﻿944 Dr. F. H. Newman on a 



It is desirable to recall two alternative conditions which 

 have been introduced into the proof of the formula. They 

 are that cv/k shall be equal to unity or else very large. If 

 these conditions are not satisfied, it may be shown that the 

 formula becomes 



£.-'(^?). ^ 



where cv/k is expected to be of little importance in the two 

 extreme cases mentioned. 



Whether the simpler formula is true for gases on the one 

 hand and for very viscous fluids on the other, and whether 

 cv/k is important for intermediate circumstances, is a matter 

 for experimental investigation. From data already available 

 it is known that the simpler form is fairly satisfactory for 

 gases, being indistinguishable in this case from the grouping 

 of variables deduced by Boussinesq for inviscid fluids. This 

 is shown graphically in a curve between lih/kO and ^L/p, 

 given elsewhere *, for the cooling of wires in a stream of air 

 (H/& and vl/v in the notation of the actual graph). 

 Corresponding data for liquids are not yet available, but an 

 isolated result has been given by Worthington and Malone f 

 for the cooling of a wire in water (cv/k — 1 ; i/ = 0*01006), 

 and this yields a result, H/& = 21*6, vl/v = 38, which is in 

 satisfactory accord with the curve mentioned for air. In the 

 analogous case of natural convection, cv/k has been shown to 

 be of little importance for a wide range of viscous fluids. 



July 1922. 



LXXXI. A Sodium-Potassium Vapour Arc Lamp. By F. H. 

 Newman, D.Sc, A.R.C.S., Bead of the Physics Depart- 

 ment, University College, Exeter %. 

 [Plate V.] 



METALLIC arcs operated in vacua give very intense 

 radiation, and the lines in the resulting spectra are 

 very narrow, whereas with a substance placed between the 

 poles of a carbon arc, working under ordinary conditions, 

 broad lines are obtained, which often show much reversal, 

 the centres of the lines being comparatively faint. This is 

 the case when the sodium D lines are excited, and a bunsen 

 flame, to which salt has been added, is not a suitable source 

 of sodium radiation. As the amount of salt is increased, the 



* Phil. Mag. xli. p. 899 (1921). 



t Journ. Frank. Inst, clxxxiv. p. 115 (1917). 



X Communicated by the Author. 



