438 BREAKDOWN OF GENETIC SYSTEMS 



this way there would need to have been a double division of the 

 univalent chromosomes such as is never regularly found in plant 

 hybrids. Gustafsson (1935) considers that unreduced embryo-sacs 

 such as would be capable of development without fertilisation do 

 not, in fact, arise in this way. Where " normal " embryo-sacs 

 have been described in Thismia and Artemia they have probably 

 arisen by reduction and will be incapable of parthenogenetic 

 development (v. Table 68). In most parthenogenetic plants the 

 normal meiosis is replaced by a single division, as it is in hybrids 

 where pairing has failed. There are then two possibilities of 

 development corresponding to the three of sexual plants. We 

 should expect that those having a normal type in the sexual repro- 

 duction of their relatives would have the egg-cell developing from 

 one of the two cells produced by the non-reducing meiosis. This 

 is so, as a rule, in Taraxacum (Gustafsson, 1935) and Chondrilla 

 (Poddubnaja-Arnoldi, 1933, 1934). But in many other groups the 

 embryo-sac develops from both the daughter-cells, in other words 

 the mother-cell develops directly into the embryo-sac (Fig. 130), 

 e.g., in E'U-Hieracium (Rosenberg, 1917), Antennaria (Stebbins, 

 1932 b ; Bergman, 1935 a), Ochna (Chiarugi and Francini, 1930), 

 and Ixeris (Okabe, 1932). In these cases there has evidently been 

 a shift in the development of the embryo-sac. Failure of cell-wall 

 formation begins earlier than in the corresponding sexual species, 

 and as is usual in such circumstances, the change is variable in its 

 effect ; most forms exceptionally have development from one 

 instead of from both the products of meiosis. Such variations have 

 been most accurately described in Artemisia (Table 68, Chiarugi, 

 1926). 



The behaviour of the polar nuclei is variable in parthenogenesis. 

 These nuclei in sexual reproduction fuse with a pollen nucleus to 

 give the triple endosperm nucleus. Clearly in parthenogenesis 

 the same three to two proportion in the nuclear content of endo- 

 sperm and embryo cannot usually be established as in sexual 

 reproduction. It has been found that in Erigeron annuus 

 (Holmgren, 1919) the polar nuclei fuse, while in Balanophora (Ernst, 

 1 91 4) they do not. Zephyr anthes is remarkable in one generative 

 nucleus of the pollen fusing with a polar fusion-nucleus, so that the 



