AND THE THEORY OF LIGHT. 27 



bright azure which tinges the mountains of the distant 

 landscape." He adds : " As we ascend in the atmosphere 

 the deepness of the blue tinge gradually dies away; and 

 to the aeronaut who has soared above the denser strata/ 

 or to the traveller who has ascended the Alps or the 

 Andes, the sky appears of a deep black, while the blue 

 rays find a ready passage through the attenuated strata 

 of the atmosphere.^^ What is to be remarked here is, 

 that the particles or molecules of air in the higher strata 

 of the atmosphere are not of sufl&cient thickness, to use 

 Arago's expression, to reflect blue. They reflect no colour 

 at all ; they appear black. If we adhere to the argument 

 of Arago and Humboldt, supported by the authority 

 of Newton, we are, in this case, compelled to acknow- 

 ledge that it is the vapours of the atmosphere in the 

 lower strata which reflect blue, and that the air when 

 pure reflects no colour whatever, or that it reflects dif- 

 ferent colours according to its difierent degrees of density. 

 On the latter supposition it cannot be the size of mole- 

 cules which is the cause of colour, but their compactness 

 or the distance of one molecule from another, or, in other 

 words, the ratio of the molecules which reflect light to the 

 spaces which reflect no light ; or otherwise the ratio of 

 light to the shade, which is exactly the principle on which 

 I explained the card experiment and the candle experi- 

 ment ; at all events, that the blue is not owing to the size 

 of the molecules of air is evident from the facts supplied 

 by science itself in these examples. 



44. There is a respect due to an old and an established 

 doctrine which even criticism ought to venerate, a respect 

 which demands that if we attempt to pull down we shall 

 also endeavour to build up. But the colour of the sky 

 cannot be experimented on in its native element. Our 

 information must be drawn chiefly from accurate observa- 

 tion, assisted of course by such experiments as can be 



