324 MESSES. C. CUTHBEKTSON AND E. B. R. PEIDEAUX 



These results are not concordant, and the causes of the discrepancies were not 

 completely disentangled. It will be sufficient, therefore, to indicate, briefly, their 

 probable nature, without attempting detailed criticism which the figures will not 

 bear. The first four experiments are rendered nugatory by the absence of any means 

 for destroying the ozone produced with the oxygen, which, as will be shown later, 

 invariably accompanied the fluorine (see Column 9). 



No correction can be introduced into the figures for this source of error ; for the 

 proportion of ozone to oxygen, produced under the conditions of the experiment, is 

 not known, and the quantity of oxygen present is itself doubtful, since the nature of 

 the reaction between ozone and mercury is not beyond dispute.* 



Any correction for ozone would reduce the value found for the index of fluorine. 



In the fifth and sixth experiments the measurements of the volumes of residual 

 gases proved insufficiently accurate, and were complicated by the presence of ozone 

 produced by the action of the fluorine on the solution of soda. 



In the last three experiments these sources of error had been eliminated, and we 

 are forced to suppose that the method of absorbing the fluorine over mercury is open 

 to some grave objection, possibly the formation of an oxyfluoride of mercury. It is 

 certain that some source of error is to be sought in the process of absorption over 

 mercury, since, in these experiments, the calculated values for the index of fluorine 

 given in Column 10 are more discordant than those given in Column 4 for the 

 observed refractivity of the mixture of gases. 



But, though this series did not give values sufficiently concordant to warrant the 

 belief that the true index of fluorine was being measured, some important inferences 

 could be drawn from the results. The number of values which ranged below the 

 lowest figure obtained in Paris confirmed the opinion that, on that occasion, some 

 other gas or gases were present. On the other hand, the absence of any very low 

 value, in spite of the variety of methods employed, indicated that the refractivity of 

 fluorine was to be sought in the neighbourhood of the figure 200, and was by no 

 means so low as students of the refraction equivalents have surmised. 



But the most interesting point observed was the presence, in the residuals, of a 

 larger proportion of oxygen than could be accounted for by the amount of air present. 

 This was observed to be the case in all the experiments shown in Tables II. and III., 

 as well as in others specially designed to test the point ;t and it was ultimately 

 proved, beyond reasonable doubt, that the oxygen was produced by the intermittent 

 electrolysis of traces of water in the electrolytic tube, and not by subsequent reactions. 

 Our experience was that the proportions of oxygen and fluorine liberated were not 

 sensibly altered by prolonging the experiment for two or three hours. 



Having established this fact, we were enabled to make dispositions for the series of 



* ANDREWS and TAIT, 'Phil. Trans.,' 1860, p. 114; SHENSTONE and CUNDALL, ' C. J.,' 51, p. 623; 

 E. C. C. BALY, ' B.A. Reports,' 1897, p. 613. 



t It is hoped that the details of these experiments may be published on another occasion. 



