+--------+--------------------------------------+ | | Individual Correctness of Judges in | | | Estimating | | Judge |--------------------------------------+ | | Intelligence | Neatness |Sociability | +--------+--------------+----------+------------+ | I | .51 | .11 | .39 | | II | .11 | .10 | .08 | | III | .15 | .29 | .05 | | IV | -.27 | .06 | .49 | | V | .08 | .24 | .08 | | VI | .43 | .41 | .28 | | VII | .04 | .11 | .02 | | VIII | .39 | -.09 | .32 | | IX | .22 | -.08 | .00 | | X | .30 | .02 | .55 | +--------+--------------+----------+------------+ |Average | .19 | .11 | .22 | +--------+--------------+----------+------------+ These random samples of individual judicial capacity show at once how unreliable individual judgment is in these matters. The individual judges vary widely among themselves and they also depart widely from the established order. Moreover, a judge who may happen to show a reasonable degree of correctness in judging sociability may be very far away from correctness in judging the other traits, or may, indeed, judge in quite the reverse of the correct order. To have accepted the verdicts of a single judge would not only have been manifestly unfair to the individual but also hazardous to the employer. The combined impressions of twenty-five judges is here required for the correlations for even half of the traits to reach over .38. The second point to be noted is that even under these circumstances the coefficients are far from perfect, even for those traits in which they are the highest. Only if beauty, snobbishness, or vulgarity are the traits which are crucial, are judgments of the photographs reliable enough to be worth considering. It would appear that the vocations which depend markedly on these characteristics are exceedingly few. And even here, although the reliance on coefficients of .55 might in all probability aid the employer in decreasing the percentage of the snobbish or the vulgar among his employees, grave injustice would most certainly be done to those many individuals who constitute exceptions and keep the correlations from being perfect. Only when correlation coefficients are very high can their indications be applied in the guidance of individuals (as distinguished from the selection of groups) with safety and justice. Dean Schneider reports an attempt to verify the principles of a certain system of physiognomics by putting them to an actual test. He writes: