GEOWTH AND MORPHOLOGICAL CHARACTERS. 69 



Summing up the data bearing on average skeletal dimensions in C. 

 rufescens, C. porcellus, and three generations of hybrids, we may say that : 



(1) C. rufescens is smaller than C. porcellus. 



(2) The J wild hybrids were larger and more vigorous than either 

 parent species. 



(3) The I wild were smaller than the ^ wild and possibly showed a 

 general tendency to be smaller than the guinesir-pig, particularly in the 

 male sex. 



(4) The I wild and the guinea-pig were of the same size and practi- 

 cally indistinguishable. 



Coefficients of Vaeiabilitt of Dimensions. 



C. rufescens is specifically smaller than C porcellus. We do not 

 know whether the smaller species lacks factors for size, or whether it 

 has factors inhibiting growth, or whether there are any "factors" 

 involved at all. If we suppose that the difference in size is due to 

 one or many completely dominant factors, then the Fi should be like 

 the dominant parent; and crossing the Fi and F2 generations back to 

 this parent should give only the dominant form. But if we suppose 

 the difference to be due to multiple, incompletely dominant factors, 

 then the Fi generation should be a blend, and the F2 should show an 

 increased variability, as was shown on pages 50-51. It has been 

 pointed out by East (1910) that "as dominance becomes less and less 

 complete, the Mendelian classes vary more and more from the formula 

 (3+1)° and approach the normal curve, with a regular gradation of 

 individuals on each side of the mode." In order to ascertain whether 

 the hybrids were more variable than the parents, the coefficients of 

 variability were calculated (see tables 67 and 68). 



The variability of C. rufescens is unknown. The number of adult 

 skeletons available in our own experiments was far too small to use 

 as data. If we analogize with the tame C. porcellus, it is probable 

 that the wild is not very variable. 



The coefficients of variability of the guinea-pig were extremely small. 

 The highest coefficient of any dimension in either sex was only 3.73 per 

 cent ±0.20 (measurement 9, table 67). Only 6 of the 32 coefficients 

 were 3 per cent or more. Furthermore, they were very uniform, for 

 the lowest was exactly 2.00 per cent ±0.12; and they range, therefore, 

 from 2 per cent to 3.73 per cent. Compared with the parent stock 

 used in experiments on maize (ShuU 1910, East and Hayes 1911), or 

 with the stock used in experiments on gourds and beans (Emerson 

 1910), these coefficients in the guinea-pig are very small. In the case 

 of maize, the coefficients of variabihty of the parents were sometimes 

 as large as 14 per cent. Emerson gave a coefficient of variability as 

 26.9 per cent for the shape of one parent (scaUop) in summer squashes. 

 This in no way reflects on the results and interpretations of these 



