Heat across Layers of Gas. 



437 



by surprise that they repeated and varied their experiments, 

 till they were fully satisfied that the existence of the increased 

 escape of heat was proved. 



With protoxide of nitrogen (Ng 0), a gas which has the 

 same specific gravity as carbonic anhydride, their observations 

 gave similar results. Between tensions of 35 millims. and 12 

 millims. the total rate of cooling remained nearlj' constant; 

 and, as in carbonic anhydride, it was slightly increased by 

 further diminishing the tension from 12 down to 4 millims. 

 This slight increase was less in the protoxide of nitrogen than 

 in the carbonic anhydride — about -^ of the whole amount in 

 the former gas, about ^2 i^ ^^® latter. 



The observations with these gases show that the form of the 

 curve represented in fig. 5 is in an appreciable degree different 

 in different gases, and that there are some gases in which the 

 increase of the escape of heat by penetration when the tension 

 is decreased will, within certain limits of temperature, exceed 

 the decrease of the escape of heat by convection. 



15. We now come to the very remarkable results .which 

 were obtained with hydrogen. They are plotted down in fig. 8. 

 A Table of some of the results of the observations, and the 

 following particulars recorded by De la Provostaye and Des- 

 ains enable us to construct this figure. The loss of heat by 

 radiation from the silvered bulb was only -^ of the whole 

 amount (presumably at 760 millims. tension). The rate of 

 cooling at 20 millims. tension was found to be JJ of that at 

 760 millims. The rate of cooling at a tension of 12 millims. 

 was rather more than ^ of that at 20 millims. At 4 millims. 

 it fell to about one half ; and nevertheless, when the rate of 

 cooling at this lowest tension was compared with that of a 

 similar bulb placed in the open air, it was found to exceed the 

 latter in the proportion of 4 to 3. We thus get from the ob- 

 servations* all parts of the figure except those entered in dotted 

 lines. The curve representing the loss by convection must lie 



* The following observations are recorded by De la Provostaye and 

 Desains : — 



Tensions 



760 



millims. 



477 

 millims. 



57 

 millims. 



20 

 millims. 



• 4.4 



miUims. 



Times of cooling.. 



m s 

 12 46 



m s 

 13 20 



m s 

 13 40 



m s 

 14 49 



m s 

 27 24 



The rates of cooling will be as the reciprocals of these times, i. e. as 

 the numbers 652, 625, 610, 562, and 304 respectively ; and to these num- 

 bers the corresponding ordinates of the upper line in our figure have been 

 made proportional. 



