480 



NA TUR£ 



[September 17, 1903 



very powerful magnetic field underwent the same kind of 

 change ; for instance, each quadruple group changed to 

 sextuple, the second and fourth lines in each group be- 

 coming double. Lines in spectra which have not the same 

 constitution behave differently. Recently Runge and 

 Paschen have arrived at the same conclusion ; and, further- 

 more, have established homology in the spectra of sodium, 

 copper, and silver ; also between aluminium and thallium. 

 Indium is almost certainly homologous with aluminium and 

 thallium, but it was probably not investigated on account 

 of its rarity. Marshall Watts has pointed out that a re- 

 lationship exists between the lines in the spectra of some 

 elements and the squares of their atomic weights, from 

 which it is possible to calculate the atomic weight of an 

 .element if that of another in the same homologous series 

 is known, and the oscillation frequencies of corresponding 

 lines are known. 



The knowledge of spectra we now possess enables the 

 determination of atomic weights to be controlled with quite 

 as much efficiency and certainty in many instances as by 

 specific heat or vapour-density determinations. 



The first application of the observed homology in spectra 

 was directed towards the question of the atomic mass of 

 beryllium, for which purpose the lines in the ultra-violet 

 spark spectrum of this element were first photographed 

 and measured. The nature of the evidence on the subject 

 adduced at the time was in outline as follows : — 



" If, as Nilson and Petterson suggest, the position of 

 beryllium is at the head of a series of triad rare earth 

 metals, the element scandium (at. wt. 44) and yttrium 

 (at. wt. 89) must be members of the same group. If this 

 be the case the spectra of the three elements must have 

 certain characters in common, for the series of which 

 aluminium and indium are the first and third terms yield 

 strictly homologous spectra. As a matter of fact no two 

 spectra could be more dissimilar than those of beryllium 

 and scandium." 



Having compared the photographs and wave-length 

 measurements of a large number of spectra of the elements, 

 I felt justified in making the following remarks : — • 



" The spectrum of beryllium exhibits no marked analogy 

 with the calcium, the magnesium, or the aluminium spectra, 

 all of which are members of well-defined homologous series. 

 There is nothing similar in it to the boron, silicon, or 

 carbon spectra, nor to those of the scandium, yttrium, or 

 cerium. The spectrum of lithium is most closely analogous 

 to that of beryllium in the number, relative positions, and 

 intensities of the lines. This leads to the conclusion that 

 beryllium is the first member of a dyad series of metals, 

 to which in all probability calcium, strontium, and barium, 

 as a sub-group, are homologous, its atomic mass being 

 92, its place is above magnesium." Subsequently Nilson, 

 and also Humpidge, by chemical evidence and from vapour- 

 density determinations of certain compounds, substantiated 

 the conclusion previously arrived at by Emerson Reynolds, 

 that the atomic mass of beryllium was not 13-8 but 9 2. 



The next practical application of the spark spectra was 

 to the analysis of rhabdophane, a mineral found many years 

 ago in Cornwall and described by Heuland in 1837 as a zinc 

 blende of a peculiar character. 



This mineral I found to contain neither zinc nor sulphur, 

 and therefore it is not a blende. It is, in fact, a phosphate 

 of the formula R,0,.P205.2H,0, in which the oxides of 

 cerium, didymium, lanthanum, and yttrium may wholly or 

 in part replace each other. The didymium absorption spec- 

 trum is well seen both by reflection from the surface and 

 transmission through thin sections of the mineral. - The 

 spark spectrum of the yttrium chloride obtained from 

 rhabdophane was compared with that observed by Thal6n 

 and ascribed to yttrium. Of the fifty-one lines in the spec- 

 trum of yttrium thirty-eight were absent from the yttrium 

 obtained from rhabdophane, and it was concluded that the 

 purest yttrium was that which yielded the simplest spec- 

 trum. This was the first occasion of the finding of yttrium 

 in any British mineral. Quite recently a confirmation of 

 this view has been obtained by comparing this spectrum 

 with lists of the arc lines of yttrium and ytterbium which 

 have just been published by Kayser (1903). 



Penfield analysed a mineral found in the United States 

 which he named scovellite : it proved to be identical in 

 species with rhabdophane. 



NO. 1768, VOL. 68] 



Flame Spectra at High Temperatures. 

 What are commonly known in the chemical laboratory 

 as flame spectra are chiefly those of the metals of the 

 alkalies and alkaline earths ; also of gallium, indium, and 

 thallium. The researches of Mitscherlich and Lecocq de 

 Boisbaudran first showed that copper, manganese, and gold 

 gave flame spectra. Lockyer, Gouy, and Marshall Watts 

 also investigated flame spectra. 



In 1887 I used iridium wires one millimetre thick, twisted 

 into loops upon which fragments of minerals were heated 

 in the oxygen blowpipe flame. Natural silicates yielded 

 spectra not only of alkalies but of the alkaline earths, and 

 also distinct manganese spectra. Baryta, strontia, and 

 lime gave spectra when insoluble compounds such as the 

 sulphates were thus examined at high temperatures. Iron, 

 cobalt, and nickel gave spectra even when compounds such 

 as the oxides were heated strongly. But iridium, though 

 infusible, is somewhat volatile, and contributes a line spec- 

 trum to the flame. In 1890 thin slips of the mineral kyanite 

 and even pieces of tobacco pipe were used instead. Ex- 

 perience with this method of working went to show how 

 the flame spectra of oxides of calcium, strontium, and 

 barium could be separated from those of lithium, sodium, 

 potassium, rubidium, and caesium, as observed in the 

 Bunsen flame. Furthermore, that even the most volatile 

 of these substances could be made to yield a continuous 

 coloration from a single bead of salt for a period exceeding 

 fifteen minutes, and extending to one or two hours, so that 

 measurements of the lines might be made with some degree 

 of certainty. 



In order to study the flames emitted from furnaces during 

 metallurgical operations, and particularly from the mouth 

 of Bessemer vessels, it became necessary to ascertain what 

 really were the lines of the elements observed under different 

 conditions at a high temperature, and accordingly system- 

 atic methods of study were developed from the previous 

 somewhat tentative experiments. 



In all the flame spectra obtained by the oxyhydrogen 

 blowpipe the ultra-violet line spectrum emitted by water 

 vapour which had been discovered by Huggins and by 

 Liveing and Dewar was visible on the photographs by 

 reason of the combustion of the hydrogen in the hydro- 

 carbon, or the hydrogen gas itself, when burnt along with , | 

 oxygen. The flame spectra are always shorter than those 

 obtained from the arc or from condensed sparks. After an 

 extended examination of spectra produced by the oxy- 

 hydrogen blowpipe from solid substances, the knowledge 

 obtained was applied to the examination of the flames 

 coming from the Bessemer vessel during the " blow " during 

 all periods from the commencement to the termination. 

 These observations were made at the London and North- 

 western Railway Steel Works at Crewe ; and at Dowlais, 

 in South Wales. In collaboration with Mr. Ramage, a 

 large number of these complicated spectra were photo- 

 graphed at the North-Eastern Steel Works, where the 

 Thomas-Gilchrist process is carried out. The spectra were 

 fully described and measured, with the result that every one 

 of the lines and bands was accounted for. A new line 

 belonging to potassium was discovered to have peculiar 

 properties. Gallium was proved to be present in the Cleve- 

 land ore from Yorkshire, in the finished metal, in clays 

 and in all aluminous minerals, even in corundum. Also, 

 by very accurate determinations of the wave-lengths of its 

 principal lines, gallium was proved to be a constituent of 

 the sun. Moreover it was found in several meteorites. 

 Pure gallium oxide was separated, by analytical methods, 

 from iron ores and other materials ; and the proportion of 

 the metal in the steel rails made by the North-Eastern Steel 

 Company, of Middlesbrough, was determined and found to 

 be one part in thirty thousand. This Yorkshire steel is 

 richer in gallium than any other substance from which it 

 has been extracted ; for instance, the Bensburg blende, sup- 

 posed hitherto to be the richest ore, contains only one part 

 in fifty thousand. 



By observations on the spectra, the thermo-chemistry of 

 the Bessemer process of steel manufacture was studied, and 

 the temperatures attained under varying conditions were 

 estimated. The demonstration of the great volatility of 

 most metals, and of many metallic oxides in an undecom- 

 posed condition, at the temperature of the oxyhydrogen 

 blowpipe and of the Bessemer flame was of special interest. 



