168 LORD RAYLEIGH ON THE COLOURS OF THIN PLATES. 



One discrepancy between the diagram and the above descriptions will at 

 once strike the reader. According to the diagram, the red and purple of the 

 first order are superior to those which follow, whereas Herschel says that the 

 red of the first order hardly deserves the name. Judged by the standard of 

 the spectrum red at (24), this criticism would apply to them all; but the 

 question is as to the relative merits of the various reds. The explanation 

 depends upon considerations of brightness, of which the curve takes no account. 

 If we refer to Table III., we see that at 1846 the red component is 1*20, but 

 that at the corresponding point for the red of the second order (between 3600 

 and 3800) it rises to about 30. The deficiency of brightness in the first order 

 goes a long way by itself to explain the apparent inferiority, for dark red gives 

 rather the impression of brown ; but if there is the slightest admixture of white 

 light, the comparison is still more unfair. It would be useless, for example, to 

 take the colours from an air-plate between lenses. The feebly luminous red of the 

 first order is then drowned in a relatively large proportion of white light, which 

 tells much less upon the brighter, though less pure, red of the second order. This 

 complication does not arise when soap-films are employed, and the red of the 

 first order is evidently much improved ; but the rapidity of transition at this 

 part of the scale renders observation difficult. The best comparison that I have 

 been able to make is with the aid of a beautiful mica combination kindly lent 

 me by Eev. P. Sleeman. When this is examined in a dark room between 

 crossed nicols, and lighted brilliantly from a part of the sky near the sun, the 

 red of the first order is seen in great perfection, and I had no difficulty in 

 believing it to be superior to that of the second order. It is not very easy to 

 bring the rivals into juxtaposition under equal brightnesses ; but there is, I 

 think, no reason to doubt that the first order would come off victorious. The 

 composition of the lights will be understood by reference to Table II. 



§ 10. The only colours which can be said to make any approach to spectrum 

 purity are the yellows of the first two orders, and the blue and green-blue of 

 the second and third orders respectively. There is a corresponding difficulty 

 in obtaining good greens by absorption. To do so it is necessary that the 

 transmitted spectrum should terminate at two pretty well-marked points ; in 

 the case of red the difficulty is much less, all that is requisite being that the 

 transmission should increase rapidly as the refrangibility of the light diminishes. 



Besides the absolute brightness, there are two other circumstances which 

 may influence the estimation of the colours of thin plates as normally presented. 

 It is probable that in some cases the colours are much affected by contrast with 

 their neighbours. To this cause we may attribute the difficulty in observing 

 the transition between the reds and blue-greens of the fourth and higher 

 orders. As the nearly neutral transition-tint is approached from either side, 

 the effect upon the eye is improved by contrast, so as largely to compensate 



