Landslides and Allied Features in the Dunedin District in Relation to Geological Structure, Topography, and Engineering.

Transactions and Proceedings of the Royal Society of New Zealand, Volume 70, 1940-41, Page 249

Landslides and Allied Features in the Dunedin District in Relation to Geological Structure, Topography, and Engineering. By W. N. Benson [Read before the Otago Branch, June 11, 1940; received by the Editor, June 26, 1940; issued separately, December, 1940.] The valuable memoir recently published by Sharpe (1938), and the increasing realisation of the disastrous consequences of soil-erosion and landslides within New Zealand have called forth several papers during recent years. The present contribution has arisen from the detailed mapping of the Dunedin district, wherein the variety of rock-formations and topography is reflected in the variety of the landslide-phenomena. Outline of Geology. The basement-rock is quartz-ablite-sericite-chlorite-schist which has a dominant though not invariable north-eastward dip at moderate angles. On a surface carved in this during Cretaceous times there rests locally a Mid (?) Cretaceous series of incompletely leached feldspathic sands and conglomerates. Over a much greater portion of the district the formation lying directly on the schist consists of quartz-gravel and sand, either loosely compacted or cemented by silica or haematite, and interstratified with lignite. On the submergence of this there was deposited thereon locally a Senonian (?) pebbly limestone, and in another still more restricted area a richly fossiliferous Danian (?) sandstone. Both of these are followed by the Abbotsford glauconitic mudstone 700–800 ft. thick containing casts of Cretaceous foraminifera and a Cretaceous fish. An incoherent “Loose sandstone” rests on this, 200 ft. thick in the southern portion of the district, but thinning out completely in the north, where locally the Abbotsford mudstone is followed directly by the very fine grained, grey, somewhat calcareous, Burnside mudstone 150–200 ft. thick, containing Upper Eocene foraminifera (Finlay and Marwick, 1940). Sharply separated from these is a richly glauconitic greensand, usually less than six ft. thick, merging upwards into the Caversham sandstone, about 900 ft. thick, to which a Lower Miocene age has been assigned (op. cit.). This in turn passes up into the Goodwood–Dowling Bay series of shales and limestone (late Lower Miocene), of which but little is exposed within the Dunedin district. A detailed account of all these formations except the last has been given by Ongley (1939). After sedimentation ceased, a general emergence of the region ensued, some faulting and a gentle flexing with a predominately south-easterly tilting, after which prolonged erosion resulted in the formation of a “Late Miocene” peneplain cut obliquely through the marine sediments in the eastern half of the district, but becoming nearly coincident with the stripped Cretaceous peneplain in the west (Cf. Benson, 1935).

In Pliocene times, volcanic activity broke out accompanied by slight crust-folding, and a series of effusions, explosion-breccias and minor intrusions were produced. The rocks comprised basalts, dolerites, trachybasalts, trachyandesites, trachytes, phonolites, and small amounts of less familiar rock-types, the sequence and relation of which will be discussed elsewhere (See Benson, 1934, and for earlier statements, Marshall, 1906, 1914). The sequence of effusions was interrupted at intervals by energetic explosions, particularly on two occasions, when the ejectamenta, washed down the gentle lava-slopes, formed widespread mudflows and flood-plain deposits which serve to divide the series of lava-flows into those of the first, second and third eruptive phases, each phase comprising a wide range of rock-types. The gentle crust-movements which had been intermittently in progress before, during, and after the eruptive activity culminated at its close in vigorous folding and faulting of the area covered by the volcanic rocks. The hills and valleys thus formed have since been modified by the erosive activities of the consequent series of streams which drained the dislocated land-surface. Further west the schist surface with its thin incomplete cover of sediments was thrown into broader warpings, forming a gently flexed plateau and an anticlinal ridge, both trenched by the antecedent Taieri River and separated from one another by an alluviated syncline, the Taieri Plain. Another strongly accentuated and faulted syncline was formed within the area of volcanic rocks and, now drowned, has become Otago Harbour. These features may be inferred from the series of sections given on the map herewith (Fig. 1), in which, for the sake of clarity, no distinction is drawn between the various types of igneous rocks, and the accumulations of blown sands are not indicated. While Sharpe's classification of landslides is assumed herein as the basis for genetic treatment, it will be convenient to consider the recent mass-movements of rock-material and detritus as they concern the above-mentioned series of rock-formations, either singly or in combination. Basement Rocks. The basement schists have afforded many examples of rock-slides, debris-avalanches, or more slowly creeping masses of detritus, where they have been deeply entrenched by steep-sided valleys. Of these, two which are important in connection with engineering, call for mention. Where the gorge of the Taieri leaves the western margin of the region mapped, its locally narrowed channel is spanned by a pipe-bridge conveying from distant uplands part of the Dunedin water-supply. The gorge here follows the strike of the schistosity-planes of the rocks which dip south-eastwards at angles of 20° to 40°. During the cutting of the gorge the valley-sides became locally steeper than the schistosity-planes, and where this happened, percolation through joints leading into these planes of easy separation, and the extension of weathering therein, resulted in dislodgment of huge joint-bounded blocks of schist which slid down the

Fig. 1.—General Geological and Structural Map of the Dunedin District.

hill-slope for some distance without change in the inclination of their planes of schistosity. Many such blocks are scattered over the hill-side between a high residual bluff and the river 700 ft. below; but far more have been heaped into a scrub-covered aggregate of irregularly tilted blocks in the bottom of the valley into which they have been piled to a depth of about fifty feet, have produced the channel-constriction spanned by the pipe-bridge, and have caused the filling of the valley with alluvium for a distance of one and a half miles above the constriction. In contrast with this presumably rapid sliding of slightly weathered, large joint-blocks, is the slow creep of thick masses of more or less decomposed rock-debris, clays, gritty detritus and large or small residual blocks down hill-slopes also approximately parallel to the schistosity-planes. This was well displayed at Waipori twenty miles south-west of the locality last described. Here, water diverted from the distant plateau has been lead by tunnel to a point 680 feet above an electricity-generating station at the bottom of the gorge of the Waipori River, to which station it was conveyed by four large pipes anchored at intervals to huge blocks of schist projecting above the general hill-slope, and so massive that originally no doubt was entertained as to the solidity of their foundation. In 1929, however, the thawing of an unusually heavy snowfall was followed by small slumpings of the slope-surface, and slight movements noticeable at the expansion-joints in the pipes between the several anchor-blocks. These movements were carefully measured by the resident engineer every few hours until after a fortnight had passed they ceased to be perceptible. A subsequent resurvey of the positions of the anchor-blocks relative to the datum point at the generating station, indicated the total down-slope movement that had occurred at each, the greatest being 4.85 inches. The measured movements in the expansion-joints between the six lower anchor-blocks on the slope prove to have been irregular because of differential movements, but those between the sixth block and that immediately above (which last moved altogether only half an inch) show a speed of movement during the fortnight, decreasing from its commencement at a very regular rate, following extremely closely the curve y = 6·287e–.186x–2·252e–.919x, where y is the distance in inches from the inferred position of final rest, and×the period in days from a point of time. The writer is indebted to Mr. J. G. Alexander, Dunedin City Engineer during 1929, for calculating this curve from the observational data collected by his officers. Its double exponential form suggests the normal decrease in the rate of movement of a body subjected to frictional restraint, modified by an increase in that restraint resulting from the drainage of lubricating water from the moving rock-debris; but it is impossible to give physical significance to the several constants. Since it was clear that the anchor-blocks had no solid foundation, the lines of pipes on the hillside were replaced by others in a tunnel, in the placing of which it was found that the mass of rock-debris and partially decomposed rock on the middle portion of the hillside was nearly 200 ft. thick.

Fig. 2.—Landslides in Abbotsford Mudstone, south of Saddle Hill, with dates of last movement of the masses of displaced material.

Cretaceous Conglomerates and Sandstones. The feldspathic conglomerates and sandstones which occupy a small area on the hill-slopes south of Mosgiel yield readily to weathering, and the debris covering these rocks shows very evident signs of slumping or creeping throughout its extent. The quartz-sands are usually too firmly cemented or occur on slopes too gentle to permit sliding; but, when incoherence is associated with very steep topography, rock-falls may occur. The best example of this is afforded by the white cliffs of incoherent sandstone (“The Chalkies”) surrounding the amphitheatrical head of Dodd's Gully which dissects the fault-line scarp brought into relief by the deepening of the Silverstream Valley. The material removed by the fall of sandstone and its former cover of basalt, together with that removed from the steep gully-sides, has been built into an extensive alluvial fan where the gully enters Silverstream Valley. Where the “Miocene” peneplain was so nearly coincident with the Cretaceous peneplain that at most a few feet of sandstone separated the Pliocene basalts from the schist, artificial cuttings through even gentle slopes reveal a layer or “stone line” (Sharpe, p. 24) of residual blocks of basalt lying near or on the surface of the schist, and covered by several feet thick of creeping roughly banded sands. Abbotsford Mudstone. The Abbotsford Mudstone is more prone to slide than almost any other formation in the Dunedin district. Its steeper slopes are often covered with “terracettes” (Sharpe) and even the massive unweathered rock may creep in semi-plastic fashion (See below). The detailed map of a single area (Fig. 2), namely, that between Saddle Hill and the sea, shows what may be an extreme case of repeated and extensive sliding. The mudstone here dips seaward with an inclination a little less than that of the land-surface (6°), and was covered by a thin flow of basalt of which small patches remain. The hedges planted when the land was cleared first in 1850–60, serve to distinguish between landslides earlier or later than those dates. According to local report most of the renewed movements of the material of former landslides have resulted from the discharge on to it of drains constructed to de-water higher portions of the hillside. The largest slide (see Plate 34), however, commenced to move after abnormally wet weather early in 1939, and with the thawing of snow six months later the subsoil was converted into a semi-fluid mass of mud, which, within a few days, moved for a distance of over three quarters of a mile down the valley, leaving at its head more than an acre of hummocky, torn turf and a slump-scarp forty feet high. It carried two hundred yards length of “metalled” roadway a like distance down its course, and built at its termination a “bulging crevassed dome,” from the base of which, as it became more fully charged with water during the next few

months, there protruded a tongue-like mass of mud which is now extending slowly further down the valley* Note added November 9, 1940. Since January, 1940, when the map in Fig. 5 was completed, this has moved about 100 yards further down the valley, and is now within 170 yards of the roadway. During the unusually dry winter and spring months it has, however, become well drained and dried, and movement seems to have ceased.. Where, as at Abbotsford, the mudstone has been covered by massive, jointed sheets of lava through which more steeply sided valleys have been cut, irregular slumping of the residual blocks of lava and of the clay derived from these and from the mudstone, has given rise to an uneven topography like that of hummocky moraines (Grange, 1921). As is not unusual in such cases (cf. Sharpe, pp. 56, 59, 74) a glacial origin has been claimed both for these masses and for others of similar origin near Seacliff (Park, 1910, 1924). Loose Sandstone and Burnside Mudstone. The Loose Sandstone has but a small extent of exposure, and its direct product of landslides and rock-falls is small. It is, however, indirectly accessory to the slipping of the overlying Burnside Mudstone, where, as at Puketeraki (Fig. 3), it has been reduced to a few inches thickness, and, when charged with water, acts “like ball-bearings” beneath the layer of mudstone, the more so as the layer below it is also an impervious mudstone. But even without this aid, the Burnside Mudstone may move easily in semi-plastic fashion on gentle slopes wherever it is exposed, as is seen very clearly at and northwards of Puketeraki Railway Station. All the survey-pegs controlling the original setting out of the railway on the Burnside and Abbotsford Mudstone between here and the Waikouaiti River were found to have been displaced when the line was resurveyed about fifty years later. Caversham Sandstone. The Caversham Sandstone is more massive and jointed than the older sandstones, and water soaking down to its base so lubricates the surface of the underlying mudstone that it slides easily thereon, even on very gentle slopes and on both a small or a large scale. Where a sloping surface of sandstone is overlain by jointed lava, water percolating through the lava dissolves the calcareous cement from the sandstone converting it into a mass of incoherent grains, on which the overlying residual blocks and clay may slide easily. That such removal of cementing material is, in part, the explanation of landslides which have displaced the railway-track between Seacliff and Puketeraki (Fig. 3† Where possible the estimated amounts of track-displacement are shown.) was recognised by G. J. Williams (1929). Where there is a constant “dribble” of residual clay down the face of a steep sandstone slope beneath a lava-cover, the turf thereon may become so coherent that the removal of the cementing material from the underlying sandstone results in the formation of “terracettes.”

Abbotsford Mudstone Burnside Mudstone Coversham Sandstone Fig. 3.—Geological Map and true-scale section of the Puketeraki area illustrating: (a) Weathering detritus slipping on calcareous sandstone; (b) Burnside Mudstone slipping on a thin bed of Loose Sand above Abbotsford Mudstone; (c) Large slumping strips of calcarcous Caversham Sandstone, pierced by railway tunnel, sliding on Burnside Mudstone.

Volcanic Rocks. Debris-avalanches without number occur from time to time on the steep slopes of the volcanic rocks. The exceptionally “greasy” character and depth of the residual clay on phonolite renders it especially prone to slipping, but basaltic tuff and agglomerate being permeable and thus deeply weathered yield readily on deforested slopes. Instances of the blocking of roadways by slips from such fragmental rocks occurred adjacent to Papanui Inlet in 1927 and 1937. In each case the surface slope was about 17° (30%) and the depth of detritus moved (height of slump-scarp) 10–15 ft. Cornish Head. Landslides, in which several geological formations are concerned, are to be seen on Cornish Head (or Waikouaiti Head) at the north-east corner of the region mapped. Here, basalt rests on a tuffaceous mud-flow conglomerate below which is Caversham Sandstone resting on Burnside Mudstone, the upper surface of which is nearly horizontal and about forty feet above sea-level. As a result of the yielding of the mudstone to the weight of the overlying rocks, the headland has been broken into a series of backward-rotated slumped blocks, which are still in very slow movement as may be recognised on revisiting the region after an interval of some years. A clearly exposed “surface of stability” (Sharpe, p. 166) is to be seen, whereon the sandstone (from which its basal glauconitic layer has been torn) slides over the Burnside Mudstone obliquely to its stratification down a slope inclined seaward at 15° to the horizontal. Within the headland this surface of stability is bent up into a nearly vertical position corresponding with the exposed high slump-scarp cutting across the sandstone and overlying formations. The curve is thus similar to the form deduced by Macdonald (1915) and Becker (1916) for such “stability surfaces.” Puketeraki Railway Tunnel. Some idea of the rate of these slumping movements may be obtained from the records of displacement at the Puketeraki railway tunnel, which through the courtesy of the Otago District Engineer for Railways, the writer was permitted to study (See Fig. 3). The tunnel was cut through a northward-facing bluff of Caversham Sandstone, and enters into what is actually the face of a slump-scarp a short distance above the underlying mudstone. The tunnel is 516 ft. long and rises southwards with a slope of 1 in 66, crossing several major as well as minor surfaces of slump-movement. The displacement along these was soon made obvious by the disruption of the tunnel-lining, and necessitated repeated repair, regarding and realignment. Comparison made in 1932 between the elevations of points taken at 10 ft. intervals along the axis of the tunnel (crown of the tunnel-arch) and those determined at the time of its construction in 1878 (the northern portal being used as the datum-point), showed that there were three main slump-faults between which strips, respectively 50, 150 and 100 ft. wide, subsided with a backward rotation at an average rate of 0° 1′ 7”, 0° 0′ 24”, and 0° 0′ 5·8” respectively per year during the 54-year interval. The

greatest differential movement (12 inches) occurred between the middle and inner strip, and this slump-fault is continued upward from the tunnel into the small slump-scarp about 18 inches high on the top of the bluff over the middle of the tunnel. This scarp was thus scarcely in evidence when the construction of the tunnel was commenced. In order to obtain more nearly the absolute measurements of the displacement of the tunnel-axis a line of sight from a datum-point fixed in the station yard in 1928, and controlled by azimuth-readings to distant trigonometrical stations, was taken as the base-line for measurements of the position of seven points at various intervals along the crown of the tunnel-arch, and repeated measurements were made at intervals between 1928 and 1936. Taking those made before June, 1934, it may be deduced that the annual rate of backward rotation of the middle and inner slumping strips had then increased to 0° 0′ 42″ and 0° 8·2′ respectively, and that the average rate of subsidence at six of the points observed by reference to the seventh point near the north portal (most nearly stable* This apparent stability resulted from the approximate equality of the upward movement of the outer edge of the northern rotating slump-strip, and the downward movement of the strip as a whole. during the 54-year interval) had increased by 80%, these increases probably resulting from the increased vibration caused by heavier and more frequent trains. The determination of the eastward as well as the downward component of the displacement showed that each slumping strip had a double movement of rotation, the backward rotation about an axis parallel to the face of the scarp, and a seaward slipping also with backward rotation about a line nearly perpendicular thereto; for the resultant direction of lateral slipping is inclined more steeply than the bedding plane of the Burnside Mudstone on which the movement takes place. The average rate of monthly slipping (varying between 0·08 and 0·29 inches) during the several intervals between the successive measurements accords closely with the value of 0·0138 R2.70 inches, where R is the mean of the average monthly rainfall during these intervals at the two official recording stations nearest to Puketeraki (respectively 12 and 16 miles to the N.N.E. and S.S.W.). If R should approximate to the average monthly rainfall at Puketeraki, the above relation may indicate at least qualitatively the dependence of the rate of slipping on the amount of water percolating down the fissures in the Caversham Sandstone to lubricate the sloping surface of the Burnside Mudstone on which it rests. Between July, 1934, and September, 1936, when the measurements were discontinued, the deviation-cutting and embankment were being built, and the use of the tunnel was abandoned. The displacements of the observed points greatly diminished, and after April, 1935, appeared to cease, in part, no doubt, because of the excellent drainage of the fissures afforded by the new cutting, but possibly also because the extra load imposed by the weight of the 87,000 cubic yards of sandstone, which makes the new embankment

on the mudstone in the vicinity of the datum point for observations, may have so increased its tendency to creep seaward as to have reduced below the limits of observational error the movement of the tunnel in reference to that of this datum point. Puketeraki-Seacliff-Omimi. The southward-tending hillslopes between Puketeraki, Seacliff, and Omimi (Fig. 4) afford many evidences of landslides and soil-creep. They were formerly covered by volcanic rocks resting either on the Caversham Sandstone or on the inlier of Burnside Mudstone which had been exposed by “Miocene” peneplanation. Most of the lava has now been removed, but remnants thereof remain capping the sandstone which around the heads of Seacliff and Omimi Creeks Fig. 4.—Block-diagram of the Seacliff-Omimi District.

has broken into a series of lunate backward-rotating slumping strips, moving forward on the surface of the mudstone which slopes seaward at an inclination of about 6°, and leaving behind arcuate slump-scarps. In the middle portion of the valley of Seacliff Creek the residual basaltic material and clay derived from both it and from the mudstone has been heaped into moraine-like masses creeping down a slope of 3°–4°. A long building erected in 1882 stretched from stable ground on to the apparently stable but actually creeping mass, the movement of which was soon made obvious by the opening of transverse fissures in the building. Before the last of the displaced portions of the building was demolished in 1937 it had been moved relatively to the stable portion a distance of eight feet at an average annual rate of at first 1·6 and later 2 inches per year. The displacement of the adjacent railway line on the full “flood” of the creeping material has been nearly twice as rapid, and calls for frequent re-alignment.* Note added November 9, 1940: While this was passing through the press there appeared an article by W. S. Putnam and R. P. Sharp (Landslides and Earthflows near Ventura, Southern California, Geographical Review, vol. 30, No. 4, October, 1940, pp. 591–600) which illustrates and describes in quantitative detail, with reference to destructive effects, landslides which, though on a larger scale, are essentially similar to those here discussed. Though the middle portion of the watershed of Omimi Creek was cut in mudstone, its lower portion discharged through a pitching syncline of lava (kulaite). The middle portion of the valley-floor became reduced to an extremely gentle slope leading to the upper edge of the lava-sheet over which the creek discharged. But as soon as the recession of the coastal cliffs enabled the creek bottom to be lowered below the lava into the underlying mudstone, its valley was quickly rejuvenated, sank into the mudstone throughout most of the length of the lava-covered syncline, and is now rapidly cutting small gorges headward into the previously senile area. As a result, almost the whole of the land-surface about the lower and lower middle portion of the valley of Omimi Creek has become mobilised, and the effects of slumping and creeping on pastureland, roadway or railway have been almost continuous during the period of European occupation. (See Fig. 4.) Swampy Hill. Evidence of more prolonged surface-creeping resulting from the semi-plastic yielding of mudstone may be seen near Burns and Williams Creeks draining the north-eastern face of Swampy Hill at the head of the Waitati watershed. The region was on a tectonic hinge-line during the crust-warping immediately prior to the “Miocene” peneplanation, was after this covered by mudflow deposits derived from the explosive ejectamenta erupted at the close of the first eruptive phase, and was then covered by the basaltic flows of the second eruptive phase, and the basalts of the third eruptive phase and, as these were tilted eastward, their south-eastern depressed portions were covered by the phonolite erupted during the later part of that phase. This phonolite rose in part through the fault-fissures traversing the eastern flanks of

Fig. 5.—Geological Map of north-eastern face of Swampy Hill, with true-scale section and inset detail.

Swampy Hill (Fig. 5). Cedar Knob, an intrusive mass of phonolite, may have been one of these feeding masses. The present drainage seems to have arisen as a series of consequent streams at an early stage of the development of the Swampy Hill anticlinal ridge, and to have been carried up with it, so that some of the streams still show in their head-waters features indicative of greater maturity than existing in their steep middle course (e.g., Nichol's, Morrison's and Ferguson's Creeks). What was originally the highest portion of Williams' Creek also had a fairly matured valley on the north-eastern flank of Swampy Hill, but has since had a curious history. With the rise of Swampy Hill as a north-easterly pitching anticlinal ridge, the heavy cover of lava on the north-eastern flank so pressed on the underlying semi-plastic mudstones that when the Waitati Stream cut through the lavas and into this weak foundation, it yielded, bulging the outer slopes, behind which it broke into a series of lunate backward-rotating slump-strips, separated by slump-faults causing arcuate slump-scarps extending a mile and a half backwards from the lower edge of the lava-cover almost to the summit of Swampy Hill. Though the highest portion of Williams' Creek was at first deepened because of the formation of a slump-scarp across it, its upper middle portion became alluviated because of its backward rotation, while its lower middle portion leading down across the lower slump-strips was deeply entrenched in the flanks of the still rising anticline, and was apparently superposed on and into the intrusive phonolite of Cedar Knob. This mass seems to have acted as a prop supporting the southern margin of the major slumping mass, for the next movement was an extension of the slumping north-eastward, enlarging the upper portion of Burns' Creek into an irregular fault-angle depression, and drawing into it the headwaters of Williams' Creek, which, being thus locally rejuvenated, have cut a V-shaped gorge fifty feet deep into their formerly matured valley. As will be seen from the true-scale section in Fig. 5, the movement of the westward (inner) margin of the major slump-strip has continued steadily until the slightly dissected eastward-facing scarp of Swampy Hill, exposed by this subsidence, is now nearly five hundred feet high, and the backward rotation of the slumped block itself has amounted to about ten degrees. The northern margin of the major slumping sheet was further broken into a group of narrower backward-rotated lunate strips giving rise to a series of gently sloping or alluviated reaches or peat-swamps drained by narrow gorges cut through the successive slump-scarps to a depth (in one case) of over a hundred feet. The study of the most noteworthy of these fault-bounded depressions in order to determine its suitability for a municipal storage reservoir was the occasion for the writer's detailed examination of this area in 1927. (See inset detail Fig. 5.) A number of pits in the alluvium showed that even the sub-recent peat had been displaced about four feet by slump-faults concealed beneath the younger alluvium. Two hundred yards east of the area of the inset map two modern slump-faults have exposed vertical scarps of uneroded soil several feet high. Traced north across Burns' Creek one of

Fig. a.—Slump-scarp and head of slip, north-east of Saddle Hill Fig. b.—Toe of same slip. (Otago Daily Times photo.) Fig. c.—Slump-scarp near head of Whare Creek. (Dr. C. R. Laws photo.)

these is continued into an arcuate trough, now 8–16 feet deep, probably a tension-fissure filled by lateral slipping of soil. A much more striking example of the opening of such a tension-fissure occurs 200 yards further east where a massive fracture-bounded block of lavas over a hundred feet thick and fifty acres in extent* Owing to the density of vegetation on this sliding block the boundary of the dolerite thereon is merely sketched in Fig. 5. has moved bodily eastward, leaving behind a cliff-bounded trough some 20–40 yards wide. Here and there a rocky tree-crowned fragment, torn from and left behind the forward sliding-block, rises above the low, tangled vegetation growing on the newly exposed sliding surface, the inclination of which, though small, cannot be exactly determined. The lower or eastern margin of the lava-cover, or of the tufaceous mud-flow beneath it, rests on Caversham Sandstone. A line of pipes designed to divert water from several streams into a reservoir beyond the area mapped in Fig. 5 follows the margin of the volcanic rocks, and is continuously moving. At one point on this pipe-line, it has been proved that displacement of 16 feet downwards and 26 feet laterally occurred during the last twenty years as the latest phase of the long continued movements in this region. In general, the north-eastern face of Swampy Hill within the area bounded by the well-marked tectonic faults exemplifies in more elaborate detail the features described by Russell (1900) where the Columbia lavas of western U.S.A. in sheets 400–500 feet thick (or more) resting on clay, sand or volcanic lapilli have been broken into a series of slowly slumping blocks separated by steep escarpments. Flagstaff. A final area may be mentioned wherein the evidence that the scarp to be described results from the semi-plastic yielding of Abbotsford Mudstone, rather than from recent rejuvenation of a formerly active fault, may not be quite conclusive. The head-waters of Whare Creek, draining the western side of the Flagstaff ridge, emerge from the valleys cut out of lavas on to the gently sloping surface of marine sediments. This surface was deeply covered with lava-blocks and clays which extended downwards to the Silverstream and were in part residual flood-plain deposits on the gentle slope. With the continuation of the rise of the Flagstaff-Swampy anticlinal ridge and the associated depression of the Taieri syncline, the concomittant, deeper incision of the Silverstream permitted the rejuvenation of its tributary streams, Whare Creek and those adjacent thereto, and the consequent dissection of the wide-spread sheet of residual and drift material, a rejuvenation which has extended back to within a mile of the main ridge. The upper portion of Whare Creek is, however, crossed by a scarp which extends for nearly a mile parallel to the ridge, raising the almost undissected mudstone with its cover of residual and drift material 100–150 feet above the remnants of the deeply dissected residual and drift material further west. Though the rock exposed in the scarp beneath this covering sheet consists of the very soft Abbotsford Mudstone, the formation of the scarp was so recent that it

Fig. 6.—Geological Map of the headwaters of Whare Creek, west of the Flagstaff-Swampy ridge, with true-scale section.

has not been dissected save by the antecedent valley of Whare Creek, which has cut a gorge through it. (See Plate 34, Fig. C.) The fact that this scarp is confined to the region occupied by the mudstone, and does not extend southwards into other formations, suggests that the semi-plastic yielding of the Abbotsford Mudstone in the lower portion of the slopes downwards towards Silverstream with consequent formation of a slump-scarp rather than the continuation with faulting of the anticlinal rise of the Flagstaff-Swampy ridge may have been the origin of this peculiar scarp. It is appropriate to acknowledge that the writer's detailed mapping of this feature was aided by the earlier work of Hurst (1928). In conclusion, acknowledgments are due to the former and present Dunedin City Engineers, Mr J. G. Alexander and Mr G. S. Scoular, and their staffs, to the City Electrical Engineer, the Otago District Engineer of Public Works, the District Engineer of Railways, to the Director of the Dominion Meteorological Bureau for the opportunity to acquire some of the information utilised in this paper, to the staff photographer of the Otago Daily Times, and to Dr. C. R. Laws for the photographs on Plate 34. List Of Papers Cited. Becker, G. F., 1916. Mechanics of the Panama Canal Slides, U.S. Geol. Survey, prof. Paper, no. 98, pp. 253–261. Benson, W. N., 1934. The Geology of the Dunedin District, N.Z., Abs. Proc. Geo. Soc. London, Dec. 14, 1934. — 1935. Some Land-forms in Southern New Zealand, Australian Geographer, vol. 2, no. 7, pp. 3–22. Finlay, H. J., and Marwick, J., 1940. The Divisions of the Upper Cretaceous and Tertiary in New Zealand, Trans. Roy. Soc. N.Z., vol. 70, pp. 79–135. Grange, L. I., 1923. An Account of the Geology of Green Island Coalfield, Trans. N.Z. Inst., vol. 53, pp. 157–174. Hurst, J. A., 1928. Geology and Petrology of the Silverstream and Upper Waitati Valleys, Unpublished Thesis, Univ. of Otago. Macdonald, D. F., 1915. Some Engineering Problems of the Panama Canal and their Relations to Geology and Topography, U.S. Bureau of Mines Bull., 86. Marshall, P., 1906. The Geology of Dunedin (New Zealand), Quart. Journ. Geol. Soc., vol. 62, pp. 381–423. — 1914. The Sequence of Lavas at the North Head, Otago Harbour, Quart. Journ. Geol. Soc., vol. 70, pp. 382–406. Ongley, M., 1939. The Geology of the Kaitangata-Green Island Subdivision. Geol. Surv. Branch, N.Z. Dept. Sci. Ind. Research Bull., no. 38. Park, J., 1910. The Geology of New Zealand, Whitcombe and Tombs, pp. 198–200. — 1924. Evidences of Pleistocene Glaciation at Abbotsford, near Dunedin, Trans. N.Z. Inst., vol. 55, pp. 599–600. Russell, C., 1900. Preliminary Paper on the Cascade Mts. in Northern Washington, U.S. Geol. Surv. 20th Ann. Rept., ii, pp. 193–200. Sharpe, C. F. S., 1938. Landslides and Related Phenomena, Columbia University Press. Williams, G. J., 1929. Geology of the Seacliff District, Unpublished Thesis. Univ. of Otago.