624 report — 1884. 



and let us suppose that it neither loses nor gains heat by its ends. Let II' be the 

 quantity lost by emission from the surface per unit of time. Let e be the emissivity 

 or quantity of heat lost per unit area of the cooling surface, per unit difference of 

 temperatures between the cooling surface and the surroundings, per unit time ; and 

 tf being, as has been said, above the temperature of the -wire, let 6° be the tempe- 

 rature of the surroundings. Then : — 



W=ndl.e.(t— 0) (3) 



But when the wire has acquired a permanent temperature, with the current flowing 

 through it, there is as much heat being lost at the sides as is being generated by the 

 current. In this case II = 11' and we obtain the expression for e : — 



My experiments consist in measuring the strength of the current and the tempe- 

 rature of the wire, the latter being effected by measuring the electric resistance of 

 a known length of the wire while the current is flowing through it, and hence 

 inferring the temperature. These being: known, and likewise the temperature of 

 the surroundings, we have all the data for finding e, the emissivity of the surface in 

 absolute measure, 



The experiments of Mr. D. Macfarlane giving ernissivities in absolute measure 

 are well known, and are of undoubted accuracy. They were communicated to the 

 Royal Society (1'roc. Roy. Soc., 1S72, p. 93); and the results are quoted in Prof. 

 Everett's 'Units and Physical Constants ' (chap. ix. § 137). These experiments 

 were made with a copper globe about 4 centimetres in diameter, suspended in a 

 cylindrical chamber, with top and bottom, about 60 centimetres in diameter and 60 

 centimetres high. The results may be briefly summed up as follows : Mr. Macfarlane 

 rinds an emissivity of about ~^ of the thermal unit 0. G. S. per square centimetre 

 per second, per degree of difference of temperatures between cooling body and 

 surroundings for a polished surface, with an excess of temperature of a little more 

 than 60° C ; and, for a blackened surface, the same emissivity with an excess of 5° (! 

 or under. 



Using round wires of small diameter (0'85 millimetre and under), and with the 

 surfaces either brightly polished or in the common dull condition of a wire 

 fresh from the maker, I have found a much larger emissivity than ~^, I have 

 obtained for wires of different sizes different values of e varying from j- 1 ,,- down to 

 ~^, which was obtained with a wire of - 40 mm. diameter, and with an excess of 

 temperature of 24° C. It seems to be showu by all the experiments I have made 

 that, other things being the same, the smaller the wire the greater the emissivity. 



To do away with the part of the emissivity which is due to convection and con- 

 duction by the air, I have commenced experiments on loss of heat by small wires 

 in the nearly perfect vacuum afforded by the modern mercurial air-pump. This 

 part of the subject was experimented on long ago by Dulong and Petit, and within 

 the last few years by 'Winkehnann and by Kundt and Warburg; lastly, and much 

 more perfectly, by Mr. (Jrookes {P)-oc. Roy. Soc, vol. xxxi., p. 239), though in no 

 case, 1 believe, were the ernissivities in absolute measure determined. The con- 

 clusion come to by all these experimenters is the same, namely, that there is a 

 decrease of emissivity due to lowering of the air pressure, this decrease being very 

 small for a reduction down to one-half, or one-third, of the ordinary atmospheric 

 pressure, but becoming very great as the vacuum approaches completeness. The 

 Aery interesting experiments of Mr. Crookes seem to show that, even with the 

 lu^h vacuum which he obtained, the effect of the residual gas in carrying oft' heat 

 from the cooling body was far from being annulled. 



The following table shows the emissivity of a copper wire with bright surface 

 half a metre long, 0-40 millimetre in diameter, and sealed into a glass tube about 

 1-5 centimetre in internal diameter. 



