1^94-1 271 [Dofl-e. 
liko expansions ^villi detritus would tlien commence, every tributary as 
well as the main river carrying down its contribution, wliich, wherever 
these river expansions were of considerable length, Avonld be thrown doAvii 
and form deltas and terraces. Gradually the smaller and narrower ones, 
more especially in their upper reaches, would become partially or wholly 
filled with detrital materials except in the passages of the channels of the 
river. As the embankment of the glacial drift referred to, holding in 
the lakes, became eroded and the river's passage through them deepened 
coincidentally with a lowering of the waters, these partially filled lake- 
like expansions would assume the river-like form, and eventually become 
a part thereof , and terraces appear along tlie sides. Finally, by erosion 
and the deepening of the river's channel, it would reach a comparatively 
even flood plain, and as it became more and more confined to a narrow 
channel, other and lower terraces would appear along its border from 
time to time. In this way might the terrace have been formed Avithout 
supposing the existence of enormous floods and rapidly dissolving 
glaciers. Indeed the vicAV here given does not require that the rivers be 
very much greater in volume tlian they noAV are."' 
I propose a modilicatioii of the plan given above for the 
formation of alhivial terraces in glaciated valleys, that seems to 
me to accord better with the facts as we find them in such large 
rivers as the Connecticut, in wliich no system of lakes could well 
account for the formation of all the terraces, though it may 
account for some of them. The Connecticut River occupies such 
a well-marked valley that it must have been the drainage channel 
of a large amount of the water caused by the melting of tlie great 
glacier that overlay some j>ortion of its valley. According to the 
plan that I have suggested, part, if not all, of the waste in the 
terraces must have Vjeen laid down during the immediate j^resence 
of the ice. Afterwards a decrensed volume and a rising land will 
account for the rest of the work done in postglacial times. In 
other words, the upper terrace plain is due to a glacial accident 
in the river's history, and the upper escarpment Avas formed as the 
river cut down tOAvard base-level after the land rose AAdien relieved 
from the weight of the ice. The whole terrace is therefore 
complex in its origin. The later terraces formed as the river 
sank its channel deeper into the glacial waste, each terrace plain 
representing the temporary level of the stream and each escarp- 
ment showing intermittently rising land. Such an hypothesis is 
framed in accordance with the plan for the development of a 
I Op. cit., p. CI et seq. 
