( 880 ) 
nor with the results calculated from the isotherms in § 2, is due to 
the fact that the spiral s, was so narrow as to cause a considerable 
fall of pressure along the tube; and at this temperature the pressure 
fall is too great to allow of the correction being calculated with 
sufficient accuracy. 
This correction became smaller when the narrow tube, s,, was 
replaced by two of 2.5 mm. bore in parallel. The following results 
were thus obtained: 
TABLE V. 
Saries XIV. Glass valve. Air at 0°C. 6/4/08. 
(p,—-1) ioe 
6 atm. 1.054°C. 
OE Be 7 EE 
W US SO) rs 
21; 462 „ 
BU 6.34 „ 
The correction, however, is still too great to allow of its being 
applied with sufficient accuracy. 
In order to ascertain exactly at what pressure expansion was taking 
place, the apparatus was so altered that a manometer might be 
attached directly to the expansion valve at the end of the second 
spiral. This, of course, does not show the true hydrostatic pressure, 
but calculations of the speed of the gas stream before expansion 
showed that the difference was negligible. 
In the series of expansions undertaken after this change, only one 
value was obtained, for, on raising the pressure, the valve, which 
up till now had worked admirably, gave way, and further experiments 
had to be abandoned. That one value, however, viz. : 
Series XV. Glass valve. Air at 0°C. 8/4/08. 
(p,—1) TT) 
5 atm. £5670. 
gave a cooling effect of 0.272° C. per atmosphere pressure difference, 
in good agreement with the Jourw-KeLviN and calculated values. 
§ 10. Advantage of the experience gained in these preliminary 
investigations showing the special difficulties to be guarded against 
was taken in the construction of a new apparatus. In fig. 3 the 
apparatus is shown im situ in the cryostat into which liquid hydrogen 
may be introduced. The gas whose JouLe-KeLviN effect is to be 
