222 nutting: helium as light standard 



current sensibly as does the total light. Helium shows no trace 

 of any secondary spectrum in the visible region. 



Neither the voltage nor frequency of alternation of the excit- 

 ing current affect the light emission by a measurable amount. 

 Voltages of 1, 2, 5 and 10 thousand were used and direct, alter- 

 nating 60-cycle and 900-cycle current. 



Gas density does not affect the specific light emission over a 

 range from 3 to 8 mm. pressure provided the bore of capillary 

 is not much over 2 mm. With 25 m.a. current through 2 mm. -bore, 

 the light is sensibly independent of gas density from 3 to 8 mm. 

 pressure. Chiefly for this reason these specifications and 4 to 7 

 mm. pressure were adopted for the standard tubes inasmuch as 

 gas density is difficult to control and determine in such tubes. 



The correction to the light emitted by a centimeter length 

 of capillary for slight departures from 2 mm. in the bore was 

 carefully determined. It is —0.010 c. p. /cm. for each +0.1 mm. 

 difference in bore from 2.00 mm. 



A test for reproducibility was made on a set of forty tubes. 

 The capillary portions of these tubes were specially prepared and 

 selected by Baudin to secure the utmost possible uniformity of 

 bore and wall and freedom from striations. The tubes were made 

 up and filled at the Bureau of Standards. 



Color reproducibility was determined by intercomparing the 

 spectra of these tubes line by line with a spectrophotometer. 

 The slight variations observed were well within the errors of 

 observation. 



Photometric observations on the mean horizontal candle power 

 per centimeter length of capillary were made by four experienced 

 observers. The computed probable error in the c. p. /cm. for 

 all observers and all the tubes was 0.16 per cent. The average 

 deviation of the tubes from their mean value was 1.15 per cent, 

 the maximum deviation 3 per cent. The four different observers 

 read on an average 1.2 per cent below, 0.1 per cent above, 1.5 

 per cent above and 0.5 per cent below the mean of the four. 



The report on reproducibility tests is to be published in the 

 Bulletin of the Bureau of Standards. The preliminary work 

 was reported on in the Bureau of Standards Bulletin 4: 511. 1908; 

 the relation between line intensity and current in same journal, 

 7: 65. 1911. 



