390 Transactions. 



owino- to the erosive action of the sea : this, too, tends to maintain the 

 appearance of the interior walls of the crater as having been causedjiby 

 explosion. The steep-graded valleys carry their grades down below water- 

 level till they merge into the floor of the harbour, emphasizing the unity 

 of origin of the subaerial and submarine portions. In the western part of 

 the basin the interior slopes are occupied by long tongue-like peninsulas, 

 nearly two miles in length from the point where they are joined on to the 

 walls of the crater-ring. The valley-floors between them are very flat 

 and grade insensibly in the floor of the harbour, and they show the form 

 they should possess were the harbour a river-valley and its head enlarged 

 by water action with just that amount of modification effected by the in- 

 filling of sediment into the submerged portion. The sources of the streams 

 which drain inwards are distributed along the inner crest of the crater- 

 walls, but the streams join the main valley at points in close proximity 

 to each other. Where the valleys are cut in the lava-flows dipping out- 

 wards the valleys have just the same features as those on the northern side 

 of the harbour. 



This locality also illustrates the breaking-down of the wall of the crater 

 by the combined action of streams attacking it from the inside and the 

 outside slopes. The streams responsible for the formation of Gebbie's and 

 McQueen's Valleys, as well as those occupying the valley within the harbour 

 area where Teddington lies, have completely removed the andesites, deeply 

 dissected the rhyohtes, and exposed the underlying greywackes over a con- 

 siderable extent of country. The forms of the two external valleys are so 

 strongly reminiscent of the lower reaches of Lyttelton Harbour that they 

 are probably due to a common cause. 



On following the harbour round beyond this gap past a remnant of the 

 old crater-ring the interior slopes are modified from the true caldera form 

 by the gentle slopes reaching up to Mount Herbert accordant with the flow 

 of the lava-streams from that more recent centre of eruption ; but even 

 here, where small exposures of the underlying beds can be seen, they suggest 

 a land surface analogous to that on the northern side of the harbour. To 

 the east of these flows, and eroded along their margin, lies another drowned 

 valley, called Purau Bay ; but from this on to the southern head the cha- 

 racters are those which we should expect from the erosion of a stream flowing 

 across the strike of gently inclined beds dipping uniformly in one direction. 

 This portion of the harbour thus has a trench-Like cross-section, a feature 

 made more pronounced by the attacks of the sea on lava-flows where the 

 jointing is almost vertical. The same feature is seen in the lower portions 

 of Gebbie's and McQueen's Valleys, except that the influence of the sea in 

 forming a wave-cut cliff is not so prominent. 



A feature to be considered in arriving at a conclusion as to the origin 

 of the cavity is that the valley-heads almost entirely occupy lower parts 

 of the crater-ring and do not lead up to the highest peaks. Although this 

 suggests that the valleys have had a portion of their upper courses beheaded 

 by explosion or by collapse of the cone, such as happened in the case of 

 Mount Mazama,* in the United States, it is quite possible to explain the 

 phenomenon as an effect of normal stream erosion on a cone which has not 

 sufEered such a catastrophe. The approximation of valley-heads on the 



* J. S. Dillox and H. B. Patton. Geology and Petrography of the Crater Lake, 

 National Park, U.S. Geol. Surv. Prof. Paper No. 3, 1902. 



