Art. XXVI —The Geology of Banks Peninsula.* This paper is intended to be introductory to one dealing with the petrology of the district, which will be submitted at a later date

Transactions and Proceedings of the Royal Society of New Zealand, Volume 49, 1916, Page 365

Art. XXVI —The Geology of Banks Peninsula.* This paper is intended to be introductory to one dealing with the petrology of the district, which will be submitted at a later date By R. Speight, M Sc., F.G.S., Curator of Canterbury Museum [Read before the Philosophical Institute of Canterbury, 3rd November, 1915; received by Editors, 30th December, 1916, issued separately, 5th November, 1917.] Plates XXIV-XXVI. Table of Contents. A. Introductory. B. General Topography of the Area.     1. Lyttelton Harbour and Neighbourhood     2. Akaroa Harbour and Neighbourhood.     3. Little River Valley, Kaituna Valley, &c. C. Geological History.     1 Trias-Jura Greywackes and Slates at Gebbie's Pass and Head of Lyttelton Harbour, Occurrence of Sandstone at Governor's Bay, Little Quail Island, Charteris Bay.     2. First Volcanic Phase Rhyolites—their Occurrence at Gebbie's Pass, Head of the Bay, Charteris Day, and Quail Island     3 Second Volcanic Phase.         (a.) Lyttelton: Extrusion of Andesitic and Normal Basalts, the Dyke System         (b) Akaroa; Basic Flows, Dykes, Syenite.         (c.) Little River Valley: Not a Separate Volcano     4 Third Volcanic Phase         (a) Mount Herbert Volcano         (b) Mounts Fitzgerald and Sinclair to be assigned to Second Phase     5. Fourth Volcanic Phase. Quail Island     6 Post-volcanic History.         (a) Evidence of Greater Elevation, followed by Depression and Slight Recent Rise         (b) Results of Erosion and the Formation of Calderas A. Introductory. Banks Peninsula as a geological locality has attracted considerable attention at various times, but no detailed account of its geological features has appeared since Haast published in 1879 his Geology of Canterbury and Westland. In this work he summarizes the results of his observations and investigations as Provincial Geologist, reports of which he had furnished to the Provincial Government from time to time, and which were published in its Gazettes A petrological account of the rocks occurring in the Lyttelton Tunnel, based on collections made by Haast while it was being constructed, appeared in Filhol's Mission de l'ile Campbell, which was issued in Paris in 1885, this containing as well a summary of the geological history of New Zealand Since then various petrological papers have been published dealing with the rocks of the district, among which may be mentioned “The Microscopical Investigation of some Eruptive Rocks from Banks Peninsula, New Zealand,” by B Kolenko (Neues Jahrb f. Min, Geol., Pal, 1885, Band 1, p. 1, translation published in N Z. Journ of Science, vol 11, 1885, p 548), and “The Igneous Rocks of New Zealand,” by Hutton (Journ. and Proc Roy. Soc. N S W., 1889), containing reference to many rocks from the peninsula; while the following papers have appeared in the Transactions of the New Zealand Institute: “The

Tridymite-Trachyte of Lyttelton,” by P. Marshall (vol. 26, 1894); “An Olivine-andesite of Banks Peninsula” (vol. 25, 1893), “On a Dolerite Dyke from Dyer's Pass” (vol. 26, 1894), and “On a Soda Amphibole Trachyte from Cass's Peak, Banks Peninsula” (vol. 40, 1908), by the present author. There are also a few other papers on general geological subjects connected with the area, such as a “Note on the Silt Deposit at Lyttelton” (vol. 15, 1883), by Captain Hutton; “Note on an Artesian-well System at the Base of the Port Hills” (vol. 33, 1901), by S. Page and E. B. R. Prideaux; and, lastly, a paper “On a Remarkable Dyke on the Hills near Heathcote” (vol. 13, 1881), by A. D. Dobson. With the exception of several slight references in the writings of others, this is all that has been written on the district; and as no general account of this interesting locality has been issued since 1879, and no comprehensive account of its petrology has yet been published, no excuse need be put forward for this paper. B. General Topography of the Area. (See fig. 1 and Plate XXIV.) Banks Peninsula was discovered by James Cook on the 14th February, 1770, while on his first voyage of discovery. He thought it was an island, and named it after the celebrated naturalist who accompanied his first expedition No landing was made, the nearest approach being about three to four leagues distance. He says with regard to it, “It is of a circular figure, and about twenty-four leagues in compass. It is sufficiently high to be seen from a distance of twelve to fifteen leagues, and the land has a broken, irregular surface, with the appearance of barrenness rather than fertility Yet it was inhabited, for we saw smoke in one place and a few straggling natives in another.” Cook's mistake as regards its being an island can be easily understood when one considers the difficulty of picking up the low-lying land on the south and west from a ship over twenty miles away off shore. The peninsula lies in latitude 43° 32′ S. and longitude 173° 30′ E., and forms a rough elliptical salient on the central portion of the South Island of New Zealand. Its diameter in a N.W.-S.E. direction is about twenty-five miles, and its breadth at right angles thereto about eighteen miles It is bounded on the north-east, east, and south by the sea; but on the other side, looking toward the Canterbury Plains, the place of the sea is taken by the estuary of the Avon and Heathcote Rivers on the northwest, by low-lying and swampy land on the west, the latter passing into marsh and shallow lake on the south, where the flat expanse of Lake Ellesmere lies behind a great shingle-spit whose proximal end is attached to the south coast of the peninsula. Near this point another body of brackish water, known as Lake Forsyth, is ponded back in the lower reaches of one of the main valleys stretching south About forty miles away to the west is the long line of the Southern Alps, and sloping up to their foot lie the Canterbury Plains, formed by the coalescing and overlapping fans of the shingle-charged rivers issuing from the mountain tract. The most striking features of the peninsula are Lyttelton and Akaroa Harbours, occupying old volcanic craters of the “caldera” type, surrounded by crater-rings, which are broken at one place so as to allow of the entrance of the sea into the depressed floors of the craters. The centres of the two craters occupy, as it were, the two foci of the elliptical area constituting the peninsula. The former lies to the north of the area, and stretches in

Photographic reproduction of relief map of Banks Peninsula, from original in Canterbury Museum by S C Farr. This gives a fairly accurate representation of the relief of the region.

an E.N.E.-W.S.W. direction, with its opening to the north-east, while the latter runs south and south-east from near the centre of the peninsula, and forms one of the finest deep-water harbours in New Zealand. The chief summits round the edge of the crater-rings range between 1,800 ft. and 2,500 ft. in altitude, but the highest points in the district are near the centre of the mass on a ridge connecting the rings of the two craters, and formed of territory belonging to some extent to both volcanoes. Immediately to the south-east of Lyttelton is the highest summit, Mount Herbert Peak (3,014 ft.), and five miles south-east from it is Mount Fitzgerald (2,710 ft.), and a mile and a half farther on is Mount Sinclair (2,763 ft.), the last-named lying just outside the Akaroa basin, on the dividing ridge which separates two important valleys—viz., Little River Fig. 1. and Pigeon Bay—from each other These two latter peaks are situated almost in the heart of the peninsula, and form its geographical centre, so to speak. Stretching down from the elevated ridges around the two foci, with rude radial orientation, are numerous valleys whose lower reaches on the eastern side are occupied by the sea and form a series of deep bays, the principal being Port Levy, Pigeon Bay, Little Akaloa, and Okain's Bay, facing north-east, Le Bon's Bay and Long Bay, facing east; and Peraki, which lies to the west of Akaroa entrance, facing south-west; while on the landward side the corresponding indentations are filled with alluvial material, the chief being Gebbie's Valley, leading to the head of Lyttelton Harbour; Kaituna Valley, lying to the south of Mount Herbert and cutting far back into the major axis of the ellipse; and beyond it Little River, whose lower reaches are occupied by Lake Forsyth. All these valleys are short, the

longest ten miles in length, and their streams are diminutive torrents which fail altogether or carry but little water in dry weather. In their lower reaches their floors are flat, owing to the aggradation of the streams incompetent to carry their load, and to material which has been swept in by wave and tide and deposited in the sheltered waters at the head of the submerged valleys. This action is well seen in Okain's and Le Bon's Bays, whose streams are tidal for some distance near their mouth; but in these the filling-in has not reached such a mature stage as in the case of Peraki This effect is more marked on the southern and eastern coasts, which experience to a greater degree the strong northerly drift sweeping up the coast and dropping its load of coarse material on the southern margin of the peninsula, while the fine material is carried farther and contributes markedly to the filling of the lower reaches of the bays, if it be not swept into the deeper water off shore and thus be removed beyond the sphere of influence of waves and shore currents. When viewed from the sea the land presents, as Cook said, a bold, irregular surface, and the exposed headlands are terminated in high sea-cut cliffs, which reach a height of nearly 800 ft. on the eastern edge of the land. Here they are exposed to the full force of the gales from the south and east, under whose influence the waves have cut back all the headlands which project in that direction, in marked contrast to the spurs which project into the plain, whose terminations are hardly truncated at all, except near the coast, where within comparatively recent times they have been subject to the action of the sea. When the peninsula was first discovered by Europeans it was almost completely clothed with forest, the only bare patches being the tops of the highest hills and on the extremities of the spurs reaching down to the sea; but this has been cleared off as settlement progressed, so that patches of forest more than a few acres in extent are few and far between The rainfall amounts to between 30 in. and 40 in. per year, and is well distributed over the entire period, so that under its influence and with the advantage of a rich soil the hills are excellently adapted for pastoral purposes, and are noted for the rich grasses with which they are covered and for the excellent stock they produce. After this general description I pass on to the more particular account of the principal physical features—viz, Lyttelton and Akaroa Harbours; and, after dealing with them, to the more important minor ones, such as Little River Valley, Kaituna Valley, Port Levy, and Pigeon Bay. 1. Lyttelton Harbour. (Plate XXV, figs 1 and 2) Lyttelton Harbour is about eleven miles long by three wide in its widest part, which is opposite the town of Lyttelton, its general width being from one and a half miles to two miles; the entrance is one mile in width The northern side of the harbour is but slightly indented, and the land rises steeply to the crater-ring, the highest points being Mount Pleasant (1,638 ft.), just behind the town of Lyttelton; the Sugarloaf (1,630 ft), farther west; Cass Peak (1,780 ft.) and Cooper's Knob (1,880 ft.), towards the western end of the harbour. A little distance beyond this elevation the crater-ring is completely broken down, and the divide between the inside and the outside slopes of the harbour is for a space of about three miles reduced to approximately 400 ft., and in two places, first at the head of Gebbie's Valley and again a little farther east at the head of the parallel McQueen's Valley, the ridge is reduced below that height and forms two passes, which are known from

Fig 1.—Lyttelton Harbour, from near Cass's Peak, showing Quail Island and stream-eroded valleys on the interior of the caldera. Fig 2—Lyttelton Harbour from summit of Mount Herbert, showing drowned-valley topography of upper portion of the harbour and part of the long spur reaching down from Mount Herbert.

Fig 1—Akaroa Harbour from summit of Mount Bossu, looking north, showing stream-eroded valleys in interior of the caldera Fig. 2.—Akaroa Harbour from summit of western side of the crater-ring, the stream-eroded slopes of Barry's Bay valley in the foreground, and the entrance to the harbour in the distance.

Fig 1—View of Porter River, showing two beds of limestone with interstratified tuff. The angle of the river is just at the point of inflow of Home Creek Fig 2—Great fall of limestone rocks at lower gorge of Porter River The small debris slope on the middle, as well as the surface of the large block above it, is full of shells The shell bed as it lies in position is near the top of an inaccessible cliff to the right The tuff beds underlying the upper limestone are to be seen on the left side of the picture, just across the stream, and the limestone appears in position above them

Fig 1—View of the basin taken from the summit of Castle Hill, looking south-east The limestone in the foreground is dipping towards the observer, the fields in the middle distance are probably based on greensands and tuff beds, and the distant hills are the greywackes of Mount Torlesse, the ends of the spurs of which are faceted either by glacial action or by faulting—in all probability the latter Fig 2—View of the basin looking south-west from the hill between Broken River and Porter River The limestone in the foreground forms here the lower gorge of the Porter River, but it appears again farther up-stream after making a swing round to the left, and is seen again in the distance forming the eastern face of Castle Hill The Pareora beds are seen between the two gorges; the characteristically terraced landscape reposes on these beds The general conformity of the stratification is clearly seen Text fig. 4 follows along the eastern face of Castle Hill as seen in the plate

the names of the valleys which they head. Their heights are as follows: Gebbıe's Pass, 360 ft.; and McQueen's Pass, 350 ft. Immediately to the east of McQueen's Valley the crater-ring rises to 1,843 ft. in Dyke Hill, so called from two prominent dykes on its northern slope, and this is succeeded by Kaituna Pass, on the eastern side of which rises like a bastion the steep slopes of Mount Herbert (2,805 ft.); this passes into Herbert Peak, from whose summit a long gentle slope according with the angle of the lava-streams which form it reaches down to the harbour. This is cut off on the eastern side by the valley which leads down to Purau Harbour, and is then succeeded by the remnants of the crater-ring which he between it and the southern side of the entrance At the head of the harbour there are two well-marked peninsulas which determine the position of three considerable indentations of the shore-line—viz, Governor's Bay, Head of the Bay, and Charterıs Bay—the first known as Manson's Peninsula and the second as Potts Point; while on the eastern side of the Mount Herbert slope there lies Purau Bay, another deep indentation. These are all instances of depression topography, and the bays are drowned valleys with their lower reaches prolonged beneath sea-level at the same grade as the exposed floors of the valleys. In the middle of the harbour, in practically the centre of the crater-ring, lies Quail Island, 184 acres in area, seven-eighths of a mile long by half a mile broad, and its highest point 282 ft above sea-level, with steep cliffs 200 ft. high facing the harbour, but with gentler slope on the other sides. It lies off the end of the peninsula dividing Charteris Bay from the Head of the Bay, and at low tide it is possible by wading to cross the space between the two. About half-way over there is a rise in the connecting ridge which is permanently above sea-level and is known as Little Quail Island. The harbour is very shallow in its upper parts, and at low water there are extensive mud-flats, the water gradually deepening towards the entrance, where it is about 10 fathoms deep; the floor is almost flat, the deep water continuing right up to the rocky wall-like shores. In one place, just opposite to the entrance to the breakwater, a rocky area rises above high-water mark, and at another place, about a mile nearer the heads, opposite Ripa Island, a solid rock rises to about 10 ft. of the surface. With these two exceptions the floor of the harbour is entirely covered with mud and any irregularities are completely masked. Lyttelton Harbour was described by Haast as a caldera, the entrance between the heads forming the barranco, and his explanation was adopted by Hutton. Allowing for the modification of the southern wall by the slope reaching down from Mount Herbert Peak, the name, as used in its widest sense, is applicable. It has, however, been considerably modified by stream action: the crater-wall has been broken down completely at Gebbie's Pass, and the majority of the topographic features are due to stream erosion and not to the effects of a paroxysmal explosion. This point will be dealt with more fully when the geological history is considered. 2. Akaroa Harbour. (Plate XXVI, figs. 1 and 2) Akaroa Harbour is of characteristic caldera-like form, with the entrance at the heads forming the barranco. It is fiord-like in some of its features, eleven miles in length and three miles wide in its widest part. The entrance is a mile wide, and flanked by perpendicular cliffs, the southern being 525 ft. in height and the northern about 300 ft. Inside the northern entrance,

however, there are cliffs which rise to 500 ft. The summit of the crater-ring is in a remarkable state of preservation, and nowhere does it sink below 1,000 ft. in height, while the prominent peaks rise to just over 2,000 ft. The most important of these are: On the western side—Mount Bossu (2,386 ft.), Carew's Point (2,598 ft), Saddle Hill (2,758 ft), French Peak (2,675 ft.), Rocky Peak (2,297 ft); and on the eastern side—Duvauchelle Peak (2,406 ft.), Okaın's Peak (1,880 ft), Laverick's Peak (2,478 ft.), Brasenose (2,375 ft), and Flag Peak (2,668 ft), immediately above the town of Akaroa. The inside of the crater-ring shows well-developed signs of stream erosion, and the valleys so formed are prolonged beneath sea-level. These are best developed near the head of the harbour, where we have the indentations known as French Farm Bay, Barry's Bay, Duvauchelle's Bay, Robinson Bay, and German Bay—all valleys with their lower extremities drowned by the sea, and divided from each other by ridges which are cut back at their extremities to some extent by wave action Other indentations of the shore which do not show this so markedly are French Bay (near which the town of Akaroa is situated), Wamui and Lucas Bays, two embayments in the western side The most prominent of all the ridges which stretch into the harbour is the pear-shaped peninsula of Onawe, one mile in length, which rises at its termination into a rounded hill, 348 ft in height, from which the peninsula stretches back with decreasing height and narrowing cross-section till at the isthmus joining it to one of the ridges which descend from the crater-ring it is almost cut across by the sea The general soundings of the harbour as disclosed by charts indicate a gradual deepening from the northern end to the entrance, and suggest that the floor is merely an extension of the gradient of the valleys which are formed at the head The topographic features thus present a striking analogy to those of Lyttelton Harbour, just referred to, but Akaroa is the more typical caldera of the two 3. Little River Valley, Kaituna Valley, Etc Little River Valley, the chief one on the southern side of the peninsula, was considered by Haast to be one of the eruptive centres from which its rocks were poured out In the opinion of the present author its formation can be attributed almost entirely to stream erosion The main valley is about ten miles in length, the lower portion, six miles in length with a breadth of approximately one mile, being occupied by Lake Forsyth, which is separated from the sea by a narrow bar of shingle When Europeans first visited the spot it was open to the sea and used as a boat-harbour by the Maoris, but the continual drift of shingle up the coast has now completely cut it off from the sea The lake is shallow, and the water is brackish Just above the head of the lake the valley divides, the eastern branch, called the Okute Valley, reaching up to the crater-ring of Akaroa between Saddle Hill and French Farm Peak It is further divided into the Western Valley, which runs north-west towards the head of Port Levy and is excluded from the slopes which belong to the Akaroa crater-ring The main valley, however, continues with several small branches, and drains a considerable portion of the north-western slopes of the Akaroa volcano, as well as a large portion of the ridge which stretches west from it and passes through Mount Sinclaır and Mount Fitzgerald and divides the Little River Valley from Pigeon Bay and Port Levy on the north Mount Sinclair is, however, the most striking physical feature at its head The walls of the valley are steep, and are formed of long ridges whose terminations are between 300 ft and 400 ft

in height, but which rise in places to close on 2,500 ft., and it is specially noteworthy that the highest points of these ridges do not occur near their proximal end Kartuna Valley is about seven miles in length, and runs in a general south-westerly direction along the south-eastern side of Mount Herbert. Its head reaches back to the main divide of the peninsula which joins Mount Herbert on to the crater-ring of Akaroa. Its walls are steep, since it has followed generally the line of flow of the lavas from Lyttelton—i.e., it has the characteristics of a stream-valley running with the dip of the beds Its floor is flat and deeply covered with alluvium, but about one-third of the way up from the entrance it contracts somewhat where the valley takes a right-angled turn, and farther up it opens out into a wide amphitheatre-like head which a number of valleys of the region present; it thus reaches to the head of the Little River Valley on the east To the north of the main divide he Port Levy and Pigeon Bay, the former to the east of the Lyttelton crater-ring, but eroded for the most part from the lavas poured out of that vent; but a part of the walls on the eastern side of the valley no doubt belong to Akaroa. Its head is divided up into separate valleys like Little River, and it presents few of the features which would suggest that it was a separate centre of volcanic action Pigeon Bay has similarly been eroded out of the flanks of the Akaroa volcano, and it closely resembles in form and depth the adjoining bay on the west. It has, however, a broader-headed valley, which has cut into the crater-ring of Akaroa on the eastern side, and farther west reaches to the main divide of the peninsula, where he the high peaks of Mounts Sinclair and Fitzgerald Neither Pigeon Bay nor Port Levy is truly radial in arrangement, and both appear to have been eroded to some extent out of ground where the flows from the two volcanoes mutually interfered with each other and were probably intercalated along the line of junction. The drainage areas of Little Akaloa, Okam's Bay, and Le Bon's, and also that of Peraki, on the south side of the peninsula, reproduce on a smaller scale the features of Pigeon Bay and Little River, and have steep-walled sides and amphitheatres at their heads C. Geological History. 1. Trias-Jura Sedimentary Series. (See fig. 2.) The oldest rocks exposed on the peninsula consist of a series of slates and greywackes with beds of chert and jasperoid rock occurring to the south of Lyttelton Harbour. They form almost the whole of the spurs which stretch down from the Gebbie's Pass ridge towards the head of the harbour, also the points round the indentation known as the Head of the Bay, and stretch in an easterly direction across the base of Potts Peninsula, extending for half a mile up the valley and on its eastern side, projecting as a narrow fringe under the later volcanics. They form the basement beds under the flat-topped ridge running north from Mount Herbert towards Potts Peninsula, being exposed on both sides under the cap of basalt forming its summit Greywackes and slates also occur on the western flanks of Mount Herbert across the divide between the harbour and McQueen's Valley, forming the rocks round the heads of the streams running therefrom, and reappearing as inlırs on the western side of McQueen's

Valley, and also on the end of the spur dividing that valley from Gold Valley. These rocks are well developed on the western end of the Gebbie's Pass ridge, and form the country round the heads of the streams running therefrom both to the north and the south. The precise limits of this series are at times difficult to determine, since a covering of soil masks the surface and good exposures are rare except where the rock has been quarried for road-metal or building purposes. These beds are so siliceous in places that many years ago they were prospected for gold, and numerous shafts and Fig 2. drives were made, but without success The strike of the beds is generally in a north-and-south direction, with variations to the east and west about that mean; a specially heavy band at the western head of Gebbie's Valley shows a north-west orientation Up to the present no fossils have been discovered by which their true age may be determined, but judging from their lithological characters merely they are of the same age as those which occur in the mountain

region of Canterbury, and have been definitely determined as Jurassic from plant fossils found at Mount Potts, Clent Hills, and Malvern Hills, and therefore should be assigned to the Maitai system of Marshall, which is of Trias-Jura age Overlying these submetamorphic rocks is a series of different texture, having more the character of a freestone, which has been classified previously as later than the rhyolite volcanic phase,* but which more detailed examination has convinced me represents a deposit antedating the rhyolite eruptions. These beds are developed at the head of Governor's Bay, where they are at least 30 ft. thick, and may be much more, as the upper and lower boundaries of the occurrence are nowhere visible. The rock is here pinkish in colour, free in the grain, and without well-marked bedding-planes. Under the microscope it is highly quartziferous, with some admixture of feldspar and occasional shreds of mica, and therefore is similar in mineral composition to a greywacke, and may have been derived from the same land-mass as furnished the associated greywackes. This bed has, however, yielded no fossils by which its age and true relations can be determined. Although the contacts of this occurrence are not clear, exposures of* J. Von Haast, Geology of Canterbury and Westland, 1879, p 327. similar rock at Little Quail Island, at the end of Potts Peninsula, and about half-way along its western side show that it undoubtedly underlies the rhyolite, and farther round the head of Charteris Bay near the wharf, and on the peninsula to the west of the bay, this facies of the sandstone also underlies the rhyolite, and overlies directly, without the intervention of volcanic material, the basement slates and greywackes. It is therefore very probable that the Governor's Bay occurrence is in the same stratigraphical position, but whether it rests conformably or unconformably on the greywackes it is impossible to say at present. This rock has been quarried for building purposes in several places, especially at Governor's Bay, Little Quail Island, and on the eastern and western sides of Charteris Bay. Where any stratification can be observed the beds strike N E.-S.W., and in places this has been so perfect that large flags suitable for paving and for the making of grindstones were easily obtained. Round the shores of Charteris Bay and at Little Quail Island these sandstones are completely intersected by the trachyte dykes of a later age, which will be referred to hereafter. 2 First Volcanic Phase. The Jurassic rocks were deeply eroded before the first phase in the volcanic history of the district began, and the unconformity is most marked, since the first volcanic beds he right across the edges of the Jurassic sedimentaries The matter produced at this stage is exclusively of rhyolite, and is typically developed near the head of Lyttelton Harbour. Rhyolite rocks are visible about half-way along the cliffs of Quail Island facing the town of Lyttelton, extending in one place almost to the summit of the island, and they reappear on the south side, where they form the solid material of the foreshore and the slopes immediately behind it, flows and beds of agglomerate stretching without intermission, except from dykes, from near the wharf at the east end of the island, behind the quarantine buildings, to its extreme south-westerly point. The rhyolites are exposed along the eastern shore of Charteris Bay, above the narrow band of sedimentaries occurring there, and again on the southern shore of this bay,

and after a break they continue west, forming the greater part of the peninsulas stretching into the upper part of the harbour On the eastern one the rhyolite occurs only on its distal end, the proximal end being of sedimentaries overlaın by basic volcanics, but on the western peninsula the rhyolite extends from its extremity right up to the base of the steep slope immediately below the prominent peak of Cooper's Knob, on the Lyttelton crater-ring. After an interval of sedimentaries it appears again in the vicinity of Gebbie's Pass, forming the prominent hills in its immediate vicinity, stretching down on either side of the stream running therefrom, and forming the main mass of the spur between Gebbie's Valley and McQueen's Valley, an isolated block occurs on the eastern side of the latter valley, and a small outcrop also occurs on the east side of the Teddington Valley, to the north-west of Mount Herbert, just opposite Mr Wilson's house The only evidence of a wider extension of the rhyolite under the covering of later basics is furnished by a tuff containing rhyolite pebbles on the east side of Onawe Peninsula, Akaroa, this being the only known occurrence on Banks Peninsula outside the Lyttelton area. In several places, as at Quail Island, near the summit of Gebbie's Pass, and on the shore of Charteris Bay, the lowest beds exposed consist of agglomerate, composed of subangular boulders of all sizes up to 3 ft in diameter In McQueen's Valley there is a well-developed breccia with pitchstone fragments. Among the rhyolitic material are occasional foreign elements, such as pebbles of greywacke and fragments of an andesite differing in character from the olivine-bearing andesites or andesitic basalts that were extruded later. Where this came from it is quite uncertain, since no exposure of similar rock in position has been located The rhyolitic material includes banded rhyolites with irregular wavy bands of mıcıospherulites gaınetıferous rhyolite mica rhyolite of various shades of colour—green white, and pink—as well as fragments of pitchstone The mass has weathered into fantastic forms, with large globular cavities a striking feature Wherever these fragmentary deposits occur they are apparently at the base of the series but in some cases solid rhyolite flows rest directly on greywacke and slate The rock is typically of white colour, but is occasionally pinkish In some places it shows dark-coloured phenocrvsts of smoky quartz, and in others sanidine crystals 1½ m in length stand out on weathered surfaces In many places however, phenocrysts are not visible to the naked eye At the back of Quail Island it is markedly spherulitic in places, with spherulites up to 3 m in diameter but here, as in other places, it appears to have been markedly sılıcıfıed after eruption, no doubt by the action of warm waters laden with siliceous material percolating through the mass of rock. The alteration produced by this action is in some instances very pronounced, and as it has involved other rocks as well it is sometimes extremely difficult to determine the relations to the beds with which they are in contact I have found this specially so on the shores of Charteris Bay The rhyolite is penetrated by dykes of trachyte and basalt belonging to a later period and also by dykes of rhyolite and pitchstone which are apparently contemporaneous with the rhyolite-flows The most note-worthy of these occur near the crest of the Gebbie's Pass ridge, on the summit of the hill west of the road The pitchstone dyke strikes east and west and is 50 ft in width, and close alongside it to the north is a dyke of rhyolite with similar orientation and well-developed columnar structure Another rhyolite dyke forms the summit of a hill to the north-west of this

On the southern side of the ridge Haast notes a dyke running north-east across the gully up which the road goes It appears as a well-defined mass of markedly silicified rhyolite resembling a dyke in its field occurrence, but in all probability is a flow which has been tilted by earth-movements subsequent to its extrusion. It is difficult to locate the centre from which the rhyolites were erupted. The dykes—excluding from consideration those of later date, which are of no value in this connection—do not show any radiating arrangement (fig 3), but the flows and ash-beds on Quail Island show a well-defined dip to the north, and those in Governor's Bay, though somewhat irregular. dip to the west and north-west in general, and those in Charteris Bay to the east, so that it seems probable that the centre should be located somewhere near the area now occupied by the Head of the Bay. The present thickness of the deposits, allowing for denudation, does not point to any mass of material having been poured out at all comparable with that from other centres of rhyolitic activity in the province, and as the direction of the flows at the commencement of volcanic activity would in many cases conform to the shape of the land surface on which the volcanic material was poured out a few observations of the inclinations of beds in a case like that under consideration may give a somewhat unsatisfactory result. This is well exemplified on the west side of Potts Peninsula, where the rhyolite-flows overlie immediately the earlier sedimentaries, and exhibit a marked variation in the direction and amount of inclination. The clearest case of bedding is to be observed on the south side of Quail Island on the point between the two groups of buildings, and also on the south side of the point to the south of the Leper Station In these places the strata dip to the north and north-east at angles of about 20°, so that we can say for certain that the volcanic focus lies to the south or south-west of Quail Island An interesting deposit of fossil leaves and stems of dicotyledonous plants is mentioned by Haast* J Von Haast, loc cut., p. 326. as occurring on the south side of Gebbie's Pass in sandy shales associated with coarse sands and loose conglomerate. These have a strike in an E N E - W N W direction—that is, parallel to the great band of rhyolite thought to be a dyke by Haast. This is distant only a few chains, and its steep inclination accords with that of the leaf-beds, which dip to the S.S E. at very high angles, approximately 90°. It appears to me, therefore, that the inclination of both the band of rhyolite and the leaf-beds is to be attributed to tilting after deposition as a result of folding or other deformational movement. As far as can be ascertained, the material of which the sands and conglomerate are composed is entirely volcanic in origin, and it is probable that it represents the lowest beds of fragmentary rhyolitic material which occur elsewhere, and that the plants are the remains of the vegetation on the land surface on which the volcano was originally established, and formed in much the same way as the deposits of timber in the pumice to the east of Ruapehu and Tongariro, in the North Island, which is a remnant of the old forest covering of the land in the vicinity of the volcanoes. This idea is supported by the fact that a short distance farther up the valley is a small inlier of Trias-Jura sedimentaries, which can be only a few feet lower in the stratigraphical sequence than the plant-beds, and which no doubt lie close under them at the spot where they occur No good sections are available, however, and the field relations are obscure

Outpourings of rock of identical lithological character occur in other localities of the Canterbury Province—e.g., at Malvern Hills, Rakaia Gorge, and Mount Somers—and it is only by reference to them that we can fix the age of the Lyttelton rhyolitic eruptions The evidence for the period of the former is as follows: In these localities rhyolites lie invariably on an eroded surface of Jurassic strata, while over them he coal-measures containing the following fossils, bones of Ichthyosaurus, shells of Inoceramus, Conchothyra parasitica, Trigonia, belemnites. The fossil-content of these beds fixes them as Cretaceous. At the base of this series lie a conglomerate formed largely of rhyolitic pebbles and beds formed of detrial matter of rhyolitic origin, so that the rhyolites formed a land surface in Cretaceous times. However, at Rakaia Gorge and at Mount Somers coal-seams occur interbedded with rhyolitic tuffs, from which it is evident that eruptions were taking place while the coal was being deposited. It seems certain, therefore, that the rhyolite eruptions took place at the close of the Cıetaceous period on the margin of the mountain region of Canterbury, and, although it is not safe to draw inferences as to age based purely on lithological resemblance in the case of volcanic rocks separated by a distance of forty miles, it is the only evidence that we have to rely on at present, and therefore we may say, at least tentatively, that the age of the Lyttelton rhyolite is Cretaceous also. 3 Second Volcanic Phase (a) Lyttelton The second phase of volcanic activity was marked by the pouring-out of basic lavas which built up the Lyttelton and Akaroa cones. Structurally they are both composite cones of the normal type, as can be seen from the magnificent sections afforded by the sea-cut cliffs. It will be best to consider first of all the case of Lyttelton, as it is more easily examined, not only because ıoadmaking and excavations are more common near the centres of population, but because the Lyttelton Tunnel has been driven in a radial direction through the wall of the cone, and records of the inclination of the flows and samples of the various rocks encountered were made by Haast,* J. Von Haast, loc cit, p. 355. and a petrological description of these rocks is given by Filhol† H. Filhol, Mission de l'île Campbell, Géologie, 1885, p. 66.. These show that Lyttelton has all the features of a composite cone built up of layers of lava and fragmentary material, the latter being thicker as the interior of the volcano is approached along the line of the tunnel. Judging from the quaquaversal dip of the flows observed on the cliffs and on the sides of the deep-cut valleys, as well as from the records of the tunnel, the centre of the harbour corresponded to the centre of activity, but owing to the occupation of the bottom of the crater by the sea the site of the actual vent cannot be definitely located. It can perhaps be fixed with reasonable certainty as a deduction from the orientation of the dykes which traverse the walls of the crater-ring. Their arrangement is markedly regular, and when they are plotted they are found, with few exceptions, to radiate from a small area just south of Quail Island, which may therefore be regarded as the actual centre of disturbance and the precise locality of the vent. The exposed limits of the rocks of this series to the north and west of Lyttelton Harbour are easily determined, seeing that they continue down to

the level of the plains; but in places artesian bores have struck them some distance out from the toot of the hills, so that their actual extension in that direction is quite uncertain. To the east of Gebbie's Pass they form the ridges between Gebbie's Valley and Kaituna Valley, and extend therefrom to the north along the western flanks of Mount Herbert, and probably continue round the western heads of the streams discharging into Charteris Bay. After a considerable gap, where they are covered with later rocks, they appear again at the head of the Purau Valley and form all the mass of hills lying between Purau Bay and Port Levy. They may reach as far eastward as the ridge between Port Levy and Pigeon Bay, since the flows exposed on the eastern shore of the former have a distinct inclination to the east; but as it is apparently impossible to separate in that locality the flows coming from the direction of Akaroa and those from Lyttelton the boundaries in that part of the periphery of the volcano are uncertain. No doubt at one time they extended right across the entrance to the harbour, and across the gap at Gebbie's and McQueen's Passes, but have been removed by erosion in those sectors. Remnants of these covering beds are to be seen on the ridge between Gebbie's Valley and McQueen's Valley. Rabbit Island, or Motukarara, is an isolated fragment lying about half a mıle off the western bounding ridge of Gebbie's Valley, surrounded on all sides by lake sediments, and proving by its position a former wider extension of the limits of the volcano in that direction. This remnant owes its preservation to the resistant character of the rock of which the low hill is composed. With one exception, there are no apparent difficulties in interpreting the structure of the locality as far as these rocks are concerned. On the south side of McQueen's Pass and extending across McQueen's Valley from its western side to the east side of Gold Valley is an occurrence of basic rock associated with the rhyolite whose position is difficult to account for except on the supposition that it is older than the rhyolite. These basic rocks rest on cherts, and are apparently overlain by the acid variety. The appearance may however, be deceptive, but it should be noted. If it represents an old volcanic flow subsequent to the rhyolite, then the surface contours must have been extraordinarily steep when the extruded material flowed over them. At the only places where contacts of the rocks of the Lyttelton series of volcanics with underlying beds are visible they rest either on greywackes, slates, and sandstones of the Trias-Jura series, or on rhyolites, their relations to each of these being well seen on the east and west sides of the Gebbie's-McQueen's Valley depression, where the original covering beds have been removed by erosion and the basement series exposed. The lava-flows and ash-beds of which the cone is constructed are exclusively basic in character, they vary from fine-grained basalts to those in which feldspar phenocrysts form a considerable bulk of the rock. On account of this feature and their consequent high percentage of silica (up to 55 per cent) they have been classified as andesites, but they contain normally a considerable amount of olivine, so they should more properly be called basalts of andesitic habit. Haast records the presence of trachyte-flows in the Lyttelton Tunnel, but this is probably incorrect, as in my experience trachytes occur only in the form of intrusions, and the occurrence in the tunnel may be merely an overflow from a dyke or a sill. The general high percentage of feldspar phenocrysts seems to indicate that the lava-flows were fairly viscid, and therefore the inclination of the beds is steeper than would be expected in the case of a normal basalt.

From a consideration of their average inclination, especially in the case of those recorded by Haast from the Lyttelton Tunnel, some estimate may be arrived at as to the former height of the Lyttelton volcano. The average inclination of the flows in the tunnel is almost exactly 15°, and this value corresponds with that obtained from observations of prominent flows exposed on the sides of valleys eroded deeply into the flanks of the volcano. Taking this value as approximately correct, and also the distance of the outer fringe of the hills from the centre of the volcano as approximately six miles, the height of the cone must have approached 8,000 ft, and if we make due allowance for a probable greater inclination of the flows and an increased thickness near the vent, and also for the depression of the land which has occurred since volcanic activity waned, it is possible that the cone approached, if it did not actually exceed, 10,000 ft in height. The present form of the cone is no doubt entirely different from that which it presented at the close of this volcanic phase. Instead of the usual moderate-sized crater at the top there is now a great hollow, and evidence suggests that this form had developed to some extent before the next phase in the volcanic history began. The actual cause of the formation of these vast cavities will be dealt with later, but the following three working hypotheses have been put forward to account for them:— (i.) They have been formed by explosion or collapse of the cone, the former indicating a revival of volcanic activity. (ii.) The form is due to peripheral faulting causing subsidence of the original crater. (iii.) They have been eroded by the action of streams, or by the sea, or more probably, in some cases at all events, by a combination of both processes. The Dyke System (see fig 3)—A most striking feature in connection with this volcanic cone is the dyke system. The great majority of these intrusions are of trachyte, decidedly alkaline in composition, and frequently containing as a consequence alkaline augıtes and hornblendes, though common hornblende trachytes also occur, as well as ordinary fine- and coarse-grained basalts and andesitic basalts. Many of the trachyte dykes are very massive, such as those on the Lyttelton-Sumner Road, notably the tridymite trachyte, dykes near the head of Heathcote Valley, the great dyke at Rapaki, those near Kennedy's Bush, and on Dyke Hill, near Kaituna Pass—the last four of which are each approximately 60 ft in thickness. There are besides many others of smaller size, so that the volume of intrusive trachyte as deduced from exposures at the surface exceeds in volume all other dykes. There does not appear to be any definite grouping of the trachytic and basic dykes into pairs so that one might be assumed to be complementary to the other, therefore any explanation of their different chemical composition on the basis of a divergence from a common magma requires that the differentiation took place at some depth and not near the surface of the volcano. There is, however, a general tendency for groups of trachytic or groups of basic dykes to occur in particular localities. For example, along the Lyttelton-Sumner Road there is a special occurrence of trachytes; the same remark applies to Heathcote Valley, near the Bridle Path; then, again, there are the large trachyte dykes at the head of the Rapaki Valley, and numerous smaller ones near Victoria Park and near Kennedy's Bush. Basaltic dykes occur freely to the west of Dyer's Pass Road, near Cooper's Knob, and in places near McQueen's Pass in greywacke and rhyolite, forming well-defined groups, but, all the same, they are

occasionally found sandwiched in between trachytes, so that a definite pronouncement must be taken with considerable reserve. Dykes of both species appear at the very crest of the crater-ring, but it is remarkable that the trachytes, though apparently soft and non-resistant. stand out freely as massive walls sometimes as much as 70 ft. above the surrounding country. As might be expected, the basic dykes are more common on the outskirts of the volcano, while the massive and more viscid trachytes are as a rule short in length and somewhat lenticular in shape. There are however, exceptions to this general rule. On the south side of Quail Island the trachyte dykes are unusually numerous, and they are apparently the only ones occurring. The foreshore from the outer wharf to the extreme south-westerly point of the island is intersected by them, nearly sixty occurring in the space of a mile in length. Fig. 3—Map showing the system of radiating and divergent dykes belonging to Lyttelton Harbour They vary in size from mere ribbons up to masses 12 ft across. In some cases injection appears to have taken place twice up the same fissure, or the original dyke appears to have been disrupted and injection taken place up the fissure so formed. Most of the dykes are vertical or nearly so, but occasionally they are flat like sills and are injected along the planes of flow of the rhyolıte. Friction breccias are also sometimes in evidence. The orientation of the dykes in this locality varies through a right angle, so that looking south from the shore they he in the quadrant extending from south-south-east to west-south-west, with one or two having a more east-and-west trend. The great majority, however, are directed between south and south-west. Owing to the variation, they occasionally cross, but there is no regularity in the direction of those crossing later from which a conclusion may be drawn as to the existence of more than one focus of eruption, the variation in all probability being due to proximity to the single centre of

disturbance. All these dykes are white in the hand-specimen, except where they are stained by oxide of iron, which has been derived from the scanty amount of ferro-magnesian mineral originally present in them, and they are no doubt connected genetically with those occurring in the later basalts, and are not in any way related to the rhyolites in which they are included. Round the shores at the head of the harbour the dykes exhibit extreme irregularity in orientation. On the western shore of Governor's Bay they strike S.E.-N.W., but on the point running north from Allandale they not only exhibit the same direction, but others have a more easterly trend, while others again, especially towards the end of the peninsula, lie almost N.-S.; even here, however, large dykes have a S.E.-N.W. orientation. On the east side of this peninsula a large dyke runs parallel with the shore—that is, N.-S.—but it is cut by another running E.N.E.-W.S.W. On following the shore to the south their direction averages about S.E.-N.W., but there are several large ones with a N.-S. direction, with a few running E.-W. In the rhyolite quarry on the beach a large dyke strikes S. 15° W. In Charteris Bay there is the same variation in orientation. On the road near the wharf the dykes strike generally N.W.-S.E. but close to the wharf a dyke runs N.E.-S.W. On the southern shore of the bay their direction is various, but chiefly in the sector N.W.-N.E. and S.E.-S.W. There are isolated occurrences outside even these wide limits, for example, a dyke exposed on the road over the hill to Teddington has a strike E. 25°S.-W. 25° N. On Potts Peninsula there is also marked irregularity in direction. At its extremity, opposite Quail Island, they strike N.E.-S.W. N.-S., and also N.W.-S.E., with an occasional one E.-W. On the west side of the peninsula the majority strike E.-W., but numerous others intersect them at all angles, so that they do not appear to radiate from any common centre. In the Gebbie's Pass locality the only dykes, apart from the rhyolites, that I have been able to locate are of basalt, with one possible exception in McQueen's Valley. These are oriented in the direction of Quail Island—that is, in the normal direction of those on the outskırts of the volcanic cone. In this sector it is evident that the tough and resistant greywackes and slates proved too strong for the more viscid trachytes, although the liquid basalts were able to penetrate them freely. A specially interesting occurrence is that of the tridymite trachyte on the Lyttelton-Sumner Road, described by Marshall.* P. Marshall, Tridymite Trachyte of Lyttelton, Trans. N.Z. Inst., vol 26, 1894, pp. 368–87. The inclination of the mass at an angle of 40° to the vertical, plastered, as it were, against the inner side of the crater, certainly encourages his opinion that it was a flow and not a dyke, but an opportunity of more complete examination afforded by a clearing of the ground has confirmed me in the opinion that it is intrusive in origin. Its lower surface has a dyke contact and not that of a flow, and, further, it is oriented in the proper direction for dykes in that locality. It is cut by two other dykes, both trachytic in character, although one, a hornblende trachyte, is more distinctly basic than the average of this class. Another interesting trachyte dyke occurs at Governor's Bay where it forms the sea-cliff for several chains, and owing to the action of the weather has developed a most remarkable spheroidal weathering† J. Von Haast, loc. cit, p. 335. Close alongside this is a glassy trachyte dyke of distinctly alkaline character and with a texture analogous to that of pitchstone.

Since nearly all the dykes except those at the head of the harbour, which are almost entirely trachytic, have a radial orientation, and therefore do not intersect, no satisfactory conclusion can be come to as to the relative age of the trachytic and basic varieties—if, indeed, they do belong to different periods of intrusion. The only intersections that I am aware of are those of trachytic dykes, and they do not furnish any sound grounds for differentiating between the earlier and later trachytes on the score of composition. (b.) Akaroa. Basic Flows, Trachyte Dykes, and Syenite—The general features of the Akaroa volcano are analogous to those of Lyttelton. It has been constructed in the same way of alternating flows of basic rock and fragmentary material, but fine- and coarse-grained basalts of normal habit form the great majority of the flows, basalts of andesitic habit being rare. Some of the flows, too, seem to be specially subject to weathering agents, so that there is a thicker covering of fertile soil than in the case of Lyttelton; but hard resistant basalts are very common, and seem to determine by their presence the marked shelves of flat ground which are evident on the sides of the harbour, and also the summits of the higher peaks, elevations such as Mount Bossu, Saddle Hill, French Peak, and Brasenose being composed of flows of lava showing little signs of decay, their slopes facing the harbour being precipitous and bold, while their external slopes are comparatively gentle and accordant with the angle of the lava-flows. Judging from the present spread of the base and the inclination of the lava-streams, the volume of the Akaroa cone far exceeded that of Lyttelton. Like this it had also a series of radiating dykes, which are generally trachytic in character, but neither their number nor the volume of the output appears to have approached those of Lyttelton. This may, however, be deceptive, since the maturity of erosion and depth of dissection are not so marked in the case of Akaroa, and therefore dykes have not been so extensively exposed. In the neighbourhood of Onawe they depart from the radiating arrangement and form on the shore platform a criss-cross pattern, with no regular orientation, except perhaps that more run in a N.E.-S.W. direction than in any other; but this conclusion may be erroneous. A most interesting occurrence in the Akaroa area is a coarse-grained plutonic rock—a hornblende syenite of peculiar type—which forms the rounded hill terminating the pear-shaped peninsula of Onawe. The contacts between this and the surrounding rocks are everywhere obscured by soil and debris even down to low-water mark, so that its field relations cannot be worked out satisfactorily. It apparently represents a volcanic rock which has consolidated at a deeper level and has been exposed owing to erosion, and is no doubt connected genetically with the trachyte dykes of the area, which form an intersecting network at the base of the peninsula referred to previously. But one at least of these dykes on the western side intersects the syenite as well as the later basic volcanics, so that the syenite may be a fragment of an older land-mass, and not directly connected with the later volcanic period. Just where this trachyte dyke intersects it there is a small exposure of an extremely basic facies of the rock containing numerous grains of magnetite, and the sand on the beach close to it is largely composed of this material, as well as hornblende and feldspar. The form of the end of the peninsula is one specially characteristic of granite rocks, and the material in position is deeply weathered.

The remaining portion of this peninsula is composed of basalt penetrated by trachyte dykes, with one small exposure on the eastern side of a well-stratified tuff containing pebbles of rhyolite. The field relations of this occurrence are obscure, but it certainly underlies basalt-flows, and it is further interesting as indicating a probable extension in a south-easterly direction of the Gebbie's Pass rhyolites under the covering of basic rocks. (c) Little River Valley To this epoch Haast* J Von Haast, loc cit, p. 343. and also Hutton† F. W. Hutton, Sketch of the Geology of New Zealand, Quart Journ. Geol Soc., vol. 41, 1885, p. 216. would also assign the formation of a caldera in the Little River Valley. Although an explosion may have been responsible for the formation of this valley in its initial stages, there is no actual evidence of its having been an independent centre. There is no appearance of a quaquaversal dip of the lavas issuing from a centre located somewhere within its basin, rather they are all disposed as if they flowed from the direction of Akaroa. The dyke system is that belonging to Akaroa and radiating from the centre of that volcano. There is, besides, no difference in the lithological nature of the lavas from the two localities. Further, its shape, with three long trailing spurs directed in some measure towards the middle of its basin, suggests most strongly the action of water. The fact that in certain parts of the western side cliffs face the valley and simulate to a minor degree the form of the caldera is entirely explained by streams cutting parallel with the strike of the beds, which must of necessity produce at times a form which resembles the steep interior faces of a caldera. The same remarks apply to Pigeon Bay, also regarded by Haast as a caldera, except that it is divided into two subordinate valleys, and not into three as in the case of Little River. I regard both as eroded by streams on the flanks of a volcanic cone. 4. Third Volcanic Phase. (a.) Mount Herbert Volcano. The third distinct phase in the history of the area was marked by out-pourings of lava from the neighbourhood of Mount Herbert, in all probability from a centre in Kaituna Valley or on its north-western boundary. This may perhaps be regarded as a third and subsequent vent, analogous to the two great caldeıas but on a smaller scale. By these eruptions the highest peak on the peninsula was constructed, the greatest development of the lavas being on Mount Herbert Peak and the ridge which extends in a westerly direction from it and forms the flat-topped mountain known as Mount Herbert. The tops of these elevations are formed of massive flows of lava, lying very level, individual flows being from 80 ft. to 100 ft. in thickness, and exhibiting in places columnar jointing on a large scale. The flows form successive tiers of ramparts round the head of the valley above the Head of the Bay and Charteris Bay, and again on the southern side fronting Kaituna Valley; but between Charteris Bay and Purau they form a long gentle slope with an average inclination of 10°, both the slope of the ground and of the lava-flows being in agreement. The termination of the flows has been cut back into a sea-cliff, about 50 ft. in height, for about a space of a mile on the southern shore of the harbour between Charteris Bay and Purau and there is an outlying fragment on the eastern side of Purau forming a low ridge behind the Ripa Island fort.

The flows belonging to this series rest almost entirely on the basic rocks of the previous series all round the heads of the valleys reaching down to Purau, Charteris Bay, and the Head of the Bay on the north side, but on the east side of Charteris Bay they rest on rhyolites; in Port Levy and Kaituna Valley they again rest on the older basic rocks. (See fig. 2.) Where the underlying rocks of the Lyttelton system are exposed they give the idea that even at that period the cavity resembled to some extent that at present existing and that it was of the caldera form, and this suggests that a long period of time elapsed between the two basic series of flows This point is also emphasized by the juvenile character of the drainage which has established itself on the long gentle slope running down from the summit of Mount Herbert Peak The period of these eruptions is therefore in all probability late Tertiary, and may agree with those which took place at Timaru, which are in all probability of Phocene age, seeing that the lavas he there on an eroded surface of Upper Miocene rocks and that underneath” the flows fossil bones of Dinornis have been obtained. The lava-flows from Mount Herbert are exclusively basalts, most of which are extremely fine-grained in texture, but a few are coarse in character No system of dykes similar to that associated with Lyttelton or Akaroa occurs in connection with this outburst from Mount Herbert. (b.) Mounts Fitzgerald and Sinclair. To this stage in the volcanic history the two authorities cited above assign outbursts at Mount Fitzgerald and Mount Sinclair, which lie on the ridge connecting the two great cones, and, indeed, form its highest summits. The lavas of these two peaks are exactly the same as those erupted from Akaroa, and there appears to me to be no justification for calling them separate volcanoes Their lava-streams show an inclination which leads one to think that they flowed from Akaroa; and, though they appear to he beyond the limits of the crater-ring, they are no farther from the centre of the caldera than similar heights on the eastern side of the harbour about which there has never been any doubt They apparently he far out, but this effect is due to the fact that at the head of Pigeon Bay the external stream erosion has cut far into the edge of the crater-wall—in fact, a portion of the internal drainage of Pigeon Bay Peak and its neighbourhood, which should go towards Akaroa Harbour, flows into Pigeon Bay, whose streams are thus capturing the heads of those on the other side of the ridge. Thus it is that the crest-line of the hills round the harbour takes a remarkable bend in at this point, and the headwaters of Pigeon Bay on its western side have cut in circumferentially along the strike and apparently separated the elevations from the walls of the crater There appears to me to be no reason why they should be regarded as separate centres of eruption, and after careful examination I must withdraw my former endorsement of Haast's and Hutton's opinion* R. Speight, On a Soda-amphibole Trachyte from Cass's Peak, Banks Peninsula, Trans. N.Z. Inst., vol. 40, 1908, p 176 5 Fourth Volcanic Phase. Quail Island. The final stage in the volcanic history of the peninsula was the formation of Quail Island (Plate XXV), which represents in all probability a secondary cone within the crater-ring of Lyttelton. The basement of the

island consists of rhyolitic material described previously, but the lavas capping this are entirely of basalt. The lowest beds exposed on the north-west corner of the island are of basalt, but at of the base of the cliffs facing the harbour is a bed of irregular angular material, 80 ft thick where it is best developed, which may form the lowest member of the series in this locality The fragments are of all sizes up to 1 ft in diameter. chiefly of basalt, rhyolite, and trachyte, but with occasional pieces of greywacke, no doubt torn by explosive action from the underlying substratum in close proximity to the vent. Where exposed on the shore platform at the base of the cliffs the material is especially coarse, but in the cliff itself appear layers of fine-grained ash with well-developed stratification Since there is an absence of traces of sea action, as well as no evidence of marine fossils, it is reasonable to assume that the stratification is due to sorting by the wind and not to water, a mode of formation of wide-spread occurrence where volcanic cones have been constructed on a land surface. Fig 4—Section, north side of Quail Island Length, 40 chains 1 Rhyolite with dykes of trachyte 2 Older basalt, amygdaloidal in parts. 3 Volcanic breccia and tuff 4 Later basalt, columnar and piismatic 5 Loess At the edge of the shore platform the layer of agglomerate is overlain by solid flows dipping north, an arrangement suggesting that the sea has cut into the heart of the cone and exposed the beds in close proximity to the vent This is all the more probable since at the extreme northeasterly point of the island the lava-flows dip south—i e, in the contrary direction Another fragmentaıv layer is exposed on the shore at the base of the low cliff just north of the outer wharf. the fragments in this case consisting almost entirely of basalt, and as this lies almost on top of solid rhyolite in position, the junction being in close proximity but obscured, this may represent the oldest bed of the Quail Island series The massive layer referred to previously probably outcrops on this face of the island as well, but the section is difficult to make out, as would be expected where the extruded material near the vent consists of irregular layers of fragmentary material, as well as small and irregular lava-flows Over the thick layer of fragments is a horizontal sheet of rudely prismatic basalt, 50 ft thick, which is well exposed in the cliff facing Lyttelton This abuts against the underlying basalt on the east, and against the rhyolite with its trachyte dykes on the west, its continuity is here broken, and it is impossible to say with certainty whether the flows exposed on the cliffs on the north-western point of the island belong to a later flow or not. The upper surface was deeply eroded before the succeeding layer of fragmentary

matter was deposited. This layer is similar in character to the layer underlying the prismatic jointed flow, and it may be connected with the layer interposed between the two basaltic flows to the north-west of the island. The last basaltic outburst has formed another markedly horizontal sheet, which in all probability caps the greater part of the top of the island, and extends down to sea-level on its western side. It shows almost perfect columnar structure in its lower parts, especially where exposed on the cliffs, but the upper layer is rudely prismatic, the line of junction between the two parts being quite distinct, although there is apparently no actual break in the flow The columns are vertical, and in places the upper part of the flow overhangs and large pieces of rock have broken away and have fallen at the base of the cliff near sea-level. On the southern shore of the island boulders from this flow form a considerable portion of the loose material of the shore-line, though the flow does not appear to have reached the sea-level. Professor J. P. Iddings and Dr. P. Marshall have drawn my attention to the fact that some of these boulders are probably of alkaline type. Haast considered that a hollow at the top of the island might have formed the crater;* J. Von Haast, loc. cit, p. 348. but both the eastern and western extremities of the two uppermost flows as exposed in the cliff abut against older volcanic material, either solid or fragmentary, and the inclination of the flows on the shore platform is generally to the north, so that the two distinct horizontal sheets occupy the actual floor of the crater. It is extremely probable that the northern side of the cone was breached either by the outflow of lava or by inroads of the sea, or perhaps by a combination of both processes, the reef which lies between the island and Lyttelton being perhaps the remains of a flow which extended in that direction. There is at present no evidence as to the date of these eruptions, but they probably date from the latter part of the Tertiary era. 6. Post-Volcanic History. (a.) Evidence of Greater Elevation. The subsequent history relates to the destruction of the volcano by stream dissection, and to the depression of the land which allowed the sea to invade the floor of the caldera. The former action was actively continued when the land was higher, on lines which were in all probability determined when the volcano was in active operation. Besides the evidence for former greater height deduced from the drowned valleys and the sea-invaded calderas, there is proof available from the records of the bores put down in the artesian area which fringes the volcanic mass immediately on the west.† R. Speight, A Preliminary Account of the Geological Features of the Christchurch Artesian Area, Trans. N.Z. Inst., vol. 43, 1911, p. 420. The land in this region was at least 700 ft. higher when these beds were laid down, as is evidenced by the occurrence of layers of peat at various levels down to 700 ft. beneath sea-level. These are found as far down as boring has been continued, and there is no reason why they should not be found lower still. An elevation of the land by even this amount would drain the caldera-floors, convert the drowned valleys both inside and outside the calderas into dry land,

and would extend the fringe of land a considerable distance beyond the present outer margin of the peninsula. Another proof of increased height may be obtained from the character of the vegetation of the mountain-tops. If they were once higher we should expect to find survivals of an alpine or at all events a subalpine flora. This is exactly what we do get. R. M. Laing* R. M. Laing, On a Subalpine Element in the Flora of Banks Peninsula, Trans. N.Z Inst., vol. 46, 1914, p. 57. has shown that certain forms occurring on the summits are closely related to those found on the Alps fifty miles away to the west This author, however, says in conclusion, “Whether this florula is to be regarded as a collection of waifs and strays or the remnant of a more widespread flora of glacial times I shall not endeavour to discuss here” Dr. Cockayne is of the opinion that the latter explanation is probably the more correct one of the two. There is no doubt, however, of the greater height, and this corresponded with the time of the severe glaciation which this country experienced in Pleistocene times and later. It is generally agreed, however, that increased height of the land, and not a marked refrigeration of the climate, was the cause of the extension of the ıce; but there is no evidence that the peninsula had a covering of ıce or even nourished glaciers in its higher valleys. Their form gives no suggestion that this was the case, and there are no accumulations of angular boulders which might be called morainic in character The only evidence of the neighbourhood of glaciers is afforded by the covering of loess which is found widely distributed at all levels and in various positions in the area It is found on the extremities of the ridges dividing the bays on the north and east, on the southern slopes which reach down towards Lake Ellesmere, and on the western slopes above the Canter-bury Plains, and it is found very thick in places within the calderas It is found at all levels up to 2,000 ft, but it is specially thick on the ridges, and, above all, on those within the Lyttelton caldera, which he at the back of Quail Island, where the capping is in places 25 ft thick This is well seen on the ridge between Governor's Bay and the Head of the Bay, in a road cutting. Captain Hutton† F. W. Hutton, Note on the Silt Deposit at Lyttelton, Trans. N.Z. Inst., vol. 15, 1883, p. 411. considered this a marine deposit, and if so it would imply a depression of the land at least 2,000 ft below its present level The general absence of marine fossils, the presence of remains of Dinornis and other land-birds, as well as its peculiar distribution, is against a marine origin being assigned to it. It is in all probability a rock-flour formed by glacier erosion, transported out on to the plains by glacial torrents, and distributed by winds during a time of drier and more steppe-like climate. On the slopes of the hills it would receive an admixture of clay derived from the weathering of the silicates contained in the volcanic rocks, so that it should not be regarded as a typical loess such as exists in Europe and America, but as a “pseudo-loess,” as it has been called by Heim ‡ A Heim, Neujahrsblatt, 107. Zurich, 1905, p. 38. Although the later major movements of the land have been in a downward direction, there is evidence of a recent rise of from 2 ft to 3 ft at least In some of the sheltered bays there are beaches with Recent marine shells some distance above high-water mark. The best illustration of this is found on the south side of Quail Island, where a well-

developed beach with shells occurs at the foot of a wave-cut cliff, now over a chain from the sea at its maximum distance. This evidence is supported by that obtained from the neighbourhood of Sumner, where the marine cliffs are now removed a considerable distance from the present shore-line. A part of this effect may be attributed to the progression of the shore by the deposit of sand, which has developed in places a dune system; but this is hardly sufficient of itself to account for all the phenomena. With this movement is connected in all probability the system of low terraces which fringe the Avon and Heathcote Rivers, immediately north of the peninsula. These streams have reached a second base-level below an original one, which was disturbed by this slight elevation of the land. Pronounced elevatory movements are in evidence on the coast-line from thirty miles farther north, near Amberley, Motunau, Amuri Bluff, and Kaikoura, and it may be that the peninsula is on the outskirts of the area affected, since the upward movement in its neighbourhood is very slight. (b.) Results of Erosion and the Formation of Calderas. A consideration of the form of the harbours strongly supports the contention that stream erosion combined with wave action in cutting back the cliffs is responsible for their principal landscape features. The importance of stream erosion is very evident when we consider the depth to which valleys have been eroded on the outer slopes of the volcano. In numerous instances they have cut back till a very narrow ridge separates the inner and the outer slopes. This is well exemplified at Heathcote Valley in the dividing ridge above the tunnel, and at the head of the valley leading to Dyer's Pass; and in two places—viz., at the entrance and at Gebbie's Pass—the wall has been completely broken down. It has been mentioned previously that the heads of the valleys generally lead to the lowest parts of the crater-ring, and that this suggests that a former stream system has had the upper parts of the valley removed either by explosion or by collapse It is possible, however, that it merely means the approach of the heads of two valleys from opposite sides of a ridge, and that the combined attack on the crest has resulted in its being lowered at that spot; and, further, the form of the valleys is a modern development, whereas the formation of the caldera by explosion belongs to some date in Tertiary geological time, and therefore such a feature is not likely to exhibit such a perfect arrangement now. The extent of denudation and its marked effect on the form of the mountains, reducing their height, is emphasized by the facts that the radiating ridges are frequently higher some distance away from the crater-walls, and that these higher portions are frequently capped by rock of more resistant nature, and that when its slope is prolonged crater-wards it reaches a much higher level than even the highest parts of the crater-ring in the vicinity. This is markedly true of the Akaroa volcano, for there we have eminences reaching well over 2,000 ft. on the spurs which divide adjacent radial valleys; Instances of this are Pigeon Bay Peak and View Hill, on the northern flanks, and there are as well numerous unnamed elevations on the ridges to the south. While a volcano is in process of construction streams will establish themselves on its outer slopes, with radial orientation, their heads cutting farther and farther back towards the crater, so that the valleys will be more bunched together towards the centre of the cone than on its margins, and a struggle for existence will occur among them, those reaching farther back

robbing the less extended valleys of the supply of water at their heads and cutting in behind them. The result of this will be that the spurs which stretch down from the summit will broaden out as they reach lower levels, and the valleys established on their peripheral portions will in many cases be short. This feature is well displayed in the spurs which reach out in a northerly and westerly direction into the plains, where wave action has exerted but little effect; but on the outer rim of the volcano, where sea action has been intense, the broader terminal portions have been removed, and the effect is not so pronounced, since the lower portions of the short streams have been cut away, and the water either enters the sea by short steep ravines or by falls. As these valleys run in the direction of the dip of the beds and at right angles to the strike, they exhibit the special features of transverse valleys —that is, they are trench-like in character, with steep sides; there is always a tendency to cut subsequent valleys at right angles, especially in their higher portions, but owing to the variation in the direction of dip and the discontinuity of particular layers these do not exhibit the regularity usually displayed by valleys cut in sedimentary strata. This may explain the amphitheatre-like form which characterizes the heads of many of the valleys, and where this does not occur it explains the branching heads of the valleys. This peculiarity has been noted by J. D. Dana with regard to the volcanic island of Tahiti,* J. D. Dana, Characteristics of Volcanoes, 1890, p. 376. where he states that the valleys of that dissected island are frequently narrow in their lower portions but open out into distinct amphitheatres at their heads. This gives an appearance resembling a crater or caldera, and it is this which probably influenced Haast so strongly in suggesting that the Little River and Pigeon Bay Valleys were really calderas, and the sites of independent volcanic vents. Owing to the dominant valleys enlarging the upper part of their basins into an amphitheatre-like form, the heads of the adjacent smaller valleys are robbed of their proper supply of water—that is, the amount rightly belonging to the whole of the sector of the volcano which they should drain, and not the outer portion of that sector only Although the smaller valleys may initially have been truly radial in orientation, in process of time they depart slightly from that direction and their heads point towards the high bounding ridge of the adjacent large valley. Further, owing to the more pronounced erosion in the dominant valley, the walls tend to encroach on the higher reaches of the smaller valley and to capture some portion of its higher reaches, which were primarily eroded by the water which discharged by means of the smaller valley. It thus exhibits what may be called a recessive character As a result the smaller valleys are often headed by distinct saddles. This is well seen to the east of the Little River Valley. In the former direction, towards Peraki, all the valleys reaching down to the small indentations of the coast rise not from the crater-ring of Akaroa, but from the high eastern wall of the Little River Valley; while to the west, between this valley and the large Kaituna Valley, we have first Price's Valley and then Birdling's Valley reaching back successively to the valley-wall of Little River, the second valley as it touches the crest farther up being far the longer of the two (see Plate XXIV). In both these cases the subordinate valleys are headed by depressions in the dominant ridge, not deeply incised, it is true, but the effect is still visible, and in

process of time, as the dissection becomes more advanced, it will result in the division of the ridge into a series of isolated blocks and thus aid in the destruction of the cone. The more energetic attack on the heart of the volcano owing to the bunching-together of the heads of the streams accounts for the form of the remnants of dissected cones, which consist usually of a central resistant plug forming the neck, and then some distance away isolated remnants of the lava-flows in their proper positions. While the streams flowing on the outside surface of the cone are usually consequent in character, with a tendency to subsequent tributaries, those on the inside of the crater will have the character of obsequent streams with steep grade, obstructed by falls and rapids when they flow over ledges of harder rock. In a normal crater the erosion of obsequent streams will not be important while volcanic action is going on; but if the crater is enlarged by explosion or by breaching this erosion may achieve important results, and specially so if there is any concentration of the drainage of the internal slopes in the early stages. This will naturally arise in the case of a cone not subjected to explosion if it has been breached by a lava-flow from the summit. Thus a well-organized direction of drainage will be promoted from the beginning, and it is very probable that a number of calderas may owe their initial development to this circumstance. The direction of drainage from the enlarged crater will generally follow the line of one of the consequent streams of the outer slope if it be not breached, since a low place in the wall of the crater-ring will usually head a valley. This is well exemplified at the present time by Ruapehu, for the low part of the crater-ring faces the eastern side, and the drainage from the basin on the top of the mountain will naturally concentrate towards that point. In this case the depression in the height of the wall may be due to the accidents of explosion, or to the attack of the wall from the outer slope, or to both causes. When this direction is once firmly established erosion will tend to enlarge the interior of the crater by sapping back the walls, the form of the crater being to a large extent preserved, the steep scarp slopes of the interior being maintained in just the same way as the scarps of tilted sedimentary strata are preserved. Thus the form of the crater will be kept till it reaches the dimensions of a normal caldera, and the barranco will be the valley due to the erosion of the concentrated drainage. Should any modification of the structure occur from any cause, or should denudation expose material of different character from that of a normal cone or with different stratigraphical arrangement, then a departure from this form will ensue, an instance being seen at the head of Lyttelton Harbour, where the sedimentary strata and rhyolite beds of different arrangement from the remainder of the volcano have caused modifications in the form of the resulting cavity. Should the crater-wall be broken in more places than one by the explosion or by normal volcanic action, or should several streams cut back their heads till they have invaded the crater, then similar features will be produced along other lines and more than one barranco may drain the caldera. If we take the conditions obtaining now in the case of Lyttelton, we observe that all along the northern side of the harbour the valleys have the features of obsequent streams, and as they cut back their heads and diminish their average grades they will retain this character, but with diminishing characterization, as the grade becomes less pronounced. The spurs in this portion of the harbour are usually terminated in steep cliffs

owing to the erosive action of the sea: this, too, tends to maintain the appearance of the interior walls of the crater as having been caused by explosion. The steep-graded valleys carry their grades down below waterlevel till they merge into the floor of the harbour, emphasizing the unity of origin of the subaerial and submarine portions. In the western part of the basin the interior slopes are occupied by long tongue-like peninsulas, nearly two miles in length from the point where they are joined on to the walls of the crater-ring. The valley-floors between them are very flat and grade insensibly in the floor of the harbour, and they show the form they should possess were the harbour a river-valley and its head enlarged by water action with just that amount of modification effected by the infilling of sediment into the submerged portion. The sources of the streams which drain inwards are distributed along the inner crest of the crater-walls, but the streams join the main valley at points in close proximity to each other. Where the valleys are cut in the lava-flows dipping outwards the valleys have just the same features as those on the northern side of the harbour. This locality also illustrates the breaking-down of the wall of the crater by the combined action of streams attacking it from the inside and the outside slopes. The streams responsible for the formation of Gebbie's and McQueen's Valleys, as well as those occupying the valley within the harbour area where Teddington lies, have completely removed the andesites, deeply dissected the rhyolites, and exposed the underlying greywackes over a considerable extent of country. The forms of the two external valleys are so strongly reminiscent of the lower reaches of Lyttelton Harbour that they are probably due to a common cause. On following the harbour round beyond this gap past a remnant of the old crater-ring the interior slopes are modified from the true caldera form by the gentle slopes reaching up to Mount Herbert accordant with the flow of the lava-streams from that more recent centre of eruption, but even here, where small exposures of the underlying beds can be seen, they suggest a land surface analogous to that on the northern side of the harbour. To the east of these flows, and eroded along their margin, lies another drowned valley, called Purau Bay; but from this on to the southern head the characters are those which we should expect from the erosion of a stream flowing across the strike of gently inclined beds dipping uniformly in one direction. This portion of the harbour thus has a trench-like cross-section, a feature made more pronounced by the attacks of the sea on lava-flows where the jointing is almost vertical. The same feature is seen in the lower portions of Gebbie's and McQueen's Valleys, except that the influence of the sea in forming a wave-cut cliff is not so prominent. A feature to be considered in arriving at a conclusion as to the origin of the cavity is that the valley-heads almost entirely occupy lower parts of the crater-ring and do not lead up to the highest peaks. Although this suggests that the valleys have had a portion of their upper courses beheaded by explosion or by collapse of the cone, such as happened in the case of Mount Mazama,* J S. Dillon and H B. Patton, Geology and Petrography of the Crater Lake, National Park, U.S Geol. Surv. Prof. Paper No. 3 1902. in the United States, it is quite possible to explain the phenomenon as an effect of normal stream erosion on a cone which has not suffered such a catastrophe. The approximation of valley-heads on the

inside and outside of a cone would probably arise as erosion proceeded without the intervention of explosion or collapse of the interior. The general features of the Lyttelton crater are reproduced in Akaroa Harbour. The steep interior slopes towards the entrance, the peninsulas at the head with their drowned valleys lying between them, the gradually deepening of the harbour and its tributary bays towards the entrance, the accordance of the grade of the floors of the valleys with that of the harbour, the wave-cut cliffs gradually getting higher as the outlet is approached, and the perpendicular walls which guard the entrance agree exactly in both cases. Akaroa has, however, a more perfect crater-ring: it is only broken where the sea enters, and its regularity is not destroyed by subsequent eruptions breaking out on the edge of the crater, as has happened in the case of Lyttelton. If we make due allowance for the special features which exist in the case of Lyttelton and are absent in Akaroa, we cannot but conclude that the same causes have contributed in each case to establish the caldera-like form, and the dominating one is the effect of stream action on a crater which has been enlarged by a paroxysmal explosion of moderate intensity or by breaching, but not one which has produced the great cavities which now form the harbours as they are. The stream-erosion theory of the formation of calderas has been advanced by Gagel in connection with the Palmas Caldera, and I believe that he did not regard explosion as a contributing agency. I have not been able to obtain his paper, but have only the reference to it in Professor R. A. Daly's Igneous Rocks and their Origins. Haast and Hutton, on the other hand, regarded the cavity as having been produced by explosion alone. If, however, explosion were entirely responsible for the formation of the cavity there should be accumulations of fragmentary matter round the vent approximately equal in volume to that of the cavity. The amount of material thrown out to make a space five miles across and at least 2,000 ft deep should certainly have left traces in the surrounding country, even after allowing for denudation. There is no such accumulation under the subsequent lava-flows from Mount Herbert, where it would be protected from denuding agents. It is not intended to deny that in the early stages moderate explosions did exert some effect. When the small size of the necks of old volcanoes is considered (see Sir A. Geikie's Ancient Volcanoes of Britain), the largest recorded in Fife being a mile in diameter, and the average far below this, we must conclude that unless the original crater is enlarged in some way water erosion by itself would in all probability be unable to produce a cavity of the shape usually arrived at. Subsidence of the floor due to faulting has been suggested by Dutton* C E. Dutton, Hawaiian Volcanoes, 4th Ann. Rep. U.S. Geol. Surv., 1884, p. 105. as a prime cause of the formation of calderas, especially those of the Hawaııan group. When, however, we examine the line of junction of the Lyttelton lavas with the underlying greywackes at the head of the harbour and the Gebbie's Pass ridge there is no sign of dislocation following the base of the crater-ring; the long tongues of rhyolite stretching down therefrom into the upper part of the harbour and almost reaching the actual centre show no sign of a break, such as should occur were the line of fault to follow the bounding walls of the depression. If such a line did occur it should certainly be found cutting across the direction of flow of the rhyolite lava-streams in close proximity to their junction

with the greywacke. There is also no sign of dislocation in the discordance of the grades of the streams which would cross such a line: perhaps this point is of no great importance. I am therefore of the opinion that the determining cause of the formation of such great hollows as Lyttelton and Akaroa Harbours is prolonged water erosion and not paroxysmal explosions, although an explosion of moderate intensity or the breaching of the cone by a lava-flow may have initially determined the directions of the streams which established themselves on the surface of the volcano. The landscape features of these two harbours are reproduced almost exactly in the case of Carnley Harbour, in the Auckland Islands, the comparison with Akaroa being especially striking,* R Speight, Physiography and Geology of the Auckland, Bounty, and Antipodes Islands, Subantarctic Islands of New Zealand, 1909, p. 708. except that Carnley is on a much larger scale, and instead of having one peninsula it has three large peninsulas forming long trailing spurs directed towards its interior. The agreement as to nature of the lavas, arrangement of the flows, height of the encircling hills, form of the internal and external slopes, and the shape of the great cavity suggest a similar origin and a similar geological history. No doubt the hollow was formed primarily by a moderate explosion or by breaching of the cone, and was enlarged subsequently by stream erosion and a glaciation of moderate intensity—the latter absent in the case of Akaroa —at a time when the land was higher. The enlargement of the cavity by erosive agents and the removal of the periphery of the mountain by the action of the sea have resulted in the partial break of the walls at one place on the western side and in their complete breakdown in another, and through the gap fierce tides and heavy waves pour into the western portion of the basin. The proper entrance is placed at the east, being exactly similar in form and cross-section to that of Akaroa and Lyttelton, with a distinctly fiord-like character, a result perhaps due partly to glacier action, but certainly attributable in great part to the action of water as described previously.