The average composition of the igneous rocks is a point of considerable interest in its relation to the problem of their differentiation, and several attempts have been made to solve it by the collation of analyses. Thus, Mr. Clarke estimated the American rocks to average 59.77 per cent. of silica (which may be taken as representative), and Mr. Harker came to very similar conclusions as regards the British rocks, obtaining 58.46 as his figure. The process followed was to add up the results of all the obtainable analyses and take their mean. If each class of rock analysed occupied the same average amount of space—if, for example, the basic intrusions were approximately equal in bulk to the acid ones—such a process would give results of considerable value. As it is, however, very little consideration will show that unless due weight is attached to the relative abundance of the different classes, the results will be very far removed from the truth. Even in Britain, where the development of igneous rocks is comparatively insignificant compared to the sedimentary ones, there are quite enough exposures of the different types to demonstrate this fact. If a geologically coloured map be examined, and the nature of the various patches of igneous rock be enquired into, the immense preponderance of granite becomes obvious, even though the basaltic lavas make a great show on account of their horizontal extension. In fact, the Dartmoor granite mass, if it be assumed to extend to a depth of only one mile, would probably suffice to weigh down the scale against all the other non-granitic igneous rocks combined.