W. Ferrel — R \ Law of Thermal Radiation. M.s 



tfssr 1 -r„ 



K =the constant ot conductivity ol air at r:=0°, 

 nr=the temperature coefficient <>i' the conductivity, we have 



ME, 



, r. 

 r. loe 

 r. 



A+ar,+±ad\<$ (0) 



e 



Stefan has used the value of Iv o = '00o054, where the unit 



rime is the second, and a='0027. 



11. It is readily seen from tliis formula that with very small 

 values of r, the values of // hecome very large. For instance, 

 the last experiments of Schleiermacher, referred to in §4, were 

 made with wires O-to.Y ,U1 in diameter, the internal diameter of 

 the tube being 24*2 mm . Putting, therefore, in centimeters 

 r,=s*02 and r t =l*2, the preceding formula gives for <J=1, and 

 r t =0, Z» = 0'O00tU. But we have found in §22, paper A, for 

 the heat lost by radiation from a unit of glass surface with 

 <5 = 1, E=*0056 nearly on the average from all experiments, 

 and by means of several formula, which reduced to the second 

 unit is '000093, a quantity which differs but little from the 

 values found experimentally by Winkelmann and by Kundt 

 and Warburg. But the radiativity of platinum is only about 

 ± of that of glass, and hence for platinum we have, for the 

 second unit of time, E = '00001 very nearly. Hence, in these 

 experiments of Schleiermacher the heat of the wire lost by 

 conduction with ordinary, or even with very small air pressure, 

 would have been 64 times that lost by radiation, where the 

 difference between the temperature of the wire and that of the 

 tube is small. It is reasonable to suppose, therefore, that it is 

 necessary for the tension of air in the tube to be reduced to 

 an exceedingly low one, in order that the heat conducted may 

 be neglected in comparison with that lost by radiation, and 

 this is especially so in the case of a bright wire, in which the 

 radiation is comparatively small. And if the tension is not so 

 reduced, but a part of the heat lost by the wire is due to con- 

 duction, it is evident that the rate of increase with increase of 

 temperature is diminished, since the increase in the part due to 

 conduction, as is seen from (6), is but little more than the first 

 power of o, the difference between the temperature of the 

 heated body and the enclosure, and very uncertain for high 

 temperatures, and since the value of the temperature coefficent 

 has been determined for ordinary temperatures only, while 

 from a mere inspection of the various formulae which should 

 represent the rate of loss of heat, or from the experimental 

 values of JStt in the preceding tables, it is seen that the rate 

 of increase with increase of temperature is much greater. In 

 this case a smaller value of a in Weber's formula would be re- 



