« 
64 Variation in Chilomonas 
In view of the striking difference in type between the two series it is some- 
what surprising to find them so nearly alike in variability. For none of the 
characters can the differences in the variation constants be said to be significant. 
It is worth noticing, however, that, with a single exception (the absolute variation 
in breadth) the differences between the variation constants, both absolute and 
relative, are negative. That is to say, the individuals of Series A, living under 
■imfavourahle conditions, are slightly more variable than those of Series B, living 
under favourable conditions. The differences are so small in comparison with 
their probable errors, however, that no particular stress is to be laid upon this 
fact. The conclusion to which we must come from the present data is that there 
is no marked difference in variability between individuals living under conditions 
which in the one case were very favourable and in the other case very un- 
favourable to the continued existence of the race in the active condition. What 
slight preponderance exists is in favour of greater variation under unfavourable 
conditions. 
We may next examine the correlation between length and breadth of body for 
the two series. The raw material is' given in Tables I and II, whence, calculating 
S (xy) 
the coefficient of correlation by the usual formula r = — — , we set : 
Series A (Unfavourable conditions) r = '683 + "025 
Series B (Favourable conditions) r = -617 ± -032 
Difference = -066 + -041 
These values of the coefficients are high, indicating a closer relationship 
between length and breadth of body in this simple protozoan than would have 
been predicted, I think. Just as in the case of the simple variation, however, 
there is no certainly significant difference between the two series in respect to 
degree of correlation. What difference there is is in favour of higher correlation 
under unfavourable conditions, but no great stress is to be laid on the difference. 
Since biometric investigations on Protozoa are as yet not especially numerous, 
it seems desirable to examine the regression for these two characters, length and 
breadth, to determine whether it is linear or not. The equations to the regression 
coefficients are as follows : 
Series A, Length on breadth. 
Series A, Breadth on length. 
