THIRD SERIES.— VARIABLE POLARIZABILITY OF HEAT. X79 



When this experiment was performed with dark heat (which, according to Mel- 

 LONi, ought to give the greatest effect), not the slightest movement of the Galva- 

 nometer-needle was observable on removing the screen, during a far longer space 

 of time than is ever in practice allowed for the absorption of heat. This experi- 

 ment ought to be considered quite conclusive. 



12. M. Melloni had hinted that the different dimensions of the Sources of 

 Heat, and the various angles under which the rays fell on the mica plates must 

 materially affect the results ; and, as I was quite convinced that operating with 

 parallel rays was the most correct method, I proceeded to repeat my experiments 

 on his plan, with a salt lens placed in front of the source of heat so as to render 

 the rays parallel ; I also removed the polarizing and analyzing plates to a consider- 

 able distance from the pile, and afterwards varied their distance in order to see 

 whether any adequate explanation of the discrepancy could thus be obtained. 



18. The apparatus was arranged in the following way. P, the pile ; A, a 

 square pasteboard tube to protect it from currents of air ; I and K, the analyzing and 

 polarizing plates ; B, a moveable screen ; L, a rock-salt lens, in the focus of Avhich 

 is placed S, the source of heat. In these experiments the distance from the centre 

 of the pile to the centre of the first mica plate or P I was 12 inches ; PS = 24 

 inches. With this apparatus I found my former conclusions fully confirmed. 



The apparent polarization was somewhat increased, as I had anticipated from the 

 rays falling more nearly at a constant angle when previously rendered parallel ; 

 but the different polarizability of the different kinds of heat was even more dis- 

 tinctly marked than ever ; whilst the distance of the mica plates from the pile was 

 now such as to reduce to insignificance any effect of secondary radiation, had such 

 before been sensible. 



14. In prosecuting these experiments, most of which were repeated man}'- 

 times under various circumstances, I remarked more distinctly than formerly the 

 influence of particular states of combustion of the source of heat upon the index 

 of polarization, and the accidental variations to which this gives rise on different 

 days, and even during the progress of an experiment. Heat from brass about 700° 

 I have generally found the most uniform on different days, though there occasion- 

 ally occurs in a series of experiments, considerable deviations from the mean. The 

 LocateUi Lamp seems subject to greater variations, and the Argand still more ; in- 

 deed, I have found it so impossible to maintain an Argand-lamp in a uniform 

 state of combustion, even for a quarter of an hour, that I have lately abandoned the 



