63 
gauge and whose content of gas was subsequently determined by 
titration. The gas was generated at the middle and either ends of the 
frame and samples taken from all of these and opposite points after 
different lengths of time. A solution of cyanide and sulphuric acid in 
proportion of 1 part cyanide to 14 parts acid to 24 parts water and an 
amount equal to two-tenths gram potassium cyanide per cubic foot 
air space was used in all of these tests. The amount of cyanide remain- 
ing in the residual solution was determined in each instance and 
deducted from the amount calculated to be present if the gas were 
completely generated and diffused. With this frame the residuum 
averaged about 5 per cent; with a large box, mentioned below, about 
8 per cent, and the percentage was found to decrease the more time 
elapsed after generation. With the generator and intake at the same 
end after three minutes it was found the amount varied from 70 to 197 
per cent of the calculated amount with an average of 13 per cent. In 
other words, the gas diffused irregularly, but remained largely in the 
end where generated. The generator was then placed at the middle 
and the intake at the end. Samples were taken after ten minutes, 
thirty minutes, one, four,and twelve hours. In all of these the gas 
was found to diffuse irregularly and in no instance was the diffusion 
nearly complete. As far as could be detected the frame was tight, 
but a loss of gas seemed to occur in some way, whether by leakage or 
transfusion is a question. At any rate, after twelve hours not a trace 
of gas was found. The average percentages of the calculated amount 
for the four other periods were 46, 41, 27, and 13 per cent, respectively, 
showing a regular decrease the longer the time, but never a complete 
diffusion, the highest amount after ten minutes’ time being 70 per cent. 
From these tests it seemed evident that by itself the gas would not 
diffuse evenly in such a frame. A rough fan was therefore put in the 
center, and a diaphragm placed horizontally through the center to 
within a foot of each end, so that a thorough circulation might thus 
be secured mechanically. 
By use of this mixer with the generator and intake at opposite ends 
after five minutes 72.4 per cent of the calculated amount was secured, 
and this with fairly constant results, only varying from 68.5 to 75 per - 
cent in four trials. As this fan was far from a perfect device, and 
also promoted leakage, we concluded that these tests showed fairly 
that in such a shaped frame the gas diffuses entirely irregularly and 
never completely. In other words, it would be necessary in actual 
usage to generate a very much larger amount of gas than necessary 
were it diffused, in order to secure a sufficient amount in all parts to 
be effective, with the danger of burning the plants at the point of gen- 
eration if used too strong. Kvidently the only way to determine this 
was by tests in the field upon insects, as such a fan arrangement was 
clearly impracticable. 
