Geographical Collections. 207 



results from this disposition that tlie generality of the Pyrenean vallies' originat- 

 ing from the crest of the principal chain, are transverse with respect to it, but 

 longitudinal with respect to the transverse chains, which also originate from the 

 main crest, while the vallies dividing the mountains, and forming the lines of the 

 crest of the perpendicular chains, transverse with respect to them, are parallel 

 witli regard to the main chain, and to the vallies dividing the lateral or parallel 

 chains. Most of the transverse vallies of the Pyrenees, as previously remarked 

 by Ramond and Charpentier, present at their origin a vast basin in the form of 

 an amphitheatre, or a succession of basins, which rise one above the other, so 

 that the valley, instead of offering an uniform slope, rises as if by so many stages 

 towards the crest of the chain. This arrangement occasions the moimtain tor- 

 rents descending from above, to form cataracts or falls ; but this, as in fact the 

 diminution in size of these basins, or oules (oUa) as they are termed in the lan- 

 guage of the country, we have always observed to be connected with other geog- 

 nostic changes. 



The vallies which divide the parallel chains, are alone truly longitudinal. 

 They occur generally in the direction of the strata of the mountain, and the dif- 

 ference between the direction of the valley and that of the strata affects the forma- 

 tion of caverns. Charpentier has remarked in the Pyrenees, that the entry of 

 vallies terminating in the plain, is sometimes large, sometimes narrow ; but val- 

 lies terminating in another valley, are almost always narrowed at their origin, of 

 which fact he does not, however, appear to have seen the geognostic causes, as the 

 junction of one or more vallies generally takes place in a basin, and the extent of 

 the latter is proportioned to the number and to the size of the outlets which ter- 

 minate in them. 



The sum of the maximum of elevation of the crest marked in the peaks or cul- 

 minating points, and of the minimum marked by the transverse vallies and 

 cols ports or passages, gives the mean height of the chain of hiUs. The deter- 

 mination of the mean height of the line of the crest by the mean height of the 

 cols ports or passes, is even, according to the Baron de Humboldt, an abstract 

 idea ; and vague when there is grouping of mountains and no continuous chain. 

 And we think that a nearer approximation would be gained to the mean height 

 of the crest by a comparison of the maximum and minimum of elevation of the 

 protuberances themselves, than by an hasty calculation founded upon the heights 

 of such ridges or passes, whose depths are sometimes connected with accidents 

 posterior to the formation of the chain. Some countries of mountains, as the 

 Himmaleh, are traversed by large rivers. Chains (Dovrefeldt, &c.) are often 

 divided by profound rents, which are sometimes empty veins, (Jameson, Von 

 Buch,) while the basin of the crest may in other cases be filled with deposits of 

 the coal formation, or other secondary or more modern deposits, (Alps, Liban.) The 

 data upon which the calculation of the mean height of the crest of the Pyrenees 

 has been founded are more or less empirical ; for the country of mountains known 

 under that name, consists of a series of parallel and lateral chains, from which 

 the principal is difficult to be distinguished. When a country of mountains, as 

 the Grampians, consist not of one continuous crest, but of a series of crests, more 

 or less parallel to one another, traversing the country at angles to the line of di- 

 rection of the chain, the data of the calculation should be founded on the mean 

 height of the culminating points and minimum of crest in each chain, which 

 alone can give the mean height of the whole range, and in this case the trans- 

 verse or divergent chains should be entirely neglected. From this disposition of 

 countries of mountains, and one which appears very common, some apparent ano- 

 malies take place. Thus a chain that is divergent and transverse to the chains 

 whose crests are to give the data for the calculation of the mean height of the 

 range, may be parallel to the line of that range while at right angles to the prin- 

 cipal crest. Charpentier had remarked on the Pyrenees, that the point of de- 

 parture of transverse or lateral branches from the main, or even from lateral 

 chains, is generally marked by an increase in elevation, (culminating point,) as 

 the extremity of these branches, when not lost in the plain, is generally a peak of 



