364 Tricotylous Races. 
seedling will have two such divisions and so become a 
tetracotyl or hemi-tetracotyl. If, for instance, 50 divi- 
sions are distributed over 100 seedlings, with 200 cotyl- 
edons, how often may we expect a single plant to present 
two such divisions? 
In the same way the expectation of pentacotyls may 
be calculated. Without going closely into this calcula- 
tion, it is obvious that the proportion of tetracotyls will, 
on the whole, increase with that of the tricotyls indepen- 
dently of course of the nature of the species in question. 
As a matter of fact we do not observe such an independ- 
ence. Some species are relatively rare in tetracotyls 
whilst others produce them more abundantly. Thus An- 
tirrhinum- ma jus never gave more than 1% to 2% of 
tetracotyls (Fig. 63 D, p. 345), although the proportion 
of tricotyls was as much as 79%. Ocnothera hirtclla, 
Scrophularia nodosa, and Cannabis sativa are also poor 
in tetracotyls. The latter produced only 1 to 3.5% of 
them, even when the whole value amounted to 63% (in 
20 individual records). On the other hand, other species, 
or at any rate the races of them which I observed, pro- 
duced tetracotyls abundantly. 
I have grouped together well over 100 separate rec- 
ords from my cultures of 1894-1896, in which the heini- 
tricotyls, tricotyls and tetracotyls were recorded sep- 
arately for each sowing, which almost always consisted 
of about 300 seeds. From these I have especially cal- 
culated, besides the percentage composition in split-leaved 
seedlings, the proportion of these to the tetracotyls ; and 
I give below the number of tetracotyls per 100 tricotyls 
in the wider sense of that term. This proportion varied 
in Amarantus speciosus, for 2-10% tricotyls, and in Can- 
nabis satira for 6-52%, between 1 and 7%. In Mcr- 
