82 THE INDIAN JOURNAL OF VETERINARY SCIENCE AND ANIMAL HUSBANDRY [ I, II

         85 days S. P. has been selected as the standard, as with this S. P. and a nor-
mal gestation period of 280 days the cow will calve at yearly intervals. Since a
long S. P. increases the lactation milk yield and a short one depresses it, a per-
centage as given in Table X for corresponding S. P. must be added to lactation yield
for all S. Ps. below 85 days, and a percentage subtracted from it for all periods higher
than 85 days. For the sake of comparison, the elimination corrections for Penrith
as obtaired by Sanders [1923] and recalculated to a standard of 85 days S. P. are
also included in Table X. It will be seen that the two sets of figures agree admira-
bly well though our corrections are a trifle lower for the S. Ps. below 85 days and
a trifle higher for the S. Ps. above 85 days.

         The correction factors given for varying lengths of S. P. shew very clearly the
very great potency of S. P. in influencing the total lactation yields for a reference
to Table X will show that milk yielded by a cow in a lactation must be raised by
26 per cent. if she is served 9.5 days after calving, and lowered by 28.5 per cent. if
she is left empty for 409.5 days after calving, in order to see what she would have
produced with a S. P. of 85 days. It must, of course, be mentioned that these
correction factors for different S. Ps. are strictly true for the mid-value of the
intervals. But since the variation of milk yield with S. P. is a continuous one, more
accurate results would be obtained if the corrections for the S. Ps. lying in the
intervals, the means of which are only given in Table X, be interpolated by ordinary
linear interpolation ; especially will this be necessary for very low S. Ps. when the
mean lactation yields are rising very rapidly with successive S. Ps.

         CAUSES OF VARIATION OF MILK YIELD WITH S. P.

         It is generally realized that the length of lactation is determined to a very
great extent by the length of its S. P. The problem has been the subject of study
at different places all over the world with more or less similar results regarding the
extent of correlation existing between these two variables. Ellinger [1910?], for
instance, dealing with Red Danish breed got a very high value of correlation co-
efficient between these two variables r = +.943 ╤ .005. Sanders [1927, 1], how-
ever, working on the same problem got the following results :—

         As will be seen, the value of r is quite high though it is much lower than that
of Ellinger, and close agreement between r and n justifies the use of a linear regres-
sion equation. Both these sets of correlation coefficients, however, seem to the
author rather high and therefore over-estimating the true physiological rela-
tionship existing between these two variables. In fact, Sanders himself admits