In the spring of 1922 the Grand Junction High School was asked by the North Central Association of Colleges and Sec ondary Schools to participate in an experiment concerning the relation of the size of classes to school efficiency. The purpose of the experiment was to ascertain, if possible, whether the size of a class has any appreciable effect upon the knowledge the pupils acquire and the work they do. The only requirements made by the North Central Associa tion were that the experiment must be conducted in at least three classes, all of which must be taught by the same teacher and all of which must cover the same work. The size of the classes was to be approximately forty, twenty-five to thirty, and less than twenty. The pupils, as nearly as possible, were to be selected without regard to their intelligence scores. The North Central Association provided no particular plan and no particular tests. We were therefore left to conduct the experiment in our own way. Accordingly, in the fall of 1922, we planned the project for the junior English classes, and organized three of these classes as follows : No. 1 with 20 pupils ; No. 2 with 34 pupils ; and No. 3 with 44 pupils. We selected the pupils for each class with no thought of their intelligence quotients. Some four weeks after school began, however, we gave them the Terman Group Test of Mental Abil ity and computed their intelligence quotients. We then found that the medians for the intelligence quotients varied only a frac tion of one percent. The first and third quartiles were also nearly the same for all three classes. After thoroughly establishing our classes, our method of con ducting the experiment was merely to proceed with the year's 126