STRUCTURE AND LIGHT OF THE SKY. 263 



we should have the solar light reflected ; in two others un- 

 reflected. In fact, out of such a solitary beam, traversing 

 the unilluminated air, we should be able to extract every 

 effect shown by our incipient cloud. In the production of 

 such clouds we virtually carry bits of the sky into our 

 laboratories, and obtain with them all the effects obtainable 

 in the open firmament of heaven. 



The real sky is, as I have said, less perfect than our 

 artificial one may be made. For, mingled with the infini- 

 tesimal particles which constitute the true matter of the 

 sky, there are others too coarse to scatter perfectly po- 

 larized light at right angles to the solar beams. Hence, 

 when the brilliancy of the sky is diminished to the utter- 

 most, there is still a residue of light; the extinction is 

 partial, and not total, as in the case of our incipient cloud. 

 Let us consider this matter. The perfect polarization can 

 only be produced by excessively minute particles ; imagine 

 them growing gradually larger as they actually do in our 

 experiments. The extinction by the Nicol is perfect as 

 long as the polarization is complete. But what would you 

 expect? Manifestly, that after a time the polarization 

 would cease to be perfect. But here again the relation of 

 the size of the particles to the size of the waves must come 

 into play. In relation to the blue waves the particles are 

 larger than in relation to the red ; the blue waves, there- 

 fore, will be the first liberated from a condition dependent 

 on the smallness of the particles. They will first escape 

 from the trammels of polarization ; and on their liberation 

 they exhibit an azure far purer and more brilliant than that 

 produced by the first precipitation of the particles. Could 

 we overarch ourselves with a sky of this color for a single 

 day, it would make us discontented with our present lack- 

 lustre firmament ever afterward. It will be observed that 

 in all these cases reason and experiment go hand in hand, 

 the one predicting, the other verifying; every such verift- 



