Although this sequence of events may be reasonable, I know of no direct evidence of the former presence of a lake in the Trelissick Basin. There are no stratified gravels or silts that I am aware of, and the floor is not level. This may not be an insuperable objection to the hypothesis, as the terraces and the floor itself may have been

Sketch Map Showing Relative Postion of Trelissick Basin and S.Esk Valley

warped as the result of earth movements, or the basin may have been tilted as suggested by Hutton, although he puts his tilting much earlier. I can see no difficulty, however, in the basin having been a waste-filled plain, like the Hanmer, Maruia, or Upper Buller Plains, or the Mackenzie Country, while glaciers filled the neighbouring

valleys. No doubt each of the streams issuing from the mountain tract to the west of the basin had its fan, and the coalescence of these formed the floor of the area, and thus its approximately—for it is only approximate—level surface can be explained, the departure from absolute horizontality being due to the varying conditions of the streams issuing from the Craigieburn Range or from the front of the ice to the north or perhaps to the south. These streams would differ in size and in the amount of waste material they carried, and thus their grade would vary, causing any unevenness in the main floor existing now.

The hypothesis of the former existence of a lake must be considered in the light of the fact that a lake did occupy the floor of the Waimakariri Valley in the reach downstream from the Cass to the mouth of the Esk. (See Map B.) Its former presence is evidenced by glacial silts and by level terraces marking its old shore line. Glacial silts, some of them varved, occur in the lower reaches of the Poulter, but I have seen none in the Esk. The traces of this lake indicate that it was at a higher level than the present floor of Slovens Creek, but this would not account for the existence of a lake in the Castle Hill basin. This lake, too, was ponded back behind a rock barrier just below the mouth of the Esk, now deeply incised by the main stream of the Waimakariri. The hollow above the barrier may have been structural in origin, but it was certainly modified by glacier action. If due to movements of the crust they must either have been pre-glacial or contemporaneous with the occupation of the upper part of the valley by the glacier, since glacial silts occur in it. It is not maintained that the present hollow was in existence then, but a hollow did exist for certain. The movement responsible for the formation of the barrier must have commenced, if, indeed, it had not concluded, before the end of the glacier advance.

There is, however, in the Esk Valley a great deposit of stratified gravels, best developed on what is called the Tableland, a flat area 3000 feet above sea-level, and lying three to four miles above the confluence of the Waimakariri and Esk (Plate 55 and map B). This was once covered with forest, but a fire has destroyed most of it, though a bush-covered extension of the plateau continues for two miles further up the Esk in the form of a high terrace. The beds of which the Tableland is composed lie flat, that is, parallel to its upper surface; they are mostly of gravel, but this is interstratified with beds of sandy clay and sand. They rest on greywacke, and the river has now cut its course through the gravels and incised the solid basement. It might be noted here that the destruction of bush was followed by marked erosion of the gravels on the edge of the plateau facing the Esk, and “badland” topography has become typically developed. (See Cotton's Geo-morphology of New Zealand, p. 191.)

These gravels occur on both sides of the stream, and extend north-east beyond the Tableland at a lower level across the lower courses of the Nigger and Ant Streams, past Anna Creek and Grant (Cattle) Creek nearly as far as Anderson's Creek. They reach north-

west nearly a mile up the Ant and Anna, in the neighbourhood of which they form an extensive flat terrace lying at a height of from 2700 to 2800 feet. The drop to this terrace from the top of the Tableland is sudden, and there are no intermediate shelves. This is incised in turn, and there are numerous terrace remnants at various levels till the bed of the Esk is reached. The suddenness of the drop from the Tableland and the extent of the terrace just referred to are remarkable (Plate 55).

The gravel deposit is most extensive, and therefore indicates some special conditions favouring its accumulation. In general the bedding is flat, but along the northern boundary of the terrace the beds had been tilted prior to its formation. The strike is here W.N.W.-E.S.E. approx., and the dip southerly at angles of from 25° to 30°. The lowest beds exposed in the Ant consist of whitish and greenish-grey sands, sandy clays, and irregular gravels interstratified with sands. In the Anna, which lies parallel with the Ant and a short distance to the east, green-grey sands and chocolate coloured clays occur interstratified with subangular gravels, the largest pebbles of which reach 8 inches in diameter. These beds extend north-east to the main Esk between Cattle and Anderson's Creek on the slopes of the Puketeraki Range, and have the same strike and dip throughout. The inclination is due to subsequent deformation and not to foresetting. There are also heavy deposits of gravel in the valley of the Poulter, above and below the junction of the Cox.

There is no precise evidence as to the age of the Esk gravels, but they may be tentatively assigned on lithological grounds to the Kowai Series of North Canterbury. Marine Tertiary beds do occur in the locality, but the relations of the gravels to them is obscure. Large fragments of rock containing fossils were found during a recent trip in a tributary of the Ant on the slopes of the Candlesticks at a height of approx. 3500 feet, and Dr Marwick has assigned these to the Mid-Tertiary, and almost certainly to the Awamoan. On the right bank of the Esk, about a mile below the Anna Hut, there is an exposure consisting of the following sequence:—

These beds strike N.E.-S.W. and dip N.W. at high angles. They are affected by a fault which runs in the same direction. The exposure is obscure owing to the height of the river and the cover of terrace gravels, but at the downstream side of the occurrence, which, after all, is only about 4 chains in length, there are irregular cemented gravels resting on greywacke in close proximity, but their

precise relationship is obscure as well. I am told by Mr R. Turnbull, the lessee of the Mount White Station, that fossils have been found in close proximity to these beds, though I saw none on my various visits. Unless they are disturbed locally owing to faulting, their pronounced dip and the difference in strike from that of the gravels in the vicinity, though it must be admitted not the immediate vicinity, indicate that they underlie the gravels unconformably.

The flatness and even stratification of the gravels indicate that they were laid down in comparatively still water, and it is thus possible that a lake occupied the lower part of the course of the Esk. There is at present no rock barrier indicated that would pond back this water from the Waimakariri side to the height of the top of the Tableland, but there is a certainty that the ice at its maximum extension could furnish such a barrier, for there are clear evidences of ice at a height exceeding 3500 feet in the Craigieburn region of the Waimakariri basin, and this would be quite sufficient to furnish the required barrier if the ice extended on to the slopes of the Puketeraki Range below the junction of the Esk. There is little doubt that this was the case, and if the lower reach of the Esk were free from ice at the same time, there would be opportunity for the deposition of these gravels under the required conditions. A difficulty in accepting this explanation in its entirety is the determination of the precise date of these gravels. In order for it to be satisfactory, they must date from the time when the central Waimakariri was occupied by ice and the lower reach of the Esk free from it and occupied by water. The synchronism cannot be demonstrated. There is therefore no definite support from the valley of the Esk of the contention that the Trelissick Basin was once a lake, but the evidence is in support of the hypothesis that the barrier of greywacke to the east of the basin was in existence in pre-glacial times or at any rate before the ice had commenced its last retreat, and that it has been cut through subsequently.

It is noteworthy that the main coalesced drainage from the Waimakariri intermont follows along the base of two ranges, viz., the Puketeraki Range and the Mount Torlesse Range, and there is comparatively little drainage in the middle of the area, only two streams being worth mentioning, viz., Winding Creek and Slovens Creek, both of minor importance. This may reasonably be explained by supposing the main part of the basin to have been filled with ice, and the streams therefrom discharging along the margin of the main flow, just as the Murchison River discharges along the margin of the Tasman Glacier. The impinging of the ice against the slopes of Mount Torlesse would provide a temporary barrier against the cutting down of the river channels in the Trelissick area, and thus help in the formation of terraces. All the same, the dominating agencies in this area connected with the formation of terraces appear to be the retardation in the cutting down of weak beds by solid rock barriers, such as the limestone and greywacke to the east of the Trelissick Basin, and the latter downstream from the mouth of the Esk, as well as the warping of the crust on an axis