In a paper read at the 1903 meeting of the British Association and afterwards published in this Magazine, the method of calculating the chemical constitution of the igneous rocks by the collation of analyses was criticized, and the results were compared with those based on actual field-work in an area where such rocks were very abundant. Mr. F. W. Clarke, whose results and those of Mr. Harker had interested me in the problem, has recently published a fresh calculation of the results of the rock analyses made in the Laboratory of the United States Geological Survey. It is not proposed to discuss these results in detail, and it will suffice for my present purpose to take the question of silica percentage alone. Mr. Clarke's average for all the 1,358 rocks analyzed up to the end of 1903 works out at 60·91. Mr. Clarke, however, admits that the ‘salic’ rocks are more abundant than the other classes, and he further grants that my criticism of his previous results is entitled to some weight, though he thinks that a wider range of observation would probably modify my opinion that granite is practically the average igneous rock.