Manchester Memoirs, Vol. xlviii. (1904), No. 24. 9 



and that the same is true of the " extracted " recessives : 

 this can be illustrated in our imitation hybridization 

 experiment by placing half the RR's in one hat and half 

 in another ; it is evident that nothing but reds can be got 

 from " matings " from these two hats. 



It is also a fundamental part of the Mendelian principle 

 (in fact it seems to me to be its foundation-stone) that the 

 " extracted " hybrids will produce the same kind, and 

 proportions of the three kinds, of offspring, as the first 

 hybrid ; for if half the R Ws are put into one hat and half 

 into another it is evident that random matings will give 

 25% RR 50% i^i^and 25% WW as before, and, what is 

 more, that they will continue to do so (so long as we keep 

 up the number of counters) for however long we continue 

 the process, that is say for howsoever many generations it 

 is carried on. 



(6) A point of difference between Galton's and 

 Mendel's theory. 



I do not propose to discuss here the difference between 

 the Mendelian principles and the statistical conception of 

 inheritance, but to consider one part of the hypothesis 

 put forward by Mendel, which is at variance with Galton's 

 theory. I refer to the phenomenon of segregation. We 

 have seen what Mendel says {see Bateson :02, p. 57). But 

 this is flatly contradicted by the Galtonian generalization, 

 according to which the greater number of generations a 

 given hybrid is from the first hybrid {i.e., of course, also 

 from the parents of the hybrid) the fewer pure recessive 

 and dominant forms is it likely to produce when mated 

 with another hybrid of its own generation (Darbishire :04, 

 pp. 23 et seq.). 



I refer to this point (which at first sight may appear 

 insignificant, but in reality is not) because it seems 

 to me to afford a means of deciding; between the relative 



