]\gt ; Mr. C. Spurge .
On the Effect of Polish on the " " On the Effect of Polish on the Reflexion of Light from Surface of Iceland Spar By C. SPURGE , B.A. , Catherine 's College , Cambridge .
Communicated by R. GLAZEBROOK , M.A. , F.R.S. Received November 18 , \mdash ; December 16 , 1886 .
Revised March 3 , 1887 .
I. Introduetion .
The optical effect of polishing the surface of a transparent body received the most complete investigation at the hands of and till very recently Seebeck 's were almost the only made on the subject .
Seebeck 's method consisted in observing a Nicol the light of a lamp reflected from the surface of the By means of a divided circle , the angle of polarisatiou was and it was from an alteration in this angle that a change in the of the surface was inferred .
But it has been since shown by that , when polarissd is incident upon the surface of transparent body , the reflected light is in general not plane but to measureable degree elliptically polarised , and consequently tbere is angle of incidence at which the light can be completely quencbed by Nicol .
It follows that , as regards our present state of Seebeck 's investigation is to some extent incomplete , and also there is some uncertainty in the determination of the angles polarisation , which may affect our conclusions as regards the state the surface , especially sincs the difference produced by according to Seebeck not very large .
Both Sir David BrewsterS M. Jamin were of the opinion that Seebeck 's experiments should repeated , and the latter promised to consider the effect of later on but appears never to have done so .
Mr. Glazebrook pointed out to me that the subject presented a suitable field research , and , at his instance , I undertook the present investigation .
My object has been to attain greater accuracy than hitherto employing for an analyser a quarter undulation } ) ] in addition to Nicol , so as to make the extinction of the reflected light complete .
The angle of incidence of the polarised light falling on surface of the crystal was kept constant , in order to measure direct as possible the alteration produced by change of polish .
the azimuth of the major axis and the ratio of the axes of polarised Iight were calculated .
These quantities * ' Poggendovff , Annalen , ' vol. 20 , 1830 , : vol. 21 , 1831 , p. 290 .
Sir J. Conroy , ' Boy .
Soc. Proc Feb. , ] 'Annales de Chimie , ' vol. 29 , 1850 , p. 263 .
S 'Edinb .
Journ. Sci vol. 5 , 1831 .
Reflexion of Light from lcelund Spar .
with two independent tests of a change of surface , and completely the nature of the light , so that a knowledge of their values and after polishing .
enables us to state the precise alberation in the reflected light , a question which has never been and is , I believe new to this paper .
II .
Apparatus .
A series of preliminary experiments were made to discover what apparatus was best suited for the investigation .
I found that , whether a Nicol or a Nicol and a quarter undulation plate were used as analyser , it was best to polarise the light before incidence .
Also , observations showed me that a Nicol and a quarter wave plate were a more sensitive arrangement than a simple Nicol , that in each case the light was polarised before incidence .
I have therefore deemed it necessary to employ the Nicol and quartsr wave plate arrangement in ordel to secure all the accuracy that is possible by completely quenching the reflected light .
The instrument I used was an elliptic analyser kindly lent for the purpose of these experiments by Professor Stokes .
A very full description of the instrument will be found in the ' Phil. Mag. ' for 1851 , but the following abbreviated account is given in order to explain the way in which it was used during the course of the experiments .
The elliptic analyser consists of a brass annulus attached to a vertical stem which fits into a hollow cylindrical foot .
When the foot is placed on a table , the plane of the annulus is vertical .
Within the annulus turns a brass graduated disk ; and the angle through which it turns is read off by means of verniers engraved on the annulus .
These verniers are therefore fixed .
The disk is pierced by a central aperture on the side of which opposite the incident light is a screw thread , so that a cell containing a quarter wave plate can be screwed into the disk .
In front the disk carries a hollow cylinder tnrned in the lathe with the disk itself .
Round the cylinder turns collar into which is screwed a tube containin the analysing Nicol .
The collar carries a pair of level edged verniers by which the angle may be read ofl through which the Nicol has been turned .
These verniers are therefore moveable .
Thus the quarter wave plate moves in azimuth , carrying the Nicol along with it , and the Nicol has likewise an independent motion in azimuth .
In observing , the light is extinguished by a combination of the two movements , in which oase the elliptically polarised light is comverted by the quarter wave plate into plane polarised , which is then quenched by the Nicol .
There are two principal positions in which the light can be qnenched , and , since either Nicol or quarter plate may be reversed by turning through , there are four subordinate positions corresponding to .
C. Spurge .
On the Effect of Polish on the each principa ] position .
The position of the Nicol is determined the readings of the two moveable and that of the quarter wave by the readings of the two fixed verniers .
Thus each position is determined by eight readings , and in the tables follow , each number is the mean of eight readings .
Suppose that are the mean readings of the fixed , mean readings of the moveable verniers , then the quantities , which the object of the presenb investigation to determine , are ratio of the axes of the ellipse , I , the azimuth of the major axis of ellipse , and these are given by the formulae\mdash ; , and I These equations determine absolutely , but I will be from an arbitrary zero which will remain fixed so long as the plate is not uuscrewed from its containing tube , but which changed by a constant amount , if for any reason it is unscrewed rescrewed up .
A subsidiary quantity is , the retardation of the crystal which may be determined by means of the equation , .
The source of light employed was an Argand burner , the rays which were polarised by means of a Nicol before incidence .
III .
Adjustments .
The tube of the polariser was levelled and its axis placed in a line with the centre of the flame .
The height of a small brass table , which the crystal was placed was adjusted so that the reflected passed ffirough the tube of the analyser .
A number of observations were made to determine the best angle of incidence , the angle at which the extinction was most rapid .
The best position of the analyser having been found , the centre the tubs of the analyser was adjusted to the same height as the of the tube of the polariser and the centre of the face of the The tube of the analyser was so directed that a ray of light from centre of the flame passing along the axis of the polariser was so as to enter at the centre of the tube of the analyser and leave at centre of the tube .
As the present experiments were directed to discover a which at the outset was recognised as possibly small ] , especial was taken to secure fixity of position in the parts of the and in the position of the face of the of Iceland spar .
Reflexion of from lcetand Spar .
instruments were firmly attached to a laboratory table , and commencing the moveable parts were examined and tightly up .
Round the base of the table and the foot of the analyser small quantity of meIted paraffin was poured so as to form connecting links from one part of the apparatus to another : From time to time these links were examined and found to be unbroken .
To attach the crystal to the table , hard electrical cement was used .
The plan finally adopted was to place the crystal on the table , and to fix it by pouring a small quantity of melted wax down the back of it .
Since the crystal was removed from the table to polish its surface , some means of restoring it to its original position were needed .
The following optical method was employed .
A circnlar diapbragm with a central pinhole was fitted to the brass tube containing the polarising Nicol .
If the pinhole were slightly eccentric , the position of the hole would change as the disk rotated in its plane .
To any such alteration , a radius was drawn on the diaphragm which was set so that it was horizontal and always pointed in the same direction .
In front of the source of light was placed a screen having a smatl hole at the same height as the pinhole in the diaphragm .
The position of the screen was defined by lines drawn on the table .
Thus only a single ray of light was allowed to fall on the surface of the crystal , , that passing through the apertures in the screen and diaphragm .
As these aperturss could always be replaced in the same position , the direction of this incident ray was a fixed horizontal straight line .
In a similar manner a circula diaphragm with central pinhole was fitted to the brass tube containin the analysing Nicol , and some distance in front of this tube was placed a screen with a pinhole at the same height as the pinhole in the diaphragm .
The position of the screen was defined as before by lines drawn on the table .
A radius was drawn on the diaphragm , and also , since the tube itself was moveable , a was made on it so that it could be turned into the same position .
The crystal was placed on the brass table so that its plane was vertical and passed through the centre of the circular top .
On placing the eye opposite the aperture in the screen facing the elliptic analyser , it was found that a bright dot of light was visible .
Thus the horizontal incident ray already mentioned must have been reflected by the crystal surface so that it passed through the apertures in the screen and the diaphragm fitted to the analysing tube .
These apertures could be replaced in the same position , and therefore the direction of the reflected ray was a fixed horizontal line .
Conseqnently the normal to the surface of the crystal bisecting the angle between the incident and reflected rays was a fixed direction .
Supposing the crystal to have been taken down for polishing , it Mr. C. Spurge .
On the Effect of Polish on the couId be restored to position by placing the diaphragms and their proper stations , and setting the crystal so that a bright dot light was visible to the eye in front of the last screen .
The only possible changes in position that this method allows are displacement parallel to the table and a rotation of the face of crystal in its own plane .
The former was prevented by means fixed marks on the table , and the latter by taking every precaution to leave the base of the crystal in contact with the table unchanged , and later on by the use of a template .
A series of experiments were made to determine whether the diaphragms and screens could be removed and replaced in exactly the same positions .
For this purpose the screens and diaphragms were removed one at a time and replaced , using only the setting lines .
It was found that the dot of light remained visible , while the slightest displacement of the screens caused it to disappear .
During each set of experiments the screens and diaphragms were frequently replaced to determine if the crystal remained unmoved .
The surface of the crystal was shielded during the day by a box with aperbures , and completely covered at night .
In taking the Teadings the quarter wave plate and Nicol were turned into such .posibions that the centre of the field as dark as possible .
IV .
Experiments made with a Natural Face .
Observations were now made with the light flected from that natural face of the crystal which seemed the best .
The resuIts are given in Tables I , II .
The observations made with the crystal were always consecutive , none being rejected after the first satisfactory observation had been taken .
Taking from Table I the mean values of , we have\mdash ; Thus , and we obtain for the values of the quantities which determine the nature of the polarised light , , We have to find whether these two quantities are altered by polishing .
The subsidiary quanbiby is given by\mdash ; We are calculating merely for the purpose of verification , and are not using it to determine the naturo of the polarised light .
Take , then , to be the least positive angle which satisfies the last equation .
Thus .
This is the 1nean value of .
To estimate the error in determining , take the 12th set of observations in Table I. We have\mdash ; whence .
There is thus a difference of in the extreme value of from the mean .
The cause of this apparently large variation in the values of will be considered later on .
Mr. C. Spurge .
On the Effect of Pohsh on the V. Determination of the of the to the Faee Polished .
The crystal was taken down from its position on the brass and , in order to ensure that the face from which the light reflected was ground parallel to itself , the inclinations of the face two other faces which were left untouched , were obtained before after polishing .
The measurements were made in the usual way the spectrometer .
It was found that a fair imags of the slit could obtained by reflexion at each of the faces .
Three times the was completeIy dismounted and measurements of each angle The angles are almost unalt and the differences are within limits of ) erimental error .
Lines were drawn round the sides the crystal parallel to the edges , and after polishing remained parallel , which was an additional confirmation .
The conclusion which we arrive is that the polished face was parallel to the face .
VI .
Method of Polishing the Crystal .
The natural face which had been the subject of the recorded in Tables I and II was polished .
The polishing was formed by myself to ensure an exact knowledge of the treatment received .
I am indebted to Professor Threlfill for the nse of the and materials .
As previous experimenters seem to have a difficulty in obtaining a surface polished in the same manner , it be well to state the exact mode of polishing the surface .
That the crystal might be polished under the same conditions pressure , a rectangular block of lead was cemented to the crystal .
polishing especial care was taken not to press on the crystal wards , but to exercise only lateral pressure .
The crystal was polished with emery on a pIate of glass which had been plane by grinding on a slate .
The emery was prepared as Reflexion of Light from Iceland Spur .
very fiqe emery was scattered over a tub of water and allowed ; after standing for a number of minutes , the liquid was off , and the ssdiment , which was deposited on standiug for a number of minutes , was preserved for use .
kinds of were used , .
:\mdash ; Stood 1 minute , depositsd in 5 minutes .
\ldquo ; , 30 , , ours .
2 hours , , 40 , The crystal was polished about twenty minmtes with each kind in a perfectly quiet room to avoid dust being deposited on the glass and causing flecks in the crystal .
Next a bed of refined pitch was prepared having its upper surface perfectly plane .
Upon this the crystal was polisbed with rouge for about three hours .
Finally , the surface was carefully cleaned by washing it in a stream of water .
VII .
periments made with the same Face Polished .
The screens were carefully set in their places , and by means of these the crystal was fixed in the same position as formerly by the method describsd towards the end of Section .
Then the observations recorded in Table III were taken consecutively .
Taking from able III the mean values of , we have\mdash ; Thus , and we obtain for the values of the quantities , which we are seeking in order to determine the nature of the polarised Comparing these values with those formerly obtained which follow Table II , we see that the eifect of is to cause a small alteration of tlIe ratio of the axes and in the inclination of the major axis of the ellipse .
Thus the ratio of the axes has been changed from to , while the inclination been changed from to , an alteration of about 16 ' .
These results also show that the reflected light is exceedingly nearly plane polarised .
Again , the Tables for the subsidiary quantity so that The value of obtained bsfore polishing was\mdash ; ll These values ars not the same , but this has no bearing on the polishing , inasmuch as is a constant of the instrument and is independent of the crystal .
of Light from Iceland Spar .
order to test the truth of the results we have arrived at , prism , the azimuth of hich could be read offi by a divided to 3 ' , was mounted on the side of the table opposite to that on the elliptic analyser was , in such a position that , when the angle incidence was the same as before , the light was rsflected along the containing the Nicol .
It was found that the light reflected the surface could be reduced to a minimum but could never be eclipsed .
The minimum was very small but quite perThis might be due to the fact that the light was not homoBut further experiments showed that , even with orange glasses there was a perceptible minimum .
Next the inciwas increased and diminished in succession by , and the result obtained .
We , therefore , conclude as before , that the 1vas slightly elliptically polarised .
VIII .
Discussion of the Determinahon of the Refardation of the Quarter Wave Plate .
We have now to inquire into the discrepancy of the values of which have bsen found before and after polishing , and Now , is the ratio of the tangents of two angles , one of which is very nearly .
Consequently , a very small error in this last angle will a considerable change in the value of .
We therefore come to this conclusion , that it is not a good plan to determine the insbrumental constant by reflexion from Iceland spar , because the light reflected is so slightly elliptically polarised .
To determine more accurately the value of , a steel plate was substittlted for the crystal .
At first , white light was used as before , but the image found to be strongly coloured .
The difference in readings for blue and red rays was seYeral degrees .
After trying a spectroscope and coloured glasses , a ruby glass was selected as the best .
Then the observations recorded in Table were made .
Mr. C. Spurge .
On the of on the Table V we may find with what accuracy we can work the elliptic analyser .
The greatest difference between the mean and a single is , so that the error of determination of is less than a centage .
The mean value of I is , and the greatest error or 14 ' .
The meaIl value of is , and the greatest error is what less than 26 ' .
Let be the space retardation measured in for wave-Iength , then\mdash ; Thus error in the determination of expressed in is\mdash ; As in the experiments , this is very nearly the in which the of the analyser has been determined by fessor Stokes .
The present results agree well with his accuracy .
" " The mean error of single observations amounted to about the determination of bhe azimuth of the principal axis , about three four thousandths in the ratio of the minor to the major axis , and little more than a thousandth part of an undulation in the tion of We notice that in the mode of expressing in degrees an error be made more apparent , for the mula shows that space retardation is divided by the small quantity .
Thus , the difference between the two sets of observations in leng- , we find that the error of detelmination of is only per cent. of a wave-length .
* ' Phil. Nag vol. 2 , 1851 .
of Light from lceland Spar .
value of which has been determined is for red light , whereas experiments made on the crystal face were performed with white It was not possible to determine accurately for blue rays , as glass absorbed too much light .
We cannot , therefore , employ value of as determined by the steel plate to accurately correct observations made with the crystal .
Nevertheless , let us see what effect of the substituhon will be .
We have Since is supposed known , eliminate the smaller quantity , and we find This formula shows that if there be any change in the ratio of the rxes thers must be a change in the value of .
Looking at the Tables I , III , we see there is such a change .
Using the last written form , ula and the mean values of from Tables I , III , we find that for a natural face , and for the polished face Thus , even when the value of for red light as determined by the steel plate is used , we find that the numerical ] indicate that the character of the change produced by polishing remains the same .
In the case of the natural face , taking 12 observatio1ls of from Table I , the greatest deviation from the mean is less than 9 ' , leading by formula A to a deviation from the mean of less than 6 per cent. In the case of the polished face , the greatest deviation will be , according to Table III , about 10 per cent. The value of has been determined for only one colour , and consequently we cannot calculate its value accurately for light of mean wavelength , so as to be able to compars the value of obtained from the steel plate with the value of obtained from observations with the crystal .
Let us , however , examine in a general way what the effect of change of wave-length on the value of is .
Let be the tbickness of the selenite plate , the ordinary and extraordinary indices for a line of the spectrum .
Mr. C. Spurge .
On he of Pohsh on the Then and for a line Consequently , Now decreases from the lines A to , and varies by over 30 psr cent. The variations in the diflerence are much less if any analogy holds with Iceland spar and quartz , which the variation is less than 4 per cent. * Both causes tend to crease as the wave-length diminishes .
Thus , somewhere about the rdation is , and will approximate to the value derived Table I for white light , as we approach E. This discussion shows us that while the steel plate aifords a accurate means of determining for light of given refrangibility , is to a large extent a purely ntal constant , such that value has no effect on the character of the resuIts , and , when we cider that white light was used in the previous experiments , the of determined by Tables I , III , would seem to be confirmed by steel plate observations .
We therefore accept the resuIts of Tables as determining the alteration in the polarisation of the light .
IX .
Statement of Results .
The of Tables are brought together in Table VII the sake of future reference and comparison , so as to exbibit a tical view of the final result up to this stage of the work .
X. Variation of Surface State of a Polished Crystal with the Time .
It is a point of importance to determine if the state of the of a polished crystal is so permanent that it does not alter with the * Budberg , ' Poggendorff , Annalen , ' vol. 14 , p. 45 ; Glazebrook , ' Phil. Trnns .
, ' 1879 .
Reflexion of Light from lceland for otherwise the experiments made would not have very great unless the time elapsed since polishing were specified .
I found preceding experimenters had made investigations on this point to a few bodies , and had come to the conclusion that the did not alter with the time .
Thus Seebeck found such a refor some glass experimented upon , Sir John Conroy has that in the case of surfaces the surface state is a permanent one , not being destroyed by contact with a liquid or considerable amount of rubbing with a chamois leather .
It therenaturally occurred to me to examine this question for Iceland this purpose a simple analyser consisting of a Nicol and a circle was set up on the side of the ] opposite to tlJe analyser , and placed in such a position that the light could be to a minimum .
The observations taken are recorded in VIII .
Thus there is in the period of six weeks no time variation of a polished surface , for the differences between the means are quite within the limits of experimental error .
Also this result is confirmed by the general character of the observations made with the elliptic analyser , which often extended over some weeks , during which no change was noticed .
XI .
of State of a Polished with Change of Temperature .
An attempt was made to secure greater accuracy by altering the arrangement of the , rts of the elliptic analyser .
For this purpose the screw fixing the vertical stem was loosened , and the circular rim rotated through two right angles about the stem as axis .
The Nicol mas unscrewed from its collar , and ths cell containing the quarter wave plate from the disk .
The quarter wave plate was now connected with the collar and the Nicol with the disk .
In this mode of * ' Poggendorff , Annalen , ' vol. 20 .
'Roy .
Soc. Proc vol. 31 , 1881 .
Mr. C. Spurge .
On the of Poluon the arrangement , therefore , the Nicol moves in azimuth , quarter wave plate with it , while the quarter wave plate has an independent motion in azimuth .
After taking 700 readings I became convinced that the this mode of using the analyser was less than previously .
sequent investigation of the theory , made by tracing the surfaoe of a point whose coordinates represented the intensity of the mitted light , and the azimuths of the Nicol and quarter wave seemed to confirm this conclusion .
It is well known that in the case of some bodies the application heat considerab ] influences the mode in which they reflect Thus , in the case of flint glass , Sir David Brewster produced as an alteration as in the polarising angle by varying the To test the alteration in the case of Iceland spar , a small tray with ice was placed on the top of the crystal , all the rest of the ratus being carefully protected , especially the woodwork .
I that , leaving the quarter wave plate and Nicol untouched , no ceptible effect was produced by lowering the temperature of , and also that no difference could be detected when tions were taken .
I conclude that in the case of Iceland spar moderate ranges the effect of temperature is insensible .
XII .
New Series of Since the changs in the mode of using the elliptic anaJyser scribed in the last section had diminished the accuracy of the vations , it was considered best to revert to the original of the parts of the analyser .
The instrument was therefore The screw which fixsd the stem of the analyser was unloosened , the disk rotated through two right angles about the stem as axis .
tube was set to in the same direction as before by means of diaphragms and scresns .
The quarter wave plate was from the collar and the Nicol from the disk .
The quarter plate was now connected with the disk and the Nicol with the Thus the arrangement of the parts of the analyser was now like that described in Section II .
Since the quarter wave plate has been unscrewed , we must longer expect the values of I to be the same as in Table will be a constant difference between the preceding series of and those which follow , because the arbitrary zero from which I measured is now changed .
For these reasons it is necessary to a new set of observations to serve as a standard of reference .
Sir John Conroy has attempted to determine the effect of a crystal of Iceland spar by observing with a simple analyser * ' Phil. Trans 1815 .
Reflexion of Light from Iceland Spar .
lamp reflected from the surface immersed in water .
* With to a comparison of my own work with his rssults , a second burner was placed close to the former .
The brass table was unclamped and rotated till the face of the crystal was in such a position that the incident light was reflected to the opposite side of the table to that on which the elliptic analyser was placed .
The reflected light was observed with a simple analyser consisting of a Nicol , the azimuth of which could be determined by means of a divided circle and verniers to 3 ' .
The observations made with the elliptic and simple analysers are given in Tables IX and X respectively .
Mr. C. Spurge .
On the of Polish on the Taking the mean values of from Table IX , the analyser gives whence , and we have for the values of the which fully determine the nature of the reflected light , and The previous experiments , the results of which ars recorded , give , and The reason of the difference of the values of I is the change of the index error .
In order to compare the values of I found in a subse .
quent part of the paper with those already obffiined , we must there .
fore subtract .
The changes in may ascribed to two in resetting the instrument so that it was not exactly in its former positio1l , and error of experiment .
It should be noted that the difference produced by the reversal of the instrument is mot at all compaxable with the differencs prodnced by polishing .
In the case of the readings taken with tbs simple analyser , the mean reading will correspond to the azimuth of the major axis of tho ellipse , determined by the elliptic analysel .
Thus from Table X ' XIII .
ination of the of Rotation of the Crystal Face in its own Plane .
It was suggested that the setting of the was such as to a tion o face in its own plane , which might possibly be the cause of the differences hitherto observed .
A wedge of angle was constructed , and upon this the crystal was placed .
So deIicate was the mode of setting by screens that , even with the labour of hours , I unable to turn the crystal into such a position that the reflected ray emerged through the pinholes .
I therefore took thin sheet and gently wore its surface away so as to obtain a fine wedge which drawn under a of the crystal .
By means the reflected ray with some was rendered visible .
observations are divided in the following table int sets of two three .
]]iach set was taken after the crystal had been aken off wedge and replaced upon it in position by means of the screens .
this means the accuracy of setting can be judged .
Mr. C. Spurge .
On the Effect of Polish on the Table VII shows us that the eflect of polishing is to decrease to , a diminution of .
Since then a 427'produces an increase of ; to produce an effect equal of polishing by rotation , the crystal face would have to its own plane .
No such error of settin was possible .
Again , , and The last reading is of an altogether differen magnitude to has hitherto occurl.ed .
A template was constructed and used in future observations to test the accuracy of setting .
It was found that the setting by the was always correct on afterwards being examined by the template , and when it was out of adjustment by the screens the template did not fit .
These three grounds seem to be more than sufficient for the rejection of the hypothesis of a rotation of the crystal face in its It may be of some interest to calculate the value of the change the ratio of the axes produced by a rotation of .
We thereforo calculate Thus whence and The valne before rotation was XIV .
of the Effect of Repohshing the same Face of the Crystal .
gation of the of Means of a Series of Obser vations with an or Simple ser. Compan..son of the Relative the Elliptic Simple Analysers .
Much of the valus of the observations on polishing must depend on the fact whether the crystal can be polished in the same manner , that is , polished and repolished so as to obtain the same resuIts .
The crystal was , therefore , submitted to the same process of polishing as has been described already in Section .
The observations with the same face thus repolished are recorded in Table XII .
Mr. C. Spurge .
On the Effect of Polish on the The observations in Table made with the elliptic analyser divided into sets of 12 , for object is not only to determine the change produced by repolishing , but also to discover the a mean of 12 sets of observations with a view to fixing the of working with the elliptic analyser , also the errors of means of the sets of observations recorded in Tables I\mdash ; IV .
convenience of comparison the means are exhibited together Table XIII .
We now proceed to consider the accuracy of the sets of obser vations with the elliptic analyser .
Table XIII shows that the greatest deviation of a set of 12 observations from the mean is less than per cent. for the ratio of the axes .
The best set differs from the mean by under per cent. Again , Table XII shows that the greatest deviation from the mean for I is or under 3 ' , whereas the least deviation from the mean is under 1 ' .
We take , then , the upper limits as the errors of observations .
Let us now consider the effect produced by repolishing .
Before repolishing , the calculation following Table shows that\mdash ; After repolishing , XIII shows that\mdash ; The difference between the two values of is 2# per cent. , which may be covered by the limits of errors of observation .
Again , before repolishing\mdash ; Aftel repolishing , Table XII shows that\mdash ; There is thus a dffierence of or , which is covored limits of errors of observation .
Light Jrom Iceland Spar .
Table VII shows that the effect of the first polishing was to I from to , a diminution of ; but the of the second polishing has been to increase I from to , an increase of .
Thus the effect of the polishings has not been cumulative so as to cause the original deviation of polishing b ) increase still further on second polishin .
On the contrary , I has approached by a small quantity the value it would have for a natural face .
The inference may be drawn that the first polishing produced all the effect that polishing can bring about , and that the second polishing diflers from the first by a which there is some ground for as an error of experiment .
But if this is so , it may be thought that the values of , the second independent test of a change in the reflected light , should also alter in the same manner and sense as those of I have done .
Now , Table shows that the effect of polishing is to diminish from to , a diminution of .
Table XIII and the calculation following Table IX show that the effict of repolishing is to cause to from to , an increase of .
Thus changes sympathetically with I. In Table are contained the resuIts of the observations made with the simple analyser , and we have first to investigate the limits of accuracy .
In regard single observations there can be no doubt of the correctness of Sir John Conroy 's observation that " " in order to obtain accurate results with observations of this kind , it is necessary to make a large number of , ervations , and take their mean * Thns , in Table XIV the means of tens of readings ars recorded , and purposes of comparison the means of fifties are tabulated .
The results in Table show a greatest error of from the mean for a 100 readings , and of about 4 ' for 10 .
Before polishing the value of I as given by Table X is\mdash ; ' After polishing , Table gives\mdash ; ' There is thus a diffel'ence of 8 ' .
This is covered by the larger limit above , but not by the lower .
This is not surprising when the errors and fluctuations of single observations taken into account , and also the number of observations .
The parallelism of the polished and repolished faces was tested by means of the lines round the sides .
As these lines gave no perceptible inclination , it did not seem worth whiIe to measure the angles with a spectrometer , since the differences caused by repolishing are so small .
Thus , we come to the conclusion that the repolishing produced little , if any , alteration in the nature of the reflected light , and that this alteration is very small compared with the change produced Reflexion of Light from Jceland Spa a natnral face .
Also , this inference is supported by two as regards the elliptic analyser , and by the resuIts obtained with simple analyser .
Sir John has come to a different conclusion .
" " It did not seem worth while to make any further experiments with artificial surfaces , as it seemed certain the results would be untrustworthy Conroy 's experiments were made with the surface immersed in water to secure a grealer rotation .
I am not sure that the greater rotation would compensate for the loss of light which must ensue when reflexion takes place at the surface of two media of nearly the density and refractive Another cause of error has been indicated by Conroy .
There was some uncertainty whether the face was polished parallel to itseIf .
Though it wouId not altogether account for the large changes observed by Conroy , yet I found in the cass of a crystal , with which I performed a first set of experiments for two months , that a perceptible alteration of the inclination of the face was sufficient to reduce the elliptically polarised light to plane polarised .
Thus as regards my own experiments such a change of inclination would have been quite fatal , and I abandoned the crystal .
There is another point which is worthy of attention .
An extreme ill-luck seems to have befallen those who have had their crystals polished by others .
Thus Seebeck found that in the case of a glass polished by an artist there was a difference of , but that he himself could polish it so that the difference was only .
I think most likely this may be the probable cause of the Iarge alterations observed by Sir J. Conroy .
XV .
liixperiments with a Face parallel to the Repolished Face .
To make the whole series of experiments as concIusivs as possible , I now attempted to make a series of experiments with a nabural face as at tlJe b of the paper .
Unfortunately the crystal did not cleave very readily .
In splitting the base of the crystal came into two or three pieces , which were fixed on to it again as accurately as possible in their former position .
As the sides of the crystal also came off , the inclination of the face to the repolished face could not be determined .
There was some doubt as to whether the face of the crystal was in the sams absolute position .
The surface was not very good , being somewhat broken , and as the reflexion took place from part of the crystal nearer an edge , it had to be placed so much on one side of the brass table that the template could not be used .
Then the observations recorded in Table were taken .
Thus we find that and We have now to compare these results with those recorded in Table , taken some fifteen months previously .
Originally wo had\mdash ; whereas the present value is\mdash ; The agreement is very close , and , as might be expected from a compalison of Table with the results of Table IX , th latter number is somewhat larger than the former on account of the resetting of the instrument .
Again , first series of observations show that the effect of polishing was according to Table to change I from to , a decrease of , whereas the second series of experiments show that the result of repolishing has been according to Tables XII and XV to change I from to , a decrease of This resuIt is correct as regards sense , but not so satisfactory as regards magnitade .
It must , however , be remembered that the breaking up of the base of the crystal would most probably have an influence on the value of I. We are now in a position to sum up the resuIts of this investigation .
The process of polishing a natural face of a crystal of Iceland spar with emery and rouge does most certainly alter the state of the surface .
This alteration is evinced by a change both in the ratio of Reflexion of Light lcelamd Spar .
axes and in the azimuth of the major alcis of the elliptically larised light , and was observed in the case of two different crystals hich were made the subject of experiment .
Since th light reflected from the surface of the crystal is exceedingly plane polarised , the absolute value of the change in the ratio of axes is small ; but the relative change is considerable , for is by polishing from to .
Also the change in the of the major axis is not very large .
As regards the disturbing causes , it is found that temperature and do not cause any very perceptible alterations in the surface stats a polished crysta ] .
The experiments prove a result unnoticed by Seebeck , that an emery-rouge polished surface gives perfectly concordant results on repolishing , and in this respect is quite as satisfactory as the chalkpolished surface recommended by hi .
This conclusion is supported both by the elliptic and simple aIlalysers .
And in general the results of the paper tend to confirm the views of Seebeck rather than those of Sir J. Conroy , for Seebeck in his paper prefers polished surfaces because of the liability of the natural surface to tarnish .
In conclusion , my best thanks are due to Mr. Glazebrook for his advice , and to Professor J. J. Thomson for placing at my disposal a room and apparatus in the Cavendish Laboratory .
" " urther Experiments on the Distribution of Micro-organisms in Air ( by Hesse 's Method By PLRCY F. FRANKLAND , Ph. D. , B.Sc. , F.C.S. , and T. G. HART , A.R.S. .
Communicated by Prof. FRANKLAND , D.C.L. , F.R.S. Received November 22 , \mdash ; Read December 9 , 1886 .
[ PLATE 3 .
] In a previous commumcation entitled " " The Distribution of Microorganisms in Air a number of experiments have been recorded one of us on the relative abundance of microbes in the air of various places and of the same place at different times .
The determination of the aerial micro-organisms in these experiments was made by means of Hesse 's apparatus , the method of using which was there fully described .
Since the publication of the above experim we have been extending our investigations by means of this method , and the resuIts which we have obtained form the subject of the present communication .
In addition to the determination of the number of micro-organisms XLII .

