1901.] and Melting Points to Atomic Mass. 3 



Upwards of thirty metals yield spectra when heated in the oxyhy- 

 drogen flame : these metals also yield arc and spark spectra. The 

 flame spectra are the simplest : the spark spectra are the most com- 

 plicated. 



Under the conditions which obtain in flames, the metals are usually 

 heated to a temperature not very far removed from their boiling 

 points, while the electrical conditions are probably simpler than obtain 

 in the electric arc or spark. There are present in the Bunsen flame, in 

 addition to the vapour of the metal, various gases such as : hydrogen,, 

 oxygen, oxides of carbon, steam and nitrogen, all heated to a high 

 temperature. Some of the chemical changes which take place in the 

 flame are reversible, and the metallic vapours are consequently in 

 presence of gases, some of which are combining together, some are 

 the compounds formed, and some of the latter are undergoing disso- 

 ciation. Similar changes occur in the oxycoal-gas flame. The reactions 

 in the oxyhydrogen flame are simpler, but similar spectra of metals 

 are obtained from both of these high temperature flames. 



It is possible, by varying the temperature of the flame, to obtain 

 the simplest or fundamental spectra of many of the metals, and also 

 to study the order of appearance of additional lines as more complex 

 spectra are produced. 



The metals of the alkalies and alkaline earths give spectra in the 

 Bunsen flame. The former yield line spectra, the latter yield spectra 

 composed of lines and bands. The bands and some of the lines have 

 been attributed to the oxides of the metals. Indium gives a spectrum 

 composed of two lines, and thallium a spectrum of one green line in 

 the Bunsen flame. No other metals give spectra of importance in this 

 flame. 



Spark spectra of salt solutions, produced without the aid of a Leyden 

 jar, are very similar to the arc and high temperature flame spectra of 

 the metals present in those salts. 



Flame spectra, then, furnish purely experimental data with which to 

 begin an investigation of the laws which govern the distribution of 

 the lines in spectra, and by which to study the relations of the 

 physical and chemical properties of the metals to their spectra. 

 Further experimental data are furnished in studying more complex 

 spectra by : — 



(a.) The self-reversal of lines. 



(b.) The character of the lines — sharp or nebulous ; continuous or 

 discontinuous. 



(c.) Hartley's discovery of the constant differences between the 

 oscillation frequencies of the components of doublets and 

 triplets. 



(d.) The Zeeman effect. 



B 2 



