[smith] determining REFRACTIVE INDICES 147 



had to be altered about 2 mm. On heating the cell to 800°C. a very 

 considerable change in deflection took place, amounting in one in- 

 stance to 30". Thus the effect due to the quartz plates alone is 

 comparable with the whole effect when filled with Thallium, which 

 was from 50 to 90", depending upon the source of light. 



It was found also that at one given heating* the deflection was 

 very steady, and the probable error of the mean of 20 or more readings, 

 taken at one heating, was never more than 0.3", while readings taken 

 on different days, both hot and cold, differed sometimes as much as 

 10", and the differences between hot and cold readings correspondingly 

 more. 



The entire tube was then enclosed in an electric furnace and 

 maintained at a brilliant red heat for 11 hours, in an attempt to 

 anneal the plates. This caused no improvement whatever. 



It was then decided that, while better agreement than that given 

 above could probably be obtained, no reliance whatever could be 

 placed on values which are the difference of two readings which are 

 themselves unreliable within say 10 per cent. Consequently the 

 method was abandoned. 



Some mention should be made of the approximate agreement 

 with other experimenters obtained in the case of Mercury vapour. 

 The average temperature used in this case was about 320°C. as com- 

 pared with 800°C., while the total deflection was somewhat greater, 

 2' in some cases. Consequently the effect which has caused us to 

 abandon the method is in this case comparatively small, and moderate 

 agreement was obtained. 



Refractive Indices of Metallic Vapours 



2. Interferometer Method 



The work outlined above is now being continued by means of a 

 Jamin refractometer of a type designed by Michelson and first used 

 by H. G. Gale^. The instrument was made by the workshop staff 

 of the University of Toronto from blueprints furnished by Messrs. 

 Hilger, who made the fluorite optical parts. Fig. 1 shows a plan of 

 the instrument. The interference is obtained by means of the four 

 fluorite mirrors MM, two half-silvered and two fully-silvered on the 

 front surface. The two interfering beams, as shown in the figure, 

 pass through tubes T, which may be filled with gas or vapour, then 

 through the fluorite compensating plates CC, and are recombined in 



^Phys. Rev. 14, Jan., 1902. 



