OP ARTS AND SCIENCES. 225 



melting point is quite constant in successive measurements, and it ap- 

 pears to present greater uniformity than many of our commercial 

 standards. There is little doubt, moreover, that by the use of a 

 wire longer, and perhaps of larger diameter, the results would be 

 improved, as there would be less liability to error from the cooling 

 of the ends of the wire by contact with the metal supports (an action, 

 however, which can easily be removed altogether), and from accidental 

 minute inequalities in the length of the wire used. It would also be 

 desirable to use platinum of better quality than is usually found in 

 commercial wire, and to have special care exercised in drawing to 

 insure accurate cylindricity. 



It is also evident that such a wire as I have used cannot at best be 

 employed except as a secondary standard, since the ratio of the lumi- 

 nosities of the two gauges of wire employed is not at all in the ratio 

 of the projection of their surfaces. 



Still further, a difficulty may arise in the use of a melting plati- 

 num wire or strip even as a secondary standard from variations in 

 the physical condition of the metal due to cracking, occluded gases, 

 or other causes. No certain evidence of this was observed with the 

 samples of wire employed, but I am not at all sure that it may not 

 occur with different specimens of commercial wire. From a few 

 preliminary experiments of a series still in progress it appears proba- 

 ble that successive heating and cooling, or continued heating of the 

 wire even under ordinary atmospheric pressure, tend slightly to raise 

 its point of fusion, an action which Edison has shown * to be carried 

 to an extreme degree when continued heating in vacuo is employed. 

 Thus, in one experiment the wire was heated by the current and 

 cooled from one to ten times, and then carried up to its melting point, 

 photometric readings being taken as in previous experiments, with the 

 following results. The wire was .006 inches in diameter. Each value 

 given is the mean of five measurements. 



Mean Reading. 



Without reheating 37.65 



Reheated once 37.42 



« twice 37.42 



«* 3 times 37.62 



« 4 " 37.26 



« 5 " 37.06 



" 10 " 37.26 



* Proceedings of American Association, 1879, vol. xxviii. p. 173. 



VOL. XXII. (n. S. XIV.) 15 



