246 JOUENAL OF THE KOYAL HOETICULTUEAL SOCIETY. 
This proportion I found to be confirmed in very many experi- 
ments, e.g. : — 
A, ivith artificial crossing. 
Dominant 
Eecessive 
Eec. 
Year of 
Crossing 
Agrostemma Githago 
nicaensis 
24 per cent. 
1898 
Chelidonium majus 
laciniatum 
26 „ 
1898 
Hyoscyamus niger 
pallidus 
26 „ 
1898 
Lychnis diurna 
L. vespert. (white) 
27 „ 
1892 
„ vespertina (h&iYy) glabra 
28 „ 
1892 
CEnothera Lamar ckiana 
brevistylis 
00 
AA „ 
Papaver somnif. Mephisto Danebrog 
28 „ 
1893 
„ „ nanum 
(single) 
double 
24 „ 
1894 
Zea Mays (starchy) 
saccharata 
25 „ 
1898 
B, with free crossing. 
Dominant 
Eecessive 
Eec. 
Year of 
Crossing 
Aster Tripolium 
album 
27 per cent. 
1897 
C hrysanthemum 
Boxburghi (yellow) 
album 
23 „ 
1896 
Coreopsis tinctoria 
brunnea 
25 „ 
1896 
Solanum nigrum 
chlorocarpum 
24 „ 
1894 
Veronica longifolia 
alba 
22 „ 
1894 
Viola cornuta 
alba 
23 „ 
1899 
Average of all these trials, 24-93 per cent. 
The experiments covered usually some hundreds, sometimes up 
to 1,000 examples. With many other species I obtained confirmatory 
results. 
The differentiation of the remaining 75 per cent, of the two groups 
cited is by no means so simple. It demands that a number of examples 
with the dominant character be fertilised with their own pollen, and that 
in the following year the progeny of each plant be cultivated and counted. 
I have made this trial in 1896 with Papaver somniferum Mephisto x 
Danebrog, and thereby obtained as the composition of the first generation 
of 1895:— 
Dominant (Mephisto) 24 per cent. 
Crosses (with = 25 .per cent. Daneb.) 51 ,, 
Recessive (Danebrog) 25 ,, 
This result agrees with the formula above given ; or rather, from these 
figures the formula was originally deduced. 
The dominant and the recessive characters show themselves in the 
progeny as constant, so far as they were isolated by the separation. The 
crosses, however, separate themselves again, according to the same law. 
They yielded in this trial on tlie average 77 per cent, with dominant, and 
23 per cent, with recessive characters. 
This relation has remained the same in the course of years. I have 
continued this experiment still through two further generations. The 
50 per cent, crosses split up, the 25 per cent, dominants remain 
constant. 
