Light traverses Transparent Media. S3 "> 



the light from the sky came directly, and at the distance of a 

 few yards from this window, I held up before me the plate (a 

 small one) with a piece of black velvet behind it; then holding 

 a piece of fine platinum wire (a very fine steel needle would 

 answer as well,) directly before the pupil of one of my eyes, I 

 viewed with the same eye the images, produced by the sur- 

 faces of the glass, of the fine line of light which was shown by 

 the platinum wire. When the surfaces are even very slightly 

 inclined to each other, it is immediately detected by their two 

 images being distinctly separated ; but when both surfaces are 

 accurately perpendicular to the same incident rays, the two re- 

 flected images become blended into one. Knowing it to be a 

 very difficult task to produce two such flat and parallel sur- 

 faces on a small scale, I did not attempt to grind and polish 

 a plate myself, but chose rather to avail myself of the kind- 

 ness of a friend, to select a small plate, to suit my purpose; 

 and I considered myself fortunate in finding one amongst a 

 large number of plates in which no inclination of the surfaces, 

 could be detected in any direction. One of the best methods 

 for examining the truth of flat surfaces, I find to be that of 

 pressing together alternately the surfaces of the two pieces of 

 glass, when, if there be any appretiable curvature, it will be de- 

 tected by an appearance of Newton's rings ; care must, how- 

 ever, be had that the curvature does not arise from a too 

 great degree of pressure. 



My two pieces of glass from the plate just mentioned were 

 attached, the one to the pivot of a divided circle, and the other 

 to the arm turning on this pivot, which carried the reading 

 microscope. The circle was something less than 18 inches 

 in diameter; and from the examination I made of the values 

 of the divisions, I think the angles, in the experiments which 

 were deduced from the divisions for even tens of degrees, could 

 not be wrong more than five or six seconds. For the other 

 divisions it is possible the error might be several times as 

 much, but nothing, as I "have proved, which could affect funda- 

 mentally the results. The heliostat I used was the one I have 

 described in a former number of the Magazine, (Lond. and 

 Edinb. Phil. Mag. and Journal, vol. ii. p. 6,) and the lens and 

 two mirrors were attached to it, so that the line e p, fig. 1, 

 might be horizontal. The plane of the circle was adjusted 

 accurately to be parallel to this line, and perpendicular to the 

 horizon, by three screws forming the feet of its support. The 

 two glasses being cemented to plates of brass were also ad- 

 justed by three screws each, so as to be perpendicular to the 

 plane of the circle. The distance ef in the figure was mea- 

 sured by means of a chased steel screw set accurately in a 

 brass frame ; this screw was set perpendicular to the horizon, 



