A. Barrington, a. Lee and K. Pearson 
261 
data fur which are given in the memoirs cited below*. The results are given in 
Table X. We see that we have here a very considerable range of values, that 
the Basset Hounds and the Stud-book Greyhounds agree well with the mean 
value for manf, but that the Thoroughbred Horses and Schedule Greyhounds 
exceed these values. The only common factor that we see to explain this 
difference lies possibly in the earlier age of record in the last two cases. The hair- 
colour changes with growth in the case of children |, but there is probably greater 
uniformity in the hair-colour of siblings in the first year of life than at any later 
age — our records are for children in the bulk between 10 and 14 years old — and 
this is a stage hardly comparable with that of yearlings and puppies, as far as 
dcfiniteness of hair-colour is concerned §. 
It will be seen from these results that the suggestion made in Biometrika, 
Vol. II. p. 391, that the high value of the greyhound fraternal correlation is due to 
the action of individual prepotency on members of the same litter, is not confirmed. 
We consider that in the case of the dog we may safely say that there is no marked 
difference between the degree of resemblance of siblings of the same and of different 
litters. After considerable labour we found it impossible to collect enough data from 
the stud-books for horses to test the same point on twin and non-twin foals of the 
same parents. Both twins rarely survive to be recorded, and twin foals themselves 
are infrequent. The point can and will be dealt with on the basis of Darbi- 
shire's material for mice. So far, however, even if a co-uterine environment does 
increase the resemblance of brethren from the same litter in some species, it 
certainly does not appear to do so for all, and at any rate the high value of 
fraternal correlation in the case of the greyhound is not due to this source. It 
is peculiar to the nature of the record. We believe as in the case of the horse 
it may possibly be due to a mere temporary influence on the somatic characters of 
the embryonic co-environment. This disappears when the record of the colour is 
taken at a later stage. Should this view be the correct one the stud-book records 
would after all turn out to be of more value than the schedule records. We 
should then conclude that judged from pigmentation the inheritance in grey- 
hounds is given by : 
•466 for parental correlation, and -529 for fraternal correlation (different litters). 
These numbers would be in excellent agreement with the corresponding mean 
values '460 and •519 found for physical measurements in man||. 
Phil. Trans. Vol. 195, A, pp. 93 and 10(3. Biometrika, Vol. iii. pp. 154 et scj. The hair-colour 
returns have been taken from school children ; these children give -53 for eye-colour — a result close to 
Greyhound and Basset Hound values for different litters. 
t The mean value in the case of man for 9 series of physical measurements in the adult is -Bid 
(Biometrika, Vol. ii. p. 519) and for 12 series of head-measurements in children is -513 {Biometrika, 
Vol. III. p. 140). 
X The correlation is not nearly so great as some have supposed, but of this more on another occasion. 
§ In cattle the like difficulty arises. Some are recorded as mere calves and the colour changes a good 
deal after losing the first coat. Thus a bull calf recorded as a ' fawn ' may be ultimately a ' dark fawn ' 
or ' red ' bull. 
II Biometrika, Vol. ii. pp. 378 and 390. 
