( 305 ) 



Ohm. If the ft-alvanoilictci' is carefully packed in codon Wool It 

 appeared (hat <1ui-infi' the lime required for one measuremeul the 

 temperature was sul'ticienllv conslani and hence the total variation 

 iu \l\ and 11.; did iu»l exceed 0.1 ()liiii. If Ihcii we take foi- the 

 smallest of the «, and «, JOOO Ohms, (he inaccuracy due (o (his 

 uncertainty need not therefore he larjier than Vi.mmmi- '' '^ '•^'^ely, 

 moreover, that the tirst adjustment af(er the measm-cment where «^ 

 and «.J were added to ]\\ and IT.^ will be re[)eated in most cases 

 so that w'e then can more or less judge of those variations in ITi 

 and Tr,. 



T Ct 



In the relation — = (1 + /^) — , ^ ii^ still to he determined. To this 



'>'i't «1 



end \)\ means of commutator (\ (PI. I, tig. 3), the conducts of 



the galvanometer are interchanged, so that the turns which fii-st 



were parallel to r,,- are iu)w arranged in parallel to r^,, and i-exersely. 



A new determination vields - = (1 — ,>) - (>' is considered small). 



r,n it, 



■'■'• /'"i "'A 

 If [1 is uidvnown we tiiul — = M 1 — r • 



Besides we can determine [ï, and lind for it 



^i = ~^- ^ 



a, «', 



If a rapid determination of the ratio is desirable, as it was 

 naturally always the case with the measurements at low tempera- 

 tures, it is better to determine [3 beforehand. We must then have 

 convinced ourselves by preliminary experiments that [3 remains 

 sufliciently constant. This was the case for observations matle at Jiot 

 too long intervals. Thus 11 determinations from June 14 to July 12 

 yielded as smallest value 0.0007G4, as mean value 0.000780 and as 

 largest value 0.00083 (?). 



Thus supposing the value of {3 to be known we can deri\e the 



ratio — ^ from a single determination of (c, and «.^ ; «i and «.^ were 



read on the resistance boxes /i', and 11 g. 



§ 8. The determination of the temperature. The determination of 

 the temperature, now that ^ve do not measure the resistances separa- 

 tely but determine the ratio at once, need not be made with the 

 accuracy whiclj was required for the test of the i)latinum thermo- 

 meter. It was obtained by measuring the resistance of the platinum 



