Raymond Pearl 
233 
12'9 per cent, of the original mean*. These values give some idea of the great 
changes which are produced in Paramecium as a result of environmental influences. 
That this increase in mean length in the present case is due to environmental 
influences and is in no way connected with the conjugation epidemic is evidenced 
by the fact that the conjugants show an increase as well as the non-conjugants 
though not by so great an amount. Furthermore the same sort of change 
has been found in other casesf. Further discussion of this interesting and 
important matter of environmental influence on variation in Paramecium may 
be deferred for the present, as I hope to return to it in a special paper based 
on data collected ad hoc. 
In connection with this increase of mean length during the period of exami- 
nation, it will be noted that both the absolute and relative differences between 
conjugants and non-conjugants increase, being lowest for Series A and highest 
for Series I). In other words, the conjugants and non-conjugants are more 
differentiated from one another at the end of the conjugation epidemic than they 
were at the beginning. This is due to the fact that the environmental influences 
produce a greater effect on the non-conjugants than on conjugants. The signi- 
ficance of this fact on the theoretical side will be discussed later. 
Examining the coefficients of variation in Table V., we see that the differences 
— both absolute and relative — are practically constant for all three series. A, G, 
and B. In other words, though the means of both conjugants and non-conjugants 
increase, the amount of variation in proportion to the size does not sensibly 
change. This seems to be a rather important result, when it is remembered 
that Series C and D must have included among the non-conjugants measured 
a larger proportion of recent ex-conjugants than did Series A. It points distinctly 
to the conclusion that the biological significance or purpose of the conjugation 
process is not to bring about increased variability. Because, clearly, if conjuga- 
tion were followed by greater variation in the ex-conjugants, we should expect 
the non-conjugant distributions of series G and D (containing more ex-conjugants) 
to differ sensibly more from the conjugant distributions in variability than does 
that of series A. Furthermore, if we compare the coefficient of variation for 
length of series E (= 8"529 + "357) with the values of the same constant for 
the non-conjugants of series A, G and D (= 8-185 + -271, 9-123 + -309 and 
8"937 + '759 respectively), it clearly is not significantly greater. But the indi- 
viduals of series E must have been, in considerable proportion, ex-conjugants. 
Putting all the results together, there seems no doubt that the primary signifi- 
cance of conjugation is not to produce variability, whatever else it may be. 
* The increase, is of course, apparently somewhat greater in the case of the non-conjugants if 
we take only the period to August 30th, thus making D the last series, since Series D shows a slightly 
higher mean than Series E. Since, however, D is such a short series it seems likely that we shall 
come nearer the actual facts by taking the mean of Series E to represent the maximum for the period 
and culture under discussion. In any event the difference between the means of series D and E for 
length is not significant in comparison with its probable error. 
t Cf. Pearl, E., and Dunbar, F. J., loc. cit. 
