Another result came out in connection with these experiments on memory, which, apart from its practical interest, may serve to show an additional resource of experimental psychology. In making up the results of a series of experiments it is very important to observe the way in which the different cases differ from one another. Some cases may be so nearly alike that the most extreme of them are not far from the average of them all; as we find, for example, if we measure a thousand No. 10 shot. But now suppose we mix in with the No. 10 some No. 6 and some No. 14, and then take the average size; we may now get just the same average, and we can tell that this pile is different from the other only by observing the individual measurements of the single shot and setting down the relative frequency of each particular size. Or, again, we may get a different average size in one of two ways: either by taking another lot of uniform No. 14 shot, let us say, or by mixing with the No. 10 a few very large bullets. Which is actually the case would be shown only by the examination of the individual cases. This is usually done by comparing each case with the average of the whole lot, and taking the average of the differences thus secured--a quantity called the "mean variation." In the case of the experiments with the squares, the errors in the judgments of the students were found to lie always in one direction. The answers all tended to show that they took, for the one originally shown, a square which was really too large. Casting about for the reason of this, it was considered necessary to explain it by the supposition that the square remembered had in the interval become enlarged in memory. The image was larger when called up after ten or twenty minutes than it was before. This might be due to a purely mental process; or possibly to a sort of spreading-out of the brain process in the visual centre, giving the result that whenever, by the revival of the brain process, the mental image is brought back again to mind, this spreading out shows itself by an enlargement of the memory image. However it may be explained, the indications of it were unmistakable--unless, of course, some other reason can be given for the uniform direction of the errors; and it is further seen in other experiments carried out by Messrs. W. and B. and by Dr. K.[9] at a later date. [Footnote 9: Dr. F. Kennedy, demonstrator, now professor in the University of Colorado (results not yet published).]