Journal of Petrology Advance Access originally published online on December 13, 2006
Journal of Petrology 2007 48(3):563-587; doi:10.1093/petrology/egl072
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Infiltration Metasomatism of Cumulates by Intrusive Magma Replenishment: the Wavy Horizon, Isle of Rum, Scotland
1Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ, UK
2Institute of Theoretical Geophysics, Centre for Mathematical Sciences, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, UK
3BP Institute (Bullard Laboratories), Madingley Road, Cambridge CB3 0EZ, UK
RECEIVED NOVEMBER 25, 2005; ACCEPTED NOVEMBER 10, 2006
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONThe Eastern Layered Intrusion of the Rum Layered Suite comprises paired peridotite and allivalite (troctolite and gabbro) layers forming 16 macro-rhythmic units. Whereas the majority of these macro-units are believed to have formed by a process of crystal–liquid differentiation involving successive accumulation of crystals from multiple picritic replenishments of the chamber, the Unit 9 peridotite is interpreted as a layer-parallel picrite intrusion. Closely correlated with this discontinuous peridotite body is a distinctive feature generally known as the Wavy Horizon, which divides the overlying allivalite into a lower troctolite and an upper gabbro along a well-defined undulating contact. We propose that the Wavy Horizon is a metasomatic feature formed consequent to the removal of clinopyroxene from an original gabbroic mush. Foundering of the mush into the picritic sill resulted in the replacement of the original interstitial liquid by one saturated only in olivine (± plagioclase). Progressive through-flow of this liquid resulted in the stripping out of clinopyroxene from the lower parts of the allivalite. We interpret the Wavy Horizon as a reaction front, representing the point at which the invading liquid became saturated in clinopyroxene. The distinctive pyroxene-enriched zone immediately above the Wavy Horizon could have formed when mixing of the interstitial liquids on either side of the reaction front formed a supercooled liquid oversaturated in pyroxene, as a result of the curvature of the olivine–plagioclase–clinopyroxene cotectic. The presence of many such approximately layer-parallel features, defined by differences in pyroxene content, in the Eastern Layered Intrusion of Rum suggests that such an infiltration–reaction process was not unique to Unit 9.
KEY WORDS: cumulate; infiltration metasomatism; Rum; Eastern Layered Intrusion
|
INTRODUCTION |
|---|
The evolution of the crystal mush at the base of mafic magma chambers is a complex problem. In large chambers, in which the crystal pile can attain sufficient thickness to permit gravity-driven compaction (e.g. McKenzie, 1984
), the consequent upwards flow of liquid through the crystal mush may result in infiltration metasomatism as a result of the juxtaposition of fluids and solids out of chemical equilibrium (e.g. Irvine, 1980; Robins, 1982; McBirney, 1987; Boudreau & McCallum, 1992). In this contribution we present evidence that supports the action of infiltration metasomatism related to the emplacement of a relatively primitive picrite liquid under a pre-existing gabbroic cumulate pile in the periodically replenished Paleocene Rum magma chamber (Inner Hebrides, Scotland). |
GEOLOGICAL SETTING |
|---|
The Rum Layered Suite (Fig. 1a) is subdivided into the Eastern Layered Intrusion (ELI), the Western Layered Intrusion and the Central Intrusion, together with numerous subsidiary peridotite, gabbro and dolerite plugs (Emeleus, 1997
). The ELI is composed of a 750 m thick succession of 16 macro-rhythmic units (Brown, 1956; Volker & Upton, 1990; Fig. 1b), each comprising peridotite overlain by allivalite (the local name for feldspathic cumulates, which may be troctolitic or gabbroic). Numerous late-stage peridotite plugs occur in the Layered Suite itself, and in the Torridonian sediments to the north (Wadsworth, 1994; Emeleus, 1997; Holness, 1999). The Layered Suite is cut by numerous basaltic dykes and rarer picrite dykes. The latter are believed to be representative of the replenishing magma in the Rum chamber.
|
To a first approximation, these alternating layers can be attributed to crystal settling following successive intrusions of replenishing picritic magma (e.g. Brown, 1956
; Dunham & Wadsworth, 1978; Upton et al., 2002). However, there is clear field evidence (Butcher et al., 1985; Bédard et al., 1988) that the Unit 9 peridotite is an intrusive body emplaced into allivalite (the area labelled A in Fig. 1b cuts across the allivalite layering). Bédard et al. (1988) suggested that many of the other peridotite bodies of the ELI were also sill-like intrusions. However, Holness (2005) presented textural evidence that shows that the Unit 9 peridotite is likely to be the only major layer-parallel intrusive peridotite body in the ELI.
The intrusion of hot reactive picritic liquid into an allivalite crystal mush would juxtapose melt and crystals out of chemical and thermal equilibrium, thereby leading to mass mobilization and metasomatism. Bédard et al. (1988) and Hallworth (1998) suggested that the effects of such metasomatism could account for the enigmatic feature known as the ‘Wavy Horizon’ in Unit 9 (Fig. 2). This undulatory contact between troctolite and an overlying gabbro cuts across textural, grain-size and modal layering with little or no disturbance (Fig. 2a and b). Although early discussion of this feature revolved around an origin as a loading structure (Young & Donaldson, 1985; supported by Volker & Upton, 1990, 1991; but disputed by Bédard & Sparks, 1991), later interpretations are that it is a metasomatic reaction front.
|
Bédard et al. (1988
) suggested that the overlying gabbroic rocks formed as a consequence of metasomatic pyroxenization of a troctolitic mush by interaction with downwards-migrating basaltic melt segregated from solidifying picrite sills within the ELI. Hallworth (1998), Lo Ré (2004) and Holness (2005) suggested that the metasomatism was caused by the immediately underlying intrusive Unit 9 peridotite, with upwards migration of reactive metasomatizing fluids that stripped out pyroxene from an original gabbro. In this contribution we present a model for the formation of undulatory metasomatic contacts above bodies of reactive liquids emplaced into the crystalline material on the chamber floor. Our preferred explanation of the Wavy Horizon is the preferential removal of pyroxene from a gabbroic mush to leave a troctolitic residue. We suggest that this process is likely to play a significant role in the development of the crystal pile in open-system, frequently replenished magma chambers.
|
FIELD OBSERVATIONS |
|---|
Unit 9 as defined by Brown (1956
) comprises a laterally impersistent lower peridotite layer of variable thickness (0–25 m), overlain by an allivalite layer 15–25 m thick (Fig. 1b). However, as the Unit 9 peridotite is interpreted as an intrusive body (Butcher et al., 1985; Bédard et al., 1988; Holness, 2005), the over- and underlying allivalites are actually parts of a single original body termed the Unit 8/9 allivalite by Holness (2005). The area studied in detail comprises the eastern end of the peridotite body (labelled A) in Fig. 1b.
The Unit 8/9 allivalite on the north flank of Hallival forms a prominent NW–SE-trending shelf and escarpment. The overlying peridotite of Unit 10 forms a boulder-strewn slope rising from the back of the shelf. At its SE end, the allivalite shelf widens and is punctured by an irregular peridotite dome (labelled A in Fig. 1b) that becomes approximately layer-conformable towards the NW. Localized replacement of the allivalite by peridotite occurs along a fingered replacement front at the contact. No chilled margins are evident. Locally, the upper parts of the Unit 9 peridotite contain abundant xenoliths of allivalite (Bédard et al., 1988, fig. 14).
The Unit 8/9 allivalite far from the Unit 9 peridotite (sampled along Traverse f, Figs 1 and 3) is dominated by a coarse-grained (1–3 mm) gabbro. It is modally layered on the centimetre to metre scale (for modal compositions see Fig. 3), and although much of the unit is non-laminated, some layers have a lamination defined by alignment of plagioclase grains. Immediately above the Unit 9 peridotite, the Unit 8/9 allivalite comprises a lower troctolite and an overlying gabbro [shown in the detailed logs 9C, 9D and 9E of Bédard et al. (1998, fig. 17), and sampled by Traverses g and h, shown in Figs 1 and 3]. The troctolite and the lower regions of the gabbro have a well-developed foliation, and are layered in terms of grain size and modal mineralogy on scales from millimetres to tens of centimetres. Their mutual contact, which is the Wavy Horizon itself, is sharp (over 
1 mm), and broadly conformable with the igneous lamination on the outcrop scale. In the vicinity of Traverses g and h, it is undulatory (hence the epithet Wavy Horizon), with a wavelength of 50–100 cm and amplitude of 30–40 cm (Fig. 2). The troctolite culminations are commonly more pointed than the convex-downwards gabbroic lobes, and occasionally spread laterally to form overhanging heads (Young & Donaldson, 1985). Where the contact between troctolite and gabbro is undulatory it is generally marked by a great abundance of clinopyroxene (up to 60 vol. %) in the basal 10–30 cm of the overlying gabbro. Olivine-rich laminae, the planar foliation defined by plagioclase alignment, occasional anorthosite schlieren, and the grain size and texture of individual layers are commonly unaffected as they pass from troctolite to gabbro across the Wavy Horizon where it cuts across primary igneous lamination. Individual plagioclase-rich layers pass through the contact with little or no deflection, retaining their plagioclase-rich character within the gabbro.
|
|
|
|
|
Where the contact between the trocolite and overlying gabbro is planar (in the vicinity of Traverse e, and where it crops out further south on the eastern flanks of Askival) there is no thick clinopyroxene-rich layer in the lowermost gabbro; instead, the contact itself is marked by a 1 cm thick clinopyroxene-rich (up to 60 vol. %) layer.
The undulatory contact between troctolite and gabbro may be traced, with almost continuous exposure, for 600 m to the west of the peridotite dome. The Unit 8/9 allivalite is poorly exposed further west than this, but scattered blocks containing the Wavy Horizon can be found as far as GR NM38986, 97339. The Wavy Horizon disappears into homogeneous allivalite immediately SE of the dome, but reappears to the south where the next lens of the peridotite sill begins (locality B, Fig. 1b). Near the SE end of the Wavy Horizon (at Traverse e, Fig. 1b), the troctolite changes over 10 m to a distinctive poikilitic facies with layer-parallel clusters of centimetre-scale clinopyroxene oikocrysts [the poikilitic gabbro of Bédard et al. (1988)]. The Wavy Horizon becomes a layer-parallel, planar contact between poikilitic gabbro and a granular gabbro [log 9E of Bédard et al. (1988) and Traverse e, Fig. 3], which is not as pyroxene-rich as that further west. Underlying the poikilitic gabbro is a horizon of granular gabbro [log 9E of Bédard et al. (1988) and Traverse e, Figs 1 and 3], which is clinopyroxene-rich at its lower, undulatory, contact with underlying laminated troctolite (Figs 3 and 4). This lower, subsidiary, Wavy Horizon (marked by a dashed line in the stratigraphic log of Traverse e in Fig. 3) crops out over a strike of a few tens of metres and is truncated by the intrusive peridotite to the NW.
|
The main Wavy Horizon overlies the discontinuous lenses of the Unit 9 peridotite throughout the exposed stratigraphy [e.g. Traverse G of Bédard et al. (1988
, fig. 7), to the east of the Askival Plateau], and is absent where the peridotite is absent. The distinctive poikilitic gabbros described above are found above all the terminations of the discrete peridotite intrusive lenses.
We present the results of a petrographic and geochemical study of samples collected along three traverses across the Unit 8/9 allivalite above the Unit 9 peridotite (Traverses e, g and h shown in Figs 1 and 3), a traverse across the Unit 8/9 allivalite underneath the intrusive body (Traverse i, Figs 1 and 3), and another across the entire Unit 8/9 allivalite far from the intrusive Unit 9 peridotite (Traverse f, Figs 1 and 3). Textural variations in samples from these traverses have previously been described by Holness (2005). Traverses g and h, separated along strike by 50 m, are correlated using olivine-rich laminae (typically 2–5 mm thick), which are truncated by the northwestern side of the peridotite dome, precluding stratigraphic correlation further to the SE.
Rare picrite dykes intrude the Rum Layered Suite, and are believed to be representative of the magma that periodically replenished the Rum chamber (e.g. Upton et al., 2002). The majority contain abundant olivine phenocrysts, and have been described by McClurg (1982) and Upton et al. (2002). As part of this study we present compositions of two small, granular-textured and fine-grained picrite dykes from the Unit 3 allivalite near Loch Coire nan Grund. The locations of the picrite dykes are shown in Fig. 1a.
|
ANALYTICAL TECHNIQUES |
|---|
Mineral compositions were determined using energy-dispersive electron microprobe spectrometry at the Department of Earth Sciences, University of Cambridge, using either a Cameca SX50 electron probe with a Link AN10000 analyser or a Cameca SX100 electron probe microanalyser (details of which analyses pertain to each instrument are presented along with the full analytical dataset in Electronic Appendices 1–4, available for downloading from http://petrology.oxfordjournals.org). Operating conditions were an accelerating potential of 15–20 keV, with a nominal beam current of 15–20 nA and 60 s live counting time. Relative accuracy is
2% for major elements (>5% element present) with 3
detection limits of: Na 0·25%, Mg 0·15%, Al and Si 0·10%, and K–Zn 0·05% (Reed, 1995). In each sample, 3–10 olivine grains, 6–9 plagioclase grains, 3–20 spinel grains, and up to 10 clinopyroxene grains were analysed. Plagioclase grains were analysed at both core and margin to detect chemical zoning. The proportions of Fe2+ and Fe3+ in Cr-spinel were calculated by converting the chemical analyses into cationic proportions (normalized to 24 and neglecting the minor elements Si, Ca, Na, K, P, V and Ni), and assuming that Mn2+ occupies divalent lattice sites, with Ti4+ participating in the substitution 2Fe3+ = Ti4+ + Fe2+. The cationic proportions of Fe2+ and Fe3+ were then calculated using |
|
Mean mineral compositions for each of the five traverses are presented in Tables 1–5
.
|
Whole-rock major element oxide concentrations of picrite dykes were determined by X-ray fluorescence spectrometry at the Department of Geology, Leicester University. These concentrations were used to calculate the CIPW norm of the dykes (Table 6). Modal proportions of olivine, clinopyroxene, plagioclase and spinel in allivalites were determined by point counting (1000–2000 points per thin section).
|
|
PETROGRAPHIC DESCRIPTIONS |
|---|
Peridotite
The Unit 9 peridotite comprises a framework of equidimensional to slightly elongate, subhedral or rounded olivine grains, typically 1–2 mm in diameter, with cumulus, euhedral, Cr-spinel, surrounded by poikilitic plagioclase and clinopyroxene (Fig. 5a). Primary hydrous minerals, such as brown mica and amphibole, occur in rare segregations interstitial to the olivine framework. The range of modal proportions of olivine, plagioclase and clinopyroxene is shown in Fig. 6a.
|
|
Allivalite—far from the Unit 9 peridotite
The Unit 8/9 allivalite far from the intrusive peridotite body (Traverse f) and that underlying the Unit 9 peridotite (Traverse i) are indistinguishable. The allivalite is generally coarse-grained (1–3 mm grain diameter), and contains abundant clinopyroxene and plagioclase, and subordinate cumulus olivine (Fig. 5b, with modal compositional range shown in Figs 3 and 6a). Igneous laminations are generally absent, especially in the upper half of the allivalite body (i.e. the stratigraphic equivalent of the allivalite overlying the Unit 9 perioditite). Plagioclase forms either randomly aligned, low aspect-ratio, elongate laths or sub-equant grains. Some layers, however, especially in the lower parts of the stratigraphy, have a well-developed igneous lamination defined by preferred alignment of high aspect-ratio, elongate, plagioclase laths; these layers tend to be relatively fine-grained. Clinopyroxene is poikilitic in some horizons, forming large, compact grains enclosing a few small, randomly oriented plagioclase laths. In the very rare troctolitic horizons (which are < 1 cm thick) clinopyroxene sometimes forms wedges interstitial to plagioclase or rims surrounding cumulus olivine. Rarely olivine is poikilitic. Chrome spinel is absent.
Allivalite—overlying the Unit 9 peridotite
Troctolite
The lamination in the troctolite of Traverses e, h and g is defined by strong preferred alignment of highly elongate (0·1–0·5 mm long) plagioclase laths (Fig. 5c). Olivine grains are generally subhedral, tabular to equant (and aligned with the plagioclase if non-equant), with a size range 1–3 mm. Diopsidic pyroxene forms optically continuous, interstitial wedges between plagioclase laths. It also forms mono-crystalline rinds separating olivine grains from plagioclase (Holness, 2005; Fig. 5c). Cr-spinel occurs as sparsely distributed euhedral grains (
0·2 mm), commonly either enclosed by, or immediately adjacent to, grains of interstitial pyroxene (Fig. 5d). There is no textural evidence for postcumulus reaction between olivine and Cr-spinel (e.g. Henderson, 1975). Evidence for compaction is given by undulose extinction in the olivine and by kinking and deformation twinning in the plagioclase (Fig. 5e). Modal layering is present in places, with occasional olivine-rich laminae 2–4 mm thick.
Above the southeastern end of the Unit 9 peridotite lens marked A in Fig. 1 (Traverse e), the stratigraphic equivalent to the upper parts of the troctolite is a poikilitic facies containing large (< 2 cm) pyroxene oikocrysts that envelop rounded grains of olivine and randomly oriented plagioclase, in addition to abundant interstitial pyroxene. Below the poikilitic facies the clinopyroxene is more granular. At Traverse e, the rocks below the Wavy Horizon are generally more pyroxene-rich than the stratigraphically equivalent parts of Traverses g and h (Fig. 3).
Gabbro
The Wavy Horizon is overlain by 3–4 m of gabbros comprising clinopyroxene [which may be equant, elongate with a low aspect ratio, or weakly poikilitic (enclosing plagioclase)], together with subordinate amounts of olivine primocrysts (sub-equant to elongate) and plagioclase (low-aspect ratio laths). The grain size of olivine is indistinguishable from that of the underlying troctolite, although the plagioclase grain size is generally larger than in the troctolite. Cr-spinel is absent. A moderate to weak lamination is present, defined by preferred orientation of all non-equant grains (Fig. 5f).
The upper few metres of the Unit 8/9 allivalite, overlying this laminated gabbro, are texturally indistinguishable from the equivalent range at Traverse f (see Fig. 5b), although they do contain localized laminated troctolitic horizons (Fig. 3). These troctolitic horizons notably do not contain Cr-spinel (i.e. similar to the troctolites of Traverse f, but different from the troctolites underlying the Wavy Horizon). The transition between the two types of allivalite, as defined by the loss of the pervasive preferred orientation of elongate grains, occurs some 4–5 m above the Wavy Horizon at Traverse e (the uncertainty is due to insufficient sample spacing, but the change is present in the five topmost samples of the traverse), 3 m above the Wavy Horizon at Traverse g, and 2–4 m above it at Traverse h (in both g and h it occurs below the topmost two samples).
|
TEXTURAL MATURITY IN THE EASTERN LAYERED INTRUSION |
|---|
Holness (2005
) and Holness et al. (2006) showed that the median value of the clinopyroxene–plagioclase–plagioclase dihedral angle, 
cpp , in both gabbros and troctolites from the Eastern Layered Intrusion is a measure of textural maturity that can be used to constrain the sub-solidus cooling history. Rocks that cool slowly are texturally mature and have a dihedral angle population that approaches the expected sub-solidus equilibrium population, with a median of 114°. The median dihedral angle in more rapidly cooled, texturally immature, rocks is much lower, and is essentially unchanged from that inherited by interstitial clinopyroxene.
The extent of textural maturity is constant throughout the Unit 8/9 allivalite at Traverse f and in the lower parts of Traverse i, with median angles in the range 88–93°, apart from a basal excursion to lower values caused by late-stage infiltration of liquids from the underlying Unit 8 peridotite (Holness, 2005, 2006; Holness et al., 2006; and Tables 2 and 5). The degree of textural maturity is consistent with plagioclase, olivine and clinopyroxene as liquidus phases (supported by textural evidence for all three forming cumulus grains), and a constant cooling and solidification rate (Holness, 2005; Holness et al., 2006). The allivalites immediately below the Unit 9 peridotite have higher angles than the range 88–93°, and this localized increase is attributed to the localized thermal effects of the picrite intrusion (Holness, 2005).
The median value of cpp departs significantly from 90° in the Unit 8/9 allivalites overlying the peridotite [data from Holness (2005) reproduced in Fig. 7a and Tables 1, 3 and 4]. The median clinopyroxene–plagioclase–plagioclase angle varies from a minimum of 74° a few metres above the peridotite to a maximum of 110° a few metres above the Wavy Horizon. The textural maturity reverts to the expected value of 90° in the upper reaches of the allivalite, approximately corresponding to the textural transition to coarse non-laminated gabbro, apart from a single sample (B0023) from the actual contact with the Unit 10 peridotite, which has an angle of 110°. The median value of cpp in the gabbros shows a positive correlation with the modal percent of clinopyroxene (Fig. 7b).
|
The median value of
cpp in the poikilitic gabbros underlying the Wavy Horizon at Traverse e varies according to whether it is measured in the central part of the oikocrysts, or either on the oikocryst margins or at the junction of grains interstitial to plagioclase. The oikocryst margins and interstitial wedges have median values of 70–96°, corresponding to the values observed in the stratigraphically equivalent troctolites, whereas median values from the central regions of the clinopyroxene oikocrysts only a few millimetres away are in the range 103–106°. This pattern is also observed in poikilitic allivalites from Unit 10 [the type unit for the ELS (Brown, 1956)] in which cpp internal to oikocrysts is 101° compared with 86° for that outside (M. B. Holness, unpublished data), and it can be attributed to the coupled effects of commensurate rates of textural maturation and pyroxene growth and a closure temperature for textural maturation that is close to the solidus (Holness et al., 2005, 2006). For the purposes of comparison with the troctolites we use the value of cpp measured from the interstitial wedges. |
UNIT 8/9 MINERAL COMPOSITIONS |
|---|
Olivine
The variation in forsterite content of the olivine in the Traverse f allivalites is shown in Fig. 8a, the mean compositions are reported in Table 2, and the full dataset is given in Electronic Appendix 1. No olivine zoning was detected in any of the samples examined as part of this study. Within each sample the olivine lies in the range Fo75–80, but varies by up to 2 mol % Fo. There is a pronounced decrease from Fo82 to Fo75 in the basal few metres. The Ni content within each sample varies by up to 1000 ppm, with most analyses in the range 620–1300 ppm (Fig. 8b). The basal few metres display a steep decrease from more Ni-rich compositions.
|
The Fo content of olivine in Traverses e, g, h and i (mean compositions reported in Tables 1, 3, 4 and 5, with the full dataset given in Electronic Appendix 1) is significantly higher than that in the stratigraphically equivalent parts of Traverse f (Fig. 9a). The Fo content of Traverse e olivine is generally slightly lower than that in Traverses g and h (Fig. 9a), and all three traverses show a slightly higher Fo in the lower part (below the Wavy Horizon) compared with the upper. All traverses show a prominent increase of Fo within the topmost few metres. This increase occurs over distances < 1 m for Traverses e, g and h, but over
5 m for i.
|
The olivine in most samples from Traverses g and h has higher Ni than olivines from stratigraphically equivalent parts of Traverse f (Fig. 9b). That in Traverse e is intermediate between olivine from Traverses f, g and h. Ni contents of olivine in Traverse i are indistinguishable from those in Traverse f, except immediately below the Unit 9 peridotite, where they increase (Fig. 9b). The olivine in the troctolites below the Wavy Horizon is slightly more Ni-rich than that in the overlying gabbros.
Plagioclase
A preliminary investigation of plagioclase from Traverses f and g was undertaken to constrain the differences between allivalite far from the intrusive peridotite body and that overlying the peridotite. Almost all grains analysed have a compositional difference between the grain core and margin; the data are shown in their entirety (with no distinction made between central and marginal analyses) in Fig. 10a. Average analyses are given in Tables 2 and 3, and the complete dataset (with the position of each analysis point relative to the centre of the grain) is given in Electronic Appendix 2.
|
The plagioclase composition in Traverse f shows a decrease at the base from An75–80 to An60–75 and then a slow increase back to values of An75–80, in an approximation to the pattern observed in the olivine composition (Fig. 8). In contrast, the plagioclase in Traverse g shows no variation with stratigraphic height, although compositions are more An-rich than the plagioclase of Traverse f.
The difference in composition between the cores and the rims of individual grains is shown in Fig. 10b. In general, the plagioclase grains of Traverse f show a large compositional difference between core and margin (consistent with the optically visible differences in thin section), with a predominance of normal zoning. The plagioclase grains of Traverse g have a smaller composition range, with a greater proportion showing reverse zoning. The sample from the top of the traverse, i.e. that with textures indistinguishable from those seen in Traverse f, is dominated by strongly normally zoned grains.
Cr-spinel
The compositions of Cr-spinel from the Unit 9 peridotite and overlying troctolite (none is present in the Unit 8/9 gabbros) are plotted in Fig. 11, which shows the relative proportions of the trivalent cations in the formula [Mg2+,Fe2+]8[Al3+,Fe3+,Cr3+]16O32 (for the primary data see Electronic Appendix 4). The data are superimposed on the compositional trends in Rum spinels (Henderson, 1975). The common origin of the two trends represents the primary composition of Cr-spinel, with the Al trend caused by reaction with olivine and plagioclase, and the Fe trend formed by interaction of spinel with evolved interstitial melt (Henderson, 1975). The Cr-spinel compositions from the Unit 9 peridotite and the overlying troctolite show evidence for only limited post-cumulus reaction with interstitial liquids (Fig. 11).
|
Clinopyroxene
Clinopyroxene in the Unit 8/9 allivalites overlying the Unit 9 peridotite can be divided on textural grounds into four distinct populations (Fig. 12): the interstitial pyroxenes in the troctolites; the pyroxene primocrysts in the overlying gabbros; the pyroxene oikocrysts in the poikilitic gabbro underlying the Wavy Horizon at Traverse e; and their associated interstitial pyroxene (for the full dataset see Electronic Appendix 3). The troctolite and gabbro of Traverses g and h contain pyroxenes with similar Mg-numbers [Mg-number = MgO/(MgO + FeOtot)] in the range 84–88 (Tables 3 and 4), but the troctolite pyroxenes have generally higher TiO2 and slightly lower Al2O3 compared with the gabbros (Fig. 12a and b). Pyroxene in the Unit 9 peridotite immediately adjacent to the overlying troctolite has a similar Mg-number, generally higher Al2O3 and lower TiO2 than that in the troctolite. Pyroxenes from Traverse e generally have lower Mg-number (83–86) compared with those of Traverses g and h, but the interstitial pyroxene has higher TiO2 and lower Al2O3 compared with the central regions of the oikocrysts (Fig. 12a and b).
|
For comparison, pyroxene compositions from sample 6074, a poikilitic gabbro from Unit 10 (collected by G. M. Brown and housed in the collection at the Earth Sciences Department at Oxford), are also shown, subdivided into oikocryst and interstitial grains (Fig. 12c). These two textural types form a continuous trend on a plot of TiO2 vs Mg-number. The arrow in Fig. 12c shows the progressive changes in clinopyroxene composition for successively more evolved liquids derived by progressive solidification of a plausible parental magma. This was calculated assuming that the parental liquid entering the Rum magma chamber was the same as that of the groundmass of the picrite dyke M9 (McClurg, 1982
; Table 6), using the program MELTS (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998) with fO2 set at QFM [quartz–fayalite–magnetite buffer; following Upton et al. (2002)] and a pressure of 200 bars [calculated using the pressure estimate of 150 bars for contact metamorphism in Kinloch Glen (Holness, 1999), a plausible palaeosurface for the volcano, and assuming a basaltic overburden]. The compositions of the pyroxenes in sample 6074 are consistent with the growth of interstitial grains from progressively more evolved liquids. Variation in clinopyroxene TiO2 content with stratigraphic height in Traverses g and h is shown in Fig. 13a, with distinction made between troctolitic and gabbroic allivalites, and distances normalized to the position of the Wavy Horizon (to allow for the undulose nature of this contact). The troctolite clinopyroxene is generally more TiO2-rich than that in the gabbros, although pyroxenes in both rock types show a decrease in TiO2 with height (this decrease for the gabbros occurs only in the lower part of the section). The range of pyroxene TiO2 content below and at the Wavy Horizon is much greater than that in the gabbros higher in the traverse. Both the uppermost Unit 9 peridotite and the lowermost Unit 10 peridotite contain pyroxene with TiO2 similar to that of the Unit 8/9 gabbros (Fig. 13a). A detailed examination of the variation in pyroxene composition across the Wavy Horizon itself is shown in Fig. 13b. The pyroxene changes composition over a distance of a few millimetres. This is particularly evident for TiO2 and Cr2O3. The Mg-number remains unchanged across the boundary.
|
|
THE UNIT 9 PERIDOTITE AS AN INTRUSIVE BODY |
|---|
The field evidence for the intrusive nature of the Unit 9 peridotite (e.g. Butcher et al., 1985
; Bédard et al., 1988) is supported by consideration of its modal mineralogy (Fig. 6). The synthetic, pseudo-ternary system forsterite–anorthite–diopside is shown in Fig. 6a (after Osborn & Tait, 1952), and is a good analogue for basaltic systems (Morse, 1994). The three minerals forsterite (Fo), anorthite (An) and diopside (Di) each have a primary field of crystallization separated by univariant boundary curves that converge at what approximates to a ternary eutectic. The spinel (Sp) field is non-ternary as spinel does not lie in the plane of the diagram. Spinel is thus a transient, high-temperature phase, which dissolves during liquid evolution to the eutectic. In the simplified diagram (Fig. 6b), spinel is omitted for clarity. Perfect fractional crystallization of a hypothetical picritic parent liquid of initial composition p forms a dunitic cumulate, followed by a troctolite (Ol + Plag) and then a gabbro (Ol + Plag + Cpx) (Fig. 6b). Liquid trapping in such a crystal accumulation results in bulk-rock compositions that are transitional between idealized cumulates.
The bulk compositions of the picrite dykes from Table 6 are also shown in Fig. 6a, and fall into two groups. The aphyric dykes (and the groundmass of the olivine-phyric dyke M9) lie within the olivine field, close to the pseudo-ternary eutectic. The bulk composition of M9, together with the bulk composition of the olivine-phyric picritic dyke R124 (Hallworth, 1998), is indistinguishable from that of the Unit 9 peridotite. In contrast, the range of bulk compositions of the Unit 10 peridotite (from Brown, 1956; Tait, 1984, 1985), which is clearly a cumulate body, is more olivine-rich than both the olivine-rich picrite dykes and the Unit 9 peridotite (Fig. 6a). We interpret this as support for an origin of the Unit 9 peridotite as an intrusion of olivine-phyric liquid with a bulk composition similar to that of M9 and R124, rather than by accumulation of olivine crystals during fractional crystallization.
M9 contains 58 vol. % olivine phenocrysts, comprising several populations of olivine, suggesting that the dyke intruded with a high crystal load [at the viscous limit of Marsh (1981)]. The composition of the intruding liquid (i.e. minus the crystal load) that formed the Unit 9 peridotite was likely to have been in the range given by the modal compositions of aphyric picritic dykes on Rum (including that of the groundmass of M9) shown in Fig. 6a (and Table 6). The subsequent discussion is based on the assumption that the liquid composition was similar to that of the M9 groundmass, which, with the addition of 50–60 wt % olivine as phenocrysts (Fig. 6), would give the observed modal composition of the Unit 9 peridotite.
|
THE WAVY HORIZON AS A METASOMATIC FEATURE |
|---|
The close spatial association of the Wavy Horizon with the underlying discontinuous, intrusive, Unit 9 peridotite throughout the Eastern Layered Intrusion on Rum provides strong evidence for a direct genetic relationship between the intrusion of a picrite body as a sill into a pre-existing allivalite and the development of the Wavy Horizon in the overlying material. Although the Unit 10 allivalite contains some troctolitic horizons (Fig. 6a), troctolites are rare in the Unit 8/9 allivalites far from the Unit 9 peridotite but occur immediately overlying the Unit 9 peridotite (Figs 3 and 6). Furthermore, the gabbros that overlie these troctolites are texturally distinct from those of Traverse f, although their average modal pyroxene content is no different from that of Traverse f (Fig. 3).
Another major difference between the allivalites overlying the Unit 9 peridotite and those elsewhere in the Eastern Layered Intrusion is in their extent of textural maturity (Fig. 7). The high dihedral angles above the Wavy Horizon are inconsistent with simple crystal accumulation and steady growth of the crystal pile on the chamber floor, even for a three-phase liquidus assemblage [for which the median cpp is 90° (Holness, 2005; Holness et al., 2006)]. The low angles in the troctolites below the Wavy Horizon are lower than would be expected for a cumulate in which only olivine and plagioclase were cumulus phases [for which the expected range is 80–84° (Holness, 2005, 2006; Holness et al., 2006)]. The Unit 8/9 allivalite above the Unit 9 peridotite has therefore experienced a sub-solidus thermal history distinct from that expected for a cumulate in the Rum chamber.
The disruption to the expected values of textural maturity and the modification of the gabbro texture relative to that of the unaffected allivalites of Traverse f suggest that the effects of the intrusive peridotite extend 3–5 m above the Wavy Horizon, or up to 8 m from the top of the peridotite (Traverse e shows effects extending some 14 m from the top of the peridotite, but this may not be a true distance, as the sub-surface geometry of the peridotite body is uncertain in this locality).
A complete model of the intrusive peridotite must account for the following observations: (1) the low abundance of clinopyroxene in the allivalite (troctolite) immediately overlying the peridotite; (2) the development of igneous lamination in the allivalite overlying the peridotite; (3) the pyroxene-rich zone at the boundary between troctolite and gabbro (i.e. the Wavy Horizon); (4) the absence of significant fabric distortion across the undulating contact; (5) the undulation of the Wavy Horizon; (6) the mineral chemistry; (7) the sub-solidus thermal history recorded by variation of dihedral angles, which reflects the degree of textural maturity.
|
MODEL: FOUNDERING GABBROIC MUSH INTO A PICRITE SILL |
|---|
The crystal mush above a lateral intrusion of liquid is likely to have some internal strength and behave as a coherent body (e.g. Irvine, 1980
; Shirley, 1986; Philpotts et al., 1999). If injection were rapid, the magmastatic pressure exerted by the picrite would result in an initial upwards displacement of this coherent body of gabbroic mush (Fig. 14a and b). What subsequently occurs, on a purely mechanical level, depends on the relative densities of the coherent mush and the picritic liquid in the sill.
|
|
The density of the intruding picrite can be constrained by assuming that the parental liquid was similar to the quenched groundmass of the picrite dyke M9 (McClurg, 1982
; Table 6) and using the program MELTS (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998), assuming a pressure of 200 bars and fO2 set initially at QFM (Upton et al., 2002). This liquid, which is saturated with 0·8 wt % H2O, has a density of 2·66 g/cm3 at its liquidus temperature of 1450°C. The liquid density decreases to 2·61 g/cm3 at 1160°C, the point at which plagioclase becomes a liquidus phase (again calculated using the program MELTS with the P and fO2 given above).
The minimum bulk density of the coherent crystal mush can be approximated assuming the solid fraction consists of 60 vol. % plagioclase, 10 vol. % olivine and 30 vol. % clinopyroxene (i.e. average concentrations from Traverse f gabbros), with a liquid fraction not exceeding 60 vol. % [which is the maximum permitting a self-supporting framework (Irvine, 1980; Shirley, 1986; Turbeville, 1993)]. Assuming densities of 3·45 g/cm3 for olivine, 3·22 g/cm3 for clinopyroxene and 2·73 g/cm3 for plagioclase (Deer et al., 1966), and a density of 2·62 g/cm3 for the interstitial liquid [calculated assuming equilibrium crystallization of liquid M9 (with 0·8 wt % H2O) to form a gabbroic mush in equilibrium with olivine of Fo83], this results in a minimum bulk density for the crystal raft of 2·78 g/cm3. If the gabbroic mush contained less than 60 vol. % liquid then the bulk density approaches the maximum density of 2·91 g/cm3 for a fully solidified gabbro. The coherent crystal mush was therefore denser than the intruding picrite and would have sunk into it after emplacement, replacing the resident liquid in the mush with the reactive picrite as it did so (Fig. 14c).
As the invading picrite formed a series of discrete bodies, rather than a continuous, chamber-wide intrusion (see the discrete bodies labelled A and B in Fig. 1), the geometry of the sinking gabbroic mush has two possible end-members. The first is when the crystal raft becomes locally detached from the surrounding, non-invaded, horizons and founders as a discrete block. The second is when the crystal raft locally sags into the sill, retaining mechanical coherence along strike.
If the raft sinks purely under its own weight as a mechanically discrete block, the rate of descent is controlled both by the difference in density between the bulk mush and the picrite and by the permeability of the mush. Assuming that the raft is coherent then, according to Darcy's law, the speed of the interstitial liquid (
) relative to the matrix has the form
|
|
is the porosity, 
m is the bulk matrix density, p is the picrite density, g is the acceleration due to gravity, µ is the liquid viscosity, and k is the permeability of the matrix. By volume conservation, the interstitial velocity of the melt is related to the descent speed of the matrix, U, according to the relation |
|
2d2/1600, where d is the grain size of 1 mm, von Bargen & Waff, 1986), this implies a time-scale of 0·5–5 years for the picrite to invade of the order of 1 m into the gabbro. Elliott et al. (1997) calculated the permeability of a mush with a pore structure controlled entirely by the juxtaposition of planar-sided grains (an ‘impingement texture’) to be as high as 10–9 m2 for a grain size of 1 mm. This results in a time-scale of weeks to months for the picrite to invade 1 m into the mush. For a geometry in which the gabbroic mush retains its integrity on a chamber-wide scale, the rate of descent into the sill will be controlled by the viscosity of the crystal raft, with no descent possible at the edges of the sill and the maximum rate of descent above the centre of the sill. For infinitely wide sills, the rate of descent of the crystal raft overlying the central parts of the sill will approach the rates given above, which are thus maximum estimates.
The crystal raft did not founder completely into the picrite sill—there remains 25 m of peridotite below it—because it came to rest on the compressed accumulated olivine crystal load (Fig. 14c and d). If the incoming picrite were similar to M9, it would have been carrying 50 vol. % olivine phenocrysts. The heat loss consequent to intrusion would promote further olivine crystallization. From the position of the olivine–plagioclase cotectic (Osborn & Tait, 1952; Fig. 6a) it is probable that up to 60 vol. % olivine could have crystallized from the picrite before plagioclase arrived on the liquidus. It is likely that this olivine accumulated on the floor of the sill, with a packing density similar to that of the Unit 10 peridotite; that is, 30% porosity (Fig. 6a, using the Lever Rule and averaging the bulk composition of the Unit 10 samples). The present-day 25 m thickness of the peridotite thus suggests an original thickness of 30 m for the sill. Assuming 50% porosity in the crystal raft, the picrite could have displaced the resident liquid in the lower 10 m of the Unit 8/9 allivalite, consistent with the scale of the observed textural and compositional changes in the vicinity of Traverses g and h.
The picrite liquid would have been hotter than the gabbro into which it intruded. A magma with a composition similar to the quenched groundmass of the picrite dyke M9 has a liquidus temperature of 1450°C if saturated with 0·8 wt % H2O [again calculated using the program MELTS (Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998), assuming a pressure of 200 bars and f O2 set initially at QFM (Upton et al., 2002)]. We have few constraints on the thermal structure of the gabbro—this would depend on the amount of adcumulus growth and the extent to which the interstitial liquid was able to exchange with the bulk magma of the chamber.
The thermal time constant of a 25 m thick sill (defined as a2/
, where a is the sill thickness, and is the thermal diffusivity, taken to be 10–6 m2/s) is of the order of a few years. Given a length-scale of infiltration (i.e. the distance over which textural and compositional modification has occurred) of order 10 m and infiltration rates as above gives a minimum infiltration time-scale of 0·5–50 years. The lower bound is similar to the thermal time constant of the sill and suggests that the picrite sill would attain thermal equilibrium early in the development of the Wavy Horizon. Given the difficulties of constructing a realistic thermal model, and the simplified treatment of constant porosity above, we present a greatly simplified model, which assumes that the dominant process forming the Wavy Horizon was chemical re-equilibration of infiltrating liquid and the solid phases, which were essentially in thermal equilibrium.
|
THE CHEMICAL EFFECTS OF DISPLACING THE RESIDENT LIQUID IN THE MUSH |
|---|
Modal composition
In the early stages the picrite intrusion (given by the composition P1 in Fig. 15a) would have been saturated only in olivine. Rapid cooling in the body of the sill (mainly by conduction, as it is not likely that such a crystal-rich body would have been vigorously convecting), and in the liquid displacing the original resident liquid, would result in olivine crystallization and the eventual arrival of plagioclase on the liquidus (although the absence of plagioclase primocrysts in the peridotite precludes this happening in the bulk of the sill). For illustrative purposes, the bulk composition of a typical sample of the original allivalite is shown by B1 in Fig. 15a. This comprises a eutectic liquid, L1, in equilibrium with solids of bulk composition S1, with an arbitrarily chosen liquid fraction of
40 wt %. The porosity is likely to have varied on a local scale from close to 0 wt % to 60 wt %. Similar schematic representations can be used to account for about half the sampled original allivalite shown by the Traverse f samples in Fig. 6; the other half has a bulk composition lying within the plagioclase primary phase field (Fig. 6) and will be discussed later. The replacement of the eutectic liquid, L1 , by the picrite liquid, P1 , results in a new bulk composition, B2. The picrite displacing the resident liquid in the sinking mush would become co-saturated in olivine and plagioclase, and its composition would move to the olivine–plagioclase cotectic (liquid P2 , Fig. 15a). This would be accompanied by a change in the solid composition to that of the troctolite S2 as a result of the removal of clinopyroxene (Figs 14d and 15a). In the example shown in Fig. 15a, the reaction of clinopyroxene is accompanied by the crystallization of olivine from the displacing interstitial liquid. However, for the allivalite example given Fig. 15a, there is a net increase in porosity in the mush consequent to infiltration (which will be followed by compaction), whereas for that given in Fig. 15b (for which the amount of liquid in the original mush was 50 wt %), there is a net decrease in the porosity of the mush (as a result of the crystallization of olivine in excess of the dissolution of clinopyroxene). On cooling, further evolution of the system results in the cotectic crystallization of olivine and plagioclase from the displacing interstitial liquid in the allivalite (and the rotation of the solid–liquid tie-line about the bulk composition, B2), until the liquid reaches the eutectic, where it becomes saturated with clinopyroxene. This point is denoted by the dashed line in Fig. 15a, at which point the solid bulk composition is at S3.
For the case in which the original bulk composition, B1, lay in the plagioclase primary phase field, the progressive infiltration of picrite would result in reaction of clinopyroxene, together with the crystallization of olivine from the interstitial liquid, if the modified bulk composition, B2, lies in the olivine primary phase field (Fig. 15c). This would happen for horizons with both a solid composition, S1 , close to the olivine–plagioclase cotectic, and a high volume fraction of liquid. For horizons with a low volume fraction of liquid and a solid composition lying far from the olivine–plagioclase cotectic, replacement of the interstitial liquid by picrite will not displace B2 from the plagioclase primary phase field (Fig. 15d). In this case, crystallization of plagioclase and olivine from interstitial liquid subsequent to the complete reaction of clinopyroxene crystals results in movement of the solid composition, S2 , to S3 , towards the olivine apex (Fig. 15d). Given the stratigraphic variation of bulk compositions of the unmodified allivalite of Traverse f, we would expect these different cases to develop at different levels of the infiltrated mush.
As a consequence of the above, a compositional, or reaction, front would develop, with a troctolitic mush below it (containing liquid saturated only in olivine and plagioclase), and unaltered gabbro above it (with a stratified interstitial liquid, with the displaced original liquid overlying the modified infiltrating liquid). Above the reaction front, the invading liquid would be in equilibrium with the solid fraction, and would essentially have arrived at the original composition of the displaced original resident liquid (L1 in Fig. 15). The infiltration front (Fig. 14d) is at a higher level than the reaction front, and marks the junction between the original, displaced liquid and the invading (but now modified) liquid. Because the displacing liquid is in, or close to, chemical equilibrium with the solid phases at this level, the infiltration front is unlikely to have a strong chemical signature.
Once the crystal raft had come to rest, no more infiltration would occur and the interstitial liquid would solidify. Crystallization below the reaction front would result in a troctolite containing small amounts of interstitial clinopyroxene (formed by the evolution of the interstitial liquid towards the pseudo-eutectic), whereas that above the reaction front would result in a bulk composition little different from that of the unmodified gabbros of Traverse f.
According to this simple model, the contact between troctolite and gabbro should be marked by a return of the modal composition to that for the unaffected allivalite of Traverse f. Instead, we see a concentration of clinopyroxene in the basal few decimetres of the gabbro (and the basal centimetres of the minor Wavy Horizon at Traverse e). This, together with the geometry of the contact, which led Young & Donaldson (1985) and Volker & Upton (1990) to suggest that the Wavy Horizon represents load structures, could be explained by the mixing of two liquids each co-saturated in olivine, plagioclase and clinopyroxene, but with differing degrees of chemical evolution.
The pseudo-eutectic in the simplified Ol–Plag–Cpx phase diagram can be more clearly understood in a projection from plagioclase, in which it becomes a plagioclase–clinopyroxene (Ca-rich pyroxene)–olivine cotectic ending in the eutectic at which the liquid is in equilibrium with olivine, Ca-rich pyroxene and Ca-poor pyroxene (Fig. 16). Walker et al. (1979) pointed out that the curvature of this cotectic means that if two liquids, each at different points on the olivine–plagioclase–clinopyroxene cotectic, are mixed, the composition of the mixed liquid can enter the primary phase field of clinopyroxene (plus plagioclase) in which it is supercooled (Fig. 16). We suggest that something similar to this might have occurred once the crystal raft had come to rest. If the liquid immediately below the reaction front (which had only just arrived at the olivine–plagioclase–clinopyroxene cotectic) mixes with liquid that had been resident in the gabbroic part of the mush above the reaction front for some time (and had thus evolved some way down the olivine–plagioclase–clinopyroxene cotectic), then the mixture would be momentarily in the clinopyroxene (+ plagioclase) phase field (Fig. 16). This would promote crystallization of clinopyroxene (and dissolution of olivine) and would form the pyroxene-rich band that is such a remarkable feature of the Wavy Horizon (Fig. 14e).
|
The load-cast-like structures on the Wavy Horizon, with convex-downwards lobes of gabbro [Fig. 2, and described in detail by Young & Donaldson (1985
)] are present only where the gabbro immediately overlying the Wavy Horizon is enriched in clinopyroxene. Where the Wavy Horizon is planar, and layer-parallel, clinopyroxene enrichment occurs only within a centimetre of the contact between the gabbro and troctolite. This pattern is consistent with clinopyroxene enrichment being due to mixing of interstitial liquids: where the front between the two liquids was planar, no mixing occurred and hence there was no excess clinopyroxene crystallization.
Mineral compositions
A comparison of the olivine and plagioclase compositions in the allivalite immediately under- and overlying the Unit 9 peridotite with those far from it (Fig. 9) suggests either that the picrite intrusion had a far-reaching and significant effect on mineral composition or that there is significant original lateral compositional variation within the Unit 8/9 allivalite. We believe the latter to be correct.
If the allivalite below the sill were not completely solidified, it is likely that the resident eutectic liquid would have been displaced by the denser picrite (see Tait, 1985). This can account for the increased Fo and Ni contents in olivine observed at the top of Traverse i (Fig. 9), and also that at the top of Traverse f (caused by the influx of the Unit 10 picrite). This suggests that the elevated Fo in the lower parts of Traverse i is not related to picrite invasion. Comparison with Traverses e, g and h (which would have directly overlain Traverse i before the picrite intruded) suggests that the generally elevated Fo in the upper parts of the Unit 8/9 allivalite is similarly unconnected to the picrite [apart from the elevated Fo at the very top of Unit 8/9, which can be directly correlated with downwards infiltration of the Unit 10 picrite (Fig. 9; see Tait, 1985)].
The difference between the Ni content of olivine in Traverses e, g, h and i and that of f is minor, although there is a general pattern of increased Ni relative to Traverse f. There is an increase in Ni in the topmost few metres of i, consistent with downwards percolation of picrite. There is also a marked increase in the range of olivine Ni contents in the troctolites below the Wavy Horizon compared with those in the overlying gabbros (mirrored in the range of TiO2 in the pyroxenes). The pattern of more evolved primary compositions of the Unit 8/9 allivalite towards the SE is mirrored in differences between Traverse e and Traverses g and h, with slightly lower Fo and Ni in e (Fig. 9), generally more evolved clinopyroxene compositions (i.e. higher TiO2 and lower Al2O3, Fig. 12), and more evolved plagioclase compositions (Fig. 10a).
These significant lateral variations in the original compositions of the cumulate phases preclude direct comparison of Wavy Horizon rocks with unmetasomatized stratigraphic equivalents. However, we suggest that the small increase in Fo content of the olivines in the troctolites relative to the gabbros is consistent with re-equilibration with a more primitive liquid. Furthermore, the pattern of zoning in the plagioclase suggests some metasomatic alteration of an original population, which, by comparison with Traverse f, was likely to have been dominated by significant normal zoning (consistent with crystallization from evolved interstitial liquids in the mush) (Fig. 10b). In contrast, the plagioclase population of Traverse g contains many more grains with reverse zoning, albeit with only small compositional differences between cores and margins (Fig. 10b). The topmost samples from this traverse (those that have similar textures to Traverse f and lie above the putative infiltration front) contain many grains with significant normal zonation (Fig. 10b). We suggest that this difference in zoning patterns reflects recrystallization consequent to fluid infiltration.
In contrast to the olivine, the clinopyroxene in the troctolites of Traverses g and h is more evolved (with a higher TiO2 content) than that in the immediately overlying gabbro, consistent with it crystallizing from small amounts of trapped liquid. Whereas the pyroxene composition in the gabbros far above the Wavy Horizon (i.e. above the infiltration front; Fig. 14d) is likely to reflect the original cumulus composition (albeit with modification as a result of interaction with trapped interstitial liquids), consistent with the narrower range of compositions (Fig. 13a), that immediately above the Wavy Horizon should reflect the composition crystallizing during reaction consequent to mixing of the two interstitial liquids as proposed above.
As discussed above, the lower Mg-number of the clinopyroxene in the poikilitic gabbros of Traverse e is probably inherited from a more evolved precursor gabbroic mush. However, the poikilitic pyroxene has similar TiO2 and Al2O3 to that in the gabbros in Traverses g and h, whereas the interstitial pyroxene is similar to that in the troctolites. The larger cpp in the oikocrysts compared with the interstitial pyroxene, the relatively primitive oikocryst compositions compared with the interstitial pyroxene, and the randomly oriented included plagioclase grains tell us that the oikocrysts grew early, before compaction. The generally higher modal pyroxene content of the troctolite below the Wavy Horizon at Traverse e suggests that compaction of the metasomatized mush was not as effective as that of the mush in Traverses g and h. The nucleation and growth of oikocrysts before compaction is consistent with this. Field observations of poikilitic gabbros in the regions where the Wavy Horizon disappears above the ends of the discontinuous peridotite bodies suggest that these higher modal pyroxene contents and oikocrystic growth are general features of the ends of the picrite sills.
|
THE TEXTURAL EFFECTS OF DISPLACING THE RESIDENT LIQUID IN THE MUSH |
|---|
Lamination and recrystallization
The basal 10 m of the Unit 8/9 allivalite overlying the Unit 9 peridotite is recrystallized relative to the inferred original textures observed in Traverse f. The recrystallization resulted in grain-size reduction (notably of plagioclase) and the formation of lamination, particularly in the troctolite. We suggest that the lamination in the troctolite, together with the internal deformation of both plagioclase and olivine, and the almost complete absence of trapped liquid, was the result of significant compaction consequent to clinopyroxene reaction. This suggests that the bulk composition of the original mush was generally such as to promote an increase in porosity consequent to infiltration metasomatism. However, the reduction in plagioclase grain size is consistent with at least some crystallization of new plagioclase from the invading picrite, and the difference in zoning patterns of the plagioclase at Traverse g compared with Traverse f is consistent with this.
The weaker lamination in the gabbros overlying the Wavy Horizon suggests recrystallization between the reaction and infiltration fronts, and the reason for this is not clear. It is possible that the invading liquid was not in complete chemical equilibrium with the solid phases, even when saturated in clinopyroxene, thus promoting some dissolution and recrystallization.
Textural maturity
The sigmoidal shape of the dihedral angle data (Fig. 7) suggests that several factors controlled the sub-solidus history of the allivalite. The greater textural maturity of the gabbros compared with normal cumulates indicates that they cooled more slowly while they were above the closure temperature for sub-solidus maturation. This closure temperature is close to the solidus, meaning that changes in latent heat of crystallization can have a significant effect on textural maturity (Holness et al., 2006). The strong correlation between textural maturity and the amount of clinopyroxene in the gabbros (Fig. 7) is consistent with the increased textural maturity in the gabbros above the Wavy Horizon being a consequence of increased pyroxene growth after mixing of the two liquids, which resulted in a localized slower cooling rate in the immediate sub-solidus as a result of the release of the latent heat of crystallization. The absence of significantly enhanced textural maturity at Traverse e, where there is no major pyroxene enrichment at the Wavy Horizon, supports this suggestion.
The troctolites below the Wavy Horizon are texturally immature compared with equivalent troctolitic cumulates elsewhere in the Eastern Layered Intrusion (Holness, 2005; Holness et al., 2006), but have a similar maturity compared with gabbros affected by late-stage injection of cool evolved liquids [e.g. the base of Traverse f (Holness, 2005, 2006); the base of Unit 10 (Holness, 2006)]. We suggest that the low values of cpp in the troctolites reflect the upwards movement of interstitial liquid derived from the picrite sill during the last stages of evolution of the Wavy Horizon (consistent with the relatively evolved composition of the interstitial clinopyroxene). The sigmoidal shape of the median dihedral angle data thus reflects the localized effects of latent heat release in the gabbros and the effects of relatively late-stage liquids expelled from the picrite into the troctolite (inset in Fig. 7a). The slightly higher angles at the base of the troctolite could plausibly be the effects of heating from the picrite sill.
|
DISCUSSION AND CONCLUSIONS |
|---|
We have argued that the wholesale sinking of a coherent, evolved, gabbroic crystal raft into a layer-parallel intrusion of picrite resulted in the complete replacement of the interstitial liquid. The juxtaposition of cumulus minerals and reactive melt resulted in significant post-cumulus re-equilibration, notably the complete removal of clinopyroxene from the lower parts of the mush below a reaction front. Subsequent to the wholesale sinking of the raft, the multiply saturated liquid above the reaction front mixed with the liquid below the reaction front. The mixing of the two liquids formed a new liquid over-saturated in clinopyroxene, resulting in the generation of a zone of highly pyroxene-enriched gabbro at the contact. Such a model can account for the Wavy Horizon overprinting the igneous lamination, the concentration of pyroxene in the lowermost gabbros (which is evident where the Wavy Horizon is wavy, and absent when the Wavy Horizon is planar), and the closely related elevated textural maturity.
The Wavy Horizon is not so clearly defined at the ends of the picrite sills. The underlying allivalite has not been stripped of its pyroxene and the pyroxene-rich zone in the gabbros is not well developed. Additionally, there is a lower Wavy Horizon developed at Traverse e, which divides granular gabbros with eutectic bulk composition (Fig. 3) from the underlying troctolite. These effects probably reflect the waning influence of the infiltrating picrite, with onset of pyroxene crystallization before significant compaction, and little compositional contrast between the two liquids, precluding significant pyroxene crystallization in the lower gabbros. The lower Wavy Horizon itself may represent a solidified porosity wave, generated during compaction, or a later wave or cycle of infiltration, as suggested by McBirney (1987).
The model described here may have implications for a more extended application on Rum, given the field evidence for intrusion of replenishing picritic melt as sills and dykes within a partially solidified cumulate pile in the Rum chamber in at least one other unit (e.g. Renner & Palacz, 1987). Throughout the Eastern Layered Intrusion there are a number of wavy horizons separating underlying pyroxene-poor from overlying pyroxene-rich layers, but without an associated peridotite body (notably one in the Unit 8 allivalite near Askival). It is possible that these features formed from a directly analogous process to that described here but where the crystal raft sank to the base of the intruding (?aphyric) picrite.
|
SUPPLEMENTARY DATA |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
This research was funded in part by the European Community's Human Potential Programme under contract HPRN-CT-2002-000211 [EUROMELT]. Scottish Natural Heritage granted permission to conduct fieldwork on the Isle of Rum (project RG 458). M.A.H.'s fieldwork was funded by Anglia Polytechnic University. R.E.S. acknowledges receipt of an NERC studentship. Monica Price at the Oxford Museum of Natural History was most helpful in providing access to the collections. Reviews by Steve Sparks, Mike Cheadle, Henry Emeleus and Marjorie Wilson greatly improved earlier versions of the manuscript.
*Corresponding author. E-mail: marian{at}esc.cam.ac.uk
|
REFERENCES |
|---|
Asimow PD and Ghiorso MS. (1998) Algorithmic modifications extending MELTS to calculate subsolidus phase relations. American Mineralogist 83:1127–1131.[Abstract]
Bédard JH and Sparks RSJ. (1991) Comment on: ‘The structure and petrogenesis of the Trallvall and Ruinsival areas of the Rhum ultrabasic complex’. By J. A. Volker & B. J. G. Upton. Transactions of the Royal Society of Edinburgh: Earth Sciences 82:389–390.[Web of Science]
Bédard JH, Sparks RSJ, Renner R, Cheadle MJ, Hallworth MA. (1988) Peridotite sills and metasomatic gabbros in the Eastern Layered Series of the Rhum Complex. Journal of the Geological Society, London 145:207–224.
Boudreau AE and McCallum IS. (1992) Infiltration metasomatism in layered intrusions—an example from the Stillwater Complex, Montana. Journal of Volcanology and Geothermal Research 52:171–183.[CrossRef][Web of Science]
Brown GM. (1956) The layered ultrabasic rocks of Rhum, Inner Hebrides. Philosophical Transactions of the Royal Society of London, Series B 240:1–53.
Butcher AR, Young IM, Faithfull JW. (1985) Finger structures in the Rhum Complex. Geological Magazine 122:503–518.[Abstract]
Deer WA, Howie RA, Zussman J. (1966) ednAn Introduction to the Rock Forming Minerals 2nd (Longman, Harlow).
Dunham AC and Wadsworth WJ. (1978) Cryptic variation in the Rhum Layered Intrusion. Mineralogical Magazine 42:347–356.[Web of Science]
Elliott MT, Cheadle MJ, Jerram DA. (1997) On the identification of textural equilibrium in rocks using dihedral angle measurements. Geology 25:355–358.
Emeleus CH. (1994) 1:20 000 Solid Geology Map of Rum 2nd (Scottish Natural Heritage, Edinburgh).
Emeleus CH. (1997) Memoir for 1:50 000 Geological Sheet 60 (Scotland). British Geological Survey. Geology of Rum and the Adjacent Islands(The Stationery Office, London).
Ghiorso MS and Sack RO. (1995) Chemical mass transfer in magmatic processes. IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology 119:197–212.[Web of Science]
Hallworth MA. (1998) Laboratory investigations of geological fluid flows. Ph.D. thesis , Anglia Polytechnic University, Cambridge.
Henderson P. (1975) Reaction trends shown by the chrome-spinels of the Rhum layered intrusion. Geochimica et Cosmochimica Acta 39:1035–1044.[CrossRef][Web of Science]
Holness MB. (1999) Contact metamorphism and anatexis of Torridonian arkose by minor intrusions of the Rum Igneous Complex, Inner Hebrides, Scotland. Geological Magazine 136:527–542.[Abstract]
Holness MB, Cheadle MJ, Mckenzie D. (2005) On the use of changes in dihedral angle to decode late-stage textural evolution in cumulates. Journal of Petrology 46:1565–1583.
Holness MB. (2005) Spatial constraints on magma chamber replenishment events from textural observations of cumulates: the Rum Layered Intrusion, Scotland. Journal of Petrology 46:1585–1601.
Holness MB. (2006) Textural immaturity of cumulates as an indicator of magma chamber processes: infiltration and crystal accumulation in allivalites from the Rum Layered Series. Journal of the Geological Society, London (in press).
Holness MB, Nielsen TFD, Tegner C. (2006) Textural maturity of cumulates: a record of chamber filling, liquidus assemblage, cooling rate and large-scale convection in layered intrusions. Journal of Petrology 46:141–157.
Irvine TN. (1980) Magmatic infiltration metasomatism and double-diffusive fractional crystallisation, and adcumulus growth in the Muskox intrusion and other layered intrusions. In Hargraves RB (Ed.). Physics of Magmatic Processes(Princeton University Press, Princeton, NJ) pp. 325–383.
Lo Ré FC. (2004) Microstructure, crystallography, and geochemistry of the Unit 9 allivalite, Eastern Layered Series, Rum Intrusion, Scotland. Ph.D. thesis, University of Wyoming, Laramie pp. 190.
Marsh BD. (1981) On the crystallinity, probability of occurrence, and rheology of lava and magma. Contributions to Mineralogy and Petrology 78:85–98.[CrossRef][Web of Science]
McBirney AR. (1987) Constitutional zone refining of layered intrusions. In Parsons I (Ed.). Origins of Igneous Layering(D. Reidel, Dordrecht) pp. 437–452.
McClurg JE. (1982) Petrology and evolution of the northern part of the Rhum Ultrabasic Complex. Ph.D. thesis, University of Edinburgh.
McKenzie DP. (1984) The generation and compaction of partially molten rock. Journal of Petrology 25:713–765.
Morse SA. (1994) Basalts and Phase Diagrams: an Introduction to the Quantitative Use of Phase Diagrams in Igneous Petrology(Krieger, Malabar) pp. 129–149.
Osborn EF and Tait DB. (1952) The system diopside–forsterite–anrthoite. American Journal of Science, Bowen Volume 413–433.
Philpotts AR, Brustman CM, Shi JY, Carlson WD, Denison C. (1999) Plagioclase-chain networks in slowly cooled basaltic magma. American Mineralogist 84:1819–1829.[Abstract]
Reed SJB. (1995) Electron microprobe analysis. In Potts PJ, Bowles JFW, Reed SJB, Cave MR (Eds.). Microprobe Techniques in the Earth Sciences(Chapman & Hall, London) pp. 49–89.
Renner R and Palacz Z. (1987) Basaltic replenishment of the Rhum magma chamber: evidence from unit 14. Journal of the Geological Society, London 144:961–970.
Robins B. (1982) Finger structures in the Lille Kufjord Layered Intrusion, Finnmark, Northern Norway. Contributions to Mineralogy and Petrology 81:290–295.[CrossRef][Web of Science]
Shirley DN. (1986) Compaction of igneous cumulates. Journal of Geology 94:795–809.[Web of Science]
Tait SR. (1984) Fluid dynamical processes in the formation of layered igneous rocks. Ph.D. thesis , University of Cambridge.
Tait SR. (1985) Fluid dynamic and geochemical evolution of cyclic unit 10, Rhum, Eastern Layered Series. Geological Magazine 122:469–484.[Abstract]
Turbeville BN. (1993) Sidewall differentiation in an alkalic magma chamber: evidence from syenite xenoliths in tuffs of the Latera caldera, Italy. Geological Magazine 130:453–470.[Abstract]
Upton BGJ, Skovgaard AC, McClurg J, Kirstein L, Cheadle M, Emeleus CH, Wadsworth WJ, Fallick AE. (2002) Picritic magmas and the Rum ultramafic complex, Scotland. Geological Magazine 139:437–452.
Volker JA and Upton BGJ. (1990) The structure and petrogenesis of the Trallvall and Ruinsival areas of the Rhum ultrabasic complex. Transactions of the Royal Society of Edinburgh: Earth Sciences 81:69–88.[Web of Science]
Volker JA and Upton BGJ. (1991) Reply to comments by J. H. Bédard & R. S. J. Sparks. Transactions of the Royal Society of Edinburgh: Earth Sciences 82:391.[Web of Science]
Von Bargen N and Waff HS. (1986) Permeabilities, interfacial areas and curvatures of partially molten systems: results of numerical computations of equilibrium microstructures. Journal of Geophysical Research 91:9261–9276.
Wadsworth WJ. (1994) The peridotite plugs of northern Rum. Scottish Journal of Geology 30:167–174.
Walker D, Shibata T, DeLong SE. (1979) Abyssal tholeiites from the Oceanographer Fracture Zone. II. Phase equilibria and mixing. Contributions to Mineralogy and Petrology 70:111–125.[CrossRef][Web of Science]
Young IM and Donaldson CH. (1985) Formation of granular textured layers and laminae within the Rhum crystal pile. Geological Magazine 122:519–528.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. B. HOLNESS and B. WINPENNY The Unit 12 allivalite, Eastern Layered Intrusion, Isle of Rum: a textural and geochemical study of an open-system magma chamber Geological Magazine, May 1, 2009; 146(3): 437 - 450. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Gaeta, T. Di Rocco, and C. Freda Carbonate Assimilation in Open Magmatic Systems: the Role of Melt-bearing Skarns and Cumulate-forming Processes J. Petrology, February 1, 2009; 50(2): 361 - 385. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. B. Holness, S. A. Morse, and C. Tegner Response to Comment by McBirney, Boudreau and Marsh J. Petrology, January 1, 2009; 50(1): 97 - 102. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||