Voiv. 6, 1920 
GENETICS: S. WRIGHT 
327 
both cases being almost exactly intermediate. Sex also makes no sig- 
nificant difference in variability. In the case of the means, however, 
there is in all cases the sex difference already mentioned. The mean 
and standard deviations are measured on a scale in which unity is the stand- 
ard deviation of color potentialities for areas within the skin of a single 
average guinea-pig. It is interesting to find that even the random-bred 
guinea-pigs differ less among themselves in average color potentiality than 
do the spearate skin areas of a single animal. 
TABLE 2 
Correlations between Parents, between Parent and Offspring and between 
Litter Mates in Random-Bred Stock and Inbred Family No. 35, 
DURING 1916, 1917 and 1918 
EXPERIMENT B (random-bred) 
FAMILY 35 (inbred) 
No. 
Correlation 
No. 
Correlation 
Sire-Dam 
105 
+0.019 ± 0.066 
73 
+0.029 ± 0.079 
Sire-Son 
492 
+0.231 ± 0.029 
235 
+0.013 =t 0.044 
484 
+0.194 ± 0.030 
236 
+0.082 =t 0.044 
498 
+0.251 =t 0.028 
235 
+0.042 ^ 0.044 
Dam- Daughter 
488 
+0.165 ± 0.030 
236 
—0.080 ± 0.044 
Average, Parent-OfFspring 
1962 
+0.211 ± 0.015 
942 
+0.014 ± 0.022 
Brother-Brother 
390 
+0.219 =fc 0.033 
182 
+0.090 ± 0.050 
Brother-Sister 
437 
+0.228 =t 0.031 
203 
+0.062 =b 0.047 
406 
+0.180 ± 0.032 
194 
+0.064 ± 0.048 
Average, Litter Mates 
1233 
+0.214 ± 0.018 
579 
+0.069 =t 0.028 
The matings in the correlation between sire and dam are each weighted by the num- 
ber of offspring. The probable errors in this case are based merely on the number of 
matings. The probable errors in the other cases are based on the number of entries 
in the tables. Owing to repetition of individuals, they are probably somewhat too 
small. The correlation between litter mates in experiment B is based on 894 individuals 
in litters in which two or more were graded. The probable error based on this number 
is =1=0.022 instead of =t 0.018 as given in the table. Similarly there were 426 individuals 
in such litters in family 35, giving a probable error of ±0.031 instead of =1= 0.028. 
The negligible correlations between the parents (+0.019 in the control 
stock and +0.029 in the inbred family) show that matings actually were 
made at random as far as color is concerned. In the control stock, both 
the average correlation between parent and offspring (+0.211) and that 
between litter mates (+0.214) are unquestionably significant. In the 
inbred family, there is virtually no correlation between parent and off- 
spring (0.014) while that between litter mates is of doubtful significance 
(+0.069). 
There is evidently very little genetic variability left in family S5, but is 
a good deal in the control stock. A closer analysis can be made by con- 
