334 University of California Publications in Agrioilt nnil Sciouun | Vol. 2 



lop-ieally similar. In fact, tlio summaries of T.shikawa (1916) and 

 Tischlor (191G, 1922) contain very few genera with either the same 

 number throughout, or even a single ground number. Even in the 

 Liliaceae certain species have been reported as having chromosome 

 numbers different from that typical of the genus. Time and further 

 work alone will tell how many of these exceptions are real and how 

 numy are due to error. At jiresent few genera have been much studied, 

 and even Avhere a large number of counts have been published, the same 

 error may appear in a Avhole series of observations. For instance, in 

 both Tritkum and Rosa numerous species were included in recent 

 summaries as having 8 and 16 pairs of chromosomes. 11 has been 

 shoAvn by Sakamura (1918) and Sax (1918, 1921) for Tritinnn, and 

 by Tiickholm (1922) for Fosa, that 7 and not 8 is the ground number 

 for both genera. Another very real source of error in any attempt to 

 generalize from summaries lies in the fact that few cytologists are 

 trained taxonomists. Our experience with Crepis indicates that seeds 

 Avhich are obtained from the most reputable sources may be incorrectly 

 labeled, and, unless the seeds are grown and the plants classified, we 

 cannot al"\vays be positive that they even belong to that genus, much less 

 to the species to which the sender has attributed them. While lists of 

 chromosome numbers include such errors as are indicated above and 

 are, therefore, not suitable as a basis for very sweeping generalization, 

 no one can doubt that chromosome number and, in some cases, size and 

 shape, are good specific characters. We venture the prediction that 

 chromosome number and size will sometime be given with taxonomic 

 descriptions, 



Crepis contains species with 3, 4, 5, 6, 8, 9, and 20 pairs of chromo- 

 somes; but 3, 6, 8, 9, and 20 are much less frequent numbers than 4 

 or 5, each of the former characterizing only one of the twenty-one 

 species represented in table 3. A similar condition has been described 

 for a closely related genus, Lactuca (Ishikawa, 1921), most of the 

 species having 5, 8, 9, or 12 as the haploid number, while single species 

 have 7, 16, or 24. It is especially interesting that Ishikawa finds that 

 his grouping of species according to chromosome number and size cor- 

 responds very strikingly with the taxonomic classification of Nakai 

 (1920). In Lactuca, as in Crepis, great differences in chromosome 

 size exist, and because of this and the numerical differences, Ishikawa 

 is inclined to think that Lactuca is really an assemblage of genera. 

 It is particularly interesting that two varieties of L. dentata have 12 

 pairs, while one has 7 pairs of chromosomes. 



