12 BULLETIN 988, U. S. DEPT. OF AGRICULTURE. 
this was not sufficient time to remove all the carbon dioxid from the 
system, as was shown by the high percentage of the gas found is the 
residual air. It is possible that in this circumstance the air in the 
hive had a larger percentage of carbon dioxid than that of the 
sample analyzed. On the 22d the air was passed through the soda 
lime for nearly three hours prior to the end of the period, in which 
case the air in the hive had greater opportunity to become like that 
of the system. Even with a carbon-dioxid content of at least 6 per 
cent, which was the case on the 21st, the quantity of the gas carried 
over in the hive to the next period would be much less than 1 liter, 
which would still leave a wide difference between the figures for 
carbon-dioxid production in these two days. There is nothing in the 
data at hand to suggest a reason for this difference. It is interesting 
to observe that the total of carbon dioxid produced for these two 
days was almost identical with that of the two days preceding them, 
when the carbon-dioxid concentration of the air was low. 
The proportion of oxygen in the air at the end of each period is 
also shown in the table. These figures simply show the condition at 
a given time each day, but they give no definite idea of the propor- 
tion of oxygen in the air during the whole day. This would vary 
hour by hour with the admission of oxygen, the absorption of water 
vapor and carbon dioxid, and with changes in the temperature of 
the air, but on the whole would be somewhere in the range between 
the proportion at the end of one period and that at the corresponding 
time in the period preceding or following. The figures therefore 
show that there was a continual increase in the proportion of oxygen 
from the 18th to the 20th, then a decrease to the 23d. 
The low proportion of oxygen in the air at the beginning of the 
experiment was due to the fact that air rather than oxygen was 
supplied to the system to replace the carbon dioxid and water vapor 
removed during the preliminary period and to maintain a sufficient 
quantity of air in the system while the apparatus was being chilled 
before the experimental conditions were established. After the ex- 
periment began, replacement was made by oxygen until the 20th, 
when the requisite volume was again maintained by admitting air, 
in order to reduce the proportion of oxygen in the air of the system. 
No effect that could be ascribed to changes in the oxygen content: 
of the air was observed until the last day of the experiment. On 
that day not only water vapor and carbon dioxid, but oxygen also 
was removed from the system by passing the circulating air through 
a solution of potassium | pyrogallate before returning it to the 
chamber. This was continued until the proportion of oxygen in 
the air, which was only 18 per cent at the beginning of the period, 
was very greatly reduced. After a few hours the circulation of air 
was stopped and the water vapor and carbon dioxid allowed to ac- 
