196 BOTANICAL GAZETTE [SEPTEMBER 
for comparison; but it is naturally better to set up the experiment in 
duplicate and thereby reduce the sources of error. In many instances 
I used three flasks of live material and one of dead. Since the effi- 
ciency of each flask as an insulator depends upon the completeness 
of the evacuation of the space between the walls and upon the silvering 
it is evident that the flasks themselves will not be exactly alike, and 
that an average result is likely to be better than any single one. If 
proper pains are taken, the efficiency of each flask can be determined 
in advance; but unless the experiments are conducted in a constant 
temperature, as they should be, there is little use in doing this. 
Miss BertHA A. Witz, a graduate student and assistant in 
physiology in this university, did most of the actual work of setting 
up and recording the results of these experiments; but as we have 
worked constantly together, the experiments are ours rather than the 
work of either one of us. As the work progressed, experience showed 
us how the experiments should be improved in method, but the reasons 
for these improvements will be more evident if I describe one. 
Experiment 1.—An unweighed quantity of dry peas was soaked 
for 24 hours in tap-water. They were then rinsed in boiled distilled 
water two or three times and divided into two unequal lots, the 
smaller of which was then covered with a fairly concentrated aqueous 
solution of corrosive sublimate for at least half an hour in order 1 
kill these peas. The other lot was divided into equal parts, which 
were poured into two Dewar flasks of about 250° capacity, the one 
silvered, the other unsilvered. The dead peas were poured into 
another silvered flask of the same size. The flasks were cotton 
plugged and suspended on strings (not wire), in such a manner that 
they would not touch any object, metallic or other. This was done 
to avoid the changes in temperature which might otherwise result. 
A thermometer reading only to degrees was pushed through the 
cotton so that the bulb was as nearly as possible in the center of each 
mass of peas. The data will be found in the accompanying 
(p. 197). 
Various things are evident in this first rough experiment. 
continued for nearly nine days and, in spite of the fluctuating ®" 
perature of the room, the temperature in the silvered flask contaimné 
live peas rose until the last day very steadily. Comparing this ™ 
It was 
g tem- 
