ON THE MEASUREMENT (.)F TEMPERATURE. 
manojiieter tii1)e was 1 aOO milliins. abov-e the level of the dead-space, so that when 
the pressure on the gas in the thermometer was 1000 millims. there was still a 
considerable vacuous space above the mercury. The manometer tube was boiled out 
and the apparatus filled with mercury before the thermometer was fixed to its stand, 
or tl)e bulb sealed to the capillary tube leading to the dead-space. For this purpose 
the apparatus was set up temporarily in clamps, and the top of the manometer tul^e, 
which was drawn out to a fine capillary (fig. 4, A), was sealed to a tube leading to the 
mercury pump. The capillary tube connected witli the dead-space was also closed. 
After thoroughly exhausting the apparatus, and heating it to remove trace of 
water condensed on the glass, mercury was admitted through the stop-cock n. As 
the mercurv entered the manometer tube it was boiled bv heat inn the tube Avitb a 
». c' o 
Bunsen burner. By this method, which we have frecpiently employed, the lisiug 
surface of the mercury I’emains in contact with mercury vapour only, and as the 
exhaustion is continued throughout the operation a very perfect barometric vac\ium 
is finally obtained. By admitting air to the tube h through the stop-cock i, and 
lifting the reservoir attaclied to the tube /, the mercury in the manometer tube was 
raised nearly to the top. By heating, mercury vapour was made to pass through the 
constriction, and condense in the descending tube. The constriction was then heated 
in the blow- 2 :)ipe, and the manometei' was sealed. The mercury was then allowed to 
fall in the manometer tube. 
I'he bulb was not sealed to the rest of the aijjjaratus till this ojDeration had 
been completed, as it was found that it was im 2 )ossible to jjrevent mercury vapour 
condensino’ in it. The connection was now made bv sealing; the stem with a mouth 
blow-jiipe. It was unnecessary to make any correction for the distortion of the stem, 
as the volume of 1 millim. of its length amounted only to 0‘000444 cub. centim. 
The instrument was now fixed to its iDermanent supj^ort, as in fig. 4. This 
consisted of a stout wooden jdank, 3 centims. thick and 15 centims. wide at the 
bottom, cut away for G centims. of its width from the le^■el of the dead-S 2 )ace ip)wards 
to allow of the illumination of the scale from behind. Tlie sip^jDort was stayed rigidly 
at the to}) to the wall and to a ])eam running transversely across the I'oom, so as to 
})revent all lateral movement. 
We do not intend to give an account of the exijeriinents which Ave carried out 
before we commenced the series of measurements on whicli Ave base our residts. It 
Avill suffice to state that about four months were spent detecting and eliminating 
sources of error such as that due to lack of rigidity in the ajpAaratus. 
12. Introduction of the Gas into the Apparatus. 
The gases em^Dloyed in these exj^eriments Avere hydrogen and helium, and it will be 
convenient to consider the method by Avhich they Avere prejDared and j^urified at a 
later stage. The gas AA-as introduced into the thermometer through the stop-cock /. 
