APPENDIX.
379
These columns may be used for two purposes.
The one is to calculate a table like that in p. 34, where
I have simply lumped 11 of Quetelet’s grades into I,
so that my classes A and a correspond to his grade 11 in
column N, my classes B and b to the difference between
his grades 22 and 11, my C and c to that between his grades
33 and 22, and so on.
The other is as a test, whether or no a group of events
are due to the same general causes ; because, if they are,
their classification will afford numbers that correspond with
those in the table ; otherwise, they will not. This test has
been employed in pp. 30, 31, and 33. The method of
conducting the comparison is easily to be understood by
the following example, the figures of which I take from
Quetelet. It seems that 487 observations of the Right
Ascension of the Polar Star were made at Greenwich
between 1836 and 1839, and are recorded in the publica¬
tions of the Observatory, after having been corrected for
precession, nutation, &c., and subject only to errors of
observation. If they are grouped into classes separated by
grades of 0.5 sec. the numbers in each of these classes will
be as shown in Column III. page 380. We raise them
in the proportion of 1,000 to 487 in order to make the
ratios decimal, and therefore comparable with the figures
in Quetelet’s table, and then insert them in Column IV.
These tell us that it has been found by a pretty large
experience, that the chance of an observation falling within
the class of — 0.5 sec. from the mean, is 150 to 1,000; of
its falling within the class of — 1.0 sec. is 126 to 1,000;
and so on, for the rest. This information is analogous to
that given in Column M of Quetelet’s table, and we shall
now proceed to calculate from IV. the Column V. which is
analogous to Quetelet’s N. The method of doing so is,
however, different. N was formed by adding the entries in
M from the average outwards ; we must set to work in the