Preferred Orientation of Olivine Crystals in Peridotites, With Special Reference to New Zealand Examples.

Transactions and Proceedings of the Royal Society of New Zealand, Volume 72, 1942-43, Page 280

Preferred Orientation of Olivine Crystals in Peridotites, With Special Reference to New Zealand Examples. By Francis J. Turner, University of Otago. Introduction. During the last eight years several papers dealing with preferred orientation of olivine in peridotites, olivine-nodules of basalts, and olivinites of metamorphic origin have appeared (e.g. Andreatta 1934, 1935; Ernst 1935; Phillips 1938). From these it would seem that strong preferred orientation of the component crystals commonly develops during deformation of olivine-rocks, and that in the case of peridotites such deformation may originate during emplacement of a largely crystalline intrusive mass. Certain orientation rules have already been established for olivine in rocks of different origins. But before satisfactory conclusions can be reached as to the mechanism and causes of the orienting processes, or the mode of development and petrofabric significance of such frequently observed phenomena as undulose extinction and lamellar structure in olivine, additional data must be obtained from rocks of widely different age and geographic occurrence. The present paper is a contribution to this end, and includes fabric data for typical New Zealand peridotites and olivine-nodules, with which are compared olivine fabrics of certain well-known European rocks of like composition. Parting, Translation-lamellae and Undulose Extinction in Olivine. Olivine has a distinct cleavage parallel to (010) and a less perfect cleavage parallel to (100). In the olivine of peridotites this second plane of mechanical weakness (perpendicular to γ) may be rendered more obvious by the presence of minute rod-like opaque or semi-opaque schiller inclusions which tend to be sharply oriented in (100) of the host crystal (cf. Harker 1919, p. 84). Among the rocks discussed in this paper this feature is particularly well shown in some of the fresh black peridotites of Rum (e.g. 7497* Unless otherwise indicated, numbers refer to specimens and sections in the collections of the Geology Department, University of Otago.) described by Harker (1908), and a partially serpentinised dunite (5308) from the vicinity of Bluff (Service 1937, p. 212, no. 208). The black peridotite of Abhuinn Fiadh-innis, Rum, has been described by Phillips (1938, p. 134) as showing a remarkable development of lamellae parallel to (100), but in the writer's sections of this rock it is difficult to estimate the relative importance of fine schiller inclusions and of actual planes of parting in determining this sharp,

finely lamellar appearance, as seen under high-power magnification using a universal stage without the analyser. Iddings (1906, p. 370) mentions inclusions of ilmenite arranged parallel to (100) in the olivine of some basalts, so that in general the presence of schiller inclusions of this type in olivine is not necessarily connected either with plutonic conditions of crystallisation or with deformation. Certainly of deformational origin, however, are closely-spaced, discontinuous but sharply-defined lamellae parallel to (100) that may be observed in sections of many New Zealand dunites when viewed between crossed nicols. In general appearance these recall the translation-lamellae of deformed quartz figured in a recent paper by H. W. Fairbairn (1941, pl. 2, fig. 1), and are so interpreted by the present writer (cf. Ernst, 1935, p. 153). The olivine lamellae are very closely spaced, usually are restricted to parts of any particular grain, and not infrequently terminate before reaching the grain boundary. In a random section as seen between crossed nicols with an ordinary microscope the lamellae fall into two alternating series with a difference of 5° to 10° in extinction position, thus simulating polysynthetic twinning (cf. Harker, 1919, p. 84). Careful measurements on a universal stage show, however, that the crystallographic orientations of adjacent lamellae are always only slightly different; indeed, the lamellar structure often disappears when the plane of the lamellae is brought parallel to the axis of the microscope-tube. Even where the difference in optic orientation is a maximum, stereo-graphic plotting of α β and γ for each set of lamellae shows that the only twin relationships that could be responsible for the observed phenomena would be irrational types, viz. either (a) normal twinning with the twin axis inclined at about 5° to γ (= crystallographic a), or (b) parallel twinning with the twin plane inclined at 85° to γ and the twin axis at 3° to 5° to α or β. Twinning must therefore be eliminated as a possible explanation, and translation-gliding remains the most likely mode of origin for lamellar structure sub-parallel to (100) in olivine. Buerger (1930, p. 13) has noted that in general translation-gliding is commonly accompanied by slight changes in optic orientation along the translation planes.* Similar sharply defined lamellae parallel to (100) are developed in crystals of enstatite associated with lamellar olivine in South Island peridotites. Though this is highly suggestive of origin by translation-gliding in such cases (cf. N. F. M. Henry, Min. Mag., vol. xxvi, pp. 179–189, 1942), universal application of this explanation to cover all cases of lamellar hypersthenes in norites, dolerites, etc., is in the writer's opinion still unwarranted. Undulose extinction, so commonly observed in the olivine of peridotites in general, is particularly prominent in the dunites and other peridotites of the western portion of the South Island of New Zealand. Within any large undulose superindividual, the boundaries of the subindividuals or extinction-bands are invariably subparallel to (100) and not infrequently take the form of sharply defined surfaces of rupture. If the lamellar structure referred to in the preceding paragraph is also present, it maintains constant orientation within any one subindividual, but there may be slight differences of orientation between lamellae of adjacent subindividuals.

Considerable attention has recently been given to the phenomenon of undulose extinction, especially as it occurs in deformed grains of quartz (e.g. Sander 1930, pp. 173–180; Knopf 1938, pp. 170, 171; Hietanen 1938, pp. 33–35; Fairbairn 1941, pp. 1273–1276). It is generally agreed that the undulatory bands result from a permanent distortion of the space-lattice, which not infrequently leads at last to rupture and slight translation parallel to the boundary surfaces of the subindividuals. According to Hietanen (loc. cit.) the primary mechanism by which undulose extinction develops in bands parallel to the vertical axis in quartz is translation and flexure on the basal plane; Fairbairn on the other hand favours a hypothesis of translation governed by an invariable glide line (the edge m:r) and variable irrational glide-planes inclined at low angles to the basal plane. In both cases the causal translation movement is pictured as operating on planes perpendicular to the trend of the undulose bands. If a like mechanism is assumed for the origin of undulose extinction in olivine, certain conditions follow immediately with regard to the relative parts played by glide-directions and glide-planes in the deformation. The possibility of deformation governed by a definite glide-plane without any predominating glide-direction must be discarded for olivine. For in such a case the boundaries between the extinction bands of any crystal could be parallel to any plane perpendicular to the glide-plane; actually they are restricted to a single crystallographic plane, (100). All that is necessary to produce the effects observed in olivine is the presence of a controlling invariable glide-direction parallel to γ (a crystal axis); translation could occur on any plane containing γ, i.e. on any plane in the zone [100], or it could be restricted to an invariable plane in that zone. It is interesting to note at this point that in olivine the direction of maximum atomic density, and hence the most likely potential glide-direction, is the a crystal axis (cf. Bragg 1937, p. 147); furthermore on the similar grounds the most likely glide-plane is (010). The crystallographic planes and directions of structural weakness with which petrofabric data should be compared are as follows:—(010); ⊥ α; most distinct cleavage; on theoretical grounds the most probable glide-plane. (100); ⊥ γ; indistinct cleavage; most pronounced plane of schillerization and accompanying parting; plane of microscopically observable translation-lamellae, and undulose banding. [100]; = γ; on theoretical grounds the most likely glide-direction; glide-direction assumed for development of undulose extinction by flexure-gliding in a plane or planes perpendicular to the resultant banding. [001]; = β; on theoretical grounds a probable glide-direction. Banding and Fissility in Peridotites. Parallel banding, usually interpreted as a primary flow structure, is well known in peridotites. In many New Zealand dunites thin, discontinuous black bands rich in chromite, usually only a few

millimetres thick, alternate with wider green olivine-rich bands poor in chromite. Flow banding is also well exhibited in the Tertiary peridotites of Rum and Skye as described by Harker (1904, 1908). This includes both small-scale alternation of layers respectively rich and poor in a spinellid mineral (e.g. Harker, 1904, pp. 71, 72), and banding of a much more striking nature, clearly conspicuous in the field, such as that resulting from variation in the feldspar content of interlaminated sheets of peridotite and allivalite on the island of Rum (Harker, 1904, pp. 75, 76, pl. ii; 1908, pp. 72–74). Harker's interpretation of all these parallel structures as the result of magmatic flow is supported by marked dimensional parallelism of the undeformed plagioclase tables in many of the allivalites and troctolites, and by the definite tendency for elongated olivines to show dimensional parallelism in Hebridean peridotites investigated by Phillips (1938, p. 131). There are strong objections to any hypothesis of origin of banded structures in the unserpentinised dunites of New Zealand and the Hebridean province, other than that of laminar flow in a partially, perhaps almost completely crystallised magma. Laminated structure very commonly develops in crystalline schists by metamorphic differentiation through the medium of aqueous pore-solutions, but an analagous mechanism could hardly be responsible for the structure of dunites in which hydrous minerals such as serpentines, chlorites, talc, etc., are completely lacking. The same objection, together with the absence of recognisable replacement textures weighs against the possibility of late magmatic introduction of chromite in the present case, though this mode of origin is probable for some chromite-rich rocks associated with New Zealand peridotites, and has also been suggested for particular instances of partially serpentinised banded chromite-dunites in other parts of the world (e.g., Rynearson and Smith, 1941) Cataclastic origin may also be ruled out, for many of the rocks concerned show no other certain evidence of deformation than a slightly undulose condition observable in some olivine grains, while the chromite or picotite of the dark layers typically is in sharp undeformed octahedral crystsals. A totally different type of banding not infrequently noted in basic plutonic rocks is that shown by norites and gabbros of the Bushveld and Stillwater type (e.g., Hess 1938). The controlling magmatic process, rhythmic gravitational settling of heavier and lighter crystals in a liquid subject to periodic slight turbulence (Hess 1938, p. 266), cannot, however, be applied to banded dunites that have been emplaced in the form of vertical or steeply dipping sills as in the case of the great peridotite masses of the South Island of New Zealand. While accepting the hypothesis of flow-banding, the writer would draw attention to certain petrographic details of a picotite-dunite (7496) from South of Alt a'Chaoich, Skye (Harker 1904, pp. 71, 72, Fig. 13), that make it difficult to picture the exact sequence of events during crystallisation and flow of the magma in question. The light coloured bands consist of olivine, with accessory scattered grains of poorly translucent chromite, while the dark layers are made up of idiomorphic grains of a distinctly more translucent picotite wholly

enwrapped by highly allotriomorphic interstitial basic plagioclase (An76±3) a mineral which is completely lacking in adjacent olivine-rich bands. These two assemblages of minerals can hardly be products of simultaneous crystallisation from the same magma. Less conspicuous than flow banding, but nevertheless of fairly general occurrence in peridotites, is a slight fissility two types of which are to be distinguished:— (a) In dunite-mylonites, such as the rock from Milford Sound described in a later section, the fissility is a schistosity of cataclastic origin resulting from dimensional parallelism of elongated fragments of ruptured olivine grains. (b) More general in its occurrence is fissility governed by dimensional parallelism of elongated grains that do not appear to be of cataclastic origin. Such a fissility could be the result of flow in a partially crystalline mass, especially if the rock tends to split paralle to the trend of flow banding, as appears to be the case with Hebridean peridotites and allivalites. However, there are other possible modes of origin that might at least play an auxiliary role—namely, deformation of grains by flattening (as in the flattened quartz grains of many granite gneisses), or on the other hand elongation of grains during crystallisation, along surfaces of maximum ease of growth such as might be provided by flow surfaces in the later stages of consolidation (growth orientation controlled by Wegsamkeit). Fabric analysis should yield some information as to which of these orienting process is responsible for fissility in any given case of a fissile peridotite. Olivine Fabrics in Peridotites and Related Rocks. (a) Previously Recorded Fabric Data for Olivine. Space-lattice orientation of olivine crystals with (010) parallel to the schistosity (i.e. α perpendicular to the schistosity) was first recorded by Andreatta (1934, 1935) for olivine-rocks believed to be of metamorphic origin.* In Andreatta's diagrams there is also a much less pronounced tendency for γ to lie perpendicular to the shear-planes. The same orientation rule was confirmed by Ernst (1935) for Norwegian olivine-schists. From the even spread of the β and γ directions within the schistosity-plane Ernst (op. cit., p. 149) concluded that in olivine crystals (010) is a glide-plane but that there is no glide-line within this plane. The same writer established the existence of identical fabrics in olivine-nodules enclosed within basaltic rocks from various German localities, and cited this fact as evidence supporting the view that such nodules are not products of magmatic segregation but represent fragments of older olivine-rocks that have been caught up in the basaltic magma prior to eruption. The frequent development of translation-lamellae in the olivine of such nodules was also cited by Ernst (op. cit., p. 153) as evidence of deformation dating from some period before immersion in the basaltic magma. Measurement of grain-size in sections cut perpendicular to the fabric plane in which (010) of the olivine tends to lie, shows that the olivine-nodules described by Ernst have a slight but distinct schistosity determined by dimensional orientation of somewhat lensoid grains.

The Hebridean peridotites and allivalites described by Phillips (1938), though apparently unaffected by any deformation subsequent to intrusion, are slightly fissile rocks with distinct dimensional orientation of the component grains. There is usually strong preferred lattice-orientation, with (010) parallel to the plane of fissility just as in the rocks described by Ernst; γ is said to be evenly distributed within the plane of fissility, though one diagram (Phillips, op. cit., Fig. 2) shows a pronounced maximum within the γ-girdle. In the Hebridean peridotites, as in the case of Ernst's olivine-nodules, strong development of translation-lamellae parallel to (100) is said to be accompanied by weakening of the α-maximum and development of an α-girdle (Phillips, op. cit., p. 134). Phillips concludes that the “stresses acting during the emplacement of an olivine-rich intrusive already largely crystalline” can develop an olivine fabric identical with that previously considered diagnostic of olivine-schists of metamorphic origin; furthermore, the presence in basalts of olivine-nodules with such a fabric could be explained by assuming derivation from peridotite intrusions belonging to an earlier phase of the same igneous cycle. It may be mentioned here that some confusion has arisen from Fairbairn's (1937, p. 36) misquotation of Andreatta as finding (001) parallel to the shear-surfaces, and from a statement by Eskola (1939, p. 307) that Phillips has recorded orientation of olivine grains with (100) in the shear-planes (doubtless referring to Phillips's description of translation-lamellae parallel to that plane). (b) Fabrics Resulting from Movement of Olivine Crystals Immersed in a Liquid Medium. To determine to what extent a space-lattice orientation controlled by a primary dimensional orientation might develop by movement of olivine crystals surrounded by a liquid medium, fabric diagrams were prepared for the olivine of two basaltic rocks showing pronounced flow structure, and for a specimen of peridotite from the gravitationally differentiated sill of Lugar in Ayrshire. In these, as in most of the other fabric diagrams in this paper, the α, β and γ directions of 50 olivine grains measured in several traverses of each thin section have been plotted separately. This number is usually sufficient to demonstrate the essential features of preferred orientation in that it allows clear recognition of maxima and girdles in the diagram. Individual submaxima are probably of no significance, however, and it must be remembered that maxima which actually appear as groups of submaxima would be more completely filled in if one or two hundred grains were measured. Where, as with oriented olivine fabrics, the degree of preferred orientation is high and the pattern is simple, it is more profitable to measure in as many rocks as possible the minimum number of grains necessary to bring out the pattern of the fabric, than to investigate a few specimens in greater detail. In Figs. 1–3 the orientation of olivine in a Tertiary mugearite (P.5561† The two rocks P.5561 and P.5564 are from the collections of the N.Z. Geological Survey.) from Jeffrey's Hill, Dunedin district, is recorded. The

rock was selected as being likely to show the maximum effects of flow on orientation of olivine, since the slender laths of andesine present have a pronounced fluxional arrangement, and the olivine crystals have a simple uniformly developed prismatic habit with obvious elongation parallel to the a crystal axis (γ). The only prominent faces are those of the brachydome (021); sections parallel to a therefore have elongated rectangular outlines (0.4 mm. × 0.2 mm.) while those perpendicular to a are simple rhombs with interfacial angles of 81° and 99°. Figs. 1–3 nevertheless fail to reveal any significant space-lattice orientation of the olivine crystals, while the same holds true for diagrams prepared for a coarser mugearite (P.5564) from the nearby locality of Scroggs Hill. Separate plotting of the axes of coarser grains (0.5 mm. to 1 mm.) and those of the smaller granules (0.1 mm. to 0.3 mm.) in P.5564 also failed to bring out any trace of preferred orientation of space-lattice in either type of olivine crystal. A specimen from the hornblende-peridotite layer of the differentiated teschenite sill of Lugar, Ayrshire, was selected for fabric measurement as an example of an aggregate of olivine crystals, the fabric of which has been determined by gravitational settling and has not been affected by subsequent deformation (Tyrrell 1917, pp. 112, 125–127). The resulting diagrams, Figs. 4–6, show no indications of preferred orientation of the olivine, which makes up about 60% of the thin section and takes the form of rounded equant grains averaging 0.2 mm. in diameter. (c) Fabrics of Banded Dunites. Chromite-orpicotite-dunites with marked parallel flow banding are represented by 7494, 7495 (both from Dun Mt., Nelson), and 7496 (glen south of Alt a'Chaoich, Skye). In all three specimens black layers a few millimetres in thickness, consisting mainly of chromite or of picotite, alternate with wider olivine-rich bands in which chromite is but a sparsely distributed accessory. The Nelson dunites consist of olivine, chromite, and minor enstatite; in the rock from Skye, on the other hand, the associations olivine-chromite and picotite-plagioclase are alternately developed in adjoining bands. The grains of olivine are rounded, and for the most part slightly undulose, but the sharp lamellar structure associated with a strongly undulose condition in other peridotites here described is altogether lacking, as also is schiller structure parallel to (100). As seen in sections cut at right angles to the banding, many of the grains of olivine in 7494 and 7496 are slightly but distinctly elongated in the plane of banding of the rock, but in 7495 the grains are more nearly equant and dimensional parallelism is lacking. The fabric of the typical banded dunite of Nelson (7495) is illustrated in Figs. 7–9 based on 50 sets of measurements. The α-diagram shows a strong maximum approximately at right angles to the plane of banding (s). As is commonly the case with peripheral maxima in general, elongation of the maximum along the circumference of the projection is in the main apparent and due to elliptical distortion of small circles of the sphere as projected upon an equal-area projection. The diameters of the sectors enclosed by the 8% and 4% contours in Fig. 7 correspond respectively to angular distances of 22° and 45° measured radially, as compared with 30° and

60° measured along the circumference of the diagram. The actual tendency for the maximum of Fig. 7 to spread in an incipient girdle is therefore slight. In Fig. 9, on the other hand, there is a distinct though incomplete girdle of γ-axes, indicating a strong tendency for γ to lie within or close to s and to be concentrated especially within a limited sector in this plane. In Fig. 8 β approximates much less closely to the plane of banding than does γ in Fig. 9, and tends to fall within a poorly defined maximum corresponding to the break in the γ-girdle. From this lack of correspondence in the distribution of β and γ it would appear that orientation of the olivine crystals is governed by some more complex rule than the simple tendency for α to lie perpendicular to s that is indicated by Fig. 7. This conclusion also emphasises the advisability of recording orientation diagrams for all three indicatrix axes rather than that for α alone. Replotting of α, β and γ for selected grains whose orientation appears definitely related to s in section 7495, brings out the following facts:— (1) The area of maximum concentration in Fig. 7 contains the poles of 23 α-axes out of a total of 50 measured. (2) Out of 50 measured grains 20 have β inclined to s at angles of 20° or less; for 19 of these α falls within the maximum of Fig. 7. (3) A total of 41 grains have γ inclined to s at angles of 20° or less; for 21 of these α falls within the maximum of Fig. 7. (4) In Text-fig. 1, α, β and γ for the 23 grains having α within the maximum of Fig. 7 are plotted separately. For these grains β and γ are distributed just as within the girdles of Figs. 8 and 9. Thus the area of concentration of γ in Fig. 9 is shown to be an imperfectly filled elongated maximum, all parts of which bear the same relation to the simple maximum of the α-diagram. Text-Fig. 1. Banded chromite-dunite, 7495, cf. Figs. 7–9. Plot of α (crosses), β (dots), γ (circles) for 23 grains of olivine for which α falls within the maximum of Fig. 7. The vertical line is the projection of the plane of banding.

Taking these additional data into consideration, it is possible to state the orientation rules for olivine in the dunite 7495 as follows:— (1) α tends to lie perpendicular to the plane of banding, s. (2) γ tends to lie sub-parallel to s regardless of whether α conforms to rule (1) or not. (3) γ (and therefore β) tends to be concentrated over a limited range of direction in s. The orientation of olivine in the second banded chromite-dunite from Nelson (7494) is recorded in Figs. 10–12, which show distinct maxima for α, β and γ in mutually perpendicular directions; of these the α concentration is most clearly defined and is normal to the plane of banding. A somewhat different condition is illustrated in Figs. 13–15 which represent the orientation of 50 grains of olivine in a banded picotite-dunite from Alt a'Chaoich, Skye. In the α-diagram there is the usual maximum at 90° to s, with a tendency to elongate into an incomplete girdle; but in contrast with the two cases described above, the β-girdle parallel to s is more strongly developed than that for the γ-axes. (d) Fabrics of Fissile Non-banded Dunites. Most typical New Zealand dunites, as exemplified by specimens from the type area of Dun Mt., Nelson and the Olivine Range of South Westland are non-banded rocks in which chromite (or picotite) is of only accessory rank. The majority of the specimens seen by the writer are non-fissile rocks that break irregularly beneath the hammer, but in a few instances a slight fissility is apparent. The fabric of one of these slightly fissile dunites from Dun Mt. (7493) has been investigated and compared with those of a slightly felds-pathic fissile dunite from Alt a'Chaoich, Skye (7498) and a foliated black peridotite from Rum (7497). In the rock from Skye grains of olivine showing faintly undulose extinction but no trace of lamellar structure are locally enwrapped by small, highly allotriomorphic crystals of basic plagioclase that make up less than 1% of the total composition. The fabric (Figs. 16–18) differs but little from that of the banded picotite-dunite (Figs. 13–15) from the same locality. The α-maximum perpendicular to the plane of fissility (s) is somewhat stronger in Fig. 16 than that of Fig. 13, and both β and γ tend to be distributed evenly in equally well defined girdles parallel to s. The maxima in these β- and γ-girdles are fortuitous, for they do not correspond in partial diagrams each prepared from 25 sets of measurements. It is probable, though not necessarily the case, that banding and fissility in the dunites of Alt a'Chaoich are similar in origin and tectonic significance. In the fissile dunite from Dun Mt. (7493) the grains universally show very marked undulose extinction which in many instances has developed to the stage when the extinction bands of a super-individual are separated by sharply defined fractures (approximately perpendicular to γ). Translation-lamellae perpendicular to γ are commonly observable between crossed nicols. Both these phenomena

are more pronounced than in any other rock described in this paper. The orientation of α, β and γ for fifty distinct grains of olivine is depicted in Figs. 19–21; where, as frequently is the case, the positions of the indicatrix axes vary considerably within a super-individual, the mean position of the axis in question has been plotted as a single point on the projection. The most clearly defined feature of the olivine fabric is a sharp girdle of γ-axes in a plane inclined at a low angle to the plane of fissility (since the latter is hard to locate precisely in a hand-specimen, it is possible that the γ-girdle actually coincides with the plane of fissility). In Fig. 20 there is a very strong β-maximum perpendicular to the plane of the γ-girdle, and for the corresponding crystals α falls within the girdle of γ-axes (e.g., points adjacent to Y in Fig. 19). For other crystals, however, α is inclined to the plane of the γ-girdle at angles 60°–70° (X in Fig. 19) while β approaches to within 20° or 30° of this plane (e.g., points near X, Fig. 20). Orientation of olivine in 7493 is therefore governed by at least two principles: (1) β lies normal to s, where s is a plane slightly inclined to or possibly coincident with the megascopic plane of fissility. About half the grains measured approximate to this orientation. (2) γ lies approximately parallel to s. This rule affects all crystals, whether β is normal to s or not. To illustrate the relationship of undulose extinction in super-individuals to the fabric of the rock as a whole, the measured range of α, β and γ in the component parts of sixteen strongly undulose superindividuals in shown separately in Figs. 22–24. (The mean position of α, β and γ for each of these has already been included in the projections from which Figs. 19–21 were constructed). The strongest divergence within a superindividual is almost invariably shown by γ and the other indicatrix axis (sometimes α, sometimes β) that falls close to the γ-girdle. Furthermore, these axes tend to diverge subparallel to the girdle, so that lines joining the poles of a given axis (e.g., γ) in a superindividual trend subparallel to the projection of the γ-girdle. On the other hand, there is no observable regularity in the direction of divergence of that axis (α or β) which is steeply inclined to the γ-girdle (e.g., 2, 3, 7, 8 in Fig. 22; grains 4, 5, 6, 16 in Fig. 23). The maximum observed divergence was 30° for γ in grains 6 and 13; corresponding divergences of α and β respectively were 22° and 15° in the one case, 2° and 24° in the other (all values are regarded as correct only to ± 3°). From comparison of Figs. 19–21 with Figs. 22–24 it would appear that no profound change in fabric can be attributed solely to grain deformation that has been sufficient to cause the very marked undulose extinction recorded for olivine in this rock. The γ-girdle of Fig. 21, and its imperfect equivalent in Fig. 19, have not been strongly affected, though the attenuated maxima within the girdle of Fig. 21 may well have been weakened (or accentuated) during this phase of deformation. The extent to which the maximum of Fig. 20 may have been weakened (or emphasised) during development of undulose extinction is illustrated in Text-fig. 2. For each of the nine undulose

grains of Fig. 23 (1, 4, 5, 6, 10, 11, 14, 15, 16), that approximate to the rule β ⊥ s, two extreme positions of β have been selected and re-plotted so as to give respectively the greatest (A) and the least (B) possible scattering of points on the projection. Text-Fig. 2. Dunite, 7493, cf. Figs. 20, 23. Plot of 9 β-axes (grains 1, 4, 5, 6, 10, 11, 14, 15, 16) from the maximum of Fig. 23. Positions of β within each undulose crystal have been selected so as to give greatest (A) and least (B) possible scattering of points. Contours 5, 3, 1 points per 1% area of projection. The third example of a fissile peridotite is the “black olivine-rock” (7497) from south of Abhuinn Fiadh-innis, Rum, described by Harker (1908) and subjected to fabric analysis by Phillips (1938, p. 134 and Text-fig. 6). The writer's specimen is only poorly fissile and has a distinct lineation, the direction of which varies slightly within the plane of fissility. It consists of faintly undulose grains in which schiller structure and sharply defined parting parallel to (100) are very well shown, but which show no trace of the translation-lamellae that are so conspicious between crossed nicols in many New Zealand dunites. The grains vary in size, but have no dimensional orientation in relation to the plane of fissility. The fabric diagrams (Figs. 25–28) show a more definite preferred orientation of the space-lattice than that recorded by Phillips, though less clearly defined than, and of a different type from, that so far encountered in other Hebridean peridotites. The α-diagram (Fig. 25) shows an incomplete girdle about a point (approximating to the pole of the megascopic lineation) lying in the plane of fissility (s), and a minimum perpendicular to s; positions of the maxima within the girdle vary in partial diagrams. In Fig. 26 β tends to approach a point in s that coincides with the centre of the α-girdle. On the other hand γ (Fig. 27) appears to lack regular preferred orientation. The writer finds it difficult to reconcile this last feature with the conclusion of Phillips (op. cit., p. 134) that the peculiarities of fabric in this rock are related to the unusually perfect development of parting normal to γ, except in so far as the well developed schiller structure and poorly oriented fabric might both be connected with absence of stress after intrusion.

(e) Fabrics of Peridotites Lacking Both Banding and Fissility. Specimen 1380 (Turner 1933, p. 258) is typical of the fresh non-fissile non-banded harzburgites of South Westland. Olivine makes up more than 80% of the total composition. The grains range from 0.5 mm. to 2 mm. in diameter, and commonly show distinct undulose extinction accompanied by translation lamellae-parallel to (100) when observed between crossed nicols. No preferred dimensional orientation of grains could be detected. In spite of the absence of megascopically observable parallel fabric, the orientation of the space-lattice is strongly marked and conforms to a simple pattern (Figs. 28–30). A single strong maximum in the α-diagram tends to elongate along an arc the centre of curvature of which is the γ-maximum of Fig. 30. In the plane perpendicular to the concentration of α-axes, β and γ are restricted to clearly defined sectors (Figs. 29 and 30). As with similar fabrics recorded for banded dunites, the maximum for γ is more sharply defined than that for β, and the tendency for γ to lie in the plane perpendicular to the α-maximum is more marked than is the case with β (cf. with Figs. 7–9). Measurement of the range of α, β and γ in each of three large, strongly undulose crystals gave 20°–25° as the maximum divergence of any indicatrix axis; no regularity in direction of divergence was noted. A representative of the Nelson dunites lacking banding and fissility is 7505, which is composed of fresh equant grains 0.1 mm. to 0.5 mm. in diameter with strongly developed undulose extinction and translation-lamellae normal to γ. The section is traversed by many cracks of highly variable orientation, but there is one set of curving subparallel cracks that are particularly prominent, continuous and constant in trend, and apparently independent alike of the orientation of the crystals through which they pass and of the presence of undulose extinction or translation-lamellae in these. The space-lattice orientation of the olivine crystals is illustrated in Figs. 31–33. This is the only instance, among the rocks investigated by the writer, of a fabric in which the α-axes fall in a sharply defined nearly complete girdle. There is clearly a tendency for β (Fig. 32) to occupy the same girdle, though the orientation is less sharp than in the case of the α-girdle. In Fig. 33 γ is strongly concentrated in a single maximum perpendicular to the girdle of α and β. In the absence of macroscopically visible s-surfaces the fabric might represent a single set of s-surfaces to which γ is normal, or a B-tectonite fabric with γ parallel to the B-axis. The presence of abundant supparallel cracks inclined at high angles to γ strengthens the latter alternative, if it is assumed that the cracks in question are approximately parallel to the ac fabric plane, the commonest orientation for tension-fractures in B-tectonites. The third example of a peridotite without megascopically visible s-surfaces is a dunite (5308) from the vicinity of Bluff, previously described by H. Service (1937, pp. 207, 212, no. 208). The rock occurs as a small mass, possibly a late dyke, associated with the well-known norites that occur extensively in this region. The thin section consists of equant grains of olivine 1 mm. to 5 mm. in diameter usually showing pronounced schiller structure parallel to

(100). In contrast with the West Coast dunites the Bluff rock appears not to have been affected by stress since consolidation, for not only is there no trace of undulose extinction in the olivines, but the serpentine partially replacing the olivine grains is chrysotile with mesh-structure instead of the antigorite almost universally present in serpentinised peridotites from the West Coast. The fabric (Figs. 34–36) is poorly defined and somewhat resembles that of the black peridotite from Rum illustrated in Figs. 25–27. In the α-diagram is a vaguely defined maximum (present in each of two partial diagrams which are combined into Fig. 34), while β clearly tends to lie in a girdle perpendicular to the α-maximum. On the other hand there is no preferred orientation of γ (Fig. 36). (f) Fabric of Cataclastic Dunite. The peridotites described by Marshall (1905) from Anita Bay, Milford Sound, include a steeply dipping sill of dunite interbedded with amphibolites of high metaphoric grade. The dunite (7491) shows no trace of serpentinisation, but has evidently undergone extreme cataclasis, which has reduced the grain-size to between 0.03 mm. and 0.05 mm. and imparted a distinct fissility to the hand specimen. Sections cut perpendicularly to the plane of fissility (s) show that this structure is determined partly by dimensional parallelism of the larger elongated granules and partly by a tendency for subparallel streaks of coarser and finer material to alternate. There is also a tendency for scattered granules of chromite to lie in discontinuous strings parallel to s. The individual grains of olivine in the writer's sections are free from schiller structure, translation-lamellae and undulose extinction, nor is there any tendency for grouping of grains into superindividuals. In sections of the same rock described by Marshall (op. cit.) undulose extinction and granulation are recorded in residual larger grains enclosed in the fine-grained matrix. The fabric of 7491 is illustrated in Figs. 37–39, based on measurements of 50 grains in several widely separated traverses across a single thin section. Complete lack of orientation of the space-lattice is attributed to the destructive process of cataclastic as opposed to plastic deformation of other rocks. (g) Fabrics of Olivine-nodules Associated with Basaltic Rocks. Three New Zealand specimens of olivine-nodules from widely separted localities and different geological environments have been selected for fabric analysis, and the results have been compared with data obtained from a nodule (7499) from Prussia. The first example (7503) is typical of the numerous small nodules (3 cm. in diameter) that are scattered among the basaltic lapilli of a Pleistocene tuff-cone, Onepoto, Shoal Bay, near Auckland city. In comparison with associated nodules, 7503 is a somewhat finegrained rock (0.5 mm. to 1 mm.) with no trace of dimensional orientation of the constituent grains of olivine. Many crystals show well defined undulose extinction, with sharp ruptural boundaries between adjacent sectors of any individual, but translation-lamellae were not observed. A little enstatite (2V = 80°; sign +) and nearly opaque chromite are also present. The fabric diagrams (Figs. 40–42) show pronounced preferred orientation of a somewhat unusual type in

that the only orientation rule is for the α-axes to fall into a well defined girdle indicated by the broken line in Fig. 40. The β- and γ-diagrams show poorly developed submaxima which appear to have no significance. The slight concentrations of β and of γ in the vicinity of the normal (X) to the α-girdle are no greater than can be accounted for by the fact that × is a possible position for any direction in the βγ plane of all crystals for which α lies in the girdle of Fig. 40. The second investigated specimen (5100; Figs. 43–45) is a large nodule (10 cm. in diameter) from the Pliocene basalt of Kokonga, in the outer portion of the East Otago petrographic province. Nodules are rare in the basalts of this region. The rock has the composition of olivine-rich lherzolite containing some enstatite, diopsidic augite and picotite in addition to the major constituent. The grain-size varies from 0.5 mm. to 3 mm., and there is no obvious preferred dimensional orientation of grains. Translation-lamellae were not observed, but undulose extinction is conspicuous in many of the larger crystals of olivine. At first glance it would appear that the grains plotted in Figs. 43–45 lack any significant preferred orientation; and this conclusion is strongly borne out by applying the statistical test described by H. Winchell (1937, pp. 21–27), for the distribution of points in any one diagram agrees almost exactly with a theoretical random distribution.‡ A circular net divided into 148 equal squares was superposed upon the “scatter diagram” of axial points. In the case of the β diagram (Fig. 44) the number of squares containing 0, 1, 2 and 3 points respectively was found to be 106, 39, 4 and 1. Corresponding figures for a random distribution are 106, 36, 5, 0. The calculated probability index, P, = 0.45. Nevertheless it will be noted that the three poorly defined maxima marked × in Figs. 43–45 represent three mutually perpendicular directions in the fabric, and it is a simple Fig. 3. Text-Fig. 3.—Olivine-nodule, 5100, cf. Figs. 43–45. Plot of α (crosses), β (full circles), and γ (open circles) for eight grains having β close to the maximum × of Fig. 44. Shaded areas show the positions of maximum × for α, β and γ respectively. Fig. 4. Text-Fig. 4.—Olivine-nodule, 6655, cf. Figs. 46–48. Plot of α (crosses), β (full circles) and γ (open circles) for 13 grains having β close to maximum × of Fig. 47. Solid triangles indicate points marked × on Figs. 46–48; broken lines show small circles 25° from X.

matter to test whether or not this is fortuitous, by re-plotting the three axes α, β and γ for each of the crystals (eight in all) in which β lies close to × in Fig. 44 (Text. fig. 3). For each of these crystals it is found that both α and γ closely approach maximum × in their respective diagrams, and a slight but nevertheless significant preferred orientation involving 16 per cent. of the measured grains is in this way established. On the other hand, when α, β and γ are re-plotted for an additional 13 grains, having γ close to × in Fig. 45, the corresponding α and β directions are well scattered within the girdle perpendicular to the γ-maximum, and approach maximum × (though not closely) in three cases only. The third case is that of a typical large nodule (6655) from the Oligocene marine breccia of Kakanui, North-East Otago. The breccia is of pyroclastic origin and consists of fragments of tachylite and basalt with which are associated plentiful coarse crystals of brown hornblende and large nodules of peridotitic composition, and a variety of rocks consisting of such minerals as garnet, hornblende, augite, biotite, etc. (Thomson 1907). Biotite-schists and other meta-morphic rocks are also represented in the breccia. The nodule 6655 is a lherzolite containing abundant picotite, and petrographically resembles the rock from Kokonga just described, except that the grain is rather coarser in 6655. Undulose extinction is conspicuous and translation-lamellae quite lacking in the grains of olivine. The orientation of all available grains (35 in number) is shown in Figs. 46–48, each of which has a rather vaguely defined concentration of axes round a single point X. By re-plotting α, β and γ for 13 crystals whose β axes approach parallelism with × (Text-fig. 4), it is found that for 11 of these grains α and γ lie not far distant from maxima × of Figs. 46 and 48 respectively. It is therefore concluded that the grains of olivine tend to some extent to develop a preferred orientation in which α, β and γ all occupy constant positions in the rock fabric. The fabric of the olivine nodule (7499) enclosed in basalt from Prussia is not figured here since it conforms to the descriptions given by Ernst (op cit.). There is a strong concentration of α-axes (maximum = 18% per 1% area, in 50 measured grains) with a β-γ-girdle perpendicular to this. It should be noted, however, that within this there are definite mutually perpendicular maxima (10%–12%) for β and γ respectively. Conclusions. The conclusions put forward below are based on the petrofabric studies embodied in this paper, taking into account also the findings of the previous authors cited. Further work of the same nature will be necessary to verify the extent to which they may be applied, or have to be modified, in interpreting olivine fabrics in general. (1) Flow in basic lavas containing suspended prismatic olivine crystals fails to produce preferred space-lattice orientation of olivine (Figs. 1–3) even though it may have given rise to pronounced fluxional arrangement of the accompanying laths of feldspar. This raises the possibility that dimensional parallelism of the feldspars in volcanic rocks may be governed by growth of crystals in the surfaces of flow of the viscous magma according to Sander's principle

Figs. 1–3. Olivine (50 grains) in mugearite, P.5561. Contours 8, 4, 2%. Broken line indicates the mean trend of elongation (γ) of plagioclase laths, within the measured section. Figs. 4–6. Olivine (50 grains) in hornblende-peridotite, Lugar sill. Contours 8, 4, 2%.

Figs. 7–9. Olivine (50 grains) in banded chromite-dunite, 7495, Dun Mt. Section perpendicular to trend of bands (broken line). Contours, 12, 8, 4, 2%. Figs. 10–12. Olivine (50 grains) in banded chromite-dunite, 7494, Dun Mt. Section perpendicular to trend of bands (broken line). Contours, 12, 8, 4, 2%.

Figs. 13–15. Olivine (50 grains) in banded picotite-dunite, 7496, Glen South of Alt a'Chaoich. Skye. Section perpendicular to trend of bands (broken line). Contours, 12, 8, 4, 2%. Figs. 16–18. Olivine (50 grains) in fissile dunite, 7498, Glen south of Alt a'Chaoich. Skye. Section parallel to plane of fissility. Contours, 12, 8, 4, 2%.

Figs. 19–21. Olivine (50 grains) in slightly fissile dunite, 7493, Dun Mt. Section approxi-mately parallel to fissility. Contours, 12, 8, 4, 2%; maximum concentration, 16% in Fig. 20. Figs. 22–24. Olivine (16 undulose grains) in fissile dunite, 7493, Dun Mt. Dots connected by straight lines represent observed positions of the indicatrix axis in any one crystal.

Figs. 25–27. Olivine (50 grains) in black olivine-rock, 7497, south of Abhuinn Fiadh-innis, Rum. Section perpendicular to the plane of fissility (broken line) and approximately perpendicular to lineation. Contours. 12, 8, 4, 2%. Figs. 28–30. Olivine (50 grains) in harzburgite, 1380, Olivine Range, South Westland. Contours 12, 8, 4, 2%; maximum concentration 20% at × in Fig. 30; 16% in Fig. 28 and at Y in Fig. 30; 14% in Fig. 29.

Figs. 31–33. Olivine (50 grains) in dunite, 7505, Dun Mt. Nelson. Contours 12, 8, 4, 2%; maximum concentration, 16% in Fig. 33, cc = mean trend of most regular fractures. Figs. 34–36. Olivine (50 grains) in dunite, 5308, near Bluff, Southland. Contours, 12, 8, 4, 2%.

Figs. 37–39. Olivine (50 grains) in cataclastic dunite (7491), Anita Bay, Milford Sound. Section perpendicular to plane of fissility (broken line). Contours, 8, 4, 2%. Figs. 40–42. Olivine (50 grains) in nodule (7503) in basalt, Onepoto, Shoal Bay, Auckland. Contours, 8, 4, 2%. Maximum concentration in each case 10%. × = normal to the girdle (broken arc) of Fig. 40.

Figs. 43–45. Olivine (50 grains) in nodule (5100) in Pliocene basalt, Kokonga, East Otago. Contours, 8, 4, 2%. Maximum concentration in each case, 10%. Figs. 46–48. Olivine (35 grains) in nodule (6655) in tachylite breccia, Kakanui. North-East Otago. Contours, 6, 4, 2, 1 points per 1% area.

of Wegsamkeit, as well as by rotation of pre-existing crystals into the same surfaces. (2) Settling of olivine crystals under the influence of gravity as they separate from basic magma under static conditions also fails to produce preferred orientation of the olivine space-lattice (Figs. 4–6). (3) The megascopically conspicuous regular banding of certain fresh chromite-rich dunites is believed to have originated during flow of a “magma” in an advanced state of crystallisation. A feature of the fabric, constantly associated with banding, is preferred orientation of olivine governed by external form of grain, in that individual crystals tend to be elongated parallel to the banding. Therefore it is probable that this dimensional orientation and any consequent preferred orientation of the space-lattice are also determined by flow prior to complete solidification of the magma, rather than by movement and actual deformation of grains after the mass had become completely crystalline. A high degree of penetrative intergranular movement, such as would necessarily affect a largely crystalline dunite “magma” in process of injection as pictured by Bowen (1928, p. 167), would be expected to give rise to preferred orientation of grains according to their external form, and this would also involve orientation of space-lattice if the grains in question conformed uniformly to a well defined crystallographic habit. The usual habit of magnesian olivines is a prismatic form flattened normal to b (α) and elongated parallel to c (β). Laminar flow of the largely crystalline mass (with its lubricating interstitial liquid) along approximately plane surfaces (such as is indicated by the regular subparallel layers of the banded dunites) should give rise to a fabric the dominant feature of which would be a strong concentration of α-axes normal to the plane of flow (Figs. 7, 10, 13). A secondary feature, depending upon elongation of the crystals parallel to β, is the tendency for β to lie in the flow-surface even in grains for which α departs from the ideal orientation, so that the β-girdle perpendicular to the α-maximum is more clearly defined than is the γ-girdle. This condition is shown in Figs. 14, 15. The opposite tendency for the γ-girdle to be more pronounced than the β-girdle, shown by one of the two banded dunites from Dun Mt. (Figs. 8, 9), cannot be explained in terms of the above-described mechanism, and is attributed to later deformation. (4) In the absence of visible banding in dunites poor in chromite, there is often no way of determining whether a simple orientation pattern, with a high concentration of a-axes and a fairly even spread of β and γ in the girdle normal to this, has originated by laminar flow prior to complete solidification of the magma or by deformation of the already solid mass. If magmatic flow is the controlling process, the space-lattice orientation should be accompanied by dimensional orientation of crystals with their long axes aligned in the surfaces of flow, and distinct fissility perpendicular to the α-maximum might be obvious in the hand-specimen. However, according to Ernst (op. cit.) dimensional orientation of olivines elongated at right angles to α is also conspicuous in schists whose fabric is of deformational origin. (5) The fabrics of three Hebridean peridotites have been described in this paper. One (7496) is a banded type with a simple

fabric believed to have been determined by magmatic flow as discussed under (3) above. The second is a fissile non-banded rock (7498, Figs. 16–18) with a closely similar fabric which is therefore attributed to the same general process. The third rock (the black peridotite, 7497, from Rum) has a rather weakly defined fabric of quite a different type, the main features of which are an α-girdle normal to the plane of fissility, and a β-maximum perpendicular to the α-girdle and approximately parallel to a weak megascopic lineation (Figs. 25–27). This pattern is just what would be expected if, during flow of a largely but not completely crystalline mass, the crystals rotate with their long axes (β) aligned in the flow surface and normal to the direction of flow. The presence of a megascopic lineation strongly supports the possibility that such a mechanism has indeed controlled development of the fabric in this particular instance. Of the four Hebridean peridotites investigated by Phillips (1938) two have simple fabrics dominated by an α-maximum (Phillips, op. cit., 3, 4, 7, 8); in one other the α-maximum and β-γ-girdle are accompanied by a definite concentration of γ in the latter (Figs. 1, 2); the fourth fabric shows no preferred orientation of olivine (Fig. 6). Both Phillips' results and those of the writer may be explained on the assumption that the fabrics of Hebridean peridotites are due to flow of largely but not completely crystalline olivinerich magmas, and have not been influenced to any extent by post-crystallisational deformation, of the rocks concerned. This is much the same conclusion as was arrived at by Phillips (op. cit., p. 134), except that the present writer would emphasise his opinion that deformation of crystals has played little or no part in development of the observed preferred orientation. (6) The great steeply inclined dunite sills of Nelson and South Westland have been injected along zones of major dislocation during profound orogeny. The rocks concerned may therefore be expected to bear the imprint of post-crystallisational deformation superposed upon a fabric determined by flow prior to complete consolidation. The two effects will often be difficult to separate since both owe their origin to the same set of deforming forces acting upon the intrusive mass. (a) In specimens 1380 and 7505 there is strong preferred orientation of the olivine space-lattice but no recognisable dimensional orientation, so that the fabrics may be assumed with some confidence to have been determined mainly by deformation after complete solidification of the rocks in question. The orientation pattern of 1380 (Figs. 28–30) is marked by strong mutually perpendicular concentrations of α, β and γ respectively. The α-maximum is most pronounced and is somewhat elongated in a plane perpendicular to the mean trend of γ. If, as appears likely both on theoretical grounds and from fabric studies of Andreatta and of Ernst on olivine-tectonites, α tends to be oriented at right angles to the principal slip-plane of the fabric, then the direction in which γ is concentrated is the b fabric axis about which an incipient α-girdle is in process of development. This would point to β (crystal axis c) as the preferred glide-direction within a glide-plane (010). This conclusion is borne out by the fabric of 7505 (Figs. 31–33) in which the α-girdle is fully developed in a plane normal to a strong γ-maximum; in this

rock, too, micro-fractures inclined at high angles to the mean trend of γ (b fabric axis) are conspicuous and could be interpreted as ac fractures such as are commonly present in girdle-tectonites. (b) The following sequence of events is suggested in explanation of the fabrics of 7494 (Figs. 10–12) and 7495 (Figs. 7–9):—Laminar flow in the still partially liquid peridotite “magma” gives rise to a dimensional orientation of olivine crystals with which a strong concentration of α-axes normal to the flow-surfaces (s) may be correlated. As deformation of the intrusive mass continues after it has completely solidified, slip-movements develop principally in the original flow-surfaces, with which, moreover, the best marked crystallographic glide-plane (010) already coincides in many of the grains of olivine. However, before gliding can occur in any given grain it is necessary that the grain should rotate until the preferred glide-direction (β) is brought into approximate coincidence with the direction of slip-movement (a fabric axes) for the rock as a whole. Therefore one of the component movements accompanying the earliest stages of deformation of the solid rock is a rotation of both β and γ about the normal to s, with resultant concentration of β and γ in two mutually perpendicular directions in s. This condition is exemplified by Figs. 10–12, the fabric in this case being more weakly defined than is usual in West Coast peridotites. A further complication in the fabric would be introduced if there were several sets of slip-surfaces inclined at low angles to s, or if the latter were somewhat curved by external rotation about the b axis of the fabric. Both possibilities are commonly encountered in deformed rocks and would have the same effect upon the orientation of the olivine space-lattice—namely, spreading of the α-and β-maxima in the plane normal to the b axis of the fabric (direction of concentration of γ). A strongly developed fabric of this type is shown in Figs. 7–9, in which it will be noticed that whereas the γ-maximum is elongated within s, the β-maximum is much less distinct and is dispersed not only in s but also in the plane of elongation of the α-maximum. (e) The strongly defined unique orientation pattern of Figs. 19–21 cannot be related to the megascopic fissility-planes of the rock in question on any such hypothesis as those put forward above. It would seem that the fissility is due to persistence of original surfaces of magmatic flow, and that the space-lattice orientation is the expression of a subsequent deformation under a new set of kinematic conditions. The unusually sharp γ-girdle is interpreted as a B-girdle such as is commonly developed in many tectonite fabrics, and the β-maximum on this assumption represents a concentration of β-axes parallel to B. Further interpretation is somewhat speculative, for more than one gliding mechanism could give rise to an orientation pattern of this one type, e.g.: (1) glide-plane = (010) (i.e., ⊥ α), glide-direction = γ; (2) glide-plane = (100) (i.e., ⊥ γ), glide-direction = α. On account of the sharpness of the γ-girdle and wide range of orientation of both α and β, gliding parallel to γ (theoretically the most probable glide-direction in olivine) in several crystallographic glide-planes, of which (010) is specially favoured, would appear the most likely orienting mechanism.

(7) In most peridotites having orientation patterns showing what is believed to be the influence of deformation subsequent to complete solidification [cf. (6) above], undulose extinction is conspicuous in the constituent grains of olivine. Since this property is considered to be an expression of combined flexural gliding and incipient rupture of the space-lattice, its origin is almost certainly bound up with the process of deformation by which the orientation pattern was imprinted upon the fabric as a whole. It has been suggested above that in many of the investigated rocks olivine grains, originally oriented by magmatic flow so as to bring α normal to the flow-surfaces (s), have subsequently rotated until β approaches the principal direction of slip for the deforming rock. As long as the opposing force of friction on the intergranular boundary surfaces did not exceed a limiting value, this rotation could be in the nature of differential movement of the individual grains. Once the friction exceeded this limiting value, intragranular flexural gliding on space-lattice surfaces inclined at high angles to s (the principal slip-surfaces of the rock fabric) could achieve the same result. For example, in the case of a grain with α approximately normal to s, flexural gliding on (001) with γ (crystal axis a) as glide-direction could rotate both β and γ within s. Deformation of the individual grains is therefore pictured as a complex process involving flexural gliding on (001) with γ as glide-direction, rupture and some differential movement on (100), and finally (when β is brought sufficiently close to the principal glide-direction for the fabric as a whole) translation-gliding on (010) parallel to β. It has long been recognised that some such process as flexural gliding, or a combination of gliding movements on two mutually perpendicular sets of crystallographic glide-planes, must be assumed in order to account for the high degree of preferred orientation of minerals that is commonly attained during comparatively slight deformation that cannot have involved extensive rotation of individual grains (e.g., cf. Schmidt, 1932, pp. 173–176; Eskola, 1939, pp. 297, 298, 304, 305). Finally attention is drawn to the single case (7493, Figs. 19–21) where gliding parallel to γ on various crystallographic planes, but especially on (010), is thought to have been the most influential process in development of the olivine fabric; as might be expected, undulose extinction is here more strikingly shown by the olivine grains than in any other rock investigated. (8) When under high confining pressure a rock is strained beyond the elastic limit, it may deform plastically by combined plastic deformation of individual grains and recrystallisation through the agency of aqueous pore-solutions. In this way a cooling mass of peridotite, intimately penetrated by magmatic waters containing silica or carbon dioxide, may by continued deformation be converted to a schistose antigorite-serpentine. In the absence of water, however, recrystallisation of olivine or crystallisation of new minerals like serpentine or talc appears to be impossible. And when deformation advances beyond the stage where internal adjustment to the stress system can be achieved by movements on crystallographihc glide-planes and intergranular boundary surfaces, rupture of the olivine grains and differential movement of the fragments so produced become the essential factors in a deformation that may now be described as cataclastic. The ultimate product of a process of this sort is a fine-grained cataclastic dunite such as 7491, with a distinct

schistosity (s-planes of shearing) marked by alternation of bands of different grain-size and by dimensional parallelism of the elongated crystal fragments. Lack of preferred orientation of the olivine space-lattice is a characteristic feature of the fabric (Figs. 37–39). (9) Mineralogically the olivine-nodules from New Zealand Tertiary and Pleistocene basalts are identical with lherzolites and harzburgites of plutonic origin. The constituent minerals include enstatite, chrome-diopside, picotite and chromite, none of which has been recognised in the enclosing basalts. In two of the three specimens used for fabric analysis (5100, 6655; Figs. 43–48) the orientation pattern, though very weak, conforms to a type here attributed to deformation of a solid rock, while the third fabric (7503; Figs. 40–42) with its pronounced α-girdle and absence of dimensional orientation of grains must have originated in a like manner. This conclusion is supported by prevalence of undulose extinction in all three cases. It is therefore suggested that all three nodules are fragments torn from masses of solid peridotite, which have formed under plutonic conditions but have never flowed in the form of injected bodies of largely crystallised peridotite magma. If these parent peridotites were formed as accumulations of early crystals separating from the primitive basaltic magma at sufficient depth for the conditions of temperature and pressure to be quite different from those obtaining during later crystallisation of the basalts themselves, the load of overlying rock and magma might well be sufficient to give rise to the weak deformation recorded in the fabric. The European nodules described by Ernst (op. cit.) on the other hand include rocks with strongly developed fabrics in which well developed girdles appear to indicate deformation of a more pronounced nature, especially in that the olivine of such rocks often shows strongly developed undulose extinction and translation-lamellae. The petro-fabric evidence, taken in conjunction with the widespread occurrence of olivine-rich nodules in basalts in many parts of the world, points to deep-seated solid accumulations of olivine crystals and in some cases perhaps intrusions of peridotites as the source of the nodules in question. As Phillips has pointed out (op. cit., p. 134) it is not necessary to assume the existence of a deep-seated metamorphic terrane as a source of olivine-nodules. Nor, however, can they be pictured as mere local aggregations of crystals separating from the basaltic magma just prior to eruption. (10) The following characters of peridotite fabrics are considered as criteria of origin by flow in a largely but not wholly crystalline magma: s-planes marked by alternation of layers respectively rich and poor in chromite, or by parallel orientation of elongated olivine grains; concentration of α normal to s-surfaces of flow, and fairly even spread of β and γ within the plane of flow, the β-girdle sometimes being sharper than the γ-girdle; any concentration of β or γ within the plane of flow must be accompanied by microscopic or megascopic lineation due to dimensional parallelism of elongated grains as seen in sections parallel to the s-planes in question. (11) Characteristic criteria of origin of a peridotite fabric by deformation of the solid rock include the following: presence of a complete or incomplete α-girdle at right angles to any visible flow

surfaces, accompanied by concentration of γ normal to the α-girdle and parallel to the b axis of the fabric; less commonly presence of simple mutually perpendicular maxima for α, β and γ respectively; occasionally development of a γ-girdle at right angles to a β-maximum which is parallel to b of the fabric; conspicuous undulose extinction and sometimes translation-lamellae normal to γ; cataclastic structure, schistosity due to shearing, dimensional parallelism of crystal fragments and lack of preferred orientation of the olivine space-lattice. Acknowledgement. My thanks are gratefully recorded to the Council of the Royal Society of New Zealand for a research grant from the Hutton Fund to defray expenses involved in this research (including part cost of publication), and to Professor J. A. Bartrum, who kindly supplied specimens of olivine-nodules from Auckland. Literature Cited. Andreatta, C., 1934. Analisi strutturale di rocce metamorfiche. V. Olivinite. Periodico di Mineralogia, Anno V, pp. 237–253. —– 1935. La formazione gneissico-kinzigitica e le oliviniti di Val d'Ultimo. Mem. del Museo di Storia Naturale delle Venezia Tridentina, iii, fasc. 2. Bowen, N. L., 1928. The Evolution of Igneous Rocks. Princeton University Press. Bragg, W. L., 1937. Atomic Structure of Minerals. Oxford University Press. Buerger, M. J., 1930. Translation-gliding in Crystals. Amer. Min., vol. 15, pp. 1–47. Ernst, T., 1935. Olivin knollen der Basalte als Bruchstücke alter Olivinfelse. Nachr. Ges. Wiss. Gottingen, Math-Phys. Kl., Gruppe IV, Bd. 1, Nr. 13. pp. 147–154. Eskola, P., 1939. Die Metamorphen Gesteine. Die Entstehung der Gesteine (T. F. Barth, C. W. Correns, P. Eskola), Berlin, J. Springer. Fairbairn, H. W., 1941. Deformation lamellae in Quartz from the Ajibik Formation, Michigan. Bull. Geol. Soc. Amer., vol. 52, pp. 1265–1278. Harker, A., 1904. The Tertiary Igneous Rocks of Skye. Mem. Geol. Surv United Kingdom. —– 1908. The Geology of the Small Isles of Inverness-shire. Mem. Geol. Surv. Scot. —– 1919. Petrology for Students, 5th ed. Cambridge University Press. Hess, H. H., 1938. Primary Banding in Norite and Gabbro. Trans. Amer Geophys. Union, Vulcanology, pp. 264–268. Hietanen, A., 1938. On the Petrology of the Finnish Quartzites. Bull. Com geol. Finlande, no. 122. Iddings, J. P., 1906. Rock Minerals. New York, Wiley. Knopf, E. B., 1938. Structural Petrology, Part I. Mem. Geol. Soc. Amer., no. 6. Marshall, P., 1905. Magnesian Rocks at Milford Sound. Trans. N.Z. Inst., vol. xxxvii, pp. 481–484. Phillips, F. C., 1938. Mineral Orientation in Some Olivine-rich Rocks from Rum and Skye. Geol. Mag., vol. lxxv, pp. 130–135. Rynearson, R. A. and Smith, C. T., 1941. Chromite Deposits in the Seiad Quadrangle, Siskiyou County, California. U.S. Geol. Surv. Bull., 922-J. Sander, B., 1930. Gefügekunde der Gesteine. Vienna, J. Springer. Schmidt, W., 1932. Tektonik und Verformungslehre. Berlin, Borntraeger. Service, H., 1937. An Intrusion of Norite and Its Accompanying Metamorphism at Bluff, New Zealand. Trans. Roy. Soc. N.Z., vol. 67, Pt. 2, pp. 185–217. Thomson, J. A., 1907. Inclusions in some Volcanic Rocks. Geol. Mag., vol. iv, pp. 490–500. Turner, F. J., 1933. The Metamorphic and Intrusive Rocks of Southern Westland. Trans. N.Z. Inst., vol. 63, Pts. 2 and 3, pp. 178–284. Tyrrell, S. W., 1917. On the Picrite-Tcschenite sill of Lugar, Ayrshire. Q.J.G.S., vol. lxxii, pp. 84–131. Winchell, H., 1937. A new method of Interpretation of Petrofabric diagrams. Amer. Min., vol. 22, pp. 15–34.