OP ARTS AND SCIENCES. 343 



For the third series of observations the following preparations were 

 made. A steel bar was prepared G mm. shorter than one meter. 

 Steel plugs were inserted at each end, each having a projection of 

 3mm. The ends of the bar were then made as hard as possible, and 

 the plugs were ground off till the defining surfaces were one meter 

 apart. Careful and repeated observations showed that these end sur- 

 faces were truly parallel. The bar was then mounted in a trough of 

 galvanized iron having a depth of 5 inches and a width of 5^ inches. 

 The trough was originally prepared for a yard, and, in the first series 

 of observations which follow, the bar, which is placed near the bot- 

 tom of the ti'ough, extends 4.5 cm. beyond each end. At the con- 

 clusion of the first series the trough was lengthened, and in the second 

 series the projection at each end was only 3 mm. 



In planning the execution of this series of observations, there were 

 certain difficulties about the use of the Universal Comparator for this 

 purpose which could not be easily remedied; and it was found neces- 

 sary to make use of Comparator No. 1. Since it had been found that 

 the Tresca bar T was extremely sensitive to small changes of temper- 

 ature, and since it was possible to maintain a nearly constant temper- 

 ature in the comparing-room of the Observatory for short intervals of 

 time, it was arranged to determine the absolute coefficient of T by 

 comparison with the bar S in melting ice, and then to determine the 

 relative coefficients between T and the remaining standards from com- 

 parisons with the Universal Comparator. 



Partly as an experiment, however, simultaneous observations for the 

 absolute coefficients of bars R.^ and C. S. were made. A study of the 

 tables which follow will show that the probable value of the results 

 obtained is nearly proportional to the mass of the bars. 



The various steps of the observations are as follows. 



The line bar having been placed in position and adjustment upon 

 the comparator during the afternoon, the comparing-room was closed 

 till the next morning. During this time the temperature in the room 

 generally became very steady. For low temperature observations, 

 both windows were left open during the night. About sunrise the 

 next morning the bar S, after having been submerged in melting ice 

 for at least twenty minutes, was removed from the room in the cellar 

 in which the pier of the Russian transit instrument is situated, and 

 placed between the stops of the comparator. The comparison with 

 the line-meters was then made in the manner already described. The 

 ice-trough was then immediately removed to the cellar, and allowed 

 to remain for a few minutes, Ay hen the operation just described was 

 repeated, provided always that a steady temperature was maintained 



