Go to National Library of New Zealand Te Puna Mātauranga o Aotearoa
Volume 54, 1923
This text is also available in PDF
(2 MB) Opens in new window
– 90 –

Art. 6.—Note on the Hanging Valleys of the Upper Rangitata Valley.

[Read before the Philosophical Institute of Canterbury, 7th September, 1921; received by Editor, 31st December, 1921; issued separately, 1st February, 1923.]

Plates 8, 9.

The question of the origin of hanging valleys has attracted so much attention, and their investigation has resulted in the expression of such divergent opinions by experienced observers, that the present author ventures to submit a note on his observations on this landscape-feature in one district of the mountain region of New Zealand. As a result of repeated excursions he has traversed all the main valleys of the Canterbury rivers and their chief tributaries to their sources, some of them many times, and also paid visits to the fiord region of Otago, as well as to several valleys of Westland once exposed to ice-action; and he should therefore be in a position to consider the arguments advanced by earth-morphologists whose experience has been chiefly gained in the mountains of the Northern Hemisphere.

The solution of the problem is no doubt dependent on the efficiency of ice as an erosive agent, a question concerning’ which there are varied opinions. Authorities like Davis (1909), Gilbert (1904), and Penck credit glaciers with enormous powers of excavation; whereas Garwood (1902, 1910), Freshfield (1888), and Bonney — British observers who have had. long experience in high glacier-covered mountains and the valleys leading therefrom—maintain that, instead of actively eroding the” hollows they occupy, glaciers actually furnish a protection against the more powerful action of running water; and these opinions are supported in America by Fairchild (1905), Russell (1905), and others.

According to the first group of authorities, hanging valleys are due to the greater activity of the ice occupying the main valley in comparison with that in the tributaries, this activity being attributable to the greater volume and thickness of the ice in the former, the result being that the floor of the main valley is overdeepened as compared with that of its tributary, so that when the ice-flood has subsided the tributary river enters the main valley with discordant grade. De Martonne (1910) attributes the formation of hanging valleys chiefly to lateral sapping, though he admits a certain amount of overdeepening, but not on the bars (verrous) across the main valley, near which hanging valleys are also found. Certain members of the second group of authorities regard glaciers in the tributary valleys as protecting their floors, while the rivers of the main valley lower their bottoms by normal stream erosion.

A consideration of the satisfactory nature of one or other of these explanations arose during two visits to the upper Rangitata in company with three of my colleagues, when we were fortunate in having ample opportunity of observing more perfect suites of such valleys than it had been my fortune to see previously.

One group of these occurs on the left (eastern) bank of the Havelock River, on the western slope of the Cloudy Peak Range, a spur running from

– 91 –

the main divide between the Clyde and the Havelock rivers, the two main feeders of the Rangitata. Cloudy Peak (7,870 ft.) is a fine rocky mass near the end of the spur, forming a counterpart to Mount d'Archiac (9,279 ft.), which lies just opposite, on the western side of the Havelock Valley, on a spur of the Two Thumb Range (see map). Two of the hanging valleys are in the most perfect condition; they have the characteristic U-shaped cross-section of a glacial trough with enlarged head, and enter the main stream by waterfalls (Plate 8, figs. 1, 2).

A striking feature of the locality, however, is the association with these of other valleys which enter the main valley at grade, and also of cup-shaped hollows, not worthy of the name of “ cirque,” which are obviously connected genetically in some way with the action of ice. Neither of the main theories advanced by the authorities cited above accounts for their origin satisfactorily, although these explanations may be quite sufficient in other cases.

The other group I refer to occurs in the Clyde and Lawrence valleys, the last-named being the most easterly of. the streams which form the Rangitata River. On the western side of the Clyde, on the flanks of Cloudy

– 92 –

Peak Range, is a series of hanging valleys of the initial, simple type, so perfect in arrangement and in features that they might almost be sketches in a text-book; in fact, the well-known diagram by Davis, so often cited in this connection, might serve as an illustration of them. I have, unfortunately, no photograph of them. Farther up the Clyde are other valleys of the more, developed type seen on the Havelock. I had a better opportunity, however, for examining closely those occurring up the Lawrence Valley. This is a most remarkable valley for showing the remains of glaciation in their most perfect form. Old terminal moraines forming complete dams across the valley, except where breached by the river, lateral moraines, faceted spurs, and other evidence of glacier-action occur, almost as fresh as if they were formed yesterday. The only remnant of the glaciers responsible for them is the small valley glacier, nearly a mile in length, at the head of the river, and small cliff and corrie glaciers on the shady south-eastern slopes of the Jollie Range, which flanks the valley to the north-west—these glaciers now occupying the upper parts of the hanging—valley basins which were excavated at the time of the former ice—flood.

From six to nine miles above the junction of the Lawrence with the Clyde occur three of these hanging valleys in perfect state of development (Plate 9, figs. 1, 2), discharging into the main valley by waterfalls. These valleys are of the well—matured type. Associated with them, are hollows similar to those in the Havelock. The series is, however, more complete, so that every stage occurs between the incipient shell—shaped depression and the maturely developed hanging valley; and the association seems strongly to suggest that the same causes operating now will in process of time produce the matured hanging valley from these hollows, without calling in the overdeepening effect of the main-valley glacier.

Before dealing with this question, however, it will be best to give a brief summary of the general geological conditions of the area under consideration.

The rocks in the valley of the Havelock consist of greywackes and slaty shales, the latter approaching a true slate in character with slightly lustrous cleavage-surfaces, but with the cleavage always parallel to the stratification. In the Lawrence Valley the rocks are more truly greywackes, of texture varying between the coarse-grained type and the fine black slaty variety. Owing to the proximity of the area to the Mount Potts fossil-beds in Rocky and Tank gullies, the matrix of which is similar to the rocks encountered at the head of the main river, it must be concluded, tentatively at all events, that the strata are of the same Trias-Jura age. The submetamorphic character of the strata appears more pronounced towards the west—that is, towards -the main divide at the head of the Havelock and Clyde rivers—and also towards the proximal end of the Two Thumb Range in the neighbourhood of Mount d'Archiac.

The beds have a general northerly strike, with slight local variations on either side of this mean, but in the Lawrence district the direction is more commonly N.E.-S.W. The angle of dip is generally high.

The geological history of the area is analogous to that of other parts of the Southern Alps—viz., that after a long-continued period of sedimentation during Mesozoic times the beds were folded up into a mountain-chain of a somewhat simple Alpine type; this was-reduced to a peneplain at the end of that era, covered with a discontinuous veneer of Tertiary sediments, and raised at the close of the Tertiary era into an upland, which during its elevation and subsequently was exposed to vigorous stream and glacier erosion. The original streams which established themselves on the surface appear to have been consequent, with general

– 93 –

directions across the strike, but their positions have been modified by the formation of structural basins and lines of tectonic weakness, so that there are departures from this general arrangement. However, in the upper reaches of the valleys the control of the grain of the country on the direction of the rivers appears more complete, so that there is a pronounced tendency for the tributaries and also the headwaters of the main rivers to develop their valleys along the strike of the beds, and especially along the strike of their weaker members. Thus it is that the heads of all the main rivers of central and northern Canterbury have an almost identical arrangement. The directions of the Waimakariri and its tributary the White River are reproduced almost exactly in the Havelock and its tributary the Forbes. The Lawrence, too, has developed parallel with the strike, a tendency which is very pronounced “n its higher reaches. The Clyde, however, runs across the strike, maintaining the original direction taken by the river on what was in all probability the old peneplain surface. Emphasis is here laid on the control exerted by the direction of the stratification on that of the rivers, since the’ same dominating influence is to be observed in the formation of the hanging valleys.

It is fairly evident that before the onset of the Pleistocene glaciation the drainage had arrived at a stage of early maturity, and after the area has experienced the modifying effects of this glaciation the present tributaries still show an accordance in grade with that of the main valleys. No doubt filling-up of the valleys with the waste. deposited from overcharged streams may have tended to mask any discordance, but in the absence of any sign of this whatsoever it is reasonable to conclude that if it occurs beneath the veneer of waste it must be very slight or it would show somewhere. It is only in the middle and upper reaches that hanging valleys occur, and they are but slightly developed in the former section of the river's course. This suggests that the former glaciers, apart from their abrasive action, did not modify the shape of the floors of the valleys as much as has been generally supposed. I do not mean by this that I consider glaciers incompetent to do so, but in the case of the Canterbury region of the Southern Alps this has not taken place to any great extent. The absence of profound modification may be accounted for by supposing the duration of the ice-advance to have been short, but due consideration should also be given to the possible obliteration of the traces of glaciers by the action of frost on the mountains farther east.

In the upper reaches, besides the chief tributaries, numerous smaller ones enter the main valley in perfect accordance, and associated with both of these are hanging valleys. It is this association which is difficult to account for on any theory of differential erosion in the main valley as compared with that in the tributary. If one tributary has had its lower reaches removed and the floor of the main valley adjacent thereto lowered so that its junction is discordant, why has its neighbour, placed in an analogous position, not been similarly modified? There is no difficulty, however, as far as this point is concerned with those valleys which have had their discordance reduced in post-glacial times.

I might also say here that there is the entire absence of the development of the glacial stairway in the valleys of Canterbury, although they appear such a striking feature of the valleys of the European Alps and of other countries. There is nothing here analogous to the cascades in the floor of the Ticino Valley as described by Garwood (1909), or the breaks in the floors of the Alpine valleys with the paternoster lakes on the flat shelves as described by Nussbaum (1910). Even in the Sounds region, with one or two striking examples to the contrary, this feature is only occasional.

– 94 –

The explanation which credits the formation of hanging valleys to deepening by rivers while the tributaries are protected by ice also fails, since there is almost entire absence of evidence of deepening, but strong evidence in favour of aggradation. In no case that I am aware of, where a glacier at present reaches the level of the valley-floor, is the stream deepening its bed in front of the glacier-face; in every case the stream is overloaded with sediment, and is aggrading rather than eroding its bed.

These two hypotheses, therefore, do not afford a satisfactory solution of the problem as far as hanging valleys in the region under consideration are concerned. The suggestion made by the present author is that hanging valleys are at times due to progressive head-erosion of the rocks along belts of structural or lithological weakness.

The hanging valleys in the Havelock district are certainly associated with such belts, and have been eroded principally along the strike of the beds. This is clearly seen in that valley whose head is in the vicinity of Cloudy Peak (Plate 8, fig. 1). Cup-shaped hollows in an earlier stage of development are associated with these more perfect forms, an instance of which is given in Plate 8, fig. 2, to the left of the typical hanging valley. If sapping-back of the head-wall along a belt of weakness should occur here, then the typical form would be developed; and if the mouth or lip over which the water falls into the main valley were marked by the presence of a hard rock or bar of more resistant material, then the discordance would be perpetuated: whereas if the stream issuing from the tributary glacier, or from the tributary valley after ice had disappeared, established itself on an easily eroded bed, then the floor would rapidly be adjusted to the main valley.

These small valleys with grades adjusted posterior to the glaciation are entirely distinct in origin from the main tributaries whose valleys were formed in preglacial times, and which have maintained their accordance. The process of adjustment of the grade in post-glacial times is well illustrated by the photograph of the hanging valley in Plate 9, fig. 2.

In the Lawrence Valley there are excellent examples of the influence of structure on these elevated glacial troughs. Plate 9, fig. 1, gives an illustration of a valley developed along the line of strike; the direction of stratification clearly shows in the picture. In Plate 9, fig. 2, is a similar valley whose lip has been eroded across the strike; but at the back of the picture the development of a trough parallel to the strike is quite apparent, a result being produced analogous to the development of a subsequent stream drainage parallel to the strike. The form which now ensues is a kind of double-headed valley. In the incipient hollows alongside these typical valleys, and grading into them, as regards their main features the same process is to be seen, but the result is not yet so pronounced. The lengthened member of the double-headed trough often forms a pass between adjacent and parallel main valleys.

The fact that so many of the hanging valleys have been eroded on the strike of the beds results chiefly from the fact that the main river directions were initially determined across it. Thus the remarkable and regular series on the west bank of the Clyde are arranged at right angles to the Clyde for this reason; but where the strike-direction varies locally, and also in the higher reaches of the rivers where the tendency of the main stream is to set itself parallel to the strike, the hanging valleys across the strike are increasingly numerous.

If such valleys arise from the development of glacier-filled hollows the level of the lip will be controlled initially by the height at which

– 95 –

such commence to form. On passing up a valley glacier these hollows are observed first of all at considerable elevations above the glacier, on the fringe of the area which is being glaciated. The hollows there formed are generally small in size, but as one goes up the glacier' they become bigger. Intercalated with them may be actual tributary glaciers, in many cases occupying the floors of the valleys of the tributaries of the old river-system. Their surfaces will be accordant with that of the main ice-stream, though their beds may be discordant. If the ice disappeared these might form hollows with a moderately denned lip, but the load of waste might completely mask it if the discordance were small. However, tributaries, distinct from these, may occur which do not join the main stream at surface grade, but cascade” in as icefalls. Compare, for instance, the Ball Glacier with the Hochstetter Icefall on the Tasman, and the Cockayne Glacier with the Heim on the Lyell. The latter member of these pairs would no doubt leave hanging valleys were the ice to disappear, and yet they would not be dependent for this feature on differential erosion.

Such elevated hollows are enlarged by glacier erosion, chiefly by the sapping-back of their heads till they may contain glaciers of tolerable size. The erosion of the lip of the hollows does not appear to be at all marked, either as the result of glacier erosion or subsequent stream-action. The fact that the lip is so markedly preserved in many cases is rather a testimony to the inefficiency of stream erosion as compared with glacier erosion, at all events under certain conditions. In some cases, no doubt, the bar at the lower end of the cirque is deeply incised, generally as a result of structural weakness, and the tributary may enter the main valley at grade, and this may account for the occurrence of discordant and accordant minor tributaries.

The possibility that hanging valleys might be eroded along a belt of weak rocks was considered by Davis (1900, p. 283), but he decided against it even in the critical case of the Linth Valley, leading into the Wallen See. But he evidently considers it a possibility in connection with corries, for he says, when dealing with these basins on page 305, “ Unless the erosion of the corries has been guided by differences in rock-structure there does not seem to be reason for their possessing a basined floor at this stage of development; but if a change of climate should now cause the trunk glacier to disappear, while many of the blunt-headed branches .remain in their; corries, each little glacier thus isolated will repeat the conditions of erosion above inferred for the trunk glacier.” Davis thus contemplates that rock-structure may have an important effect on the, formation of corries, but it is hard to see why the formation of an extended basin should be delayed till change of climate. Why should this process not go on at the time when the trunk valley is filled with ice?

Some such mode of formation of hanging valleys is described by Russell (1905, p. 79). He says, “ It is evident from the inspection that typical mountain-side glaciers are engaged in excavating depressions for themselves, which at an early stage in the process have the essential features of cirques, and at a later stage develop a valley of the typical U-shaped cross-profile, flat bottom, & c., with a cirque at its head. These glaciers may- be said to burrow into the mountain-side by headward extension, chiefly, as is judged, by ‘quarrying.’ The lower limit to which they are able to excavate their beds, or the local base level, is determined by melting. A glacier of the type referred to deepens its bed to this

– 96 –

level, and, given time enough, extends it headward until a flat-bottom-trench is produced. Should the glacier then melt, a hanging valley would be left, the mouth of which would open out on the slope on which the glacier originated.” This, statement seems' to fit the conditions here admirably; but in the case under consideration the backward sapping will be more rapid if carried on along the strike of beds of structural or lithological weakness.

An important circumstance to consider is the cause which determines the height at which hanging valleys lie above the floor of the main valley. If they are to be attributed to overdeepening of the main valley or to cutting back the lower courses of tributaries', then the heights should be accordant in a general sense, unless there is some special reason to the contrary. Thus the Stirling Fall and the Lady Bowen Fall, of Milford Sound, show some approximation in height, but this is entirely different from that of Harrison's Arm or Smbad Valley—all reference to elevation being made, of course, to the bottom of the sound as the datum-level. If these hanging valleys are due to differential erosion, then their preglacial form and gradient must have been quite different This general accordance in height is tacitly admitted by the diagrams used by Davis to illustrate his position. He also says (1909, p. 340), “ The depth of glacial erosion in the main valley is roughly indicated by the discordant altitude of the hanging lateral valleys, as “well as by the height of the main-valley-floor steps; but in both cases allowance must be made for the glacial erosion of the lateral valleys, or of the bench at the top of the rock-steps.” I think from this that he evidently “would expect all hanging valleys in one reach of the main valley to be approximately accordant.

Both Gilbert (1904, p. 154, fig. 77) and Garwood (1910, p. 329) certainly indicate that the floor of hanging valleys in the same main valley are occasionally accordant, so that it is possible that the objection made here as to the variation in level does not always hold; but if we admit the correctness of Russell's statement cited -above, this, too, will result in approximate accordance of the lower lips of a suite of hanging valleys of approximately the same size.

If it be granted that some hanging valleys may be developed from corrie basins, then the height of their lips will be dependent on the level at which corries can form. Geikie (1902, p. 233) notes that this is related to the height of the snow-line, and its level will, in general, be accordant under similar conditions of slope and aspect in the same region; therefore the accordance in the level of the floors of corrie glaciers and the hanging valleys developed from them is what we might reasonably expect, even if they are not to be regarded as due to the overdeepening of the main river—valley. The general lowering in height of the corrie-level when followed westward up the valleys of Canterbury is what might be expected, since the snow-line falls the nearer approach is made to the main divide. It may be noted here that the table given by Penck, and quoted by Geikie on page 233 of the work just cited, as to the level at which corrie-lakes occur in New Zealand—viz., from 600 to 1,200 metres—gives too restricted an upward range, since well—developed cirques and lakes are found at heights exceeding 4,500 ft. (1,400 m., approximately).

Hanging valleys of a type similar to those occurring in the upper Rangitata can be seen in the other valleys of Canterbury—as, for example, on the higher levels on both sides of the Bealey River, in the Waimakariri between the mouths of the Bealey and Hawdon, also in the Wilberforce

Picture icon

Fig. 1.—Hanging Valley, havelock River: cloudy peak in the background : showing valley of simple type eroded along the strike of the beds.

Picture icon

Fig. 2—Hanging Valley, havelock River: showing typical valley with mcipient hollow to the left

Picture icon

Fig. 1.—Hanging Valley, Lawrence River: showing development parallel to the strike of the beds, also the lip of the valley in process of being reduced.

Picture icon

Fig. 2—Hanging Valley, Lawrence River: showing the valley, originally established across the strike, developing another head parallel to” the strike; the lip is in a further stage of reduction.

– 97 –

Valley. It cannot be urged that these have been formed by. sapping-back along lines of structural weakness, since some, of them lie across those lines, but the method of formation is analogous to those in the Havelock. Some of them certainly show a tendency to enlarge their heads along the; strike when their general direction crosses it at right angles.

Although it is here urged that many of the hanging valleys of the alpine area of Canterbury are not to be attributed directly to differential erosion, there are some cases which may be explained in that way. Andrews Greek, 1 a tributary of the Waimakariri on the north side, between the junctions of the Hawdon and Poulter, may be attributed to this cause, since there is good evidence of overdeepening in this portion of the Waimakariri Valley; but the possibility ‘of the discordance being due to’ tectonic movements in this case must not, be overlooked, though I do not. think it probable.

There are other cases, however, which need special consideration. The formation of Arthur's Pass is most easily explained by supposing that it is due to two hanging valleys which have sapped back along the—strike of the beds, one from the west—coast side and the other from the Bealey side of the range, and the drainage has ultimately been captured by the former, leaving both ends of the valley in a hanging condition. An -exactly analogous case “occurs in Browning Pass- not the well-known pass at the head of the Wilberforce, but one which leads from the valley of the White River, one of the feeders of the Waimakariri. Another type of hanging valley near the main divide is to be- seen in Browning Pass itself; also in Walker's Pass, at the head of the Hawdon; and again on the western side of the upper Waimakariri River to the south of Mount Armstrong. One description applies in all these cases.

The main stream comes from a northerly direction parallel to the strike of the beds, and flows along the floor of a characteristic deep glacial trough. The walls, notably the western, are precipitous, and streams sometimes cascade over them from hanging—valley mouths. The floors of these hanging valleys cut across the strike, but they have been reduced to such an extent that they furnish - fairly easy routes across the main divide to the valleys of the western slope, and the streams occupying them are partially, or wholly diverted to the west, as in the case of Browning Pass.

The marked discordance in these cases is to be attributed to the facility with which stream and glacier erosion has proceeded along the strike, so that overdeepening has taken place in these cases, whereas erosion across the strike has been slow. The - erosion of the main valley was in all probability due to stream—action initially, but serious modification of the valley has followed, and the present landscape-features are those attributable, to ice-action rather than to stream-action. It is conceivable, however, that stream erosion may maintain a steep-sided valley in these cases, since the beds dip at high angles and hard bands occurring at regular intervals' in the greywacke series help to maintain that approach to perpendicularity which was impressed on the valley—walls by ice-action.’

At various places in the above discussion I have suggested that the modification of the valleys of the Southern Alps by glaciation has hardly been of the profound order originally indicated by Hector and Haast, and endorsed by later geologists, including myself. My reasons for doubting the statement are—

  • (1.) The general absence of discordance in, the grade of-the tributaries at the point of junction with the main streams.

– 98 –
  • (2.) The possibility of attributing hanging valleys to causes not demanding profound deepening of the main valley.

  • (3.) The comparatively slight modification of many of the spurs reaching down into the main valley. Truncation and partial truncation are admitted, but this does not demand any abnormal change.

  • (4.) The difficulty of restoring the preglacial stream-system if profound modification be granted, though the restoration is easy if the modification has been slight. This reason has certainly no great weight of itself, but taken in conjunction with others it is of some importance.

  • (5.) The thickness and extent of the morainic deposits are not such as one would expect had the valleys been modified profoundly, this neces-sitating the removal of an enormous volume of rock. In the Mackenzie country, near Lake Coleridge and the gorge of the Rakaia, in the valley of the Ashburton near Lake Heron and Hakatere, and at other places, the morainic material is a veneer resting on older river-gravels, or on Tertiary and other beds, rather than a massive deposit. The area covered may-be large, but the total volume of such deposits bears no comparable relation to the amount of material which would have to be removed were the modification by glaciation at all great, and certainly not so much as would be necessary if the hanging valleys had been formed entirely by overdeepening of the main valleys. Of course, a good deal of the solid taken away must have been in the form of flour, sand, and gravel; still, the amount of coarse fragmentary matter should be greater had such excavation been done by ice.

In conclusion, I wish to say that the object of this note is to draw attention to the possibility of processes other than overdeepening of the main stream having been responsible for the formation of hanging valleys. No doubt certain of these are due to this cause, but not all, and each case should be considered on its merits in view of the possibility of more than one cause operating.


A recent visit to the Wakatipu-Wanaka district has impressed me with the fact that local circumstances must be carefully considered when advancing an explanation for the formation of hanging valleys, and that, while some may reasonably be attributed to overdeepening of the main valley of a preglacial stream - system, other hanging valleys must be credited to causes which have no connection with this previous drainage.


Bonney, T. G., 1893. Do Glaciers excavate? Geog. Jour., vol. 1, pp. 481–99.

Davis, W. M.. 1900. Glacial Erosion in France, Switzerland, and Norway, Proc. Boat. Soc. Nat. Hist., vol. 29, “No. 14, pp. 273–322.

Davis 1909. Glacial Erosion in North Wales, Quart. Jour. Geol. Soc, vol. 65, pp. 281–350.

De Martonne, E., 1910. L'erosion glaciaire et la formation des vallees alpines, Annales de Geographie, No 106, pp. 311–12.

Fairchild, H. L., 1905. Ice-erosion Theory a Fallacy, Bull. Geol. Soc. Amer., vol. 16, pp. 13–74.

Freshfield, D. W., 1888. Note on the Conservative Action of Glaciers, Proc. Hoy.geol. Soc, vol. 10, pp. 779–89.

Gabwood, E. J., 1902. Hanging Valleys of the Alps and Himalaya a, Quart. Jour. Geol. Soc, vol. 58, pp. 703–14.

Gabwood, 1910. Features of Alpine Scenery due to Glacier Protection, Geog. Jour., vol. 36, pp. 310–36.

Geikie, James, 1902. Earth Sculpture.

Gilbert, G. K., 1904. Alaska, vol. 3, Glaciers and Glaciation.

Nussbatum, F., 1910. Die Taler der Schweizeralpen.

Russell, I. C, 1905. Hanging Valleys, Bull. Geol. Soc. Amer., vol. 16, pp. 75–90.