RICHARDS AND MARK. — THERMAL EXPANSION OF HYDROGEN. 133 



adjustment of the lower level into contact with the lower point (1) of the 

 barometer would bring the upper level also back to its proper place. In 

 practice the reading of the upper mercury level was taken within a minute 

 after the adjustment was made. Under ordinary atmospheric conditions, 

 no barometric change sufficient to invalidate the setting takes place in so 

 short a time. Thus atmospheric changes could be neglected, unless the 

 atmospheric barometer was changing with unusual rapidity, in which case 

 measurements had to be suspended. 



The temperature correction of the Lord Rayleigh barometer is based 

 on the difference between the change in length of the mercury column, 

 due to change in the density of the mercury, and the change in the length 

 of the iron rod. This amounts to 0.136 mm. per degree centigrade, which 

 is to be subtracted from the height of the mercury meniscus when the 

 temperature of the barometer is above the standard (20° C). The tem- 

 perature of the barometer was taken as the mean of that recorded by a 

 standardized thermometer at the bottom and another at the top. In 

 most cases no correction at all was applied, because the room was kept 

 at constant temperature and the air in it was thoroughly stirred by a large 

 electric fan. Nevertheless it is probable that slight fluctuations of a few 

 hundredths of a degree often occurred, and it would have been better to 

 surround the iron stem with a very accurate constant temperature bath. 

 The accuracy was increased by taking many observations over several 

 hours, the average of which would of course be more precise than a 

 single setting, because it would represent an average temperature ; 

 but it is probable that this point was the weakest one in the present 

 series of measurements, and accounts for the most of the variations in the 

 preceding table as well as for those in the final tables of results. This 

 conclusion is based upon the evidence found in the foregoing table that 

 the agreement between successive readings of pressure at a given time 

 was closer than between the averages at different times. These differ- 

 ences could hardly be due to the electrical contact arrangement, since 

 the levelling bulbs were adjusted by means of very finely threaded screws 

 and the current used to show the contact was of too low strength and too 

 small voltage to allow perceptible sparking. Moreover, the tubes were 

 so wide and the platinum points were so sharp that no difference in 

 position was noted between the making and breaking of contact. It may 

 be noted that the variation between the readings, undesirably large as it 

 is, does not exceed that observed by Chappuis in his admirable work 

 on the constant-volume hydrogen thermometer. The measurement of 

 pressure more accurately than this is an exceedingly difficult problem. 



