testing thevalidityof the petrological hypothesis … · testing thevalidityof the petrological...

Testing theValidity of the Petrological Hypothesis‘No Phenocrysts, No Post-emplacementDifferentiation’

RAIS LATYPOV*DEPARTMENT OF GEOSCIENCES, UNIVERSITY OF OULU, PO BOX 3000, FIN 90014, OULU, FINLAND

RECEIVED DECEMBER 12, 2008; ACCEPTED MAY 6, 2009ADVANCE ACCESS PUBLICATION JUNE 17, 2009

An extravagant hypothesis ‘no phenocrysts, no post-emplacement dif-

ferentiation’ has been put forward by Marsh, in a series of papers,

for the development of maficûltramafic intrusions.This hypothesis

is based on an assertion that the majority of these intrusions are

structureless and undifferentiated because they lack residual granitic

rocks.To explain this, the hypothesis postulates that phenocryst-free

magmas are not able to differentiate in crustal chambers because all

evolved interstitial liquid is locked in solidification fronts, producing

compositionally uniform magmatic bodies. Layered, well-differen-

tiated intrusions are attributed to the successive emplacement of

magma pulses with phenocrysts of different phase and chemical com-

positions, rather than to slow magma cooling and fractional crystalli-

zation, as conventional models imply. Such phenocryst-laden

magma pulses are supposed to be derived from an underlying mag-

matic mush column. However, structureless, undifferentiated,

maficûltramafic bodies simply do not exist in nature. All well-

studied maficûltramafic bodies, with or without residual granitic

rocks, that crystallized from parental magmas of non-eutectic compo-

sition, tend to reveal clear evidence of internal compositional differ-

entiation in terms of crystallization sequences (e.g. Ol, Opx,OpxþPl, OpxþPlþCpx), mineral compositions (e.g. An in

plag, En in cpx) and compatible/incompatible major and trace ele-

ment geochemistry (e.g. Mg-number, Cr, rare earth elements).This

is especially evident in layered intrusions that represent the key evi-

dence against the hypothesis. Essentially, by denying the ability of

magma to differentiate in intrusive bodies, the hypothesis forbids

magmatic differentiation in any sub-chamber related to the entire

magmatic mush column. As a result, magma pulses in which pheno-

crysts progressively change in composition cannot be derived from the

column to form layered intrusions.The hypothesis is thus contradic-

tory, baseless and fundamentally flawed. It should be abandoned in

favour of a classical fractional crystallization model based on the pio-

neering experiments of Bowen and amply confirmed over almost a

century by subsequent studies of layered intrusions. The classical

model was, is, and will most probably remain, the best explanation

for the origin of differentiated magmatic bodies.

KEY WORDS: layered intrusions; mafic sills; liquidus phase equilibria;

crystallization sequences; compositional trends

I NTRODUCTIONThe key aspects of the petrological hypothesis ‘no pheno-

crysts, no post-emplacement differentiation’ are summar-ized in several publications (Marsh, 1988, 1989, 1990, 1991,1996, 2000, 2004, 2006; Zieg & Marsh, 2005). Its popularityamong some igneous petrologists (e.g. Gibb & Henderson,1992, 2006; Miller & Miller, 2002; Simura & Ozawa,2006; White, 2007; Aarnes et al., 2008; Galerne, 2009)raises concern, as it is in sharp contradiction with muchof what we know about the mechanisms of magma differ-entiation operating in differentiated sills and layered intru-sions. It should be noted that this is not the first attemptto raise objections to this hypothesis. In particular, Sparks(1990), Huppert & Turner (1991), and Worster et al. (1990)have pointed out the misuse of fluid dynamics in controver-sial interpretations of crystallizing convective systemsbased on this hypothesis. In particular, the hypothesisstates that convection is almost absent in crystallizingmagma chambers (Marsh, 1988, 1989), an assertion that isin sharp contradiction with both a theory of fluid

*Corresponding author. Telephone: þ358 8 553 1452. Fax: þ358 8 5531484. E-mail: [email protected]

� The Author 2009. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 50 NUMBER 6 PAGES1047^1069 2009 doi:10.1093/petrology/egp031

dynamics (Huppert & Turner,1991;Worster et al.,1992) andnatural observations (Sparks, 1990). Despite these criti-cisms the hypothesis is actively promoted and has beenrecently employed to reinterpret the origin of classicalobjects of igneous petrologyçmafic^ultramafic layeredintrusions (Marsh, 1996, 2006). The significance oflayered intrusions to igneous petrology cannot be over-emphasized: they represent ‘fossilized’ natural labora-tories that constrain many fundamental principles ofigneous petrology. Traditionally, these intrusions areviewed as products of slow cooling and fractional crys-tallization of basaltic magmas in crustal chambers (e.g.Wager & Brown, 1968; Parsons, 1986; Cawthorn, 1996)(Fig. 1a). In contrast, a new hypothesis considerslayered intrusions simply as accidental mechanicalaccumulations of ‘tramp phenocrysts’ entrained from anunderlying mush column by numerous phenocryst-laden

magma pulses (Marsh, 1996, 2006; Zieg & Marsh, 2005)(Fig. 1b). The new hypothesis thus challenges and in factdenies the fertile theory of cumulate igneous rocks begunby such petrological luminaries as Grout, Hess, Wager,Deer, Brown, and Wadsworth. This is done without men-tioning such sources, but instead merely by promoting analternative paradigm that makes no sense to students oflayered intrusions. Because the potential to resolve the con-flict of views with the help of numerical modelling andsimple experimental studies appears to have beenexhausted in the earlier discussions (Marsh, 1990, 1991;Sparks, 1990; Huppert & Turner, 1991), the best remainingway is to return to the original information recorded inthe rocks themselves. They are the final ‘Court of Appeal’.In accordance with this strategy, the validity of thehypothesis ‘no phenocrysts, no post-emplacement differentiation’and its underlying postulates will be tested against

Fig. 1. Schematic illustration of two contrasting models for the origin of layered mafic^ultramafic intrusions: (a) the classical model thatimplies ‘single pulse’ magma emplacement followed by fractional crystallization with formation of a single-cyclic layered intrusion; by ‘singlepulse’, is meant that the intrusion crystallizes from a single large body of homogeneous magma produced by complete mixing of numeroussmall magma pulses in the chamber prior to onset of crystallization; (b) the ‘no phenocrysts, no post-emplacement differentiation’ model that impliesmultiple emplacement of phenocryst-laden magma pulses (one pulse per layer) forming a single-cyclic layered intrusion. Modified afterLatypov & Chistyakova (2009). Hereafter in figures and text: Pl, plagioclase; Ol, olivine; Fo, forsteritic olivine; Fa, fayalitic olivine; Opx, ortho-pyroxene; Pig, pigeonite; Cpx, clinopyroxene; Qtz, quartz; Ne, nepheline; Kfs, alkali feldspar; Mag, magnetite; Ilm, ilmenite; Ap, apatite;Chr, chromite; Am, amphibole; Fe-Bust, ferrobustamite; Bt, biotite; Zrn, zircon.

JOURNAL OF PETROLOGY VOLUME 50 NUMBER 6 JUNE 2009

1048

observational data from maficûltramafic sills and layeredintrusions.

A SHORT INTRODUCTION TOTHE HYPOTHESI S AND ITSUNDERLY ING POSTULATESThe hypothesis in question has been defined in threesomewhat different, but mutually complementary, ways:

(1) ‘The central or eruptible melts of sill-like bodies formed of

magma initially free of phenocrysts undergo little, if any, differ-

entiation.This is the Null Hypothesis.’ (Marsh, 1996, p. 12).(2) ‘In contrast to the convection^fractional crystallization model,

the Null Hypothesis states that any body of magma emplaced

free of crystals will crystallize into a compositionally and

mineralogically uniform layer of rock.’ (Zieg & Marsh, 2005,

p. 1445).(3) ‘Instantaneous injection of crystal-free magma does not form

exotically layered, well-differentiated intrusions.The Sudbury

testimony is clear: no phenocrysts, no layering. This is the

Null Hypothesis.’ (Marsh, 2006, p. 290).

In this paper I use the shortest wording of the hypothesisthat most precisely summarizes its essence: ‘no phenocrysts,

no post-emplacement differentiation’ (Marsh, 1996, p. 13).Although not formulated in explicit form, the hypothesisconsists of three major elements that are referred below toas postulates.Postulate 1:The world abounds with featureless, undifferentiated

maficûltramafic sills and large intrusions (Marsh, 1988,p. 1721; Mangan & Marsh, 1992, pp. 605^606; Marsh,2006, p. 289). Intrusions are considered to be undifferen-tiated if they lack residual rocks of granitic composition.This means that any maficûltramafic body without grani-tic rocks, irrespective of the presence of phase, modal andcryptic layering, is classified as undifferentiated. Becauseresidual granites are rare in maficûltramafic intrusions,most of them, according to this specific definition, fall intothe group of undifferentiated bodies. In an attempt toaccount for this phenomenon, it became necessary to putforward the purely physical interpretation referred to aspostulate 2.Postulate 2: Phenocryst-free basaltic magmas are not able to dif-

ferentiate in crustal chambers (Mangan & Marsh, 1992, p. 618;Marsh, 1996, pp. 11^13; Zieg & Marsh, 2005, p. 1445). Thisis because rapidly growing solidification fronts capture allthe interstitial liquid so that it cannot mix with the mainmagma body. The inability of magma to differentiate isfurther substantiated by the belief that all crystals settlingfrom the roof are completely dissolved in the hottermagma and that convection is non-existent in crystallizingmagmatic chambers; convection ceases as soon as magmaloses its initial superheat and reaches liquidus temperature.As a result, crystallization takes place from magma that is

essentially stagnant throughout the entire chamber. Thismeans that phenocryst-free magma emplaced into a crus-tal chamber of any size will crystallize into a solid bodywith an essentially homogeneous composition from top tobottom. To explain the existence of layered intrusions thatshow clear evidence of magmatic differentiation, thehypothesis introduced another simple physical explanationreferred to as postulate 3.Postulate 3: Layered intrusions form by successive emplacement of

phenocryst-laden magma pulses (Marsh, 1996, p. 6; Marsh,2004, p. 499; Zieg & Marsh, 2005, p. 1446; Marsh, 2006,p. 289). In practice, this means that each stratigraphiczone of a layered intrusion that is distinguished by a spe-cific assemblage of cumulus phases (e.g. Ol, Opx, OpxþPl,OpxþPlþCpx) must be produced from at least one sepa-rate magma pulse carrying this particular assemblage ofphenocrysts. As cumulus minerals in such zones commonlybecome progressively more evolved stratigraphicallyupwards (e.g. olivine Fo80^Fo50), a single stratigraphic zonemust actually be produced from a series of magma pulsescarrying phenocrysts of systematically changing composi-tion with time. It was suggested that such phenocryst-laden magma pulses are derived from magmatic mushcolumns underlying layered intrusions. How such magmapulses with phenocrysts of progressively changing phase,modal, and chemical composition can be produced inthese columns has not been explained.

TEST ING THE VAL IDITY OF THEPOSTULATES UNDERLY ING THEHYPOTHESI SPostulate 1: The world aboundswith featureless, undifferentiatedmaficûltramafic sills and large intrusionsIt should be emphasized that the entire hypothesis isdriven by this key observation expressed in the followingways:

‘The world abounds in basaltic sills up to 350 m thick with no

structure; essentially uniform from top to bottom’ (B. D.Marsh, personal communication, 2008).‘The field evidence, at least for thin to moderately sized bodies

(5500 m thick) would suggest not . . . .These bodies are undif-

ferentiated, showing no cumulates and no systematic significant

vertical chemical zonation’ (Mangan & Marsh, 1992, pp.605^607).‘Thick diabase sills and gabbroic bodies like Skaergaard have

made precious little progress along the liquid line of descent

toward eruptible rhyolite . . .Even very large mafic, ultramafic

and more hydrous bodies do not show much tendency to make

silicic liquids. Why is it that the clearest examples of magma

chambers, ones we can touch, show little differentiation . . . ?’

(Marsh, 1988, p. 1721).

LATYPOV MAGMA DIFFERENTIATION MECHANISM

1049

As this postulate is the backbone of the entire hypoth-esis, it will be subjected to particularly detailed examina-tion. There appear to be three principal misapprehensionsthat gave rise to this postulate.

Misinterpretation of the concept of magmatic differentiation

It is important to realize that much confusion arises fromthe fact that the hypothesis implicitly introduced, and con-sistently follows, a unique definition of magmatic differen-tiation. Traditionally, magmatic differentiation isconsidered to be the process by which diverse rock typesare generated from a single parental magma (e.g. Hess,1989; Hall, 1996; McBirney, 2007). In contrast, accordingto the hypothesis, magmatic differentiation is principallyconcerned with the ability of basaltic magmas to produceeruptible rhyolite. This idea is not explicitly formulatedbut clearly follows from the general concern about a greatnumber of mafic^ultramafic intrusions revealing no resid-ual granites and that are therefore referred to as showinglittle, if any, chemical differentiation (see quotes aboveand below). Following this logic, a magmatic body is inter-preted as differentiated only if it produces significantamounts of residual granitic liquid: if this is not the case,the body is not differentiated. It is therefore no wonderthat most known differentiated mafic sills lacking graniticrocks are classified as undifferentiated. Even such excep-tionally well-differentiated bodies as the Skaergaard intru-sion and the Basistoppen sill in East Greenland areinterpreted as showing no internal differentiation:‘Skaergaard does show a phenomenal degree of crystal sorting or frac-

tionation, but in terms of differentiating to produce significant

amounts of siliceous liquid, it is an exceptionally poor example . . . .

What about the extent of differentiation in sills? In terms of progres-

sion toward eruptible proportions of silicic magma, it is exceedingly

limited . . .This is also apparently true for the Basistoppen sill of

the Skaergaard which aside from granophyres shows very little,

if any, systematic chemical differentiation with solidification’(Marsh, 1990, p. 849). Such specific characterization ofmagmatic differentiation is deeply rooted in the erroneousbelief that all basaltic magmas must necessarily evolvetowards residual granitic liquids: ‘that basaltic magmas eventu-ally give rise to granitic magma through separation of crystals and

melts is well established from a purely chemical perspective’(Marsh, 2002, p. 2211).Whether a parental magma will evolve towards a resid-

ual granitic liquid or not is primarily dependent on its ini-tial composition. There are several cases where parentalbasaltic magmas cannot generate residual granitic liquid,regardless of the extent of their differentiation. This ischaracteristic of parental magmas that lie close to theplane of silica undersaturation Ol^Cpx^Pl that representsa thermal barrier between nepheline-bearing and ortho-pyroxene-bearing systems (Yoder & Tilley, 1962) (Fig. 2).One of the most obvious cases is when the parental

magma is a nepheline-normative olivine basalt.Its evolution towards the granitic eutectic E1(QtzþPlþOpxþCpx¼L) is impossible because uponcrystallization it will move away from the plane of silicaundersaturation Ol^Cpx^Pl towards the nepheline olivinegabbro eutectic E2 (NeþPlþOlþCpx¼L) (Fig. 2). As anexample, one can mention �20 mafic sills with traceamounts of interstitial nepheline [e.g. Shiant Isles MainSill belonging to the Little Minch Sill Complex of north-ern Skye (Gibb & Gibson, 1989; Gibb & Henderson, 2006;Latypov & Chistyakova, 2009)]. Another case is whenparental magmas of olivine basalt composition are locateddirectly in the plane of silica undersaturation Ol^Cpx^Pl.(Note that for multi-component basaltic melts this is avolume, rather than a plane.) Crystallization of suchmagmas in the chambers does not bring them towards sat-uration either in nepheline or orthopyroxene, let alonequartz.The reason for this is very simple. Because the com-ponents of these minerals were absent in the parentalmagma from the very beginning, these minerals cannotappear in the residual melts however efficient fractionalcrystallization may have been in the chamber. In otherwords, all basaltic magmas that compositionally falldirectly into this plane (volume) will remain there up tothe end of their crystallization. One of the best examplesof intrusions produced from such parental magmas is thesuite of �300 gabbro^wehrlite intrusions in the Pechengaarea, Kola Peninsula, Russia (Gorbunov et al., 1985;Hanski, 1992; Latypov et al., 2001) that are distinguishedby the absence of both nepheline and orthopyroxene. Yetanother example is when the parental magma is repre-sented by an olivine basalt with negligible amounts of nor-mative orthopyroxene (51^2wt %). Under mostfavourable conditions, such magmas upon crystallizationof their trapped interstitial liquid may probably reach theolivine gabbronorite peritectic P1 (OpxþCpxþPl¼LþOl) giving rise to rocks with trace amounts ofinterstitial orthopyroxene. A good example of this case isthe �30 Noril’sk-type differentiated mafic/ultramaficintrusions of northern Siberia, Russia (Czamanske et al.,1995; Krivolutskaya et al., 2001; Latypov, 2002, 2007).To sum up, there are numerous examples of intrusivebodies formed from magmas of silica-undersaturated, oli-vine basalt composition. Such intrusions are commonlydistinguished by trace amounts of interstitial orthopyrox-ene (e.g. Noril’sk-type intrusions) or nepheline (e.g. manyof the Skye intrusions) or by the absence of both of theseminerals (e.g. the Pechenga intrusions). Despite theabsence of residual granitic rocks, these intrusions are allmarkedly differentiated, as indicated by their crystalliza-tion sequences (Fig. 3).There are also several instances of basaltic magmas that

fail to produce granitic rocks, despite the absence of phaseequilibria obstacles. The simplest case is when magmas


1050

need to experience almost perfect fractional crystallizationto reach the granitic eutectic. However, perfect fractionalcrystallization is practically never achieved in nature.This is indicated by the common occurrence of ortho-

and mesocumulates in mafic^ultramafic intrusions, imply-ing that interstitial liquid (up to 25^40 vol. %) remainsburied in the cumulate pile (Wager & Brown, 1968).In practice, this means that some basaltic magmas that

Fig. 2. Coupled phase diagrams Ol ^ (PlþCpx) ^Qtz and Ol ^ (PlþCpx) ^Ne showing the spatial confinement of compositional fields ofNoril’sk-type intrusions, Pechenga intrusions and Skye mafic sills to the critical plane of silica undersaturation Ol^Pl^Cpx. This location indi-cates that their parental magmas correspond to silica-undersaturated olivine basalts (Yoder & Tilley, 1962). The evolution of the basalticmagma compositions parental to the Skye mafic sills and Pechenga intrusions towards the granitic eutectic (E1) is forbidden by phase equilibria,resulting in the absence of residual granitic rocks in these intrusions (Fig. 3). The evolution of basaltic magma compositions parental toNoril’sk-type intrusions towards the granitic eutectic (E1) is theoretically possible, but cannot be practically realized because of imperfect frac-tional crystallization.Their parental magmas reach only the olivine gabbronorite peritectic P1 (OpxþCpxþPl¼LþOl) during the crystalliza-tion of trapped liquid, giving rise to rocks with trace amounts of interstitial orthopyroxene. Phase diagrams are from Dubrovskii (1998). E1,QtzþOpxþCpxþPl¼L; E2, OlþCpxþPlþNe¼L; P1, OpxþCpxþPl¼LþOl. Cotectic and reaction lines are indicated by one and twoarrows, respectively. Arrows point toward decreasing temperature. Hereafter in figures and text: E, eutectic point; P, peritectic point; L, melt.Data for compositional fields from the following sources: Noril’sk-type intrusions, Latypov (2002, 2007); Pechenga intrusions, Latypov et al.(2001); Skye sills, Gibson & Jones (1991) and Gibb & Henderson (2006).


1051

Fig. 3. Stratigraphic sections and cumulate stratigraphy through three mafic^ultramafic sills: Talnakh sill, Siberia, Russia; Pilguja« rvi sill,Fennoscandian Shield, Russia; and Shiant Isles Main sill, NW Scotland. The sills show no residual granitic rocks because they crystallizedfrom parental magmas of silica-undersaturated, olivine basalt composition (Yoder & Tilley, 1962) whose evolution towards a granitic composi-tion is either forbidden by phase equilibria (Shiant Isles Main sill and Pilguja« rvi sill) or unlikely because of imperfect fractional crystallization(Talnakh sill) (Fig. 2). The fact that granites are absent from these intrusions cannot, therefore, be used to argue that the intrusions are ‘struc-tureless and undifferentiated’, as implied by the hypothesis ‘no phenocrysts, no post-emplacement differentiation’. All these intrusions are well differen-tiated, as clearly indicated by their crystallization sequences. The thick continuous and thin discontinuous lines indicate cumulus andintercumulus minerals, respectively. Data from the following sources: Talnakh sill, Krivolutskaya et al. (2001); Pilguja« rvi sill, Latypov et al.(2001); Shiant Isles Main sill, Gibb & Henderson (2006) and Latypov & Chistyakova (2009).


1052

could theoretically have produced, say, 5^10 vol. % ofgranitic rocks, fail to do so because most of the residualgranitic liquid is dispersed throughout the cumulate pile.As an illustration, the parental magmas to many layeredintrusions are characterized by a crystallization trend ofOl (dunite), Opx (orthopyroxenite), OpxþPl (norite) orOpxþCpx (websterite), OpxþPlþCpx (gabbronorite),PigþPlþCpx (pigeonite gabbro), OpxþPlþCpxþQtz

(quartz gabbro), OpxþPlþCpxþQtzþKfs (granite).However, imperfect fractional crystallization results inrock sequences in these intrusions commonly being com-pleted by gabbronorite or pigeonite gabbro. Numerousmafic^ultramafic sills and layered intrusions worldwidebelong to this group: among them are the Monchegorsk,Penikat, Portimo, Kemi, Burakovka layered intrusions(Alapieti & Lahtinen, 2002), Mound Sholl, Munni Muni,Radio Hill and Andover layered intrusions (Hoatson &Sun, 2002), Great Dyke, Bushveld and StillwaterComplexes (Cawthorn, 1996), and many others allaround the world. It should be stressed that, despitethe lack of mappable residual granites, all theseintrusions are well differentiated, as indicated by theircrystallization sequences. One typical example of such anintrusion, the Karikjavr layered intrusion, Russia, is illu-strated in Fig. 4. Its differentiated nature is evident fromthe stratigraphy, comprising, from bottom to top, Ol cumu-lates (dunite), Opx cumulates (orthopyroxenite), OpxþCpx

cumulates (websterite), and OpxþPlþCpx cumulates

(gabbronorite).On the other hand, there are many examples of mafic^

ultramafic sills and layered intrusions that have residualgranitic or quartz syenitic rocks and whose existence isignored by the hypothesis. They include, among others,the Akanvaara, Koitelainen, and Keivitsa layered intru-sions (Mutanen, 1997), the Bjerkreim^Sokndal andFongen^Hyllingen layered intrusions (Wilson & S�rensen,1996; Wilson et al., 1996), and Muskox layered intrusion(Irvine & Smith, 1967). The ultramafic and mafic zonesof these intrusions are culminated by several hundredmetre thick granophyre or quartz syenitic units thatrepresent the products of late-stage crystallization ofresidual liquids. The tholeiitic parental magmas of theseintrusions have apparently reached the granitic/syeniticeutectic. Based on the specific definition of magmaticdifferentiation (Marsh, 1988, 1996, 2006), these bodiesshould be classified as well differentiated. As an illustra-tion, a section through the uppermost megacyclic unit ofthe Bjerkreim^Sokndal layered intrusion, Norway, is illu-strated in Fig. 5. This43 km thick megacyclic unit displaysa continuous stratigraphic transition from mafic rocks(troctolites, norites and gabbronorites) to felsic residualrocks (mangerite, quartz mangerite and charnockite).There is no evidence for mineral compositional breaksbetween mafic and felsic rock sequences, indicating

closed-system fractionation of basic (jotunitic) magmafrom the troctolitic cotectic with only three liquidusphases to a granitic eutectic with up to 10 liquidusphases (Wilson & Overgaard, 2005). The extensivefractional crystallization was accompanied by someassimilation (Tegner et al., 2005), but such intrusions pro-vide irrefutable evidence that basic magmas commonlyfractionate towards the granitic eutectic when parentalmagma composition and cooling conditions areappropriate.

Fig. 4. Stratigraphic section and cumulate stratigraphy through theKarikjavr layered intrusion, Fennoscandian Shield, Russia.The intru-sion shows no residual granitic rocks because fractional crystallizationof the basaltic parental magma was not effective enough to reach thegranitic eutectic. It should be emphasized, however, that despite thelack of residual granitic rocks, the intrusion is markedly differentiatedas indicated by its crystallization sequence. The thick continuouslines indicate cumulus minerals. Data from Bakushkin (1983).


1053

To sum up, the evolution of many basaltic magmastowards a granitic composition is forbidden by phase equi-libria (e.g. Skye sills and Pechenga intrusions, Fig. 3).Many basaltic magmas can evolve towards granitic com-positions only with near-perfect fractional crystallization.But because such conditions are rarely achieved in nature,such magmas do not commonly produce residual graniticrocks (e.g. Karikjavr layered intrusion, Russia, Fig. 4).Some magmas with sufficient initial silica do produceresidual rocks of granitic composition (e.g. Bjerkreim^Sokndal layered intrusion, Norway, Fig. 5). The majorimplication is that it is misleading to use the absence orpresence of granitic rocks in mafic^ultramafic intrusions

as an indicator of the extent of their magmaticdifferentiation.

Misapplication of indices of magma differentiation

Since the pioneering experiments of Norman Bowen, it hasbeen firmly established that the best index of magmaticdifferentiation of basaltic liquids is the successive appear-ance of liquidus phases upon crystallization (e.g. Ol, Opx,OpxþPl, OpxþPlþCpx). Another important index ofmagmatic differentiation is the progressive change incomposition of these solid solution minerals towards low-temperature end-members (e.g. olivine Fo80^Fo20; plagio-clase An80^An20). Yet another very sensitive index of

Fig. 5. Stratigraphic section and cumulate stratigraphy through the uppermost megacyclic unit IV (MCU IV) of the Bjerkreim^Sokndallayered intrusion, Fennoscandian Shield, Norway. The megacyclic unit displays a remarkably continuous transition from primitive rocks (troc-tolites, norites and gabbronorites) to evolved residual rocks (mangerite, quartz mangerite and charnockite). There is no evidence of mineralcompositional breaks in MCU IV rock sequences, indicating closed-system fractional crystallization of basic magma. The existence of suchintrusions represents irrefutable evidence that basic magmas may freely fractionate towards the granitic eutectic with an appropriate parentalmagma composition and suitable cooling conditions. The thick continuous and discontinuous lines indicate the cumulus (liquidus) and intercu-mulus minerals, respectively. Data fromWilson et al. (1996).


1054

magmatic differentiation is the behaviour of highly com-patible (e.g. Cr, Ni) and incompatible [e.g. rare earth ele-ments (REE), Zr, Y] trace elements. However, none ofthese conventional indices of magmatic differentiation areused to assess the internal differentiation of mafic sills inpapers supporting the hypothesis. This is particularly evi-dent in the study of the Peneplain Sill, Antarctica (Marsh& Wheelock, 1994; Marsh, 1996), a matter that has alreadybeen covered by Latypov (2003a) and Latypov et al.(2007). In short, the ‘undifferentiated’ nature of this maficsill has been illustrated by stratigraphic variations of threemajor oxides, SiO2, MgO and CaO (Fig. 6), which arevery poor indices of magmatic differentiation in magmasof near-eutectic composition (see below). There is littledoubt that the sill will reveal clear signs of cryptic layeringif more diagnostic indices of magmatic differentiation areused instead of the reported oxide variations. This confi-dent assertion derives from the observation that, evenfrom oxide variations, the Peneplain Sill has a very well-developed marginal reversal that exhibits a clear progres-sion from about 4wt % MgO in the chilled margintowards 7 wt % MgO at a height of about 100m (Fig. 6).This maximum is followed by a slight, but steady, decreasein MgO throughout the central portion of the sill to5 wt % at a height of about 300m. These data are fullyconsistent with the observations of Gunn (1962), whofound an increase in normative An from 61 to 67% in themarginal reversal, and a clear decrease from 67 to 57% inthe central portion of the Peneplain Sill. Unfortunately,despite the repeated use of this sill in support of thehypothesis (e.g. Marsh & Wheelock, 1994; Marsh, 1996,2000; Hort et al., 1999), the whole-rock geochemical datafor this body are still not published, and readers cannotform their own opinion about the compositional profilethrough the Peneplain Sill.To sum up, reliance on inappropriate indices of mag-

matic differentiation, together with disregard of geochemi-cal data, do not provide sufficient evidence for the‘undifferentiated’ nature of mafic sills, such as thePeneplain Sill, Antarctica. Inability to recognize composi-tional zoning in such sills is most probably a consequenceof the inappropriate choice of parameters to illustrateliquid differentiation rather than the absence of differentia-tion itself.

Misuse of evidence to illustrate the widespread occurrenceof undifferentiated intrusions

It is curious to note that the whole idea about the wide-spread existence of undifferentiated maficûltramaficintrusions essentially stems from the study of a single spe-cific bodyçthe Peneplain mafic sill, Antarctica. Closeexamination of the modal and chemical composition ofthis sill (Fig. 6) and other mafic sills (Latypov, 2003b,p. 1627) clearly indicates, however, that they crystallizedfrom parental basaltic magmas of near-eutectic

(cotectic) composition [e.g. Opx(Pig)þPlþCpxþL orOlþPlþCpxþL]. This gives an immediate answer as towhy these mafic sills show only slight modal and major ele-ment variations: their parental magmas simply have verylittle ability to differentiate further as they already repre-sent near-eutectic residual liquids of mantle-derivedmagmas. The near-eutectic compositions of parentalmagmas of mafic sills do not allow them to produce anextended range of compositions along a liquid line of des-cent (apart from variations related to appearance of suchminor phases as apatite or ilmenite). In principle, theentire sections of mafic sills formed from such near-eutecticparental magmas must consist of cotectic plôpx^cpx(gabbronorite) or pl^cpxôl (olivine gabbro) rocks, closelycorresponding to the parental magma composition. It istherefore no wonder that magmatic bodies up to hundredsof meters thick may show little evidence of internal differ-entiation in terms of mineral assemblages (but still showvariations in mineral compositions and trace element con-centrations). All the examples of mafic sills that are citedby Mangan & Marsh (1992, pp. 606^607) as ‘undifferen-tiated, showing no cumulates and no systematic significantvertical chemical zonation’ essentially crystallized fromnear-eutectic parental magmas. To use magmatic bodiesformed from such magmas as a basis for claiming the uni-versal existence of undifferentiated intrusions is thereforemisleading.Special consideration must be given to the Sudbury

Intrusive Complex, as it has been recently cited as ‘dramaticconfirmation of the null hypothesis, namely, that magma emplaced

instantly and free of crystals crystallizes into a nearly homogeneous

sheet’ (Zieg & Marsh, 2005, p. 1446). Sudbury is an intru-sive body that was produced in a matter of minutes by ameteorite impact. Because the initial melt was heated toat least 17008C, it was initially free of phenocrysts (Zieg& Marsh, 2005). For this reason this body is exceptionallysuitable for testing the hypothesis ‘no phenocrysts, no

post-emplacement differentiation’. The hypothesis predictsthat the complex must be featureless and undifferentiated.The available data suggest, however, otherwise (Fig. 7).From bottom upwards, Sudbury is composed of five majorunits: sublayer, mafic norite, felsic norite, quartz gabbro,and granophyre, indicating a remarkable progressive frac-tionation of a parental magma along a crystallizationtrend: Opx þ Chr (� Ol), OpxþPl, PlþCpxþAm, Pl

þCpxþAmþMag, PlþCpxþAmþMagþAp, PlþCpx

þAmþMagþApþIlm, PlþCpxþAmþMagþApþ Ilm

þKfsþQtz. The successive appearance of cumulus phasesis marked by significant changes in the major, minor andtrace element composition of the bulk rocks. For instance,the appearance of cumulus plagioclase is marked by adecrease in MgO (Fig. 7); cumulus clinopyroxene by anincrease in CaO and Sc; cumulus magnetite by an increasein Fe2O3; cumulus apatite by an increase in P2O5; cumulus


1055

Fig. 6. Generalized stratigraphic section (a) through the Peneplain Sill (Antarctica) showing variations in whole-rock SiO2, MgO, and CaOand cumulate stratigraphy with height. It has been interpreted as an undifferentiated intrusion by Marsh (1996).The intrusion comprises a mar-ginal reversal and a layered series. The line of demarcation between these units runs through the crossover maximum exhibiting the highestMgO content in the entire section through the sill. The figure highlights two major points. First, the ‘undifferentiated’ nature of the sill is illu-strated with inappropriate indices of magmatic differentiation. Instead of oxide variations, variables such as whole-rock Mg-number,An(norm), highly incompatible element concentrations, or mineral compositions should be used to illustrate compositional variations. Second,the sill reveals no phase and modal layering because it crystallized from a parental magma with gabbronoritic composition that represents anear-eutectic residual liquid of a mantle-derived magma (b). The parental magma composition is deduced from petrography because the orig-inal geochemical data for the Peneplain Sill remain unpublished. (See text for further discussion.) Modified after Marsh (1996).


1056

ilmenite by an increase inTiO2; cumulus alkali feldspar byincreases in K2O and Rb; cumulus quartz by an increasein SiO2 (Lightfoot et al., 2001; Therriault et al., 2002; Zieg& Marsh, 2005). The disappearance of cumulus orthopyr-oxene in the quartz gabbro is related to a peritectic reac-tion with primary igneous amphibole (Therriault et al.,2002).This crystallization sequence is accentuated by nota-ble systematic changes in all indices of magmatic differen-tiation. In particular, the mafic part of the complex(mafic norite, felsic norite and quartz gabbro) reveals asignificant upward decrease in whole-rock MgO and Mg-number from about 16 to 3wt % and from 70 to 30, respec-tively (Fig. 7; Lightfoot et al., 2001), as well as a decrease inthe anorthite content of plagioclase from about An70 toAn35 (Therriault et al., 2002) and in Mg-number of pyrox-enes from about 80 to 50 (Naldrett et al., 1970). In addition,a basal sublayer reveals a well-developed reverse

fractionation trend in terms of all the above-mentionedparameters (e.g. Naldrett et al., 1970; Pattison, 1979) andcan therefore be regarded as a marginal reversal, a typicalfeature of most mafic^ultramafic intrusions (Latypov,2003a).The regular geochemical and mineralogical variations

from the felsic norite through the quartz gabbro to grano-phyre have been interpreted as strong evidence for thecrystallization of Sudbury from a single melt system(Therriault et al., 2002). A systematic increase in abun-dance of REE and the substantial uniformity in REE pat-terns between the Sudbury lithologies (Therriault et al.,2002), remarkably homogeneous incompatible trace ele-ment ratios in the entire section (e.g. Ce/Yb, Th/Nd;Lightfoot et al., 2001), the oxygen isotope compositions(Ding & Schwarcz, 1984), sulphur isotope ratios (Thodeet al., 1962), homogeneity in Os, Pb, Sr and Nd isotope

Fig. 7. Stratigraphic section, compositional variations and cumulate stratigraphy through the Sudbury Igneous Complex illustrating its well-differentiated nature in terms of crystallization sequence and geochemical indices of magma differentiation (MgO and Mg-number). TheSudbury Intrusive Complex provides the ultimate test for the hypothesis ‘no phenocrysts, no post-emplacement differentiation’ because it crystallizedfrom a highly superheated and therefore phenocryst-free parental magma, but nonetheless shows remarkable evidence of magmatic differentia-tion. The existence of the Sudbury Intrusive Complex provides the strongest refutation of the hypothesis: it demonstrates the ability of a pheno-cryst-free magma to freely differentiate in a crustal magma chamber forming a well-differentiated intrusion showing pronounced phase,modal and cryptic layering. (See text for further discussion.) Modified after Lightfoot et al. (2001). The cumulate stratigraphy is based on petro-graphic and rock chemical data from Lightfoot et al. (2001), Therriault et al. (2002) and Zieg & Marsh (2005).


1057

compositions (Faggart et al., 1985; Dickin et al., 1999), andcompositional variations in apatite (Warner et al., 1998)are also strongly indicative of a single melt system. Theevidence therefore seems very clear: the SudburyIntrusive Complex is a well-differentiated mafic bodyproduced by fractional crystallization of a single meltsystem (Fig. 7).To prove the opposite, Zieg & Marsh (2005) undertook

the following steps:

(1) the mafic norites (along with the sublayer) that arecharacterized by the most primitive compositionswere excluded from the section without anyjustification;

(2) the crystallization sequence showing the successiveappearance in the stratigraphy of eight liquidusphases Pl, Cpx, Am, Mag, Ap, Ilm, Kfs and Qtz wasignored;

(3) systematic changes in geochemical indices of magmadifferentiation were either left without comment (e.g.whole-rock MgO, REE) or these indices were notused at all (e.g. whole-rock Mg-number, normativeAn%);

(4) systematic changes in mineral compositions (e.g. Anin plag, Mg-number in pyroxenes) across the maficpart of the complex (based on previous studies) werenot included in the petrogenetic consideration;

(5) without providing any mineralogical, geochemical,and petrological evidence, the complex was dividedinto felsic norite and granophyre layers that werestated to have formed from two separate magmascrystallizing inwards from the margins, with quartzgabbro representing a rock formed at the level wherethe two solidification fronts finally met.

Attention was then focused on two layers of felsic noriteand granophyre, which, when viewed separately, show lim-ited compositional variations in terms of some oxides (e.g.SiO2, Na2O, TiO2, P2O5). The conclusion was reachedthat these two layers are essentially uniform in com-position and that the complex itself is therefore not dif-ferentiated. This was taken as strong evidence thatphenocryst-free basaltic and granitic magmas are not ableto differentiate in crustal chambers. It is hardly possible toagree with such a reinterpretation of the complex, as evena passing look at MgO and Mg-number variations acrossthe mafic part of the complex (Fig. 7) shows that it is farfrom being ‘essentially uniform in composition’ (Marsh,2000, p. 197). In general, this is an instructive example ofhow a world-famous, well-differentiated intrusion can betransformed into an ‘undifferentiated’ body by manipula-tion of the available data. Using such methods any layereddifferentiated intrusion can be reinterpreted as ‘featurelessand undifferentiated’. Such artificially created ‘undifferen-tiated’ bodies do not, however, provide evidence in favour

of the widespread occurrence of featureless, undifferen-tiated intrusions.To conclude, the postulate that undifferentiated mafic

intrusions are ubiquitous is a myth that has emerged fromincorrect interpretations of the concept of magmatic differ-entiation, the non-application or inappropriate applicationof indices of magmatic differentiation, and the inappropri-ate use of petrological evidence. No convincing examplesgiving support to this postulate have been presented.In reality, when examined in detail, all magmatic bodiesranging in size from several dozen meters to several kilo-metres thick that are formed from parental magmas ofnon-eutectic basaltic composition show more or less well-developed magmatic differentiation trends in terms ofcrystallization sequences, mineral compositions (e.g. Anin plag, En in cpx) and compatible/incompatible trace ele-ment geochemistry (e.g. Mg-number, Cr). This is illu-strated using three examples of well-differentiatedintrusive bodies with different sizes: (1) the 45m thickIvanovskii mafic sill; (2) the 660m thick Basistoppenmafic sill; (3) the 2250m thick Kivakka layered intrusion(Fig. 8). I argue that there are practically no undifferen-tiated plutonic bodies; there are only poorly or improperlystudied ones. Even thin mafic dykes, just a few centimetersthick, when studied in detail, may show remarkably well-developed internal differentiation (e.g. Chistyakova &Latypov, 2009a).

Postulate 2: Phenocryst-free basalticmagmas are not able to differentiate incrustal chambersThis postulate is intended to provide a simple physicalexplanation for the previous one. It implies that the mainmagma body of all sill-like intrusions formed frommagma initially free of phenocrysts undergoes very little,if any, differentiation (Mangan & Marsh, 1992, p. 618;Marsh, 1996, pp. 11^13; Zieg & Marsh, 2005, p. 1445; seealso quotes above). This is because intrusive bodies largelycool by conduction, with no compositional or temperaturechange in the bulk magma until the arrival of the solidifi-cation front. The hypothesis allows only one specificmechanism that leads to so-called bimodal magma differ-entiation (Marsh,1996, 2002). It involves gravity instabilityand collapse of the upper solidification front that resultsin segregations of interstitial liquid from rocks to formcoarse-grained silicic lenses in the upper portions of largediabase sills, lava lakes and gabbroic intrusions. The mech-anism is of little interest to the present discussion as itdoes not concern differentiation of the main body ofmagma in the chamber.According to the hypothesis, the magma in the chamber

does not differentiate for three major reasons: (1) allevolved interstitial liquid is trapped within solidificationfronts; (2) all crystals falling from the roof are dissolved in


1058

Fig. 8. Stratigraphic sections, compositional variations and cumulate stratigraphy through three mafic^ultramafic intrusions with differentsizes: (a) 45m thick Ivanovskii mafic sill, Fennoscandian Shield, Russia; (b) 660m thick Basistoppen mafic sill, Greenland; (c) 2250m thickKivakka layered intrusion, Fennoscandian Shield, Russia. The figure illustrates that magmatic bodies that are formed from parental magmasof non-eutectic composition tend to show well-developed magmatic differentiation in terms of crystallization sequences, mineral compositionsand compatible/incompatible trace element geochemistry. The thick continuous and discontinuous lines indicate cumulus and intercumulusminerals, respectively. Data from the following sources: Ivanovskii sill, Latypov (2003a); Basistoppen sill, Naslund (1989); Kivakka layered intru-sion, Koptev-Dvornikov et al. (2001). Cumulus mineralogy of fayalite gabbro from the Kivakka layered intrusion is based on unpublished databyYana Bychkova.


1059

the hotter magma below; (3) convection is absent in themain magma body. Close inspection of these threeassumptions shows that none of them is valid.The statement that all evolved interstitial liquid is

buried in cumulate rocks has not been supported by anypetrographic data from actual intrusions (Marsh, 1996,2006). Meanwhile, the statement is at sharp variance withtextural observations from most mafic sills, and especiallylayered intrusions. If the statement were true, non-cumulate rocks with more than 40^50 vol. % of interstitialmaterial would be the dominant rock type in such plutonicbodies. This is clearly not the case. The common occur-rence in layered intrusions of meso- to adcumulates with7^25 vol. % and 0^7 vol. % of interstitial material, respec-tively, (e.g. Wager & Brown, 1968; Parsons, 1986;Cawthorn, 1996) strongly indicates that most interstitialliquid must have somehow been expelled from the cumu-late pile into the main magma body. Transfer of this inter-stitial liquid may have been achieved by processes such asdiffusion (e.g. Morse, 1986a), compaction (e.g. Shirley,1987; Meurer & Boudreau, 1998a, 1998b) and composi-tional convection (e.g. Sparks et al., 1984; Morse, 1986a;Jaupart & Tait, 1995; Tait & Jaupart, 1996). If so, mixingof interstitial liquid with the main magma body must inev-itably result in magmatic differentiation and crystalliza-tion of rocks with progressively evolving composition.The statement that all crystals falling from the roof are

dissolved while settling through the hotter central regionof a magma chamber is based on numerical calculations,not on any direct observations of magma itself (Mangan& Marsh, 1992; Marsh, 1996). There are two major pro-blems with these calculations. First, these papers fail tomake reference to earlier, more rigorous numerical model-ling by a research group led by the fluid dynamicist andthermodynamicist M. Y. Frenkel’. These workers demon-strated that crystals settling from the roof may surviveresorption and reach the bottom, resulting in pronounceddifferentiation even in small mafic sills of about 100mthick. This happens because the dissolution of the leadingportions of settling crystals causes a marked decrease inthe temperature of the melt that allows the following por-tions of crystals to undergo progressively less dissolution,finally reaching the temporary floor of the chamber(Frenkel’ & Yaroshevsky, 1978; Koptev-Dvornikov et al.,1979; Frenkel’ et al., 1988, 1989). Because later, conflictingresults by Mangan & Marsh (1992) have not been com-pared with the earlier ones, they cannot be considered asrigorously proven. Second, there is another, much moreeffective way, of delivering crystals from the roof to thefloor that allows crystals to escape resorption in the hottermagma and lead to differentiation. This is two-phase con-vection (Grout, 1918) that involves crystalþ liquid suspen-sions that can move through the magma several orders ofmagnitude faster than small, single crystals can settle

(Morse, 1986a, 1986b; Trubitsyn & Kharybin, 1997). Thestandard causal cycle here is as follows: heat loss throughthe roof, nucleation, two-phase convection, transport ofthe two-phase packet or plume to the floor, growth of crys-tals on the floor, compositional convection of light solute,and as a result, fractionation of magma in the chamber(e.g. Irvine, 1980; Morse, 1986a, 1986b, 1988; Irvine et al.,1998).The statement that convection is non-existent in magma

chambers comes from experimental data indicating thatthermal convection ceases as soon as the liquid loses itssuperheat and reaches its liquidus temperature (Brandeis& Marsh, 1989, 1990). Because most natural magmas arenot superheated, the conclusion is made that crystalliza-tion should take place in crustal chambers from almoststagnant magmas (Marsh, 1989, 1991). Such reasoning isflawed because it ignores the fact that thermal convectionis not the only type of convection operating in crustalmagma chambers. There is another, and much more effec-tive, type of convectionçcompositional convection, whichhas been intensively studied petrologically, theoreticallyand experimentally for several decades (e.g. Morse, 1969,1982, 1986a; Sparks et al., 1984; Jaupart & Tait, 1995; Tait &Jaupart, 1996). Compositional convection is caused by den-sity instabilities arising in (1) a thin chemical boundarylayer just above the mush, or (2) the interstitial liquidwithin the mush layer (Jaupart & Tait, 1995). The implica-tion of these studies is that compositional convectionalone can keep a magma body in vigorous convection andbe an effective factor in magmatic differentiation [seedetailed discussion by Campbell (1996)].It may be instructive to examine the reasoning that

underlies the second postulate. Essentially, it comes fromthe belief that what is true for lava lakes and flows mustbe also valid for sills and layered intrusions. To give anexample, at Kilauea compositionally primitive (picritic)magmas charged with olivine crystals evolve chemicallyby dropping out olivine. Their composition evolves from�25% MgO and 49% SiO2 to �7% MgO and 52%SiO2 (Marsh, 1996), and then suddenly stops changing.Crystal fractionation ceases. Nothing approaching rhyoliteor even dacite appears. Why? The answer that is given isthat fractionation stops because the rate of advance of thesolidification front is high enough to hamper the return ofinterstitial liquid into the main mass of magma (Marsh,1996). This seems entirely plausible because the cooling oflava lakes is a relatively fast process, but this conclusion isthen applied to all plutonic bodies regardless of size(Marsh, 2006). This is a crossover point at which a state-ment that is applicable for volcanic petrology loses validitywhen applied to plutonic petrology. Plutonic bodies, suchas mafic sills and layered intrusions, crystallize and solid-ify hundreds to thousands of times slower than lava lakesand flows. This permits the operation of processes that are


1060

almost non-existent in volcanic bodies, such as compactionand compositional convection, which allow the effectivereturn of evolved interstitial liquid from the cumulate pileinto the main magma body (Sparks et al., 1984; Jaupart &Tait, 1995; Campbell, 1996), allowing it to differentiate.The important lesson to be learned from this case was for-mulated by Morse (1988, p. 222): ‘One repeatedly learns thatlayered intrusions are not volcanoes and that many plutonic systems

are unlike vented systems. It is, therefore, not prudent to reason back-

wards from volcanoes to explain layered intrusions, for then every-

thing begins to look like a volcano.’To sum up, the postulate that phenocryst-free basaltic

magmas are not able to differentiate in crustal chambersis physically ungrounded. The evidence from the rocksthemselves unequivocally tells us that an essential portionof evolved interstitial liquid is expelled from the cumulatepile. Mixing of this liquid with the resident magma mustcause it to fractionate, especially when the magma in thechamber is kept in vigorous motion by compositional orother forms of convection. The ubiquitous occurrence innature of differentiated magmatic bodies (Figs 3^6) itselfprovides irrefutable evidence that basaltic magmas can,and do, evolve in crustal chambers. The most glaring dis-proval of the postulate comes from the Sudbury IntrusiveComplex. The Sudbury testimony is very clear:phenocryst-free magmas are able to freely fractionate incrustal chambers to form well-differentiated intrusions thatshow remarkable phase, modal and cryptic layering (Fig. 7).

Postulate 3: Layered intrusions formby successive emplacement ofphenocryst-laden magma pulsesLayered intrusions show very well-developed internal dif-ferentiation that must somehow be explained. It has beensuggested that large ‘exotically layered’ intrusions are acci-dental mechanical accumulations of ‘tramp crystals’ thatare brought in with prolonged pulses of magma derivedfrom underlying magmatic mush column(s):

‘What, then, are the conditions that give rise to large, exotically

layered, sheet-like intrusions such as Bushveld, Stillwater, or

Dufec? The most obvious answer is that these bodies form from

exactly the opposite initial conditions: i.e., magma amassed

through a long series of injections of compositionally similar

magmas carrying highly variable amounts of crystals of signifi-

cant average size. The larger the volume of the ultimate magma

chamber, the more protracted and varied the sequence and nature

of the individual injections required to fill it.’ (Zieg & Marsh,2005, p. 1446).

It is thus not slow cooling and fractional crystallization,as conventional models imply, but the number and natureof magmatic pulses that lead to the ‘exoticness’ of layeredintrusions.

In other words, the layered intrusions are not remark-able natural laboratories that have long been providing uswith invaluable information about the ways magma frac-tionates with cooling and crystallization, but simple physi-cal containers in which numerous magma pulses arrivingfrom unknown deep sources dump detrital crystals. Thislatter interpretation provides no place for such fundamen-tal concepts of igneous petrology as parental magmacomposition, liquidus phase equilibria, phase crystalliza-tion sequences, mineral solid solutions, etc. They are allignored. One can hardly accept such an ungroundedreinterpretation of classical objects, as it takes us backto the dark ages of pre-Bowen igneous petrology, whenlittle was known about the governing role of liquidusphase equilibria in magmatic differentiation. Ironically,the fatal flaw of this explanation is that it is excludedby the hypothesis itself. The second postulate states thatphenocryst-free magmas cannot differentiate in intru-sive chambers. From this one can immediately inferthat such magmas cannot differentiate in any sub-chamber related to the entire underlying magmaticmush column(s). From whence can we then obtainmagmas with compositionally different phenocrysts toform layered intrusions? This question has not yet beenaddressed.The invalidity of the third postulate becomes especially

obvious when applied to specific cases of differentiatedbodies. Let us consider that most classical object of igneouspetrology, the Skaergaard layered intrusion: this reveals aremarkably continuous and complete crystallizationsequence recorded simultaneously in the Marginal BorderSeries, a Layered Series and an Upper Border Series(Fig. 9a). The various zones of the Skaergaard record theprogressive fractional crystallization of the original tholeii-tic basalt magma. Each zone is defined by the addition ofa new cumulus phase (e.g. Cpx, Mag, Pig, Ap, and Fe-Bust) or by temporary removal of an already crystallizingphase (e.g. Ol and Pig) as a result of a peritectic reactionwith the magma.The mineral solid solutions also show sys-tematic and extreme compositional changes upward inthe Layered Series, reflecting the internal changes in tem-perature and composition with the magma chamber(Wager & Brown, 1968). As the second postulate does notallow magmas to differentiate in the chamber, then theappearance and disappearance of each liquidus phasemust be related to a new emplacement of magma with therequired phenocryst assemblage. The Layered Series ofthe Skaergaard intrusion would thus require the successiveemplacement of at least eight separate magma pulses car-rying different assemblages of phenocrysts that now occuras cumulus phases in these zones (Fig. 9a): the firstmagma pulse carried olivine and plagioclase phenocrysts,the second magma pulse had olivine, plagioclase and clin-opyroxene phenocrysts, and so forth.


1061

Fig. 9. Stratigraphic section and cumulate stratigraphy through the Skaergaard layered intrusion, East Greenland (a). The three petrologicalunits, the Layered Series, Upper Border Series, and Marginal Border Series, crystallized concurrently along essentially parallel trends of differ-entiation. The Marginal Border Series has zones (LZ�, MZ� and UZ�) corresponding to all the units of the Layered Series except UZc. The‘Tranquil Zone’ (MGZT) of the Marginal Border Series has an equivalent in the Upper Border Series (UBST) and corresponds to the HiddenZone (HZ) below the Layered Series. The Layered Series formed on the floor, while the Upper Border Series crystallized under the roof andMarginal Border Series at the wall. The two fronts of crystallization converged at the Sandwich Horizon (SH).The figure shows that formationof the Layered Series of the intrusion in the frame of the hypothesis ‘no phenocrysts, no post-emplacement differentiation’ requires the successiveemplacement of at least eight separate magma pulses, each carrying a different assemblage of phenocrysts from eight independently operatingmagmatic mush columns. It should be noted that to produce the phase equilibria-controlled rock sequence of the Layered Series, the magmapulses must be emplaced strictly in the sequence from (1) to (8). However, as multiple magma emplacement is an unpredictable process, onecan envisage a huge variety of hypothetical Layered Series with chaotic compositional profiles that can be produced from random intrusion ofphenocryst-laden magma pulses from eight independently operating magmatic mush columns. None of them will be related to the phase equili-bria-controlled rock sequence of the Layered Series of the Skaergaard intrusion. For illustration, one of such hypothetical Layered Series is pre-sented in (b). Diagram is not to scale and is based on McBirney (1996).


1062

Fig. 10. Stratigraphic section, compositional variations and cumulate stratigraphy through the Basistoppen mafic sill, Greenland (a).The figure shows that formation of the sill in the frame of the hypothesis ‘no phenocrysts, no post-emplacement differentiation’ requires the suc-cessive emplacement of at least five separate magma pulses, each carrying a different assemblage of phenocrysts, derived from five indepen-dently operating magmatic mush columns. It should be noted that to produce the phase equilibria-controlled rock sequence of the Basistoppensill, the magma pulses must be emplaced strictly in the sequence from (1) to (5). However, as multiple magma emplacement is an unpredictableprocess, one can envisage a huge variety of chaotically layered mafic sills that can be produced from the random intrusion of phenocryst-laden magma pulses from five independently operating magmatic mush columns. None of them will resemble the rock sequence developed inthe Basistoppen sill. For illustration, one such ‘random’mafic sill is presented in (b). Data from Naslund (1989).


1063

These magma pulses cannot apparently be derived froma single magmatic mush column, as it can contain onlyone particular phenocryst assemblage (e.g. olivine and pla-gioclase phenocrysts). This is because the second postulatenot only forbids differentiation of magmas in the intrusionchamber but also forbids differentiation in the entire mag-matic mush column. If so, magmas with different assem-blages of phenocrysts can only be derived from separatemagmatic mush columns. For the case of the LayeredSeries it requires the simultaneous operation of at leasteight independent magmatic mush columns with composi-tionally different phenocrysts. This is not the end of thestory, however. Each of these magma pulses would bringinto the chamber phenocrysts of only one specific composi-tion that upon settling will produce a homogeneous cumu-late layer. How then can the fact that the minerals withineach cumulate zone tend to show progressive upwardchanges in composition be explained? This question is notaddressed by Marsh and his co-workers, but one can imag-ine the answer: namely, successive emplacement ofmagma pulses carrying phenocrysts of the same assem-blage (e.g. olivine and plagioclase) that become moreevolved with time.To provide the observed smooth changesin mineral compositions, one would need very manysmall magma pulses even to produce a single zone, andmany more to produce the entire sequence of the LayeredSeries. Thousands of such phenocryst-laden injections arethought to have been involved in the formation of largelayered intrusions (Marsh, 2000, p. 203).It should be noted that the upward progression in min-

eral and rock compositions of the Layered Series entirelycorresponds to what can be expected from the fractionalcrystallization of a single pulse of tholeiitic magma con-trolled by phase equilibria (e.g. Thy et al., 2006). To repro-duce such a sequence by multiple emplacements,thousands of pulses must be intruded in a sequence strictlyconsistent with liquidus phase equilibria. It is sufficient tohave only one magma pulse emplaced out of the correctorder to fail to produce a section of the Layered Series.Following this logic one can envisage an abundance ofhypothetical Layered Series with chaotic compositionalprofiles that were produced by the random intrusion ofphenocryst-laden magma pulses from eight independentlyoperating magmatic mush columns. As an illustration,one such hypothetical Layered Series is presented inFig. 9b. Its chaotic compositional profile is in striking con-trast to phase equilibria-controlled sequence of theLayered Series of the Skaergaard intrusion (Fig. 9a). Inaddition, one can also question the very existence of mag-matic mush columns with highly evolved assemblages ofthe liquidus minerals observed in these bodies (e.g. plagio-clase, ferrobustamite, magnetite, apatite and iron-rich oli-vine in the UZc). Such an evolved magma cannot bederived directly from a mantle source. It can only be

produced by protracted differentiation of basaltic magmaunder crustal chamber conditions, which, ironically, areforbidden by the second postulate.It has been argued, however, that the Skaergaard intru-

sion is not an appropriate body for testing the hypothesisbecause of the important role played by its steep sidewalls, which are not integral to the model of sheet-likemagma chambers (Marsh, 1990, p. 849). Let us consider,therefore, the 660m thick Basistoppen mafic sill(Fig. 10a), which is believed to have been emplaced as asingle pulse of magma in the upper part of theSkaergaard layered intrusion (Naslund, 1989). Despite itssmall size, the Basistoppen sill has one of the most exten-sive differentiation sequences known. It is composed offive major units that record the progressive fractional crys-tallization of the basaltic parental magma as predicted byliquidus phase equilibria. Each zone is defined by theappearance of a new cumulus phase or by the temporarydisappearance of olivine as a result of peritectic reaction.From the base upward these zones are: a Gabbro PicriteZone, containing cumulus olivine and Fe^Cr spinel (twocumulus phases); a Bronzite Gabbro Zone, containingcumulus orthopyroxene, clinopyroxene and plagioclase(three cumulus phases); a Pigeonite Gabbro Zone, contain-ing cumulus plagioclase, clinopyroxene, pigeonite, andmagnetite (four cumulus phases); a Fayalite Gabbro Zone,containing plagioclase, clinopyroxene, magnetite, ilmenite,apatite and iron-rich olivine (six cumulus phases); and aGranophyre Zone, containing cumulus plagioclase, clino-pyroxene, magnetite, ilmenite, quartz and K-feldspar (sixcumulus phases). The compositions of olivine, plagioclaseand pyroxene change systematically from the bottom tothe top of the sill, with the range of solid solution mineralcompositions comparable with those in the Skaergaardand Bushveld intrusions.Applying the logic used for the Skaergaard intrusion,

one can infer that the Basistoppen sill was produced bythe emplacement of phenocryst-laden magma pulses thatbecame progressively more evolved with time. Five majorzones of the Basistoppen sill would thus require the succes-sive emplacement of at least five separate magma pulseswith different assemblage of phenocrysts from five simulta-neously operating independent magmatic mush columns(Fig. 10a). To create the observed smooth changes in min-eral compositions would require thousands of smallmagma pulses that become more evolved with time. As inthe above case, to reproduce such a sequence by multipleemplacements, pulses must be intruded in a sequencestrictly consistent with liquidus phase equilibria: onemagma pulse emplaced out of the correct order woulddestroy the observed sequence. Following this logic onecan imagine a huge number of chaotically layeredmafic sills that were produced by the random intrusion ofphenocryst-laden magma pulses from five independently


1064

operating magmatic mush columns. As an example, one ofsuch hypothetical mafic sills is presented in Fig. 10b. Itschaotic compositional profile has nothing in commonwith the phase equilibria-controlled rock sequence of theBasistoppen sill (Fig. 10a). Thus, the hypothesis does notwork for bodies that are produced in sheet-like magmachambers either.One more conclusive piece of evidence against the third

postulate is provided by the Hasvik layered intrusion,Norway (Tegner et al., 1999). It shows a remarkable inverserelation between plagioclase An% and whole-rock Sr iso-tope compositions upsection in a 1�6 km thick layeredsequence evolving from olivine gabbro, through oxide gab-bronorite to apatiteôxide norite.This has been interpretedin terms of coupled assimilation and fractional crystalliza-tion resulting from a thermal link between the latent heatof crystallization and the heat consumed to heat and meltthe surrounding country rocks (Tegner et al., 1999). This isstrong evidence for magma differentiation within thechamber and this is essentially impossible to reconcilewith the third postulate. Not only must repeated magmarecharge events match the mineral compositions (crypticlayering) and the phase layering, but also the Sr isotopedata. In addition, the Hasvik layered intrusion reveals adearth of modal layering, so there is no evidence forrecharge with phenocryst-bearing magmas in the ‘logic’ ofthe third postulate.This discussion would not be complete without mention-

ing an example of a natural magma body with a chaoticcompositional profile. Remarkable examples of such chaot-ically layered sections are provided by the Rum Complex,NW Scotland, which crystallized as an open-system, sill-like chamber that was repeatedly replenished by numerousmagma pulses of picritic and basaltic composition(Emeleus et al., 1996). This resulted in the formation of sev-eral maficûltramafic cyclic units showing chaotic alterna-tions of peridotite and troctolite layers, indicating littlecontrol by phase equilibria (e.g. Renner & Palacz, 1987;Holness & Winpenny, 2008). This beautifully illustrateshow maficûltramafic bodies would appear if they wereproduced by the random emplacement of magma pulseswith or without phenocrysts. The fact that this is not thecase for the vast majority of mafic sills and layered intru-sions indicates that their phase-equilibria controlled sec-tions are not produced by multiple pulses in open-systemchambers, but rather by fractional crystallization ofmagma under closed-system conditions.To sum up, the postulate that layered intrusions form by

the arrival of magmas bearing progressively more evolvedloads of phenocrysts is excluded by the hypothesis itself,because it forbids magma fractionation not only in theintrusions, but also in the entire magmatic mush col-umn(s). One might call this situation an epicycle of philo-sophical delusion. Without being able to separate residual

liquid from solid phases, the hypothesis does not allow cre-ation of an evolved phenocryst-free liquid at any depth inthe Earth. It cannot produce, in particular, the basalticliquids of near-eutectic composition that are the parentalmagmas for most gabbroic sills and intrusions in theworld. The bulk composition of magmas arriving in thecrust can vary widely but clearly this variability does notarise in the mantle source: separation of crystals fromliquids or vice versa must occur somewhere and somehow.That such a process is implied at deep levels and is pre-cluded by the second postulate at shallower levels is amajor inconsistency of the hypothesis. The groundlessnature of the hypothesis is particularly well illuminatedby the Skaergaard layered intrusion and Basistoppen sill,whose compositional profiles are impossible to reconcilewith the emplacement of phenocryst-laden magma pulsesfrom magmatic mush column(s). This is because it wouldrequire a highly unlikely situation of repeated recharge ofmagma from underlying mush column(s) that always con-tains phenocrysts of the right composition to mimic frac-tional crystallization from a single parental magma. TheSkaergaard intrusion and Basistoppen sill can thereforeonly represent the result of slow cooling and fractionalcrystallization of magma under closed-system conditions(Wager & Brown, 1968; Naslund, 1989; McBirney, 1996).This conclusion can be extended, with some reservations,to most differentiated sills and layered intrusions.

CONCLUSIONThe analysis above shows that the petrological hypothesis‘no phenocrysts, no post-emplacement differentiation’ completelyfails the test against observational data from maficûltramafic sills and layered intrusions. The evidence fromthe rocks themselves unambiguously disproves this hypoth-esis. All three major postulates underlying the hypothesisare therefore invalid.First, the world does not abound with structureless,

undifferentiated, maficûltramafic intrusions. Apart fromthose bodies with parental magmas of near-eutectic com-position (e.g. OlþPlþCpxþL), all maficûltramaficbodies show marked differentiation, especially evidentwhen proper indices of magmatic differentiation areapplied (e.g. crystallization sequences, An%, Mg-number). The lack of residual granitic rocks in maficûltramafic bodies is not an indication of their undifferen-tiated nature; this may simply mean that the chemicalcomposition of their parental magmas (e.g. nepheline-normative olivine basalts) was not appropriate for theirevolution towards residual granitic melt. The differentiatednature of such maficûltramafic bodies is evident fromtheir crystallization sequences (e.g. Ol, OlþPl,OlþPlþCpx) and mineral compositional trends (e.g. oli-vine Fo80^Fo40).


1065

Second, there is nothing to prevent the chemical evolu-tion of the resident magma in a chamber by mixing withevolved interstitial liquid expelled from the underlyingcumulate pile. Strong evidence for such interstitial liquidtransfer is provided by the common occurrence of meso-and particularly adcumulates in layered intrusions.The chemical evolution of magma in the chamber proba-bly takes place by the combined operation of crystal set-tling, compaction and compositional convection in thecumulate pile and vigorous convection in the mainmagma body. Magma in a chamber can be subject to com-positional convection, irrespective of the role of thermalconvection.Third, layered intrusions are not mechanical accumula-

tions of ‘tramp phenocrysts’ of different phase and chemi-cal composition entrained from underlying mushcolumns. Bodies with such systematic crystallizationsequences can only be produced by the fractional crystalli-zation of magma in the chamber. Essentially, instead of sol-ving the problem of layered intrusions directly within acrustal magma chamber, the hypothesis transfers the prob-lem to a deeper level within the Earth. In doing so, it doesnot explain how phenocryst-laden magmas with therequired crystallization sequences (e.g. Ol, Opx, OpxþPl,OpxþPlþCpx) and compositional trends (e.g. olivineFo80^Fo20) can be produced there and delivered in the cor-rect sequence to form layered intrusions. Moreover, thehypothesis itself clearly excludes the formation of suchmagmas, as it forbids magma differentiation not only inmagma chambers themselves, but also in any sub-chamberrelated to the entire magmatic mush column(s).The hypothesis ‘no phenocrysts, no post-emplacement differenti-

ation’essentially represents a set of simple physical solutions(no return of interstitial liquid, no convection in the cham-ber, complete resorption of settling crystals, etc.) to a non-

existent petrological problemçthe widespread occurrenceof undifferentiated mafic^ultramafic intrusions. Thehypothesis is not valid, as such bodies do not exist innature and natural magmas are able to differentiate irre-spective of whether they contain intratelluric phenocrystsor not. Phenocrysts have nothing to do with the ability ofmagma to differentiate during cooling and crystallization.The more realistic situation concerning magma differenti-ation in crustal chambers can be better summarized as‘post-emplacement differentiation forever’. The ultimate test forthe hypothesis is provided by the Sudbury IntrusiveComplex, which crystallized from a highly superheatedand therefore phenocryst-free parental magma, but none-theless shows remarkable evidence of magmatic differentia-tion. Ironically, instead of representing ‘dramaticconfirmation’ of the hypothesis (Zieg & Marsh, 2005), theSudbury Intrusive Complex, in fact, provides its mostpotent refutation. The Sudbury testimony is very clear:phenocryst-free magmas are able to freely fractionate in

crustal chambers to form well-differentiated intrusionsthat show pronounced phase, modal and cryptic layering.The hypothesis ‘no phenocrysts, no post-emplacement differenti-

ation’ represents an attempt to erect a mathematicallysophisticated petrological model for the origin of mafic^ultramafic intrusions that fails because it is based on asuperficial study of real rocks. Such prerequisite attributesof any successful petrological model as detailed petro-graphic descriptions, mineral compositional data, whole-rock geochemical data, thorough consideration of parentalmagma composition, mineral crystallization sequences,and liquidus phase equilibria are notably absent fromMarsh’s papers [the only exception is the recent paper byBe¤ dard et al. (2007), which appeared long after the formu-lation of the hypothesis]. As a result, the hypothesis addslittle to understanding of real magmatic processes, as it ispoorly constrained by actual observations. To prove itsvalidity, the hypothesis is presented in terms of equationsand numerical calculations that tend to obscure the under-standing of very simple petrological phenomena. It maybe noted in this connection that, as the philosophy ofscience tells us (Platt, 1964, p. 352): ‘manyçperhapsmostçof the great issues of science are qualitative, notquantitative, even in physics and chemistry. Equationsand measurements are useful when and only when theyare related to proof; but proof or disproof comes first andis in fact strongest when it is absolutely convincing withoutany quantitative measurements.’It may be instructive to comment on the likely back-

ground that resulted in the appearance of such an extrava-gant hypothesis. Magmatic differentiation always includestwo factorsçchemical and physical. The chemical factorestablishes the compositional differences between solidand liquid phases upon crystallization, whereas the physi-cal factor preserves these differences by segregating or frac-tionating the phases so that they can form distinctiverocks. Ignoring either of them may lead to misleading con-clusions. This is what has essentially happened duringthe formulation of the hypothesis ‘no phenocrysts, no post-

emplacement differentiation’: the process of magmatic dif-ferentiation was reduced to a simple physical story withoutconsideration of the chemical factor of magmatic differen-tiation. As a result, it comes into glaring opposition withliquidus phase equilibria, as has been repeatedly illustratedhere using many examples of mafic sills and layered intru-sions. The hypothesis may probably be valid for volcanicpetrology dealing with solidification of lava flows andlava lakes in which the chemical factor of differentiationbecomes inefficient because of fast cooling. As for plutonicpetrology, it is hardly applicable even for small maficdykes, let alone sills and layered intrusions. This confidentassertion derives from our recent centimeter-scale geo-chemical study of dolerite dykes, which has revealed thatcrystal^liquid differentiation processes already operate in


1066

dykes less than a half meter thick (Chistyakova & Latypov,2009b, 2009c). In conclusion, I believe that the future ofigneous petrology lies in approaches that combine boththe chemical and physical aspects of magmatic differentia-tion. A good example of such research is the pioneeringnumerical modelling of several Siberian mafic sills by M.Y. Frenkel and co-authors that was carried out several dec-ades ago (Frenkel’ & Yaroshevsky, 1978; Koptev-Dvornikov et al., 1979; Frenkel’ et al., 1988, 1989; see arecent review by Ariskin & Yaroshevsky, 2006). We shouldcontinue research along these lines to assure further prog-ress in our understanding of how magma chambers work.

ACKNOWLEDGEMENTSThis paper would never have been written without strongencouragement, support and much good advice frommany of my colleagues who specialize in the field ofmafic^ultramafic sills and layered intrusions, as well as influid dynamic processes in basaltic magma chambers, andwho share a similar opinion with respect to the petrologi-cal hypothesis in question. Among them are RichardWilson, Brian Robins, Ed Ripley, Richard James, SteveSparks, Ian Campbell, Steve Prevec, Michail Dubrovskiiand Andrey Lavrenchuk. I would like to express my sin-cere gratitude to all of them. I am also grateful to AlexeyAriskin, Eero Hanski and Jury Podladchikov for their dis-agreements and criticisms of some aspects of this study,which helped me to reshape the paper and sharpen itsarguments. The manuscript was improved by very con-structive and supportive reviews of Marian Holness,Christian Tegner, Grant Cawthorn and Tony Morse,whose contribution is highly acknowledged. The researchwas supported by Fellow Research Grant from theFinnish Academy of Science.

REFERENCESAarnes, I., Podladchikov,Y. & Neumann, E.-R. (2008). Post-emplace-

ment melt flow induced by thermal stresses: implications for differ-entiation in sills. Earth and Planetary Science Letters 276, 152^166.

Alapieti, T. T. & Lahtinen, J. J. (2002). Platinum-group elementmineralization in layered intrusions of Northern Finland and theKola Peninsula, Russia. In: Cabri, L. J. (ed.) The Geology,

Geochemistry, Mineralogy and Mineral Beneficiation of Platinum-group

Elements. Canadian Institute of Mining, Metallurgy and Petroleum, Special

Volume 54, 507^546.Ariskin, A. A. & Yaroshevsky, A. A. (2006). Crystallization differen-

tiation of intrusive magmatic melt: Development of a convection^accumulation model. Geochemistry International 44, 72^93.

Bakushkin, E. M. (1983). Some geological peculiarities of the Ni-bear-ing Karirjavr massif, Kola Peninsula. Sovetskaya Geologiya 3, 65^72,(in Russian).

Be¤ dard, J. H. J., Marsh, B. D., Hersum, T. G., Naslund, H. R. &Mukasa, S. B. (2007). Large-scale mechanical redistribution oforthopyroxene and plagioclase in the Basement Sill, FerrarDolerites, McMurdo Dry Valleys, Antarctica: petrological,

mineral-chemical and field evidence for channelized movement ofcrystals and melt. Journal of Petrology 48, 2289^2326.

Brandeis, G. &Marsh, B. D. (1989).The convective liquidus in a solidi-fying magma chamber: a fluid dynamic investigation. Nature 339,613^616.

Brandeis, G. & Marsh, B. D. (1990). Transient magmatic convectionprolonged by solidification. Geophysical Research Letters 17, 1125^1128.

Campbell, I. H. (1996). Fluid dynamic processes in basaltic magmachambers. In: Cawthorn, R. G. (ed.) Layered Intrusions.

Developments in Petrology 15, 45^76.Cawthorn, R. G. (ed.) (1996). Layered Intrusions. Developments in Petrology

15, 531 pp.Chistyakova, S. Yu. & Latypov, R. M. (2009a). Fine-scale chemical

zonation of small dolerite dykes, Kestio« Island, SW Finland.Geological Magazine (in press).

Chistyakova, S.Yu. & Latypov, R. M. (2009b). On the development ofinternal chemical zonation in small mafic dykes. Geological

Magazine (in press).Chistyakova, S.Yu. & Latypov, R. M. (2009c). Two independent pro-

cesses responsible for compositional zonation in mafic dykes of the—land^—boland Dyke Swarm, Kestio« Island, SW Finland. Lithos(in press).

Czamanske, G. K., Zen’ko, T. E., Fedorenko, V. A., Calk, L. C.,Budahn, J. R., Bullock, J. H., Jr, Fries, T. L., King, B.-S. W. &Siems, D. F. (1995). Petrographic and geochemical characterizationof ore-bearing intrusions of the Noril’sk type, Siberia; with discus-sion of their origin. Resource Geology, Special Issue 18, 1^48.

Dickin, A. P., Nguyen,T. & Crocket, J. H. (1999). Isotopic evidence fora single impact melting origin of the Sudbury Igneous Complex.In: Dressler, B. O. & Sharpton, V. L. (eds) Geological Society of

America. Large meteorite impacts and Planetary Evolution II 339, 361^371.Ding, T. P. & Schwarcz, H. P. (1984). Oxygen isotopic and chemical

compositions of rocks of the Sudbury basin, Ontario. Canadian

Journal of Earth Sciences 21, 305^318.Dubrovskii, M. I. (1998). DifferentiationTrends of the Standard Alkalinity

Olivine-normative Magmas and Corresponding Rock Series. Apatity: KolaScience Center, 336 pp. (in Russian).

Emeleus, C. H., Cheadle, M. J., Hunter, R. H., Upton, B. G. J. &Wadsworth, W. J. (1996). The Rum Layered Suite. In:Cawthorn, R. G. (ed.) Layered Intrusions, Developments in Petrology

15, 403^439.Faggart, B. E., Basu, A. R. & Tatsumoto, M. (1985). Origin of the

Sudbury Complex by meteoritic impact: neodymium isotopic evi-dence. Science 230, 436^439.

Frenkel’, M.Ya. & Yaroshevsky, A. A. (1978). Fractional crystallizationof an intrusive magma: simulation of the thermal history and dif-ferentiation of a tabular intrusion involving settling of the solidphases. Geochemistry International 15, 1^19.

Frenkel’, M. Ya., Yaroshevsky, A. A., Ariskin, A. A., Barmina, G. S.,Koptev-Dvornikov, E. V. & Kireev, B. S. (1988). Dynamics of in situ

differentiation of mafic magmas. Moscow: Nauka, 216 pp. (in Russian).Frenkel’, M. Ya., Yaroshevsky, A. A., Ariskin, A. A., Barmina, G. S.,

Koptev-Dvornikov, E.V. & Kireev, B. S. (1989). Convective^cumu-lative model simulating the formation process of stratified intru-sions. In: Bonin, N., Didier, J., Le Fort, P., Propach, G., Puga, E.& Vistelius, A. B. (eds) Magma^Crust Interactions and Evolution.Athens: Theophrastus, pp. 3^88.

Galerne, C. (2009). Emplacement mechanisms and magmatic differ-entiation induced by magma flow in sill intrusions in sedimentarybasins. PhD thesis, University of Oslo. Faculty of Mathematicsand Natural Sciences, pp. 178.

Gibb, F. G. F. & Gibson, S. A. (1989). The Little Minch Sill Complex.ScottishJournal of Geology 25, 367^370.


1067

Gibb, F. G. F. & Henderson, C. M. B. (1992). Convection and crystalsettling in sills. Contributions to Mineralogy and Petrology 109, 538^545.

Gibb, F. G. F. & Henderson, C. M. B. (2006). Chemistry of the ShiantIsles Main Sill, NW Scotland, and wider application for petrogen-esis of mafic sills. Journal of Petrology 47, 191^230.

Gibson, S. A. & Jones, A. P. (1991). Igneous stratigraphy and internalstructure of the Little Minch Sill Complex, Trotternish Peninsula,northern Skye, Scotland. Geological Magazine 128, 51^66.

Gorbunov, G. I.,Yakovlev,Yu. N., Goncharov,Yu.V., Gorelov,V. A. &Telnov, V. A. (1985). The nickel areas of the Kola Peninsuka. In:Papunen, H. & Gorbunov, G. I. (eds) Nickel^Copper Deposits of theBaltic Shield and Scandinavian Caledonides. Geological Survey of Finland,

Bulletin 333, 41^109.Grout, F. F. (1918). Two phase convection in igneous magmas. Journal

of Geology 26, 481^499.Gunn, B. M. (1962). Differentiation in Ferrar Dolerites, Antarctica.

New ZealandJournal of Geology and Geophysics 5, 820^863.Hall, A. (1996). Igneous Petrology. Harlow: Longman, 551 pp.Hanski, E. J. (1992). Petrology of the Pechenga ferropicrites and coge-

netic, Ni-bearing gabbro^wehrlite intrusions at Pechenga, KolaPeninsula, Russia. Academic dissertation. Geological Survey of

Finland, Bulletin 367, 192 pp.Hess, P. C. (1989). Origin of Igneous Rocks. Cambridge, MA: Harvard

University Press.Hoatson, D. M. & Sun, S.-S. (2002). Archean layered maficûltrama-

fic intrusions in the West Pilbara Craton, Western Australia: asynthesis of some of the oldest orthomagmatic mineralizing systemsin the world. Economic Geology 97, 847^872.

Holness, M. B. & Winpenny, B. (2008). The Unit 12 allivalite, EasternLayered Intrusion, Isle of Rum: a textural and geochemical studyof an open-system magma chamber. Geological Magazine doi:10.1017/S0016756808005797, 1^14.

Hort, M., Marsh, B. D., Resmini, R. G. & Smith, M. K. (1999).Convection and crystallization in a liquid cooled from above: Anexperimental and theoretical study. Journal of Petrology 40,1271^1300.

Huppert, H. E. & Turner, J. S. (1991). Comments on ‘On convectivestyle of vigor in sheet-like magma chambers’ by Bruce D. Marsh.Journal of Petrology 32, 851^854.

Irvine, T. N. (1980). Magmatic density currents and cumulus pro-cesses. AmericanJournal of Science 280-A (JacksonVolume), 1^58.

Irvine, T. N. & Smith, C. H. (1967). The ultrabasic rocks of theMuskox intrusion, Northwest Territories, Canada. In: Wyllie, P. J.(ed.) Ultramafic and Related Rocks. NewYork: JohnWiley, pp. 38^49.

Irvine, T. N., Andersen, J. C. �. & Brooks, C. K. (1998). Includedblocks (and blocks within blocks) in the Skaergaard intrusion: geo-logical relations and the origins of rhythmic modally gradedlayers. Geological Society of America Bulletin 110, 1398^1447.

Jaupart, C. & Tait, S. (1995). Dynamic of differentiation in magmareservoirs. Journal of Geophysical Research 100, 17615^17636.

Koptev-Dvornikov, E.V.,Yaroshevsky, A. A. & Frenkel, M.Ya. (1979).Fractional crystallization of an intrusive magma: reality of the sedi-mentation model. Geochemistry International 16, 85^101.

Koptev-Dvornikov, E. V., Kireev, B. S., Pchelintseva, N. F. &Khvorov, D. M. (2001). Distribution of cumulative mineral assem-blages, major and trace elements over the vertical section of theKivakka intrusion, Olanga group of intrusions, Northern Karelia.Petrology 9, 1^24.

Krivolutskaya, N. A., Ariskin, A. A., Sluzhenikin, S. F. &Turovtsev, D. M. (2001). Geochemical thermometry of the Talnakhintrusion: assessment of the melt composition and the crystallinityof the parental magma. Petrology 9, 451^479.

Latypov, R. M. (2002). Phase equilibria constraints on relations ofore-bearing intrusions with flood basalts in the Noril’sk region,Russia. Contributions to Mineralogy and Petrology 143, 438^449.

Latypov, R. M. (2003a). The origin of marginal compositional rever-sals in basicûltrabasic sills and layered intrusions by Soret fractio-nation. Journal of Petrology 44, 1579^1618.

Latypov, R. M. (2003b). The origin of basicûltrabasic sills with S-,D- and I-shaped compositional profiles by in situ crystallization ofa single input of phenocryst-poor parental magma. Journal of

Petrology 44, 1619^1656.Latypov, R. M. (2007). Noril’sk- and Lower Talnakh-type intrusions

are not conduits for overlying flood basalts: insights from residualgabbroic sequence of intrusions. Applied Earth Science (Transactions

of the Institutions of Mining and Metallurgy: Section B) 116, 215^225.Latypov, R. M. & Chistyakova, S.Yu. (2009). Phase equilibria testing

of a multiple pulse mechanism for origin of maficûltramafic intru-sions: a case example of the Shiant Isles Main Sill, NW Scotland.Geological Magazine (in press).

Latypov, R. M., Smolkin,V. F. & Alapieti,T.T. (2001). Differentiationtrend and parent melt composition of the Ni-bearing gabbro^wehr-lite Pechenga intrusions, Kola Peninsula, Russia. Petrology 9,381^397.

Latypov, R. M., Chistyakova, S. Yu. & Alapieti, T. T. (2007).Revisiting problem of chilled margins associated with marginalreversals in maficûltramafic intrusive bodies. Lithos 99, 178^206.

Lightfoot, P. C., Keays, R. R. & Doherty,W. (2001). Chemical evolu-tion and origin of nickel sulfide mineralization in theSudbury Igneous Complex, Ontario, Canada. Economic Geology 96,1855^1875.

Mangan, M. & Marsh, B. D. (1992). Solidification front fractionationin phenocryst-free sheet-like magma bodies. Journal of Geology 100,605^620.

Marsh, B. D. (1988). Crystal capture, sorting, and retention in convect-ing magma. Geological Society of American Bulletin 100, 1720^1737.

Marsh, B. D. (1989). On convective style and vigor in sheet-likemagma chambers. Journal of Petrology 30, 479^530.

Marsh, B. D. (1990). Reply to discussion of ‘Crystal capture, sorting,and retention in convecting magma’ by Sparks, R. S. L. GeologicalSociety of America Bulletin 102, 849^850.

Marsh, B. D. (1991). Reply to comments on ‘On convective style ofvigor in sheet-like magma chambers’ by Huppert, H. E. & Turner,J. S. Journal of Petrology 32, 855^860.

Marsh, B. D. (1996). Solidification fronts and magmatic evolution.Mineralogical Magazine 60, 5^40.

Marsh, B. D. (2000). Magma chambers. In: Sigurdsson, H.,Houghton, B., Rymer, H., Stix, J. & McNutt, S. (eds) Encyclopediaof Volcanoes. NewYork: Academic Press, pp. 191^206.

Marsh, B. D. (2002). On bimodal differentiation by solidification frontinstability in basaltic magmas, I: Basic mechanics. Geochimica et

Cosmochimica Acta 66, 2211^2229.Marsh, B. D. (2004). A magmatic mush column Rosetta Stone: the

McMurdo Dry Valleys of Antarctica. EOS Transactions, American

Geophysical Union 85, 497^502.Marsh, B. D. (2006). Dynamics of magmatic systems. Elements 2,

287^292.Marsh, B. D. & Wheelock, M. M. (1994). The vertical variation of

composition in the Peneplain sill and Basement sills of the DryValleys: The Null Hypothesis. Antarctic Journal of the United States

29, 25^26.McBirney, A. R. (1996). The Skaergaard intrusion. In: Cawthorn, R.

G. (ed.) Layered Intrusions. Developments in Petrology 15, 147^180.McBirney, A. R. (2007). Igneous Petrology, 3th edn. Boston, MA: Jones

and Barlett, 550 pp.


1068

Meurer, W. P. & Boudreau, A. E. (1998a). Compaction of igneouscumulates part I: geochemical consequences for cumulates andliquid fractionation trends. Journal of Geology 106, 281^292.

Meurer, W. P. & Boudreau, A. E. (1998b). Compaction of igneouscumulates part II: Compaction and the development of igneousfoliations. Journal of Geology 106, 293^304.

Miller, C. F. & Miller, J. S. (2002). Contrasting stratified plutonsexposed in tilt blocks, Eldorado Mountains, Colorado River Rift,NV, USA. Lithos 61, 209^224.

Morse, S. A. (1969). The Kiglapait layered intrusion, Labrador. Geological

Society of America, Memoirs 112, 204 pp.Morse, S. A. (1982). Adcumulus growth of anorthosite at the base of

the lunar crust. Journal of Geophysical Research 87, A10^A18.Morse, S. A. (1986a). Convection in aid of adcumulus growth. Journal

of Petrology 27, 1183^1214.Morse, S. A. (1986b). Thermal structure of crystallizing magma with

two-phase convection. Geological Magazine 123, 205^214.Morse, S. A. (1988). Motion of crystals, solute, and heat in layered

intrusions. Canadian Mineralogist 26, 209^244.Mutanen, T. (1997). Geology and ore petrology of the Akanvaara and

Koitelainen mafic layered intrusions and the Keivitsa^Satovaara layered com-

plex, northern Finland. Geological Survey of Finland Bulletin 395, 233 pp.Naldrett, A. J., Bray, J. G., Galparrini, E. L., Podolsky, T. &

Rucklidge, J. C. (1970). Cryptic variation and the petrology of theSudbury Nickel Irruptive. Economic Geology 65, 122^155.

Naslund, H. R. (1989). Petrology of the Basistoppen Sill, EastGreenland: a calculated magma differentiation trend. Journal ofPetrology 30, 299^319.

Parsons, I. (ed.) (1986). Origins of Igneous Layering. NATOASI Series C:

Mathematical and Physical Sciences 196, 666 pp.Pattison, E. F. (1979). The Sudbury layer. Canadian Mineralogist 17,

257^274.Platt, J. R. (1964). Strong inference: certain systematic methods of

scientific thinking may produce much more rapid progress thanothers. Science 146, 347^353.

Renner, R. & Palacz, Z. A. (1987). Basaltic replenishment of the Rummagma chamber: evidence from unit 14. Journal of the Geological

Society, London 144, 961^970.Shirley, D. N. (1987). Differentiation and compaction in the Palisades

sill, NewJersey. Journal of Petrology 28, 835^865.Simura, R. & Ozawa, K. (2006). Mechanism of crystal redistribution

in a sheet-like magma body: Constraints from the Nosappumisakiand other shoshonite intrusions in the Nemuro peninsula,Northern Japan. Journal of Petrology 47, 1809^1851.

Sparks, R. S. L. (1990). Discussion of ‘Crystal capture, sorting, andretention in convecting magma’ by Marsh, B. D. Geological Society ofAmerica Bulletin 102, 847^848.

Sparks, R. S. J., Huppert, H. E. & Turner, J. S. (1984). The fluiddynamics of evolving magma chambers. Philosophical Transactions ofthe Royal Society of London, Series A 310, 511^534.

Tait, S. & Jaupart, C. (1996). The producing of chemically stratifiedand adcumulate plutonic igneous rocks. Mineralogical Magazine 60,99^114.

Tegner, C., Robins, B., Reginiussen, H. & Grundvig, S. (1999).Assimilation of crustal xenoliths in a basaltic magma chamber: Srand Nd isotopic constraints from the Hasvik layered intrusion,Norway. Journal of Petrology 40, 363^380.

Tegner, C., Wilson, J. R. & Robins, B. (2005). Crustal contaminationin basalt and jotunite: constraints from layered intrusions. Lithos83, 299^316.

Therriault, A. M., Fowler, A. D. & Grieve, A. F. (2002). The SudburyIgneous Complex: a differentiated impact melt sheet. Economic

Geology 97, 1521^1540.Thode, H. G., Dunford, H. B. & Shima, M. (1962). Sulfur isotope

abundances in rocks of the Sudbury district and their geologicalsignificance. Economic Geology 57, 565^578.

Thy, P., Lesher, C. E., Nielsen, T. F. D. & Brooks, C. K. (2006).Experimental constraints on the Skaergaard liquid line of descent.Lithos 92, 154^180.

Trubitsyn, V. P. & Kharybin, E. V. (1997). Convection in magmachambers due to inversion of vertical distribution of deposited crys-tals. Fizika Zemli 5, 47^52.

Wager, L. R. & Brown, G. M. (1968). Layered Igneous Rocks. Edinburgh:Oliver & Boyd, 588 pp.

Warner, S., Martin, R. F., Abdel-Rahman, A. F. M. & Doig, R. (1998).Apatite as a monitor of fractionation, degassing, and metamorph-ism in the Sudbury igneous complex, Ontario. Canadian

Mineralogist 36, 981^999.White, C. M. (2007). The Graveyard Point intrusion: an example of

extreme differentiation of Snake River Plain Basalt in a shallowcrustal pluton. Journal of Petrology 48, 303^325.

Wilson, J. R. & Overgaard, G. (2005). Relationship between theLayered Series and the overlying evolved rocks in the Bjerkreim^Sokndal intrusion, southern Norway. Lithos 83, 277^298.

Wilson, J. R. & S�rensen, H. S. (1996).The Fongen^Hyllingen layeredintrusive complex, Norway. In: Cawthorn, R. G. (ed.) Layered

Intrusions. Developments in Petrology 15, 303^329.Wilson, J. R., Robins, B., Nielsen, F. M., Duchesne, J. C. & Vander

Auwera, J. (1996). The Bjerkreim^Sokndal layered intrusion, south-west Norway. In: Cawthorn, R. G. (ed.) Layered Intrusions.

Developments in Petrology 15, 231^255.Worster, M. G., Huppert, H. E. & Sparks, R. S. J. (1990). Convection

and crystallization in magma cooled from above. Earth and

Planetary Science Letters 101, 78^89.Yoder, H. S. & Tilley, C. E. (1962). Origin of basalt magmas: an

experimental study of natural and synthetic rock systems. Journalof Petrology 3, 342^532.

Zieg, M. J. & Marsh, B. D. (2005). The Sudbury Igneous Complex:viscous emulsion differentiation of a superheated impact meltsheet. Geological Society of America Bulletin 117, 1427^1450.


1069

testing thevalidityof the petrological hypothesis … · testing thevalidityof the petrological...

Documents