410 
Study of Blood Corjniscles of Tadpole 
(5) Lastly an attempt was made to ascertain to what extent the blood 
corpuscles of an individual are characteristic of that individual. The difficulties 
here are undoubtedly considerable. At the outset of the investigation it was not 
anticipated that the red blood corpuscles would change in size with the growth of 
the individual, and it might well have been expected that if this should prove 
to be the case, the index would still be practically free from the growth factor. 
In the next place a good deal more inquiry is needed as to the influence of the 
locus of withdrawal and the treatment of the blood corpuscles, before very definite 
conclusions are drawn. 50 blood corpuscles drawn from the heart after obliterating 
the central nervous system in the case of Frog No. 3, when examined without any 
reagents, gave a value of the mean length = "0262 mm. A second series of 50 blood 
corpuscles from the heart mixed with normal salt solution and exposed to vapour of 
chloroform for 20 to 30 seconds before measurement, gave mean length '0260 mm. 
The mean length of 1000 corpuscles of adult frogs 
= -0256 + -00004. 
The error of a random sample of 50 would be about '0002. It cannot therefore 
be asserted that the difference of the two samples of the same blood is due to 
treatment ; it is just what might be expected in two random samples. Further 
the two sample means appear to be significantly different from the mean of the 
general population of corpuscles. Samples of at least 50 to 100 (and better the 
latter) ought to be taken, and uniformity observed in time of measurement after 
extraction and in actual treatment. The immense amount of labour involved is, 
however, a very serious consideration in such large sampling. 
The problem of possible heterogeneity must also be borne in mind, but I think 
we may definitely assert that the blood corpuscles here dealt with certainly did 
not belong to any two well-marked classes, they failed to give real solutions in 
all attempts at analysis. It is of course conceivable, in fact not improbable, that 
the red corpuscles as a population are in various stages of growth, and till more is 
known as to the place and manner of production of these corpuscles it would be 
profitless to hazard any guess as to how this may affect the variability and in- 
dividuality of the population. This may account in part for the relatively low values 
found for homotyposis in these free cells as compared with my unpublished data 
for homotyposis in fixed cells. 
Emphasising these preliminary points, 1 will put together the results so far 
reached for the homotyposis or degree of resemblance of blood corpuscles in the 
same individual. 
The probable errors are based on the number of corpuscles, not on the number 
of individuals or of pairs dealt with. 
It will be obvious from this table that there is a sensible degree of individuality 
in the corpuscles of the same individual, the raw values of the coefficients are 
sufficient to demonstrate this. Further, by correcting for change in the blood 
