]\gt ; Wave-lengths of Hydrogen Determination of th .
Series Constant .
By W. E. CURTIS , A.R.C.S. , D.I.C. , B.Sc. , F.RA .
S. , Demonstrator of Astrophysics at the Imperial College of Science , South Kensington .
( Communicated by A. Fowler , F.R.S. Received June ll , \mdash ; Read June 25 , 1914 .
) Introductory .
The Balmer or series spectrum of hydrogen possesses a special interest and .
importance for several reasons .
In the first place hydrogen occupies a unique position among the elements , on account of its chemical properties and atomic weight .
Secondly , the spectrum itself is notable as having been the first series to be recognised and represented by a formula , by Balmer in 1885 .
Many other series have since been found , but in no other case is the formula so simple in character or capable of representing the series with such great accuracy .
Since Balmer published his " " law\ldquo ; improved wavelengths have been obtained by Ames , and the series has been greatly extended ( in the chromospheric spectrum ) by Dyson , Evershed , and Mitchell , but the formula remains completely satisfactory ; that is say , the observed and computed wave-lengths never differ by a greater amount than may reasonably be attributed to experimental error .
From the above considerations it is clear that this spectrum cannot fail to offer a particularly attractive field for the work of the theoretical investigator who seeks to throw light on the mechanism of radiation .
The problem in the case of the hydrogen series is evidently reduced to its simplest terms ; a satisfactory solution once obtained may , perhaps , be capable of extension to the series of other elements , for these series can all be represented by generalisations of Balmer 's formula .
It thus becomes desirable to determine the wave-lengths of the lines in question with the utmost possible exactness with a view to testing the formula and making such modifications of it as may be found necessary .
Thia was one of the objects in view when the present research was undertaken .
Another was the evaluation on the International system of wave-lengths of the constant occurring in the formula , a number of considerable ) auce , since , as was shown by Rydberg , it may be used .
in all formulae representing series of lines .
Further , it was desired to find what degree of accuracy could be attained with the instrument available ( a 10-foot concave grating ) , using the ew International standards of .
Finally , Mr. W. E. Curtis .
-lengths of Hydrogen the very frequent occurrence of the hydrogen lines in the spectra of vacuum tubes renders a precise knowledge of their positions very useful , as a check is thereby afforded on the results for other lines in the spectrum .
In connection with the first of the above objects , the testing of the formula , it may be pointed out that measurements of the first few lines of the series are , in all probabiIity , all that are required .
For , supposing Balmer 's formula to be found insufficiently exact , analogy with other series would require that a formula of the ydberg type be triea , viz. , Since Balmer 's law is such a very close approximation , must be very small and A very nearly equal to .
Now the effect of upon the wave-length , or , varies as , and hence decreases very rapidly in the direction of the limit .
Thus , for example , the value of is 54 times as great for as for .
It is evident then that if the value of is in question the data are to be obtained from the first few lines of the series , so that the above-mentioned results of Dyson , Evershed , and Mitchell , capable of furnishing a fairly good value for , are in this particular connection very little advance on those of Ames .
A summary of the wave-lengths hitherto determined is given in Iable V below .
It will be seen that the deviations of the observed values from those computed amount at times to as much as , so that there seemed to be room for a substantial increase of accuracy .
Exactly what degree of ecision was attainable by the means employed in the present investigation was not known when the work was commenced\mdash ; certainly not more than .
It is thought that this limit has very nearly been at any rate in the case of the four lines , and Conditions of Experiment .
The source of light throughout was a hydrogen vacuum tube ( of the ordinary Plucker form ) excited by means of a 12-inch coil .
A motor break was used in conjunction with the latter , and the primary current was usually between 2 and 4 amperes .
The capillary portions of the various tubes ranged between and 3 mm. in bore , and the pressures employed were from $ to 5 mm. A number of experiments were made with a view to determining the most suitable conditions for the purpose .
The condensed discharge was out of the question , for although it gives a very pure series spectrum , suppressing the secondary or compound line spectrum , the lines are much too broad for accurate measurement except at very low pressures , when the luminosity and of the Series Constant .
becomes very feeble .
With the uncondensed discharge a bright series spectrum is obtainable from the capillary , but the secondary spectrum is also present .
The relative brightness of the latter can be diminished by increasing the current density\mdash ; most conveniently effected by using a narrower capillary \mdash ; but it is not practicable to carry this process beyond a certain limit , owing to the broadening of the lines which results .
This broadening effect increases with the order of the line , so that for the higher members of the series it was impossible to use such a high current density as was practicable for .
In the latter case the tube could be made so brilliant that an exposure of one minute was sufficient , whereas in order to obtain the sixth member , six hours was the minimum necessary , even when using the fastest plate obtainable .
The light from the middle portion of the capillary was utilised , the observations being made perpendicular to its length .
Ames appears to have used an end-on form of tube , but this method is open to objection on account of the marked development of the secondary spectrum which always occurs at the ends of the capillary .
As this is associated with an actual diminution in the intensity of the Balmer series , the advantage of viewing the capillary axially is less than might appear ; when , as in the present case , the spectrum obtained is astigmaetic , so that there is no advantage in concentrating the light into a small image , the method described above is preferable .
The was prepared by electrolysis of a saturated solution of carefully purified barium hydroxide which was very kindly supplied by Prof. H. B. Baker , F.R.S. The spectrum of the gas so obtained showed no impurity except a slight trace of oxygen , which seemed to be derived from water vapour , as it was not present when the gas had been passed over anhydrous phosphoric pentoxide .
This was not done , however , as the intensity of the series is greatly diminished thereby , and that of the secondary spectrum increased .
Hydrogen which has been very thoroughly dried by standing over for some time gives a comparatively feeble Balmer series , It was also noticeable that the reIative brightness of the series lines tended to decrease after the discharge had been passing for some time .
With the heavier currents used this falling off was very appreciable in a quarter of an hour or so , and necessitated the refilling of the tube with fresh hydrogen .
The repetition of this process being rather troublesome during long exposures , an apparatus was constructed by means of which a continuous flow of hydrogen could be maintained .
The gas was passed from the generator into a reservoir having a capacity of about a litre and then into the vacuum tube through a length of very fine capillary tubing .
The other end of the vacuum tube was connected to the Gaede pump , which was kept working all the time , through a short piece of capillary tubing of medium bore .
The apparatus Mr. W. E. Curtis .
Wave-lengths of Hydrogen proved very convenient , and enabled the brightness of the .
series lines to be maintained unaltered for as long as desired .
It was only necessary to .
the reservoir once in several hours , when the pressure in the discharge tube was getting rather too low .
This method of working possesses another advantage , in that the flow of hydrogen quickly removes the carbonaceous gases which are always given off when the discharge is first passed .
After a few minutes ' running the Angstrom bands completely disappeared , and were not seen again .
Determination of Wave-lengths .
The photographs were taken in the second and third orders of the 10-foot concave grating of the Royal College of Science .
A description of the mounting , which is on the Littrow plan , has been given by Eagle .
* A variety of plates were used , according to the wave-length and intensity of the line to be photographed .
The following Table gives some details:\mdash ; Table I. The iron arc used to give the comparison spectrum was about 6 mm. long and carried a direct current of about 5 amperes , obtained from the 110-volt mains .
The electrodes were about 1 cm .
in diameter .
The light from the whole of the arc was utilised , the electrodes only being screened off from the slit .
The comparison shutter of the instrument permits of the juxtapOsition of three spectra .
Accordingly , the hydrogen spectrum was put in the middle , with iron arc comparisons overlapping above and below .
One arc spectrum was exposed before the hydrogen , and the other after .
The two were measured and reduced separately , so that any shift of the late or grating during the exposures was immediately evident from the discordance of the two results .
When the exposures are long it is quite necessary , as will be seen later , to have such a check on the stability of the instrument .
The Hilger micrometer used for the measurements had a screw of 1 mm. pitch and 150 mm. in length .
Its errors were investigated and corrections applied , but these were so small as to be hardly appreciable in the majority of cases .
*'Astrophys .
Journ vol. 31 , p. 120 ( 1910 ) .
Lines and Determination of the Series Constant .
The scale of the spectrum is about .
per millimetre in the second order , and .
per millimetre in the third order .
Over a range of not more than 15 A.U. wave-lengths may be calculated from micrometer readings by means of a linear formula ; if the range is greater than this , a correction , which is approximately parabolic in form , must be applied .
For a range of 100 A.U. the correction at the centre is of the order of , so that for this type of mounting it is particularly desirable to use standard lines as close together as possible .
The secondary International standards , averaging 50 A.U. apart , are much too few and far between .
The publication of tertiary standard wave-lengths by Burns*was therefore very welcome , and his results were of the utmost service .
In the early part of the work some six or eight iron lines were employed for each reduction , with a view to checking the internal agreement of Burns ' values .
This was found to be so uniformly satisfactory that it was judged permissible , in order to lessen the labour of measurement , to rely upon a somewhat smaller number , usually four or five , but never less than three .
The measurements were made with the greatest care ; six independent sets of readings were taken on each line , the plate being reversed end for end after the first three .
The final mean readings , after correction for micrometer errors , were reduced to wavelengths with the help of an arithmometer , and a curve of corrections constructed if necessary .
Two difficulties were encountered in the wave-length determinations .
First , the measurement of the hydrogen lines was by no means easy , on account of their width .
This averaged about .
for for and for , and , very much more than that of an ordinarily good iron line .
Second , the results derived from the two comparison spectra were often seriously discordant .
This was a very fruitful source of trouble\mdash ; the various attempts which were made to locate its origin occupied , in fact , most of the time spent on the research .
These discrepancies were small , only reaching in two cases , but quite certainly existed , that is , they could not be attributed to errors of measurement .
Further , they were most marked in the case of long exposures , a point of some significance .
Thus the determination of and practically free from trouble of this nature .
The case of will receive separate consideration .
Sources of Error .
It is perhaps worth while to summarise the chief sources of error in connection with wave-length determinations by this method\mdash ; * Zeitschr . .
Wiss .
Phot vol. 12 , 6 , p. 219 1913 ) ; 'Lick Obs. Bull vol. 8 , No. 247 ( 1913 ) .
Mr. W. E. Curtis .
Wave-lengths of Hydrogen ( 1 ) Incomplete illumination of the grating .
( 2 ) Curvature , or other irregularity of the slit ( only effective in astigmatic instruments ) .
( 3 ) Unevenness of plate surface ( only effective when diffracted beam does not fall normally on plate ) .
4 ) Relative displacements of slit , grating or plate , whether arising from\mdash ; ( a ) Mechanical disturbance ; or ( b ) Temperature change .
( 5 ) Change of refractive index of air , arising from variation of temperature or pressure .
( 6 ) Inaccuracy of setting on lines , systematic and otherwise .
( 7 ) Inaccuracy of micrometer screw , including such as may result from temperature variations during the measurement .
( 8 ) Incorrect values for the standard wave-lengths used .
Since the discrepancies in question were so small , although large enough to be of importance for the present purpose , and since only the first two of the above possibilities could be absolutely removed , it will be understood that it was a matter of some difficulty to detect the source of the trouble .
Complete illumination of the grating was secured in the usual manner , by using in each case a lens ( of sufficient aperture ) to form an image of the source on the slit .
The influence of the shape of the slit was quite easily eliminated by using the same portion of the latter for all three exposures .
With regard to the influence of the plate surface it may be pointed out that , on the average , displacements of this kind would be just as likely to occur to the iron as to the hydrogen lines ; in fact , more so , since more of the former were measured .
In point of fact , however , no shifts of any consequence were encountered in the case of the iron standards used .
Moreover , an examination of the plates tailed to reveal any correspondence between poor surface and discordant results .
The changes of wave-length produced by variations in atmospheric temperature and pressure are of the order of .
per 1o C. and .
per milhmetre respectively .
The former source of displacement was always negligible in the present instance , and it was only in a very few cases that the pressure shift was large enough to be worth taking into consideration .
After allowing for it , if necessary , there still remained a number of serious discrepancies to be accounted for .
As far as the measurement was concerned , systematic errors of setting should be eliminated , assuming constancy of personal equation .
Accidental erzors of setting were unavoidable , of course , but after some experience it quite possible to form a rough estimate of their probable magnitude .
For example , in the case of a good plate an accuracy of about Lines and of the Series Constant . .
was to be expected in setting on an average iron line .
The uncertainty was always larger for the hydrogen lines , and varied greatly with their breadth and diffuseness .
The permanent errors of the micrometer were investigated and allowed for .
In order to guard against the effect on the screw of varying temperature three separate sets of settings were made , as already mentioned .
The discrepancies in question could hardly have been due to variations in wave-length of the iron lines ; the conditions under which they were produced ( length of arc , current , etc. ) were kept as constant as possible throughout the experiments , and the practice of measuring a number of lines furnished a further safeguard against individual shifts .
There remains to be considered the possibility of disturbance of the spectrograph .
The mounting of the College instrument would appear to possess sufficient stability to render accidental shifting of either slit , grating or plate very unlikely , especially as it rests on piers which are independert of the floor of the laboratory .
The uniformly concordant results in the case of the shorter exposures were reassuring on this point .
Further , several long exposures were taken during the night , using every conceivable precaution against disturbance of bhe instrument , but without securing perceptibly better results .
Variation of temperature would appear to be a much more likely origin of the trouble .
Taking as the coefficient of expansion of the speculum metal , we find that in the neighbourhood of a shift of ( such as not infrequently occurred ) result if the temperature of the grating changes by C. between the two iron exposures .
Thus in order to preclude variations of more than , the smallest which can be detected in the present case , it is necessary that the temperature shall remain constant to within C. To test the point a platinum thermometer was employed .
It was of the grid type , and was placed above the grating and as close to it as possible .
A continuous record of the temperature , on a scale of 1 inch to 1o C. , was obtained by means of a Callendar recorder .
It was found that temperature variations of sufficient nitude to account for the observed shifts did undoubtedly occur , and an attempt was made to correlate the two , but without much success .
In several cases there was an approximate correspondence , but usually there was not .
But for several reasons it did not seem feasible to make a really satisfactory investigation of the point at issue .
For one thing the hydrogen lines , as already noted , are broad and increasingly diffuse towards the limit of the series .
Hence , so far as the first six lines are concerned , the probable error of measurement is greatest for and , which on accolmt of their feeble intensity require the longest exposures and are thus most susceptible to the effects of varying temperature .
Moreover , it was not VOL. XC.\mdash ; A. 2 X Mr. W. E. Curtis .
Wave-lengths of Hydrogen usually possible to measure satisfactorily more than one line of the series on one plate , since its neighbours were on the one side over-and on the other under-exposed .
Thus the determination of the shift depended entirely on one line , a very unsatisfactory state of affairs .
It is hoped to resume the study of the question under more favourable conditions , that is to say in connection with a spectrum containing a number of fine lines .
Whether or not the aoements found in the present case are to be ascribed to temperature changes , it is certainly likely that the latter will provide a serious obstacle to the accurate determination of wave-lengths when the exposures are long .
The instrument employed is very compact : being entirely contained in a box 11 feet by 2 feet by 2 feet , the walls of which are double , with a lagging of wool between .
One would thus expect the temperature to remain much more nearly constant than in the case of other forms of mounting , e.g. Rowland 's , so .
that it will be interesting to compare the present results with those obtained with other instruments under similar circumstances.* It was particularly desirable , for the reasons already explained , to determine the position of with the utmost accuracy .
Unfortunately , however , the probable error of this result is larger than that of any of the other five lines measured .
This is in part due to the scarcity of standard lines in the in question , only four being available , and those some distance apart .
Using the first and last for the derivation of alinear formula , the curve of corrections obtained from the other two departed very appreciably from the parabolic form which holds for shorter ranges .
A cubic formula was accordingly used to determine the correction required for .
In to this source of uncertainty some evidence was obtained of variability of the wave-le1igths of these standards .
The possibility of such variations is also suggested by the rather discrepancies between the values obtained for these lines by different observers .
-length Resulis .
In combining the results for any one line the two values derived from each plate were weighted according to the agreement between them ; if they * Since the above was written , a paper which bears on the point has come to hand Birge , Astrophys .
Journ ) .
The author was investigating the first positive band system of the nitrogen spectrum , and the exposures were very long up to 120 hours ) .
He found shifts in exact correspondence with the observed temperature changes , but he was able to control the latter the grating mounled in a specially construted room ) in such a way as to neutralise the disturbing effects of variations of atmospheric pressure .
Burns , Lnnes and Deterrnination of the Series Constant .
613 disagreed by more than .
both were rsjected .
The probable errors given have been derived by means of the usual formula , viz. , . .
It may be remarked that in the case of the obable error stated is too small , since it does not include the uncertainty arising from the causes mentioned above .
The results are summarised in the following Table:\mdash ; Table II .
These wave-lengths were reduced to vacuum by applying corrections taken from Kayser 's 'Spectroscopie , ' vol. 5 , p. 514 .
If now Balmer 's law is accurately obeyed the value of given by the expression should be the same for each line within the limits of observational error .
The figures given below appear to indicate that such is not the case .
Table III .
A small systematic departure from Balmer 's law is obviously suggested , the nature of the deviation being such as to point to a Rydberg formula as being preferable .
Before calculating such a formula , however , a careful examination of the value for was made , in order to make sure that the deviation in question could not reasonably be attributed to error in the derived wave-length .
The effects of errors of measurement and method Mr. W. E. Curtis .
Wave-lengths of Hydrogen of reduction of the standard wave-lengths and of the correction to vacuum were considered , and an estimate formed of the maximum total error on all these acoouuts .
This seemed to be about .
in the extreme and very improbable case when all the errors have the same sign .
To fit to a Balmer formula derived from the other lines requires a correction of about , so that one is forced to the conclusion that the deviation is genuine .
Duplicity of the Hydrogen The determination of an exact formula for the Balmer series is unfortunately complicated by the fact that the lines are very close doublets , so close , indeed , that they appear single on all the plates obtained in the present work , except that has shown a suggestion of resolution in a few cases .
Michelson applied his interferometer to and in 1892 and found them to be double , the separations being and .
respectively , but the components were so broad as to render the resolution very difficult .
The lines have since been examined by a number of observers with very varying results .
There is a general consensus of opinion that and are double , but considerable disagreement as to the precise separations of the components .
Thus for Houstoun*gives , Fabry and Buisson , and Paschen and .
Further , the components appear to be unequal in intensity , the less refrangible being slightly the strongerS ( the .
ratio is given by Michelson as about 10 : 7 ) .
There is thus some uncertainty as to what point of the compound line the present measurements ] refer ; it will -presumably be slightly to the less refrangible side of the point midway between the components .
In any case it seems evident that the simple Balmer formula is exact when applied to the " " centre of gravity\ldquo ; of the lines , although there is just a possibility that it may hold for the less refrangible components .
An attempt was made to derive wave-lengths for the components and calculate separate formulae for them , but the data available seem too uncertain to permit of reliable results being obtained in this way .
Corrected Formula for Hydrogen The general formula of Rydberg is as follows:\mdash ; 'Phil .
Mag ( 6 ) , vol. 7 , p. 466 ( 1904 ) .
'Journ .
de Phys June , 1912 .
'Ann . .
Phys vol. 39 , p. 897 ( 1912 ) .
S Fabry and Buisson , Lines and of the Series Constant .
As the deviation from the Balmer or single constant formula is so small , it is natural to try next if the observations can be fitted by means of a modified Rydberg formula containing only two constants .
If is put .a least squares solution gives and .
This formula ( II ) represents the lines within , with the exception of which it gives too high .
It is distinctly preferable , however , to put , as the formula ( I ) then gives a fit quite within the limits of experimental error .
The values obtained for the constants by the method of least squares are Table A comparison of the results of observation with the two formulae is given in the figure .
The ordinates represent the difference between a standard value of and the value derived from each line by assuming Balmer 's law .
The superiority of the formula adopted is clearly shown in spite of the smallness of the wave-length deviations .
Mr. W. E. Curtis .
-lengths of Hydrogen The figure also gives the results obtained by means of a formula ( III ) suggested by Dr. Bohr in a letter to Prof. Fowler , viz.:\mdash ; The departure from the simple Balmer formula , which has already been derived theoretically by Dr. Bohr , *depends upon a correction of the mass of the electron due to its velocity .
It would seem that this formula is not of a suitable type to represent the experimental results .
The calculation of a general Rydberg formula following results:\mdash ; The improvement obtained by introducing the third constant is so small as to be scarcely perceptible , even when the fourth decimal place ( in wavelengths ) is taken into account .
It will be noted that the values of and are practically the same as before , while is exceedingly small .
It thus seems quite legitimate to adopt the two-constant formula already given .
With regard to the constants and , it seems pretty certain that the former must lie somewhere between and .
The value of cannO be relied upon with such confidence , as it depends almost entirely on the wave-length of , which is rather more doubtful than the others .
But is certainly positive , and very probably lies between and The above formula gives the convergence frequency of the series as , corresponding to a wave-length ( in air ) of I.A. , or on Rowland 's scale .
on with other Observations .
The following Table gives a comparison of the wave-lengths calculated from this formula with the results of various observers .
The former values refer to air at C. and 760 mm. , and the latter have been reduced to the International scale .
The observations of Dyson , Evershed , and Mitchell were made on the chromosphere during total solar eclipses , the two former employing prismatic instruments and the latter plane and gratings .
As the pressure in the upper chromosphere must be very low , we should expect the wave-lengths of the hydrogen lines to be practically identical with those determined from vaouum tube work .
The results of Dyson and Evershed support this view , * Phil. Mag vol. 26 , p. 1 Lines Determination of the Series Constant .
' Preliminary Table of olar Wave lengths .
'Phil .
Nag .
Phil. Trans , vol. 206 , p. 403 ( 1906 ) .
S 'Phil .
Trans , vol. 201 , p. 457 ( 1903 ) .
'Astrophys .
tourn vol. as , p. 407 1913 ) .
while those of Mitchell might at first sight appear to be opposed to it .
The average deviations are as Dyson .
Evershed , , , , , Mitchell .
, , , , , There is thus a considerable systematic deviation in the case of Mitchell 's * Neglecting Nos. 22 , 24 , 28 , 32 , 33 , 36 , which , according to Mitchell , are blended with other lines .
Mr. W. E. Curtis .
Wave-lengths of Hydrogen results , but nothing to correspond with this in the other two cases .
The explanation probably lies in the difficulty of determining wave-lengths from objective or slitless spectra such as these .
This difficulty is especially marked in the case of strong lines .
Since Mitchell 's list contains many more lines than either of the other two it is to be presumed that his hydrogen lines are stronger , and hence more difficult to measure .
It will be seen that discrepancies are particularly large for the earlier ( i.e. the strongest ) members of the series , as would be expected on this view .
Series Relationships in the Eydrogen Spectrum .
It is frequently possible in work on series to obtain valuable evidence with regard to the one under discussion from an examination of the series associated with it .
For example , if the Rydberg-Schuster relations hold\mdash ; and they are exceedingly nearly exact , if not absolutely so\mdash ; the convergence frequency of the Diffuse series is equal to the " " variable part\ldquo ; of the first Principal line .
It may readily happen that the latter , on account of its strength , sharpness or position in the spectrum , is better , ted to accurate investigation than the Diffuse lines themselves , so that it may be practicable to obtain in this way a better formula than is derivable merely from internal evidence .
It is unfortunate that in the present case no assistance of this kind is available , on account of the special character of the hydrogen spectrum .
Up to the present the Balmer series has usually been regarded as of the Diffuse type .
In 1896 Pickering found a series of lines in the spectrum of Puppis , which it was quite natural to clude was the associated Sharp series .
There was , of course , the anomaly that it could not be tained in the , but the numerical evidence of identification seemed so strong as to outweigh this objection .
The case seemed to be further strengthened by Rydberg 's well known calculation of the stellar line 4688 as the first line of the Principal series , and to be finally established by Prof. Fowler 's success in producing this line , together with several other members of Rydberg 's calculated series , in the laboratory .
Prof. Fowler 's most recent work* on series of enhanced lines , however , seems to leave no room for doubt that both the " " Puppis\ldquo ; and the " " 4688\ldquo ; series are really enhanced line series of helium .
The same conclusion had previously been reached from theoretical considerations by Dr. Bohr ( loc. cit The hydrogen spectrum thus regains its apparently anomalous position , and we are to enquire whether the Balmer series is really a Diffuse series , and , if so , what has become of the associated ShalT and Principal *Bakerian Lecture , 1914 , ' PhiL Trans , vol. 214 .
Lines of the Series Constant .
series .
The evidence as to the nature of the Balmer series is very meagre .
The fact that the less refrangible component of is the stronger favours the view that the series is of the subordinate type .
The diffuse character of the lines naturally suggests the Diffuse type , but this is by no means an infallible criterion .
The electrical resolution of the lines appears to correspond with that of the Diffuse series of helium .
There is no further evidence available at present\mdash ; the Zeeman effect , for example , throws no light on the matter .
Let us suppose , however , that the series is of the Diffuse type .
Analogy with all other known cases leads us to expect that the Sharp series will be developed under precisely the same conditions of experiment , and , since it must have the same convergence frequency as the Diffuse , the lines will fall in the same region of the spectrum , so that they should be readily observable .
As they have never been seen , however , it is reasonable to suppose that in the case of hydrogen the Sharp and Diffuse series practica , ' the Balmer series representing the superposition of the two .
There not at present any direct evidence bearing on the question , but the ve manner of regarding the Balmer series has the advantage of bringing the hydrogen spectrum into line with other series spectra , and converting it from an exception into a limiting case .
If this view is correct , we should expect the Principal series to be given either by , or by , the uity arising from the very small value of in the Balmer series .
In the former case the Principal would be superposed on the two Subordinate .
In the latter we have the series recently obtained by Lyman in the Schumann region .
Similarly , the Fundamental series would be given either by or by respe , ctively .
The former series is in the infra-red ; two of the lines and have been obtained by Paschen .
The question of the precise nature of the Balmer doublets must be left unsettled for the present .
and are the only members which have been resolved , and the results obtained are so discordant that conclusion can be drawn from them .
Bohr makes the interesting suggestion that the lines are normally single , but are resolved into doublets under the action of the electric field in the vacuum tube .
If this were so the nce of the results of various workers would be very simply explained , as arising from differing conditions of experiment .
But for any one case the ratio of the separations of and should be , taking Stark 's results for these lines .
Paschen and Back 's measurements give * In some other cases they lie very close together , e.g. , for 'Phil Mag ( 6 ) , vol. 27 , p. 606 ( 1914 ) .
Wave-lengths of Hydrogen for this ratio , in sufficiently good agreement with the above .
On the other hand , Michelson 's results give and Fabry and Buisson 's , so that it does not seem possible to account for the discrepancies in this way .
Summary .
1 .
The wave-lengths in I.A. of the first six lines of the hydrogen series have been determined with an accuracy of about 2 .
Balmer 's formula has been found to be inexact .
The results may be represented by a modified Rydberg formula containing only two constants , thus : where and 3 .
An accuracy of .
is attainable in the third order of a 10-foot concave grating if the exposures are short ( say less than half an hour ) .
With longer exposures accurate determinations become very difficult if the temperature of the instrument cannot be controlled .
4 .
The tertiary iron arc standards determined by Burns were tested in the special regions under ation , and found very satisfactory .
The author 's best thanks are due to Prof. Fowler for the interest he has taken in the work and for measuring some of the plates .
/

