174 BOTANICAL GAZETTE [SEPTEMBER 
and barley. Glass cages of various sizes and types, provided with 
cotton ventilators to filter the air and caps to keep out the rain, 
were employed. Some were attached to a suction pump and arti- 
ficially ventilated; others, on all sides except the north, were fitted 
with double panes of glass, between which was passed a continuous 
current of water. Despite all these efforts to produce a normal 
environment for the plants, the air inside the cages was always 
2—6° degrees hotter than the air outside, and the light considerably 
diminished. The plants were always abnormal, often attaining 
two or three times the height of their outside neighbors. In a 
few cases a single winter cereal plant was placed in each of several 
glass tubes early in the spring, long before any rust appeared 
outside. As a rule, however, seeds were planted in pots of steril- 
ized soil and placed in the bottom of the cages. As all the air 
entering the cages passed through cotton filters, no spores could be 
carried in from the outside. Although the majority of his results 
were negative, a considerable number of infections was obtained 
with both winter cereals and annuals. This he considered fully as 
much as could be expected, since the plants were grown under 
abnormal conditions. A similar set of experiments was planned 
by MAssEE (25), who planted wheat seed infected with P. rubigo- 
vera in two pots of soil and kept them covered with bell jars pro- 
vided with cotton wool filters. In one pot 26 per cent of the 
plants rusted, and in the other 47 per cent, while not a pustule 
appeared on the controls. 
The amount of rust developing in the grain field seems to vary 
somewhat with the date of sowing. Both the early and late grain, - 
according to Errksson and HENNING’s tables for the different 
cereals, are usually less rusty than those sown at an intermediate 
date. GaLLoway (17) called attention to this fact in 1893, when 
all his duplicate plots of grain, sowed ten days later than the 
originals, were free from rust. Moreover, he says, ‘‘ Examining 
the weather records for ten days preceding the rust, we find nothing 
to warrant the belief that the simultaneous appearance of the 
fungus the first week in May in widely separated spots was due to 
peculiar climatic conditions.” 
ERIKSSON’S well known mycoplasm theory was advanced to 
