7/27/2019 j.1525-1314.1992.tb00083.x

1/12

1. metamorph ic Geol., 1992, 10, 283-294

Porphyroblast rotation: eppur s i muove*?C . W. P A S S C H I E R , R . A. J . T R O U W , ' H. 1 . Z W A R T AND R . L . M. V I S S E R SInstitute of Earth Sciences, PO Box 80.021, Budapestlaan 4, Utrecht 3508 TA , The Netherlands'Departamento de Geologia, Universidade Federal do Rio de laneiro, CP27970 Rio de ianeiro, Brazil

A B S T R A C T In a number of recent papers, the theory has been postulated that porphyroblasts as a rule do not rotatewith respect to geographical coordinates, and can be used to determine the original orientation of olderfoliations. Complex inclusion patterns in spiral garnets have even been used to advocate a new model oforogenesis, involving several alternating phases of horizontal shortening and extension. Criticalassessment of the assumptions and data used to support the theory of irrotational porphyroblasts revealsnumerous flaws. Millipede structures, used as proof for flow partitioning, can also form by other flowgeometries. Evidence quoted to support irrotational behaviour of porphyroblasts is unsound. Porphyrobl-asts do occur in sets with a preferred orientation of the internal foliation trace, but these cannot be shownto represent original orientations. Microstructures which resemble truncation planes in spiral garnets areused as evidence that these structures developed by several phases of deformation and as proof forperiodic extension and horizontal shortening in orogenesis. They can, however, also be explained byintermittent growth of a rotating porphyroblast during a single phase of deformation. Finally,porphyroblast sets in which orientation is a function of aspect ratio indicate that porphyroblast rotationwith respect to kinematic axes does occur in at least some situations.Key words: millipede structure; porphyroblast rotation; spiral garnet.

INTRODUCTIONA basic problem in geology is that rocks a re commonly theonly available data source from which a sequence of eventsis to be reconstructed; only an end-product can be studied,without access to intermediate stages. This appliesespecially to structural geology, where deformationmechanisms, flow and strain patterns, and deformationhistories have to be inferred from finite geometries inrocks. The situation is comparable to that of reconstructingthe sequence in which this manuscript was written andmodified from the completed paper as it now stands. Thebest possible way to come to reliable conclusions is byexperimental work and careful study of simple geometricalsituations, in which as many parameters as possible havebeen fixed. Even then, we may have overlooked an aspectof the geometry considered to be insignificant, which mayprove to be crucial for correct interpretation. Structuralgeology proceeds by small steps, through reinterpretationof geometries as our understanding of rheology and ofprogressive deformation in crystalline materials grows, andnew data become available.

In a group of recent papers, established concepts of theinterpretation of microstructural porphyroblast geometriesin metamorphic rocks have been criticized (Bell, 1985,

* Italian: 'but she moves nonetheless', the words reportedlyspoken by Galileo Galilei on 22 June 1633 at the offices of theHoly Inquisition in Rome, after having been forced to renouncethe idea that the Earth rotates around the Sun.

1986; Bell et al., 1986; Bell & Johnson, 1989, 1990;Steinhardt, 1989; Johnson, 1990a, b; Hayward, 1990).These papers have made the useful contribution of callingattention to the fact that many apparently rotationalfabrics can also be explained as non-rotational. However,they advocate the statement that porphyroblasts as a ruledo not rotate with respect to geographical coordinates. Ifthis statement is correct, inclusion patterns preservedwithin porphyroblasts could be used to deduce the originalorientation of early foliations. On the basis of thesestatements, Bell & Johnson (1989) and Johnson (1990a)reinterpreted spiral-shaped inclusion patterns in garnets asbeing the result of progressive overgrowth over folds of upto eight deformation phases. They claim that these phasesreflect periodic shortening and gravitational collapse of anorogen and that porphyroblasts can be used in this way askeys in the reconstruction of orogenesis. Because of theinherent problems with the interpretation of geometries indeformed rocks outlined above, it seems that such aradical reinterpretation of basic concepts in orogenesisjustifies a critical analysis of the underlying assumptions,methods and data. In this paper, we proceed to assess thevalidity of the most important statements used by severalof the aforementioned authors. We use the followingterminology (Fig. 1):

S, (i for internalepattern of elongate inclusions,reflecting an overgrown foliation within a porphyroblast asobserved in thin section;S, (e for external)-trace of a foliation in the matrixaround porphyroblasts as observed in thin section;

283

7/27/2019 j.1525-1314.1992.tb00083.x

2/12

284 C . W. P A S S C H I E R E T A L .

mica cap-a mica-rich domain adjacent to an objectsuch as a porphyroblast;

strain shadow-a domain with low mica contentadjoining an object such as a porphyroblast (also known asa pressure shadow);

cleavage lamella(e)-the domain of a spaced cleavage orcrenulation cleavage which is rich in elongate grains ofminerals such as micas or amphiboles;

microlithon-the domain of a spaced cleavage orcrenulation cleavage which is rich in equidimensionalgrains of minerals such as quartz, feldspar or carbonates:this domain may contain relic fold closures;

truncation plane-surface in a porphyroblast whichforms a sharp boundary between domains in which S, havedifferent orientations: Si is discontinuous over a truncationplane;

deflection plane-plane in a porphyroblast where Sishows a sharp change in orientation: this structureresembles a truncation plane but S, is not discontinuous;

kinematic frame-set of characteristic directions in adeforming rock, such as the bulk instantaneous shorteningand extension directions and the flow plane in the case ofprogressive simple shear deformation;

syntectonic growth-growth in an actively deformingrock: also known as synkinematic growth.

C O M M E N T S T O S T A T EM E NT S ONP O RP H YR O B L A S T R O T A T I O NStatement (1)-porphyroblasts do not rotatebecause they are mechanically fixed in the rock bystrain shadows (Bell, 1985, 1986; Bell eta/., 1986;Bell & J ohnson, 1989; Bell & Cuff, 1989;Steinhardt, 1989; J ohnson,1990a, b; Hayward,1990)Rigid objects in a linear or non-linear viscous fluid willrotate with respect to the kinematic frame of the flow if thefluid deforms by a homogeneous non-coaxial flow typesuch as simple shear (Jeffery, 1922; Bretherton, 1962;Ghosh & Ramberg, 1976; Gierszewski & Chaffey, 1977).The angular velocity of the object depends on the vorticityof the bulk flow and, for non-spherical objects, on objectorientation and aspect ratio. In simple shear flow,maximum and minimum angular velocities are reachedwhen the long axis of the object lies normal or parallel tothe simple shear flow plane (Ghosh & Ramberg, 1976;Passchier, 1987). If the foliation i s parallel to this flowplane, the object can rotate with respect to a fixedfoliation.

Porphyroblast geometries where S, is oblique to S, andwhere S, and S, constitute one continuous foliation (Fig.2a) have traditionally been interpreted along these lines asrigid objects which rotated with respect to a foliation fixedto the flow plane of simple shear (Fig. 2b; Clough, 1897;Flett, 1912; Turner, 1948; Ramsay & Huber, 1987, p. 633).However, if the foliation is not fixed with respect to thekinematic frame, both the foliation and the porphyroblast

Fig. 1. Schematic diagram showing terminology as used in thispaper.can rotate and porphyroblasts with S, oblique to S, canform in other ways. Ramsay (1962) pointed out that suchporphyroblast geometries can even form by rotation of afoliation with respect to a stationary spherical object incoaxial (pure shear) progressive deformation (Fig. 2c). Bell(1985) and Bell ef ul. (1986) have suggested that thesituation is even more complex. They state that sphericalobjects can also be stationary in non-coaxial progressivedeformation if flow is partitioned around t h e object. If arigid strain shadow is present adjacent to a porphyroblast,the strain shadow and porphyroblast may act mechanicallyas a single elongate rigid object (Bell, 1985; Bell et al.,1986). Such a combined porphyroblast-strain shadowaggregate will rotate at a very slow rate in non-coaxial flowin some orientations, leading to oblique S J S , patterns(Fig. 2d; Passchier, 1987). Possible natural examples ofsuch a situation have been reported by Fyson (1975, 1980).The local occurrence of such protected porphyroblastsdoes not imply, however, that rotation of equidimensionalporphyroblasts does not occur, although this is claimed byBell (1985, 1986), Bell et al . (1986), Bell & Johnson(1989), Bell & Cuff (1989), Steinhardt (1989), Johnson(1990a, b) and Hayward (1990).

Flow partitioning as described above is commonlyreferred to in the literature as deformation partitioning;although this is not strictly correct we will adhere to thisuse here i n order to avoid confusion.

If a porphyroblast overgrows a straight foliation, somestrain i s needed for the development of a strain shadow,and until this strain has accumulated, the porphyroblast isunprotected and may therefore rotate. Bell & Rubenach

7/27/2019 j.1525-1314.1992.tb00083.x

3/12

P O R P H Y R O B L A S T R O T A T I O N 285

Fig. 2. (a) Schematic diagram of aporphyroblast with S, obl ique to S, in thesurrounding matrix. Such a structure canform by a large numbe r of progressivedeformation paths including: (b) rotat ion of 1 . 1he porphyroblast with respect to astationary foliation in simp le shear flow; (c)rotation of the fol iation aroun d a s tat ionaryporphyroblas t in pure she ar f low, (d)rotation of the fol iation aroun d a s tat ionaryporphyroblas t protec ted by d s t rain shado win simple shear flow Strain shad ow in grey 7(1983), Bell et al . (1986) and Bell & Johnson (1989)address this problem by suggesting that the porphyroblastgrows in an existing microlithon, which remains unde-formed. In this case, S, in the porphyroblast should beexactly mimicked in the adjacent strain shadow, which israrely the case in nature. The statement that the foliationtrace in the strain shadow was destroyed by ongoing orsubsequent deformation can be countered by the argumentthat, in that case, the porphyroblast would no longer beprotected by the strain shadow, and may thus haverotated.

We conclude that it is rarely possible to determine froma superficial investigation of porphyroblasts with obliqueS,/S, relationships whether the porphyroblast or theexternal foliation has rotated, or both (compare Fig. 2a, b,c & d); careful investigation of the structure is needed, andin a number of cases no solution may be possible. In allcases, it is safe to refer to the structure as being due torelative rotation of S, and S, .Statement (2)-millipede microstructures areevidence for deformation partitioning ntodomains of coaxial and non-coaxial progressivedeformation (Bell & Rubenach, 1980; Bell, 1985,1986; Bell &J ohnson, 1989)Bell & Rubenach (1980) were the first to describe unusualS,/S, geometries in and around porphyroblasts which theynamed millipede microstructures (Fig. 3a). Such milli-pede structures are taken to be indicative of deformationpartitioning into domains of coaxial and non-coaxialprogressive deformation (Bell & Rubenach, 1980; Bell,

1985, 1986; Bell & Johnson, 1989). However, experimentsby Ghosh & Ramberg (1976) and computer modelling byMasuda & Mochizuki (1989) show that asymmetricmillipede-type structures can also form around objects thatrotate with respect to the kinematic frame of progressivedeformation (Fig. 3b). Millipede structures seem to resultfrom deflection of foliation planes around rigid objects ingeneral, and are not necessarily indicative of deformationpartitioning of the type envisaged by Bell & Rubenach(1980).

Millipede-like structures can also result from sectioning.Schoneveld (1979) has shown that millipede-like structuresmay occur in thin sections of porphyroblasts withspiral-shaped inclusion trails, if these are cut parallel towhat he interpreted as the rotation axis. Some of theexamples in Bell & Johnson (1989) may be of this kind.Only serial sectioning can show the true nature of thesepatterns. We would like to stress, however, that not allmillipede structures can be explained as a sectioning effect.

Statement (3)-porphyroblasts have not rotatedwith respect to each other because they showidentical orientation o f Si over a large area,despite later deformation (Bell, 1985; Bell e t a / . ,1986; Bell &J ohnson, 1989,1990; Steinhardt,1989; J ohnson, 1990a, b; Hayward, 1990)One of the main advances in the study of porphyroblasts inthe last decade has been the discovery by Fyson (1975,1980), Bell (1985) and Bell et al . (1986) that Si in differentporphyroblasts can have a remarkably similar orientationwithin an outcrop, or even over a larger area, despite

7/27/2019 j.1525-1314.1992.tb00083.x

4/12

286 C . W . P A S S CHI E R E J A L .

pure shear -no rotation of objectrotation of foliation

c

dextral simple shearobject rotation

sinistral simple ahear -rotation of object and foliation

A.Fig. 3. (a) Typical millipede microstructure after Bell &Rubenach (1980). These structures are interpreted by Bell (1985,1986) to form by flow partitioning around a stationaryporphyroblast, as indicated; (b) millipede structures can also formby passive deformation of a foliation around a rigid object inabsence of flow partitioning of the type envisaged by Bell (1985,1986); three possible settings ar e shown, two of which aresituations where the porphyroblast rotates with respect to thekinematic frame.

deformation which post-dates porphyroblast growth. Thebest documented example is by Fyson (1975, 1980). Thepreferred orientation of inclusion patterns does not imply,however, that the porphyroblasts did not rotate at all; theymay have rotated due to flattening associated with thedevelopment of the later foliation. Let us investigate thedata in the five papers that are cited most frequently asevidence for non-rotation of porphyroblasts (Fyson, 1975,1980; Steinhardt, 1989; Johnson, 1990a, b).

The papers by Fyson (1975, 1980) show a strikingpreferred orientation of Si in biotite porphyroblasts overan area with a folded foliation. However, close inspectionshows up to 45" variation in orientation of Si inporphyroblasts over the investigated area, which may bedue to porphyroblast rotation.

Johnson (1990b) claims to observe irrotational be-haviour of porphyroblasts in the Otago schists. He doesnot present the full three-dimensional orientation of Sipatterns, however. The rose diagrams in Johnson (1990b)are plots of the strike of steeply dipping S, which show aconsistent orientation of the mean values; close inspection,however, shows that the spread in orientation of the strikein each diagram is considerable, up to 60"; in addition, thedip of Si in porphyroblasts is totally unconstrained. Clearly,these kinds of da ta cannot be used as proof for constantorientation of Si. In fact, the same type of relativelyconstant orientation of Si strike would be expected from apopulation of rotated porphyroblasts with spiral inclusionpatterns, with subparallel rotation axes. In that case,stretching lineations should lie approximately perpendicu-lar to the strike direction of Si . Inspection of the map fig.9(b) in Johnson (1990a) shows that locally the lineationsare indeed approximately perpendicular to S, traces in therose diagrams.

Steinhardt (1989) gives evidence for constant orientationof Si in porphyroblasts in schists in South Australia. Hisfig. 4 hows the large-scale distribution of traces of Si in theplane of thin sections, plotted in a stereogram. However,because his thin sections are not randomly orientated,their orientation influences the Si distribution in thediagram and the significance of this plot is thereforedoubtful. Only his fig. 5 shows the full orientation ofplanar Si in six porphyroblasts for a double fold whichpost-dates the porphyroblast growth. Of the six S, planes,five are of similar orientation, but these are all fromporphyroblasts which lie in the same fold limb; the only Siplane in a porphyroblast on another limb has a differentorientation. Clearly, the value of this material as evidencefor non-rotation of porphyroblasts is highly questionable.

Besides the fact that presently published evidence fornon-rotation of porphyroblasts is dubious, t h e papersdealing with the subject (Fyson, 1975, 1980; Bell, 1985;Steinhardt, 1989; Bell & Johnson, 1989, 1990; Johnson,1990a, b; Hayward, 1990) commonly apply unsoundargumentation, based on statements which cannot betested. On the one hand, the authors claim thatporphyroblasts do not rotate, since Si in porphyroblasts hasthe same orientation over large areas, in spite of intenselater deformation (Fyson, 1975, 1980; Steinhardt, 1989;Bell & Johnson, 1989, 1990; Johnson, 1990a, b; Hayward,1990). On the other hand, if porphyroblasts are discussedwhich do have a variable orientation of Si over an area,this is interpreted as an effect of folding of the foliationprior to porphyroblast growth or to deformation of theporphyroblasts, but no t to rotation of porphyroblasts(Bell, 1985; Johnson, 1990b). By this interpretation, thebasic statement of the theory of irrotational porphyroblastscannot be tested.

7/27/2019 j.1525-1314.1992.tb00083.x

5/12

PORPHYROBLAST ROTATION 287

(a) original tin31

Fig. 4. Effect of deformation on theorientation of porphyroblast populations.Schematic map views show porphy roblasts(rectangles) with S, parallel t o their longaxes (inset top right) in strain shadows(white) flanked by cleavage lamellae (solidlines). Sh ape of strain shadows an d cleavagelamellae is indicated schematically. S, an dthe porphyroblasts have a strong preferredorientation with respect to geographicalcoordinates indicated by orientationdistribution graphs under the maps; n =number of measurements. (a) Simple shearof the material by slip on cleavage lamellaewhile the strain shadows remainundeformed may cause rotation of thewhole set of porphyroblasts and the mainfoliation (S,) into a new geographicalorientation; (b ) when a similar set ofporphyroblasts is deformed by non-coaxialflow throughout the specimen ,porphyroblasts are reorientated with respectto the cleavage lamellae. In both cases, therotation of S, with respect to geographicalcoordinates cannot be deduced from thefinite fabric in the rock.

n Si

0 90 180North East S outh

final1 N

1;90 180North East SouthStatement (4)-porphyroblasts are fixed togeographical coordinates (Steinhardt, 1989; Bell &J ohnson,1989,1990; J ohnson,1990a; Hayward,1990)From the discussion of statement (3 ) it will be clear thatthe orientation of S, in porphyroblasts in a volume of rockmay have a distinct mean orientation, but that the spreadin orientation of S, in different porphyroblasts can berather high. It is difficult in such situations to decide whichorientation of S, is the original orientation of the foliationover which the porphyroblasts grew. The mean value couldbe taken as such, but one can never be sure thatsubsequent deformation did not modify or shift theorientation distribution of S, while a clear mean value isstill preserved (Fig. 4). Non-coaxial flow in cleavagelamellae without internal deformation in strain shadows asenvisaged by Bell (1985, 1986), Bell et al. (1986) and Bell& Johnson (1980) could cause rotation of all porphyro-blasts without changing their orientation with respect to

each other (Fig. 4a). Penetrative non-coaxial flow afterporphyroblast growth can also cause a shift in porphyro-blast orienta tion, without affecting the shape of theorientation curve of Si (Fig. 4b). Besides this influence ofdeformation on porphyroblast orientation, any large-scalerigid body rotation of volumes of rock will cause rotationof the included porphyroblast population with respect togeographical coordinates. Reconstructions of the originalorientation of foliations from porphyroblast populationsare therefore unreliable.Statement (5)-porphyroblasts can remainirrotational with respect to geographicalcoordinates through several phases ofdeformation (Bell & J ohnson,1989,1990; J ohnson,1990a; Hayward, 1990)Natural examples show that porphyroblasts can remainrelatively stationary with respect to each other during onephase of foliation development after growth if the

7/27/2019 j.1525-1314.1992.tb00083.x

6/12

288 C . W . P A S S C H I E R ET A L .

Fig. 5. Porphyroblasts with rigid strain shadows c an be protecte dfrom rotation by these strain shadows (a) ; however, when they areaffected by later shortening oblique to the foliation, theporphyroblasts may lose the protection of these strain shadows (b )until new strain shadows are developed related to the newlydeveloped foliation (c). In stage (b) porphyroblasts areunprotected a nd are free to rota te. Strain shadows-greyornamentation, S, In porphyroblasts-striped ornamentation.

porphyroblasts are equidimensional and progressivedeformation is coaxial or, in other cases, if theporphyroblasts lie in elongate strain shadows (statement 1;Fyson 1975, 1980). It seems unlikely, however, thatporphyroblasts can remain irrotational with respect togeographical coordinates through several phases ofdeformation which form overprinting foliations in obliquedirections. The main problem is that shortening at a highangle to the strain shadows around a porphyroblast will

(c) ,deflection plane

Fig. 6. Schematic diagram of the developm ent of deflectionplanes in a porphyroblast which is periodically growing while itrotates in a non-coaxial How: (a) porphyroblast core which hasgrown over a planar foliation is rotated and mica caps develop atupper left and lower right; (b) when these mica caps aresufficiently develop ed, they a re overgrow n by the porphyroblastmineral; (c) deflection planes a re produced in the process. Furtherrotation of the porphyroblast can cause development of new micacaps and a repetition of the process.

tend to destroy this shadow, thereby destabilizing theporphyroblast which is then free to rotate until a newstrain shadow has been established (Fig. 5). Bell &Johnson (1989, 1990), Johnson (1990a, b) and Hayward(1990) did not address this problem.

Fig. 7. A garnet porphyroblast in which Si is clearly continuous with S, at top left and bottom right, but with two wings at top and bottomwhere S, has a d ifferent orientation from Si in the core; deflection planes partly sep arate the core and the wings. These could be m istakenas truncation planes. Development of this porphyroblast m ay correspond to the sequ ence in Fig. 6. Bafios, Equador. Scale bar = 1 mm.Fig. 8. Staurolite crystal synkinematic with respect to the crenulation cleavage, the lower part of which overgrew a cleavage lamella,showing that the mode l proposed by Bell el al . (1986) restricting porphyroblast growth to microlithons is not valid in this case. LukmanierPass, Switzerland. Scale bar =0.5 m m .Fig. 9. Garne t porphyroblasts in mica schist that overgrew cleavage lamellae and part of a m icrolithon. SHo Felix do Araguaia, Goias,Brazil. Scale bar =0.5 mm .Fig. 10. Detail of a garnet porphyroblast as in Fig. 9 which overgrew tw o cleavage lamellae and part of microlithons. The garnet shows arelative rotation of 45" with respect to the crenulation cleavage. SHo Felix do Ara guaia, G oias, Brazil. Scale bar =0.5 mm.

7/27/2019 j.1525-1314.1992.tb00083.x

7/12

P O R PH Y RO B L A S T R O T A T I O N 289

7/27/2019 j.1525-1314.1992.tb00083.x

8/12

290 C . W . P A S S C H I E R E T A L .

Statement (6)-porphyroblasts in pelit ic rocksnucleate during progressive deformation inmicrolithons, grow unti l they impinge ondeveloping cleavage lamellae, and stop growingthere (Bell & Rubenach, 1983; Bell eta/., 1986;Bell&J ohnson,1989)The idea that porphyroblast growth is syntectonic andlimited to microli thons which correspond to domains ofcoaxial progressive deform ation (state me nt 2) is of crucialimportance for the development of the non-rotat ionalmodel proposed by Bell and co-workers. O n theoreticalgrounds, Bell & Rubenach (1983) , Bel l et al . (1986) andBell & Johnson (1989) have rejected the possibil i ty ofgrowth of a porphyroblast over cleavage lamellae or micacaps. They claim that pressure solution during deformationwill inhibit such growth. H owe ver, exam ples of syntectonicporphyroblasts which have indeed at least part ial lyovergrown cleavage lamellae (Figs 8-10) or even compl e t edeveloping crenulation cleavages including numerouslamellae (Fig. 11) ar e not uncommon in nature.Statement (7)-sharp deflections of Si inporphyroblasts are truncation planes and can betaken as evidence for polyphase deformation (Bell& J ohnson,1989; J ohnson,1990a, b)Some porphyroblasts contain more complex inclusionpatterns than a single straight or curved S, pat tern . Suchporphyroblasts may contain truncation planes, where onepattern of S, i s t runcated by another pat tern (Fig . 1;Thompson et a l . , 1977; Karabinos, 1984) . Truncat ionplanes apparently form by a relat ive change in theorientation of the porphyroblast with respect to thesurrounding foliation, possibly with local partial solution ofthe porphyroblast , followed by new porphyroblast growth.As a result, S, in the newly a dded s egm ent will have adifferent orientation than that in the core of the crystal(Rosenfeld, 1968; Karabinos, 1984). Bell & Johnson(1989) and Johnson (1990a, b) claim t o observe trunc ation

planes in spi ral garnets , and use the presence of theseplanes as evidence that spi ral garnets form by a sequenceof separate deform at ion phases. In th is in terpretat ion theydeduce u p to eight deform at ion phases dur ing all of whichthe garnets cont inued growing. However , few realtruncation planes are visible in their examples. Si is onlylocally discontinuous and in most places only deflected.Such st ructures are bet ter descr ibed as deflection planesrather than truncation planes (Fig. 6). Deflection planescan form dur ing a single phase of progressive deformationif porphyroblast growth is not radially con tinuou s (Fig. 6).Mica caps can form by dissolut ion of quar tz and feldsparwherever the surface of a porphyroblast l ies oblique to theinstantaneous shor tening direct ion of the de format ion(Fig. 6a, b). A fter t he m ica cap is sufficiently de velo ped, i tmay be overgrown by the porphyroblast and a deflectionplane of S, bearing some similari ty to a t runcat ion plane i sformed (Fig. 6c). Figure 7 shows an example of a garnetwhich ma y have for me d in this way.

Statement (8)-spiral garnets do not form b yrotation during a single phase of deformation , butare an effect of polyphase deformation andgrowth (Bell &J ohnson,1989,1990; J ohnson,1990a; Hayward, 1990)Garnets with spiral inclusion patterns have tradit ionallybeen in terpreted as objects which grew dur ing rotat ionwi th respect to the kinemat ic f rame of bulk deformat ion(Flett , 1912; Miigge, 1930; Rosenfeld, 1970; Schoneveld,1979) . The new interpretat ion by Bel l & Johnson (1989)has been inspi red by the occurrence of deflection planesinterpreted as t runcat ion planes, and mi l l ipede- typeinternal structures in some spiral garnets (Bell & Johnson ,1989; Hayward, 1990); each set of al leged truncationplanes is explained as the effect of a separate phase ofdeforma t ion, and a large number of deform at ion phases istherefore proposed for rocks with this type of porphyro-blast . Mi l l ipede s t ructures, however , can form aroundobjects that ro tate wi th respect to t he kinemat ic f rame ofprogressive deformat ion (s tatement 3 ) , and deflection

Fig. 11. Central part of a garn et porphyroblast t ha t overgrew a progressively developing crenulation cleavage (S3). The garnet isnon-rotational and demo nstrates that if garnets synkinematically overgrow tightening folds, growth is not ne cessarily restricted tomicrolithons between cleavage lamellae. Note that the ope n folds in the centre of the crystal deform an older crenulation cleavage (S2).Vermont, USA; specimen courtesy of C. Scho neveld. Width of micrograph=30 mm.Fig. 12. A spiral garnet with 120" relative rotation of S, with respect to S,. Note how the strain shadows are incorporated and bend withthe garnet quite differently from the evolution imagined for such 'rotational' garnets by Bell & Johnson (1990a). Itumirim, Minas Gerais,Brazil. Scale bar = 1 mm.Fig. 13. Fully deve loped spiral garnet with continuous quartz spirals linked with the ou tside strain shadows. The specime n fits therotational model (Schone veld, 1977, 1979) quite w ell, but no t the model proposed by Bell & Johnson (1990a) because of the lack oftruncation planes and m illipede structures. R elative rotation of garnet with respect to foliation is estimated to be 300" in a clockwisesense, which makes it more prob able that the ga rnet rotated in space instead of the foliation. CaCaipava do Sul, Rio Grande do Sul,Brazil. Scale bar =0.5 mm .Fig. 14. Garnet with well-developed continuous quartz and opaqu e spirals interpreted to represent strain shadows and mica capsincorporated during growth (Sc honeve ld, 1979). The g arnet ro tated approxim ately 270" in a clockwise sense with respect to the matrixfoliation. Aiuruoca, Minas Gera is, Brazil. Scale bar = I mm .

7/27/2019 j.1525-1314.1992.tb00083.x

9/12

P O R P H Y R O B L A S T R O T A T I O N 291

7/27/2019 j.1525-1314.1992.tb00083.x

10/12

292 C . W. P A S S C H I E R E T A L .

planes can form by syntectonic overgrowth of mica caps(statement 6). We think that inclusion patterns in garnetsas shown in Bell & Johnson (1989) and Hayward (1990)can also be explained as syntectonic growth of rotatinggarnets, for the following reasons.

Schoneveld (1979) has shown that generally two spiralsare present in such garnets; one defined by quartz,interpreted as relics of the strain shadow absorbed by thegarnet, and the other by opaques, interpreted as a relic ofovergrown S, in mica caps. Beautiful, continuous spirals ofthis type are common in nature (Figs 12-14; Schoneveld,1977, his figs 3, 4, 5, 6, 8, 9 ; Powell & Vernon, 1979, theirfigs 4-6). It is difficult to imagine how a pattern ofsubsequent phases of deformation and growth (Bell &Johnson, 1989, their fig 20) could form such regular spiralshapes. Schoneveld (1979) proposed that spiral patternsdevelop when garnets rotate at constant angular velocityand grow continuously and in a radially symmetric way.However, if during a single deformation event growth islocalized and periodic rather than radially symmetric andcontinuous (Fig. 6), or if the angular velocity of the garnetis variable due to variable bulk strain rate, inclusionpatterns similar to those in Bell & Johnson (1989) andHayward (1990) could result. The large number ofdeformation phases postulated by these authors cantherefore be explained as an equal number of growthphases during a single deformation event. The presence ofcontinuous quartz spirals in many of the figures in Bell &Johnson (1989, their figs 1, 4, 6g, 8c, d) supports this idea.The quartz spirals can be explained by a model ofporphyroblast rotation and syntectonic growth (Schone-veld, 1979; Schulz, 1990), but seem incompatible with amodel of polyphase deformation and growth as shown inBell 8c Johnson (1989, their fig. 20); it is difficult tounderstand why the new pressure shadows never overlapeach other and stop exactly where the one from thepreceding phase ended. The millipede patterns withoutquartz spirals presented by Bell & Johnson (1989, their figs12-14) can be explained as sections parallel to the garnetrotation axis (Schoneveld, 1979). Notice in the figures ofBell & Johnson (1989) that wherever clear quartz spiralsare present, millipedes are absent or asymmetrical whilesymmetric millipedes only exist in the absence of quartzspirals. The arguments in Bell & Johnson (1989) andHayward (1990) should be based on serial sectioning toshow the three-dimensional structure of S ,, rather than onsingle two-dimensional sections. This will show, forexample, whether the millipede structures in Bell &Johnson (1989) are real millipedes, or a sectioning effect(see statement 2 ) . To date, three-dimensional patterns ofS, in garnets have only been published as stereographicimages by Schoneveld (1979, his figs 14 & 19). Thesepatterns do not contain deflection or truncation planes.

A final consideration refers to the period of porphyro-blast growth. In the Bell & Johnson (1989) model,porphyroblasts grow more or less continuously during upto eight successive deformation phases that are associatedwith an equal number of inversions of the kinematicframework of an entire mountain belt. The model shows

considerable uplift followed by collapse related to eachtwo phases, that would change the metamorphic conditionsat any given point considerably. It seems unlikely thatgarnet growth could continue through such an evolution.

F A B R IC S I N D I C A T I V E OF R O T A T I O NFrom the content of this paper it will be obvious thatporphyroblast microstructures are difficult to interpret interms of kinematic analysis. Nevertheless, we can proposea few structures which may be used as evidence forrotation of porphyroblasts with respect to each other andtherefore of rotation of at least some porphyroblasts in thekinematic frame of bulk progressive deformation. Onepossible argument in favour of relative rotation ofporphyroblasts with respect to the kinematic frame of bulkdeformation in the rock is a dependence of objectorientation on object shape (Fig. 15). Zwart & Calon(1977) described elongate chloritoid crystals in a schistwith only one foliation, from the Urseren zone,Switzerland, that show opposite orientations of slightly

Fig. 15. (a) Schematic representation of porphyroblastmicrostructure in which elongate porphyroblasts h ave S, withopposite relative rotation sense to S,, depending on porphyroblastorientation. This can be interpreted as opposite sense of rotationin response to flattening normal to S,, as outlined by Zwart &Calon (1977). (b) Schem atic three-dimensional representation ofthe dependence of the relative rotation angle between S, an d S, onporphyroblast orienta tion. This effect has been observed inBosost, Spain, by Zw art (1962) an d it is explained by dependenceof porphyroblast rotation rate on porphyroblast orientationL=biottte lineation.

7/27/2019 j.1525-1314.1992.tb00083.x

11/12

P0 R PH Y RO B L A S T R 0 T A T 0 N 293

s-shaped inclusion patterns for crystals with oppositevergence of the long crystal axis and S, (Fig. 15a).According to the authors, the only possible explanation isrotation of the crystals during coaxial flattening oblique tothe pre-existing foliation. Crystals which have their longaxis perpendicular to the foliation as a rule do not showsigns of relative rotation, as would be expected (Zwart &Calon, 1977, their fig. 5 ) . Their conclusion is corroboratedby the presence of asymmetric pressure shadows aroundthe crystals which have developed on the lee side of therotating porphyroblast. Schoneveld (1979) described andexplained a similar relationship for amphibole crystalsfrom Norway. In Bosost, Spanish Pyrenees, elongateandalusite porphyroblasts in mica schist have their longaxes in the plane of the foliation (Zwart, 1962). The anglebetween S, and S, in oblong andalusite porphyroblastsdepends on the orientation of andalusite in the foliationplane; crystals which have their long axes parallel to abiotite lineation in the foliation plane have parallel S, andS, (Zwart, 1962); crystals with long axes normal to thelineation have S,/S, angles up to 110" (Fig. 15b). Allintermediate angles are encountered in the same rock.Vissers (1989) described a similar relationship betweenaspect ratio of garnet porphyroblasts and the S,/S, anglefor samples from the Betic Cordillera, Spain (Vissers, 1989,his fig. 6). Such a relationship would be expected for rigidobjects rotating with respect to kinematic axes, but not forstationary objects. Recently, Vissers & Mancktelow (1992)and Busa & Gray (1992) have presented new data whichsupport rotational behaviour of porphyroblasts. In allthese cases there is no clear preferred orientation of S, inthe rock, and rotation of at least some of the crystals withrespect to the foliation and the kinematic frame of thedeformation seems the most likely explanation.

CONCLUSIONS(1) Porphyroblasts in which the inclusion patterns are

relatively constant in orientation over a large volume ofrock despite later deformation do exist in nature. They canbe explained as porphyroblasts which have grown over aplanar foliation in a rock (Fyson, 1975, 1980), which issubsequently affected by development of a crenulationcleavage that anastomoses around the porphyroblasts. Insuch a situation, the porphyroblasts may shift and slightlyrotate with respect to each other, without destruction ofthe preferred orientation of the inclusion patterns. Thispreferred orientation, however, is not necessarily aninherited unchanged orientation with respect to geographi-cal coordinates.

(2) Rotation of porphyroblasts with respect to thekinematic frame of bulk deformation in a volume of rockcan only be proven in cases where a clear relationshipexists between aspect ratio and orientation of elongateporphyroblasts, and the relative rotation angle between S,and S,. All other cases of porphyroblasts with straight S,oblique to S, are presently inconclusive.

(3) Inclusion patterns in spiral garnets, even if they

contain deflection planes or millipede structures, cannot beused as sound evidence for multiple phases of horizontalshortening and extension; the inclusion patterns can beexplained by continuous or periodic garnet growth duringrotation of garnets in a single deformation phase. A newmodel for orogenesis as proposed by Bell & Johnson(1989) is therefore unfounded.(4) In well-studied mountain belts like the Alps and thePyrenees, no independent evidence exists for repeatedchanges in shortening and extensional events as envisagedby Bell & Johnson (1989). Because of the uncertaintiesinvolved in the interpretation of porphyroblast microstruc-tures, it seems hazardous to use these patterns as exclusiveevidence for new concepts of orogenesis.

FINAL STATEMENTAs matters stand, we do not claim that we understand thecomplete behaviour of porphyroblasts in anisotropicmedia, and we recognize that Bell and co-workers havemade a major contribution in a field which has been sadlyneglected for at least a decade by most other researchgroups. With the present state of affairs, however, it isunwise to use porphyroblasts for determination of shearsense without cross-checking with structures whosedevelopment is better understood. The use of porphyro-blasts to determine the orientation of earlier foliationsseems unfounded. Samples must be carefully investigatedand individually interpreted as to the kinematicsresponsible for their development. More detailed ex-perimental work must be done, such as the investigation of'natural experiments', experiments with analogue materialsin general non-coaxial flow fields, and finite elementanalysis. Evidence for rotation or non-rotation ofporphyroblasts could be obtained where porphyroblastscan be found in association with fibrous veins which trackthe incremental extension direction; no such situationshave come to our knowledge, but if they exist they wouldbe extremely useful to decide whether porphyroblasts orfoliations have rotated with respect to the kinematicframe. Only further work can tell us what geometricpatterns in porphyroblasts are important and reliable, andwhich are not.ACKNOWLEDGEMENTSWe thank the research group Casus Belli for usefuldiscussion. R.A.J.T. acknowledges the Brazilian nationalcouncil for research (CNPq), process no. 202267190-3, forfinancing his post-doctoral stay at Utrecht University.Reviews by V. L. Hansen and P. Karabinos andsuggestions by J. P. Platt, N. Mancktelow and B. Ildefonseare gratefully acknowledged.

REFERENCESBell, T. H., 1985. Deformation partitioning and porphyroblastrotation in metamorphic rocks: a radical reinterpretation.Journal of Metamorphic Geology, 3, 109-1 18.

7/27/2019 j.1525-1314.1992.tb00083.x

12/12