Atmospheric Ozone in the Spectra of Sun and Stars .
and that when m is even , | \lt ; pm|2 = 0 .
The second plate neutralises the reflection from the first plate , the fourth plate that from the third , and so .on .
The simplest case under this head is when 8 = ^ A , == A. A variation of the latter supposition leads to a verification of the general formulas worth a moment 's notice .
We assume , as above , 8 ' = s'A , but leave $ open .
Since e*m ' = \#177 ; 1 , ( 9 ) and ( 10 ) become _ eiikS_e-hikS q-1-77 / 6gx and these are of the form ( 39 ) , if we suppose a = rfx\gt ; = The reflection \lt ; f)m from m plates is derived from rby merely writing bm for b , that is , , " iimks for eiikS } leaving \(f\gt ; m\ equal to |r| , as should evidently be the case , at least , when S ' = 0 .
Absorption Bands of Atmospheric Ozone in the Spectra of Sun and Stars .
By A. Fowler , F.R.S. , Professor of Astrophysics , and the Hon. R. J. Strutt , F.R.S. , Professor of Physics , Imperial College , South Kensington .
( Received June 13 , 1917 .
) [ Plate 3 .
] Introductory .
In a paper communicated to the Royal Society in 1890 , Sir William Huggins gave an account of his discovery of a new group of lines in the photographic spectrum of Sirius , situated near the limit of atmospheric transmission.* A photograph taken with a long exposure was described as showing no strong lines after the termination of the hydrogen series until about A 3338 was reached , at which place there appeared the first of a group of at least six lines , all of which were nearly as broad as those of hydrogen .
The third line of the group , about A , 3278 , appeared to be the broadest .
The sixth line occurred almost at the limit of the photograph , where the spectrum was very faint , and it was not possible to determine whether this was the last member of the group .
The wave-lengths of the lines , which were only considered to be roughly approximate , were given as 3338 , 3311 , 3278 , 3254 , 3226 , and 3199 .
In 1915 , a copy of an excellent photograph of the spectrum of Sirius , * ' Eoy .
Soc. Proc. , ' vol. 48 , p. 216 ( 1890 ) .
Profs .
A. Fowler and Hon. R. J. Strutt .
which had been taken with a small quartz prismatic camera at the Royal Observatory , Edinburgh , was kindly forwarded to us by Prof. Sampson .
' In this photograph the Huggins group is well shown ( Plate 3 , fig. 3 ) , and it was at once suspected , from the general appearance of the spectrum , that a system of bands , and not a group of lines , was in question .
From our knowledge of the physical conditions in the atmosphere of Sirius , as indicated by other parts of the spectrum , the existence of bands in this star seemed highly improbable , and it was an obvious inference that they might be produced by absorption in our terrestrial atmosphere .
This view was strongly supported by photographs of the solar spectrum which we obtained with a small quartz spectrograph near sunset , as compared with similar photographs taken when the sun was at a considerable elevation .
The spectrum of the low sun showed very decided indications of bands occupying about the same positions as those in the spectrum of Sirius , but , on account of the superposition of solar lines , the wave-lengths of the bands could not be very certainly derived .
The published results of experiments on the absorption of atmospheric gases suggested that ozone was the most probable source of the bands under consideration .
Ladenburg and Lehmann* had , in fact , noted a number of bands in the absorption spectrum of ozone , extending from 3220 to \ 3566 , and thus covering a part of the region occupied by the bands observed in Sirius .
The wave-lengths given by Ladenburg and Lehmann , however , were not in sufficient agreement with those given by Huggins , or with the provisional wave-lengths communicated to us by Prof. Sampson , to establish the identity of the ozone and Sirian bands .
As no other probable origin occurred to us , it was decided to make a new experimental investigation of ozone absorption , and , after some unavoidable delay , the experiments were recently undertaken .
The new observations have definitely established that ozone is the origin of the Huggins group of bands in the spectrum of Sirius , and that the same bands are also present in the solar spectrum , with an intensity varying with the sun 's altitude .
While this investigation was in progress , our attention was drawn by Prof. Sampson to a dissertation by Otto Kohl , f in which the author gives results of measurements of a number of photographs of ultra-violet stellar spectra which had been taken at Potsdam by Prof. Hartmann in 1904. .
In this paper , the discovery of the bands near the end of the spectrum was wrongly attributed to Prof. Hartmann , but the additional fact was brought out that they occurred in different stars , irrespective of spectraL * ' Ber .
Deutsch .
Phys. Gesell .
, ' vol. 8 , p. 125 ( 1906 ) .
t ' Ast .
Mitt .
Konig .
Sternwarte zu Gottingen , ' No. XYI ( 1913 ) .
Atmospheric Ozone in the Spectra of Sun and Stars .
579 types .
Thus , the bands were noted in a Aurigae , a Canis Majoris , a Lyrae , / 3 Orionis , and f Tauri , as well as in the solar spectrum .
There were considerable divergences in the measured positions of the hands , but , in view of their diffuseness and their varying breadths in different spectra , it could reasonably be concluded that they were common to the different stars .
The occurrence of the bands in different stars naturally suggested that the whole group had the same origin , and that they might possibly be due to absorption by the earth 's atmosphere .
There was no evidence , however , for attributing them either to atmospheric oxygen , nitrogen , water-vapour , or carbon dioxide , and there appeared to be no agreement whatever between the stellar bands and the bands of ozone , as recorded by Ladenburg and Lehmann .
The alternative supposition that the bands might be of stellar origin likewise received no support from comparisons with metallic spectra , and it was concluded that chemical identification would be necessary to decide whether they were of sidereal or telluric origin .
It should be mentioned that the bands also appear very distinctly in photographs of the spectra of Capella , Vega , and Eegulus , taken at Edinburgh on April 12 , 1915 , and recently forwarded to us by Prof. Sampson .
Previous Observations of Ozone Absorption .
The general features of the absorption spectrum of ozone have long been known .
In the visible spectrum there are several ill-defined bands , which were first described by Chappuis* in 1880 .
These bands were also obtained , but less strongly , by Living and Dewarf in 1889 ; and , more recently , their positions have been re-determined by Ladenburg and Lehmann.| The strongest of these bands , according to Chappuis , extends from \ 6095 to \ 5935 .
A region of strong absorption in the ultra-violet was discovered by HartleyS in 1880 .
The central part , where the absorption was most intense , was estimated to be at \ 2560 , and the band was observed to extend in both directions as the amount of absorbing gas was increased .
Hartley concluded , however , that the band would not extend indefinitely with increase in the amount of ozone , and that nearly the maximum effect was produced when all rays with a wave-length less than 2930 were absorbed ; that is , it was thought that further additions of ozone would only slightly increase the absorption .
* * * S * ' Comptes Rendus , ' vol. 91 , p. 985 , and vol. 94 , p. 858 .
t ' Roy .
Soc. Proc. , ' vol. 46 , p. 222 ( 1889 ) .
+ Loc .
cit. , p. 129 .
S 'Jour .
Chem. Soc. , ' vol. 39 , pp. 57 and 111 .
Profs .
A. Fowler and Hon. R. J. Strutt .
Quantitative investigations of the broad ultra-violet absorption band have been made by Fabry and Buisson * who used a mercury arc as the source of light , and determined the coefficient of absorption for 10 different wavelengths between 2352 and 3341 .
The wave-length of maximum absorption was placed at X 2550 , in close agreement with Hartley , and ozone was stated to be as opaque for light of this wave-length as are the metals , mass for mass , for visible light .
The ultra-violet bands which form the chief subject of the present communication occur near the less refrangible boundary of the great absorption band .
The only previous observations of bands in this region appear to be those of Ladenburg and Lehmann , to which reference has already been made .
Burning magnesium was used as the source of light , and it was stated that with small concentration of ozone there was merely a continuous absorption extending to about X 3160 , while with a large amount of the gas a number of relatively narrow bands were observed near the limit of absorption towards the red .
It was further observed that bands appeared at successively longer wave-lengths as the concentration of ozone was increased .
With the greatest concentration obtained by Ladenburg and Lehmann , the complete absorption reached as far as X 3388 .
The bands in question probably escaped the notice of Hartley , because the electric spark which he employed as the source of light in his experiments was not well adapted to show them .
Fabry and Buisson make no reference to the bands , and it is probable that they also did not observe them on account of the unsuitability of their source of light .
For the sake of completeness , it should be added that the absorption of ozone in the infra-red has been investigated by K. Angstrom , f who observed bands which were also found in the solar spectrum .
Experimental Procedure .
For laboratory observations of the absorption bands a source of light is required which yields a spectrum as continuous as may be , and extending far into the ultra-violet .
The absorption bands to be looked for , as will appear later , are very faint over a part of their range , and emission lines or bands , due to the source of light used , add greatly to the difficulty of observing them .
After trial of various sources , those selected for use were : ( 1 ) the incandescent crater of the carbon arc ; and ( 2 ) the light of burning magnesium ribbon , fed by a clockwork lamp .
The carbon arc does fairly * ' Jour , de Phys. , ' vol. 3 , p. 196 ( 1913 ) .
t 'Arkiv.for Matematik ' ( Stockholm ) , vol. 1 , pp. 347 and 395 ( 1904 ) .
Atmospheric Ozone in the Spectra of Sun and Stars .
581 well to about \3000 , but beyond this the continuous spectrum thins out , and lines due to impurities have a disturbing effect .
Burning magnesium is less convenient for long exposures , but its continuous spectrum has been photographed as far as X 2200 , and it shows no lines or bands between that and X 3600 , except the strong reversed line at X 2852 .
The ozone used was prepared in the ordinary way by means of a Siemens ozoniser .
Oxygen from a cylinder was bubbled through sulphuric acid , which gave a rough idea of the rate of flow ; then it passed through the ozoniser to the observation vessel , through which a continuous flow was maintained .
The observation vessels were of various kinds .
Tor great thicknesses , glass tubes were used , closed by quartz plates cemented on with sodium silicate .
The length of these tubes varied from 8 feet to 18 inches .
For small thicknesses , the ozone stream was delivered by displacement into silica flasks or test-tubes placed in front of the slit of the spectroscope .
No strict control was kept over the quantities of ozone in use , but the rate of passage and other conditions in the generator were kept fairly constant , so that the length of the vessel afforded a convenient means of varying the amount of ozone producing absorption .
In other experiments , when the long tube was in use , the tube was first filled with ozone , the ozoniser was then put out of action while the gas stream was left on , so that ozone was slowly displaced by oxygen .
During this process , a succession of exposures were made , allowing the spectroscopic changes to be traced .
The greatest concentration of the ozone used was probably less than 1 per cent. Ladenburg and Lehmann were not able to trace any bands when using dilute ozone .
They had recourse to the troublesome process of concentrating it by liquefaction .
We have not found this to be necessary .
The instrument chiefly employed for the photography of the spectra was a small quartz spectrograph by Hilger , giving a linear dispersion of 58 A. per millimetre at X 3200 .
This small dispersion was well adapted for the purpose in view , as most of the bands are rather diffuse , and very precise determinations of their positions are not to be expected .
Some of the photographs were taken with copper arc comparisons , and the wave-lengths of the ozone bands were determined in the usual manner by the Cornu-Hartmann interpolation formula .
Photographs of the bands were also taken with a much more powerful instrument , giving a dispersion of 7 A. per millimetre at X 3200 , but the bands were so diffuse under these conditions that there was no gain in accuracy as compared with the smaller spectrograph .
The latter has also the VOL. xciii.\#151 ; a. 2 z Prof. A. Fowler and Hon. P. J. Strutt .
advantage of giving spectra comparable in purity and dispersion with those of the instruments employed for the stellar spectra .
The bands were lost altogether when it was attempted to photograph them in the 3rd order of a 10-feet concave grating .
The New Observations .
The bands in the visible spectrum have not been considered in this investigation , but as regards the ultra-violet region the new observations are in general agreement with those of previous investigators .
Additional features , however , have been noted .
The great absorption band in the ultra-violet is illustrated in fig. 1 , Plate 3 .
These photographs were taken with a cadmium spark as the source of light , and show the broadening of the band which accompanies an increase in the amount of absorbing gas .
In opposition to the statement of Ladenburg and Lehmann , the system of narrow bands was found to be represented even with minute amounts of ozone .
They were , in fact , photographed as far as X 2700 when the layer of dilute ozone was only 1 cm .
thick .
As the thickness of gas was increased , the more refrangible members of the group were obliterated by the increased extent of the great band , and new members continued to appear just beyond the boundary towards the red .
There was , however , a general increase in the distinctness of the bands when the region about X 3100 to X 3300 was reached .
The least refrangible band observed was a very faint one at X 3432 .
It would seem that Ladenburg and Lehmann only measured the spectrum with very large concentration , as they recorded no band more refrangible than X 3220 , and their last band towards the red was about X 3566 .
The present investigation suggests that the narrow bands continue throughout the whole range of the great absorption band , though they become rather inconspicuous even at X 2700 .
The successive appearances of the spectrum with increasing thicknesses of ozone are illustrated in fig. 2 , Plate 3 .
Eeferring to these spectra , the actual bands seen in ( a ) do not appear in or ( c ) because , on account of the general absorption , no light in the region of X 2800 can get through .
And again , the actual bands seen in ( c ) do not appear in ( a ) or ( b ) , because there is not enough ozone to bring them out .
As it was not found possible to obtain the desired range of bands in a single experiment , photographs taken with different concentrations of ozone were used for purposes of measurement , and the results combined .
The wave-lengths and intensities of all the bands on the red side of X 3080 which were sufficiently distinct for measurement are included in Table I. The strongest bands are assigned intensity 10 , and the faintest intensity 1 .
Atmospheric Ozone in the Spectra of Sun and Stars .
583 Bands which appear as pairs or groups are indicated as such by connecting brackets , and those which are of more than average breadth are indicated by the letter " b " following the intensity number .
Table I.\#151 ; Wave-lengths of Ozone Absorption Bands .
x. 3482-2 3421 4 3402 -6 f 3377 -7 \lt ; 3374 -1 [ 3365 -2 3346 -0 3338 -5 3331 -2 / 3311 -5 \ 3304 -1 I 3284 -0 \lt ; 3279 -8 [ 3272 -0 / 3255 -5 13249-7 Intensity .
5b 8b Id X. f 3232 '8 .I 3227 -2 [ 3221 -5 f 3206 -8 ^ 3201 -0 [ 3194 -8 3188 -8 3181 -5 f3177 -0 P 3171 -6 I 3162 -6 \ 3156 -1 3137 -4 3114 -3 f 3105 -o t 3096 -5 3089 -5 Intensity .
8b 8b 4 2b 8 lOi 8b 5 4 8b The wave-lengths of the 12 bands given by Ladenburg and Lehmann , without further description , are 3566 , 3503 , 346 , 3435 , 3409 , 3376 , 3343 , 333 , 3295 , 327,3235 , and 322 .
The first three are out of range of our observations , and the remainder are not in good agreement with our measures .
The band 3295 does not appear in our list , and that at 3235 is only represented by a very faint band at 3232*8 .
Moreover , the wave-lengths 327 and 322 represent very imperfectly the corresponding groups in our photographs .
The strong pairs 3311*5 , 3304T and 3255*5 , 3249*7 , which are included in Table I , do not appear at all in Ladenburg and Lehmann 's list .
It should be noted that the photographs taken with high dispersion gave no indications of the resolution of the bands into close component lines .
A similar result was obtained by Chappuis in the case of the bands which occur in the visible spectrum .
Ozone Bands in the Spectrum of Sirius .
The Edinburgh photograph of the spectrum of Sirius , of which an enlarged copy is reproduced in fig. 3 , Plate 3 , is on a scale nearly identical with that employed in the laboratory experiments , the dispersion being 56*5 JL per millimetre at X 3200 .
In accordance with the description given by Huggins , there are no strong lines between the end of the hydrogen series and the beginning of Prof. A. Fowler and Hon. R. J. Strutt .
the group of bands which occurs near the limit of the spectrum .
Only one distinct line was , in fact , certainly recognised on the photograph in this intermediate region , and by extrapolation from the hydrogen lines this was identified with the iron line 34408 .
This was subsequently used as a standard , in combination with Hy ( \4340'7 ) and in the determination of the wave-lengths of the bands .
The resulting wave-lengths are included in Table II ; where they are compared with the positions assigned by Huggins and by Kohl , the latter being taken as the means of the two wavelengths stated to represent the edges of each of the bands .
The last column of the Table gives the strongest bands of ozone according to our measures .
Table II.\#151 ; Ultra-violet Bands of Sirius compared with Ozone .
Spectrum of Sirius .
Huggins .
Kohl .
Fowler \amp ; Strutt .
X. 1 X. 1 Int. A. Int. 3338 3337 -2 2 3311 j r 3311 -9 3 5 1 3303 -7 2 1 f 3285 -0 1 3278 3280 *6 i 3279 *8 5 j [ 3272 -9 1 3254 j 13254 -8 2 i [ 3249 -7 5 3226 3224 '8 2 f3227 -5 [ 3220 '0 5 5 55 3199 3199 2 2 \lt ; f 3201 *7 [ 3194 -8 4 4 3176 6 1 4 f 3177 *9 [ 3172 5 5 2 3158 -2 3157 ?
4 5 3138 ?
45 Ozone .
Kemarks .
A. Int. 3338 -5 4 Kohl gives a " line " at 3337*5 .
3311 -5 55 3304 -1 3 3284 -0 2 3279 -8 85 3272 -0 3 3255 -5 3249 -7 5 8 j* Kohl gives a " line " at 3249*4 .
3227 -2 10 3221*5 10 3201 -0 85 \ There is possibly a metallic 3194 *8 6 J line in Sirius , about 3198 .
3177 -0 85 These are only vaguely sepa3171 '6 4 rated in Sirius .
f Estimates of intensity in star 3156 -1 8 J very uncertain on account 3137 *4 105 j of faintness of continuous L spectrum .
The bands , as a whole , are more diffuse in the photograph of Sirius than in the laboratory spectra , and , making due allowance for the difficulties attending the measurement of band spectra , the agreement in the measured positions of the bands of Sirius with those of ozone may be considered very satisfactory .
The proof of the identity of the two sets of bands , however , does not depend so much upon close coincidences in the measured positions of individual bands as upon a general agreement of the wave-lengths and intensities as a whole .
This agreement is more convincingly demonstrated by the direct comparison of the spectrum of Sirius with that of ozone , as in fig. 3 .
A perfect correspondence of the two spectra in the region occupied Atmospheric Ozone in the Spectra of Sun and Stars .
585 by the bands is clearly indicated , there being no strong lines in the spectrum of Sirius to cause complication .
Ozone Absorption in the Solar .
The presence of the Huggins group of bands in the spectrum of the low sun was indicated by numerous small-scale photographs taken during 1915 .
Measurements of the bands were rendered difficult by the superposition of metallic lines , but the general accordance of wave-lengths was sufficient to prove their identity with the bands appearing in Sirius .
Kohl had previously noticed the bands in the sun , as we afterwards learned , but he apparently made no investigation of their varying intensity , and his list included three bands which were not indicated by him as appearing in the spectrum of Sirius .
Conclusive evidence that the bands are of telluric origin is furnished by photographs of the high and low sun , such as are reproduced in fig. 4 .
These have been selected from a series of 17 negatives obtained between noon and sunset on 1917 , May 23 , and show the spectrum corresponding to altitudes of the sun 55 ' , 30 ' , 10 ' , 4 ' , and 1 ' .
It will be observed that as the sun 's altitude is reduced there is a general advance of the complete absorption towards the red , accompanied by a gradual strengthening of the bands , so that the appearance of the spectrum about sunset is very different from that near noon .
It follows that the bands are produced by absorption in our own atmosphere .
As in the case of Sirius , the identity of the bands with those of ozone is more convincingly demonstrated by direct comparison of the solar and experimental spectra than by attempts to prove exact agreement of wavelengths .
The spectrum of ozone given in the lowest strip of fig. 4 was obtained with a concentration closely corresponding to the maximum atmospheric absorption , and it will be seen that ozone accounts completely for the difference between the spectra of the high and low suns in the ultra-violet region.* The foregoing investigation gives further substantial support to Hartley 's suggestion that atmospheric ozone is the effective agency in fixing the limit towards the ultra-violet to which the solar spectrum can be observed .
The extension of the complete absorption towards the red , and the successive appearances of bands of longer wave-lengths , as the sun 's altitude becomes lower , correspond precisely with the phenomena observed in the experimental spectra as the quantity of absorbing ozone is increased .
* The appai-ent lack of exact coincidence about X 3195 is due to an adjacent group of solar lines near the more refrangible edge of the corresponding band of ozone .
586 Atmospheric Ozone in the Spectra of Sun and Stars .
It may be that the recognition of these narrow absorption bands of ozone in the ultra-violet spectra of the sun and stars will facilitate investigations relating to atmospheric ozone , as the bands occur in a part of the spectrum which is easily photographed , and only a small spectrograph is necessary to show them .
The authors desire to record their appreciation of the kindness of Prof. Sampson in freely placing at their disposal the Edinburgh photograph of the spectrum of Sirius , which was the origin of the present investigation .
They are also indebted to Mr. J. Brooksbank , B.Sc. , for a number of photographs of the solar spectrum , and for assistance in preparing the plate for reproduction .
DESCRIPTION OF PLATE .
Fig. 1.\#151 ; To illustrate the great absorption band of ozone in the ultra-violet .
( a ) Cadmium spark , without ozone .
( b ) Absorption due to 1*2 cm .
dilute ozone .
( c ) Absorption due to 3'5 cm .
dilute ozone .
( d)Absorption due to 7'5 cm .
dilute ozone .
Fig. 2.\#151 ; To illustrate the advance of the great absorption band towards the red , and to show the appearance of the narrow bands at successively longer wave-lengths , as the thickness of absorbing ozone is increased .
The source of light was burning magnesium , and X 2852 appears as a reversed line .
( a ) 1 cm .
of dilute ozone , showing faint bands from X 2700 to X 2840 .
( b)45 cm .
of dilute ozone , showing stronger bands from X 3000 to X 3200 .
( c ) 175 cm .
of dilute ozone , showing strong bands from X3150 to X 3432 .
Fig. 3.\#151 ; The spectrum of Sirius compared with ozone .
( a ) Spectrum of Sirius , photographed at the Royal Observatory , Edinburgh , on 1915 , February 8 ; altitude of star about 15 ' .
( b ) Absorption bands of ozone in the region X 3150 to X 3432 .
Fig. 4.\#151 ; Solar spectra , compared with ozone .
Enlarged 5*7 times .
( a ) Sun , altitude 55 ' ' ( b ) Sun , altitude 30 ' ( c ) Sun , altitude 10 ' \gt ; 1917 , May 23 .
( d ) Sun , altitude 4 ' | ( e ) Sun , altitude 1 ' J ( / ) Ozone absorption bands .
Fowler and Strutt .
Roy .
Soc. Proc. , A , vol. 93 , PI .
3 .

