No. 565] CHANGES PRODUCED BY SELECTION 



7 



material used in pure line studies. It has characters that 

 can be estimated readily and accurately and which arc 

 affected only slightly by external conditions. It is easily 

 grown, is naturally self-fertilized, reproduces prolifically, 

 and is known in many markedly different varieties. In 

 fact, it is an ideal subject for work of this kind. 



The investigations were not patterned after the stand- 

 ard type set by Johannsen wherein the constancy of suc- 

 cessive generations of pure lines grown from selected 

 extremes were tested, since even if it were possible to 

 gather a quantity of data at all comparable to that col- 

 lected by Johannsen (:09) and Jennings ( :08) in their 

 brilliant investigations, the criticisms mentioned above 

 might still be made. The plan chosen was that of cross- 

 ing two varieties of tobacco which differed in a character 

 complex easily and precisely determined, and of selecting 

 extremes from a number of families of the F 2 generation. 

 If Johannsen 's views be incorrect, such continued selec- 

 tion should affect each family in the same degree. If his 

 conclusions be justified, selection should reach an end- 

 point in different generations in different families, and 

 there should be no relation between the number of genera- 

 tions required to reach this end-point and the progress 

 that is possible. 



There should be no need of a historical summary of the 

 previous investigations that have been interpreted as cor- 

 roborating or refuting Johannsen 's conclusions. Such 

 summaries have been made in other papers. It should be 

 mentioned, however, that the classical researches of Pearl 

 ( ill) on the inheritance of fecundity in the domestic 

 fowl have been so planned and executed that certain of 

 the criticisms directed against Johannsen mentioned above 

 are not justified, yet Pearl finds himself thoroughly in 

 accord with the Danish physiologists position. 



Several hundred varieties of Xicotiana tabacum exist 

 which differ from each other by definite botanical char- 

 acters, yet only two general characters suitable for our 

 purpose were found. We desired to confine our observa- 

 tions to quantitative characters that were influenced but 



