| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 9 | Pages 1619-1656 | 2003
© Oxford University Press 2003
The Origin of Basic–Ultrabasic Sills with S-, D-, and I-shaped Compositional Profiles by in Situ Crystallization of a Single Input of Phenocryst-poor Parental Magma
GEOLOGICAL INSTITUTE, KOLA SCIENCE CENTRE, APATITY, 184200, RUSSIA
* Corresponding author. Present address: Institute of Geosciences, P.O. Box 3000, FIN—90014, University of Oulu, Oulu, Finland. Telephone: +358 8 553 1452. Fax: +358 8 553 1484. E-mail: rais.latypov{at}oulu.fi
RECEIVED FEBRUARY 8, 2002; ACCEPTED MARCH 7, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONREVIEW OF SILLS WITH...PHENOCRYSTS IN SILLS:...A PROPOSED IN SITU...IMPLICATIONS OF THE PROPOSED...PALISADES SILL: AN ORIGIN...CONCLUSIONSREFERENCES |
|---|
An attempt is made to develop an in situ crystallization model based on the concept of Soret fractionation to explain the origin of commonly observed S-, D-, and I-shaped compositional profiles in sills formed from a single pulse of phenocryst-poor parental magma. The model envisages that the various compositional profiles observed in sills can be interpreted in terms of different combinations of four principal units—Basal Zone and Layered Series forming the floor sequence, and Top Zone and Upper Border Series constituting the roof sequence. The Basal and Top Zones represent mirror images of the Layered and Upper Border Series, respectively, and therefore are referred to as basal and top reversals. It is proposed that the formation of basal and top reversals takes place through the non-equilibrium Soret differentiation of liquid boundary layers which form as a consequence of the temperature gradient imposed by the cold country rock. In contrast, the Layered and Upper Border Series originate predominantly through the crystal–liquid boundary layers developing in equilibrium conditions. The model permits the production of S-, D-, and I-shaped compositional profiles from the same magma composition. All that is necessary to produce a specific shape of compositional profile is an appropriate temperature gradient imposed by the cold country rock on the liquid boundary layers of a parental magma of a given composition.
KEY WORDS: sills; compositional profiles; in situ crystallization; Soret fractionation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONREVIEW OF SILLS WITH...PHENOCRYSTS IN SILLS:...A PROPOSED IN SITU...IMPLICATIONS OF THE PROPOSED...PALISADES SILL: AN ORIGIN...CONCLUSIONSREFERENCES |
|---|
The study of basic–ultrabasic sills may provide some of the best insights into the processes leading to the differentiation of natural silicate magmas. Sills usually show complete rock sequences including roof and basal chilled zones, which provide information about phase, modal and cryptic layering of magmatic bodies. Apart from those examples where multiple intrusions are involved, the bulk composition of sills can provide a reliable estimate of the parental magma composition, which is of paramount importance for any model of magma differentiation. The course of emplacement and solidification is also commonly clear in sills, allowing resolution of the conflict over whether the final rock sequence was produced by a single injection or by multiple injections.
Three main types of basic–ultrabasic sill have been identified on the basis of the shape of the modal mineralogical variation, or MgO and SiO2 compositional profile as a function of stratigraphic height (Gibb & Henderson, 1992; Marsh, 1996); these are sills with S-, D- and I-shaped compositional profiles (Fig. 1). The well-known S-type sills are the most differentiated bodies, which are characterized by high concentrations of olivine near the base (Gray & Crain, 1969; Fujii, 1974; Frenkel' et al., 1988, 1989; Marsh, 1989). The term D-type sill was introduced to distinguish bodies in which modal olivine contents tend to decrease towards the upper and lower margins in a roughly symmetrical manner (Gibb & Henderson, 1992). The name I-type sill (Marsh, 1996) is used for bodies showing little field, petrographic or geochemical evidence of differentiation and displaying neither well-developed cumulate zones nor late-stage differentiates (Gunn, 1966; Froelich & Gottfried, 1988; Mangan & Marsh, 1992).
|
There have been several attempts to resolve the problem of the existence of sills with S-, D- and I-shaped compositional profiles (see discussion by Simkin, 1967; Gibb & Henderson, 1992; Marsh, 1996). Different mechanisms have been suggested to explain this phenomenon. Among them are a gravity-induced settling of crystals, either already present in the magma at the time of emplacement (Gray & Crain, 1969; Fujii, 1974; Marsh, 1989) or newly grown in the magma chamber (Frenkel' et al., 1988, 1989; Woster et al., 1990, 1993), flow differentiation (Komar, 1972; Marsh, 1996), multiple injection of distinct magmas with different compositions (Gibb & Henderson, 1992; Czamanske et al., 1994, 1995), and the convective flux of refractory components within the crystal–liquid mush of the boundary layer during in situ differentiation (Tait & Jaupart, 1996). All of the above processes are probable and, in some circumstances, can contribute to the final shape of the compositional profiles. They are unlikely, however, to represent the dominant explanation for the appearance of sills with different compositional profiles because they are unable to explain the formation of the basal and top reversals that represent the distinctive characteristic feature of many sills (Latypov, 2003).
In this paper a model is developed that can explain the origin of all types of compositional profiles in sills formed from a single pulse of phenocryst-poor or phenocryst-free parental magma. By a single pulse I imply here that the sills are filled to their present width continuously with the same magma composition on a time scale much shorter than that of solidification. The model represents a combination of the model of in situ crystallization in thermal boundary layers of Tait & Jaupart (1996) with a mechanism for the origin of marginal reversals by Soret fractionation (Latypov, 2003).
|
REVIEW OF SILLS WITH DIFFERENT COMPOSITIONAL PROFILES AND CURRENT MODELS FOR THEIR FORMATION |
|---|
|
TOPABSTRACTINTRODUCTIONREVIEW OF SILLS WITH...PHENOCRYSTS IN SILLS:...A PROPOSED IN SITU...IMPLICATIONS OF THE PROPOSED...PALISADES SILL: AN ORIGIN...CONCLUSIONSREFERENCES |
|---|
Description of sills with S-, D- and I-shaped compositional profiles
In the following discussion I shall adhere to the conventional classification of sills into those with S-, D- and I-shaped profiles (Fig. 1) although I find that this division is rather schematic and does not embrace all the types of compositional profiles that are observed in nature.
S-shaped profiles
The best descriptions of the typical S-shaped sills are provided by detailed studies of differentiated alkaline sills of Skye, Scotland (Drever & Johnston, 1967; Simkin, 1967), picrite–gabbro–dolerite sills of the Noril'sk region, Russia (Distler et al., 1979; Czamanske et al., 1994, 1995; Turovtsev et al., 2000), and dolerite sills of Antarctica (Gunn, 1966). Modal variations in vertical sections through most alkaline S-shaped sills (Fig. 2a–c) typically show an olivine-poor lower zone followed by a rapid, but gradational, increase in the amount of olivine to a picrite unit that constitutes one-half to three-quarters of the sill thickness. The picrite unit with 30–50 vol. % olivine is succeeded abruptly by an olivine-poor picrodolerite–analcime gabbro unit with 
10 vol. % Ol, which continues to the upper margin. Such discontinuities between rocks relatively rich and poor in olivine are observed in nearly all alkaline dolerite sills and are commonly rather sharp (Simkin, 1967). Gibb & Henderson (1989) and Gibson & Jones (1991) reported that the sharp breaks are also marked by interior chills, whereas Simkin (1967, p. 64), in contrast, argued that there is no evidence for interior chilling at these discontinuities.
|
Similar modal variations are observed in most sills in the Noril'sk region, where a relatively sharp, unchilled contact between picritic gabbro–dolerite and overlying olivine gabbro–dolerite is marked by an abrupt decrease in olivine content from 50–60 vol. % to 10–20 vol. %. Passing upwards from olivine gabbro–dolerite to olivine-bearing gabbro–dolerite, and then to prismatic and magnetite gabbro, the amount of olivine decreases further until it completely disappears. The transition from picritic units to underlying taxitic gabbrodolerite in these sills is characteristically gradational. An important peculiarity of many Siberian picrite–gabbrodolerite sills (Lebedev, 1957; Zolotuhin & Vasil'ev, 1967; Ivanov et al., 1971; Ryabov, 1992; Turovtsev et al., 2000) and some alkaline sills (Drever & Johnston, 1967; Gibb & Henderson, 1996; Foland et al., 2000) is the occurrence of upper olivine-rich units (Fig. 2d–f). It is noteworthy that the modal and mineral compositions of such units cover practically the whole range of rock compositions from the entire intrusion section (Duzhikov & Distler, 1992; Gibb & Henderson, 1996). This type of compositional variation, which I refer to as a ‘double-humped’ profile, is classified here as a variety of the S-shaped profile.
Typical S-shaped compositional profiles are also exhibited by Antarctic dolerite sills (Gunn, 1962, 1966). These sills usually include a thick layer of hypersthene gabbro or norite centred somewhat below the middle of the sills and commonly making up a third of the entire bulk of the sill. This lithology grades abruptly or gradationally both above and below into a pigeonitic dolerite in which orthopyroxene is entirely absent. This transition is accompanied by a marked decrease in whole-rock mg-number, and Cr and Ni concentrations. A generalized stratigraphic section through one such sill, at Lake Vanda, showing a profile intermediate between S- and D-shape, is shown in Fig. 3.
|
D-shaped profiles
Some typical examples of D-shaped sills, which are distinguished by thick olivine-rich central units, and roughly symmetrical olivine-poor upper and lower margins, are illustrated in Fig. 4. The most primitive bulk compositions in such sills are not at the margins but near the middle; the most evolved compositions are at the chilled margins (Marsh, 1996). Compositional profiles through the D-type Lugar sill, Scotland, are shown in Fig. 5. Here, the central olivine-rich unit consists mainly of picrite and minor kaersutite nepheline gabbro and nepheline gabbro. It is noteworthy that the four uppermost analcime gabbro units are mirror images of the four bottom units, both being the most evolved rocks in the sill in terms of major and trace elements.
|
|
I-shaped profiles
The existence of I-shaped profiles through sills and lava flows, which exhibit very little modal or whole-rock chemical variation, has been noted by many petrologists (Gunn, 1966; Froelich & Gottfried, 1988; Mangan & Marsh, 1992). It is repeatedly stressed that the bodies displaying I-shaped profiles are remarkably undifferentiated, show no cumulates and have no systematic vertical chemical zonation despite being sometimes up to 400–500 m thick (Fig. 6). It should, however, be noted that many sills of interest are I-shaped only in terms of modal composition or major elements such as SiO2 and MgO. In terms of mg-number and trace elements, they can reveal fairly well-developed compositional profiles. As an example, the New Mountain sill, Antarctica, despite having an I-shaped modal profile, exhibits a typical S-shaped cryptic profile in terms of mg-number, Cr and Ni (Fig. 7). It is therefore incorrect to consider that these types of sills do not exhibit evidence of differentiation. Most of them, if not all, show evidence of fairly well-developed differentiation, especially in terms of trace elements, which are the most sensitive indicators of magma evolution for this type of sill.
|
|
Current models for the formation of S-, D- and I-shaped compositional profiles in sills
Four principal mechanisms have been invoked to explain the olivine-rich rocks at the base of S-shaped sills. Settling of phenocrysts or newly formed phases is the most traditionally cited mechanism to produce large concentrations of olivine at the bottom of igneous bodies (Wager & Brown, 1968; Gray & Crain, 1969; Fujii, 1974; Frenkel' et al., 1988, 1989; Marsh, 1988, 1989). The hydrodynamic migration of phenocrysts away from the margins of the intrusion during flow is considered to result in olivine-poor margins whereas gravity aids the production of the abrupt transition between picrite and olivine-poor overlying rocks (Simkin, 1967). The occurrence of basal reversals, upper picrite units and especially a relatively sharp contact between picritic and gabbroic units in the sills of the Noril'sk region are usually taken as evidence for their formation by two- or multiphase injection of distinct melts (Rogover, 1959; Smirnov, 1966; Ivanov et al., 1971; Czamanske et al., 1994, 1995) or emplacement of immiscible basic and picritic melts (Marakushev et al., 1982; Ryabov, 1992). Similar features in basic alkaline sills have also been attributed to multiple emplacement of compositionally distinct, but genetically related magmas (Gibson & Jones, 1991; Gibb & Henderson, 1996; Foland et al., 2000). Recently, Tait & Jaupart (1996) have shown that S-shaped profiles in sills could also be produced by the convective flux of refractory components into a crystal–liquid mush within a boundary layer in the course of in situ differentiation of phenocryst-poor magma.
The most common explanation for the origin of sills with D-shaped profiles is flow differentiation involving the segregation of heavy particles (phenocrysts) towards the centre of the flow—the Bagnold effect (e.g. Gunn, 1966; Simkin, 1967; Marsh, 1996). The tendency towards D-shaped profiles in sills is also accounted for by magma intrusion in a series of pulses carrying increasingly higher contents of olivine phenocrysts (e.g. Henderson & Gibb, 1987; Gibb & Henderson, 1992).
Only one explanation has so far been proposed for the origin of sills with I-shaped compositional profiles. This involves the injection of phenocryst-poor magma that crystallizes predominantly through the upper and lower solidification fronts, preventing any evolution of the magma because of the capture of all interstitial liquids (Mangan & Marsh, 1992; Marsh, 1996).
Because of space restrictions, it is impossible to present here a detailed discussion of all these published concepts. Brief summaries of some of the main arguments and evidence are presented below, which lead to the conclusion that existing models are not adequate to explain the characteristic features of S-, D- and I-shaped compositional profiles in sills.
Arguments against the formation of sills by crystal settling and flow differentiation
- The fact that sills with I-shaped modal profiles can exhibit well-developed S-shaped cryptic profiles in terms of mg-number and trace element variation (Fig. 7) testifies strongly that crystal settling and flow differentiation are not responsible for the formation of S-shaped profiles. Some other physicochemical process must govern the generation of the S-shaped profiles.
- Neither crystal settling nor flow differentiation can produce adequate separation of orthopyroxene from pigeonite in the Antarctic dolerite sills (Fig. 3), as these minerals have practically the same density. A physicochemical rather than a mechanical process must be responsible for such a regular distribution of pyroxenes.
- The occurrence of two maxima in olivine concentration within some sills (Fig. 2d–f) is difficult to reconcile with a gravity-controlled settling crystal model involving a single magma pulse.
- Detailed petrographic investigations (e.g. Frenkel' et al., 1988; Czamanske et al., 1995) show that the Layered Series in sills is primarily composed of cumulates with nearly cotectic mineral proportions. This does not support a crystal settling mechanism, which generally leads to non-cotectic compositions of rocks as a result of appreciable separation of dense crystals from light ones during their gravitational settling to the floor (Morse, 1980).
- The roughly cotectic composition of rocks comprising marginal reversals suggests that, in most cases, the intruding magmas were phenocryst poor and consequently cannot produce any appreciable concentration of phenocrysts at the base of the sills (Latypov, 2003).
Arguments against the formation of sills by multiple magma injection
- The most impressive evidence for multiple magma injection into sills is the occurrence of pronounced reversals in normal fractionation trends, which are not expected from basic magmas undergoing classical fractional crystallization. Internal contacts between individual units of intrusions that are commonly marked by grain size and textural changes and often interpreted, therefore, as interior chills (e.g. Henderson & Gibb, 1987) provide less reliable evidence for multiple magma injections. This is because the distinct internal contacts can be principally explicable in the frame of a single intrusive episode if the dynamics of magma cooling in the chamber are taken into account (Huppert & Sparks, 1989). The compositional reversals mentioned above are clearly recognized from (a) abrupt changes in the order of phase crystallization and (b) prominent shifts towards less evolved compositions of cumulus minerals. There are numerous documented examples of such reversals in large basic–ultrabasic layered intrusions (e.g. Irvine, 1970; Alapieti, 1982; Todd et al., 1982; Faithfull, 1985; Eales & Cawthorn, 1996; Latypov et al., 1999a, 1999b). For instance, a major shift in the crystallization sequence from the Ultramafic Series (Ol
Ol + Opx Opx Opx + Pl Opx + Pl + Cpx) to the Lower Banded Series (Ol + Pl Pl + Opx Pl + Opx + Cpx) is recognized in the Stillwater Complex, testifying that the latter series was formed from a new, distinctly different, magma that was emplaced into the chamber (Todd et al., 1982). There is, however, no indication of such compositional reversals at the contact between picritic and gabbroic units in most alkaline (Gibson & Jones, 1991; Gibb & Henderson, 1996; Foland et al., 2000) and picrite–gabbrodolerite sills (Duzhikov & Distler, 1992; Czamanske et al., 1994, 1995). They are also absent at the contact between norite and gabbro units in dolerite sills from Antarctica (Gunn, 1966). As an illustration, the internal stratigraphy of the Noril'sk-type intrusions is broadly predictable from in situ differentiation of a single pulse of basic magma. The overall trend of crystallization correlates well with the following sequence of phase appearance: Ol Ol + Sp Ol + Pl + Sp Ol + Pl Ol + Pl + Cpx Ol + Pl + Cpx + Opx (Likhachev, 1965, 1977; Nekrasov & Gorbachev, 1978; Zen'ko, 1983). The transition from the picritic units to the overlying layered gabbroic series of the Noril'sk sills reveals no indication of reversals, neither in the sequence of phase crystallization nor in mineral compositions.
- Convincing arguments against a multiple magma injection model for the alkaline sills of north Skye have been presented by Simkin (1967, p. 64). He pointed out that the ‘multiple intrusion of liquids of differing compositions would require initial intrusion of basaltic liquid (basaltic chilled margins) followed, with no significant time lag (absence of interior chilling), by a more basic, perhaps ultrabasic, liquid at the base of each sill. A rather delicate mixing of these two liquids would seem necessary to explain the gradational dolerite-to-picrite transitions at the base, a relatively cool part of the sill where abrupt, if not chilled, junctions would seem more likely. However, in the upper central, and presumably hottest, part of the sill the junction is invariably sharp. Addition of a third pulse of basaltic liquid might explain the upper discontinuities, but would require three separate intrusions in the same time and space sequences throughout all north Skye sills.’ It remains to be added that acceptance of the multiple injection model would require at least three separate intrusions in the same space sequences not only for alkaline sills (nepheline-normative magma), but also for picrite–gabbrodolerite (olivine-normative magma) and gabbroic sills (quartz-normative magma). This leaves little room for doubt that a model based on accidental events such as multiple magma injection is untenable for the explanation of the features of S-shaped sills that are commonly observed in igneous bodies throughout the world regardless of their parental magma composition.
- The fact that some I-type sills, lacking any commonly cited ‘evidence’ for new magma pulses (basal reversals, sharp contact between lower picritic and gabbroic units and occurrence of upper picritic units), possess S-shaped cryptic profiles (Fig. 7) strongly indicates that the formation of the S-type sills is not necessarily related to multiple magma injection.
- The occurrence of numerous S- and D-type sills with chilled margins that are much more evolved than the bulk sill composition is often taken as strong evidence for multiple magma injection (e.g. Simkin, 1967; Froelich & Gottfried, 1988; Czamanske et al., 1995). This stems primarily from the widespread notion that a single homogeneous liquid should yield a chilled margin representative of the bulk composition of the sill. However, a new model for the formation of marginal compositional reversals in sills and layered intrusions (Latypov, 2003) suggests that relatively evolved chilled margins can be produced, even from a homogeneous liquid, by the operation of Soret diffusion along the intrusion margins. The evolved nature of chilled margins, therefore, cannot be regarded as an indisputable argument for the origin of sills by multiple magma injection.
Arguments against the formation of sills by in situ crystallization
Whereas in situ crystallization in thermal boundary layers is a principal component of the model presented below for the origin of sills with different compositional profiles, this process can be applied successfully only in combination with the process of Soret fractionation (Latypov, 2003). The reasons why an in situ crystallization model cannot be employed in the way that it was proposed by Tait & Jaupart (1996) are as follows:
- the in situ crystallization model is based on the assumption that the chilled margins of intrusions are representative of the composition of the parental magma, which has not experienced fractionation. However, numerous data on sills and layered intrusions show that this assumption is erroneous as the compositions of chilled margins, along with the lower parts of basal zones, are generally much more evolved than the bulk intrusion composition (Latypov, 2003). This apparent lack of mass balance between the chilled margins and the bulk composition of intrusions, which is one of the essential properties of marginal reversals, finds no explanation in the framework of the in situ crystallization model.
- The in situ crystallization model makes no provision for the generation of top compositional reversals that are so characteristic of sills with double-humped compositional profiles (Fig. 2d–f).
A line of attack on the problem of sills with I-shaped compositional profiles
It has been proposed that I-shaped profiles result from the injection of phenocryst-poor magma where crystal growth occurs predominantly within the solidification front, inhibiting differentiation (Mangan & Marsh, 1992; Marsh, 1996). This explanation seems plausible, particularly for thin sills in which the rate of advance of solidification fronts is high enough to hamper the return of interstitial liquid into the main mass of magma. However, in my opinion this is not the main reason controlling the generation of I-type sills. Close examination of the modal and chemical composition of some typical I-type dolerite sills and lava flows indicates that they crystallized along the Opx + Pl + Cpx + L cotectic in the quartz-normative section of the phase diagram Ol–Cpx–Pl–Qtz (Fig. 8). This gives an immediate answer as to why I-type sills show little modal and major element variation: the four-phase cotectic liquids simply cannot produce either basal one-phase cumulates or a well-developed Layered Series. In principle, the entire section of such sills must comprise cotectic pl–pig–aug cumulates (+ intercumulus quartz), closely corresponding to the parental magma composition. Only at the level of the Sandwich Horizon can a minor amount of quartz-rich pegmatoidal material be produced from fluid-rich residual liquid of eutectic composition 
. Obviously, the parental magmas for quartz-normative dolerite sills are not the only ones that are able to produce I-shaped compositional profiles. Any magma with a near-eutectic composition can result in the formation of such types of sills. For instance, one can predict that parental magmas close to the eutectic point Ol + Cpx + Pl + L will produce an I-shaped profile with small olivine contents (10–15 vol. %) throughout the entire section of the sill. Olivine-poor sills with a roughly I-shaped modal profile have been reported in the literature (e.g. Gibb & Henderson, 1978; Ryabov, 1992).
|
|
PHENOCRYSTS IN SILLS: INTRATELLURIC VERSUS IN SITU FORMATION |
|---|
|
TOPABSTRACTINTRODUCTIONREVIEW OF SILLS WITH...PHENOCRYSTS IN SILLS:...A PROPOSED IN SITU...IMPLICATIONS OF THE PROPOSED...PALISADES SILL: AN ORIGIN...CONCLUSIONSREFERENCES |
|---|
There is a rather common opinion among petrologists that intratelluric phenocrysts formed at depth before magma intrusion play an important role in the chemical evolution of sills as recorded in their compositional profiles (Frenkel' et al., 1988, 1989; Marsh, 1988, 1989, 1990, 1991, 1996; Mangan & Marsh, 1992; Chalokwu et al., 1993; Czamanske et al., 1994, 1995; Ariskin, 1999; Hort et al., 1999; Ariskin & Barmina, 2000). Marsh (1996, p. 13) put forward the clearest formulation of this idea: ‘no phenocrysts, no post-emplacement differentiation’. In a companion paper (Latypov, 2003) I have suggested, in contrast, that most sills form from phenocryst-poor magmas. I have also noted that the well-developed compositional profiles observed in a number of I-type sills formed from phenocryst-poor magmas are not consistent with the critical role of phenocrysts in magma differentiation. The possible role of intratelluric phenocrysts in the formation of S- and D-type sills requires, however, more detailed examination. There are at least two principal lines of evidence that are taken as indicative of the formation of S- and D-types of sills from parental magmas enriched in phenocrysts. These are the occurrence of intratelluric phenocrysts in the (a) basal and middle cumulate zones of sills and (b) chilled margins.
Phenocrysts in basal and middle cumulates
Phenocryst sizes
Marsh (1996) has reported that many basaltic sills of varying sizes contain dense accumulations of large (2–10 mm) olivine or orthopyroxene phenocrysts near the base (S-type sills) or in the middle (D-type sills). He also suggested on theoretical grounds that such anomalously sized phenocrysts alone precludes them from having grown in situ. To explore this issue I have compiled crystal size data from many literature sources that cover approximately 400 basic–ultrabasic sills (Table 1). At least three points are evident from the examination of this table.
- Crystal sizes throughout the I-type Peneplain, Ivanovskii, Mikado, Tasmanian and Dæmmevands sills, which are reported to have formed from phenocryst-free or phenocryst-poor magmas, reach 1–3 mm, and in some cases attain 4–12 mm.
- Crystal sizes in rocks of S- and D-type sills that were undoubtedly formed in situ from a late-stage evolving melt (pyroxenite, gabbro, dolerite, diorite, analcime gabbro) can reach 3–6 mm (e.g. Basistoppen sill), and in some cases 10–15 mm (e.g. Koli sill, Pechenga intrusions).
- Crystal sizes of olivine, orthopyroxene and clinopyroxene from the basal or middle cumulate zones of most S- and D-type sills vary from 0·1 to 3 mm. The only exception seems to be the York Haven section of the York Haven sill, where orthopyroxene is reported to attain 5–10 mm.
|
The main conclusion that can be drawn from the available data is that crystals in basal and middle cumulate zones in most S- and D-type sills are commonly <3 mm in size. Formation of crystals of such size, on the basis of theoretical calculations (Marsh, 1996) and observations on sills formed from phenocryst-free magmas (Table 1), does not require pre-eruption growth. Crystals can attain such size by in situ growth in sills. This means that the basal and middle cumulates in basic–ultrabasic sills can be principally produced in situ from phenocryst-poor or even phenocryst-free parental magmas. In addition, the data show that crystals that grow in sills during the middle and late stages of magma differentiation can attain 10–15 mm. This suggests that the 2–10 mm size limit theoretically proposed for the identification of intratelluric phenocrysts in sills (Marsh, 1996) needs to be reconsidered.
Phenocryst size/frequency variations
Strong evidence for the existence of two generations of crystals (intratelluric and in situ phenocrysts) in sills would be a discovery of bimodal crystal size populations. Wyllie & Drever (1963) and Gibb (1968) have carried out such investigations of olivine size distribution in small picritic intrusions of the Isle of Skye, Scotland. Both studies have shown that these intrusive bodies have a continuous gradation in the size of olivine rather than two distinct size groups. These results do not completely preclude the existence of two generations of olivine crystallization, however, because bimodal distributions can fail to occur if olivine of a second generation is deposited on the surfaces of first generation crystals in addition to forming new crystals (Gibb, 1968). It is important, however, to note that the lack of bimodal size distribution does not, at least directly, support the idea of the existence of two generations of crystals in sills.
Phenocryst morphology
Czamanske et al. (1995) carried out perhaps the only detailed investigation of the morphology of phenocrysts throughout sill sections. They presented convincing evidence for the existence of two morphologically different generations of olivine in Noril'sk type intrusions. The first-generation olivine, having a mainly idiomorphic form and containing no inclusions of other minerals, occurs solely in the picritic units of the intrusions and is identified as intratelluric phenocrysts. The second-generation olivine is represented mainly by (1) large oikocrystic or xenomorphic grains, containing, or being penetrated by, laths and microlites of plagioclase and (2) small to very small, rounded and idiomorphic grains, which are common only in the gabbroic units of the intrusions and are identified as an in situ type of phenocryst. Although such a subdivision of morphologically different olivines into products of intratelluric and in situ stages of crystallization is possible, it is certainly not the only explanation. It is conceivable that both morphological generations of olivine are produced in situ. The first-generation olivine in the picrites crystallized as a sole liquidus phase while the parental magma fractionated through the primary volume of olivine. This is why this olivine contains no inclusions of other minerals. The second-generation olivine crystallized during liquid fractionation along an Ol + Pl + L cotectic (Ol 15%, Pl 85%) and at an Ol + Pl + Cpx + L eutectic point (Ol 10%, Pl 60%, Cpx 30%) where olivine is a modally subordinate phase. Nucleation and crystallization of olivine here can be suppressed by that of modally predominant plagioclase and clinopyroxene. This can explain why the second-generation olivine contains, or is penetrated by, crystals of plagioclase.
Phenocrysts in chilled margins
Porphyritic texture
The observation that crystals of two distinct average sizes (phenocrysts and groundmass crystals) are present in chilled margins is traditionally interpreted by petrologists in terms of a two-stage cooling history imposed by external factors. It is commonly supposed that the phenocrysts form slowly at depth whereas the groundmass forms on ‘quenching’ against cold country rocks. It became, however, clear from experimental studies on lunar samples that the two-stage crystallization implicit in a porphyritic texture does not necessarily imply a two-stage cooling history. It was demonstrated that a porphyritic texture could develop from single-stage cooling of basaltic liquids both at constant linear and decreasing nonlinear cooling rates (e.g. Lofgren et al., 1974; Walker et al., 1976; Lofgren, 1980). In particular, a bimodal grain-size distribution of pyroxene was obtained during single-stage cooling in all the dynamic crystallization experiments as well as the natural rocks. The necessary condition for the formation of a porphyritic texture is when the liquidus or near-liquidus phase(s) (olivine or pyroxene) are modally dominant and have a significant temperature interval of crystallization (up to 30°C) before the appearance of another liquidus phase. In this case, as the liquid cools, the first generation of olivine (or pyroxene) nucleates and grows, without competition from other crystallizing phases, to typical phenocryst size (Lofgren, 1980). Upon reaching a cotectic line a new groundmass generation of smaller olivine (or pyroxene) and plagioclase starts to form although the precise mechanism of its production remains obscure (Dowty, 1980). Formation of the first generation of phenocrysts can be facilitated by a small amount of nuclei of sub-micron size ranges present upon injection of magma in the chamber. The nuclei can represent the unmelted remains of crystalline material that mixed with the magma during transport to the surface. The nuclei can also be formed by homogeneous nucleation during the long pre-emplacement history of the magma (Lofgren, 1983). In general, simple conductive cooling of a magma body that solidifies because of heat loss through its margins seems a suitable environment for the production of a porphyritic texture (Walker et al., 1976).
Phenocryst morphology
Drever & Johnston (1957) first postulated that direct growth rather than resorption was responsible for the skeletal development of crystals in chilled margins. They concluded that there was ample evidence to show that large skeletal olivine phenocrysts could form in situ from a rapidly cooling magma and that such phenocrysts were not necessarily derived from elsewhere. Later Donaldson (1976) provided strong experimental support for this idea. He also showed that skeletal crystal shapes reflect not a rapid cooling rate, but rather a rapid growth rate caused by the high MgO content of the melt. Skeletal olivines grow at much slower cooling rates in high-MgO melts (picrites) than in more MgO-poor melts (basalts). On this basis, abundant hopper olivine in the porphyritic picritic minor intrusions of the Scottish and West Greenland Tertiary igneous provinces has been interpreted as an in situ phenocryst phase (Donaldson, 1976).
It should be emphasized that hopper olivine is not the only morphological type of olivine phenocryst that can form at sill margins. The hopper olivine in olivine-normative basalt forms at a relatively high cooling rate of 2·7–15°C/h; subhedral and granular olivine grows at <2·7°C/h (Donaldson et al., 1975; Donaldson, 1976). Cooling at such relatively slow rates is estimated to take place at the margins of small dykes and lava flows (Baker & Grove, 1985; Kretz et al., 1985). Hence it would be incorrect to think that relatively rapid cooling near the chilled margins of sills will almost certainly produce skeletal olivine (e.g. Gibb, 1968, 1972). The predominance of granular and scarcity of skeletal olivine phenocrysts in the chilled margins of some basic–ultrabasic sills suggests that the cooling rate was <2·7°C/h. An important factor in the formation of granular rather than skeletal phenocrysts can also be a small amount of nuclei present upon the injection of magma in the chamber. Growth on pre-existing nuclei makes the shapes of liquidus phases less dependent on variations in cooling rate. It was also experimentally demonstrated, by using plagioclase, that initially skeletal crystals can fill in to become tabular with continued growth at a cooling rate of 2°C/h (Lofgren, 1980). This important observation presents a real mechanism for how skeletal crystals can be readily transformed into granular ones leaving little trace of the initial crystal morphology. A much longer time period for crystallization of chilled margins in comparison with laboratory experiments, as well as crystallization of natural magmas under conditions of fluid pressure that significantly increase the rate of chemical diffusion, suggest that such a morphological transformation is possible.
An additional explanation for the rare occurrence of skeletal minerals in chilled margins comes from the experimental work of Kouchi et al. (1986). They have shown that stirring, or shear flow, during crystallization, dramatically changes both the nucleation rate and crystal morphology of plagioclase and clinopyroxene compared with experiments under static conditions. At amounts of supercooling >45°C, stirring increases the nucleation rate by 10 times and skeletal crystal morphologies are never observed. Thus the rare occurrence of skeletal crystals in the chilled margins of sills may also indicate that crystallization takes place during continuous magma flow that is provided by vigorous thermal convection in the chamber. A similar explanation has been adopted for the interpretation of the texture of quench rocks between macrorhythmic units in the Kap Edvard Holm gabbro complex (Tegner et al., 1993). Overall, it is clear from the above discussion that ample opportunities exist in nature for in situ growth of subhedral rather than skeletal crystals during the cooling of sill margins.
Phenocryst sizes and abundances
Tables 2 and 3 present a compilation of data on the shapes, sizes and abundances of phenocrysts observed in the chilled margins of sills and in slow-rate cooling experiments, respectively. Close examination of the data shows that crystal shapes (from hopper to granular) and sizes (up to 1–2 mm, with the exception of 6–7 mm long skeletal crystals) that have been attained at a cooling rate of 2–3°C/h are comparable with those documented in chilled margins. Thus, one can conclude that the size and shape of crystals is not always a criterion of critical significance in deciding whether or not their precipitation ante-dated or post-dated magma intrusion. There seems therefore to be no strong obstacle to the suggestion that most phenocrysts observed in chilled margins can be produced in situ. The total amount of phenocrysts in chilled margins, judging from the abundance of in situ formed hopper phenocrysts in the Chukotat group of sills, Quebec (Table 2), can reach as much as 10–40 vol. %.
|
|
Phenocryst composition
I am unaware of any studies presenting compositional criteria for distinguishing intratelluric phenocrysts from those formed in situ in chilled margins. It is reasonable to expect that the inner zones of true intratelluric phenocrysts are characterized by the highest mg-number (ol, opx, cpx) and An contents (pl) and show relatively smooth zoning profiles, whereas the outer parts of the phenocrysts, which are contemporaneous with the groundmass, will become more evolved towards the crystal margins. Analysis of experimental data shows, however, that similar compositional features can be attained by in situ phenocryst growth at a low rate of cooling. In particular, Walker et al. (1976) did not observe any decrease in the Fo content of the cores of olivine as a function of cooling rate (0·5–2000°C/h), temperature (1325–600°C), and pressure (0–12 kbar) for a constant bulk composition. Also, Grove & Bence (1977) failed to find any evidence of major element zoning in pyroxene, caused by cooling rates up to 150°C/h. Both observations were attributed to heterogeneous nucleation of minerals on Fe-metal particles present in the melt as a result of the iron-capsule technique employed (Lofgren, 1980). Thus, if as was suggested above, in situ phenocrysts in chilled margins nucleate and grow on pre-existing nuclei, then the core compositions will be hardly distinguishable from those of intratelluric phenocrysts. Compositional zoning also seems a poor guide for distinguishing intratelluric from in situ phenocrysts. Donaldson et al. (1975) reported no zoning in an olivine from a 2·7°C/h cooling rate experiment, suggesting that equilibrium is achieved at such a cooling rate. Lofgren et al. (1974) and Grove & Bence (1977) found that at cooling rates <3°C/h, olivine had time to react with the liquid to form pyroxene as the equilibrium phase relations suggest. At a cooling rate of
2°C/h the inner third to half of the plagioclase crystals show little zoning and only the outer portion is zoned (Lofgren, 1980). Thus there seems to be no reliable compositional criterion that would permit one to distinguish intratelluric phenocrysts from those formed in situ.
Concluding remark
The size, morphology and composition of phenocrysts in the basal and middle cumulates as well as in the chilled margins of sills present no compelling evidence for their intratelluric origin. It is conceivable that most of these phenocrysts actually form in situ. Those phenocrysts that are distinguished by especially large sizes could indeed represent intratelluric fractions of phenocrysts. In most cases, however, their amount is unlikely to exceed a few percent.
|
A PROPOSED IN SITU CRYSTALLIZATION MODEL FOR THIN SHEET-LIKE MAGMA BODIES |
|---|
|
TOPABSTRACTINTRODUCTIONREVIEW OF SILLS WITH...PHENOCRYSTS IN SILLS:...A PROPOSED IN SITU...IMPLICATIONS OF THE PROPOSED...PALISADES SILL: AN ORIGIN...CONCLUSIONSREFERENCES |
|---|
It follows from the above and a companion paper (Latypov, 2003) that current models of magma differentiation in the high-level magma chambers do not provide an adequate explanation for the origin of the different compositional profiles observed in sills. The obvious question is: What mechanism can resolve the problem? A close look at this problem yields the conclusion that the best candidate is an in situ crystallization model. There are several variants of this model (e.g. Campbell, 1978; McBirney & Noyes, 1979; Cawthorn & McCarthy, 1980, 1985; Sparks et al., 1984; Langmuir, 1989; Marsh, 1989; Nielson & De Long, 1992). I favour one of the most advanced variants that was developed by Jaupart & Tait (1995) and Tait & Jaupart (1996). As previously noted, however, this variant does not provide a reasonable interpretation for the origin of marginal compositional reversals in sills. To overcome this drawback I have attempted to modify the in situ crystallization model by including the effect of Soret fractionation to account for the formation of compositional reversals in sills (Latypov, 2003).
The basic principles of the model can best be illustrated by considering a sill with a double-humped compositional profile. For the sake of convenience let us take a basic–ultrabasic sill (Fig. 9a) whose parental magma composition X' lies in the primary olivine phase field in a simple silica-undersaturated system Ol–Pl–Cpx (Fig. 9b). The trend of crystallization of the parental magma X' can be readily traced in the phase diagram as shown by the letters located along the compositional profile for whole-rock mg-number, Cr and Ni (Fig. 9a). The model suggests that the sill is principally composed of floor and roof sequences that meet at the level of the Sandwich Horizon (Fig. 9a). The floor sequence of the sill is proposed to comprise the Basal Zone and the Layered Series, whereas the roof sequence contains the Top Zone and Upper Border Series. The lines of demarcation between the coupled units of the floor and roof sequences run through the lower and upper crossover maxima (Latypov, 2003) exhibiting the most primitive mineral and rock compositions in the entire section of the sill. The maxima do not coincide with any lithological boundary, being located completely within the modally homogeneous dunite layers, and hence can be identified only with the help of mineral compositional data. The Layered and Upper Border Series are composed of the lithological succession dunite, troctolite and olivine gabbro, which is strictly consistent with the expected trend of crystallization of the parental magma X': Ol Ol + Pl Ol + Pl + Cpx. The Basal and Top Zones consist of the succession olivine gabbro, troctolite and dunite, showing the trend of crystallization Ol + Pl + Cpx Ol + Pl Ol, which is the exact opposite of that of the parental magma X'. The Basal and Top Zones, being mirror images of the Layered and Upper Border Series sequences, respectively, are therefore referred to as basal and top reversals.
|
The whole scenario for the formation of the sill with the double-humped compositional profile is subdivided into eight principal stages T1–T8 (Fig. 10), which are presented below.
|
Stage T1: chilled margin remelting
The emplacement of the first batch of fluid-saturated phenocryst-free or phenocryst-poor magma X' into cold country rock leads to the immediate formation of a chilled margin corresponding compositionally to the initial parental magma. The initial chilled roof margin, however, can subsequently be partly or completely melted back when heat flux from the vigorously convecting magma in the chamber becomes equal to or exceeds the conductive heat flux in the country rock (Huppert & Sparks, 1989). Remelting of a chilled margin at the floor is considered to be less effective but can still occur by magma flow through the chamber in which heat is advected to both boundaries (Huppert & Sparks, 1989). A common lack of mass balance between the chilled floor margins and bulk compositions of sills is taken as an indication that their marginal facies usually form after prolonged and extensive meltback of the initial chilled floor (Latypov, 2003).
Stages T2–T4: basal and top reversal formation
The formation of basal and top reversals takes place through the non-equilibrium evolution of liquid boundary layers as a result of a temperature gradient imposed by the cold country rock (Latypov, 2003). The boundary layers tend toward a stationary non-equilibrium state that is attained when the compositions of melts in the boundary layers become adjusted to an imposed temperature gradient in accordance with liquidus phase equilibria. To achieve such a compositional adjustment the system transfers high melting point components (HMPC) from the liquid boundary layer into the main magma body and low melting point components (LMPC) in the opposite direction. Mass transfer across the short distance of a liquid boundary layer occurs mainly by Soret diffusion, which is aided by thermal convection in the main magma body. Soret fractionation under the initial highest thermal gradient produces a liquid boundary layer comprising, in a stationary state, compositions that span in reverse sequence a nearly entire liquid line of descent of the parental magma X'. Compositional oversaturation in the main magma body that arises as a result of a supply of HMPC from the liquid boundary layer is uniformly distributed throughout the magma chamber via convection. Nucleation and crystallization occurs predominantly in the stagnant lowermost regions of the boundary layers composed of the most evolved liquid compositions. Upon the progressive decrease in magnitude of the thermal gradient related to the growing cumulus pile, and an overall fall in the temperature of the whole magma body, the compositional characteristics of the liquid boundary layers in the stationary state become more and more primitive. Given favourable kinetic conditions, the stagnant crystallizing region of the liquid boundary layers in this process progressively travels all the way along the liquid line of descent of the parental magma.
According to the above predictions, the trend of crystallization of the basal and top Soret-produced liquids should be exactly the reverse of that of the parental magma X': Ol + Pl + Cpx (E1, olivine gabbro) Ol + Pl (E1–Y, troctolite) Ol (Y–X', dunite) (Figs 9 and 10). The crystallization of these liquids will result in clearly pronounced basal and top reversals—mirror images of the future Layered and Upper Border Series. The reversals are distinguished by: (1) an apparent lack of mass balance between the chilled margins composed of olivine gabbro (a eutectic point E1) and the initial parental magma composition (point X'); (2) the order of phase crystallization; (3) a trend of mineral compositions (whole-rock mg-number, Cr, Ni) opposite to those expected from the fractional crystallization of parental magma X'.
Opportunities for effective Soret fractionation in the liquid boundary layers remain only as long as a high thermal gradient persists locally during thermally juvenile magmatic stages (Lesher & Walker, 1991). Thus, when the thermal gradient in the boundary layers wanes, Soret fractionation will no longer be able to cause magmatic differentiation and its effect will be insignificant. The stratigraphic levels where the Soret effect ceases to be active are referred to as the lower and upper crossover maxima (Fig. 9a, points X'') forming from magma more primitive (but not hotter!) than the parental magma X'. This magma arises as a result of the progressive removal of low melting fractions from the initial composition X' while the basal and top reversals crystallize along the path E1 Y X' (Fig. 9b). This is why the crossover maxima represent the most primitive compositions of minerals and rocks observed in the entire sill. Obviously the mineral compositions (e.g. Fo % in olivine) that occur slightly above and below the level of the lower and upper crossover maxima can be reliably used for estimation of the parental magma composition.
It should be noted that the crystallization of magma in the marginal zones of intrusions generally occurs relatively rapidly, as indicated by the common fine- to medium-grained textures of most marginal rocks. Crystallization will therefore proceed without much release of intercumulus liquid from the crystallizing boundary layer into the main magma body so that the interface between solid and melt can be treated here as a nearly flat boundary. However, from the level of the crossover maxima, where rocks are already medium to coarse grained, the interface between solid and melt can no longer be considered flat. At this time the thermal boundary layers are likely to represent a crystal–liquid mush from which, under favourable conditions, interstitial liquids can escape into the main reservoir of the magma.
Stages T5–T8: Layered and Upper Border Series formation
Upon reaching the crossover maxima, the formation of the Basal and Top Zones in non-equilibrium conditions gives way to the formation of the Layered and Upper Border Series in equilibrium conditions. Thus the importance of the crossover maxima is that they mark the boundary where a non-equilibrium style of evolution of the magma in the chamber gives way to an equilibrium one. The future evolution of the magmatic system can be fully described in the framework of an in situ crystallization model developed in recent publications (e.g. Sparks et al., 1984; Tait et al., 1984; Tait & Jaupart, 1992, 1996; Jaupart & Tait, 1995). The model treats the magma chamber as a central mass of nearly crystal-free convecting magma, which gradually loses heat and crystallizes inwards from its margins. The transition from crystal-free magma in the central part of the chamber to completely solidified rock in the outer parts is thought to occur through a boundary layer of crystal–liquid mush, with the proportion of interstitial liquid decreasing systematically in the direction of falling temperature. The model presumes that the evolved intercumulus liquids within the boundary layer at the floor will be unstable as a result of enrichment in SiO2 and volatiles, which tend to reduce its density. In most circumstances, this compositional effect dominates the opposing thermal effect so that relatively primitive, denser liquid from the overlying reservoir infiltrates down through the crystalline matrix of the boundary layer, driving out the lighter, more evolved interstitial liquid. It is the transfer of evolved liquid into the overlying reservoir through density-induced compositional convection at the floor that is considered to be the main mechanism by which crystallization at the margins results in differentiation of the magma body as a whole. It is worth noting one apparent difference in the style of component redistribution before and after the crossover maxima. Before the crossover maxima (stages T2–T4), liquid differentiation occurs through the flux of LMPC from the interior of the magma body towards the chamber margins, accompanied by flux of HMPC in the opposite direction. After the crossover maxima have been reached, the direction of transfer of LMPC and HMPC is reversed (stages T5–T7).
During the formation of the Layered and Upper Border Series the magma in the chamber follows through a normal order of phase appearance Ol + L (X''–Y, dunite) Ol + Pl + L (Y–E1, troctolite) Ol + Pl + Cpx + L (E1–E3, olivine gabbro). This trend is fully compatible with the location of the parental liquid within the primary field of olivine (Fig. 9b). The modal compositions of the rocks are strictly constrained by the topology of the liquidus phase diagram. In this context it should be emphasized that an abrupt compositional transition from dunite (oCpa) to troctolite (poCa) is a consequence of the shift in magma composition from the primary volume of olivine (Ol + L) to the olivine–plagioclase cotectic (Ol + Pl + L). This cotectic is characterized by the predominance of plagioclase (70–85 wt %) over olivine (15–30 wt %). This is why, after the formation of dunite by exclusive precipitation of Ol within its primary phase volume, the olivine abundance in the overlying troctolite, crystallizing on the Ol + Pl + L cotectic, must drop abruptly to as little as 15 wt % (stage T6). It is therefore not reasonable to take this common phenomenon as evidence for the formation of the picritic and gabbroic zones of basic–ultrabasic sills from two distinct magmas (Smirnov, 1966; Ivanov et al., 1971; Marakushev et al., 1982; Ryabov, 1992; Czamanske et al., 1994, 1995). Upon further crystallization, the liquid will reach the eutectic point E1 where olivine gabbro is produced. Here the proportion of olivine in the phase assemblage falls to 5–10 wt % (stage T7). Further gradual decrease in the olivine content of olivine gabbro could take place as a result of the progressive contraction of the plagioclase field. The latter can result from increases in fluid pressure (Yoder, 1965) and in the albite content of plagioclase (Bowen, 1915) in the course of liquid fractionation along the path E1 E2 E3. The Layered and Upper Border Series, growing contemporaneously from the roof and floor and meeting at the Sandwich Horizon, will exhibit similar trends of phase crystallization. The thickness of the Upper Border Series, however, will be much smaller, as vigorous thermal convection in the main mass of magma will keep the upper thermal boundary layer thinner than the lower one. Thermal convection provides a good explanation for the fact that, in most well-described sills and intrusions, rock sequences that have crystallized from the roof are typically 6–7 times thicker than those that formed at the floor (Jaupart & Tait, 1995).
|
IMPLICATIONS OF THE PROPOSED IN SITU CRYSTALLIZATION MODEL FOR THE ORIGIN OF VARIOUS COMPOSITIONAL PROFILES |
|---|
|
TOPABSTRACTINTRODUCTIONREVIEW OF SILLS WITH...PHENOCRYSTS IN SILLS:...A PROPOSED IN SITU...IMPLICATIONS OF THE PROPOSED...PALISADES SILL: AN ORIGIN...CONCLUSIONSREFERENCES |
|---|
The validity of any model can primarily be judged from its ability to explain or/and reproduce actual observations. In this respect the proposed model adequately duplicates the shape of the compositional profile for complex, ‘double-humped’ sills (compare Fig. 10 with Fig. 2d–f). Significantly, the model can account for the three main features of double-humped sills, namely, the basal reversal, the upper reversal and the presence of a relatively sharp contact between the ultramafic and mafic parts of the Layered Series, by in situ crystallization of a single input of phenocryst-poor parental magma. The model also explains the commonly developed, puzzling, phenomenon of the evolved nature of chilled margins compared with the bulk composition of sills, which is not easily explained by other models of magma evolution. Taken together these observations provide strong support for the proposed model and allow it to be used to explain the origin of other types of compositional profile observed in basic–ultrabasic sills.
The shape of the double-humped modal profile (Fig. 10) is defined from the base upward by rock sequences that compose four principal units: basal reversal, Layered Series, Upper Border Series and top reversal. Such subdivisions can be useful in understanding the origin of other types of compositional profiles. Figure 11 summarizes the different modal profiles observed in sills that can be interpreted as representing combinations of the four units. The obvious question is: Why are some units missing from these sill profiles? Does this reflect initial differences in parental magma composition or in conditions of magma evolution in the chamber? Finally, the most intriguing question: is it possible to produce all these profiles from the same magma composition given different conditions of magma evolution in the chamber?