Comments on the Method; Statistical Computations

Abstract

The individual weighings were in fact rather inaccurate. The amount of food eaten by a mouse in a 2-hr period may be in the order of some tenths of grams. The possible weighing error made when estimating the weight in tenths of grams therefore is not really small in comparison to the weight differences in which we were interested. In principle one might think the maximal weighing error to be ± 0.1 gram, but our data show that this is too optimistic: subtracting present weight of a food grating from its previous weight, we sometimes got negative numbers; usually in such a case we got −0.1, which might be expected if the mouse had not eaten at all; but sometimes we got differences like −0.2, −0.3 or even −0.4 which seem inexplicable because our readings did not seem to us that inaccurate.