ON THE MEASUREMENT OF TEMPERATURE. 
based, difterences of temperature can only l^e expressed as fractions of the al^solute 
teinperatnre at which the processes in the cycle are performed. The accuracy of 
a thermometric measurement sliould then l^e as expressed not as AT but as AT/T ; 
this leads us to the conclusion that an error of 0°-01 at the boiling-point of hydrogen 
is equivalent to an error of 0°'15 at the normal temperature. 
i). The Vapour Pressures of Liquid Hydrogen. 
Pressiu’es in millimetres 
of mercury. 
Temperatures on the 
hydrogen scale. 
Temperatiu’es on the 
helium scale. 
1 
800 
20° 41 
i 
20-60 
790 
20 • 36 
20 - 55 1 
780 
20-32 
20-51 
770 
20-27 
20-46 
760 
20-22 
20-41 
750 
20-17 
20 - 36 i 
740 
20-13 
20-31 
7.30 
20-08 
20-26 
720 
20-03 
20-21 ; 
710 
19-98 
20-16 
700 
19-93 
20-12 
650 
19-68 
19-87 
600 
19-41 
19-61 
550 
19-13 
19-33 
500 
18-82 
19-03 
450 
18-50 
18-70 
400 
18-15 
18-35 
350 
17-78 
17-98 
300 
17-36 
17-57 
250 
16-90 
17-11 
200 
16-37 
16-58 
150 
15-74 
15-95 
100 
14-93 
15-14 
50 
— 
14-11 
10. Discussions of the Results. 
As has already Iteen pointed out in Part II. of this research, the temperatures 
measured on the scale of the two thermometers differ by Off at the temperature 
of lic|uid oxygen. It is not, therefore, surprising to find that at the temperatures 
of liquid hydrogen the two scales differ to an even greater extent. A glance at the 
vapour-pressure curves of liquid hydrogen on the helium and hydrogen scales will 
show that the temperatures on the helium scale lie almost exactly 0‘20 above those 
on the hydrogen scale. Such a difference was, as we have already pointed out, to he 
expected (Part II., p. 152), hut in the present state of our knowledge of the 
properties of these gases at low temperatures further discussion of these results is 
impossible. 
VOL. cc.— A. 
