A STUDY OF THE SEGREGATION OF A QUANTI- 

 TATIVE CHARACTER IN A CROSS BETWEEN 

 A PURE LINE OF BEANS AND A MUTANT 

 FROM IT. 



By I. LEITCH. 



(With Four Text-figures.) 



The service which Mendel and his rediscoverers rendered to the study 

 of heredity in demonstrating that certain qualitative characters behave, 

 relatively to each other, as separate entities, has been paralleled by 

 Professor Johannsen, Jennings and others for quantitative characters. 

 Until their researches were made, size, for instance, was regarded in the 

 light of Darwin's and Galton's work as something capable of endless 

 variation — the net result, in any particular case, of a summation of causes, 

 natural selection, fitness, and other equally indefinite concepts. What 

 Professor Johannsen did was to delimit certain quantitative characters; 

 to show that, although to some extent influenced by environment, still 

 they are definite characters, capable of accurate description and analysis. 

 What makes the problem of quantitative differences appear more difficult 

 than that of differences in colour and other such qualities is that they 

 cannot be described and classified on inspection, but the two problems 

 approach in cases where, for instance, cumulative colour factors operate, 

 and quantity of colour enters the field. There appears indeed to be no 

 reason to regard the problems as essentially different. 



Professor Johannsen established for quantitative characters the prin- 

 ciple of working with homozygous, " pure," material as the basis of all 

 analysis. He demonstrated for his pure lines of beans and barley that 

 such material breeds true ; that, in a homozygote, such characters as size 

 and weight are as constant as gi'eenness or smoothness ; and he made 

 familiar the exact description and comparison of such characters by means 

 of their distribution curves. For instance, the sizes of the beans in two 

 plants, or pure lines, or aggregates of beans, will be described fully and 

 accurately in any given case by the distribution curves of their lengths 

 and breadths, and the comparison of different sizes will be a question 

 merely of the comparison of distribution curves. These curves have not 



