260 Ultra-violet Transparency of the Lower Atmosphere , , and Relative Poverty in Ozone .
By the Hon. R. J. Strutt , F.R.S. ( Received December 19 , 1917 .
) [ Plate 2 .
] Many years ago it was suggested by Hartley* ' that the limit of the solar spectrum towards the ultra-violet was attributable to absorption by atmospheric ozone , which , as he showed , would give rise to a general absorption beginning at about the place where the solar spectrum ends .
In a recent paper by Prof. A. Fowler and myself , f the evidence for this view was very much strengthened .
For it was shown that just on the limits of extinction the solar spectrum shows a series of narrow absorption bands which are eventually merged in the general absorption , and these narrow bands are precisely reproduced in the absorption spectrum of ozone .
For my own part , I do not feel any doubt that ozone in the atmosphere is the effective cause limiting the solar spectrum .
The work just cited naturally led to an investigation of the absorption by the atmosphere of ultra-violet light from a distant terrestrial source .
This research , though more than once suggested , does not seem to have been undertaken before .
Various sources of ultra-violet light were tried in the course of the investigation .
Burning magnesium proved inconvenient , because of the rapidly falling intensity of the spectrum in the far ultra-violet .
A cadmium spark was used for short distances up to 1200 yards , and has the advantage of showing strong lines right through and even much beyond the region of strongest ozone absorption .
This would have been the best source for longdistance experiments also , if facilities for producing a powerful spark and skilled assistance in maintaining it for long hours had been at hand , which they were not .
I was compelled to use a source which would work without attention , and a quartz mercury vapour lamp answered this purpose well , though its strongest l ines are not of quite such short-wave-length as might be desired .
The source of light ( spark or mercury lamp ) was placed behind a quartz lens of -U inches diameter , arranged to focus it on the distant station .
Since * 'J .
Chern .
Soc. , ' vol. 39 , p. Ill ( 1881 ) .
t ' Roy .
Soc. Proc. , ' A , vol. 93 , p. 577 ( 1917 ) .
i Ultra-violet Transparency of the Lower Atmosphere .
261 the difference of focus for the visual and extreme ultra-violet regions is very marked , a choice must be made between them ; and since the furthest ultraviolet rays are most weakened by atmospheric scattering , the focus was adjusted for them .
In the case of the cadmium spark , the line 2313 was chosen .
Adjustment was made during day time .
The spark-gap was fixed at the calculated distance from the lens .
The image of the distant station , as seen by daylight , was viewed by means of an eyepiece , and brought on to the spark-gap .
To do this , it was essential , temporarily , to stop down the lens to a small aperture , as otherwise the focus adjusted for the far ultra-violet was too indistinct for the visual rays .
After adjustment , the diaphragm was removed , and everything left undisturbed till night had fallen .
Nearly the same method was used lor the mercury lamp .
Here , owing to the large size of the source , chromatic aberration is of less importance , and it suffices to focus for the visual rays .
The image of the distant station cannot , however , be well examined with the lamp in place , and it was necessary to substitute cross-wires , which were carried on a support like that of the lamp , and which could be brought into the same position relative to the lens by means of stops .
The mercury lamp was set up at Whitelands , near Hatfield Peveril Station , Essex , where it could be supplied by a private electric installation .
The longest range obtainable was across the Clielmer Valley to Little Baddow , a distance of almost exactly four miles ( 645 kilom .
) .
Over most of the range the path of the beam was some distance above the ground , and out of the reach of low-lying mists .
A small portable prismatic camera was used for photographing the spectrum .
It was provided with a 60 ' quartz prism , and a quartz lens of 1 inch aperture and 5 inches focal length .
A small telescope with crosswires was fixed at the proper angle on the top of the instrument , to serve as a finder .
When the distant light was on the cross-wires , the spectrum was in focus .
The plate holder had of course to be sloped , to allow for the chromatic aberration of the lens , and this made it necessary to recover the exact direction of the source for which the instrument had beeu adjusted originally .
The camera was designed for broadening the spectrum by slow tilting in the course of the exposure , but this was not found advantageous in practice .
Broadening in this way means loss of intensity , and is not required for a bright-line spectrum.* The monochromatic images appear as round dots , not lines , but this is no particular disadvantage .
* It is , of course , much more necessary for chirk-line stellar spectra .
262 Hon. R. J. Strutt .
Ultra-violet The camera was taken each night to the place of observation and set up on a portable tripod in the open air .
Dew on the prism was sometimes a source of trouble .
Extra rapid plates were used , and in order to determine the limits of transmission , it was necessary to expose till the strong lines in the near ultra-violet were lost in the blur of over-exposure .
A spectrum of the cadmium spark taken at 3600 feet ( 1100 metres ) showed no definite indication of ozone , the whole spectrum being transmitted to \ 2313 , right through the region , near \ 2536 , where ozone absorption is a maximum .
The best of the mercury lamp spectra are reproduced in Plate 2 .
Referring to that plate , Is o. I is the spectrum of the lamp at short range , the wavelengths of some of the chief lines marked .
The extreme range ever observed in the solar spectrum is also shown by the arrow .
No. II is a mercury lamp spectrum , taken with two hours ' exposure , on a clear night , at a distance of 4 miles ( 6'45 Idiom .
) .
It will be seen that the spectrum extends as far as wave-length 2536 .
The length of this spectrum cannot be compared directly with the length of the solar spectrum , taken near the sea-level , for the equivalent thickness of air traversed in that case would be more than 5 miles , and therefore considerably more than the 4 miles range of the experiment.* We can , however , compare it with the photographs of the solar spectrum obtained by Simony on the Peak of Teneriffef at an altitude of 3700 metres ( 12,200 feet ) .
Now , at this altitude , the atmospheric pressure , reckoned by the usual logarithmic formula , would be about 0'644 of its value at the earth 's surface .
Taking the " height of the homogeneous atmosphere " at 0 ' C. as 26,200 feet , the equivalent thickness of air at N.T.P. , measured in a vertical direction above the mountain , would be 16,900 feet .
The observations are stated to have been made in August , when the sun 's declination cannot have been less than 9 ' north .
Assuming the observations to have been made near noon , as no doubt they would be , the sun 's zenith distance in latitude 28 ' north would be not more than 19 ' .
For such moderate distances from the zenith , the thickness of air traversed may be taken as inversely proportional to the cosine of the zenith distance .
This makes the thickness of air traversed not more than 17,900 feet .
This is decidedly less than the thickness of air through which the mercury spectrum was photographed .
Reduced to 0 ' C. , the latter was about 20,100 feet .
* The lie of the ground did not allow me to get a longer range , t See Cornu , ' Comptes Rend us , ' vol. Ill , p. 941 ( 1890 ) .
Lower Atmosphere , and its Relative Poverty in Ozone .
263 We have , then , the following results to consider:\#151 ; Thickness of air .
1 Ultra-violet limit .
feet .
29,000* 17,9a ) 20,100 2948 fcoiar specirum irom uctu sluci*./ * ex Solar spectrum from Peak of Teneriffe . .
\rf\#187 ; vmirv Inmn \#171 ; mfiPtriim 2922 2536 iUCrtlU V lamp apetu * 1 * This figure is a little vague , as the data for calculating zenith distance are not given .
it is evident that the lower air is far more transparent than the upper air to ultra-violet rays , if equal masses are considered .
This conclusion is opposed to that of Cornu.* His method was to observe near sea-level the limit of the solar spectrum for various zenith distances , obtaining in this way an empirical formula connecting the spectrum limit with the equivalent thickness of average air traversed ( reduced to N.T.P. ) .
He then went up to a high altitude above sea-level , and found that the increased length of the spectrum for a given altitude of the sun was in accordance with what was to be expected from his formula , on the assumption that the lower air had the same absorbing power as the average .
He concluded , therefore , that this latter assumption was justified .
The results now brought forward , however , clearly prove that it was not justified .
Cornu 's method is not an advantageous one , as great strain is thrown on the accuracy with which small variations in the position of the spectrum limit are determined .
This limit is only a moderately definite quantity , and however carefully the photographic and instrumental conditions may be defined and reproduced , the limit will depend to some extent on the amount of atmospheric haze .
These considerations may perhaps account for what I think must , in any case , be considered an erroneous conclusion , for an indirect inference that the lower atmosphere will not transmit rays of short wave-length cannot stand against direct experimental proof that it will do so .
I have spoken so far of the long-distance mercury spectrum as not extending beyond wave-length 2536 .
But there is no evidence , nor is it probable , that this is a limit in the same sense as the observed end of the solar spectrum .
It is known that the solar spectrum cannot be appreciably lengthened by multiplying the exposure many times .
But the line 2536 was barely visible in a photograph which was given one hour 's exposure , whereas it comes out clearly with two hours ' .
Circumstances prevented me from trying longer exposures , but there would be no inherent difficulty in exposing for 10 hours on a winter night .
Moreover , a more powerful lamp * 'Comptea Re ml us , ' vol. 90 , p. 940 ( 1880 ) .
*264 Hon. R J. Strutt .
Ultra-violet Transparency of the might be used .
With these and other improvements , the spectrum of the lamp might probably be prolonged to the point where oxygen absorption begins to be important .
According to the observations of Fabry and Buisson , * the line.2536 is more strongly absorbed by ozone than any other in the mercury spectrum .
Since this radiation gets through , it is evident that the apparent limit of the photographed spectrum is not conditioned in any way by ozone .
Although the ultra-violet spectrum is transmitted as far as X 2536 , yet if we compare the long distance spectrum No. II with No. Ill taken at short distance , it will be noticed how rapidly the latter falls off in intensity as the wave-length diminishes .
The exposures were adjusted to give about the same intensity in the yellow and green lines in each spectrum .
There is a cause quite distinct from selective absorption by ozone or any other special constituent of the air , which will account , at any rate in part , for an effect of this kind .
I mean the scattering of light by " small particles .
" These particles are of two kinds\#151 ; the molecules themselves and the larger particles which give rise to atmospheric haze .
The action of the former alone would not be enough to explain the observed diminution of intensity .
It can be shown from established theory that the coefficient of absorption K , for light of wave-length X , has the value XT _ , 3NX3 " where / x is the refractive index and N the number of molecules per cubic centimetre under standard conditions .
From this we find , for X 2536 , K = 2-95 x 10-\#171 ; , which gives for the transmission at 6*44 kilom .
( 4 miles ) the fraction 045 .
A similar calculation carried out for the less refrangible part of the spectrum shows that here the loss by scattering is almost negligible .
Judging by the great increase of exposure required to extend the spectrum to X 2536 , it appears that the transmission for this line cannot be nearly so much as 045 , and therefore that scattering by pure air will not account for the observed absorption .
In the lower air , however , suspended particles , small compared with the wave-length , but large compared with the molecules , produce a more important effect than the latter .
The amount of this effect is so variable * 'Journal de Physique , ' March , 1913 .
t Lord Rayleigh , 'Phil .
Mag. , ' vol. 47 , p. 375 ( 1899 ) ; see also Schuster 's 'Optics , ' 2nd ed. , p. 326 ( 1909 ) .
Lower Atmosphere , audits Relative Poverty Ozone .
265 from day to day that only a general notion of its magnitude can be obtained from known data .
The measurements of Abbot and Foul * on solar radiation for different altitudes of the sun at Washington and at Mount Wilson , show that 1 mile of air measured vertically from sea-level on average clear days transmits 0-745 of the radiation of wave-length 3987 .
This makes the coefficient of absorption of such radiation T83 x 10"(1 cm .
ignoring changes of refractivity , and assuming as a rough approximation that the absorption coefficient varies as \"4 , we get ; for X 2536 the coefficient of absorptionT125 x 10"5cm"1 , and the transmission of this ray by 6*44 kilom .
( 4 miles ) of air would be only 0-0007 .
This value , which can only be considered as giving the roughest idea of the true transmission under the actual weather conditions of my experiment , would perhaps more than account for the observed enfeeblement of the line 2536 , even if no ozone were present at all .
Making use of Fabry and Buisson 's values ( lac .
til .
) for the absorption of \2536 by ozone , we find that this computed reduction of intensity ( 0-0007 ) , actually due to atmospheric scattering , could be produced by a layer of ozone 0 26 mm. thick .
Although it appears from what has been said that the experiments afford no positive evidence of any ozone in the lower atmosphere , yet they do enable us to give a maximum estimate of the ozone content , though one that may be much in excess of the actual value .
It is impossible to estimate from the present observations how much of the enfeeblement of the ultra-violet spectrum is due to ozone , and how much to scattering .
To do this would require very elaborate photometric studies of the transmission of the various wave-lengths , and even so the results would-probably not be very secure.f bet , us suppose , however , for a moment that scattering were put out of account .
If X2536 were enfeebled by ozone only how much ozone could be present ?
To determine this , spectrum photographs were taken through known thicknesses of the gas .
An ozone tube was arranged with quartz windows cemented on with sodium silicate , and a sulphuric acid manometer , as shown ( see figure ) .
The quartz windows were 18 mm. apart .
The ozonisei was filled with oxygefi , allowing the gas to wash through , and to bubble out through the acid below .
The pressure was then reduced by * 'Ann .
Astr .
Obs. of Smithsonian Institution , ' vol. 2 , p. 113 ( 1908 ) .
t In any case , nothing could be attempted in this way without establishing a station from which the distant light could be conveniently observed whenever the weather was suitable , over a long period .
I had to be content with a much less ambitious programme , working without assistance during a short summer holiday in the country .
VOL. XC1V.\#151 ; A. Y Hon. K. J. Strutt .
Ultra-violet Transparency of the 1^ ?
light V. INDUCTION COIL MANOMETER TUBE slight suction so that the acid stood at a convenient level in the tube .
The discharge was passed for a short time so as to produce ozone , and when the temperature had settled down , the level was read again in order to estimate the contraction .
The volume of the whole apparatus was measured as 12 c.c. Calibration of the narrow tube gave its volume as 32'5 c.mm .
per centimetre of length .
From these data it may be calculated , taking change of both pressure and volume into account , that 1 cm .
rise of sulphuric acid level means 0890 per cent , by volume of ozone present in the oxygen .
Hence , in the distance of 18 mm. between the quartz windows , each centimetre rise of the acid level means a thickness of pure ozone ( at N.T.P. ) equal to 0152 mm. A small metal diaphragm was illuminated by the mercury lamp , and the ozone apparatus which has been described was placed in front of it .
The spectrum was photographed by the prismatic camera , placed about 15 metres off .
The exposure was adjusted so that the yellow and green mercury lines , which are not much affected by ozone absorption , had the same intensity as in the long distance experiments .
Without any ozone the spectrum No. Ill was obtained .
It will be observed that the line 2536 is lost in the blur of over-exposure .
Spectrum No. IV was obtained with a rise of the sulphuric acid equal to O'7 cm .
, hence with O 'll mm. of pure ozone in the absorbing layer .
The whole spectrum is still transmitted , but with considerable enfeeblement in the region about 2900 .
No. V was obtained with a rise of the sulphuric acid equal to 1*8 cm .
, hence 0'27 mm. of ozone in the absorbing layer .
The line A 2536 is just visible on the negative , but will probably not come out in the reproduction .
It is decidedly less intense than in the long distance spectrum No. II .
A slight increase in the ozone beyond this was found to obliterate A , 2536 altogether .
We may conclude then that even if the low intensity of A 2536 in the long distance spectrum were wholly due to ozone absorption , it would be accounted for by less than 0*27 mm. of ozone in 4 miles of air .
SULPHURIC ACID W We have already seen that it is quite probable that an effect equivalent to Lower Atmosphere , and its " Poverty Ozone .
0-2o mm. ozone is really due to atmospheric scattering .
The close agreement of the two figures is , no doubt , largely accidental , but still , allowing for the somewhat uncertain deduction to be made for scattering , it cannot be said that any undoubted effect remains to be attributed to ozone absorption .
In any case it is certain that the ozone cannot exceed 0'27 mm. Now Fabry and Buisson ( Joe .
cit. ) , observing the solar intensity at A. 3000 for various zenith distances , concluded that the observed effects would be explained if the light of the vertical sun passed through the equivalent of 5 mm. of pure ozone , measured at N.T.P. At this rate , for 4 miles of average*air at N.T.P. , the ozone would be equivalent to 4 mm. of the pure gas , at least 15 times the amount that can be present in 4 miles of air near sea-level .
It appears clearly , therefore , from comparison of optical data only , that the air near the ground contains very much less ozone than the average for the whole atmosphere .
So far as the optical evidence now available goes , this might be attributed to local impoverishment of the air near ground-level , due for instance to the oxidising action on organic matter ; or alternatively it might be due to a stratum rich in ozone , on the outer confines of the atmosphere .
A repetition of the mercury lamp experiments at a high altitude on the Alps would help to decide between these alternatives .
The same general conclusion as to the relative poverty in ozone of low-lying air has been drawn previously , on the strength of chemical determinations of ozone , both by Fabry and Buisson and by Hayhurst and Bring* and also by Pring.f The chemical results of the various experimenters are , however , so hopelessly discrepant with one another that the purely optical method seems to afford a much-needed confirmation .
It is wholly free from the main difficulty besetting chemical determinations\#151 ; I mean the difficulty of distinguishing between ozone and other chemically active substances which may be present in the air , such as hydrogen peroxide and oxides of nitrogen .
Summary .
1 .
The lower atmosphere is found to be comparatively transparent to ultraviolet light .
The line X253G can be detected in the spectrum of a mercury lamp 4 miles distant .
2 .
The solar spectrum , even when observed from high altitudes when the equivalent thickness of air overhead ( reduced to N.T.P. ) is less than 4 miles , is limited by atmospheric absorption to X 2922 .
Air near the ground-level is , therefore , much more transparent to ultra-violet light than the upper air .
y 2 * 'Chem .
Soc. Trans. , ' vol. 97 , p. 868 ( 1910 ) .
t ' Proc. Roy .
Soc. , ' A , vol. 90 , p. 204 ( 1914 ) .
2G8 Ultra-violet Transparency of the Lower 3 .
Since the limitation of the solar spectrum is almost certainly due to ozone , it follows that there must be much more ozone in the upper air than in the lower .
4 .
Scattering by small particles acts in the same way as ozone to absorb ultra-violet radiation from a distant source , and this action makes quantitative estimation difficult .
Even if the observed enfeeblement of X 2536 were entirely due to ozone , 027 mm. of pure ozone in 4 miles of air would suffice to produce it .
Taking scattering into account , the quantity is probably much less , and there is no evidence from this investigation that any ozone is present in the lower air .
DESCRIPTION OF PLATE .
I. Reference spectrum of mercury .
The arrow shows the extreme limit observed for the solar spectrum under the most favourable conditions .
II .
Spectrum of mercury lamp 4 miles distant .
The greater part of the spectrum is unavoidably lost by over-exposure , in order to bring out the ultra-violet end , which extends to X 2536 .
III .
Spectrum of the same lamp at short distance .
No ozone interposed .
This and the remaining spectra are given an exposure bringing the yellow and green lines on the extreme right to the same intensity as the long distance spectrum II .
IV .
The same , through 0T1 mm. thickness of ozone .
The ultra-violet much enfeebled , though less than in II .
V. The same , through 0*27 mm. thickness of ozone .
The ultra-violet feebler than in II showing that the ozone in 4 miles of air amounts to less than 0*27 mm. Strutt .
Roy .
Soc. Proc. , A , 94 , Plate 2 .

