252 
JOUKNAL OF THE EOYAL HORTICULTUEAL SOCIETY. 
With trials of less extent the numbers, of course, appear to be more 
diverse. (Eiiothera Lamarckiana x CE. Lam. cruciata, for instance, in 
four trials yielded 0,0, 1, and 28 per cent, of plants showing the cruciata 
character. 
It follows from this that if only a few crosses be made it is a matter 
of chance in such cases what percentage results. 
In like manner, if reciprocal crosses be compared with one another on 
too small a numerical scale, the differences can apparently be very great. 
The larger, however, the number of crosses effected, the greater will be 
the agreement. I quote the percentage of nanella from four crossings of 
CEnothera nanella ? x CE. Lamarckiana S' for comparison with the per- 
centages given above of the reciprocal cross. I found 7, 21, 21, 29 per 
cent, nanella. 
There is a possibility that the variability shown in the heritage 
numbers may be determined partly by internal and partly by external 
causes. By means of a deliberate choice of the pollen and of the flowers 
to be fertilised, and also by artificial interference, it must be possible to 
influence this variability in definite directions. In this way I succeeded, 
in 1899, in reducing the yield of nanella from eight crosses between 
(E. Lamarckiana $ and (2/. nanella ^, nearly 1,800 seedlings being con- 
cerned, to 0-5 per cent., and in eight further crossings, involving about 
1,600 seedlings, I reduced it to nil. 
On the other hand, the yield of crosses displaying the younger 
character may be increased. I, for instance, obtained in extreme cases in 
the first generation from 
CE. Lamarckiana x CE. gigas 100 per cent, gigas 
CE. Lamarckiana x CE. ruhrinervis 73 per cent, ruhrinervis. 
In like manner the crossing of CE. nanella ? x CE. biennis ^ with 
respect to the first character yielded, according to the choice of the pollen, 
in some trials about 0 per cent., in others up to 96 per cent, nanella. 
From these figures the deduction follows that the non- uniformity of 
the first generation of false crosses can be easily overlooked or even be 
lacking altogether. Extreme conditions of the experiment may, especially 
when the sowings are small ones, easily result in only one type appearing. 
In such cases one must be very careful, because it cannot even be deter- 
mined by a solitary sowing whether the effected cross be one of equal or 
unequal heritage. The repetition of the experiment on a larger scale or 
the study of the second generation can alone permit of a decision. 
As examples of a ditypic first cross generation I may cite the follow- 
ing crossings : — CE. Lamarckiana $ x CE . suaveolens S (= CE.odorata, 
Hort) yielded me two types — the maternal, bearing, however, suaveolens 
flowers, and the paternal. It was the same with CE. Lamarckiana x 
CE. hirsutissima (= CE. biennis liirsutissima, Gray), and with a series of 
further crossings. 
In the cases so far cited the first generation after the cross was 
mono- or bi-typic ; it may, however, display three or more types. 
Such may be the case, among others, when two species by various 
mutations have arisen from one and the same species. In such case, the 
earlier character, which, has been lost in the new form, may reappear 
through the crossing. 
