]\gt ; It will be seen that this group of gases in the periodic Table obey laws similar to those which hold for the inert A New Type of Series in the Band Spectrum Associated with Helium .
By A. FOWLER , F.R.S. , Assistan ProfeF ; sor of Physics , Imperial College , South Kensington .
( Received January 19 , 1915 .
) A previously unknown band spectrum noticed in the course of experiments on bydrogen and helium made at the Imperial College in 1912 , and was further ated and described by W. E. Curtis in the year following An independent account of this spectrum was also br ; iven , almost at the same time , by Dr. E. Goldstein .
In each case , some hesitation was felt in attributing the new spectrum solely to helium , in consequence of the persistence of traces of in the helium tubes employed . .
Curtis found that the band spectrum was best developed , in the wider parts of the tubes , when a with small capacity and a small air-gap was passed helium at a pressure rather than that which is usual in sealed tubes of the gas .
The discharge is then quite brilliant , and by giving long exposures , amounting in some cases to seven or eight hours , 'Phil .
Mag January , 1911 , p. 45 . .
Soc. Proc , vol. 89 , p. 146 ( 1913 ) . .
Deutsch .
Phys. Gesell vol. 15 , p. 402 ( 1913 ) .
Band Spectrum with Curtis obtained an excellent series of photographs of the spectrum with concave grating of 10 feet radius , some of were in the second and orders .
with other instruments were also taken by Curtis , but he has been unable to continue the investigation in conseof his temporary enlistment .
Under these circumstances it has been desirable to make a preliminary analysis of the bands , so that might be drawn to any points of special interest without undue As previously described and illustrated by Curtis and Goldstein , the pectrum includes some conspicuous bands with heads , others with Louble heads , and a number of complex ions in which no heads ecognisable at .
A prominent fluting with a single head , beginning at 5733 , degrades to the violet , but otherwise all the which have been recognised as such are vraded to the red .
The first result of the investigation was the unexpected discovery that the double-headed bands are not arranged according to the usual of band spectra , but closely follow the law of line series .
The structure of the individual bands , vever , is essentially the same as that of bands which are distributed in the more usual way ; that is , the distances between successive lines of a series form an approximately arithmetical progression .
Mr. Cnrtis 's plates show five bands of the main doublet series , and four additional bands in the ultra-violet which plainly belong to it have since been on a small scale .
A fairly conspicuous doublet in the green , 5133 , , was not included in the main series , and it therefore became important to search for additional bands , in order to determine if there were other series which might be related to the first .
Only one other doublet , a faint one at 3634 , 3620- , was at first recognised by inspection , and it remained to be seen if others could be picked out in the more complex regions where bands of different types might be superposed .
For this purpose , all the individual lines composing the bands between and were measured , so that the special istics of the different types of bands could be ascertained , About 1300 lines in all have been measured , but the final determinations of the wavelengths , with the necessary corrections for temperature and other sources of error , will occupy a great deal of time , and it is not considered desirable to give so long a list until the wavo-lengths have been freed from systematic errors .
The provisional wave-lengths , however , are adequate for a first discussion .
As a general statement , it may be remarked that the lines composing single-headed bands show a smaller second difference than those Mr. A. Fowler .
A New Type of in the to the doublets , but in each case the second difference increases in passin from bands in the red to those in the violet part of the spectrum .
many of the band lines can be arranged in series , but comparativel few have been traced or suspected .
Two doublets are related to those about 5133 and 3634 , vever , have been identified also a few -headed bands in addition to those tabulated rtis .
No regularity in the arrangement of the single bands has been recognised , and present paper is restricted to consideration of the doublets .
Structur of Doublets .
structure of the double band , at , is illustrated in , where the intensities of the component lines are represented by the ths of the lines in diagram .
In this case , the more rible , and weaker , component of the doublet includes seven strong lines , and the less ible component at least eight .
In each case the lines to the observed heads .
The " " tail\ldquo ; of the band consists of a numbet of , of which the brightest form the obvious series which is shown in the diagram ; the calculated convergence point this series lies considerably on .
violet side of the two observed heads , and the lines cease to be visible before the convergence point is reached .
eries of the latter type are very numerous throughout the spectrum ; they appear to accompany all the bands , whether or double , and the second diHerence is apparently dependent on that in the band with which it is connected .
The provisional wave-lengths of the lines mapped in fig. 1 are given in Table I , which also shows the corresponding wave-numbers ( reduced to vacuum ) , and the first and second differences .
It will be seen that the second differences are approximately constant .
Band Spectrum with Helium .
TABLE I.\mdash ; Details of the Donblet 4625 , The Scrics of Do The bands belonging to the main series of doublets are readily nised pection of the photographs .
They all occur in regions which are free from complication by the superposition of other bands , and the last seven are apparently the only bands which ] 0CCU in the ultra-violet between 3450 and , beyond which the spectrum has not yet been examined .
The intensities of the bands decrease gradually passing along the series .
The wave-lengths ( International scale ) and wave-numbers ( corrected to vacuum ) of the heads of these bands are given in II ; the first three have been derived from rating plates , but the remainder are only approximate values obtained from taken with slmaller dispersion .
* Calculations soon confirmed the suspicion that the series of the chal.acter hitherto exclusively yarded as belonging to line spectra .
The * The instruments which would ordinarily have been used for oving on the wavelengths are detained in Russia , where they were for ervations of the eclipse of 1914 , August 21 .
VOL. XCI.\mdash ; A. .
A. Fowler .
A Type of in the series may , in fact , be represented in the usual way by the formulae 1iydberg or Hicks .
For the less ible components of the which are the stronger , the following formulae have been calculated:\mdash ; ' The adopted value of the ydberg constant is that given by Curtis for International scale .
derived from his measurements of lines of hydrogen .
The differences between the observed and calculated wave-numbers shown under I and II in Table II , and will be seen that the bands represented with quite an ordinary degree of accuracy .
The wave-number intervals between the two components of the diminish in passing to the ultra-violet , but the convergence is less rapid would be the case if they corresponded with the components of doublets the Principal se1ies of a line spectrum .
Attempts to unite the more refrangible coulponents of the doublets formulae have been less successful .
The simple Rydberg formula residuals , and even the Hicks formula with four constants is not entirely satisfactory .
following formulae have been calculated:\mdash ; - , The observed minus calculated values are shown under III , Table II , from which it will be seen that there are considerable residuals in each case .
A Ritz formula , with three constants , has been calculated ; it gives.residuals slightly larger than those iven formula In view of the observed rate of contraction of the doublets , the iven by formulae III and appear to be too high as compared with linlit 34295 for the series of less ngible heads , which is probably far from correct .
From this point of view , the limit given by formula would seem to be ] correct , relatively to that of the less heads .
OIL the other hand , the position of the band in the .
Soc. , vol. 90 , p. 605 ( 1914 ) .
Band Table II.\mdash ; The bIain Series of Doublets .
to given by is very discordant with that indicated the less refrangible band by I and II ,4711 , respectively ) , and a closer agreement in this respect is shown by III and respectively ) .
The difficulty in a satisfactory formula for the more refrangible components is doubtless associated with the unusual character of the doublet separations , wlrich resemble neither nor Subordinate series in the case of line spectra .
It may be remarked , however , that some of the known line series are not satisfactorily represented by any of the recognised f.ormulae .
There is at all events no escape from the conclusion that the doublet bands in question are arranged substantially in the same way as the lines in an ordinary line serieH .
Second of Doublets .
The first doublet of the second series is quite a conspicuous feature in the preen of the visible spectrum , and the third can be recognised without Mr. A. Fowl A New Type of Series in the difficulty on the large scale photographs .
The heads of the second however , occur among other band lines , and were only identified measurement of the individual lines ; the fourth band is very faint , prolonged exposures would evidently be required to bring out members of this series .
The positions of these bands are indicated Table III .
The less refrangible heads of the four ) served bands are represented by the following Rydberg formula , as will be seen under Table The more refrangible heads , as in the case of the main series , are not well represented by a Bydberg formula , and it is probable that even the Hicks formula would be unsatisfactory if a greater number of bands were available to test its applicability .
The values of nnder and VIlI in Table III refer to the following formulae : Corresponding to , formula gives as the wave-number of the less refrangible component of a possible infra-red member of the series , and formulae and VIII give 3721 and 3382 respectively for the more refrangible head .
Band Spectrum Associated with Helium .
nparison of Two Series .
It is remarkable that , although the two series of doublets follow the law of line spectra individually , no relation between them to any which exists between the different members of a system of line series is cerCainly indicated .
The series which has been described as the " " nain series\ldquo ; of doublets , in consideration of its brightness and extent , would probably correspond with the Principal series in the case of a line spectrun ) if it had any equivalent .
The second series would similarly correspo1ld to one of the subordinate series , and the fact that the first ternl of the occurs with a positive sign may be taken to indicate it would be equivalent to a Diffuse series .
A third series , which would corresl ) to the series in a line spectrum , has not yet been identified .
accordance with vell k ) rinciples , however , an approximate formula for the Sharp tlay be derived that for the Principal , bnt neither of Lwo or additional doublets which'n have been pected o positions so calculated .
In a of line , as expressed by the tster law , the common limit of the 1)iffuse and Sharp series fro1ll of tho set.ies by an anount equal to the wave-number of first Principal line .
In the present case the diHerence between the of the two doublet series is about , while the wave-number of the first lnembe of the main series calculated by the for1nulae I and II is betn 4700 and .
The differe t ce appears to be too great to be accounted for by the character of the formlae enJployed , ind , if so , the doublet series cannot stand in the relation of Principal Diffuse The same conclusion is ested by the fact the less of the doublets are .
in both series .
Thus , although there can be no doubt that the heads of ublet bands are according to the law of line spectra , other relations shown by the different series of a line systenl do not appear to band spectrutl associated with helium , as previonsly described Curtis oldstein , includes with single heads and bands heads .
A preliminary of this spectrum has led ) the foll o conclusions:\mdash ; ( 1 ) doublets do llot ordinary law of } ) ectra , but can be arranged ill two sel.ies of the type hitherto tSHociated with line , and call be ately represented by the ustlal Mr. A. E. Oxley .
fuence of involving the Rydberg constant .
Nine bands of the main series and four of the fainter second series have been identified .
The two series may be likened to the Principal and Diffuse series in the case of line spectra , but the usual relation between such series is not certainly indicated , and no equivalent of the Sharp series has yet been traced .
( 3 ) The doublet separations are not in accordance with those associated with line spectra ; they diminish in passing along the series , but do not vanish at the limit .
No regularity in the arrangement of the single bands has been recognised .
The author is indebted to Mr. F. S. Phillips and to Major-General du Guard Gray , C.B. , for photographs of the new band spectrum supplementing those obtained by Mr. Curtis .
The Influence of Constitution on gnetic ASusceptibility .
Part III.\mdash ; On the Molecular Field in nagnetic S By A. E. OXLEY , M.A. , M.Sc .
, Coutts Trotter Student , Trinity College , , Mackinnoll Student of the Royal Society .
Commnnicated by Prof. J. J. Thomson , O.M. , F.R.S. Received Jmle 24 , 1914 .
) ( Abstract .
) The work is a continuation of in ns , vol. 214 , .
, which co1ltains Parts and II .
The suggestion made at the end of Part II , p. 143 , that the local molecular field in diamagnetic crystalline substances may be comparable with the ferronetic molecular field , has been justified .
Estimates of the order of magnitude of this field have been obtained from the following sources:\mdash ; ( a ) Th. change of spccific susceptibilit , ?
accompanying crystallisation .
The extent of this change may be interpreted , on Langevin 's theory of diamagnetism , as produced by a local molecular field of the order of intensity gauss , which comes into operation on crystallisation S4 ) .
This intense local field distorts the molecules and alters the periods of vibration of the contained electrons .
From the nature of the structure which has been postulated for a netic molecule , this field is of an alternating character , the distance over which it is unidirectional comparable with the

