Page:Popular Science Monthly Volume 86.djvu/508

This page has been validated.


IN a series of articles entitled "A Statistical Study of American Men of Science," printed in Science in 1906 and 1910 and as an appendix to "The Biographical Directory of American Men of Science," methods were explained by which the thousand leading scientific men of the United States had been selected and arranged in the order of the merit of their work. Studies were made of the measurement of scientific performance, of the origin and distribution of scientific men, and of the changes which occurred during an interval of several years. Data have now been gathered in regard to the families of the men of science previously selected. In our present state of ignorance a statistical study of any homogeneous and objectively chosen group should be of value, both as a contribution to psychological and vital statistics in general and for comparison with other groups which may be similarly studied. Scientific men form a desirable group for such study as, on the one hand, they may be assumed to be willing and competent to supply the information and, on the other hand, knowledge concerning the conditions favorable to scientific performance may have important practical applications.

Of one thousand one hundred and fifty-four scientific men from whom information in regard to their families was requested 1,036 replied and 118 did not. Of the replies 16 were blank, sometimes accompanied by the explanation that the information was not readily attainable or the like, 7 were to the effect that the information would be sent later or the like, 13 were received too late, 25 were very imperfect, 975 were usable and in most cases complete. This is an unusually full reply to a questionnaire. For example, in answer to an inquiry in regard to noteworthy relatives addressed to 467 fellows of the Royal Society, Sir Francis Galton received 207 useful replies, and the completely available returns "scarcely exceeded 100." In such cases it is desirable that returns should be complete in order to avoid the selection of a special class. Thus, when people are asked whether they have noteworthy relatives, those having them are more likely to reply than others, and the percentage of positive replies may give no definite information in regard to the frequency. In the present case it appears, from examination of the names of the ten per cent. who failed to reply, that there was no group that would affect appreciably the result of the inquiry.