92 BOTANICAL GAZETTE [February 



been superseded in the more mature rosette by leaves of a different 

 form, but that the sharp distinction between the variation and the 

 typical form has not been obscured. The occurrence of so interest- 

 ing a variation in Lexington C led to a careful examination of the 

 strain for evidence of mutability. 



A second sowing of the same seed on good soil resulted in a 

 progeny of 720 seedlings from 1000 seeds. As soon as the seedlings 

 were well rooted, they were transplanted to square seed pans in which 

 they were widely enough spaced to allow of unimpeded growth for 

 a month or six weeks. This system was followed in all subsequent 

 work. Of course, the seeds were invariably sown on sterilized soil. 

 After the seedhngs were transplanted, the pans were frequently 

 examined for mutations, and all plants which were noticeably 

 divergent from the mass of the culture were marked for preser- 

 vation. Among the 720 seedlings of the second sowing, there were 

 only 4 round-leaved plants. Since the mass of the culture was 

 uniform, and the round-leaved plants constituted an absolutely 

 discontinuous variation from both the typical form and one other 

 pronounced variant which occurred in the culture, it was concluded 

 that they were probably mutations. In the following pages the 

 round-leaved type is called 0. praiincola mut. nummularia}'' 



In order to show the discontinuity between typical 0. pratincola 

 and mut. nummularia, photographs of two of the seed pans in 

 which this mutation occurred are reproduced as figs. 5 and 6. At 

 the time the pans were photographed, the plants were about as 

 far advanced as nos. 1-4 in fig. 3 . Comparison of the figures will 



"The writer has suggested (Amer. Jour. Bot. 1:237. 1914) that mutations of 

 experimental origin be given trinomial names such as 0. pratincola mut. nummularia, 

 in order to avoid confusion with names which must be given consideration in floristic 

 works. A trinomial nomenclature has the advantage over the binomial system pro- 

 posed by Gates (Trans. Linn. Soc. London II. Bot. 8 : 10. 1913) in that the parallelism 

 of mutations occurring in different species may be indicated by the use of the same 

 mutational designation. For example, a convenient way to show the parallelism 

 between the mutations of O. Lamarckiana and those of O. biennis would be to call 

 them 0. Lamarckiana mut. semigigas, 0. biennis mut. semigigas, etc. The trinomial 

 used in this way need imply nothing as to the specific, varietal, or formal rank of a 

 mutation, but only the manner of its origin. Nevertheless, for the sake of avoiding 

 confusion, it would be well not to give any mutation a name which had previously 

 been used in any subspecific category within the species which had given rise to the 

 mutation. 



