]\gt ; On the ersion of Gaseous Mercury , Sulphu Phosphorus , Helium .
By C. , Fellow of University College , London , and E. PARR METCALFE , B.Sc. , 1851 Exhibition Scholar .
( Communicated by Professor F. T. Trouton , F.R.S. Received February 11 , \mdash ; Read February 20 , 1908 .
) In continuation of previous work* on the refractive indices of certain elements in the yaseous state , we have measured the dispersion of the elements named above within the limits of the visible spectrum .
Jamin 's refractometer was used , and the arrangement of the instrument was that described in our previous paper .
But , as different wave-lengths had to be employed for the determination of the dispersion , the method of illumination was improved .
Light from a Nernst filament was focussed on the slit of one of Messrs. Hilger and Co. 's fixed-deviation spectroscopes , and , in the focal plane of the .
spectrum , a slit was placed capable of motion in plane .
The adjustment was calibrated by ' the wave-length of the passing the slit with the reading of the drum of the spectroscope .
By rotating the drum , light of any required wave-length could be obtained , and , by the slits , the spectrum was of sufficient purity to admit of two hundred bands being counted to one-fifth of their breadth .
With this arrangement two of procedure were employed .
At first a separate observation was made of the l.efractive index at each wavelength selected , while the substance was heated until it had completely evaporated .
But this plan is open to the serious objection that any error due to unequal disposition of the tubes , unequal , buckling of the supports , or drift due to the heating of the mirrors , is not ; and , since the dispersion is in all cases a small fraction of the refraction , such errors are of importance .
In order to avoid these causes , a second method was orked out by one of us .
M. ) which , as .
as we know , has not been previously used .
interferonleter t vacuous , the lnirrors and compcnsator are so adjusted that each of the beams passes alne thickness of glass , of air , and of silica .
When this is the case the interfel.ence pattern , viewed ) white , presents ) of a train ' , pp. , 1907 .
VOL. LXXX.\mdash ; A. Messrs. C. Cuthbertson and E. Parr Metcalfe .
[ Feb. 11 , of fringes , black and white at the centre , but rapidly becoming coloured as their distance from the centre increases .
The only truly achromatic point in the interference pattern is the centre of the bright fringe between two darkest fringes : at this point the relative retardation is zero for all wave-lengths .
It is not difficult to pick out by eye the two darkest fringes .
The adjustment is effected by rotating one mirror , or the compensator , until the cross-wire or pointer of the observing telescope lies exactly midway between these two .
The white light is then replaced by the monochromatic illumination .
If now the wave-length is changed continuously by slowly rotating the spectroscope drum , the central bright fringe remains in the same position relatively to the cross-wire ; the only effect of varying the wave-length is to change the spacing of the fringes .
This preliminary adjustment havino been made , the sion of the vapour is measured thus .
The spectroscope drum is set to some convenient wave-length , , that of the reen mercury line , which is specially suitable , both on account of the luminosity of the spectrum in that region , and because a mercury vapour lamp a handy standard for testing the accuracy of the correspondence of the drum with the wave-length of the light passing through the slit .
Then the charge is slowly evaporated in one of the interferometer tubes .
The fringes which cross the pointer are counted , and , when they have come to rest , the spectroscope drum is rotated slowly .
This will , in general , cause the fringes to move across the pointer .
As each fringe reaches the pointer the corresponding wavelength is read off the drum .
In this way we get , in one operation , the fringe readings for the particular quantity of vapour used , for thirty or forty points along the visible spectrum .
It must be remembered that the rate of variation of fringe reading with wave-length is not simply a function of the dispersion of the gas alone .
If number of fringes observed , wave-length , refractive index of the gas , then \mdash ; where is dependent only on the ' dimensions of the apparatus , and .
Of the terms on the right-band side , the first represents the effect of introducing the gas into the refractometer tube , the wave-length remaining the same ; the second expl.esses the movement of the fringes when the wave-length is varied , while the gas pressure is constant .
1908 .
] Dispersion of Gaseous Mercury , Sulphur , etc. The terms being of the same order of magnitude , they must be measured with the same degree of accuracy .
It is , of course , highly desirable that the achromatic ) oint of zero retardation , when both interferometer tubes are vacuous , should not suffer displacement during the operation of the fringes , through uneven temperature conditions or mechanical shocks .
As soon as possible , therefore , after taking a set of , the vapour is rapidly condensed in its tube , and the coloured fiinges of white are examined .
In practice it is convenient to several slow readings of the number of fringes for one wave-length , so as to obtain it correct to the nearest integer .
Any outstanding fraction is best estimated by the charge as quickly as possible ; for in this the zero has but little time to " " drift\ldquo ; before all the have passed .
When refractometer tubes of fused silica*are used , difference of thickness of the end plates is almost unavoidable , owing to the necessity of refiguring and them after they have been fused into the bore of the tubes .
This difference introduces a complication .
Suppose th an attempt is made to compensate for a slab of silica of thickness and refractive index placed in the path of one beam by putting in the other a slab of lass of thickness and index .
Then the condition for achromatism becomes where refractive index of air .
The dispel.sion of air is so small , compared with that of glass or of silica , that the last term may be ected .
So ] , for the achromatisation to extend over any range of waye-length , it is necessary that the ratio of the dispersions of the two slabs shall be constant throughout that range .
This ition is not satisfied in .
the case of silica and lass , so that perfect chromatic compensation is not possible .
The difficulty could , of course , be met completely by a compensator of the type used by Jamin , the plates made of fused silica .
But a simpler way was found to be quite satisfactory .
The silica slab was compensated for as well as possible with lass ; and , the tubes remaining yacuous , the spectrum was traversed .
The position of the zelO fringe was observed to shift by about one-tenth of a width .
This displacement was plotted against wave-length , and thus it was easy to correct for the effect in the dispersion curves of the gases and vapours dealt with .
* Cuthbertson and Metcalfe , .
cit. Messrs. C. Cuthbertson and E. Parr Metcalfe .
[ Feb. 11 , A single charge of mercury was sufficient to yield concordant results .
Numerous series of were taken by the differential method .
The curves plotted from them were found to agree well , and from 43 of the best observations we select the following values of the refraction at three points the spectrum\mdash ; The value obtained for sodium rees well with that obtained by C. Cuthbertson three years ago , and is identical with that found by E. P. Metcalfe last year .
Adopting these numbers , the ction curve is expressed in the form given by Cauchy , as .
The rate of change of index is about four times that of air , for which the formula is , according to Scheel , .
Sulyhur .
Two charges of sulphur were used .
In the spectrum of this element there is an absorption band extending from the violet almost to the red , and , with the first charge , readings could not be obtained beyond the yellow .
At , light failed after 108 bands had passed , and at only 75 bands could be read .
In order to obtain values beyond this point , the weight of sulphur had to be diminished to gramme , giving about 31 bands at , the lowest number from which fair accuracy could be obtained .
From 35 observations we deduce the following as the best values for the refractiviCy of sulphur at three points in the spectrum\mdash ; cm .
1908 .
] Dispersion of Gaseous Mercury , , etc. Using these values , the curve of refraction can be expressed by\mdash ; .
The two independent valnes obtained for the refraction at ( 1105 and 1111 ) well with that found by C. Cuthbertson in with gJass tubes and a tenlperature .
If we attempt to determine the position of the centre of ) band from the formula , we find from the first and second observations , and from the first and tbird valnes which are not very concordant .
We have previously observed tha until values of the refractivities for could be obtained , it was useless to examine closely the curiously simple ratios between the refractivities of allied elements which were obtain at the lines .
It was shown thnt , at that point of the spectrum , the atio of the refractiyities of to to , Cl to , and Are to No , was very nearly four .
It is interesting , therefore , to compare the value now for sulphur with the ctivity of oxygen for infinite wave-lengths .
The dispersion of oxygen does not appear to have been fully investigated , but Natanson has calculated the of the refractivity for from Mascart 's figures to be , and Koch has found for the number Adopting the mean of these determinations , the atio of the refractivities of sulphur to oxygen for infinite wave-lengths is , therefore , , which only differs from the number 4 by per cent. If it be remembered that the presence of the absorption band in the violet , and possibly of one of the red , must disturb the values from which the refraction curve is calculated , the coincidence is good .
It is closer than that between the indices of the two elements at the lines .
It is interesting also to compare the dispersions of the two elements .
The refraction curve of oxygen , calculated from Mascart 's , is .
philTrans , A , netic theol , and extinction ' Bull .
de l'Acad .
des Sciences de Cracovie , ' Ap. , 1907 , p. ' An. .
Physik , ' vol. 17 , p. 665 , 1905 .
Messrs. C. Cuthbertson and E. Parr Metcalfe .
[ Feb. 11 , Thus the rate of change of refraction of sulphur is about four times that of oxygen .
One of our observations in the extreme red was abnormally low , which would the presence of an absorption band in the infra red .
So far as we know , the spectrum of sulphur vapour has not been investigated in this region .
Phosph orus .
In this case also a single charge of the element was used .
From thirty-nine observations we deduce the following most trustworthy values : these figures the refraction curve .
The value now found for agrees well with C. Cuthbertson 's earlier value ( 1197 ) , but , like that of sulphur , the new value is about 1 per cent. higher than the old .
As in the case of sulphur and oxygen , it is interesting to compare the refractive index for infinite with that of nitrogen .
Scheel 's values for the dispersion of are expressed by the formula .
Multiplying the refractivity for infinite wave-lengths by four we obtain 1162 , which is identical with that found for phosphorus .
Comparing the dispersions we find that the rate of change of refraction for phosphorus is almost exactly double that of nitrogen , while that of sulphur was about four times .
that of oxygen .
Helium .
The specimen of helium used in the experiment was kindly lent us by Sir W. Ramsay , to whom our thanks are due .
It was obtained from thorianite and purified by Dewar 's process of passing the gas through charcoal cooled with liquid air .
The measurement of the dispersion of this element presents considerable difficulties owing to its smallness .
With a tube nearly two metres , and a ence of pressure of 760 mm. , 1908 .
] Dispersion of , Sulphur , etc. only about a hundred bands can be obtained in the yellow .
The dispersion is less than per cent. of the refraction in the visible spectrum , so that the whole measurable effect between the red and violet is aboub one-third of a band , and , as it is almost impossible to read to less than one-tenth of a band , accurate results were not obtained .
The principal difficulties were found in bhe change of zero of the refractometer with temperature , instability of the building , parallax in making in either the red or violet , and the determination of the wave-length employed .
In order to avoid this last source of error a further modification of the diHerential method was adopted .
Enough helium was admitted to one tube to cause about a humdred bands to cross the field .
Into the other tube air was then introduced until an exactly equal number of bands had passed in the opposite direction .
The of the employed were then from red to violet .
In these circumstances any change in the position of the band which is on the pointer is due to the difference between the dispersions of air and helium .
For the dispersion of air the formula of Ka.yser and was assumed\mdash ; The helium was repeatedly repurified , as it was found that it became contaminated by small bubbles of air creeping in during the manipulations .
Out of a large number of experiments we select six as the most trustworthy for the refraction at the line .
These are as follows\mdash ; The mean of these is , and this we adopt as the refractivity at For the dispersion the last and best series of experiments for the difference of refraction between and the following values\mdash ; Messrs. C. Cuthbertson and E. Parr Metcalfe .
[ Feb. 11 , The mean of these is , and , adopting this value , we obtain for the curve of refraction .
The difference between the values given above seems large ; but it must be remembered that they represent direct determinations of the dispersion , which is , between the limits chosen , 1/ 350 part of the refractivity .
They correspond , therefore , to an accuracy in the determination of the refractivity of any particular ray 350 times as great .
We think the refractivity can be trusted to per cent. , but the dispersion to not less than 10 per cent. Thus , the dispersion of helium bears to the refraction a ratio of one-half to one-third that which the dispersion of air bears to its refraction .
The figures given are those for a monatomic gas .
For comparison with the other elements they should be doubled .
In 1901 Sir William Bamsay and Dr. Travers found for helium , with , the value .
Our value is per cent. less .
If this difference were due to greater purity of the gas there should be a corresponding decrease in the density .
But an experiment on the specimen we used gave a density of tYainst 1 adopted by Ramsay and so that the discrepancy cannot be explained in this way , and may be attributed to our good fortune in having a larger quantity of the gas than its discoverers could command .
In a previous paper , by C. Cuthbertson , *attention was drawn to the simplicity of the ratios between the refractiyities of the five inert gases .
The following table shows that , with the new value for helium now obtained , the coincidence is even better:\mdash ; There is still room for improvement which may be found when the indices of the other gases are remeasured after being purified by absorption over cold 'Phil .
Trans , vol. 204 , p. 323 , 1905 .
1908 .
] Dispersion of Mercury , Sulphur , etc. Th1 results now obtained may be summarised as follows:\mdash ; Mercury .
Sulphur Phosphorns .
Helium .
The dispersion of mercury is about four times that of air .
The index of sulphur for infinite waves is , within 2 per cent. , four times - , hat of oxygen .
Its dispersion is , not so exactly , four times that of The index of phospl ) orus , for infinite waves , is exactly four times that of .
Its dispersion is almost exactly twice that of nitrogen .
The index of helium is , within per cent. , of the ) xistin value for the in-Tex uf rgon .
Its dispersion is about three-sevellths , hat of We have $(yreat pleasure in our cordial thanks to routon and the staff of the Physical Laboratory at University London , for assistance and adyice , and to the Royal Society for a grant in aid the research .

