The portion of Treble Mountain massif that extends from Isthmus Sound towards Bald Hills is made up of a very coarse-grained granite with pink orthoclase crystals up to 2 cm. in length, white plagioclase, clear quartz, and flakes of biotite. It would make a most handsome building stone, and could readily be quarried alongside sheltered bays, as Marshall (1929) has pointed out. This rock was described by Hutton (1899) as a syenite, but corrected accounts have been given by Marshall (1907) and Speight (1910). Slices of a specimen from Isthmus Sound (1658) and of an erratic from Te Oneroa, Otago's Retreat (1603), show that orthoclase is the chief constituent, forming 35–40% of the rock in large irregular perthitic prismoids. Plagioclase is almost as abundant in irregular grains up to 6 mm. across and in rather smaller prismoids, which are markedly zoned and apparently a little more basic than in the rock last described, for they range from Ab₅₅ An₄₅ to Ab₆₅ An₃₅ with a marginal film of Ab₇₈ An₂₂. Quartz makes up nearly 20% of the rock in grains sometimes 6 mm. across and slightly strained. There is a little interstitial microcline, amounting to 5% or less, and biotite is still less abundant, though occasionally forming plates as much as 2 mm. across. There are a few minute prisms of apatite and a little secondary epidote and sericite. According to Hector (1864) the rock of Red Head, north of Gulches Head, is of a similar character, though with a finer base, and so also is the granite from Green Islets described by Professor Park (1926), the analysis of which is quoted below.

Quartz-biotite-syenite.

A pebble (1558) from the east side of Landing Bay, near Cape Providence, appears as a modification of the last type of granite, for quartz occurs in relatively small amount and usually in shatter-zones. This syenite invades a richly hornblendic rock and consists chiefly of microperthitic orthoclase in crystals sometimes as much as 5 mm. in diameter, with irregular grains of quartz and oligoclase and rarely hornblende and biotite. The rock has been intensely sheared, yielding shatter-zones composed of crushed feldspar and quartz in sutured, ribboned, intensely strained grains, and myrmekite has developed in the feldspars adjacent thereto, or the orthoclase is replaced by microcline (Pl. 23, Fig. 2). Plates of biotite are drawn out into ropes in these shearing zones and are there associated with apatite, magnetite, sphene, and a little hornblende, while thin films of epidote occur rarely in the cleavages of the feldspar. Both as regards mineral composition and granulated structure, this rock is very similar indeed to a gneissic rock (D.6) obtained by Professor Speight (1910) at Pickersgill Harbour near the southern entrance to Dusky Sound.

Granodiorites (or Soda-tonalites).

The bulk of the plutonic rocks studied belong to this group, which is represented by 1689, a rather gneissic rock on the margin of the Kakapo massif, two miles south-west of the head of Edwardson Sound, by specimens from the vicinity of Northport (1528, 1573, 1587, 1591, 1619, 1643, 1651), by a very narrow selvedge between the greywacke and the invading granite in Brokenshore Bay (1679),

and by the rocks from the middle portion of Long Sound, namely 1564, 1566 from its northern shore and 1507, 1562, 1565, 1575, and 1647 from its southern.

The characteristic minerals of these granodiorites are oligoclase, quartz, and biotite, with rarely subordinate amounts of potashfeldspar, muscovite or hornblende and accessory apatite, zircon, iron ore, sphene, pyrites, epidote, and rarely allanite. The rocks are either massive or gneissic, sometimes being crushed with granulation or elongation of the quartz crystals in the direction of foliation. The grain-size varies between 1 mm. and 4 mm., and is usually about 2 mm. Plagioclase forms 50% to 75% of the rock and varies in composition from Ab₆₅ An₃₅, or more often Ab₇₀ An₃₀, to Ab₈₀ An₂₀. It occurs in irregular, equant grains or elongated prisms in which zonal structure is but slightly marked, whilst the grains are usually more or less clouded by minute scales of sericite. Quartz forms from 15% to 35% of the rock, averaging about 30%, the lowest figure being reached only in a highly feldspathic rock (1564). It occurs in interstitial grains moulded on feldspar or in aggregates of sutured individuals usually showing marked strain-effect. Potash-feldspar occurs only in small amount. It forms prisms of moirée orthoclase-perthite in a rock on the margin of the intrusion south of Only Island in Long Sound (1507) and appears as microcline in a sheared rock (1647) invading diorite two miles north-east of the last, in 1566 on the north shore of Long Sound immediately opposite to 1647, in 1679 from Brokenshore Bay, in 1643, a small dyke invading the diorite at Northport, and in 1619, a dyke intrusive into the schists near Mosquito Point, Northport. The presence of myrmekite along the margin of the plagioclase in the crushed portions of the rock is exhibited in 1566, 1619, and 1647. Biotite occurs in amounts varying from 5% to 25%, the latter in a dark rock (1643) invading diorite at Northport. The biotite in 1566 and most of the rocks of this type adjacent to Northport, where they are found north and west of Mosquito Point and in Great Island, has as its darkest tint a rather greenish brown; that in the most micaceous of the Northport rocks (1528) and in all those of Long Sound is at darkest a deep-red-brown. Its flakes are rarely as much as 1.5 mm. long and only rarely show sieve-structure (1651). They are more or less parallel in the gneissic types, and in the sheared ones may be bent and streaked out along the zones of shearing. They are slightly chloritised even in the freshest rocks, and in the most altered (1564) are completely replaced by pennine with the separation of rutile in imperfect sagenitic form. In 1564 the chlorite is almost colourless, its former content of iron having been removed save for the little that remains as siderite and limonite in its cleavage planes, while some muscovite occurs among the chlorite. In other rocks, e.g., 1691 from Northport, the pennine is apple-green and strings of sphene have formed in its cleavage cracks. Muscovite occurs in certain of the granodiorites irrespective of the area from which they come, but is absent from others. When present it is usually in amounts less than 5%, though it may make quite large flakes up to 3 mm. in diameter, upon which the biotite flakes are often moulded. In a group of peculiar rocks (1507, 1512, 1565,

1575) from the margin of the granodiorite half a mile south of Only Island, there has been so intimate a lit-par-lit injection of igneous rocks into the quartz-mica schists that it is difficult to make a distinction between the two sets of rocks, and, as a result, there is an abundance of extremely irregular pseudopoikilitic or else confusedly-matted sericite in portions of rock which otherwise have a granodiorite composition. This feature has been discussed at greater length in describing the injection-schists.

Text-fiq. 2.—Crystal of allanite surrounded by epidote (dotted) producing dark halo in enclosing biotite. Granodiorite (1528), Northport.

Hornblende is present in a single gneissic rock (1651) which in invaded by a vein of aplite, and was collected as a pebble in Shallow River at Northport. It forms less than 3% of the rock in grains usually less than 1 mm. long, which rarely show sieve-structure and are moulded on or intergrown with biotite. It is very strongly pleochroic with X = pale-brown, Y = deep-brownish-green, Z = deep-bluish-green. Apatite is a very common accessory mineral in rounded grains or idiomorphic prisms 0.1–0.3 mm. in diameter; in the rocks from the zone of lit-par-lit injection mentioned above, it is commonly zoned, having a grey kernel surrounded by a colourless outer layer. Zircons are fairly plentiful, rare crystals reaching as much as 0.2 mm. in length, and when included in biotite have given rise to very dark haloes. Allanite occurs, but is infrequent. A few idiomorphic almost colourless crystals of this mineral, 0.1–0.2 mm. long, appear in 1528 from Northport surrounded by a thin film of epidote, and where this is thinnest there is a broad and dark halo in the enveloping biotite (Text-fig. 2). The allanite here appears to be primary in origin, and has a birefringence of 0.013. 1566 from Long Sound, however, shows another variety of the mineral corresponding with Larsen's (1921) description of a secondary allanite produced by the alteration of the primary

mineral. It has a slightly higher birefringence, and a strong pleochroism (X = yellow-brown, Z = dark-reddish-brown) the extinction direction X being inclined about 25° to the elongation of the grain. This mineral may also be bordered by epidote, which is considerably more abundant than in other rocks of this group, amounting to nearly 2%, and occurring in more or less poikilitic masses up to 0.5 mm. across. It is in close association with biotite and very probably of primary origin. The same possibly is true of the epidote in 1651, in which allanite is also present.

The iron ore of the granodiorites is commonly ilmenite with a tendency to form elongated or platy grains often surrounded by leucoxene, which, if enclosed in biotite, may form long threads extending along the cleavage lines. In some cases the oxide appears to be magnetite only. Pyrites is present in small grains in only a few of the rocks examined, and the ores as a whole are very sparsely developed.

The classification of the rocks of this group is complicated by lax petrographic usage. Few of them have sufficient potash feldspar to fall within the limits of granodiorite as defined quantitatively by Iddings or Shand, the possible exception being 1507. The predominant plagioclase is more sodic than is normal for quartz-diorites or tonalites and there is usually no hornblende. On the other hand, there is usually more biotite than occurs in the otherwise similar trondhjemites. They are, however, similar to rocks which by an extension of the strict usage have been termed granodiorites by Lindgren, Hatch and Wells, and others, but strictly appear to fall within Shand's definition of soda-tonalite.

Related rocks in neighbouring regions are but little known. There is, however, considerable resemblance between 1528 above and a rock (D.14) obtained further north by Professor Speight (1910), whilst another from Green Islets described by Professor Park (1926) as a diorite has probably some connection with the granodiorites of the present area, but differs from them in the presence of hornblende with potassic feldspar. Seelye's analysis (see Park, 1926) shows that the closest analogues of Professor Park's rock are among the quartz-monzonites.

Quartz-mica-diorite or Tonalite.

In the members of this group, as compared with the granodiorites, quartz is present in smaller amount, there is little or no potash feldspar, the plagioclase tends to be more calcic, and hornblende may occur with or in place of biotite. The rocks occur on the margins of the batholiths or as inclusions in the granite of these latter; commonly they have been invaded by granodiorite or granite. They are represented by 1586 and 1588, which are massive rocks invaded by granite on the northern side of Northport, by 1606 and 1616, pebbles from the same region, by 1614, a dyke on the south side of Cunaris Sound near the entrance to Cliff Cove, by 1695 forming the promontory a mile north of Only Island in Long Sound, by 1633, a dyke occurring about three miles east-north-east of Last Cove in the same Sound, and by inclusions in granodiorite on the southern side of Long Sound (1572) and in granites on the mid-west

shore of Isthmus Sound (1595). There is about 65% of plagioclase, the major portion of which is Ab₆₀ An₄₀ with a narrow outer zone of Ab₇₀ An₃₀ and rarely a thin fringe reaching Ab₈₀ An₂₀. In some of them sericitisation is far advanced. The quartz varies in amount from 5 to 15% and is usually highly strained. In 1588, 1633, and 1695 biotite is the chief or only coloured mineral and makes about 20% of the rock, in plates up to 2 mm. in length which are either fresh or chloritised. In the absence of hornblende a little muscovite may be present, as in 1588. In 1572, 1586, and 1614 the biotite is associated with an equal amount of hornblende, often in very intimate intergrowth with this latter, a particularly good example being afforded by 1572. In 1572 and 1586 the hornblende has the usual pale-yellow-brown to bluish-green pleochroism, and in the latter is occasionally sieve-like. In 1614 brownish tints predominate in the fresher crystals of this mineral, but there is a tendency to change into pale actinolitic pseudomorphs. In 1595 the change to colourless tremolite is complete, and has been accompanied by the separation of magnetite in grains arranged along the cleavages, and by the formation of very finely divided carbonates and of marginal outgrowths of chlorite. This alteration of hornblende is perhaps connected with the propylitisation accompanying the formation of Bradshaw's copper-bearing lode in the immediate vicinity, but it is peculiar that the biotite surrounding the amphibole-grains should have suffered little alteration. In 1606 hornblende is much more abundant than biotite and forms slightly-schillerised brown crystals with greenish margins. Epidote often occurs in small grains, but in 1588 it makes large irregular masses up to 3 mm. in diameter containing a core of deep-brown allanite, a mineral which also appears sparsely in 1572 and 1586. Sphene is an important accessory in the strongly hornblendic rocks, and apatite is very abundant throughout the quartz-diorites in prisms sometimes as much as 0.40 × 0.05 mm. in size. The iron ore is magnetite, and occasionally there is a little pyrites.

Gneissic structures and shearing are more common among the tonalites and diorites than among the acid plutonic rocks. A striking example is afforded by a boulder (1606) from Shallow River, Northport, which macroscopically appears to be a dark fine-grained amphibolite containing lenticular patches of coarse-grained diorite. In the latter the grain-size reaches 3.0 mm. and feldspars dominate over the coloured minerals, namely hornblende and grass-green flakes of pennine which replaces biotite oriented in parallel bands. The finer-grained portions are composed of hornblende and feldspar in equal proportion, oriented and sheared along the direction of schistosity. 1616, a pebble from the same locality, is a relatively fine-grained gneissic quartz-diorite or microdiorite noteworthy for the substantial parallelism of the deep-brown to greenish-brown flakes of biotite 0.5–0.8 mm. in length, which make up about 25% of the rock along with about 5% of grass-green hornblende, generally in small prisms, and for the abundance (15%–20%) of coarse epidote. This last mineral is usually in well-formed crystals as much as 3 mm. long, often with a core of allanite, but may be in less regular masses. It is strikingly poikilitic and appears to be of primary origin.

As an appendix to these quartz-diorites may be cited 1563, an inclusion in the granodiorite near the creek north of Mosquito Point, Northport. It consists of about 70% of sericitised indeterminable feldspar in grains about 0.5 mm. long, with about 20% of pennine replacing more or less oriented biotite, a little sphene having been developed in the change. Quartz grains form nearly 10% of the rock, and others of leucoxenised ilmenite are fairly abundant. Small prisms of apatite appear and grains of pyrites, while films of the last mineral often surround the grains of quartz. Veinlets and scattered masses of calcite and an occasional grain of epidote are also present. Probably this rock is a much-altered fragment of a feldspathic quartz-mica-diorite.

Comparisons with rocks from other parts of Fiordland show that, though there are differences in texture and grain-size, there is similarity in all essentials between D4, a gneissic diorite from Duck Cove described by Professor Speight (1910), and 1586 and, to a lesser extent, 1572 noted above.

Diorites.

This group differs from the quartz-diorites in the smaller amount or absence of quartz and biotite, though the distribution of its members is the same as that of these tonalitic rocks. Diorite occurs as dykes extending out from the batholiths, as small masses at the margins of these latter, or as inclusions in them. It is represented by 1560 and 1515 from the southern shore of Long Sound 3000 yards north-east of Only Island; by 1659, included in the granites of Isthmus Sound; 1592, a vein invading the sediments half a mile north-east of Cuttle Cove; 1610, a narrow dyke a mile east of Cunaris Islands; 1559, on the northern shore of Northport; 1589a, a mile up the creek draining from Lake Caesar; 1615, a pebble collected from the same creek; and 1568 and 1678, from the eastern shores of Landing Bay near Cape Providence.

In 1568 the coloured constituent forms about 10% of the rock and is a pale-green pennine replacing biotite; with it there is leucoxenised ilmenite, much sphene having been developed by the change of both biotite and this latter mineral. The rock has been a good deal crushed, and the original mica has been arranged in and parallel to shearing-zones. The plagioclase is approximately Ab₇₆ An₂₄ and occurs in prismoids up to 4 mm. in length dusted with sericite and containing small inclusions of quartz; its grains have been separated by the shearing, and a small amount of crushed quartz fills the interspaces. The feldspar of the other diorites is usually more basic than in this rock; it is sometimes turbid owing to the presence of zoisite and sericite, and in 1559 has been partly replaced by prehnite along definite fracture lines, the change working sporadically into the feldspar from its margins. In 1515 the grain-size is a little smaller, and the shattering not so marked. The pennine is again in oriented flakes, but is not so abundant as in 1568. On the other hand, there is about 6% of epidote in aggregates of irregular grains of moderate size closely associated with biotite or strung out along fracture lines, and scattered as a cloud of minute grains through the feldspar (Ab₆₅ An₃₅). Quartz is absent. 1560 is still more basic, for zoned plagioclase makes up 75% of the rock

and is Ab₅₀ An₅₀ in the centres of the grains. It further contains nearly 20% of hornblende in prisms nearly 4 mm. long, which show the normal pleochroism, X = brownish-yellow, Y = deep-green, Z = brownish-green with a bluish tinge on the margins. Sometimes this latter mineral is lighter in colour, as in 1559 and 1589a, where it forms 45% of the rock and is associated with 3% of chloritised biotite in small oriented flakes. Its pleochroism in these two rocks is X = very pale yellow-brown, Y = grass-green, Z = weak blue-green. It forms grains about 1 mm. long sometimes moulded upon the feldspar, sometimes idiomorphic against it. In 1678, on the other hand, the hornblende forms prisms 2–4 mm. long, darker and browner than normally, and often schillerized, as also in 1560.

In the diorite (1659) included in the granite at Bradshaw's Mine, Isthmus Sound, propylitisation is more advanced than in the tonalite (1595) which forms a similar inclusion in this granite, and the hornblende has been replaced by finely-divided carbonates and distributed chlorite. In 1610, a dyke of diorite occurring a mile east of Cunaris Islands, the hornblende is present as large poikilitic masses of the pale-green actinolitic type, mingled with a little chlorite and associated with a greater amount of biotite in irregular aggregates of small confusedly-matted plates.

Ilmenite is a frequent accessory, and is often more or less leucoxenised; it is especially fresh and abundant in 1610. Separate grains of sphene are occasionally present, and in rare cases may be quite large (0.8 mm.). Epidote is not infrequent, and is occasionally accompanied by allanite (1615). Apatite is plentiful in nearly all the diorites, and is sometimes in relatively large prisms.

Gneissic Microdiorites.

The gneissic microdiorites constitute a small group of rather fine- or medium-grained rocks which are probably sill-like apophyses of the batholiths and which have most of them been subsequently invaded by pegmatite, with consequent reactions leading to the formation of biotite. 1527, which occurs at the head of the bay by Mosquito Point, Northport, where it is invaded by granite or granodiorite, has an almost granulitic texture and even grain-size (0.3–0.5 mm.) for both feldspar and hornblende. Flakes of biotite are less abundant than either of these minerals, but larger (up to 1.5 mm.), and are usually oriented in one plane and along the lines of shearing which traverse the rock are completely chloritised with the development of sphene. The scattered grains of ilmenite have also been changed to sphene along these lines. The plagioclase, averaging Ab₄₅ An₅₅, is slightly zoned, and there is about 5% of interstitial quartz. The hornblende is rather pale in colour and slightly schillerized.

Interesting features are exhibited by a series of boulders obtained from Shallow River, Northport. Of these, 1548 shows a contact between microdiorite and pegmatitic granodiorite. The former is composed of labradorite, slightly schillerized hornblende, and a smaller amount of biotite together with a little ilmenite and apatite, the longer axes of all these minerals being more or less oriented. There is also some interstitial quartz. The invading pegmatitic granodiorite occurs in long lenticles consisting essentially of quartz and

andesine 2–4 mm. across, sieve-plates of biotite (3–6 mm.), large prisms of hornblende and sieve-garnets over 6 mm. across. In addition, it contains ilmenite, a little sphene, tawny yellow epidote, and zircon. In 1692 and 1693, a vein of aplite invading gneissic microdiorite breaks across the foliation of the latter, either penetrating or causing the feldspathisation of the invaded rock for a distance of one or two centimetres from the intrusion. A thin section (1692) taken 10 cm. from the vein shows a normal gneissic microdiorite, consisting of 45% of andesine (Ab₆₀ An₄₀), 28% of hornblende, 24% of biotite, and 3% of quartz in oriented grains about 0.3–0.7 mm. long, with accessory sphene, ilmenite, rare epidote, and a few large feldspar phenocrysts or phenocrystic aggregates. Immediately adjacent to the aplite, however, the hornblende has completely disappeared, and the plates of biotite which continue the foliation fray out among the crystals of quartz and oligoclase (Ab_68–72 An_32–28) which constitutes the vein of granodiorite in proportions approximately 40% and 60% respectively. This replacement of hornblende by biotite immediately adjacent to an acid intrusion is a well-known phenomenon (cf. Bowen, 1926, p. 198). Indeed, the coarse flakes of biotite in the rocks 1548 and 1527 described above may owe their development in large measure to the influence of the adjacent granodiorite intrusions.

Though these rocks are not precisely similar to any described by Professor Speight (1910) from the adjacent Sounds, there are points of an analogy between 1548 and D.12 from an unrecorded locality in Dusky Sound, for the latter rock (of which no description has yet appeared) has also streaks of large oriented flakes of biotite in a general base of gneissic hornblendic microdiorite.

Hornblende-norite.

This type of rock is represented by 1649 and 1687 from the north side of Last Cove, about two and three-quarter miles east of Last Cove. 1649 is a grey rock of medium grain-size, of which slightly sericitised plagioclase (Ab₃₅ An₆₅) in prisms 1–2 mm. long makes up about 50%. Hypersthene forms about 30% of the rock, occurring in rather dusty faintly pleochroic ragged grains about 1 mm. long and usually roughly prismatic, though sometimes moulded on the plagioclase. Though the pleochroism is slight, the negative optical character is distinctive Brown hornblende also occurs in poikilitic plates usually small, but sometimes up to 3 mm. in diameter. It is a little schillerized, and has usually altered marginally into a colourless tremolitic amphibole in optical continuity with it and without observable change in the extinction angle, though with some indication of increased birefringence. Where this alteration has been complete, there are individual masses of the colourless mineral, either in compact grains or tending to fray out into confused aggregates of fibres. Such colourless amphibole may project into the hypersthene obliquely to the cleavage of the latter, or may lie enclosed within its cleavage-planes, or may form poikilitic masses enclosing it. Red-brown biotite may also occur in little flakes in the cleavage planes of the hypersthene, or in larger ones lying within the amphibole, quite unaffected by this change of host, or again may be present in flakes independent alike of hypersthene and amphibole.

There are a sprinkling of pyrites and a few small grains of sphene. In 1687 the biotite, like the hornblende, may occur in extended poikilitic patches enclosing grains of feldspar and pseudomorphs of pyroxene, though the actual amount present is small. The hornblende is partially bleached as before, and the feldspar rather more sericitised, but the hypersthene has been entirely replaced by talc. The occurrence of a large mass of pyrite suggests that hydrothermal solutions may have effected the last change. It is worthy of note that the schiller-inclusions in the hornblende consist of very fine brown translucent fibres in a close rectangular lattice.

The two rocks just described come from an apparently small mass of norite which has invaded the most highly metamorphosed of the schists. They exhibit, however, no sign of strain effects, and there is, therefore, no direct analogy with the development of colourless amphibole in hornblende schists at points of special stress as noted in the Singhbhum district (Dunn, 1929, p. 93–5), though the bleaching of the margins of a hornblende in an epidiorite invaded by granite-gneiss in the same region may be noted (loc. cit., p. 98). Again, a rather similar change of brown hornblende has been noted by Turner (1933) in a diorite-pegmatite (1333) invading hornfels in the Arawata River in South Westland; here again the feldspar is only slightly sericitised, and no pressure-effects occur. On the other hand, in the flaser-gabbros of the Lizard, though the secondary hornblende may be colourless “often the original brown hornblende is not modified, but remains even when the diallage with which it is surrounded is entirely transformed” (Flett, 1912, p. 88). In view of the small amount of information as yet available, further consideration of the relation of these two types of amphibole must be postponed.

It is, however, of interest to note the distribution of possible analogues to this hornblende-norite in adjacent regions. These are briefly a mica-norite from the Darran Mountains (Cleddau River), Milford Sound, containing augite rather than hornblende (Marshall, 1907), a similar but gneissic rock from Bligh Sound (Speight, 1910), with which may be mentioned a hypersthene-amphibolite from the same Sound (Speight, 1910), and a dioritic rock with a little enstatite from Harrison Cove, Milford Sound (Hutton, 1891). Hutton (loc. cit.) noted the presence of a plagioclase-hypersthene-rock (“enstatite-diorite”) in Wet Jacket Arm, Dusky Sound, and Speight has collected but not yet described a quartz-diorite (D.8) containing a minor amount of brightly pleochroic hypersthene with augite and hornblende, occurring at an unrecorded locality in the same Sound. None of these rocks appears to be closely analogous with the Long Sound norite, and the norite of the Bluff Hill (Wild, 1911) is again different in several respects from any of the rocks cited. There is, therefore, no clear indication as yet of the magmatic affinities of the Long Sound norite, and it must still remain an open question as to whether it belongs to the same series of plutonic intrusions as the diorites, granodiorites, and granites, though much less gneissic than the earliest and most basic members of that series, or whether it may be relegated to a series of intrusive masses injected at a much