As part of an investigation of methods of selecting prospective teachers for training, it became necessary for the authors to construct two prognosis tests of teaching ability. One is known as Teachers Interest and the other as a Prognosis Test for Prospective Teachers. This article gives the procedure used in studying these tests. Questions concerning reliability and validity, particularly the latter, must be left for a later and much longer discussion. These tests, and five others, were given in the fall of 1927 to all students entering the New York State normal schools and to students in about half of the training classes. (The training classes are oneor two-year institutions that prepare teachers for the rural field.) Criteria for validating these tests will consist of the teachers' marks at the end of each semester, the results of general objective examinations given to all students, the rating on practice teaching, and the rating on teaching following the completion of training. The whole investigation will, there fore, cover a period of four or five years. The intelligence and achievement tests used in this study had pre sumably been analyzed and revised before being published. The two newly constructed tests had not, however, undergone this procedure. Since the analysis and revision of a prognosis test involves a peculiar technique, it is of interest to the research public to follow the steps: first, the determination of the difficulty of the individual items ; second, an analysis of the test, item by item, in the light of these data ; third, the elimination or revision of items in the test on the basis of this analysis ; and fourth, a statistical comparison between the original and revised form. The Teachers' Interest Test consisted originally of 110 true-false statements concerning teaching practices and classroom procedure. The underlying idea is that the student entering the normal school has had, as a pupil in the elementary and high school, opportunity to observe classroom procedure, and that the student's interest and probable ability in teaching may be measured by the extent and accuracy of the observa 36