igos.] JENNINGS— HEREDITY IN PROTOZOA. 517 



line or race, and in the same environment, then the coefficient of 

 variation depends largely upon the stages of growth it contains. By 

 taking specimens nearly in the same stage of growth we were able to 

 reduce the coefficient of variation in length in some cases to 4.521, 

 in breadth to 6.976, while by taking collections including various 

 ages, under similar conditions, coefficients were found as high as 

 13.729 for length and 13.292 for breadth (Table X.). The most 

 carefully selected lots contain specimens differing a certain amount 

 in age, otherwise the coefficient of variation could be still further 

 reduced in this way. Specimens beginning fission or undergoing 

 conjugation include few growth stages, hence they show a low coeffi- 

 cient of variation. The coefficient for those beginning fission is less 

 than for conjugants (see page 453). 



The coefficient of variation for a given line is tremendously 

 affected by environmental conditions. Thus, we see this coefficient 

 changed in twenty-four hours, by a change in environment, from 

 7.003 to 12.767 for length; from 12.473 to 28.879 for breadth. 

 Different environments give us all sorts of values between such 

 extremes. 



It is evident that no particular coefficient of variation can be 

 considered characteristic of Paramecium, or of any line of Para- 

 mecium; certainly not unless the conditions as to growth, envir- 

 onment, etc., are very precisely defined. Wo. have seen that the 

 variations found among different individuals of the same pure line 

 do not show themselves to be heritable. This, along with all the 

 rest of the evidence, indicates that if all conditions of growth and 

 environment were made identical throughout a sample of Paramecia 

 belonging to a pure line, the coefficient of variation would be very 

 near to zero. In other words, all the variations that we have been 

 able to detect with certainty in a pure line are due to growth and 

 environment. Presumably other variations (congenital and heredi- 

 tary) must occur at times, but they appear to be so rare that it is 

 difficult to detect them and they would have little effect on the 

 coefficient of variation. By properly varying the conditions, we may 

 get in a pure line all coefficients of variation in length, from a limit 

 near zero up to 20 or more. 



4. The ratio of breadth to length (serving to partly define the 



