]\gt ; Dr. L. N. G. Filon .
Measurements of the [ Feb. 24 , On both points of view , it must be supposed that the luminosity due to -rays accompanies the dissociation of the active centres and cannot be ascribed to the recombination of the dissociated parts , for there is an undoubted chemical alteration of the material due to continued bombardment .
Whatever point of view is taken , it is clear that for willemite and zinc sulphide the active centres have dimensions comparable with a molecule .
This excludes the suggestion put forward at one time that the scintillations of zinc sulphide due to -particles were a consequence of the mechanical cleavage of small crystals .
I am indebted to Mr. Marsden for the use of the data obtained by him , and for his assistance in calculation of the decay curves .
Measurements of the Absolute Indices of Glass .
By L. N. G. FILON , M.A. , D.SG .
, Fellow and Lecturer of University College , London .
( Communicated by Prof. F. T. Trouton , F.R.S. Received February 24 , \mdash ; Read March 3 , 1910 .
) 1 .
In June , 1907 , the author described* a method by which the doublerefraction in strained glass could be measured by observing the deviation of a ray of light passing through a slab of glass under flexure .
If a slab or beam of glass of rectangular cross-section be bent in a vertical plane under a bending moment , and if a plane wave be transmitted through the glass in a direction perpendicular to the plane of flexure , the light is broken up into two components , one polarised horizontally ( i.e. perpendicular to the cross-section and along the line of stress ) and the other It was shown in the paper referred to that the variation in the index of refraction of the lass , due to the stress , produces in the vertically polarised ray an upwards deflection radians , ( l ) and in the horizontally polarised ray a deflection radians , ( 2 ) ' Roy .
Soc. Proc , vol. 79 , pp. 440\mdash ; 442 .
191.0 .
] Absolute of in where is the thickness of the glass , I the second moment of area of the cross-section about the " " neutral axis and and are the stress-optical coefficients for the vertically and horizontally polarised rays respectively : the stress-optical coefficient for any ray being the increase in the index of refraction for unit tension , or , what is equivalent , the additional retardation introduced per unit thickness per unit tension .
In the above , is reckoned positive when the slab is bent concave downwards .
2 .
This effect is complicated in practice by the fact that , when a positive bending moment is applied , the upper half of the slab is in tension , and therefore contracts laterally , whereas the lower half is in compression and expands lateralIy .
This transforms the rectangular slab into a prism of very small angle ; the additional upwards deflection thus produced is readily calculated and found to be , ( 3 ) the natural refractive index of the glass for the light used , 's modulus , Poisson 's ratio .
Putting in for I its value ( 1/ 12 ) , where height of the slab , the deflections actually obsel , are .
( 4 ) 3 .
In order to apply the correction change of shape , we immerse the glass in a liquid\mdash ; cedar oil was actually used\mdash ; of which the refractive index is known .
This is done by placing a saddle piece ABCD ( fig. 1 ) fornled of parallel pieces of Dolass AB , CD cemented to a bridge , across the slab , the liquid being held by capillary action in the position shown the shading .
When the cross-section of the slab is distorted as shown in fig. 1 , the cedar oil FIG. 1 .
wedges formed deflect the rays , the deflection being subtracted from that produced by the change of shape of the glass .
lf be the refractive index of cedar oil for any given line of the spectrum , then the deflections observed for light of this are Dr. L. N. G. Filon .
Measurements of the [ Feb. 24 , From ( 4 ) and ( 5 ) , If we denote by the difference , which is got directly from obseryation , we have , ( 7 ) , formulae giving the scress-optical coefficients .
In the actual experiments was measured for the -lines , for which for cedar oil .
4 .
The arrangement of the experiment is shown diagrammatically in fig. 2 .
The light was reflected down a vertical collimator and then diffracted FIG. , source of light ; , mirror ; , lens ; , occulting shutter ; , collimator with slit ; , grating ; , slabs of glass ; , nicol ; camera ; , plate-holder .
horizontally by a reflection grating so as to pass through the glass slabs\mdash ; two of which were used to obtain a greater effect .
The spectrum was then raphed with a camera of about 50 cm .
focus .
A nicol was interposed in the path of the rays of light between the slabs and the camera .
By turning the nicol the horizontally or cally polarised component could be separately observed ; or , if desired , the nicol could be removed and both components observed simultaneously .
An arrangement of weights and levers ( not shown in fig. 2 ) kept the slabs under a constant upward bendin moment , as large as was consistent with the safety of the slabs .
On moving a lever over , a downward bending moment was brought into play .
was roughly equal to 2 , so that the addition of practically reversed the sign of the moment on the slabs without altering its magnitude .
In this way a deflection was obtained twice as great as if the bar had been loaded from the unstressed state , and this without the risk of rupture .
5 .
In taking the observations , half of each exposure was taken with the moment applied and half without .
Two images of each line were thus obtained , and the displacement between the two images on the photographic plate was measured with a micrometer .
1910 .
] Absolute Indices of Refraction 575 .
To avoid confusion of the two spectra , one was displaced laterally with regard to the other by moving an occulting sbutter across the slit at the same time that the moment was applied .
The occulting shutter was so adjusted that the } ) ectra ran side by side , yet partly , so to allow the shift of the lines to be measured .
Two plates were taken for each glass , one in horizontally , the other in vertically polarised light .
The exposures on each plate searched the whole range of the spectrum which would affect the plate .
" " Panchromatic\ldquo ; plates , manufactured by Messrs. Wratten and , of Croydon , wele used .
These are fairly uniformly sensitive throughout the visible spectrum .
The scale value connecting shift on the plate with angular deviation was obtained by taking away the grating and it by a double mirror consisting of two mirrors cemented together and a known small angle with one another .
The two mirrors two images of the slit on the plate : the distance between these two responded to a deviation 2 of an The light used was an arc lamp burning yellow " " flare\ldquo ; carbons .
These have their core mainly filled with calcium and give a very bright and steady calcium spectrum , with several of the hter strontium .
lines 6 .
In order to obtain as large an effect as possible , large slabs of cm .
, height 1 cm .
, and thickness 6 cm .
were made for me by Messrs. Zeiss , of Jena .
As two were employed in each experiment the traversed 12 cm .
of glass .
Two glasses have becll examined by this method , , the Jena numbers and 3199 respectively .
The first is a borosilicate , the second a so-called " " Uviol\ldquo ; glass , which transparent to ultra-violet light .
The kers were unable to give the composition of , and as they are at present required for there work , they have not yet been analysed .
On a occasion a glass bearing the Jena Catalo , ue number was examined by the author .
Its analysis gave the following per cent. ; per cent. ; .
cent. per cent. ; per ccnt .
; these , however , can at best indication of the composition of the ) resent slabs , which , probably , may cliffer a good deal from the earlier ones .
7 .
The stress-optical coefficients and have been calculated for a number of spectrum lines , and are given in Tables I and II below .
are negative , so the ) give and .
These are expressed in a unit of cm.2 per gramme weight .
Dr. L. N. G. Filon .
surements of the [ Feb. 24 , It would be very desirable to have a short name for this unit , which is constantly occurring in this work .
It happens to be of the order of the stress-optical coefficients for ordinary glasses , and is therefore a convenient practical unit .
I propose to call it a brewster , after the first discoverer of the stress-optical effect .
The displacements measured varied between and revolutions of the micrometer screw .
The latter was graduated to hundredths .
The mean of five readings was taken and the average probable error should not have exceeded of a revolution .
The errors due to irregularities plate , photographic irradiation , or stretching of the film , will have raised this somewhat , but the probable error of or should not exceed ) of a brewster on the Table I.\mdash ; Glass Table II.\mdash ; Glass 3199 .
[ N.B.\mdash ; In the above table , entries in the column headed are occasionally missing the column headed , and conversely .
This is due to the fact that a which was suitable for measurement on one plate was unsuitable for measurement on another , owing to imperfeot or overlapping images .
] 1910 .
] Absolute lndices of Refraction in Strained .
577 The results given in the above tables are exhibited on the diagrams of figs. 3 and 4 , the base-line being shifted so as to bring the curves of and close to one another for purposes of comparison .
FIG. 3 .
\mdash ; Glass FIG. 4.\mdash ; Glass 3199 .
8 .
With regard to the above results we note the following points .
Since and are negative , both rays are accelerated tension and retarded by pressure .
being unerically greater than , the ray polarised in the direction of stress is the most strongly affected .
The other , however , is affected to a sensible extent .
If we now examine figs. 3 and 4 , we see that in both and show a slight numerical as we move towards the violet .
But in 3199 , shows a numerical decrease towards the violet , whereas shows a numerical increase in the same direction .
These general variations in and appear to be of the same ordel of magnitude in absolute value ; but the variation is much ooreatel in than in to the smaller valuc of itsel In addition , bove gene1 al variation is broken by a number of local irregularities .
most important of these occurs in between 5189 and 5042 , where is alarge and sudden jtunp in of of a blcwster .
This has been tested by careful re-lneasulement .
It confirms a result 578 Absolute Indices of Refraction in obtained some years ago by the author , *when an abrupt oscillation of this sort was discovered in the differential stress-optical coefficient of two other borate glasses .
This gave a dip about , and rise about , and there seems little reason to doubt that we have here the identical irregularity .
In the paper referred to it was conjectured that this might be analogous to the bend in the curve of refraction produced by an absorption band , and possibly due to a free period of some constituent of the glass .
At the time it seemed doubtful whether this constituent was , the bend not having been detected in other borate ative evidence , however , is hardly conclusive and the presence of the bend in rather that it may be characteristic of borosilicates .
The other irregularities are not so marked .
Many appear so erratic .
that at first sight one is tempted to put them down as due to errors of observation and to draw through the observations a smooth curve ( shown by the full curve in figs. 3 and 4 ) .
There is no doubt that they are very much of .
the magnitude of the errors of observaGion , and it is hard to say of any one of them that it is certainly significant .
Yet after consideration it may seem that they should not be so easily dismissed .
For example , the values of for show a little maximum at .
Taken by itself this means nothing : the line formed by joining successive observations is in this neighbourhood quite irregular and jagged\mdash ; but , curiously enough , the curve of shows a little dip at precisely the same place , as though some common physical factor had affected the results .
Further , in the paper in ' Phil. Irans .
' quoted above , showed a decided peak in the neighbourhood of 4500 for certain borate glasses .
All these points , taken ether , make one hesitate to condemn this little sinuosity as entirely irrelevant .
Again , there are slight inequalities in in figs. 3 and 4 , near 5100 , which are insigl in themselves , but acquire significance from their occurrence at the same place as the characteristic bend previously noticed .
Finally , in the curve of fig. 4 it is difficult to resist the conclusion that the groups of three observations round 5890 , 6100 , and 6450 respectively are not disposed as they are by pure chance , but really imply two crests and a trough in the curve , the more so that tloe three other curves show a rise from 5800 to 6100 .
If all these pieces of evidence be taken together and weighed , the conclusion is almost forced upon one that we have here before us a number of bends in the curves , due to various free periods of the constituents of the glasses .
The dotted lines of figs. 3 and 4 show a possible course of the curves 'Phil .
Trans , vol. 207 , pp. 293\mdash ; 301 .
Some gnetic Disturbances at Kew .
but more delicate observations would be necessary before the oscillations , if they exist , can be definitely identified .
9 .
Incidentally the method provides us with the elastic constants the glasses used , for ?
can be calculated when is known : and if take Poisson 's ratio to be 1/ 4 ( which is very nearly 's value for the ratio in glass ) , we find E. We have 10 .
In conclusion , I desire to express my best thanks to Prof. F. T. Trouton , F.R.S. , and to the Staff of the Physical Laboratory of University where the work has been carried out , for their kind assistance and courtesy .
Phenomena of Jlagnetic at By CHREE , Sc. I ) .
, .D .
, F.R.S. , Superintendent of the Department of the Natioual Physical LaboIatory .
( Received January 5 , \mdash ; Read February 10 , 1910 .
) ( From the National Physical Laboratory .
) ( Abstract .
) A recent papel* discussed the diurnal inequality of Kew magnetic declination derived from 209 of the most disturbed days of the ] eveu years 1890 to 1900 .
The present paper discusses the corresponding phenomena for the same days in the other magnetic elements .
The most prominent features of disturbed days are the nature of magnetic changes , and the frequency of large oscillations .
These more obvious features have hitherto engrossed attention .
It is found , however , that the irregular changes on disturbed days are panied by diurnal changes , which are specially striking in yertical force .
In declination and horizontal force the of the regular diurnal inequality on disturbed days is in the average month of the year about double that on quiet , but in vel.tical force the former range is fully four times the latter .
Disturbance affects the type as well as the range of the diurnal inequality .
The former influence is 'Phil .
Trans , vol. 208 , p.

