416 Prof. H. BufF on the IViermal Conductivity 



two faces of the pile became incompletely developed, the de- 

 flections of the needle being indeed greatest in those cases in 

 which the rays passed most directly to the pile, i. e. when they 

 traversed a vacuum or h^^drogen. And such behaviour is 

 clearly shown in the experiments. 



A still more important source of error lay in the consider- 

 able distance of the source of heat from the thermopile as 

 compared with the small diameter of the glass cylinder. 



Almost simultaneously with Magnus, Tyndall published a 

 comprehensive memoir on the diathermancy of gases and va- 

 pours *. His method differs essentially from that first adopted 

 by Magnus, in that the rays had to pass through a polished 

 plate of rock-salt before they reached the vessel containing 

 the gas under investigation. This vessel consisted of a hori- 

 zontally placed metal tube 4 feet in length, polished inter- 

 nally and hermetically closed at both ends by plates of rock- 

 salt. The heat-rays emanating from a Leslie's cube at a dis- 

 tance of 8 inches, after passing through the air, entered the 

 tube through one of the rock-salt plates, issued again by the 

 second, and, having traversed a second layer of air, fell upon 

 the conical reflector of a thermopile. 



By this method all the heat-rays for which air is atherma- 

 nous were evidently absorbed ; and hence those entering the 

 tube would pass through it unhindered when the tube was 

 filled with dry air. This would also be the case for the chief 

 constituents of air, oxygen and nitrogen, as well as for hy- 

 drogen or even a vacuum, since these media allow all rays to 

 pass. In this way the result would be obtained, that the four 

 gases exhibit the same behaviour as a vacuum towards heat- 

 rays from a dark source. 



In a subsequent series of experiments, Tyndall employed a 

 small chamber fixed air-tight onto one of the rock-salt plates, 

 its opposite sides being similarly closed by one of the faces of 

 the cube. During the experiment the chamber was exhausted ; 

 so that the heat-rays, before entering the tube, could not suffer 

 any loss except by their passing through the rock-salt. Never- 

 theless the results differed very little from the preceding ones. 

 A small but not constant fraction of the incident rays (accord- 

 ing to Tyndall about 0'33 per cent.) was lost. Such a result 

 was indeed to be expected ; for since the rays had still to tra- 

 verse a layer of air before reaching the thermopile, only those 

 rays could arrive there for which air is diathermanous. But 

 the rock-salt also may have retained some of the incident rays. 

 Melloni had, it is true, pronounced rock-salt to be completely 



* Phil. Mag. [IV.] vol. xxii. p. 169. 



