Joly — On a Hydrostatic Balance. 353 



If it "be desired, however, to render the arrangement nearly 

 permanent, so that the operation of filling need but very seldom be 

 repeated, the effects of temperature in expelling water or drawing 

 in air must be met in some way. In the balance of fig. 2 this is 

 done by providing the expansion reservoir shown surrounding the 

 tubulure, and which communicates with the interior of the sphere 

 by the narrow tube nearly reaching to the bottom of the reservoir, 

 as shown in the figure. The large surface of water exposed in this 

 reservoir bears to stand at a level above or below, by a couple of 

 millimetres, the surface level of the water in the tubulure, as in the 

 well-known experiment on capillarity in communicating tubes of 

 very unequal bore. Hence, with rise of temperature the reservoir 

 receives the expelled water ; with fall of temperature it parts with 

 some of its contents, and no water is lost. The annular reservoir 

 communicates with the air by a very small perforation, and the 

 loss by evaporation is very small. 



To enable the balance to be readily filled, the ring by which it 

 is suspended is arranged to screw out of a little tubulure communi- 

 cating with the interior. The balance is filled in a few seconds by 

 screwing out this ring, and immersing the sphere in a vessel of 

 water; when no more bubbles ascend through the tubulure, the 

 ring is screwed home, while the tubulure is still beneath the surface 

 of the water. On withdrawal a little water runs out at the lower 

 tubulure, till the head in the reservoir has been syphoned down to 

 a position of equilibrium with the surface tension at the tubulure ; 

 the head is now still further reduced by applying a little bibulous 

 paper to the tubulure, in order to provide for a subsequent rise, in 

 temperature. 



words, there would be no expulsion of water or entry of air with atmospheric varia- 

 tions of temperature. Thus for a spherical float in a spherical chamber, and assuming 

 any desirable radius, r, for the float, let x be required radius of outer vessel ; also let 

 ff, b, and whe the co-efficients of cubical expansion of glass, brass, and water, equating 

 the increments of volume for a rise of one degree — 



x 3 b — r 3 g + (x 3 — r 3 ) iv ; 

 taking # = 0-000025; b = 0-000054 ; w = 0-00014; 



x 3 = 1-337 x r\ 



But this affords unfortunately rather closely approximating values for x and r, as, for 

 example, if r = 2-9 cms. (vol. = 100 ccs.), then x = 3-2 cms. Nor can I find materials 

 affording much better results. 



SCIEN. PROC. R.D.S. VOL. V. PT. V. 2B 



