ON LIGHT. 289 



fifth pale bluish-green, white, pink. After these the 

 colours grow paler and paler, alternately bluish-green 

 and pink, and can hardly be traced beyond the seventh 

 order. 



(72.) None of these tints are pure prismatic colours. 

 To see them to the best advantage the bubble with its 

 glass shade should be placed out of direct sunshine, 

 where only dispersed light, such as that of a cloudy sky, 

 shall fall on it. Or, the illumination of the rings may be 

 effected by a thin semi-transparent paper, or a groxtnd- 

 glass screen interposed between them and the incident 

 light. And if, instead 'of illuminating this with the direct 

 light of the sky, the coloured rays of a spectrum, formed 

 by passing a sunbeam through a glass prism, be thrown 

 upon it, the composite nature of their tints will be at 

 once apparent. If all the rays but those at the red end 

 of the spectrum be excluded from the illuminating beam, 

 the rings will appear wholly red, separated by black 

 intervals, and much more numerous. And if, now, the 

 colour of the illuminating light be changed, so as to pass 

 in succession through the whole prismatic scale of tints 

 orange, yellow, green, &c., from the red to the violet 

 the colour of the rings will undergo a corresponding 

 change, the dividing intervals preserving their blackness, 

 but their number still continuing greater than in white 

 light. But, besides this, a very remarkable phenomenon 

 will be observed. The rings contract rapidly in diameter 

 as the colour of the illumination changes, being a maxi- 

 mum for a red and a minimum for a violet illumination ; 

 and if, by a slight movement given to the prism, the 



T 



