accept himself as less gifted, or even to accept life in general as necessarily a matter of repeated failures. In a similar connection, James E. Russell says,[40] "the boy who becomes accustomed to second place soon fails to think at his best." Such psychological results in regard to habits and attitude accruing from repeated failures are both certain and insidious. And an education which purports to be for all and to offer the highest training to each must abandon the inculcation of attitudes of mind so detrimental to the individual and to the very society which educates him. 4. THE PERCENTAGES THAT THE NON-GRADUATE GROUPS FORM OF THE PUPILS WHO HAVE EACH SUCCESSIVELY HIGHER NUMBER OF FAILURES By merely adding the columns of totals for Tables VIII and IX, we are able to obtain the full number of pupils who have each number of failures from 1 to 25. We may readily secure the percentages for the non-graduates in each of these groups by referring again to the numbers in the totals column of Table IX. The following series of percentages are thus obtained. THE PERCENTAGE FORMED BY NON-GRADUATES WITH 0, 1, 2, 3, ETC., FAILURES ON THE TOTAL NUMBER WHO HAVE 0, 1, 2, 3, ETC., FAILURES No. of Failures 0 1 2 3 4 5 6 7 8 Per Cent 68.4 65.7 68.5 77.2 69.0 68.0 70.6 67.3 63.5 No. of Failures 9 10 11 12 13 14 15 16 17+ Per Cent 61.8 63.6 69.0 61.2 66.0 65.3 70.0 61.5 69.4 That these percentages would be higher for the non-graduates than for the graduates (that is, above 50 per cent) would certainly be expected by a glance at their higher numbers in every group of their distribution. But it would hardly be expected by most of us that the percentages would show no general tendency to rise as the failures per pupil increase in number, yet such is the truth as found here. The reverse of these facts was found by Aaron I. Dotey, with a smaller group of high school pupils[41] (1,397), studied in one of the New York City high schools. Still he also asserts that failure in studies is not a cause of elimination to the extent that it is generally supposed to be. We may gain some advantage for judging the general tendency of the extended and varied series of percentages above, by computing them in groups of larger size, thus yielding a briefer series, as follows: (A CONDENSED FORM OF THE PRECEDING STATEMENT) No. of Failures 0 1 to 4 5 to 8 9 to 12 13 to 16 17 to 25 Per Cent 68.4 67.6 67.3 63.9 65.7 69.4