Part I.—Physiography.

position of the highest peaks with regard to the mountain axis of the South Island is to be observed further south, where Mount d'Archiac (9,279 ft.) in the Two Thumb Range, Malte Brun in the Maltebrun Range, and Mount Cook itself lie to the east of the main watershed, and excel in elevation the neighbouring peaks on it. The crest of the range also lies far to the east of the structural or tectonic axis of the Island, which no doubt follows up through the schistose belt between the sea and the present main range about twenty miles from the west coast. This suggests that the present configuration has been the result of excessive erosion acting for a long period of time on the western wing of the geanticline of which the range has been built, far more profoundly than it has acted on the eastern wing.

Fig. 1.—Map of Mount Arrowsmith District.

(b.) Relation to Rainfall and Conditions of Erosion.

The mountains in this locality he right across the direction of the prevailing westerly winds of this latitude, and the bulk of the moisture is intercepted on their western slopes. The conditions are exactly the same as these further north at Otira, which, according to the records of the meteorological station recently established, has a yearly rainfall of a little over 200 in., while at Bealey, about twelve miles away in a straight line, it is only 100 in. Although no statistics are available for the Rakaia region, the effect is clearly visible in the character of the plant covering, which changes from the rain forest of Westland to the markedly xerophytic tussock steppe of the region to the east of Mount Arrowsmith. Intermediately there is the totara forest of the Upper Rakaia, which has followed the rain just across the main divide from the montane rain forest containing totara on the higher hill country of Westland.

These different conditions of rainfall have probably lasted for a very long time, and their effect is evidenced by the lower elevations to the west, where the subaerial denudation has been excessive, owing to the heavy rains causing frequent floods and the great amount of snow forming large and powerful glaciers. There is evidence from other parts of the Alps that the dominant western erosion has resulted in the capture by the western streams of the upper tributaries of some of the rivers of the eastern watershed. It is only in some such way that the formation of Arthur's Pass can be explained. This action is likely to continue when the marked difference in the height of the floors of the eastern and western valleys is considered—for example, the floor of the Bealey Valley near the tunnelentrance is at an altitude of 2,435 ft., while the western end, at Otira, five miles and a quarter distant, is only 1,583 ft.—that is, it is 852 ft. lower. The same is generally true for the other valleys of that part of the Southern Alps, a feature well brought out by Mr. Edward Dobson's original surveys for a road across the range. This remarkable physical peculiarity can be most easily explained by the greater efficiency of eroding agents on the western side of the range. The heads of the valleys have been sapped back by glaciers, and the valleys have been deepened by ice and water action so that they have been able to encroach on their eastern neighbours; and the marked overlapping of adjacent streams on either side of the crest of the range intensifies this effect when capture of even one small tributary has taken place. The more rapid erosion on the west will lower the range progressively from that direction, so that the eastern region will become of relatively higher relief, and in future geological time Mount Hutt and Mount Torlesse. on the eastern border of the mountainous country, will, if similar metreorological conditions continue, become its highest elevations.

(c.) Present Form of Mountain Region—a Dissected Peneplain.

The rocks of which the area is composed consist of greywackes, slates, and mudstones of Lower Mesozoic age, which have been folded subsequently by mountain-building movements into folds whose general axes run in a N.N.E.-S.S.W. direction. Local variations appear to be frequent, so that at times the direction is almost E.-W., and again become N.W.-S.E. This variation is apparently due to change in the direction of the thrusts by which the area was folded. The date of the folding is probably Upper Jurassic, but it may have been Upper Cretaceous. A feature of the mountainous region which has thus been produced is the approximately uniform height of the great majority of peaks. A very large number of these are between 6,800 ft. and 8,200 ft. in height, with very few above or below these limits except in the Mount Cook region. This suggests that the whole area has been reduced to a level platform either by marine denudation or, more probably, has been base-levelled by a former stream system, and a few peaks, like Mount Arrowsmith, which dominate the rest are the residual elevations on the peneplain. The higher mountains which no doubt once existed further west, and have been removed by erosive agents, would represent the higher mountains on this peneplain. It is possible, however, that the present prominent elevations are the remains of an old divide which existed on it, and from which streams once flowed east and west.

In advancing this theory for the present form of the mountain region of Canterbury, I am quite aware that this is contrary to the generally

3. Drainage Systems.

(b.) The Rakaia Valley.

The largest river of the district is the Rakaia, which takes its origin in the Lyell and Ramsay Glaciers, on the eastern side of the main divide, and runs first easterly for twenty miles and then turns south-east. It receives on its north bank numerous streams which rise in small glaciers on the main range and run in a southerly direction in parallel valleys till they meet the Rakaia. Further down, beyond the limits of the area under consideration, it receives two important branches—the Mathias and the Wilberforce. On the south bank several fair-sized streams rise in glaciers on the slopes of Mount Murray, but the main tributary is the Lake Stream. This is formed from the overflow waters of Lake Heron—hence its name—but its chief supply comes from the Cameron River. This rises in the Cameron Glacier, on the south-east face of Mount Arrowsmith, runs in a south-easterly direction for about twelve miles, and forms an extensive fan on the north-west side of the lake. At this part of its course it changes its regular channel frequently, running into the lake at times, but at present it joins the Lake Stream about two miles below its outlet from the lake. This body of water receives several streams on its eastern side, the largest being the Swin. The Lake Stream runs between north and north-east and has a very slight fall for several miles, but then it becomes swifter and makes a descent of 200 ft. before joining the Rakaia.

The Rakaia River and most of its tributaries are overcharged with detritus, and have filled up to a fairly uniform surface the floor of the glacial trough through which they run (Plate III, fig. 1). The width of the flood-plane is about two miles, and on this the river forms many diverging and anastomozing branches, in the way so characteristic of streams laden with waste. In winter it is frequently dry, owing to the freezing of the water as its sources; but in spring and summer it becomes a large river, and often impossible to cross on horseback, although the splitting-up into different streams renders it somewhat more easy to negotiate. Near the head the valley narrows somewhat, and there the river runs in a solid body of water capable of rolling down stones a ton in weight. In this part of its course no terraces are formed except those of temporary nature, but these last long enough at times to allow of their being covered with grass and scrub; still, they are liable to rapid destruction if the river in the course of its wanderings impinges for long against their unconsolidated edges (Plate III, fig. 1).

(c.) The Lake Heron Valley: its Features and Origin.
(See map, p. 318.)

The valley through which the Lake Stream runs is a very striking physical feature of the district. It is continued in a south-easterly direction for nearly thirty miles, and extends across the Ashburton River as far as the Rangitata. It intersects these streams nearly at right angles, and bears little relation to the present principal lines of drainage. It severs the eastern mountainous district of middle Canterbury from the main range of the Southern Alps, and well merits some distinctive name. Haast called a part of it the Upper Ashburton Plains; but this name will hardly apply to the whole extended valley. I shall refer to it in this paper as the Lake Heron Valley, as that name has been used repeatedly by the recent Commission which examined the Canterbury runs for the purpose of closer settlement. At the upper end of the valley, about five miles from the Rakaia,

4. Lakes.

5. Present Glaciers.

recent origin, cut for about three miles through a solid rock barrier, the height of the lip above the floor of the basin on its up-stream side being over 200 ft. It is extremely likely that a lake once occupied this basin, which has been drained by the river cutting down its bed through the solid obstruction. It seems impossible to explain the peculiar relation of basin and barrier on any other assumption than that large valley glaciers have under certain circumstances great powers of basal excavation. No locality that I am acquainted with shows this better.

Immediately outside this ice-eroded basin the moderately elevated country has been dissected and the drainage-directions changed, largely as the result of glacier erosion; but I have not examined the country in sufficient detail to speak definitely on the matter, though I feel certain that it will furnish very interesting material bearing on the much-discussed question of the efficiency of glaciers as erosive agents.

(b.) Rakaia Glaciers.

The glaciers of the main Rakaia basin are (1) those on the flank of Mount Murray, (2) those at the head of the river, and (3) those on the north side in the system of valleys which rise in the main divide and run towards the south-west.

Fig. 3.—Map of Rakaia Glaciers.

On the north side of Mount Murray there are small cliff glaciers heading a stream known as the Little Washbourne, which joins the Rakaia about five miles above the outlet of the Lake Stream. Another glacier, of slightly larger size, heads a ravine on the north-west of Mount Gould, and almost

exactly opposite Whitcombe Pass. But by far the most important are the glaciers at the actual head of the Rakaia, from which the river derives a great part of its water. The furthest west of these is the Lyell Glacier.

(i.) Lyell Glacier.
(See fig. 3, and Plate IV, fig. 1.)

The Lyell Glacier was discovered by Dr. von Haast in the year 1862. He saw it from Mein's Knob, but did not actually visit it, although he must have sent on some one to take an aneroid reading of the height of the terminal face. Mr. G. J. Roberts, the late Chief Surveyor for Westland, crossed the end of it when he made his survey of the Ramsay Glacier and its neighbourhood, but he does not seem to have done more. It is thus an unknown glacier, although within a reasonable distance of settled parts of Canterbury. The present writer, with two students, made an exploration of it in December last, and the following facts about it are the result of observations made on that occasion.

The Lyell Glacier extends from Mount Tyndall^* in an easterly direction for nearly five miles till it reaches to within a mile and a half of Mein's Knob, the bluff which fronts the Ramsay Glacier on the south side of the Rakaia River. It is bounded on the south by Mount Goethe, and on the north by the main range of the Southern Alps, and then by a spur from that range stretching in an easterly direction towards Mount Kinkel. The floor of the valley is a little over a half a mile wide on the average; it is wider than lower down, immediately between Mount Kinkel and Mein's Knob, where the valley takes a turn to the north towards the Ramsay Glacier. In former times the Lyell Glacier overrode the end of the spur where Mein's Knob now is, and truncated it-partially, leaving the knob with the characteristic shape produced by this mode of glacier erosion. Jim's Knob, on the opposite side of the river, has been formed in the same way by the Ramsay Glacier. The river which issues from the present Lyell Glacier may well be called the Lyell River, the name “Rakaia” being given to the stream formed by the junction of the Lyell with the twin stream from the Ramsay, the present confluence being between Mein's Knob and Jim's Knob. The two streams from the Ramsay and the Lyell appear to be of about equal size.

The Lyell River issued at the time of our visit from near the north side of the terminal face. A large body of water came from an ice-cave near the middle, ran by a tunnel under the ice in a northerly direction, and added considerably to the volume of the main stream. Behind the ice-cave the cliffs rose to a height of 60 ft. A small creek coming from the east side of Mount Goethe enters the valley on its south side immediately below the end of the glacier. This marks roughly the present position of the face. The floor of the valley is kept fairly clear of morainic accumulations by the transporting action of the powerful stream issuing from the glacier. About a half a mile down a high mound still remains which belonged to a former lateral moraine; but even this shows signs of being rapidly removed by the river. It is hard to tell from the present form of the terminal face whether the glacier is retreating at the present time, but the southern side shows signs of a collapse, which suggests that this is the case.

south-west. The mountain called Tyndall by Haast, and pictured in his report on the Rakaia, is not the same Mount Tyndall to which he gave the name from the valley of the Rangitata, nor yet again from the Godley. It is also probable that the Mount Tyndall of the excellent Westland survey is none of the mountains to which Haast gave the name. This is somewhat unfortunate, and appears to result from the mistake made originally by Haast, who though that he saw the same peak from each of these three great valleys. Mr. Earle has recently pointed out that the mountain called “Tyndall” by Haast when he explored the Rakaia is probably a peak not really marked on the maps, but one to which a new name should be assigned. Dr. Teichelmann has still more recently confirmed this observation in a letter to the author, and says that this peak is called variously McCoy Peak and Mount Nicholson, and that it is not on the main divide, but lies to the east of it.

The Lyell Glacier presents no features of features of special importance. Its surface is fairly smooth and little crevassed, the roughest portion being that near the confluence of the Cockayne Glacier and its disturbing influence. The lower portion is covered with moraine, which comes principally from those mountains not covered with ice and perpetual snow, but from those specially subject to the action of frost. This weathering-agent attacks the exposed surfaces of the rocks by means of their frequent joints and bedding-planes, and produces in this region a particularly large amount of angular material, which is poured on to the surface of the glaciers by numerous “shingleslips.” Although the Lyell has a thick covering in many places, its moraine is small as compared with that of its near neighbour, the Ramsay.

(ii.) Ramsay Glacier.
(Plate IV, fig. 2.)

This glacier takes its origin in the snowfield on the west side of Louper Peak, between it and Mount Whitcombe. It stretches for six miles in a S.S.W. direction between Mounts Ramsay and Kinkel on the west and the Butler Range on the east, and maintains a direction parallel to that of the valleys on the north of the Rakaia. It receives from the west two important tributaries—(i) the Clarke Glacier, which rises between Mount Westland and Mount Whitcombe, and runs in a north-westerly direction to a low saddle leading to the Upper Wanganui River on the western side of the range; and (ii) the St. James Glacier, which rises between Mounts Kinkel and Ramsay. Although the upper portions are comparatively free from debris, the lower three miles is more covered with moraine than any glacier with which I am acquainted. Even the Mueller and Tasman fail to come up to the Ramsay in respect to the size of the blocks and the completeness of the covering. Not only is there an abnormal amount of small material, but angular masses of the size of cottages occur piled together in disorderly heaps. Most of this comes from the precipitous faces of Mounts Ramsay and Whitcombe, which are so steep that little snow can cling to their bare crags, and are therefore rapidly broken up by the action of frost. The amount of material which comes from the Butler Range on the east is of comparatively little importance.

On the east side or the terminal face the Ramsay branch of the Rakaia rises from an ice-cave, but water certainly soaks through from all the face between Mein's Knob and Jim's Knob. The glacier is here just over half a mile wide.

There are clear signs of the decrease in size of the glacier, as abandoned lateral moraines marking old ice-levels occur in places along the valleywalls, and it is extremely probable that at a comparatively recent date it stretched right across the Lyell River till it impinged on the lower slopes of Mein's Knob. This river would then take a course through a tunnel partly under the ice and partly under the edge of the knob. A large part of this has at a fairly recent time slipped away from the face of the bluff and left a narrow chasm which affords a path round the shoulder of the knob, overlooking the river.

(c.) Absence of Terminal Moraines from Present Glaciers.

The glaciers of this region, like the great majority of those in New Zealand, are not forming any well-marked terminal moraines. Neither the Fox or the Franz Josef on the west coast not the Tasman or Mueller on the east have any sign of a terminal moraine; those formed by the Mueller are really lateral moraines formed by a glacier crossing a valley. In all the cases that I have observed the debris which reaches the terminal face is removed by the transporting agency of powerful streams. The Lyell Glacier has no true terminal moraine, and, although the Ramsay Glacier is heavily laden with waste from the neighbouring hills, there is no accumulation in the form of irregular heaps or a barrier at its end; in fact, there is no sign of such a moraine in the Rakaia Valley till the plains are reached. The Cameron Glacier is not forming a terminal moraine at present, although a very well-marked one lies some 800 yards away from the present face. In the Rangitata Valley an exceptionally distinct one occurs about five miles from the terminal of the glaciers. These moraines were formed when the glacier reached further down the valleys than now, but there must have been some difference in the then conditions which promoted the accumulation of débris across a valley, when no such accumulations are forming now. The actual circumstance under which terminal moraines are formed seem somewhat obscure. They are taken for granted as a feature of every glacier, but such is not the case. Why is it that the Fox and Franz Josef have formed huge terminal moraines some distance away from the ice, a little further down the valley, and are not forming any now? If the Ramsay Glacier, heavily encumbered as it is, were to disappear, and the loads of waste that lies on it were to coat the surface of the ground, there would be no sign of the heaps which characterize true terminal moraines. The question must resolve itself into one of supply and demand. If the glacier furnishes material in such quantity that the river can remove it, then no moraine will form. If for any cause the supply becomes greater or the volume of the river less, then accumulations will take place. At the present time the material supplied to the Ramsay Glacier is excessive and a moraine should result, but no moraine is forming. On the Lyell the amount is not really great, and its front is swept quite clear. It may be that our rivers have such a steep bed that they are equal to removing even the fullest supply that the glaciers can furnish. But when the case of the two great West Coast glaciers is considered this explanation does not appear quite so satisfactory. In former times, when forming great moraines, they had the same steep slope as now. The same remark also applies to the Cameron. We must suppose, therefore, that a little time ago the conditions were such as promoted the formation of enormous supplies of waste. This may be explained by an increase in the height of the

6. Former Glaclation.

came through the Rakaia Gorge on to the plains near the Point Station. Haast says that they extended several miles on to the plains, but I am inclined to think that the evidence for this extreme extension is of a somewhat doubtful character, and what he took for morainic accumulations are fluvio-glacial deposits such as are formed by the powerful streams issuing from the edge on an ice-sheet or glacier. In the neighbourhood of the Points Station there are very extensive areas covered with reassorted detritus arranged in irregular heaps of the Drumlin type, showing that the glacier reached almost to that point, and so must have been over sixty miles in length.

In the Rangitata Valley there are undoubted signs of glaciation where the river debouches on to the plains.

Thus the valleys of the Rakaia and Rangitata were filled with exceedingly large glaciers of the ordinary type, but there is no indication that they approached even distantly to the character of an ice-sheet. In the Lake Heron Valley, with its flat floor and wide cross-section, and its relatively higher altitude, they possessed in certain respects a remote resemblance to an ice-sheet of very small dimensions. The area of land in this valley formerly covered with glaciers extended twenty miles in length by at least eight in breadth; but it must be regarded as a great basin filled principally with snow and névé fields at the height of the glaciation rather than a true ice-sheet.

(b.) Old Moraines.

The chief morainic accumulations are those of the Hakatere and its tributary valleys. Here for over several square miles the floor is covered with irregular heaps of angular material forming the terminal moraines of ancient glaciers; fairly extensive accumulations occur at various points in the Cameron Valley and at its outlet, in the Upper Ashburton Valley, as well as on the north side of Lake Heron, on the slopes of the Sugarloaf, but these are insignificant when compared to the square miles of débris lying to the south of the lake, and forming the great dam which acts as the containing-wall on that side. This great deposit stretches across the floor of the valley, and also extends down it for several miles towards the Clent Hills Station and the Ashburton River, especially on the north-west side of the valley. At one place this has been cut through by a former river-channel which drained the lake, and has left as proofs of its former existence high terraces on either side of its bed. In the floor of the great deserted channel a tiny stream meanders, an insignificant remnant of the great river which at one time flowed from the front of the glacier and for a time served as the means of discharge for the surplus water of the greater Lake Heron.

The next extensive deposit is that which stretches from Hakatere Station right through to the Potts River. For miles the floor of this streamdeserted valley is covered with heaps of angular material. It forms a great barrier across the upper end of the Pudding Stone Valley and on the low saddles which lead to the small valleys behind Mount Possession Station. These were terminal moraines of the ice as it halted in its retreat from the eastern hills of the area.

Apart from these, the only morainic accumulations of any extent are those lateral moraines which mark the high ice-level on the valley-sides, They are common up the Rangitata and in the valley leading from Hakatere to the Potts, and occasionally in other places. A frequent position is

rounding the shoulder of a spur, or in the wider part of a valley, where in some corner or indentation in the side they have not been exposed to the full action of denuding and transporting agents. They sometimes occur near the mouth of tributary streams coming into a larger valley. These moraines are at times the terminals of the tributary glaciers, but in some cases they are undoubted lateral moraines of the main glacier. The incoming stream of ice has pushed the main glacier over, and accumulations have taken place on both sides of the tributary, especially on its upper side, and the stream of water which has succeeded the glacier has pushed over the main river in its turn, thus preserving the moraines at that spot, even if they have been removed from other parts of the valley by the transporting action of the river.

A peculiarity may sometimes be observed in the arrangement of the blocks of the lateral moraines which is occasionally useful for differentiating them from terminal moraines. In the latter case there is no order or arrangement—the blocks lie quite at haphazard; but in the case of lateral moraines the blocks usually lie with their long axes parallel to the direction of motion, and also the upper ones overlap the lower ones like pebbles in a stream. This arrangement is certainly rude at times, but it is quite distinct, and is brought about by the movement of the glacier causing a drag on the heavy accumulations between it and the valley-sides where the latter are not in close contact with the ice, such as at those parts where the valley is slightly wider, or where the glacier is showing signs of decreasing activity near its terminal face.

Thin coatings of moraine also lie on the tops of truncated and semitruncated spurs left by the receding tide of ice. An excellent example of this occurs on the downs behind Prospect Hill, in the angle between the Rakaia and the Lake Stream (Plate V, fig. 2).

(c.) Ice-planed Slopes.

These are a special feature of the Lake Heron Valley, all the north-west side of which is smoothed and terraced to a remarkable extent (Plate V, fig. 1), recalling the photographs which appear in Professor Park's bulletin on the Wakatipu district. The similarity of the landscape in the two districts is really surprising. Judging from the shape and slope of the glacier-shelves, the ice must have come from the north, even in that part of the valley adjacent to the Rakaia. This undoubtedly proves that a section of the Great Rakaia Glacier overflowed from the present Rakaia basin, came up the valley of the Lake Stream, and left its marks on the ice-planed hills towards Hakatere. As it overflowed the Clent Hills towards the Stour River it modified the landscape similarly.

(d.) Roches Moutonnées.

Ice-shorn rocks are a feature of certain parts of the land-surface. They occur in the Lake Heron Valley between the Clent Hills Station and Lake Heron, where their long scour slopes presented towards the north clearly indicate the direction from which the ice has come. In the Rakaia they appear at the sides of the river-beds, but lower down we find Double Hill and Little Double Hill, typical roches moutonnées, in the actual floor of the valley. In the Rangitata, the Jumped-up Downs, and the isolated hills in the floor of the valley between them and the Potts are also excellent examples of this result of glaciation. Clearly marked striae are rarely seen but if the surface coating of soil and débris were removed they

would appear plentifully, since the rocks are hard and resistant enough to retain the finest markings. Roches moutonnées of small size are very infrequent, and the larger ones grade into the general ice-shorn slopes and truncated spurs.

(e.) Truncated and Semi-truncated Spurs.

These are exhibited in all stages of development in the Rakaia Valley. Its sides have been straightened so that their alignment is almost perfect, and the spurs exhibit the triangular facets which result from the shearing-off of their ends. I have shown elsewhere that such a land-form is stable, and persists after other features resulting from glaciation have been destroyed by denuding agents. Even in the Ashburton Gorge these faceted spurs occur in an almost perfect condition. In the earlier stages, if the valley is not very deep, the mode of truncation of the spur appears to be the that a series of strips is planed off the ends from above downwards. Shelves left in an unfinished condition suggest that this is a common procedure, although erosion at the base of the spur and along its whole face must occur as well. These processes heighten the steep slope that the end of the spur presents to the main valley, and it must become so high in time that the glacier cannot overtop it, so that the valley can be widened only by scouring away the end and by eroding the base.

The angle of slope of the end of the spur is related to the width of the valley and the amount of ice passing through a particular cross-section. Where the valley is wider in its pre-glacial stage and the supply of ice is small the walls flare more and the angle of slope is not so steep. Where the valley is narrow and the supply of ice is great the angle approaches the perpendicular. An exceptional instance of this action is seen on the north side of Milford Sound, where the Lion Rock is vertical for hundreds, even thousands, of feet. The height of the face of the spur is increased by the deepening of the valley as the ice erodes its bed, and the slope tends to become more and more steep as the valley is more deeply cut and the stream of ice more confined, a result quite analogous to that produced by a river when actively eroding its bed. In some cases the spurs are only partially truncated, and specially so where a strongly marked ridge is overridden by ice coming in from a tributary or round a pronounced bend in the valley. Both Mein's Knob and Jim's Knob, at the head of the Rakaia River, owe their form to this action, and in their case the result has been intensified by the mutual interference of the Lyell and Ramsay Glaciers as they forced their way together through the somewhat narrow valley just below their junction. The ice has thus been crowded up on the slopes of the spurs at the sides, and they both show the well-defined notch cut close to the valley wall by the more active erosion of the ice as it swings round a corner. If this process is carried further it produces a rounded knob separated from the main portion of the spur. Prospect Hill (Plate V. fig. 2.), at the junction of the Lake Stream and the Rakaia, owes its from to a part of the Rakaia Glacier turning out of the main valley in the direction of Lake Heron and cutting off the projecting spur on the western side. It is a splendid example of a semi-detached knob, and the downs behind it are the remnant of the spur which has been reduced, but still remains at a height of 600ft. above the present floor of the main valley. In a former plateau country in which the valleys are well developed the ice frequently crosses saddles and cuts them down to a greater or less extent. In this way ridges are often partially or completely separated from the mountain mass which they were originally

connected with. Such ridges will generally lie parallel to the direction of the ice-flow, and will be abraded and reduced in height if the ice can pass over them, and reduced in length where it cannot pass over them. Thus all stages appear between the long spur and the conical hill. In the valley of the Waimakariri and in the Lake Coleridge basin on the north side of the Rakaia River the intermediate forms occur in perfect sequence; but in the Lake Heron district the destruction has been more complete, and the spurs grade into low flat roches moutonnées or into ice-planed slopes. An instance of such a cut-out spur can be seen in Shaggy Hill, five miles north of Lake Heron, where lay the original divide between the Ashburton and Rakaia basins, which has been lowered, and one of the spurs reduced to an immense roche moutonnée. An advanced stage in the destruction of such a ridge occurs in the Sugarloaf, at Lake Heron. This mass is evidently the remains of a spur which divided two valleys, an which projected above the stream of ice as a nunatak even at its maximum, for its top shows no sign of glaciation, while its sides and northern end have been rasped and smoothed away.

(f.) Valley-features

The profiles and cross-sections of the valleys are those produced by the action of glaciers on a matured stream system. Their special features, such as the alignment of the valley-walls, the truncation of the spurs, and the presence of glacier-shelves, roches moutonnées, and moraines, have been just dealt with. The surface features of the Lake Heron Valley have been little altered since the glaciation, and this suggests that it cannot have occurred at a distant period of time, geologically speaking. The Rakaia Valley has been subject to the action of a great river overcharged with detrital matter, so that the U-shaped floor has been filled to considerable depth with material derived from the glaciers at its head and from tributary streams and shingle-slips. The U-shaped form has thus been modified and the floor has been flattened by the deposits of an aggrading stream. The actual depth of the deposit is unknown at present, but it must amount in places in hundreds of feet.

The cross-sections of the valleys present some interesting features. Above the level of the truncated spurs, or above the level to which the glaciers filled the valleys—and this applics to the tributaries with equal force—the slopes become concave in shape (Plate VI, fig. 1). These are to be found in the whole mountain area of Canterbury. At higher levels they are occupied for a part of the year with snow, and they seem to owe their origin to its weathering action. By its presence the surface of the rocks is maintained in a moist condition, and slow but sure chemical and other erosion takes place, a shell-like hollow being eventually formed. The snow forms these hollows, in which afterwards, as weathering action proceeds, thick drifts accumulate. If the climate grows more rigorous or other circumstances promote the progressive accumulation of snow, then a small glacier forms. Such conchoidal slopes, developed under former more severe climatic conditions, are the favourite location of our moisture-loving alpine plants and our mountain meadows, with their rich displays of Ranunculus and Ourisia.

(g.) Corrie Glaciers in Relation to the Formation of Passes and the Dissection of Spurs.

In former times, when the conditions favoured the accumulation of thick drifts of snow, these concave hollows were gradually filled with ice

Canterbury which has been subject to glaciation in the past, or where glaciers occur now. The conclusions I have come to were made quite independently, and this may perhaps lend additional weight to them, as they are the result of observations in a country far distant from that where Professor Hobbs made his. Professor W. M. Davis, of Harvard University, has lately informed me in a letter that Matthes was the first to point out the relation of corrie glaciers to saddles and passes, in a report on the glaciation of the Big Horn Mountains (United States Geol. Survey, 21st Annual Report, 1899–1900).

(h.) Glacier Potholes.
(Plate VI, fig. 2.)

A very interesting and suggestive landscape-form occurs at the junction of the Lake Stream with the Rakaia. This is a round hollow over 300 ft. deep and about half a mile across, with the whole interior cut into horizontal or nearly horizontal shelves. These have evidently been produced by the erosion of ice moving in an approximately circular direction in horizontal planes. There is a remarkable resemblance to a pothole in an ordinary river, and the occurrence serves to emphasize the close resemblance that exists under certain circumstances between the erosion of a river and that of a glacier.

If the similarity in the action of the two streams be granted, the explanation of the formation of the hollow is quite easy. A portion of the great Rakaia Glacier running from west to east turned south just above the Lake Stream, rounded the spur at Prospect Hill, and turned up the valley towards Lake Heron. The ice impinged against the massive hill to the east of that valley, and just at the junction an eddy was formed which scoured out the hole, the horizontal terraces being a result of that gyratory motion. When the ice disappeared a rock-bound lake occupied the hole, which has since been emptied through the degradation of the barrier by the outflowing stream. An exactly analogous case occurs at the junction of the Cass River with the Waimakariri. The Waimakariri Glacier at the time of its greatest power flowed from west to east, and on reaching Goldeney's Saddle, about six miles below the Bealey, overrode the end of the spur and turned south-east into the valley towards Lake Pearson and the head of Sloven's Creek, following the line of the Midland Railway. On the east side of Goldeney's Saddle it formed a rock-bound basin exactly similar to the one in the Rakaia, but not quite so perfect. The Cass River now runs through it, and passes through a notch in the rim towards the Waimakariri. The whole locality presents a remarkable resemblance to that of the Rakaia. Goldeney's Saddle and its semi-detached knob corresponds exactly to Prospect Hill. There is a mountain mass on the east side of the inflowing stream, against which a part of a great glacier impinged. The rock-bound and ice-scoured pool with horizontal terraces lies between the two, no doubt in both cases forming a lake on the retreat of the ice, and then this has been emptied in both cases by the erosive action of a stream coming in from the south. Both occurrences emphasize the erosive action of the ice under similar conditions, and its capacity to scour out rock-bound basins in circumstances such as would favour the formation of eddies if water and not ice were the moving element. This action explains the formation of many of the rock-bound ponds and small lakes which occur freely in countries which have been glaciated. Numerous instances of

this action can be seen in the Sounds district on the south-west of the South Island of New Zealand, and those at the head of George Sound may be specially cited as furnishing excellent examples of the phenomenon.

(i.) Efficiency of Glaciers as Eroding Agents: Evidence furnished by the Locality.

A consideration of the glaciation of the area would be incomplete without some reference, however slight, to the evidence bearing on the much-discussed problem of the efficiency of glaciers as erosive agents. It is admitted, even by those who admit least, that glaciers act as flexible rasps, and remove the minor inequalities of the land-surface. The landscapes of the area under consideration give abundant evidence of this, but they are not so decided on the major question of the power of glaciers to excavate the beds on which they lie. It appears to the present writer that corrie glaciers, instead of acting as protective agents, as suggested by some observers of wide experience, do actively erode their beds, and also their side walls and their heads, especially the last, much more rapidly than the streams which issue from them erode their beds. Also, small valley glaciers have this power as well, and enlarge the amphitheatres at their heads at a more rapid rate than the rivers which they give birth to erode their valleys, so that these are narrow in their lower reaches, whereas their upper portions, which have till recently been filled with ice, are wide and of basinlike form. In this case, however, full consideration must be given to the possible neutralization of the erosive power of a stream overloaded with waste, as many of these streams are. Still, after making full allowance for this, there appears to be ample proof that ice in corrie glaciers and in the smaller valley ones related to them does not act as a protecting but as a powerful erosive agent.

There are other matters suggested by the landscape-forms of the Rakaia Valley which require very careful consideration in this connection, and apparently point in the opposite direction. First of all, there is a marked absence of waterfalls and hanging valleys. On the northern side of the main river the numerous parallel streams which rise in the main divide and flow south enter at grade (Plate VI, fig. 2), and the same is true of those on the southern bank, although they are few in number, and, with the exception of the Lake Stream, are much smaller. At first sight it would appear that the glaciers have only acted as a rasp, and modified but slightly the features of the previous valley system, were their efficiency as erosive agents not clearly indicated by the basin which occurs in the main valley of the Rakaia below the intake of the Lake Stream. Here there is a great hollow over twenty miles in length and 500 ft. deep in places, shut in at its lower extremity by a rock wall, which was once occupied by a great glacier and was subsequently either partially or wholly filled with silt. It was finally drained by the outflowing river cutting down a narrow gorge through the lip of the basin, a result hastened by the pouring-in of enormous supplies of waste as the glaciers retreated up the valley. This landscape-form is reproduced perfectly, but on a smaller scale, in the Upper Ashburton Valley. In this case it seems impossible to attribute the rock-barred hollow to any other cause than basal excavation by a glacier. Exactly similar features can be seen in the valleys of the Rangitata and the Waimakariri, and they no doubt occur in the less-advanced condition in the valleys of the Tasman and Godley, in the basin of the Waitaki, further south. Here lakes still occupy basins once filled with ice; but they also, like the basins further

7. Changes in Drainage in the Rakaia Valley.

The case of the change of drainage in connection with the Lake Heron Valley has been referred to several times previously. In the pre-glacial river system the Cameron undoubtedly drained south toward the Ashburton, and in all probability a small tributary ran north to the Rakaia from a divide between that river and Lake Heron. This divide was lowered by glacier-action, and the drainage was reversed. The change resulted largely from two causes—viz., (1) the piling-up of a barrier across the Lake Heron Valley at the south end of the lake, and (2) the lowering of the bed of the Rakaia by the erosion of its great valley glacier. The Lake Stream has thus been given a high gradient, and it has therefore rapidly removed any barriers that may have existed to the north of the lake, and has degraded the rocky ridge which formed the lake's containing-wall on that side, so that its size has been materially reduced from that which it had immediately succeeding the retreat of the glaciers. The great swamp north of Lake Herson is the old bed of that lake in its extended form.

The overdeepening of the bed of the Rakaia also allowed the small rock-bound lake at the junction of the Lake Stream and the main river to be emptied as well.

In other parts of the Rakaia Valley drainage anomalies occur, the most remarkable being that of Lake Coleridge. This lake occupies a valley parallel to that of the Rakaia, down which the Wilberforce River and glacier once flowed. A great terminal moraine was left by this glacier, which was also added to by a lateral one from the main Rakaia, across the south end of the lake, blocking the drainage in that direction. The effect was accentuated by the lowering of a saddle in the ridge between the upper end of the lake and the Rakaia Valley, so that at one stage the glacier flowed over this and reduced it so far that the Wilberforce River deserted its old bed and joined the Rakaia ten miles further up stream. The lake followed this direction too, and now discharges at its upper end by a small stream which joins the Harper River and flows into the Wilberforce. The Harper River also once flowed down a valley to the east of the Lake Coleridge Valley and parallel with it; but it, too, turned over a low saddle in the direction of the Rakaia, and flowed through the Wilberforce gap. Other small streams have behaved in a similar way, and have turned from their own valleys through gaps in the ridges that once separated their own valleys from the Coleridge Valley. These gaps were in all probability formed primarily by the action of corrie glaciers in the manner described previously. These remarkable changes in drainage seem to have been made possible by the overdeepening of the Rakaia Valley by its more powerful glacier below the level of the parallel valleys, thus providing a lower temporary base level for the tributary streams. The same fact is also true to a modified extent of the valley to the east of Lake Coleridge, which has been eroded deeper than they have by the large glacier which occupied its floor.

While the Rakaia has been gaining in the upper part of its course as a result of glacier interference, it has lost lower down. Between the Rock-wood Hills and the Big Ben Range there is a wide velley which was occupied