of Light by 'Potassium Vapour. 291 



visible spectrum. The cylinder of vapour was maintained 

 successfully, by means of a heating coil, in a steel tube with 

 glass ends. The behaviour of the vapour in this respect 

 was found to be similar to that of sodium. 



The arrangement of the apparatus was as follows-: — 



A steel tube, length 76 cm., internal. diameter 3 cm., was 

 fitted with brass water-jackets at its ends and a short side 

 tube for connexion with an exhaust-pump. 



On the centre part of this tube a heating coil of about 

 100 turns of nickel wire was wound on a layer of asbestos 

 cloth — the length of this heating coil being 33 cm. 



A nickel wire was soldered on to the outer surface of the 

 steel tube halfway between its ends and a second nickel wire 

 on to one of the brass water-jackets, thus constituting a 

 thermo-couple for the measurement of the temperature at 

 the centre of the heating coil. 



Inside the tube was fitted a piece of D cycle tubing with 

 the plane side placed vertically so as to divide the tube into 

 two parts along its whole length. 



In one half of the tube and about halfway along it a few 

 small pieces of metallic potassium were placed. A pair of 

 thick optically plane parallel glass plates were then fixed to 

 the two ends with sealing-wax. 



The whole tube was mounted on a light metal framework 

 provided with levelling-screws and the height adjusted to 

 that of a Hilger spectroscope. 



When a current of 2*5 amps, was passed through the 

 heating coil the potassium vaporized, and by adjustment of 

 the current steady temperatures up to 400° C. could be 

 obtained. The temperatures were calculated from readings 

 of the millivoltmeter attached to the thermo-couple which 

 was standardized. 



The air and hydrogen were pumped from the tube to a 

 pressure of about 1 mm. of mercury. 



The tube was placed between the plates of a Jamin inter- 

 ferometer — the dimensions of the plates being 2x2x5 cm. 

 The plates were mounted on an optical bench about 90 cm. 

 apart. Light from a carbon arc passing through a leus was 

 reflected from the front and back surfaces of the first plate, 

 so that the two interfering beams travelled along ihe two 

 halves of the steel tube. The interference fringes formed 

 after reflexion at the second plate were focussed on to the 

 slit of a Hilger wave-length spectroscope of the Constant 

 Deviation type. 



The plates were adjusted until the fringes were horizontal, 

 and the spectrum, as seen through the telescope, was then 



X 2 



