274 BOTANICAL GAZETTE [ocToBER 
striking in the hybrids, whereas those of the father, O. biennis, 
have only a less influence on the general type. 
In table XI the figures of table IX are combined with the per- 
centages derived from my Gruppenweise Aribildung, by taking the 
means of the two crosses for each case given on pp. 251 and 26r. 
TABLE XI 
O. canaXO. biennis COMPARED WITH O, laia AND O. scintillans 
. P t f ty 
Forms wiachaee owas Dad ie of O. Lamarckiana Mutants 
O, cana 00. Wiens. i es coe 49 Co een eit cr asian” 
O. Seeman t Ss ahs iia ss 60 36 4 
©. Ina O, Neen os eee ee. 57 Ao ea ce eee 
PRON 5 OO vas os ss Cees 58°. 43 (5 SA ave eee 
In this table we see that the characters of O. cana, even as those 
of O. scintillans and O. lata, are repeated in about one-half of the 
progeny, but not in the other half. We may consider this as the 
simplest case. In the other crosses the proportions of cana are 
17, 25, 28, with a mean of 23 per cent in table IX, and 21, 40, 42, 
mean 34 per cent in table X, and these figures may be assumed to 
point to a splitting into nearly equal parts with a loss on the side 
of the weaker form. Exactly the same behavior occurred among 
the progeny of the self-fertilized i tee of O. cana, as we have 
seen previously. 
et us now compare O. cana with the two allied forms in those 
crosses where the progeny splits into the twin hybrids O. /aeta and 
O. velutina, as shown in table XII. 
The types of O. laeta and O. velutina have been compared in 
each case with the twins derived from O. Lamarckiana by the same 
father. The comparison embraced the whole lifetime from the 
germination in February until the production of the fruits in 
September. No differences have been observed. 
From table XII we see that the splitting percentages are 
practically the same, whether the pollen is taken from O. biennis 
Chicago, O. Cockerelli, or O. Hookeri. For this reason I have given 
* DeVries, Huco, On twin hybrids. Bor. GAz. 44:401-407. 1907. 
