52Q Prof. R. TV. TVood on Separation of Close 



must be used, placed between the two halves of the con- 

 denser, since the rays which traverse the block must be 

 parallel. If this is not the case, different pencils will traverse 

 different thicknesses, and will be differently polarized. 

 Moreover, one half of the light is lost at the start by the 

 polarizing nicol. This difficulty was overcome by employing 

 a large double-image prism, and subsequently analysing by 

 a double-image prism. In this way, with proper orientation 

 of the prisms, the two images containing only D 2 light were 

 superposed, the D 1 images (of one half the intensity) lying 

 to the right and left. By this expedient the D 2 image had 

 the full intensity, except for the loss by reflexion from the 

 six transparent surfaces of the prisms and quartz block. 



A rotation of 90° of the plane of polarization is produced 

 by a quartz plate '032 mm. in thickness for sodium light, 

 consequently the plate must be plane-parallel to within con- 

 siderably less than this distance, otherwise D x will be passed 

 by some parts of the field and D 2 by others. If the dif- 

 ference in thickness changes by '032 mm. in passing from 

 one edge of the block to the other, one edge will transmit D 2 

 only, and will appear brighter than the other edge which 

 transmits only D x , while the centre of the plate will transmit 

 both D 1 and D 2 in a state of circular polarization. Here 

 both wave-lengths will be passed by the analysing nicol, 

 and the intensity will be intermediate between the values at 

 the edges. If the thickness varies at a more rapid rate, 

 bright and less bright bands will cross the field, the bright 

 bands containing D 2 , the less bright D x . 



The calculated value "032 mm. was verified with a frag- 

 ment of a quartz plate 30 mm. in thickness and slightly 

 wedge-shaped, placed at my disposal by Mr. Twyman, 

 manager of the firm of Adam Hilger & Co., who also 

 loaned me the quartz echelon used in the preliminary in- 

 vestigation. With this block between crossed nicols seven 

 bands were counted (counting both dark and bright). This 

 means that we pass from D x to D 2 transmission seven times 

 in crossing the plate. Multiplying the calculated thickness 

 •032 mm. by seven gives us 0*224, while the difference in 

 thickness of the plate at the two edges, as measured with 

 the spherometer, was found to be 0*243. 



An investigation was also made with an echelon of quartz 

 placed with its elements horizontal between the nicols. It 

 was illuminated with a sodium flame, and an image of the 

 steps thrown upon the slit of a spectroscope. Each element 

 of the slit covered by an echelon step is thus illuminated by 

 light which has traversed a different thickness of quartz. 



