342 PROCEEDINGS OF THE AMERICAN ACADEMY 



clock-room, since the relation of tlie microscope to the line bar would 

 not be affected by a contraction in the length of the bar. Several 

 pointings for coincidence with the other defining line of the meter were 

 then made, after which the end-measure bar was returned to its former 

 position. An interval of several hours was always allowed to elapse 

 before a second comparison was made. 



In the reduction of the observations, allowance was made both for 

 the small deviation from a constant temperature in the end-measure 

 bar, and also for the inferred rate of change between the time the bar 

 was brought into the comparing-room and the time the observation for 

 contact was made. Usually two or three contacts were made at inter- 

 vals of thirty seconds, in order to obtain the observed rate of change. 



Many observations of this kind were made, during the year 1881, for 

 temperatures in the comparing-room ranging between 30° and 70°. 

 The following values of the absolute coefficients were obtained. On 

 account of the obvious imperfection of the method, it does not seem 

 worth while to print the details of the work. It will be seen, however, 

 that these values do not differ widely from those found from the method 

 which will be presently described. 



Absolute Coefficients. 



For T', lG.58/<. For i?„ 10.03 jw. For ^,, 17.55 ^u. 



The experience acquired in the observations thus far made was 

 wholly in favor of the plan of deriving the coefficient of expansion of 

 a line-bar for air contact, by comparison with an end-bar at a constant 

 temperature. In no other way could an accurate comparison be so 

 quickly made. But it is essential to the success of this method that 

 the end-measure bar shall remain at a constant temperature during the 

 entire series of comparisons with the line-standard at wide ranges of 

 temperature. Only a few ooservations were needed to show that the 

 end bar could be kept at an invariable temperature by keeping it con- 

 stantly submerged in melting ice. It was found necessary to take only 

 two precautions: first, that the bar should have a covering of ice about 

 two inches in depth ; and second, that the ice-trough should not remain 

 in a high temperature for a longer time than about ten minutes. Ex- 

 periment showed that, if there was too little ice in the trough in which 

 the bar was submerged, the expansion beyond the normal length of 32° 

 might amount to as much as 8 /x. It was also found that, if the ice- 

 trough remained in the comparing-room at a high temperature for a 

 time exceeding ten or fifteen minutes, an expansion amounting to 

 about 5 IX was liable to occur, notwithstanding a vigorous stirring of 

 the ice. 



