258 
On Inheritance of Coat-Colour in the Greyhound 
Grandparental correlations with regard to change of sex : 
Mean value no change of sex : -350. 
Mean value one change of sex : -336. 
Mean value two changes of sex : "306. 
It is further true for the selection in record values of the parental corre- 
lations *. Thus we are inclined to think that there really exists a weakening of 
inheritance with change of sex in the case of the greyhound, but that it is so small 
that it is liable to be screened by the probable errors of the results. 
On the whole, for direct inheritance of pigmentation, the greyhound falls quite 
well into the system formerly found for man and horse. 
(7) Collateral Reseinblance in the First Degree. From the material available 
on Mr Howard Collins' schedules we could only obtain the degree of resemblaiiee 
between siblings of the same litter. There was no possibility of getting siblings 
from different litters except by appeal to the stud-books with their selection for 
record. Since selection for record depends on the personal equation of the 
breeder, and the points he selects may to some extent involve coat-colour, the 
tendency accordingly ought to be to emphasise rather than reduce the fraternal 
correlation. We find, however (Table VIII), that when we compare the un- 
selected record of the schedules for the same litter with the selected record of 
the stud-books for different litters that the values are much reduced, i.e. from "676 
to '529 on the average. Our first idea was that this great reduction must be due 
to the fact that we are in one case considering siblings from the same and in 
the other case from different litters, or that there must be a sensible difference 
in the degree of resemblance in the two cases. This was opposed to our experience 
in the case of Basset Hounds f. To confirm the supposition in the case of the 
greyhound, we worked out from the stud-books all three sibling relationships for 
the case of siblings from the same litter. To our great surprise the result was 
only slightly in excess of the result obtained for siblings from different litters. 
In other words, for the greyhound if the same material was used in both cases, 
the conclusion reached for the Basset Hound seemed to apply, i.e. there was no 
very marked difference in fraternal resemblance when the siblings were and were 
not from the same litter. We were thus left in considerable doubt as to what 
interpretation could be put on the fact that fraternal correlation between members 
of the same litter when found from the schedules and from the stud-books was 
so markedly different. It seemed difficult to believe that it could be due to 
selection for record, because it appeared to us that this ought, as far as it 
influenced colour, to tend to increase the correlation. On examining our totals 
for the two cases we noted one point which might conceivably throw light on the 
matter. In the schedules we had 651 pairs of parents, giving rise to 7484 pairs 
* Tlie means for nnselccted and corrected selected values of parental correlation give -539 for no 
change of sex and -524 for one change of sex, a slight difference, 
t RoijaJ Soc. Proc. Vol. GG, p. 158. 
