4 DENsiTir.s oi' (ixynr\ and iiYnROGEN 



another hour; the weight collected loi' the air tlis])lace(l l)y the gh>Iie ami liy the 

 brass weights coiilil now lie fonn>iitetl. 



4. AUXILIAKV WEIGHTS lOK IIYDKO.VI \ I h U I li . II I Ncs. 



The aihlitioiial weights required to sink in watti my laigest globes, amounted 

 to more than twenty kilogrammes. This amount was made up of biass weights, 

 each of which could be weighed on the balance just mentioned, while their 

 sum was nearly twenty times its greatest admissible load. Kacli one of these 

 weights was weighed while suspended in water of known temperature, so as to 

 determine, not the absolute weight of the mass of brass, but its effective weight 

 when used in the hydrostatic weighing of a globe. A scale pan with which to 

 suspend the weights from the globe was also weighed in the same manner. The 

 jai' of distilled water, in which the weights were immersed, was placed in the case 

 already mentioned, with a thermometer divided into tenths of degrees, and of 

 known index error. Weighings could be made to single milligrammes, but weigh- 

 ings on different days differed by many times that amount. 



The temperature of 18° C. was taken as a standard temperature to w Iiicli the 

 weights were reduced. If a biass w^eiyht is immereed in water of some lower teni- 

 perature, its own contraction will add to its effective weight, Avhile the contraction 

 of the water \v'\]\ lessen the effective weight. The difference of these contractions, 

 referred to the assumed standard temperature, will give the loss of weight of the 

 unit volume of brass. The specific gravity of the weights was 8.4, so that the 

 effective w^eight of the unit volume of Ijrass when immersed was 7.4 units of weight. 

 If, tlierefore, we divide the difference betw'een the expansions of water and >>( 

 brass, both counted from 18° C, by the factor 7.4, \ve shall obtain the change of 

 the effective weight of the mass of brass which, when innnersed in water at 18° C, 

 has an effective weight of one kilogramme. 



The following table gives the volumes of water and of bi'ass at different tem- 

 peratures referred to the volume at IS" taken as unity, the dirt'ei'ences, and the dif- 

 ferences after division by 7.4, expressed as grammes pei- kilogianune. Tlic corn'c- 

 tions taken with chauijed sii^n will reduce the weii^ht observed at some other 

 temiierature to the weight at 18°. 



Ten of these auxiliary weights were ( ylinders tw'enty-tive millimetres in di 

 ameter; nine were disks about fifteen centimetres in diameter. The weitrht of each 

 was determined by hydi'ostatic weighings on four different days; it was suspended 

 by a tine brass wire in water whose temperature was noted, and the observed 

 weight was reduced to the weight which would have been observed at 18° C. by 



