114 Mr. H. H. Poole on the Dielectric Constant and 



crossed nicols, and a suitable one selected, free from flaws 

 and variations of thickness. Owing to the almost perfect 

 cleavage these latter always occur abruptly, forming escarp- 

 ments across the sheet, due to faulty splitting. These 

 escarpments were at once visible between crossed nicols, 

 owing to the variation in tint on either side. The thickness 

 was at first measured with a spherometer rending to about 

 O'OOl mm., but errors due to abrasion of the surface by the 

 spherometer leg, and possibly also to slight buckling of the 

 mica sheet, made it hard to be certain of the thickness to 

 within 0'002 mm. Accordingly an optical method was 

 adopted. 



A Nernst filament was mounted a few centimetres in front 

 of the sheet, and the image of its reflexion in the mica was 

 thrown on the slit of a grating spectrometer by means of a 

 lens, the arrangement being such that the reflexion was very 

 nearly normal. Under these conditions the spectrum is 

 crossed by a number of dark bands due to interference. 

 The condition that a wave-length X should be absent is 

 2/jut = n\, where t is the thickness, //, is the index of refraction 

 for the wave-length X, and n is any integer. If X' is the 

 wave-length of the centre of the next dark band towards the 

 red, then 



or, if 8X be the width of one complete band, 



-*&' 



t = 



As the wave-lengths of a great number of dark bands can 

 be accurately measured, we can find BX with very fair 

 accuracy for any required part of the spectrum. If we 



knew /j, and — ~ for the given specimen of mica and the 

 aX 



given part of the spectrum we could find t, but, as there 



was some uncertainty as to the dispersive power of mica, the 



method was only used as a comparative one. A sheet of 



mica about 0*06 mm. thick was mounted in front of the 



spectrometer, and the value of 8X found for various parts of 



the spectrum from red to green. The mica was then cut up 



into twelve pieces the total thickness of which was measured 



