]\gt ; Sir J. Conroy .
[ Feb. 15 , .
THE PRESIDENT in the Chair .
The Presents received were laid on the table , and thanks ordered for them .
The following Papers were read : \mdash ; I. " " Some Experiments on Metallic Reflection .
No. III .
On the Amount of Light Reflected by Metallic Surfaces By Sir JOHN oNROy , Bart. , M.A. Communicated by Professor STOKES , Sec. B.S. Received November 1 , 1882 .
As far as I am aware the only experiments that have been made on the amount of radiant energy reflected by metallic surfaces at different angles are those of Pottor Edinburgh Journal vol. iu , 278 ) and Jamin Ann. de Chim .
et de Phys [ 3 ] , , 296 ) for light , and for radiant heat , those of Forbes Phil. Mag [ 3 ] , viii , 246 ) and of MM .
De la Provostaye and Desains ( ' ' Ann. de Chim .
et de Phys [ 3 ] , , 276 .
Potter used a 's photometer ; the transparent screen being made of white paper , behind which two lamps were placed , the light from one of which always fell upon one half of the screen , whilst that from the other either fell directly upon the screen , or after reflection from the metallic plate .
The observations were made by measuring the distance at which the first ] had to be placed in order that both halves of the screen should appear equally bright .
The two illuminated portions of the screen were not actually in contact , being separated by a dark shadow , an arrangement which , to a certain extent at least , must have interfered with the accuracy of the determinations .
The mirror having been placed close to the lamp , the light incident upon its surface must have been very divergent , and as , in addition , the angle at which it was placed could not , owing to the construction of the apparatus , have been very accurately determined , the values of the angles of incidence can only be considered as approximations .
In M. Jamin 's experiments the reflecting surface was half glass and half meta ] , the line of separation being vertical , and the incident light polarised in a plane to this ; the reflected light was examined with a double image prism which was rotated until one of the images due to the light reflected by the meta ] , and one of those 1883 .
] Some Experiments on Metallic Reflection .
27 dne to the light reflected by the glass appeared equally bright , i.e. , till the ordinary image of the light reflected -by one-half of the mirror was equal to the extraordinary image of the light reflected by the other half .
The refractive index of the glass being known , the amount of light reflected by it at any angle could be calculated by Fresnel 's formulae , and thus the percentage of light reflected by the metal determined .
The numbers obtained by M. Jamin agree well with those deduced by calculation from Cauchy 's theory , and also with those experimentally determined by MM .
Dc la Provostaye and Desains , by means of the thermopile .
Unfortunately , however , the experiments on this point made by these two eminent French physicists were not very numerous , and the method used by M. has even described by M. Verdet d'Optique Physique 11 , ) as Un procede indirect qui n'est pas susceptible d'une grande perfection .
Under these circumstances I trust that some experiments which I have recently made on the subject may be thought worthy of publication , although , owing mainly to the difficulties inseparable from all photometric determinations , the observations are not as co1lcordant as could be wished .
The method used essentially that of Potter , the experiments being made by comparing photometrioally ount of light reliected by a polished metallic surface different angles with that which fell directly on the photometer when the reflecting surface was removed .
This method is , of collrse , only applicable to the white metals , and order to obtain nything ] accurate results the mean of a considerable number of observations must be taken .
Two similar paraffine lamps , with flat wicks , were used , one arranged to slide along a board about two metres long , to which a scale divided into millimetres was attached , and the other supported by a metal ring fastened to one of the arms of a Babinet 's goniometer .
An endless cord , which passed round a pulley with a handle at end of the board , enabled the first-mentioned lamp to be moved and placed at different distances from the photometer .
It was originally intended to use a Bunsen 's disk , but it was found that owing to the small size of the beam of reflected light , it was not possible to make satisfactory measurements with it , and after various arrangements had been tried , a modification of Ritchie 's photometer was finally adopted .
Two pieces of white paper were so placed that whilst both were visible to the observer , one } ) eing slightly in of the other and overlapping it to a small extent , each received light from one only of the lamps , and when equally illuminated the edge ot the front paper vanished .
Two triangular blocks of wood 4 centims. 28 Sir J. Conroy .
high , were screwed to a rectangular board about cetlbims .
centims. , in the position shown in the figure , and pieces of white paper , 3 centims. by 3 centims. , held against the hypotenuse of each of these triangular prisms by india-rubber bands .
The whole rrangement was enclosed in a box with three apertures , which was painted , both internally and externally , a dead black , and was placed the end of the horizontal board .
The light from the sliding lamp entered by the left-hand aperture , }whilst that which fell directly on the paper , or after reflection from the metallic surface , entered to the right-hand one ; the third aperture allowed the papers to be seen by the observer , who was about 6 decims .
from it .
the actual size .
The goniometer was fixed on the other side of the photometer , with its vertical axis in the prolon.ation of the median line of the board , and bad a vertical stage , to which the reflecting plate could be readily fastened .
The second lamp , which , as before mentioned , was carried by a plate attached to one of the arms of goniometer , was always placed with its flame edgewise , i.e. , radiaJly , and the beam of light mited by means of a diaphragm with an aperture 25 millims. high and 5 millims. wide , placed at a distance of centims. from the centre of the flame and between it and the axis of the instrument .
The sliding lamp placed with the flat side of the flame towards the photometer .
The experiments were made by first turning the goniometer until light fell on the paper of the photometer ; the position of the sliding lamp was then altered until both papers appeared equally illuminated , and the distance of the lamp from the central line of the photometer observed .
Four such observations were made , in the first and third the sliding lamp being placed too near the photometer and the distance increased , and in the second and fourth the lamp placed too far off and the distance diminished until the illumination of the two papers becomes equal .
1883 .
] Some Experiments on Metallic Reflection .
29 The metal mirror was then clamped to the stage , which had previonsly been carefnlly adjusted , in order tha.t the surface of the mlrror might be vertical and in the axis of the lns trument , and the whole instrument rotated on its outer axis until the reflected beam fell on the photometer .
Four readings were then made , as has already been described , the mirror removed , and four more readings made of the intensity of the light emitted by the lamp ; the mirror replaced , the angle of incidenc.e aItered , and four more made , and so on .
It was found necessary to make the observations in this way , as although two lamps were trimmed as nearly as possible alike , considerable fluctuations in helr relative intensity not unfrequently occurred , and in order to diminish as far as possible this source of error , the mean of eight observations , four before and four after the measurement of the intensity of the refiected light , was taken as the intensity of the light incident upon the mirror .
The metal used measured about 8 centims. by 5 centims. , were accurately plane , and had all been polished with putty powder , it having been previously ascertained Proc. Roy .
Soc. , 'l vol. 31 , p. 486 ) that bhe optical constants for metallic surfaces depend to a extent upon the nature of the substance with which they have beeu polished .
It had been originally intended to use light polarised in and perpendicularly to the plane of incidence , and to determine the ratio of the reflected to the incident light in either case , but it was found impossible to make any satisfactory measurements , owing to the great ss of light .
The intensity of the could have been creased by the use of a larger Nicol , but it seemed so very doubtfu whether sufficient light would be thus obtained , that it was thought best to abandon the use of polarised light .
Ordinary light being equlvalent to two beams of light of equal tensity polarised at right angles to each other , if the total amount reflected at any angle , and the ratio of the intensities of the light polarised in and perpendicularly to the plane of incidence when light polarised at an allge of with that plane is incident upon the surface of the plate at the same angle , are kuown , the lective power the plate for light ] arlsed in an perpendicularly to the plane of incidence can of course be ] calculated .
The table gives a series of measurements made with a silver plate .
The numbers in the first column are the distances in centimetres of the sliding lamp from the photometer when the light of the other lamp fell direct on the paper , and those in the third when the light was reflected by the mirror .
The means of these observat ons are contained in the second and fourth columns , the angles of incidence 111 the fifth oolumn , and the ratios of the reflected to the incident light , the latter be , ing baken as 100 , in the sixth column .
As the intensity Sir J. Conroy .
[ Feb. Mean .
98 .
114.4 98 .
98.1 99 .
114.5 98 .
113.8 97 .
97 .
1 97 .
110.3 109.9 110.4 . . . . . . . . . . . . . . . .
73.10 . . . . . . . .
73.38 . . . . . . . . . . . . . . . .
74.15 . . . . . . . .
74.46 . . . . . . . .
76.94 . . . . . . . .
76.11 . . . . . . . . . . . . . . . . . . . . . . . .
79.60 1883 .
] Some Experiments on Metallic Reflection .
31 1 as the square of the distance from the source , of light varles lnverse the percentage reflecte ted the late is obtained by dividing the numbers contained in the secon column by those in the fourth , squaring , and multiplying by 100 .
Two other series of observations were made with the silver plate , the meastlrements being about as concordant as those given above .
Table I contains the results of these three series and their mean .
Table I. Note.\mdash ; The mirror was ghtly tarnished when the series of measurements were made .
Similar asurements were made with steel , tin , and speculum metal plates .
The resuIts are given in Tables II , III , and Table II .
Sir J. Conroy .
[ Feb. 15 , Table III .
Angle of Incidence .
Percentage amount of reflected by Tin Mirror .
Table The principal incidences and azimuths for the four mirrors wero determined in the described in the paper on metallic reflection which has already been referred to .
Roy .
Soc vol. 31 , p. 486 ) , a soda flame being used as the source of light .
Four observations were made in each position of the retarding plate , two with the principal section of the polarising Nicol on the right , and two with it on the left of the plane of incidence .
The means of several sets of eight observations each are given in Table 1883 .
] Some Experiments on Metallic Reflection .
Table V. Principal Principal Incidence .
Mean .
Azimuth .
Mean .
Silver . . . . . . . .
' Steel . .
Tin Speculum lvIetal The tables show that the amount of light reflected increases with .
the angle of incidence .
In the case of speculum metal , however , after first , the amount of light appears to diminish slightly , and after passing through a minimtlm at about to increase again .
These results are not in accordance with the experiments of Potter or of M. Jamin .
Potter found that the amount of light reflecbed diminished as the angle of incidence increased , being a maximum for perpendicular incidence ; a result that was confirmed by the experiments of M. Jamin , who showed that at angles greater than amy at which Potter had made observations , tho amoullt of reflected light increased The values of the principal cidences and azimuths given in Table V were used for calculating the amount of which , according to Cauchy 's theory , should have been reflected by the plates .
His formulae are\mdash ; For polarised in the plane of incidence : For light polarised perpendicularly to the plane of incidence .
and are given by the equations\mdash ; ( 1 ) arc , 2 arc .
and being two , and and two variables , rmined by the relation : \mdash ; ( 2 ) arc , VOL. XXXV .
34 Sir J. Conroy .
[ Feb. 15 , At the angle of polarisation , lvhen principal azimuth and the principal incidence .
The forms which have been given to these expressions by MM .
de la Provostaye and Desains Ann. de Chim ( 3 ) , 30 , p. 279 ) , are more convenient for calculation , and therefore were used .
These authors remark in their paper that it is generally sufficient to take , and therefore and this was done in calculating out the intensities .
The incident light having been unpolarised , half the sum of the intensities of the light polarised in , and perpendicularly to the plane of incidence was taken as the theoretical intensity of the reflected light .
Amoun of Light Reflected by the four The table shows that with the speculum metal mirror the observed and calculated intensities agree fairly well , but that such is not the *According to Lundquist Pogg .
Ann 152 , p. ) , Jamin himself appealB to have done so ; in this case the formulae ( 1 ) for detelmining and become spectively\mdash ; 2 arc , 2 arc .
1883 .
] Experiments Metallic caso with the other three mirrors .
All four mirrors had appeared crfectly untarnished and bright when the observations were made , but as silver , steel , and tin nlirrors had been polished some months they were used , it was thoug.ht possible that slight films formed on their surfaces , and that the difference in the calcnlated and observed results was due to this cause .
The silver mirror vas therefore returned to the mftkcr , er , to be repolished with putty power .
amount of light reflected by it was determined the same day it was leccived back , and it was found that its reflective power was slightly diminished for incident upon its surface at angles of , and , and what increased for light incident at , the number . . . . . .
68.54 These result agreeing fairly well with the means of the nllmerous observations which had previously been made , it was not necessary to make any there determinations , as it clear that he difference between the observed and caIculated values could not be due to a film on the surface of the mirror .
The surface of the silver mirror not .
very , it was returned to the maker to be polished with rouge , that being stated to be the best material for .
silver ; the resuIt was not very satisfactory , as the surface appeared less good than before\mdash ; that this was really the case was confirmed by the reflective power of the mirror being diminished .
The results of three series of observations are recorded in table , and also two deternlinations of the values of the principal incidenccs and azimuths .
Sir J. Conroy .
[ Feb. 15 , Principal Incidence .
Principal Azimuth . . . . . . . . . . . . .
The theoretical intensities calculaed by Cauchy 's formulae from the value of the principal incidence and principal azimuths are\mdash ; Incidence .
Observed .
Caloulated . . . . . . . . .
These numbers do not agree any better than those previously btained , and it therefore seems necessary to assume that Cauchy 's formulae ( at least in their simplified form ) do not express the facts of the case , except , perhaps , for speculum metal .
The values for the intensity of the light calculated by the formulae given by Professor James MacCullagh ( ' ' Collected Works p. 133 ) decrease slowly up to a large angle of incidence .
and then increase again just as is the case with the similar formulae of Cauchy .
Professor Stokes suggested that very probably the discrepancy between the observed and calculated results was due to imperfect polish , as the differences were greater with the soft metals , silver and tin , to which it is more difficult to give a good polish than for the hard metals , steel and speculum metal , and also as the observed intensities fell short of the calculated ones at moderate incidences , whilst sometimes even exceeding them at hi incidences , for which deficiencies of illumination due to defects of polish might possibly be to disappear .
Professor Stokes also gested a method for examining the polish of the mirrors .
In accordance with his suggestion a cylindrical tinned iron canister , closed at one end , about 9 centims. in diameter and 27 centims. deep , was blackened internally and supported on a table at an angle of about with the horizon , and with the lower of the open end about 4 centims. above the surface of the table , which lvas covered with a black cloth .
The table was placed out of doors , so that here might be plenty of light coming from all round , and the metal plates laid on the black cloth in front of the canister , so that an observer in front see , by reflection , into the perfect darkness .
If the polish were perfect , the surface of the metal plate would appear perfectly black ; and if such should not be the case , the illumination of the plate would afford an estimate of the defect of polish .
The reflection in the plate was examined through a small hole in a 1883 .
Some on Metallic .
37 black screen , in order to prevent any light diffiused from the observer 's face being reflected by the plate into the ( .anister , and thus its perfect blackness .
Two of the plates were placed side by side and examined together , and in this way it was ascertained that their order of polish was , steel , speculum metal , silver , and tin ; there being but differenoe between the polish of the steel and the speculum metal , and a considerable diflerence between the speculum metal and the silver , and gain between the silver and the tin .
The experiment was originally made with the silver mirror polished with putty powder , and was repeated after it had been polished with rouge ; the order remained the same , but it was thought that there was a greater difference between the speculum metal and the silver in the latter case .
Professor Stokes also suggested that the surface of perfectly clean mercury would fnrnish a standard .
Some mercury was , therefore , cleaned by being well shaken with pounded sugar , and then filtered three or four times through a cone of writing paper , with a small perture at the apex .
A small porcelain basin was blac , kened both internally and externally , and nearly filled with the clean mercury , and the reflection of the canister in it and in the plates compared .
The refiection in the steel appeared quite as black as that in the mercury , whilst that in the speculum meta ] appeared slightly less black .
In a preliminary experiment the steel was thought to be blacker than the mercury ; the difference , if any , , however , very slight , and was probably due to a thin film having formed on the surface of the mercury which , in that case , had only been cleaned by filtration through paper .
The experiment was therefore repeated with mercury which had been cleaned as has alreadybeen described , but it was still found that , as compared with the mercury , the polish of the steel was sensibly perfect .
The difference between the calculated and observed results for the steel at least cannot , therefore , be due to imperfect polish , and the discrepancy is almost too great to be accounted for by errors of observation , especially as the intensities actually observed increase with the incidence , whilst theoretically they ought to diminish , and then increase again .
The experiments show ( unless there is some error due to the 1nethod of observation , and therefore common to all the determinations ) , that the amount of light reflected by silver , steel , and tin gradually increases with the ange of incidence ; that with speculum metal after first increasing it diminishes slightly , and then increases again ; that the results obtained by the method described in this paper are not in accordance with the experiments of Potter and M. Jamin , or with the values calculated by the formulae of Cauchy and MacCullagh ; 38 Sir J. Conroy .
[ Feb. 15 , that with the silver and the tin mirrors the difference between theory and observation may be due to imperfect polish , but that such can hardly be the case with the steel .
Under these circumstances it appears desirable that a fresh series of observations should be made by some independent method , which would either confirm or disprove the resuIts contained in this paper , and this I hope to attempt .
Received December 18 , 1882 .
All previous determinations of the reflective power of metals havlng been made in air , it appeared desirable to make some observations with the sbeel and speculum metal mirrors in water , the polish of these mirrors being satisfactory .
A glass trough about centims. square and centims. deep , was filled with distilled water and placed on the stage of the goniometel .
in such a position that the light from the lamp passed normally through two of its opposite sides , and then fell on the photometer .
The distance of the sliding lamp was then altered till both papers appeared equally illumlnated ; four such readings were made , and then the mirror , which had been previously adjusted , placed in the trough and at an angle of with the incident light .
The trough was then so adjusted that the incident and emergent light passed normally through two of its adjacent sides , the goniometer turned until the light fell on the photometer , and the readings made in the usual way .
The mirrol .
was theri removed and the intensity of the light which passed the trough again determined .
The partial reflections from the glass sides of the trough , and the length of the path of the light in the water ( the reflecting surface coinciding with the diagonal of the trough ) , being the same in both cases , the differenco in the intensity of the light could only be due to the loss caused by reflection .
The light not being parallel , but forming a slightly divergent beam , the eflect of introducing the trough of water between the source and the photometer was equivalent , in an optical sense , to slightly reducing the distance between them ; it was therefore necessary that the light when falling directly on the photometer , and when doing so after reflection , should have traversed in both cases an equal thickness of water .
This condition prevented observations being made at angles other than The resuIts of four observations with the specnlum metal , and three with the steel mirrors , show that , as might have been anticipated , the percentage of light reflected was less than when the mirrors were in air ; the numbers are\mdash ; 1883 .
] on Iiletatlic Reflection .
39 Metal .
Steel .
58.73 . . . . . .
59.43 . . . . . .
44.74 . . . . . .
44.20 lfean . . . . . . . . . . . .
With the exception of the first determination made with the spcculum metal mirror , which is the mean of eight , each observation is the mean of four readings .
principal incidences and azimuths of the mirrors when in water were determined by the same method that had been used for ascertaining ths values of these constants in air ; the mirrors being surrounded by a cylindrical glass vessel filled with water .
P.IP .
ASpeculumMetal. . .
Steel .
P.A. . . . .
Mean . . . . . .
These values were troduced ito Cauchy 's , and the theoetical intensity of the hght reflected at determined .
Speculum Metal .
Steel .
Observed . . . . . . . . . .
Calculated . . . . . . . . .
4898 The diffel.ences between the observed and calculated values are nearly the same as when the measurements were made in air .
Note by the Communicator .
Received February 13 , 1883 .
The differences between the results of theory and observation as to the intensity of reflected light are in several of the above experiments so considerable , that we are led to ask whether there may not be someLing in the experiments to which it may be referred .
A slight inaccuracy in the calculated numbers is produced by the neglect of the polarisation due to oblique reflection from the paper employed .
With glazed paper this might be considerable ; but naturally a paper would be selected with as dull a surface as possible , and the author assures me that the polarisation was only just perceptible .
It could be easily allowed for by measuring the amount of polarisation produced by the oblique reflection .
When light of given intensity polarised first in and then perpendicularly to the plane of incidence is incident obliquely on the paper , Jet the intensities of the reflected Sir J. Conroy .
[ Feb. 15 , light be as to 1 .
Then it will be easily seen that the theoretical intensity for common light reflected first from the metal , and hen from the paper , as determined hy the method of the paper , will be instead of .
The former exceeds the latter by\mdash ; which is positive , since is a little greater than 1 , and is greater than .
This excess is very small , but as far as it goes it tends rather increase , on the whole , the between theory and observation , since the obsel'ved intensity nearly always falls short of calculated .
The imperfection of the polish in the case of such soft metals silver and tin has , doubtless , much to do with it .
But the polish of the steel seems to have been practically perfect , and yet this metal showed discrepancies , though not so great .
But I think there are strong reasons for believing that the ordinarily received formulae for metals can only give a more or less approximate resuIt .
MacCullagh was the first to show that by substituting for the refraotive index in Fresnel 's formulae a complex imaginary , and then interpreting the formulae as Fresnel has done in a somewhat analogous case , results were obtained agreeing , at any rate approximately , with those deduced from observation .
Cauchy afterwards ve formulae substantially the same , as they differ only in algebraic development , but made an important advance in the physical theory by connecting the coefficient of in the complex imaginary with an intense absorbing action of the medium .
But metals are not the only bodies to which the formulae of Fresnel do not apply .
More than fifty years ago Sir George Airy ; howed that in the case of diamond a considerable quantity of light polarised perpendicularly to the plane of incidence was reflected at the angle which made the nearest approach to a polarising ; and that on increasing the angle of incidence through the angle of maximum polarisation there was a apid retardation of phase .
Similar phenomena were afterwards observed in other transparent substances of high refractive index , and more recently M. Jamin has observed them in transparent substances in general a few exceptions .
The effect increases on the whole with the refracve index of the substance , but not in such a manner as to allow us to suppose that it is a function of the refractive index .
Hence two independent constants are required to define for a given kind of homogeneous light the optical character of a transparent substance .
1883 .
] Some on IlfetaUic Reflection .
The propert of a substance , as for the sake of a name it may be called , increasing on t , he whole with the index , and consequently on the whole with the density , there can be little doubt that if metals , retaining their actual density , were transparent like diamond , they would exhibit this property in a greatly exalted degree .
On the other hand , the metallic properties connected with intense absorption are exhibited by many non-metallic substances , such for example as the colouring matters derived from aniline .
Hence we have very strong reaso1l for believing that there are two distinct and independent in a metal by virtue of either of which , if it stood alone , there would be a change of phase and persistence of intensity light polarised perpendicularly to the plane of incidence as the angle of incidence was increased .
The to these two wou no two would not follow the same laws as regards their dependence on the angle of incidence .
The coefficients expressing these two properties , together what answers to the index of a transparent substance , make three physical constants which are required to define a metal optically , even in relation to homogeneous light .
If we confine our attention to an experimental determination of the difference of phase and ratio of intensity for light polarised in and perpendicularly to the plane incidence , and if , iecting altogether the adamantine property , we determine the two constants in tlJe ordinary formulae for metallic reflection so as make them agree with observation in two suppose by giving a difference of phase of for an observed angle , and an observed ratio of intensities at that angle , it may be readily imagined that the ordinary formulae may be to a certain extent formulae of interpolation , giving the difference of phases and ratio of intensities for other angles of incidence without any very material error ; and yet when we come to a botally.erent kind of observation , such as that of determining the ratio of the intensity of incident to that of reflected light , that the formulae may be found to be very distinctly in error .
Hence observations of this latter class seem deserving of more attention than have lately been bestowed upon them , lest from too great reliance on the accordance between theory and observation as ards the difference of phase and ratio of intensities when we compare light polarised in and perpendicularly to the plane of incidence , we should be led unduly to trust the formulae , for giving correctly the ratio of intensities for incidcnt and reflected light.\mdash ; G.G.S.

