Excavation of' Valleys. 19 



200 and 300 feet in thickness had filled an ancient excavation, 

 which in this instance had been effected through gneiss capped 

 with basalt, as in Mount Perrier, through tertiary marls cap- 

 ped with a similar basaltic plateau. The exact depth of the al- 

 luvium removed cannot be precisely stated, but the height of 

 some parts, which ascend to the foot of the basaltic plateau, is 

 above 250 feet, and one of the vertical cliffs composed of allu- 

 vium, of depth unknown, is, as we before stated, about 140 

 feet high. This new excavation forms a sinuous valley about * 

 two and a half miles in length, sometimes expanding so widely, 

 that the Sioule runs in sweeping curves without reaching either 

 bank, in a valley which may be compared, in point of size, 

 to a large majority of those in our tertiary and secondary for- 

 mations, where it is equally difficult to conceive that such pro- 

 di^ous effects can have resulted from causes apparently so 

 feeble. The mode of operation whereby so narrow a stream 

 has eaten out so wide a valley^ may still, however, be detected. 

 Here and there are the remains of ancient land-falls, the ruins 

 of alluvial cliffs formerly undermined. These are now covered 

 with vegetation, and constitute under-cliffs or taluses. By such 

 obstructions, the stream has been deflected from time to time 

 against new points of attack, where it is now undermining steep 

 precipices of clay, and at the top of these are large fissures 

 threatening future land-slips. The solid gneiss moreover, now 

 occasionally laid bare at the bottom of the valley, causes sudden 

 flexures in the stream. 



In proportion as the distance between the opposite banks in- 

 creases, (and it often fails in its present bendings to reach either 

 side), the shifting of the course of the Sioule is becoming less 

 frequent. As the rate of waste therefore diminishes with the 

 width of the valley, the mode of its formation may become 

 at a remote epoch an obscure and difficult problem. We refer 

 the operation in question to the Sioule, because no extraordi- 

 nary inundation can have co-operated. For the " cheire" of 

 Come, which flowed down before the formation of the new val- 

 ley had begun, occupies so low a level, that had the waters risen 



• See DumaresL's map of Auvergne, the earliest work of merit on the phy- 

 sical structure of this district, and one of the first successful efforts at com- 

 bining geographical and geol(^cal information. 



