VARIATION AND MENDELIAN INHERITANCE 385 



recent systematists to the subspecies rubidus Osgood. Those 

 from CaUstoga, Berkeley, and La Jolla are assigned to gambeli 

 (Baird), while ones from Victorville are assigned to the charac- 

 teristically desert race sonoriensis (Le Conte). 



In table 1 we find the mean values given for most of the char- 

 acters which have been subjected to measurement.^^ It must 

 be explained, however, that the mean values here given have, for 

 most characters, been 'corrected' in such a way as to be com- 

 parable with one another. This was rendered necessary by the 

 fact that the mean body length (total length minus tail length) 

 differed considerably among the various collections, owing largely 

 to the inclusion of differing numbers of immature individuals. ^^ 

 Since most of the characters here considered are rather strongly 

 correlated with body length, their mean values in these different 

 sets would obviously have not been directly comparable.^'' 



The figures (or most of them) have accordingly been corrected 

 for each series in such a way as to give their most probable value 

 had the mean body length of the series in question been 90 mm. 

 This was accomplished, I need hardly say, by the use of the 

 so-called regression coefficient. ^^ As a matter of fact, the cor- 

 rections which were applied were in most cases small in com- 

 parison with the differences between the various races. They 

 were largest in the Fort Bragg series, which contained a greater 

 proportion of immature animals than did any of the others. 



1* Weight and skull width have been omitted, for reasons which need not here 

 be discussed. Foot-pigmentation has only recently been included among the 

 characters measured, so that figures are not yet available for these races. Body 

 length is not included for reasons stated in the next paragraph. 



15 All animals below 80 mm. in body length were, however, arbitrarily excluded. 



'^ In a more complete presentation of these results, I shall include the original 

 averages, but this does not seem necessary for the present. 



1^ The correction is obtained by the equation x = r — y, in which x represents 



the difference sought between the corrected and obtained values for a given 

 character, r the correlation between the character in question and body length, 

 <r and crij the standard deviations for this character and for body length, re- 

 spectively, and y the difference between the standard value (90 mm.) and the 

 actual mean body length of the lot in question. Of course, all these standard 

 deviations and correlation coeflBcients had to be first computed for each race 

 and sex separately. 



