ROSA CANINA 461 



their occurrence and of their form. Erlanson (1931) has shown 

 that metaphase pairing is of the partially terminalised type in Rosa. 

 Uniformity, we now know, means that while in other groups of Rosa 

 the chromosomes may be united by chiasmata at one or both ends, 

 in the CanincB they are always united at both ends, and this is an 

 indication of regularly high chiasma frequency and homozygosity as 

 to the pairs (Erlanson, 1933). Pairing behaviour is identical in 

 embryo-sac and pollen mother-cells (Fig. 134), but the mechanism of 

 distribution is entirely different. In the pollen mother-cells the 

 bivalents and univalents lie in one equatorial plate. The bivalents 

 divide first and then, as is usually the case, the univalents lying 

 between the separating groups of daughter bivalents divide and 

 follow them to the poles with very little irregularity. At the 

 second division the same process is repeated ; the daughter bivalents 

 divide again and the daughter univalents also proceed to divide a 

 second time (like the unpaired chromosomes in Pygcera hybrids). 

 The delay, however, is greater than at the first division. Most of 

 the univalents fail to reach the poles and many are left on the 

 plate. In consequence many nuclei of irregular size are formed and 

 the 4 nuclei which have received the 7 chromosomes derived from 

 the bivalents have in addition a varying number of chromosomes 

 derived from the univalents. Examination of the first mitosis in 

 the pollen grains of R. tomentella oUusifolia and R. seraphini 

 (2n ^= $x — 35) showed the following chromosome numbers [cf. 

 Table 47) : — 



7 8 9 10 II 12 13 14 15 16 17 18 19 20 21 22 

 923137411— I— II— I— I 



These do not necessarily correspond with the numbers originally 

 received, but they indicate the irregularity of the process and 

 the relatively high proportion that receive the simple haploid 

 complement. 



Several embryo-sac mother-cells are found in each ovule of these 

 roses, and they are therefore exceptionally favourable for study. 

 At metaphase the unpaired chromosomes, instead of lying in the 

 same plate as the bivalents, are all, or almost all, grouped at the 

 micropylar pole of the spindle, where they are joined at the first 



