34 Dr. Foote and Dr. Mohler on Ionization and 



potentials for electrons in vapours o£ zinc, cadmium, sodium, 

 and potassium. The present paper is a continuation of this 

 work^ using vapours of magnesium and thallium. 



The experimental arrangements have been full}' described 

 in the above-referred to papers. One modification has been 

 introduced for the work with thallium in that an equi- 

 potential surface was used for the hot cathode. This method 

 of producing a more nearly uniform distribution of velocities 

 of the electrons, first suggested by Groucher*, has proved 

 most satisfactory. A helix of tungsten wire is threaded through 

 a small porcelain cylinder, and the equipotential surface from 

 which the electrons are accelerated consists of a platinum 

 or preferably, for thallium, a nickel cylinder, coated with 

 lime, fitting tightly over the porcelain. This surface is heated 

 to about 1000° C. by the current through the tungsten helix, 

 and a thermionic current of several milliamperes may be 

 maintained with an accelerating potential of a few volts. 

 The metal and ionization-chamber were heated in a glazed 

 porcelain tube to about 600° C. for magnesium and 900° C. 

 for thallium. Hard glass and iron tubes were also used for 

 magnesium. Both of these metals attack porcelain very 

 rapidly at these temperatures. At temperatures above a 

 glow the thermionic emission from the cylinders of the 

 ionization-chamber and from the walls of the porcelain 

 tube become very troublesome, and for this reason the 

 accuracy obtained in earlier work upon metals of lower 

 boiling-point is not to be expected. The apparatus was 

 evacuated by means of a Stimson and a Langmuir conden- 

 sation-pump to a pressure of less than 0'002 mm. Hg as 

 read by a McLeod gauge. 



Fig. 1, curves 2, 3, 5, 6, 7, and 8, and fig. 2, curves 1, 3, 4, 

 5, 6, 7, and 8, for magnesium, represent the current between 

 the inner net and outside cylinder, through a retarding field 

 of about 0*7 to 1 volt, as a function of the accelerating field 

 applied between the hot wire and net. The total current 

 from the hot wire as a function of the accelerating field is 

 represented by fig. 1, curves 1 and 4, and by fig. 2, curves 2, 

 2 a, 9, 10, and 11. The analysis of these curves is given in 

 Tables I. and II. The partial current curves are characterized 

 by successive maxima and minima due to inelastic collision of 

 resonance. We have chosen as before the points at which 

 such inelastic collisions begin to show an appreciable effect 



* Goucher, Phys. R. viii. p. 561 (1916). 



