]\gt ; On the Structure of the Series of Hydrogen Lines .
307 ( 10 ) , ( 11 ) , Shakespeare , unpublished .
( 13 ) , Shakespeare , Daynes and Lambourn , " " A Brief Account of some Experiments on the Permeability of Balloon Fabrics to Air 'Advisory Committee for Aeronautics .
' ( 14 ) Dewar , ' Proc. Roy .
Inst March , 1918 .
18 ) Carslaw , ' Fourier 's Series and Integrals , ' p. 263 ( 1906 ) .
( 19 ) Shakespeare , ' A New Permeability Tester for Balloon Fabrics .
' Advisory Committee of Aeronautics .
20 ) Shakespeare and ynes , " " Further Notes on the Hydrogen Permeameter ' Reports of Advisory Committee for Aeronautics .
' ( 21 ) , ( 22 ) Dewar , see ( 13 ) .
On the Structure of the Balmer Series of Hydrogen Lines .
By T. B. MERTON , M.A. , D.Sc .
( Communicated by Prof. A. Fowler , F.R.S. Received March 23 , 1920 .
) Introdnciory .
It is well known that the spectrum of hydrogen comprises the series of .lines generally known as the Balmer series , which is a prominent feature in the spectra of the stars and nebulae , and the secondary spectrum , consisting of a vast number of lines whose presence in celestial spectra has not yet been established .
The co-ordination of the secondary spectrum would appear present great difficulties on account of its complexity , but it is perhaps remarkable that a precise knowledge of the Balmer series , which was at one time considered to be the most simple series known in spectroscopy , should still be wanting .
Balmer showed , at a time when the ] exity of the lines was not yet known , that the wave-lengths could be represented with considerable accuracy by the formula , where takes successive integral values .
Balmer 's formula is a special case of the formula of Rydberg for the wave-number , where is the wave-number , N Rydberg 's " " universal\ldquo ; constant , and and are fractions appropriate to the series ; this is identical with Balmer 's formula when and are put equal to zero .
Curtis* bas deduced from a series of accurate measurements of the leading lines of the .Balmer series that the value of Rydberg 's constant is equal to , and that the wave-numbers of the lines cannot be represented to the degree of accuracy attained in his 'Roy .
Soc. Proc vo ] .
90 , p. 605 ( 1914 ) ; vol. 96 , .
] ( 1919 ) .
Dr. T. R. Merton .
On .
the Structure of the measurements by the Balmer formula .
Curtis was able to represent his experimental values accurately by the formula where , and takes the value These results depend on the measurement of the " " optical centre of gravity\ldquo ; of the lines , for it has been shown by Michelson*and in numerous later investigations that the lines are complex .
Michelson deduced from the visibility curves obtained with the interferometer that the lines were close doublets , the separation of the components being 0 A. in the case of the line , and about half that value in the case of .
In the case of .
the intensities of the components were estimated to be in the ratio 10 : 7 , the stronger component being on the less refrangible side .
The widths of the components , particularly in the case of , are of a nitude comparable with their separation , and there is no doubt that this fact is partly responsible for the discordant values obtained by different illvestigators .
Thus Houstoun gives for A. , Buisson and , and Paschen and BackS .
The line appears at best as a very diffuse and difficult doublet , and I am not aware that the doubling of has ever been seen , its structure having been deduced from the ance of the fringes seen with the interferometer .
In a recent investigation the separations of the components of and have been determined from quantitative photometric measurements of the intensity distribution given by the overlapping components , the values found being A. for , and A. for , the former value being in exact agreement with the value iven by Buisson and Fabry .
These values imply that the Balmer series is to be looked on a principal series , in which the wave-number separations of the components converge to the limit of the series , rather than as a diffuse or sharp series , in which a constant wave- number difference between the components is preserved throughout the series .
It must , however , be remembered that difl'erent lines in the spectrum are subject to very great variations in their relative intensities under different physical and electrical onditions of excitation , and that under certain circumstances the structure of the lines may be subject to variations .
In this case a number of suggestions which have been made as to the complete representation of the series would not be excluded .
'Phil .
Mag ( 5 ) , vol. 34 , p. 282 ( 1892 ) .
'Phil .
Mag ( 6 ) , vol. 7 , p. 456 ( 1904 ) .
'Journ .
de Phys June , 1912 .
S 'Ann .
der Phys vol. 39 , p. 897 ( 1912 ) .
Merton and Nicholson , ' Phil. Trans , vol. 217 , p. 237 ( 1917 ) .
Balmer Series of Hydrogen Lines .
It has been pointed out*that if a single series of lines were represented by the formula used by Curtis , another series of lines might be looked for on the basis of the combination principle which would be represented by and there might also be a series The lines of the first and second of these series would exhibit a constant difference in wave-number of , corresponding to wave-length separations in and of A. and A. respectively , whilst wave-number differences between the first and third of these series would not be constant , and would give rise to wave-length differences of A. for and A. for .
These considerations involve no new assumptions and are based on the well established combination principle , but Sommerfeld has carried out an investigation of the origin of the Balmer series on the basis of the quantum theory , which demands an even more complex structure for the line .
Sommerfeld 's is an extension of Bohr 's theory of the hydrogen spectrum , the hydrogen atom being supposed to consist of an electron in circular or elliptical orbital motion about a positively charged nucleus , the steady states being defined by discrete values of the angular momentum and , in the case of elliptical motion , the eccentricities of the ellipses being limited certain definite values .
The precise value of the separation of the components , or principal groups of components , of the line is an important constant in the numerical results deduced by Sommerfeld .
Although his conclusions do not appear to be in agreement with the experimental results of a previous investigation , it will be shown in a later section that the phenomena are more complex than has been } ) osed herto , and that it is probable that the structure of the lines varies under different conditions of excitation .
Width of the Hydrogen Lines .
The sharpness of all spectrum lines is affected by the broadening due to the motion of the radiating particles in the line of sight , operating in accordance with Doppler 's principle .
Under certain conditions , notably when the spectrum is excited by powerful condensed electric discharges , the lines * Merton and Nicholson , 'Sitz .
der K. Bayr .
Akad Munchen , 1916 .
Merton and Nicholson , cit. Dr. T. R. Merton .
On the Structure of the may be affected by an entirely different cause of broadening , but the motion of the atoms in the line of sight sets a very definite inferior limit to the widths of the lines .
The theory of this effect has been given very completely by Lord Rayleigh , has shown that the distribution of intensity in the widened line is represented by the well known.probability law and that the width of the line may be conveniently expressed by defining the " " half-width\ldquo ; as the value of when , I being equal to unity at the maximum of intensity .
According to this definition Lord Rayleigh shows that where is the absolute temperature of the gas and the mass of the radiating particles in terms of that of the hydro , atom .
In their investigations with the interferometer Buisson and Fabry toc .
have shown that in the case of gases of the helium group vacuutl tubes at low pressures , and excited by uncondensed discharges , the widths of the lines are entirely , or almost entirely accounted for by the motions of the radiating particles , and that the temperature is that of the walls of the tube or not much higher .
Putting C. the limiting half-widths of the hydrogen components for are found to be , and putting the separation of the components at , and their relative intensities in the ratio 10 : 7 the intensity of the minimum between the components should be about 31 per cent. of the mean of the intensities of the two components .
It is evident that the line should be seen as a fairly sharp doublet , which is certainly not the case under ordinary conditions .
On the basis taken , above , the limiting half-widths of the components of should be somewhat less than , so that it should not be difficult to see the line as a doubtlet if the distance between the components were .
or more .
The appearance of at once shows that these limits are not attained .
Either there is in this case some cause of broadening which does not affect the lines of the rare gases or some other explanation must be sought .
The former view seems unlikely , for the only cause of broadening to be suspected would be of the same type as that which occurs to a greater extent when ths gas is excited to luminosity by condensed discharges , and which is now enerally recognised to be due to the influence of the electric field of neighbouring charged particles on the atoms , operating in accordance with the Stark effect .
It is true that this effect is reater for hydrogen than for other elements , but the effect in the case of helium is comparable in magnitude and the excellent reement between the experi ' Phil. Mag vol. 29 , p. 274 ( 1916 ) .
Balmer Series of Hydrogen Lines .
mentally measured widths of helium lines and the widths calculated on the basis of the Doppler effect in the experiments of Buisson and Fabry ( loc. cit would seem to leave little room for broadening of this type .
The Echelorating .
A large number of observations of the and were made with a Lummer Gehrcke plate , and results of some value were obtained .
It was found , however , that obselvations could be more easily made with an e'chelon diffraction grating , and the stlucture of the lines has been investigated with an instrument of this type , having 35 plates mm. in thickness , the step of the grating being mm. The instrument was made by Messrs. Adam Hilger , and was mounted in the usual ma1mer .
This instrument possesses certain advantages over the Lummer Gehrcke plate in the investigation of broad lines ; with the latter instrument the ratio of the power to the distance between successive orders is extremely small , a fault from which the echelon suffers to a much smaller though still somewhat ob.ectionable degree .
Preliminary analysis of the light from the vacuum tubes was effected by means of a deviation spectroscope , the slit of the e'chelon spectroscope being situated at the focus of the telescope lens of the constant deviation instrument .
These instruments , after the first adjustment , gave no further trouble .
Measurements could be made with a micrometer eyepiece , or by substituting a camera for the eyepiece , the spectrum could be photographed .
The theory of the e'chelon has been given by Michelson , in its application to the problems under discussion certain considerations which do not usually affect the use of the instrument must be taken into account , and may , in some cases , be used to obtain information which is not usually afforded by measurements of this kind .
In the separation of the lines of a doublet it is generally found convenient to calculate the difference in wave-length corresponding to the separation of successive orders ; the distance between successive orders and between the components of the doublet being measured with a micrometer , the wave-length separation of the doublet can at once be found .
The difference in wave-length corre- spondin to the separation of successive orders is by the formula , where is the wave-length of the light , the thickness of the plates , and the refractive index of the glass for wave- length Accurate measurements of the refractive index of the glass for the lines * Astrophys .
Journ vol. 8 , p. 36 ( 1898 ) .
Dr. T. R. Merton .
On the Structure of the , and were furnished by the , and it was found that the relation of refractive index to the wave-length could be accurately represented by the formula , due to Hartmann , , where and are constants , being thus equal to .
The above method of determining the difference in wave-length of the components of a doublet is , however , only strictly applicable in the case of lines of ficant width , a condition by no means satisfied by the hydrogen components .
This arises in the following manner : The distribution of intensity by the e'chelon is given by the formula where denotes the step of the grating and the angle of diffraction .
Thus when , and when .
If the maximum of a particular order falls at the two neighbouring orders are absent since the angular separation of successive orders is equal to .
Putting the formula for the intensity distribution becomes rotation of the echelon a line may be set in the single order position , when , and only one order is visible , or in the double order position when two orders of equal intensity ( with Saint orders on either side ) can be seen at , or in any desired intermediate position .
It is now evident that a line which is not in the single order position will suffer an apparent displacement , in so far as the maximum of intensity is concerned , towards the point by an amount depending on the width of the line .
Assuming that the width of the line is due to the Doppler effect , the intensity distribution in the line is given by , where may , for conyenience , be expressed in circular measure .
The apparent maximum cf aline does not occur at the point defined by but at the point at which , or This gives the solution for the displacement ; but 1for practical purposes the following application is important .
It has been found that the jhelon can be set in the double order position with considerable precision .
It can be easily shown that a small error in the setting is unimportant .
Let the distance between the maxima of the two orders in the double order position be P. Then the displacement of each order in the direction of is given by .
Then , where It has been found that this equation can readily be solved , with an accuracy transcending the requirements of practical measurement , by the following simple interpolation formula : Let be the half-width of the line in radians , i.e. , the value of when .
Then .
Since and is measured on the micrometer , it remains only to Balmer Series of Hydrogen Lines .
calibrate the micrometer in radians .
This can be readily accomplished by measurements of the separation of successive ders of such lines as the satellites of the green mercury line , A. , whose width is so small that the value of is insignificant .
Since the micrometer measures linear distances at the focus of the telescope objective , a small correction may be necessary if the difference in focal length of the telescope lens for the green line of mercury and the line umder consideration is appreciable .
This be determined with precision by measurements of the lengths of the lines with the micrometer , the length of the lines } limited by the wedge used for varying the effective length of the dehelon slit .
Let be the length of the mercury line , A. , on the plate in terms of the micrometer readings , and the length of a line of wave-length , and let be the difference between the micrometer readings of successive orders of a mel'cury satellite and , the corresponding distance for .
Then .
It thus appears ) the corrections to be applied in the case of broad lines may be turned to adyantage , since they give an estimate of the physical widths of the lines .
It is also possible to measure the relative intensities of the components of a doublet with some precision .
The procedure in the case of a doublet and is as follows : The value of , and thence the value of is determined measurements of the separation of the orders in the double order position for .
The separation of the components is then measured , and from the value of the displace1nents of the maxima the true centres of the liIles can be calculated .
many cases the rences between the displacements is insignificant , and the separation of the components in wave-lengths can at once be found when the true separation of successive orders in terms of the micronleter divisions has been determined from measurements of the separation of the orders for a line of negligible width .
For the measurement of the relative intensities , the is set to the single order position for , and the crosswires of the micrometer eye-piece are set on to the maximum of the line .
The echelon is then rotated until the maxima of and appear to be of the same intensity .
Let their positions , reduced to radians , be and .
The relative intensities and are then given by .
Strictly speaking , the displacements and should be taken into account , the true centres of the lines being situated at and , the relative intensities being given by the formula Dr. T. R. Merton .
On the Structure of the but the accuracy attained in photometric measurements does not generally call for the use of the more r expression .
Experimental Observntions .
In the investimation previously referred t which related to the structure of the lines and , the vacuum tubes which were used were filled with at a low pressure , without any very special precautions .
I have since iced on numerous occasions that the behaviour of the line , as seen under a high has been very capricious .
Sometimes the doubling was fairly distinct , whilst at other times , under conditions of electrical excitation which did not appear to be dissimilar , it seemed as if all definite structure had vanished , and only a broad unsymmetrical line could be seen .
Attempts to control these changes by varying the pressure in the discharge tube and by altering the conditions of excitation were atory , but there remained the possibility that the changes which occurred might be referred to the influence of other gases which were present in the discharge tube .
It has been shown in previous that tlJe presence of impurities , notably helium , may give rise to alterations in the distribution of intensity in the spectrum , and that , under certain conditions , the seconda1y spectrum of hydrogen may be modified to a profound extent .
Further investigation in this direction at once showed that the structure of the Balmer lines is affected by the presence of impurities , and , since it has been found possible to observe the lines and distinct doublets , which , by a change in the conditions , are transformed into broad unsymmetrical lines , in which no definite structure can be detected , I have attempted in the first place to descI'ibe qualitatively the character of these changes , and in the second to isolate the conditions for obtaining the sharpest resolution of the lines , and to measure with the greatest possible accuracy the separation of these components under these conditions .
Qualitative .
Observations have been made at the temperature of the laboratory , and also with the capillaries of the vacuum tubes cooled to the temperature of liquid air .
With tubes of the form , the latter condition was attained by making a trough of indiarubber round the capillary of the tube , and keeping this trough filled with liquid air during the experiment .
That this procedure and Nicholson , .
cit. Merton and Nicholson , 'Phil .
Trans , vol. 220 , p. 137 ( 1919 ) ; 'Roy .
Soc. Proc , vo ] .
96 , p. 112 ( 1919 ) ; Merton , ' Roy .
Soc. Proc , vol. 96 , p. 382 ( 1920 ) .
Balmer Serles of Ilydroyen was effective will be seen } the results of quantitative measurement .
Vacuum tubes containing very pure hydrogen were prepared by heating in a flame of a palladium tube , which was connected with the previously exhausted tubes ; by repeatedly admitting hydrogen in this way , and it again by pumping , the impurities could be shed out , and it is believed that the tubes filled in this way were very free from contamination .
It is known that , in very pure , the intensity of the lines of the Balmer series is greatly reduced relatively to those of the secondary spectrum , and , in the tubes prepared in the manner described , the effect is very striking .
With these tubes the line appears as a fairly broad unsymmetrical line , showing no evidence of doubling , and the appearance of is very similar .
At the temperature of liquid air , there is , perhaps , just an indication of structure in has already been pointed out that the " " half-width\ldquo ; of the components of at room temperature should be A. if the broadening is entirely due to the motions of the radiating particles .
At the temperature of liquid air , the ' half-width\ldquo ; should be reduced to A. corresponding values for are .
and A. respectively .
With a small trace of impurity , such as water vapour , or in tubes showing the Angstrom cafbon bands , traces of structure in are apparent at ordinary temperatures , though still shows no sign of resolution .
In tubes filled with water vapour , appears as a fairly distinct doublet .
Cooling with liquid air , as , indeed , might be -expected , is ineffective in improving the definition , the necessary impurities being , in fact , frozen out by this procedure .
It is important to note that the total width of the doublet seemed to be greater than the width of the unresolved line seen in pure hydrogen .
In tubes containing a mixture of hydrogen and helium at a few millimetres pressure the resolution of is comI ) let , while still shows no definite structure .
With these tubes , however , the advantages of cooling to the temperature of liquid air are fully , for the helium , unlike cal.bon compounds and water vapour , is not removed from the mixture , and under these conditions appears as a sharp doublet , perfectly resolve , and almost reminiscent of the appearance of the 1 ) lines in a diluted sodium flame when seen under a lower dispersion .
appears as a perfectly definite doublet but the components are not completely resolved and are hazy in appearance .
The proportions of helium and hydrogen necessary for exhibiting these phenomena are not closely restricted , and no definite changes can be observed over a wide range of pressure so long as the helium is present in excess .
It is perhaps remarkable that the doubling of at ordinary temperatures is Dr. T. R. Merton .
On the Structure of the quite clean with tubes filled at pressures as rreat as 40 mm. of mercury , proportion of helium to in these tubes being estimated to be in the ratio 80 : 1 .
The qualitative observations appear to be in accord with the results of the previous investigation , * but they show that these results refer to vacuum tubes containing hydrogen with a small trace of imptl.ity ; with the tubes containing a mixture of hydrogen and helium it is at once evident that the structure of both and , and not merely the width of the has a striking .
The appearance of is in fact the most criterion for the purity of the hydrogen , the colour of the discharge becomes almost white at a in the purification at which spectrum lines due to impurities appear to be entirely absent .
A small trace of impurity can be renloved by passing a heavy current through the tube so that a deposit of aluminium from the electrodes is formed on the walls of the tube .
This layer of sputtered aluminium , as is well known , removes traces of carbon compounds and other impurities , and it is thus possible to watch the change in the line the process of purification .
cmtitative .
It has already been stated that the conditions of sharpest resolution are attained by cooling a mixture of helium and at a few millimetres pressure to the temperature of liquid air , and it has been under these conditions that the quantitative measurements here recorded have been made .
An enlarged reproduction of showing and is given in fig. 1 .
The enlal.gement , showing one order only , was made with mm. mm. FIG. 1 .
* llerton and Nicholson , .
cit. Series of Lanes .
the aid of a cylindrical lens in the optical train to eliminate the rain of the plate .
Measurements were made with a Hilger photo-measuring micrometer of three plates of and four of , and the results given for each plate were taken from the mean of six sets of micrometer dings .
A plate was also taken of the reen line for purposes of calibration .
The method of reducing the measurements has been iven in a previous section , and the numerical results , using the same notation , are as follows:\mdash ; Mercury Separation of orders in micrometer divisions Length of lines in micrometer Qions Separation of components .
Separation of orders .
Length of lines .
Mean 890 4478 If the components were of vanishingly small width the separation of the should be radian , radian .
arent maximum of 1st component is at Apparent maximum of 2nd component is at From the value the value of the constant in the intensity distribution formula I is found to be and from the reiation True maximum of 1st component is at ) True maximum of 2nd component is at Difference radian .
Since , the separation of the components is equal to It so happened that , in the double order position , the intensities of the two components appe.ared to be identical , and the relative intensities could Dr. T. R. Merton .
On the Structure of the be calculated without further measurement , To a relevant degree of accuracy , the relation of the intensities is thus given by The intensities are thus in the ratio Separation of components .
Separation of orders .
Length of lines .
Mean 618 4560 radian , radian .
Apparent maximum of 1st component is at Apparent maximum of 2nd component is at True maximum of 1st component is at True maximum of 2nd component is at Difference radian .
Since , the separation of the components is equal to Whilst the sharpness of the components of made accurate micrometer settings an easy matter , those of were very difficult to measure ; this is shown by the inferior reement in the case of of the results for different plates , and a measurement of the ] ative intensities of the components has not been attempted .
Discussion .
The results obtained in the section may be summarised as follows:\mdash ; .
Separation of components Half-width of components Intensities of the components are in the ratio 10 : .
Separation of components Half-width of components Series of Hydrogen Lines .
In the case of the two components are very sharply defined , and any additional components must be either of vanishingly small intensity or at an exceedingly small distance from the main components under the conditions of the experiment , for the value found for the half-width , , is very close to the theoretical value A. On the other hand , the half- width of the components of , though of doubtful accuracy , is much in excess of the theoretical value , a result which rrht well be accounted for by the disturbing influence of one or more weak additional components .
If the series , umder the conditions of experiment , consisted of doublets of constant wave-number difference , a difference in wave-length of A. between the components of would lead to a corresponding difference between the components of of about A. The experimental value for A. , would not appear to ative this arrangement , when it is borne in mind that accurate measurements of the components of are difficult , and , as St. John and Ware*have shown , micrometer measurements of incompletely resolved lines lead to values of the separation which are too high .
The observed separation and the halfwidth found for show that the components are , in fact , nearer to the limit of resolution than the raphs would lead one to expect , the contrast being somewhat enhanced by the photographic plate .
Although it is impossible to attach great weight to the exact numerical values obtained for , it is eyident that the separation is of a magnitude appropriate to a diffuse or sharp series .
It is believed that considerable may be attac the value given for the separation of the components of , for they were completely resolved and well defined , and were therefore not exposed to sources of error , which may perhaps have affected the accuracy of measurement in the case of These results , which were obtained with tubes containing a mixture of hydrogen and helium , are not in accordance with obseryations made with tubes containing pure hydrogen , or hydrogen with a small amount of impurity .
The separation of the components of appears to become smaller as the gas is purified , and the minimum between the components becomes brighter , to an extent which seems quite out of proportion .
The appearance of both and in hydrogen of the highest of purity does not suggest unresolved doublets with broad components of the separations given above , but is rather in accordance with doublets of a somewhat smaller separation in the case of , and a considerably smaller separation in the case of , the components being in neither case resolved .
Taking all these results into consideration it seems that the structure of ' Astrophys .
Journ ( 1916 ) .
320 On the Structure of the Series of Hydrogen Lines .
the lines must be complex , for it does not appear possible to explain observations on the basis of simple doublets .
If , however , it may be assumed that the lines consist of three or more components which vary in their relative intensities under different conditions , the changes which have been observed can be easily explained , and would , I think , be in harmony with Sommerfeld 's .
cit. theoretical deductions if the changes in the relative intensities of the components under different conditions is not incompatible with his theory .
There is a considerable amount of experimental evidence which affords some justification by for this assumption , for there are several instances in which this very phenomenon has been observed .
King*has found that the red lithium line , A. , may appear as a doublet or as a triplet under different conditions both in the arc and also in the tube furnace , in which a change in the structure of the line due to an electrical resolution.would appear to be excluded .
has shown that under different conditions in vacuum tubes the cadmium line , A. , may appear either as a doublet or as a triplet , and Gale and have observed similar phenomena in the satellites of the green mercury line A. There is thus a considerable amount of evidence that the fine structure of lines is not invariable , and though it cannot be claimed that the exact nature of the variation in the structure of the lines and been estabIished it seems probable that each of these lines consists of three ( or perhaps more ) components which show very great variations in their relative intensities under different conditions ; one of these components in and one in appear to be separated from the main components of the lines by a constant difference in wave-number , these components being prominent in impure hydrogen , and especially in a mixture of hydrogen and helium , whilst the other components , which are enhanced in very pure hydrogen , are separated from the main components by differences in wave-number which are more appropriate to a principal series .
This explanation would not , I believe , be inconsistent with Sommerfeld 's theory , and would be in complete harmony with the experimental results .
'Astrophys .
Journ vol. 44 , p. 169 ( 1916 ) .
'Astrophys .
Journ vol. 9 , p. 116 ( 1904 ) .
'Astrophys .
Journ vol. 31 , p. 78 ( 1910 ) .

