]\gt ; PROCEEDINGS OF THE ROYAL SOCIETY .
EBJrATICAL AND PEySICAL ScIBycBS .
Optical Dispersion : A of the Maxima of Absorption SelectVve Reflection for Substances .
By T. H. HAVELOCK , M.A. , D.Sc .
, Armstrong College , Newcastle-on-Tyne .
( Communicated by Prof. Sir J. Larmor , Sec. R.S. Received August 8 , \mdash ; Read November 9 , 1911 .
) Introduction and Summnry .
In the comparison of optical dispersion formula with experimental results , it is important to consider the meaning of the various constants , and in particular of the natural wave-lengths which occur in them .
With the formula in the simple form , the quantities are the wave-lengths of the natural vibrations and have been identified in practice indiscriminately with the wave-length of maximum absorption or that of selective reflection obtained by the method of residual rays .
This interpretation is a]lowable if the absorption is very small .
For a more general type of formula such as , the corresponding wave-length , on a similar argument , for small absorption is a certain wave-length , which is larger than ; in a previous paper* numerical examples were given of the difference between the wave-lengths and .
The present paper includes absorption more generally in the formulae , so as to determine the position of the various maxima more accurately .
Using dispersion formulae of a general type , expression is found for 6 ' Roy .
Soc. Proc , 1911 , vol. 84 , p. 512 .
VOL LXXXVL\mdash ; A Dr. T. H. Havelock .
Optical Dispersion : [ Aug. 8 , the position of the maximum of , the coefficient of extinction ; further ; a quartic equation is obtained , three of whose roots give the maximum of , and the minimum and maximum of the refractive index .
The next step is to consider numerical examples of these formulae .
We choose first a substa1lce for which the dispersion and absorption have been measured , namely , iodeosin ; for this we calculate the positions of the various maxima and compare with experimental results .
The remaining examples are carbon disulphide , rock salt , and sodium vapour .
For these substances dispersion formulae are known in a form not involving absorption ; the formulae have been completed by introducing coefficients of suitable value , and the maxima of absorption and of selectiye reflection have then been determined .
The results illustrate the differences that may exist laetweeu these wave-lengths and the wave-lengths of the simple dispersion formula ; in the infra-red especially , where reflective power is large .
the lnaximum of selective reflection is proved in eneral to be displaced considerably from the natural wave-length or from the position of maximum absorption .
Derivation of Formuloe for Maxima .
Although there are various physical theories of the mechanism of absorption , such as those of Lorentz and Planck , the final formulae are of the same form as those obtained by a simple frictional term into the equation of motion of the vibrating particle .
In terms of frequency , where the time factor has been made , the complete dispersion formula is , ( 1 ) where .
The quantity is a constant : two special types are obtained by making it zero , or is the frequency of the natural undamped vibrations .
If we consider the vibrating particle to be of lnass and to carry a charge , we have equal to , where is the number of such particles in unit.volume .
In what follows we are concerned specially with values of in the neighbourhood of a frequency .
We shall assume that the region of absorption in question is removed from other similar regions , so that in the above summation all ) terms except one may be replaced by a quantity independent of ; hence .
in the vicinity of a frequency , we write 1911 .
] of of Absorption .
Solving for , and separating real and imaginary parts , we find ; ; ; ; .
( 3 ) We make the following substitutions:\mdash ; ; ; .
( 4 ) previous expressions reduce to ; .
( 5 ) Equations of this form , with unity and with different variables , have investigated by various writers ; in studying the extinction curve of as a function of wave-length approximate methods have been used .
For example , Planck* makes three types as the maximum of is a large , medium , or small number .
The last case is the usual approximation which treats as a number , the result a curve for symmetrical on both sides of a maximum which occurs near a frequency given by .
In the first case , when is large , Planck illustrates the unsymmetrical nature of the curve , and after various assumptions , states that its maximum is given by Konigsberger and also point out that the maximum of does not occur at the frequency of the present notation ; as a result of an experimental study they conclude that in strongly absorbing substances , in the visible spectrum , the maximum of is about 10 from towards the shorter wave-lengths .
They state that no formula has been derived for the position of the maximum , on account of the complexity of the equations involved .
Without limiting the values of the coefficients we can proceed in the following manner .
Solving the equations ( 5 ) we have ; ; .
( 6 ) * M. Planck , .
Akad .
Berlin , ' 1903 , vol. 1 , p. 480 .
Konigsberger and Kilching , ' Ann. der Physik , ' 1909 , vol. 28 , p. 889 ; also 1910 , Dr. T. H. Havelock .
Optical Dispersion : [ Aug. 8 , We see that the maximum of occurs when has its maximum ; that is , when We shall find it convenient to write ; , ; The equation for the maximum of becomes .
( 7 ) The only root which comes into consideration is , ( 8 ) We shall find that is small compared with ; if in addition is small we have often a sufficient approximation in .
( 9 ) For the maximum of we differentiate the expression .
Expressing , and in terms of and rationalising the equation , we find eventually a quartic equation in .
( 10 ) With values of and such as occur in practice , the coefficients in brackets in this equation are all positive .
The equation has extraneous roots which have entered in rationalising the equation for a maximum of it ] is easily seen that it is the smaller positive root which corresponds to a maximum of .
With the same notation as before , the equation becomes We can see what the other roots of this equation signify by finding the positions of maximum and minimum refractive index ; for this we proceed in the same manner by differentiating After some reduction , one obtains again the same equation ( 11 ) .
The equation has , in practice , four real roots , two positive and two .
One of the negative roots is clearly greater than 1 numericalIy ; remembering 1911 .
] Comparison of the of Absorption .
that is , we see that this root does not enter into consideration .
For the other three roots we find the following correspondence:\mdash ; Larger positive root .
Minimum of Smaller .
positive root Maximum of Negative root , numerically Maximum of For a first approximation for the smaller roots one can write , as the more important terms of the equation , Hence , approximately , .
( 12 ) Taking the radicle with the positive sign , we have ; while with the negative sign we obtain For the larger positive root we have approximately .
( 13 ) Of these approximations , that for is generally very good ; the others give at least the first significant , and one can then obtain the roots from the complete quartic to any desired accuracy .
For the reflective power we have is a maximum when 2 is a minimum , that is when or .
( 14 ) The procedure of the two previous cases leads to an equation which is much too complicated to be of service ; in the examples which follow , the maximum of is found ) graphing its values when calculated from the formula .
If were zero , would occur at .
From the above equations , if the absorption is small or is large , so that could occur between and ; otherwise the maximum of comes between the maxima of and For comparison with other formulae and with experimental data , it is convenient to change from frequency to wave-length .
From ( 5 ) we obtain ; ; Dr. T. H. Havelock .
Optical Dispersion : [ Aug. 8 , where , and and are the quantities already used in the equations for the maxima .
We may bring together the various wave-lengths which are to be compared .
Suppose the dispersion formula is in the form ( 2 ) , or .
( 16 ) Then is the wave-length of the natural undamped vibrations of a single particle if isolated .
is an associated wave-length for the medium as a whole , and is the limiting position of maximum absorption when is small and the region of absorption is narrow .
is greater than , and is equal to We have , further , the actual maximum of , which is displaced towards the shorter wave-lengths from and is given by in the present notation ; similarly we have the maximum of and of the reflective power .
It is of interest incidentally to express some of the constants in the manner of Drude , considering the natural vibrations as due to a particle of luass , charge being the number of such particles per unit volume , and the number per molecule of the substance .
In ( 16 ) above we have .
The equation ( 16 ) , neglecting absorption , transforms to where and are given by ( 3 ) .
Hence we have Expressing in electromagnetic units , and with for the molecular weight and for the density of the substance , also with 9660 as the value of for hydrogen , we obtain Simpler fornoe which been used are found by putting zero , or by writing equal to unity .
Iodeosin .
lConigsberger and Kilching ( loc. cit. ) have made recently a quantitative study of various strongly absorbing media .
It is sufficient to take one example , iodeosin , for which they give the values of and throughout a 1911 .
] Comparison of the ' of Absorption .
region of absorption in the visible spectrum ; they estimate that this is associated with one vibrating electron per molecule .
We take the values of , and which they calculated from their results , namely , cm .
; Fl.om the observed values is seen that the maximum of is and occurs at 510 .
The minimum and maximum of are at 480 and 550 respectively .
The maximum of is at about ; and by calculating and graphing the reflective power , we find its maximum is 18 per cent. at rather less than 540 .
From equation ( 8 ) we have , and this gives as the maximum of .
The equation ( 11 ) becomes The approximations of ( 12 ) and ( 13 ) give for the three roots we need , .
From the quartic equation one has more accurately the values .
For the minimum and maximum of these give and 548 respectively ; and for the maximum of we find 510 These calculations agree well with the observed values , except for the minimum of ; of course from the method of calculation in this case any considerable lack of agreement is due to and not being represented with sufficient accuracy by simple dispersion formulae with the above values of the constants .
Using a Lorentz type of formula , Konigsberger and Kilching calculate for the natural undamped vibrations as .
We have then the following sel'ies of wave-lengths:\mdash ; and We notice that in this case , corresponding with the fact that in this , as in equation ( 14 ) .
Carbon ulphide .
We consider now a substance for which quantitative measurements of the selective absorption are not available ; we confine our attention to the principal region in the ultra-violet .
In a previous paper various dispersion formulae for were examined ; using values of the refractive index given by Flatow , namely:\mdash ; one obtained a formula , , ( 18 ) with ; ; ; Dr. T. H. velock .
Optical [ Aug. 8 , This is not given as a complete formula ; it ignores , for example , the minor region of absorption near 325 , but it suffices for the present purpose .
We have to introduce a suitable value for , denoting absorption , and then work out the various maxima .
In order to do this , we first transform ( 18 ) to the form , ( 19 ) so that we can compare it with the complete formulae in ( 15 ) .
Making the calculations we find ; ; ; cm .
From ( 17 ) , with equal , we have This indicates one electron per molecule , in contrast with Drude 's calculation , which gave two electrons per molecule .
In Drude 's form is zero ; if we write and use Flatow 's values of and for at , we have We have now to introduce a suitable value of , so that ( 19 ) is replaced by the two equations .
A simple way would be available if we knew the reflective power at ; { 'or at this place we have and Another method would be to use a knowledge of the position of any of the maxima , for the quartic equation ( 11 ) in is a quadratic for .
Tn the present case we adopt another method .
etaining the above values of and unaltered , for simplicity , means that we regard as inappreciable at wave-lengths at which these constants were determined .
Now the index is given by Flatow as at a wave-length 267 .
The formula ( 19 ) gives a calculated value much too high , and by trial we estimate so as to reduce this value nearer to the observed one at 267 , without having much effect at It is sufficient to take , that is ; we obtain then a calculated index of at 267 .
This value of is probably of the right order of magnitude , and we proceed with the constants of ( 15 ) so determined .
the maximum of we find , and .
We have also ; in this case , as in the previous one , the maximum of coincides practically with the wave-length .
The quartic equation becomes 1911 .
] of the Maxima'of Absorption .
The roots required are , and .
Hence is ; and the minimum and maximum of are at 198 ari In fig. 1 the values of have been raphed from the formulae : one notices the unsymmetrical nature of the curves , descending more steeply towards the longer wave-lengths , and the maximum of being sharper the minimum{ .
From the calculated values one obtains the curve of power ; its maximum is about 40 per cent. at 215 , a smaller wave-length than for the maximum of .
The only experimental result available in respect is an observation by Flatow .
Using a method of repeated reflection from carbon disulphide pressed between plates of quartz or fluorspar , he concluded that the maximum of selective.reflection was the vicinity of 230 .
One might apply certain corrections allowing for the presence of the quartz plate , for stance , by using the relative index of refraction ; but any such correction would tend to inorease htly the above estimate .
Possibly the absorption coefficients are too large in the formulae we have used ; but in any case further observations by means of residual rays in the ultra-violet Are desirable .
Dr. T. H. Havelock .
Dispersion : [ Aug , 8 , There is more satisfactory agreement in the case of the absorption maximum , The observations refer to a capillary layer of a solution of one part olume of carbon disulphide in ten parts of alcohol ; it was found that almost complete absorption occurred from about 190 to , with a meau position of 210 We must modify the constants so as to make the formuIae suitable for such a solution .
This problem deserves fuller investigation , but for the present we use the method of estimating the effect .
We assume that is additive , so that , for volumes and of two substances making a total volume of solution , we have Instead of equation ( 18 ) we have a similar one with different constants , , the transformation to the form ( 19 ) giving the scheme ; ; ; ; ; ; ; ; .
( 20 ) In the present case we may ignore any absorption due to the solvent , and also assume / as constant in the range in question .
For the dissolved substance we take the values of , and the same as before , in ( 18 ) ; taking the volumes in the ratio of 1 to 10 , we find the values ; ; If we give the same value as before , we have and From equation ( 12 ) we find and consequently the maximum of is at the .
This agrees well with the observedmean position of 210 for the solution .
The curve for for the has been graphed and is shown in fig. 1 .
Rock Salt .
For this substance we consider first 's dispersion formula , * 'Roy .
Soc. Proc , 1908 , vol. 81 , p. 367 .
1911 .
] Comparison of the of Absoption .
For this form , with general instead of equal to 2 , we can easily modify the previous expressions for reduction to the required form .
We find in the vicinity of , with the same notation as in ( 19 ) , ; ; For the value of from ( 17 ) we have , and ; we find .
This might be interpreted as two electrons per ulolecule .
Drude 's calculation gave a value of To estimate the maximum of we take , so that ' We obtain , and the maximum of occurs at .
The reflective power has not been calculated in case ; apart from this we have the sequence , and We consider now more fully the region in the infra-red .
Making similar calculations , we find ; ; cm .
; cm .
If we calculate in this case we find a value 536 , indicating a vibrating particle of mass comparable with that of the molecule , if we hold to the simple interpretation .
We have no direct guide to an estimate of .
We keep to the previous method of ; we shall find that this gives values of the reflective power similar to those obtained by experiment .
Reverting to the primitive form of the dispersion formula in ( 1 ) , it appears that making constant is equivalent to making the quantity vary directly as the natural frequency ; appears in the equation of the vibrating particle in the term , a term which is frequently written in the form Having assigned the values of the constants , we calculate the various maxima .
We find and quartic equation becomes The required roots are , and .
These give and the minimum and maximum of at and respectively .
Values of and have been calculated and the curves are shown fig. 2 ; on account of the larger magIlitudes the unsymmetrical features are very marked .
The values of the reflective power have also been graphed ; we find a maximum of about 76 per cent. at 52 .
Recent experiments with residual rays give a double maximum with a mean position of Other observations have given the following estimated values of in at at at at Dr. T. H. Havelock .
Optical Dispersion : [ Aug. 8 , and at 108 ; these have been marked by circles in fig. 2 , and one sees that the calculated curves agree well in a general way with these results .
Probably the value , of is too large , and one obtain a less rounded curve by taking a smaller value , but one does not lay much stress on actual agreement , because the curves are meant to be illustrative in the first instance .
With the above formula , Maclaurin identified with the position of maximum reflection obtained by residual rays , taking that to be We have seen that , compared with the simpler form of dispersion formula , the w is replaced by , equal to , which would be the position of maximum absorption if that were small .
Putting in a more suitable value for the absorption , we have found that the maximum is at , only slightly displaced to the shorter wave-length from ; however , under the same conditions we find that the maximum of selective reflection- occurs at 52 Similar calculations could be made for any of the dispersion formulae which have been given for rock salt .
These are generally in the form , that is , is zero and coincides with .
Martens in such a mula put the wave-length for th6 infra-red equal to the observed position of maximum reflection , but we have seen that 1911 .
] Comparison of the Absorption .
absorption coefficients are introduced the maximum of occurs at a considerably less wave-length than the of the formula .
Paschen*determined directly from the valuqs of the refractive index , and claims that his formula with represents the dispersion better , in spite of apparent disagreement with reflection maximum .
If we put Paschen 's formula into the form used above , for the vicinity of 60 , we find , and .
These values are not much different from those of Maclaurin 's formula when expressed similarly .
It is unnecessary to repeat the calculations ; if we introduced a quantity as before , we should find the maximum of reflective power to be in the neighbourhood of 50 It seems probable then that Paschen 's formula , when interpreted aright , does not disagree with the results of reflection experiments .
Absorption Vapours .
The rption in the previous cases is enormously larger than that of gases and vapours , so that the differences between the various maxima in the latter are very minute and enerally quite beyond the limits of observation .
For dense mercury vapour W. has been able to observe selective reflection , with the interesting result that the maximum reflective power occurs at about one Angstrom unit towards the shorter wave-lengths from the absorption maximum .
The particular line in question is near 2536 A.U. , and it shows powerful anomalous dispersion .
The vapour was contained in a fused quartz tube , and Wood suggests that the displacement is entirely a spurious effect due to the relative index of refraction of quartz and the vapour being nearly unity on the side of the longer wave-lengths but quite large towards the shorter wave-lengths from the absorption line .
There are no data available for the dispersion of mercury vapour by which one might test this numerically .
As a similar example we use the values given by Wood for dense sodium vapour ; for a single mean absorption at 5893 A.U. , the dispersion formula is with cm .
This formula is only approximate ; to obtain any approach to observed values near the D.lines , Wood alters the value of .
However , one can use it as an example of the numerical magnitudes of the quantibies .
It has been estimated that in less dens sodium vapour the logarithmic decrement of the F. Paschen , ' Ann. der Physik , ' 26 , p. 130 , 1908 . .
W. Wood , PhiL Mag 18 , p. 187 , 1909 .
A of thoe Absorption .
natural vibrations is of the order ; in the present case we shall take , a value which ives moderate reflective powers .
From the formula ( 12 ) we find that the maximum of is at less ) 5893 Iby about A.UQ .
Working out values of and we obtain a maximum reflective power of per cent. at we divide the values .of by a mean value for the index of quartz , say , we can graph the reflective power taking account of the effect of the wall of the tube ; this shifts the maximum rther to the shorter wave-lengths by about We obtain thus a total of about , of which one-half is due to the wall of the tube .
This displacement is several times smaller than that observed in mercury vapour , but there are certain possibilities to noticed .
In sodium vapour with the above value of it has been estimated there is not more than one vibrating electron for every 12 molecules ; in mercury vapour the ultra-violet line showed very strong anomalous dispersion , and is probably a more dominant natural vibration a larger value of .
In addition , the absorption extended over several Angstrom Jllits .
Hence the conclusion is that while the displacement observed by may be exaggerated by the presence of the quartz wall , it is probably in part a true effect\mdash ; a displacement of the reflection maximum from the absorption maximum towards the shorter wave-lengths .

