STRUCTURE AND LIGHT OF THE SKY. 261 



obliquely to it are transmitted. It is easy, therefore, to see 

 that, from the position in which the prism must be held to 

 transmit or to quench the light of our incipient cloud, we 

 can infer the direction of the vibrations of that light. You 

 will be able to picture those vibrations without difficulty. 

 Suppose a line drawn from any point of the " cloud " per- 

 pendicular to the illuminating beam. The particles of ether 

 along that line, which carry the light from the cloud to the 

 eye, vibrate in a direction perpendicular both to the line 

 and to the beam. And if any number of lines be drawn 

 in the same way from the cloud, like the spokes of a wheel, 

 the particles of ether along all of them oscillate in the same 

 manner. Wherefore, if a plane surface be imagined cutting 

 the incipient cloud at rigbt angles to its length, the vibra- 

 tions discharged laterally are all parallel to this surface. 

 This is the plane of vibration of the polarized light. 



Our incipient blue cloud is a virtual Nicol's prism, and, 

 between it and the real prism, we can produce all the 

 effects obtainable between the polarizer and analyzer of a 

 polariscope. When, for example, a thin plate of selenite, 

 which is crystallized sulphate of lime, is placed between 

 the Nicol and the incipient cloud, we obtain the splendid 

 chromatic phenomena of polarized light. The color of the 

 gypsum-plate, as many of you know, depends upon its 

 thickness. If this be uniform, the color is uniform. If, on 

 the contrary, the plate be wedge-shaped, thickening grad- 

 ually and uniformly from edge to back, we have brilliant 

 bands of color produced parallel to the edge of the wedge. 

 Perhaps the best form of plate for experiments of this 

 character is that now in my hand, which was prepared for 

 me some years ago by a man of genius in his way, the late 

 Mr. Darker, of Lambeth. It consists of a plate of selenite 

 thin at the centre, and gradually thickening toward the 

 circumference. Placing this film between the Nicol and 

 the cloud, we obtain, instead of a series of parallel bands, a 



