326 
Journal of Agricultural Research 
VoL V, No. 8 
methods before the resistance coils became short-circuited were always 
in very satisfactory agreement, but this was hardly a sufficient test of 
the accuracy of the resistance method, because the measurement of 
temperature difference by the electric-resistance thermometers is much 
superior to that by the mercury thermometers in sensitivity and pre¬ 
cision. With the third type of resistance bulb a more satisfactory 
method of checking was provided. A differential thermoelement, with 
several junctions of copper and constantan wire in each end inclosed in 
thin glass tubing, was mounted with one end in the water just leaving the 
ingoing thermometer and the other end in the water just entering the 
outgoing thermometer. The terminals of the thermoelement were 
connected with binding posts on the observer’s table, from which con¬ 
nection could be made with a potentiometer, by means of which tem¬ 
perature differences could easily be measured to an accuracy of o.oi 
degree. Measurement of the increase in temperature of the water 
flowing in the heat absorber by means of this apparatus afforded a real 
check on the measurement with the resistance thermometer. 
MEASUREMENT BY TEMPERATURE DIFFERENCE RECORDER 
As a matter of fact* this method of measurement could be employed 
instead of the resistance-thermometer method when the readings are to 
be made and recorded by the observer. Either method was more con¬ 
venient and decidedly more sensitive than the mercury thermometers, 
and by use of it the temperature difference was actually measured to 
o.oi degree, whereas in reading the mercury thermometers the tem¬ 
perature was only estimated to o.oi degree. The particular advantage 
in the resistance thermometers was in the opportunity to use with them 
a device which gives automatically a practically continuous record of the 
difference between the temperature of the water entering and that of the 
water leaving the heat absorber. A device of this character which has 
been employed for five years in the investigations with the present res¬ 
piration calorimeter has proved very satisfactory indeed and relieves the 
observer of a considerable amount of drudgery, while it entirely elimi¬ 
nates the possibility of error due to personal inaccuracy in recording 
data regarding the temperature differences. 
Like the resistance thermometers described above, the two coils used in 
this device have the same resistance, approximately 25.5 ohms, at the 
same temperature, and the same change in resistance with the same change 
of temperature, but the bulbs differ somewhat in mechanical construction 
from the earlier type. The platinum resistance wire is not in a thin, flat 
coil in a flat sheath, but is in a helical coil in a narrow annular space be¬ 
tween two metal tubes with thin walls. The wire is wound upon the 
inner tube, and the outer tube fits close against it, an electrical insulation 
of thin sheet mica separating the wire from each tube. The space between 
