n the possibility of
failing, for the same pupils and the same semesters, considered by age
groups? The summary line of Table I gives the total failures according
to the ages at which they occurred. The number of pupils sharing in
each group of these failures is also known by a separate tabulation.
Then the full number of subjects per pupil is taken as 41/2, since
approximately 50 per cent of the pupils take five or more subjects each
semester and the other 50 per cent take four or less (see p. 61). With
the number of pupils given, and with a schedule of 41/2 subjects per
pupil, we are able to compute the percentages which the failures form
of the total subjects for these failing pupils at the time. These
percentages are given below.

THE PERCENTAGES FORMED BY FAILURES AT EACH AGE ON THE POSSIBILITIES OF
FAILING AT THAT AGE AND TIME, FOR THE SAME PUPILS

Ages 13 14 15 16 17 18 19 20 21

% 36.6 38.0 37.9 40.9 40.8 41.2 41.3 42.0 42.7

[Footnote: These percentages are computed from the data secured in
Table I, as noted above.]

There is an almost unbroken rise in these percentages from 36.6 for age
13 to 42.7 for age 21. Not only do a greater number of the older pupils
fail, as was previously indicated, but they also have a greater
percentage of failure for the subjects which they are taking. It seems
appropriate here to offer a caution that, in reading the above
percentages, one must not conclude that all of age 14 fail in 38 per
cent of their work, but rather that those who do fail at age 14 fail in
38 per cent of their work for that semester. The evidence does not seem
to indicate that the maturity of later years operates to secure any
general reduction of these percentages. The prognostic value of such
facts seems to consist in leading us to expect a greater percentage of
failures (on the total subjects) from the older pupils who fail than
from the younger ones who fail. If it were possible to translate the
above percentages to a basis of the possibility of failure for all
pupils, instead of the possibility for failing pupils only, the
disparity for the different ages would become more pronounced, as the
earlier ages have more non-failing pupils. But this we are not able to
do, as our data are not adequate for that purpose.

4. THE INITIAL RECORD IN HIGH SCHOOL FOR PROGNOSIS OF FAILURE

For this purpose the pupil record for the first ye