THE PLACE OF ORIGIN OF VARIATIONS 



103 



single fertilized ovum, are closely alike, while those which arise 

 from different ova (even in the case of peas in the same pod) 

 differ as much, or almost as much, as offspring born at suc- 

 cessive births or seasons. This indicates that the part of the 

 germ-tract at which variations normally arise is the last, or one of 

 the last, cell-divisions that precede the ripening of the germ-cells ; 

 for if variations occurred earlier, not only identical twins, but whole 

 batches of offspring would tend to be identical, especially when 

 the reproduction was parthenogenetic. 



1 66. The following diagram may help to elucidate the 

 argument. 



A 



AAAMAAMAAUAAMAA: 



-Z 



167. Suppose A is a fertilized ovum from which the cells of the 

 germ-tract B, C, D . . . Y, Z, are derived. Z represents the 

 descendant germs. Granting that variations arise at a particular 

 point in the tract, then, if they arise at A, B 1 will differ from B 2 , 

 as will their descendants ; but all the descendants of B 1 will be 

 alike, as will be those of B 2 . If variations arise at B, there will 

 be four batches of identical descendants. If at C, eight. And so 

 on. But if the point at which variations arise is at Y, there will 

 be no identical germs, except (especially in the case of complex 

 organisms) as a very unlikely coincidence. On the other hand 

 if variations arose all along the germ-tract, then, though all the 

 germs would tend to differ, the lengthening of the germ tracts, as 

 by pseudo-generations, would tend to increase variability, which, 

 as we see, does not happen. 



1 68. Now, the evidence is that all these laws (laws in the 

 scientific sense of uniformities in the sequences of events), unlike 

 the laws of physics, have exceptions. Thus somatic cells some- 

 times vary spontaneously, as when a branch bearing nectarines 

 appears on a peach tree. Again, when a cancer appears in the 

 tissues of a man, the variation seems frequently to be caused, 



