In an earlier report, one of the authors describes 28 temperament factors for which there is sufficient consensus in the literature to call them "established". From 1 to 5 distinct bipolar subscales were suggested to mark each factor; 12 or 16 item subscales were written, each balanced in terms of the two defined poles and in terms of numbers of positively and negatively stated items. This yielded a total of 1,332 items for 87 subscales as intended markers for the 28 factors. This study determines, by means of factor analyses using the newly constructed subscales, how well the 28 putative factors stand up. A total of 4,000 naval recruits were tested with a complex design that permitted determination of the correlations among all 87 subscales but that required no individual to respond to more than 320 items. A second sample of 153 female and 92 male students at the University of Oregon were given the entire battery of items. After item analyses and reliability determinations were made, 77 subscales survived. Correlation matrices were computed separately for the two samples and factor analyses performed; 22 minres factors were retrained in each sample and rotated to maximum obliqueness using the direct oblimin method. Results are discussed separately for the two samples and compared across the two. (Author/RC)