504 Dr. Gr. J. Stoney's Analysis of 



third of the millionth of the millionth) of a second of 

 time. 



The micro-jot is the millionth part of a jot. 



j\ the air- jot of time, is the time that the ray of mean refran- 

 gibility takes to advance one tenth of a millimetre in air. 

 Accordingly j 1 = fj, l j, where /x x is the refractive index of 

 air for the ray of mean refrangibility. If we regard 

 the ray whose wave-length is 5000 tenthet-metres as the 

 mean ray, then (jl 1 is very nearly 1*000280. (See British 

 Association catalogue of oscillation-frequencies in the 

 Brit. Assoc. Eeport for 1878.) 



X is the wave-length-in-air of a ray, expressed in tenthet- 

 metres, as determined by Rowland*. 



k is the " inverse-wave-length" i. e. 10 6 /^. It is the number 

 of wave-lengths-in-air which occupy one tenth of a milli- 

 metre. 



T is the periodic time of the waves of a ray of light, 

 expressed in micro-jots. Hence T = /ul\, in which /a is 

 the refractive index of air for the ray of wave-length \. 



N is the oscillation-frequency in each jot of time, i. e. the 

 number of the oscillations of the waves of a ray in each 

 jot. It = k/jjl. 



Of these quantities K" is obviously the one which is best 

 adapted to theoretical investigation. It is, however, \ that is 

 observed. From this k can be accurately deduced, but N = k/jjl 

 cannot be accurately obtained till we know the value of fi for 

 different parts of the spectrum. We may, however, in a 

 theoretical investigation, use, instead of N, any quantity pro- 

 portional to it, e. g. ftjN, where ^ has the value assigned to 

 it above. We shall call this n. It is the oscillation- frequency 

 in each " air-jot " of time. Accordingly 



n is the oscillation-frequency in each air-jot of time. It 



= fl 1 N = K./l 1 /fl. 



Now Ketteler's observations on the dispersion of air, though 

 not sufficient for the general determination of yu,, are enough 

 to satisfy us that /n does not anywhere differ more than a very 

 little from fi 1} its mean value. And accordingly, in comparing 

 our results with observation, we shall regard faf/i as unity, 

 and treat k (the quantity furnished by observation) as if it 

 were identical with n (the quantity required by theory). 

 With this we must be content until adequate determinations 



* X is accordingly about l/6000th more than the wave-length as deter- 



o 



mined by Angstrom. 



