96 GENETIC STUDIES ON A CAVY SPECIES CROSS. 



tests rather than in the total of all tests. The series of percentages of 

 ultimate recessives expected on the basis of eight factors (see table 77) is: 



0.00 0.39 10.01 34.36 59.67 77.57 88.16 



One must admit that there is a remarkable similarity between these 

 three series for the first 5 hybrid generations at least — such a close 

 resemblance that one wonders whether it is chance coincidence or 

 whether there actually were 8 allelomorphic pairs involved, such that 

 the ultimate recessives in each generation segregate out with many 

 motile sperm. One would be forced to conclude that further factors 

 were necessary to give real fertility in addition to mere numbers and 

 motility, for it was shown that males with many motile sperm were 

 not necessarily fertile. The great range of possibilities between no 

 sperm and all motile sperm would, on this h>"pothesis, be due to recom- 

 binations of factors. Individuals homozygous in 6 or 7 recessive factors 

 would be almost fertile, for they would have segregated out most of 

 the disturbing "wild chromosomes" and have replaced them with 

 homologous pairs entirely from the tame source. 



Such an hj^pothesis is suggestive and alluring, but other critical 

 considerations are necessary. The probable errors for the percentages 

 were calculated, but are not given. I am indebted to Dr. H. L. Rietz 

 for valuable suggestions regarding these. They would be extremely 

 difficult to handle and very misleading. The probable error of any 

 generation would have to be calculated on the supposition that the 

 females of the preceding generation were normally distributed, or 

 else one would have to take the error of all preceding generations 

 into account. It is logically impossible to suppose that the females 

 of any generation (except Fi) could have been normally distributed. 

 On this hypothesis we would suppose that the wild and tame had 

 8 factors or chromosomes which were incompatible in the Fi males, 

 and this led to disturbances in the maturation of the sperm, but did 

 not affect the females. We might represent the factors from the wild 

 as AABBCCDDEEFFGGHH, and those from the tame as aabbccdd- 

 eeffgghh. The ^ wild would be Aa Bb Cc Dd Ee Ff Gg Hh. The fertile 

 Fi females should then produce 256 kinds of gametes, but only one of 

 these, abcdefgh, would have segregated out the disturbing elements 

 from the wild. Now, when this gamete met its mate from the tame, 

 also abcdefgh, it should have given fertility in the F2 males. But the 

 expectation of this combination based on random sampling is 1 in 256. 

 The number of F2 males (22) actually procured was far too small to 

 expect an ultimate fertile recessive male. One would, however, expect 

 recombinations which had eliminated some of the disturbing elements. 

 Such were actually obtained, for 2 F2 males showed a few deformed, 

 immotile sperm. (See tables 74, 75.) If the ultimate recessive, fertile 

 males actually lacked all disturbing elements from the wild, then in 



