208 Mr. W. F. Barrett on Sources of Error in 



source ; that is, the introduction into the cell of the bisulphide 

 of carbon augmented the heat falling on the thermo-pile. When 

 the source of heat was cut off by a metal screen, the needle moved 

 swiftly back to zero and then rapidly up to 67° on the opposite 

 side. From this deflection the absolute increment in the trans- 

 mission caused by the liquid could at once be calculated. And 

 the mean of eight closely concordant experiments showed that 

 when bisulphide of carbon was poured into a rock-salt cell, so as 

 to form a layer r ^ of an inch thick, 9 per cent, more heat 

 reached the pile through such a layer than through the empty cell. 



Before passing on to further experiments, I will reply at once 

 to two inquiries which may suggest themselves in explanation of 

 this phenomenon. The one is that the temperature of the bisul- 

 phide of carbon might be higher than that of the cell, and that 

 hence its introduction would produce the effect noticed. The 

 other is that the exterior surfaces of the plates of rock-salt might 

 be partially rounded by the frequent polishing which they re- 

 ceive ; so that when the bisulphide of carbon was introduced the 

 combination might have acted as a liquid lens. But direct ex- 

 periments instituted for the purpose of testing these and other 

 possible experimental errors show that they play no part in the 

 phenomenon observed. 



Other liquids beside bisulphide of carbon were now tried in 

 layers of equal tenuity; and I was surprised to find a liquid 

 which even surpassed the bisulphide of carbon in this apparently 

 anomalous action on radiant heat. This was the case with bi- 

 chloride of carbon. When this liquid was poured into the rock- 

 salt cell so as to fill the space *02 inch thick, the needle moved 

 from zero to 16° on the side of heat — equal to an increased trans- 

 mission of 12'4 per cent. At the same thickness bisulphide of 

 carbon increased the transmission 9 per cent. But the fact has 

 now, I believe, been well established that the bichloride of carbon 

 is the most diathermic liquid as yet known. Of chloroform, a 

 liquid considerably less diathermic than bisulphide of carbon, the 

 same film of half a millimetre gave an increased transmission of 

 4*5 per cent. With this liquid (among those examined at this 

 thinness) the effect ceased ; for benzole, a more powerful absorbent 

 than chloroform, gave an absorption of 20 per cent. Of Sulphu- 

 ric ether, a still more energetic absorbent, the same thin layer of 

 half a millimetre absorbed 30 per cent, of the total emission from 

 a red-hot spiral. 



These experiments show that, in passing from feebler to more 

 powerful absorbents, the increased transmission, most strikingly 

 observed in the case of bichloride of carbon, becomes more and 

 more disguised until finally it disappears and absorption predo- 

 minates. A similar effect, as might be anticipated, takes place 



