HOMOTYPOSIS IN THE VEGETABLE KINGDO?^!. 
311 
brothers ; in my own measurements of 1000 families I have taken only two Iwothers 
from each family. We cannot a ‘priori expect the results to come out exactly the same. 
Indeed, we might a priori expect the latter method to give rather higher results 
than the former, for it introduces more individual families, and if we tabulated all 
the brothers of one family only or all the like organs of one organism, we should find 
noticing l)ut a spurious correlation growing indefinitely small as the niunher of 
brotliers or organs was indefinitely increased.Thus I anticipated that the 
Hampden Shirleys would give a lower correlation than the Chelsea Shirleys, the 
wild poppies from the top of the CTilterns a higher value than those from the 
bottom, &c. 
As our series of poppy plants were in no case very large, it was impossilDle to directly 
test this result by taking only a single pair from each plant, even if it had been 
possible after the counting ajid marking to select a pair of capsules at random. I 
accordingly adopted a third or intermediate method. Miss Fawcett had recorded 
her capsules under the letters c, &c., and not grouped them directly under 
the number of the stigmatic bands. Accordingly, it was possible to take a with h, 
c with d, e with /, &c., and thus repeat no single capsules, but get a number of 
pairs sensibly taken at random. Thus large and vigorous plants will give more 
pairs than feeble ones, and yet much fewer than if we take every possible pai]'. 
This series will he referred to as Series (ii.)"'L Its general result confirms the anti¬ 
cipation referred to above. It will be noticed tliat nothing of this kind can he 
done in the case of leaves of trees, where we cannot gather every leaf One great 
advantage of the 2G leaves from every tree is the means it j^rovides of testing the 
accuracy of the lengthy process of tahulation and calculation. Each leaf is used 25 
times, and thus the mean and standard-deviation of the total number of jmirs ought 
to be tbe same as those of the total series of leaves (say, 2600 for 100 trees). This 
valuable method of verification is lost when we do not take tlie same numl)er of 
like organs from each individual. It then becomes of considerable interest to ascer¬ 
tain how far the mean and variability of the total number of organs ol)served are 
identical with the mean and variability of the total numl)er of pairs of organs 
observed. Tliis comparison is made in Sei'ies (v.)'"-^, the mean and varial)ility of the 
4443 capsules being compared witli tliose of tlm corresponding 197,478 pairs. The 
whole of the tahulation and aritlimetic involved in Series (ii.)''-""' is mine, Imt Series (v.)^'*^ 
is again due to Dr. Lee. 
I now proceed to sum up the results obtained from the above seven series. 
I worker! out the theory of tliis some time ago, and may puhlisli it on anotlier occasion. 
•X 
