212 Prof. Forbes on the Refraction a?id Polarization of Heat, 



36. I fitted up two other bundles of mica- plates, in square 

 pasteboard tubes of the kind described, which were marked 

 E and F, the other plates being occasionally substituted, in 

 order to verify the results, and to show that no accidental pe- 

 culiarity of the plates could account for the differences ob- 

 served. My experiments were usually made thus. The tube 

 E was fixed to the pile ; the tube F, containing the other 

 plate, had an index, which pointed to 0° when the two plates 

 were parallel, to 90° when they were at right angles, &c. Five 

 observations were taken ; at 0°, 90°, 180°, 270°, and again at 

 0°. The mean of the first and last were taken ; then the 

 mean of this, and the indication at 180°, and the difference 

 between this and the mean at 90° and 270°, was considered 

 as the polarizing effect. An example will best illustrate this : — 



1834, Nov. 26. — Brass heated hy Alcohol: 5^ inches from 

 centre of Pile. 



Deviation. 

 Analysing plate (E) at 0°; polarizing plate (F) at 0° . . 6J° 



90 ..5i 

 180 ..7 



270 . . 6 



Mean at 0° . . . 6°'9 

 180° ... 7-0 



Mean, 6 -91 Ratio 100 : 81, orl9 per cent. 



Mean at 90° and 270°, . . 5 '6 J polarized. 



The general concordance of these experiments will be gathered 

 from the following list of results. 



37. With non-luminous heat from ir<z55 about 700° ; ratio 

 of effect, when plates E and F were parallel and crossed, 

 100:78; 100:76; 100:80*; 100:81 (from five observa- 

 tions each), with plates E and A (from three observations 

 each), 100:74; 100:59; 100:68; 100:60; with A and B, 

 ratios 100: 78; 100:72. 



38. With non-luminous heat from mercury, about 500°, 

 plates E and F; 100 : 77 ; 100 : 90, plates E and A ; 100 : 88; 

 with A and B, 100:78. 



39. But even with heat from water below the boiling-point, 

 I was able, by the improved method of observing the galva- 

 nometer, art. (5), (6), to establish completely the polarizing 

 effect. One series of six comparisons (conducted as in (20),) 



* Plate B was used to polarize in this experiment. 



