352 C. H. Oslenfeld. 
the origin of new forms in the sub-genus Archierncium must go on 
otherwise, at least in the absolutely apogamous forms. A Swedish 
botanist, Samuelsson, has shown that some forms have a very 
limited distribution and that their areas fall within the area of 
nearly related, more widely distributed forms, and further that they 
seem to have a centre from which the distribution extends, all the 
evidence pointing to this centre as the place where the form in 
question has arisen. Presuming that his forms also are apogamous, 
the forms of restricted distribution must have arisen by means of a 
variation which because of apogamy is at once fixed, since it cannot 
be mixed by crossing with the original form. If this supposition is 
correct, it should be possible always to obtain new forms by culti¬ 
vating a sufficient number of individuals of a species, and, pro¬ 
bably, it would be best if the species chosen for such ap 
experiment were a form of wide distribution, i.e., a comparatively 
old form. Working on this hypothesis I have cultivated rather 
large numbers of individuals—as large as the space at my disposal 
would permit—of certain forms, and have always used those whose 
constancy I had tested beforehand by castration and by sowing the 
seeds. Also the seeds for the experiment were obtained from 
heads which had been castrated ; thus no kind of fertilisation had 
taken place. The number of individuals obtained reached between 
100 and 300 in each experiment, and while three experiments did 
not show any variation, the fourth one with a form of H. tridentatum 
from Denmark gave the result that 152 individuals were alike and 
like the parent, but one individual differed in several respects with 
regard to the involucral bracts, the length and direction of the 
corollas, etc. This variant was castrated, and the seeds obtained 
from it were sown. The seeds have germinated ; but only a few of 
the plants have yet flowered. Thus I do not know what the whole 
offspring will prove to be ; but the few flowering individuals look 
exactly like their parent, keeping its distinctive marks. 
The material for showing the origin of new forms by means of 
variations is consequently somewhat meagre at present; but to the 
experiment just described and to the facts put forward by Samuelsson 
must be added the above mentioned variation of a cross in the F„- 
generation. Further a single variation found in a culture of H. 
aurnntiacum originated as a cross between two different races of 
this species: 167 individuals were all alike and like the mother, but 
one differed from the others and did not show any character of the 
“ father,” which undoubtedly has no influence. The distinctive 
