114 Mr. Spencer U. Pickering on the 



of some change of curvature at about 30 per cent, similar 

 to that indicated by Topsoe's values, though the present 

 determinations are not sufficiently numerous to establish 

 it satisfactorily*. (I have not attempted to indicate the 

 changes of curvature in the woodcut, owing to the restricted 

 scale of plotting necessarily adopted.) 



Fig. 1. — Densities of HBr and HC1 solutions. 



Per cent. acid. 



The more numerous results with stronger solutions, 44 to 

 65 per cent., suggested the existence of two breaks at 52 and 

 60 per cent, respectively, neither of which, however, were at 

 all well marked. To examine them more fully I first deter- 

 mined the mean error of the results by my graphic method 

 (Ber. xxiv. p. 3332, and xxv. p. 1100) : this gave 0*000735, 

 a value greater than that deduced from the titrations, as was 

 to be expected, owing to two different samples of acid (con- 

 taining rather different amounts of free bromine) having been 

 used in making the two series of determinations. When 

 the drawing representing the existence of the two above- 

 mentioned breaks was examined in the manner already 

 described in these pages (Phil. Mag. 1892, xxxiii. p. 438), it 

 was found to represent the mean error of the points, and also 

 their "total" error, to be 0*9 time that of the known experi- 

 mental error, and it must, therefore, be regarded as agreeing 

 very well with this error ; whereas a drawing as a single 



* Both Berthelot's and Thomsen's values (the latter especially) for the 

 heat of dissolution show strong indications of some break between 20 and 

 40 per cent., but they are not numerous enough to give its exact position. 



