32 INORGANIC EVOLUTION. [CHAP. 



large number of elements, very many comparisons are rendered possible. 

 All these suggest that if dissociation is really in question, in some 

 cases at least more than two simplifications in the line stage are 

 necessary to explain the facts. It is possible that the effects at first 

 ascribed to quantity may be due to the presence of a series of molecules 

 of different complexities, and that this is the true reason why " the more 

 there is to dissociate, the more time is required to run through the 

 series, and the better the first stages are seen."* 



After this general statement of the changes in spectra observed to 

 accompany change in the quantity and kind of energy used in the 

 experiments, I propose to refer briefly to the most recent work on this 

 subject, touching the changes observed on passing from the arc to the 

 spark in the case of many of the metallic elements. By the kindness 

 of Mr. Hugh Spottiswoode, the photographs of the enhanced lines have 

 been obtained by the use of the large induction coil, giving a 40-inch 

 spark, formerly belonging to Dr. Spottiswoode, P.K.S. I am anxious 

 to express here my deep obligation to Mr. Hugh Spottiswoode for 

 the loan of such a magnificent addition to my instrumental stock-in- 

 trade. 



The spark obtained by means of the Spottiswoode coil is so luminous 

 that higher dispersions than those formerly employed can be effectively 

 used, and in consequence of this, the detection of the enhanced lines 

 becomes more easy; their number therefore has been considerably 

 increased. 



At the higher temperature enhanced lines have been found Lo maKe 

 their appearance in the spectra of nearly all the metals already ex- 

 amined. Lithium is one exception. 



Neglecting then all changes at the lowest temperatures, but 

 including the flame spectrum, four distinct temperature stages are 

 indicated by the varying spectra of the metals ; for simplicity I limit 

 myself to iron as an example. These are : 



1. The flame spectrum, consisting of a few lines and flutings only, 

 including several well-marked lines, some of them arranged in triplets. 



2. The arc spectrum, consisting, according to Rowland, of 2,000 

 lines or more. 



3. The spark spectrum, differing from the arc spectrum in the 

 enhancement of some of the short lines arid the reduced relative 

 brightness of others. 



4. A spectrum consisting of a relatively very small number of lines 

 which are intensified in the spark. This, as stated above, we can 

 conceive to be visible alone at the highest temperature in a space 

 efficiently shielded from the action of all lower ones, since the enhanced 



* Pr.jc. Roy. Soc., 1879, No. 200. 



