REFRACTIVE INDICES OF THE ELEMENTS. 333 



assuming a mean value of '000008 for a and a mean range of temperature of 

 250 C. - 1 5 C. = 235 C. for all the experiments. Hence 



,-*,) = T0038, 



and this correction is made on the mean value obtained for the refractivities of 

 Hg, P and S. 



Degree of Accuracy. 



The degree of accuracy aimed at was no greater than one per cent. The 

 inquiry was undertaken with the object of discovering whether any simple numerical 

 relation subsisted between the refractivities of members of the various chemical 

 families, similar to that observed in the cases of the inert gases and the 

 halogens. If such a relation did subsist, it was not likely that experiments made 

 with light of a wave-length arbitrarily selected* would exhibit a higher degree of 

 concordance, since the dispersion, within the limits of the visible spectrum, is generally 

 at least as great as 1 per cent, of the refraction. A higher degree of accuracy could 

 only have been attained by the sacrifice of more time than was advisable, having 

 regard to the time spent on each experiment, which generally took about a week to 

 prepare. 



I much regret that the number of experiments, especially in the case of sulphur, 

 was not larger ; but it seemed at the time that more would be gained than lost by 

 attacking a new substance, as soon as it became clear that concordant results were 

 being attained. 



With these general observations I pass to a record of the results obtained. 



Mercury. (Atomic Weight = 200'3.) 



Mercury was the first substance with which experiments were made. No special 

 precautions were taken to obtain the substance absolutely pure, since the percentage 

 of impurity in good, clean mercury must be very small. 



As has been stated above, the plan first adopted was to put excess of mercury in 

 the refractometer tube, and to calculate the density of the vapour present from an 

 observation of the temperature of the tube at its coldest part, by means of the table 

 connecting the temperature and vapour-pressure of mercury given by RAMSAY and 

 YOUNG. Though the method ultimately .proved a bad one, owing to the difficulty 

 of measuring the temperature accurately, I give below the details of all the 

 experiments performed by this method, since they are sufficiently concordant to 

 corroborate roughly the results subsequently obtained by the balance method, and to 

 show that the true value of the index of mercury is of a different order from that 

 obtained by LE Roux. 



* The figures arc for sodium light throughout, and the wave-length is taken at 5893. 



