.JxjLT 29, 1921] 



SCIENCE 



93 



non-binomial, unsystematic jwpular compila- 

 tion. 



The volume is effusively dedicated to 

 " Wilhelmo VI Hessio Landgravio," whose 

 titles and virtues his " devotus cliens " ex- 

 pounds at length. 



David Starr Jordan 



SPECIAL ARTICLES 



ON A METHOD OF ESTIMATING THE NUMBER 



OF GENETIC FACTORS CONCERNED IN 



CASES OF BLENDING INHERITANCE 



In the early days of rediscovered Men- 

 delism Bateson^ suggested the idea that what 

 was then kaown as blending inheritance might 

 be a variety of Mendelism in which dominance 

 was wanting, but in which several or many 

 independent factors were involved. This sug- 

 gestion was found to be in good agreement 

 with much experimental work on quantitative 

 characters subsequently carried on by Nilsson- 

 Ehle, Tammes, Emerson and East, and others. 

 It is now generally accepted as the most prob- 

 ably correct explanation of all varieties of 

 intermediate or blending inheritance. Ac- 

 cepting this as a working hypothesis, have 

 we any means of discovering how many fac- 

 tors are involved in cases of blending inheri- 

 tance? Surely the number must be very dif- 

 ferent in different cases. 



Noteworthy features of blending inheri- 

 tance are the following: (1) F^ is intermediate 

 between the pure parental races, but not more 

 variable than the more variable parent. (2) 

 Ej is likewise intermediate in character but 

 is more variable than Fj or either parent. (3) 

 In F, and subsequent generations the varia- 



bility decreases from the maximum suddenly 

 attained in F,. 



In all varieties of inheritance, whether typi- 

 cally Mendelian or blending, the maximum 

 variability is to be found in the F„ generation. 

 In ordinary Mendelian inheritance we are 

 able to detect the number of genetic factors 

 concerned by the number of phenotypes which 

 are distinguishable in F^ and by their numeri- 

 cal proportions. The Fj generation is in 

 strong contrast with the F, generation to 

 which it gives rise, for Fj^ is of a single type, 

 if the parent races were pure. 



In Mending inheritance also, it is the F^ 

 generation which affords a clue to how many 

 genetic factors are involved, not by the forma- 

 tion of clearly distinguishable types (for there 

 is but one), but hy the amount of the varia- 

 bility of that single type in F^ as compared 

 with f J. 



To make this clear, let us consider the nu- 

 merical series commonly employed, by ex- 

 positors of the multiple factor hypothesis, for 

 explaining the increased variability of F^ in 

 blending inheritance. If two pure races differ 

 from each other by a single genetic factor 

 (which does not show the phenomenon of domi- 

 nance), and if these two pure races are 

 crossed, F^ will be intermediate. Y^ will also 

 be intermediate in part, but the parental 

 classes will also reappear, and there will thus 

 be three distinguishable classes in F^, which 

 correspond with the two parental types and 

 the Fj type respectively. The classes will be 

 numerically as 1:2:1, as in the familiar 

 case of the blue Andalusian fowl. 



Now suppose that the pure parent races dif- 



Fj Distributions in Size Classes, when Inheritance is Blending and Involves from One to Sia 

 Independent and Equivalent Factors 



1 Report I. to the Evolution Committee of the Royal Society, London, 1902 



