( 337 ) 



The tables I and II also give new data about the accu racy of the 

 readings of the resistance and of the gas thermometer 1 ). According 

 to table I the error of the resistance remains below 0. c 01. As to 

 the readings of the gas thermometer, the firsl two data of table II 

 inter se yield a difference smaller than 0.°02. With the difference 

 + 0.031 52 at -217° correspond +0,028 and +0.01(5 of table II 

 in Comm. N°. 95° and tabic I in Comm. N°. 10K This also agrees 

 with the accuracy of O. c 02 derived in § 7 of Comm. N°. 95 e . This 

 seems not to be the case for the value of 18 Pebr. '08 which 

 deviates rather much from the formula. It will depend on later 

 determinations what part of this deviation must be ascribed to the 

 formula AI. 



§ 4. The vapour pressure determinations in tli>- neighbourhood of 



tin' boiling point of <>,ri/<i<'n. 



We have used the cryostat described in Comm. N°. 94'" (PI. V 

 Sept. 1906). The temperature was determined and regulated with 

 the resistance Pt'j placed in the bath. The deviations of the galvano- 

 meter were so small that no correction was required for it. In the 

 apparatus A the mercury was raised successively in the lower and 

 the upper end of the manometer m 1} in B first the gas of one reser- 

 voir was condensed, afterwards of the two reservoirs. This is indicated 

 in the following table with "little" and "much". Table III contains 

 a determination at a small excess of pressure, another at a pressure 

 of a little below 760 m.m. and a determination for control also at 

 an excess of pressure. The pressure is reduced to O C. The last 

 column contains the deviations from the mean for each series. 



The resistances of the thermometer read on the Wheatstonk bridge 

 were for the three series respectively : 



34.433 , 34.098 and 34.433 Q. 



If we compare the results obtained with the condensation of a 

 little quantity of gas with those with the condensation of a large 

 quantity we cannot find any systematical deviations, which speaks 

 for the purity of the gas used. In case impurities should occur their 

 influence on the vapour pressure derived with the -mailer conden- 

 sation is sure to be less than the influence on the difference just 

 mentioned. It may therefore be neglected. 



] ) Comp. Comms. No. 95c, 95« and I'll". 



