110 Prof. D. Mendeleeff on the Variation in the 



much greater than that adopted by us *, and since this reduc- 

 tion affected the third decimal in the volumes, it may be 

 presumed that the above-determined densities of water contain 

 an error at least in the fourth, or perhaps even in the third 

 decimal place. 



Having made this reservation, it will be possible to com- 

 pare the densities found by experiment with those calculated 

 from formula No. 1, adopting the above-mentioned values for 

 the constants A, B, and C. Thus : — 



120°. 140°. 160°. 180°. 200°. 

 From Hirn's experiments, S t = 09429 0-9257 09071 0-8*66 0-8627 

 By calculation, formula No. 1= 0"9433 0-9262 0-9073 0-8864 0-8635 



Difference -00004 -0-0005 -00002 +0-0002 -0-0008 



The difference, therefore, between the results obtained by 

 experiment and by calculation, for temperatures ranging from 

 100° to 200°, does not exceed the possible error in the deter- 

 minations made by Hirn, which are distinguished by the 

 highest degree of accuracy yet attained in this province. 



With respect to the influence of pressure on the density 

 of water, we must, inter alia, make the following remarks : — 



(1) A most important addition to the study of the properties 

 of water w T ill be introduced by determining with the greatest 

 possible accuracy its compressibility between — 10° and + 200°. 



(2) In accurate determinations of the density of water (and 

 of other liquids) the pressure must be determined and a cor- 

 rection introduced for it. 



(3) The normal density of liquids (also of gases) must be 

 reckoned at the normal pressure of 760 mm. of mercurv (at 

 lat. = 45°). 



(4) For the theory of the subject it would be highly important 

 to make a series of determinations of the density of water from 

 0° to 100° and upwards at some fixed and considerable pres- 

 sure, in order to judge of the manner in which S and V are 

 dependent on t and y (pressure). At present, wdiilst we are 

 ignorant of the true nature of this dependence, we may take 



if 0(f) be found for j? = 1 atmosphere. From the theory of 



* Pagliani showed that with the majority of investigated liquids //, 

 increases very rapidlv with the temperature ; for instance, with normal 

 propyl alcohol, at 0°, \i =0-0000086, and at 100° /z=0-0000158. Judging 

 from the variation of the properties of water, there is reason for thinking 

 that at 200° its coefficient of compressibility will be, for example, twice as 

 great as at 100°. 



