IV.] FOUNDATION OF A THEORY OF HEREDITY. 201 



true unbroken continuity. The nucleus of the egg-cell directly 

 gives rise to the nuclei of the pole-cells, and there is every 

 reason to believe that the latter receive unchanged a portion 

 of the idioplasm of the former, and with it the tendencies of 

 heredity. But in all other cases the germ-cells arise by division 

 from some of the later embryonic cells, and as these belong to 

 a more advanced ontogenetic stage in the development of the 

 idioplasm, we can only conclude that continuity is maintained, 

 by assuming (as I do) that a small part of the germ-plasm per- 

 sists unchanged during the division of the segmentation nucleus 

 and remains mixed with the idioplasm of a certain series of 

 cells, and that the formation of true germ-cells is brought about 

 at a certain point in the series by the appearance of cells in 

 which the germ-plasm becomes predominant. But if we accept 

 this hypothesis it does not make any difference, theoretically, 

 whether the germ-plasm becomes predominant in the third, 

 tenth, hundredth, or millionth generation of cells. It therefore 

 follows that cases of early separation of the germ-cells afford 

 no proof of a direct persistence of the parent germ-cells in 

 those of the offspring; for a cell the offspring of which become 

 partly somatic and partly germ-cells cannot itself have the 

 characters of a germ-cell ; but it may nevertheless contain 

 germ-idioplasm, and may thus transfer the substance which 

 forms the basis of heredity from the germ of the parent to that 

 of the offspring. 



If we are unwilling to accept this hypothesis, nothing re- 

 mains but to credit the idioplasm of each successive ontogenetic 

 stage with a capability of re-transformation into the first stage. 

 Strasburger accepts this view; and he believes that the idio- 

 plasm of the nuclei changes during the course of ontogeny, but 

 returns to the condition of the first stage of the germ, at its 

 close. But the rule of probability is against such a suggestion. 

 Suppose, for instance, that the idioplasm of the germ-cell is 

 characterized by ten different qualities, each of which may be 

 arranged relatively to the others in two different ways, then 

 the probability in favour of any given combination would be 



/l\io I , . 



represented by the fraction (-1 = : that is to say, the 



re-transformation of somatic idioplasm into germ-plasm will 

 occur once in 1024 times, and it is therefore impossible for such 



