92 Outline of Genetics 



that had fluctuated continually, never having been brought to as 

 small a variability as have most other characters. The question 

 was raised whether Castle's assumption that this variability was 

 merely due to fluctuation was altogether justified. Might not 

 the variability have been due to varying doses of cumulative 

 factors? Suppose for the moment that this were the case; it 

 would not be surprising that Castle could develop two diverse 

 strains by selection, for selection would result in piling up the 

 cumulative factors in one direction or the other. Castle's 

 rejoinder was that if this were a cumulative factor situation, why 

 had none of the extremes appeared in the non-selected stock, 

 which instead bred approximately true ? The answer was made 

 that the extremes did not appear in the pure bred stock merely 

 because of the mathematical limitations. If one is deahng with 

 six cumulative factors, and the so-called pure stock has an inter- 

 mediate number of doses, there could not be much chance of 

 getting out the extremes in a single generation. It would be 

 necessary to secure over 4000 progeny to have an even chance of 

 getting one such extreme; or about 50 progeny to get anything 

 that would very noticeably approach the extreme. It would seem, 

 therefore, that Castle's chances to determine this would be very 

 small. Rats certainly do not produce 4000 progeny in a single gen- 

 eration; in fact, they produce much less than 50; therefore Cas- 

 tle's "pure stock " went on in the intermediate condition, and only 

 by selection could he pile up the factors and reach either extreme. 

 Thus far the explanation seemed satisfactory. Castle 

 showed, however, that the coat pattern condition behaved in 

 crosses as a simple Mendelian unit; that is, it did not split up into 

 complex ratios, but came out as a recessive in a regular 3 : i ratio. 

 This really involved no difliculty. Suppose Castle crosses one 

 of his pure strain rats having the hooded character with another 

 race that has some pattern character that conceals the hooded 

 character. If this other character is a .simple Mendelian one, the 

 result of the cross would be the ordinary monohybrid ratio; that 

 is, in the F2 generation from such a cross the ratio of hooded to 

 non-hooded (with the " hood-conceahng " character) would be 1:3, 

 which, in fact, is exactly what Castle got. At the same time, 

 the amount of pigmentation, determined by numerous cumulative 



