638 TRANSACTIONS OK SECTION A. 



of the spark, though they are not invariably present in every oscillation ; generally 

 not more than about ten apppear in a single complete oscillation. These intervals 

 are the same in different photographs of the spark under the same conditions, but 

 the individual streamers are not always reproduced in the same oscillation in 

 different photographs. 



Department of Cosmical Physics. 

 The following Papers were read : — 



1 . On Neiv Methods of ohtaininy the Spectra of Flames. 

 By G. A. Hemsalech, M.Sc, D.Sc. 



The apparatus exhibited is a simplified form of the one which the author, in 

 conjunction with M. de Watteville, has devised for the production of flame 

 spectra. A glass bulb encloses a spark gap formed between electrodes of the 

 metal whose spectrum it is desired to obtain. The lower part of the bulb is pro- 

 vided with one or more tubular openings, aud the top communicates with a Bunsen 

 burner in such a way that the air feeding the burner flows through the bulb. 

 When powerful electric sparks (condensed) are allowed to pass between the elec- 

 trodes, metal vapour issues from the latter and difl'uses into the surrounding aii'. 

 This vapour generally combines with the oxygen and nitrogen of the air, and the 

 resulting product forms very finely divided particles, which to a great extent are 

 in the ultra-microscopic state. These particles remain suspended in the air and 

 are carried along with it into the Bunsen flame, where they give rise to the cha- 

 racteristic flame .spectrum of the element under examination. In many cases it is 

 preferable to supply the flame with the spray from a salt solution of the element. 

 The solution is contained in a small porcelain crucible placed at the bottom of the 

 glass bulb, and a condensed spark is taken from the surface of the solution by 

 means of a Cu. or Pt. electrode. The spectra thus obtained are very similar to 

 those given by the Gouy apparatus, and the new method has the advantage of not 

 necessitating the use of compressed air. The application of these methods to the 

 case of the oxy-coal gas, hydrogen, air-hydrogen, and oxy-hydrogen flames has 

 enabled M. de Watteville and the author to study the flame spectrum of an 

 element under various temperatures and other conditions. An examination of the 

 iron spectrum carried out in this way has disclosed the presence of ' spark ' or 

 ' enhanced ' lines in the cone of the Bunsen flame. On passing to flames of higher 

 temperature these lines diminish in relative intensity, and they are nearly all 

 absent from the oxy-hydrogen flame. Thus it appears that for iron the 'dis- 

 sociating power ' of the inner cone of the Bunsen flame is greater than that of any 

 other flame here considered. 



2. Polar Lines in Arc Spectra. By Dr. W. G. Duffield. 



In a paper which appeared in the ' Astrophysical Journal' (27, 260, 1908) a 

 list of lines occurring at the tips of the electrodes of a continuous current iron 

 arc were chronicled by the writer. Direct comparison with a spark discharge 

 showed a close resemblance in the ultra-violet between these lines and those 

 occurring in the spark ; this resemblance gradually disappeared as the region of 

 longer wave-length was approached, the lines under discussion becoming fewer 

 in number. The term ' polar ' lines was suggested to distinguish those lines 

 appearing at the tips of the poles from those in the centre of the arc or spark 

 (for which the term ' median ' lines seems suitable), and avoids the ambiguity 

 introduced by the term ' spark ' line when applied to the spectrum of a normal 

 arc. Not all of the polar lines occur in the spark ; a few are absent from it and 

 others are weak in it. The paper also discusses the efiect of density, pressure, 

 chemical action, potential gradient, temperature, and molecular velocities upon 



