GASEOUS SPECTEA UNDER HIGH DISPERSION. 457 



APPENDIX II. 



PROF. ALEX. S. HERSCHEL'S LATER REMAEKS ON PLATE LXXVIIL, AND AN 

 IMPROVABLE POINT IN ITS SCALE OE REPRESENTATION.— May, 1884. 



Two distinct 'spectra closely resembling each other, together form the Green band of 

 Carbonic oxide figured in the. Plate LXXVIIL; one of which consists of single linelets, and 

 the other of slowly opening double ones, or of linelets coupled together in close pairs. If the 

 whole unilinear spectrum is shifted together to the left until its first line coincides with the 

 leading one at the band's least refrangible edge, all its lines fall nearly into coincidence either 

 with the middle place, or else with one or other side-line of the several linelet pairs of which 

 the remaining bilinear portion of the spectrum is composed. 



An ideal spectrum is placed for comparison above and below the two component spectra 

 of the band, forming an arithmetical series of micrometer-revolution, or of sensible dispersion 

 intervals, representing with a suitable scale-unit of measurement, the series of natural numbers 

 1, 2, 3, 4, &c. The necessary data for replacing this array of gradually increasing micrometrical 

 intervals by a similar, more scientific arithmetrical progression having a wave-number unit 

 instead of a micrometrical dispersion one for prime measure of its successive terms or intervals, 

 was not exactly procurable in the state of the instrument's adjustment ; but the small regular 

 differences which are noticeable in the Plate between the two observed spectra and the 

 arithmetical comparison series of micrometrical line intervals, are, it may be remarked, of 

 exactly the description in direction and in varying magnitude which the provisional substitution 

 in the Plate of a micrometrical for a wave-number series of successive intervals would 

 correspond to, and serves sufficiently to account for. Were such a replacement of the 

 provisional array by a corresponding wave-number one made with perfect certainty and 

 correctness, it would seem to be a safely legitimate assumption to conclude, that the small 

 visible departures of the ideal from the observed spectra which their comparison together 

 exhibits on the Plate, would all, then, be quite satisfactorily obliterated and removed. 



A. S. H. 



To which I, as the Observer, may probably be allowed to add, — not " quite removed." For 

 wherever there is numerical observation aiming at exactness, there will always be errors of the 

 observer to some extent. But I must confess I have been well pleased to see the smaller 

 amount of the apparent errors of observation, when one Natural System of spectral lines is 

 compared with another, as in the lowest compartment of Plate ; — than when either one of them 

 is contrasted with the artificial screw-unit scale, as shown first at the top of the plate, and then 

 near the bottom of it ; viz., Plate LXXVIIL— C. P. S. 



