[Vor. 8 
346 ANNALS OF THE MISSOURI BOTANICAL GARDEN 
variety, Kentucky burley. The inoculations were made on 
March 2 and final notes were taken March 16, though the plants 
were actually observed until April 11. No observations were made 
on temperature and humidity, but conditions in the greenhouse 
were such as to encourage rapid growth. These experiments 
were conclusive in showing that the virus or disease agency does 
not under our filtration conditions penetrate through agar of the 
consistencies employed, nor does it diffuse through a parchment 
membrane. On the other hand, the infected particles pass 
readily through the spherical atmometer cup. In this connection 
it should be observed that while the diffusion experiments lasted 
for a period of 4 days it has been shown (Allard, 716) that there 
is little, if any, lessening of pathogenicity in solutions subjected 
to more or less fermentation. The fact that the infected juice 
from within the diffusion shells invariably induced the disease 
is sufficient evidence that the growth of foreign organisms was 
not a factor worthy of consideration. The method of inoculation 
employed in the above experiments was the same as that de- 
scribed below for the more elaborate work here reported, and the 
reader is referred to the later description for the method employed. 
It should be stated that several of the porous spherical at- 
mometer cups have been tested in this laboratory under similar 
conditions and have been found invariably to prevent the. passage 
of vegetative cells or of spores of Bacillus subtilis, and the subse- 
quent results will show that this particular filter possesses finer 
pores than the Mandler diatomaceous filter. The indications 
furnished by Beijerinck as to the capacity of the virus to pass 
certain porous filters was again confirmed. On the other hand, 
Beijerinck claims a very slow diffusion, or penetration, of the 
virus into agar. The concentration of the agar is not noted. 
For the present the writers are unable to discuss the merits of 
this claim, since our own experiments represent direct filtration 
results, and the agar employed was probably denser than that 
used by Beijerinck. 
FILTRATION OF THE DISEASED JUICE THROUGH ULTRAFILTERS 
After the preliminary work reported above it was clear to the 
writers that it would be desirable to filter the diseased tobacco 
