- © The Mineralogical Society
Reinvestigation of the top 25 m of the 165 m thick Shiant Isles Main Sill requires reinterpretation of its internal structure, especially of the last unit emplaced – the granular olivine picro-dolerite. New mineral analyses allowere-assessment of the magmatic processes involved in the intrusion and differentiation of this classic multiple sill. C-shaped vertical mineral composition profiles, which occur in olivine-rich, alkaline basaltic sills, reflect reaction between phenocrysts and variable composition liquids. A case is made for vertical migration of interstitial liquids within and across the boundaries of intrusive units, for the mixing of such liquids from adjacent units, and for their influence on mineral chemistry on both sides of the interface.
The Shiant Isles lie between the Outer Hebrides and the Scottish mainland 24 km N of Skye and are formed of several Tertiary alkali basaltic sills intruded into Jurassic sedimentary rocks. The Main Sill is 165.6 m thick and is a classic example of a differentiated multiple sill (Walker, 1930; Drever and Johnston, 1965; Gibb and Henderson, 1989). Gibb and Henderson (1996) used cored drill holes and outcrop samples to define the ‘stratigraphy’ and mineral compositional variations throughout the sill. Four intrusive units were identified, in order of emplacement:
Olivine teschenite; 2 m thick, emplaced as a magma carrying phenocrysts of olivine with mg [= atom. % Mg/(Mg+Fe+Mn)] ≥83 and possibly of plagioclase (An ≥76).
Picrite; 24 m thick. The olivine teschenite was intruded concordantly by a magmatic mush of liquid and olivine phenocrysts (mg ≥83) to form a picrite sill with a D-shaped modal olivine profile. Also present in the picrite magma were discrete spinel micro-phenocrysts with mg ≥12 and cr+al [= mol.% (chromite + spinel)/(chromite + spinel + ulvospinel)] ≥28 (note that coexisting spinel inclusions within olivines are more primitive with (cr+al) >50). Small amounts of plagioclase phenocrysts (An ≥86) might also have been present.
Picrodolerite-crinanite; 140 m thick. A substantial intrusion of magma, carrying about 10% olivine (mg ≥83) together with some calcic plagioclase (An ≥85) as phenocrysts, was emplaced high in the picrite unit before the host rock was completely solidified, splitting the picrite into upper and lower leaves. The olivine phenocrysts settled towards the bottom to form the picrodolerites.
Granular olivine picrodolerite. This unit was believed to be <2 m thick and intruded into the top of the picrodolerite-crinanite unit as a magma containing over 30% modal olivine microphenocrysts, again before the host rock was fully solid. The olivine in the granular olivine picrodolerite is rather more evolved (mg ≥77) than in the other units and does not contain inclusions of spinel.
Textural relations are consistent with the bulk of the plagioclase and probably all of the clinopyroxene having crystallized within the sill after emplacement. By contrast, most of the olivine (and spinel?) crystallized earlier and was transported as phenocrysts in the intruding magmas, often with flow differentiation causing axial migration to form olivine-rich ‘plugs’ in the central parts of the feeder conduit(s) (Gibb, 1968; Komar, 1972). During transport and after emplacement, these phenocrystic olivines became slightly zoned towards less mg compositions, but the strong zoning towards very Fe-rich compositions (Johnston, 1953; Gibb and Henderson, 1996) only occurs in ophitic olivines and clearly formed by in situ fractional crystallization.
Further study of the upper part of the sill during a Sr and Nd isotopic investigation of the sill (Foland et al., 2000) requires redefinition of the granular olivine picrodolerite unit. In this paper we present new microprobe data for the top 25 m of the sill, an updated composite section, and a reassessement of the petrogenetic implications and pre- and post-emplacement processes.
Revised section through the sill
Gibb and Henderson (1996) considered the granular olivine picrodolerite [GOPd] to form a 1.7 m thick unit of relatively homogeneous picrodolerite characterized by ~34% of small subhedral olivines which impart a distinctive granular appearance to the rock; the Fe-Ti oxides also have a granular habit. This olivine-rich picrodolerite has fairly sharp upper and lower boundaries against similar, but less olivine-rich, rocks with mineral constitutions and textures not unlike finer-grained variants of the crinanite.
Detailed re-examination of the Upper Discontinuity and the subjacent part of the sill has revealed that the GOPd is in fact 3.96 m thick (extending from 162.10 m down to 158.14 m in Fig. 1⇓) and the olivine-rich variant forms the middle part of the unit. (Specimen heights given in italics refer to heights in the composite section shown in Fig. 1⇓ and are consistent with those given in Gibb and Henderson (1996)). Immediately below the medium-feldspar picrite at the Upper Discontinuity is a thin (6–9 mm) zone of very fine-grained rock containing up to 50% of small granular olivines and below this is a similar rock but with only 15% olivine. The grain size of the latter increases downwards over a few cm but it remains a relatively fine-grained rock characterized by acicular plagioclase laths and small granular olivines. Originally interpreted as chilled crinanite (Gibb and Henderson, 1989), this is now recorded as the upper chilled margin of the GOPd unit. This fine-grained GOPd extends down for 0.7 m where it gives way to the more easily recognized olivine-rich variant for 1.7 m before reappearing to form the bottom 1.6 m of the unit. Over the lowermost few cm the rock again becomes very fine-grained and represents the lower chilled margin of the unit.
The GOPd is markedly heterogeneous with mm- to cm-scale regions with textures distinct from those of the granular host rocks. Usually these regions have fairly sharp boundaries against the host rock and are interpreted as xenoliths. Most of the xenoliths are coarser than the host rock and are of crinanite or pegmatitic crinanite, suggesting that they were derived from, or are genetically related to, the underlying picrodolerite-crinanite unit. The crinanitic nature of the xenoliths suggests that they could have been picked up from the upper parts of the crinanite during lateral emplacement of the GOPd. Some of the crinanitic regions have gradational boundaries and could be either partially reacted xenoliths or patches representing interstitial liquid which had migrated upwards from the still unconsolidated upper part of the picrodolerite-crinanite unit. Henceforth, the terms ‘crinanite xenoliths’ and ‘crinanitic patches’ are used to distinguish regions with sharp and more diffuse boundaries respectively.
The fairly abrupt changes in modal olivine content across the internal boundaries of the GOPd unit can be attributed to fluid mechanical processes accompanying intrusion of the magma with the most olivine-rich part representing axial concentration of small olivine phenocrysts during plug flow. The asymmetrical location of this plug in the upper part of the GOPd (Fig. 1⇑) suggests that it formed during magma uprise in a feeder system, followed by lateral emplacement into the upper part of a slightly earlier-intruded, less-olivine-rich portion of GOPd magma.
Recognition of the true nature and extent of the GOPd means that: (1) the Upper Discontinuity is in fact the junction between picrite and chilled GOPd and not between picrite and crinanite as previously thought; (2) the GOPd was intruded between the upper picrite and the picrodolerite-crinanite unit; and (3) the original upper contact of the third intrusion (picrodolerite-crinanite) with the second intrusion (picrite) has been swept away by the subsequent emplacement of the GOPd. This interpretation explains the irregular but sharp nature of the bottom of the medium-feldspar picrite at the Upper Discontinuity where it has the appearance of having been broken off mechanically, suggesting it was either solid or a semi-rigid crystal mush (Marsh, 1996) at the time that the GOPd was intruded.
Mineral chemistry and petrogenetic implications
The mineral analyses database for the sill (Gibb and Henderson, 1996) was thought to contain only one GOPd specimen (SC1047 (160.9 m)) but two of the rocks then classified as crinanites (SC1038 (162.0 m) and SC1054 (159.4 m)) are now recognized to be GOPds. The database for the top 25 m of the sill has now been augmented by a fourth GOPd (SC1060T (158.4 m)), two picrites (SC1019 (163.7 m) and SC1027 (163.2 m)) and one of the highest crinanites (SC1081 (156.2 m)). Most of the new data were obtained by energy dispersive EPMA.
Many of the magmatic minerals show exceptionally wide ranges of zoning with some rocks apparently having crystallized under conditions of almost perfect in situ fractional crystallization (Johnston, 1953; Gibb, 1973; Gibb and Henderson, 1996). Under these circumstances it is difficult to obtain the absolute extremes of composition, especially as many of the crystals have poikilitic textures. In addition, deuteric and/ or subsequent hydrothermal alteration of the crystal margins might conceal the full primary compositional ranges.
For the silicates, the compositional ranges represent maximum intra-grain variations, but individual spinel grains tend to be fairly homogeneous and the ranges represent inter-grain compositional differences. The latter probably reflect different amounts of late-magmatic and/or subsolidus re-equilibration between the spinels and surrounding silicates (Henderson and Suddaby, 1971; Sack and Ghiorso, 1991). Average compositions appear to define the clearest compositional trends for the spinels. Representative analyses for minerals from the upper picrite, GOPd, and upper part of the picrodolerite-crinanite unit are given in Tables 1⇓–3⇓⇓. The compositional relations will be summarized for each mineral type and then discussed in more detail in the context of the intrusive units and their petrogenesis.
Olivines and pyroxenes
The Fe-Mg zoning ranges of olivines and clinopyroxenes from the upper picrite, GOPd and upper part of the picrodolerite-crinanite are plotted in Fig. 2⇓. The picrites contain the most primitive olivines and the crinanites the most evolved, with the GOPd olivines being intermediate. The topmost samples from the upper picrite (163.7 m and 163.2 m) have clinopyroxenes with lower mg values than those of most of the picrites and picrodolerites lower in the sill, and trend towards the evolved compositions shown by clinopyroxenes in the olivine teschenite margins of the sill (Fig. 2⇓; cf. Figs 5 and 6, Gibb and Henderson, 1996).
The mgmax [maximum mg in zoning range] values for the olivines from the upper teschenite and upper picrite are generally higher than those for the coexisting clinopyroxenes which is also the case for the lower picrite, lower teschenite and the lowermost picrodolerites. This relationship is not an equilibrium Fe/Mg distribution and reflects the fact that the more primitive olivine phenocrysts were transported from a deep crustal magma chamber, whereas the clinopyroxenes did not crystallize until after emplacement.
The kaersutitic amphiboles from the upper picrite have similar compositions to those in the lower picrite and picrodolerite, except that those from the upper picrite are richer in K than those from the lower picrite (Table 1⇑; cf. Table 3⇑, Gibb and Henderson, 1996).
Representative data are plotted in Figs 2⇑ and 3⇓ from which it can be seen that the overall Ca-Na zoning ranges are remarkably similar in all three units (cf. Figs 6 and 8, Gibb and Henderson 1996). Many of the rocks have plagioclase with very similar zoning ranges down to about An20, with similar Or contents, e.g. Or1.5–2.5 at An40 (Fig. 3⇓). At this An content, the plagioclases from the upper picrites are significantly more potassic (Or2.0–2.5) than those from the lower picrite (Or1.0–1.5). Also the second lowermost GOPd (SC1054) and the topmost crinanite (SC1081) have plagioclase with more elevated K contents (Or3.5), while the GOPd from the bottom of the unit (SC1060) is higher still at Or6.5. These distinctive elevated Or contents become apparent at quite high An contents early in the zoning trends (from at least An60). It seems that the rocks immediately above and below the bottom of the GOPd are unusually enriched in K.
Fe-Ti oxides and FeS
The upper picrites and the GOPd samples SC1047 and SC1060 contain coexisting spinel and ilmenite (Table 3⇑) similar in composition to those from lower in the sill (Table 5, of Gibb and Henderson, 1996). The temperatures and fO2 values calculated using the method of Powell and Powell (1977) for coexisting magnetite and ilmenite cover the ranges 826 to 896°C and 10−16.7 to 10−13.7 atm. (Table 3⇑). These values fall within the range reported by Gibb and Henderson (1996) with fO2 varying from ~ 1 log unit less oxidizing than the quartz-fayalite-magnetite buffer down to the magnetite-wüstite buffer.
Pyrrhotite is present in all of the upper picrites but occurs in only one sampled lower picrite, near the bottom of the sill. Pyrrhotite also occurs in GOPd SC1060 and in the host and crinanite xenoliths present in GOPd SC1047. Gibb and Henderson (1996) reported that pyrrhotite is the main opaque mineral in the olivine teschenite margins of the sill and is ubiquitous in the pegmatite segregations in the upper part of the crinanite but that it occurs in only two of the 15 samples studied from the picrodolerite-crinanite unit. The localized occurrence of pyrrhotite suggests that it formed during the later stages of magmatic fractionation with the sulphur possibly being introduced in aqueous fluids derived from the sedimentary country rocks (Gibb and Henderson, 1996).
Olivine teschenite and picrite units
Variations in mineral composition with height in the upper part of the sill are shown for olivine, clinopyroxene, plagioclase and spinel in Fig. 2⇑. Throughout the olivine teschenite and upper picrite, mgmax of olivine varies only between 80.7 and 82.9: this is consistent with the earlier suggestion that the first two magmas to be intruded contained phenocrysts of olivine with mg ≈ 83. The upper picrite olivines tend to be more zoned than those from the lower picrite.
The spinels from the upper picrite are the least evolved with mg values suggesting that they were microphenocrysts in the intruded magma along with the olivine phenocrysts. Curiously, these values are somewhat higher than for the spinels in the lower picrite and this might be related to re-equilibration reactions involving pyrrhotite which leave the residual spinels enriched in Mg, Cr and Al relative to Fe. By contrast, the lower mg and cr+al values for the spinel in the upper olivine teschenite show that it is a late-crystallizing phase in this unit.
The Anmax [maximum mol.% CaO/(CaO+Na2O+K2O)] in the upper picrite varies only from 75.3 to 76.4 which is significantly smaller than the range of Anmax in the lower picrites and slightly more An-rich than values in the overlying olivine teschenite. However, with the possible exception of one of the lower picrites, SC642 (3.67 m) (Fig. 6, of Gibb and Henderson, 1996), the plagioclase zoning ranges are remarkably similar throughout the whole of the picrite and olivine teschenite units which is consistent with the bulk of the plagioclase crystallization having occurred within the sill in the presence of fractionating liquids of broadly similar compositions.
The clinopyroxenes from the upper picrite have mgmax values which decrease upwards while those from the overlying olivine teschenite have slightly higher mgmax values than those of the topmost picrite (Fig. 2⇑). The upward ‘evolution’ of clino-pyroxene composition in the upper picrite mirrors the downwards trend at the bottom of the lower picrite. Since the clinopyroxene crystallization occurred in situ after emplacement, the variation of clinopyroxene mgmax with position in the original picrite sill suggests that the liquid nearer the edges of the D-shaped modal olivine profile was somewhat more evolved than that at the centre, with the marginal magma approaching the adjacent olivine teschenites in composition.
The C-shaped clinopyroxene composition profile for the entire picrite unit is the opposite of what would be expected in a sill formed by fractional crystallization of a single intrusion of homogeneous magma where inward crystallization from the margins would result in the later, more compositionally evolved minerals occurring in the central part of the intrusion. Such a ‘normal’ fractionation pattern occurs, for example, in the main central unit of the Palisades Sill (Walker, 1973; Shirley, 1987). However, the C-shaped profiles found in differentiated alkaline basaltic sills (e.g. Gibb and Henderson, 1978; Marsh, 1996) appear to be related to either (or both) the presence of olivine-rich central zones or temporal variations in the chemistry of the intruded magma. Both are almost certainly functions of the way in which compositionally-zoned, lower-level basaltic magma chambers are tapped with the entrainment of olivine phenocrysts in the liquid.
As mentioned above, the original bottom of the upper picrite and top of the crinanite with their mutual contact must have been swept away by intrusion of the GOPd. The uppermost crinanites remaining in the sill show ranges of olivine, pyroxene and plagioclase zoning which are smaller than in crinanites lower in the unit but are comparable with those for the picrodolerites (Fig. 2⇑; Figs 5 and 6, from Gibb and Henderson, 1996). However, olivine mgmax values for these uppermost crinanites are more Fe-rich than for any of the underlying crinanites and picrodolerites. Also, the olivine from the highest crinanite analysed (SC1081 (156.2 m)) is surprisingly homogeneous with a relatively evolved composition (mgmax = 51.2). For the two highest crinanites analysed, at 156.2 m and 152.6 m, the differences between the mgmax values of the olivines (51.2 and 57.2) and coexisting clinopyroxenes (76.9 and 78.4) are much smaller than for rocks lower down the picrodolerite-crinanite unit and in the picrite unit. In basic intrusions the mg value for olivine is invariably lower than that of coexisting clino-pyroxene, with the difference increasing as the rocks become more Fe-rich (Wager and Brown, 1968; Wadsworth, 1988). It is therefore likely that the olivine-pyroxene relationships in the uppermost crinanites are a consequence of both minerals co-precipitating in situ, although reaction with migrating more-evolved interstitial liquids might subsequently have changed them. The pegmatitic horizons and patches within the upper crinanites show that such liquids existed during the later stages of solidification of the unit. It is also possible that reaction with volatile-rich, evolved liquids has resulted in more homogeneous minerals in the uppermost crinanites than those characteristic of the crinanites lower down. In contrast, the olivine-pyroxene relationships lower in the picrodolerite-crinanite unit and in the picrite are likely to have been affected, if not controlled, by the abundant phenocrysts of relatively primitive olivine present in the magma at the time of emplacement.
Granular olivine picrodolerite unit
The mineral compositions in this last intruded unit need careful evaluation because of the way they vary with height and because of the heterogeneous nature of the rocks. The compositional ranges of minerals from the finer, more even-grained textural regions are distinguished in Fig. 2⇑ from those of coarser, crinanitic patches and crinanite xenoliths. Distinct textural variants are not obvious in SC1054 (159.4 m), but there appear to be two populations of spinels (Fig. 2⇑). The crinanite xenoliths and crinanitic patches have olivines with lower mg values, and spinels with significantly lower mg and cr+al values, than those from the host GOPd. Plagioclases from the crinanite xenoliths and crinanitic patches tend to be less An-rich than plagioclases from the host, albeit overlapping, whereas the clinopyroxenes have more similar composition ranges.
The following discussion focuses on the compositions of minerals from the GOPd itself rather than on those from the crinanite xenoliths and crinanitic patches. The mgmax in the olivine and average mg values for the spinel in the GOPd are lower than those throughout the picrites suggesting that these microphenocryst phases crystallized from a more-evolved liquid than in the case of the picrite. In contrast, the compositions of clinopyroxene and plagioclase are similar in these two intrusive units, consistent with these minerals having crystallized after emplacement, from liquids of similar compositions. Although the average plagioclase compositions and zoning ranges in the GOPd are similar to those in the subjacent 6 m of crinanite, the olivine, spinel and clinopyroxene relations all point to the GOPd magma being more primitive than the crinanite.
Within the GOPd, the mgmax of the olivine and the average mg of the spinel show clear C-shaped trends with height with the most primitive compositions occurring in the inner, olivine-rich core of the unit (SC1047 (160.9 m)). Further, the sample from the bottom of the unit is slightly more evolved than that from the top. The clinopyroxene mgmax shows a similar but smaller increase towards the centre of the unit. The average cr+al values in spinel also show central enrichment and the bottom sample is again more evolved than that at the top. The C-shaped mineral compositional trends with height are similar to that described above for the clino-pyroxene in the entire picrite unit. This implies that the liquid phase of the GOPd magma with its flow-differentiated ‘plug’ of olivine (and spinel) phenocrysts was compositionally zoned with the outer parts being more evolved than the central zone. In contrast to the picrite unit, this compositional variation within the liquid appears to have exerted some influence on the olivine trend, the difference between the two units possibly being related to olivine content.
The fact that the sample from the bottom of the GOPd is more evolved than that at the top suggests that either the melt filling the lower part of the GOPd was more evolved than that being emplaced at the top, or that some other post-intrusive process was operational, possibly involving uprise and reaction of evolved interstitial liquid from the underlying unit. This could also have affected the olivine trend.
Migration and mixing of interstitial liquids between the upper crinanites and GOPd
It is clear that highly evolved interstitial liquid compositions developed during the differentiation of the picrodolerite-crinanite unit. The pegmatites present in the upper crinanite (Figs 1⇑ and 2⇑) are believed to have formed by migration of such liquids into lower pressure regions within the upper crinanite solidification front (Marsh, 1996; Gibb and Henderson, 1996). The two pegmatites in Fig. 2⇑, at 157.8 m and 154.1 m, contain low mg clinopyroxenes trending towards aegirine-rich varieties, low-Ca anorthoclases, K-rich alkali feldspars and low mg and cr+al spinels. Pyrrhotite is present in both samples. Thus the pegmatites and their constituent minerals are markedly more-evolved than their host crinanites.
The late-magmatic kaersutitic amphiboles in the upper picrites are much more potassic (average atom.% K/(K+Na) = 14.4) than those in the lower picrites (5.7) and the plagioclases from the upper picrite are significantly more K-rich than those from the lower picrite. These differences in K-contents could be attributed to uprise of highly-evolved interstitial liquids from the crinanite into the overlying picrite, before the intrusion of the GOPd. These liquids might also have introduced the sulphur necessary for the formation of the pyrrhotites common in the upper picrites but rare in the lower picrites. These uprising liquids would have reacted directly with pre-existing minerals as well as mixing, mechanically and/or by diffusion, with locally developed residual liquids to produce intermediate interstitial liquid compositions.
Further, it is possible that evolved, pegmatitic interstitial liquid might still have been present in the underlying upper crinanites at the time of intrusion of the GOPd and might have migrated upwards into the GOPd. The crinanitic patches in the GOPd have mineral compositions (Fig. 2⇑) intermediate between those in their GOPd host and those in the pegmatites and resemble those in the uppermost crinanites. They could be accounted for by crystallization from mixtures of GOPd interstitial liquid and uprising interstitial (pegmatitic) liquid from the crinanite. As such they would represent ‘relics’ marking the flow paths of the migrating interstitial fluids. Since the olivine-rich zone formed during intrusion of the GOPd was not significantly disrupted, an efficient process of intermixing of interstitial liquids is indicated rather than any in situ ‘whole’ magma mixing. Any upward movement of interstitial liquid from the crinanite must have been accompanied by downward movement of interstitial liquid from the GOPd. Although differing in detail, this migration could be similar to the “diapiric melt transfer” described by Helz et al.(1989) for the Kilauea Iki lava lake in Hawaii.
The extent of intermixing of the interstitial liquids would have decreased with distance from the interface between the units. It would therefore be expected that any reactions between upward migrating interstitial liquids and the pre-existing minerals would be most marked at the bottom of the GOPd. The olivine and spinel in the lowest GOPd analysed (SC1060 (158.3 m)) are the most evolved in the unit and are consistent with this model but, curiously, the clinopyroxene mg and plagioclase An do not appear to show any effects of such reactions. A further complication arises from the fact that the plagioclase in SC1060 is much richer in K than any other plagioclases from within the sill, including those from the pegmatites (Or2.0 at An40) and upper crinanites (Fig. 3⇑). This implies that there must have been a ‘richer’ source of K than the uprising interstitial liquids. One possibility is that K was introduced from the sedimentary country rocks either via a hydrothermal fluid phase or by bulk contamination during the passage of the GOPd magma through the country rock. The distinctive elevated Or contents arise very early in the zoning trend (from at least An60) implying a high temperature origin. This appears to favour bulk contamination. Contamination of the lowermost GOPd magma with sediment-derived K would also account for the elevated Or contents of plagioclases from GOPd SC1054 and crinanite SC1081. Such contamination, with the amount of K decreasing upwards in the GOPd unit, explains the relationship between SC1060 and SC1054, while exchange between interstitial liquids from the bottom of the GOPd and the underlying crinanite would extend the contamination to the rocks below the GOPd (e.g. SC1081).
Conclusions and emplacement model
All four units were emplaced with phenocrysts of early crystallized olivine and spinel suspended in magmas which were tapped from one or more underlying crustal magma chambers. The olivines in the first three units to be intruded (olivine teschenite, picrite and picrodolerite-crinanite) are more primitive (mg83) than those in the GOPd (mg77), and those in the picrites and lowermost picrodolerites contain inclusions of spinels which are more primitive than spinels forming discrete grains in the same rocks. The absence of spinel inclusions in the GOPd olivines might be related to their more evolved provenance. The average modal amounts of olivine in each unit are approximately: 10–15% (olivine teschenite); 40% (picrite); 10–15% (picrodolerite-crinanite) and 20% (GOPd). The different amounts of olivine suspended in the intruded magmas are likely to be a consequence of either the way in which a zoned lower level magma was tapped, or extraction from a complex magma chamber system similar to the ‘cedar-tree laccolith’, “magmatic mush column” postulated by Marsh (1996) to occur below highly active volcanic systems. The different sub-chambers in the latter type of magma system, although fed from essentially the same source in the upper mantle, could be at different stages of fractionation so that tapping from different levels could provide magmas with different amounts of olivine phenocrysts, at different stages of evolution, and with different liquid compositions.
The liquid phases of the magmas transporting the olivine and spinel from depth seem to have been at different stages of evolution when the four units were emplaced. The order of increasing evolution is picrite → picrodolerite → GOPd → olivine teschenite. Also, the picrite and GOPd liquids seem to have been more evolved at the margins than in the inner parts of the units. Some of these compositional differences may be inherited from the source magma chamber(s) but differentiation en route in the feeders could also have had a significant effect, as could the effect of phenocrysts on subsequent reactions during the crystallization of the liquids.
The D-shaped modal olivine distributions in the picrite and GOPd were not significantly disturbed after intrusion, indicating a lack of turbulent convection (Gibb and Henderson, 1992) and only modest amounts of post-intrusive gravity settling. By contrast, the phenocrystic olivines in the picrodolerite-crinanite unit settled readily after emplacement to produce the differentiation into crinanites and picrodolerites and the enrichment trend at the bottom of the latter (Fig. 1⇑). However, not all of the olivine was emplaced as phenocrysts. The ophitic olivines in the crinanite must have crystallized in situ and it is inevitable that some overgrowth occurred on the phenocrysts during crystallization of the liquids. The textural and compositional relations of the clinopyroxene and plagioclase necessitate that the bulk of these phases crystallized after emplacement of the different units. The zoning ranges to almost pure Fe and Na end-members, respectively, in the crinanite indicate that this crystallization took place under conditions of almost perfect fractionation within much of the picrodolerite-crinanite unit.
During the later stages of fractionation of the picrodolerite-crinanite unit, evolved interstitial liquids were segregated into ‘gravitational sags’ in the upper crinanite solidification front. This stage had been reached prior to the intrusion of the GOPd with some of the interstitial liquids rising into the overlying upper picrite, as evidenced by the crystallization of more potassic kaersutite and plagioclase than in the lower picrite.
The intrusion of the GOPd removed the upper contact between the picrodolerite-crinanite and picrite units and some fragments of solidified, or near solidified, crinanite were caught up as xenoliths. These xenoliths have sharp boundaries suggesting that there was little reaction with the enclosing GOPd magma. The GOPd also contains more evolved ‘crinanitic patches’ with indistinct boundaries which are interpreted as relics from the uprise of evolved crinanitic liquids from below which mixed with the GOPd liquid.
Plagioclases from the bottom of the GOPd and top of the picrodolerite-crinanite unit have K-contents significantly higher than those from other parts of the sill, including the pegmatites. This would seem to require a source of K richer than the interstitial liquids generated by fractional crytallization. One possibility is that the local sedimentary rocks might have introduced K (and other) contamination during the uprise and emplacement of the GOPd.
We thank Colin Donaldson for perceptive and constructive comments on an earlier version of this paper, David Plant (Manchester) for assistance with microprobe analysis, and Mike Cooper (Sheffield) for drafting the figures. We also thank Adam Nicolson for permission to undertake fieldwork on the Shiant Isles.
- Manuscript received 23 October 1999.
- Modified version received 1 March 2000.