6 W. Ferrel — Law of Thermal Il<> illation. 



pressures, gives the effect of convection only, which entirely 

 vanishes before the low tension of 2 or 3 mm is reached. He 

 therefore restores this correction, which is very small, and then 

 the corrected rates of cooling P, in degrees per minute, are 

 those given in the second column of the following table, cor- 

 responding to the values of d in the first, which in this are the 

 temperatures of the cooling body since that of the inclosure 

 was r n =0. 



6 



R 

 1-74° 



2-02(l-0077 (? -l) 



0-925(? 4 -l) 



R 

 1-48° 



l-592(l-0082 (5 -l) 



0-730(g 4 - 2 -l) 



80 = 



+ •03 



+ •08 



•00 



+ •06 



100 



2-30 



— •03 



00 



1-96 



-05 



-•02 



120 



3-02 



-•03 



-•03 



2-60 



—•06 



-02 



140 



3-88 



— •02 



-•04 



3-38 



.-•04 



-•06 



160 



4-89 



+ •01 



— 04 



4-31 



•00 



— •05 



180 



6-10 



+ •08 



+ •01 



543 



+ •08 



+ •02 



200 



7-40 



+ •04 



-•01 



664 



+ •05 



•00 



220 



8-81 



-•10 



-•10 



7-95 



-•08 



— •09 



240 



10-69 



-•04 



+ •08 



9-74 



-•01 



+ •11 



These rates are satisfied by the expressions at the head of 

 columns 3 and 4, the former being that of Dulong and Petit's, 

 and the latter that of Stefan's law, with the residuals, observa- 

 tion minus computation, given beneath in each column. 



5. Stefan has given a formula for computing the rate with 

 which a spherical body within a spherical inclosure is cooled 

 by heat conduction, which is equivalent to 



(n; 



3r. 



v=- 



A<*0 +£«(*+*„))> 



in which, besides the notation already adopted, 



v = the rate of cooling in degrees per minute, 



)\ = the radius of the cooling body, 



r 2 = that of the spherical inclosure, 



k = the conductivity of air at temperature t=0, 



a = the temperature coefficient. 



He puts & =0'00324, which corresponds to his coefficient 

 •000054: where the second is the unit of time. He also puts 

 a=0-0027, c=0-0332 and *=13-6. Hence we have cff=0-4515. 

 The values of r x and r 2 in Dulong and Petit's apparatus were 

 respectively 3 cm and 15 cm . If with these constants and -data 

 the values of v in (11) are computed for the several values of 

 o in the first column of the preceding table and deducted from 

 the second column, we get the values of E, in the fifth column, 

 which arise entirely from radiation. But these rates now are 

 not accurately represented by either Dulong and Petit's or 

 Stefan's law, with any given numerical coefficient, but they 



