The Heavy Minerals of Some Cretaceous and Tertiary Sediments from Otago and Southland

Transactions and Proceedings of the Royal Society of New Zealand, Volume 66, 1937, Page 255

The Heavy Minerals of Some Cretaceous and Tertiary Sediments from Otago and Southland By C. O. Hutton and F. J. Turner, University of Otago, Dunedin. [Read before the Otago Branch, May 19, 1936; received by the Editor, May 20, 1936; issued separately, December, 1936.] Introduction Fiordland Region Material Examined Mineral Composition Provenance Middle Waiau Valley North-Western Otago Material Examined Petrography of Concentrates Heavy-mineral Assemblages in Relation to Stratigraphy Provenance Southern Otago Conclusion Literature Cited Explanation of Plate Introduction. During petrographic examination of some Tertiary sedimentary rocks collected from the vicinity of Lake Manapouri, the heavy mineral concentrates were compared with others from Tertiary strata in various other parts of Otago and Southland. These concentrates were found to vary considerably; and in the absence of previous work of this kind upon New Zealand rocks, other than the investigations of G. J. Williams (1934) upon tin-bearing gravels of Stewart Island, it was thought advisable to place upon record the data collected by the writers. These results indicate that a detailed survey of the heavy minerals present in the younger rocks of southern New Zealand should throw light upon problems connected with the later geological history of this region. The material examined was obtained from the following general localities:— (a) Fiordland region: Lake Manapouri, Eglinton Valley, Preservation Inlet. (b) Waiau Valley: Clifden. (c) North-West Otago: Lake Wakatipu (Bob's Cove Beds), Crown Range Saddle. (d) Eastern Otago: various localities within a fifteen-mile radius of Dunedin. (e) South Otago: Chatton (8 miles east of Gore), Waikaia Valley. The majority of rocks were only slightly cemented, and were crushed readily with a few blows in a steel mortar. The crushed material was screened with a standard 100 I.M.M. sieve, and the

fine residue treated in a separating-funnel with bromoform (G. slightly greater than 2.9). The heavy concentrate so obtained was washed with benzene, dried, and treated with a hand-magnet to remove magnetic iron-ores, after which further separations were made, when necessary, with Clerici solutions of various strengths (G = 3.4., 3.6 and 4.1). Portions of each heavy fraction were immersed in clove oil (μ = 1.54) in a watchglass and examined microscopically in the usual way. In most cases cheek samples were also mounted permanently on glass slides in Canada balsam. The microscopic examination was conducted with a view to (a) identification, (b) observation of grain-form and other local peculiarities and (c) approximate estimation of relative abundance of the constituents present in the heavy fractions. For these purposes it was not found necessary to estimate specific gravity accurately, while refractive indices were measured in a few instances only. In one or two cases, notably in determination of cassiterite, confirmatory chemical tests were carried out. Fiordland Region. Material Examined. Representative specimens (2736, 2738–2741, 2743) were obtained from the Tertiary marine sandstones that are exposed continuously in low cliffs for some miles on either side of Stony Point, towards the eastern end of Lake Manapouri (Park, 1921). The strata in this locality undulate gently with no obviously dominant direction of strike, and include coarse beach-conglomerates several hundreds of feet in thickness overlain by coarse arkosic sandstones inter-stratified with beds of finer texture. The principal constituents of the conglomerates are large boulders of “granitic” gneiss, amphibolite-gneiss and allied rocks, and obviously derived from local sources; similar rocks outcrop from beneath the Tertiary cover a few miles west and north-west of Stony Point. A calcareous well-cemented grit of the same general facies as some of the rocks from Stony Point was collected from near the summit of Freestone Hill, a low residual mass a few acres in extent rising above the Pleistocene gravels 2 miles south-west of Manapouri Boarding-house. Arkosic sandstones are also well developed among the Tertiary rocks east of Lake Te Anau. The specimen selected (1494) is typical of the coarse light-coloured sandstones which outcrop on the east side of the Eglinton Valley road about 28 to 30 miles north fo the Te Anau Hotel. These rocks are not far above the coarse basal conglomeraets of the Tertiary sequence which here rest directly upon the almost unmetamorphosed greywackes and braccias of the Te Anau Series. The conglomerates themselves contain many large boulders of granitic, dioritic and other plutonic rocks obviously derived from the adjacent Fiordland complex. Finally a specimen of white feldspathic sandstone (2762) collected by Professor W. N. Benson from Coal Island, Preservation Inlet (Benson, 1934), has been included for comparison with the rocks of Manapouri and Te Anau. This rock also occurs no great distance above the local basement, which is made up chiefly of slates greywackes, granites and hornfelses.

Plate 22

Mineral Composition. In every case the light fractions consist mainly of nearly fresh feldspar and quartz. Muscovite is widely distributed, but is seldom plentiful. The minerals present in the heavy concentrates (G. > 2.9) are listed in alphabetical order in Table I. In this and subsequent tables the abundance of each mineral is indicated with reference to the heavy concentrate obtained from the sieved portion of the original specimen. It is therefore not possible to use the tables for accurate comparison of the quantities of a particular mineral present in different rocks. For example, while sphene is listed as “abundant” in 2736, and “fairly abundant” in 2738, the amounts of sphene in the two rocks are approximately equal, the total percentage of heavy minerals (notably biotite) being much higher in 2738 than in 2736. Table I. Heavy Minerals in Sandstones from Fiordland Region. Specimen Number Allanite Apatite Biotite Cassiterite Chlorite Epidote Garnet Hornblende Magnetite Monazite Pyrite Sphene Tourmaline Zircon Undetermined Opaque 2736 r a fa r m a m 2738 r m p m r m m r fa m m 2739 m p r m m r fa m m 2740 fa a fa m m a m 2741 r m p m r r fa m 2743 r r a r fa r m m a m 2746 r m a fa a r r fa r r 1494 m p m r fa m m m 2762 m a r m r m a r m Heavy Minerals in Sandstones from Fiordland Region. 2376, Sandstone interbedded with conglomerates, ½ ml. S. of Stony Pt., Manapouri. 2738, Sandstone, ¼ ml. W. of Stony Pt., Manapouri. 2739, Fine phase of sandstone, ½ ml. W. of Stony Pt., Manapouri. 2740, Sandstone, E. point of Circle Cove, Manapouri. 2741, Coarse phase of sandstone, ¼ ml. E. of Circle Cove, Manapouri. 2743, Sandstone, Head of Circle Cove, Manapouri. 2746, Freestone Hill, near Manapouri. 1494, Coarse Sandstone, Eglinton Valley, 28 mls. N. of Te Anau Hotel. 2762, Coarse Sandstone, Coal Island, Preservation Inlet. p = predominant (> 60%*These figures are rough percentages (estimated by inspection) of constituents of the heavy residues (G. > 2.9).); a = abundant (20–60%); fa = fairly abundant (5–20%); m = minor constituent (< 5%); r = rare. Allanite occurs as clear irregular broken grains with high refractive index, low birefringence, and distinct pleochroism in various shades of brown. The optic axial angle is large and the sign indeterminate in the few grains available. Though these data are insufficient for certain determination, the mineral appears to be identical with undoubted allanite present in gneisses and pegmatites of the Lake Manapouri region.

Apatite. Typically apatite is colourless and clear, but sometimes may be clouded locally with minute dusty black inclusions. The hexagonal prismatic form is often distinctly preserved (Figs. 18, 19), but details of the terminal faces are usually somewhat obscured by slight rounding of the grains; in other cases the initial form is completely lost (Fig. 20). Biotite is the most abundant constituent of the heavy concentrates. Two varieties are common. In sandstones from the neighbourhood of Stony Point and Circle Cove, Lake Manapouri, the biotite has a characteristic brownish-green or olive-green tint for basal vibrations, and is identical with the greenish biotites that are widely distributed in certain of the basement gneisses. To check this determination, refractive indices were determined as follows:— Green biotite from arkosic sandstone (2738) : β = 1.625 ± .005. Green biotite from fresh biotite-gneiss (2448) : β = 1.620 ± .005. The green colour is thus not to be regarded as a result of incipient alteration to chlorite. In the rocks from Freestone Hill, Eglinton Valley and Preservation Inlet the-biotite is the deep reddish-brown variety commonly present in the acid members of the Fiordland complex. Cassiterite was identified with certainty in a single rock (2738), where it is surprisingly plentiful. The grains usually are rounded, but the pyramid habit is sometimes distinct, while striated faces and traces of twinning occasionally show in reflected light. The mineral is almost opaque except in very thin splinters where it may be faintly translucent and deep red-brown in colour. It was distinguished chemically by failure to dissolve in fusion-mixture, fused sodium bisulphite or boiling nitric acid, but was readily taken into solution by fusing with a mixture of sodium carbonate and sulphur; an aqueous solution of the product of fusion gave confirmatory tests with mercuric chloride and ammonium molybdate respectively. Chlorite occurs occasionally as rare pale-green flakes resulting from decomposition of biotite. Epidote. Clear, irregular broken grains usually with strong yellowish-green absorption-tints and distinct pleochroism; less commonly a colourless or pale-yellow form is also present (1494, 2762). The optic axial angle is large and the sign invariably negative, indicating a fairly high content of Fe2O3 even in the paler varieties. Cleavage lines are absent or only imperfectly developed (Fig. 12). Garnet is absent from the Manapouri sandstones, but small angular grains were recorded in specimens 1494 (pink) and 2762 (pink and colourless). Hornblende is absent or rare except in specimens 1494 and 2746, where it is abundantly developed as angular or slightly rounded cleavage-fragments. Typically the Z absorption-tint is deep bluish-green as in the amphiboles of the Manapouri gneisses; but in the Freestone Hill rock (2746) there are, in addition to this, two other varieties with brownish-green and pale bluish-green colours respectively. Grains of the latter type usually enclose plentiful inclusions of opaque black ore.

Magnetite. Magnetic iron-ore is always present in small quantities. Some of this may be ilmenite, but the grains are irregular and show no trace of leucoxene, though frequently stained with reddish oxidation-products. Monazite. Rare grains were identified in two concentrates (2741, 2746). These are tabular with rounded outlines (Figs 21 and 22), and are crossed by highly perfect and less perfect cleavages intersecting at right angles. They are interpreted as (100) tables nearly parallel to the optic axial plane, with perfect (001) and difficult (010) cleavages (cf. Winchell, 1933, p. 138). Consequently the optic sign cannot be determined from the interference figures yielded by grains of this type. The colour is deep golden-yellow with faint pleochroism, the minimum absorption being for vibrations parallel to the perfect cleavage (i.e. parallel to X). The extinction is approximately straight with reference to the cleavages and the refractive index and birefringence are high. The mineral is easily distinguished from associated epidote by its striking colour, the details of pleochroism and the presence of two cleavages. Pyrite usually shows good crystallographic development, as octahedrons, pentagonal dodecahedrons or rarely cubes, but rounded and less commonly angular grains were also observed. When relatively plentiful it is probably authigenous e.g., in the standstone 2740 which is associated in the field with beds containing abundant pyritic concretions. A detrital origin is possible for some of the angular grains. Sphene. After biotite this is the most abundant mineral in the heavy concentrates. It occurs as angular irregular clear grains devoid of inclusions and typically pale-yellow in colour (Figs 23 and 24), though occasionally it may be colourless (2746) or have a dirty or clouded appearance (e.g., 2743). It is distinguished from associated epidote by higher refractive index and birefringence, and especially by the small optic axial angle, positive signs and strong axial dispersion as indicated by interference-figures yielded by sections more or less perpendicular to Z. In the Preservation Inlet rock (2762) the colour is much deeper and there is distinct pleochroism from pale-yellow to deep-reddish-brown, much as in the “red-brown” sphene described by Benson and Bartrum (1935, p. 131) as occurring in the rocks of the Long Sound Series of this region. The pale-yellow variety common in the Manapouri sediments appears on the other hand to be identical with the coarse yellow sphene which is so abundant in some of the local gneisses. Tourmaline was recorded only in the Preservation Inlet sandstone (2762), where it takes the form of sparsely scattered prismatic enhedra, strongly pleochroic in reddish-brown tints and sometimes showing peripheral zones of lighter colour. Zircon is ubiquitous as a minor constituent of the concentrates (Figs. 2, 3 and 4). For the most part the grains are colourless well-formed prisms with a variety of pyramidal faces, the most frequent combination being a second-order prism terminated by simple pyramids of the first order. The acicular habit frequently

displayed by zircons from other districts is lacking. Partially rounded crystals (Fig. 5) are not uncommon, but traces of the prismatic form always are distinct. Most of the crystals contain minute rod-like inclusions or else what appear to be relatively large, irregular almost vermicular cavities. Undetermined Opaque Ores. Some of the residues contain small quantities of opaque non-magnetic ores. Much of this may be ilmenite, but in one specimen (2739) the presence of striated metallic faces and brightly-reflecting cleavage-surfaces suggests the possibility of wolfram. Provenance. The distinctive features of the heavy-mineral concentrates just considered are as follows:—Biotite is the most abundant mineral and often makes up 80% or more of the concentrate. Sphene is plentiful, while epidote, apatite, zircon and magnetite follow in the order given. Equally characteristic is the general absence of tourmaline and garnet. The consistently angular or euhedral shape of the mineral grains, their unweathered condition, and the presence in quantity of the unstable minerals biotite, apatite and sometimes hornblende all point to a nearby source of origin for the sediments. From the mineralogical analogies enumerated in the previous section it is highly probable that this source lay in the Fiordland complex of plutonic, gneissic and schistose rocks. On this assumption the presence of cassiterite and monazite is noteworthy, for so far as the writers are aware these minerals have not yet been recorded in situ in this region, though both might be expected to occur there. Further, detrital monazite was recently found by Williams (1934, p. 353) in the Quaternary tin-bearing gravels of Stewart Island. On the other hand the absence or rarity of tourmaline and garnet is a most unusual feature for sediments derived from such a source as that afforded by Fiordland. A partial explanation of the peculiarity lies in the rarity of tourmaline in the basement rocks of Lake Manapouri as revealed by petrographic studies in progress; but these same investigations have also shown that garnet is not uncommon as a constituent of pegmatites and aplites extensively developed towards the head of that lake. The general absence of hornblende coupled with universal abundance of biotite provides a further anomally. Hornblendic rocks are widely distributed among the basement rocks as exposed around Lake Manapouri, and local patches of black sand concentrated on the present beaches always consist largely of this mineral. Finally attention is drawn to two significant local variations:— (a) In the Preservation Inlet rock the biotite is red-brown, the sphene is a distinctive reddish pleochroic varicty, and brown tourmaline and minor garnet are present. These peculiarities all accord with the mineralogy of the local basement rocks as described by Benson and Bartrum (1935).

(b) The sandstone from Eglinton Valley differs from the typical Manapouri rocks in the brown colours of the biotite and the presence of abundant associated green hornblende. The sandstone from Freestone Hill is similar in these respects. It would thus appear that, while the sediments listed in Table I may all be regarded as derivatives of the gneisses, schists, etc., of the Fiordland region, different local sources within this distributive province have supplied detritus to different areas of deposition. At least three such local sources are indicated for the sediments discussed above. Middle Waiau Valley. Three specimens of sandstone from the fossiliferous mid-Tertiary beds of Clifden, Waiau Valley, were examined (2747. band 6B, west bank Waiau River, Clifden; 2748, band 7A, west bank Waiau River, Clifden; 2749, cutting on branch road, 1 mile beyond Clifden racecourse). Nos. 2748 and 2749, though obtained from localities separated by about one mile, belong to the same geological horizon, while 2747 is from somewhat older beds; the age of all three is post-Hutchinsonian but pre-Awamoan (Allan, 1933, p. 104). The heavy fractions from bromoform separation amount to about 10% of the total bulk. The grains invariably display a high degree of rounding (Figs. 8 and 9) in contrast with the angular grains of the Fiordland rocks. The constituent minerals are listed in Table II with the exception of glauconite, the presence of which in the heavy fraction appears to be determined by the degree to which it has been oxidised. Table II. Heavy Minerals in Sandstones of Middle Waiau Valley. Specimen Number Apatite Augite Biotite Chlorite Epidote Garnet Hornblende Hypersthene Magnetite Pyrite Sphene Zircon 2747 m r m r p r r 2748 fa m ? r p m m r r r 2749 fa m ? r p fa m r r Hornblende, always predominant in the heavy fractions, includes both blue-green and brownish-green varieties, the former sometimes enclosing schiller-inclusions of opaque ore. Occasional grains in 2747 are partially bleached as if by incipient replacement by tremolitic amphibole. Pyroxene makes up about 10% of the concentrate in 2748 and 2749, and includes a pale-green or colourless augite and some-what less plentiful strongly pleochroic hypersthene (X = deep pink, Z = pale green). Both types commonly exhibit well-developed schiller-structure (Figs. 8, 9 and 10). Biotite when present is a strongly pleochroic red-brown type. Yellowish, semi-opaque ironstained grains of what may be epidote were recorded in 2748 and

2749, but it is not unlikely that these are actually partially altered glauconite. In 2747, however, there are clear colourless grains of undoubted clinozoisite with low birefringence, positive optic sign and large optic axial angle. Colourless zircon, pale-yellow or colourless sphene and pinkish garnet are notable only for their rarity and the rounded outlines of their grains. The heavy-mineral assemblages of the Clifden rocks are noticeably different from those of the Fiordland region (compare Tables I and II), and have obviously been derived from a distinct distributive province. The most significant feature is the presence of pyroxenes which, taken in conjunction with abundance of hornblende (including a brownish-green variety), suggests derivation from an area in which the Southland “norites” were exposed. However, the Clifden sediments have not yet been found to contain the distinctive pyroxenes of the clinohypersthene and enstatite-augite groups found by Service (1934a) in the “norite” of Bluff and now known to be widely distributed in similar rocks throughout southern New Zealand. The presence of small amounts of biotite, zircon, sphene and garnet may be explained by assuming either that their source lay in a relatively distant area of Fiordland gneisses, or that they were rewashed from older sedimentary rocks such as the greywackes exposed to-day in the Takitimu and Longwood Ranges. North-western Otago. The material from this region consists of three specimens collected from the mid-Tertiary strata involved by faulting in the basement schists near Bob's Cove, Lake Wakatipu (Park, 1909). These are silty sandstone (2742), fine grey siltstone (2744) and concretionary sandstone (2763), which lie stratigraphically at distances of about 700, 900 and 400 feet respectively above the base of the series. (As a result of overturning during involvement, the beds now lie in reverse order to the true stratigraphic sequence.) Table III. Heavy Minerals in Siltstone and Sandstones, Lake Wakatipu District. (1) Bob's Cove Beds, Lake Wakatipu. (2) Summit of Crown Range Road, North-west Otago. Locality Specimen Number Apatite Biotite Chlorite Epidote Garnet Magnetite Monazite Pyrite Sphene Tourmaline Zircom (1) 2742 r r m m a m m m a 2744 r r m m a fa m a 2763 fa r a fa m r m r fa (2) 2745 a r m fa a The crop of heavy minerals is small, not more than about fifty grains being obtained by treatment of a twenty-gramme sample with bromoform; but the assemblage is distinctive (see Table III) when compared with those obtained from the rocks of Fiordland and

Clifden. Hornblende and pyroxenes are altogether absent, while biotite is absent or inconspicuous; the mineral listed as biotite in 2742 might actually be stilpnomelane, for it is a deep-brown strongly pleochroic type. Zircon and magnetite are relatively abundant, while epidote, garnet, and sphene are constantly present as minor constituents. The presence of tourmaline in two of the three specimens is regarded as significant. It occurs as strongly pleochroic, euhedral, singly-terminated prisms or broken prismatic grains, crowded with minute black specks and identical in appearance with the deep blue or brown tourmaline that is so widely scattered as an accessory mineral in the schists of Otago. The zircons are well-formed rather stout prisms with little sign of rounding (inclusions as described for zircons from Fiordland rocks). The garnet for the most part is in highly irregular, angular pink or colourless grains with marked conchoidal fracture, though one or two small rhombic dodecahedrons were also noted (Fig. 25). Angular yellow epidotes are very plentiful in 2763, but are less conspicuous in the other two concentrates. Sphene also occurs as irregular yellowish grains rendered cloudy by swarms of minute inclusions. Pyrite when present appears to be of authigenic origin and may take the form of perfect spheroidal micro-concretions (2744). With the exception of monazite the detrital minerals determined in the Bob's Cove rocks are all known to be present in similar form in the local schists, and there can be no doubt that the sediments under consideration are composed of detritus worn from a region underlain by members of the Central Otago schist group. Text-figure 1. Monazite from No. 2763. The presence of monazite in 2763 is thus of considerable interest. The mineral is much paler than the detrital monazite from the Fiordland area, and the grains are sharply angular or euhedral with no trace of rounding. It is pale yellow, non-pleochroic and highly transparent compared with the cloudy darker-coloured associated sphene. In one well-preserved grain (Text-fig. 1) the crystal form corresponds well with one figured by Dana and Ford (1932, p. 700,

fig. 1013). The cleavage parallel to 001 (b) and the optic orientation of this crystal (X parallel to b), together with the small axial angle and positive sign observed in other grains, leave no doubt as to identification. The grains of monazite, though rare, can hardly have travelled far, and derivation from the local schists themselves is highly probable. In Table III the concentrate obtained from a specimen of white sandstone (2745) from near the summit of the Crown Range is included for comparison with the Bob's Cove rocks. In 2745 epidote (mainly a colourless clinozoisitic type with dusty black inclusions) is more abundant and tourmaline much more plentiful than in the Bob's Cove beds; on the other hand garnet (colourless, angular) is rare and sphene appears to be absent in the Crown Range rock. The zircon occurs principally as rather stout cloudy prisms showing some degree of rounding; a few grains are perfectly rounded, while one or two are slender and clear. The tourmaline includes the same varieties as are present in the rocks from Lake Wakatipu. On the whole, as regards heavy mineral content, 2745 resembles certain sediments from the Dunedin district (Taratu Coal-measures) rather than the much less distant rocks of the Bob's Cove group. Eastern Otago (Dunedin District). Material Examined. The downward sequence of younger sedimentary rocks in the Dunedin district (Ongley, 1933) is as follows:— (7) Goodwood or Dowling Bay limestone*This occurs north of the area mapped by Ongley, and has been described by Finlay and McDowall (1923) and Service (1934). (6) Caversham Series (b) Caversham sandstone (a) Greensand (5) Burnside Series (b) Burnside mudstone (a) Loose sandstone (4) Wangaloa Series (b) Abbotsford mudstone (a) Glauconitic sandstone (3) Brighton Series Shell-limestone (2) Taratu Series Quartz-conglomerates and sandstones (1) Kaitangata Series Schist-conglomerates and sandstones The units listed above range in age from Cretaceous to Miocene and have a total thickness of some 6,000 ft., of which the two lowest series together make up about 4,000 ft. In the immediate vicinity of Dunedin the basal beds, lying upon an eroded surface of ancient schists, mostly belong to the Taratu Series. The schists themselves are low-grade regionally metamorphosed quartzo-feldspathic and pelitic rocks of the Chlorite Zone. Specimens typical of each of the above lithological types, with the exception of those belonging to the Brighton Series, were examined for heavy minerals, with the result shown in Table IV.

Table IV. Heavy Minerals in Cretaceous and Tertiary Sediments from Dunedin District. Horizon Specimen Number Andalusite Apatite Biotite Chlorite Epidote Garnet Hornblende Kyanite Magnetite Pyrite or Marcasite Rutile Sphene ?Topaz Tourmaline Zircon Zoisite 7 2760 fa fa r fa fa m r fa 2752 fa m m r a a 6b 2755 fa r a m m m r a 2759 fa r p r r r r r 6a 2771 r m r p r m 5b 2765 r m p r r r 2750 r m r r r r fa a r 2753 r fa r r r fa r r a a 5a 2754 a r r r fa r a a 2758 r r r r a m a a r 2769 r r fa r fa a 4b 2751 r m r m m r m p 2768 r r m fa p 4a 2761 m r r fa a m r a 2764 m fa r a m a 2766 m m a a 2 2767 m m a a 2770 m m a a 2774 r r r p 2775 r r r r p r 1 2776 r r r r p 2777 r r r p Numbers in the first column refer to the stratigraphical sequence on page 265. p = predominant, a = abundant, fa = fairly abundant, m = minor constituent, r = rare. Petrography of Concentrates. The distribution and distinctive features of the minerals represented in the heavy concentrates are summarised below:— Andalusite is a rare constituent of three concentrates, all obtained from loose sandstones of the lower part of the Burnside Series. The grains typically are highly angular, clear and unweathered and are readily distinguished by their distinct pleochroism from pale pink (X) to colourless or pale green (Z). The sign could not be determined in the few grains available, but the optic axial angle is certainly large. Apatite is plentiful only in the Caversham sandstone and to a less extent in the glauconitic sandstones (Boulder Hill beds) at the base of the Wangaloa Series. Usually the crystals, though slightly corroded, retain their stout prismatic form and may even show well-defined faces (cf. Figs. 18 and 19), but in a few cases (e.g., 2752) the outlines of the grains are rounded or irregular. Clear colourless grains predominate. In 2755 swarms of black specks are enclosed in the otherwise colourless crystals.

Biotite. In several rocks from various horizons there are rare flakes of a golden-brown micaceous mineral which, though identified as biotite, might equally well be stilpnomelane, a mineral now known to have a wide distribution in the schists of Otago. Chlorite was recorded in only two rocks—Goodwood limestone (2760) and Caversham sandstone (2755) respectively—in both of which it is one of the important constituents of the heavy concentrate. It takes the form of rather coarse, ragged pale-green flakes, which may enclose finely granular sphene or iron-ore. Epidote is widely distributed in minor to considerable amounts, and in one specimen (Caversham sandstone, 2759) is the predominating mineral in the heavy concentrate. It includes two varieties: the one is colourless, poorly birefringent and optically neutral or negative, evidently rich in the clinozoisite molecule, while the other, highly birefringent, faintly pleochroic and greenish-yellow is colour, is obviously a more ferruginous type. Dusty black inclusions are common, especially in the colourless grains. Garnet is usually present as a rare or minor constituent, except in the concentrates from the sandstones of the Taratu Series. It typically occurs as clear pale-pink or less commonly colourless angular or rounded grains, but euhedral crystals of small size were occasionally noted. Hornblende. Very rare grains of blue-green hornblende were observed in three specimens of loose sandstone (Burnside Series) and one of Caversham sandstone. Kyanite. In concentrates obtained from three specimens of loose sandstone of the Burnside Series, rare grains of colourless kyanite are consistently present. The crystals are unweathered but always worn, the outlines being rounded or sub-prismatic (Figs. 14 and 15). The high refractive index, low birefringence, good cleavage and optic orientation are distinctive. The grains conform to the characteristic tabular habit parallel to (100) with elongation parallel to the c axis (cf. Wichnell, 1933, pp. 205, 206), and thus show two sets of cleavages intersecting at 90°, the extinction angle being 30° with reference to either set (elongation positive in prismatic forms). Such grains also give a centred bisectrix figure with negative sign when viewed in convergent light. Magnetite. Rounded or subangular sometimes weathered grains of magnetic iron-ore are present in greater or less amount in all the specimens investigated, with the exception of the Burnside mudstone and the overlying greensand at the base of the Caversham Series. Pyrite. Minute concretionary masses of pyrite or marcasite of undoubted authigenic origin are exceedingly plentiful in the concentrates from the Burnside mudstone and the basal Caversham greensand, and fairly abundant in the Goodwood limestone. Otherwise pyrite is absent or rare in the sediments of the Dunedin district. Rutile. Deep-red-brown or yellowish-brown grains of rutile are found in small numbers in concentrates from most horizons. When distinguishable the typical form is a rather slender prism terminated

by a single set of pyramids, but rounded or irregular outlines are more usual. In all cases the form suggests detrital rather than authigenic origin (Figs. 27, 28). Sphene is sparsely distributed as small, rounded, clouded grains in two concentrates only (2760, 2771). Topaz (?). The mineral listed as topaz is represented by a single colourless, glassy grain obtained from a glauconitic band near the top of the Wangaloa Series (2751). It appears from its smooth surface to be a cleavage-flake, nearly rectangular in outline, with medium refractive index (γ = 1.63 − 1.64) and rather low birefringence, and in convergent light yields a perfect centred uniaxial interference-figure with positive sign. Apart from the uniaxial condition, these properties are identical with those of topaz and are distinct from other minerals with which the writers are familiar. The known variability in the optic axial angle of topaz seems to warrant doubtful identification with that mineral. Tourmaline is consistently abundant in the sandstones of the Taratu Series and the Burnside loose sandstone, and also is present as a minor or rare constituent of rocks from all other horizons except those of the Caversham Series. Several forms and varieties are represented:— (a) Sharply euhedral slender prisms (averaging 0.2 mm. in length) terminated (at one end only) by flat pyramids, and invariably crowded with dense masses of minute black inclusions; the colour is blue, purplish, or commonly blue tipped with brown at the pyramidal end. Tourmaline of this type is widely distributed in the schists of Otago, and may be panned from almost any local specimen of schist. (b) Clear, euhedral rather stout, intensely pleochroic prisms, sometimes doubly terminated, and usually devoid of inclusions. The colour most frequently is yellowish-brown, but clear deep-blue is not uncommon; rarer colours are pale-blue, bluish-brown, blue-green and deep-coffe-ebrown (2754, 2768). (c) Relatively coarse, clear angular fragments of yellowish-brown or deep-blue colour, sometimes reaching 0.25 mm. in diameter (2753, 2754, 2751). (d) Clear blue, sharply euhedral slender prisms, usually terminated at one end only, and devoid of inclusions (specimens 2774–2777). Of these type (a) is distributed through most of the concentrates examined, but types (b) and (c) appear to be confined strictly to the Burnside loose sandstone and the Abbotsford mudstone. Zircon is on the whole the most plentiful heavy mineral in the Dunedin sediments; it is present in every concentrate, usually in abundance, and in the two specimens of Abbotsford mudstone rises to predominance. Typically it is colourless and unzoned, but a pale-pink tint is sometimes perceptible in zircons from the Taratu sandstones, while strongly zoned crystals (Fig. 1) were occasionally

noted (2751). The following forms, listed in order of abundance, are recorded:— (a) Sharply euhedral stout prisms about half as wide as long, terminated by simple bipyramids, usually enclosing irregular linear cavities or rod-like inclusions; sometimes cloudy or smoky (2770) in appearance, more rarely clear and glassy. (b) Crystals similar to the above, but with slightly corroded outlines. (c) Well rounded grains sometimes circular in outline, usually dusty or smoky in appearance (2760, 2759, 2751, 2769). (d) Sharply euhedral acicular crystals (ranging up to 0.7 mm. × 0.1 mm.) with simple pyramidal terminations (cf. Fig. 6), usually glassy-clear and devoid of inclusions (e.g., 2751, 2768). (e) Rather small angular broken grains, either clear (2768) or dusty (2769) in appearance. Types (c) and (e) are common only in concentrates from the Burnside loose sandstone and the various members of the Wangaloa Series. Zoisite. The mineral identified as zoisite occurs as extremely rare grains in three concentrates (2750, 2758, 2759, 2775). These take the form of tables perpendicular to the acute bisectrix, showing a single perfect cleavage and having a high refractive index (estimated roughly as > 1.7). In 2750 it is colourless, but in the other two cases is pale golden yellow with distinct pleochroism (Y > X). Convergent light tests show that the mineral is optically positive, with a small axial angle (in 2759 mean value of 2E* Estimated by the Becke method. = 35° ± 5°), the optic plane being normal to the cleavage. The most striking property is strong dispersion of the optic axes (r > v) as shown by the interference figure (in 2750 the mineral is almost uniaxial for violet light). These properties agree well with those given by Winchell (1933) for iron-bearing or β-zoisite, except for the unusual yellow colour and pleochroism. Iddings (1906, p. 382), however, records zoisite as pleochroic from bluish-green to yellow in thick grains. In general appearance the zoisite from Dunedin is strikingly similar to the monazite of the Bob's Cove sandstone (No. 2763), but is distinguished by the optic orientation and the extreme dispersion. Heavy-mineral Assemblages in Relation to Stratigraphy. The assemblages of heavy minerals characteristic of the various stratigraphic units are summarised below. The minerals are listed in each case in order of average abundance; names in brackets refer to species which may or may not be present. Kaitangata Series: zircon (95–98%), tourmaline of types (a) and (d), magnetite (garnet and apatite). The rarity of tourmaline of type (a) and the absence of epidote distinguish these beds from those of the overlying Taratu Series in the same locality.

Taratu Series: zircon, tourmaline, epidote, magnetite. Abundance of tourmaline of type (a) and complete absence of accessory constituents such as garnet and rutile are distinctive features, though perhaps of only local significance since all three specimens recorded in Table IV were collected from the same general locality, viz., within a mile of Scrogg's Hill. A fourth specimen (quartzsandstone, near Fernhill coal-mine) was completely lacking in heavy minerals.* Sediments of this series (Nos. 2772, 2773) obtained from the Taratu Mine, about 40 miles southwest of Dunedin, differ from the Scrogg's Hill rocks in absence of tourmaline and epidote and presence of appreciable amounts of garnet in the concentrates. Wangaloa Series. (a) Basal glauconitic sandstones (Boulder Hill beds): zircon, magnetite, apatite, epidote, tourmaline of type (a), (rutile, biotite). Apatite and tourmaline are not usually associated in other Dunedin sediments. The presence of small numbers of slender clear zircons distinguishes the assemblage from that typical of the Caversham sandstone. (b) Abbotsford mudstone: zircon, tourmaline, rutile, magnetite, biotite (epidote, garnet, (?) topaz). The presence of plentiful tourmaline of types (b) and (c) and rarity of type (a) are distinctive peculiarities otherwise observed only in concentrates from the Burnside loose sandstone. The predominance of zircon is noteworthy. This mineral is very varied and in addition to the usual stout prismatic crystals, includes many clear acicular prisms, perfectly rounded grains, and sometimes highly angular fragments. The correspondence between the two concentrates examined is striking in view of the fact that one specimen (2751) was collected from a glauconitic band near the top of the series in the neighbourhood of Puketiraki, while the other (2768) is the typical fine-grained mudstone from Abbotsford itself. Burnside Series. (a) Loose sandstone: zircon, tourmaline, magnetite, garnet, (epidote, rutile, andalusite, kyanite, hornblende, biotite, pyrite, zoisite). This is the most constant and distinctive assemblage in the Dunedin rocks. It resembles that of the Abbotsford mudstone in abundance of tourmaline of types (b) and (c) and in the variety of habits displayed by the zircon; tourmaline, however, is much more plentiful in the loose sandstone, while zircon is not of predominant rank. Acicular, clear sharply euhedral zircons, and grains showing a high degree of rounding are always conspicuous. Though rare, kyanite and andalusite are significant constituents, in that occasional grains of one or other mineral have always been noted in the concentrates, though as yet unknown from any other horizon in the district. (b) Calcareous mudstone: pyrite, epidote, zircon, apatite, (rutile, tourmaline). The complete predominance of concretionary pyrite (possibly marcasite) separates this assemblage from any other except that obtained from the immediately overlying greensand of the Caversham Series, which is almost identical in composition.

Caversham Sandstone: zircon, apatite, epidote, magnetite, (chlorite, garnet, biotite, rutile, hornblende, tourmaline, zoisite). The assemblages vary considerably, but the constant presence of relatively plentiful apatite and rarity or absence of tourmaline seem to be distinctive. Typically the zircon is entirely of the stout prismatic euhedral type; though usually the most abundant constituent, it may sometimes sink to accessory rank, while chlorite or epidote may locally dominate the assemblage. Goodwood Limestone: zircon, chlorite, epidote, magnetite and pyrite are equally plentiful in the one concentrate recorded. Provenance. With the exception of pyrite the minerals listed in Table IV are of detrital origin. Some have obviously been derived from the nearby schists, while a few appear to have come from some other source. Tourmaline of type (a) (see above), epidote, garnet, chlorite and much of the zircon belong to the first category. Apatite, magnetite and sphene are also known to be very widely distributed in the Otago schists and most probably therefore are to be referred to this ultimate source. Hornblende has been observed by the writers as a rare relict mineral in schists from eastern Otago, while brown rutile and zoisite may be found later when more is known of the petrography of these rocks. Some differnt source must be assumed, however, for kyanite and andalusite. While the former might conceivably have been rewashed from more ancient sediments such as greywackes that outcrop not far south of Dunedin, this can hardly be possible for the clear fresh andalusite in view of its recognised instability in sedimentary rocks (cf. Milner, 1929, pp. 126, 438). The presence of these two minerals in the Burnside loose sandstone suggests the existence of a mass of high-grade metamorphic rocks invaded locally by granite or some allied plutonic rock, within the distributive province that supplied the detritus concerned. To-day the nearest exposures of such rocks are in Fiordland and South Westland, which are too distant to be regarded as likely sources for detritus deposited in the Dunedin area. Clear tourmaline of types (b) and (c) above, such as is common in the Burnside loose sandstone and underlying Abbotsford mudstone, has not been recorded from schists of the Otago type. Similar tourmalines are common, however, in the high-grade schists adjacent to granitic intrusions in South Westland and Fiordland, and have been described by Williams (1934) in alluvial sands derived from the granite-schist complex of Stewart Island. Since the vertical distribution of clear tourmaline in the Dunedin sediments is limited and includes the loose sandstone to which andalusite and kyanite are restricted, its presence further supports the writers' views as to the source of these latter minerals. Finally it is also possible that other rocks than the local schists furnished some of the zircon found in the rocks under consideration, though the data are too incomplete for generalisation. Two significant

facts may be mentioned, however. In concentrates panned from local schists the common stout prismatic type of zircon has often been obtained, but neither rounded grains nor slender clear-cut acicular crystals have yet been found in these rocks. Dr W. N. Benson has drawn the writers' attention to the occurrence of rounded dusty detrital zircons in Ordovician quartzites from south-west Fiordland. The evidence brought forward above appears strong enough to warrant the following conclusions as to the geological history of the Dunedin area:— (1) The constant scarcity of tourmaline of type (a) in the Kaitangata sediments indicates that they were derived from a terrain in which schists of the Otago types were not prominent. Professor Benson has suggested to the writers that the Kaitangata beds of the Dunedin district were probably laid down at a time during which greywackes still to a large extent covered the schists now exposed in this region. (2) The sediments of the Taratu Series are of purely local origin, and were derived from the low-grade schists upon which they usually lie. The same facts probably apply for the lower beds (Boulder Hill beds) of the Wangaloa Series. (3) The detritus in the upper part of the Wangaloa Series (Abbotsford mudstone) and the loose sandstone of the Burnside Series was furnished partly by the local low-grade schists, and partly by high-grade metamorphic rocks invaded by granites such as outcrop to-day in south-western New Zealand. (4) The distributive province which supplied the material now comprising the Burnside mudstone, Caversham Series, and Goodwood limestone, did not include rocks of this latter distinctive type. (5) Petrographic evidence afforded by the heavy minerals indicates major changes in conditions of distribution of detritus at the following stratigraphical levels:— (a) At the junction of the Kaitangata and Taratu Series. (b) At the base of the Abbotsford mudstone. (c) At the junction of the loose sandstone and calcareous mudstone of the Burnside Series. (d) Immediately above the basal greensand of the Caversham Series. The second of these is not supported by the general lithology (Ongley, 1933), but the third and fourth correspond approximately with the commencement and close of the general marine transgression that accompanied deposition of the Burnside mudstone. The major break assumed by Ongley (1933) at the junction of the Abbotsford mudstone and the Burnside loose sandstone perhaps explains the presence of kyanite and andalusite at the latter horizon alone. If this is so, the possibility that the distinctive clear tourmalines found in both beds were deposited primarily in the Abbotsford mudstone and later were concentrated in the loose sandstone during erosion of the earlier sediment must be borne in mind.

Southern Otago. Two specimens representing the Tertiary rocks that outcrop extensively in Southern Otago north of the Mataura River were examined. The results may be summarised thus:— No. 2756, fossiliferous sandstone, Chatton:—Epidote, zircon and magnetic iron-ore are all plentiful. The epidote is a yellow highly birefringent type, clear except for staining with limonite, while the zircon includes sharply euhedral as well as rather rounded grains. Greenish-brown (rarely green) hornblende is not uncommon, as broken cleavage-prisms. Rare angular grains of clear colourless sphene and pale-pink garnet were also noted. The assemblage is probably derived from the local greywackes on which the Chatton sandstone rests at no great distance from the fossiliferous outcrop. No. 2757, fossiliferous sandstone, Waikaia:—Epidote minerals make up 90% of the concentrate, the commonest variety being a colourless clinozoistic type occurring as subidiomorphic tables of rectangular outline, either clear or crowded with minute black inclusions; clear irregular yellow grains and more opaque granular clusters of a more highly ferriferous epidote are also plentiful. Clear-cut blue prisms of tourmaline (devoid of inclusions) and worn or euhedral zircons are fairly abundant. Hornblende and sphene are rare, while a single highly refringent grain of clear green spinel was also recognised. In the absence of complete data, no speculations as to provenance are offered. Conclusion. When the results discussed above are compared, it is at once apparent that in every locality investigated the sediments appear to have originated locally. This conclusion is highly significant since it implies that in the southern part of the South Island the erosion surface upon which the younger sedimentary rocks rest did not approach the condition of peneplain. Indeed a considerable degree of relief is indicated. Literature Cited. Allan, R. S., 1934. On the System and Stage Names applied to Subdivisions of the Tertiany Strata in New Zealand. Trans. N.Z. Just., vol. 63, pp. 81–108. Benson, W. N., 1934. The Geology of the Region about Preservation and Chalky Inlets, South-west Fiondland. N.Z.—Part I. Trans. N.Z. Inst., vol. 63, pp. 393–432. Benson, W. N., and Bartrum, J. A., 1935. The Geology of the Region about Pieservation and Chalky Inlets, South-west Fiordland, N.Z.—Part III, Petrology, Trans Roy. Soc. N.Z., vol. 65, pt. 2, pp. 108–132. Dana, E. S., and Ford, W. E., 1932. A Text-Book of Mineralogy, 4th ed., New York, John Wiley and Sons. Finlay, H., and McDowall, F. H., 1923. Fossiliferous Limestone at Dowling Bay, Trans. N.Z. Inst., vol. 54, pp. 106–114. Iddings, J. P., 1906. Rock Minerals, New York, John Wiley. Milner, H. B, 1929. Sedimentary Petrography, London, Murby. Ongley, M., 1933. Kaitangata-Green Island Subdivision, Ann. Rept. N.Z. Geol. Surv., no. 27 (n.s.), pp. 12–18.

Park, J., 1909. Geology of the Queenstown Subdivision, N.Z. Geol. Surv. Bull., no. 7. ——1921. Geology and Mineral Resources of Western Southland, N.Z. Geol. Surv. Bull., No. 23. Service, H., 1934. The Geology of the Goodwood District, North-east Otago, New Zealand, N.Z. Jour. Sci. and Tech., vol. xv, no. 4, pp. 263–279. ——1934a. Note on the Occurrence of Clinohypersthene and Enstatite-augite in the “Norite” from Bluff. New Zealand, Trans. Roy. Soc. N.Z., vol. 64, pt. 2, pp. 147–150. Williams, G., 1934. The Auriferous Tin Placers of Stewait Island, New Zealand, N.Z. Jour. Sci. and Tech., vol. xv, no. 5, pp. 344–357. Winchell, A. N., 1933. Elements of Optical Mineralogy—Part II, 3rd ed., New York, John Wiley. Explanatation of Plate 23. Fig. 1. Zircom No. 2751 × 90. 2. " No. 2738 × 50. 3. " No. 2738 × 50. 4. " No. 2738 × 145. 5. " No. 2746 × 55. 6. " No. 2754 × 55. 7. " No. 2759 × 70. 8. Hypersthene No. 2748 × 48. 9. " No. 2749 × 48. 10. " No. 2749 × 43. 11. Epidote No. 2759 × 75. 12. " No. 2741 × 65. 13. Allanite No. 2743 × 50. 14. Kyanite No. 2733 × 42. 15. Kyanite No. 2754 × 50. 16. Tourmaline No. 2742 × 55. 17. " No. 2758 × 82. 18. Apatite No. 2740 × 55. 19. " No. 2740 × 250. 20. " No. 2739 × 100. 21. Monazite No. 2741 × 110. 22. " No. 2741 × 120. 23. Sphene No. 2739 × 40. 24. " No. 2739 × 60. 25. Garnet No. 2742 × 55. 26. Rutile No. 2761 × 250. 27. " No. 2761 × 200. 28. " No. 2761 × 175.

Plate 23