]\gt ; Conduction through RareJied Gases.\mdash ; II .
By FREDERICK SODDY , M.A. , , and ARTHUR JOHN BERRY , B.A. , Physical Chemistry Laboratory , University of Glasgow .
( Received August 25 , \mdash ; Read November 10 , 1910 .
) In a previous paper* measurements were described of the thermal conductivity of twelve different gases in what may be termed a free-path vacuum that is , at pressures so low that the molecules conducting the heat from the hot to the cold surface do not , as a rule , experience mutual encounters .
It was that at sufficiently low pressure the conductivity of all the gases was proportional to the pressure , and the conductivity ( defined as the calories dissipated per second , per mm. pressure per square centimetre of hot surface , per 1o difference of temperature between the latter and its surroundings ) was compared with the theoretical conductivity , as calculated approximately from the molecular heat and mean molecular velocity of the gas , by means of the kinetic theory , on the assumption that the heat interchange between the molecule and the surface it impinges upon was perfect .
For argon and neon the ratio and respectively ) was in gratifying agreement with this assumption , but for all the other gases the ratio was less than unity .
For these the ratio was greater than , except for helium and hydrogen , for which the very low values and respectively were found .
The suggestion was hazarded that possibly for these light gases the interchange of heat is imperfect , owing to the greater velocities at which the molecules move .
In the present paper some of the earlier measurements have been repeated with the original apparatus , with better provisiou for keeping the temperature of the surrounding water jacket uniform .
Then the apparatus was rebuilt to allow of the measurements of the conductivity to be taken over a wide range of temperature , both of the hot and cold surface , and the effect of variation of the temperature on the ratio was examined .
Measurements were confined to the three ses\mdash ; hydrogen , helium , and argone Incidentally the effect on the conductivity of hydrogen of using a hot palladium surface instead of one of platinum was examined .
The result of these new has been to negative the suggestion already referred to , that the discrepancies between the found and calculated conductivities might be due to imperfect interchange of energy on impact .
The ratio appears to diminish as the temperature at which the experiments are 'Roy .
Soc. Proc 1910 , , vol. 83 , p. 254 .
Conduction of Heat through Rarefied performed decreases and to increase as the temperature is increased , and no anation of this can at present be suggested .
eriments with the Original Apparatus.\mdash ; The conductivities of helium , argon , neon , and hydrogen were redetermined by the original method and apparatus .
In the whole of the work to be described the water round the apparatus was kept at nearly constant temperature by first circulating through a considerable length of metal tubing immersed in a thermostat , usually at .
The values of and of are shown in Table I , ether with the results obtained previously .
They agree fairly well with results already given , that of hydrogen slightly Table I.\mdash ; Original Apparatus .
Strip at Further experiments then carried out on helium and argon , but the strip was now heated to different temperatures .
The electrical arlangements were also somewhat modified .
The two ratio arms of the bridge were two 1-ohm coils of stout Eureka wire .
The other arm of the bridge was a set of variable resistances of the same , so that the strip could be heated to various known temperatures .
In measuring the fall of potential along the strip , an Elliot voltmeter , with an 8-inch dial reading from to volts , was nently connected across the potential leads .
To com- pensate for the resista1lce of instrument , an equal resistance ( 389 ohms ) was put in parallel with the corresponding arm .
In this series the variable resistance was set so that at the balance point the strip should assume temperatures .
with the following :\mdash ; vater at , the points of the following liquids , ether } , water , ethylene dibromide , and aniline This was accomplished with the apparatus filled hydrogen and jacketed at the desired tenlperature , by the variable resistance until balance was obtained with a momentary feeble current .
The of the strip at the working temperature was found * This temperature was assumed from esistance of the strip , which was found to be a linear function of the temperature .
It if ; probable that the boiliug-point of the aniline had been raised by decomposition .
Messrs. F. Soddy and A. J. Berry .
[ Aug. 25 , by reading the fall of potential across the potential ] eads when balance was produced with a.known heating current , the walls of the apparatus being cold .
The results are included in Table II .
They showed that but little change in Table II.\mdash ; Original Apparatus .
Temperature of Strip varied .
the ratio was produced by changes in the difference of temperature between strip and wall from to , the temperature of the wall being always about .
The low conductivity of argon and the comparatively small range over which its conductivity is a linear function of the pressure make the determination of the ratio less accurate than for helium or hydrogen .
For helium the ratio remained sensibly constant over the range examined .
This was unexpected .
On the view suggested tentatively to account for the smallness of the ratio , that perfect interchange of energy is not secured with the faster molecules by a single impact , it is to be expected that the ratio would decrease as the temperature of the strip was increased .
Comparison of the Conductivity of drogen with Heated Surfaces of Palladium and Platinum.\mdash ; On the same view it had been deduced , since hydrogen is occluded by heated palladium more than by platinum , that the value of for hydrogen should be greater for a heated palladium surface than for one of platinum .
The point was simply tested in the following apparatus employed consisted of two parallel tubes carrying , axially , similar wires of platinum and palladium respectively .
The tubes were joined together at their upper extremities and immersed in the thermostat , in order that the same gas might be investigated under identical conditions of temperature and pressure .
The two wires formed two arms of a bridge , the rest of the bridge consisting of a stretched wire along which a sliding contact moved .
A sensitive galvanometer was inserted between the slider and the two ratio arms of the bridge .
Care was taken to avoid 1910 .
] of Heat through Rarefied Gases .
heating the wires to such an extent as to cause them to touch the sides of the tubes .
Argon and hydrogen were ated by this apparatus .
Experiments were performed in water at , and also with the apparatus cooled in a mixture of solid carbon dioxide and ether giving a temperature of about .
Both gases were ated over a fairly vide range of pressure , and the behaviour with both was absoluteIy identical , so far as could be seen .
The balance point remained at the middle of the wire , only to a very slight extent as the pressure of the gas became reduced , due to a slight difference between temperature coefficients of of the platinum and palladium wires .
These experiments , therefore , ative the prediction that hydrogen ) ected to exhibit a higher conductivity when tested with a surface of than with one of inum .
new tube , with the platinum strip employed in the appal'atus , was made up in a suitable for work at low temperatures .
To allow ansion and contraction of the with change of temperature , a spiral of hard copper wire was employed at the top , instead of a at the bottom as in the original apparatus .
But since resistance of this was a considerable fraction of that of the platinum strip , it was no longer possible to work with the strip as of a Wheatstone bl'idge .
The diameter of the new tube was very nearly the same as the old one ; it actually cm .
, and the potential leads were cm .
apart .
One of the set of determinations that ] been made with the old apparatus was repeated closely as possible with the new , and the curves obtained were practically identical , showing that the electrical properties of the strip were unchanged .
The method of was ollows : strip , with its potential leads connected with the voltnleter , was put in series with a constant esistance of stout Eureka wire ( shunted with the resistance of ohms , equal to that of the voltmeter ) , of such a value as to be equud to the resistance of the portion of the sl , rip between the potential leads at the } temperature , and with a rheostat for the heating current till this equality was obtained .
Adjustment was made by the use of a commutator , which exchanged the voltmeter and similar 389 ohrns resistance .
The rheostat was then simply adjusted until the volbmeter dings w constant for the two positions of the commutator .
With the apparatus sul'rotlnded by liquid air , the strip was worked at two temperatures , ( 1 ) at , and ( 2 ) C. Helinm and hydrogen were ated at both tempelatures .
The resistance of the strip between the potential leads was carefully determined at liquid air temperature and in a bath of solid carbon dioxide and ether .
This was done by observing with a potentiometer the fall of potexltial Messrs. F. Soddy and A. J. Berry .
[ Aug. 25 , across the potential leads of the strip , produced by a known momentary feeble current , when the apparatus was filled with hydrogen and immersed in the cooling bath .
It was found that the resistance of the strip was very nearly a linear function of the temperature from to .
At low temperatures the linear portion of the curve , expressing the relation between the heat dissipated and the pressure of the gas , is curtailed , and it was not found possible for this reason to work at low temperatures with argon , but experiments were done with hydrogen and helium .
The results are given in Table III .
Table III.\mdash ; Rebuilt Appal.atus immersed in Liquid Air .
From the expression before deduced , in which and are constants , while varies directly as the square root of the absolute temperature and inversely as the absolute temperature , it follows that varies inversely as the square root of the absolute temperature .
It had been deduced that the value of for helium at liquid air temperature should approach unity , since at that temperature its molecular velocity is similar to that of neon at room temperature .
The ratio for both helium and hydrogen was , however , found to be distinctly smaller at low temperature .
As before , variation of ths difference of temperature between the wall and the strip appeared to make little , if any , difference .
In these low-temperature measurements the radiation loss was negligible .
Further experiments were then performed with the apparatus jacketed with steam and the strip at 19 in one set , and 26 in another , and with the apparatus jacketed with aniline and the strip at In the experiments at higher temperatures , the radiation loss became of considerable importance .
It was determined in each case by extrapolating Where is the molecular heat at constant volume , the mean molecular velocity , the number of molecules in 1 .
of gas at the temperature of the wall and mm. pressure , and the number in the gram-molecule .
1910 .
] Conduction of through reJied Gases .
the curves to zero pressure , except in the experiment with the strip at and the bath at .
For this case the separate curves did not give a sufficiently concordant result , and the loss was determined directly in a calcium vacuum .
In attempting to push the temperature higher by jacketing with sulphur vapour heated in a quartz tube , the copper wire spring softened , and the apparatus was rendered useless .
The results are shown in Table Table \mdash ; Rebuilt Apparatus .
High Temperatures .
For convenience , the whole of the results are collected for each gas separately in Table .
The measurements are } in ascending order of bath temperature , the erent measurements at the same bath temperature being arranged in ascending order of strip temperature .
It appears that rise in temperature of both surfaces causes the ratio to increase , while an increase in the difference of temperature between wall and strip causes it to diminish .
Helium is perfectly in accord with this view , and so is hydrogen , though here only four measurements have been done .
For argon , also , the results , they appear at first sight irregular , are not inconsistent with the view that two opposing influences are at work , and sometimes the one and sometimes the other predominates .
It is clear , from the fact that a value as high as for has been obtained for argon , that this conforms to theory hardly better than the others , and that the agreement between the theoretical and experimental values found for it in the last paper was partly fortuitous .
The reasons for the discrepancy between the tbeory and the experimental results remain unexplained .
Admittedly , the theory , which assumes all the molecules to possess the teml ) erature of the wall , is imperfect , btlG it may be questioned whether this can make any very serious diflerence in the calculated values .
It is interesting to note that the conductivity found Messrs. F. Soddy and A. J. Berry .
[ Aug. 25 , gases at low pressure val.ies far less with the nature of the gas and with the temperature than is to be expected from kinetic considerations .
Thus the extreme values of found in these experiments re and , whereas the extreme values of over the same range are and Table Collected Results .
El .
Addendum , October 15 , 1910 .
Sir Joseph Larmor has called our attention to some recent papers by M. Knudsen on the dynamics of rarefied gases which reveal a source of error 1910 .
] Condnction of Heat through in the oing measurements of the of the gas .
First , we may be allowed to state that we had already considered the ents of the pressure to be the least satisfactory , and had decided , if an opportunity of continuin .
the work arose , to do away ether with the ange .
This could readily be done by having the tube .
the hot strip , the volume of which was determined once for all , connected by capillary tubing with a definite volume en.closed between two taps , for definite quantity of any gas to the apparatus .
By the apparatus with calcium at the start and ' the heat dissipation after each admission of one such quantity of gas , the experimental work would be much silnplified and the results more definite .
For in a previous paper*it has been pointed out that when the free path of the molecule becomes comparable with the diameter of the tube there is , properly , no pressure in the hydrostatic sense , since flow does not operate to equalise the concentlation .
It stated that under these conditions " " of pressure in apparatus connected to the gauge by a orifice not , strictly , pressure readi all Knudsen has deduced from the kinetic theory that between two closed sels at difl'erent temperatures containing the same gas , and connected by a tube , of diameter small compared with the mean free path of the gas molecule , there must exist a diffelence of pressure , coiven by where , and , are the pressures and tenlperatures of the two vessels respectively .
In fact , Osborne Reynolds , in work on thermal nspiration , had , long previously , alried at the same relation .
Knudsell brings in support of this remarkable deduction many striking experiments , and there can be little doubt some common physical measurements , as well as those discussed in the oing pages , are vitiated by this effect .
In the measurements , with the tube containing the heated strip jacketed at and low teulperatures , the pressure of the which remained at room temperature , would not represent the pressures in the tube .
the tubing was some 3 or 4 mm. wlde , yet this was always compared with the mean free path of the boas lnolecules .
As there to be no reason why the above relation cannot be exactly applied to our } ) , we have corrected the ratio , iven in Table , by multiplying it by , where is the lperature F. Soddy and T. D. Mackenzie , ' Roy .
Soc. Proc 1908 , , vol. 80 , 1 ) .
102 .
" " Eine Revision der ] ihewichtsbdigung der Gase ' Annalen der Physik , ' 1910 , vol. 31 , 205 .
Phil. } , p. 727 .
VOL. LXXXIV .
Conduction of Heat through reJied G of the room , assumed throughout to be C. , and the absolute temperature of the bath .
The values so corrected , and multiplied by the factor described below , have been added in italics as another column to Table .
It will be seen that the results now have quite a different appearance , not unfavourable to the original hypothesis of incomplete interchange of energy on impact .
For helium , the value of is distinctly higher at liquid air temperature than at room temperature , but remains almost constant from room temperature up to .
The low value with the bath at and the strip at 26 is exceptional .
For , the value of is also highest at the lowest temperature , and falls regularly , so far as can be seen , with increasing temperature .
Argon is still somewhat irregular .
Over the range investigated , to , temperature appears to have little influence .
As already remarked , these results are probably not so accurate as those with the other gases .
Finally , reference must be made to a recent paper by M. Smoluchowski , * who , in the last section of his paper , deals with the results obtained by us in our first paper , from the theoretical standpoint .
He gives an exact calculation for the conduction of heat in a rarefied gas on Maxwell 's assumption that some of the impinging ecules are reflected with their original velocity , while others are emitted with a new velocity corresponding to the temperature of the surface .
He points out that in our roughly calculated formula for , the numerical factor should be instead of 1/ 6 , so that all our values for should be multiplied by the factor In addition , the value of , to correct for imperfect equalisation of the mean energy of the impinging molecules to that of the molecules in the wall , must be multiplied by the factor , where is derived from the formula in which , and denote the temperature of the wall , of the impinging and of the emitted molecules .
With only the first correction in the values of , this ratio ranges for the 12 gases examined from for argon , to for , showing that the interchange of energy is always imperfect .
Smoluchowski does not discuss the interesting question as to how this coefficient is likely to vary with the velocity " " Contributions to the Theory of Transpiration , Diffusion , and Thermal Conduction in Rarefied Gases 'Bull .
International de l'Acad .
des Sciences de Cracovie , ' 1910 , No. 7 , p. 295 .
Prof. Larmor had sent me the same correction as M. Smoluchowski 's , remarked to him independently by Mr. Sydney Chapman .
A slight change has beeu made in the decimal value of the factor .
of Potential .
or temperature of the molecule .
He considers our earlier results to range very much as is to be expected , the of energy with the molecules of the platintlln wall being less perfect for the lighter molecules than for } eavier ones , and also for polyatomic than for nlonatomic , for the reason that intramolecular is less disposed to equalisation by impacts than energy of progressive motion .
It is clear that further experimental work must be done in the manner suggested befo the seary data are available .
The periment on the conduction of palladium platinum ires tells against the idea of imperfect interchange of en , but it ) no means lsive .
Moreover , Prof. remarks that the recent experiments of on electrified water drops in an field , when eted by the of E. make very } , so that , for air on water at any rate , litCle correction of that kind arises .
An .
for Photo Potential .
By GEORGE .C.Sc .
, lf .
A. , Superintendent of Eskdalemtlir ( Communicated by Dr. .
Shaw , F.R.S. October 11 , \mdash ; Road November ) , 1910 .
) The difficulties continuous ation of electrical potential gradient of the atmosphere very great ; but it is not my intention in this paper to discuss these .
object is to call attention to the part of an more ) to experiments made with an instrument that pronlises to be of considerable in wolk of this kind .
A recording voltmetel ' ought to have a uniform scalc value over the range for which it is intended to be used , it must possess a ) of insulatiou , and it ought to be capable of acting efficiently for periods ) having to be taken to } ) ieces for cleaning .
The general excellence of the Dolezalek electrometer and its rlce olnechanical sested to me its use as a , although I do not that it beon The instrunleut designed to measure very small differences of potential Boy .
Soc. Proc , 1

