making Experiments upon Elliptic Polarization. 247 



ments we compute, by the last formula, the intensity of re- 

 flexion for speculum metal at a perpendicular incidence, in 

 which case /* = 1, we shall find I = -583. This is considerably 

 lower than the estimate of Sir William Herschel, who, in the 

 Philosophical Transactions for 1800 (p. 65), gives "673 as the 

 measure of the reflective power of his specula. The same num- 

 ber, very nearly, results from taking the mean of Mr. Potter's 

 observations.* It might seem therefore that the formula is in 

 fault ; but I am inclined to think that the metal which I em- 

 ployed had really a low reflective power. Its angle of maxi- 

 mum polarization was certainly much less than that of the specu- 

 lum metal used by Sir David Brewster,f who states the angle 

 to be 76, whereas in my experiments it was only about 73-|- ; 

 and 'any increase in this angle, by increasing the value of M, 

 raises the reflective power. On the other hand, the maximum 

 value of |3 (when a = 45) was greater than that given by Sir 

 David Brewster, namely, 32 ; and any increase in /3 tends also 

 to increase the reflective power. Now it is not unreasonable to 

 suppose that the highest values of both angles may be most 

 nearly those which belong to the best specula ; and accordingly, 

 if we take 76 for the incidence of maximum polarization, and 

 retain the maximum value of /3, namely, 34 37", which results 

 from my experiments, we shall get M = 3-68, ^ = 66 16', and the 

 value of / at the perpendicular incidence will come out equal to 

 662, which scarcely differs from the number given by Herschel. 

 It is clear from what precedes that the optical constants are 

 different for different specimens of speculum metal, and this is 

 no more than we should expect, from the circumstance that the 

 metal is a compound, and therefore liable to vary in its optical 

 properties from variations in the proportion of its constituents ; 

 but I am disposed to believe that the same thing is generally 

 true, though of course in a less degree, of the simple metals ; 

 so that in order to render the comparison satisfactory, tne 

 measures of intensity should always be made on the same spe- 



* Edinburgh Journal of Science, New Series, VOL. in. p. 280. 

 t Philosophical Transactions, 1830, p. 324. 



