The Coastal Plateau.
The stereogram (Part I, Plate 41) and profiles (Plate I; Plate 2, B, D, E) of the region under consideration show that the edge of the elevated Fiordland Peneplain is separated from the sea by a lower sloping coastal plateau, a striking feature, which has been described by Hector and McKay. Thus Hector (1863) states concerning the coast between Dusky Sound and Cape Providence, “The shore is bounded by rocky cliffs a few hundred feet in height, from the summit of which there is a gentle slope for a distance of six miles to an elevation of fifteen hundred feet backed by smooth wooded ridges, the summits of which are three thousand feet above the sea. This slope is divided by a stream which comes down to the sea at West Cape … and its uniformity is broken by a few sharp cones.” Similar features continue east of Preservation Inlet. McKay (1896) notes that “the south coast from Puysegur Point to the Big River presents between the granite mountains and the shore an area five to ten miles in breadth which declines gradually or in terrace-like steps from twelve hundred feet to within two hundred feet of sealevel, the land as a rule terminating in a line of cliffs. Above the general level stands a line of rocky projections which have received the name of hummocks, but which are really hills of considerable size. … The passage across it of the several small streams and lesser rivers has been the means of cutting deep gorges and canyon-like valleys, or, in the case of larger streams, broader valleys, the sides of which are sculptured into gullies and ridges.” The terraces are, however, evident only in the south-eastern portion of the region, between Wilson's and Big Rivers, where attention was drawn to them by Hutton in 1875, but the present writers had no opportunity of studying them at close quarters.
At Puysegur Point the surface of the plateau is apparently unbroken by terracing, but is convex, its slope near the shore being nearly 6°, but becoming almost horizontal further from the sea (see Plate 2 E). The inner margin of the plateau has the line of “hummocks”* referred to by McKay running near and parallel to it, but is not marked by any abrupt cliffs. There is instead a pronounced change of slope around the base of the Bald Peaks, which may be traced from their south-east to their north-west aspect, and continues to the base of Arnett Peak overlooking the sharp bend in Long Sound, thus leaving a strip of the Coastal Plateau to the east of Revolver Bay which probably reaches over 1500 feet above the sea. It is not possible to recognise any systematic break of slope along the sides of Long Sound that would indicate the continuation therein of this coastal bench, though it seems to be represented on the southern flanks of Treble Mountain by the sloping shelf upon which lie the lagoons shown in the map. The shelf is, however, vaguely defined and becomes visible from Preservation Inlet only when the illumination of the mountain slopes comes from an appropriate angle. It is well defined on the south-western spurs of Treble Mountain and is sharply distinct where interrupted by Southport and Preservation Inlet, though its landward limits are very indefinite, owing to the gentle slope of the mountain. It may be traced northwards along the eastern coast of Chalky Inlet, continuing with increasing elevation into the prominent spur above its junction with Cunaris Sound. No clear indication of remnants of this coastal bench can be seen along the sides of the upper part of Edwardson and Cunaris Sounds, though there are hints of it to the north-east of Northport, where a gently sloping mammilated bench averaging about a thousand feet in height above the sea appears to have resulted from the glacial scouring of the remnants of the Coastal Plateau to a depth of several hundred feet. The western slopes of the Brothers and the Kakapo Mountains slope gradually down into the Coastal Plateau to the north of Cape Providence, and it is apparent from the topographic map of Dusky Sound that the change of slope is fairly abrupt at the base of Mount Bradshaw, and is continued by the western slopes of Resolution Island, the low Five Fingers Peninsula being also a portion of the Coastal Plateau. The broken ridge constituted by Gulches Peninsula and Passage and Great Islands lies mostly below the summit level of the Coastal Plateau, but the outer seaward slopes of Chalky and Coal Islands continue its level from north of Cape Providence to the south-east of Puysegur Point (Plate 3 D; and Map, Text-fig. 4).
Hypotheses of Origin.
The Coastal Plateau may conceivably have originated in one of several ways.
Hypothesis No. 1.—Andrews (1911) was of the opinion that it represents the downwarped continuation of the Fiordland Peneplain.
[Footnote] * A notable instance is the “Knob” near the head of the Wilson River, a rocky peak rising abruptly over 100 feet above the general plateau level. (See distant view, centre of Plate 3 A.)
Thus he says, “Earth forces raised a peneplain here in recent times to form two high plateaus.” The grounds for this conclusion are comprised within the following sentences taken from the same paper. “Whenever two peneplain or old age surfaces are found associated in resistant rock structures … and the two such surfaces are situated the one above the other, and the two are separated by a youthful or mature topography, it may be considered that they were formerly continuous, but are now discontinuous, owing to earth processes other than those due to erosive activities. Such earth processes may be either warping or faulting.”
In Andrew's view, therefore, the steep western rise in the profiles traced in Plate 2, D, E, would be the result of a pronounced local dislocation. That this view is inadequate seems probable for two reasons. The first is that the higher plateau has been so highly dissected that it is represented only by sharp-cut, narrow divides separated by huge valleys, but wide interfluves remain between the early mature or youthful valleys that traverse the lower plateau, even where it is cut in comparatively yielding Tertiary sediments. This absolute difference in degree of dissection is strong if not conclusive evidence that the two surfaces are not coëval. Secondly, the inner margin of the lower surface is strongly sinuous, for it enters for some distance into the major embayments of the upper surface; this makes difficult of acceptance the view that the two were separated merely by a line of warping.
Hypothesis No. 2.—A second possible explanation is that the Coastal Plateau resulted from partial peneplanation of a coastal strip of the Fiordland Peneplain consequent upon uplift and warping subsequent to its formation. Professor Speight has kindly drawn the writers' attention to Gilbert's (1904, pp. 129–134) account of low-level peneplains along a portion of the Alaskan shore-line which rise on their landward margin to a height of three or four hundred feet above the sea. They have been cut in relatively weak slates and mica-schists lying between the shore and more resistant granites and quartzites further inland, which form an older high-level plateau, from which mature valleys open out on the lower peneplains. The greater part of the Coastal Plateau near Preservation Inlet is considerably higher than the Alaskan low-level peneplain, and there is no relation between its limits and those of the various geological formations. Granites, as well as schists, slates, quartzites, and Tertiary sediments are in turn truncated by its surface. But though there is an abrupt rise from the Coastal Plateau to the level of the Fiordland Peneplain at the Bald Peaks and Mount Bradshaw (Plate 2 D), the long slightly concave slope rising from the plateau remnant above Southport up to the level of the Fiordland Peneplain in Treble Mountain may well have formed during a partial peneplanation such as Gilbert envisaged (Plate 43, Section; Plate 3 D; Plate 4).
Hypothesis No. 3.—The Coastal Plateau, though now uplifted, may be deemed comparable in origin with the strand-flats of Norway. The Fiordland Peneplain, it may be assumed, after elevation was
dissected by many small valleys, as well as the few large inlets, so that on partial submergence it had an exceedingly indented shoreline. At a time of considerable refrigeration, preceding, however, the last expansion of the ice, this very broken coast-line was so eroded by wave-action fortified by excessive frost-work that the many promontories and islets were consumed and a wide plattorm or strand-flat was produced. The detritus was washed into the valleys and fiords that traversed the area and was later removed from these by the normal marine, fluviatile, and glacial processes which were active during and after the subsequent uplift.
The formation of fiords and the coastal dissection generally before the cutting of the strand-flat in this manner are strongly urged by several Norwegian writers such as Nansen (1922, pp. 46–7). Holtedahl (1927, pp. 146–167) believes that this process, though probably effective, is not always of dominant importance. He supposes instead that, after some marine planation, the low shelves thus formed may become covered with a coastal ice-sheet and broadened by the headward cutting of its feeding cirques. As the hemicycle of glaciation advanced to its maximum, continental glaciers descending the great fiord valleys would overflow the coastal lowland, plane its surface, and smooth away the minor irregularities left by the cirque-heads in its landward bounding slopes. Comparison between the Fiordland Coastal Plateau and the Norwegian strand-flat fails, however, in several respects. The maximum elevation of the former is about ten times that of the strand-flat described by Nansen, and, in place of being nearly horizontal, its surface slopes sometimes with increasing declivity seaward. It is not best developed near the mouths of the main inlets, but between them, and its planation cannot have been conditioned by the “preceding splitting up of the landmass into peninsulas and islands,” as is necessary for the Norwegian strand-flat. It might be held that the “hummocks” and “conical hills” are analogous to the cliffed islands which make the Norwegian “skjaergaard,” but this cannot be discussed, as the writers were unable closely to examine the residuals in question.
The absence or obliteration of anything corresponding to a sea-cliff on the landward side of the plateau and its deep dissection by extended and consequent streams do not accord at all well with the idea of the origin of the surface in glacial times in the manner described, nor does there seem in Nansen's and Holtedahl's accounts to be any Norwegian analogy for the merging of the coastal platform into several flights of terraces, prominent on its seaward face as well as on the sides of the valleys, such as seem to occur near Big River (but of these only distant views were obtained).
Hypothesis No. 4.—The Coastal Plateau may be an uplifted surface which has had a twofold origin. It may be supposed that after a moderate elevation of the Fiordland Peneplain a low partially planed surface was produced subaërially in the manner described by Gilbert (See 2 above). Slow subsidence then permitted the sea to transgress across it, and to plane the gently undulating surface truncating alike the Tertiary and Ordovician sediments and the
granites, a process which must have been slow enough to permit the obliteration of nearly all traces of the small coastal cliffs that were formed. The “hummocks” and “conical hills” may then represent wave-cliffed residuals of the higher portions of this lowland which formed off-shore islets at the time of greatest submergence when the Fiordland Peneplain had been lowered to within two thousand feet of sea-level (See middle of Plate 3 A). Coastal detritus and river-alluvium accumulated in the drowned river valleys or was exported seawards. Then re-elevation of such character occurred that the wave-cut surface was raised to form the present Coastal Plateau rising from near modern sea level to a height of over fifteen hundred feet. It was more or less completely stripped of its cover of unconsolidated detritus as it rose through the littoral zone, and, in consequence of irregular rate of uplift, under favourable conditions developed terraces which modified its even slope. Streams that had extended across this plateau from the adjoining former coast and the new consequent streams cut deep canyons, with re-excavation of old valleys buried under unconsolidated detritus, this process being aided by the betrunking of the streams as the sea-cliffs receded before the attacking waves. Even the huge accumulations of detritus in the valleys of the former major streams were removed, and, thus rejuvenated, these latter proceeded further in the adjustment of their headwaters to the structures that they traversed.
Later, as the climate changed, glaciers advanced down the main valleys, scouring them out to their present depths, which are very great within the narrow parts of the valleys, though not so great near their mouths, where the flow, diminished by melting and ablation, deployed over wider channels. Within the region of glaciation erosion upon the valley-sides removed the last remnants of the pre-glacial high-level bench. Thus, it may be supposed, there were developed the raised Coastal Plateau, the deep sounds, and broad shallow inlets.
So far as it refers to the effects of glaciation, the hypothesis accords more or less with Andrews' earlier views (1905; 1906) and gives, moreover, a hint as to the origin of the “double slope,” to which he directed attention as occurring in many of the New Zealand Sounds (Andrews 1906) and which may represent residuals of the high-level bench.
On either side of the outlets of the major valleys the Coastal Plateau was covered by a thin piedmont ice-sheet, creeping slowly seaward and depositing ground moraine and outwash-gravels. These would temporarily protect the plateau from dissection by rivers, though such erosion would become active again as soon as the ice retreated.
The assumption that débris-filled valleys were completely scoured out after the Coastal Plateau had been elevated suggests that some valleys might occur in which this process was not carried to completion. An instance appears to be afforded by the Cascade Valley, two hundred miles further north. According to Turner (1930a), the valley was cut along a fault-zone in the Fiordland Peneplain,
and, during a period of subsidence, became partly filled by coarse gravel which spread out into a broad delta-fan. Uplift followed to such an extent that the landward margin of the fan was raised to 1800 feet above the sea. The gravel, however, had become strongly cemented. The revived Cascade River cut a deep broad valley across only the southern flank of its former delta-fan. Though it has since been truncated by coastal recession, notched by consequent streams, and partly covered by a thin sheet of ice (an overflow from the Cascade Glacier), a broad quadrangular block of the delta-fan still remains to form the Cascade Plateau, and residual masses of the cemented gravel, resting on the sides of the Cascade Valley, extend for some miles upstream.
It is not yet clear what evidence exists for the continuation of the Coastal Plateau between Cascade River and Dusky Sound. Park (1887) spoke of terraces from 100 to 300 feet high extending southwards to Martin's Bay, and, according to the verbal statements of Mr E. James, a strip of relatively low land runs between the high mountains and the coast almost as far south as Milford Sound. Yates Point, five miles north of the entrance to Milford Sound, as noted by one of us (Benson), has a form very suggestive of the existence there of a feature comparable with the Coastal Plateau (See Text-fig. 5). Further south, Hutton (1875, p. 80) noted a series of narrow terraces on either side of the entrance to Doubtful Sound, the highest being at an elevation of about 800 feet.
The degree of maturity attained in the main Preservation and Chalky valleys after the uplift of the plateau, but prior to the glaciation, demands attention. The existence at the present time of remnants of the Coastal Plateau level, represented by the benches on the slopes high above the entrances to Long, Cunaris, and Edwardson Sounds, and the rarity of small tributaries deeply incised into the sides of the Sounds, shows that the tributaries could not have been developed to such an extent that their dividing ridges survived the lateral scouring-action of the great glaciers. The major tributary valleys, namely, Richard Burn and Lumaluma Creek, which were clearly important prior to the uplift, were thereafter deepened by subaërial and glacial erosion to an extent almost comparable with the development of the main valleys. Between the extremes furnished by the small tributaries and these last are several streams draining catchment areas of moderate size each containing portions of the marginal slope of the Fiordland Peneplain and of the Coastal Plateau at its foot. In all instances the general course of the stream runs a little south of west parallel to a series of joints and fractures; valley-development may consequently have been favoured, even though the rock traversed is granite in all but one case. The upper portions of the valleys of Blacklock Stream and Dawson Burn, the only two which have been observed by us, have obviously been glaciated. Blacklock Stream, with a catchment of five square miles, has had its lower portion so cut back that it now hangs several hundred feet above Long Sound. The adjacent valley of Dawson Burn (See Text-fig. 6 B), with a catchment of ten square
Text-figure 6.—Views of three valleys cut in granite.
A. Valley of Gray River seen from off Kisbee Beach: a pre-glacial valley cutting a V-gorge 1400 feet deep through the edge of the Coastal Plateau and draining about eight square miles of westward-facing catchment. Bald Peaks (3500 feet) in background. Ice coming south from Revolver Bay (behind the hill in the left) divided above the low ground in the middle distance, part of it flowing towards the observer into Preservation Inlet, and part continuing southwards, rising on to the Coastal Plateau, and depositing thick masses of moraine on the slopes rising to the right.
B. Valley of Dawson Burn entering the southern end of Long Sound, draining from Caton Peak (3784 feet) with a westward-facing catchment of 10 square miles. Glacially modified in its upper portion; an open V-gorge in the lower portion. Remnant of Coastal Plateau on right. Main ice-flow from left to right. Traced from photograph.
C. Unnamed valley entering Edwardson Sound near its head. Eastward sloping catchment of about four square miles area, rising behind Inaccessible Peak (3600 feet). Main ice-flow from right to left. Traced from photograph.
miles, is so much more deeply recessed that its enclosing spurs have escaped complete truncation even though the valley opened on to the convex side of a right-angle bend in the course of the main ice stream. The same may be remarked concerning the Gray River valley (See Text-fig. 6 A), which has a catchment of eight square miles and enters the south-eastern angle of the Preservation Inlet depression through an open gorge cut to a depth of over 1400 feet in the edge of the Coastal Plateau. In both these cases the lower valleys are V-shaped, show little or no sign of glacial modification and appear to have been formed subaërially and adjusted to a base-level rather below the present sea-level.* At the time of their formation, therefore, the floor of the main valley, now occupied by Long Sound, must have been excavated to a depth below the present sea-level at least as far from the coast as the mouth of Dawson Burn, and, thereafter the crust movements amounted in all to moderate subsidence. The increasing body of evidence that in New Zealand, as well as in other countries, there was more than one epoch of Pleistocene glaciation (cf. Willis 1932) raises the question as to whether such a deepening of the main Long Sound valley was accomplished in pre-glacial times, or whether it could have resulted in part from the excavation performed during an early epoch of the glaciation. This involves the decision as to whether such deep gorges could be cut in inter-glacial times. On the views of Garwood (1910; 1932) this might be affirmed, and attention might be drawn to the contrast between the forms of these westward-facing tributary valleys and that of the small glaciated valley discharging eastward into Edwardson Sound near its head (Text-fig. 6 C). It is difficult to obtain quantitative conceptions from the available literature, but though a considerable amount of subaërial gorge-cutting in riegels and valley-steps and benches during inter-glacial times is recognised by Penck and Brückner (1909), De Martonne (1910–11), Nussbaum (1910) and others, there is little to indicate in the works cited that streams with such small catchments as the Gray and Dawson could excavate in granites such large valleys during inter-glacial times.† Moreover, the even slopes of the sides of these valleys lend no support to an hypothesis of multicycle origin.
The extent of development of Kohe Creek is even greater than that of these other streams. Possessing a catchment of eight square miles, this stream rises on the west of Treble Mountain. It is not
[Footnote] * Gilbert (1904, pp. 151–6) describes V-shaped valleys slightly modified by ice and more or less truncated at their mouths entering the upper portions of the Lynn Canal almost at grade and concludes that “there was a pre-glacial, comparatively narrow valley through the Lynn Canal, the floor of the valley being below present sea-level. Lateral V-gorges were tributary to it, and were largely adjusted to it at grade. The Pleistocene glacier broadened the river valley, truncated the side spurs and the tributary gorges, and at the same time materially deepened the valley for the whole breadth of the trough.” Capps (1931) holds that here “faulting had much to do with the establishment of the pre-glacial drainage lines.”
[Footnote] † Atwood and Mather (1932, pp. 58–9, Plate 17 C) described the gorge of a large stream which has been cut into schist and granite to a depth of about 2000 feet since the earliest Pleistocene glaciation.
noticeably glaciated in its upper portions and is deeply incised into the gently sloping flanks of this mountain and into the Coastal Plateau, whilst it enters Southport in a widely open early mature valley which has received a number of tributaries apparently entering it at grade. A flat delta extends a short distance into its valley. It also must be considered to have been formed in pre-glacial times adjusted to a sea-level rather lower than at present existing.
The absence of deep valleys dissecting the southern slopes of Treble Mountain indicates that during the same time the tributaries of the main valley through the Preservation Inlet area can have had only short valleys dissecting the surface of the Coastal Plateau and adjusted to the shatter-zones of which Isthmus Sound, with Revolver and Useless Bays may be the evidence. These valleys did not extend back into the higher slopes of Treble Mountain. Gray River possibly found an outlet at this time through a valley ancestral to Otago's Retreat, though the fact that it was adjusted to the same base-level as Dawson Burn, which entered a much more powerful stream, renders this unlikely. The seaward slope of the Coastal Plateau is sufficient reason for the lack of noteworthy northward-flowing tributaries.
Similar tributaries to the main Chalky Valley may have been developed along fracture-zones in the neighbourhood of Northport, but were not incised to any great extent in the flanks of the Kakapo Range, say, near Mount Inaccessible (Text-fig. 6 C). Traces of the old plateau level appear as a glacially scoured bench extending a couple of miles north-east of Northport, while on the eastern side of the Inlet the Coastal Plateau, which is well developed south of Kohe Creek, passes northwards into a faintly-marked bench above the mouth of Cunaris Sound. It is noteworthy that whether or not the eastern opening of Chalky Inlet was in existence in preglacial times, the pre-glacial maturing of Kohe Creek betokens a pre-glacial origin for the Southport depression, though not necessarily the existence of an outlet valley there.
Hypothesis No. 4, now under discussion, thus leads to the conclusion that at the commencement of glacial times sea-level stood rather lower than at present. The main streams, and a few of their tributaries draining remnants of the Fiordland Peneplain and rejuvenated by the uplift of the Coastal Plateau, had cut deep valleys extending far up their courses, but their minor tributaries had formed only short youthful gorges, dissecting their old valley floors and guided largely by fracture zones, and had seldom become recessed into the higher slopes. They had, however, cleared out most of the detritus that had accumulated in their former valleys. It follows, however, that the plexus of small gorges formed in the Preservation Inlet area must have taken a large share in the total excavation that had been accomplished in that area.* Accordingly, when the glaciers advanced down the main valleys and spread over
[Footnote] * Apart from a moderate degree of glacial modification the present features of Kiwi Burn Valley (see Plate 2 A), especially in its lower portion, afford a small-scale analogy with the inferred form of the Preservation Valley immediately prior to glaciation.
the Coastal Plateau, they occupied in the latter not only their main outlet channels, but also a plexus of small valleys which greatly facilitated the work of erosion.
On the assumption that erosion is competent completely to account for the present land forms, it should first be observed that the glacier discharging from Long Sound rose high above the level of the remnants of the Coastal Plateau. The scoured form of the promontory by Arnett's Peak, and of the plateau remnant above the mouth of Dawson Burn, and the broad channel of high-level flow from the end of Long Sound into the Gray River catchment which McKay had seen, afford evidence of this. There would consequently have been a vigorous discharge into the Isthmus, Useless, and Revolver valleys cutting down the divides at their heads, so that in the case of the last the ice crossed the divide to reinforce the flow in the Gray River valley. As the ice advanced thus on to the dissected coastal plateau with its greatest force concentrated at three points, conditions were propitious for the development of a wide terminal basin, even though the glacier became considerable deployed. Excavation was further facilitated by the fact that the formations on to which the glaciers emerged from the rather sparsely jointed granites were the abundantly jointed Ordovician quartzites and argillites, more susceptible to glacial plucking than any other formation in the region. It is significant that the eastern boundary of the depression occupied by Preservation Inlet almost coincides with that of the sedimentary rocks, while its southern shore, and its northern possibly, coincide with the heads of the small pre-glacial insequent gorges, for there are no important recesses made by large tributaries streams crossing these boundaries. The channels in the floor of the inlet follow the lines that might have been expected for the outlet streams, so that the various islands may be considered to be incompletely consumed resistant residuals of the former secondary divides. They consist chiefly of massive quartzite rather than argillite. Their small size compared with the granitic isthmus to the east betokens the greater resistance to glacial plucking and scouring offered by the latter.
The features of Otago's Retreat need special consideration. It is enclosed between steep almost rectilinear walls truncating obliquely the strike of the Ordovician and Tertiary sediments. Great bluffs of quartzite project from the eastern wall near the middle and at the northern end of the channel, and these bands of resistant rock are continued into Coal and Crayfish Islands. In a shallow recess between them lies Te Oneroa Beach, in which the rocks are largely argillites as at the Morning Star Mine. Immediately opposite this is the deepest part of the inlet, but the depth decreases rapidly near the quartzite bar on the southern side. The channel narrows seaward, and is half closed by a promontory composed of Tertiary arkosic sandstone. That the top of this promontory is the same height as the coastal plateau on either side of the channel seems to indicate that it has not been over-ridden by any effectively eroding mass of ice.
Certainly, the widespread morainic matter on the plateau on either side of Otago's Retreat may indicate a temporary extension of the ice-sheet over it, but the presence thereon of more or less stratified gravels shows that the conditions were often those of an outwash apron. Holtedahl (1929, p. 141) comments: “If there is any tendency for the ice to dig deeper in one sort of rock than another … conditions do not at any rate point in favour of the looser rocks; one would rather say that hard but jointed rocks give a more angular and complicated design to a fiord … so there is a greater chance of getting in these rocks an uneven irregular profile with very great depth. Evidently the plucking effect of a glacier is of paramount importance.” Matthes (1930, pp. 89–103) also emphasises (and, according to von Englen, 1933, pp. 590–592, possibly over-emphasises) the importance of close-spaced jointing for the promotion of glacial erosion.* Matthes holds that in the Yosemite region cross-walls, glacier-stairways, and roches moutonnées are all the products of selective glacial erosion at points where monolithic, hence obdurate, granite masses are interposed between much jointed, hence quarriable, rock. Hence in our region there is no reason to suppose that a relatively thin sheet of ice over-flowing from the main mass in Otago's Retreat should have on the Tertiary sediments as strong an effect as the thicker mass had on the harder but more jointed Ordovician sediments. At the same time, the seaward constriction of the “Retreat” suggests that it may have been formed by a lobe of the main glacier in the Inlet which had overflowed through a low saddle at the head of a small consequent stream traversing the seaward slope of the Coastal Plateau. The valley of this stream, possibly, was the outlet of Gray River at some stage of its history. The glacial lobe scoured out a small terminal basin, and did not extend beyond the promontory. As the tidal Big River fifteen miles east of the “Retreat” discharges from Lake Hakapoua through a valley cut in Tertiary sediments, so it may be that there was never any discharge of ice from the Inlet into the sea through Otago's Retreat. If this were so, the present opening of the latter would have to be explained by river-erosion and cliff-recession under wave-attack, together with some regional subsidence. Any terminal moraine must then have been removed in the formation of the outwash apron. Alternatively, it may be suggested that a terminal basin may have been formed by an ice-lobe in an early glacial period, the opening into the sea being made in the manner mentioned, but during an interglacial period, and enlarged by ice-action during a later glacial epoch. There is no need, however, for such an hypothesis. The small notch at the base of the promontory through which the lighthouse road has been made could easily have been cut by a small stream escaping from the ice-front.
[Footnote] * In Yosemite this holds within a single lithological unit. “Where the rock (granite) is massive or only sparsely divided by fractures, the glacier … can reduce it only by abrasion, a slow and relatively feeble process; on the other hand, where the rock is abundantly divided by natural partings the glacier will quarry out entire blocks and excavate at a relative rapid rate.” (Matthes 1930.) The writers have not yet had access to Ljunger's (1930) paper on the morphological significance of fractures.
If either of these views as to the origin of Otago's Retreat be accepted, the implied wasting away of the main Long Sound ice so near the coast may explain the failure of the main ice-stream in the Inlet to broaden the opening between the resistant masses of Crayfish Island and Cavern Head, though there is no reason to doubt that there was a considerable discharge of ice into the sea through it.
Hypothesis No. 5.—Though the hypothesis last discussed seems adequate to explain the present topography of the region, it is desirable to consider Andrews' (1911) final conclusion concerning the region: “The earth forces raised a peneplain here in recent times … and dropped a centre block to form the (Preservation) Inlet which has since been modified by glacial erosion.” At one stage in these studies this view seemed to the writers to be inescapable. The narrow rectilinear channel of Otago's Retreat, half-closed by a promontory of weak rocks, and parallel to well-known lines of fracturing and perhaps faulting (the hypothetical fault along the west coast of Fiordland), seemed best explained as the result of trough-faulting, though the need to suppose that the fault-strip did not extend the whole length, but left a “bridge,” subsequently reduced by erosion, at the southern end, imports a special feature which, though not fatal, lessens the probability of this hypothesis. Again, the rectilinearity of the boundaries of Preservation Inlet itself and in particular of the long wall extending from Revolver Bay southwards, together with the need of explaining the early development of a base below the present sea-level to which the Gray and Dawson valleys could become adjusted before the glacial period, offers support for the hypothesis that this broad depression resulted in part from a pre-glacial subsidence of fractured crust-blocks, such as has occurred on a large scale in many of the intermontane basins of the South Island of New Zealand. It still remains possible that such movements may have been instrumental in the dismemberment of the coastal plateau in pre-glacial times, but the features of Preservation Inlet alone seem inconclusive as evidence of recent subsidence of this character.
Further, the western side of Southport coincides with the most strongly marked fault in the district studied, and in topographic form is either a fault-scarp or fault-line-scarp. The eastern side has a much gentler slope, and north of the mouth of Kohe Creek rises gradually up to the level of the Coastal Plateau. The section (Part I, Plate 43) suggests this may indicate the stripping away of Tertiary rocks involved in a fault-angle depression during the pre-glacial dissection of the Coastal Plateau. If such a fault continued across to join the fracture-zone along the western shore of Edwardson Sound (thus outlining the granite massif), it is conceivable that for a time the outlet of the Chalky drainage may have been through here. This suggestion need not assume any dislocation of the Coastal Plateau, but an explanation of the maturity of Kohe Creek involving the pre-glacial subsidence of narrow crust-blocks in Southport and Chalky Inlet would require such movements. Once more, however, the evidence is inconclusive.
The glacier in Chalky Inlet had probably a greater catchment than that in Long Sound (though the exact position of the divide between them is yet unknown). Moreover, it received by the through valley at Last Cove about a third of the total stream of ice moving down Long Sound (estimated by a comparison of approximate cross-sections), and there is such abundant proof of intense glacial scouring as far down as the junction of Edwardson and Çunaris Sounds that the overdeepening of these Sounds requires no further explanation. Overflow into a valley draining the Coastal Plateau between Chalky Island and Gulches Head, and its subsequent enlargement would seem a sufficient explanation of the eastern outlet of Chalky Sound, even if the main pre-glacial stream discharged through Southport (which need not have been the case). The glacial enlargement of Southport to its present form (Plate 3 F) would, however, follow only if there were maintained a significant difference of surface-levels between the Chalky and Preservation glaciers, which the abundant opportunity for westward discharge of the former would tend to prevent, though McKay cites evidence of a noteworthy flow of ice through this opening at one stage. If his description of the locality be correctly interpreted, the edge of the Coastal Plateau nearest to the Seek Cove-Southport neck bears a covering of morainic matter, indicating that the thickness of the ice here was for a time over 1500 feet.
There is sufficient analogy between the conditions in Preservation Inlet and those which must have obtained in the western part of Chalky Inlet to render it unnecessary to assume extensive blockfaulting for the explanation of this western depression. Though there was here some deploying of the wasting glacier, three main streams must have come through Northport, Return Passage, and Bad Passage respectively, in each case through channels developed in granitic rocks. They passed out on to an area of normal, well jointed Ordovician sediments lying between Cape Providence and Chalky Island, which would have been very susceptible to plucking. The total volume of rock that was removed would have been comparable with that excavated from Preservation Inlet. The retention of a remnant of the Tertiary rocks now forming Chalky Island may be explained by Holtedahl's (1929) comment cited above.
General Conclusion.—The amount of topographical and geological detail available in this region of varied features has invited an attempt to make a fairly comprehensive application of the method of multiple working hypotheses to the elucidation of its geomorphogeny, in the light of the latest accessible discussions of the origin of the Norwegian fiords and American glacial topography. It has lead to the conclusion that in general the glacial excavation of a series of pre-glacial valleys adjusted to a more or less fractured series of diversified rock formations, together with small submergence of the region, is competent to account for the bulk of the features observed. And, further, though there are a number of facts suggesting that the differential subsidence of relatively small fractured crust-blocks may have influenced the pre-glacial relief, the evidence that this was the
case is not complete. Nevertheless, there is considerable probability that a major fault-plane determining the west coast of Fiordland obliquely truncates the coastal plateau in the region of Dusky Sound (See Part I, Text-fig. 2).