XXXV.] COLOURS AND THEIR COMBINATION. 109 



perceiving light extends over but one.^ The number of luminous 

 undulations to the second of the faint purple-red rays beyond the 

 violet is about twice that of the extreme red rays at the opposite 

 extremity of the spectrum, and their ■wave-length consequently 

 about one-half. Satisfactory observations on the subject are 

 difficult, in consequence of the faintness of the purple rays, 

 which makes them often difficult even to see, and still more 

 difficult to identify with the red. !N"eYertheless, the transition 

 from violet to red through purple is stated to be perfectly visible 

 with good light and a good instrument ; so that the colours of 

 the spectrum really occur in the following recurrent order, ^ like 

 that of the notes of music : — 



Eed, orange, yellow, yellowish green, green, bluish green, 

 azure, indigo, violet, purple, red again. 



If there is any doubt of the actual recurrence of the same 

 tint of colour in the spectrum, though I believe there is none, 

 this theory of an octave in the spectrum would still be in the 

 highest degree probable ; not only from the analogy of sound, 

 but also from the visible fact that violet and purple are inter- 

 mediate colours of tint between blue and red, just as orange is 

 visibly intermediate between red and yellow. 



It is obvious that this fact of an octave in the spectrum is an 

 ultimate fact of sensation, not to be accounted for by any law 

 of habit, or accounted for at all. 



If the series of colours in the spectrum is recurrent, it is The series 

 using only another word for the same fact to say that it is ?f colours 

 circular. A circidar arrangement of the colours of the spectrum spectrum 



is circular, 



1 According to Sir John Herschel (Good Wonls, August 1S05), the 

 luminoTis vibrations of the extreme red number 399,401,000,000,000 to 

 the second, and those of the exti-eme violet 831,479,000,000,000 ; so that 

 the latter are a little more than twice as numerous as the former, and the 

 power of vision extends through a little more than an octave. No such 

 determinations, however, can be anythiug more than approximative. The 

 measurement of the lengths of the waves, and of their luimber to the 

 second, is not subject to any great error, but it is a matter of great uncer- 

 tainty where the extreme red and the extreme violet are ; for the illumi- 

 nated part of the spectrum graduates into darkness, and its extreme limits 

 will be talcen as at different points, according to the goodness of the light 

 and of the in.stvument, and perhaps also according to the peculiarity of tlio 

 observer's eyes. 



~ For the fact of this veeurrencp, see Professor Gras.smann's paper 

 already leferred to. 



