392 THE GRAMMAR OF SCIENCE 



segmentation a difficult, although I do not say impossible 

 task. In counting the petals of marsh marigolds also I 

 have found, perhaps, once or twice per thousand heads, 

 an extra petal growing out of the stalk about an inch 

 below the true flower. Abnormal variations thus un- 

 doubtedly exist ; how many of them exist in such 

 numbers as to be capable of giving rise to a new variety, 

 Jiow Many are indeed fertile at all} are points which must 

 be fully determined before it can be asserted that evolu- 

 tion is largely the product of such abnormal variation. 

 Every variation, unless frequent, very advantageous or very 

 fertile, is sure to be swamped. The frequency and fertility 

 of normal variations are easily ascertainable, but these 

 are matters wherein the statistics of abnormal variation 

 are at present rather to seek. 



8 7. — Correlation 



Hitherto we have been discussing the type and vari- 

 ation about the type in the case of a single character. 

 We have seen that the racial variation is greater than the 

 individual variation, that capsules on the same poppy plant 

 are more alike to each other than they are to the capsules 

 of a second plant, or the leaves of one beech tree to each 

 other than to those of a second beech tree. This resem- 

 blance of the like organs of the same individual is a 

 special case of correlation, and we now want a quantitative 

 measure of such correlation. The answer is again a 

 question of probability. If I pluck two beech leaves off 

 a tree, and one has 1 8 veins, what is the most probable 

 number of veins upon the other? It will more nearly 

 approach 1 8, be more closely associated with that 

 number, than if it had been gathered from another tree. 

 If I pick a capsule of 7 stigmatic bands from one poppy 

 and one of 1 3 stigmatic bands from a second, then if 



1 The infertility of many abnormal variations is also a fairly general rule. 

 If they are, as appears frequently the case, e.g. in human giants and dwarfs, due 

 to pathological causes, i.e. diseased conditions, this infertility is perhaps to 

 be expected. 



