362 STUDIES, SCIENTIFIC AND SOCIAL chap. 



unchanged through a large proportion of the period com- 

 prised in the geological record. The four limbs of verte- 

 brates were already established in the fishes of the 

 Devonian period, as were the four wings and six legs of 

 true insects in the cockroaches and archaic orthoptera of 

 the Carboniferous : and almost all subsequent changes 

 have resulted from modifications of these early types. 

 The earliest mammals of which we have sufficient know- 

 ledge have the typical five-toed feet, and the earliest birds 

 appear to have had the same progressive series of toe- 

 joints as now prevails. 



We are thus irresistibly led to the conclusion that, 

 among all the possible forms of variation now occurring, 

 those affecting the number of important serial parts 

 among higher organisms, are those which have the least 

 possible relation, to whatever modification of species may 

 noio be going on around us, or which has been going on 

 during a large portion of geological time. Yet it is to 

 variations of this nature, a large proportion of which are 

 mere malformations or monstrosities, that the bulky and 

 learned volume we are discussing has been devoted. The 

 author of this book puts forward these malformations and 

 irregularities, mixed up with a proportion of normal 

 variations, under the misleading name of " Discontinuous 

 Variations," as if they were something new, and had been 

 ignorantly overlooked by Darwin and his followers ; and 

 he loses no opportunity of telling us how important he 

 thinks they are, what difficulties they enable us to over- 

 come, and how they are the beginnings of the establish- 

 ment of a sure base for the attack on the problems of 

 evolution. In so doing he has entirely failed to grasp 

 the essential features which characterise at least 99 per 

 cent, of existing species, which are, slight differences from 

 their allies in size, form, proportions, or colour of the 

 various parts or organs, with corresponding differences of 

 function and habits, combined with a wonderful amount 

 of stability in the numerical relations of serial parts, ex- 

 tending sometimes only to genera, but more usually to 

 families, tribes, orders, or even to whole classes of the 



