62 GENETIC STUDIES ON A CAVY SPECIES CROSS. 



The Coefficients of Variability. 



It is indicated on pages 48-55 that a number of recent papers on 

 size-inheritance postulated multiple factors for size with incomplete 

 dominance. According to this theory, a cross between a pure large 

 race and a pure small one would result in a blend, in the absence of 

 disturbing influences such as the vigor of heterozygosis, environment, 

 and the like. If the Fi generation were then crossed inter se, one 

 should obtain an increased coefficient of variability and, with sufficient 

 numbers, recover the parental forms. If, however, the Fi generation 

 were crossed back to either parent, one should obtain a range from the 

 Fi to that parent with the mode in between. The usual method of 

 procedure would be to mate the Fi generation inter se in order to 

 obtain a m.aximum coefficient of variabilitj^ as the best evidence of 

 segregation and recombination of size factors. But this was impossible 

 in these crosses, for the males were entirely sterile. Two alternatives 

 remained, either to cross the Fi females back to the guinea-pig or to 

 the small wild C. rufescens. The latter would have been preferable, 

 but not enough cases were successful to give data of value, hence all 

 results were based on crossing the Fi females back to the guinea-pig. 

 The F2 males were likewise sterile and consequently the F2 females 

 had to be crossed back to the guinea-pig. This meant that conditions 

 made it necessary to resort to the class of matings least advantageous 

 for a study of size-characters. 



The study of the average weights at different ages is quite insufficient 

 to show the complete relation between the size of parents and hybrids, 

 for they do not indicate the dispersion of the individuals from the 

 average of the group considered; or, in other words, averages do not 

 give evidence of segregation and recombination of possible unit factors 

 for size. Therefore, the coefficients of variability of the weights of 

 the parents and hybrids were calculated from the individual smoothed 

 curves for six different ages ranging from 100 days to 380 days (see 

 tables 61 and 62). It must be stated at the outset that the data and 

 results are very unreliable, for the numbers are small, although breeders 

 of mammals must be content with such; and environment affects 

 growth and weights greatly. 



The coefficients of variability for weights of the males and females 

 (tables 61 and 62) give no clear, pronounced evidence that the hybrids 

 of the second generation were more variable than those of the first or 

 than the guinea-pig parent, and hence there is no evidence of segre- 

 gation and recombination of factors. It is true that some classes of 

 hybrids were verj^ slightly more variable than either original parent 

 species, but it is difficult to know whether this was due to real inherent 

 variability or to experimental error. Furthermore, such differences as 

 do obtain are not wholly consistent with an explanation that postulates 



