FRESH-WATER STREAMS. 643 



walls of rock. The island of Tahiti affords grand exhibitions of such 

 gorges, knife-edged ridges, and mountain amphitheatres. 1 



A model of this system of erosion is often admirably worked out in 

 the earthy slopes along a road-side, — the little rill having its cascade- 

 head, then its torrent-channel, and, below, its flat alluvial plain, inter- 

 sected by the little winding water-channel ; some of the ridgelets worn 

 away in their upper parts, until two or more little valleys coalesce ; 

 then, at times, the head of the coalesced valleys widened into an am- 

 phitheatre, and the walls fluted into a series of alcoves and buttresses. 



Through this simple method, erosion by running water, nearly all 

 the valleys of the world have been made. 



The flood-grounds widen through encroachments on the hills along 

 the margin of the flood waters. But they may have their limits either 

 side varied by changes of level in the land, subsidences widening 

 them, and elevations narrowing them ; or by a change for a long geo- 

 logical period in the amount of water supplied by precipitation, an 

 increase widening them by new encroachments, and a decrease nar- 

 rowing them, and producing, finally, a new plane at a lower level. 



On some large rivers the flood-grounds extend more than fifty miles 

 from the low-water channel. On the Mississippi, abreast of Ten- 

 nessee, they are in some parts over fifty miles wide ; on the Amazon 

 (up which the tides go 400 miles), over a hundred miles ; on the Par- 

 aguay there are lagoons 300 miles in length. 



The nature of the rocks causes modifications in the results of erosion. 

 If there are harder beds at intervals, in the course of the stream, or 

 any impediment to even wear, the impediment, by resisting erosion, 

 becomes the head of a waterfall and precipice, whose height increases 

 rapidly from the force of the falling waters, until some other similar 

 impediment below limits the further erosion. Many waterfalls and 

 rapids are thus made in the cascade-portion of a stream; and they 

 are not absent from the river-portion. Another effect of this cause is 

 that the stream is set back for some distance above a waterfall, and 

 has in this part more or less extensive flood-plains. 



Fig. 1092 represents some remarkably slender columns of Tertiarj- sandstone, from 

 the Report of Dr. Hayden for 1873. There are here two layers harder than the rest; 

 and one has been left to make the top of the taller column, while another caps a shorter 

 series. These examples of nature's modeling are very numerous in Colorado, over 

 what has been called Monument Park. The erosion is due to the rains, or the rills they 

 produce, and the latter part to the gentler action of rain-drops, together with the action 

 of the winds and frosts. Lyell has described a remarkable example of erosion by rains, 

 of a thick deposit of reddish indurated mud, containing scattered bowlders, reallv a 



1 See the Author's Expl. Exped. Geol Rep., pp. 290, 384, and Am. J. Scl, II. ix. 48 

 and 289, 1850. 



