lOO 



in their characters" — (Hurst, p. 891.) The pollen is only partly 

 fertile, but the plants reproduce apomictically, and within their area 

 flourish exceedingly. Here belong the familiar and variable dog- 

 roses of England, Harrison and Blackburn- give a table showing 

 the percent of fertile pollen in the R. canina group, ranging from 

 those (R. sitbcristata, R. fiigax, R. coerulea) in which the whole 

 of the pollen aborts, to those (^R. senticosa, R. mollis) in which it is 

 75 to 90 percent good. They tested many of these plants by cas- 

 trating and bagging them, and without exception seeds were set. 

 Comparison may be made with the condition in Hieracium, which is 

 likewise polymorphic. The formulae have been worked out for 

 a number of these irregular polyploids, and the following may serve 

 as examples : 



R. mollis Smith, CDDE, male gametes D, female CDE. 



R. rubrifolia Vill. ADDE, male gametes D, female gametes 

 ADE. 



R. canina L. iVABDE, male gametes A, female gametes ABDE. 



R. ruhiginosa L. ABBCD, male gametes B, female gametes 

 ABCD. 



Finally descriptions are given of numerous experiments in 

 crossing the various roses, and it is stated that these appear to fully 

 confirm, or to be fully in agreement with, the theory set forth. 

 In a former paper Hurst^ sets forth his ideas concerning the evo- 

 lutionary significance of all these phenomena but since then he has 

 somewhat modified his views. The following statement is quoted 

 from a paper which he read before the Linnean Society in 1926, 

 a typewritten copy of which he has kindly sent me : "The original 

 decaploid species would most probably arise by duplication of an 

 ancient diploid species under luxuriant conditions, just as duplicated 

 forms have arisen under cultivation. This would be followed, in 

 geological time, by differentiation of the five double septets of 

 chromosomes and characters by duplicational segregations and 

 gene mutations, thus giving rise within itself to the potentiality to 

 throw ofif numerous new septet species by losses of septets. In 

 this way evolution would be an alternating process, from diploid to 



■ The Course of Pollen Formation in Certain Roses, with some deduc- 

 tions therefrom. Memoirs Horticultural Soc. New York, July, 1927. 



^ Chromosomes and Characters in Rosa and their significance in the 

 origin of species. Genetics, XXXVIII (1925). 



