460 WOLFF, SHOEMAKER, BRIGGS : RESISTANCE STANDARDS 



sation of moisture being eliminated by a current of cooled and 

 dried air directed at the end of the tube and at the stroking off 

 plate. Weighings were made in a special balance room, on a 

 StiJckrath balance sensitive to a hundredth of a milligram. Six 

 fillings of each tube were made, the mean of the average devia- 

 tions of the individual fillings from their respective means being 

 but ± 4 parts per million. 



The problem of determining Lo, the length of the axis of a 

 tube at 0°C., was reduced, through the use of suitably ruled end 

 pieces of platinum-iridium, from one of comparing end standards 

 to that of comparing line standards. Comparisons were made 

 directly with corresponding known intervals on a nickel-steel 

 meter. The length constant added, due to the end pieces, was 

 determined by abutting the end clips and measuring the interval 

 between the lines on them by comparison with a subdivided 

 decimeter standard. The probable error in the lengths as de- 

 termined, all things couvsidered, did not exceed five ten-thou- 

 sandths of a millimeter. 



Electrical comparisons of the mercury units were made by the 

 Thomson bridge method, the mercury units and five sealed 

 manganin standards being substituted in turn in the same bridge 

 arm. The ratio coils and the manganin standards were contained 

 in a thermostatically controlled oil bath, while the mercury units 

 were in the ice bath adjoining. Connections from the tubes to 

 the bridge were made by inserting heavy copper conductors in 

 the glass terminal protecting tubes of the end bulbs employed, 

 the tubes being partly filled with mercury. 



Seven fillings of each tube for electrical comparison were 

 made in 1911. The international ohm as defined by the four 

 mercury standards was found to be 25.5 millionths smaller than 

 the international ohm as represented by the manganin coils at 

 that time. The average deviation of the four tubes from their 

 mean was but =t 5 parts per million. 



A second and third series of electrical comparisons, made in 

 June and December, 1912, showed the mercury standards to have 

 changed with respect to the wire standards. Redeterminations 



