THE SPLIT OR Y COMB. 



Again, mating two extracted oo combs of F, should jdeld, in F„ two 

 types of families in equal frequency, as follows : 



1. oo comb, 100 per cent. 



2. 00 comb, 75 per cent; no comb, 25 per cent. 



Single combxY comb should give families of the types: 



1. Y comb, 100 per cent. 



2. Y comb, 50 per cent; I comb, 50 per cent. 



3. Y comb, 50 per cent; oo comb, 50 per cent. 



4. Y comb, 25 per cent; I comb, 25 per cent; oo comb, 25 per cent; 



absent, 25 per cent. 



Mating oo comb and Y comb should give the family types : 



1. Y comb, 100 per cent. 



2. Y comb, 50 per cent; oo comb, 50 per cent. 



3. Y comb, 50 per cent; I comb, 50 per cent. 



4. Y comb, 25 per cent; oo comb, 25 per cent; I comb, 2h per cent; no 



comb, 25 per cent. 



comb and oo comb should give the following types of f amiUes : 



Y comb, 100 per cent. 

 I comb, 100 per cent. 



Y comb, 50 per cent; oo comb, 50 per cent. 

 I comb, 50 per cent; no comb, 50 per cent. 



Now, what do the facts say as to the relative value of these three 

 hjrpotheses? Abundant statistics give a clear answer. In the first place, the 

 progeny of two Y-combed Fj parents is found to show the following distri- 

 bution of comb types: Y comb 471, or 47.3 per cent; I comb 289, or 29.0 

 per cent ; oo comb 226, or 22.7 per cent ; and no comb 10, or 1 per cent. 

 The presence of no comb in Fj speaks for the second hypothesis, but instead 

 of the 6.25 per cent combless expected on that hypothesis only 1 per cent 

 appears. There is no close accord with expectation on the second hypothesis. 



Coming now to the F3 progeny of two Y-combed parents, we get the 

 distribution of famiUes shown in table 3. 



Table 3. 



Finally, 



1. 

 2. 

 3. 

 4. 



An examination of these families shows not one composed exclusively 

 of Y-combed individuals nor those (of significant size) containing Y-combed 

 and l-combed or oo-combed individuals exclusively, much less in the 



