1916] DEV RIES—DIMORPHIC MUTANTS 275 
the mean for each group in the last columns of the table, and the 
reliability of this mean evidently depends strongly on this fact. 
TABLE XII 
SPLITTING INTO Jaeta AND velutina 
: — Veicans. | Peacean. MEANS FOR THE GROUPS 
ons ce ae ee ee 
MOTHER : type of laeta velutina 
O. canaXO. b. rime 25 II 63 
O. canaXO. Cockerelli. . 17 17 63 23 14 61 
O. canaXO. Hookeri...... 28 14 58 
oO. Stet xO. b. Chicago. 33 32 32 
O. scintill. xO. Cockerelli. . 21 49 29 } 3° 36 3° 
0. scintill. XO, Hookeri. . . 35 28 29 
O. lataXO. b. Chicago. . 21 24 55 
O. lataXO. a 21 31 48 27 25 46 
O. lataXO. Hookeri...... 39 18 43 
O. be. b. - 
MMMM oe oo oe eek an: 19 81 
O. Str gar Cock- 
Sr reagan pire 0 Tac Nateg oh II 89 esc 18 82 
oO. LamarckianaXO. Hook- 
sie SOtLA Wh Mian cial eua phan <p as 23 77 
O. nanellaXO. b. Chicago.|........ 41 59 
O. nanella xO. Cockerelli.. Sik ae 38 G2 ey 41 59 
O. nanellaXO. Hookeri....|........ 45 55 
Theoretically a splitting into 4 groups of equal size should be 
expected, namely into cana-laeta, cana-velutina, laeta, and velutina.* 
Evidently one of the first two groups is suppressed. This con- 
clusion holds good for O. cana just as for O. scintillans, but in the 
case of O. lata the fourth group is sometimes visible, a few speci- 
mens of the /afa-like hybrids assuming at the same time the marks 
of O. laeta (about 1 per cent, Gruppenweise Artbildung, p. 255), 
whereas the remainder are clearly Jata-velutina. It is probable 
that the same group is suppressed in both the other cases, and for 
the same reasons, which are as yet unknown. The fact that there 
are so often more velutina than should be expected probably has 
the same cause as the exuberant occurrence of this form in the 
crosses of O. Lamarckiana (82 per cent). The analogous crosses 
of O. nanella, given in the lower part of the table, show that these 
DE Vries, Huco, On triple hybrids. Bot. Gaz. 47:1-8. 1909. 
