BETWEEN THE FREEZING AND BOILING-POINTS. 
191 
In the tests to be described, an external resistance of at least 40 ohms was inserted 
in the battery circuit, which was supplied from one accumulator. The current was 
never as much as ’02 ampere in each thermometer, and current-heating could he 
safely neglected. In view, however, of a possible elfect of current-heating on the 
differential readings, the F.I. of thermometer D was determined for different external 
resistances, but no effect could be measured. The current was left continuously 
running during a test, and was always the same as that used in the calorimetric 
experiments. 
Three Fortin barometers, supplied by Elliott Bros., of London, were used to 
obtain the steam-point. They were Nos. 571, 572, and 573. They had all been filled 
originally by Elliott. During the preliminary part of the experiments, barometer 572 
was used, but owing to an accident was replaced later by 571. Barometer 571 was, 
however, re-filled later as a check, by boiling out with mercury, and was compared 
with 573, to which most of the later steam readings were referred. As a check also 
the scale of barometer 572 was verified, and found correct to within T millim. This 
was of sufficient accuracy for the determination of the F.I., as the mercury height 
could be measured accurately only to T millim. with the vernier supplied with the 
scale. It was possible to estimate to '02 millim. with a little care. 
A comparison of 572 re-filled with 573, made on February 26, 1900, is given below. 
No. 573, 75'415 centims., t = 17*5; No. 572, 75'410 centims., t = 17'3. 
Second setting, 
No. 573, 75'415 centims. ; No. 572, 75'400 centims. 
/ 
The temperature of the mercury column was taken from a thermometer embedded 
in the barometer case. I decided to adopt the readings of the highest barometer 
as likely to be most accurate. It is probable that barometer 573 was correct to 
'01 centim. in its readings over different dates. An error of '01 centim. in setting 
and reading the height of mercury would produce an error of about '004° on the 
steam point, which is about the order of agreement given in the measurement of 
the F.I. from time to time. In repeating readings of the F.I.. where it was not 
necessary to alter the setting of the barometer, much closer agreement than this was 
attained. The barometer readings were reduced to 0° C. and latitude 45°, and 
corrected for temperature by the formula 
H 0 = H (1 - -0001614?:) (1 - -000033) + -0002£ centim. 
The essential scheme followed for the determination of the F.I. has already been 
laid down. The following tables contain the results of the tests made on the 
different instruments used in these experiments. Owing to the importance attached 
to thermometer E, it is given first place. For this thermometer it will be necessary 
to divide the tests into four groups. 
