( 545 ) 



Correction for the temporafuro of' tho mernirv : 



If the pressure amoiiiifs to say 0.5 atin., measured as tlie (liflerenee of 



two mercnrv oolumiis of 7fi and 38 em. respeelix ely, the (empci-aliii-e of 



the k)iif? eolumn must be known precisely to within .5, if an acc»ii-acy 



1 . 

 <^t ^^^^ is required. We must assume llial this is possiltle it" the 

 öOOO ' 



mercnrv cohimns top^ether with tlie thermometers are packed in wool. 



The difference between the indications of the thermometer at the 



lower end and at the npper end of tlie barometer Bar. did not as 



a rule amount to more than ()°.5 C, hence we may accejit that the 



uncertainty caused by this in the knowledge of the temperafuiv did 



not exceed by much ()\25 C. For the diiference of the temperature 



of the mercury in J/ and E a correction is applied. 



The capillary depressions are largest in the Aolumenomcfer, and 

 of the order of say 0.07 mm. The uncertainty here is surely much 

 smaller than the errors resulting from the windows not being vertical. 



Corrections for the difference in level between the meniscus in 

 Bill' below, and in M, and also those for the reduction of the pres- 

 sure immediately above the mercury in E to the mean pressure of 

 the gas above it, may be applied with sufficient accnracy. 



Measnrement of the temperature : 



During the time of one measnrement (25 minntes) the tem])(Matui(^ 

 by means of a thermostat of Schalkwijk \i can certainly be kept 

 constant at 0°.03 C, while if the water at constant temperature 

 flows rapidly enough throngh A', the temperatures at the lower and 

 at the npper end did not differ more. The temperature was read on 

 a thermometer Th connected with the stirrer Er, which thei-efore 

 conld be placed in different positions.. The thermometer was 

 graduated in 0.05, so that 0^01 could be easily read; it has 

 been repeatedly compared with a standardthermometer tested at 

 the Keichsanstalt. As therefore the eri(»i' in the measnrement of 

 temperatnre does not amount to 0\03 C., the accnracy is certaiidy 



1 1 



0.03 X = 



^300 10000 



A j)Ortion of tli(^ xolunie ^in the tubes EIkj^ and Ehij^. which 



have an inner bore of 1.2 mm.) is n(»t enclosed in the jacket. This 



\-olume is a little less than 1 cc. Foi" its tempei'atnre which is pnt 



e({ual to the temperature of M a correction is applied. If in a 



measni-emejit in the tirsi bulb A'A^ w c assume that in the temj)e]"atnre 



of that [)ortion there is an eri-or of 5', the en-oi' in the mean 



1) Gomp. Comm. N". 70, 'YW-^v Proc. IV May -lo, 1901, p. 129. 



