Ch. XV.] UPPER EOCENE STEATA. 203 



gravelly and sandy beds of Lake Superior no pebbles of modern volcanic 

 rocks can be included, since there are none of these at present in the 

 district. If igneous action should break out in that country, and pro- 

 duce lava, scoriae, and thermal springs, the deposition of gravel, sand, 

 and marl might still continue as before ; but, in addition, there would 

 then be an intermixture of volcanic gravel and tuff, and of rocks precip- 

 itated from the waters of mineral springs. 



Although the freshwater strata of the Limagne approach generally to 

 a horizontal position, the proofs of local disturbance are sufficiently 

 numerous and violent to allow us to suppose great changes of level since 

 the lacustrine period. We are unable to assign a northern barrier to the 

 ancient lake, although we can still trace its limits to the east, west, and 

 south, where they were formed of bold granite eminences. For need 

 we be surprised at our inability to restore entirely the physical geography 

 of the country after so great a series of volcanic eruptions ; for it is by 

 no means improbable that one part of it, the southern, for example, may 

 have been moved upwards bodily, while others remained at rest, or even 

 suffered a movement of depression. 



Whether all the freshwater formations of the Limagne d'Auvergue 

 belong to one period, I cannot pretend to decide, as large masses both of 

 the arenaceous and marly groups are often devoid of fossils. Some of 

 the oldest or lowest sands and marls may very probably be of Middle 

 Eocene date. Much light has been thrown on the mammiferous fauna by 

 the labors of MM. Bravard and Croizet, and by those of M. Pomel. The 

 last-mentioned naturahst has pointed out the specific distinction of all, or 

 nearly all, the species of mammalia from those of the gypseous series near 

 Paris, although many of the forms are analogous to those of Eocene 

 quadrupeds. The Cainotherium^ for example, is not far removed from 

 the Anoplofkerium, and is, according to Waterhouse, the same as the 

 genus Microtheriiim of the Germans. There are two species of marsujDial 

 animals allied to Didelphys, a genus also found in the Paris gypsum, and 

 several forms of ruminants of extinct genera, such as Amphitragulas elegans 

 of Pomel, which has been identified with a Rhenish species from Weisse- 

 nau near Mayence, called by Kaup JDorcatherium nanum ; other associ- 

 ated fossils, e. g., Micr other ium Reuggeri, and a small rodent, Titanomys^ 

 are also specifically the same with mammalia of the Mayence basin. The 

 Hycenodon, a remarkable carnivorous genus, is represented by more than 

 one species, and the oldest representative of the genus Machairodus has 

 been discovered in these beds in Auvergne. The first of these, Hycenodon, 

 also occurs in the English Middle-Eocene marls of Hordwell clifi', Hamp- 

 shire, considerably below the level of the Bembridge limestone, with 

 Paleotheria. Upon the whole it is clear that a large portion of the 

 Limagne rocks have been correctly referred by French geologists to their 

 Middle Tertiary, and to that part of it which is called Upper Eocene 

 in this work. 



Cantal. — A freshwater formation, of about the same age and very 

 analogous to that of Auvergne, is situated in the department of Haute 



