]\gt ; INDEX SLIP .
GODPEEY , Charles.\mdash ; On the Application of 's Double to Optical Problems .
Phil. Trans. , , 1900 , pp. 829-362 .
, Natural , tion by CircuIar of random sequence of pulses .
, Charles .
Phil. Trans. , , vol. 195 , 1900 , pp. 329-362 .
Spectrum of Incandescent Ga ; Widening of Lines of .
GODrREY , Charles .
Pltil .
Trans. , , vol. 195 , 1900 , -362 .
Vibrator , Effect of Natural Light on .
GODFREY , .
Phil. Trans. , , vol. 195 , 1900 , ) .
[ 32 ] X. On the ])of Fonrier .lntcgrals to CHARLES Stu erxity of by ofessor J. J. Received Jnne Bettd J , 1899 .
CONTENTS .
INTRODUCTORY : Page SS1-3 .
Light .
tion I Circular unctions . . . .
331 S4 .
Only -polarised Light to be con sidered . . . . . .
} 85 .
GouY on " " Luminous Motion\ldquo ; . . . . .
S6 .
Integral . . . . . . .
332 S7-8 ' Criticism . . . . .
333 THE FUNDAMENTAL THEOREM : $ 9-10 .
SCHUSTER'S Theorem . . . . .
334 S11 .
Discussiou to be confined to " " \ldquo ; Light . . . . . .
335 S12 .
Our cognisance of Light is ) effects over a length of time which depends upon the means by lvhich the Light is being detected . . . .
$ 13-17 .
All observable effects of Constant Light may be estimated by summing the effects of the mo1lochromatic elements into which it is resolved by Double Integral . . . . . . . . .
335 RADIATIONS COMPOSED OF A RANDOb1 AGGREGATE : SS18-20 .
Radiation of Incandescent Gas .
LonrlEL't{ theory of widening of spectrum lines as due to damping of molecular vibrations .
Trlohlso\ldquo ; s theory of Rontgen puIses . . . . . . . . . .
338 21 .
The effect of a sequence of similar pulses as tre Lord RAy IGrI .
S22 .
questions suggested .
Degree of crowding of the ) ulses . . . .
339 .
The sequence of pulses will be equivalent to light of composition determined by the Fourier analysis of a single pulse , subject to the condition that there are many compleie pulses in the smallcst observable interval of time .
Difficulty connected with long waves . . . . . .
339 S28 .
Effective width of a pulse whose strict mathenlatical e infinite width . . . . . . . . . . . . . . . .
34 S29 .
Crowded pulses whose width is . . . . . . .
342 S30 .
Composition of aggreg of unequal pulses .
343 VOL. CXCV .
24.12.1900 .
MR. C. ( GODFREY ON THE APPLICATION OF FOURIEIt'S RoNTGEN RAys AND ORDINARY LIGHT : S31 .
Professor J. J. THOMSON'S theory of Magnetic Pulses .
Lord RAYLEIGH'S suggestion that these pulses will be equivalent to simple waves of short length . . . . . . . . . .
343 S32 .
The energy distribution to which the Thomson pulses are equivalent 343 SS33\mdash ; 35 .
nergy chiefly in the ultra-violet , but perhaps one-thousandth of it in the visible spectrum .
Other considerations . . . . .
344 .
SIR GEORGE STOKES'S theory .
Absence of Diffraction . . . . .
344 TION OF AN INCANDESCENT GAS ; WIDTH 0F SPECTRUM : S37 .
Plane Waves .
No absorption taken into account . . . . .
345 .
Doppler effect .
Mutual action of molecules during encolmter . . . . .
S39-40 .
Finite length of train of waves emitted by a molecule between two collisions .
This cause will affect the width of the spectrum line when the radiation of the gas is analysed . . . . . .
345 S41 .
Fourier analysis of a train of sine-waves . . . . .
346 S42 .
Summation for all the molecules of the gas .
Expression for the total radiation . . . . .
347 S43 .
MICHELSON'S Experiments on the Visibility of Interference Fringes .
Visibility function for present case . . . . . .
349 S44 .
Limiting case of zero pressure . . . . . .
350 S45 .
Practical evaluation of integrals . . . . .
351 SS46-48 .
Comparison of present results wibh the widths observed by MICHELSON .
Observed widths in all cases greater than calculated widths .
ison of esent resuIts with the resuIts deduced from DOPPLER'S principle by Lord RAYLEIGH .
Spectrum line now foumd to be similar in structure , bnt 10 per cent. narrower . . . . . .
35 $ 49-53 .
Justification of preceding result . . . . . .
354 or DAbIPING ON WIDTHS OF SPOCTRUNI LINES : SS 54-65 .
LonrmEL'S Theory . . . . .
355 S56 .
Fourier analysis of damped sine-wave . . . . .
355 SS57-58 .
Estimate of magnitude of damping effect . . . .
356 CHARACTER OF ETHER-MOTION IN NEARLY HOMOGENEOUS S59 .
Interference fringes with white light ; with nearly homogeneous light 357 860 .
Nearly homogeneous light may be regarded as a simple )ration with slowly-varying amplitude and phase .
Connection of this rate of variation with the width of the spectrum line . . . . .
357 S61 .
Interference phenomena looked at from the two points of vie . . . .
358 S62 .
Interference as limited by the lengths of the free paths of the molecules in the incandescent gas . . . . . . . . . . . . . .
358 NATURAL LIGHT ON A VIBRATOR : S63 .
Equation of motion of the vibrator .
Effect of light on vibrator depends on the spectrum of the light and on the constants of the vibrator . . . . .
359 S64 .
SELLMEIER'S treatment of the effect of the irregularities in nearly homogeneous light .
His erroneous conclusion that the natural vibrations of the molecules will not be aroused . . . . .
359 TNTEGRALS TO OPTICAL 3 .
xamination of his argtlment . . . . .
.360 69 .
a fricfionless vibrator approximately monochromatic light will arotlsc periods of tion in tensity . . . .
} SS70-71 .
Sol of the by lncnns of . . . . .
} INTRODUCTORY .
S1 .
THE )ject of the following work is to make some progress with the mathematical representation of the motions which go to compose natural light , S 2 .
It has always been recognised that interference phenomena forbid us to regard any natural radiation as consisting of an unending train of simple waves , such as may represented by sine functions .
At the same time , the equations of optics find their simplest solution in circular functions .
It is desirable to enquire how far we may resolve a natural luminous motion with a sum of simple wave-trains by means of " " Theorem of Double Integrals This ) cedure was first suggested by S3 .
have often been entertained as to the permissibility of this process .
Writers have been sceptical as to the physical meaning and independence of the simple waves thus introduced .
In the following pages will be found an attempt at a strict justification of the nlethod .
It is based upon two principles that we are cognisant of light only by means of the integral effects produced by the light during an interval of time which depends upon the nature of the detector in use ( the eye , photographic plate , \amp ; c. ) ; ( ii .
) that we are not concerned with simple wave-lengths , but rather with short ranges of wave-length , whose integrated energy we observe .
The former principle is generally accepted ; the latter has been put forward with great force by Lord RAYLEIGH .
S4 .
In what follows we shall deal solely with planle and plane-polarised light .
S5 .
The matter at issue cannot be introduced better than by a quotation from GolJY:\mdash ; ' ' On sait que la theory ondulatoire , les explications qu'elle donne des phenomenes optiques , a pour objet immediate le mouvement simpfe , dans lequel la vitesse vibratoire d'un point quelconque est donnee par un cquation de la form Jonrn .
Physiqnc , Lord , ' Phil. vol. 27 , 1889 .
, ' J. de Ph ser. 2 , vol. 5 , p. 354 .
It is clearly immaterial whether we spcak of velocitics ccments of clastic medium , or of electric and magnetic forces .
The real ) jects of discllssion vcctors , which can be interpreted in yarious ways .
2 .
112 signant le temps , ) , et des constants .
Cette equation definite un suite entierement illimitee de vibrations pendulaires , d'une bsolue , dont la periode est " " Si ces conditions de regularity et de duree no se trouvent pas rigoureusement alisees , l'equation du mouvement est different , et , par suite , un problem nouveau se trouve pose , la solution fournie par la theory pour un mouvement simple n'etant applicable en .
S'il s'agit , par exemple , d'un phenomene d'interference ou de action , on voit immediatement que tout irregularite et tout interruption entraine un perturbation dans l'effet produit , comme on l'a remarque depuis temps .
Il en sera encore de , abstraction faite des interferences , toutes les fois que l'on aura a considerer des milieux doves de dispersion .
En effet , le mouvement vibratoire dansles divers milieux devant toujours satisfaire aux equations differentielles des petits mouvements de ces milieux , il n'est nullement permis de fair subir un alteration , quelle qu'elle soit , au mouvement vibratoire et de supposer en suite que ce mouvement se comportera , dans les Inilieux doves de dispersion , colnlne s'il n'etait pas altere .
Ainsi , par exemple , on n'est pas en droit de supposer le mouvement no comprend qu'un nombre de vibrations limit , et qu'il se agera dans les divers milieux comme s'il formait la suite entierement illimitee definite l'equation .
" " D'autre part , on a remarque ' depuis longtemps qu'aucnne source lumineuse no peut produire un serie de vibrations indefinie et parfaitement reguliere , no t-ce qu'en raison du renouvellement incessant des particules incandescentes .
Ainsi aucun mouvement lumineux reel , meme le moins complexe qu'on puisse supposer , no rentre rigoureusement dans le cas du mouvement simple que considere la theory ondulatoire .
" " Cette difficult est presentee dos l'origine de cette theory .
On repond en upposant que les sources lumineuses produisent des series de vibrations , mais troublees de temps a autre par des perturbations subites ou de court duree .
Si la se'rie , entre deux perturbations , comprend un grand nombre de vibrations , on peut prevoir que l'effet moyen d'un pareil mouvement differera peu de celui d'un mouvement simple .
Mais cet apercu , reposant sir un hypothese , no peut ervir de base a un etude rationnelle du problem qui nous occupe , et nous verrons que , pour les sources donnant des spectres continus , on peut se fair un idee beaucoup moins etroite au mouvement lumineux .
De plus , cet apercu no nous apprend rien sir les effets des perturbations else emes , qui paraissent jouer un role important dans la constitution des spectres fournis par les vapeurs et les gaz incandescents .
The general process to which GOIJY alludes is the analysis of any disturbance whatever by means of FOURIER'S theorem .
He considers a function which is defined within a given interval of time ; this is analysed into a sum of circular functions of time ; the periods of the terms being the interval itself and all sub-multiples of it .
DOUBLE INTEGRALS TO OPTICAL PROBLFMS .
It seems unnecessary to restrict the function by definin it for a finite interval alone .
We may at once discuss a motion given for all values of the time , to By the theorem of FOURIER'S double where ) Tlue disturbance is tOrus analysed into a sum of elementary } ) , of which is typical .
Each of is a ) circular function ; the esults of the undulatory are directly applicable to it .
The periods of the elementary vibrations have all yalues zero to infinity .
Now , if the disturbance could be analysed as a number of ] circular functions with distinct periods , the separate elements would have meaning , as in the harmonic analyses of tides , vibrations of musical instruments , \amp ; c. The question we have to answer is , have the simple elements meaning in the limit , when their number is infinite , and the sum becomes an integral ?
S 7 .
We will at once notice an obvious criticism ; this was , in fact , offered by POINCARE , soon after Gouy 's article appeared .
Each of the component vibrations exists unchanged through all time .
This is true whatever be the nature of the disturbance we itre analysing .
But this disturbance may , for instance , be zero , except within a certain definite interval of .
Take the case of a flash of light .
Now a spectroscope , says M. POINCARL ' , will separate the component vibrations laterally ; they may be examined separately .
Hence a spectroscope will enable us to see the an ilne 1 ) efore it is dled , an infinite time after it is extinguished .
The analysis ltlust therefore be acious .
The answer to this objection is as follows .
No spectroscope possesses1 infinite analysing power .
A given point at the focus of the telescope will be illuminated by light of a whole range of periods .
Or , to look at the matter from another point of *On of mathematics the )ility of FOURIER'S theorems to is ) ject to certain limitations .
These limitationls exist when the metions possess infinitc sets of ities , or sets of tions , infinities of types .
Now , in concrete physical cases , we find Ilcither infinities nor disco1ttinuities .
It is true that infinities and discontinnities occur in colnmonly nsed to represcnt quantities .
But the presence of such tnres is due to the racter of the method ; a fnnction more cIosely the propcrties of the physical quantity qncstio1t wonld ) ithont infinite ' Jiscontinuous tures .
POINCARF ' Spcctrcs \ldquo ; ' C. 120 , pp. 757-7G2 , 189 ; sco , ' ' 130 , , 1895 , .334 MR. C. ON THE APPLICATION OF FOURIER'S view , a perfectly monochromatic train of waves will , by virtue of diffraction , minate , not a point , but a small area of the focal plane .
The different elements of the Foul will not be distinguished separately ; they will to some extent be superposed and recombine .
The result will be , at each point of the focal plane , disturbance not altogether different from the original motion in duration and character .
We see then that the Fourier alySiS may after all have meaning and , and not lead to a paradox such as PoINCARE anticipated .
It must be noticed that this recombinatio of the different elements of the integral is essentially connected with the phase-relation which exists between the said simple elements .
8 .
We have seen that objection will not prevent us from regarding the original ether-motion as mathematically to the Fourier integral .
Whatever services the analysis can render us we may safely accept .
It will be found that the different simple elements of the Fourier integral cannot in be said to have any independent physical existence .
On the other hand , part of the following essay is an attempt to prove that in certain cases the different Fourier ] ements can be regarded as having such physical existence .
A special case of this nature is that of a steady emission , such as the radiation of an incandescent gas .
We shall inquire to what extent such radiations are equivalent to mixed light , a continuous spectrum of composition determined by Fourier analysis .
The Theorem .
S 9 .
We will now introduce a theorem proved by fessor S. A ticular case was given Lord The theorem is as :\mdash ; where In there words , if ut * Professor SCHUSTER , ' Phil. Mag vol. , p. 533 , 1894 .
Lord RAYLEIGH , ' Phil Mag vol. 27 .
1889 .
DOUBLE RALS TO OPTfCAL OBLLhIS .
where ) functions of then ) The integrated value of depends therefore upon the distribution of in the separate elements , and upon the difference of of corresponding elements in the integrals .
S 10 .
SCHUSTER needs the theorem in order to prove that , in case of bifurcated beam oflight in terfel.ing with itself , " " the amount of interference depends on the distribution of energy only , and not on any assumption respectin the regularity or of vibration The proposition has , however , very much wider S 11 .
For the present , the discussion will be confined to the ticular case of bstant light , .
light which does not present any fluctuations othel ' time-features .
S 12 .
All cognisance of radiation other than the long waves of HLRTZ is by average effects .
We average ovel a length of time great compared with the periods of vibration .
This is true whatever be the means used to perceive and register the diations , whether by direct visual pel.ception , or by chemical effect , photographic or other , or by heating eHect ( , or by luminescence which the radiation excites .
For Hertziall waves , on the other hand , the features of a single wave be discovered .
In discussing the qualitative effect of constant light , with a view to discriminating between different wave-lengths , we are concerned solely with the integral over a interval of time .
It is doubtful how far we at liberty to consider the molecule as a simple vibrator .
But , in so far as this assumption is justified , we may prove that the observed effects of constant light will depend on nothing but the partition of energy lnong the different elements of the equivalent Fourier integral .
The will of course determine whether the light shall be " " constant\ldquo ; no ; but , if that condition is fulfilled , their further influence will not be perceived .
S 13 .
Energy.\mdash ; The of the light motion depends upon ) By the theorem of SCHUSTER RAYLhlGH , this is equal to where .
C. GODFREY ON THE APPLICATION OF path difference ends on Now Influ nce of Light on : of action , dispersion and absorption , can be explained ( subject 10 above vation ) by considering the ) oflight waves on a vibrator .
tion of motion of the vibrator is The " " \ldquo ; solution of this equation will be of the form , bitraIy .
The above solution is to hold for time tu .
We must therefore put the complete solution be of ( energy of th motion excited will depend upon , the work done by the light depends upon .
Applying SCHUSTER'S , ' ' vol. 37 , p. 533 .
hal this is solution .
rfhis involvef , the process of differentiating inside the integral .
Now , the condition that inter of being so treitted is , that the function represented shall } ) free from discontinuities and shall vanish at .
No mathematical discontinuities will in a physical problem ; and , if necessary , the conditions at infinity may be satisfied by troducing into } ( such as where is small ) which shall ensure dying away at both extremities , ttnd which at the same time ffect the Fotlrie ' resolntion in any marked GRALS TO .3.37 theorem , we find that the terms connecting different periods drop out ( a familiar property of ordinary harmonic analysis ) , and the of bsorption is dependent on The phase has again disappeared .
effects are directly dependent upon absorption .
So again with physiological effects .
S 14 .
In the case of chemical and electrical effects produced by light we probably have some kind of dissociation .
This is perhaps true of luminescence also .
It may be that we do not yet understand the mechanism of dissociation .
But , if the dissociation arises from separation of ions as their light-excited vibrations become large , the vibrator analogy will apply here as well .
Doubtless some molecules will split up sooner and others later ; for the individual molecule the precise timing of its owll vibrations with the phase of the incident light will be all-important .
But on the average of a large number of molecules , the amount of dissociation will perhaps depend on the rate of absorption of energy by a vibrator typifying the average structure .
It is necessary to repeat that our assumption of a linear equation for the vibration lnolecule cannot be regarded as more than a first step towards a solution of a difficult problem .
In the words of Sir GEORGE STOKES , " " Linearity applies to the small disturbance of the single elastic medium\mdash ; the it does not follow that linearity applies to all the roduced in a complex system of molecules S 15 .
Let us considel the application of the treatment to the spectroscopic analysis of light .
The light emergent the instrument in a given direction is compounded of diflerent wave-lengths .
The element of the integral will contribute a coml ) onent this expression depends upon ] the structure of the instrument , and the direction chosen , as well as upon .
The change of phase depends upon the same causes .
We must note , however , that neither nor depends upon The emergent light will be Now we have seen that the phase only enters as determining the constancy of the .
If the light which comes into the instrument is constant , so also is the emergent beam .
The phase has no furthel part to play ; hence , the ) VOL. OXCV . .
2 X MR. C. GODFREY ON TH1 CATION OF FOURIL 3 of constanc light depends solely tloe the distribution of lements of the integral .
S 16 .
A dispersive medium , apart from its possible selective absorption of the different wave-lengths , will always alter the relative phases of the different elements .
The transmitted light will thus be altel.ed .
But the preceding work has shown that these phase-changes will not the sensible properties of the light .
S 17 .
We have lrrived at the conclusion that the different simple components of constant light not only } ) osed , but also independent as regards all enel.gy diations cof Aggregat of Pnlscs . .
A .
problem in optics is that of the composition of an sequence of pulses of a given type .
The question occurs in dealing with the radiation of an incandescent gas .
The pulse here consists of the train of waves given off by the molecule during its free pa.th ; after an encountel the train will be entirely changed , and practically independent of the former train .
Again , what is the total effect on radiation of the damping to which the vibrations of the molecules are subject ?
The question was raised by LONMEL .
* This author was content to analyse as a Fourier integral , and assume that the different elements are independent .
This , of course , will not be true for the simple pulse which LOMMEL considered .
It is true that the motion can be reconstructed by means of an infinite series of vibrators whose amplitudes follow the law of the expansion .
But the phases of these vibrators will not be independent ; they must be carefully adjusted to give the requisite effect .
S 19 .
We shall find that , when we deal with an infinite and irregular succession of such pulses , the energy properties do completely specify the motion .
The disturbing influence of phase will disappear ; in the Fourier integral representing the complete motion , the phase will be a rapidly-fluctuating function of the wave-length , and all distinctive phase-properties will average out .
The omission to deal with a sequence of pulses has exposed LOMMEL'S analysis to adverse criticism .
It will seen that a.more complete treatment confirms the results which he obtained as regards the widening of spectrum lines through damping .
S 20 .
Another case in point is that of Rontgen rays .
These are satisfactorily covered by Professor J. J. THOMSON'S theory of electric pulses .
The pulses are of given type ; each one may be analysed by FOURIER'S theorem .
We find a certain energy-wave-length curve ; in dealing with an infinite succession of pulses the phase , ' Wied .
3 , 251 , 1878 .
RALS TO OPTICAL .3.39 relations disappear , and we are left with the energy curve to completely specify the properties of the sequence of pulses .
These statements will be justified below .
S21 .
The subject has been opened by Lord in his paper on " " The Complete Radiation at a given Temperature He proposes to regard this as an irregular sequence of pulses of the type Now ( 8 ) and the whole energy of the pulse ( 22 ) The intensity corresponding to the limits and is therefore " " If an infinite number of impulses , similar but not necessarily equal to ( 8 ) , and of arbitrary sign , be distributed at random over the whole rallge from to , the intensity of the resultant for an absolutely definite value of would be indeterminate .
Only the probabilities of various resultants could be assigned ; and if the value of were changed , by however little , the resultant would again be indeterminate .
Within the smalest assignable range of there is room for an finite number of independent combinations .
We are thus concerned only with an average , and the intensity of each component may be taken to be proportional to the total number of impulses ( if equal ) without regard to their phase-relations .
In the aggregate vibrations , the law according to which the energy is distributed is still , for practical purposes , that expressed by ( 22 ) .
S 22 .
This important paragraph suggests the whole theory .
But when we come to take a closer view of it , it will be that there : certain questions which still remain to be solved .
Suppose , for instance , that the elementary co1lfined to a small range of time , such as are the pulses in Professor THOMSON'S theory , how many of those pulses must be present in order to give the properties which Lord RAYLEIGH associates with an infinite ssion ?
Again , we might suppose that different consequences would follow from different degrees of crowding among the pulses .
They may be so close , on the average , that a great number of them are everywhere found oyerlapping ; or , again , they may be so thinly scattered as to be , on the average , far apart in comparison with the space occupied by each .
Experiment does not allow us to say which of these suppositions is correct ; we may enquire what is the test which determines the applicability of Lord RAYLEIGH'S theorem to the aggregate .
S 23 .
In order the sequence of pulses may , for us , be equivalent to a spectrum , 'Phil .
hIag 27 , 1889 .
an obvious condition is that the pulses shall not be so far apart as to be separately distinguishable .
Photography can fix second ; hence there must be many pulses in second .
This is a condition certainly fulfilled by Rontgen rays .
The coarser the means of observation which we use , the more thinly may the pulses be scattered .
The results which we are about to investigate may be true , for a certain radiation , in the present state of experimental science ; but will cease to be true for that particular kind of radiation when our instrumental means shall have been so improved as to enable us to distinguish structure in that radiation .
It shall be shown that this is the only condition necessary in order that a random sequence of similar pulses may be equivalent to radiation of a spectral composition given by the analysis of a single pulse .
S 24 .
We have already proved that we are concerned simply with an integral effect over a time of the order of the shortest observable interval .
If we are content to view the radiation with the eye , or to use a slow photographic plate , may be taken as great as we please .
If , on the other hand , we are investigating the radiation with the shortest possible exposure , may be reduced as far as our experimental skill will allow .
Let us examine the urier composition of a numerons sequence of random similar pulses .
Suppose the pulse to be the angle a definite function of .
We are to examine the Fourier integral , equivalent to . . .
where , . . .
define a large number of points of time distributed at random in an interval T. The breadth of the pulse is to be small compared with T. The resultant integral is Consider the quantity A. First , suppose that the time of vibration , , is small compared with or Draw from an origin lines of length making with the prime vector angles \amp ; c. Then the bounding lines of the angles DOUBLE TO PltOBLLMS .
:341 lie at random all round an origin .
The phase .
the resultant will be arbitrary , while the mean value of its modulus is Now this is the amplitude of the compound harmonic motion -t .
If we pass to a frequency , where is small , the new phases will differ but slightly from the old ; but if is fin ite or great , the new phases will differ finitely from the old , and the resultant for have no apparent connection as regards phase with the resultant for Let us consider what happens we observe this radiation .
First , we can only observe the content ) a certain interval of time , which we have taken to be ; we receive into our apparabus the total energy of all the pulses in T. Now theorem expresses the connection between the total energy of a radiation , and its expression as a Fourier integral ( see p. .3.35 ) .
If the Fourier expression for the resultant of the pulses in the present case is uf the total is We have just seen that is not a definite fUnction of , but partakes of the andom character of the sequence of pulses .
At this point we make use of principle ; that we are ot concerned with ) articular wave-lengths , but rather with the ayerage energy over of -length .
Bearing in the average value of , we see that , for practical ) oses , we have : spectrum whose intensity of energy for period is .
S25 .
The fluctuations in the energy-wave-length will be less rapid as we descend to the longer of the spectrum .
As we have just seen , when the time of vibration is small compared with , the fluctuations are so crowded as to be indistinguishable ; the eye , or sensitive plate , will take the meancurve .
But if the time of vibration is large compared with range of angle included in the set . . .
will be but small ; the resultant will possess a phase intermediate between the extreme values , and a modulus of almost ? .
As we ) , continuously to quicker vibrations , the modulus will diminish from to zero , and so ) ; the divergences of the energy curve from being no longer rapid but ( a broad and ) eoretically distinguishable scale .
It will be found , owever , that for short pulses the amount of energy in the slow * Lord , ' Phil , ' Ang .
, 1880 ; or ' Theory of ' ed. 1894 , p. 40 , MR. C. ODFREy ON THE APPLICATTON OF waves is inconsiderable .
The prepotent part of the energy resides in those quicker waves for which the energy curve is of the normal form .
( Compare the magnetic pulses of Professor THOMSON , treated in the next chapter ) .
S26 .
To ecapitulate , the pulses of the sequence will not be separately distinguishab]e ; their effect depends upon the integral of energy over an interval of time comparable with ; the phase of the Fourier element will have no further effect ; and all the observable properties of the sequence will be bound up with the energy function .
S 27 .
Hitherto it has been assumed that the interval comprises a large number of complete pulses , these being for the moment supposed not to be of infinite breadth .
In general the boundaries of the interval will find themselves in a pulse ; there will be a number of incomplete pulses near each end .
But these are few compared with the whole number ; they will not perceptibly affect the aggregate .
The spectrum is independent of as regards composition , if is large .
The total intensity , however , varies as ; thus a variation in the crowding of the pulses causes a corresponding variation of the brightness of the spectrum\mdash ; a result which might have been expected .
S28 .
Let us consider how these results are affected when the individual pulses are of infinite breadth .
Suppose that we examine the type which Lord RAYLEIGIT suggested , The displacement becomes comparatively small when the distance from the centre of the pulse is great compared with .
In fact , we are tempted to regard these pulses as practically equivalent to pulses of finite breadth or thereabouts .
Suppose that the least observable interval comprises a large number of central points of pulses .
Suppose also that is great compared with .
The interval will contain a contribution from each of the infinite succession of pulses .
But , since is small when is great , only those pulses which contribute finite displacements will affect the aggregate content of T. Now the centres of these will lie either in the interval , or at a distance from its extremities of order .
As is small compared with we shall practically be concerned only with the large number of pulses which lie almost entirely within T. We are , therefore , justified in regardin as the effective order of breadth of these strictly infinite pulses .
S29 .
If the pulses ( supposed finite ) are crowded , so that they overlap largely , we shall not find the characteristic spectrum unless the time-interval which we are investigating is large compared with the breadth of a pulse .
If cuts into many pulses , but is not large compared with the breadth of each , we shall lose sight of individual pulses , it is true ; but the energy function will be largely affected by the incomplete pulses .
In other words , if we can shorten the exposure till it is comparable with the duration of a pulse , the spectrum observed will begin to show DOUBLE TO deyiations from the normal ) ectrun taken ltuch observed by the eye .
S30 .
The extension to ates of pulses which not similar is obvious .
Suppose for instance that we have a sequence of pulses of constant displacement , the lengths of the pulses varying , while at the same time the } ) Ol.tions of different lengths are given .
The pulses of lengths between and , say , , of the whole .
They may be taken equal pulses ; suppose that they give an energy function .
Then the whole energy of the mixture is Ordinary S 31 .
explain Rontgen radiation by supposing it to consist of a succession of electro-magnetic pulses .
pulse is practically a pulse of constant magnetic force , lasting for a short time .
The thickness of a pulse is com- parable with the diameters of the particles posing the cathode strea1 ] ] .
Lord RAYLEIGH has pointed out that these pulses may be regarded as simple waves of short wave-length .
He did not explicitly consider the properties of a sion of ) ulses .
Perhaps on account of this insufliciency of statement , Professor has not fully accepted the above-mentioned view .
He has held that the Fourier analysis of a pulse has no physical .
Now this is a valid objection to the identification of the single pulse with ordinary light of any composition ever .
The different elements of the integral will possess definite ; they are no sense independent .
On the other hand , it has been in the course of the ) resent essay that the succession of will actually be equivalent to a spectrum of definite composition .
The homson pulses will certainly possess the propel .
tie of brief comparison with the shortest observable interval of time ; will be number of them in such interval ; it follows that , as the instrument averages ovel of wave-length , phase properties will be lost ; the processes of time-and wave-length averaging will efface all distinction between the succession of pulses and that mixture of light which is determined by the analysis of the single pulse .
S 32 .
We proceed to consider the energy-distribution in the scale of wave-length .
We must express as a Fourier integral a function of which is zero from to from to ; zero from to We find ofessor THOMSON , ' Phil. Mag February , 1898 .
Lord RAYLEIGH , ' Nature , ' Apri128 , 1898 .
Professor J. J. THOMSON , ' Nature , ' May 6 , 1898 .
MR. C. GODFREy ON TH APPLICATION OF Accordingly distribution of energy in the spectrum given by the succession of such pulses is that shown in the curve Fig. l.\mdash ; Energy Cnrve of Rontgen Rays .
If we take the ) articles of the cathode stream to be at least as great as molecules , .
the thickness of pulse , is small compared with the wave-length of visible light ( see THOMSON'S paper ) ; may be taken as , where is the wave-length of yellow light , say .
In the scale of fig. 1 is very near to .
It appears that long waves have the greatest amplitude ; practically the same amplitude is maintained onwards through the visible spectrum , and in fact till we approach to wave-lengths comparable with the diameter of the molecules .
S 33 .
The measure with which we are concerned , however , is not amplitude , but integral energy through ranges of -length .
Considering this , we see at once that the short waves are all-important .
The total energy of the pulse is of order The energy contained in waves of length from infinity to is of order .
Remembering that the visible includes an octave , we may say , roughly , that of the energy of the radiation will reside in the visible part of the spectrum ; and , of the rest , practically the whole in waves of length comparable with the diameter of the molecules .
It is noteworthy that waves of length equal to the thickness of the pulse or sub-multiples thereof will be excluded from the spectrum .
S 34 .
Inequalities in the thickness of the pulses will slightly modify the features of the equivalent spectrum .
Such inequalities will arise partly from the fact that different pulses arrive in slightly different directions ; they come from different parts of the glass ( an effect diminishing with distance ) .
Furthermore , the particles are not stopped at a single impact in the molecules of the glass .
It also appears that , if Rontgen rays can be made powerful enough , they will the eye as ordinary white light .
S 35 .
Professor THOMSON'S magnetic pulses are all negative .
A mixture of negative and positive pulses will present the same features except in so far as the long-waves are concerned .
If the negative and positive are present in equal quantities , the amplitude of the infinite wave will vanish .
S36 .
It is to be remarked that Professor THOMSON'S magnetic pulses differ in one inlportant respect from the thin pulses by which Sir GEORGE STOKES* has sought to * WILDE Lecture , ' .
Manchester Phil. boc 1897 .
DOUBLE INTEGRALS TO OPTICAL PROBLEMS .
explain Rontgen rays .
In the former , we have the magnetic force great and negative throughout the pulse ; in the latter , positive and negative are to be so balanced that the force integrated through the thickness of the pulse shall vanish .
On this property , together with that of the thinness of the pulses , Sir GEORGE STOKES* bases his proof that there will be no sensible action .
The consequence of this property will be zero amplitude for the infinite wave-length .
Practically this means that the energy in the visible spectrum is very much smaller than for the Thomson pulse .
It will be of order instead of ; a proportion of of the whole , instead of .
Now diffraction depends chiefly on waves whose lengths are of this order ; very much shorter waves will not be diffracted , but will penetrate matter ; and in any case would give much smaller diffraction patterns .
Pulses of both the proposed forms will be sensibly free from diffractive properties ; those of STOKES in a much higher degree than those of THOMSON .
Radiation of ndescent G S 37 .
As an example of the composition of a large number of independent pulses of uniform type , we will take the case of radiation from an incandescent gas .
We will suppose the mass of gas to be at a great distance , and to have no visible diameter ; we shall thus be enabled to co1lsider the radiation as composed of planewayes travelling at right angles to their own wave fronts .
Furthermore , the amount of gas is to be so small that the emission is not sensibly afl'ected by absorption .
The gas is to consist of molecules all having the same period of free vibration .
S 38 .
The light received by the spectator will not be homogeneous .
One reason for this is the Doppler effect .
The velocities of the 1nolecules in the line of sight will alter the period of the light received .
Another cause will doubtless be the altered vibrations of two molecules when very near to one another .
This will perhaps become important at high pressures , but we will not further consider it at present .
S 39 .
Lastly , we have to take into account the fact that the train of single waves emitted by each vibrating molecule is not infinite in length , but has a definite beginning and ending .
The effect of this cause is investigated below .
The } ) pler effect is included in the same piece of analysis .
We shall arrive at the remarkable result that the limiting width of the spectrum line when the pressure is indefinitely diminished is less by some 10 per cent. than the width calculated by Lord RAYLEIGH , who took into account nothing but the Doppler effect .
S 40 .
The vibration of a molecule will be altered by collision with another .
The velocity will also be altered , in both magnitude and direction .
The vibration received by the spectator from this molecule will therefore be suddenly and fortuitously altered in period , amplitude and phase .
The total radiation received Lecture , ' Proc. Manchester Phil. ' 1897 .
Lord RAYLEIGII , ' Phil. Mag vol. 27 , , 1889 .
VOL. CXCV.\mdash ; A. 2 Mr C. GODFREY ON THE APPLICATION OF FOURIER'S from the whole gas will consist of a great number of finite trains superposed .
We must consider those as practically independent .
It is true that each individual is connected in one respect , namely , instant of beginning or ending , with two others , the train emitted by the same molecule before and after .
But this element of regularity will be overwhelmed by the independence of the different molecules .
All we have to do is , to find the Fourier integral equivalent to a finite train of waves , to find the distribution of energy in the scale of frequency , and to sum up the energy for all possible trains .
Fourier of Train mplete Sine Waves .
S41 .
The general theorem is In the present case , except from to , within which limits Thus dco If we consider the quantities , we see that the latter attains to a maximum value ) } at , and that the former is small in comparison since is generally a considerable number .
We shall be concerned only with the values of near to ; accordingly the first term shall be neglected .
We shall then have a distribution of energy , neglecting numerical coefficients which do not alter the distribution .
Let ; then at a distance from the maximum ( being reciprocal wave-length ) we have energy proportional to where is the length of the train .
DOUBLE INT GRALS TO OPTICAL PROBLEMS .
For this single train , then , we have ] off from a ximum at according to the law ordinary pattern is the the rhtest part measured scale -length .
Summation for Molecules locities both in Line of S 42 .
For these we have a definite position of maximum brightness , and definite resultant velocity ) .
We have to integrate for the different of train .
Now TAIT* has shown that , of all moving with velocity , a fraction penetrates unchecked to distance , where We may write this function of as follows:\mdash ; . . . . .
( ii .
) , where number toms in unit volume , diameter of atom .
From this we see , of molecules moving with velocity , a ction have free paths between and Now , such a molecule will emit an undisturbed train of waves of length between and ?
, where , and V is the velocity of light .
Hence , of all molecules moving with velocity , a fraction , will give free paths between and *TAIT , ' Edinb .
Trans vol. 33 p. 72 .
Returning to expression for the energy of a single train of length , we see that with the ates of molecules now under consideration ( definite thwart and line-of-sight velocities ) we have for a proportion of energy We integrate for definite velocity in line of sight , possible velocities The proportion of molecules with thwart velocities between and is .
Hence , omitting the from ( iii .
) ( it does not affect the distribution of energy ) , we have where Lastly , we introduce possible velocities in of sight .
Here the Doppler effect enters ; the mid-point of the spectrum will be different for different .
Let be the distance from the centre of the final spectrum line ( measured , as before , in reciprocal wave-lengths ) , we have This integral , regarded as a function of , gives the distribution of light in the spectrunl .
To make further progress , we will change variables of integration from and to and .
We must remember that is a function of DOUBLE INTEGRALS TO OPTICAL PROBLEMS .
We have or , changing the order of integration , visibility Curve .
S43 .
Professor MICHELSON* has shown that , although the breadths of elementary spectrum lines cannot in general be examined directly , yet the application of his interference method enables one to much more detailed information .
The light is made to interfere with itself , at a relative retardation of the two half streams .
Interference bands are ) roduced and their " " visibility\ldquo ; estimated for diflerent values of , the path-difference .
From the visibility-curve thus constructed we can work backwards to the breadth of the spectrum line , and find out something about the distribution of light in this breadth .
MICHrLSON has shown that , if represent the intensity of light for position in the spectrum , and and then V is the visibility-function , in terms of , the path-difference .
MICHELSON , ' Phil. Mag vols .
31 and 34 .
MR. C. GODFREy ON THE APPLICATION OF FOURIER'S In the present case , and Now We may omit certain constant factors , and write .
We may still further simplify this by introducing the notation of page 347 .
omitting unnecessary factors , ( vii .
) , where To deduce V ( the visibility function ) from , all we have to do is to put in , and divide by the quantity thus formed .
Limiting Case of Zero Pressure .
S 44 .
If the pressure is very small , , the number of molecules in unit volume becomes small , and with it .
We are thus reduced to * TAIT , ' Edinb .
Trans vol. 33 , p. 95 .
DOUBLE INTEGRALS TO OPTICAL PROBLEMS .
S45 .
This expression as it stands is quite intractable .
We are enabled , however , to make progress by means of tables which TAIT* gives in his paper on the kinetic theory .
Our subject of ration is in TAIT'S notation .
On marking out the graph of by means of the values in column 6 of TAIT'S table , it becomes obvious that the general outline of the function is very near to that of Fig. 2 .
To make the agreement as good as possible we choose so that the two curves may attain their maximum for the same value of P. Now the latter function has its maximum at ; graphically we see that we must choose ( about ) ; . . .
In the accompanying figure , I. ( continuous curve ) is ; II .
( dotted curve ) is .
The agreement is so good that the error in the integral through using Il .
instead of I. will be only about 1 per cent. But .
is of the order of the errors in the observed visibilities , with which we propose to compare our results .
MR. C. GODFREy ON THE APPLICATION OF FOURIER'S Making this substitution , we have . . .
( viii .
) S46 .
The result is now in a form which we can interpret .
Lord RAYLEIGH* has worked out the width of the spectrum lines , taking into account the Doppler effect alone .
The visibility function deduced from his work in accordance with MICHELSON'S definition of " " visibility\ldquo ; is . . . . .
( ix .
) The two functions ( viii .
) and ( ix .
) are of the same form , and differ only in the presence of in ( viii .
) .
Now MICHELSON'S experiments gave visibility-curves agreeing in general character with ( ix .
) ; they would therefore agree equally well with the function ( viii .
) which has been found aboye .
Numerical Estirnates .
S 47 .
MICHELSON has investigated the lines of several gases with his interferometer .
He compares the " " half-widths\ldquo ; ( value of for which visibility is of maximum ) of the visibility-curves with the values deduced by Lord RAYLEIGH from DOPPLEB'S principle .
Unfortunately MICHELSON misquotes result , and has dropped a 2 ; in RAYLEIGH'S formula the path-difference is ; MICHELSON has taken it to be .
The last column of figures in page 294 of MICHELSON'S should be all doubled .
This would give the observed " " half.widths\ldquo ; in every case much less than the calculated values .
Furthermore , using the function which has been obtained in the present paper , where is the Napierian .
of 2 .
Hence for half-width . . .
But being the average velocity of the molecules .
Hence , instead of MICHELSON'S 'Phil .
Mng vol. 27 , p. 304 , 1889 .
MICHELSON , ' Phil. Mag vols .
31 and 34 .
'Phil .
Mag vol. 34 , 1892 .
DOUBLE INTEGRALS TO OPTICAL Lord RAYLEIGH'S expression for the half-width is Now ; hence the theory developed in the present paper necessitates a further addition of about 10 per cent. to the values calculated from Lord RAYLEIGH'S formula .
But the effect of this is quite obscured by the above.mentioned necessity of doubling the figures which MICHELSON deduces from theory .
The conclusion of his paper should be that Lord RAYLEIGH'S theory accounts for a certain of observed ?
ridths of spectral lines .
The fraction varies from to for the different substances examined .
If the distribution of energy in the spectral line be given by then and V If we construct the curve representing the energy of the spectrum on a scale of reciprocal wave-lengths , the " " half.width\ldquo ; in this curve will be .
Hence the " " halfwidths\ldquo ; in the energy-curve and the visibility-curve are connected by the relation S 48 .
Our result that 10 per cent. ought to be added to the visibility half-width means that the theoretical width of the spectrum line should be diminished by a similar percentage .
It might have been expected that , with pressure small and collisions comparatively few , the modifying effect of the curtailment of free trains would disappear , the free paths being now on the average long .
We might have expected that formula for the width of the lines would converge to that given by Lord , in which the Doppler effect alone is considered .
If the above reasoning is valid , there is no such convergence of the two theories when the pressure is indefinitely reduced ; the results derived from them differing by per cent. It is noteworthy that the present theory leads us to expect narrower lines than does Lord .
This result is certainly paradoxical , and calls for further justification .
The modification in theory has been to substitute for nlathem tically homogeneous light proceeding from each molecule , a radiation giving a certain continuous spectrum .
We could hardly have foreseen without a complete analysis that , for zero pressure , the integrated effect of all these spectra gives an intensity curve for the total spectral line steeper than before .
CXCV .
S54 MR. C. GODFREY ON THE APPLICATION OF FOURIER'S S49 .
The considerations may tend to remove the doubts which this result may arouse .
The whole set of molecules with- given velocities , thwart and in line-of-sight , and with given length of free path , will emit light of a certain spectral composition , namely , that given by the function ( p. 346 ) , being the length of the train of waves emitted during the single free path .
In theory , this will give a set of lines in the spectrum of the same pattern as the diffraction lines of a straight edge .
Other sets of molecules with other velocities and path-lengths will give other sets of lines ; the whole aggregate of lines overlapping and compounding to give such spectrum lines as actually exist , and were measured by MICHELSON . .
Now , by lowering the pressure of the gas , we may lengthen the average free path , and the average trains of waves emitted in a single flight , this lengthening being theoretically without limit .
The effect of this will be to narrow the curve , also without limit .
are tempted , when this happens , to substitute for the aggregate of curves the aggregate of their maximum ordinates .
If we do this , and also assume that the average length of wave-train is the same for all different velocities in the gas , we shall , in effect , be following Lord RAYLEIGH'S procedure , and we shall obtain his expression for the width of the resultant line .
S 51 .
On closer examination , it will be obyious that the molecules moving with greater yelocities emit , on the average , shorter wave-trains .
For , given the velocity , the mean free path is ( p. 347 ) , a function of ; while the corresponding train of waves has length It is not difficult to verify that this function of diminishes as increases .
Now , the molecules with greater velocities in the line of sight have , on the average , greater resultant velocities .
These , therefore , give shorter trains of waves , and smaller ordinates in the energy curve maximum .
But these molecules will provide light that goes towards the edges of the spectrum line .
The energy-curve will accordingly be steeper , and the line narrower thau would follow from the assumption that the mean free path is the same for two groups of molecules having two different velocities .
S 52 .
Furthermore , it is not allowable to substitute for the component curves their maximum ordinates , however steep and narrow these curves may become .
The maximum of .
If we substitute these maximum ordinates and then form an energy-curve by summing them all into a smooth curve , each will contribute energy proportional to .
But , in reality , the total connected with is .
Thus , trains of length ought to contribute total energy DOUBLE INTEGRALS TO OPTICAL PROBLEMS .
proportional to , not to .
In fact , the taller the component curves are , the narrower they are .
This consideration is overlooked if we allow ourselves to substitute the maxima of the curves for the curves themselves when these become narrow .
S53 .
This latter source of error , by which is substituted for , will not affect the shape of the resultant energy-curve if the length of train is taken to be dependent of the velocity .
But we have seen that greater velocities give shorter trains .
And this error tends in the same direction as the other ; for it gives too great prominence to the longer trains , i.e. , to the smaller velocities , which velocities send light to the middle of the spectrum line .
Hence the eHect of the error is to make the resultant curve too steep .
Accordingly , the only accurate way of investigating the limiting width for zero pressure is to form the general energy function as on p. 349 , and then to proceed to the limit by diminishing the number of molecules in unit volume .
Effect ] Damping the Widths of S 54 .
It has been urged by LOMMEL* that , whereas the vibrations of an atom are undoubtedly damped by radiation , the light emitted by a simple gas will be to some extent continuous .
The same idea has been recently developed by JAUMANN .
Both of these writers rely on a Fourier analysis of the vibration pt Their results have not been accepted generally ; as has been pointed out in the course of the present paper , their procedure will have no physical meaning for the single train of waves with which they deal .
But we have also seen that a vast aggregate of independent emissions of this will really give the result which LOMMEL suggests .
S55 .
The trains of waves emitted by the molecules of a gas cannot ictly be represented by .
As a matter of fact , such a motion as will never be allowed to go on indefinitely ; it will always be checked at a certain stage by a new collision .
Nevertheless , if the radiation is so rapid that the vibration of a molecule has generally become insignificant before the next collision occurs , we shall not be making a serious error by allowing to represent the train of waves .
We will proceed to investigate the effect on the assumption that the damping is so rapid as to allow this procedure .
S56 .
Both LOMMEL and JAUMANN make an erroneous application of theorem .
The analysis should be as follows:\mdash ; * LOIVIMEL , ' Wied .
Ann vol. 3 , p. 251 , 1878 .
JAUMANN , ' Wied .
Ann vols .
53 and 64 .
for to and pt for to ( The phase will not make any important difference in the energy-function .
) Now is small since the damping is gradual ; accordingly , both and will be of order unless is near to .
In that case , the latter of the two expressions will attain to the order .
We are justified in approximating to the extent of neglecting the former expression .
S57 .
The energy of the train of waves will depend on This function define the spectrum to which a vast concourse of such damped trains is equivalent .
Now this will be a widened line in the spectrum .
The " " half-width " " will be of order in frequency .
The half-widths which MICHELSON has observed for irresoluble lines are of order If is of this order , is finite .
Now 1 is comparable with the of vibrations .
Again , there are on the average vibrations in the free path .
In time the vibrations are reduced in the ratio ; if is finite , the reduction of the energy may be very noticeable .
S 58 .
We are thus led to conclude that the vibrations of molecules may be very considerably damped in the course of their free paths and yet no widening of lines be produced beyond what is actually observed .
It will be remembered that the kinetic theory , without damping , gave widths varying from one-quarter to two-thirds of the widths obseryed .
It seems not impossible that , for small densities , the residual width is to be ascribed to radiative damping .
When the density becomes considerable , the mutual effect of neighbourin molecules will doubtless become so lmportant as to obscure both dampin and Doppler effect DOUBLE INTEGRALS TO OPTICAL PROBLEMS .
of the dEther Motions in Homogeneous Light .
S59 .
It has been thought that the possibility of producing a large number of interference fringes from white light is an indication of a certain regularity in aether lnotion corresponding to such light .
This view has been abundantly refuted by , RAYLEI , and SCHUSTER .
* The fringes cannot be produced without the use of a spectral apparatus ; and the number of the fringes is an index , not of the regularity of the white light , but of the resolving power of the spectroscope .
A large number of fringes can also be produced without a ) spectroscope , by using radiations hich naturally possess a high degree of homogeneity .
The number of these fringes is a test of the homogeneity ; and in this case , it is also a test of the regularity of the aether motion .
S 60 .
We have justified the representation of light by a Fourier integral of the form where are functions of It can be shown that , for light of long duration , will fluctuate rapidly in terms of For approximately homogeneous light of mean frequency , we will use the notation In this expression will be insignificant , .
except for values of small compared with The above integral can be written pt ut \mdash ; pt The ranges of integration ] practically be confined to a region on either side of zero , small compared with Each of the integrals ut ut is a function of , whose variations are slow with those of pt .
The expression pt ut may be taken to denote a simple vibration of period , whose amplitude varies slowly .
The zeros of the expression will be , effectively , the zeros of pt .
statements will apply to ut GouY , 'J .
de Ph series 2 , vol. 5 , p. 354 , ' RAYLEIGH , Art .
" " Wave Theory 'Encycl .
Brit. ' ; 'Phil .
Mag vol. 27 , p. 460 , 1889 ; SCHUSTER , ' Phil. Mag. , vcl .
, p. 509 , 1894 Finaly , the motion which we have analysed as may be described as a simple vibration whose amplitude phase vary slowly .
We might equally well say that the amplitude and period vary slowly ; the latter within narl .
OW limits .
In passing it is interesting to note that all the effects of white light would be produced by an approximately simple vibration whose period varies slowly , but within wide limits ; the whole range of variation being traversed a great mbel .
of times in the course of the shortest obseryable interval of time The great betweell second ( the period of vibration ) and second ( the shortest observable time ) will give room for a rate of variation small compared with the one measure , and great compared with the other .
S61 .
Returning to the nearly homogeneous light , let be the effective range of integration in This of course means that becomes small outside the limits .
The width of the spectrum line will be of order .
Now the two integrals just quoted will give irregularly sinuous time-curves , the average extent of a sinuosity being of order Thus , the varying simple vibration will have entirely changed iu amplitude and phase after a time of order .
It will therefore be impossible to produce sensible interference with time-differences of Dlore than .
This is another aspect of the fact that for lines of width ( measured in frequency ) , the maximum path-difference for interference is of order S 62 .
We can look at the same matter from yet another point of view .
We may 90 back to the composition of the radiation from a gas .
This we have seen to be built up of finite trains of waves .
For the moment , let us omit the Doppler effect and take all the trains to be of the same period .
In an interval during which only a proportion of molecules collide , the amplitude and phase of the composite ations is but little altered .
But after an interval comparable with the mean free time of molecules , most of the molecules will have obtained new and independent vibrations ; the composite motion will be entirely altered in amplitude and phase .
For the discharge tubes used by MICHELSON this time is of order periods .
Again , isolating the Doppler effect , we deal with the superposition of infinite simple train .
We hav ' , already seen that the composite vibration will have a materially altered amplitude and phase after a time equal to the reciprocal of the range of frequency in the component trains .
For bIICHELSON'S experiments it pens that this time is again some periods .
DOUBLE INTEGRALS TO OPTICAL In the actual radiation the two causes coexist ; the joint effect is of the same order ; the radiation will attain independence after every periods .
In accordance with this fact , we have interference up to path-difference of some wave-lengths ; while the width of the lines corresponds to a fraction ' of the frequency of the light .
Effect of Natural Light on Vibration .
S63 .
We have already proved that the complete solution of is being such a function of time as can actually represent a natural radiation ( see footnote , page 386 ) .
Now we have already pointed out the hypothetical character of this treatment of the molecule as a simple vibrator .
Nevertheless the method has a historical interest , and may be regarded as a foreshadowing of the truth ; it may be worth while to sketch the result of applying our analysis .
The composition of the exciting light is of the light emitted by the molecule ' while the rate of absorption is dependent on We will take the light to be constant .
Its effect on the vibrator is seen to depend entirely upon its spectrum .
5 .
64 .
This is equally true for nearly homogeneous light .
The effect of the irregularities in the light may be deduced from the observed widening of the spectrum line .
Now SELLMEIER treated of this problem in the paper which laid the foundations of the modern theory of dispersion .
* He recognised that no natural radiation is a perfect train of simple waves , and he investigated the effect of the irregularities upon a vibrator .
The period of the light was to differ from that of the natural vibrations of the molecule .
He came to the conclusion that the irregularities in the light would , ' Pogg .
Ann 1872 , vol. 146 , p. 520 .
not arouse the natural vibrations of the molecule .
He held that the motion of the molecule would be mainly in the mean period of the incident light .
This conclusion we take to be erroneous .
S65 .
LMEIER does not adopt the Fourier method ; he builds up the incident light by means of a great number of finite trains of simple waves , all having the same phase and period , but of amplitudes , durations , and positions so adjusted as to give the actual fluctuation of amplitude which is present in the resultant motion .
In passing , it may be noticed that this arrangement will give no irregularity of phase in the light motion ; whereas we have shown that such irregularity will generally be present .
The defect may be remedied by removing the condition that the component trains shall be of the same phase ; but this consideration not alter SELLMEIER'S reasoning in any essential .
* SELLMEIER supposes the motion of the vibrator to be free from all damping .
We will show that in this case his conclusion ought to be that the natural periods of the vibrator become continually more and more prominent without limit .
S 66 .
His reasoning is as follows .
Each new train of waves , it strikes the vibrator , arouses:\mdash ; .
a vibration in the period of the incident light ( forced ) , ii .
a vibration in the period proper to the vibrator ( natural ) .
The forced vibration ( i ) will be of the same phase as the exciting train , and of amplitude proportional to that of the said train .
The natural vibration will generally have an amplitude of the same order ; its phase will , however , be different . .
The motion of the vibrator at any time results from the superposition of all the vibrations previously started .
It will be partly forced and partly " " natural The forced vibrations at any instant will clearly differ from the incident motion at that instant by a numerical factor only .
The natural vibration , on the other hand , is compounded of members , whose phases are practically fortuitous .
This is easily seen as follows .
The phase of natural vibration aroused by the beginning of a new train depends upon the point of time at which this event takes place .
The difference of a fraction of a period in the position of this point of time produces a finite change in the phase of the natural ration aroused .
But , in building up the slowly varying vibration by means of simple trains , the instant at which each start is , to a few periods , immaterial .
S 68 .
SELLMEIER concludes from this that the natural vibration will be insignificant ompared with the forced vibration .
This mistake arises from the fact that *It will be convenient to get rid of the endings of these trains .
This we may do by supposing that a train , once started , is mending ; and by introducing a train of equalamplitude and opposite phase , whose beginning is so fixed as to extinguish the former train at the right moment .
By this arrangement we shall only have to leai with beginnings .
OUBLE I TO OPTICAL he really considers the effect of only single waxing and waning of the incidenl , light .
It is true that , on the average , the amplitude of the sum of a large number of vectors of random is small compared with the sum of the amplitudes .
At the same time the energy is , on the average , equal to the of the component energies .
In the ) resent case the right deduction is , that the enel.gy of the natural vibration will vary with the numbel of the component vibrations ; in other words , will vary as the time elapsed since the light began to act .
It will become greater without limit .
It is easily seen from SELLMEIER'S analysis that there is no tendency for the natural vibrations excited by successive fluctuation to counteract one another .
As ards the forced vibrations , on the other hand , the phases are , so to speak , arranged so that there shall be no accumulation of energy .
S69 .
For a frictionless vibrator , then , common homogeneous light will give a continually-increasing motion in the natural mode ; this feature bein entirely due to the irregularities .
But if there were a friction , however small , the motion would be from mounting up indefinitely .
The vibrations started by the component trains would not persist ; in fact we may expect to find that the natural vibration settles down to a definite state , depending on the damping and the nature of the regularities .
S 70 .
whole of this matter becomes quite simple on the application of Fourie Let us first to solve We are tenlpted to take solution This expression , however , has no definite value .
The integl.and involves an infinity at ; furthermore , the infinity is of such a nature that the integral upon The above integral , in fact , is not a solution of the equation .
We are forced to include a frictional term in the equation .
But this corresponds to the actual properties of the vibrator ; we have shown that , but damping , the natural atio would continually increase ; a state of things unknown among the observed effects of light .
S71 .
On page 336 it has been shown that light of composition will excite vibration of composition .
V0L .
TION OF FOIJ 1 DOUBLE TO OPTICAL PROBLEMS .
For nearly homogeneous light , of period ' is small if is finite .
The integrand is therefore unimportant , except for values of neal to , where we may neglect , and use .
near to the emission is lf the light itted f the is nalysed by a ) , theoretically a spectrum hold be lines being at equencies p and Let the ] -width of the bright line in incident light ; then will also be the 1dth of the line in the emitted light .
The total intensity of the line in the emitted light is therefore of order The half-width of the line will be ; the total intensity of the line is of order The ratio of these intensities is - ' Now is , the incident light at being by hypothesis invisible .
On the other land and are both small .
It therefore appears that , so far as the theory of the vibrator carries us , the natural vibration may be as prominent as the forced vibrator ; the natural vibration varying inversely as the index of damping .
Whethe ] ' no it is strong enough to be visible , depends upon the spectrum of the incident light and the constants of the vibrator .
\mdash ; . .

