the Laws of Chromatic Dispersion. 255 



The observer gives the foregoing as the means of several ob- 

 servations, stating that, while the individual observations some- 

 times varied as much as ]/, he does not think that the mean 

 values can differ much above 30" from the truth. A difference 

 of 30" corresponds to one of 0*00035 on the index of B, and of 

 0-0003 on that of G. 



In testing by these observations the general formula 



it is needful, owing to the absence of H, to vary the method of 

 procedure, and to supply the index of H by calculation. The 

 exponent n must in this instance be found by trial. Then in 

 determining e n and a n we must make 



(5B W + 3(T + D n ) - (5G W + 3F" + E ra ) 



r V "B + *C + "D ' \ ^G + ^ + »W 



and 



1/ B W + C W + D W +E W + F , + G tt _/ r B ,, _ 1 _C w J _D W , E w , F ra Q n \\ _ 



With the exponent w=3*l we obtain from these formulae 

 log e n =0-1972574 and a* = 0-008186. 



The index of H calculated from the formula fi=zX n -i a n 



e 



is 1 '697760, and the differences between the other indices cal- 

 culated from that formula and their observed values are — 



These differences all lie within the assigned limits of experimental 

 error. 



The extrusions or displacements of the fixed lines from their 

 normal relative positions, are with the calculated indices as 

 follow : — 



b x~ c x~ <**+ e x + f x + g x - h x - 



0001675 0-000615 9001103 0001832 0001624 0-000031 0002238. 



This series corresponds to the regular type, thus removing 

 the anomaly presented by this medium, according to the obser- 

 vations of Professor Powell. Gladstone's observations were 

 made on the same specimen of the oil of cassia redistilled ; and 

 his statement that the line H is imperceptible raises a strong 

 probability that Professor Powell, in determining the index of 

 that line, had been deceived. This probability is enhanced by. 



