But it could be applied quite easily to most questions. Suppose you wanted to determine beyond question which of two methods of teaching a given subject was the better. We shall assume for the moment that you have unlimited time and money to experiment. It may be thought that we could settle this simply by teaching one person according to one method and another person according to the other, and that we could determine the relative merits of each method from the progress made by each pupil. This, however, would be practically of no use whatever. One pupil might be naturally brighter than the other, and so would naturally learn quicker, even were he taught by an inferior method. To make the experiment of any use we should first take two _groups_ of pupils—the larger the better. For it is obvious that if we take a great number of pupils and place them in two groups the differences between the individuals will tend to offset one another. Let us say the subject is one in which the progress can be quantitatively measured, say typewriting, and let us suppose there are fifty pupils in each group. If after a given time _all_ the pupils in one group had attained a greater speed with accuracy than _all_ the pupils in the other, the test would be almost unquestionable. This would be even more conclusive if the groups were reasonably well balanced. For if all of one group were men and all of the other were boys, the men might make more rapid progress than the boys even with a less efficient system. But it should be easy to divide classes and groups so as to have a reasonable balance of intelligence between them. The probable result of any experiment would be that in neither class would all the pupils make more progress than all the pupils of the other, though you might find that the preponderating majority in one class improved faster than those in the other, and this would probably be sufficient to indicate the superiority of one method, even though one or two pupils in the second group progressed faster than one or two in the first.