dykes Mengel et al., 1996. The NE – SW set constitute the main set of intrusions and include
the two dykes dated by Nutman and Kalsbeek 1996 at ca. 2.04 Ga. North of Itilleq Fjord,
Kangaˆmiut dykes were emplaced into Archaean E – W trending amphibolite facies zones.
3. Mineralogy and petrology
The mineralogical and petrographic variations within the Kangaˆmiut dyke swarm have been well
documented from
previous studies,
Windley 1970 showed that the dykes near Sukkertoppen
Fig. 1 were often composites of hornblende quartz dolerite, amphibolite and garnet amphibo-
lite, with
microporphyritic hornblende-quartz
margins. Cross-cutting
relationships between
Kangaˆmiut dykes show that the development of internal mineral foliations within the earlier intru-
sions are syn-intrusive. Fahrig and Bridgwater 1976
also noted
the existence
of zoned
Kangaˆmiut dykes.
To the
north, between
Kangaˆmiut village and Søndre Strømfjord Fig. 1 an additional texture is observed in some of the
most highly fractionated intrusions, which contain garnet – albite – quartz – diorite pods. It is impor-
tant to realise that these pods are sometimes situated within unsheared portions of the dyke,
suggesting their origin to be igneous rather than metamorphic Fahrig and Bridgwater, 1976, and
that at least some of the dykes crystallised from wet magmas Bridgwater et al., 1995.
Recent work has shown petrographic differ- ences associated with relative age and dyke trend
Mengel et al., 1996; the older E – W dykes con- tain olivine, clinopyroxene orthopyroxene and
plagioclase. The NNE-trending dykes contain ig- neous clinopyroxene, plagioclase, rare orthopy-
roxene
and primary
hornblende in
chilled margins. The NE-trending dykes contain igneous
clinopyroxene, plagioclase, igneous hornblende phenocrysts in chilled margins and felsic patches
of dioritic composition, within the centres of some wider intrusions.
Petrographically, the Kangaˆmiut dykes range from those with fresh igneous textures including
those with primary hornblende phenocrysts to those with recrystallised static or dynamic meta-
morphic textures essentially amphibolites Men- gel et al., 1996. The grade of metamorphic
overprinting generally increases northwards. Ini- tial metamorphic reactions resulted in the forma-
tion
of amphibole,
biotite and
garnet. Metamorphic pyroxene is recorded in the north-
ernmost areas of the swarm around and to the south of Itertoˆq Fjord, representing upper amphi-
bolite – granulite metamorphism Mengel et al., 1996.
Structurally the Kangaˆmiut dykes show strong evidence for syn-shear emplacement. Escher et al.
1976 recorded consistent stepping directions, oblique offsets of country rock bands across
dykes and oblique internal foliations. They con- cluded that the E – W and NNE – SSW trending
dykes were injected in conjugate shears, although more recent work suggests they were intruded
under a sinistral transpressive regime Mengel et al., 1996; Hanmer et al., 1997. Similarly Windley
1970 noted that some Kangaˆmiut dykes con- tained unfoliated chilled margins with strongly
deformed dyke centres, suggesting that the dykes had acted as loci of shear whilst still hot and
rheologically weak. Very similar field relationships were observed during this study.
Structural and mineralogical work therefore strongly suggests that Kangaˆmiut dyke emplace-
ment, shear zone formation and aqueous fluid transport were coeval or near coeval processes
Korstga˚rd, 1979; Bridgwater et al., 1995, at least within
the later
ca 2.04
Ga NE-trending
Kangaˆmiut intrusions. Mengel et al. 1996 fur- ther noted that that the two earlier dyke sets
could not be positively be classed as being of Kangaˆmiut age at the present time due to a lack
of precise age data.
4. Geochemistry and petrogenesis
4
.
1
. Analytical techniques Samples were crushed and ground in an agate
swing-mill to a fine powder before analysis by
X-ray fluorescence spectrometry XRF at the University of Leicester. Major elements were de-
termined on glass discs fusion beads, and trace elements on powder pellets, using analytical pro-
cedures detailed in Marsh et al. 1983 and Weaver et al. 1983. Selected samples were also
analysed for the rare earth elements REE La, Ce, Nd, Sm, Eu, Gd, Dy, Er, Yb and Lu by
ICP – MS on a Prototype VG Elemental Plas- maQuad II + at the Centre for Analytical Re-
search in the Environment, Imperial College Silwood Park. Standard operation procedures
were used. Samples and reference materials were first dissolved using nitric acid into aqueous solu-
tions. A procedural blank was also prepared and analysed. Each sample and reference material was
measured three times and the results averaged. Reference materials were run in between batches
of five samples to check for machine drift.
Pb isotope results from two dykes were gener- ated at the University of Oxford, see Table 2 for
details.
4
.
2
. Geochemical results Representative
chemical analyses
of the
Kangaˆmiut dykes are given in Table 1. Multi-ele- ment mantle normalised diagrams of these analy-
ses Fig. 2 show signatures comparable with other continental dyke swarms and flood basalt
provinces: the patterns show ‘lithospheric’ charac- teristics, being enriched in incompatible elements
when compared to primordial mantle or chondrite abundances Weaver and Tarney, 1981; Arndt et
al., 1993. They show Nb depletion relative to the light rare-earth elements LREE and large-ion-
lithophile LIL elements, but significantly not the extreme Nb depletions which are associated with
island arc basalts see Thompson et al., 1983 for comparison of continental flood basalt and island
arc signatures.
The consistency between patterns for both LIL- and for high-field-strength HFS elements show
that despite undergoing amphibolite-facies meta- morphism, the chemistry of the dykes has not
been strongly affected by element remobilisation.
4
.
3
. Inter-dyke crystallisation processes Despite the extensive fieldwork and mineralogi-
cal studies done on the Kangaˆmiut dyke swarm, geochemical analysis has hitherto been restricted
to major elements, some trace elements and REE. These provided the first indication that the dyke
swarm maintained a tholeiitic fractionation trend throughout most, if not all, of their magmatic
evolution Windley, 1970; Escher et al., 1975; Fahrig and Bridgwater, 1976; Bridgwater et al.,
1995. Hall and Hughes 1990, 1993 noted that the Kangaˆmiut dyke swarm showed no real differ-
ence in this respect from the extension-related dolerites and gabbros of the MD dyke swarm,
despite their evidently different mineralogy and tectonic environment of formation. Figs. 3 – 7
show that for major, LIL and transition metal elements, the Kangaˆmiut dykes have very simple
and consistent chemical trends. For example, the correlation of CaO and MgO Fig. 3 is very
similar to that described by Cox 1980 for conti- nental flood basalt geochemical evolution and
attributed to the fractionation of clinopyroxene and plagioclase. Modelling of these trends using
the most primitive Kangaˆmiut dyke sample 416009, based on Mg c and Zr concentration;
Table 1 as the most primitive magma showed that the trends could be precisely mimicked by the
fractional crystallisation of a 1:1 clinopyroxene – plagioclase crystal assemblage. In total, the range
in highly incompatible trace elements such as Zr suggest that approximately 82 fractionation oc-
curred, a value close to several other dyke swarms and flood basalt sequences e.g. Ahmad and Tar-
ney, 1991. The validity of this hypothesis was further checked using multi-element mantle nor-
malised diagrams Fig. 7 for incompatible ele- ments and REE. The diagrams show very good
agreement between the modelled solutions and the evolved samples. It is evident from the REE data
that garnet was not involved in the crystal frac- tionation assemblage, as even small amounts
would strongly deplete the heavy rare earth con- centrations of the most evolved dykes.
For comparative purposes, trends using a crys- tal fractionation assemblage consisting of a 1:1
mixture of hornblende and plagioclase were also
Table 1 Representative analyses of Kangamiut Dykes, W. Greenland
416025 416009
416001 416004
416005 416022
416018 416007
Sample 416019
416024 57
32 20
24 25
53 49
56 25
Loc. No. 50
SiO
2
53.2 51.8
49.7 50.9
51.2 50.8
51.4 50.7
57.4 49.1
2.70 0.84
1.37 1.63
2.29 1.65
1.95 2.65
TiO
2
1.73 3.34
13.2 Al
2
O
3
11.9 14.3
13.6 12.7
12.3 13.1
13.0 12.7
12.7 Fe
2
O
3
14.9 17.4
10.9 14.1
15.7 17.6
15.6 15.5
13.8 17.7
0.26 0.18
0.20 0.23
0.25 0.23
0.19 0.21
MnO 0.20
0.23 4.32
MgO 3.98
8.91 6.43
5.53 4.51
5.23 4.15
2.80 3.78
8.05 12.81
9.68 9.48
8.52 9.48
CaO 7.38
8.23 5.81
8.04 2.8
2.1 3.1
3.0 2.9
2.7 2.8
4.8 Na
2
O 4.2
3.3 0.88
K
2
O 0.68
0.21 0.48
0.47 0.62
0.46 1.38
1.08 1.34
0.40 P
2
O
5
0.28 0.07
0.14 0.15
0.25 0.14
0.32 0.32
0.42 0.09
0.19 B
0.05 B
0.14 0.36
B 0.13
B 0.25
B 0.26
H
2
O 0.32
B 0.38
Total 99.9
100.1 100.2
100.0 100.1
100.4 100.0
100.1 100.4
99.9 Trace elements in ppm
determined by XRF 39
40 38
34 41
36 31
14 Sc
29 26
351 V
415 247
311 347
391 382
350 229
390 19
247 71
61 32
Cr 57
58 25
21 42
54 56
55 50
50 55
55 50
Co 37
46 31
Ni 15
101 62
31 24
43 24
13 41
112 136
191 100
250 161
Cu 336
34 128
308 85
40 63
61 78
64 74
75 Zn
76 82
22 Ga
27 19
22 23
24 21
27 27
28 22
13 15
Rb 16
22 19
16 44
31 54
171 143
218 178
164 166
386 550
Sr 218
177 40
Y 55
19 26
35 33
37 29
53 66
203 54
103 118
102 125
Zr 199
169 243
282 16
6 7
7 8
7 3
25 Nb
19 21
528 Ba
214 147
139 144
177 141
485 352
438 2
– –
– –
1 2
– Mo
3 3
– 6
– 4
7 12
8 Pb
5 14
– Sn
– –
– –
3 7
6 2
6 –
– 2
4 1
– 10
6 Th
3 5
Rare earth elements in ppm determined by ICP-MS 41
29 9.68
12.08 18.94
11.87 16.08
39.51 34.77
37.63 La
80 13.70
30.33 29.60
30.68 32.08
86 84.44
Ce 73.85
90.75 –
2.58 4.07
4.16 4.249
– 4.48
Pr 10.49
9.60 11.62
34 10.50
17.53 18.21
19.58 Nd
19.36 39
42.15 40.57
50.30 Sm
– –
2.69 4.11
4.74 5.13
5.27 8.32
9.86 11.66
– 0.90
1.42 1.60
1.87 1.64
Eu 2.62
– 2.93
2.93 –
2.96 4.30
5.184 5.90
– 5.70
Gd 7.76
10.35 11.99
Tb –
– 0.52
0.73 0.92
1.04 0.95
1.04 1.66
2.02 –
3.33 4.16
5.47 6.28
5.82 5.21
Dy 9.08
– 11.39
– 0.61
0.88 1.16
1.27 1.25
– 1.01
Ho 1.83
2.33 –
Er –
2.01 2.55
3.28 3.75
3.65 2.71
5.27 7.41
– –
0.36 0.46
0.52 0.54
0.36 0.73
Tm 0.99
– –
1.87 2.52
3.08 3.35
3.44 –
2.28 Yb
4.89 6.46
– Lu
0.27 –
0.35 0.45
0.50 0.49
0.33 0.65
0.93
Table 1 Continued 416024
Sample 416025
416009 416001
416004 416005
416022 416018
416007 416019
57 32
20 24
25 53
Loc. No. 49
56 25
50 416003
416004 416008
416010 416011
416013 416002
416010 Sample
416020 416021
416023 Metased
Loc. No. 22
21 24
27 33
37 40
33 50
51 54
Av n = 17 50.6
SiO
2
50.9 51.2
48.4 50.4
50.4 50.4
49.1 50.7
51.2 51.1
49.0 1.63
2.09 2.43
2.02 2.20
1.94 1.57
2.37 TiO
2
1.09 1.65
2.69 1.0
13.2 Al
2
O
3
13.8 12.4
12.1 12.4
13.2 12.5
11.9 13.0
13.1 11.8
11.4 14.7
Fe
2
O
3
14.7 16.9
19.0 16.5
14.7 16.0
17.8 14.6
15.4 18.2
12.8 0.21
0.25 0.26
0.24 0.22
0.23 0.22
0.24 MnO
0.22 0.22
0.26 0.19
MgO 5.70
6.34 5.21
5.29 5.54
5.63 5.87
5.87 6.96
5.60 4.08
13.37 7.68
8.75 8.74
9.40 9.57
9.78 9.67
9.51 CaO
10.6 9.76
8.13 9.40
3.0 Na
2
O 4.4
2.6 3.2
3.0 3.1
2.7 2.7
2.5 2.5
2.7 2.1
0.50 K
2
O 0.86
0.46 0.39
0.39 0.53
0.42 0.41
0.24 0.40
0.66 0.72
0.15 0.18
0.18 0.17
0.33 0.19
0.16 0.17
P
2
O
5
0.09 0.14
0.33 0.10
1.32 0.02
B 0.27
B 0.16
B 0.07
0.14 B
0.15 H
2
O B
0.19 B
0.25 B
0.17 0.18
0.44 101.35
100.1 100.0
100.1 99.9
100.2 100.2
99.9 100.0
Total 99.9
100.2 99.94
Trace elements in ppm determined by XRF Sc
33 37
50 34
43 37
41 44
43 36
44 34
323 497
572 448
356 369
350 582
V 318
365 413
269 Cr
82 71
118 48
80 123
128 98
98 108
26 968
52 49
59 56
49 55
52 66
Co 60
49 51
66 73
Ni 58
33 34
37 47
47 42
59 58
21 419
249 Cu
208 213
126 102
82 100
110 161
167 59
67 62
73 78
68 64
64 65
74 Zn
58 64
90 56
22 25
26 22
24 21
Ga 21
22 17
21 24
18 18
19 12
11 18
13 23
16 Rb
7 14
23 15
236 Sr
250 160
173 150
188 137
129 130
159 136
211 25
Y 36
38 41
38 41
43 34
22 34
58 22
124 131
139 126
151 143
126 121
Zr 65
127 197
65 18
8 14
9 12
8 Nb
10 13
3 5
15 6.4
544 149
180 136
169 115
173 128
Ba 61
137 210
139 24
La 21
17 18
19 21
15 14
12 13
18 14
41 Ce
40 46
59 53
54 37
50 –
47 67
38 23
23 25
24 32
24 19
29 Nd
4 20
33 19
14 4
– 9
10 8
Pb –
– –
7 16
– 4
2 3
2 –
10 –
– Sn
– –
8 –
6 Th
19 5
– 2
2 5
– –
– 5
5
modelled for TiO
2
and K
2
O with respect to Zr content Fig. 4. The higher distribution coeffi-
cients of Ti, K and Zr for hornblende result in much more constrained fractionation trends and
weaker increases in TiO
2
and K
2
O concentration with increase in Zr. To increase contents of all
three elements would have required much larger proportions of plagioclase fractionation, which in
turn could be expected to result in much stronger depletions of Sr with increase in Zr content Fig.
6. Similarly, a higher plagioclase to hornblende ratio in the fractionation assemblage would re-
quire hornblende to have considerably higher dis- tribution coefficients for Y. The distribution
coefficients values for Cr, Sc and Ni would also have to be considerably higher and unrealistic
when compared to the published ranges of values see Rollinson, 1995.
4
.
4
. Intra-dyke crystallisation processes The simple tholeiitic trends recorded by the
dataset overall is surprising considering the re- markable across-width textural variation within
Fig. 2. Incompatible element a and rare earth b multi-ele- ment mantle normalised diagrams for the representative sam-
ples from the Kangaˆmiut dyke swarm, as presented in Table 1. Primordial Mantle abundances from Sun and McDonough
1989. Chondrite values from Nakamura 1974. Ho value estimated from Y concentration.
nocrysts. This zone became more melanocratic and finer grained towards the centre of the dyke,
though the centre of the dyke was still more leucocratic and coarser grained than the dyke
margin.
This dyke was selected for special study as it contained many of the best examples found dur-
ing this study of the textural features noted in previous studies of the Kangaˆmiut dykes e.g.
Windley, 1970. Analysis of the marginal and leucocratic zone of this intrusion showed substan-
tial chemical variation. Zr content varies between 102 – 243 ppm, and the centre of the dyke is
substantially more siliceous 57.3 wt. than the
Fig. 3. Biaxial plot of a CaO vs MgO and b Fe
2
O
3
vs MgO showing fractionation trend of Kangaˆmiut dykes. Dashed line
connects analyses from the margin [m: sample 416005] and centre [c: sample 416007] of a sheared Kangaˆmiut dyke. See
text for full details.
modelled for TiO
2
and K
2
O with respect to Zr content Fig. 4. The higher distribution coeffi-
cients of Ti, K and Zr for hornblende result inmany Kangaˆmiut dykes. One such dyke is a 30
m wide intrusion situated on the north side of Itilleq Fjord Fig. 1. This intrusion consisted of
undeformed, dioritic margins with a macroscopic mineralogy of metamorphic plagioclase, horn-
blende and garnet. The western margin of the dykes was well exposed and consisted of a 0.5 m
wide marginal zone made up on hornblende, pla- gioclase and glomeroblastic garnets. This was fol-
lowed centrewards by a 2 m zone with low strain which contained garnets 1 – 2 mm in diameter.
Finally this zone passed into a 2 m leucocratic zone containing highly strained plagioclase phe-
Fig. 4. Biaxial plots showing fractionation trends for a TiO
2
vs Zr ppm; and b K
2
O vs Zr ppm. Line with white filled circles correspond to fractionation compositions derived
from sample 416009. The fractionation crystallisation assem- blage used consisted of a 1:1 ratio of clinopyroxene and
plagioclase. Numbers adjacent to circles give amount of frac- tionation. Line with dark filled circles: same as above but for
1:1 crystal fractionation assemblage of plagioclase and horn- blende other details as for Fig. 3.
elements within this dyke are closely comparable to the general trends of the Kangaˆmiut dyke
swarm and the modelled fractional crystallisation solution.
In order to test this hypothesis the crystallising assemblage used in modelling inter-dyke chemical
variation was applied to this dyke, using the marginal sample 416005 as the parent magma
composition. Zr was again taken to be perfectly incompatible in order to estimate the amount of
fractional crystallisation required i.e. 60.
Simple clinopyroxene + plagioclase fractional crystallisation
produced poor
fits with
the analysed composition from the leucocratic zone
sample 416007. Concentrations of TiO
2
and K
2
O were too high whereas LREE concentrations are too low Fig. 8a, b. If hornblende is added to
Fig. 5. Biaxial plots showing fractionation trends for: a Y ppm vs Zr ppm; and b Cr ppm vs Zr ppm. Other details as
for Fig. 3 and Fig. 4, except line modelling hornblende-plagio- clase fractionation not shown.
margin 50.9 wt.. Such a trend in silica enrich- ment is uncommon in continental flood basalts,
where silica content is buffered during fractiona- tion Cox, 1980. The centre of the dyke is de-
pleted in Fe
2
O
3
and TiO
2
relative to its margin. Similar trends are not evident within other
Kangaˆmiut dyke samples within this study Table 1, and are usually associated with calc – alkaline
fractionation. The presence of albite-rich plagio- clase phenocrysts within the dyke centre may
explain the higher Na
2
O and Sr contents but lower CaO contents than would be expected from
the modelled fractionation trends. However, for most elements the evolutionary trends of trace
Fig. 6. Biaxial plots showing fractionation trends for: a Sr ppm vs Zr ppm; b Sc ppm vs Zr ppm; and c Ni ppm vs Zr
ppm. Other details as for Fig. 3 and Fig. 4.
ation. These results therefore show that horn- blende fractionation dominated intradyke pro-
cesses, but considerably more fractionation than using a ‘tholeiitic’ fractionation assemblage of
clinopyroxene and plagioclase is required to pro- duce a given range in concentrations.
4
.
5
. Crustal contamination The influence of crustal contamination as a
cause of trace element enrichment in continental flood basalts and mafic dyke swarms remains a
matter of controversy after many years of debate. Although it is adhered to by many petrologists
e.g. Thompson et al., 1982, 1983; Seifert et al., 1992; Arndt et al., 1993, others have preferred
Fig. 7. a Multielement diagram and b rare earth element diagrams for samples 416009 and 416019 including modelled
solution for 82 clinopyroxene-plagioclase fractionation, us- ing data from Table 1. Primordial mantle abundances from
Sun and McDonough 1989. Chondrite values from Naka- mura 1974. Ho value estimated from Y concentration.
the fractionating assemblage, much better fits were obtained for most elements. The best fit was
obtained using 32 hornblende, 20 clinopyrox- ene, 18 plagioclase and 1 magnetite. The lack
of HREE depletion again concurs with the petro- graphic evidence that garnet growth is post-mag-
matic and that it did not play any part in fraction-
Fig. 8. Incompatible element and rare earth multi-element mantle normalised diagrams for samples 416005 and 416007
and modelled solutions for: 1 60 clinopyroxene – plagioclase fractionation triangles and 2 71 clinopyroxene – plagio-
clase – hornblende – magnetite fractionation diamonds. Pri- mordial Mantle abundances from Sun and McDonough
1989. Chondrite values from Nakamura 1974. See Table 3 for distribution coefficients used.
Bridgwater et al. 1995 preferred crustal con- tamination as the enriching mechanism on iso-
topic grounds. 2. Contamination may be more subtle. For ex-
ample, the concept of contamination by selec- tive
element diffusion
Watson, 1982;
Blichert-Toft et al., 1992, rather than by bulk mixing, has made it a lot more difficult to
discount crustal contamination on mass bal- ance arguments.
Although some recent highly integrated studies using many different analytical techniques have
successfully resolved the influence of both pro- cesses e.g. Shirey et al., 1994; Hawkesworth et
al., 1995; Gibson et al., 1996, when using only major and trace elements such problems must be
taken into consideration. However, the very good fit of fractional crystallisation modelling obtained
for a variety of elements does suggest that large amounts of contamination did not occur during
or after fractionation of the magma. The fact that the dykes were intruded over a wide area into
different rock types, but have similar spidergram patterns also suggests that significant amounts of
contamination did not occur.
It is much more difficult to rule out massive selective contamination of primary magmas prior
to fractionation of the Kangaˆmiut dyke magma. It is possible for contamination to have occurred
in sub-crustal magma chambers e.g. Cox, 1980. However, the general absence of depleted MORB-
type magmas in the continental crust does suggest that the mantle source regions of flood basalts are
enriched prior to their ascent into the crust, as contamination within the crust is unlikely to be a
highly efficient process unless magmas are able to flow turbulently and hence mix thoroughly — an
unlikely occurrence in tholeiitic basaltic melts Kerr et al., 1995a.
4
.
5
.
1
. Pb isotopic ratios of Kangaˆmiut dykes Pb-isotopic ratios in mafic rocks are a particu-
larly sensitive indicator of possible crustal assimi- lation
and contamination
during magma
emplacement, given the order of magnitude lower concentration of Pb in mantle derived rocks and
the strongly contrasting and distinctive isotopic signatures developed in old continental crustal Pb
pre-contamination of the mantle source mantle metasomatism as a more viable enrichment
mechanism e.g. Sheraton and Black, 1981; Ellam and Cox, 1991; Hergt et al., 1991; Tarney, 1992
or both Gibson et al., 1996. The difficulty in resolving which process is dominant has two prin-
ciple causes.
1. In some instances the use of different ap- proaches has led different researchers to come
to strongly contrasting conclusions. For exam- ple, Hall and Hughes 1987 suggested mainly
on trace element grounds that the enriched character of the Early Proterozoic High-Mg
boninitic in West Greenland, could be best explained by mantle metasomatism, whereas
reservoirs e.g. Moorbath and Welke, 1969; Dickin, 1981. In a related study, whole-rock Pb-
isotopic ratios for 20 samples from two composite dykes south of Sondre Strømfjord are presented
in Table 2 and Fig. 10. In the
207
Pb
204
Pb vs
206
Pb
204
Pb diagram, these data define a regression line corresponding to an age of 2.08 9 0.10 Ga
2s; MSWD = 22, which is indistinguishable from the more precise U – Pb zircon ages of ca.
2.04 Ga Kalsbeek and Nutman, 1996a,b. The single stage m
1
value
238
U
204
Pb of 7.9 for this regression is typical for mantle derived magmas in
this region. In the
208
Pb
204
Pb vs
206
Pb
204
Pb dia- gram, a roughly linear trend corresponding to a
ThU ratio of about 4 typical of mantle-derived mafic rocks is observed. Plotted for comparison
in both diagrams are whole-rock Pb-isotopic ra- tios from Archaean gneisses in the Nagssugtoqid-
ian orogen Whitehouse et al., 1998. These data show a considerable range of compositions,
reflecting their complex crustal history. At the time of emplacement of the Kangaˆmiut dyke
swarm at ca. 2.04 Ga, the composition of these gneisses would lie to the left of their present
position in the diagram, along isochrons parallel to that defined by the dykes themselves, retaining
a broad range in
207
Pb
204
Pb ratios which has clearly not influenced the isotopic compositions of
the dykes to any significant extent.
4
.
6
. Mantle melting and metasomatism Metasomatic modification of mantle sources
has been preferred by many petrologists to crustal contamination due to the much lower degree of
Table 2 Pb isotope data for Kangaˆmiut dykes
a
Sample c k
2
m
2
m
1
2.04 Ga
208
Pb
204
Pb
207
Pb
204
Pb
206
Pb
204
Pb 65°56.5N 53°29.0W
158074 15.438
18.088 38.127
7.82 9.65
3.91 7.87
4.47 158076
3.69 16.161
15.219 35.871
24.585 45.189
7.75 27.10
3.85 158077
16.216 7.97
19.68 158078
3.95 21.824
15.984 42.373
158078 3.96
19.69 7.99
42.409 15.992
21.828 5.86
7.86 36.598
3.98 15.281
16.680 158079
158081 7.86
21.097 17.73
3.08 15.834
39.915 19.066
15.561 39.218
158082 7.82
12.27 3.91
158083 12.37
19.104 15.565
39.189 7.82
3.86 15.540
18.823 3.89
158084 11.62
7.84 38.914
18.899 15.530
39.084 7.80
11.83 158086
3.95 66°01.4N 53°28.5W
15.203 35.918
7.79 5.02
3.37 16.367
158119 18.258
15.518 38.346
158120 7.94
10.10 3.94
158121 18.994
15.593 39.196
7.90 12.08
3.96 158123
19.041 15.601
38.859 7.91
12.21 3.66
20.667 15.777
40.491 158125
7.85 16.57
3.62 14.87
7.92 15.732
20.033 40.196
158126 3.85
20.509 158127
4.11 16.15
7.91 41.167
15.790 4.05
15.86 7.89
40.941 20.399
15.763 158129
14.33 7.87
39.597 3.60
15.681 19.832
158128
a
Samples were analysed using a VG-54E mass spectrometer at the University of Oxford using previously described analytical procedures Whitehouse, 1990. Pb-isotope ratios have been corrected for mass fractionation of ca. −0.15 per atomic mass unit,
based upon replicate analyses of NBS 981 Pb, and have an overall analytical error of ca. 9 0.1 2s. Model parameters: m
1
and m
2
are the modelled
238
U
204
Pb ratios respectively before assuming a single stage of evolution and since 2.04 Ga; k
2
represents corresponding post-2.04 Ga ThU ratios, assuming a pre-2.04 Ga k
2
value of 4.
Fig. 9. Log NbY vs log Nb plot showing Kangaˆmiut dykes and modelled partial melts: all partial melts from Fertile
MORB Mantle FMM composition of Pearce and Parkinson 1993. Numbers next to symbols and lines denote percent
mantle melting. Mantle mineralogical composition: grey cir- cles-14 Cpx, 43 Opx, 43 Ol. Grey squares — 14 Amph,
43 Opx, 43 Ol. Black circles — 5 Cpx, 47.5 Opx, 47.5 Ol. Black squares — 5 Amph, 47.5 Opx, 47.5 Ol.
Solid enclosure-data field for Labrador Kikkertavak dykes. Dashed enclosure — data field for West Greenland sediments
collected during this study.
were calculated firstly for 1 – 9 partial melting for peridotitic sources containing 14 and 5
clinopyroxene. In each case the rest of the source was assumed to be comprised of equal amounts of
olivine and orthopyroxene. Following the ap- proach of Pearce and Parkinson 1993, clinopy-
roxene was taken to melt at twice the rate of olivine and orthopyroxene.
Fig. 9 shows these batch melt compositions plotted with Kangaˆmiut dyke samples and the
previously discussed modelled cpx-plag fractional crystallisation solution. A line for 50 olivine
fractionation is also shown. It is clear from the figure that the Nb and Y concentrations of pri-
mary Kangaˆmiut dyke magmas could be pro- duced from moderate degrees of melting of a
peridotite composition similar or perhaps slightly less depleted than the Fertile Morb Mantle com-
position used. This approximation is not unrea- sonable when considering the evidence that by
Paleoproterozoic times a degree of upper mantle depletion had occurred e.g. DePaolo, 1981; Ben
Othman et al., 1984.
The evidence that the Kangaˆmiut dyke magmas contained primary hornblende suggests that the
magmas may have become ‘wet’ from mantle modified by subduction-related metasomatic pro-
cesses. Thus identical melting calculations were performed using the same compositions and melt-
ing rates but replacing clinopyroxene with amphi- bole. The compositions derived had very low
NbY ratios and were unsuitable as primary mag- mas for the Kangaˆmiut dykes Fig. 9. The lack of
very strong island-arc type Nb anomalies or inter- dyke calc – alkaline fractionation trends, and their
similarities in NbY and Nb values to other dyke swarms which do not show evidence of intradyke
calc – alkaline
fractionation e.g.
the Early
Proterozoic Kikkertavak
dyke swarm
from Labrador; Fig. 9 also suggests that the magmas
and their source were originally no more strongly hydrated than other CFB-type magmas. However,
this does not rule out the presence of small quan- tities of hydrous minerals such as amphibole or
phlogopite being present in a metasomatised mantle source.
input of crustal material required. For example, Hergt et al. 1991 showed that up to 3 input of
sediment subducted into the mantle could account for the trace element signature of Mesozoic
Gondwanaland basalts. In other cases, mantle metasomatism is preferred because the local
crustal types may not make suitable contaminants for the production of the dyke geochemical signa-
ture e.g. Weaver and Tarney, 1981.
When considering the case for metasomatism, it is first necessary to approximate the chemical
composition of the mantle source prior to metaso- matism. Studies of mafic magmatism in arc ter-
ranes using island arc basalts and boninites Pearce and Parkinson, 1993 has shown that this
could be successfully modelled using elements least likely to be enriched by sediment, fluid or
crustal input. Therefore, an initial approach was to calculate batch melting curves for the element
Nb and the ratio NbY for a variety of possible mantle sources. In each case the trace element
composition of the mantle used was the Fertile Morb Mantle FMM composition of Pearce and
Parkinson 1993. The composition of batch melts
When considering the nature of a chemical component potentially added via subduction to
the mantle source prior to magmatism, the best analogues are likely to be found amongst the
substantial amounts of paragneisses preserved within the Nagssugtoqidian orogenic belt Fig. 1.
Fig. 10.
208
Pb
204
Pb and
207
Pb
204
Pb vs
206
Pb
204
Pb diagrams illustrating whole-rock Pb-isotopic compositions of the Kangaˆmiut dykes in this study open circles. Data from Archaean gneisses in the Nagssugtoqidian orogen are shown for comparison crosses
and squares. Plotted symbols are generally larger than the analytical error. Growth curves assume m
1
= 7.9 and k
1
= 4. Data from
the southern Nagssugtoqidian orogen, highlighted as square symbols.
Table 3 Distribution coefficients used in modelling for the major, trace and rare earth elements
a
K
2
O TiO
2
Nb Zr
Y Sr
Sc Ni
Cr La
Plagioclase 0.17
0.04 0.01
0.05 0.03
1.8 –
– –
0.14 0.40
0.01 0.10
0.90 0.06
0.04 2.5
Clinopyroxene 4
5 0.8
0.96 Hornblende
1.50 0.8
0.50 1.0
0.46 –
– –
0.27 7.50
0.80 0.27
0.20 –
– –
– Magnetite
– –
Nd Sm
Eu Gd
Tb Ce
Dy Er
Yb Lu
0.08 0.08
0.32 0.10
0.03 0.14
0.09 Plagioclase
0.08 0.07
0.08 Clinopyroxene
0.31 0.15
0.50 0.51
0.61 -
0.68 0.65
0.62 0.56
0.34 0.91
1.01 1.10
1.4 0.34
0.64 Hornblende
0.48 0.97
0.89 –
Magnetite –
– –
– –
– –
– –
a
Values from Henderson 1982 and Rollinson 1995
Thus 14 samples of metasediments were collected from the Kangaksiak and Nordre Strømfjord re-
gions Fig. 1 and analysed in order to ascertain their average composition. Following Hergt et al.
1991, this average composition was added to the FMM mantle composition in the ratio 3:97. Cal-
culations show that magmas compositionally sim- ilar to the Kangaˆmiut dykes could be produced
through second stage fractional melting of the source, using 5 melting increments. However,
uncertainty over the ages of the paragneisses and the strong likelihood that the metasomatic com-
ponent will be formed by the complex addition of fluids or melts to the mantle wedge rather than by
simple bulk mixing means that we can only ad- vance metasomatism as the most likely cause of
the
incompatible-element enrichment
in the
Kangaˆmiut dyke magmas.
5. Discussion