604 Stoney — Cause of Double Lines in Sipectra. 



The spectra of the heavier monads, Cu, Ag, and Au, have not been sufficiently 

 explored to be used here for purposes of illustration. They appear to consist of 

 double lines, one of the constituents of which is often faint, and has been recorded 

 as a satellite, indicating that the elliptic partials in these cases ai'e open ellipses 

 ajDproaching in form to the circle. In the spectra of elements of higher atomicity 

 triple lines present themselves, the discussion of which lies beyond the scope of 

 the present Paper.* 



It appears from the investigation developed in this Paper, that when the lines 

 of a sjpectrum are double, it is possible to extract from the observations a great 

 deal of information as to each of the elliptic motions which, when put together, 

 make up the actual motion of the electron. It remains to consider whether it is 

 possible to combine them, and so to ascertain what the actual motion is. 



Where, as in the case of hydrogen, such a law as Balmer's can be empirically 

 obtained, there can be no doubt that all the lines (or pairs of lines) connected by 

 so explicit a law, arise from the successive partials of the actual motion of one 

 electron. Neither can there be a reasonable doubt where, by graphical processes 

 or by using approximate empirical formulae like that of Professors Kayser and 

 Runge or that of Professor Rydberg, it is found possible to pick out the lines 

 belonging to a natural series, especially when, as generally happens, the lines 

 80 indicated are found by observation to have characteristics in common. The 

 whole of such a series we may with confidence refer to the motion of one of the 

 electrons in the molecules of the gas. In both these cases observations with the 

 spectroscope will give much information about the several partials of the motion 

 of the electron. 



But all this information falls short of being sufficient to enable us to give 



hydrogen, but has not yet been detected in the spectra of the others because of its very low refi-angibility 

 in them, and of which P, D, and S lie so far in the ultra-violet in the spectrum of hydrogen that they have 

 not yet been observed. 



* In the present Memoir we are dealing only with perturbating forces that are feeble. If the pertur- 

 bating forces were comparable with the forces which produce the orbit which we select as the dominant 

 orbit, triple lines might arise. For example, a motion represented by 



X = a cos {t,t + a) . cos {-qt + j3) . cos 6t, 



in which ^, t] and 6 have nearly equal values, would produce a triple line in association with a single line 

 far separated from it. 



It thus appears that even a vibration in a straight line may be such as will produce triple lines. And, 

 of course, more complex motions can generate them under fewer restrictions. But in all cases, if (as is 

 always possible) the motion be resolved into motion in a simpler 'dominant' orbit affected by pertur- 

 bations, the perturbations, or some of them, must be of large amount, i. e. must have periodic times which 

 are comparable with that of the motion in the ' dominant ' orbit. In fact, slow perturbations give rise to 

 close equidistant lines, so that triple lines, other than those that are equally spaced, can arise only in 

 cases where the orbit corresponding to our dominant orbit is not predominant over some of its perturbations. 



