382 NATURAL SCIENCE. j UNE , 



times hollowed into lofty domes. It would, therefore, be very unlikely 

 for the roof of a cavern to fall in all at once. 



But supposing it fell in piece by piece, how is it that the roofs of 

 those which are now canons should have disappeared so completely, 

 while the roofs of the many other caverns which do exist in the lime- 

 stone have hardly fallen in at all ? It is true there are swallow-holes 

 and avens, which are holes in the roofs of caverns lying 200 or 300 feet 

 below the surface ; but if this was the way in which open canons 

 began to be formed, one would expect to find them in all stages of 

 formation— some lengths fallen in and some still roofed over — so that 

 natural tunnels and arches should be frequent along some of the 

 canons. But this is not the case ; no such tunnel or arch occurs in 

 the course of any one of the rivers. The canons are everywhere 

 clean-cut, open ravines. 



Again, the main rivers do not take their rise within the causses 

 district ; it is true that one or two streams spring out of caverns 

 within that district, and that all are largely fed by such springs, but 

 the three chief rivers, the Tarn, the Joute, and the Doubies, rise 

 outside the limestone area and cut right through it, separating one 

 causse or plateau from another. The canons of these streams are, 

 therefore, merely the continuations of the valleys which they have 

 made for themselves in the gneiss and granite. 



Finally, there is no necessity for imagining the previous 

 existence of a cavern. The streams running off the older rocks would 

 be quite capable of excavating these gorges by ordinary mechanical 

 means if they were first set running over the surface of the limestone 

 plain from a watershed at a higher level. That they did once run 

 over this surface and carved out the channel which has become a 

 canon, few geologists would doubt. The causses most certainly did 

 not always exist as an isolated limestone district ; they are part of a 

 great formation which extended over the whole of southern France, 

 from the Bay of Biscay to the Maritime Alps, stretching completely 

 over the heights of the Cevennes, and spreading northward for some 

 distance up the slopes of the Central Highlands. Much of the lime- 

 stone formation may have been removed before the rivers came into 

 existence, but when the rivers began to cut their channels, the surface 

 of the limestone plateau doubtless extended northward and eastward 

 till it met the rising plane of the older rocks which form the Cevennes 

 and the Central Highlands. At the present time, the highest parts of 

 the causses hardly reach 4,000 feet above the sea, while the Cevennes 

 rise in places to over 5,000 feet. Consequently, if the canons did not 

 exist, and if the space between the eastern escarpment of the causses 

 and the peaks of the Cevennes were bridged, or rather filled up, with 

 limestone, the watersheds would remain where they now are, and the 

 streams would run over the surface of the limestone plateau. 



So also with the rivers Lot, Dordogne, and Vesere, which traverse 

 the causses of Guyenne. They rise in the highlands of Correze and 



