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



as the walls of the cell ; for a similar effect will take place, in a 

 greater or less degree, if the density of the body introduced ex- 

 ceed that of the medium it replaces. 



It was easy to see the truth of this reasoning by tracking the 

 course of a pencil of luminous rays through the empty and full 

 cell. The light of the red-hot spiral that formed the source of 

 heat was received on a semitransparent screen placed behind a 

 glass cell half an inch deep. The beam was seen on the screen 

 as a luminous circle 1*7 inch in diameter. Filling the cell with 

 bisulphide of carbon, the diameter of this circle contracted to 

 16 inch. When the cell was half full, the effect of the compa- 

 rison between the upper and lower half of the T? . Q 

 circle was very striking; this is shown in 

 fig. 3. With the cell 3 inches deep the 

 luminous circle c c was diminished from 2*4 

 inches, when the cell was empty, to 1*8 

 inch, when it was filled with bisulphide of 

 carbou . 



This observation explains the effect no- 

 ticed when bisulphide or bichloride of carbon 

 was poured into deep glass cells ; for though 

 sometimes concentrated by a rock-salt lens, the rays from the 

 source were always more or less divergent, and overlapped the 

 area of the pile. Lessening the divergency of the incident rays 

 ought to diminish the effect ; aud this, by experiment, was proved 

 to be the case. Hence also with parallel rays no concentration 

 of the beam could take place on the introduction of the liquid, 

 and with convergent rays a decreased amount of heat should 

 reach the pile when a dense medium occupies the cell. 



But why should not an increase occur with rock-salt as well as 

 with glass cells ? This anomaly puzzled me a good deal at first ; 

 for, on consulting the Table of refractive powers, there does not 

 appear to be any reason why such a difference should exist. I 

 think, however, the answer to this question is mainly to be found 

 in the difference between the diathermancy of rock-salt and 

 glass. Rays of heat pass through the former substance with 

 but little alteration in their quality or quantity ; but in passing 

 through the latter, their character is changed as well as their 

 amount lessened ; after passing through one plate of glass they 

 traverse a second plate with less absorption. It may be that the 

 bisulphide of carbon absorbs somewhat the same class of thermal 

 rays as glass ; the heat-tint or thermochrose of the two bodies is 

 possibly alike. Assuming this to be the case, when a cell with 

 glass sides is used, the absorption of the liquid would be so far 

 reduced that the action of the causes tending to increase the 

 transmission would become evident. ' On the other hand, with a 



