of Water and other Liquids. 405 



the results only imperfectly. In examining determinations in 

 the manner here described many different scales must be 

 employed ; and the scales must be altered to suit the accuracy 

 of the determinations of different experimenters, and of the 

 same experimenter when working at different temperatures. 



The illustrations which I give of the results are confined 

 almost exclusively to the differentials. 



The following details may be noted as to the various 

 results obtained. 



Pierre's values {Ann. Chini. Phys. xv. p. 325) exhibit great 

 regularity up to 20° ; they are very numerous, and extend 

 down to — 13°. The direct differential up to 20°, and that from 

 the smoothed experimental curve from 20° to 95° (Table HI.), 

 is shown in fig. 3, Plate VII. The direct differential for these 

 higher temperatures is given on a less open scale in fig. 12, 

 Plate V11I. The irregularity which suggests two changes at 

 0° and — 2°*5 respectively is so peculiar that it must be 

 regarded with considerable distrust : but some one change, 

 however, somewhere about this temperature seems certain, 

 since the results from —13° to 10° cannot be represented by 

 any one line or curve. The change at about 18° is also 

 doubtful, since the curve from 50° to 18° could be extended 

 down to 10°(but not lower) with but little increase in the 

 apparent experimental error. 



Despretz {Ann. Chim. Phys. lxx. p. 1) made four series of 

 determinations between —10° and 17°. That one which gives 

 the most regular results indicates a possible change at about 

 10° (see fig. 4) ; and the mean values which he deduced from 

 the four series gives a similar indication, though at a some- 

 what higher temperature (fig. 5). There appears to be some 

 irregularity below 0°, but the nature of the figure here is 

 uncertain. One series of determinations extending to 100° 

 which he made by another method gave no certain indications 

 of any change at higher temperatures, and are not sufficiently 

 numerous to show any changes which might exist at lower 

 temperatures. They have not been reproduced here. 



The differential obtained from the smoothed curve repre- 

 senting Kopp's results (Pogg. Ann. lxxii. p. 45) is shown in 

 fig. 6. Tables III. and IV. contain the numerical values. 

 They suggest a change at about 60° and 10°. 



