and gn the Origin of iM Valleys (\fAuvergiie. 223 



stream of lava, which, blqck^ up the course of the riYer,,,4nd 

 compelled it to flow in quite a diflerent direction. 



"The baffled waters of the Sioule,*" says Mr Scrope, "here 

 as at Pont Gibaud, obstructed by the rocky dike thrown across 

 their channel, must have given rise to a lake by their stagna- 

 tion, and would probably have ended, as in the other instance, 

 by wearing away a passage parallel to their former one, had not 

 the hill forming their western bank, not in this instance com- 

 posed of granite, but of a soft alluvial tufa, yielded, at some 

 distance up the stream, to the excessive pressure of the dammed 

 up waters. An immense excavation, still subsisting, was broken 

 across this hill, through which the lake emptied itself into the 

 bed of the Monges at no great distance, and through which the 

 Sioule still joins this latter stream, about three miles above their 

 former confluence.'"" 



Now, that the accumulated waters of the Sioule, when arrest- 

 ed by this barrier, should have undermined, and thereby forced 

 a passage through a rock of so very soft and yielding a nature 

 as that of the argillaceous hill alluded to, is by no means sur- 

 prising ; neither need we doubt that the river may have worn a 

 channel through the subjacent gneiss to the depth of 12 feet 

 since the period at which its direction was changed. 



But when the same conclusion is extended, as some geolo- 

 gists seem disposed to do, to the gorge of gneiss or the valley 

 excavated in the plateau of ancient basalt, through which the 

 river afterwards flows, I would ask, what proof have we, that, 

 in these latter cases^ the valley was not of anterior date, and 

 whether, if it had not been already in existence, the river ought 

 not to have surmounted the impediment opposed by the lava 

 current of Come, rather than to have worked its way through 

 the more elevated and equally unyielding barrier to its left? 



The most, therefore, we ^re entitled to conclude from the 

 phenomena exhibited near Pont Gibaud with respect to die 

 action of rivers upon the compacter kinds of rocks, is, that one 

 recent lava-stream, namely that portion of the coulee from the 

 Puy de Come which is seen near the town, has, since an epoch 

 more remote than that of the earliest records of the country, 

 been worn by the action of the stream to the depth of about 

 50 feet. 



