]\gt ; lU86 .
] On the Intensity of flected Light . .
' On the Intensit .
of Light reflected from certain Surfaces at nearly Perpendicular Incidence By LORD RAYLEIGH , M.A. , D.C.L. , Sec. B.S. Received October 6 , 1886 .
[ PLATR 3 .
] In the presenb commllnication I propose to give an account of a photometric arrangement presenting some novel features , and of some resuIts found by means of it for the reflecting power of and silver surfaces .
My attention drawn to the subject by an able paper of Professor Rood , , in giving some results of a photometric method , comments upon the lack of attention bestowed by experimentalists upon the veritication , or otherwise , of Fresnel 's formulae for the reflection of light at the bounding surfaces of transparent media .
It is true that polarimetric observations have been made of the ratio of the intensities with which the two polarised components are reflected ; but even if we suppose ( as is hardly the case ) that these measurements are altogethel confirmatory of Fresnel 's formulae , the question remains open as to whether the actual intensity of each component is adequately represented .
This doubt would be set at rest , were it shown that Young 's formulae for perpendicular incidence ( to which Fresnel 's reduce ) , , agrees with experiment .
Professor Rood 's observations relate to the effect of a plate of glass when interposed in the conrse of the light .
He measures , in fact , the transmission of light by the plate , and not directly the reflection .
No one is in a better position than myself for appreciating the advantages of this course from the point of view of experiment .
In the first place , the incidence can easily be made strictly perpendicular , in which case no question arises of a separate treatment of ttJe two polarised components of ordinary light .
And , what is much more important , the interposition of the plate leaves the course of the light unchanged , and thus allows the aIteration of intensity to be determined in an accurate manner with the simplest arrangements .
On the other hand , measurement of the transmitted , instea of the reflected light , is open to grave objection on more than one ground .
It may be doubted whether the influence of absorption is altogether negligible , even when the thickness of the plate is as small as that mentioned by Professor Rood , mm. But the feature which strikes me most unfavourably is the necessary magnification of error , when we deduce the proportion of light reflected * ' Amer .
Journ. Sci vol. 49 , ( lfarcl , ) ; vol. 50 , ( July ) .
is to be compared with the .
The difference is now some per cent. , and is thus by no means insignfficant .
In the other case given by Professor Rood the discrepancy is greater still , amounting to 7 per cent. It may be remarked that in both cases the amount of the reflection appears to be in excess of that given by Young 's formula .
But the cause may lie in the assumption that the whole failure of transmission is due to .
And whatever the explanation may be , we can hardly agree with Professor Rood when he concludes that these experiments show " " that the reflecting power of glass with the above indsx of retraction , conforms in the closest manner to the predictions of theory In the hope of being able to deal directly with the reflected light , I made a great many trials of various devices during the spring of 1885 , but finding anything satisfactory .
Indeed , at one time , I had almost come round to the opinion that the difficulties of measuring the reflected light were so great that Professor Rood had shown a wise discretion in declining to facs them , and that after all the best resuJts would perhaps be reached through measurements of the transmitted light checked by the use of plates of different thick nesses so as to eliminate absorption .
* If , indeed , we give up the perpendicular cidence , the objection founded upon the relatively small quantity of light reflected may be met ; for at an incidence of , about half the light ( polarised in the plane of incidence ) is reflected .
In such an experiment it would of course be necessary to determine accurately the angJe of incidence .
The difficulties referred to have their origin in the necessary aIteration in the course of the light by the act of reflection .
The dirsct and reflected light cannot be interchanged in any simple manner , and the shift necessary to bring about the substitution may easily lead to systematic error .
In the apparatns presently to be described ) to which I was finally led , the difficulty seems to be fairly was not until that Sir John Conroy was at work in this direotion .
Light reflected at nearly Perpendicular Incidence .
277 overcome so far as regards the accuracy of the results , but at the cost of several tiresome adjustments , impeding the ready trial and interchange of various reflectors .
My differs in several respects from that generally used for photometric purposes .
Before describing it in detail , it may be worth while to indicate some of the considerations which led me to design it .
The photometers in ordinary use may be said to depend upon the principle of diff sion .
If the candle , or lamp , be drawn back from the screen to double the original distance , the brightuess of the screen as perceived by the eye is supposed to be quartered .
This implies that ( within certain limits ) the brightness of the screen is independent of the ma , gnitude of the source of light ( the lotal radiation being given ) , or that the light diffused by the screen in a particular direction ( towards the eye ) is independent of the direction of incidence .
, the light incident in a definite direction is supposed to be diffused through a considerable angle with some approach to uniformity .
There is no doubt that with proper arrangements this condition may be with sufficient accuracy for practical purposes .
My object in formulating it is to show that the use of a diffusing screen iu photometry is necessarily atteuded by an enormous reduction of light .
For our present purpose this loss of light is a serious matter .
Weakened to by reflection from glass , the light of an ordinary candle 01 lamp is hardly sufficient to illuminate a diffusiug screen properly , unless placed so close that measurement of the distances becomes uncertain .
The difficulty might perhaps be got over by the use of iucandescent electric lamps , but such were not at my command .
WheIl , in Sir John Conroy 's experiments , reflecting surfaces under test are metaIlic , or when ( as above suggested ) the observation relates to the transmission of light by an oblique plate of transparent material , the illumination given by a lamp may be adequate .
In my apparatns all the reflections are regular , and there is no further loss of light than the characters of the surfaces entail .
An incidental advantage is that the accurate flatness of surfaces demanded by methods in which illumination is infel'red from distance , is here unnecessary .
The apparatus was first set up during the summer of 1885 ; but the glasses then at my disposal were not good enough , and when the giass mirrors , \amp ; c. , necessary for satisfactory working came into my possession , the season was so far advanoed that I decided to postpone operations until the following summer .
* ' Roy .
Soc. Proc vol. 85 , 1883 , p. 26 .
ground .
Under these circumstances , and when there is no unshine , direct or reflected , falling upon the ground glass , the latter may be looked upon as a tolerably uniform source of diffused light .
This uniformity , however , is not relied upon ; but the arrangements are so made that the parts of the field compared are contiguous or identical , and are seen by rays which eaVe at the same angle .
It will be convenient first to describe generally the course of the light , and afterwards the manner in which the adjustments were effected .
Proceeding from the ground glass ( A ) , fig. 1 , the light falls upon the transparent plates , , at which nearly equal parts are reflected .
These plates are of worked glass about 6 inches by 4 inches , and are placed at the polarising angle .
By this means are obtained two beams of polarised light of nearly equal , and of constant relative , brightness .
The light transmitted by both plates is stopped at a screen , and takes no furbher part .
On the right hand side the light reflected at is erain reflected by a mirror of worked glass , silvered behind , at , and assumes the direction CDF .
and are alernative positions of the same mirror ( also of worked glass , silvered behind ) .
When the glass under test Fi is in use , the shifting mirror is in the position , and the light follows the course CDEFH .
In the contrary case , there is one reflection instead of two , and the ray takes the finally identical course CDFH .
At this central ray is reflected at the extreme of a speculum of silver-on-glass in the direction HI , to a small cbserving telescope , is focussed upon this edge .
By adjustments that will presently be explained , it is secured that the reflections at and shall take place under the same angle , and efore with the same ( moderate ) loss of li ; but when is in nse the brightness is diminished some twenty times .
To compensate this in the other position the disk is introduced .
It consists of a blackened disk of tin , from which ( along a sufficient length measured radially ) a sector is cut out , so that when the disk is *Slron sulphuric acid is an excellent detergent for this purpose .
1886 .
] Light reflected at nearly Perpendicular lncidenoe .
oaused to revolve the view is cut off and blackness substituted for nbout nineteen-twentieths of the whole time .
When the speed exceeds about twenty-five revolutions per second , there is no perceptible flickering , and the light is seen with a simple diminution of brightmess .
The idea of the method is so to adiust the angular opening that the effects of the glass under test and of the disk shall be equal .
The two brightnesses last considered onIy be seen successively .
They are tested with a comparison light , reflected at from the same primary beam .
After reflection at a silvered mirror , , and then at a blackened glass , , this light falls upon a silveron-glass speculum at , and passes thence to the observing telescope at I. In setting up the apparatus a control over the brightness of the comparison light is obtained by varying the angle of incidence upon In order that the line of division between the two fields , as seen from I , may be quite sharp , it is necessary that the final reflector be a speculum .
To obtain a reflecting surface , perfect up to the very edge , a piece of silvered glass is carefully cut ( on the glass side ) with a diamond .
If the operation is propelly performed , the silver is left undisturbed , and when the plate is inclined , as at , no part of the glass substratum is visible .
In adjusting the apparatus the object aimed at is to cause the central ray , ABB ' , issnin .
' from A on the ground glass , to assume uItimately the position , whether it proceed by the course on the left , or by either of the alternative courses on the right .
As may be supposed , this is more easily said than done .
All the reflectors require to be adjusted so as to be perpendicular to the plane in which the central ray is to travel .
This plane is conveniently taken horizontal , so that every reflector has to be vertical .
* The central ray is de[ined by diaphragms at , points in the same horizontal straight line .
At A the light is admitted through a smaU aperture only .
At the holes are cut in thin cardboard screens held in definite positions up to the glasses .
It was found to have them rather large\mdash ; about half an inch in diameter .
In setting up the apparatus the glasses are readily put into position , accuracy being required only in the levelling .
The line CDF is now defined , and the next step is the more difficult one of fixing the two positions for the stand carrying the mirror D. This stand is ( like all the others ) provided with levelling screws , but these must not be used in passing from the one position to the other .
A heavy * The levelling of the reflectors was effected with the aid of straight edge of a long board , adjusted until it coincided with the prolongation of its image .
My assistant , Mr. Gordon , is expert at this adjustment of the edge to perpendicularity with the reflecting surface .
The the latter is then tesled application of a spirit-level to the edge of the board .
VOL. XLI .
particular care except in the levelling .
The central ray , as fixed by the diaphragms , should fall near the middle of the surface .
The other positions would also be somewhat arbitrary were it not for the necessity of securing the same angle of incidence and reflection in the two positions .
To assist in this a small frame of brass wire is provided , carrying two pointers , and so arranged that it oan always be placed in an absolutely defiuite position with respect to the mirror .
By means of hooks it makes two contacts with the back , and two with the upper edge of the mirror .
Of the other two contacts required to make up the necessary six , one is with the lower part of the face of the mirror , and the other with one of its vertical edges .
Of the pointers , one ( in the path of the incident ra .
leads upwards , and the other ( in the path of the reflected ray ) leads downwards .
By bending them suitably their extremities may be brought into the path of the central ray , so that when the eye is placed in such a position ( H ) as to see the central point A in the middle of the ( apparently elliptical aperture this central point is just enclosed between the barely meeting pointers .
By so choosing the second position , , that this condition is again satisfied to an eye looking along , we secure not only the same angle of reflection but the use ( for the central of the same part of the glass .
In making the adjustment we may first bring the pointer on which the incident ray strikes into the already determined line , and then rotating the apparatus about the vertical through this point , bring the second pointer to coincide with the reflected ray .
We have now to consider how to fix the position of , the reflector under examination .
Replacing the shifting mirror into its first position , , we mark the line of the central reflected ray , by needles , standing up from the table and as far apart as convenient .
Transferring the shifting mirror to the position , we have so to place that the reflected ray shall coincide with the same line as before .
For this purpose not only must the azimuth of be correct , but its plane must be brought into the intersection of the already determined *Auxiliary lighting , with a candle or otherwise , is sometimes necessary in order see these apertures properly .
Light reflected at nearly pendicular Incidence .
281 lines DE , FE .
A levelled slab of glass is provided , on which to rest the feet of the stand carrying E. The mirror is ItOW brought into a vortical plane , and may then be shifted on the slab without loss of this adjustment .
The remaining double adiustment is best made .
By rota , tion about any vertical axis , the central ray may be caused to pass over one of the needles .
If it fails to pass over the other , the axis of rotation must be shifted backwards and forwards until a suitable rotation allows satisfaction of both conditions .
The ray now follows in both cases the course FGH , and the mirror , with ths sharp edge , may next be pushed in so as just to catch the ray in question and send it to the observing telescope ( half of a small opera glass ) at I. The adjustments for the auxiliary light on the left hand side are a simpler matter .
All the mirrors being levelled , the central ray is brought to the point , in the prolongation of IH .
Nothing then remains but to tnrn the final ( vertical ) mirror round until the reflected ray coincides HI .
When the eye looks in along this line , the bright spot should be seen in the same position from both mirrors .
To guard against accidental displacements , the movable pieces were usually secured with a little sealing wax .
A diaphragm at limits the field of view , and is so placed that the aperture is bisected by the division line H. It is not necessary to do more than allude to various screens employed to cut ofl stray light and render the room as dark as possible .
The principal trouble experienced , that of making and ining the adjustments , is connected with the rather large scale of the apparatns : which made it difficult to use a single levelled bed for all the movable pieces .
The question is thus suggested , what is it that fixes the absolute scale P And the rather unexpected answer must be\mdash ; the diameter of the pupil of the eye , which is the only linear quantity concerned .
* In order to understand this it is necessary to bear in mind that although in describing the adjustments we speak of a single ray only , we are of necessity really dealing with a complete beam .
The observation of a match requires that the two parts of the field of view have finite angular magnitudes , and from every point of the field there must proceed a pencil of rays limited by the pupil , or by the telescope .
If all these rays are to be treated as sensibly parallel during their passage through the apparatus , certain limitations must be observed .
For easy observation the field of view should subtend at the eye an angle of not less than a degree , so that if no telescope be employed the defect of parallelism must exceed this amount .
The linear scale of the apparatus is not thus fixed , however , for we * The wave-length of light may be regarded here as infinitely small .
The employment of a telescope introduces a certain modification , which it may be worth while to state somewhat fully , as the principle is of general application .
The extreme angle between the rays of the beam may be regarded as made up of two parts : ( 1 ) the angle subtended at the object-glass by the aperture in the diaphragm ( K ) near the final mirrors ( upon which the telescope is focussed ) ; ( 2 ) the angle subtended by the object-glass at the diaphragm .
If\mdash ; diameter of pupil , diameter of aperture in diaphragm , distance between telescope and diaphragm , magnifying power of telescope , angular diameter of field of view presented to the eye , then and the extreme angle between the rays of the beam\mdash ; We may here regard and as given beforehand ; and we see that with a given the first term may be reduced without limit by increas ing , and that then the defect of parallelism is to , the diameter of the pupil .
If and can both be increased withouI limit , we may approach as nearly as we please to a state of things in which all the rays concerned are parallel .
The preservation of full brightness throughout is eady secured by the supposition that eifective aperture of the object-glass is ma .
The reasoning set forth above shows at any rate that the size of apparatus cannot be reduced below a certain point , but I do affirm that mine was not unnecessarily large .
In addition to its othe1 advantages , the use of a telescope gives facilities for obtaining a good ocus upon the division line , an adjustment of great importance the easy recognition of small differenc.es of brightness .
The necessarily finite magnitude of the field of view involves a certa imperfection in this , and probably other methods of photometry .
We 1886 .
] Light p.efleeted at nearly ence .
283 indeed secure that the lights seen in immediate juxtaposition come from the same part of the ground glass , but a corresponding perfeclion of adjustment does not apply to other parts of the field .
If we suppose onrselves to be looking through the telescope at the ground glass , the part seen to the right of the division line really lies to the right on the ground glass .
On the other side there is a distinction , according to the two positions of the shifting mirror .
When the revolving disk is , in use , the circumstances on the right hand side of the apparatus correspond to those on the left , and thus the part of the field seen to the left of the division line really comes from the left on the ground glass .
The ground glass is thus seen much as if it were looked at directly , in spite of the separation of the light into two parts following distinct courses .
On the other hand , when the additional reflector ( under examination ) is brought into play , there is another inversion , and the part of the ground glass seen to the left comes really from the right of the central line .
In this case , therefore , it is the same part of the ground glass which is seen in both final The distinction here pointed out would be of no consequence if the field were absolutely uniform , or if it were possible to compare the parts seen in immediate juxtaposition , without regard to the parts a little further removed .
But if the original field vary slightly in brightness from right to left , it will be a.question how far the eye would select for the match continuity of brightness across the division line , or how far it would demand equality in the average brightnesses of the two parts presented .
It now remains to describe certain accessolies .
the observations it is necessary to have some means of varying the relative brightnesses of the two parts of the field without removing ths mirrors or altering the width of the slit in the revolving disk .
For this purpose a plate of glass , capable of rotation about a vertical axis , was introduced into the path of the light on the right hand side of the apparatus ( between the second and third reflections ) .
As the angle of incidence upon this plate increases , a greater proportion of the light is reflected and thrown away , and a less proportion is transmitted to the eye .
The observation consists in varying the azimuth of this plate until the match is satisfactory , after which the obliquity of the plate is measured .
The transmission by the plate at the measured obliquity can then be found approximately from Fresnel 's formula .
* It may , convenient table is given by Pickering ( ' Phil. Mag vol. 47 , 1874 , p. 129 ) .
If A be the proportion of light reflected at a single surface , the transmission through a transparent plate is given by . .
The whole reflection is thus , from which Pickering 's table is calculated .
An erratum may be noted .
For the reflection bold be , not , the value of A being supposed to be .
while .
There are some other minor inaccuracies .
meant of the mirrors on the left , with an alteration in the angle of incidence upon the black glass In taking an observation the adjustment of the relative brightnesses was facilitated by a device which may now be described .
If the attempt be made to secure an absolute match between the two parts of the field in view , a doubt is apt to arise as to whether the disappearance of the division line is due to the success of the .
adjustment or to fatigue of the eyes , leading , as in my case it very rapidly does , to imperfect focussing .
This difficulty is less felt when the adjustment is under the immediate control of the observer , who can then satisfy himself of the sensibiveness of his eye by making the displacement ; but in the present experiments ( on account of the distance of the telescope ) it was convenient to employ an assistant .
glass plate , perpendicular to the path of the light , and attached to a sort of pendulum , was therefore provided on the left hand side , in such a manner that by pulling and letting go a , ring it could be introduced or withdrawn at pleasure .
The effect of the plate would be to stop some 8 per cent. of the light , and the adjustment was so made that with glass in the ( apparent right of the field was as much too dark as it was too bright when the glass was out .
The difference of brightness , amounting according to the above estimate to 4 per cent. , was always fully apparent , and probably no setting more than 2 per cent. in error would be allowed to pass , giving , as such wou ld do , a diiference of 6 per cent. on the one side and of 2 per cent. upon the other.* Since the auxiliary light is eventually eliminated , it makes no difference , of course , whether we take for the compare son the full light , or the mean of the lights with and without the interposition of the plate .
A 2 per cent. error in single settings may lead to a per cen .
4 er cent accuracy of the settings falls much short of that attained by Professor Rood .
: Light reflected at nearly Perpendicular lncidence .
error in the comparison of the effects of the reflector and of the disk ; and accordingly ( since this may occur in either direction an 8 per cent. discrepancy in the results is possible .
This , however , would be very unlikely , and with a two- or three-fold repetition of the individual settings would be practically out of the question .
The revolving disks , used to diminish the light on the right hand side of the apparatus in about the same degree as by the or under test , were cut from tin plate , about 9 inches in diameter , and carefully .
The angular apertures were finally calculated from measurements of the chord of the arc , and of the radius .
It is important that the disks be thoroughly blackened , in view of the assumption that no light reaches the eye except during the passage of the aperture .
Here is one reason why it is desirable to keep the room as dark as possible .
The disk should also be properly balanced .
On one occasion a curious and at first puzzling effect was observed .
The division line , which should present no visible width , sensibly widened , appearing sometimes darker than the nearly balanced adjoining fields , and sometimes , though more rarely , appearing relatively bright .
The explanation is to be found in a vibration of the mirror , whose edge forms the division line , in a horizontal direction perpendicular to the line of sight , the vibration being communicated from the revolving wheel through the floor to the table upon which the mirrors stood .
It is evident that if the two lights under comparison were equal , not merely on the average , but at every moment of time , such a movement of the mirror would have no disturbing influence , and could not make the division line visible .
But it is otherwise when one of the lights is intermittent , and the vibrations of the mirror are executed ( as here they must be ) in the same period .
For suppose that at the moment when the division line is advanced , so as to invade still further the field from the back mirror , the light is reaching the eye through the aperture in the disk .
In this case the parts near the edge of the vibrating mirror will be sending to the eye the full light due to this part of the field .
During the remainder of the vibration , no light should reach the eye , but if this mirror retreats , the back mirror sends its continuous light from the same apparent place , so that when the angular in the disk is small , it is possible for the part of the field over which the division line vibrates to present an almost doubled brightness , combining in fact the illumination of the two parts of the field .
A diiferent phase relation may evidently lead to an abnormal diminution of brightness in the same region .
These effects disappeared when the disk was better balanced .
with black varnish , the effect of which , however , in annulling the second reflection , was far from complete .
With this glass , carefully cleaned ( but not repolished ) , six sets of observations were made , four by myself and two by Mrs. Sidgwick .
Each set consisted of three or four settings with the glass in operation , and about the same number with substitution of the disk .
The following is a set of readings by myself on August 7th , 1886:\mdash ; Face of Prism ( I ) .
Reflection .
Revolvingdisk . . . . .
38.2 . . . . . .
42.7 . . . . . .
39.1 . . . . . .
43.5 Mean .
The angles here given are the obliquities of the adjnstable glass plate used to graduate the intensity .
According to Pickering 's table , calculated from Fresnel 's formula , the effect of this plate at would be to reflect per cent. of the light incident upon it .
The light transmitted is therefore per cent. In like manner the light tramsmitted by the plate at an obliquity of is per cent. In order to complete the calculation of the reflecting power of the glass surface , we must know the proportion of light transmitted by the revolving disk .
Measurements gave for the chord of aperture of this disk in fiftieths of an inch , corresponding to a radius of .
The angle of aperture is thus .
Accordingly the factor expressing the reduction of light by use of the disk is\mdash ; The reflection from the face of the glass prism is thus\mdash ; 82.5 1886 .
] reflected at nearly Perpendicular Incidence .
The following is a summary of the resuIts obtained at this time:\mdash ; Face of Prism ( I ) .
Lord Rayleigh .
Mrs. Sidgwick .
Aug. 4 . . . . . . . . . .
Aug. 4 . . . .
5. . . . . . . . . .
, , 5 . . . .
7 . . . .
, Mean . . . . . . . .
Mean . .
Final mean . . . . . .
Since this number is very nearly the same as that due to the disk alone , we see that the resuIt is scarcely at all dependent upon the correctness of the assumed effect of the oblique plate .
It now remains to make comparison with the reflection as given by Fresnel 's formula , .
:\mdash ; ( A ) ' where , and is the refractive index .
The index was determined in the usual way from the angle of the prism and from the minimum deviation ( D ) .
The value of was found to be .
The mimmum deviation of soda light on one side was , and on the other .
Thus , and The angle of incidence ) , which was the same in all the observations with the various reflectors , was measured by determining the angle between the incident and reflected ray .
This measurement does not require great precision , for a change of a whole degree in the value of would alter the reflection about 1 per cent. only .
I found\mdash ; With these values of and , we find\mdash ; about 10 per cent. in excess of the reflection actually observed .
order to satisfy myself that the deficiency of reflection was real and permanent , this prism was remounted after a thorough cleaning , and further observations were taken , as summarised in the following table .
The revolving disk was the same as before:\mdash ; Lord Rayleigh .
On the Intensity of Prism of Crown Glass ( I ) , Remounted .
i Lord Rayleigh .
Mr. Gordon .
Nean .
Aug. 24. . . . . . . . . . . . . . .
25 , morn. . . . . . . . . . .
25 , even.* . . . . . . . . .
Mean . . . . . . . . . . . .
The difference between and the mean previously found , , has no significance .
In consequence of the detection of a greatly augmented reflection from another glass surface Ir , below ) , as the result of a repolish with pntty powder , this surface also was submitted to similar treatment .
Immediately afterwards , on August 30th , a much increased reflection was observed , the numbers by two observers being\mdash ; ; mean , The disk gave , as before , a transmission ; so that the bers for the repolished face depend too much upon the assumed effect of various obhquities of the inclined plate to be fully orthy , even were they sufficiently numerous to against accidental errors .
But they proved , unequivocally , a considerable increase in reflecting .
power as the result of the repolish .
In view of these resuIts , a new disk was prepared of angular aperture about , and , COIlsequently , with a transmission equal to .
The numbers obtained with this are shown in the table:\mdash ; Prisms of Ground Glass I ) Repolished .
Lord Rayleigh .
Mr. Gordon .
Aug. 30 , aft . . . . . . . . . . .
31 , morn . . . . . . . .
31 , aft. . . . . . . . . . . . .
Mean . . . . . . . . . .
Final mean The observed result now agrees remarkably well with that calculated from Fresnel 's formula ; but unfortunately it depends more ( for about 5 per cent. of its value ) than could be wished upon the use of the oblique plate .
* The considerable discrepancy shown in this set of readings was probably caused by insufflciency of light .
1886 .
] Light reflected at nearly Perpendicular Incidence .
Prism of Crown Glass ( II ) .
So soon as it appeared that the reflection from the face of prism fell so much short of what was to be expected in accordamce with Fresnel 's theory , I tried another prism whose surface was still older than that of ( I ) .
The event proved a still more marked deficiency .
With the aid of the disk giving transmission , the following numbers were btained : Prism ( II ) , before Repolishing .
Lord Rayleigh .
Mrs. Sidgwick .
Mr. Gordon .
27 . . . . . .
Aug. 26 . . . . . .
\mdash ; . . . .
Mean . .
Although somewhat dependent upon the assumed effect of the oblique plate , this number is far too low to be consistent with anything obtainable from Fresnel 's formula with an admissible index .
* This circumstance suggested a repolishing of the surface , which , however , was superior to that of ( I ) , so far as could be judged from close inspection in a favourable light .
The repolishing was executed by Mr. Gordon by means of a disk of wood charged with putty powder and mounted in the lathe .
Observation now demonstrated a remarkable improvement in the reflecting power , as the following numbers will show : Prism II ) , after Repolishing .
Mrs. Sidgwick .
Lord Rayleigh .
Mr. Gordon .
Aug. 28 , morn. . . . . . . . . . .
28 , aft . . . . .
\mdash ; . . . . . . . .
30 , morn. . . . . . . . . . .
Mean . .
Here again too much depends upon the oblique plate , the transmission of the disk being only ; but there can be no doubt of an increase in the reflection of something like 30 per cent. we may argue from the number obtained from prism ( I ) after repolishing , * If be the proportions of light reflected at incidence , then Fresnel 's formula is equivalent to\mdash ; by which is found .
The index is then given by .
In the present case so that , since that after some months or years the reflectio may fall off from 10 to 30 per csnt .
, and that without any apparent tarnish .
The question as to the cause of the falling off , I am not in a position to answer satisfactorily .
Anything like a disintegration of the surface might be expected to reveal itself on close inspection , but nothing of this kind could be detected .
A superficial layer of lowerindex , formed under atmospheric influence , even though no thickerthan inch , would explain a diminished reflection .
Possibly a combined examination of the lights reflected and transmitted by glass surfaces in various conditions would lead to a better understanding of the matter .
If the superficial film act by diffusion or absorption , the transmitted light might be expected to fall off .
On the other hand , the mere interposition of a transparent layer of intermediate index would entail as great an increase in the transmitted as falling off in the reflected light .
There is evidently room here for much further investigation , but I must content myself making these suggestions .
Plate Glass Silvered Behind .
This glass was silvered chemically by the milk-sugar process , and by transmitted light showed the sky of a normal deep blue colour .
The film was not polished .
In determining the efficiency of this and other good reflectors , the bIack glass mirror replaced by one silvered behind .
The first trial without a revolving disk gave for the reflecting power .
This result , of course , depended entirely upon the assumed influence of various obliquities of the adjusting plate .
disk was therefore prepared with two opposite projecting teeth , in which the ratio of aperture to circumference turned out on careful urement to be .
This number , therefore , represents the transmission of light by the disk .
Using his disk I found the following values for the reflecting power of the mirror for light incident upon it at an angle of :\mdash ; Aug. 11 1'2. . . . . . . . . . . . . .
Mean . . . . . .
This result relates , like all the others , to light polarised in the plane of incidence .
Mirrors of this kind are durable , and not being exposed to tarnish are more convenient than specula , whenever the double reflection is not objectionable .
The high reflecting power is a satisfactory feature .
1886 .
] reflected at nearly Perpendicular Incidence .
Silver-on-Glass Speculum .
This was the silver side of the same glass as the last , polished with vash leather and a little rouge .
The milky film was not perfectly removed .
Four observations , not over concordant , probably in consequence of variation of reflecting power at different parts of the surface , gave\mdash ; Lord Rayleigh .
Mrs. Sidgwick .
Mr. Gordon .
Aug. 14 . . . . . .
Aug. 16 . . . . . .
Aug. 14. .
16 . .
Mean . .
The surface was then repolished and remounted with the following resuIts : Lord Rayleigh .
Mr. Gordon .
Aug. 18 . . . . . . . . . .
Aug. 18 . . . .
19. . . .
, , 19 . . . .
21 .
0.938 lIean . . . . . .
Nean . . . .
Mean . .
The increase in efficiency may have been due to a more careful selection of the best polished central part as much as to actual improvement in the polish of the speculum as a whole .
The transmission of the disk used with this surface is Sir John Conroy*found an even higher number as the reflecting power of silver films for light polarised in the plane of reflection , and incident at of Black Glass .
A plate of opaque glass has the advantage that the influence of the hinder surface is eliminated without more ; but , on the other hand , it lends itself less readily to determinations of index , The following resuIts were obtained with such a plate:\mdash ; Mrs. Sidgwick .
Bayleigh .
July 29 . . . .
30 . . . . . . . . . .
Aug. 2 . . . . . . . . . . .
3 . . . . . . . . . . .
lIean . . . . . . . . .
Final mean . .
* Roy .
Soc. Proc vol. 37 , 1884 , p. 38 .
glass might be flint , a conclusion confirmed by determinations of the refractive index .
These were made by two methods : ( 1 ) by observation of the polarising angle in air , ( 2 ) by observation of the angle at which total reflection sets in when the mirror is immersed in bisulphide of carbon .
The first is , perhaps , simpler in respect of experimental arrangements , but it is open to the objection that the inference of the refractive index from the polarising angle is somewhat theoretical .
The black glass was mounted upon the turntable of an ordinary goniometer .
In the focus of the collimator was placed a wire , seen dark in a bright field of view .
Various positions of the turntable were then tried , such that on rotating a Nicol held at the eye the dark patch appeared to pass somewhat to the right or to the left of the collimator wire .
After each observation the web of the telescope was set to coincidence with the collimator wire , and a reading taken .
Success depends in some degree upon the use of a suitable light .
Sunshine diffussd through ground glass answered the purpose very well .
Right .
Left .
Central . . . . . . . . . . . .
The table gives a set of circle readings .
In the first column patch was to the right of the collimator wire , in the second to the left , and in the third there no appreciable deviation .
We may , therefore , take as the reading for the polarising angle , with a probable error not exceeding 3 ' or 4 ' .
The reading for a direct setting of the telescope upon the collimator wirs was , so that the polarising angle is .
Whence according to Brewster 's law\mdash ; 1886 .
] Light reflected at nearly Perpendieular Incidence .
293 This relates to white light .
To find the index of refraction by the method of total reflection , the mirror was mounted vertically in small tank of plate glass , cemented with glue and treacle , and containing bisulphide of carbon .
The mirl.or , as shown in fig. 2 , was parallel to one of the sides of FIG. 2 .
the tank , and a cover was provided to check evaporation .
A uniform field of homogeneous light could be obtained from a salted spirit lamp , , with the aid of a plate of ground glass , , and a collimating lens , C. The eye looking in along such directions as , is able to mark with considerable accuracy the direction in which total reflection begins .
By the aid of plumb-lines , \amp ; c. , this direction and that of the face of the mirror ( seen from above ) were marked upon a board , and it appeared that the angle GLK , between the face of the mirror and the direction GLF of the first totally reflected light was A beautiful variation in the experiment may be made by replacing the spirit lamp with a , and subsequently analysing the reflected light by a direct vision prism .
For this purpose a screen a slit should be interposed as near as conveniently may be .
As the incidence of the light ] ) the black glass becomes more grazing total reflection sets in , but first at the violet end of the spectrum .
; whence for the index of the glass relative to soda light , cos1 The amount of reflection according to FresnePs formula , with an incidence of and an index , is , a little less than that actually observed .
The agreement is as good as could be expected , but it should be noticed that this was merely cleaned and not repolished with putty powder .
If repolishing were to produce as much effect in this case as upon the acute-angled prism ( I ) , Fresnel'sormula would be left considerably in arrear .
* P.S. Nov. 9 , 1886 .
am indebted to Mr. Glazebrook for a determination of the refractive index of the prism of crown glass II .
He finds .
The introduction of this into Fresnel 's formula gives for the reflecting power .
" " A Theory of yoltaic Action By J. BROWN .
Communicated by Lord RAYLEIGH , Sec. R.S. Received October 4 , 1886 .
[ PDATES 4 AND 6 .
] 1 .
From a series of experiments made mo re or less continuously during the last five years the ollowing conclusions have been drawn : \mdash ; That the difference of potential near two metals in contact as observed either by the -metallic condenser ( Volta 's ) method , of the results here given were communicated to the British Aesociation at , where also was read a paper by Sir John Conroy on the same subject .
ProcRoy \mdash ; ' \fnof ; g b4

