154 B. WILLIS—DRIFT PHENOMENA OF PUGET SOUND. 
sisting of the hollows, which are inlets, and the plateaus; and (2) minor 
features of glacial and aqueous origin. 
To account for the peculiar physiographic type of Puget sound, two 
hypotheses are suggested—the erosion hypothesis and the construction 
hypothesis. 
The erosion hypothesis: As a result of general glaciation, the Puget 
Sound basin was filled with drift up to an approximately even plain. 
With retreat of the ice after one or several epochs of glaciation, streams 
were established, which cut their channels in the drift down to a level 
represented by the depths of the Sound. In consequence of depression 
of the region, the valleys were submerged and the present inlets estab- 
lished. 
The construction hypothesis: Glaciation of the Puget Sound basin was 
confluent, not indigenous. The distinction is important in its bearing 
on the effects of advance and retreat of the ice. Where glaciation is 
indigenous the accumulation begins at high altitudes, and the summits 
are covered with névé before the lower valleys are filled with ice. In the 
epoch of lessening glaciation the summits are last bared ; but where gla- 
ciers become confluent in a region in which they do not originate, the 
valleys are first occupied and filled, and the hills are subsequently buried. 
When the ice-mass diminishes, the hilltops are laid bare while yet the 
valleys are the seats of glaciers. Glaciers first occupied the valleys of the 
Sound basin, but becoming dammed rose till the ice overtopped divides. 
During the episode of retreat hilltops appeared as nunataks, and this con- 
dition extended to divides. In ponded waters about these nunataks 
accumulated stratified drift. Stagnant ice lingered in the valleys to the 
latest stages of retreat, and, melting, perpetuated their relations to the 
heights. By repetitions of this process the antecedent divides were built 
out to plateau forms, and the axes of the original valleys were maintained 
as relatively shallow hollows, which are thus the casts of glacial tongues. 
Oscillations of level and the effects of erosion played a subordinate part. 
The distribution and depths of the hollows apparently do not corre- 
spond to those relations which should exist if the inlets were drowned. 
Consideration of the post-Pleistocene work of streams shows that the 
effects of corrasion are very slight as compared with the magnitude of 
the hollows. 
The White River alluvial cone is so adjusted to sealevel as to indicate 
that oscillations of regional extent have been very moderate during post- 
Glacial time. 
