]\gt ; INDEX SLIP .
WALKEE , G. W.\mdash ; On the of the Refractivo Index of Gases on Temperature .
Phil. Trans. , A. , vol. 201 , 190$ , pp. 4$54ae .
active I of Chses ; Dependence on nxzr , G. W. Phil. bazls .
, A. , , 1903 , pp. 435-455 .
[ 435 ] XI .
On the Dependence of the Refr(xctive Index of Gases on Temperature .
GEORGE W. LKER , , A.R. C. Sc. , Fellow of Trinity College , by Professor J. J. THOMSON , F.R. Received February 26 , \mdash ; Read March 26 , 1903 .
THE importance of this question was first impressed on me in the course of some theoretical investigations on refraction in gases and the closely related property of electric susceptibility .
A comparison of the actual temperature effect on a property of a body , with a theoretical formula professing to explain the property , is a very severe test , and one which has proved fatal to many theories .
According to GLADSTONE and DALE'S law , of which most theories of refraction are particular cases , the refractiye power of a gas is proportional to its density ; or , as a formula , where is the refractiye index , is the density , and a constant depending on the gas , but independent of temperature .
If , then , the gas closely obeys BOYLE'S and laws , we must have where is the pressure , is the temperature , and the coefficient of expansion of the gas at constant pressure .
lf the pressure is kept constant , we must have ( 1 ) Several observers have attempted to test this point .
MASCART , VON LANGES made observations on the refractive 'Annales de cole Normale Supe'rieure , ' Series 2 , vol. 6 , 1877 , p. 9 .
WIEDEMANN , 'Annalen der Physik , ' vol. 11 , 1880 .
'Travaux et Me'moires du Bureau International des Poids et Mesures , ' vol. 6 , 1888 .
S POGGENDORFF , ' Annalen der Physik , ' vol. 153 , p. 488 .
( 341 .
) 14.7.03 ORGL } ON TIIE NDENCE OF THE index of air different temperatures and made use of the formula ( 1 ) to calculate Their results are briefly as follows : LORENZ and BENOIT obtained a value of equal to the ordinary coefficient of expansion for air , VON obtained a value consider bly less , while MASCART obtained a value considerably greater .
LORENZ does llot indicate degree of accuracy he obtained , while BENOiT and VON LANGE do not appear to have obtained as great accuracy as MASCART .
MASCART imented on a number of gases , and in almost every case obtained a value of ppreciably greater than the corresponding coefficient of expansion of the gas .
range of temperature was from about C. to C. The disagreement between the results of the aboye-mentioned experimenters in the case nf air , and the somewhat limited range of temperature used by MASCART , led me to think that a repetition of the experiments on a few gases would be of value .
I set myself the task of obtaining an accuracy of 1 in 500 over a range of temperature from C. to C. , and I think the results show that this accuracy has been attained and in some cases surpassed .
The method used was JAMIN'S interference method , which I shall briefly describe , although it is well known ( see fig. 1 ) .
The rays of light from a monochromatic flame fall on a thick glass block , whose faces are optically plane and parallel , the back face being silyered .
Two parallel beams of light are thus produced and through the two tubes filled with the gas and reach a second block of glass identical with the first .
The two beams unite on emerging from the second glass block and produce interference bands , which may be observed through a telescope .
When the pressure in one of the tubes is altered , the bands move across the field of view .
As will be proved later , the number of bands displaced for a given difference of pressure enables us to calculate the refractiye index of the gas .
The glass blocks which I used were made by FELDER UND HERTEL in Munich .
The were centims. , the faces , , being optically plane and parallel .
One of the blocks was placed on an adjustable screw stand , so that the necessary adjustments might be mada The other block was placed on a heavy block of hard wood .
The tubes for holding the gas were made of brass , and were about 100 centims. long centim. diameter , and had soldered to them at each end a stuffing box ( see fig. 2 ) .
The tubes were soldered to an outer jacket , which was also made of brass , and was tightly wound on the outside with a thick layer of cotton wool .
The ACTIVE I GASES ON vertical tubes at each end of the jacket admitted the introduction of a thermometer fitted through a rubber cork .
Steam or water entered at and was pumped out at the corres o hole at the other end of the jacke is an optically plane and parallel plate of glass , 17 millims. diameter , millims. thick .
The four plates were all cut from the same plate of worked glass by HERTEL .
is a piece of hollow cork to reduce eddics of cold air .
Fig. 2 .
I experienced very great difficulty in making the joints between the glass plates and the brass tubes absolutely air-tight under the varying conditions of temperature and pressure .
I succeeded finally by using a rubber washer millim. thick between the glass and the brass , and then painting bicycle enamel round the junction .
This material dries rapidly and hardens , but still with sufficient elasticity to avoid straining of the glass .
It is not porous , nor does it melt or even soften at 10 C. It is , moreover , soluble in ether , so that the glass plate can be recoyered unimpaired .
Small brass tubes ( see fig. 3 ) passed through the side of the jacket and were screwed and soldered , one to each of the long brass tubes .
These served to connect the tubes with the manometer for recording the pressure .
These small brass tubes passed through short brass tubes of slightly larger diameter , soldered to the jacket , thus allowing play during altel.ation of tempel.ature .
The joint was made by a short piece of thick rubber tube , wired , and painted over with black enamel .
The steady temperatures required were obtained as follows : water was drawn through the jacket by means of a water pump .
This gave temperatures about C. Higher temperatures , such as C. , or C. , were obtained by the water through lead spirals of different sizes , mersed in a saucepan of water kept boiling .
Temperatures from C. to C. were obtained by water under reduced pressure in an old mercury bottle , and drawing the steam through the jacket by the water pump .
The arrangements for altering the pressure in the tubes of gas and measuring the differences are shown in fig. 4 .
The glass tube connected the brass tube with Fig. 4 .
one limb of the manometer .
The branch led to a tap , through which different gases could be passed into the apparatus .
The branch led to the second brass tube , and the branch led to the second limb of the manometer .
The branch was connected to a glass reservoir of about litre capacity , the tube being drawn to a fine capillary just before entering the reservoir .
was connected with a second reservolr by a ubber tube .
On lowering , mercury ran from to , and thus the pressure could be altered .
The tap being shut , a rence of the two tubes was produced , which could be measured on the manometer .
is a glass bulb introduced to keep the pressure in more nearly constant than it would otherwise be .
manometer was over an inch in diameter in the wide portions , and thus capillary error was avoided .
A steel scale was hung between two limbs and the level of the mercury , read by means of a telescope of a cathetometer placed in front of the manometer and about 150 centims. from it .
Whereyer it was possible , the glass joints were made by means of a blow-pipe , and only other joints were at the reservoir and at the small tubes connected with and .
These were made with thick rubber tube , wired , and painted with black enamel .
The glass taps and were fitted with mercury seals .
Very great care was taken in testing to see that the whole apparatus was absolutely air-tight .
REFRACTIyE INDEX OF GASES ON TEMPERATURE .
As a source of light , I used an ordinary Bunsen burner , placed about 150 centims. from the first glass block , and a small piece of bicarbonate of soda was held in the flame .
This gave a brilliant yellow flame for a long time .
The position of the interference bands was observed in a telescope with a micrometer scale in the eye-piece .
Theory of the surement .
MASCART and others have established that for pressures in the vicinity of atmospheric pressure the refractive power is proportional to the pressure , or where is a function of the temperature .
This is only true in cases where law practically holds .
In the case of such gases as ammonia , where the deviation from BoYLE'S law is ' correction is required .
Let length of tube , the wave-length of Na light , the initial pressure in the tubes , the final pressure in first tube , the final pressure in second tube , and the number of bands displaced , then The measurements were made as follows :Steam or water was allowed to run through the jacket for over an hour until the temperature was steady , and no drift of the bands was observed when the tap was open .
The bands were then adjusted so that a band was on the cross wire in the telescope .
The two limbs of the manometer were read , and also the two thermometers in the jacket , and the thermometer hung beside the manometer .
The tap was then shut , the reservoir lowered , and the tap opened .
When about 100 bands had passed , the tap was shut and the position of the band observed , the manometer read , and also the three thermometers .
was then raised , the tap opened , and the mercury allowed to flow back to .
The tap was shut when the original position was attained the readings again made .
This method provides a test of any possible drift of the bands in one direction due to creeping changes of temperature .
The proper temperature to take is readily seen to be the temperature as recorded when the pressure has been reduced , and not the mean of the initial and final temperatures .
With regard to the accuracy I consider that I was able to estimate millim. oq the manometer scale .
An interference band being a fuzzy thing and not sharp , I found it impossible to estimate more than of a band .
The bands appeared about 5 millims. apart in the eye-piece , and the breadth of a band about 1 millim. , or th of the distance between two bands .
It is therefore useless to have a screw micrometer reading to of a band when the eye cannot judge more than .
In the case of air at C. , 100 bands corresponds to a difference in level in the limbs of the manometer of about 16 centims. We may therefore consider the quantity to } ) accurate to 1 part in 500 .
With regal.d to the therlnometel .
S , they were made by R. , in tingen , divided in half degrees from C. to 10 C. They could easily be read to degree , but this accuracy is not necessary .
I had one of the thermometers standardised at Kew and compared the others with it under the same conditions as in the experiments .
The thermometers used in the experiments were placed in a bath at constant temperature , the same amount of stem being exposed as in the actual experiments , while the standardised thermometer was completely immersed .
Atmospheric Ai , : The air in the laboratory was used , and dried by means of phosphoric pentoxide .
The tap was kept shut so that the same air was in the apparatus throughout the experiments .
Tt would serve little purpose to give all the readings taken ; and I shall confine myself to a few specimens .
Throughout the initial pressure was as nearly as possible atmospheric pressure .
The thermometers in the jacket were marked 7 and 9 , and the thermometer placed under the scale of the manometer marked 6 .
15th November , 1901 .
The correction for each of the thermometers at this temperature as C. GASES ON TEMPERATURE .
441 Temperature grade .
Diagram I. Taking means we get From ( 1 ) and ( 2 ) ( 3 ) , , ( 4 ) ( 5 ) , , ( 6 ) Mean temperature Temperature of Ratio .
of iacket , manometer , C. .
CCI , \mdash ; A. 19th November , 1901 . .
Reading fenti danometer.ince .
Making the necessary corrections on thermometer readings we get From ( 1 ) and ( 2 ) ( 3 ) , , ( 4 ) Mean temperature Temperature of Ratio .
of jacket , manometer , C. C. 5 .
52 .
16.9 5 .
52 .
16.7 25th Noyember , 1901 .
From ( 1 ) and ( 2 ) ( 3 ) , , ( 4 ) Mean temperature Temperature of Ratio .
of jacket , manometer , C. 4 .
100 .
13.4 Mean REFRACTIyE OF GASES ON ]RATURF .
44:3 The next table gives a complete statement of the values of the ratio pressure at different temperatures .
It is convenient to reduce these values to what they would be if the mercury in the manometer was at C. and the tubes , of the length at C. The coefficient of expansion of brass was taken as 000019 .
The values of the corrected ratio are given in the fifth column .
The next columns are the values obtained by multiplying the corrected ratio by the factors , and respectively , being the temperature Centigrade .
The values are calculated to the nearest 5 in the third decimal place .
DRY Atmospheric Air .
The superiority of the coefficient 00360 is clear from the numbers .
The results are also shown in the diagram on page 441 .
Moreover , the agreement between the results on the 15th and shows that no measurable alteration in the gas had taken place .
I think the numbers justify one in taking the ratio as 6 .
The length of each tube was centims. between the inner surfaces of the glass plates .
The wave-length for Na light may be taken as metre .
The standard atmosphere as 76 centims. of mercury at C. Hence we get for the refractive index of dry atmospheric air Carbon Dio.xide .
The gas was made by warming a bulb containing sodium bicarbonate and drying by means of phosphoric pentoxide .
The whole apparatus was exhausted to under 1 centim. pressure by means of an oil-pump , and then the bulb containing the bicarbonate gently warmed until atmospheric pressure was attained .
This process of exhausting and refilling was repeated about six times , so that the apparatus might be considered filled with practically pure Observing for nearly a week at about , I noticed a gradual diminution in the value of the refractive index , which was comparatively rapid at first and became very slow in a few days .
The whole change was about 1 per cent. I consider that this was due to gas , probably air , coming off the walls slowly , and later results seem to support this view .
I refilled with pure , and now there appeared no change .
The results are given in the following table .
Throughout the initial pressure was maintained as nearly as possible at atmospheric pressure by adjusting the reservoir B. CARBON Dioxide , put in 28th February , 1902 .
OF GASES ON MPERATURE .
The results on 14th and 17th March are quite anomalous and beyond ordinary error of observation , and my inference is that more impure gas had come off at the highest temperature .
I therefore refilled with fresh , keeping the tubes at about C. while filling .
The following table gives the results obtained on the new gas .
NEW Carbon Dioxide , put in 18th March , 1902 .
No alteration appears to have taken place in the gas during the experiments .
The results are also in very close agreement with the results on the former gas .
Both sets are shown on the curye for We may take the ratio as and the refractive index for is Ilydrogen .
The gas was prepared from zinc and moderately diluted hydrochloric acid .
The apparatus was exhausted and filled about seven or eight times at the temperature of the room .
The gas was dried by phosphoric pentoxide .
Observations for a week at about C. indicated a gradual increase in the refractive index , which I attribute to carbon dioxide coming off the walls .
When this had ceased , the apparatus was exhausted and kept exhausted for a few hours , while the temperature of the tubes was maintained at about .
I hoped in this way to remove impure gas from the walls , but it will be seen from the results that Temperabure Diagram II .
I was not quite successful .
The exhaustion and re-filling with new hydrogen was repeated about four or five times .
REFRACTIyE INDEX OF ON ATURE .
The results are given in the following table:\mdash ; HYDROGEN put in 21st April , 1902 .
The results on the , and 24th April are fairly consistent , but on raising the temperature to over the values are distinctly higher .
From the 28th onwards the agreement is quite satisfactory .
I conclude that some impurity had come off at the higher temperature , and after that the composition remained constant .
The results from the 28th onwards give a ratio 3 .
The value of the ratio at for the original hydrogen may be taken as the mean of the first two observations on April .
This gives .
The amount of impurity is thus 13 in 3141 , or 1 part in 240 .
This would not alter the value of the temperature coefficient to the present order of accuracy .
I did not think it worth while to put in new hydrogen , as the purity could hardly be relied on by this method of preparation .
We take the ratio for the original hydrogen as ' which gives for the refractive index Temperacure grade .
Diagram III .
Amrnonia .
The gas was prepared dry by the following method , for which I am indebted to Dr. SCOTT , of the Davy-Faraday Laboratory .
A strong solution of ammonia , in a glass flask , was gently warmed , and the gas passed first through a tube containing dry caustic potash and next through a tube REFRACTIVE INDEX OF GASFS4 ON containing dry calcium chloride .
The calcium chloride absorbs large quantities of the gas .
One end of the tube was then sealed up and the other attached to the apparatus , and the whole exhausted with the oil-pump .
On gently warming the calcium chloride , the ammonia gas was liberated .
The former method , of exhausting and ] several times while the tubes were kept at about C. , was adopted ; and after the gas had been in the apparatus for a few days , the process was repeated .
More care must be taken in the case of ammonia , since the gas does not strictly follow the ordinary gaseous law .
If be the pressure and the temperature , the refractive index may be written .
Hence if be the pressure in the tubes , , , final , , , , one tube , , , , , , , , , second tube , the number of bands displaced According to MASCART* for ammonia per centimetre of mercury .
We must , therefore , take care that the value of ) does not vary to any extent throughout the series of measurements .
This point was carefully ttended to , and the value of was equal to 120 centims. throughout , the variation not exceeding 2 centims. My main object being the coefficient , and not so much the absolute value of , I did not make any measurements with another value of .
This I now regret ; but NIASCART'S value may be used , as he made experiments specially on this point , and had much greater ranges of pressure than my apparatus was arranged for .
In his papers I cannot find that he measured the temperature coefficient .
'Comptes Rendus , ' vol. 86 , 1878 , p. 321 .
VOL. CCI.\mdash ; A. 3 AMMONIA Gas , put in 13th June , 1902 .
I was unable to make a final observation at C. , owing to the fact that one of the glass joints had cracked during the night of the 25th June .
However , the results of 17th and 25th June are a fairly good test that no change in the gas had taken place .
We may take the ratio as Throughout centims. of mercury per centimetre of mercury ( MASCART ) .
We thus obtain for the refractive index ' being expressed in centimetres of mercury at C. At .
and 76 centims. pressure FRACTIVE INDEX OF GASES ON RATURE .
Temperabure Diagram IV .
Sulphur Dioxide .
The gas was obtained from a syphon of the liquefied gas and dried by means of a phosphoric pentoxide bulb inserted in the apparatus .
The former process of exhausting and refilling at a temperature of about adopted .
MR. GEORGE W. WALKER ON TH DEPENDENCE OF THE As in the case of ammonia , the pressure conditions must remain the same throughout .
My first experiments gave a temperature coefficient about 00415 ; but , after having been at a temperature about C. , there seemed to have been a considerable tion of gas , so that I could not obtain the former conditions at lower tures .
I therefore put in new gas and kept a record of the pressures .
Tbere was still a ption of the gas , although not so great as before .
Whether this was due solely to the walls of the apparatus or to the phosphoric pentoxide I am not in a position to say .
NEW Sulphur Dioxide , put in 16th August , 1902 .
The last two tions were made in order to obtain the coefficient of increase with We have Hence .
accuracy attained seems much greater than in the case of the former gases ; to be safo take the ratio as REFRACTIVE INDEX OF ON TEMPERATURE .
This gives for the refractive index At 76 centims. pressure and C. Temperature CenCIgrade .
Diagram The giyes a comparison of the results with those of former :\mdash ; REFRACTIVE Index for Na Light at 76 centims. Pressure and C. The following gives a comparison of MASCART'S temperature coefficients with those obtained in this paper:\mdash ; The values of the temperature coefficient of refractive index obtained are , in every , less than those obtained by MASCART .
It is somewhat futile to attempt to explain the difference ; but perhaps the following points are worthy of attention .
In my apparatus the tubes were about 1 metre long and the two rubber washers together about millim. thick , while MASCART used tubes about 25 centims. and his rubber washers were probably 1 millim. thick each .
He does not mention thickness , but LORENZ , who appears to have used an almost identical apparatus , used washers millims. thick each .
The somewhat irregular behaviour of rubber under varying conditions of temperature and pressure may have produced errors in MASCART'S observations , from which I consider that mine are entirely free .
I have already referred to the apparent escape of impurities from the walls of the apparatus .
MASCART makes no reference to this int , and gives no indication of how he tested the constancy of composition of the gas during the experiments .
It is true he analysed the gas chemically after the experiments , but this is hardly accurate enough for the point in view .
A very useful tnble of the results of different observers is given by BRUHL , 'Zeitschrift fur Physikalische Chemie , ' vol. 7 , 1891 .
REFRACTIyE INDEX OF GASES ON TEMPERATURE .
The difference between the temperature coefficient of refraction and the coefficient of expansion has naturally attracted attention ; but I do not propose to discuss the matter theoretically in this paper , mainly because I am now taking ) experiments on the temperature coefficients of the dielectric constants , which I hope will give me a more complete basis for generalization .
In conclusion , I wish to express my reat obligation to Professor }rsoN for having placed the appliances of the laboratory at my disposal , and for his kind intel'est in what has necessarily been a very tedious work .

