192(i] Bahcofl-Leshii : Chromosomes and Taxonomie Fclationships 387 



is the most probable method Avhich we know occurs. It should be 

 emphasized in all such discussion, however, that there is no known cajie 

 of a stable combination of chr&mosofnes which has been observed to 

 originate in thiis way. Similarly, no case of changed individuality of 

 the chromosomes which would account for stable types like C. setosa, 

 neglecta, and parviflora has been reported to have occurred experi- 

 mentally. Chromosom.e fragmentation is known to occur following 

 trisomy, but whether such types ever become stabilized with a pair of 

 fragments added to the normal specific complex, or whether a chromo- 

 some complex can lose a considerable section of a pair of chromosomes 

 and the plants lacking this part be viable and fertile, is unknown. Our 

 strain of C. Marschallii is peculiar in that, when we obtained it, certain 

 plants contained 9 chromosomes in the root-tip cells, comprising the 

 usual complex for the vesicaria group of species plus a very short 

 unpaired chromosome. The source of this small extra chromosome is 

 quite uncertain, although it is known to be an addition to the complex. 

 Navashin (1925) presented a figure of C. Marschallii that is like 

 vesicaria and lacks the small chromosome. Some of our 9-chromosome 

 Marschallii plants were very fertile, and among their progeny one at 

 least has two such small chromosomes. This matter is being studied 

 further and will be reported upon separately. Should such a plant be 

 fertile, we might understand how such differences in chromosome 

 groups could arise in a genus. 



Navashin (1925cf) has emphasized the importance of minute 

 ''Trabanten" or satellites attached to the tips of certain chromosome 

 pairs in Crepis species. He believes that shape of chromosome and the 

 presence or absence of satellites is "weit wichtiger filr die Charakter- 

 istik des Kernes bzw. der Art, als die Zahl der Cltromosomen und deren 

 Dimensionen sind. ' ' He groups together in class " D " all chromosomes 

 having satellites although in C. Dioscoridis, one of 19 length units bears 

 the satellite, while in C. parviflora he finds it upon one of about 10 

 length units. But in our material, which was fixed in C. A. U., 

 Trabante7i were not alwaj^s present, and sometimes resembled the 

 strands and masses of nucleolar material which are frequently found 

 being extruded from the chromosome plate. Consequently size, which 

 is relatively far less variable and more easily evaluated, was selected 

 as the best criterion of relationship, and it has thus far proved a very 

 good one as tested by species-hybridization. That shape relationships 

 may help in differentiating two pairs of chromosomes of the same size 

 in certain species of Crepis is clearly indicated by Navashin 's figures, 



