Karl Pearson 
405 
corpuscles were taken in either case, we should screen the true homotyposis owing 
to this change of corpuscle size with growth. 
We can still further illustrate this point by finding the correlation between 
total length, that is ' body ' + ' tail,' and the length of the corpuscle. We have 
r =— '23, showing a somewhat less, but still quite sensible, negative correlation, 
or the size of the corpuscle decreases with increasing total body length. I now 
proceeded to correlate the total length with the body length, and found r = -93. 
Now it is a noteworthy result of these numbers that -93 x — -25 = — 23, or well 
within the limits of random sampling, we have : 
Correlation of body length and cell length x correlation of total length 
and body length = correlation of total length and cell length. 
That is to say, the partial correlation coefficient of total length on cell length 
for a constant body length is zero. We may therefore infer that the length of 
the tail has little or no influence on the length of the corpuscles, which are affected 
only by the growth of the body. 
In Dr Warren's measurements, one ocular unit for the length or breadth of 
the corpuscle = '00268 mm., and the mean length of 1775 blood corpuscles of 
tadpoles was 7"85o4 working units = •0211 mm. The late Professor Weldon 
measured for me, in 1901, 50 red blood corpuscles from each of 20 frogs. The 
method of procedure was somewhat different to that of Dr Warren, the frog 
being killed with chloroform and the blood taken from the heart ; the frogs were 
from the neighbourhood of Oxford, whereas the tadpoles came from Hendon. 
Professor Weldon also drew the corpuscles, using a camera lucida. The mean 
cell length of these 1000 blood corpuscles of the adult frog was "0256 mm. 
Comparing this value with that found for the tadpole at various stages as well 
as the mean value, it would seem highly probable that notwithstanding change 
of local race, the data point to the increase of size of the blood corpuscle of the 
frog with or after metamorphosis. Dr Warren also measured the length of the blood 
corpuscle in 25 corpuscles taken from each of 29 " toadpolls," * or tadpoles of Bufo 
vulgaris taken from Kew lake. The mean value found was '0212, which is in very 
close agreement with the mean result for the tadpoles from R. temporaria. 
The following table sums up our results on corpuscle length, 
TABLE II. 
Length of Red Blood Corpuscles in mm. 
Species 
Number 
Mean 
Standard Deviation 
Coefficient of 
Variation 
R. temporaria (adult) 
1000 
•02560+ -00004 
•00192+ -0000.3 
7-50+ -11 
H. temporaria (tadpole) 
1775 
•02105 + -00004 
-00246 + -00003 
11-69+ -13 
B. vidgaris (toadpoll) 
725 
-02118+ -00006 
-00251+ -00004 
11-85+ -21 
* I should like to reintroduce the original of tadpole, i.e. toad-poll, as a convenient popular term for 
the tadpole of B. vulgaris; tadpole being reserved for R. temporaria. 
