,^o8.J JENNINGS— HEREDITY IN PROTOZOA. 453 



seems extraordinary that a process following the same laws should 

 in some cases be measured by hours, in other cases by months, in 

 others by years. 



3. Effects of Growth on the Observed Variation. 



A random sample of an ordinary culture of Paramecium contains 

 specimens falling in all parts of the growth curves represented in 

 Diagram 5. If we measure the various members of such a sample, 

 as was done by Pearl (1907), we shall then find many variations in 

 size, which variations consist to a considerable extent of different 

 growth stages. Not all the observed variations are due to this 

 factor, but its importance is very considerable. This will best be 

 appreciated by running through the columns headed " coefficients of 

 variation" in Table X. If we take samples including specimens 

 falling in the early parts of the growth curve, when the absolute size 

 is small but the changes with growth are very marked, then the 

 coefficients of variation in length are high ; thus in rows 4 and 5 they 

 are 15.494 and 13.729, respectively, while in the random sample of 

 the same culture the coefficient is but 8.834 (row 6). On the other 

 hand, if we take specimens restricted to a very small portion of the 

 curve, the coefficient of variation becom.es very low; thus in a lot 

 whose age falls between 18 and 28 minutes the variation is but 4.521 

 (row 8) ; at the age of 4.20 to 5 hours is 5.043 (row 17), though the 

 variation for a random sample of this same culture is 13.262 (row 19). 

 The eft'ects of growth on variation are shown to the eye in Diagram 



4, p. 440. 



Variation at Fission. — The effects of growth on the observed 

 variation are likewise seen when we compare random samples with 

 individuals that are at a definite stage in the life history. Thus, if 

 we take specimens at the beginning of fission, when the constriction 

 first appears, we find the coefficient of variation very low, as com- 

 pared with those of random samples of the same cultures. This is 

 readily seen in the following tabulation of the coefficients of varia- 

 tion for the four cultures of Table X. in which the specimens begin- 

 ning fission were studied (see next page). 



Variation in Conjugants. — Again, the same thing appears when 

 we compare conjugating individuals with random samples of the 



