( 374 ) 



Prom the indication of the float the value of u^ is found, iij and 

 uj' are chosen such that the circumstances are as closely as possible 

 equal to those for which the distribution of temperature in the 

 crjostat is determined. ■ We get now the table III, in which H^ is 

 the zero point pressure. 



From these data with formula (3), where the value 0,0036627 

 of Comm. N°. 60 ^) was assumed for a, follows for the approximate 

 value of the temperature : 



«= — 252°.964 

 and after application of the correction for the quadratic term : 

 t = — 252°.964 + 0°.035 = — 252°.93. 



§ 7. Accuracy of the determinatmis of the temjjerature. 



In order to arrive at an opinion about the error of the observations 

 with the hydrogen thermometer, we determine the differences of the 

 hydrogen temperatures found in different observations in which the 

 resistance was adjusted to the same value, reduction having been 

 applied for small differences left. 



The mean error of a single determination derived from the diffe- 

 rences of the readings of the thermometer, which succeed each other 

 immediately, is on an average ± 0^.0074, from which we derive 

 for the mean error of a temperature ± 0°.0043, assuming that on an 

 average 3 observations have served to determine a temperature. As 

 a rule no greater deviations than 0°.02 were found between the 

 separate readings of one determination. Only once, on Oct. 27^^^ '05 

 (cf. Comm. N". 95^ Tab. I) a difference of 0°.04 occurred. Even at 

 the lowest temperatures only slight deviations occur. Thus on May 

 5th '06 two of the observations in the neighbourhood of the boiling 

 point of hydrogen (cf. Comm. 95'^ Tab. VI„ observation N". 30, and 

 Comm. N". 95^ Tab. I) yielded : 



3"20' — 252°.926 

 3"58' — 252°.929 

 the two others with another resistance : 



2u35' _ 252°.875 

 3u 7' — 252°.866'') 



Determinations of one and the same temperature on different days 



1) From the values of « found by Ghappuis at different pressures and from 

 Berthelot's calculations follows by extrapolation from Ghappuis' value for 

 p = 1000 m M. a = 0.00366262 for _p = 1090 m.M., from Travers' value of a, for 

 700 m.M. with the same data ;e = 0.00366288 for p=^1090 m.M. 



2) At both these temperatures the indications of the resistance thermometer were 

 not made use of, but only the pressure in the cryostat was kept constant. That in 

 spite of this the readings of the thermometer differ so little is owing to the great 

 purity of the liquid hydrogen in the bath. 



