The data which are available do not seem very suitable for Dr Whipple's method of analysis. As explained in my original paper, the groups into which the rock-temperature observations were divided depended entirely upon the method adopted to get rid of the effect of variations of air temperature at the surface, and the resulting means were for dates which had no connection with the long-period temperature fluctuations which were later disclosed. A series of observations extending over nearly a hundred years was thus reduced down to eight mean temperatures for periods of various lengths.