BETWEEN THE FREEZING AND BOILING-POINTS. 
217 
For each 15 minutes 8 readings in all were obtained of the differential thermometers, 
which were separated by 2 minutes each, the last one being taken just before switching 
over the flow, 3 readings of the potential across the calorimeter separated by 4 minutes, 
and 4 of the potential across the standard resistance also separated by 4 minutes. 
These readings were always corrected to the middle of the interval to correspond to 
the time of average flow. The thermometer readings sometimes increased and some¬ 
times decreased during the interval, depending on the flow, but the change was never 
more than ‘02° or ’03° over the 15 minutes, and generally much less. The question 
of the lag of the thermometers and of the thermal capacity of the calorimeter, which 
amount to such large corrections in all older calorimetric methods, was reduced there¬ 
fore to a negligible quantity. The potential readings on the Thomson-Varley slide- 
box of the calorimeter and standard resistance generally fell off in a regular way 
throughout the interval, although quite often they remained steady altogether. The 
fall was seldom as much as 1 part in 1000 during the entire interval; hence a greater 
number of readings was unnecessary, since all the readings were taken at stated times, 
which were even minutes recorded from the seconds’ hand of the watch used to start 
the flow over the interval. 
In the earlier experiments, where two observers took the observations, simultaneous 
readings could be made of the temperature and potential difference every minute. 
This arrangement was of course preferable to the other, but where the conditions 
remained steady a large number of readings was really not necessary. When any 
sudden change in the conditions occurred, such as in the regulation, electric current 
or flow, so as to make the readings unsteady, these readings were either repeated 
during another interval where it was possible to restore the conditions in a short time, 
or they were abandoned altogether until such other time as the complete experiment 
could be repeated. My aim was, as far as possible, to produce a series of measurements, 
over the entire range of temperature between 0° and 100°, under as steady and 
uniform conditions as possible. 
The number of flows usually taken in a complete experiment was two, but some¬ 
times three. Other flows were tried to test the theory of the method beyond the 
limits of flow chosen for the actual experiments. In all the flows two intervals of 
15 minutes were taken, which, when worked out, gave a complete check on the 
steadiness of the conditions and the accuracy of the observations. 
None of these measurements depend very much on the absolute readings of mercury 
thermometers except the Clark cell. The temperature of the standard-resistance was 
always taken with the thermometer used in its calibration with the standard ohm, 
which in turn had such a small temperature coefficient that it made it of very little 
importance whether the thermometer used to obtain its temperature was in error by 
as much as ‘5°. As a matter of fact, by direct comparison, all the thermometers used 
agreed with our standard thermometer to within T°. The thermometer used to 
obtain the temperature of the jacket-water between 0° and 50° C., was a new Muller 
VOL. OXCIX.-A. 2 F 
