A. A. MieheUon—rnUrferena Experiment iM7 



A very considerable improvement Is effected by the substitu- 

 tion of Fresnel's "bi-prism," though in this case one of the 



chief points of interest is h.st; namely, the change of breadth 

 of the bands with alteration of angle. The phenomenoo is also 

 Complicated by the dispersion o\' the gla^>. 



The following method has the advantage of being easily 

 within the reach of even unskilled persons, and of giving the 

 required result in all its purity. 



Fasten a piece of plane glass <il>. fig. 1. (selected plate glass 

 Mill answer, though sextant glass is much better) to a right 

 angled prism of wood <1 ; then press <1 against a second pii 



of plane glass ce by means of a little soft wax, so that the angle 

 abc is a little less than a right angle. If the light from a dis- 

 tant electric arc lamp be allowed to fall on the glasses (which 

 should be silvered on the front surfaces) two rajs s / s y/ starting 

 in the same phase, after reflections aty, g and h, k, respectively, 

 will meet somewhere in the line bs in the same phase, the two 

 paths being exactly equal.* Accordingly this line is the locus 

 of the central bright fringe and the system of fringes may 

 readily be observed through a lens of about an inch focal length 

 whose axis is anywhere along this line. 



The figure is draw r n for clearness in the vertical plane con- 

 taining the source, and if the intersection of the two surfaces 

 (projected at b) were at right angles with sb it would be neces- 

 sary to deflect part of the light to one side by a transparent 

 mirror. But the appearance of the fringes is in no wise 

 changed if the apparatus be rotated through a horizontal angle 

 «. The line bs is then rotated through 2a, and there is no 

 difficulty in observing the fringes directly. 



It will be noted that no adjustments whatever are necessary. 



All that is required is that the surfaces be fairly true and the 

 angle slightly less than 90°. This last condition can be readily 

 fulfilled by making the tw T o images of a distant object nearly 

 coincide. 



*This is true whether the surfaces form equal angles with sb or not. 



