xgoS.] JENNINGS— HEREDITY IN PROTOZOA. 521 



see varied collections showing extremes of value for the coefficient 

 of correlation, from — -3138 to -j- .8500.^^ 



3. Results on Variation, Inheritance and the Effects of 



Selection. 



Our general results with regard to variation, inheritance and the 

 effects of selection are then as follows : 



In a given " pure line " (progeny of a single individual) all 

 detectible variations are due to growth and environmental action, 

 and are not inherited. Large and small representatives of the pure 

 line produce progeny of the same mean size. The mean size is 

 therefore strictly hereditary throughout the pure line, and it depends, 

 not on the accidental individual dimensions of the particular pro- 

 genitor, but on the fundamental characteristics of the pure line in 

 question. 



In nature we find many pure lines differing in their characteristic 

 mean dimensions. 



Our results with the infusorian Parauieckim are, then, similar to 

 those reached recently by certain other investigators Working with 

 pure lines of other organisms. Johannsen (1903) showed that in 

 beans and in barley many pure lines, slightly differentiated from 

 each other, exist in nature, but that selection within a pure line has 

 no effect upon its characteristics. These plants are self-fertilized, 

 so that there is no intermingling of different lines. Hanel (1907) 

 has recently found the same state of affairs in Hydra when multi- 

 plying by budding. Certain lines tend to have a higher mean number 

 of tentacles, others a lower mean number. But within a given line 

 selection of parents with more or fewer tentacles has no eft'ect on 

 the progeny; selection has no effect within the pure line. 



It is doubtless too early to draw any very positive conclusions 

 from these facts. While the results with Paramecium seem clear, 

 I intend to test them further in every way possible. It is pos- 

 sible that selection may be made on some other basis, with a better 



"This fact of course does not render the study of the coefficient of 

 correlation valueless. Its examination under varied experimental conditions 

 is of the utmost importance for determining the real effects of various agents, 

 and in many other ways it furnishes a valuable datum. 



PROC. AMER. PHIL. SOC. XLVII. I90 HH, PRINTED JANLARV I3, I909. 



