SMITHSONIAN MISCELLANEOUS COLLECTIONSVOLUME 81, NUMBER 11
Kobol^i»s jFunb
ATMOSPHERIC OZONE: ITS RELATION TOSOME SOLAR AND TERRESTRIALPHENOMENA
BYFREDERICK E. FOWLE
(Publication 3014)
CITY OF WASHINGTONPUBLISHED BY THE SMITHSONIAN INSTITUTIONMARCH 18, 1929

SMITHSONIAN MISCELLANEOUS COLLECTIONSVOLUME 81, NUMBER 11
IHobokins jFunb
ATMOSPHERIC OZONE: ITS RELATION TOSOME SOLAR AND TERRESTRIALPHENOMENA
BYFREDERICK E. FOWLE
(Publication 3014]
CITY OF WASHINGTONPUBLISHED BY THE SMITHSONIAN INSTITUTIONMARCH 18, 1929
Z-^c Bovb (§aitimoxt (pteeeBALTIMORE, HD., V. 8. k.
ATMOSPHERIC OZONE: ITS RELATION TO SOMESOLAR AND TERRESTRIAL PHENOMENABy FREDERICK E. FOWLE'The reduction of the measurements of the output of radiation fromthe sun obtained at the Smithsonian station on Table Mountain,California (altitude 2.300 m.), encountered some difficulty which didnot seem to be present at the station at Montezuma, Chile (altitude2.900 m.), in the southern hemisphere. Preliminary reductions showedthe presence of a direct relationship between the values obtained atTable Mountain for the radiation from the sim and the amount ofozone above that station. A yearly march present in the Table Moun-tain solar results, together with other irregularities, were eliminatedwhen proper allowance was made for the amount of ozone al)ovc thatstation.That ozone plays an important part in the interception of radiationcoming to us from the sun, especially at the violet end of the spectrum,has been known for some time. It exerts absorption in the followingplaces in the spectrum :
'
(1) A very strong band in the ultra-violet, 0.2300 to 0.3 lOO//.with its maximum at 0.2550/x (the Hartley band).(2) A complicated group, extending roughly from 0.3100 to0.3500JU. (the Huggins band).(3) A group in the yellow and red. 0.4500 to 0.6500/x (theChappuis band).(4) A band in the infra-red between 9 and i i//.
* A preliminary report of this research was read at the 9th annual meetingof the American Geophysical Union, April, 1928 (Ozone in the Northern andSouthern Hemispheres, Journ. Terr. Magn. and Atm. Electr. 33, 151, 1928).
* Adapted, with alterations in the wave-lengths of the infra-red band, from
" The absorption of radiation in the ui)i)er atmosphere," C. Fabry, Proc. Phys.
.Soc. 39, I, 1926.Smithsonian Miscellaneous Collections, Vol. 81. No. 11
2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8lThe longer wave-length portion of the Hartley band ( i) has beenused by Fabry and Buisson ' and others to measure the amount ofozone in the atmosphere. On June /, 1920, they found an equivalentlayer of a little more than 3 mm. at normal temperature and pressure(ntp). They estimated that at 0.2800/X the ozone absorption wouldreduce the incident solar energy to lO"^'' of its entering value. Ozone,therefore, by its absorption in this band, limits the solar spectrumat its violet end as observable at the surface of the earth. Dr.Dobson ^ uses this band for measures both of the amount and the
NO. II ATM OS I'H K R 1 C (Y/A)N E FOWLEparent to radiatiuii out-going from the earth. It was ol).serve(l in thelaboratory by Ladenburg and Lehman,' and bv the writer in thesolar spectrum."A set of atmospheric transmission coefficients, freed as carefullyas was possible from the effects of non-selective absorptions due towater vapor, dry dust, and particles associated with water vapor andcalled wet dust, was published by the writer in earlier i:)apers.' Theobservations are .shown in figure i, redrawn from Fabry's article{loc. cit.). Cabannes and Dufay* used this data to .show that the
I'lc. 2.—Almosplicric ab.sorption in Chappuis yellow ozone band (Colanse).departures from the straight line of the points at wave-lengths greaterthan 0.4729/i were caused Iw ozone present in the atmosphere. Mak-ing the assumption that the atmospheric ozone amounts to 0.32 cm.ntp., they used the differences of ordinates between the observedjjoints and the straight line, in the region of ligiu'e i, just indicated,to calculate values of the absorption coefificients of ozone for astandard dei)th of 1 cm. ntp. Figure 2 shows the 7 resulting valuesplotted as circles and also a curve showing transmission coefificients
' Ann. d. Phys. 21, 305, 1906.
'' Smithsonian Misc. Coll. 68, i, 1917.
^ Astrophys. Journ. 38, 392, 1913 ; 40, 43.=^. iQM-
* Journ. de Phys. et le Rad. Sept. 1926.
4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8lfor I cm. ozone as obtained in the lalioratory by Colange.' Theagreement is remarkable. The layer of ozone used by Colange was
1 8 cm. ntp., and from this the above curve was computed forI cm. ntp., by Bouguer's formula.The same authors," using a somewhat similar process, later utilizedpublished observations, made by Smithsonian observers at their variousstations, for further determinations of the ozone above these stations.These data had not been corrected for water vapor; also the valueswere taken from somewhat smoothed curves drawn through theplotted observed points. Further, because of gradually progressivechanges in the transparency of the sky, comparatively few days furnishobservations which are good enough for the above treatment. Onthese several accounts the investigation just cited is not fully satis-factory. In the following discussion only the original observationsare used and they are treated by a method probably nearly independentof sky changes.The results presently to be considered are to a considerable extenta by-product of spectro-bolometric observations originally made forthe determination of the radiation emitted from the sun. X'aluesfrom about i,ooo days have been utilized. In the ordinary reductionsof this work, the ordinates of the solar energy curves (generally6 curves per day) obtained with a 6o° u. v. glass prism had alreadybeen read for about half of the days used. It has been the customto read them on our plates at abscissae, among others, of i8, 20, 22,24, 26, 28, and 30 cm. towards the violet from the infra-red band, wi,at 2/t. These places correspond to wave-lengths of 0.764, 0.686, 0.624,0.574, 0.535, 0.503, and 0.475/i., respectively- This spectrum regionincludes the yellow Chappuis band due to ozone.A preliminary futile attempt was made to use these ordinates todetermine directly the depth of the ozone band. The liand is maskedby the numerous solar lines in that part of the spectrum. IndeedFabry says : " The Chappuis bands have never been observed directlyin the solar spectrum. I have often looked for them in the spectrumof the setting sun, but have never found them."However, if the several observations of any day, made at eachplace in the spectrum at different zenith distances, are used to deter-mine atmospheric transmission coefficients, and the resulting values areplotted against the corresponding deviations, the band is strongly
' Journ. de Phys. et le Rad. 8, 257, 1927.
' Journ. de Phys. et le Rad. 7, 257, 1926; 8, 353, 1927.
NO. II ATMOSPHERIC OZONE—FOWLE 5brought out as may be noted in figure 3. This figure shows resultsfor days of great, medium, and neghgible absorption in this band.The abscissae are prismatic deviations, the ordinates, atmospherictransmission coefficients for zenith sun. As the quantity of atmos-pheric ozone may be correlated to the amount of energy cut out bythis band from the radiation coming to us from the sun, the area of
30 28 26 24 22 20 30 28
SMITHSONIAN MISCEIJ-ANEOUS COLLECTIONS VOL
Calling c the corresponding energy at the selected place in the sun'sspectrum, it may be assumed that approximately the amount ofenergy absorbed from the sun's rays by ozone is{— .e) summed for spectrum places 22, 24, and 26.The accuracy of these measurements, depending, at the greatest, ondifferences of the order of (0.890—0.860), cannot exceed i part in 30,assuming no accidental errors. Further, the measurements extendover times of from one to three hours. It is presumptuous to assumealways a negligible change in the amount of ozone during suchconsiderable times. Any change in the general transparency of thesky is probably negligible, since it would alfect both the numerator andthe denominator of the above expression. It takes only 30 secondsfor the run through the part of the spectrum used, so that the timeis short to produce differential errors within this band.Because the results presently to be given differ so considerably inmagnitude and range from the values of Dr. Dobson and those asso-ciated with him, it has been thought advisable to devote considerabletime and study to the indications of the Chappuis band.Is the discrepancy due to the presence of other atmospheric lineswithin the Chappuis ozone band? A count of the number of atmos-pheric lines, designated as such in St. John's recent revision ofRowland's Solar Spectrum Table, ^ leads to the following table
:
Spectrumrange Wave-lengthrange Number of linesHoO27-29
NO. II ATMOSPIIKKIC 0Z0X1-: FUVVLE
Fig. 4.—Abscissae, ppt. 1-hO ; ordinates O,; ; 0.47 to o.6o/^.
Fic. 5.—Abscissae, ppt. HijO ; ordinates O;; ; 0.60 to o.^O^t.
8 SMITHSOXIAX MlSCELLANliOUS COLLECTIONS NOL. 8l
vapor conies considerably later in the year than that for the area-maximum of the band, yet before the time for its maximum.A far more detailed study of the transmission coefficients in theregion of this band has been made than was possible with the some-what separated measurements in the spectrum made for the solar-radiation work. Plates for two days were reread and coefficientsdetermined for each maximum and each minimum of the solar linesvisible in the observed energy curves (tig. 6, curve a). Unfor-tunately, between deviations 20 and 22, and 27 and 28, such a processwas impossible because of instrumental contingencies. The resultingcoefificients determined independently for the two days of observationsare plotted in curves b and c. This is a useful transformation, result-ing, as it does, in a spectrum, b or c, showing only atmos])hericlines, from an energy curve like a where the solar lines are domi-nant practically to the exclusion of any indication of atmosphericabsorptions.Assuming for the time being the validity of Bouguer's formula, afurther step was taken. Entering figure 2 for the corresponding wave-length wath the transmission coef^cient determined at place 24 fromthe curve c of figure 6, the amount of ozone was determined.With this amount of ozone, and the transmission coefficients at allthe maxima and minima of the curve in figure 2, an ozone bandwas computed, using the line across the top of the band in curve cof figure 6 as the basis. The result is plotted in curve d of figure 6.The agreement between c and d is better than could be expected andis indeed remarkable. Apparently because the writer is using a purerspectrum than Colange, the deflections in curves b and r are moremarked than in curve d, but the agreement in position is satisfactory.Between deviations 26 and 30, the coefficients are too small to expectany accuracy. It seems therefore highly probable that practically allof this band as observed is due to ozone.The writer, as already stated, prefers to express the results whichfollow in terms of a quantity <fairly directly coming from the observa-tions, namely, the amount of energy cut out from the incomingsolar energy by this yellow Chappuis band. These results may beapproximately reduced to amounts of ozone (ntp.) by using Bouguer'sformula Avith the constant determined by Colange (loc. cif.) asindicated by the following table
:
Band area 30 40 1 50 1 60 70 80 [ 90 100 1 cal. X lO"Ozone . 90 I . 160 ! . 200 . 230 . 260 I . 290 ! . 320 ! . 350 I cm. ntp.
NO. II ATMOSPHERIC OZONE—FOWLE
30 28 .76V26 24 2,2PRISMATIC DEVIATIONS^ 20 18Fig. 6.
lO SMITHSONIAN' MISCELLANEOUS COLLEGTIONS VOL. 8
1
The use of Bouguer's formula is unsafe for banded absorptions,except possibly for a very pure spectrum, and as an interpolationformula. Langley ' long ago showed its inapplicability in a regionwhere quite different coefificients of absorption occur, and his logic iseven more applicable in the present case where these occur in closejuxtaposition, and in banded spectra where the resolving power iscomparatively poor. Safer substitutes for Bouguer's formula maybe employed. For instance, in estimating atmospheric precipitablewater the writer always uses an absorption curve calibrated as faras possible in the laboratory. A curve approximately of the shai)c
V cMo^ AUsorbenl: "thickness
iMCi. 7-
indicatetl in figure 7 would be expected. \\'here lines of strongabsorption occur alternately with those of high transmission, thecurve of figure 7 does not tend to approach a zero value of / withincreasing absorbent, but to l)ecome horizontal for a finite value of /.Assuming Bouguer's fornnila to hold we should have a straight line,tangent to some portion of this curve. In view of the state of affairsindicated in figure 7, we should hesitate to use Bouguer's formulafor computing the amounts of ozone, unless for data requiring verylittle extrapolation from the amounts of ozone used in the laboratoryto determine the constant of the formula. It may be that these con-siderations explain certain discrepancies l)etween Dr. DobstMi's results
' Ann. Astroplij's. Observ. Smithsonian Tnst. 2, lO, 1908.
j^O. II ATMOSPHERIC OZONE FOWLE II
and mine at the same stations. He is working at a spectrum placewhere the coefficient a in the formula,/ = /„io-"^'
is very large, ranging from about i to 4. He is therefore probablyworking far down on the nearly horizontal portion of a curve suchas is indicated in figure 7 where a large change in ozone makes acomparatively small change in the observed spectrum intensity values.On the other hand, in the Chappuis band used by the writer, thecoefficient a is so small, about 0.04, that the band is very difficult toobserve visually. Therefore we may assume that the writer is measur-ing in a band where a small change in ozone produces a great changein the observed quantity. In other words, for the amount of ozonepresent in the atmosphere, the Chappuis band is a more sensitiveindicator of changes in atmospheric ozone than that employed byDr. Dobson.With these preliminary remarks, attention may be drawn to figure 8,in which recent observations made at Table Mountain with Dobson'sapparatus, and reduced by him to cm. ozone ntp. are compared withthe writer's results as expressed in areas of the Chappuis band. Theaverage amount of ozone for this interval of time as computed by thepreceding table from the writer's results is about 0.23 cm. ntp., whileDobson finds about 0.22 cm. The range of the variation found bythe writer much exceeds that found by Dobson, but nevertheless amarked correlation exists between the two series.The writer cannot leave Dr. Dobson's work without one furtherremark about his method. He states,^ " It has been shown that thereis a close connection between the amount of ozone in the upperatmosphere and the pressure conditions in the upper part of thetroposphere and the lower part of the atmosphere," and states that.
" it is remarkable that the ozone situated at so great a height " (40to 50 km., as indicated by the results of Cabannes and Dufay, 30 to40 km. by Dobson himself) should be so closely connected withvariations of pressure much lower down."Dr Dobson^ uses two methods in his evaluation of the amountof atmospheric ozone. In the first he takes as the general atmospherictransmission coefficient
' Proc. Roy. Soc. 120A, 251, 1928.
" Mp^Not. R. A. S. 86, 259, 1926. Proc. Roy. Soc. iioA, 660. 1926.
12 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8l
1 1
—
.o
NO. II ATMOSPHERIC OZONE FOWLE 13
[ 192
e 21 1928O-0.n f1 Mkr Apr May Ju'n J^l Aafe Sep oit wiv Dec-O-Oan FeU M^ A,ir Wky dJn Jl Aug S^p Oit Niv Dec
^
B HarquaHala. Arizona, o Montezuma.Chile. *TaUle Mt.Caiifornia.
^JtMtiBrukkaros.A-frica
Fig. 9.
H SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8l
where A' = /? + 8 + ax[i is the absorption coefficient (kie to small particles,8 is the absorption coefficient dne to large particles,a is the absorption coefficient due to i cm. ozone ntp.,X is the thickness of ozone atmospheric in cm. ntp.Now it seems to the writer that the very variations with atmos-pheric pressure which Dr. Dol)Son throws into x, belong fully as
NO. II ATMOSPHERIC OZONE FOVVI.E 15
of Rayleigli. Although both these assumptions may he allowable upto a certain accuracy it seems likely that from either of them avariation dependent upon the atmospheric pressure or water vapormay have been introduced.Let us now turn to the results of observations made at HanjuaHala (altitude 1,770 m.) and Table Mountain (2,300 m.) in theUnited States of America, Montezuma (2,900 m.) in Chile, andMt. Brukkaros (1,600 m.) in Africa, embodied in the followingtable and figures 9, 10, and 11. The table gives only the monthly and
Fig. II.yearly mean^; hence the plotted points, especially in the plots of yearlymeans, figures 10 and 11, depend upon a considerable number of day'sobservations but not always every successive day. The Wolfer spotnumbers and the magnetic character values here given are computedemploying only the days of radiation observations. In my preliminarypaper, already referred to, the plots related to daily values, and evenwith the few values there utilized from the 1926 and 1927 observa-tions at Table Mountain, showed a distinct correlation between theozone, the spot numbers, the magnetic character and the flocculi forthe corresponding days.
i6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8l
THE OBSERVATIONS
Month Ilarqua Hala,Ozonearea Wolferspots Magn.ch.34
NO. II ATMOSPHERIC OZONE—FOWLE 17
SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8i
Montezuma, 1926Month No. I Ozone VVolfer j MagnODS. I area ' ' Table Mountain, 1927
Jan.
NO. II ATMOSPHERIC OZONIi FOWLE I9From the data of the i)receding- tahic and the corresponding figuresseveral phenomena are notable
:
(i) There is a very decided yearly march, as lias hcen noted l)yother observers.(a) In the northern hemisphere we may take the maximum andminimum of this march as follows
:
Maximvim Miiiiimun1 92
1
March' Sept.1922 March Nov.1923 April-May Aug. ?1924 April Aug. ?1925 (April, Dobson) (Oct., Dul)son)1926 April-May Oct.1927 April-May (Ai)ri], lUiisson) Nov. (Nov.. r)nisson)''1928 May Sept.
( 1) ) and in the southern hemisphere as follows :Maximum Minimum1923 Sept. March1924 Aug.-Sept. March1925 ? Feb.1926 Aug. April ?1927 not definite not definite.whence
:
(2) In the 3'early march the maxima and minima occur at nearlythe same seasons of the year in the northern and southern hemi-spheres, though of course not in months of the same name. Themaxima occur between April and May, the minima between Augustand November in the northern hemisphere and vice versa approxi-mately in the southern.(3) A marked correlation exists between the ozone and the W'olfersun-spot numbers for the observations of the northern-hemispherestations, as indicated in figures lo and ii The range of the yearlymeans for the area of the yellow band is from 20 to 100 (see fig. 9).
' The writer is inclined to discount the appearance of the low value in May.1921, as abnormal, possibly due to erroneous observing, and to consider thegeneral march of the curve as indicating the minimum in September. Somewhatsimilar judgments occur later in the table.C. R. 186, 1229, 1918.
20 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8l(4) 111 the southern hemisphere no such strong correlation is appar-ent between the spot numbers and the ozone. The correspondingrange is only from 20 to 30. However the errors of the readingsof the area when this is small are comparatively great ; indeed theobservations do hint a slight relationship.The writer suggests that the following considerations point to afifth deduction from the observations.The ozone present in the upper air has been generally consideredas formed from the oxygen there present by the action of ultra-violetlight from the sun. Radiation of very short wave-length (less than0.1850/X,) acts upon oxygen, transforming it into ozone. It is notimprobable that radiation of this wave-length reaches the earth fromthe sun. If so, it must produce ozone in the earth's atmosphere, butonly in the highest levels, because it cannot reach the lower strata.Radiation of wave-length 0.1850/i. is completely absorbed by 10 m. ofair at ntp., and could scarcely penetrate lower than a stratum 40 km.above the earth. On the other hand, radiation lying between 0.2000and 0.3000/X decomposes ozone, and between these two oppositeactions a state of equilibrium would be established. Since the ozone-destroying wave lengths penetrate deeper into the atmosphere, thisnaturally limits the ozone layer to a high altitude.It is possible though that another agency than ultra-violet lightworks to produce ozone. The investigations of Milne ^ and Pike
'
indicate the great probability that electrified particles gain suchvelocities on the sun that they are projected outwards into space fromthat body. Mme. Curie ' has shown that the a-particles emitted fromradium salts ozonize oxygen. Electrons with a velocity of 1.80 X 10*cm./sec.^ are capable of producing ozone from oxygen. It is alsoproduced by the silent electrical discharge.Suppose then that there are two causes at work producing the ozoneof the earth's atmosphere: One portion may then be due to the ultra-violet light from the sun, and present over both hemispheres ; theother, caused by particles emitted from the sun of such a polaritythat, when they reach the earth's field, they drift towards the northernhemisphere, above which alone would the ozone due to this last causebe abundantly present. The particles would then necessarily have apositive charge, e. g., a-particles.
' Mo. Not. R. A. S. 86, 259, 1926.
' Mo. Not. R. \. S. 88, 3, 1927.^C. R. 183.
* Franck and Hertz, Verli. Doutscli. Phys. Ges. 15, 34, 1913.
.NO. II ATMOSPHERIC OZONE FOWLE 21With the assumption of these two sources for the origin of theatmospheric ozone, several of the phenomena shown by the observa-tions of this paper fall in line, and our fifth conclusion will be :(5) (a) Due to the ultra-violet light from the sun, there is a layerof ozone, varying apparently very little with the sun-spotperiod, and situated over both the northern and southernhemispheres and showing an annual march having its maxi-mum in the spring of both hemispheres and its correspondingminimum in the autumn.(b) There is another layer formed under the bombardment ofelectrical particles (probably positive ions) emitted from thesun and showing strongly a dependence u|X)n solar activity asindicated bv Wolfer's svm-spot numbers. At the only minimumof spots observed this layer appeared practically absent, themeasurements indicating the presence of the (a) layer alone.Though the corresponding marches during the year of the ozone(which the writer proposes to attribute to the first of the abovecauses) occur in diflferent months in the two hemispheres, the seasonsof maximum and minimum are the same, namely, spring and autumn.One might be led to suppose that these variations are due to somedependence upon the annual and reciprocal marches in the twohemispheres of the air-masses through which the sun's rays couldjienetrate for the formation of ozone. Further at the tropical stationat Montezuma the sun is more nearly overhead and the air-masschange smaller, which might perhaps account for the smaller annualrange there. However the maxima and minima do not occur at timeswhen the sun is farthest from or nearest to the zenith, when therewould be the greatest and least air-masses.Another circumstance might lead to an explanation of the annual)narch and its reciprocal effect in the two hemispheres so far asconcerns the times of occurrence of the maxima and minima. An-nually, as viewed from the earth, the sun's equator reaches its greatestsouthern displacement (7.25°) about March 7, and its greatestnorthern displacement about September 8. The aspect of the sun'sdisk as seen from the earth at these epochs is shown in figure 12.Since the earth subtends only about 30', as seen from the sun, undereither circumstance, the sun would have practically the same" aspectas viewed by ultra-violet light from either the northern or southernhemispheres of the earth. However, the ultra-violet light wouldprobably be strongly scattered by the particles of the solar corona,and this annual shift of the far more extensive and considerably
22 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8l
more ol)late corona might have a difterential eft'ect on the intensityof the ultra-violet light reaching the separate hemispheres.Returning again to Dr. Dobson's results, he finds much the samevalues in both hemispheres. He now has an observer in NewZealand (1928).^ He states^ that he finds very little connection ofhis observations with the sun-spot cycle, and that little apparently in areverse sense from that clearly indicated by the writer's results. Heobtains an altitude for his ozone layer from the Hartley band at 30 to40 km., whereas Cabannes and Dufay get an altitude from the
I
N pomt; N point"
Mar.7 5LPT.8
Huggin's band of 40 to 50 km. One might hazard the suggestionthat it is not beyond possibility that the band used by Dr. Dobsoncorresponds to such a state of the molecule that only ozone formedby ultra-violet light is in the proper molecular state to efifect absorp-tion ; whereas the band in the yellow is due to a molecular state whichmeasures absorption due to ozone formed by either process. It willpresently be seen how such supposition as to two layers of ozone is inline with the conclusions drawn from magnetic data relating to twoseparate strata, the lower of which is assumed to be due to ultra-violetlight.
Observatory, 51, 381, 1928.
' Proc. Roy. Soc. 114 A, 532, 1927.
NO. II ATMOSPHERIC OZOXK FOWL!-: 23Turning to the literature of Terrestrial Magnetism, the writerwas both surprised and pleased to find decided support lent by thephenomena of terrestrial magnetism to this hypothesis of two quan-tities of ozone formed l)y two separate agencies. Furthermore, thesetwo layers were not only ascril)e(l to the same two agencies as alreadystated, but assumed to be probably separate layers.If we consider magnetically (juiet days, we find a similar yearlymarch in the magnetic elements, the maxima and minima, however,occurring somewhat later, namely, in June and December at Green-wich ; and further a regular march with the sun-spot period. Thismarch is so regular as to lead to the inference that it is due to ageneral change in the whole solar disk accompanying the sun-spotperiod. Further there occur disturbed days which seem to be connectedwith specially disturbed conditions localized on the sun's disk, for theyshow a definite tendency to recur at successive rotations of the sun'sdisk.'
" There are few facts of greater significance," writes Dr. Chapman,
" with respect to the relation between magnetic changes and the sun,than the tendency shown by the earth's magnetic activity to returnto its condition at any particular time, alter the lapse of one or moreperiods of synodic rotation of the sun."There are discordances between the succession of events with theozone phenomena and those that are magnetic, so that the events maybe confused with complications not due to the same cause, but thefollowing discussion by Dr. Chapman {loc. cit.) seemed of specialinterest
:
" These conclusions regarding the ' disturbance ' solar agent have adirect bearing on the ' general ' solar agent which afifects the regulardiurnal magnetic variations over the sunlit hemisphere. If the formerconsists of electrical corpuscles, the latter cannot do so—no meredifference of mass or sign of charge would account for the completedifference of distribution of the two agents reaching the earth.On the other hand, the apparently sole alternative among possibleionizing agents, viz., ultra-violet light, seems to accord with all theproperties which the ' general ' solar agent has been shown to possess
:
for the latter affects the sun-lit hemisphere almost exclusively, itarises from the sun's surface as a whole, and its intensity varies onlygradually, from time to time, in correspondence with the generalactivity of the sun."
Dr. Chapman, Trans. Cambridge Phil. Soc. 23, 341, I9i9-
24 SMITHSONIAN MISCEI.I.ANEOUS COLLECTIONS VOL. 8lAnd he later continues, " The facts hitherto reviewed may nextbe considered in their bearing upon atmospheric questions. Onesuch question is, Are the layers affected by the two kinds of solaremissions the same or dift'erent, and if diff'erent, what is theirrelative situation ?
" Even, a priori, it would be expected that two such differentemissions as corpuscles and ether waves will have different powersof penetration into the atmosphere, though it would not be possible,on such grounds alone, to decide whether the ' absorbing ' layers werewholly distinct or not. The magnetic phenomena, however, give afairly clear indication that they are practically distinct without over-lapping * * *," and he reaches the conclusion " that the magneticdisturbance layer is situated at a higher level than the diurnal variationlayer." He infers from this that the magnetic disturbance layer (dueto ions from the sun) is situated between 90 and 120 km. and thediurnal variation layer (due to ultra-violet light), between 10 and90 km.Dr. Chapman has added a note dated July, 1919: "In a paperread (on May 22, 1919) before the Institution of Electrical Engi-neers, and shortly to be published, I have suggested that the ultra-violet radiation * * * may be some type of gamma-radiation, andthat the corpuscles are (as Vegard has urged) alpha-particles. If boththese processes originate from radio-active processes on the sun, thegamma-rays would be exj^ected to penetrate more deeply into ouratmosphere than the alpha-particles." All of which falls in with theobservations and suggestions of the present paper.
Lord Rayleigh ' has recently published observations which relateto a phenomenon possibly allied to that of ozone. These observationsare measurements of the intensity of the auroral green line in thelight of the night sky together with similar measurements of theintensity in the spectrum of the night sky on each side of this line.McLennan ' has shown that this green line owes its existence to ametastable state of the oxygen atom. Whereas the green line is alwayspresent in the light of the night sky, the negative bands of nitrogen
' Proc. Roy. Soc. ii9A, u. 1928.
-Nature, 122, 38, 1928.
NO. 1 I ATMOSPHERIC OZONE FOWLE 25
which are an important feature of the auroral spectrum, are notusually present.Rayleigh's observations (tig. 13) apparently indicate an annualmarch in the intensity of this green line with two maxima—thesmaller maximum occurring nearly contemporaneously with the singlemaximum in the ozone march, the larger with the ozone mininnun.In the southern hemisphere, as with ozone, the months of the occur-rence of these maxima are reversed but, of course, not the season.Omitting observations made at Clarcmont which Rayleigh considersfaulty, together with those for some stations with only few observa-
+1
26 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8
1
]Mean intensityPlace.
NO. II ATMOSPHERIC OZONE FOWLE 27
In the southern hemispliere no such marked rclationsliip is noted,although one may be masked by the small range and correspondinginaccuracy in the values. The range is only from 20XIO"' to50 X io~* calories.It is suggested—and such a suggestion is strengthened by magneticdata—that we are dealing with two layers of ozone. The first is dueto ultra-violet light coming from the sun and hence existing overall the stations. The second is assumed to be due to positively elec-trified particles emitted from definitely disturbed areas of the sun.This second e fleet reasonably shows a strong correlation with theWolfer sun-spot numbers. Probably because these positive particlesare deflected towards the earth's north pole this layer of ozone isfound over the northern hemisphere stations only. At sun-spot mini-mum it is negligible so far as the present measurements indicate.All the results of the present paper are based on monthly andyearlv means. A consideration of the daily values would be anotherstory. The plot published in the preliminary paper was based ondaily values for only two years at Table Mountain. The short studythen made of the daily values would indicate that what may be saidof the connection between many magnetic values and solar disturbancesmay be said of ozone; that although with monthly and yearly aver-ages, solar spottedness, for example, goes hand in hand with theamount of ozone, yet a day of many spots may pass with no increaseof ozone and vice versa.