ON THE REFRACTIVE INDEX OF GASEOUS FLUORINE. 329 



These dangers were met by reducing the quantity of glass used as tar as 

 possible, bringing the lead filings into immediate contact with the exits oi 

 the refractometer tube, drying the whole with scrupulous care, and, finally, by 

 testing the residuals for silicon fluoride. This test was carried out in three 

 cases and only in one of these was a minute quantity of SiF 4 found (0*3 cub. 

 centim.), which did not affect the index by more than one or two units. 



(3) The presence of ozone in the refractometer tube, especially if afterwards 



disintegrated, would introduce a serious error. The device employed to 

 obviate this is described above. 



(4) Owing to the form of the apparatus it was not possible to isolate the tube 



residuals and measure their index directly, and their volume was so small, in 

 comparison with that of the total residuals, that it would have been unsafe to 

 calculate the refractivity of the former from that of the latter, even after the 

 most careful measurements. It was necessary, therefore, to calculate the 

 value of the refractivity of the tube residuals from their composition, 

 ascertained by the analysis given above. But the calculation was confirmed 

 by comparison with the figures given in the preceding series of experiments 

 when the residuals were collected undiluted with air. Table II. shows that, 

 in that series, the index lay, in nearly every case, between those of oxygen 

 and nitrogen. 



It will be observed that the result given here rests on the assumption that the 

 density of gaseous fluorine is 1'319 (air = 1) ; i.e.. that the molecule is diatomic under 

 normal conditions. M. MOISSAN has twice measured this important constant. On the 

 first occasion, in 1889, he obtained four concordant values,* 1'2G4, 1'2G2, 1'2G5, 1'270, 

 the mean of which was 1'265, and he thence inferred that the gas contained a small 

 proportion of molecules of Fj. f 



In 1904 a second series of experiments, in glass bulbs, gave values of 1/298, 1'319, 

 I '313, 1'312, of which the mean is 1'310, a figure very nearly identical with the 

 value assumed. We have accepted this later determination, which was made with 

 the precautions dictated by many additional years of experience, and is supported by 

 a priori probability. 



In a recent paper by Mr. CUTHBERTSON,^ it was shown that, in four groups of 

 elements, the refractivities of the different members of the same group are related in 

 the ratios of small integers, and it was pointed out that, if this coincidence were not 

 due to chance, the refractivity of fluorine should bear to that of chlorine the ratio of 

 one to four, as those of nitrogen, oxygen, and neon do to phosphorus, sulphur, and 

 argon respectively. 



* 'Le Fluor,' p. 87; 'Ann. de Chimie et de Physique,' vol. 25, 1892, p. 131; 'C. R., ! vol. 109, 1889, 

 p. 863. 



t 'C. R.,' vol. 138, p. 729 (1904). 



t 'Phil. Trans./ A, vol. 204, p. 323, 1905, 



VOL. CCV. A. 2 U 



