310 



somes in an equal multiple of 7, must be supposed to be originated 

 (eilher directh' or in a preceding generation; from seeds produced 

 by ferliii/.alion. 



22. From the chromosome sets it can be concluded fcompare 

 point 20), which species produce apomicticnl seeds. Such roses 

 are all examined species belonging to the seclion Caninae (Caninae, 

 Åfzelianae, Villosae, Tomentosae, Riibiginosae, Stylosae, Jiinzillianae, 

 Rubrifoliae etc.) 



23. As to the kind of apomixis here present, I am not vet clear 

 on that point. However, it is not to be thought of as apogamy, 

 all the found spindles of the first division of the embryo-sac mother 

 cells having the described character of reduction. I have not found 

 any positive evidence for apospor}\ As to wether the apomictical 

 embryos are formed by nuzellar buds, I cannot saj' now, the suitable 

 slages being lacking in my material. ' The investigations regarding 

 this point are to be continued. 



24. All species and forms investigated belonging to the 

 section Caninae, i. e. the majority of the examined roses ori- 

 ginated in Europé, North Africa and West Asia, are very 

 ancient hybrids represen ting the F^ -generation, Avhich, 

 ever since its origin thousands years ago, has been main- 

 tained owing to apomictical reproduction. 



25. The anorthoploid bushes (with diverging chromosome sets) 

 mentioned in point 21 represent F^ or any låter generation of 

 crosses between two F^^-specimens belonging to different species. 



26. The power of producing seeds apomictically probably was 

 gained in connection with tliese ancient crosses. 



27. For the explaining of the peculiar chromosome conditions 

 in the Ca/7//7a-section it is necessary to assume the existence of 

 some normalh' sexual roses with higher chromosome numbers than 

 found by me, namely octoploid (x = 28) and decaploid [x = 35) 

 forms. From crosses between diploid {x = 7) and hexaploid 

 (a: = 21) forms the tetraploid villosa and' rubrifolia (7 bivalents -f 

 14 singles = 28) arose. The parents of the pentaploid roses (7 

 bivalents -f 21 singles = 35) have likely been diploid [x = 7) and 

 octoploid {x = 28) forms, and the parents of the hexaploid (7 

 bivalent -f 28 singles = 42) diploid {x = 7) and decaploid (x = 35) 

 races. It would be very interesting to know, wether normalh'^ 

 sexual roses with higher chromosome numbers than the hexaploid 

 (x = 21) still exist. 



