250 THE GERM-PLASM 



searches of Henking, vom Rath, and Hacker, indicate that even 

 the idants may become changed during the process. If we sup- 

 pose that in the mother germ-cell, when it is preparing for the 

 first reducing division, the ids become arranged in their original 

 order so as to form a long thread which doubles back on itself, 

 and thus gives rise to a ring, the latter would become trans- 

 versely divided in certain places. If the transverse divisions 

 could take place at different points, it would be possible either 

 for the old idants to be accurately restored, or for the new ones 

 to differ from them to a greater or lesser extent. 



This assumption is not, however, essential for a theory of 

 amphigonic heredity, and we may here disregard it, although it 

 will doubtless be found to apply to some extent, as was indicated 

 above with regard to such a slow and slight change of the idants 

 due to the disarrangement in the combinations of ids contained 

 in them. It must be left to future researches to follow out this 

 process in detail, and to show whether the differences in the 

 combination of ids is merely due to the halving and rearrange- 

 ment of the idants, or whether regular, or at any rate frequent, 

 changes in the composition of the idants out of ids also occur. 

 For the present we must be content with knowing that \\\tger7n- 

 cells of an indiindiial contain very inany different combinations 

 of idants^ and that a frequent repetition of amphimixis never 

 ijideed results in the germ-cells of the same parents containing the 

 same combinatio?is. It therefore follows that the combination ot 

 parental and ancestral characters continually varies, and this 

 variation is characteristic of amphigonic heredity. 



This statement also holds good for plants, in which we know 

 that a reduction to half the number of idants takes place in the 

 germ-cells. According to the researches of Guignard,* the 



* L. Guignard, Compt. rend. May 11, 1891, and ' Nouv. etudes sur la 

 fecondation,' Ann. scienc. nat. Bot. Vol. xiv., 1891, p. 163. 



Particulars regarding Giiignard's valuable researches cannot be entered 

 into here. They have not only proved that in plants the mature germ- 

 cells likewise contain only half as many ids as do the somatic cells, and 

 that the normal number is again produced by the union of the nuclei of the 

 male and female cells, but have also shown that the centrosome passes on 

 from one generation to the next. In spite of the fact that these observa- 

 tions are obviously perfectly accurate, I cannot help doubting whether the 

 reduction in the number of idants actually occurs without a nuclear divis- 

 ion, as Guignard states is the case, I have arrived at this conclusion not 



