PROCEEDINGS: PHILOSOPHICAL SOCIETY 643 



driving current is broken and the timing forks, therefore, vibrate 

 freely. A similar set-up consisting of 50-cycle forks and a 50-cycle 

 master fork is used to rule slow-moving films into hundredths of a 

 second. 



An examination of the timing line records indicates that the lines are 

 not equidistant, every alternate interval being slightly wider than 

 the intervening ones. This is due to the fact that the vanes are not 

 properly adjusted. As it was found almost impossible to adjust them 

 exactly these errors are eliminated by interpolating between every 

 other line. It is only necessary^ to do this when extreme accuracy is 

 desired. 



Further examination of the records indicates that the time exposure 

 is extremely short, certainly not greater than 0.000 01 second. Even 

 when the film is moving at a speed of 1500 cm. per second the lines are 

 quite sharp. 



In order to test the accuracy which could be obtained in the de- 

 termination of time intervals, one film was exposed on each of two 

 oscillographs simultaneously. One recording galvanometer on each 

 oscillograph was put in series with a current interrupted by a tuning 

 fork. By examining the two films, it was possible to identify the 

 individual interruptions. Measurements were then made on both 

 films to determine the time intervals between certain vibration rates 

 of the two circuits. The time intervals measured on the two films 

 varied in every case by less than 0.000 01 second. 



It therefore seems that this method makes it possible to measure 

 intervals of thousandths of a second to better than one per cent, while 

 intervals from o.i to 0.2 second can be measured with an accuracy 

 of about 0.0 1 per cent. 



The paper was illustrated by lantern slides. It was discussed by 

 Messrs. Silsbee and Beal. 



The second paper was by Mr. A. F. Beal, on Comparison of invar 

 with steel as shown by the rates of high grade watches. 



For watches such as the Bureau of Standards is testing for use as 

 chronometers by the United States Shipping Board, where they are 

 mounted so as to remain horizontal at all times, the compensation for 

 changes of temperature is the only important known cause of variation 

 that is not entirely within the control of the adjuster. Uniformity of 

 rate for different temperatures depends upon the physical constants of 

 the balance wheel and hairspring, and perfect compensation has been 

 unobtainable with the customary steel hairspring and brass-steel 

 balance wheel. If, in the balance wheel, invar be substituted for steel, 

 a combination can be obtained which will show rates that are practically 

 constant over large ranges of temperature. 



During 191 7 and 19 18, the Bureau of Standards tested the watches 

 supplied on two purchase orders placed by the United States Shipping 

 Board with two American manufacturers of high grade watches. 

 Those supplied on one of these orders (for 200 watches) had brass- 



