeO* must accompany the two
pyroxene-plagioclase assemblage. Since magnetite occurs
as a phenocryst phase in the lavas, it is not unreasonable to
conclude that this mineral is responsible for counteracfing
Fe enrichment.
An alternative mineral with low MgO/FeO* is hornblende, but its role is considered to be restricted by the
observed CaO versus Si0 2 trend. Another restriction to
the role of hornblende fractionation is that, compared
with magnetite, a much greater proportion of hornblende
is needed to separate to counteract Fe enrichment. This
would be inconsistent with the initial increase in Ti02 in
the passage from BA to BAL The Ti0 2 content in most
volcanic hornblendes is usually higher than the surrounding magma, hence extensive hornblende fractionation
would deplete the magma in Ti0 2• Since much smaller
amounts of magnetite are needed to counteract Fe enrichment, fractionation of this mineral need not result in Ti0 2
depletion.
The preferred fractional crystallisation model involves
the removal of orthopyroxene, clinopyroxene (augite),
plagioclase and subordinate magnetite from the parental
basaltic andesite (BA) to produce the transitional basaltic
andesite (BAl). Further fractionation of the same minerals may produce the andesite (A), provided magnetite
makes up a slightly greater proportion of the solid
removed. The increased importance of magnetite is necessary to produce the depletion in Ti0 2 and the slight change
in slope of the MgO-FeO* path, in the passage from BAI
to A.
It is clear from the progressive decrease in Al 20 J with
increasing O.I. (Fig. 5), that if the more acid lavas (dacites,
rhyolites) are to be derived from the andesite by fractional
crystallisation, then the process probably involves the
removal of increased amounts of plagioclase. However,
the concomitant depletion in MgO, FeO*, and Ti0 2
necessitates the sustained removal of mafic minerals.
Retention of the same fractionation scheme as proposed
for the more basic lavas would be inconsistent with the
presence of biotite as a major mafic phenocryst phase in
the acid dacites and porphyritic rhyolites.
If the acid dacites and porphyritic rhyolites are to be
derived from more basic magma compositions by fractional crystallisation, then the role of biotite must be taken
into a~count. The presence of biotite as a phenocryst
phase I~:ers that the ~ater pressure was sufficiently high
to stabilIse hydrous mmerals during intratelluric crystallisation. This suggests that hornblende may also have crystallised, P?ss!bly in pla~e of pyroxenes and perhaps also
played a signIficant role 10 the fractionation process.
The possible influence of amphibole and/or biotite
would be reflected in the behaviour of elements such as K,
Rb and Ba. In c~:mtrast with pyroxenes and plagioclase,
removal of amphibole and/or biotite from magmas will
fra~tionate K from Rb and Ba (Gast, 1968; Jakes and
Smith, 1970). Hornblende fractionation would lower the
121
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
METALAVAS IN HAIB VOLCANIC SUBGROUP
KJRb ratio of derivative liquids, while biotite fractionation
would increase the KJBa ratio. Combined hornblendebiotite fractionation would yield derivative liquids with
significantly lower KJRb and higher KJBa.
In the passage from andesite to dacite, KJRb is only
slightly lowered from 221 to 210, but decreases markedly
in passing to the porphyritic rhyolite (183, Table II). KJBa
remains constant from the andesite (29,4) to the dacite
(29,1), but shows a sharp increase in the porphyritic
rhyolite (40,5). These results suggest that hornblende and
biotite may not be dominant fractionating phases in the
passage from the andesite to the dacite and, therefore,
cannot be responsible for the observed depletion in ferromagnesian elements. It must be concluded that other
mafic silicates (e.g. pyroxenes) must be involved in the
derivation of the Haib dacite. On the other hand, hornblende and biotite appear to be important fractionating
phases in the production of the porphyritic rhyolite.
Possible minor fractionating phases include apatite and
zircon, both of which have been observed in thin section,
the latter only in the acid dacites and rhyolites. The progressive depletion of P20 S with differentiation (Fig. 5)
would be consistent with the removal of apatite. However,
the wide scatter of data points in the P20 S versus OJ. plot
precludes any precise treatment of apatite fractionation
and any further discussion is of a qualitative nature.
The porphyritic lava suite displays major element variation that is in qualitative accord with fractional crystallisation. The preferred fractionation scheme to produce the
andesites and dacites involves the progressive removal of
two pyroxenes, plagioclase and magnetite from a basaltic
andesite parent. A profound change in mafic mineralogy is
required to produce the porphyritic rhyolites and a
scheme whereby hornblende, biotite, plagioclase and
possibly magnetite are removed from a parental dacite
is preferred. A quantitative evaluation of the proposed
fractionation schemes is attempted later.
Finally, it must be stressed that the control line method
cannot prove the absence or presence of any particular
phenocryst phase in any lava; rather the method enables
the recognition of major phases which could have fractionated to produce the observed paths. Although the major
element variation trends appear to be inconsistent with
amphibole fractionation, its presence as a minor or nonfractionating phase cannot be ruled out. However, the
passive role of hornblende, as implied by the latter, is
considered unlikely.
2. Least Squares Approximations
Quantitative evaluation of the fractional crystallisation
model has been carried out using techniques devised by
Bryan et al. (1969). Least squares approximations to the
series of derivative magmas are summarised in Table IV.
They involve the removal of variable amounts of the
proposed phenocryst phases in a step-wise fractionation
scheme. Both mineral and total rock compositions have
been normalised to lOOper cent volatile free for the eight
major oxides. MnO and P20 S have been excluded because
of their low concentrations. All Fe has been expressed
as FeO. In order to overcome the problem of zoning in
plagioclase, pure end members (albite and anorthite) were
initially used; final calculations used the predicted plagioclase. A similar procedure was adopted with magnetite
and pure end members (ulvospinel and magnetite) were
initially used.
The quality of agreement is reflected in the sum of the
squares of the differences calculated for each approximation. Inspection of Table IV reveals that the proposed
stepwise fractionation scheme yields predicted major
element compositions that are very similar to those observed.
Both the relative proportions of the phenocryst phases and
the amounts removed from the successive parental magmas are not unrealistic and enhance the feasibility of the
stepwise fractionation model. The predicted plagioclase
compositions are consistent with the limited amount of
data obtained by microprobe analysis. Of particular
importance is the progressive drop in anorthite content of
the predicted plagioclase with differentiation.
The transitional basaltic andesite (BAI) closely resembles the residual liquid after approximately 41 per cent
crystallisation of the parental basaltic andesite (BA). The
dominance of pyroxenes over plagioclase in the solid removed is consistent with the predictions of the control line
diagrams discussed previously. Very small proportions of
magnetite (",5 per cent) are needed to offset Fe-enrichment caused by fractionation of the other phenocryst
phases.
Derivation of a residual liquid very similar to the andesite (A) involves about 32 per cent crystallisation of the
transitional basaltic andesite (BAI). Pyroxenes still domi-
TABLE IV
Least Squares Approximation to the Various Derivative Magmas Discussed in the Stepwise Fractionation Model
Obs. = observed composition (from Table II);
Est. = estimated composition;
X m = coefficients in the linear mixing equation;
Xs = weight fraction of each mineral component in the solid material removed (crystallate);
F = weight fraction of original liquid remaining.
Specific mineral composition used in each approximation are given next the Xs column and correspond to those listed in Table III.
BA-BAI
Si02
Ti02
A120~
FeO
MgO
CaO
N~O
~O
Obs.
Est.
58,00
0,80
15,80
7,50
5,80
7,00
2,30
2,40
58,60
0,80
15,93
7,02
5,37
7,18
2,09
2,21
0,91
I(Obs-Est)2
Xm
BA
OPX
CPX
PLAG
MT
Total
F
1,696
-0,190
-0,222
-0,251
-0,037
0,996
BAI-A
Obs.
62,60
0,68
16,20
5,60
3,30
5,20
2,70
3,40
Xs
0,271
0,317
0,358
0,053
0,590
Xm
(I)
(I)
An68
USP33
BAI 1,462
-0,145
-0,087
-0,203
-0,028
1,001
D-R
A-D
Obs.
Est.
62,20
0,76
16,69
5,28
3,05
5,75
2,43
3,51
0,96
65,80
0,60
15,90
4,40
2,10
4,00
3,00
3,90
Xm
Xs
0,314
0,187
0,429
0,060
0,684
(I)
(I)
AIlro
USPJ6
A 1,253
-0,063
-0,034
-0,141
-0,014
1,002
Obs.
Est.
71,14
0,29
14,54
2,29
1,13
2,66
2,87
4,82
65,76
0,60
15,99
4,40
2,18
4,04
2,78
4,26
0,19
Xm
Xs
0,137
0,250
0,560
0,054
0,798
(2)
(2)
An'2
USPJ3
HBL
BlOT
PLAG
MT
D 1,374
-0,105
-0,064
-0,202
-0,007
0,995
Est.
71,14
0,29
14,62
2,29
1,11
2,63
2,59
4,71
0,10
Xs
0,278 (2)
0,169
0,534 An41
0,020 USP30
0,728
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUfH AFRICA
122
nate over plagioclase in the solid removed and a slight
increase in the proportion of magnetite (from 5 per cent to
6 per cent) is sufficient to change the slope of the Ti0 2
versus Si0 2 path, so that the andesite is depleted in Ti0 2
relative to the transitional basaltic andesite.
Slightly more evolved pyroxene compositions (see Table
III) have been used in the fractionation scheme to produce
the dacite (D), which is in keeping with the well-known
change in mineral composition with differentiation. In
contrast with the previous steps, plagioclase now dominates over pyroxenes in the solid removed, which is
consistent with the progressive depletion in AI 20 3 with
continued differentiation. Derivation of a residual liquid
with a composition very similar to the dacite requires about
20 per cent crystallisation of the andesite.
Fractionation of amphibole and biotite (the latter of
which is known to occur in the rocks), together with plagioclase and magnetite from the dacite (D), may yield a
residual liquid very similar in composition to the porphyritic rhyolite (R), provided about 27 per cent crystallisation
has occurred. Plagioclase again dominates over the mafic
minerals, which is consistent with the progressive decrease
in AI 20 3 in the passage from dacite to rhyolite. It is of
interest to note that both the dacite and porphyritic rhyolite contain phenocrysts of quartz, which have not participated in the fractionation scheme. This is consistent with
the marked increase in quartz phenocrysts (and hence the
Si0 2 content) in the rhyolite relative to more basic lavas.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
3. Enrichment of Incompatible Trace Elements
Fractionation of two pyroxenes, plagioclase and magnetite in the passage from basaltic andesite to dacite suggests
that certain trace elements, because of their very low
crystal;1iquid distribution coefficients, will be preferentially enriched in residual liquids. Such elements have
been commonly referred to as "incompatible" (Ringwood,
1966) or "hygromagmatophile" elements (Allegre et aI.,
1977) and in this case include Rb, Nb, Th, U and Pb. Other
trace elements such as Ba and Zr have crystal;1iquid distribution coefficients that depart significantly from zero for
some of the minerals involved (e.g. Ba in plagioclase, Zr
and Y in c1inopyroxenes), but their average solid;1iquid
distribution coefficients may be effectively zero. Potassium will also behave as an incompatible element, since it
is not a major constituent of the postulated fractionating
minerals. Since
D~rystal/liquid 0,
(1 )
C:/C~ = I/F
r-..J
D~rystal/liquid = crystal;1iquid distribution coefficient for
trace elements i (hereafter Di ).
concentration of element i in residual liquid
C~
= concentration of element i in original liquid
F
= mass fraction of original liquid remaining.
It follows from equation (1) that respective F values for
the steps involving fractionation of two pyroxenes, plagioclase and magnetite (BA-BAI-A-D) may be estimated
from the incompatible elements by calculating the appro-
cl
=
TABLE V
Obsen'ed Degree of Enrichment in Incompatible Trace Element Abundances during Stepwise Fractionation. Included for Comparison are
Values of F Calculated in the Least Squares Approximations
BA-BAI
F
K
Rb
Nb
Th
U
Pb
Ba
Zr
Y
=
0,59
0,54
0,50
0,64
0,58
0,70
0,69
0,87
0,83
0,97
BA1-A
F
=
0,68
0,71
0,64
0,69
0,64
0,44
0,74
0,86
0,87
0,93
A-O
F
=
0,80
0,87
0,83
0,82
0,81
0,79
1,15
0,86
0,9\
0,92
D-R
F
=
0,73
0,81
0,71
0,69
0,87
0,73
0,90
1,13
0,89
0,95
priate ratios. Table V lists the various ratios for each of the
three steps involved.
In the first step (BA-BAI), F values obtained for ~, Rb,
Nb and Th are close to that obtained from the major element least squares approximation (F = 0,59). However,
other elements such as Pb, Ba, Zr and Y display degrees of
enrichment that are significantly smaller than those predicted. Similar results have been obtained for the second
step (BAl-A, F = 0,68), but Ba, Y and Zr are again anomalous. The degree of enrichment in all trace elements
except Pb, Y and Zr in the passage from andesite to dacite
(A-D) is consistent with the F value obtained by the least
squares approximation (F = 0,80). To summarise, the
behaviour of many so-called incompatible trace elements
(and K) is consistent with the proposed fractionation
scheme, but significant exceptions are the behaviour of U,
Pb, Y, Zr and Ba.
The anomalous behaviour of U may, in part, be a function of the wide variation in abundances, as displayed in
Fig. 11. Although the overall trend is reasonably systematic, the wide scatter restricts the accuracy to which U can
be estimated at any particular 0.1. U-Pb isotopic data
discussed in Reid (1979a) was found to be consistent with
geologically recent leaching of U. The wide scatter is
probably caused by partial U removal, during either surface weathering (Barbier and Ranchin, 1969) or groundwater interaction (Rosholt et al., 1973). The ratios listed
for this element in Table V are, therefore, subject to large
uncertainties and cannot be used for precise estimates of F
value.
The behaviour of Pb is slightly erratic but does not conform to that expected of a truly incompatible element. Unfortunately the distribution of this trace element between
the proposed phenocryst phases and surrounding magma
has yet to be investigated in detail. Further evaluation of
the Pb variation trend must be postponed until appropriate studies are made.
If apatite is invoked to explain the P20~ versus 0.1.
trend, then minor fractionation of this mineral could have
been responsible for suppressing the degree of enrichment
of Y in residual liquids. Y is known to enter apatite and
crystal;1iquid distribution coefficients in excess of 10 are
indicated (Lambert and Holland, 1974).
Estimated values for D:I~g are too low to suppress the
degree of enrichment in Ba, as indicated by average data
for the more basic of the Haib lavas (BA-A). Other processes which could be invoked to explain the anomalous
behaviour of Ba would normally be expected to affect
related elements such as K and Rb as well. The consistent
behaviour of K and Rb in the proposed fractionation
scheme, effectively isolates Ba as an anomalous element,
at least in the basaltic andesites and andesites. On the
other hand, Ba displays the required amount of enrichment in the passage from andesite to dacite.
Zr must be regarded in the same light as Ba, since its
behaviour requires control by processes other than simple
fractional crystallisation. It is possible that Zr may replace
Ti in magnetite, in a manner analogous to ilmenite
although very little is known about the possible range i~
valu~s. for D~~ and Dif~ (McCallum and Charette, 1978).
Preltmmary data suggest that Dfr is significantly greater
.
d
z
1m
t h an unIty an that Dm~ is probably lower than Dif~ (Erlank et al., 1978). In any case, the amount of magnetite removed throughout the proposed fractionation scheme
may not be sufficient to affect the behaviour of Zr. It is
possible that zircon may have appeared as a minor fractionating phase, at least in the more acid lavas (andesite to
rhyolit~). This min~ral has been observed as micro-phenoc!'ysts In som~ daCItes and rhyolites, but it is unlikely that
zIrcon crystallIsed early in the more basic lavas. It must be
concluded that, with the data at present available the
~ehav.iour of Zr is not consistent with the proposed 'fractIOnatIOn scheme.
123
METALAVAS IN HAIB VOLCANIC SUBGROUP
1
Table VI lists 0 values for compatible trace elements, for
the various steps in the proposed fractionation scheme. It
is assumed that individual crystal!1iquid distribution coefficients and the relative proportions of the crystallising
phases, remain constant within each step.
The appro<l:ch adopted in this study has been to estimate
values for 0 1, which effectively involves the assumption
that the behaviour of compatible trace elements has been
controlled by fractional crystallisation. The validity of the
1
assumption has been examined by comparing predicted 0
values with published data.
The behaviour of Sr is almost entirely controlled by its
distribution between plagioclase and surrounding magma.
Of the other proposed phyric phases, only clinopyroxene
has a crystal!1iquid distribution coefficient significantly
greater than zero (D~~xrvO,l, Shimizu, 1974; Hart and
Brooks, 1974). However, the contribution of clinopyroxene
to D Sr can be neglected for the purposes of this study.
Predicted values for O~rag may be obtained by assuming
that all Sr in the crystallate is contained in plagioclase.
The equation linking i f with O~rag will be
With regard to the derivation of the porphyritic rhyolite
(R), an independent estimate of F for the step D-R, using
incompatible trace elements, is difficult because of the
uncertainty as to which elements have Ohbl and Obio! near
zero. Inspection of Table V reveals that Rb, Nb and U display the greatest degree of enrichment which would be
appropriate to the predicted amount of crystallisation
(F = 0,73, Table IV), if these elements were truly incompatible. The U data are probably fortuitous, bearing in
mind the probability of recent leaching. Rb probably
enters biotite and would not, therefore, be expected to
behave as an incompatible element. Nb is probably the
only element that can be regarded as truly incompatible
and its observed behaviour is consistent with the proposed
fractionation model.
4. Compatible Trace Element Behaviour
The degree of fractionation implied by the behaviour of
major and some incompatible trace elements may be
tested further by using those trace elements that show
affinity for the crystallising phases. If the distribution of a
1
compatible trace element (0 > I) between crystals and
surrounding liquid is controlled by Rayleigh type distillation, then
X
OSr
O Sr
plag· plag =
(5)
(2)
where X plag is the mass fraction of plagioclase in the respective solids removed in the stepwise fractionation
scheme.
Values for D~rag are listed in Table VII and plotted
against the predicted plagioclase composition in Fig. 14.
Recent studies have shown that O~rag varies systematically
with temperature and/or bulk composition (Korringa and
Noble, 1971; Drake and Weill, 1975). Figure 14 includes
average trends between O~rag and plagioclase composition,
the latter parameter reflecting both the temperature of
crystallisation and bulk magma composition. Compared
with these estimated trends the predicted values for 0 ~rag
appear reasonable and enhance the feasibility of the proposed fractionation scheme. Of particular interest is the
predicted increase in O~rag with differentiation and suggests that crystal!1iquid distribution coefficients of other
compatible trace elements may behave in the same
manner.
Difficulties arise with the evaluation of ferromagnesian
trace elements (e.g. Ni, Co, Cr, V and Sc) because they will
probably enter all the proposed mafic fractionating phases
(ortho- and clinopyroxene, magnetite, hornblende,
biotite). Although several attempts have been made to
estimate individual crystal!1iquid distribution coefficients,
inspectiofl of published data reveals a discouraging spread
in values. Perhaps the most important feature upon which
the following evaluation has been made, is that the distribution of any particular trace element between coexisting
phases, behaves in a regular manner. For example, Ewart
et al. (1973) report regular distribution of Ni, Co, Sc and V
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
where the quantities are the same as those defined for
equation (I). When more than one phase is crystallising,
Oi must be replaced with Oi, the average solid!1iquid distribution coefficient, such that
(3)
where Xa is the mass fraction of mineral a in the crystallate. By rearranging equation (2) and making
the
subject,
IY
Oi
= I + 10g(C~/C~)!1og F
(4)
TABLE VI
Predicted Average Solid/Liquid Distribution Coefficients for Compatible
Trace Elements, Using Equation (4)
Sr
Ni
Co
V
Cr
Sc
Zn
Ba
BA-BAI
BAI-A
A-D
D-R
F = 0,59
F = 0,70
F = 0,80
F = 0,73
0,74
1,90
1,45
1,18
2,56
1,68
1,22
3,06
2,51
2,07
3,70
1,81
1,50
1,97
3,48
3,87
3,26
3,61
2,8
1,97
2,43
3,9
4,2
3,4
3,1
2,6
2,78
1,38
I,ll
TABLE VII
Predicted Individual CrystaVLiquid Distribution Coefficients for the Compatible Trace Elements
Co
Ni
OPX
CPX
MT
HBL
BlOT
BABAI
BAIA
3,2
2,2
6,6
-
5,4
3,7
11,3
-
A-O
7,6
5,3
16
-
-
-
D-R
BABAI
BAIA
(16)
6,6
10,4
2,9
1,4
3,8
-
5,2
2,6
6,7
-
Sr
PLAG
BlOT
HBL
2,1
-
2,8
-
V
A-O
10
5
13
-
D-R
-
(13)
8,1
10,1
Ba
3,5
-
4,5
-
-
-
-
-
-
(0,4)
6,6
0,2
Sc
BABAI
BAlA
A-O
D-R
0,4
0,8
15,6
0,6
1,4
27
-
1,3
2,7
54
-
(54)
5,2
5,4
-
BABAI
BAlBA
1,7
3,6
1,6
2,3
5,1
2,2
-
-
A-O
-
3,9
8,4
3,6
-
-
-
D-R
-
124
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
1
1
-
-
1-2
*
/
/
-
Q/
.......*/
... ~/
.-;/
-
:
/
4/,.-
-
-
An
I
1
I
I
Ab
Figure 14
Plot of 0 ~~g against plagioclase composition. Trend lines are
based on ~orringa and Noble (1971 ) (dotted line) and Drake and
Weill (1975) (solid line).
between coexisting ortho- and clinopyroxene phenocrysts
in basaltic andesites, andesites and dacites from Tonga.
These modern lavas are sufficiently similar to the more
mafic Haib Volcanic Subgroup lavas that it is considered
justifiable to use them as. analogues. The distribution pattern implied that C~p/C ~px was reasonably constant for
any i (i = Ni, Co, V and Sc). It follows that the corresponding crystailliquid distribution co effici.ents will also show
regular behaviour, such that D~p/O~px = constant. The
behaviour of Cr was erratic and no regular relationship
could be obtained; as a result this element will not be discussed further. Less data were available for the analagous
relationship between pyroxenes and coexisting magnetite,
but for the purposes of this study, the average ratio
C~pxlC ~t calculated from data for three phenocryst pairs
reported by Ewart et al. (1973) has been adopted. Analogous relationships between hornblende and biotite have
been estimated from concentration data in Dodge et al.
(1968, 1969); Dodge and Ross (1971); Joyce (1973); Albuquerque (1973, 1974). Data for the hydrous phases have
been taken from plutonic assemblages, as no information
on their volcanic counterparts was available.
Individual crystailliquid distribution coefficients for Ni,
Co, V and Sc may be calculated using equation (3). Necessary data include Oi values (Table VI), weight fractions of
the three mafic phenocryst phases (Table IV) and the
inter-mineral relationships obtained from Ewart et at.
(1973) (Table VIII). Th~ latter relationships reduce the
number of unknowns (0 1) to one per element, since the
other two Oi .values. may be e?,pressed in terms of the first.
Values for 0 ~px, 0 ~px and O:nt for the three steps considered to be controlled by the fractionation of these three minerals, are listed in Table VII. It can be seen that for any particular element, Oi must increase with differentiation, and
in this respect resembles the predicted behaviour of Sr.
There is abundant experimental evidence for the implied temperature dependence (e.g. Sewar9, 1971; Lindstrom, 1976) and it is to be expected that 0 1 for pyroxenes
will increase with decreasing temperature of crystallisation. If minerals such as pyroxenes persist throughout a
TABLE VIII
Intermineral Distribution Coefficients Obtained from Sources Listed in
the Text
°opxiDcpx
Ni
Co
V
Sc
Ba
1,44
2,00
0,47
0,46
°mtIDcpx
3,02
2,6
20
0,43
0,034
/S-
0 .......
10
0 5c
DNi
10
(···. . opx
o
8
8;""':
-
/
./
;.. ··.cPx
6:.....:
o
4
2
0
.OPX
0
oCPX
40
20
60
40
80
80
60
01
01
.
. Figure 15
Variation of 0 ~px and 0 ~px with OJ. of the .bulk rock, using data
reported in Ewart et al. (1973). Predicted 0 1 values for 00px and
ocpx compare well with the observed trends.
The variation in 0 1 with bulk composition of the Tongan
lavas has been illustrated in Fig. 15 by means of average
trends; increase in Oi with O.I. is clearly indicated. Compared with these average trends, the predicted values for
OCo, OV and oSc appear reasonable. The only exception is
the behaviour of Ni. Extremely low Ni contents
« 20 ppm) in mos~ of the Tongan lavas precluded a complete range of ONI values and only the most basic rocks
yielded any iryformation. It can be seen that the predicted
o ~~x and 0 ~~x values are significantly lower than those obtained for the Tongan lavas. However, lower and more
comparable values for 0 ~~x and 0 ~x have been reported
by most other workers (e.g. Hakli and Wright, 1967; Dale
and Henderson, 1972; Lindstrom, 1976; Mysen, 1976). It is
possible that the high values obtained by Ewart et al.
(1973) for ONi may, in part, be a function of difficulties
associated with analysing exhemely low Ni contents in the
groundmass of the Tongan lavas.
Figure )6 illustrates various derived relationships between 0 ~x and temperature, established by three independent studies. The synthetic system of Seward (1971)
§
~
40
::y~
::::~
§
§
~
OJ
T °c
N'
Dc~x
20
10
6
~
4
°hbJiDbiot
0,63
0,80
0,96
/
2
2 -0 - -
-
l-
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
o
-
I-
-4
long temperature interval, as i~plied b~ their pre.sence in
all the Haib lavas from basaltIC andeSIte to daCIte, then
o opx
i and 0 cpx.
i would be expected
to increase.
The ob-_
.
i
i'
served progressive Increase In Dop~ and. 0cpx IS encourag
ing and supports the proposed fractIonatIon scheme.
7
1/T OK
8
. .
Ni
.Figure 16
Y'anatlOn of Ocpx .as a function of temperature, according to three
mdependent studies. A = Lindstrom (1976), B
Wright (1967), C = Seward (1971).
=
Hakli and
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
METALAVAS IN HAlB VOLCANIC SUBGROUP
involved the behaviour of Ni in the binary system
CaMgSi 20 6 (Diopside)-Na 2 Si 20 5 , while the synthetic
compositions employed by Lindstrom approach more
closely those of natural rocks. The data from Hakli and
Wright are from natural rocks, but which are more basic
than the Haib lavas. Many of the synthetic liquids used by
Lindstrom often crystallised orthopyroxene and magnetite
together with clinopyroxene and, therefore, probably represent the most applicable set of data. It is noteworthy that
Lindstrom's data imply a fairly rapid increase in D ~~)( for a
relatively small decrease in temperature. For a decrease of
100 °C, D~~)( changes by over a factor of two, which is
comparable with that suspected for the Haib lavas. Temperature estimates for the onset of crystallisation of basaltic andesites, andesites and dacites span 100-200 °C, judging from published data on phenocryst-phenocryst equilibration (Ewart, 1976b; Carmichael and Nicholls, 1967) and
experimental work (Eggler, 1972; Brown and Shairer,
1968). The relationship between D~~)( and temperature established by I,-indstrom (1976) suggests that th~ predicted
change in D ~~)( (and by inference the other DNl values) in
the Haib lavas may be reasonable.
The behaviour of Ba in the passage from dacite to porphyritic rhyolite may be explained satisfactorily. The
studies of Korringa and Noble (1971) and Drake and Weill
(1975) indicate that D ~I~g will be about 0,4 for the plagioclase (An 41 ) removed during this fractionation step. If this
value for D ~I~g is adopted, individual DBa values for the
other fractionating phases (hornblende and biotite) may
be calculated using equation (3) and data from Table VIII
in the same manner as that used for the ferromagnesian
trace elements. The resulting D ~~I (0,2) and D ~i~t' (6,6) are
within the range of values compiled by Arth (1976).
Evaluation of other trace elements in the passage from
dacite to rhyolite can only be treated in general terms. The
behaviour of Zr, Th, U and possibly Pb may all be explained in a qualitative sense by minor zircon fractionation. Y is probably controlled by apatite. Ferromagnesian
trace elements such as Ni, Co and Sc are too low in abundance « 20 ppm) to yield precise information about their
respective degrees of depletion; only V warrants further
V
examination. Derivation of individual D values is hampered by lack of information on the distribution of V between magnetite and the other mafic fractionating phases
(hornblende and biotite). If a D~t of 54 is arbitrarily
adopted (that obtained for the previous step A-D), solution of equation (3) yields D ~bl and D ~iot of 5,2 and 5,4 respectively. Inspection of what little data there are on the
distribution of V between these minerals and magma suggests that the predicted D v values are high. However, they
could be lowered if D ~t was even higher than 54; some
workers (e.g. Ewart et al., 1973; Lindstrom, 1976) report
value~ for D ~t in excess of 100. Moreover, further increase
in D~t with differentiation may be expected, bearing in
mind the predicted trends for the more basic lavas in the
Haib Volcanic Subgroup.
5. Summary
Adoption of a fairly rigorous approach to the evaluation
of the fractional crystallisation model has shown that all
the porphyritic lavas in the Haib Volcanic Subgroup may
have been produced by progressive removal of the common inferred phenocryst phases, at least in terms of major
elements. Compositions of the solid materials removed
during stepwise fractionation are appropriate to two
pyroxene gabbro for the early steps (BA-D) and biotite
diorite in the final step (D--R).
Detailed evaluation of all trace elements reported in this
study is precluded by one or more of the following:
(a) Lack of knowledge of the distribution of a particular
element between the proposed phenocryst phases and
surrounding magma (e.g. Pb, Zn, Zr).
(b) Low abundance coupled with wide scatter, which pre-
125
vents a precise estimate of the degree of enrichment!
depletion (e.g. U, Ni, Co and Sc in the acid lavas).
(c) Suspected control of minor fractionating phases (e.g.
Y, Zr, Th, U and Pb).
(d) Effects of alteration on the behaviour of geochemically mobile elements may have been overlooked,
despite the petrographic and geochemical screening
carried out (e.g. Ba, Rb, Sr, U).
Bearing all the above problems in mind, the preceding
evaluation of trace elements has yielded reasonably encouraging results. Certainly the control of plagioclase on
the behaviour of Sr is well-defined. Less well-defined is the
control of pyroxenes and magnetite on the behaviour of
Ni, Co, V and Sc in the more mafic lavas. It is concluded
that, with the data at present available, the porphyritic
lava suite of the Haib Volcanic Subgroup could have been
produced by progressive fractionation of phenocrysts
appropriate to low pressure, probably in high level magma
chambers immediately underlying the volcanic pile. The
most basic lava in the Haib Volcanic Subgroup (basaltic
andesite BA) is representative of the parental magma
to the porphyritic lava suite. Finally, the above model is
consistent with Sr and Pb isotopic data, which indicate
that all the porphyritic lavas had the same initial Sr and Pb
isotopic compositions.
B. Origin of the Non-Porphyritic Rhyolites
Much of the acid fragmental material (bedded tuffs,
pumice sheets, volcanogenic sediments) has probably
been derived from the disaggregation (either by eruption
or erosion) of the non-porphyritic rhyolites. Volume estimates suggest that the non-porphyritic acid lavas (?ignimbrites) and related fragmental material make up about 40
per cent of the exposed volcanic succession.
Compared with the porphyritic rhyolites (R), the nonporphyritic rhyolites (NR) are higher in alkalis and silica,
which is reflected in higher D.1. values of 85-95 (Table I).
The nature of the phenocryst assemblage in R suggests
that the groundmass may approach the composition of NR
in terms of major elements and suggests that they may be
related by fractional crystallisation. However, a least squares
approximation to NR, by removal of the phenocryst
phases present in R, yields unsatisfactory results. Fractionation of a subset of the phenocryst assemblage in R is
not possible because the residue would still be porphyritic,
which is inconsistent with the aphyric nature of NR.
In a very general way, many of the trace element abundances are consistent with fractional crystallisation involving the phenocrysts present in R, provided the relative
proportions are retained. For example, Rb, Zr, Nb and
Th are enriched and Sr, Ni, Co, Cr, V and Sc are depleted
in NR relative to R. However, it has been shown that R
is depleted in Ba relative to the dacites, which is probably the result of biotite fractionation. It follows that
the groundmass of R is probably even more depleted in Ba
(less than 985 ppm, see Table II) because of sustained
biotite crystallisation. In contrast, NR contains 1 4001 500 ppm Ba, which precludes its derivation from R by
fractional crystallisation. The Zn content of NR is higher
than that in R and is, therefore, inconsistent with the progressive drop in the abundance of this element with differentiation, as displayed by the porphyritic lavas.
The most likely alternative to fractional crystallisation is
that the non-porphyritic rhyolites have been derived from
a separate acid magma. Unfortunately no useful isotopic
data are available that could identify the source of this
acid magma. Sr isotopic data reported in Reid (1977) may
have been affected by post-consolidation Rb and/or Sr
migration, such that the present Rb/Sr ratios of the nonporphyritic rhyolites are too high for the degree of enrichment in radiogenic Sr. As a result, the nature and source
of the non-porphyritic rhyolites is not well-defined at
present.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
126
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
C. Magnetite Fractionation and the Calc-Alkaline Series
The porphyritic lavas of the Haib Volcanic Subgroup
may have been produced by fractional crystallisation of a
basic magma (at least as basic as basaltic andesite) involving the separation of two pyroxenes, plagioclase and
magnetite. The latter mineral is considered to have been
responsible for the lack of significant Fe-enrichment in the
passage from basaltic andesite to andesite. A model for the
production of the calc-alkaline series, involving early magnetite fractionation, has been developed by Osborn (1959,
1962, 1969). This model predicts the appearance of andesite as the residual magma and alpine-type peridotite as
the complementary cumulate. The model has not yet met
with general acceptance and appears to be inconsistent
with the following petrographic and chemical features:
(i) Lack of magnetite phenocrysts in many calc-alkaline
basic to intermediate lavas (Carmichael and Nicholls,
1967).
(ii) The abnormally high oxygen fugacity necessary to
stabilise magnetite as a liquidus or near liquidus phase
(Eggler and Burnham, 1973; Thompson, 1973; Biggar,
1974).
(iii) Oxygen fugacities do not remain constant and/or high
in the passage from basalt to andesite (Mueller, 1969,
1971).
(iv) The lack of severe depletion of elements that show a
great affinity for magnetite, such as V (Taylor et al.,
1969b).
(v) The presence of Cr-spinel rather than magnetite as
the characteristic opaque phase in alpine-type peridotites (Ringwood, 1975).
Although there is abundant evidence to indicate that
magnetite is not the liquidus (or even a near liquidus)
phase in basic magmas, when this mineral does appear it
may have a profound effect on the chemistry of residual
liquids. Magnetite, though not necessarily an early crystallising phase, may appear long before the solidus is reached
and will, therefore, be able to fractionate. This implies that
basic magmas may experience an early stage of fractionation involving the removal of mafic silicates and possibly
plagioclase, which may lead to strong Fe-enrichment in
residual liquids (tholeiitic trend). At the point where
magnetite appears, however, the liquid line of descent may
change abruptly to display progressive enrichment in alkalis (and silica) without any significant Fe-enrichment (ca1calkaline trend). The abrupt change in slope exhibited by
all tholeiitic suites on an AFM diagram usually coincides
with the appearance of phyric magnetite (and possibly
ilmenite in some cases).
The compilation of Ewart (1976a) suggests that about 50
per cent of modern orogenic basic lavas (basalt, basaltic
andesite) and an even greater proportion of andesite (up to
90 per cent) contain phyric magnetite. It is clear, therefore, that although magnetite fractionation cannot be
invoked as a general model for the production of calcalkaline suites, this process may be responsible for a significant proportion of such suites. Fewer problems appear
to exist with regard to the derivation of acid andesites and
dacites by fractionation involving the removal of magnetite. Eggler (1974) has reported experimental results under
controlled oxygen fugacities and water pressures, that predict the appearance of phyric magnetite in andesites and
more differentiated lavas, which is consistent with the conclusions of Ewart (1976a) discussed previously.
If it is accepted that magnetite phenocrysts in the Haib
basaltic andesites appeared after the more abundant
pyroxenes and plagioclase, then the most basic lavas in the
Haib Volcanic Subgroup may represent a derivative liquid
produced by an early stage of tholeiitic type fractionation
which was subsequently interrupted by the appearance of
magnetite. This is difficult to test because more basic magnetite-free lavas have not been found. A possible alternative is that the basaltic magma was extensively crystallised
before any net removal of solid material occurred. Cox
and Bell (1974) have suggested that differentiating magma
may experience compensated crystal settling, whereby ~he
overall composition is maintained since crystals settlIng
from a portion of the chamber are continually replaced by
compositionally similar crystals which settle into it from
higher levels. Variation in the efficiency of compensation
will cause a net loss or gain of phenocrysts, producing
slightly fractionated or cumulus enriched magma respectively. The resulting magma chamber will be zoned with
highly fractionated phenocryst poor magma near the roof
of the chamber and cumulus enriched magma near the
floor. Magma from the central region of the zoned
chamber may be fractionated, unchanged (as required by
Cox and Bell) or cumulus enriched, depending on the
degree of compensation. The lack of tholeiitic precursors
to the Haib basaltic andesite parent magma may be due to
the delaying effect of compensated crystal settling.
D. Magma Genesis and Source Characteristics
It remains to evaluate the possibility that the parental
basaltic andesite magma (BA) represents a primary or
derivative magma. The term "primary" is used in the sense
of Carmichael et al. (1974), as describing an unmodified,
unfractionated liquid formed directly by partial melting of
a pre-existing solid.
1. Basaltic Andesite (BA) as a Primary Magma
The basic nature of BA is taken to preclude its derivation from any crustal source and the relatively high
MgO content suggests that most basaltic source materials
are also excluded. Most natural and synthetic basaltic
systems investigated experimentally do not contain more
than 10-11 per cent MgO, and to produce a liquid with 8,5
per cent MgO requires excessively high degrees of partial
fusion. Although BA has been regarded as a basaltic andesite on the grounds of its relatively high Si02 and 0.1., the
observed abundances of MgO and ferromagnesian trace
elements (Ni, Co, Cr) are more typical of basalts. Such
high degrees of melting are theoretically possible, but it is
difficult to envisage the retention of a gravitationally unstable system such as a partial melt until 80-90 per cent of
the basaltic source rock has fused; segregation and uprise
of a magma diapir will probably occur long before such
high degrees of partial fusion are attained. A more likely
model involves lower degrees of partial fusion of some
ultrabasic source material in the upper mantle.
An important prerequisite of any mantle derived primary magma is that it must be in equilibrium with the solid
residue (mainly olivine). Kesson (1973) formulated a criterion for equilibrium using the empirical relationship between olivine and mafic liquid established by Roeder and
Emslie (1970). Kesson (1973) introduced the parameter
"Mg number" (= 100 MgO/(MgO + FeO) mole per cent)
by which the distribution of Mg and Fe between upper
mantle olivine and primary melt may be expressed.
Liquids in equilibrium with upper mantle olivine (Mg
number = 87,5-92) should have Mg numbers ranging from
67 to about 77. The Mg number of BA will depend on the
amount of FeO, which in turn depends on the Fe O/FeO
ratio of the primary magma. Kesson (1973) ad~p~ed a
value of 0,25 for Fe 20/FeO in alkaline lavas and if this
figure is us~d, then the FeO content of BA is 7,59 and the
Mg number is 67, which is just within the range for primary melts. Earlier it was concluded that the average
F~20/FeO predicted by the method of Le Maitre (1976)
y!elded reasonable results in view of the presence of phyn~ magnetite, which implied Fe2 0 3 contents at least as
high as three per cent. The resulting Fe 20/FeO ratio using
t~e method of Le Maitre on BA is 0,57, which yields a
hlg~er Mg number of 71. In terms of the predicted distri~utl?n of MgO and FeO between upper mantle olivine and
lIqUId, BA ful0ls tht: requirement of a primary magma.
Another salIent feature of BA is the relatively high K20
127
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
METALAVAS IN HAIB VOLCANIC SUBGROUP
and KINa ratio, which suggests that the origin of BA is
probably related to the genesis of high K calc-alkaline
magmas and, in particular, the factors controlling the K
content of primary melts. Potassium occurs in the upper
mantle either as a trace element « I00 ppm) in the major
minerals (olivine, pyroxene, garnet, spinel), along grain
boundaries, or as a stoichiometric component in minor
phases such as phlogopite and amphibole (Oxburgh, 1964;
Gast, 1968; Erlank 1970). It is likely that K will be strongly
partitioned into the liquid during melting, and that K bearing phases such as phlogopite or amphibole will disappear
very close to the solidus. This has been questioned by
Forbes and Flower (1974) and Beswick (1976), who suggest that under hydrous conditions, phlogopite may be a
refractory phase in the upper mantle. Experimental studies of hydrous melting of upper mantle materials have
shown that, under certain conditions, phlogopite may
coexist with a mafic liquid and, therefore, buffer the K
content of primary melts (Modreski and Boettcher, 1972;
Bravo and O'Hara, 1974; Mysen and Boettcher, 1975a, b).
However, the maximum K content of undepleted upper
mantle (garnet lherzolite) is probably I 000 ppm (Engel et
al., 1965; Gast, 1965) which limits the amount of phlogopite (K = 9,45 per cent) to about one per cent, assuming
that all K is contained in this mineral. The experimental
charges studied by the above authors contained at least
three per cent phlogopite, which seriously limits their applicability unless the maximum K content cited previously
for the upper mantle is a gross underestimate.
If it is assumed that all K in the source enters the liquid
during partial melting, then in order to attain a concentration similar to that in BA (1,3 per cent ~O = 1,08 per cent
K), 9,3 per cent of the source material needs to be melted.
Under anhydrous conditions and low pressures « 15
kbar), the partial melt will be a saturated (ol-hy-normative)
tholeiitic basalt (Kushiro, 1973). Qz-normative andesitic-basaltic magmas may be produced from the same
source material under vapour present conditions up to 25
kbar, provided X ~~ (mole fraction of H 20 in vapour
phase) is greater th~m 0,6 (Mysen and Boettcher, 1975b).
Lower water contents tend to result in silica-saturated and
undersaturated partial melts. Inspection of liquid compositions reported by Mysen and Boettcher (1975b), reveal
that they are much lower in MgO « 5 per cent) than BA,
although no data on the extent of melting in their experimental charges were given.
Trace elements which show affinity for the liquid during
partial melting of mantle peridotite include Rb, Sr, Ba, Zr,
Nb and Y. The inferred concentration of these elements,
using the same procedure as that adopted for K\ are listed
TABLE IX
Comparison Between Inferred Source Mantle Abundances of
Incompatible Trace Elements and Recent Estimates of Undepleted
Mantle. AIS(Linciuded are predicted average solid-liquid distribution
coefficients (D) for some compatible trace elements, together with
recent estimates for the distribution of these elements in the major residual mantle phases (olivine and orthopyroxene).
Lashaine
K
Rb
Ba
Sr
Zr
Nb
Y
Ni
Cr
Co
V
Sc
JJG1417 a Sun + Nesbitt
Inferred
747
3,9
56,3
41,2
8,3
2,4
1,5
747
3
73
107
32
5
6
199--274
0,574>,78
",7
19--26
9,7-13
0,534>,72
4,2-5,8
Co
C1
D
Dol
Dopx
1 600-2800
2050-3350
100-116
68-110
14-20
125
420
47
200
30
14-25
5,J....g,7
2,2-2,6
0,34>,5
0,44>,6
13
<1
4-6
<1
<1
",3
",1
2-3
<1
<1
a: From Gurney et af. (1977)
in Table IX. Rhodes and Dawson (1974) and Ridley and
Dawson (1974) have recently reported trace element data
for a series of garnet peridotite nodules from the Lashaine
volcano, Tanzania. One of the nodules (B0738) was considered to have the least depleted composition and was
taken to represent the closest approach to "fertile" upper
mantle under Lashaine. The Rb, Sr and Ba contents of
B 0738, together with the highest Zr, Nb and Y contents in
the other Lashaine garnet peridotites, are also listed in
Table IX. Inferred source abundances compare well with
the Lashaine peridotite data and tend to enhance the feasibility of the primary magma model. However, the inferred
values are not consistent with other estimates of undepleted mantle peridotite, which have also been included in
Table IX for comparison. These discrepancies could be
explained away as mantle inhomogeneity, but they serve
to emphasise the difficulties of this approach to the evaluation of the primary status for magmas.
The extent of melting suggested by the K content of
BA may be used to predict the behaviour of compatible
trace elements, using equations expressing trace element
distribution during anatexis (e.g. Gast, 1968; Shaw, 1970;
Arth, 1976). Assuming that the liquid has remained in
equilibrium with the solid residue throughout the melting
interval, and that the relative proportions of minerals contributing to the melt remain constant, then
(6)
The assumption of modal melting is not strictly justifiable,
since minor phases (e.g. phlogopite, amphibole) are considered to disappear very near the solidus. However, the
phases controlling the behaviour of compatible trace elements will be the major components (olivine, pyroxenes,
garnet) which may not experience any drastic change in
their relative proportions with up to to per cent melting.
Estimates of Co for Ni, Cr, Co, V and Sc in undepleted
mantle peridotite have been compiled in Table IX and
were extracted from Harris et al. (1967) and Reid et al.
(1974) for Ni and Cr; Sun and Nesbitt (1977) and Gurney et
af. (1977) for Co, V aryd Sc. The average solidlliquid distri1
bution coefficients (0 ) for these elements, assuming that
BA was a rv I per cent partial melt, have been calculated using equation (6). Also included in Table IX are individual
crystailliquid distribution coefficients for the major residual
phases (olivine and orthopyroxene).
The range in Co values produces a corresponding spread
in Oi. Since Oi will be controlle~by the residual mineralogy, comparison with . possible .Oi values based on published estimates for O~I and O~px provides a test for the
primary magma model. Both the maximum and minimum
possible ~ values are higher than. any individual ONi
value for possible residual phases. 0 ~I has been estimated
using the relationship of Hart et al. (1977)
°
log O~i = 3,325 - 0,088 5 MgO
(7)
(1000)
3,7
67
42
10
0,3
1,6
where MgO (mass per cent) is that in the liquid (in this
case BA).
The behaviour of Ni would only be consistent with the
primary magma model if the undepleted source had the
lowest Ni content (rv I 600 ppm) and the residual source
was almost pure olivine. This is considered unlikely after
only rv 10 per cent melting.
The behaviour of other trace elements cannot be as
rigorously tested because of uncertainties in individual Oi
values. However, for Cr to be consistent with the pnmary
magma model, the residual source must contain a significant proportion of minerals with OCr much higher than
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
128
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
those estimated for olivine and orthopyroxene. This problem could be alleviated somewhat if it is postulated that
most of the Cr in the mantle was divalent (Burns, 1975).
On the other hand, if the Cr was predominantly trivalent,
then likely candidates as residual phases include Cr-spinel,
clinopyroxene and garnet. Partial melting of undepleted
mantle peridotite probably involves the progressive elimination of garnet and clinopyroxene and their proportions
will be greatly diminished in residual mantle (Gurney et
al., 1974; Reid et al., 1974). However, significant quantities
of these minerals may still remain after only", 10 per cent
melting and, therefore, be able to retain Cr in the residual
source. Nevertheless, the behaviour of Cr is probably not
consistent with a residual source containing only olivine,
as the behaviour of Ni would suggest.
Predicted OCc, 0 v and oSe are consistent with a residual
source dominated by olivine and orthopyroxene. The presence of residual clinopyroxene and garnet would probably
tend to increase 0 v and oSe (trivalent cations) but not to
the extent experienced by Cr. Ewart et al. (1973) have
demonstrated that interelement distribution coefficients
of the type oCr/V and oCrlSe are in excess of ten and the
order of magnitude difference between oCr on the one
hand and 0 v and [)Sc on the other listed in Table IX,
would be expected.
To summarise, certain chemical features of BA are consistent with a primary nature, while others are not. The
evidence is too conflicting to conclude that BA represents
a primary magma and other models involving a primary
precursor must be considered.
2. Basaltic Andesite (BA) as a Derivative Magma
The previous use of Mg numbers highlights the important role of olivine in the generation of primary near
silica-saturated magmas. This has been the subject of
much debate (e.g. Kushiro, 1973; Nicholls and Ringwood,
1973; Mysen et al., 1974) and which concerns the behaviour of the primary field of olivine in basic melts as a function of water pressure. Kushiro (1973) and Mysen and
Boettcher (l975b) argue that saturated to slightly oversaturated basic magmas could be in equilibrium with olivine at elevated water pressures (appropriate to hydrous
mantle), but could crystallise only orthopyroxene at lower
pressures; there was no need for olivine to be the residual
and liquidus phase. It follows that BA could have been
generated in a hydrous upper mantle source but subsequently ascended to lower pressure regimes so as to
inhibit the appearance of phyric olivine. O'Hara (1968)
and Nicholls and Ringwood (1973) have asserted that olivine probably fractionates during the intervening period of
ascent of primary magmas, yielding saturated to slightly
oversaturated derivative magmas, perhaps similar to BA.
Early olivine fractionation could solve the problem of
low Ni content and the lack of phyric olivine in BA. Sato
(1977) and Allegre et a/. (1977) suggest that primary mantle
derived basaltic magmas should have at least 200--300 ppm
Ni. Relatively small amounts of olivine fractionation (5-10
per cent) would be necessary to drop the Ni content to
that in BA (125 ppm). The control of such residual minerals as clinopyroxenes and garnet of the Cr content of the
primary melt will not have been obscured by the sm~ll
amount of olivine fractionation. It follows that the HaIb
basaltic andesite (BA) could represent a product of small
degrees of olivine fractionation during the ascent of a near
saturated primary basaltic magma. Basaltic precursors to
calc-alkaline basaltic andesites and andesites appear to be
an increasingly accepted fact, at least for some modern
island arcs (e.g. Solomons, Stanton and Bell, 1969; Cox
and Bell, 1974; Talasea, Lowder and Carmichael, 1970;
Leeman et al., 1978; Indonesia, Nicholls and Whitford,
1976). Table X compares possible basaltic precursors to
BA (estimated simply by adding olivine) with estimated
primary magmas for the Indonesian arc and basalts from
the Solomon arc. Predicted Ni contents in the basaltic precursors are also included in Table X and compare well
with estimates of Sato (1977) and Allegre et al. (1977).
VII. DISCUSSION
Perhaps the most important feature of regional significance is that the entire porphyritic lava suite, which makes
up at least 60 per cent of the Haib Volcanic Subgroup in
the present study area, represents juvenile addition from
the upper mantle. The mid-Proterozoic age of 2,0 GA for
this magmatic episode makes it the oldest crust producing
event in the lower Orange River region.
A critical examination of the alleged dependence
between magma composition and the prevailing tectonic
environment must be postponed until the results of the
current phase of mapping in the lower Orange River
region are available. The association of abundant acid volcanics, a high-K calc-alkaline lava suite and an underlying
closely related calc-alkaline plutonic complex is practically identical to that developed in the Andes; so much so
that the very same geochemical properties have been cited
by Taylor (1969), lakes and White (l972a) and Taylor and
Hallberg (1977) as being characteristic of "Andean type"
magmatism. The analogy with the Andes goes so far as to
include the development of porphyry type low grade copper mineralisation. It is therefore tempting to postulate the
operation of active continental margin type geotectonism
in mid-Proterozoic times in south-western Africa. Unfortunately the two important features of this type of geotectonic environment (linear magmatic chain, underlying
continental crust) cannot be independently recognised, at
least not with the present geologic information available.
It is of special interest that Taylor and Hallberg (1977)
report Andean type calc-alkaline volcanism in the Archaean Marda complex in Western Australia. This complex appears to have been formed within a cratonic basin,
TABLE X
Derivation of a Possible Primary Magma Composition by Adding Olivine to the Parental Basaltic Andesite
(BA). Recent estimates of primary basaltic magmas which have been considered parental to a calc-alkaline
suite are also included: 1 = Basalt from the Solomon arc (Stanton and Bell, 1969); 2 = Indonesian arc
(Nicholls and Whitford, 1977). Predicted Ni contents of the primary magmas have been calculated using
equation (2) in the text. Recent estimates of Ni contents in primary basaltic magmll$ are also included for
comparison (Sato, 1977; Allegre et al., 1977). For 5 per cent olivine fractionation, D ~i = 12; for 10 per cent
0
olivine fractionation, D {;ii = 10 (after Hart et aL, 1977)
BA
Si02
Ti02
A~Ol
FeO
MgO
CaO
N~O
~O
54,6
0,70
14,4
9,3
8,5
9,0
1,8
1,3
Ni
125
BA + 5
% F089
54,0
0,70
13,7
9,2
10,6
8,6
1,7
1,2
220
BA + 10% F089
53,3
0,6
13,0
9,2
12,7
8,1
1,6
1,2
360
I
50,0
0,54
13,0
10,3
11,1
11,0
2,1
1,4
Sato
235-400
2
56,0-47,3
0,7-1,4
19,0-14,5
8,1-10,8
6,7-12,1
11,2-6,5
3,1-2,4
1,7-0,3
Allegre et al.
300-450
METALAVAS IN HAIB VOLCANIC SUBGROUP
defined by underlying banded iron-formation and clastic
sediments (Hallberg et aI., 1976). The above relationships
infer that lavas similar in composition with those of the
Haib Volcanic Subgroup were erupted within cratonic
basins during the Archaean and along active continental
margins during the Phanerozoic. What happened during
the intervening Proterozoic remains an enigma and the
present setting of the Haib Volcanic Subgroup does help
to solve it. However, some credance to the concept of Proterozoic active continental margins comes in the form of
recent speculation by Hoffmann (1973) and Watters (1974,
1976, 1977). In both cases the linear distribution of the
igneous products and the proximity of older continental
crust are clearly indicated. It may not be unacceptable to
extend the analogy to explain compositionally similar
rocks in the lower Orange River region.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
ACKNOWLEDGMENTS
I am sincerely grateful to Professor L. H. Ahrens for the
opportunity to use the facilities of the Department of Geochemistry and for continued support throughout the study.
Support during field work was received from J. de Villiers,
P. Joubert, A. Kroner and H. J. Blignault of the Precambrian Research Unit and from R. B. Cooke and 1. J.
Haumann of Rio Tinto (Pty) Ltd. Instruction and advice
during the analytical program was freely given by A. R.
Duncan, J. P. Willis and R. S. Rickard. The scientific
merit of this paper was enhanced by the efforts of A. J.
Erlank and A. R. Duncan. Financial support was received
from the C.S.I.R., University of Cape Town and Rio Tinto
(Pty) Ltd.
REFERENCES
Albuquerque, C A. R. de (1973). Geochemistry of biotites from
granitic rocks, northern Portugal. Geochim. cosmochim. Acta.
37. 1779-1802.
- - - - (1974). Geochemistry of actinolitic horn blendes from
tonalitic rocks, northern Portugal. Geochim. cosmochim. Acta.
38. 789-803.
Allegre, C J., Treuil, M., Minster, J., Minster, B., and Albarede,
F. (1977). Systematic use of trace element in igneous process.
Part I. Fractional crystallisation processes in volcanic suites.
Contr. Miner. Petrol.. 60, 57-75.
Arth, J. G. (1976). Behaviour of trace elements during magmatic
processes. A summary of theoretical models and their applica- \
tion. J. Res. U.S. geol. Surv.. 4,41-47.
Barager, W. R. A., and Goodwin, A. M. (1969). Andesites and
Archaean volcanism of the Canadian shield, 121-142. In:
Me Birney, A. R., Ed., Proceedings of the andesite conference. Dept. geol. Miner. Ind.. 65, State of Oregon.
Barbier, J., and Ranchin, G. (1969). Influence de l'alteration
meteorique sur I-uranium a I'etat de traces dans Ie granite a
deux micas de St-Sylvestre. Geochim. cosmochim. Acta. 33,
39-47.
Beswick, A. E. (1976). K and Rb relations in basalts and other
mantle derived materials. Is phlogopite the key? Geochim. cosmochim. Acta. 40, 1167-1183.
Bickle, M. J., Hawkesworth, C J., Martin, A., Nisbet, E. G., and
O'Nions, R. K. (1976). An estimate of mantle composition
derived from the chemistry of ultramafic lavas. Nature. 263,
557-580.
Biggar, G. M. (1974). Phase equilibrium studies of the chilled
margins of some layered intrusions. Contr. Miner. Petrol.,
46, 159-167.
Blignault, H. J. ·(1977). The tectonic evolution of the Richtersveld
and Namaqua Provinces in the Ais-Ais and Haib River areas,
South West Africa. Bull. Precambian Res. Unit. Univ. Cape
Town. 23, 197 pp.
Bravo, M. S., and O'Hara, M. J. (1974). Partial melting of phlogopite-bearing synthetic spinel- and garnet-lherzolites, 845-854.
In: Ahrens, L. H., Dawson, J. B., Duncan, A. R. and Erlank,
A. J., Eds., Phys. Chem. Earth, 9, Pergamon Press, Oxford,
940 pp.
Brown, G. M. (1967). Mineralogy of basaltic rocks, 103-162. In:
Hess, H. H. and Poldervaart, A., Eds, Basalts. 1. Interscience,
New York, 482 pp.
- - - - , and Shairer, J. F. (1968). Melting relations of some
calc-alkaline volcanic rocks. Mem. geol. Soc. Am., 130,
139-157.
Bryan, W., Finger, L. W., and Chayes, F. (1969). Estimating proportions in petrogenetic mixing equations by least squares approximation. Science. 163,926-927.
129
Burns, R. G. (1975). Crystal field effects in chromium and its
partitioning in the mantle. Geochim. cosmochIin. Acta. 39,
857-864.
Carmichael, I. S. E. (1964). The petrology of Thingmuli, a Tertiary volcano in eastern Iceland. J. Petrol.. 5, 435-460.
- - - - , and Nicholls, J. (1967). Iron-titanium oxides and
oxygen fugacities in volcanic rocks. J. geophys. Res.. 72,
4665-4687.
- - - - , Turner, F. J., and Verhoogen, J. (1974). Igneous Petrology. McGraw-Hili, New York, 739 pp.
Cawthorn, R. G. (1976). Some chemical controls on igneous
amphibole compositions. Geochim. cosmochim. Acta. 40,
1319-1328.
Condie, K. C (1976). Trace element geochimistry of Archaean
greenstone belts. Earth-Sci. Rev.. 12, 393-417.
Cox. K. G., and Bell, J. D. (1974). A crystal fractionation model
for the basaltic rocks of the New Georgia Group, British
Solomon Islands. Contr. Miner. Petrol., 37, 1-/3.
Dale, I. M., and Henderson, P. (1972). The partition of transition
elements in phenocryst-bearing basalts and the implications
about melt structure. Proc. 24th Int. geol. Congress. 10,
105-111.
Dodge, F. C W., and Ross, D. C (1971). Coexisting hornblendes
and biotites from granitic rocks near the San Andreas Fault,
California. 1. Geol.. 79, 158-172.
- - - - , Papike, J. J., and Mays, R. E. (1968). Hornblendes from
granitic rocks of the central Sierra Nevada batholith, California. J. Petrol.. 9, 378-410.
- - - -, Smith, V. C, and Mays, R. E. (1969). Biotites from
granitic rocks of the central Sierra Nevada, California. J. Petrol.. 10, 250-271 .
Drake, M. J., and Weill, D. F. (1975). Partition of Sr, Ba, Ca, Y,
Eu2+, Eu3+ and other REE between plagioclase feldspar and
magmatic liquid: An experimental study. Geochim. cosmochim.
Acta. 39, 689-712.
Eggler, D. H. (1972). Water saturated and undersaturated melting
relations in a Paricutin andesite and an estimate of water content in the natural magma. Contr. Miner. Petrol., 34, 261-271.
- - - - (1974). Application of a portion of the system
CaAI 2Si20s-NaAISips-Si02- MgO- Fe-02- H20-C02 to genesis
of the calc-alkaline suite. Am. J. Sc.. 274, 297-315.
- - - - , and Burnham, C W. (1973). Crystallisation and fractionation trends in the system andesite-H 20-C02-02 at pressures to 10 kb. Bull. geol. Soc. Am .. 84, 2517-2532.
Engel, A. E. J., Engel, C G., and Havens, R. G. (1965). Chemical
characteristics of oceanic basalts and the upper mantle. Bull.
geol. Soc. Am.. 76, 719-734.
Erikson, E. H. (1977). Petrology and petrogenesis of the Mount
Stuart batholith-plutonic equivalent of the high alumina basalt
association. Contr. Miner. Petrol.. 60, 183-207.
Erlank, A. J. (1970). Distribution of potassium in mafic and ultramafic nodules. Ann. Rep. Carn. Inst. Wa~h.. 68, 433-438.
- - - - , Smith, H. S., Marchant, J. W., Cardoso, M. P., and
Ahrens, L. H. (1978). Zirconium. Chapter 40. In: Wedepohl, K.
H., Ed., Handbook of Geocheml:~try. Springer-Verlag, Berlin.
Ewart, A. (1976a). Mineralogy and chemistry of modern orogenic
lavas - some statistics and implications. Earth Planet. Sci.
Lett.. 31, 417-432.
- - - - (1976b). A petrological study of the younger Tongan
andesites and dacites and the olivine tholeiites of Niva Fo'ou
Island, S. W. Pacific. Contr. Miner. Petrol.. 58,1-21.
- - - - , Bryan, W. B., and Gill, J. B. (1973). Mineralogy and
petrology of the younger islands of Tonga. J. Petrol., 15,
429-466.
Fitton, J. G. (1972). The genetic significance of almandine-pyrope
phenocrysts in the calc-alkaline Borrowdale volcanic group,
northern England. Contr. Miner. Petrol.. 36, 231-248.
Forbes, W. C, and Rower, M. F. J. (1974). Phase relations of
titan-phlogopite K2 Mg4TiA12Si602o(O H)4: A refractory phase
in the upper mantle? Earth Planet. Sci. Lett., 22, 60-66.
Gast, P. W. (1965). Terrestrial ratio of potassium to rubidium and
the composition of the Earth's crust. Science. 147, 858-860.
- - - - (1968). Trace element fractionation and the origin of
tholeiitic and alkaline magma types. Geochim. cosmochim.
Acta. 32, 1057-1086.
Gill, J. B. (1970). Geochemistry of Viti Levu, Fiji, and its evolution as an island arc. Contr. Miner. Petrol.. 27, 179-203.
Grapes, R. H., and Graham, C M. (1978). The actinolite-hornblende series in metabasites and the so-called miscibility gap:
A review. Lithos. 11, 85-97.
Green, T. H., and Ringwood, A. E. (1968). Genesis of the calcalkaline igneous rock suite. Contr. Miner. Petrol., 18, 105-162.
Gurney, J. J., Harte, B., and Cox, K. G. (1974). Mantle xenoliths
in the Matsoku kimberlite pipe, 507-524. In: Ahrens, L. H.,
Dawson, J. B., Duncan, A. R. and Erlank, A. J., Eds., Phys.
Chem. Earth. 9. Pergamon Press, Oxford, 940 pp.
- - - - , Lawless, P. J., and Dawson, J. B. (1977). Peridotite and
garnet olivine websterites at Bultfontein Mine (Abstract). Sec.
Int. Kimb. Conf. Santa Fe, U.S.A.
Hakli, T., and Wright, T. L. (1967). The fractionation of Ni
between olivine and augite as a geothermometer. Geochim.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
130
TRANSACTIONS OF THE GEOLOGICAL SOCIETY OF SOUTH AFRICA
cosmochim. Acta. 31, 877-884.
Halberg, J. H., Johnston, c., and Bye, S. M. (1976). The Archaean Marda igneous complex, western Australia. Precambrian Res.. 3, II 1-136.
Harris, P. G., Reay, A., and White, I. G. (1967). Chemical composition of the upper mantle. J. Geophys. Res., 72, 6359-6369.
Hart, S. R., and Brooks, C. (1974). Clinopyroxene-matrix partitioning of K, Rb, Cs, Sr and Ba. Geochim. cosmochim. Acta. 38,
1799-1806.
- - - - , Davis, K. E., Kushiro, I., and Watson, E. B. (1976). Partitioning of nickel between olivine and silicate liquid. Abstract, 906-907. Geol. Soc. Am.. a. Meet., Denver, Colorado.
Hawkesworth, C. J., and O'Nions, R. K. (1977). The petrogenesis
of some Archaean volcanic rocks from Southern Africa. J.
Petrol.. 18,487-519.
Hedge, C. E. (1971). Ni in high alumina basalts. Geochim. cosmochim. Acta. 35, 522-524.
Hoffman, P. (1973). Evolution of an early Proterozoic continental
margin: the Coronation geosyncline and associated aulacogens of the northwestern Canadian shield. Phil. Trans. R. Soc.
Lond.. A 273, 547-581.
Irvine, T. N., and Baragar, W. R. A. (1971). A guide to the chemical classification of the common volcanic rocks. Can. J. Earth
Sci., 8, 523-548.
Jakes, P., and Gill, J. B. (1970). Rare earth elements and the
island arc tholeiite series. Earth Planet. Sci. Lett.. 9, 17-28.
- - - - , and Smith, I. E. (1970). High potassium calc-alkaline
rocks from Cape Nelson, Eastern Papua. Contr. Miner. Petrol.,
28, 259-271.
- - - - , and White, A. J. R. (1972a). Major and trace element
abundances in volcanic rocks of orogenic areas. Bull. geol.
Soc. Am.. 83, 29-40.
- - - - , and - - - - (1972b). Hornblendes from calc-alkaline
volcanic rocks of island arcs and continental margins. Am.
Miner .. 57, 887-902.
Jolly, W. T., and Smith, R. E. (1972). Degradation and metamorphic differentiation of the Keeweenawan tholeiitic lavas of
northern Michigan, U.S.A. J. Petrol., 13, 273-309.
Joyce, A. S. (1973). Chemistry of the minerals of the granitic
Murrumbidgee batholith, Australian Capital Territory. Chem.
Geol., 11,271-296.
Keller, J. (1974). Petrology of some volcanic rock series of the
Aeolian arc, south Tyrrhenian Sea: Calc-alkaline and shoshonitic associations. Contr. Miner. Petrol., 46, 29-47.
Kesson, S. E. (1973). The primary geochemistry of the Monaro
alkaline volcanics, S. E. Australia - Evidence for upper
mantle heterogeneity. Contr. Miner. Petrol., 42, 93-108.
Korringa, M. K., and Noble, D. C. (1971). Distribution of Sr and
Ba between natural feldspar and igneous melt. Earth Planet.
Sci. Lett., II, 147-151.
Kroner, A., and Blignault, H. J. (1977). Towards a definition of
some tectonic and igneous provinces in western South Africa
and southern South West Africa. Trans. geol. Soc. S. Afr., 79,
232-238.
Kuno, H. (1966). Lateral variation of basalt magma type across
continental margins and island arcs. Bull. Volc., 29, 195-222.
- - - - (1968). Differentiation of basalt magmas, 623-688. In:
Hess, H. H. and Poldervaart, A., Eds., Bamlts, 2. Interscience,
New York, 380 pp.
Kushiro, I. (1973). Origin of some magmas in oceanic and circum·
oceanic regions. Tectonophys., 17,211-222.
Lambert, R. S., and Holland, J. G. (1974). Yttrium geochemistry
applied to petrogenesis utilising calcium-yttrium relationships
in minerals and rocks. Geochim. cosmochim. Acta, 38,
1394-1414.
Leeman, W. P., Ma, M.-S., Murali, A. V., and Schmitt, R. A.
(1978). Empirical estimation of magnetitelliquid distribution
coefficients for some transition elements. Contr. Miner. Petrol., 65, 269-272.
Le Maitre, R. W. (1976). Some problems of the projection of
chemical data into mineralogical classifications. Contr. Miner.
Petrol., 56, 181-190.
Lindstrom, D. J. (1976). Experimental study of the partitioning of the
transition metals between clinopyroxene and coexisting silicate
liquidv. Ph.D. thesis (unpubl.), Univ. Oregon, 188 pp.
Lowder, G. G., and Carmichael, I. S. E. (1970). The volcanoes
and caldera of Talasea, New Britain: Geology and Petrology.
Bull. geol. Soc. Am., 81, 17-38.
MacKenzie, D. E., and Chappell, B. W. (1972). Shoshonitic and
calc-alkaline lavas from the highlands of Papau, New Guinea.
Contr. Miner. Petrol., 35, 50-62.
McDougall, I. (1962). Differentiation of the Tasmanian dolerites: Red Hill dolerite-granophyre association. Bull. geol. Soc.
Am., 73, 279-316.
Miyashiro, A. (1973). Metamorphism and Metamorphic Belts.
George Allen and Unwin, London, 492 pp.
Modreski, P. J., and Boettcher, A. L. (1972). The stability of phlogopite-enstatite at high pressures: A model for micas in the
interior of the Earth. Am. J. Sci., 272, 852-869.
Mueller, R. F. (1969). Hydration, oxidation and the origin of the
calc-alkaline series. N.A.S.A. Tech. Note 0-5400, 27 pp.
____ (1971). Oxidative capacity of magmatic components.
Am. J. Sci., 270, 236-243.
.
.
Mysen, B. O. (1976). Partitioning of ~a~anum a~d .m~kel bet,,;e~n
olivine, orthopyroxene and lIquid: Preliminary data at
20 kbars and I 025 dc. Earth Planet. SCI. Lett., 31, 1-7.
____ , and Boettcher, A. L. (1975a).. Me.lting o~ hydrous
mantle. I. Phase relations of natural pendotlte at high pressures and temperatures with controlled activities of water, carbon dioxide and hydrogen. J. Petrol., 16, 520-548.
____ , and - - - - (1975b). Melting of hydrous mantl~. II.
Geochemistry of crystals and liquids fo.rmed by anatexIs of
mantle peridotite at high pressures and high temperatures as a
function of controlled activities of water, hydrogen and carbon dioxide. J. Petrol., 16, 549-593.
____ , Kushiro, I., Nicholls, I. A., and Ringwood,. A. E: (1974).
A possible mantle origin for andesitic magmas: Dlsc~sslon of a
paper by Nicholls and Ringwood. Earth Planet. SCI. Lett., 21,
221-229.
Nicholls, I. A., and Ringwood, A. E. (1972). Production of silica~
saturated tholeiitic magmas in island arcs. Earth Planet. SCI.
Lett., 17,243-246.
____ , and - - - - (1973). Effect of water on olivine stabili~y
in tholeiites and the production of silica-saturated magmas m
the island arc environment. J. Geol., 81, 285-300.
- - - - , and Whitford, D. J. (1976). Primary magmas associated
with Quaternary volcanism in the western S~nda. arc, Indon~­
sia, 77-90. In: Johnson, R. W., Ed., Volcanism In AustralaslG.
Elsevier, Amsterdam, 640 pp.
Noble, D. c., Bowman, H. It, Hebert, A. J., Silberman, M. L.,
Heropoulos, C. E., Fabbi, B. P., and He<;i&e, C. E. <,197~).,
Chemical and isotopic constraints of the ongm of low-SI latlte
and andesite from the Andes of central Peru. Geol., 3,
501-504.
O'Hara, M. J. (1968). The bearing of phase equilibria studies in
synthetic and natural systems on the origin and evolution of
basic and ultrabasic rocks. Earth-Sci. Rev., 4,69-133.
Osborn, E. F. (1959). Role of oxygen pressure in the crystallisation and differentiation of basaltic magmas. Am. J. Sci., 257,
609-647.
- - - - (1962). Reaction series for subalkaline igneous rocks
based on different oxygen pressure conditions. Am. Miner., 47,
211-226.
- - - - (1969). The complementariness of orogenic andesite
and alpine peridotite. Geochim. cosmochim. Acta, 33, 307-324.
Oxburgh, E. R. (1964). Petrological evidence for the presence of
amphibole in the upper mantle and its petrogenetic and geophysical implications. Geol. Mag., lOt, 1-19.
Peccerillo, A., and Taylor, S. R. (1976). Geochemistry of Eocene
calc-alkaline volcanic rocks from the Kastamonu area, northern Turkey. Contr. Miner. Petrol., 58, 63-81.
Reid, A. M., Donaldson, C. H., Brown, R. W., Ridley, W. I., and
Dawson, J. B. (1974). Mineral chemistry of peridotite xenoliths from the Lashaine volcano, Tanzania, 525-544. In:
Ahrens, L. H., Dawson, J. B., Duncan, A. R. and Erlank, A.
J., Eds., Phys. Chem. Earth, 9. Pergamon Press, Oxford, 940 pp.
Reid, D. L. (1977). Geochemistry of Precambrian igneous rocks
in the lower Orange River region. Bull. Precambrian Res. Unit,
Univ. Cape Town. 22, 397 pp.
- - - - (1979a). Total rock Rb-Sr and U-Th-Pb isotopic study of
Precambrian metavolcanic rocks in the lower Orange River
region, Southern Africa. Earth Planet. Sci. Lett. (in press).
- - - - (1979b). Age relations within the Proterozoic Vioolsdrif
batholith, lower Orange River region, Southern Africa. (Submitted to Precambrian Res.)
Rhodes, J. M., and Dawson, J. B. (1974). Major and trace element
chemistry of peridotite inclusions from the Lashaine volcano,
Tanzania, 545-558. In: Ahrens, L. H., Dawson, J. B., Duncan,
A. R. and Erlank, A. J., Eds, Phys. Chem. Earth, 9. Pergamon
Press, Oxford, 940 pp.
Ridley, W. I., and Dawson, J. B. (1974). Lithophile trace element
data bearing on the origin of peridotite xenoliths, ankaramite
and carbonatite from Lashaine volcano, N. Tanzania,
559-570. In: Ahrens, L. H., Dawson, J. B., Duncan, A. R., and
Erlank, A. J. Eds., Phys. Chem. Earth, 9. Pergamon Press,
Oxford, 940 pp.
Ringwood, A. E. (1966). Chemical composition and origin of the
Earth, 287-356. In: Hurley, P. M., Ed., Advances in Earth Sciences. M.I.T. Press, 502 pp.
- - - - (1975). Composition and Petrology of the Earth's Mantle.
ryfcGraw-Hill, New York, 618 pp.
Ritter, U. (1978). Geologic evolution of the eastern Richtersveld.
Bull. Precambrian Res. Unit. Univ. Cape Town. 26, 269 pp.
Roeder, P. L., and Emslie, R. F. (1970). Olivine liquid equilibrium. Contr. Miner. Petrol., 29, 275-289.
Rosholt, J. N., Zartman, R. E., and Nkomo, I. T. (1973). Load
isotope systematics and uranium depletion in the Granite
Mountains, Wyoming. Bull. geol. Soc. Am., 84, 989-1002.
Sato, H. (1977). Nickel content of basaltic magmas: Identification
of primary magmas and a measure of the degree of olivine
fractionation. Lithos, 10, 113-120.
METALAVAS IN HAIB VOLCANIC SUBGROUP
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2010)
Seward, T. M. (1971). The distribution of transition elements in
the system CaMgSi206-N~Si2q-H20 at I 000 bars pressure.
Chem. Geol., 7, 73-95.
Shaw, D. M. (1970). Trace element fractionation during anatexis.
Geochim. cosmochim. Acta, 34, 237-243.
Shimizu, N. (1974). An experimental study of the partitioning of
K, Rb, Cs, Sr and Ba between c1inipyroxene and liquid at high
pressures. Geochim. cosmochim. Acta, 38, 1789-1798.
Smith, R. E. (1969). Zones of progressive regional burial metamorphism in part of the Tasman Geosyncline, eastern Australia. J. Petrol., 10, 144-163.
Stanton, R. L., and Bell, J. D. (1969). Volcanic and associated
rocks of the New Georgia Group, British Solomon Islands
Protectorate, Overseas geol. Miner. Res., 10, 113-145.
Sun, S. S., and Nesbitt, R. W. (1977). Chemical heterogeneity of
the Archaean mantle, composition of the Earth and mantle
evolution. Earth Planet. Sci. Lett., 35, 429-448.
Taylor, S. R. (1969). Trace element chemistry of andesites and
associated calc-alkaline rocks, 43-64. In: McBirney, A. R.,
Ed., Proceedings of the andesite conference. Bull. Dept. geol.
Miner. Ind.. 65, State of Oregon.
- - - - , and Hallberg, J. A. (1977). Rare earth elements in the
Marda calc-alkaline suite: An Archaean geochemical
analogue of Andean-type volcanism. Geochim. cosmochim.
Acta. 41, 1125-1129.
- - - - , Capp, A. C, Graham, A. L., and Blake, D. H. (I 969a).
Trace element abundances in andesites. II. Saipan, Bougainville and Fiji. Contr. Miner. Petrol., 23, 1-26.
GEOL 82/1 -
I
131
- - - - , Kaye, M. White, A. J. R., Duncan, A. R., and Ewart,
A. (1969b). Genetic significance of Co, Cr, Ni. Sc and V content of andesites. Geochim. cosmochim. Acta, 33, 275-286.
Thompson, R. N. (1973). One atmosphere melting behaviour and
nomenclature of terrestrial lavas. Contr. Miner. Petrol., 41,
197-204.
Thornton, C P., and Tittle, O. F. (1960). Chemistry of igneous
rocks. I. Differentiation Index. Am. J. Sci., 258. 664-684.
Wager, L. R. (1960). The major element variation of the layered
series of the Skaergaard intrusion. J. Petrol., 1, 364-398.
Watters, B. R. (1974). Stratigraphy, igneous petrology and evolution of the Sinclair Group in southern South West Africa.
Bull. Precambrian Res. Unit. Univ. Cape Town. 16. 235 pp.
- - - - (1976). Possible later Precambrian subduction zone in
South West Africa. Nature, 259.471-473.
- - - - (1977). The Sinclair Group: Definition and regional correlation. Trans. geol. Soc. S. Afr., 80, 9-16.
Zielinski, R. A., and Lipman, P. W. (1976). Trace element variations at Summer Coon volcano, San Juan Mountains, Colorado. and the origin of continental interior andesite. Bull. geol.
Soc. Am., 87. 1477-1485.
Department of Geochemistry,
University of Cape Town.
Private Bag.
7700 Rondebosch.
Accepted for pUblication by the Society on 3.2.1979.