432 Mr. G. J. Stoney on the Penetration of 



which De la Provostaye and Desains used, and in the follow- 

 ing diagrams is represented by the interval between two hori- 

 zontal lines. This interval is, moreover, small, because De la 

 Provostaye and Desains reduced the loss by radiation to a very 

 small amount by silvering or gilding the bulb of their ther- 

 mometer. 



The other term of Dulong and Petit's expression, which fur- 

 nishes the rate of escape by convection, is 



where p is the tension measured in millimetres of mercury, 

 K depends on the gas and on d^ and 62, and c was found to be 

 nearly J when the receiver contained hydrogen, but was nearly 

 ^ for the other gases experimented on and for atmospheric air. 

 I will return to the case of hydrogen ; but in the other gases 

 the velocity of the escape of heat by convection with given 

 temperatures of the bulb and receiver will be represented at 

 different tensions of the gas by the ordinates of a curve not 

 differing much from a parabola, since this would be the curve 

 if the index were exactly \ ; and, accordingly, curves of this 

 kind are laid down in the annexed diagrams. It is not ma- 

 terial whether a large or a small portion of the parabola is in- 

 troduced, because all parabolas are similar. 



10. By thus plotting down the results of the experiments 

 upon diagrams, we obtain the means of seeing at a glance 

 how much of the escape of heat observed by De la Provostaye 

 and Desains can be accounted for by radiation and convection, 

 ^nd how much remains to be allotted to penetration. De la 

 Provostaye and Desains made their observations in three re- 

 ceivers — a hollow sphere of 24 centims. diameter, a hollow 

 sphere of 15 centims. diameter, and a cylinder 6 centims. in 

 diameter and 20 centims, in height ; and they used in all the 

 receivers the same thermometer, which had a cylindrical bulb 

 7 centims. long and 2 centims. across*. Accordingly the in- 

 terval between the bulb and the walls of the receiver ranged 

 from 84 centims. up to 11 centims. in the largest receiver, from 

 4 centims. up to 6^ centims. in that of intermediate size, and 

 was 2 centims, in the cylinder. 



In Dulong and Petit's experiments the diameter of the re- 

 ceiver was 30 centims., and they used thermometers with sphe- 

 rical bulbs 2 and 6 centims, in diameter, so that the interval 

 was either 12 or 14 centims. 



11. With atmospheric air in the largest receiver (in which 



* The direction in which the heat penetrates, and of the Crookes's 

 stress, will be perpendicular to the isothermal surfaces within the gas in 

 the simple case which we have hitherto considered, where A is parallel to 

 B ; but it will in general pierce the isothermal surfaces obliquely if one 

 part of the Crookes's layer is more curtailed than another. 



