34^ PROCEEDINGS OF THE AMERICAN ACADEMY 



meanwhile in the comparing-room. The high temperatures in the 

 comparing-room were produced partly by heat from the furnace, but a 

 steady supply of heat was furnished from a gas stove, which was 

 placed directly beneath the bed-plate of the comparator. 



After the first series was completed, the trough was lengthened, and 

 some changes were made in the disposition of the bar S, by which its 

 apparent length was slightly changed. During the second series, the 

 position of the stops was not disturbed, and the adopted order of obser- 

 vations was strictly followed in every case. This series, therefore, may 

 be supposed to have greater weigiit than the tirst one. An attempt was 

 made tu make an equal number of observations for slightly increasing 

 and slightly decreasing temperatures, in order to eliminate to a certain 

 extent the difference between the temperature of the thermometer 

 and the temperature of the bar ; for it must always be understood that 

 the thermometer simply registers its own temperature and not neces- 

 sarily the temperature of the bar upon w^hich it is placed. This 

 arrangement was, however, only carried out to a certain extent. 



There is one point in connection with these observations to which 

 particular attention is called. If the contact surlixces of the bar aS 

 become worn by continual use, it is evident that a constant error will 

 be introduced, which will need to be investigated before the observa- 

 tions for the coefHcients of expansion can be reduced. But it will 

 be seen from the observations which follow, that there is no con- 

 clusive evidence of such a change during the entire series. This 

 result is certainly contrary to my expectations, especially as the force 

 required to make the contacts secure is considerable. The trough 

 containing the bar S, when filled with melting ice, weighs about twenty- 

 five pounds, and this entire mass is required to be moved by the pres- 

 sure of the stop attached to the microscope carriage. I conclude, 

 therefore, that with ordinary care there is no danger of wear with steel 

 surfaces which have received a temper as high as possible. 



But in order te provide against this supposed source of error, the 

 high and low temperature observations were made in close proximity. 



After the observations for the absolute coefHcients of the bars T, 

 C. S., and E., were completed, these bars, with B^, were removed to 

 the comparing-room in Harvard Hall. The series of observations for 

 the relative coefficients between these bars extends from March 22 to 

 June 28, 1883, but the series from which the equations of condition 

 were formed terminated on June 7. A few subsequent observations 

 were made as a check upon the results obtained from the equations. 



The observations and reductions, both for the absolute and the 

 relative coefficients, are given on the following pages. 



