14. ARTHUR: PROBLEMS IN THE STUDY OF PLANT RusTs 
if he succeeds in infecting weak plants, the fungus will rarely come 
to a satisfactory fruitage. Usually all growth of the rust stops 
with the death of the host, although in a few cases, especially in 
the genus Melampsora and Coleosporium, the teleutosporic develop- 
ment continues for a time after the leaves of the host have fallen 
to the ground, as among the willow and poplar rusts. This 
physiological fact supports morphological facts, which place these 
genera among the lowest of the Uredineae in the scale of develop- 
ment, and ally them to saprophytic forms. Probably the primi- 
tive Uredineae were wholly saprophytic and largely unspecialized 
as to host, but as they gained in power to attack living tissues, 
they became more and more restricted in their dietary choice. 
When we remember what remarkable chemotactic sensitiveness 
the germinating spores of many fungi possess, saprophytes as well 
as parasites, it is not difficult to account for a high state of speciali- 
zation in the selection of a living substratum within which to func- 
tion and from which to draw food-supplies. 
This specialization of the parasite must have begun and been 
carried on in large part along with the specialization of the host, 
and we should be prepared to find a parallelism in their associa- 
tion. The large and peculiar genus Ravenelia is all but confined 
to the leguminous families of flowering plants. The genus Phrag- 
midium is largely parasitic on the Rosaceae. The species of 
Gymnosporangium are found only on certain coniferous hosts, and 
their aecidia on the Pomaceae. And so we might continue, show- 
ing that there is some association, not at present wholly explain- 
able, between certain tribes and genera of rusts and certain families 
and tribes which serve them as hosts, and in many cases this 
association is very closely circumscribed. 
The knowledge of this parallelism is an important help in 
keeping us at the present time from falling into the error, so 
common among the older systematic students of the rusts, of 
assuming that forms on widely related species of hosts, however 
similar in appearance these forms may be, might belong to one 
species, without securing rigid proof. Not until recent years 
would it have been considered strange to announce that a certain 
species of rust had been found upon Borraginaceae, Asclepiadaceae 
and Compositae. A common expression has been that the form 
Aik sti eh aa 
7 CGR ee ee Paap eT ey 
a oe 
