determining the Vapour-Density of Metallic Vapours. 547 



effect due to it must be almost if not wholly negligible in 

 comparison with that of (a) and (b). This being the case, 

 the determination o£ the relative importance of the two 

 factors, pressure and temperature, is the question which at 

 once presents itself for solution. In his very exhaustive 

 article on "The Application of Interference Methods to 

 Spectroscopic Measurements " Professor Michelson has taken 

 up this problem in considerable detail. He finds that in the 

 cases where the density of the vapour is very low the effect 

 of changing pressure on the width of the spectral lines is 

 almost wholly negligible ; for hydrogen this is true even in 

 the case where the pressure is as high as 2 or 3 mm.; in 

 fact, when the relation between the breadth of the lines and 

 1/P (pressure) is plotted the influence of P is seen to become 

 vanishingly small at about 5 mm. In summing up, Professor 

 Michelson states as follows : — " It thus appears that in the 

 case of hydrogen — and probably in all other cases — the width 

 of the spectral line diminishes toward a limit as the pressure 

 diminishes, which limit depends upon the substance and its 

 temperature ; and that the excess of width over this limit is 

 simply proportional to the pressure." 



As mercury and sodium are both readily usable in vacuum- 

 tubes, the foregoing facts would suggest them at once as the 

 ideal substances for an experiment on the effect of pressure 

 and temperature on the broadening of the spectral lines. As 

 a preliminary to such an experiment a knowledge of the 

 densities of the saturated vapours at various temperatures is 

 of course necessary. In addition to making a determination 

 of the densities for such a purpose as the above, there is still 

 another and even more urgent reason, viz., the evident 

 dependence of the change in the lines in the Zeeman effect, 

 and also in some cases the reversal of the same, upon the 

 density of the light-producing vapour. It was particularly 

 with a view to the solution of this latter problem that the 

 following experiment was proposed and undertaken. 



Apparatus. — The method employed was one suggested by 

 Prof. Michelson, in which the amount of vapour filling a 

 known volume is determined by finding the amount of con- 

 densed metal in the observing flask when the latter is cooled 

 off. The apparatus consisted of three essential parts — the 

 heating-bath, the gas-bulb, and the thermometer. The bath 

 finally found most satisfactory is shown in section in fig. 1 

 (p. 548) ; a and b are two sheet-iron boxes lined inside and 

 out with heavy sheet-asbestos, and having a 3-inch air-space 

 between them ; the inner box (a) is about 14 inches on a side ; 

 around-the inside and on the bottom of a are a number of iron 



