i9o8.] JENNINGS— HEREDITY IN PROTOZOA. 503 



group gives in both cases a low ratio. But taking the averages, in 

 connection with those of Table XXVI., it is clear that the smaller 

 races are as a rule slightly broader in proportion to the length than 

 are the larger races. 



Turning now to the question whether there are differences in the 

 proportion of breadth to length in different races of the same group, 

 we have full data only for the lines g and i, as given in Table XXIII., 

 page 488. Beginning with the data for November 23 (since before 

 that date the number of individuals is small), we can make determi- 

 nations for seven dift'erent dates of the ratio of mean breadth to 

 mean length, the two sets being on each date as nearly as possible 

 under identical conditions. 



Table XXVIII. 

 Ratio of Mean Breadth to Mean Length for g and i (Table XXIII.). 



November November December December December January February 



23. Per 26. Per 7. Per 16. Per 30. Per 2. Per 5. Per 



Cent. Cent. Cent. Cent. Cent. Cent. Cent. 



g 27.011 28.238 34.091 30.370 27.797 27.686 30.853 

 i 34.291 28.325 34.655 30.159 25.430 24.747 29.386 



Thus, in the first three determinations the ratio was greatest in 

 the line i; in the last four it was greatest in the line g. Evidently 

 there is no constant difference in proportions between these two lines. 



For other lines our data are not sufficient to test this matter. 

 Our only positive result on this point then is that the smaller races 

 are as a rule proportionately broader than the larger ones. 



2. Results of Selection within Pure Lines. 



We have seen that an ordinary " wild " culture of Paramechim 

 contains many lines or races, which are dift'erentiated in size. By 

 selection it is possible to isolate these diverse lines; so that in this 

 way we can obtain cultures in which the mean size is large or small, 

 or intermediate, as we prefer. In this case selection, of course, acts 

 by isolating lines that already exist, and allowing them to propagate 

 unmixed. 



How do these diverse lines arise? Can we obtain them by selec- 

 tion within the limits of a single line? If from among the progeny 

 of a single individual we select the larger and the smaller specimens, 



