K. Pearson 
341 
I might with perfect justification have excluded several cases from my Table. 
For example I gave the correlation between cephalic indices of brothers to be -3790 
and between tempers to be -3167. I knew these were very low and based on 
unreliable material, but I gave everything I had at that date. From far larger 
and more reliable material I know now that the correlation of cephalic index 
between brothers is -4861 and of temper "5068. I contented myself with entering 
the older values with due words of warning. 
But, -perhaps, Mr Bateson means that I have excluded things to which no 
reference is made in the paper at all ? He writes : 
" Yet another and even clearer illustration. The two claws of a crab arc a 
pair of homotypes. Their homotypic correlation in respect of any character, 
length, for example, might be determined " (p. 199). 
Well, I do know the correlation between several characters which occur in the 
claws of crabs. It never once entered my mind to include them. I know further 
the correlation between right and left corresponding bones of the skeleton in man 
for a considerable number of races — for the skull, the hand, the chief long bones 
etc., etc., perhaps in all for forty or fifty cases. If such things are homotypes 
according to Mr Bateson I have been guilty of excluding correlations which would 
bring up the homotypic average to somewhere between "9 and 1 ! But these are 
not homotypes in my sense of the word, which may perhaps aid Mr Bateson to see 
that his " principle of symmetry " is not synonymous with honiotyposis. The 
correlation between right and left member is in my sense of the word orgmiic and 
not homotypic. I term the correlation between two members " organic " when its 
value is wholly or partly determined by the fact that for the welfare of the indivi- 
dual the members must within certain limits " fit," they have a function to 
perform in common and their mutual relationship has been controlled during its 
evolutionary development by the existence of this common end. In honiotyposis 
this purpose, i.e. the performance of a common function as controlling the relation- 
ship, has either no existence or is insensible — the mutual relationship is due to the 
individuality of the producer, and is practically uncontrolled by the importance to 
the individual of the homotypes performing at some time related but really diffe- 
rentiated parts in a common function. It is not vitally important to a beech tree 
that one leaf having 12 veins upon it another gathered from anywhere-else on the 
tree should also have 12 or nearly 12. But it is important to a man that if he has 
one femur of 456 mm. the other femur should be within a few millimetres of the 
same length*. Thus in right and left-hand members we have a differentiation in 
function, one member could not possibly replace the other, and if the differentiation 
were not visible at once to the biologist, it would be evident at once to the biome- 
trician in the angle means. Such cases therefore are very fully excluded by the 
definition of homotypes, and I confess I do not understand how Mr Bateson finds 
* In an early stage of evolution possibly all correlation was nearly homotypic, but selection of 
"fitting" individuals would soon emphasise the "organic " factor, i.e. it would reduce the variability of 
the array associated with individuals of a given size. 
