]\gt ; ANumerical Examination of the Optical Properties of Thin Plates .
By RICIIARD C. , LL. D. , formerly Fellow of St. John 's College , Cambridge ; ]'rofessor of Mathematics , Wellington , New Zealand .
Communicated by Professor J. Larmor , Sec. R.S. Received August 30 , 1906 .
) The optical properties of thin metallic plates have been investigated by nulnber of physicists .
* One of the earliest workers in this field was He predicted from theory and verified by experiment that if light incident on a gold leaf were plane polarised the ansmitted beam would be elliptically polarised .
With the improvement in experimental methods since MacCullagh 's day , and the gradual removal of obscurities from the theory of metallic reflection and transmission , we now expect much more than a mere general agreement between theory and experiment .
We look for an almost exact erical coincidence .
The condition of the reflected or transmitted beam is precisely described by means of two quantities\mdash ; the ellipticity and the difference of phase between the components of the light ) olarised perpendicular and parallel to the planc of incidence .
The object of the present paper is to obtain convenient formulae for these quantities and to compare them with the results of experiments , the most careful and the most recent that available .
We shall admit into our theory no principle that has not foumd yeneral acceptance , S and shall thus be enabled to decide whether such principles are sufficient to the facts .
If they fail in this , it will behove us to look for new , by a scrutiny of our so-called " " facts to indicate in what way experimental errors have brought about an apparent conflict between fact and theory .
Suppose that a ray of light is incident at angle on a metal plate .
If the plane of be taken as that of incidence , then an incident wave of * See references in Mascart , 'Traite ' vol. 2 , p. 550 , and Winkelmann , ' Handbuch der Physik , ' vol. 6 ( 2 ) , p. 1311 .
Neither of these , however , mentions the vork of Conl.oy , ' Roy .
Soc. Proc vol. 3 ] , 1881 , p. 486 . .
Irish Acad. Proc. , vol. 1 , p. 27 , and ' Collected Works , ' p. 82 .
, e.g. , ' Roy .
Soc. Proc , vol. 77 , S Of recent investigations the most elabolate is of M. Meslin , ' Ann. de Chimie et cle Phys 1890 , pp. 56\mdash ; 177 .
After giving the results of experiments , M. Meslin obtains , Jmpirically , a formula for of produced by qsion through a late .
He then makes an ingenious attempt to deduce formula from theory .
His gulnsnt , however , contains too many ) ecial hypotheses and assumptions to be final .
Optical Properties of unit amplitude will be represented by a vector of the form where is the frequency , the time , V the velocity of in air .
This incident wave will give rise to a reflected wave represented by and a refracted wave .
The dynamical equation requires that , where is a complex of ) .
Here is the " " refractiye index\ldquo ; of the metal , 2 ' .
the ratio of the velocity of light in air to that in the metal , and is the coefficient of absorption .
The refracted wave , which we shall refer to brietly the wave , will , after traversing the plate , become .
When it reaches the bounding surface it will be reflected and refracted .
In this waves will surge to and fro acl .
OSS the plate , while an infinite series of waves will return into the first medium ( air ) and constitute the reflected beam , and another series will pass into the third medium ( air ) and constitute the transmitted beam .
We shall denote by and the quantities to and when we are considering a ) from metal to instead of from air to metal .
With this notation the reflected beam is represented by ' and the nitted 1 by .
The values of the quantities and and are jive by the theory of reflection and refraction , and in this way we can easily ) that and esults can also be ined by means of Stokes ' principle of eversion , or ) noting that when the -ness of the metal is indefinitely diminished and becomes unity , then all the transmitted and none Making use of these relations , the reHected beam becones The first refracted wave was , where Hence , if be the thickness of the plate , we have where is the wave-length of the incident we have , tall , where , and tau We thus get , where and These relations suffice to determine in terms of the optical constants of the metal and the thickness of the plate .
quantity is a complex .
Its value is given by theo1y of reflection and proves to be dependent on the position of the plane of tion of the incident .
If we egaxd the transition one the other to be abrupt , then , when the incident light is polarised perpendicular]y to the plane of incidence , is given by Fresnel 's folHlula .
For ised parallel to the ) } incidence , have the corresponding We ] lave and putting where while Similarly \ldquo ; , ' ' If the incident wave ) elCprebented ) , then the eflected beam is represented by plitude and the change of phase produced by reflection .
We .
Soc. Proc 212 .
of the Optical Properties of ) Jletallic ) We thus get where This determirle , the intensity of the reflected bealn .
Vhen the of the plate is indefinitely small and .
Hence , so that there is no reflected beam .
As the thickness increases , approaches the linlit zero , so that approaches the limit \mdash ; as is to be ) ected .
It llay be to observe , vever , that the intensity of the reflected light may be greater for a thin plate than for a thick one .
wheu , , where Consider the case of direct incidence .
We then , where cqufttio1l to deterlninc becomes For the silver and gold , to be con sidered latel , ( 1 is , so when is not very small , the quantity will be very small , we } very approximately .
This .
In of silver this makes .
This is , so that our approximation is sufficiently accurate practical that can thus easily find the that gives the netals discnssed later this ) in the case of silver , and in the case of .
If the tness is rather greater than this , we reater than , but as the thikess increases , once more approaches to equality with will then ] ) thickness that makes maximum , and it is not to deterlnine this thickness .
We have ( is to ) we get , by to zero , i.e. As is small , this can easily bc by approximations .
The first gives i.e. , For silver , this makes , which is so small that the hrst roximation is sufficie ] accurate .
The thickness that gives the greatest intensity of reflected light is determined by in the case of silver , and in the caso of gold .
In both cases , however , the of is greater than The table shows the intensity of the reflected depends on the thickness of the plate in the cnse of and silver .
The intensity , of course , depends on the values of the ) tical consGants and , and in turn vary with the colour the incident .
From ) 's experiments* with well cleaned silver in light , he calculated that , and , and from sinilar iments by Conroy , with in red we find that , and ' Wied .
Ann vol. 39 , p. 481 . .
Soc. Proc 1881 , p. 486 .
These constants are deduced from tions of the Principal Incideuce and the Azimuth .
It has often been bserve that these angles vary with different conditionls of the reflecting Slll.face .
That this is to be expected is apparent if we regard the transition one nledium to the other as gradual and not abrupt .
The Principal cidence and Azimuth depend on the nature of the layer of transition , when the matter examined efully it is found that the of the transition layer lead to considerable ) in the values of the optical constants .
See ' Roy . .
Proc 190 ) .
, and post , p. 3 : , vote present we shall take the value derived from the experiments of lJrude and Conroy the hypothesis of an abrupt transition as sufficiently accurate hrHt approximation to the truth .
302 Prof R. C. Maclaurin .
We have when is satisfied , very approximntely , by .
For silver this corresponds to , and for To make a maximum we must have As a first approximation , i.e. , , so that .
For both silver and gold this will be found to make nearly unity , so that is very ] , and the first approximation is sufficiently close .
The maximum value of occurs when we have seen that when is less than , so that is reater than ; but in the case of the silver and gold , ah.eady discussed , it is not very much reater , so that the maximum is reached soon after is equal to silver the maximum occurs when and for when The following table gives the value of , ( ' xpressed as fraction of the halfll.ave-length .
These results ftre represented boraphically in below , Curve 1 correto the silver and Curve 2 to the gold .
Perhaps the most striking thing about these results is the large percentage of the change of phase due to a thick plate which is produced by a very thin plate .
In the one case 64 per cent. and in the other 70 per cent. of the total change is produced by an thin plate .
It must be borne in mind , however , that the intensity of the reflected light for such thicknesses is vanishingly small , so that it would not be practicable to measure the 1906 .
] of the Propcrtie Thin 300- that and that , we , to determine maximum of , the equation sill This is easily solved by approximations , and yields in the case of silver , and in the case of boold .
The value of , expressed as a fraction of the half lelJgth , is ooiven in the table , and the results are represented graphically in below .
As befc ) , Curve 1 corresponds to the silver and Curve 2 to the gold .
We shall now considel to extent the results already obtained as ) the intensity and change of phase of the reflected are modilied } the metal is deposited on eolass .
The second reflection and efraction in investigation of p. 297 now takes place at from to instead of at one from metal to air .
The quantities and will therefore be altered , the values given , as before , by 's formulae .
This method of attacking the problem is adopted later ( p. 313 ) , when we are with the more general case of incidence .
As , howevel we have already calculated the amplitude and phase of the iected and transmitted beam for a 1netal plate surrounded by air , it is convenient to utilise these results in the present problem .
enable to do this suppose the metal and glass sepnrated by an layer of An incident wave of unit amplitude gives rise to a reflected wave VOL. LXXVIIL\mdash ; A. -X 306 Prof. R. C. Maclaurin .
[ Aug. } where and are and already calculated for the reflected wave .
It also produces a transmitted wave and havin. .
also been calculated ) .
This wave reflected glass , glVlI ) a wave , in its tuln is transmitted the metal plate and emerges into the air as a wave say .
The complete reflected wave is thus the resultant of the two compo ents and .
If , as before , we denote it by , we have and tall If tIre incidence be direct , then is given by Fresnel 's formula where is the efractive index of the glass .
Thus and For small thicknesses is very small , while is considerable .
Under these circumstances is nearly equal to A and to .
As the thickness increases , rapidly nishes , so that and approximate more and more closely to and .
Thus , except very thin plates , the will be the same as if the metal were SUl.rounded by instead of deposited on glass .
With very thin plates , however , the two cases will be quite different .
we get the following values of 1906 .
] of the Optical Properties of Plates .
These results are replesented graphically in above .
Curve sponds to the silver and to the gold .
A comparison of Curves 1 and 4 or Curves 2 and 5 will show the influencc of on the intensity of the reflected light .
The difference of phase is given in the table .
snlallIndef .
Silver Fig. 2 above ives a graphical representation of these results , Curve 4 or the silver and Curve 5 for the gold .
As is to be expected , the ence etween metal alone and metal on ylass is very marked for small thicknesses .
all cases the change of phase is less for metal on than for metal , lone , but the difference between the two cases diminishes rapidly as the hickness increases .
After a thickness has been reached there scarcely any appreciable difference between the results , and for the sake clearness the curves for on glass have not been drawn beyond this oint .
If we compare these results with Wiener 's observations*of the change of has produced by reflection from thin films of silver deposited chemically lass , we see that there is a close reement .
The of phase akes place rapidly , not so rapidly as Wiener 's first , but lore rapidly than with his second mirror ( B ) .
With the silver , , bove , about 65 per cent. of the total chan , .
of phase is reached when , i.e. , whelr mm. .
With Wiener 's mirror almost the whole change took place within a thickness , but yith the second mirror the corresponding thickness was 12 So far we have been ) mainly with a discussion of the intensit .
phase of the reflected and light in the of direct ncidence .
We shall now return to the more general problem stated at the nset , that of the ellipticity and the differeuce of phase between he parallel and perpendicularly polarised light for any angle of incidence .
the reflected light we have seen ( p. 298 ) that .
' Wied .
Ann vol. 31 , p. We have said nothing of the ence of the glass on the transmitted beanl .
It is hown later ( p. 337 ) that for direct incidence the ellipticity is not affected by the glass .
Prof R. C. Maclaurin .
Hence Thus if be the ellipticity and the azimuth of the reSlected light we have .
As the thickness of the plate increases , diminishes and approaches the .
We shall calculate for different thicknesses of the gold and silver already discussed .
If be the of phase , we have , where is given by the formula Having obtained for a series of angles , we can represent the resnlts graphically and so determine the angle of incidence that makes equal to a quarter of the wave-length .
This is the Principal Incidence and the corresponding value of is the Principal Azirmth .
We proceed to tabulate the values of the constants occurring in these formulae for the silver whose optical constants are and Tf , we have .
and .
The other quantities are to be calculated from the formulae ; ; ; ; ; The formulae for , and have been already given on p. 298 .
From these we get the following table .
The values of and in this and later taUes refer to the case .
As these quantities are directly 1906 .
] of the ptical Pof Plates .
In this case the reflected wave , instead of , is now , where is the same as before , the correspdio quantity , when a ray of light goes from glass to metal instead of to metal .
The quantity is the same as before , and in we must replace by and by , where and is the coefficient of refraction of the ylass .
Putting taking we the set of values:\mdash ; We have Thus the ellipticity is iven by the formula and the difference of phase is where From these formulae and the constants above we derive table for These results are represented in above and .
The dotted curves in exhibit as a function the thickness of film for difi.erenb angles of incidence .
The continuous curves of that 1906 .
] of the Optical Properties of ThiMtxllic Plates .
the polarising angle of the bolass , and we thus see we can pass from vitreous reflection to metallic reflection ffil intermediate tes by the aid of layers of silver radually i in thickness .
One of the most facts about these results is the very rapid change in the Principal Incidence produced by slight thickening of the film wheu the film is very.thin .
A of only ndth of the wave-len increases the Principal Incidence from This iudicates how considerable would be the influence of a surface layer of any kind on position of the Principal Incidence and serves to explain the observation all careful experimenters that the Principal Incidence alters with different conditions the reflecting surface .
* It is also in good reelnent with experiment by on the )ject .
A .
fihn was deposited on glass by electrolysis .
The film was so thin that its existence could not ] been suspected without previous .
It was quuite ossible to estimate the thickness exactly , but it certainly not than fivethousandths of a wave-length .
with this thin film the ) erties of he reflected from were considerably modified and the Incidence had increased from to onroy made various experiwith silver films on }lass .
The films varied in thickness cm .
to cm .
Taking .
for yellow , this makes vary from to } ) .
This lies between Curves 5 and 7 of the above .
Our raph makes the P Incidence increase in this range from to , an increase of 1o ConroY found an increase of .
Although the Principal Incidence rapidly at first , it soon tends to a constant value and there is little change when the thickness exceeds mm. QuillckeS made a large number of observations with silver films and red light .
He that values of the Incidence increased with the thikess , tended to a constant value and changed very little when the thickness of the film exceeded mm. To exhibit more clearly the influence of the glass on the Principal Incidence the results for silver alone and silver on glass are set out together and their differences stated . .
' Boy .
Soc. Proc 1 , p. 211 .
Contptes Rendus , ' vol. 76 , p. 866 .
Soc. Proc 1881 , p. 486 .
S 'Pogg .
Ann vol. 129 , p. 177 .
316 Prof R. C. Maclaurin .
Numerical mination It appears from this that , except for excessively thin fihns , the Principal Incidence is greater for silver on glass than for the same thickness of silver alone .
The difference , ever , is nowhere very considerable , and it diminishes to zero as the thickness of the l1letal increases .
The ellipticity is boiven by the fornlula on p. ; it is most conveniently expressed as the gent of the , whose value is set out in the A comparison of figs. 7 and 10 will indicate clearly tlJe change in the azimuth produced by the silver on .
It will be seen ] Curves 4 , 5 , 6 , and 7 in the two ores are very similar , so that after a thick1906 .
] of the Optical ]erties of fetallic Plates .
ness has been reached it does not make a great difference ether Che silver is alone or on glass .
As is to be expected , however , with very thin films the diffel.ence between the two cases is marked .
On comparing rves 1 , 2 , and 3 in the two figures see that they are completely different .
For silver alone the azimuth incr with the thickness , whereas for silver films on glass it rapidly at first .
The Principal Incidence been obtained above it is easy to determine the Principal Azimuth from the .
The position of the Principal Incidence { Azimuth is 1narked in fig. 10 by a cross ( ) .
The tabl the Principal Azimuth for different icknesses and compares it with the case of silver alone:\mdash ; The Principal Azimuth is thus always gl'eater for silver glass than silver alone .
The difference is very considerable , but diminushes rapidly to zero as the thickness increases .
When the thickness changes from to the Principal Azimuth increases .
Conroy* found a much large increase , viz. , , in the same range .
It appears , however , that the variation of the Principal muth with the thickness .
considerable in the yhbourhood of .
When the thickness increases from to there is a change of in the Principal Azimuth .
Thus a error in the estimate of the thickness of the thinnest of Conroy 's films would for the apparent discrepancy between theory and experiment Havin discussed the of the light reflected from silver films , we proceed .
to estimate the extent to which these results are modified whenl lnlS of some other metal are employed .
So far our results have been obtained on the supposition that the transition from air to the metal is abrupt and not gradual .
This , of course , is only an to the truth ; but in the case of silver the approximation has proved to be sufficiently close for the experimental results .
When the influence of the layer of * Conroy , loc. , p. 31 thickness was calculated from the weight of a measured was necessarily some uncertainty as to the density of the silver in so finely divided a state .
318 Prof R. C. Taclaurin .
transition is considered it is found* that the difference of phase is altered by a quantity which is a maximum the Principal Incidence , where it is iven the formula .
' enables us to determine for what metals the correction due to the layer of transition is most considerable ; e.g. , it would lead us to expect ) influence of the layer would be greater with gold than with silver .
As , however , the analysis is necessarily much more complex when there is a snrface layer to be taken into account , we shall hold fast to the hypothesis of a sudden transition from one medium to the other as a sufficiently close approximation to the truth , unless we are driven from by too great a discrepancy between theory and observation .
The metal that lends itself most readily to accurate experiments with thin lates is with red , Conroy found a Principal Incidence of and a Principal Azimuth of as the mean of several determinations .
On the supposition of a sudden transition from air to gold the optical constants can be derived from these angles ) means of the formulae ' with the ] for and already ooiven ( 1 ) 308 ) .
In this way we obtain and .
With these constants and the formulae on we the following table:\mdash ; With the aid of these quantities we derive the following series of values from the formulae * See ' Roy .
Soc. Proc 1906 , p. 224 . .
cit. , p. 213 .
320 Prof R. C. Maclaurin .
Nn than for silver .
For siiver we that when when oInci JIncidence wthick poldis sfrom , howeveritsfinal value .
* Conroy experimented with gold leaves varying in thickness cm .
to cm .
For red , taking cm .
, this makes vary from to 0149 This range lies between the Curves 5 and 7 of fig. 11 , and we see that the variation of the Principal Incidence is small .
The increase is only ) a degree , whereas Conroy found an increase of .
The agreement is quite as close as could be expected considering the discrepancies between the different experimental results , uncertainty as to the optical constants and the specific gravity of the metal in the state of gold leaf , and fact that we have neglected influence of the glass on which the leaves were Conroy 's conclusion that the Principal Incidence always increases with the thickness of the llletal is supported by our theory with the optical constants that we have adopted for silver and boold .
However , Meslin 's experiments with gold a much wider range of thickness show that the law is not universal , and we shall see later that this is supported by theory when the transition layer is taken into consideration .
The azimuth is given in the table:\mdash ; These results are exhibited in below .
On .
this with above it will bc seen that there is not much ] difference between the behavioul ' of the two metals .
The Principal Azimuths for the various thicknesses are found from the figure to be , and respectively .
The crosses in the esent the positions of the Incidence and Azimuth .
From the -k It must be remembered that is not the same in the two cases , the incident being yellow for silver and red for gold .
This , of , affects the estimate of the thickness , and a slight change in this would easily account for the apparent difference between theory and experiment . .
p. 316 .
For such thicknesses the effect of the glass is small .
1906 .
] of the Optical Properties of Thin S25 The following table ives the values of the azimuth These results are represented in fig. 13 above , and on comparison ) fig. 7 we see that there is very little difference between the two metals regards the law of variation of the azimuth .
of Layer .
So far we have proceeded on the hypothesis that the transition medium to another is abrupt and not actual .
This layer transition has led us to results that quite as well as could be expected with experiment .
Unfortunately , however , the experiments that we ] ) been able to quote are all too few in number .
* They deal almost exclusiyely with observations of the Principal Incidence and Azinluth , and eve1l these the vations extend over a rather of thickness the metallic film .
We shall now turn , however , to a much more exteusive of experiments giving us the difference of phase for reflection and transmission over a wide of thicknesses .
These experiments were made with l of and we have already ( p. :318 ) been led to expect that for this the influence of the transitio1l layer on the diffel.ence of phase would be appreciable than with silver .
When we examine Meslin 's results for reflection from a thick plate of gold we find that the theory of an ab.ttl.ansiti ] ] not give us results that even approximately ( except in a very way ) with the facts of experience .
It is known that on snch a theory the difference of phase is given by the formula and the Principal Incidence by the , or its equivalent , .
* It would be a great advantage if we could obtain accurate measurements of difference of phase and azimuth both reflection and transmission .
leslin iveh us the difference of phase , but not the imuth .
See note p. 296 .
See ' Roy .
Soc. Proc 1906 , pp. 213 and 215 .
326 Prof R. C. Maclaurin .
mination [ Aug. 30 Meslin found the Principal Incidence to be , and as must lie between and , we see that must lie between and When the incidence was , Meslin found , and from the formula for we see that cannot be greater than .
This lies considerably outside the limits for fixed by the Principal Incidence and we should be to a similar inconsistency if we considered the value of at any other incidence than that of .
The crosses in below represent Meslin 's values of for different incidences and it will be seen hat they are very fairly consistent .
* The continuous urve of the represents the theoretical results obtained later when the layer of transition is considered .
We proceed to develop some formulae applicable to the problem when the transition layer is taken into account .
In that case in addition to the optical constants , and , two new constants and are required whose values depend on the law of variation of within the layer .
The quantities hitherto * At least ) to .
There is evidently a lnisprint in the values of given for and .
They are set down and 0.648 respectively .
An inspection of the figure shows that these are quite inconsistent with the other results and with the graph given by Meslin .
Probably the numbers should be and , 5 being replaced by 8 , and the two values interchanged .
Owing to the uncertainty of these results they have been neglected altogether in the subsequent determination of the optical constants of the gold .
1906 .
] of the Properties of Thin represented by 21 and are replaced by and , which given by the following , say , nately where where ; is the difference of phase that an abrupt transition would produce , and is given by the formula for on p. 325 , while is the correction due to the transition layer .
Since and must be calculated in order to determine , it will be convenient to express and in terms of these quantities .
From the bove formulae we have , whence and Similarly See ' Roy .
Soc. Proc 1906 , 222 .
Ibid. , p. 222\mdash ; 3 .
has been aced by and by to avoid confusion with quantity called in the earlier part of paper .
The quantity is printed with wrong sign in the paper referred to , and and of p. 224 should ench 1 by 328 Prof R. C. Maclaurin .
[ Aug. 30 , whence and where It thus appears that there will be no great difficulty in calculating all the quantities involved in our formulae , once the four optical constants and are obtained .
However , to obtain these four quantities accuratdy from a series of observations of the difference of phase at different angles of incidence is a problem involving some labour .
Theoretically , we coul obtain the four constants from any four observations of ; but the equations are much too complicated to be solved a set of four simultaneous equations .
ractically we must proceed by a series of approximations and in this consistent results are soon obtained , unless very great accuracy is aimed at .
From the formulae it appears that ( the correction to the change of phase due to the layer ) vanishes at normal incidence and is small when the angle of incidence is small .
We can , therefore , obtain a first approximation to and from two observations of the difierence of phas when the angle of incidence is not .
As a first approximation in such circumstances we can neglect altogether , or to it any small value that seems reasonable .
Knowing and ( approximately ) we can calculate , and from the formulae above .
obtained by calculation , and from observation for auy othel angle of incidence , thus for that incidence .
Also we have so that In this equation everything is known ( approximately ) except and , so that by using our of at two of incidence , other than those employed for obtaining the approximate values of and , we easily obtain the unknown quantities and thus obtained approximate values of the four optical constants , we can calculate in terms of them , and proceed , if necessary , to higher approximations .
If the expeximental results give us the difference of phase for more than four of incidence , we can use the ordinary rules for finding the most probable values of the optical constants , and estimate the probable errors .
In carrying the process sketched above with reference to set of 1906 .
] of the Optical operties of Thin Metallic Plates .
observations actually ayailable at present , we find that the approximations need not be carried very far , for we soon obtain results that fit in with experiment well within the limits of experimental errors .
We take Meslin 's results with red light reflected from gold ; and the second approximation in the above process yields the following values of the optical constants:\mdash ; .
From these we derive the following table , giving the values of , and the difl'erence between theory and observalion : These results are represented in above , in which the continuous ctlrve corl'esponds to the theory and the crosses to the experiment .
From the figure 01 the table we see that the agreement between theory nent is a close one , that , in view of the uncertainty as to some of the experimental results , not much could be , ained by to approximation .
We shall therefore take the constants ooiyen above the basis of our calculations .
From the iven we then del .
iv the .
table :\mdash ; We are now in a position to discuss all the problems with in the earlier part of this paper without ecting the layer of ansitio from one medium to another .
effect of this layer is to rep lace by quantities remain before , nrnst Prof R. C. Maclaurin .
[ Aug. remembered that on pp. 297 et .
refers to a tbickness of homogeneous 1net .
We shall sst out our results in terms of , but , of course , this is not the actual thickness of the plate .
To goet that thickness we must increase by twice the thickness of the layer , and it is owing to the uncertainty as to the value of this latter quantity that we cannot state the results precisely in terms of the thickness of the metallic plate .
If be the tbickness of the layer we should expect* to lie betweell and , and with the constants iven above this makes lie between and If be incident nornlally on a gold leaf whose optical constants are those just determined , then the intensity change of phase for the reflected transmitted are as follows:\mdash ; These results are esented graphically in Curve 3 of , 2 , 3 and 4 , above .
If the gold , instead of being surrounded by air , is deposited on glass , these results must be modified as in the cnse of silver on p. 305 .
The intensity and change of phase the reflected are then boiven by the table : See 'Roy .
Soc. Proc 1906 , p. 229 .
With the of variation of given on p. 230 , we have , for the gold alone , .
This makes , which is about half way between the limits found in the text .
is almost identical with the value obtained for the layer of transition in the case of reflection front diamond .
See 'Roy .
Soc. Prnc , vol. 76 , 1905 , p. 63 .
That so thin a layer should produce an appreciable change in the Principal )lcidence and Azimuth is not altogether surI)rising after what been seen on p. .
The gold under discussion here is a good example of the that may be made in deriving the optical constants of a metal from ervations of the Principal Incidence and Azimuth and neglect of the transition layer .
The Principal Incidence is and the Principal Azimuth , obtained from the formula , is ecting the layer of transition these would give instead of and instead of .
If , then we should get , instead of and instead of 1906 .
] of the Properties of Thin These results are exhibited in Curve 6 of figs. 1 and 2 above .
the first of these figures we see , by Curves 1 and 4 , or 2 and 5 , for silver or gold and an abrupt transition the intensity of the reflected light is always reater for the metal deposited on glass than for the meta ] plate alone .
The difference between the two cases is , however , snlall , and tends to zero as the thickness increases .
On Curves 3 an .
we see in this respect the layer of transition modifies the results considerably .
After a thickness has been reached , the intensity of the light from gold on glass is less than that reflected from gold alone .
To determine the ellipticity difference of phase when the is at any angle we have to modify the on by replacing by .
For the reflected light we have ' where and where Thus , where A and are the modulus and respectively of the complex quantity .
sin The difference of phase is and the ellipticity is From these formulae we get the following table for 336 Prof R. C. aclaurin .
Examination instead of being surrounded by air .
nlethod described on p. 306 above .
can be done most simply by the A beam of unit amplitude incident on the metal gives rise to a reflected and a transmitted beam S .
where , as we have seen , The beam is then incident on a glass plate and yives rise to a reflected beam and a transmitted beam .
Here , where is obtained from Fresnel 's formulae for reflection and where is the coefficient of reflection and is tbickness of the glass .
The wave leflected from the glass is transrnitted the metal plate and gives rise to a wave in the air .
Thus the 1906 .
] of the Optical of Metallic Plates .
wave is and the reflected wave is the resultant of and These results are quite general and would suffice to solve the problem iu the most complicated case when both plates ( metal and , lass ) were thin , i.e. , not large compal.ed with a wave-length .
In problem at present in hand the glass plate is thick , and under these circumstances there is a considerable simplification .
We then have and , so that the mitted beam is and the reflected one is the resultant of ) and The ransmitted wave is most readily disl ) of .
The effect of the glass is to replace by .
The factor is the same whether the light ) polarised parallel or perpendicularly to the plane of incidence .
Hence the oolass has no influence on the of phase nor on the ellipticity of the transmitted beam .
We should expect , then , that the results on p. 334 above should agree , within the limits of experimental , with Meslin 's values .
A comparison shows that there is a very fair reement as far the law of variation of is concerned .
to the uncertainty as to the thickness of the layer of transition and the difficulty of obtainino accurate measurements of the thickness of the metallic deposit , we cannot compare the numerical results of theory and experiment with much hope of exact equivalence .
As we shall see later , the experiments on reflection , when compared with theory , us a fair basis of comparison as far as thickness is concerned .
On determining in this way the thicknesses to be compared , it appears that the theoretical results , lGhough o much the same h of dependence on as Meslin 's values , are higher than these values .
However , as was obseryed on p. 335 above , a change in the optical constants makes an appreciable difference in the difference of phase of the transmitted \mdash ; and in view of this fact tlJe agreement between theory and experiment as close as could be expected .
Unfortunately , Meslin made no record of the ellipticity , so that we have no means of comparison in this The reflected light is the resultant of and this by , we have , and The difference of phase is iven by and the ellipticity by VOL. LXXVIIL\mdash ; A.

