Raymond Pearl 
227 
the various "Selected Ancestry " series with random series, whether conjugant or 
non-conjugant. The individuals in these "Selected Ancestry" series all came 
from the same original single ancestor, and each sample was reared throughout its 
history under as uniform environmental conditions as it was possible to obtain. 
It is apparent that when the table is viewed as a whole the individuals in the 
conjugant series tend to be both smaller and less variable than those in the non- 
conjugant series. In the early history of the Leipzig culture all the individuals in 
it were small, but, as will be shown later in a more direct way, throughout the 
period during which it was under observation the mean size of the individuals 
increased. At the same time the variability in proportion to size tended to 
increase somewhat. 
Turning now to the character breadth we unfortunately have at present only 
one other series for comparison with those reduced in this paper. For the present 
the longer non-conjugant series alone will be considered with reference to this 
character. The results are shown in Table III. 
TABLE III. 
Variation in Breadth of Paramecium. 
Series and Class 
Mean 
Standard 
Deviation 
Coefficient of 
Variation 
No. 
Series A, All Non-Conjugants ... 
)) "-^I )! )) 
J) )) )) 
Simpson's Series 
52-827+ -273 
54-208+ -280 
63-250+ -340 
68-125 
5-870+ -193 
5-905 + -198 
5-786+ -240 
9-155 
11-112± -370 
10-894+ -370 
9-149+ -383 
13-439 
210 
202 
132 
100 
From this table it is at once evident that in proportion to the magnitude of the 
dimension the breadth is somewhat more variable than the length in Paramecium, 
but the difference is not great. The values of the means are, for all three of the 
present series, lower than that for Simpson's, but this is only what would be 
expected from the fact that the mean lengths are lower for these particular series 
also. It would appear that, as the breadth increases in magnitude, it becomes 
proportionally less variable, but the series of data available at present are too few 
to decide whether such a relationship is usual. 
If we consider the variation analytically we have the results shown in Table IV. 
This table gives the values of mean, mode, ix.2, /ig, |Jb^, /Si, V^Sj, /S.^, 3-/32, k^, and 
the skewness* for the length and breadth of all conjugants and all non-conjugants 
of Series A. 1 have not thought it worth while to determine the analytical 
constants for any other of the present series for the reason that they are 
statistically so short, and because I hope to be able to publish eventually the 
reductions of much more extensive material on variation in Paramecium. 
* The analysis of these curves is carried out by the methods given in Pearson's memoir on Skew 
Variation (Phil. Trans. Vol. 186 A, pp. 343—414), and its Supplement (Ibid. Vol. 197 A, pp. 443—459). 
