.3,,. BIOLOGICAL THEORIES. 583 



so narrowly and in so many directions that none of the innumerable 

 essential characters which depend upon it shall vary appreciably, 

 can yet itself remain variable to any large extent. The variation in 

 the constitution of a stock must, in fact, be extraordinarily slight if no 

 variation shall be discoverable in any of the innumerable characters 

 which that constitution determines in the new individuals. This 

 constancy of stock-constitution, this constancy of the new individual 

 constitutions itself is the guarantee of that constancy in the 

 recurrence of the " like causes " to which I referred in the 

 previous paper. The efficient cause of heredity which we set out 

 to seek is found. It consists in a stock-constitution whose constancy 

 has been gradually brought near to completeness by the indirect 

 effect of natural selection upon it. The struggle for existence has 

 eliminated all stocks varying beyond certain limits. The variation, 

 whatever it may have been, has at every stage been the necessary 

 outcome of the stock-constitution, and the elimination of stocks whose 

 new individuals varied, has been at the same time an elimination of 

 stocks of unfit constitution, that is, of unstable constitution. 



Such a stock-constitution having been once established, it is 

 obvious that, except through accident or other disturbing cause, the 

 struggle for existence might^ well be suspended, even for many 

 generations, and yet no marked variation might appear. On the 

 other hand, a radical change in the environment capable of affecting 

 the constitution directly {i.e., not through natural selection) might in 

 one generation blot out the limits of the species. A species whose 

 specific constitution had, through long selection in marine waters, 

 been rendered constant under constant external conditions, might 

 well undergo a sudden or very rapid change on transference to fresh 

 water. Whether the new constitution would be such as to lead to 

 the production of new individuals capable of life and multiplication 

 or not would be, in a sense, a matter of chance. The new individuals, 

 if capable of living, might be unlike the older ones in form, and their new 

 constitution might be such as to lead to wide variation in succeeding 

 generations. If so, a new process of selection through many genera- 

 tions could alone render it again constant, and the new set of constant 

 characters would be determined largely but indirectly by the new envi- 

 ronment. A radical change of diet having altered the chemical 

 composition of the circulatory fluid in the body of one of the higher 

 metazoa, it is doubtful whether this would lead to such a change 

 in the tissues nourishing the ova in process of formation as to 

 involve the production of new ova unlike those of former generations ; 

 while we know not whether the new ova would be capable of develop- 

 ment at all, and are ignorant of the form and structure of the new 

 individuals if the ova did develop. It would tlms be as 

 gratuitous on my part to say dogmatically that increased variability 

 in future generations would be tiie result, as it is for the coutinuitatists 

 to dogmatically deny that the germ-plasm would be affected. It at 



