304 F. J. Rogers — Magnesium as a Source of Light. 



brated* using as known temperatures the melting points of 

 metals, for high temperatures,-}* silver 954°, gold 1035°, copper 

 1054°. As the melting point of copper is far below the tem- 

 peratures to be measured it is necessary to prolong the calibra- 

 tion curve beyond the points determined by observation. But 

 as this curve is almost exactly a straight line from 700° to 

 1050° it is probably safe to prolong it to, at least within 200° 

 or 300° of the melting point of platinum. 



It was soon found that burning magnesium almost instantly 

 destroyed the small platinum wire used ("01 inch diameter), 

 apparently by some chemical action, as there was no evidence 

 of fusion. To determine the temperature in the ordinary way, 

 by noting the permanent deflection of a galvanometer needle 

 when the thermo-junction was held in the magnesium flame, 

 was clearly impossible. To obviate this difficulty several indi- 

 rect methods were employed, in all of which the temperature 

 of burning magnesium was determined by comparison with 

 other flame temperatures. 



The temperatures of the flames to be used as standards in 

 this work, were determined by the writer by means of the 

 E. M. F. produced by them in the above mentioned thermo- 

 junction. These flames were air blast flame, Bunsen flame, 

 luminous gas flame (Bunsen burner with the air holes closed) 

 and candle flame. In all cases the object was to get the temper- 

 ature of the hottest part of the flame. No galvanometer 

 readings were taken, until by repeated trial, the deflection was 

 as great as could be obtained. The final results of three sets 

 of observations are given in Table II. 



Three methods employed in the estimation of the tempera- 

 ture of combustion of magnesium are described below. 



First method. — The wire of the thermo-element was passed 



through a small hole in a porcelain screen which protected the 



junction as shown in fig. 2. The platinum-iridium wire was 



2. then heated as closely as possible to the screen 



by the different flames, including the magne- 



sium flame and the corresponding galvanometer 

 "swings" were noted. Three determinations 

 by this method are given in Table II. 



Second method. — It was found that after the 

 _, J platinum-iridium wire had been heated a short 

 s time by the magnesium-flame, a negative de- 

 pt-UL. flection of the galvanometer was obtained. 

 An investigation proved that the burning mag- 

 nesium formed a black compound with the 

 wire and that this substance formed with the 

 platinum- iridium a thermo - element whose 



* By Mr. Ernest Merritt in the course of an investigation not yet published. 

 t " Physikalische-Chemische Tabellen." Landolt und Bornstein. 



