1886.] 



Radiation of Light and Heat. 



213 



1 now dismounted these filaments and subjected them to incande- 

 scence in the hot hydrocarbon vapour as before with carbon B. 

 The result was highly satisfactory, as they both took a surface much 

 brighter than carbon B had done. Again remounting them and 

 exhausting, they were placed in the photometer as before. The results 

 are given in Tables IV and VII, and the curves from these tables are 

 shown by the bright hard lines CC and DD, diagrams 2 and 3. 

 These curves appear fully to bear out the assumption arrived at in 

 the former tests, Tables I and II, and the improvement in light 

 radiation per unit of energy is especially marked in the case of fila- 

 ment C, where it may be noticed that at 2 watts of energy per candle 

 of light the same filament in its black condition was strained to yield- 

 ing sixty candles nearly, while in its polished state it had only to yield 

 thirty-seven, and still be as economical in its electrical energy as 

 before. The same filament at 3 watts per candle, when black, had to 

 give off 31 candles, equal to 270 candles per square inch of its sur- 

 face, while in its polished state it required to give only 18 candles to 

 equal 3 watts per candle, and its surface was strained only to the 

 extent of 155 candles per square inch. It is certain, therefore, that 

 its lasting power with its surface bright would be many times greater 

 at the foregoing expenditure of energy than in its black condition. 



As the filaments were still unbroken and appeared capable of yet 

 another test, I resolved to attempt the reblackening of them, and to 

 ascertain if possible if the test-curve under these conditions would 

 again revert to its former position, but I had now to re blacken over 

 the bright surface which could not be removed. The filaments were 

 again successfully dismounted, and with some difficulty again re- 

 blackened over their former polished surfaces. 



They were now tested, as is shown in Tables V and VIII, and the 

 corresponding curves are given in diagrams 2 and 3, marked CCC, 

 DDD. The recession of the curves was in both cases nearly com- 

 plete, any difference being fully accounted for by the incomplete 

 reblackening of the carbon surface. 



In carrying out these experiments I much regret not having made 

 the necessary arrangements for simultaneously testing both the heat 

 and light emitted from each filament in its blackened and bright 

 condition. I have little doubt the loss of efficiency when black was due 

 to the energy supplied being radiated in large quantities as heat 

 waves from the blackened surfaces, which these surfaces when bright 

 would not permit. This radiation of heat, however, which had not 

 been converted into light by emission from a bright surface was abun- 

 dantly manifested in the handling of the lamps. The incandescent 

 globe containing the bright filament could at all times be readily held 

 in the hand even when giving its maximum of light, while the heat 

 radiated from the filament when its surface was blackened was most 



VOL. XL. Q 



