278 Ward.— On Relations between Host and Parasite 
were removed from their seed-beds in the Gardens, after 
having been weeded out and exposed for several weeks, and 
were potted off into small pots of soil, with no extraordinary 
precautions. Each plant had three to five leaves, and the 
infection was carried out by simply placing spores on the 
lamina and marking the leaf with silk. 
The results seemed to show that spores taken from leaves 
of B. mollis , infected B. mollis and B. secalinus freely, but 
refused to infect B. inermis , while the results were doubtful 
as regards B. sterilis , B. asper , and B. arvensis. 
Repetition of the experiment gave similar results, except 
that here the infection sometimes failed on B. mollis ; and 
yet again a third experiment yielded similar results, except 
that this time it was B. sterilis which gave the doubtful result. 
It eventually suggested itself as possible that more care 
was necessary in keeping the spores taken from the leaves of 
B. mollis and those from the leaves of B. sterilis separate, for 
although I only employed those from B. mollis for the in- 
fections, I had collected both in the same tin though separated 
in paper wrappings, and did not at the time regard it as of 
such supreme importance as was subsequently found to be the 
case to keep them separate. It being admitted that the fungus 
is the same on both species this was not unnatural, at the time, 
but the sequel showed that all such assumptions must be 
avoided where these fungi are concerned. Although it is the 
species Puccinia dispersa , whether growing on Bromus mollis 
or B. sterilis , I now know that the infection capacity differs 
much in the two cases. 
In this, the first set of experiments in which my attention 
was directed to the question of adaptation or selective infection 
within the genus Bromus itself \ several sources of error are to 
be met with, the most serious being one to which sufficient 
attention has not always been paid in the past, though most 
careful observers must have thought of it. It depends on the 
recognition of the fact that Uredospores when capable of 
normal germination at all, are highly infectious spores, and 
that if the host-plants used are-— as is usually the case — ex- 
