]\gt ; viscosity of a gas is independent of its density , whatever may be the character of the enoounters taking place between the molecules .
In the typical case of a gas subject to a uniform shearing motion , we may suppose that of the three component velocities .
and vanish , while is a linear function , independent of and .
If } ) the viscosity , the force transmitted tangentially across umit of area perpendicular to is measured by .
This represents the relative momentum , parallel t , which in unit of time crosses the area in one direction , the area being supposed to move }with the velocity of the fluid at the place in question .
We may suppose , for the sake of simplicity , and without real loss of generality , that is zero at the plane .
The momentum , which may now be reckoned absolutely , does not vanish , as in the case of a gas at rest throughout , because the molecules come from a greater or less distance , where ( e.g. ) the value of is positive .
The distance from which ( upon the average ) the molecules may be supposed to have come depends upon circumstances .
If , for example , the molecules , retaining their number and velocity , interfere less with each other 's motion , the distance in question will be increased .
The same effect will be produced , without a change of quality , by a simple reduction in the number of molecules , i.e. , in the density of the gas , and it is not difficnIt to recognise that the distance from which the molecules may be supposed to have come is inversely as the On this account the passage of tangential momentum per is inversely as the density , and since the number of molecules .
is directly as the density , the two effects compensate , and upon the whole the tangential force and therefore the viscosity remain { dtered by a change of density .
On the other hand , the manner in whioh this viscosity varies with temperature depends upon the nature of the encounters .
If the molecules behave like Boscovich points , hich exercise no force upon one another until the distance falls to a certain value , and which then repel one another infinitely ( erroneously called the theory of elastic spheres ) , then , as Maxwell proved , the viscosity would be proportional to the square root of the absolute temperature .
Or again , if the law of repulsion were as the inverse fifth power of the distance , viscosity would be as the absolute temperature .
In the more general case where the repulsive force varies as , the dependence of upon temperature may also be given .
If be the velocity of meitn square , proportional to the square root of the temOn the iscority of Argon affected perature .
perature , varies as , a formula vbich includes the cases already specified .
If we assume the lready discussed \mdash ; that is independent of density\mdash ; this conclusion may be arrived very simply by the method of ' ' dimensions In order to see this we note that the only ( ( besides the density ) on which can depend } the mass of a particle , l ) the velocity of mean square , and the repulsive force at unit distance .
The dimensions of these quantities are as follows:\mdash ; ( mass)l le1lgth time ) ( mass)1 , ( length)1 time ) mass)time ) Thns , if we assnme .
we have when / lCC Accordingly where is a purely numerical coefiicient .
For a iven kind of molecule , and are constant .
( 3 ) .
The case of sudden impacts gives , already remarked , .
Henoe .
disappears , cmd the tion of dimensions shows that where is the diameter of the particles .
The best experiments on air show that , so .
as a of : kind can represent the facts , It be observed corresponds to When we rememl ) tho } ) , such xygell , hydrogen , and nitrogen , are regarded as diatomic , we mtty be to attribute the want of simplicity in the law collnecting vi temperature to the c.omplication introduced by the of symmetry in the molecules and conscquent diversities of presentation in an encounter .
It was with this idea that I thought it would be interesting to examine the influence of temperature the viscosity of argon , which in the matter of specific heat behaves as if composed of Lord Rayleigh .
single atoms .
From the fact that no appreciable part of the total energy is rotatory , we may infer that the forces called into play during our encounter are of a symmetrical character .
It seemed , therefore , more likely that a simple relation between viscosity and temperature would obtain in the case of argon than in the case of the ' ' diatomic\ldquo ; gases .
The best experimental arrangement for examining this question is probably that of Holman , in which the same constant stream of .
gas passes in succession through two capillaries at different temperatures , the pressures being determined before the first and after the second passage , as well as between the two .
But to a gas like argon , available in small quantities only , the pplication of this method is difficult .
And it seemed unnecessary to insist upon the use of constant pressures , seeing that it was not proposed to investigate experimentally the dependence of transpiration upon pressure .
The theoretical formula for the volume of gas transpired , analogous to that first given by Stokes for an incompressible fluid , was developed by O. E. Meyer .
Although not quite rigorous , it probably suffices for the purpose in hand .
If denote the pressure and volume of the gas as it enters the ) , .
as it leaves the capillary , we have ( 4 ) .
In this equation denotes the time of transpiration , the radius of the tube , its length , and the viscosity measured in the usual way .
In order to understand the application of the formula for our present purpose , it will be simplest to consider first the passage of equal volumes of different gases through the capillary , the initial pressures , and the constant temperature being the same .
In an apparatus , such as that about to be described , the pressures change as the gas flows , but if the pressurcs are definite functions of the amount of gas which at any moment has passed the capillary , this variation does not interfere with the proportionality between and .
For example , if the viscosity be doubled , the flow takes plaoe precisely as before , except that the scale of time is doubled .
It will take twice as long as before to pass the same quantity of Although different gases have been employed in the present experiments , there has been no attempt to compare their viscosities , and indeed such a comparison would be difficult to carry out by this method .
The question has been , how is the viscosity of given gas affected by a change of temperature ?
In one set of experiments the Phil. Mag. ' vol. 3 , p. 81 , 1877 .
Pogg .
Ann vol. 127 , p. 269 , 1866 .
On the cosity of Argon as affected Temperature .
capillary is at the temperature of the room ; in a closely following set the capillary is bathed in saturated steam at a temperature that be calculated from the height of the barometer .
If the temperature were changed throughout the whole apparatns from one absolute temperature to another absolute temperature we could make immediate application of ( 4 ) ; the viscosities at the two temperatures would be directly as the times of transpiration The matter is not quite so simple when , as in these the of temperature takes place only in the capillary .
A rise of tsmperature in the capillary now acts in two ways .
Not only does it change the viscosity , but it increases the } of gas which has to pass .
The ratio of volumes is thus ( 5 ) , subject to a small COl'rection for the effect of tempsrature upon the dimensions of the capillary .
It is assumed that the temperature of the roservoirs is the same in both transpirations .
The is shown fig. 1 .
The gas flows to and fro between the bulbs A , the flow from A to only being timed .
It is confined by mercury , hich cftn through connections of blown glass from A to and from to D. The bulI ) are ported upon their seats a little plaster of Paris .
The capillary is nearly 5 fcst ( 150 cm .
) in length and is connected with the bulbs by gas tubing of moder$te meter , all joillts being blown .
represents the iaoket through which can ) ssed ; its length exceeds that of the capillary by a few inches .
In order to charge the apparatns , the first step is the exhaustion .
This is effected through the tap , , with the aid of a Toppler pump , and it is necessary to make a corresponding exhaustion in and , or the mercury would be drawn over .
To this end the rubber terminal is temporarily connected with , while I leads to a commoll air pump .
When ths exhaustion is complete , the gas to be tried is admitted gradually at , the atmosphere being allowed again to exert its pressure in and D. TVhen the charge is suflicient , is turned off , after which remains open to the atmosphere , and connected to a nometer .
When a measurement is commenced , the first is to read the temperatures of the bnlbs and of the capillary ; I is then connected to ; force pump , and prcssure is applied until so much of the is over that the mercury below in assumes the positions } in the diagram .
I is then suddenly released so that the pressure asserts itself in and the gas begins to flow back into B. The bulb allows the flow a short time in which to establish itself before the time urement I the mercury passes the that in about 2 cm .
below atmosphere .
At the end ) pressure in A was 20 , and in cm .
, both above atmosphel.e .
Accordingly the driving pressure fell from 32 to 17 Three , or , in the case of hydrogen , five , observations of the time usually taken , and the agreement was such as to indicate that the mean would be correct to perhaps one-tenth of a second .
The time for at the temperaturc of the room tbout ninety seconds , and h.ydrogen econds , but these numbers are not strictly comparable .
On scosiiy of Argon as affected Temperature .
When the low temperature observations vere finished , the gas was lighted under a small boiler placed upon a shelf above the apparatns , steam was passed ough the jacket .
It was necessary to see that there was enough heat to maintain a steady issue of steam , yet 11ot so much as to risk a sensible back pressure in the jaoket .
The time of transpiration for air was now about 139 seconds .
Care was always taken to maintain the temperature of the bulbs at the same point as in the first observations .
There are one or two matters as to which an apparatus on these is necessarily somewhat imperfect .
In the high temperature ements the whole of the gas in the capillary is assumed to be at the temperature of boiling , and all that is not in the c to at the temperature of the room , assumptions not stric.tly compatible .
The compromise adopted was to enclose in the jackct the whole of the capillary and about inches at each end of the approaches , and seems sufficient to sensible error when remember the pidity with which heat is conducted in small spaces .
A second point is the assumption the instantaneous arc esented by the heights of the moving mercury columns .
If the necting -tubes are too narrow , the tance to the flo of entcrs into the question in much the same way as the flow of gas in the pillary .
In order to )tain a check upon this source of error tho tppal a has been ried .
In an earlier form the con necting were comparatively the resuIt the ratio of viscosities of hot and cold was substantially the same that subscquently ined with the improved apparatus , in which these tubes were much widened .
Even if there be a sensible residual error arising from this cause , it can hardly affect the comparison of temperaturecoefficients of gases whose viscosity is nearly the same .
I will now give an example in detail from the observations of December 21 with purified argon .
The times of transpiratiolt the ture of the room were in , .
Mean , When the capillaries were bathed in steam , the corresponding times ere .
Mean , The barometer reading ( corrected ) being , we as the temperature of the jacket C. Thus The reduction was effected by assuming ( 6 ) .
On the Viscosity of Argon as affected by Temperature .
With the above values we get As appears from ( 5 ) , the integral part of relates merely to the expansion of the gas by temperature .
If we take , ( 7 ) , we get This number is , however , subject to a small correction for the expansion of the glass of the capillary .
As appears from ( 4 ) , the ratio as used above requires to be altered in the same ratio as that in which the glass expands by volume .
The vahle of must accordingly be increased by , making The following table embodies the results obtained in a somewhat extended series of observations .
The numbers given are the values of in ( 7 ) , corrected for the expansion of the glass .
Air ( dry ) Oxygen . . . . . .
0.782 Hydrogen Argon ( impure). .
Argon ( best ) In the last nrials , the argon probably within 1 or 2 per cent. of :bsolute purity .
The nitrogen lines could no longer be seen , and scaroely any further oontraction could be effected on king with oxygen or hydrogen .
It will be seen that the temperature change of argon does not differ very greatly from the corresponding change air and oxygen .
At any rats the simpler conditions under which we may supthe collisions to occur , do not lead to values of suc as as or , discussed by theoretical writers .
I may recall that , on a former occasion , I found the viscos tie of rgon to be relatively to that of air , both being observed at the temperature of the room .
* .
Proc January , 1896 .

