122 Prof. D. Mendeleeff on the Variation in the 



8. Hagen (Abhandlungen d. Akad. zu Berlin, 1855, Math. 

 Abth. p. 1), adopting a hydrostatic method, made one series 

 of determinations of the expansion of water, which is remark- 

 able for its completeness. As has been already mentioned, he 

 employed his own determinations of the linear expansion of 

 glass and expressed its cubic expansion by 1 -f 0*00002754 t, for 

 the material he employed. Hagen evidently injured the 

 accuracy of his results by taking the expansion of glass as 

 constant, although he paid due attention to determining the 

 temperatures and weights with the greatest possible precision. 

 Hagen expressed his determination for t, from 0° to 100°, by a 

 formula which may be represented thus : 



S^l-T^A+BT 1 ' 6 ) 2 ; 



here T=t— 3*87. A and B are two constants, and the index 

 1*6 (or the power of T) was found by a series of attempts to 

 express the entire phenomenon of the expansion of water 

 from 0° to 100° in the simplest form. I may here mention 

 that on applying this formula to the aggregate of the existing 

 data, and by changing the values of A and B, I became con- 

 vinced of the impossibility of its satisfying with sufficient 

 accuracy the data already known respecting the expansion of 

 water between —10° and 200°. Moreover, it should be 

 observed that Hagen himself considers his figures as being 

 nearer the truth for the lower than for the higher values 

 of t, which fact is proved by a comparison with the results 

 given by formula No. 1. 



9 & 10. Jolly and Henrici (Sitzgsh. d. Akad. Miinchen, 1864, 

 i. p. 160), being desirous of verifying the existing data for 

 the expansion of water at temperatures above 30°, made a 

 series of determinations by a volumetric-thermometric method 

 (Jolly), and by weighing a known volume of water (Henrici). 

 The temperatures were determined by thermometers com- 

 pared with the air thermometer, and the coefficient of expan- 

 sion of glass by the true coefficient of expansion of mercury 

 as given by Regnault. The number of observations made 

 below 30° was limited. Differences occur in the separate 

 determinations of both observers to the extent of several ten 

 thousandths. 



11. Matthiessen (Journ. of the Chem. Soc. 1865, Pogg. 

 Ann. cxxviii. p. 512), by applying methods similar to those 

 used by Hallstrom and Hagen, obtained results which differed 

 considerably from theirs, which shows that hydrostatic 

 weighing and especially the determination of the linear ex- 

 pansion of glass do not afford that degree of accuracy which 

 is generally expected from them. Moreover, the results of 



