ON THE RF.FRACTTVE INDICES OF (USEOUS POTASSIUM, ETC. 14." 



series gave values consistent with each other. The vapour exhibits a strong absorption 

 band extending fmni the violet so far towards the red that no readings could be made 

 at, the wave-length 5183, as, when five or six bands had passed, the light was 



Oninplrtcly absorbed (Srr Table IV.). 



Tellurium. (Atomic Weight 128.) 



The specimen used was obtained from KAHLBAUM and, for the first experiment, 

 was redistilled. In this case also there is a strong absorption band in the violet, 

 which pi-evented readings being made beyond X = 5460. Tellurium was found to 

 evaporate with difficulty at a temperature hardly lower than that required for zinc, 

 and it was necessary to use very small quantities, so that the number of bands 

 observed was small, and the values obtained possibly less accurate. Two charges 

 were used. In the first day's work with the second charge the temperature was not 

 taken sufficiently high, with the result that the band readings are about G per cent, 

 too low ; but they are nevertheless recorded, as they show relative numbers for the 

 dispersion effect which are useful in corroborating the other results (see Table V.). 



Mercury. (Atomic Weight 200.) 



This was the first element dealt with by C. CUTHBERTSON three years ago. As a 

 check on the older method of working, another measurement of the refractivity of 

 mercury vapour was made by E. P. METCALFE, the silica refractometer tubes and the 

 divided furnace being now used. One charge only was employed, and observations 

 were taken with light of four different wave-lengths. The value of the refractivity 

 now obtained for D light (1866) is in good agreement with the previous deter- 

 mination (1857). The accuracy of the band readings seems to 1, for the I) line, to 

 within alxiut per cent., for the other wave-lengths to within about 1 per cent, (see 

 Table VI.).* 



REMARKS. 

 Dispersion. 



The present research was designed to obtain a single value for each element, and it 

 was only with the object of satisfying ourselves that the refractivity measured was 

 not affected to an abnormal degree by the existence of an absorption band for a wave- 

 length near to that by which the observations were made that readings were taken 

 with different wave-lengths. Each band reading being completely independent of 

 the rest, and the degree of accuracy being, apparently, not much greater than to 



[* Subsequent work by Mr. METCAIJE renders it probable that the true value for A = 6562 is 1} to 2 

 per cent, higher than that given in Table VI. January 24, 1907.] 



