506 FROMME: THE CULTURE OF CEREAL RUSTS 
difficult to hold in culture than the sunflower rust, a fact which is 
attributed to the slower growth of the oat plants. Reinoculation 
is necessary about every three or four weeks. P. Sorghi is more 
easily cultured than the cereal rusts and Melhus has propagated it 
both winter and summer. The amount of infection increases as 
the corn plants grow larger and the fungus spreads from one culture 
to another by natural agencies. 
I have tabulated the principal recorded observations on the 
influence of various conditions on spore germination and develop- - 
ment in the rusts in TABLE 1. These scattering and somewhat 
fragmentary records illustrate the incompleteness of our knowledge 
of rust physiology. 
CULTURE METHODS 
Preliminary experiments were made to determine how long 
single infections would maintain themselves under greenhouse 
conditions and whether they would self-propagate to any extent. 
The rusts used were Puccinia dispersa Erikss. on rye and Puccinia 
coronifera Kleb. on oats. The seedlings were grown in 5-inch 
pots and infections were secured by atomizing them with a uredo- 
spore suspension followed by covering with a bell jar for twenty- 
four hours. The incubation periods for both forms averaged about. 
twelve days in this preliminary work. The infections secured 
maintained themselves for about two weeks after the first ripening 
of the pustules. After this time the number of pustules visibly 
decreased through withering and dying off of infected leaves and 
eventually the cultures became entirely free from infection. Close 
association of non-infected with infected plants did not produce 
new infections. No marked difference in susceptibility between 
old and young plants was apparent but the younger were. found 
more desirable for inoculation because their greater compactness 
facilitated uniform covering with the spray. 
My further experiments were planned to meet various require- 
ments. First a method to maintain cultures of as nearly as pos- 
sible constant virulence, with the fewest necessary transfers, over 
extended periods of time, was tested out. Such a method is 
suited, for example, to test the possibility of maintaining the rust 
for long periods in the uredo stage and for the study of the effects 
of such conditions of growth on its virulence, incubation period, 
