1919] Frost: Mutation in Matthiola 135 



DdCc ddCc Ddcc ddcc 



Theoretical ratio (/!= 44.4) 10.4 4 5 25 



Calculated for n = 540 126 49 61 304 



Observed (/t=:540)="' 125 51 57 307 



This fit surely cannot be criticised, whatever may be thought of 

 the devices employed to obtain it ! With cross pollination the agree- 

 ment is fairly good in the case of series 20, which gives the only fairly 

 reliable data. We are assuming 16% per cent of crossover dC sperms ; 

 elimination of .60 of 16% per cent, or 10 per cent of the total, gives 

 .06%/.90 = 7.4 per cent expected crenate, as against 5.9 per cent 

 observed. Series 21 is .supposed to have 50 per cent of C eggs in 

 the ratio 5DC :ldC ; elimination of .60 of this proportion, or 30 per 

 cent of the total, would leave .20/.70 = 28.6 per cent, against 12.0 

 per cent in the very inadequate material observed. An adequate test 

 of the hypothesis obviously requires large hybrid cultures, from 

 vigorous seed sown under favorable conditions for germination. 



A scarcity of crossover crenate singles follows from the hypothesis ; 

 they constitute only one twenty-sixth of the total number of viable 

 crenate single progeny of crenate parents. No direct evidence indi- 

 cating that the crenate and double factors are ever coupled in singles 

 has yet been discovered. 



If the supposed crenate mutants are due to immediate factor muta- 

 tion, however, it seems strange that the same locus is changed more 

 readily in a singleness chromosome than in one carrying the doubleness 

 factor, in a ratio similar to the linkage ratio of later generations. 

 If the apparent mutants are really segregates from a balanced-lethal 

 combination, the observed original coupling of crenate with single 

 might be an accident of sampling involved in the original choice of 

 material; other initial parents might give t4ie reverse coupling. 



5. THE SLENDER TYPE 



This type is comparatively rare as an apparent mutant from Snow- 

 flake or early; the 3135 plants reported in table 28 gave only 4 (6) 

 mutants (2 singles and 4 doubles, 2 of the latter perhaps Snowflake), 

 a mutation coefficient not over .19 per cent. This type seems to occur 

 more frequently among progeny of crenate, a type similar in some 



-« Omitting 29 plants classed as neither crenate nor Snowflake, which as 

 probably nou-crenate should perhaps be added to Snowflake, and also 64 plants 

 (13 crenate and 51 Snowflake) with flower data incomplete. Complete data for 

 the total of 633 plants would plainly give a somewhat poorer fit, but this could 

 be improved by assuming a slightly greater elimination of Ccii zygotes. 



