others who have not published their results. In these studies two factors-usually the first two in order of variance contribution or size of latent root-appear with only rather minor variations of pattern despite noteworthy differences in the composition of the samples of persons and variables, and despite procedural variations such as those which result from investigators' having different rotational philosophies, using different criteria (on samples of different size) to decide the number of factors to extract or interpret, etc. The reliable variance of the 16 P.F. variables is seemingly not exhausted by these first two factors, however. When the 16 P.F. is factored alone, for example, application of tests for determining the &dquo;correct&dquo; number of factors usually indicates more than two factors (e.g., Karson, 1961) ; in studies employing a wider range of variables, the questionnaire variables usually help to define more than two factors (e.g., Cattell & Scheier, 1958). But it has been difficult to regard the factors beyond the first two as replicated in different studies. Samples have not been described in sufficient detail to allow development of hypotheses that would, if verified, explain the differences in factor structure in terms of known relationships between 1 This investigation was carried out during the tenure of a predoctoral fellowship from the National