( 882 ) 
The results of this and the previous series of experiments aré 
shown graphically in fig. 4, where the ordinates represent cooling 
effects and the abscissae observed pressure difference. Values calcu- 
lated from equation (3) are also plotted, showing good agreement 
between observed and calculated effects. 
§ 12. The question still remained if to the observed results a 
correction should still have to be applied for possible heat conduc- 
tion to the gas during expansion. Further experiments were therefore 
made with a view to determining the influence upon the observed cooling 
effect of the quantity of gas issuing from the valve. At very small 
valve-openings, when the quantity of gas expanding was small, the 
thermoelement registered a temperature that was markedly higher 
than that obtained with greater valve openings; but as the valve 
opening increased, the cooling effect rapidly became greater, and a 
state was soon reached when further increase in the quantity of gas 
passed through the apparatus had no further measurable influence 
upon the observed cooling effect. 
For instance, with air at O°C. expanding from a constant pressure 
of 5.16 atm. the following results were obtained : 
Litres per minute. Cooling. °C p. atm. 
5.5 0.244 
8.2 0.258 
11.4 0.272 
12.0 0.273 
12.8 0.273 
14.0 0.273 
That too small a cooling effect was observed at small valve ope- 
nings may be result of conduction of heat along the glass valve, 
and of insufficient cooling of the thermoelement (which must constantly 
receive some heat from outside by radiation and heat-conduction), 
although, even with the smallest valve openings used, the air in 
the vacuum glass is swept out in less than a second. The fact, 
however, that a final condition is reached which is independent of 
the velocity seems to show-that the correction to be applied for 
heat conduction decreases very rapidly with increasing velocity °), 
and does not come into account in my experiments. The explanation 
of this circumstance can be given only after further experiments. 
$ 18. From the foregoing it is apparent that: 
1. For experimental determinations of the Joure-KerviN effect 
') At the above pressure, the valve did not allow greater velocities to be used. 
