70 BROWN RUST 
The hosts of these are indicated by their names, but there 
is a very high probability that their distinction depends entirely 
on accidents of weather or manipulation at the time when the 
inoculation was made. 
Finally, in regard to another case, P. dispersa (sens. lat.) 
the Brown Rust of Corn, it will be seen by referring to the 
systematic part that it is divided into a number of subordinate 
forms or sub-species, which are for the most part only dis- 
tinguishable biologically ; though here the amount of difference 
is much greater than in P. graminis, and there is more to be 
said in favour of calling these forms distinct species, as is often 
done. For, as will be seen by the descriptions, in some of them 
an ecidium stage is known, in others not, though Klebahn 
remarks that in the latter cases we might possibly find the 
zecidium, if we could trace each form to its ancestral home. 
Moreover, in some of these cases, the teleutospores germinate 
in the spring, in others in the autumn. 
One of the most remarkable of these forms is P. bromina, 
on species of Bromus, from which Ward (1902) obtained such 
important results. For instance, he showed that uredospores 
taken from B. mollis always infected B. mollis and B. secalinus 
and their close allies, but not B. sterilis and its allies; while 
on the other hand those on B. steril¢s would infect B. sterilis 
and its ally B. madritensis, but rarely the other Bromes. We 
can reason, as Ward says, that uredospores from B. mollis infect 
that species readily “because their food-supplies and previous 
environment have affected their protoplasm in some way which 
makes it easier for their germ-tubes and mycelium to grow in 
tissues which afford them the same nutriment and present the 
same obstacles, as they have hitherto enjoyed or been con- 
fronted with” (p. 299). They can flourish in B. secalinus 
because here also the food-supplies etc. offered are nearly the 
same. But in B. sterilis the resistance of the plant to infection 
is sufficiently great to present a barrier which is incapable of 
being overcome except by an odd spore, here and there, varying 
from the normal. In 4, out of 148, trials, uredospores from B. 
mollis infected B, sterilis and these might then produce spores 
which could pass on to B. madritensis, although in no single 
