]\gt ; The Band Spectrum Associated with Helium .
By J. W. NICHOLSON , M.A. , D.Sc .
, of Mathematics in the University of London .
Communicated by A. Fowler , F.R.S. Received May 4 , 1915 .
A very significant addition to our knowledge of the nature of band spectra .
has been made by Prof , Fowler , has lately described the results of his examination of the band spectrum found in connection with helium and hydrogen , and believed to be a spectrum of helium .
For Halm has maintained that the formulae which must be used to represent line and band spectra are intinlately associated , and , in fact , spectroscopists have been generally inclined to suspect that the laws of line spectra have some counterpart in band spectra also .
Fowler has taken the first step in the elucidation of this connection by showing that the universal constant of Rydberg belongs ; to this individual band spectrum , which contains two series of double " " heads\ldquo ; * arranged essentially in the same manner as the lines in a series spectrum .
One feature , however , of these double " " heads\ldquo ; or doublets appears at first sight to differentiate them from the doublets found in line spectra , and one purpose of this paper is to show that the difference in character is only apparent , and that the formal analogy with line spectra extends very far .
In ordinary Diff.use or Sharp series of doublets , the intervals between the components , when expressed in wave numbers , are constant , whereas in a Principal series the intervals rapidly become smaller , and vanish at the limit of the series .
The intervals decrease , moreover , in a very regular manner .
In the band-doublets discussed by Fowler , although the intervals decrease as the series proceed towards their limits , the decrease is not very regular , as shown by the differences , and the intervals do not obviously vanish at the limits .
Without a very precise ement in series , it is not possible to of their limiting behaviour .
Fowler has shown that the less refrangible components can be arranged in series of a very ordinary simple form , that of Rydberg being almost sufficient .
But the more refrangible components cannot be arranged in a satisfactory manner , even in a Hicks series .
They are , in fact , an example of a phenomenon not unknown in line spectra , where lines which obviously belong to a series cannot be fitted in a satisfactory way into the usual formulae , because those formulae do not , in these individual cases , converge with sufficient rapidity .
* Boy .
Soc. Proc , vol. 91 , p. 208 ( 1915 ) .
'Roy .
Soc. Edin .
Trans vol. 41 ( 1906 ) .
become apparent iaper tpproximate superposition o The Band Spectrum Associated with Helium .
and this series may be most conveniently discussed first .
Its wave numbers , expressed in International units in vacuo , are and the separations are respectively .
Fowler has represented the less refrangible components by the formula , ( 1 ) which only gives errors , and there is no doubt that these components follow the usual laws of series .
The simple formula calculated for the other components is , ( 2 ) with errors , and is not satisfactory .
Fowler has , therefore , added one more constant , and , from the first , second , and fourth lines , obtains , ( 3 ) with errors .
This is more satisfactory , but not completely so .
The doubleC separations become , in the two cases , , and we are led to the question of a possible further reduction of these separations by formulae of accuracy .
In order to test this point , we may notice that in a formula where proceeds in inverse powers by a series which is not very convergent , a more accurate limit will be obtained by calculations which do not include the first line , in which the enceo is most serious , provided , of course , that all the lines are measured with an equal degree of accuracy .
This is the case for the four lines of the present series .
Calculating , therefore , from the second , third , and fourth lines and using the betber form of the series will serve almost equally well .
Adopting this form , the calculated limit becomeo This already further reduces the limiting separations by , and it no becomes , representing the best value which can be obtained by a formula , only three lines .
The next step in the demonstration makes use of all the four lines for the calculation of a formula with an extra constant , and of the more appropriate form .
The final result of this calculation is where .
( 7 ) The limit falls 10 more units , the divergence of the doublets at the limit being now only about 27 units .
This is not the most accurate formula which can be obtained from the four lines , and it is not yet very convergent for the earlier lines .
It is evident that the next term in will be important in the lines , 3 , and , therefore , that the limit is still incorrect .
Since every improvement in the formula has led to a marked depression in the limit , which now only differs by about 27 from the value for the companion series , there is every reason to believe that the limits are actually identical , and that the correspondence with line series extends further than the occurrence of the series relation .
The most accurate formula which can be obtained from Ghe four lines is calculated as shown below , and involves a further decrease in the limit .
Details of the calculation are given , as it takes the form which appears to be most convenient for general application .
the limit of the series as where is small , and ( 8 ) being a more exact limit for the other components with a simpler law , and being the ydberg constant , we find for the four lines , on calculation , 5 'Roy .
Soc. Proc , vol. 91 , p. 255 ( 1915 ) .
, ( 12 ) whence .
The correction to this value , dependent on , is , if is the corresponding correction for any number , determined by and ultimately .
( 13 ) With the extreme value of which is possible , we can show to be so small that for , the term in does not contribute a significant amount to and the contribution to is small in comparison with that to .
A similar calculation to the above , performed with , 4 , 5 , 6 , would therefore give a result very close to that for , 4 , 5 , only , and the value for combined with that for ( 2 , 3 , 4 , o ) expressed in ( 13 ) will lead to a limit more accurate than even that in ( 7 ) , although still not exact enough .
The calculation with , 4 , 5 , proceeds by writing whence and the ultimate solution by the preceding method is .
( 14 ) the values ( 13 ) and ( 14 ) for the approximation available , .
( 15 ) The limit has again decreased by units , and this decrease therefol.e continues to be systematic as formulae of increasing accuracy , and of the type necessary in line spectra , are employed .
The doublet separation is now only at the limit as against the value which accords with the best Hicks formula given by Fowler .
It is evident , moreover , from a glance at the series below , and a comparison with 7 ) , that it is not yet in any way absolute , and that limit would again fall .
This final series is where ( 16 ) That is certaiIlly much less even than can be proved at once .
For a calculation from the first three lines gives , by this method , .
( 17 ) This cannot be so accurate as ( 14 ) , which again is not so accurate as ( 13 ) .
Let us suppose that ( 13 ) , which has the least error , and uses all four lines simultaneously , is correct .
Then the errors in ( 17 ) and ( 14 ) are mainly derived from neglect of the term in , and are roughly in the ratio .
If , therefore , is the error in ( 14 ) , we may write with a close approximation ( 18 ) and , solving these three equations , This is most probable value of which can be derived from the four doublets , and it is very conclusive .
We can now hardly doubt that the limit is identical with that of the simpler series .
On this supposition , the doublet series is analogous to a Principal series in line spectra .
Without measureents of further members of the more refrangible components , however , no formula can be iven for these members of a much more satisfactory character than ( 7 ) , and the question of the relation of the constants , , to those of known helium series cannot be fated .
The rapid of these constants with the addition of an extra term to the formula precludes arly precise specification of their values at present .
First Series of DouUets .
The other series of doublets isolated by Fowler has for its leading members the way nulnbers given in the Table , where denotes the doublet separations , and , their first and second differences .
The differences appear to be quite irregular , there is a remarkable oscillation in the second differences .
A stndy of the values of indicates a definite .
towards zero , but in an ular manner .
This irregularity is exactly of the type which would be expected if the refrangible components followed a series law of the form where the coefficients , not rapidly to zero , and being alternately positive and ative , whereas in the .
series for the less refrangible components , the convergence of , to zero is rapid .
The preliminary series given by Fowler accord with this result .
For the less refrangible components , he finds a simple .
formula , to.be nearl.y satisfactory , and a corresponding Hicks formula does not show much improvement .
It must be borne in mind , of course , that only the first three doublets are measured with great accuracy , although the next three or four must be fairly accurate .
The later values are admittedly approximate .
A formula of the proper generalised Rydberg type has been calculated from the first four lines of this simple .
series , and the result is The limit is almost exactly that obtained by Fowler from the simpler series , VOL. XCI.\mdash ; A. 2 series remains .
The application of the simple and the Hicks formulae by Fowleg 3$ for this the limits , the more accurate formulQ a limit nearer to that of less refra1lgible series of components .
has also used a Hicks formula with an extra constant , giving a limit , which is cvain noticeably nearer .
But even this formula is by no means satisfactoryas a representation of the lines , and the steady decrease which already baken place in the limit with improvement in the representatiol ) , with the results of our examination of the other set of doublets , eaves little doubt that this series of doublets is also of a Principal type , if : calculation is pushed further .
For even the apparent separation doublets is by no means so large as it was in the first series with which this paper deals .
The complete lvestigation follows so closely along the lines already described for the other doublet series that there is no need to give it in detail here .
One or two of its main features may be mentioned .
If the limit of the series is taken as , the values of in flre The last crures are not very reliable beyond , and even the last three figures may be incorrect in later entries .
Fowler 's Hicks formula was derived from the first , second and seventh lines .
The seventh may not be very accurate , and the first should not be used in the problem of merely determinin the limit accurately , which is our presenLpurpose .
) The Band Spectrum Associated with .
439 3 Applying the Hicks formula to the second , third and fourth lines , as the suitable and accurate trio , in the form , .we find But on the theory of the preceding paper , a better value must be iven by nearly , for is evidently close to unity .
This leads to or very nearly zero .
On the supposition , therefore , that the second , third and fourth hnes are measured accurately , the limit of the series is almost certainly that of the less refrangible components .
Further examination on these lines need not be given , but we may perhaps notice that with we can deduce a formula with three constants from the first three lines , which satisfactory as that calculated by Fowler according to the Hicks model and with a later line .
With the same number of constants , more satisfactory formulae have been obtained throughout when the generalised Rydberg form has been used .
Summary .
1 .
The paper gives further support to Fowler 's conclusion that the heads of the bands in the band spectrum of Goldstein and Curtis follow ordinary line-series laws , by showing that the doublet separations tend to zero at the limits of the series .
2 .
Both the doublet series isolated by Fowler are strictly analogous to Principal series in line spectra .
3 .
The generalised Rydberg formula gives the most suitable representation of these series as well as of line series .

