Excavation of Valleys. 41 



of Mont Dor and the Cantal, they must be referred to a similar 

 origin ; and we agree with Mr Scrope, that the sudden rush of 

 large bodies of water down the sides of an elevated volcano, at 

 its moments of eruption, sweeping away all the loose materials 

 surrounding its crater, might be supposed to give rise to such 

 breccias ; and that the floods observed by Humboldt and others 

 to descend from the trachytic volcanoes of America, countenance 

 the hypothesis *. 



The associated strata of alluvium and breccia terminate, in 

 all the cases described by us, in abrupt escarpments towards 

 the rivers, and they must once have extended much farther, 

 but possibly not so far as to the opposite sides of the present 

 valleys. For we have seen, that where the bottom of a valley, 

 in Auvergne, or the Vivarais, has been filled with lava, the 

 rivers have usually hollowed out a new passage through the 

 granitic schists, as well as in part through the lava. So, if a 

 valley become a lake, and be filled up to a great height with 

 alluvial conglomerates and tufaceous breccias, it is highly im- 

 probable that the new excavation will coincide precisely in posi- 

 tion with the original valley, though perhaps of equal width 

 and depth ; and especially if it happen that the ancient valley 

 was bounded by tertiary marls, probably much softer than most 

 of the newly imported matters. 



For the same reason, therefore, that we so often see all that 

 remains of an ancient basaltic current, exclusively confined to 

 one and the same bank of a river, we may expect to find the 

 relics of a lacustrine formation similarly situate. 



To the hypothesis of lakes closed up by volcanic currents, it 

 will, no doubt, be objected, that we are unable to point out any 

 remnants of the supposed barriers. But how, we ask, is it pro- 

 bable, that such barriers could have survived the changes which 

 this country has subsequently experienced in consequence of 

 aqueous denudation, assisted, in all probability, by violent earth- 

 quakes.? The present valleys are much deeper than those sup- 

 posed to have been formerly converted into lakes. In the case 

 of Mont Perrier, the excavation extends to the depth of be- 

 tween 50 and 100 feet in the fresh-water strata; and at St 



* Geology of Central France, pp. 101 — 103. 



