CH. viii] DIFFERENTIATION 67 



which seem to point to the supposition that differentiation gives 

 a more correct picture of the direction of movement of evolu- 

 tionary change than does natural selection, even though we have 

 no clear vision of the mechanism that was involved in making the 

 changes that occurred. 



The family is supposed to have arisen by some well-marked 

 and sudden mutation (or conceivably a series of smaller ones, 

 probably at close intervals), which would at one stroke change 

 two or more characters and pass the line of mutual sterility that 

 commonly divides species from one another. As the characters 

 that divide the families are, after all, not so very numerous 

 (cf. Appendix I) each family must take a different combination, 

 sometimes taking one of a given pair, sometimes the other, and 

 in every kind of mixture. Many families, for example, have 

 alternate rather than opposite leaves, or superior rather than 

 inferior ovary, but only the Cruciferae have alternate exstipulate 

 leaves, bractless racemes of ^ regular flowers, sepals in two whorls 

 of two, four petals, two short and four long stamens, superior 

 ovary of two carpels, unilocular with replum, a pod-like fruit, 

 and exalbuminous seeds. As the characters run in contrasted 

 pairs (or triads), we have no information as to whether there is 

 any advantage in one side rather than the other, or in either as 

 against any possible intermediate, or indeed that any has any 

 adaptational value. There is thus no evidence to show in which 

 direction evolution moved, and we are perfectly free to select 

 that for which we think that the evidence is better. It is this 

 evidence, or rather, some of it, which we propose to bring forward 

 below. 



Nor can we say with any likelihood of accuracy that the 

 change indicated in any one pair is larger than that in another. 

 Is it a greater change from two cotyledons to one than from 

 alternate to opposite leaves? We do not know; all we have to go 

 upon is that the latter is much more common. With one or two 

 rare exceptions, there is no difficulty in supposing all Mono- 

 cotyledons to have descended from at most a few different 

 ancestors, whilst one may find alternate and opposite leaves side 

 by side in many cases of allied genera or species. There is nothing 

 inherently absurd in the idea that a family might be founded by 

 a single mutation. 



About 1902 the writer became a convert to the theory of 

 mutation, but it seemed to him completely illogical to insist that 

 mutation could only be very small, when before us, in every 



5-2 



