42 Messrs. Playfair and Joule on 



at its maximum density were taken as the unit*. In that 

 case, calculations would often be greatly facilitated ; and if 

 the temperature of the water sliould happen to be a degree 

 under or over the maximum point in any experiment, the ex- 

 treme slowness of the expansion would avert the possibility 

 of a grave eri'or. 



Intending to give still greater accuracy to our future ex- 

 periments, we deem the present a favoui'able opportunity for 

 fixing upon a good unit of comparison. We propose there- 

 fore to occupy the present series with a brief discussion as to 

 the point of temperature at which water arrives at its state of 

 greatest density. 



Although a variety of methods have been employed in the 

 investigation of this subject, they may be classified under two 

 general heads. The first of these embraces all the methods 

 involving the necessity of an accurate acquaintance with the 

 rate of the expansion of some solid body by heat. The in- 

 quiry, when made in this way, is one of great difficulty, and 

 has occupied the attention of many of our most accurate ex- 

 ])erimenters. The Florentine Academicians, Croune, Deluc, 

 Dalton, and others, compared the indications of a thermometer 

 filled with water with one filled with mercury ; Lefevre 

 Gineau and Hallstrom weighed a solid body in water at dif- 

 ferent temperatures ; and Blagden and Gilpin measured the 

 variation of volume by the variation of the weight of water 

 contained in the same vessel at different temperatures. 



The other general method does not require a previous ac- 

 quaintance with the expansion of a solid by heat : it consists 

 virtually in weighing water in water — the heavier water de- 

 scending, while the lighter ascends to replace it. This prin- 

 ciple was introduced by Dr. Hope, who applied it in the 

 following elegant manner : — He filled with water at different 

 temperatures tall glass jars having thermometers at top, 

 middle and bottom. In this way he observed that when 

 water was cooled down to 40° at the surface, it sank to the 

 bottom ; and when cooled below 40° at the bottom, it rose 

 again to the surface. Tralles, Count Rmuford, Ekstrand and 

 Despretz have repeated Hope's experiment with a similar 

 apparatus. 



Believing that the second general method is susceptible of 

 a far greater degree of accuracy than the first, we at once de- 

 termined to employ it in our own experiments. The particular 



* Since this paper w.ns communicated to the Society we have seen the 

 Annuairc du Biirrati des Longiliides for 18)5, in which, under the high au- 

 thority ofArago, \v;>ter at its point of maximum density is assumed as tlie 

 iiiiit for comparison. 



