522 Day and Sosman — Nitrogen Thermometer Scale 



Dickinson was introduced into the re-entrant tube of the gas 

 thermometer bulb itself. The resistance thermometer, which 

 was easily sensitive to a few thousandths of a degree, revealed 

 small temperature fluctuations (005°) in the rapidly circulating- 

 liquid outside the bulb but no systematic temperature differ- 

 ences. Within the re-entrant tube the fluctuations were no 

 longer felt. 



With these precautions to guard against temperature differ- 

 ences about the bulb, temperatures were measured (1) at the 

 boiling-point of benzophenone, (2) at the melting-point of zinc, 

 (3) at the melting-point of antimony.* The three thermo- 

 elements, after removal from the nitrate bath, were placed in 

 one or the other of the following : in a vapor bath of boiling 

 benzophenone, in an apparatus for determining the zinc melt- 

 ing-point, or in a similar apparatus containing antimony ; after 

 which they were returned to the gas thermometer furnace for 

 the verification of their readings. This series of operations 

 constituted a set of observations as carried out in the tables which 

 follow. Inasmuch as the gas thermometer was brought as close 

 as practicable to the temperature of the points (benzophenone, 

 zinc, etc.) selected as standards, the intermediary role of the 

 thermoelements was merely that of a transfer agent, in which 

 role the individual properties of the thermoelements do not 

 appear at all jjrovided the wires were originally homogeneous. 

 The danger of contamination of the elements and consequent 

 inhomogeneity is negligible at these temperatures. Even if 

 such contamination had crept in, it would have discovered itself 

 in differences between the readings of the elements with each 

 change in the gradient, of which differences no trace was 

 found. 



By way of providing a strictly rigorous test of the accuracy 

 of the transfer of temperature from gas thermometer bulb to 

 the reference standards and its independence of the intermedi- 

 ary thermoelement, a special arrangement was devised in the 

 case of zinc as follows : A steel bulb was made up with approx- 

 imately the dimensions of the gas thermometer bulb and sus- 

 pended in the same position in the nitrate bath. Enclosed in 

 this bulb was the charge of zinc in its graphite crucible (fig. 2). 

 In this crucible the thermoelement occupied the same position 

 which it occupied in reference to the gas thermometer bulb, 

 and all other conditions were, of course, identical. The zinc 

 melting-points were determined in this way, i. e., in a nitrate 

 bath in which there were no measurable temperature differ- 

 ences in the reo-ion about the melting zinc and with the tern- 



* The zinc and antimony were the same charges which were used in the 

 previous investigation. The analyses may be found in Pub. 157, pp. 87 and 

 88 ; this Journal, xxix, p. 159. 



