Inheritance of Acquired Characters 199 



tion, but by the divisions of a single cell, which is formed by 

 the union of two half cells, one from each parent. This was 

 illustrated in Figure 5, on page 21. 



It is clear that if any of the later (secondary) hereditary 

 characters are to become modified, this must be accomplished 

 by some modification of the primary hereditary characters, 

 -those passed on bodily from parent to offspring. For 

 if two germ cells are exactly alike, the characters inherited 

 through them are bound to be alike; two individuals that 

 are to differ in their later hereditary characters must be 

 diverse in their primary hereditary characters. If environ- 

 mental agencies are to produce diversities that are to be 

 hereditary, they must change the germ cells, in the particular 

 way required to bring about the observed later changes in 

 the body ; for the body is derived from these germ cells. 



This has always been the theoretical difficulty with the 

 "inheritance of acquired characters," if we mean by that 

 abused expression the inheritance of modifications produced 

 directly on the body by the outer world. If the form of the 

 hand is changed by certain outward conditions, how is that 

 change to modify the primary hereditary characters in the 

 germ cells, which are not directly touched by the given 

 outer conditions, in just such a way as to cause them to 

 produce the same new form of hand in the next generation? 

 Theoretical difficulties of this sort of course must not 

 be allowed to stand in the way of our recognizing how nature 

 actually does operate, if she does indeed operate without 

 regard to these difficulties. There is nothing so little worthy 

 of confidence in science as assertions that particular events, 

 not yet observed, are impossible ; such propositions have been 

 falsified a thousand times, and the careful man of science, 

 will not permit his researches to be guided by them. But in 

 this case the great weight of evidence thus far is that this 



