1919] Frost: Mutation in MatthioJa 137 



respects, and vice versa. Under favorable conditions this type may 

 closely resemble Snowflake, but is decidedly more slender in stems, 

 leaves, and pedicels. A characteristic drooping of flowers and branches 

 is well shown by two plants in figure 18 ; the single is 25b of the 

 tables. The progeny of 25b shown in figure 19 illustrate a variability 

 of the "slender" characteristics which has suggested the presence of 

 genetic differences among plants classed as slender. The leaves often 

 resemble those of crenate more closely than do Snowflake leaves. 



In the field at Ithaca flowering was markedly earlier than with 

 Snowflake, and the type seems to be earlier on the whole. The River- 

 side conditions have commonly given a decided dwarfing as compared 

 with Snowflake, though not to the extreme degree that this has occurred 

 with crenate (figs. 20 and 21). 



The results of selfing tests are reported in tables 34 and 35. The 

 distributions have the same general characteristics as with crenate, 

 with some remarkable differences. The excess of doubles with Snow- 

 flake is very much greater, the ratio being about 30 : 1 ; with slender, 

 however, the excess of singles is slight in the grand total and perhaps 

 significantly variable with different parents. 



Plant 25b-ll, the "extreme" individual of figure 19, appears to 

 give a real excess of slender over Snowflake, and of double slender 

 over single slender, though the numbers are much too small for cer- 

 tainty. The two parents classed as "extreme" are (tables 39 and 40)-^ 

 quite probably genetically different from the other slender parents. 

 It should be noted that plant 25b-6-8-6, progeny of one of the parents 

 described as "extreme," has also given a relatively high proportion of 

 slender progeny. Perhaps the "extreme" form is heterozygous for a 

 second slenderness factor similar to the original one. 



The percentages of mutant-type progeny are (table 39) much more 

 variable than with smooth, large, or crenate, and (table 40) there is 

 no good evidence of selective elimination ; both these facts may depend 

 on genetic differences among the parents tested. 



The great modifiability of the various types, including Snowflake, 

 indicated by a comparison of, for instance, figures 14, 15, and 16, 

 greatly complicates the positive determination of types. In the cul- 

 tures of 1911 H and 1913, where crowing in flats or aphis injury in 

 the fleld interfered with normal development of some plants, the im- 

 pression was obtained that the slender type occurred in several grades 



-1 In the calculation of the probability of simple sampling, f is taken as 3 

 (the number of cultures), not 2 (the number of parents). 



