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



iodine and thus rendered opaque. Notwithstanding the aug- 

 mented absorption by the iodine, I still found a larger amount 

 of heat fall on the pile in both cases on pouring the opaque 

 liquid into the cell. When a glass cell J an inch deep was used, 

 the heat falling on the pile was raised 3 per cent, by the opaque 

 liquid. When a glass cell 1*2 inch deep was employed, a similar 

 increase of fully 5 per cent, was produced, and this by the intro- 

 duction of a liquid so opaque that at this thickness it intercepted 

 all the light of the brightest sun. 



What, now, is the explanation of this singular action ? Review- 

 ing the experiments detailed in the foregoing paper, they are 

 seen to divide themselves into two groups : — 



First. Those where an increased transmission attended the 

 introduction of moderately diathermic liquid films between plates 

 either of rock-salt or of glass ; and 



Second. Those where this increase occurred with considerable 

 thicknesses of the same liquids as before, but only when the 

 liquids were enclosed between glass plates. 



The most probable explanation of the first group of facts at 

 once suggests itself — namely, that the introduction of the liquids 

 diminishes the reflection taking place from the inner surfaces of 

 the walls of the cell. In the empty cell a double reflection takes 

 place when a pencil of rays passes from one plate of the cell to 

 the other through the intervening space of air. If the cell be 

 filled with a liquid whose refractive index is the same as that of 

 the walls of the cell, this loss by reflection from the inner sur- 

 faces will be abolished; hence such a liquid, if its absorbent 

 power were nil, or extremely low, would increase the transmis- 

 sion when replacing air in the interior of the cell. In the fol- 

 lowing Table are given the refractive indices of the substances 

 employed ; as these were not determined from the specimens 

 used, they can only be regarded as approximately correct ; that of 

 glass is especially doubtful. The indices for rock-salt and bichlo- 

 ride of carbon were kindly furnished me by Dr. Gladstone, and 

 are from his own recent determinations : — 



Index of Refraction for the line A. 



Plate-glass (mean) .... 1*525 



Rock-salt 1-537 



Bisulphide of carbon (mean) . 1'616 



Bichloride of carbon . . . 1*456 



Chloroform - 1*441 



From this Table it will be seen that the index of refraction of 

 bichloride and bisulphide of carbon is sufficiently near to rock- 

 salt to greatly diminish the reflection from the interior of the 



