This means that in the long run a stranger will place a given individual in a group of twenty persons not over three or four positions away from the place to which other strangers would assign him. The individual's physiognomy, however little it may actually reveal of his personality, nevertheless suggests rather definite characteristics to those whom he meets, and to that degree determines their reaction toward him, expectations of him, and belief in him. The definiteness or agreement of these impressions seems also to vary with the trait in question; it is high for intelligence and perseverance, low for humor and deceitfulness, and intermediate for kindliness, conceit, and courage. Our own results, however, must be taken only as suggestive, rather than as general, since they may easily have been determined partly by the particular set of photographs we used and by our particular and diverse sets of judges.[3] Results of this character, and many similar ones which we are accumulating, suggest, however, an interesting set of problems. It is psychologically as interesting to inquire just what impressions people actually receive from one's physiognomy and expression, as it is to ask whether these impressions are correct. One's ultimate vocational accomplishment often depends on the first impression he creates, the type of reception his appearance invites, even though there may be no necessary connection whatever between appearance and mental constitution. Vocational success depends not only on the traits one really possesses, but also somewhat on the traits one is believed to possess. It is also interesting to observe that high correlations exist between some of the traits as judged merely on the basis of photographs. Let 1.00 be taken to indicate complete correspondence between two orders of merit, so that the highest in the one scale is also the highest in the other scale, the second in one the second in the other, and so on; then -1.00 will indicate a completely reversed order, the best in one class being the poorest in the other, etc.; a coefficient of 0 will mean only a chance relationship, i. e., none at all. Then from 1.00 through 0 to -1.00 we have represented all possible degrees of correspondence.[4] These figures are called "coefficients of correlation," and can easily be computed by proper statistical methods. In the present case the coefficients for all combinations of two traits are as follows: