TRANSACTIONS OF SECTION B, 595 



It suffices here to state that a solution containing entirely the one form may be 

 converted wholly into the other. 



The two formulse referred to are given below : — 



CH(0H).N.CH3 0H = N(CH,).OH 



^CH, CH, CH,.CH,_, 



Carbiuol'Form. ' Ammonium Base. 



Emission Spectra. 

 SiMvk Spectra and their Constitution. 



As it became necessary to make accurate measurements of absorption spectra 

 in the ultra-violet, the work of obtaining the wave-lengths of Unes in twenty 

 metallic spectra was undertaken. They were for the most part in a region 

 which, except in the case of two or three elements, had not been previously 

 explored. A small Rutherford grating was employed, combined with quartz 

 lenses with a focal length of three feet. Experience had shown that it was advisable 

 in describing these spectra to give measurements in hundredths of an inch of the 

 positions of the lines on the published photographs of the prismatic spectra in the 

 * Journal of the Chemical Society ' (March 1882), and to follow Lecocq de Bois- 

 baudran by giving a description of the character of each of the lines. In this way 

 they are easily identified, and the value of the measurements for practical purposes 

 is greatly enhanced. Prior to the publication of the work (1882), in the prosecution 

 of which Dr, Adeney was associated with me, Liveing and Dewar, who had been 

 engaged on a similar investigation, but operating in a different manner, published 

 an account of the spectra of the metals of the alkalies and alkaline earths, and 

 subsequently the hues of iron, nickel, and cobalt. They showed a rhythmic 

 grouping of the lines to be characteristic of the spectra of the alkali metals. 



In connection with the prismatic spectra which were photographed some 

 remarkable facts were noticed ; for instance, the character of the lines belonging to 

 different groups of elements was a noticeable feature, as well also their disposition 

 or arrangement, more particularly in the ultra-violet. Similarities in the visible 

 spectra of zinc and cadmium, of calcium, strontium, and barium, and in those of 

 the alkali metals had been observed by Mitscherlich, by Lecocq de Boisbaudran, 

 and also by Ciamician. As to the grouping of the lines as observed on the 

 photographs, it appeared that the spectra of well-defined groups of elements had 

 characteristics in common which were different from those of other groups. For 

 instance, the alkali metals differed from the alkali earth metals which appeared to 

 form a group by themselves. Then in marked contrast to these simple spectra 

 were those of iron, nickel, and cobalt, which though very complicated were seen 

 to be much alike. Nearest to these but differing from them in certain respects 

 were the palladium, gold, and platinum spectra. 



It was observed how these elements with certain chemical and physical 

 properties in common could be recognised as being relations owing to their family 

 likeness when their spectra were photographed. Then it was remarked that the 

 spectra of magnesium, zinc, and cadmium, had distinctive characters in common ; 

 for instance, the individual lines in these spectra were marked by similar character- 

 istics, such as a great extension of the strong lines above and below the 

 points of the electrodes. This extension was increased with the atomic mass of the 

 metal, and with the greater atomic mass in this group the volatility of the metal 

 is also greater. An arrangement of the lines in pairs and triplets was noticed, the 

 triplets being repeated, but less distinctly than in the first instance, and again 

 repeated sharply but less strongly, so that there were three different sets of triplets 

 in each spectrum. The point of greatest interest and importance was the 

 connection traced between the atomic mass and the numerical differences observed 

 in the intervals between the lines of different groups when measured by their 

 OBcillation frequencies. 



These differences were not in the spectrum of one element, but were in the lines 



Q Q 2 



