]\gt ; On Metallic Reflection and the of the Layer of Transition .
* By RICHARD C. MACLAURIN , M.A. , LLD .
, late Fellow of St. John 's Cambridge ; Professor of Mathematics , Wellington , New Zealand .
( Communicated by Professor J. Larmor , Sec. R.S. Received January ll , \mdash ; Read January 25 , 1906 .
) It is well known that when light is propagated in an absorbinglnedium , the dynamical equations and the boundary conditions are of exactly the same form as for a transparent medium .
From a mathematical point of view the only difference between the cases is that , the refrative index in a transparent medium , is replaced in the absorbing medium by a complex quantity , where is the " " refractive index\ldquo ; of the medium , i.e. , the ratio of the velocity of light in air to that in the medium , and is the ooefficient of absorption .
When dealing with the problem of reflection we shall take the plane of as that of incidence , and as the surface of separation of the two media ; the vectors representing the displacements will then be of the forms in the incident , and in the reflected wave .
Here is the angle of incidence for the frequency , and V the velocity of propagation in the first medium .
The incident wave is of unit amplitude , and if , then and represent the amplitude and change of phase in the reflected wave .
When the incident light is polarised perpendicularly to the plane of incidence , is given Fresnel 's formula where is the angle of refraction .
For light polarised parallel to the plane of idence we have the corresponding formula When appiying these mulae to the problem of metallic reflection , we have to replace by the complex .
Then becomes complex , With regard to the paper " " On Newton 's Rings formed by Metallic Reflection\ldquo ; .
Proc , vol. 76 , 1905 , , of which the proofs were corrected in England , in order to avoid delay , Professor Maclaurin writes : I have in one place iuadvertently put instead of the complex ) .
Unfortunately this affects some of the results that follow .
The errors thus introduced are not numerous , and they do not at all affect the general trend of the argument .
Prof R. C. Maclaurin .
Metallic Reflection [ Jan. 11 , , and .
With this notation we get ; Whence we get where whence ' where and .
If and be given , these equations suffice to determine , and completely .
For some purposes we are mainly interested in the ratio , and in the difference of phase between the light polarised perpendicularly and parallel to the plane of incidence .
Now , from the above , we have Thus where and the Influence of the Layer of Transition .
The last equation shows that as increases from to , increases to .
We have , when nd this equation accordingly determines the Principal Incidence .
For this angle we have where and is the Principal Azimuth .
These are substantially the formulae obtained by Cauchy when discussing the problem of me allic reflection .
Befi ) putting them to the test of modern experiments , we shall make some transformations that will be useful for some purposes .
We know from the experimental investigations of Drude and othels as to the optical constants of metals , that is always large .
From Drude 's results it is least for copper , where its value is , and greatest for zinc , where it is .
This enables us to expand some of the above functions in ascending powers of , and so obtain approximate formulae sufficiently accurate for many purposes .
We have If we multiply each side of this identity by and by , and equate real parts , we get , Also we have Let us now consider how varies as increases from to .
For brevity write , and we then have VOL. LXXVIL\mdash ; A. Prof R. C. Maclaurin .
MetallicReflection [ Jan. 11 As is large , we have , as aJirst approximation , and , so that This is a maximum , so that is then a minimum .
Hence for polarised perpendicularly to the plane of incidence the intensity of the reflected light diminishes as increases , until it reaches a minimum in the neighbourhood of , which thus determines the " " quasi-polarising\ldquo ; angle .
When we have ( approximately ) and .
For a certain class of steel we shall find later that , and .
In this case , or , gives as the quasipolarising angle , and as the minimum value of the amplitude of the reflected wave .
This ation is , of course , somewhat rough , as we lave neglected squares and higher powers of .
Proceedin to a higher order , we get In the small terms we may put the results of the first approximation and .
This gives This makes a minimum when .
With the values of and given above for steel , this fixes the quasi-polarising angle at In light polarised parallel to the plane of incidence we have As increases denominator alters little , as is small compared with , while the numerator steadily decreases .
Thus B2 increases steadily , and as the equation to determine its maxima or minim no real roots , it has no maximum or have seen that the Principal Incidence is determined by the equation .
Putting we get as the approximate equation to determine the Principal Incidence , .
is most simply solved by approximations .
The first approximation yives that .
To this order of approximation and the Influerwe of the of Transition .
Principal Incidence and the quasi-polarising angle are the same , so as a rule the Principal Inoidence will be very near the quasi-polarising The second approximation gives , nd the third , If , an error of 1/ 40 in corresponds to an error of about ve minutes in , and this is within the limits of experimental error .
In his case , so that the second approximation will give the nncipal Incidence sufficiently accurately for comparison with experimental esults .
Returning now to the exact formula with which we began , we shall see ow they fit in with modern experiments on metallic reflection .
It has been usual* to compare them with the results of Jamin 's experiments ou steel and other metals .
When this is done it is found that the theory fits in with the experiment as far as the main features of the metallic reflection are concerned ; but it is only necessary to plot the experimental results to see that they are much too dis.cordant to a satisfactory test .
Since Jamin 's time there have been many experimental investigations into the phenomena of metallic reflection .
Amongst others , we have an elaborate series of experiments by Sir John on reflection from steel and speculum metal .
Conroy found for steel that the Principal Incidence was ( with a probable error of ) and the Principal Azimuth ( with a probable error of He made four separate sets of experiments for reflection of light polarised perpendicularly and parallel to the plane of incidence .
By rejecting the most discordant of the four when there is considerable discordance , and taking the mean of the remaining ones , and by taking the mean of the four where , as is generally the case , there is no great discordance , we get the following table , the notation being the same as at the outset of this paper.\mdash ; * See , e.g. , Mascart 's ' Tr it6 d'Optique , ' t. 2 , vol. 13 .
See ' Roy .
Soc. Proc vol. 36 , p. ' Prof. R. C. Maclaurin .
fetallic Reflection [ Jan. 11 ; theory experiment .
The values of are almost uniformly between 1 and 2 per cent. larger than those found by experiment , while the values of R2 are larger by quantities varying between and 1 per cent. The differences are rather too to be put down to experiment , and the that they are all of the same sign makes it improbable that they represent experimental errors .
Moreover , on making a similar calculation for speculum metal , we find that the values are in this case always larger than those found by Conroy , the difference reatest for R2 , where it amounts to per cent. in some cases .
When dealing with transparent media it has been found that the discrepancies between theory and experiment have disappeared as soon as it has been recognised that the transition from one medium to another is , as a rule , gradual and not abrupt .
In the present paper we shall investigate the extent to which this idea will ssist us when dealing with the problem of metallic reflection .
When the layer of transition is taken into account , the formulae for and take the forms* and in our present notation ; so that we have and Here , where is the thickness of the layer , and the wavelength in air .
and are complex constants defined by the equations and .
The values of these constants depend , of course , on the law of variation of in the layer of transition .
This being unknown , we cannot determine and , but we should expect them to lie between 1 and .
If had the value in the layer , we should have and so that if ( as with steel ) the modulus of were about 13 , would be about 7 and about 2 .
But in any case , owing to the largeness of , the term in the formula for will be large compared with , except where is * See a paper by the present writer , ' Roy .
Soc. Proc , vol. , pp. and ) and the Influence of the yer of Transition .
nearly when the term introduced by the layer of transition becomes , which is small compared with .
The sequel will prove that very small , so that the changes due to the layer will be small .
Putting , neglecting the term in comparison ith , and putting in the small terms , we here is small compared with and If the modulus of be , where obtained from the formula of 212 , then is the correction to due to the layer of transition .
We have ' whence we have , approximately , whence , employing the approximate values , we have ; ; Making these substitutions in the formula for , we This makes vanish when and also when .
Prof. R. C. Maclaurin .
Reflection [ Jan. 11 , In considering the law of variation of , it is convenient to ascertain the position of its maxima and minima .
Putting we have proportional to , where .
In the small terms we may use the approximate formulae and We then have and , where Thus The equation to determine the maxima and minima is , then , With the values of and found later for steel this equation becomes Solving this , by Horner 's process , we get two real roots , and The former corresponds to and determines the position of the minimum , the latter corresponds to and determines the position of the maximum .
The graph below ( fig. 3 ) represents the march of the function in the case of steel .
and Influence of the of sition .
similarly with r we note that the modulus of is lalge with that of Neglecting the latter term in parison with the former we get If the modulus of be , so that is the correction due to the yer of transition , we have ; whence , approximately , ; : ; so that and will thus vanish when .
It would also vanish if ; but , as a rule , this will be larger than unity , and so there will be no real value of to satisfy it .
In the formula for the term will usually be negligible comparison with , so that we shall have , approximately , Now .
For almost all the metals is greater than , so that is positive .
Thus is greater than and is less than .
Owing to the factor will thus be small and it will diminish steadily with .
If we investigate the position of the maxima and minima in the same manner as we did with we are led to the following equation:\mdash ; As cannot be greater than , and the above equation in has no real roots less than , so that there are no maxima and minima .
Prof R. C. Maclaurin .
Metallic Reflection [ Jan. 11 , The graph above ( fig. represents the march of the function in the case of steeL It will be seen from this figure that the range for which and are appreciable is much larger than in the case of a rent medium , where the influe1lce of the transition layer is practicalIy confined within a few degrees of the polarising angle .
It has been observed before that for many purposes the ratio of to and the difference of phase between light polarised perpendicularly and parallel to the plane of incidence is what is wanted for comparison with experiment .
We proceed to develop some formulae suitable for end .
We have Thus , whele the moduli of and are small compared with those of A and where Now and the Influence of the Layer of Transition .
embering that and get Hence , in the notation of p. 219 , , where To determine the maximum value of we put , as before , and the approximate relations and , we have to make a maximum .
This requires Solving this by approximations we get as the first approximation .
This , as we have seen ( p. 214 ) , is the first approximation to the quasipolarising ] .
A second approximation ives p. If and , as we shall find later for a certain class of steel , the first approximation ives , and the second which is very near the Principal Incidence .
If we put , we have and the difference of phase is Here and If , as will usually be the case , be small , will be small and the correction to the change of phase will be small .
It will be greatest when is largest , i.e. , in the neighbourhood of the Principal Incidence .
Perhaps , however , the most important point to notice is that , even although the correction to the change of phase be small , it may make an preciable difference to the position of the Principal Incidence .
At the Principal Incidence we have , therefore or This is the equation to determine the Principal Incidence , and owing to the presence of on the right hand side , that term may be appreciable even although be small .
Prof R. C. Maclaurin .
Reflection [ Jan. 11 , It would seem then , that two constants ( and ) are not sufficient to describe the optical properties of a metal\mdash ; at least two more ( a and ) are required in the problem of reflection , these latter constants depending on the law of variation of in the layer of transition between the media .
This being the , we cannot derive the optical constants from observations of the Principal Azimuth and Principal Incidence alone .
The simplest method is to proceed by successive approximations .
The true values of and will ) smaller than those obtained by negecting the layer of transition and proceeding as on p. 216 .
On obtaining approximate values of and in this way , we can calculate the constants and from observation of the Principal Azimuth and Principal Incidence ( ) .
We have , ( i ) and ( ii ) These are two equations , from which and may be readily determined .
Equation ( ii ) determines , and we then have and once determined we derive from the equation We shall apply this method to Conroy 's experiments on reflection from steel .
The values of and already obtained are too high .
We shall take as the next approximation and , although the sequel will prove that these are still too high .
Taking and from Conroy 's experiments , we the following from the equations just obtained:\mdash ; ; ; ; ; ?
With and the formulae of ) give us the table : and the Influence Layer of nsition .
for water , In the case of air we get , which is within half degree of the value derived ( on p. 224 ) from Conroy 's experiments .
The orresponding value of makes and Effect of Changing the Mediurn in Contact with the Metal .
It has been found by experiment that the Principal Incidence and Principal Azimuth depend not only upon the nature of the reflecting metal , but also upon the medium in contact with it .
Quincke* made some vestigations into this matter , but the subject has been discussed much more completely in a long series of experiments on gold and silver by Sir John Conroy .
Conroy could find no simple relation between the changes in the values of the Principal Azimuth and Incidence and the indices of the media .
His results , however , are in complete accordance with the trend of this paper , as is also the observation of Drude that surface impurities tend to reduce the value of the Principal Incidence .
If the medium in contact with the metal be of refractive index , we have to replace in our earlier formulae by , keeping as before .
We have seen that for an abrupt transition the Principal Incidence is given by the formula , which is approximately equivalentS to se .
If there be a layer of transition we have the equation Now is always small , so that in the last term we may put and we then get , where , and is small .
Thus the effect of the layer is to replace by or by , while the effect of replacing air by a medium of refractive index is to replace by Hence,.if there be a layer of transition between a medium and the metal , the Principal Incidence will be determined by the equation where 'Pogg .
Ann vo ] .
188 , p. 641 .
See ' Roy .
Soc. Proc vol. 31 , p. 486 .
See Drude , ' Wied .
Ann vol. 36 , 1889 , and , 1890 .
8 See p. 218 .
Prof R. C. Maclaurin .
Reflection [ Jan. 11 An increase of will diminish and so diminish , the Principal Incidence .
The following are the means of Conroy 's experimental determinations of the Principal Incidences , the incident light being yellow:\mdash ; Gold in air 71 43 Silver in air 74 37 water 67 39 , , water 72 15 carbon bisulphide 66 36 , , carbon tetrachloride 71 39 A change in will also affect the Principal Azimuth ( ) .
We have , very nearly .
The angle is determined by the equation An increase of will diminish and therefore increase , so that as a rule will be increased slightly , although in some cases the increase of may be counterbalanced by the diminution of the factor These are the means of Conroy 's experimental determinations of the Principal Azimuth , corresponding to the Principal Incidences above : Gold in air 41 14 Silver in air 43 water 41 15 , , water 44 9 carbon bisulphide 41 41 , , carbon tetrachloride 43 40 It will be observed that these results are in general accord with the argument above .
Unfortu ately , however , we have not sufficient data to put the theory of a layer of transition to the exact test of numerical verification or otherwise .
We have seen that a knowledge of the Principa Incidence and the Principal Azimuth is not en ough to determine the optical properties of a metal , and Conroy 's results do not enable us to supply the deficiency .
Even if we had sufficient data to determine the constants and for air , we could still do little better than guess what they would become when some other medium was in contact with the metal .
We have , so that the modulus of should be increased by an increase of *Cf .
p. 230 .
and the Influence of the of nsition .
an increase of might be expected to raise and therefore also snd .
These quantities and will be er raised when is replaced , for this will increase , and so increase diminish \mdash ; should expect , then , that the diminution of , due to the of oslashuld be greater if there were a layer of transition than if there were none .
be effect on the Principal Azimuth is not so easily described .
An qrease of will raise , but it will diminish , so that we cannot say general whether will be increased or diminished .
A little investigation will show that Conroy 's experimental results are consistent with the theory of an abrupt transition from one medium the other .
Thus for gold in air with yellow light his values of the Principal Azimuth and Incidence would give and on the theory of an abrupt transition .
For gold in water these constants would lead to and } , whereas Conloy found and .
For gold in carbon bisulphide theory would give and instead of and as obtained by Conroy .
With different colours for gold and also for silver the same discrepancy between theory and experiment would also be apparent , the differences being in nearly every case in the same direction , the theoretical results being too large .
This discrepancy is just what the above discussion would lead us to expect , if there is a layer of transition between the two media .
The chief results of the present investigation are the following:\mdash ; 1 .
That in metallic reflection , if the transition from one medium to the other be abrupt , the Principal Incidence is always near the quasipolarising angle , and is given very approximately by the formula 2 .
That even wheh great care is taken to clean the surface of a metal the transition from it to the neighbouring medium is often gradual and not abrupt .
This is in accordance with experimental and theoretical investigations on reflection from trampcu.ent substances such as glass and diamond .
3 .
That the influence of this layer on the ellipticity of the reflected light and on the difference of phase between light polarised perpendicularly and parallel to the plane of incidence extends over a wider range than in the case of transparent substances .
4 .
That the thickness of the layer is of about the same order of magnitude as with transparent media .
Mr. W. Spens .
The Osmotic Pressure the [ Jan. 3 , .
That the layer affects the position of the Principal Incidence considerably , and also influences the Principal Azimuth .
6 .
That , consequently , the deduction of the optical constants of a metal from observation of the Principal Incidence and Azimuth alone is liable to considerable error .
[ In the case of steel this method leads to ( see p. 216 ) , while the wider theory yields and 7 .
That four constants are required to describe the optical properties of a metallic reflector , two of them depending on the nature of the layer of transition .
S. That with these four constants a very satisfactory agreement exists between theory and experiment , as regards both the intensity of the reflected light and the difference of phase between the lights rised p dicularly and parallel to the plane of incidence .
The Betwecm thoe Osrnotic Pressure and the Pressure in a Concentrated Solution .
By WILL SPONS , B.A. , King 's College , Cambridge .
( Communicated by W. C. D. Whetham , F.R.S. Received November 11 , 1905 ; received in Revised form January 3 , \mdash ; Read January 25 , 1906 .
) 1 .
The relation between the vapour pressure and the osmotic pressure of a solution is often investigated by considering the equilibrium of a column of solution separated at the ottom from the pure solvent by a semi-permeable membrane , and placed in an atmosphere of vapour from the solvent .
Now the hydrostatic equilibrium of the vapour column gives where is the vapour pressure of the pure solvent , the acceleration due to gravity , the height above the surface of the pure solvent , and the specific volume of the vapour .
Hence considering the equilibrium of the liquid column we get when is the osmotic pressure , the vapour pressure of the solution , that of the pure solvent , and where is the effective mean density of the column of liquid .

