x9o8.] JENNINGS— HEREDITY IN PROTOZOA. 523 



will raise the average still further by getting rid of some of the 

 smaller lines, etc. 



3. It has been noticed that in many cases continued selection will 

 not carry a character beyond a certain point. This is due (on the 

 view we are setting forth) to the fact that we have finally isolated 

 that line (or lines) of the original collection which had this character 

 most strongly marked, and since selection of the fluctuations has no 

 effect within the pure line, we can make no farther progress. 



4. The phenomenon of so-called regression finds its explanation 

 in the same way. It is found that when extremes are selected, the 

 progeny of these extremes stand nearer the mean than did the par- 

 ents, though they diverge in the same direction as the parents. The 

 reason for this may again be seen by considering such a hetero- 

 geneous collection as that of Table LXL, with the effects of selecting 

 the extremes of size. If we select the largest and the smallest indi- 

 viduals, we shall have taken (i) the largest individuals of the largest 

 lines, and (2) the smallest individuals of the smallest lines. But 

 these, when they propagate, produce, as we have seen, merely the 

 means of the lines to which they belong. The largest individuals 

 will produce then progeny that average smaller than themselves; 

 the smallest individuals progeny that are larger than themselves ; 

 both sets will then approach the mean of the original collection as 

 a whole. 



In working with populations reproducing by cross fertilization 

 among the different lines, the conditions on which these results 

 depend become quite obscured, owing to the introduction of new 

 factors, the union of different factors, the appearance of mendelian. 

 results, etc. Work with pure lines perhaps shows the real cause 

 for the observed phenomena above set forth. 



It must be admitted, then, that the work with pure lines, indi- 

 cating that selection of fluctuations within the lines is powerless, 

 leads to a simple and consistent explanation of many of the observed 

 facts. But, of course, it gives no explanation of the origin of the 

 different pure lines. Clear proof of the effectiveness of selection 

 even within a pure line would therefore be of the greatest interest, 

 and the present writer would find great pleasure in being the first to 

 present such proof. But until such proof is forthcoming, it must be 



