BY RAYMOND PEARL 31 



perience, that two sorts of variations can be distinguished, namely those 

 (a) that are represented in the germinal material and are inherited without 

 substantial modification, as in "pure lines," and those (6) that are somatic 

 and are not inherited. By anything short of the actual breeding test it is 

 quite impossible to tell whether a particular variation observed in the soma 

 belongs to the one category or to the other. As I have tried to emphasize in 

 other places, it is both to be expected on this view of inheritance, and is 

 also the case in actual fact, that the somatic manifestation or condition of 

 any character is a most uncertain and unreliable criterion of the behavior 

 of that character in breeding. Finally under the genotype concept, of 

 course, the whole array of facts brought out by Mendelian experiments find 

 their place. 



Now while certain adumbrations of the genotype concept have long been 

 current in biological speculations in regard to heredity, this general view- 

 point owes its grounding in solid facts primarily to Johannsen's work with 

 beans and with barley. It is to be noted that in these cases, as well as in 

 most of the investigations of the pure line theory which have followed 

 Johannsen's work, the organisms used have been such as reproduced either 

 by self-fertilization, or by fission, or by some vegetative process. This 

 brings us to the consideration of a question of great importance, both 

 theoretical and practical. In cases of dioecious organisms, where a "pure" 

 pedigree line in the sense that such lines are found in beans or in Parame- 

 cium by definition can not exist, has the genotype concept any bearing or 

 significance? In a general way it obviously has. Probably no one (except 

 possibly some of the ultra-statistical school) could be found who would 

 deny that in general a distinction is to be made between variations having a 

 gametic and those having merely a somatic basis. But specifically how 

 far has the genotype concept any application in case of * 'non-self ed" 

 organisms? Johannsen in his "Elemente" has thoroughly analyzed 

 Galton's material and shown that it is capable of a satisfactory and reason- 

 able interpretation on the genotype hypothesis, and East and Shull have 

 gone far in the analysis of genotypes in maize. This, however, is only a 

 beginning. There is the greatest need for careful, thorough investigations 

 of the inheritance of characters showing marked fluctuating variation in 

 organisms having the sexes separate. Here lies one of the crucial fields in 

 the study of inheritance to-day. Through the brilliant results in Mendelian 

 directions and from the study of really "pure' lines we are getting clear-cut 

 ideas as to the inheritance of qualitatively differentiated characters, such as 

 color, pattern and the like, on the one hand, and in regard to the inheritance 

 of quantitative variation in self-fertilized or non-sexually reproducing 



