]\gt ; PROCEEDINGS OF TIEI E ROYAL SOCIETY .
THEJfA TICAL AND PE l'SICAL ScIENCBS .
The A Vncleus Theory applied to the Serics Hydrogen .
By L. SILBERSTEIN , Ph. D. , Lecturer in Mathematical Physics at the University of Bome .
( Communicated by Prof. J. W. Nicholson , F.R.S. Received April 16 , 1920 ) .
The purpose of the present paper is to apply the quantum theory of spectrum emission by atomic systems containing an nucleus , given in my recent paper , * to the Balmer series of hydrogen .
, in working out the general formulae of the said theory , I have had in view chiefly the more complicated , non-hydrogenic , spectra as a possible field of its application , yet it has seemed worth while to compare also the aspherical nucleus formulae with the Balmer spectrum of , the nlorc so , as recent measurements have revealed a notable deviation of at least the first six members of this series from the simple Balmer formula .
The measurements alluded to were made by W. E. Curtis in 1914 .
and their newly revised results are tabulated in his paper of for a copy of which accompanied by helpful explanations , I am indebted to Prof. Fowler .
1 .
It will be sufficiently general for the purpose in hand to assume an axially symmetrical nucleus of unknown asphericity ( to be determined from the observations ) .
Then the series ill be given by the formulae ( 28 ) , with nucleus charge , with written for the constant term , and , 4 , 5 , etc. , for the members , etc. , of ths series .
Thus , if be the Bohr value of the Rydberg constant , and if retaining all other symbols of my quoted paper ) we write for brevity , ( 1 ) * Spectra of Atomic Systems containing a Complex ( i.e. , Aspherical ) Nucleus Phil. vol. 39 , pp. 46-66 ( 1920 ) .
E. Curtis , ' Roy .
Soc. Proc , vol. 96 , pp. 147-155 .
, p. 62 . .
L. Silberstein .
The Aspherical Nucleus the frequency formula will be , ( 2 ) where for the initial , and for the final orbit , are as on p. 55 loc. .
It will be kept in mind that , therefore , , may be either positive or negative , according as the nucleus is " " oblate\ldquo ; or " " prolate\ldquo ; in the generalised sense of these words , as explained in the quoted paper .
The variable number is the sum of the three independent introduced through the quantum integrals , and , similarly , the quantised eccentricity and inclination of the electronic orbits , appearing in , being given by , loc. cit. 2 .
According to ( 2 ) , every " " member\ldquo ; or group of the series , such as , will consist of many single lines or components .
Let us call components those lines in each group which correspond to the passage of the electron from a circular equatorial to a circular equatorial ( final ) orbit , i.e. , for and .
This corresponds to and .
Thus , and the series of the main components will have ( by 2 ) the frequenoies where , 4 , etc. , for , etc. In the absence of better knowledge , and in a first attempt , let us correlate these frequencies of the main components with those measured by Curtis , and let us see whether the two constants and can be determined so as to represent his observations by formula ( 3 ) with sufficient accuracy .
At a later opportunity , we may try , if need be , to improve this correlation of heory with observation .
Curtis has measured only the first six members of the Balmer and his vacuum , in , with their probable errors in are .
and the corresponding frequencies are with probable errors which are at once derivable from those of the are the ( electrical ) " " moments of inertia\ldquo ; of the whole nucleus charge , as on P. 61 , .
cit. The dimensions of these constants , , characterising the those o. length .
applied to the Balmer Series of Hydrogen .
Let us determine and by the method of least squares from all observations , notwithstanding that the uncertainty of a coincidence between our " " main\ldquo ; component and Curtis 's observed " " centre\ldquo ; is much greater in the case of the broad group than in the remaining ones .
The resulting Gaussian normal equations are Whence the required two constants become so that the hydrogen nucleus would be oblait With these values of the coefficients formula ( 3 ) gives which differ from Curtis 's -values by the following amounts , The agreement for , is complete , the differences beirg considerably smaller than Curtis 's ( bracketed ) probable errors .
For and the differences are about twice his probable errors ; but even these are not unsatisfactory , of different ( and-13 ) , the group being so broad that two or three thousandths of .
can perhaps be thrown upon his fixing not our main component , but a slightly different place of the group .
In fact , Paschen quotes , for , which is by greater than Curtis 's value , and gives , which agrees completely with our calculated being .
For , Paschen 's wave-length is identical with Curtis 's value , but , , PascheI ] quotes ( which is by smaller than Curtis 's value ) , giving , and thus almost annihilating the above for Under these circumstances , the agreement of ( 3 ) , ( 4 ) with observations will be found better than might be expected , especially as the six differences * If is ( in view of its excessive ) discarded from the least-square calculation , the result is } , differing but unessentially from the above value , and .
Curiously enough , these two coefficients give , with ( 3 ) , for the main component of the excluded group , identical in all eight figures with Curtis 's value , which , of course , is a mere chan result .
But these new values of make the for the remaining five groups considerably greater , and , what matters more , all of the same sign ( to wit .
all negative ) .
' Annalen der Physik , ' vol. ) , p. 935 ( 1916 ) .
Dr. L. Silberstein .
The Aspherical Nucleus are irregularly distributed , and three are positive and three ative , , in , if Curtis 's observed are adopted throughout .
With Paschen 's values for and , the agreement is , for all six groups , almost ideal .
* nately , therefore , we shall retain the values of and given under ( 4 ) .
These values will be used in ( 3 ) for the main components , and in ( 2 ) for all possible of each group of the Balmer series .
S. on 's Formula.\mdash ; Before proceeding with our subject , it may be interesting to see to what extent Curtis 's observations can be represented by Sommerfeld 's relativistic formula , which is based on the usual assumption of a spherical nucleus and a variable mass of the electron .
Sonlmerfeld himselft does not seem to have tested his formula extensively enough in this respect .
It is true that he comparison of his theoretical values of , the difference of the Rydberg constant in passing from to , or to , or , with observation .
But , first of all , the corrections\ldquo ; made for this purpose on the observed ar6 avowedly conjectul.al and by no means binding ; and , secondly , even if be granted , the agreement between his theoretical figures and corrected obse1wed ones , 0.37 1.04 0.06 is certainly not very reassuring .
Under these circumstances , a comparison with Curtis 's Tervations will not be out of place in our present connection .
Sommerfeld 's frequency formula for the " " main i.e. , the circular-circular components is , ( S ) where is Sommerfeld 's " " universal\ldquo ; constant ( not free as our ) , given by so that coefficient of the correction term in ( S ) is essentially positive and , within natrow limits , fixed in its value .
The value adopted by Sommerfeld , based on , is Now , using ( S ) and determining and , again by the method of least squares , Curtis 's six observations , exactly as in the case of our formula ( 3 ) , I find and these coefficients , substituted in ( S ) , give for ' for up to , the values , ( S ) * It may be argued that Mr. Curtis has represented all his six observations still better , to wit , by his formula ( I ) , p. 148 , .
cit. But in this formula , which is of the RydbeI type , the two constants and do not seem to have an immediate physical meaning .
Ann. .
Physik , ' vol. 51 , pp. 1-94 ( 1916 ) . .
cit. , pp. 60-61 .
Theory applied to the of which compare very disadvantageously with our differences corresponding to ( 3 ) , ( 4 ) , namely , .
( 3 , The latter aspilerical ) are considerably smaller and of different signs , whereas Sommerfeld 's formula gives all negative .
Moreover , the required value is much too far from Sommerfeld 's to be admissible .
This , however , is ) no means intended to depreciate the value of Sommerfeld 's beautiful theory , which in the eyes of the present writer is one of the most admirable works in modern physics .
But the chief experimental triumphs of Sommerfeld 's theory seem to lie in the fine-structure of " " lines\ldquo ; or groups ) , more especially those of Fowler 's principal series of ( ionised ) helium .
Notwithstanding these notable successes in the case of the helium lines , we must keep an open mind in judging Sommerfeld 's theory with regard to our present or to any other case .
4 .
By what has been said in Section 2 , we have good reasons for attributing to the atom of hydroge1l , as far as its Balmer series is concerned , the Rydberg constant given under ( 4 ) , and an oblate nucleus , of which the amount of asphericity is given by It may be interesting to estimate , according to this form ) , the difference A-B of the principal " " moments of inertia ' ' of the nucleus .
Putting , in round figures , I hnd cm.2 .
To form an idea of the meaning of this result , suppose the nucleus to be a homogeneous rotational ellipsoid of axes and , the former of these the axis of symmetry .
Then ( cf. , loc. cit. p. 63 ) , and , so the nucleus is oblate in the usual sense of the word .
If be the eccentricity of the generating ellipse , we have , and , therefore , may be interesting to mention in this connection that a much smaller deviation from this " " universal ' ( value ( to wit , as against has sufficed to K. Glitscher , a pupil of Prof. Sommerfeld , to condemn categorically Abraham 's rigid electron in favour of Lorentz 's deformable relabivistic ) one .
Glitscher.s paper in 'Ann . .
Physik , ' vo ] .
62 , p. 608 ( 1917 ) , or the account of it given in my ' Report on the Quantum Theory of Spectra , ' London , Hilger , 1920 .
The " " objections\ldquo ; against Sommerfeld 's theory , hinted at in , footnote , p. 47 of my 'Phil .
Mag. ' paper ( cited before ) , were found to be only apparent difficulties concerning the " " canonical ' nature of Sonlmerfeld 's variables .
This point has , in the meantime , been completely explained by Sommerfeld in a private communication to the author .
Dr. L. Silberstein .
The Nucleus , ( A-B ) , i.e. , for the product of the eccentricity into the major semi-axis of the nucleus , This is , comparatively speaking , surprisingly .
For it would mean , for the equ atorial semi-axis of the nucleus , at any rate , cm .
, the limit being reached for a nucleus flattened down to a disc , and , even then , its lateral dimensions would be scarcely as much as 100 times smaller than those of the whole hydrogen atom , or better , of the smallest stationary electronic orbit .
Yet , the unexpectedly large value of A-B or of does not seem to constitute a serious objection against the proposed theory .
If , as is assumed by Rutherford , the mass of the nucleus is wholly electromagnetic , and its shape is spherical or almost so , then , of course , its dimensions would be of the order of cm .
But there is certainly no evidence whatever for the purely electric nature of the nucleus ( nor for its ical form ) .
And , althougI ] Rutherford 's experiments are interpreted as to nucleus dimensions of the order of cm .
, or even cm .
, yet the elementary theory of Butherford 's experiments is in many instances in direct opposition to observation , that the low-dimension often quoted after Sir Ernest Rutherford do not seem to be definitely established .
5 .
Fine-Structure of \mdash ; For the first member of the Balmer series , the group , we have to substitute in equation ( 2 ) so that the equation for whole group will be , ( 5 ) where , as for every group of the series , .
For the lllain component of we have .
( 5 ) Lt us count the frequency distance of any component of the group the main similarly , their length distances Then , subtracting ( 5 ) from ( 5 ) , we shall have , for any component of the group .
( 6 ) * See , for instance , ' Phil. bfag , ' vol. 37 , p. 563 , table ( 1919 ) .
Theory applied to the Balm Series of The fine-structure of the whole group is thus seen to depend on only through their product , which , by ( 4 ) is , to three significant figures , cm . .
( 7 ) Since it is for the present futile to think of determining the distances of the components to within 1 per cent. , we may as well take cm .
It is scarcely necessary to say that our fine-structure formula ( 6 ) , and , similarly , those for , etc. , are entirely different from Sommerfeld 's formulae .
Passing to the evaluation of the numerous arithmetically possible components determined by ( 6 ) , it may be well to adopt a short symbol for each of them .
As such we will take which will stand for the component due to the passage from the orbit to .
Thus the main component of the group will be .
Similarly we shall write , and so on .
The sum of the upper numbers will be 3 , and that of the lower numbers , 2 : Notwithstanding the smallness of the totals in the case of , there is still quite a multitude of components arithmetically possible .
Our first task must be to reduce their number , avoiding , however , as much as possible , any arbitrary assumptions .
In the first place , since the ular momentum corresponding to any stationary orbit and , since rectilinear orbits are not in question , we shall have to consider only such combinations of the two triads of numbers , for which .
( I ) In the second place , it will be remembered that the full expression for the function appearing in our formula is and , since the perihelion longitude is not quantisable , all components for which both the inclination and the eccentricity of the initial or the final orbit differ from zero , will be more or less broad bands , according to the amount of inclination and eccentricity .
Now , even if such components or , p. 54 .
Dr. L. Silberstein .
The Nucleus bands were equally probable as the remaining ones , i.e. , if the same numbers of individual atoms were engaged in producing them , yet , their light contributions being spaced all over the broad interval , corresponding to to , these bands will be much fainter than the sharp lines , in most .
, in fact , producing only a faint , hazy round to the whole group of sharp iines .
At any rate , these broad components would be scarcely distinguishable , the more so as many of them overlap , as is shown by detailed calculations , which here may be omitted .
We shall , therefore , retain only the sharp components , i.e. , such only for which either or ( or both ) and at the same time either or ( or both ) vanish .
* We can express this restricting condition shortly by requiring This will greatly reduce the number of possible arithmetical combinations .
At the same time the iunction will be simplified down to , where , by ( II ) , either or ( or both ) vanish .
Remembering that we shall have , ultimately , valid for the sharp components of any member of the series .
For the group we have to put here , and similarly ( for ) We may subsequently consider a " " selective principle such as that given by Rubinowicz .
But , in the mtime , let us consider the still numerous components left by the application of the conditions ( I ) and ( II ) alone , in short , all sharp components of the group These , classified according to the final orbits ( in rows , and initials in columns ) * That is to say , orbits either equatorial or circular ( or both ) .
Theory applied to the Balmer Ser.ies of Hydrogen .
The first is the main .
In all we have 24 components , which may be considered as four " " sextets * The first , or the main , sextet can be shortly denoted by 2.0.0 ; similarly , the second by 1.1.0 , and so on .
ance at equation ( 6 ) will suffice to show that all these sextets are obtainable from one another by a rigid shift along the scale of freqnencies Thus it is enough to write down one of the sextets , say the first , and to indicate the rigid shifts which carry it as a whole to the position of the second , the third , and the fourth sextets .
According to ( 6 ) and ( 8 ) , and since , more enerally , the first sextet will be given by and the rigid shifts transforming this into the 2nd , 3rd , and 4th sextet will be , respectively , 1 .
1 .
, ( 1.0 .
1 ) Thus , by ( 8 ) and ( 2.0.0 ) , using our above value of , passing to , and ordering the components after their increasing wave-length , .
we have , in , for the first sextet 0 .
3 .
1 .
2 .
2.1.0 .
3 .
1 .
and for the shifts carrying this sextet rigidly into the remaining three sextets , according to the last quoted values of ; ; Let us discar for the present , the very eccentric , and therefore distant , component of ( 2.0.0 ) , and the corresponding components , ( 1.1 .
as well as the whole distant sextet ( 1.0.1 ) , distinguished by square brackets .
Moreover , in view of the actual limits of spectroscopic precision , let us retain but the third decimals of .
Then the whole material can be written as follows ( regardless of the provisional grouping into sextets which are now partly intermingled ) : \mdash ; * If the reader prefers to classify them according to the initial orbits , he can consider the whole group , with equal right , as consisting of six quartets .
At the same time we have six quartets , 3.0.0 , 2.1.0 , etc. , following from one another by other rigid shifts .
Dr. L. Silberstein .
The Nucleus Fine-Structure of These are in components , of which those joined by brackets are so close together as to be scarcely separable from one another by available spectroscopic means .
As such they are drawn on the accompanying figure .
We ma } now consider A. Bubinowicz 's selective principle Auswahlprinzip\ldquo ; mentioned before .
This requires\ldquo ; the azimuthal quantum number of an atom to jump at the utmost by a unit that number is in our ( three'Physik .
' vol. 19 , pp. 441 , 465 .
See also Chap. of Sommerfeld 's book , just published , ' Atombau und allinien , ' Brunswick , 1919 .
pplied to the Balmer Series of Hydrogen .
dimensional ) case , the factor of in the value of the ular momentum , so that Rubinowicz 's selective principle means or or .
( Rub .
) Now , although we have made no use of this well-founded principle in .drawing up our list , it so happens that all the 15 components satisfy it , as a glance at the first column of the list show.* The more advisable does it seem to retain , for the present at least , all these 15 sharp components of the roup .
We may have to discard some of them on the evidence of further ltore precise urements , which , as I learn from Dr. Merton , are now contemplated by him jointly with Prof. Nicholson .
For the present , little more can be said than that the above fine-structure is not contradicted by the available observations on .
In fact , what up to the present is known of this group is that it is separable into two conspicuous " " components apart , t the stronger component being the more red one .
Now , in a first approximation ( and especially in view of the light background due to the discarded broad components or bands ) our first four theoretical components , ether with the main which they so closely accompany , can be interpreted as the less red , and all the remaining , fused together , as the second , more red ' component The ' centre\ldquo ; of the former can be taken to lie at , and that of the latter somewhere between and , and this would give , , the observed " " separation of the two components correct at least as to its order of magnitude .
I understand from Dr. Merton 's private communications that is very likely to have a much more complicated structure than that usually described by a " " doublet a third component , in fact , appearing , under certain circumstances , between the two others .
Also a suggestion made by Prof. Fowler would make it likely for to be a typical triplet , this would answer better our configuration of components , in which there are two comparatively large gaps , one between and , and another between and But a final verdict must be postponed until more precise experimental results are available .
We will leave , therefore , for the present any further discussion of the fine-structure of the group .
Bohr 's sharpened form of the selective principle excludes the zero , and thus would exclude our last three components .
But , although certain cases speak for Bohr 's principle , there are certainly no reasons for universalizing it .
Buisson and Fabry , recently confirmed by Merton and icholson ' Phil. Trans vol 217 , , who find from to with a mean of .
Merton and Nicbolson , loc. , p. 262 .
Aspherical Theory applied to Balmer 6 .
Fi of \mdash ; The equation of this group is , analogously with .
The final orbits will be as before .
The initia ones will , of course , be more numerous .
, again , all the components , which satisfy ( I ) and ( II ) , according to the final orbits , we shall now have the octet 2.0.0 , and three more octets , 1.1.0 , 1.0.1 , 0.2.0 , obtainable from this by the same rigid shifts as in the case of .
The whole group would thus consist of :32 sharp components .
Discarding , , the very eccentric , and therefore very distant components , and proceeding in much the same way as before , we should still have as many as 21 components , whose range from to A. Of these , however , unlike the previous case , the greater part infringes against Bubinowicr. 's selective principle .
Discarding all these , to which belongs also the main itself , the components and 12 others , we are left with only six sharp components satisfying Bubinowicz 's , to wit , A. According to Bohr 's rule , would still have to discard the two underlined components .
But , in view of the incomplete experimental knowledge of thestructure of this group , it will be best to leave the matter here for the present .
For the same reason it will be better to postpone also a desirable and possible retouching of the coefficients ( 4 ) , and especially of , in view.of the manifestly acentral position of the main component within the theoretical group as well as , and presumably , etc. Such a redetermination of might still nprove the already good agreement of ( 3 ) , ( 4 ) with Curtis 's measurements .
* Which ' ' \ldquo ; all combinations with or .
Cf .
Sommerfeld:s. .
' -Atombau , ' etc. , p. 451 .

