90 



NATURE 



[^Vtf:'. 22, iSSj 



In Messrs. 1 iveing and Dewar's n^.ap of the carbon spectrum 

 [Froc. A'oy. Soc. vol. xxxiii. p. 403), and in the list of the carbon 

 lines and in the map of the iron spectrum {f/iil. Irons, part i, 

 1883), a nomber cf lines are given which are ab>ent from the 

 phoK graphs of the spectrum of graphite published in the 

 Trntisactions of the Kcyal Lublin Society and in the Journal oj 

 the Clutiiical Society (\iA. xli. p. 90). Many liundrtds of spectra 

 taken between graphite poles have failed to show a trace of 

 these lines, and as the spectra have been photographed under 

 very various coi ditions, it is scarcely likely that the lines in 

 question are really carbon Ulcs. They have now been identified 

 in the spectrum of silicon. 



Sil.i 



Spark ' 



2881-1 



2541 'O 

 2528-2 

 2523-6 

 2518-7 

 2515-8 

 2514-0 

 2506-3 



(Hartley) 



2S810 



2541-0 , 



2528-1 



2523s 

 2518-5 

 2515-5 

 2513-7 

 2506-3 



Strong, sharp. 



■Very strong, extended. 



Strong, sharp. 



2528-1 

 2523-9 

 2518-8 

 2515-8 

 2514-1 

 2506-6 

 2478-3 

 2434-S ... 2435-5 



From this it appears that, in the spectrum of the arc, carbon 

 yields but one line in the ultra-violet, wave-length 2478-3. It is 

 perhaps a liitle doubtful whether the line with wave-lengih 

 2434-8 is due to silicon or not. 



The Spectrum of Beryllium. — The researches made for the 

 purpose of this report have been useful in furnishing evidence 

 leading to a determination of the probable position of beryllium 

 among the elements. It has been proved that the spectra of 

 metallic solutions are identical with those of the metals them- 

 selves, and it is therefore obvious that characteristic spectra 

 may be obtained from concentrated solutions of nitrates or 

 chlorides when metallic electrodes are not procurable. 



It was resolved to photograph the spectrum of beryllium, a< 

 obtained from its chloride, in order to observe the cliaracter of 

 its lines and the manner of their grouping, 'ihe following were 

 the lines observed : — 



Wave-!ength 



3320-I 



3129-9 



2649-4 



2493-2 



24777 



The fir.^t two numbers differ slightly from those given in the 

 Journal of the Chemical Society (Jnne, 1883), but they are be- 

 lieved to be the more accurate. The previous measurements 

 of the lines of berylhum were two given by Thalen ( Watts's 

 " Index of Spectra "), with wave-lengths 4487 and 4575, and two 

 lines very close togethrr given in Cornn's " Map of the Solar 

 S|iectrum," wave lengths 3130 and 3130-4. It will be observed 

 that in the spark spectrum only one line corresponding to the 

 first of thes; is observed, viz. 3129-9. There is probably a dif- 

 ference in this case between the arc and the spark spectra, be- 

 cause there i^ no difficulty in distinguishing between two lines 

 differing by 0-4 of a tenth-metie, and under various conditions 

 two lines have never been observed at this point in the spark 

 spectrum. On the other hand, such differences are by no means 

 unusual. 



Regarding the views held by Emerson Reynolds, Nilson and 

 Pettersson, and Brauner on the subject of beryllium, however 

 wanting in harmony they may be in detail, they at least agree 

 in assigning a value not greater than 13-8 and not less than 

 9-210 its atomic weight. The former number implies that the 

 metal is a triad, the latter that it is a dyad. In the former case 

 it must belong either to the series of elements of which alu- 

 minium, gallium, and indium are members, or to a sub group of 

 rare earth-metals, to which yttrium and scandium belong. In 

 attempting to accommodate the element with a position in either 

 series we are met by a serious difficulty, namely, that not only 

 is the atomic weight out of keeping with the periodic law (a 

 point which cannot be ditcusfed here), but its speci rum is alto- 

 gether different from the spectra typical of either cla-s. 



There is a periodic variation in the spectra of the elements as 

 ■weU as in their atomic weights and chemical properties, and we 

 ' Proc. Roy. Soc. 2 Phil. Trans. 



cannot put the periodic law out of mind in considering the posi- 

 tion of beryllium. 



Now the spectra typical of the triad group, of which aluminium 

 and indium are the first and third terms, consist of three ptirs 

 of lines harmonically related, the intervals between the indi- 

 viduals of each pair increasing with increased refrangibility of 

 the rays in each spectrum, while the intervals between the indi- 

 viduals in each pair in different spectra increase with the increase 

 of atomic weight. The interval between each pair of lines 

 contains an isolated lay. As the atomic weight of beryllium is 

 less than that of aluminium, it should have a spectrum in which 

 the same grouping appeai-s, but the intervals between the pairs 

 of lines should be shorter, and the individuals of each pair should 

 be closer together. 



The lines of beryllium are not characteristically grouped like 

 those of aluminium and indium ; it cannot therefore belong to 

 this series of elements. If » e attempt to classify beryllium in 

 a marmer which accords with Nilson and Pettersson's views 

 (Proc. Ko\'. Soc, 18S0, vol. xxxi. p. 37), the elements scandium 

 and yttrium, with atomic weights 44 and 89 respectively, must 

 yield spectra typical of the series, aiid the similarity between the 

 spectra of the two metals, beryllium and scandium, must be 

 exceedingly close. Now Thalen's spectra of scandium and 

 yttrium, though both totally unlike the spectra of any other 

 element, have many characters in common (A'oiigl. Svenska 

 Ahademicus HaniHingar^ vol. xii. p. 4, and Com/tes Ktndiis, 

 vol. xci. p. 45) ; both spectra contain highly characteristic groups 

 of lines in the oiange and )ellow-, the lines or bands degrading 

 towards the red, and the number of lines which have been 

 measured are no fener than 103 and 90 respectively. 



From these two spectra that of beryllium is entirely different, 

 as well in the character and grouping as in the number of the 

 lines. Of the remaining rare earth-metals at present known, 

 cerium is a tetrad, didymiura is a pentad, and lanthanum a triad ; 

 their spectra are quiie oissimilar from that of beryllium. In 

 consideration of these facts it is impossible to classify the 

 spectrum of beryllium along with the spectra of the rare earth- 

 metals of the triad group. 



Let us now consider thequestijii of the dyad groups. On the 

 assumption that beryllium ha> an atomic w eight of 9-2, there is no 

 difficulty in placing it at the head of the second series, in which po- 

 sition it stands in the same relation to the sub-groups Mg, Zn, Cd, 

 and Ca, Sr, Ba, that Li occupies with regard to Na, K, Rb, Cs, 

 and Cu, A"-, Hg. Its pcsition willal-o be similar to that of B and 

 of C in their relation to the triad and tetrad metals. The spectra 

 belonging to Mg, Zn, Cd, have a very definite constitution ; they 

 con.sist of (i) a single line, (2) a 1 air of lines, (3) three to four 

 groups of tri])lets, (4) a quadruple group, and (5) a quintuple 

 group of lines. The intervals bemeeii the individual lines in 

 I heir different groupings increase with the increase in the atomic 

 weights of the elements. In fact these spectra present a con- 

 siderable addition to the body of evidence in support of the 

 view that elements whose atomic weights differ by an approxi- 

 mately constant ([uantity, and whose chemical character is 

 similar, are truly horao'ogous bodies, or, in other words, are the 

 same kind of matter in different states of condensation [Journal 

 of the Chemical Society, September, 1883, p. 390, Trans.). 

 Their particles are vibrating in the same manner, but with 

 different velocities. In the spectra of the metals Ca, Sr, Ba, 

 successive pair's of lines ai-e a strong feature, in addition to which 

 there are some other groups in the spectrum of barium. The 

 individuals of each pair are separated by smaller intervals the more 

 refrangible the lines, and by lai-ger intervals the higher the atomic 

 weights. It cannot be ^aid that the spectrum of beryllium is 

 precisely similar in constitution to either of these groups of 

 elements. 



I'liere is some slight resemblance in character to the spectrum 

 typical of the calcium group, beryllium having two pairs of lines, 

 the individuals of the first or less refrangible pair being separated 

 by a greater interval than tho^e of the second pair. It is a spec- 

 trum analogous to that of lithium, hence it was concluded that 

 beryllium is the first member of a dyad series of elements to 

 w hich probably calcium, strontium, and barium are more strictly 

 homologous than magnesium, &c. It is to be understood that 

 this is a conclusion drawn from one point of view only, and is 

 open to correction or modification \\hen fresh facts shall have 

 been discovered, but so far the classification of beryllium among 

 the dyads is confirmed and maintained by its po.sition being in 

 harmony with these spectrum observations. The metal is shown 

 to be quite < ut of pUite among the triad elements. 



