68 



NATURE 



[November 19, 1908 



density of the material) for the soft set being approximately 

 four times that for the hard. 



(2) For each set of rays the value of A/ A is constant, 

 and practically independent of the nature of the absorbing 

 material with which \ is measured, provided that in the 

 case of the soft rays secondary effects be excluded. 



(3) Secondary 7 radiation appears on both sides of a 

 plate which is penetrated by a stream of 7 rays. There 

 exists a marked lack of symmetry between the amount of 

 secondary radiation which proceeds from the two sides. 



(4) A lack of symmetry exists in the case of some sub- 

 stances between the quality of the radiation on the two 

 sides. 



(5) The last results seem very difficult to reconcile with 

 a pulse theory. On the " material " theory propounded by 

 Prof. Bragg no such difficulty arises. 



(6) The secondary 7 radiation appears to be derived from 

 the primary by a process of scattering, this process 

 generally involving a reduction in the subsequent pene- 

 trating power of the ray affected. 



{7) There appears to be reason to believe that the dis- 

 tribution of the scattered radiation depends to some extent 

 upon the hardness of the radiation which is scattered, also 

 upon the nature of the material in which the scattering is 

 produced. The softer radiation appears to be turned back 

 to a somewhat greater extent than the hard. Materials 

 of high atomic weight seem to be able to produce more 

 complete scattering than those of lower atomic weight. 



(8) The absorption of 7 radiation which has already 

 passed through a thickness of one substance by screens 

 of a different substance may not in all cases give a true 

 measure of the absorption of the original radiation which 

 has been effected by the first screens. 



J. P. V. Madsen. 



University of .^delaide, October i. 



[As there are few opportunities in .Vuslralia for an in- 

 vestigator to place his views quickly before a scientific 

 public, we print the above letter, but with it the corre- 

 spondence must cease. The subject is more suitable for 

 discussion in special journals devoted to physics than in 

 our columns. — Ed. Nature.] 



The Origin of Spectra. 



The very interesting observation of the anomalous dis- 

 persion of luminous hydrogen in the neighbourhood of the 

 Ha line recorded by Messrs. R. Ladenburg and Stanislaw 

 I.oria in N.-mure of November 5 (p. 7), and the known 

 absence of the phenomena in ordinary hydrogen, show 

 conclusively that ihe spectrutn lines of a substance are 

 not free periods of the atoms in their normal state, but 

 only of those systems produced sonwhoiv by the aj^ency 

 which gives rise to the spectra. 



The figure 1/50,000 as the number of electrons per atom 

 of course means that in the gas under experiment only 

 one atom in 50,000 was emitting the Ha line at any one 

 time. The very important remark is made that the 

 anomalous dispersion in the neighbourhood of the other 

 lines of the hydrogen series " is expected to be nnuh 

 smaller than that at the Ho line." If this be .so, it will 

 show that at any given time different numbers of atoms 

 are producing the different lines, that is to say, that the 

 spectrum is not produced in toto by each atom. Kach 

 atom (or rather the system emitting the lines) may. for 

 instance, only be emitting one line at a time. These 

 results are the same as those I have deduced from Prof. 

 R. W. Wood's work on the anomalous dispersion of 

 sodium vapour. Sodium vapour shows anomalous dis- 

 persion in the neighbourhood of all the lines of the prin- 

 cipal series, which " is very strong at D, feeble at the 

 first pair of ultra-violet lines A 3303, and almost imper- 

 ceptible at X 2852." It is also. Wood states, stronger 

 at D, than at D,. This shows that the number of atoms 

 emitting D, at any time is greater than the number 

 emitting D,, and both these are much greater than the 

 numbers emitting the higher members of the series. We 

 note that there is no anomalous dispersion in the neigh- 

 bourhood of the lines of the subordinate series of the 

 sodium spectrum showing that heat alone does not pro- 

 duce those svstems which vibrate with the periods of the 

 subordinate series, which agrees with the facts that these 



NO. 2038, VOL. 79] 



series do not appear in the absorption spectrum of sodium 

 vapour or in the Bunsen flame spectrum of sodium. 



It thus seems probable that different series of lines in 

 a spectrum are produced by entirely different vibrating 

 systems, while any system possibly only emits one line 

 at a time of its own particular series, depending upon the 

 manner in which it has been struck. It is evident that 

 the different vibrating systems obtained, and their relative 

 proportions, may be expected to vary with the nature of 

 the electrical discharge producing the spectra, and hence 

 the variation of the spectra under different conditions. 

 This may, perhaps, on the modern views, be regarded as 

 the same idea put forth many years ago by Sir Norman 

 I-ockyer in his dissociation hypothesis. 



I make these observations in order that those working 

 on the subject from the theoretical side may the better 

 see the phenomena to be explained, which are quite 

 different from ordinary dynamical vibrating systems. 



In conclusion, 1 should like to direct attention to the 

 importance of extending Messrs. Ladenburg and Loria's 

 work. By examining every line in the spectrum of an 

 clement we could, for instance, say whether a line was faint 

 because very few systems were emitting it, or whether its 

 faintness must be attributed to the fact that the vibrations 

 producing this line have only a very small amplitude. 



.\lbert E.11GLE. 



Imperial College of Science and Technology, 

 London, November 9. 



A Gall-producing Dragonfly. 



When looking through Dr. C. Houard's new worlc on 

 galls (" Les Zooc^cidies des Plantes d'Europe et du Bassin 

 de la M^diterran^e," tome i.), I was surprised to find 

 on p. 249 an entry : — " Minime borselette O. pcd. I.cstcs 

 viridis, \'an der Lind." 



.\ gall-producing dragon-fly was quite new to me, but 

 on looking up the subject I found a scries of very 

 important observations on the oviposition and larva of 

 the species in question by the .Abbe Pierre and M. de 

 Roequigny-Adanson, in the Revue scienlifique du Hour- 

 honnais et du Centre de la France, -xv. and xvi. (1902-3), 

 and the Annales et Btdlelin de la Sociiti entomologique 

 de France for 1904. .As these seem to have been entirely 

 overlooked in England, I think it may be useful to 

 epitomise them as briefly as possible. 



The eggs of Lestcs viridis are laid on the branches of 

 a great variety of deciduous trees and shrubs, but always 

 close to, or overhanging, water, and therefore probably 

 most often on alders or willows. These result in the pro- 

 duction of small galls, which are sometimes extremely 

 abundant, and which are thus described by Pierre : — 



" Un bourrelet mesure de i mm. J i 2 mm. de longueur, 

 sur 5 ou I mm. de largeur. Deux bourrelets sont associ^s 

 en chevron et forment un angle d'i peu prfes go°, ouvert 

 vers le bas du ramcau. Le sommet de Tangle presentc 

 unc pellicule corticale plus ou moins arrondie, formant 

 clapet au dessus de I'ouverture par laquelle de i & 4 

 oeufs ont 6t& ins^r(5s sur chaque bourrelet. Enfin Ics 

 chevrons distants de 2 mm., sont associ(5s en s^rie longi- 

 tudinale, de telle fa^on qu'une m6me g^nc5ratrice du 

 rameau soit sensiblement bissectrice de tous les angles." 



The emergence frojn the eggs and the structure of the 

 larva are equally curious. The new-born larva, or " pro- 

 larve, " as Pierre calls it (Ann. Soc. Enl. de France, 1904, 

 pp. 477-84, pi. iv.), resembles a coleopterous pupa, being 

 enclosed in an outer membrane which leaves it only the 

 power of leaping. If these young larva; do not fall into 

 the water on emerging from the egg, they leap about, 

 sometimes for several hours, until thev succeed in reach- 

 ing it. After reaching the water the pro-larva rests on 

 its back for two hours, and then casts the skin, a process 

 occupying from three to thirteen minutes. The larval 

 development of Lestes viridis has been compared by M. 

 Giard to that of the crickets. A similar structure of the 

 newly emerged larva has also been noticed in Epithcra 

 bitnarulata, another dragon-fly. 



I may remark that I^estes viridis, though common on 

 the Continent, is an insect of great rarity with us, and 

 not firmly established in the list of British species. 



W. F. KiRRY. 



