valley wall, which suggests that they flowed alongside the margin of the glacier as it retreated from the side of the valley, leaving a narrow strip free from ice, while the other portion of the valley floor was occupied by it. The narrow valley in which the Red Lakes lie is an illustration of this, and perhaps the lower portion of Moraine Creek behind Laing's Hill is another example. These streams, parallel to the sides of the valleys and close to them, are very deeply incised, and may therefore be associated with a narrow belt of structural weakness. A similar shelf at a lower altitude also extends along the western face of the Big Ben Range towards the Snowdon Station, and this face has features analogous to those of the bounding walls further north-west, and in addition shows practically the same alignment. It is hard to think that this correspondence is to be entirely attributable to ice erosion. If the valleys are initially due to diastrophic movements, then their cross section seems to demand the existence of parallel faults on the eastern side, one of which determined the upper clear-cut face of the mountains, and a second one which determined the edge of the ice-eroded plateau, the former perhaps continuous for the whole length of the valley, and the latter affecting only sections of it, and especially that between the Harper and the Ryton. In only one case do I know of geological evidence that the valleys are structural, and this may, after all, not be the authentic, viz., the Lake Selfe Valley; for the small occurrence of Tertiaries near the road-crossing of the Harper may owe its position to a fault running across the river, but all the same it may be due to the fault which follows up the river, as mentioned previously. Although there is no definite evidence of faulting associated with these valleys, their similarity in form and arrangement suggests that they are due to some common cause, and probably some diastrophic movement. Near the gorge, faulting does occur in the gravels in its vicinity, and there is that peculiar landscape form, known as the “Railroad,” whose formation seems explicable only on the supposition that it is due to faulting (Dobson and Speight, 1924, pp. 629–30). Both of these directions have the same orientation as the valleys. However, the date of the faulting is no doubt post-Tertiary, and some of it is post-glacial, and therefore it can have little or no bearing on the form of the valley antecedent to the glaciation, except in so far as it indicates a line or lines of structural weak-ness oriented in a direction parallel to that of the valley. It is possible that the height of the andesite capping of Round Top as compared with the height of similar beds in the gorge of the Rakaia just above the road, a difference of some 1000 ft., may be evidence of some differential movement, or it may be merely an indication of the irregularity of the surface over which the andesite was poured. If the former is the case it would support the contention that the valley was originally a graben, and that it was bounded by two fault lines, one running N.N.W. past the western