]\gt ; perpendicular to the length of the slit .
It appeared that if the width of the slit is very small in comparison with the wave-length ( ) , there is a much mOre free passage when the electric vector is perpendicular to the slit than when it is parallel to the slit , so that unpolarised light incident upon the screen will , after passage , appear polarised in the former manner .
This conclusion is in accordance with the observations of upon the very narrowest slits .
Fizeau found , however , that somewhat wider slits scratches upon silvered glass ) gave the opposite polarisation ; and I have wished to extend the calculations to slits of width comparable with .
The subject also a practical interest in connection with observations upon the Zeeman effect .
The analysis appropriate to problems of this sort would appear to be by use of elliptic co-ordinates ; but I have not seen my way to a solution on these lines , which would , in any case , be rather complicated .
In default 01 such a solution , I have fallen back upon the approximate methods of my former paper .
Apart from the intended application , some of the problems which present themselves have an interest of their own .
It will be convenient to repeat the general argument almost in the words formerly employed .
Plane waves of simple type impinge upon a parallel screen .
The screen is supposed to be infinitely thin and to be perforated by some kind of aperture .
Ultimately , one or both dimensions of the aperture will be regarded as small , or , at any rate , as not large , in comparison with the wave-length ( ) ; and the investigation commences by adapting to the present purpose known solutions concerning the flow of incompressible fluids .
" " On the Passage of Waves through Apertures in Plane Screens and Allied Problems ' PhiL Mag 1897 , vol. 43 , p. 269 ; 'Scientific Papers , ' vol. 4 , p. 283 .
'Annales de Chimie , ' 1861 , vol. 63 , p. 385 ; Mascart 's 'Traite d'Optique , ' S646 .
See also ' Phil. Mag 1907 , vol. 14 , p. 350 ; 'Scientific Papers , ' vol. 5 , p. 417 .
Zeeman , 'Amsterdam Proceedings , ' October , 1912 .
: where and is the velocity of propagation .
If we assume that the vibration is everywhere proportional to , ( 1 ) becomes , ( 2 ) where .
( 3 ) It will conduce to brevity if we suppress the factor .
On this understanding the equation of waves travelling parallel to in the positive direction , and accordingl incident upon the negative side of the screen situated at , is .
( 4 ) When the solution is complete , the factor is to be restored , and the imaginary part of the solution is to be rejected .
The realised expression for the incident waves will therefore be .
There are two cases to be considered corresponding to two alternative boundary conditions .
In the first ( i ) over the unperforated part of the screen , and in the second ( ii ) .
In case ( i ) is drawn outwards normally , and if we take the axis of parallel to the length of the slit , will represent the magnetic component parallel to , usually denoted by , so that this case refers to vibrations for which the electric vector is perpendicular to the slit .
In the second case ( ii ) is to be identified with the component parallel to of the electric vector , which vanishes upon the walls , regarded as perfectly conducting .
We proceed with the further consideration of case ( i ) .
If the screen be complete , the reflected waves under condition ( i ) have the expression .
Let us divide the actual solution into two parts , and ; the first , the solution which would obtain were the screen complete ; the second , the alteration required to take account of the aperture ; and let us distinguish by the suffixes and the values applicable upon the negative , and upon the positive side of the screen .
In the present case we have .
( 6 ) This -solution makes over the whole plane , and over the same plane the element of the aperture , and the integration is extended over the whole of the area of aperture .
Whatever functions of position may be , these values on the two sides satisfy ( 2 ) , and ( as is evident from symmetry ) they make vanish over the wall , viz. , the unperforated part of the screen , so that the required condition over the wall for the complete solution is already satisfied .
It remains to consider the further conditions that and shall be continuous across the aperture .
These conditions require that on the aperture .
( 8 ) The second is satisfied if ; so that , ( 9 ) making the values of equal and opposite at all corresponding points , viz. , points which are images of one another in the plane .
In order further to satisfy the first condition , it suffices that over the area of aperture , ( 10 ) and the remainder of the problem consists in so determining that this shall be the case .
It should be remarked that in ( 9 ) is closely connected with the normal velocity at .
In general , .
( 11 ) At a point ( x ) infinitely close to the surface , only the neighbouring elements contribute to the integral and he factor may be omitted .
Thus ; or ( 12 ) being the normal velocity the point of the surface in question .
In the original paper results were applied to an aperture , especially of elliptical form , whose dimensions are small in comparison with For our present purpose we may pass this over and proceed at once to consider The use of implies that the variation is in a fixed direction , while may be supposed to be drawn outwards from the screen in both cases .
through Fine in Thin Opaque Screens .
the case where the aperture is an infinitely long slit with parallel edges : whose width is small , or at the most comparable with The velocity-potential of a point-source , viz. , , is now to be replaced by that of a linear source , and this , in general , is much more complicated .
If we denote it by , ' being the distance from the line of the point where the potential is required , the expressions are* ( kr ) , ( 13 ) where is Euler 's constant , and .
( 14 ) Of these the first is " " semi-convergent\ldquo ; and is applicable when is the second is fully convergent and gives the form of the fimction when is moderate .
The function D- may be regarded as being derived fro1n by integration over an infinitely long and infinitely narrow strip of the surface S. As the present problem is only a particular case , equations ( 6 ) and remain valid , while ( 9 ) may be written in the form the integrations extending over the width of the slit from to .
It remains to determine , so that on the aperture At a sufficient distance from the slit , supposed for the noment to be very narrow , ) may be removed from under the inteolalb sign and also be replaced by its limiting form given in ( 13 ) .
Thus .
( 16 ) If the slit be not very narrow , the partial waves arising at different parts of the width will arrive in various phases , of which due account must be taken .
The disturbance is no longer symmetrical as in ( 16 ) .
But if , as is usual in observations with the microscope , we restrict ourselves to the direction of original propagation , equality of phase obtains , and ( 16 ) * See 'Theory of Sound , ' S341 .
VOL. LXXXIX.\mdash ; A. Lord Rayleigh .
On the Passage of Wavei remains applicable even in the case of a wide slit .
It only remains to determine as a function of , so that for all points upon the aperture ( ) , ( 17 ) where , since is supposed moderate throughout , the second form in ( 13 ) may be employed .
Before proceeding further it may be well to exhibit the solution , as formerly given , for the ease of a very narrow slit .
Interpreting as the velocity-potential of aerial vibrations and having regard to the known solution for the flow of incompressible fluid through a slit in an infinite plane wall , we may infer that will be of the form , where A is some constant .
Thus ( 17 ) becomes .
( 18 ) In this equation the first part is obviously independent of the position of the point chosen , and if the form of has been rightly taken the second integral must also be independent of it .
If its co-ordinate be , lying between ( 19 ) must be independent of .
To this we shall presently return ; but merely to determine A in ( 18 ) it suffices to consider the particular case of Here .
Thus , and ; so that ( 16 ) becomes .
( 20 ) From this , is derived by simply prefixing a negative sign .
The realised solution is obtained from ( 20 ) by omitting the imaginary part after introduction of the suppressed factor .
If the imaginary part of be neglected , the result is corresponding to 22 ) Writing , we have ( S ) Sb2 as we see by changing into in the second integral .
Since has disappeared , the integral is independent of .
In fact , * and we have , ( 23 ) as in the particular case of The required condition ( 17 ) can thus be satisfied by the proposed form of , provided that be small enough .
When is greater , the value of in ( 15 ) will no longer be constant over the aperture , but we may find what the actual value is as a function of by carrying out the integration with inclusion of more terms in the series representing D. As a it will be convenient to discuss certain definite integrals which present themselves .
The first of the series , which has already occurred , we will call , so that elow .
or , on integration by parts , Thus by which the integrals can be calculated in turn .
Thus , Similarly , and so on .
It may be that the series within brackets , being equal to approaches ultimately the limit .
A tabulation of the earliermembers of the series of rals will be convenient:\mdash ; Table I. The last four have been calculated in sequence by means of ( 25 ) .
Fine Stits in Thin Opaque Screens .
In ( 24 ) we may , of course , replace by throughout .
lf both and occur , as in , ( 26 ) where and are even , we may express by means of , and so reduce ( 26 ) to integrals of the form ( 24 ) .
The particular case where is worthy of notice .
Here .
( 27 ) A comparison of the two treatments gives a relation between the integrals Thus , if We now proceed to the calculalion of the left-hand member of ( 17 ) with , or , as it may be written , The leading term has already been found to be .
( 29 ) In ( 28 ) is equal to .
, as before , we have ( 30 ) As regards the terms which do not involve , we have to deal merely with , ( 31 ) where is an even integer , which , on expansion of the binomial and integration by a known formula , becomes co .
( 32 ) Lord Rayleigh .
On the Passage of Waves Thus , if .
If , and so on .
3 ) The coefficient of ( 31 ) , or ( 32 ) , in ( 30 ) is .
( 33 ) At the centre of the aperture where , ( 32 ) reduces to its first term .
At the edges where , we may obtain a simpler form directly from ( 31 ) .
Thus ( 31 ) For example , if ( 34 ) We have also in ( 30 ) to consider ( even ) being even .
In ( 36 ) S8i 1.2.3.4 and thus the result may be expressed by means of the integrals .
Thus if ( 35 ) If If ( 35 ) It is worthy of remark that if we neglect the small differences between the 's in ( 39 ) , it reduces to 4 , and similarly in other cases .
When is much higher than 6 , expressions corresponding ( 37 ) , ( 38 ) , ( 39 ) become complicated .
If , however , be either , or ( 36 ) reduces to a single term , viz. , or .
Thus at the centre from either of its forms ( 35 ) On the other hand , at the edges ( 30 ) In ( 30 ) , the object of our quest , the integral ( 35 ) occurs with the coefficient Lord Rayleigh .
On the Passage of Waves Thus , expanded in powers of 28 ) or ( 30 ) becomes { Scos } .
( 43 ) At the centre of the aperture , in virtue of ( 40 ) , a simpler form is available .
We have Similarly at the edges , by ( 34 ) , ( 41 ) , we have .
( 45 ) .
( 46 ) These expressions are the values of ( 47 ) for the values of We now suppose that .
The values for other particular cases , such as , may then easily be deduced .
For , from ( 44 ) we have , ( 48 ) since , log2 In like manner , if , we get still with .
( 49 ) If , we have .
( 50 ) ( 51 ) Similarly , if , we have And if , with diminished accuracy , .
( 53 ) As an intermediate value of we will select .
For from ( 46 ) Also , when .
( 55 ) When , only a rough value is afforded by , viz. , The accompanying table exhibits the various numerical results , the factor being omitted .
of .
In this case the condition ( 17 ) can be letely satisfied with , A being chosen suitably .
When is finite , ( 17 ) can no longer be satisfied for all values of .
But when , or even when the tabulated number does not vary greatly with and we may consider to be approximately satisfied if we make in the first case , ( 57 ) and in the second , .
( 58 ) The value of , applicable to a point at a distance directly in front of the aperture is then , as in ( 16 ) , .
( 59 ) In order to obtain a better approximation we require the aid of a second solution with a different form of .
When this is introduced , as an addition to the first solution and again with an arbitrary constant multiplier , it will enable us to satisfy ( 17 ) for two distinct values of , that is of , and thus with tolerable accuracy over the whole range from to Theoretically , of course , the process could be carried further so as to satisfy ( 17 ) for any number of assigned values of As the second solution we will take simply , so that the left-hand member of ( 17 ) is .
( 60 ) If we omit , which may always be restored by consideration of homogeneity , we have ( 60 ) same expression with the sign of changed .
The leading term in ( 60 ) is thus imaginaly part is independent of that variable .
The complete expression ( 60 ) naturally assumes specially imple forms at the centre and edges of the aperture .
Thus , when ( 60 ) ; ( 62 ) and , similarly , when ( 60 ) To restore we have merely to write for in the right-harul members of ( 62 ) , ( 63 ) .
The calculation is straightforward .
For the same values as before of and of , equal to , we get for ( 60 ) Table We now proceed to combine the two solutions , so as to secure a better satisfaction of ( 17 ) over the width of the aperture .
For this purpose we determine A and in , ( 64 ) so that ( 17 ) may be exactly satisfied at the centre and edges .
The departure from ( 17 ) when can then be found .
If any value of and the first tabular ( complex ) number is and the second , and for the first is and the second , the equations of condition from ( 17 ) are From ( 65 ) we get , ( 66 ) ' so that ( 67 ) Thus for we have whence and ( 67 ) The above values of and are derived according to ( 17 ) from the values at the centre and edges of the aperture .
The success of the method may be judged by substitution of the values for these in ( 17 ) we get , -for what should be , a very fair approximation .
In like manner , for ( 67 ) : and for ( 67 ) As appears from ( 16 ) , when is given , the lnodulns of may be taken to represent the amplitude of disturbance at a distant point immein front , and it is this with which we are mainly concerned .
The following table gives the values of Mod .
and Mod.2 for several values of The first three have been calculated from the simple formula , see ( 20 ) .
Table 1 The results are applicable to the problem of aerial waves , or shallow water waves , transmitted through a slit in a thin fixed wall , and to electric When is lalge , the limiting form of ( 67 ) may be deduced from a formula , analogous to ( 12 ) , connecting and .
As in ( 11 ) , in which , when is very small , we may take .
Thus , or Now , when is large , tends , except close to the edges , to assume the value , and ultimately ( 67 ) ( 69 ) which the modulus is simply , i.e. We now pass on to consider case ( ii ) , where boundary condition to be satisfied over the wall is .
Separating from the solution ( ) which would obtain were the wall unperforated , we have , ( 70 ) giving over the whole plane , S* and the point , at which are estimated , is equal to .
The form ( 71 ) secures that on the walls , so that the condition of evanescence there , already satisfied by , is not disturbed .
It remains to satisfy over the .
( 72 ) The first of these is satisfied if , so that and are equal at pair of corresponding points on the two sides .
The values of are then opposite , and the remaining condition is also satisfied if .
( 73 ) At a distance , and if the slit is very narrow , may be removed from under the integral sign , so that , ( 74 ) in which And , even if be not small , ( 74 ) remains applicable if the distant point be directly in front of the , so that .
For such a point ( 76 ) There is a simple relation , analogous to ( 68 ) , between the value of at any point of the aperture and that of at the same point .
For in the application of ( 71 ) only those elements of the integral contribute which lie infinitely near the point where is to be estimated , and for these .
The evaluation is effected by considering in the first instance a point for which is finite and afterwards passing to the limit .
Thus It remains to find , if possible , a form for , or , which shall make constant over the aperture , as required by ( 73 ) .
In my former paper , dealing with the case where is very small , it was shown that known than in case ( i ) , see ( 20 ) .
The realised solution from ( 78 ) is , corresponding to The former method arrived at a result by assuming certain hydrodynamical theorems .
For the present purpose we have to go further , and it be appropriaie actually to verify the constancy of over the aperture as resulting from the assumed form of , when .
is small .
In this case we may take , where .
From , the suffix being omitted , ; and herein const Thus , on integration by parts , In ( 81 ) and so long as is not equal to , it does not become infinite at the limits , even though .
Thus , if vanish at the limits , the integrated terms in ( 81 ) disappear .
We now assume for trial which satisfies the last-mentioned condition .
Writing we have ) .
Of the parts of the integral on the right in ( 83 ) the first yields when x. For the second we have to consider , from to where , and lastly from to .
In evaluating the first and third parts we may put at once .
And if Sin being omitted , the first and third parts together are thus where , and is to be made infinite .
It appears that the two parts taken together vanish , provided are so chosen that It remains to consider the second part , viz. , ' ( 85 ) in which we may suppose the range of integration to be very small .
Thus ( 85 ) and this also vanishes if , a condition consistent with the former to the required approximation .
We infer that in ( 83 ) , ( 86 ) so that , with the aid of a suitable multiplier , ( 73 ) can be satisfied .
Thus if , ( 73 ) gives , and the introduction of this into ( 74 ) gives ( 78 ) .
We have now to find what departure from ( 86 ) is entailed when is no longer very small .
Since , in general , we find , as in ( 81 ) , , ( 87 ) and for the present has the value defined in ( 82 ) .
The first term on the right of ( 87 ) may be treated in the same way as ( 28 ) of the former problem , the difference being that occurs now in the numerator instead of VOL. LXXXIX.\mdash ; A. term we have , and thus The latter integral may be transformed into and this by means of the definite integrals is found to be .
To this order of approximation the complete value is For the next two terms I find When , or , the calculation is simpler .
Thus , when ; , ( 92 ) the last term , deduced from , being approximate .
For the values of we find from ( 91 ) , ( 90 ) , ( 92 ) for , 1 , , 2:\mdash ; Table These numbers correspond to the value of expressed in ( 82 ) .
We have now , in pursuance of our method , to seek a second solution with another form of .
The which suggests itself with does not answer the purpose .
For ( 81 ) then gives as the leading term , ( 93 ) becoming infinite when A like objection is encountered if .
In this case The first part gives simply when becomes zero .
And ; so that , ( 94 ) becoming infinite when So far as this difficulty is concerned we might take , but another form seems preferable , that is .
( 95 ) With the same notation as was employed in the treatment of ( 82 ) we have .
( 96 ) Thus altogether for the leading term we get .
This is the complete solution for a fluid regarded as incompressible .
We have now to pursue the approximation , using a more accurate value of than that hitherto employ In calculating the next term , we have the same values of and as for ( 88 ) ; and in place of that equation we now have ( 98 ) The integral may be transformed as before , and it becomes ( 99 ) The evaluation could be effected by expressing the square bracket in terms of powers of , but it may be much facilitated by use of two lemmas .
If denote an integral function of ( 100 ) in which the doubled angles are got rid of .
Again , if be integral , .
( 101 ) ( 103 ) $ ?
ik ; * ' and thence , on introduction of the values of , for the complete value to this order of approximation , : .
( 104 ) To carry out the calculation to a sufficient approximation with the general , value of would be very .
I have limited myself to the extreme cases .
For the former , we have , ( 105 ) : and for the latter .
( 106 ) From these formula the following numbers have been calculated for the value of Table We may take the modulus of ( 108 ) as representing the transmitted vibrations , in the same way as the modulus of 67 ) represented the transmitted vibration in case ( i ) .
Using , as before , to denote the tabulated complex numbers , we have as the equations to determine A and so that .
( 110 ) For the second fraction on the right of ( 110 ) and for its modulus we get in the various cases And thence ( on introduction of the value of ) for the modulus of ( 110 ) representing the vibration on the same scale as in case ( i ) .
Table These are the numbers used in the plot of Curve , fig. 1 .
When is much smaller than 3 , the modulus may be taken to be 3 .
When is large the modulus approaches the same limiting form as in case This curve is applicable to electric , or luminous , vibrations incident upon a thin perfectly conducting screen with a linear perforation when the electric vector is parallel to the direction of the slit .
about one-third of the wave-length of yellow-green , there would be distinctly marked opposite polarisations at the ends of the spectrum .
These numbers are in good agreement with the estimates of Fizeau : " " Un ligne polarisee perpendiculairement a sa direction a paru de de millimetre ; un autre , beaucoup moins lumineuse , polarisee parallelement A sa direction , a e'te ' estimee a de millimetre .
Je dois ajouter que ces valeurs no sont qu'une approximation ; else peuvent eAtre en re'alite ' plus faibles encore , mais il est peu probable qu'elles soient plus fortes .
Ce qu'il a de certain , c'est que la polarisation parallele n'apparait que dans les fentes les plus fines , et alors que leur largeur est bien moindre que la longueur d'une ondulation qui est environ de de millimetre It will be remembered that the " " plane of polarisation\ldquo ; is perpendicular to the electric vector .
It may be well to emphasize that the calculations of this paper relate to an aperture in an infinitely thin perfectly conducting screen .
We could scarcely be sure beforehand that the conditions are sufficiently satisfied even by a scratch upon a silver deposit .
The case of an ordinary spectroscope slit is quite different .
It seems that here the polarisation observed with the finest ticable slits corresponds to that from the less fine scratches on silver deposits .

