The BAKERIAN LECTURE was delivered by JAMES CLERK MAXWELL , M.A. , F.R.S. , " On the Viscosity or Internal Friction of Air and other Gases .
" The following is an abstract .
All bodies which are capable of having their form indefinitely altered , and which resist the change of form with a force depending on the rate of deformation , may be called Viscous Bodies .
Taking tar or treacle as an instance in which both the change of form and the resistance opposed to it are easily observed , we may pass in one direction through the series of soft solids up to the materials commonly supposed to be most unyielding , such as glass and steel , and in the other direction through the series of liquids of various degrees of mobility to the gases , of which oxygen is the most viscous , and hydrogen the least .
The viscosity of elastic solids has been investigated by M. F. Kohlrausch* and Professor W. Thomsont ; that of gases by Professor Stokes , M. 0 .
E. Meyer ?
, and Mr. Graham II .
The author has investigated the laws of viscosity in air by causing three horizontal glass disks , 10'56 inches diameter , to perform rotatory oscillations about a vertical axis by means of the elasticity of a steel suspension wire about 4 feet long .
The period of a complete oscillation was 72 seconds , and the maximum velocity of the edge of the disks was about -I inch per second .
The three disks were placed at known intervals on the vertical axis , and four larger fixed disks were so adjusted above and below them and in the intervals between them , that strata of air of known thickness were intercepted between the surfaces of the moving disks and the fixed disks .
During the oscillations of the moveable disks , the viscosity of the air in these six strata caused a gradual diminution of the amplitude of oscillation , which was measured by means of the reflexion of a circular scale in a mirror attached to the axis .
The whole apparatus was enclosed in an air-tight case , so that the air might be exhausted or exchanged for another gas , or heated by a current of steam round the receiver .
The observed diminution in the arc of oscillation is in part due to the viscosity of the suspending wire .
To eliminate the effect of the wire from that of the air , the arrangement of the disks was altered , and the three disks , placed in contact , were made to oscillate midway between two fixed glass disks , at distances sometimes of 1 inch , and sometimes of -5 inch .
From these experiments on two strata of air , combined with three sets of experiments on six strata of thicknesses '683 , '425 , and '1847 inches respectively , the value of the coefficient of viscosity or internal friction was determined .
Let two infinite planes be separated by a stratum of air whose thickness is unity .
Let one of these planes be fixed , while the other moves in its own plane with a uniform velocity unity ; then , if the air in immediate contact with either plane has the same velocity as the plane , every unit of surface of either plane will experience a tangential force I , , where , is the coefficient of viscosity of the air between the planes .
The force / , is understood to be measured by the velocity which it would communicate in unit of time to unit of mass .
If L , M , T be the units of length , mass , and time , then the dimensions of , are L-M T-1 .
In the actual experiment , the motion of the surfaces is rotatory instead of rectilinear , oscillatory instead of uniform , and the surfaces are bounded instead of infinite .
These considerations introduce certain complications into the theory , which are separately considered .
The conclusions which are drawn from the experiments agree , as far as they go , with those of Mr. Graham on the Transpiration of Gases* .
They are as follows : 1 .
The coefficient of viscosity is independent of the density , the temperature being constant .
No deviation from this law is observed between the atmospheric density and that corresponding to a pressure of half an inch of mercury .
This remarkable result was shown by the author in 1860t to be a consequence of the Dynamical Theory of Gases .
It agrees with the conclusions of Mr. Graham , deduced from experiments on the transpiration of gases through capillary tubes .
The considerable thickness of the strata of air in the present experiments shows that the property of air , to be equally viscous at all densities , is quite independent of any molecular action between its particles and those of solid surfaces , such as those of the capillary tubes employed by Graham .
2 .
The coefficient of viscosity increases with the temperature , and is proportional to 1+ aO , where 0 is the temperature and a is the coefficient of expansion per degree for air .
This result cannot be considered so well established as the former , owing to the difficulty of maintaining a high temperature constant in so large an apparatus , and measuring it without interfering with the motion .
Experiments , in which the temperature ranged from 50 ?
to 185 ?
F. , agreed with the theory to within 0'8 per cent. , so that it is exceedingly probable that this is the true relation to the temperature .
The experiments of Graham led him to this conclusion also .
3 .
The coefficient of viscosity of hydrogen is much less than that of air .
I have never succeeded in filling my apparatus with perfectly pure hydrogen , for air leaks into the vacuum during the admission of so large a quantity of hydrogen as is required to fill it .
The ratio of the viscosity of my hydrogen to that of air was '5156 .
That obtained by Graham was *4855 .
4 .
The ratio for carbonic acid was found to be *859 .
Graham makes it '807 .
It is probable that the comparative results of Graham are more exact than those of this paper , owing to the difficulty of introducing so large a volume of gas without letting in any air during the time of filling the receiver .
I find also that a very small proportion of air causes a considerable increase in the viscosity of hydrogen .
This result also agrees with those of Mr. Graham .
5 .
Forty experiments on dry air were investigated to determine whether any slipping takes place between the glass and the air in immediate contact with it .
The result was , that if there were any slipping , it is of exceedingly small amount ; and that the evidence in favour of the indicated amount being real is very precarious .
The results of the hypothesis , that there is no slipping , agree decidedly better with the experiments .
6 .
The actual value of the coefficient of viscosity of dry air was determined , from forty experiments of five different kinds , to be p=-0000149 ( 461 ?
+0 ) , where the inch , the grain , and the second are the units , and the temperature is on Fahrenheit 's scale .
At 62 ?
this gives= p007802 .
Professor Stokes , from the experiments of Bailey on pendulums , has found which , with the average temperature and density of air , would give =y-=00417 , a much smaller value than that here found .
If the value of , t is expressed in feet instead of inches , so as to be uniform with the British measures of magnetic and electric phenomena , as recorded at the observatories , = : '0001 79 ( 461 + 0 ) ='08826 at 32 ? .
In metre-gramme-second measure and Centigrade temperature , =o*01878 ( 1 + 00366 0 ) .
M. 0 .
E. Meyer ( Pogg. .
Ann. cxiii .
( 1861 ) p. 383 ) makes p at 18 ?
C. =*000360 in centimetres , cubic centimetres of water , and seconds as units , or in metrical units , O==03 60 .
According to the experiments here described , pu at 18 ?
C.-='02 .
Meyer 's value is therefore nearly twice as great as that of this paper , while that of Professor Stokes is only half as great .
In M. Meyer 's experiments , which were with one disk at a time in an open space of air , the influence of the air near the edge of the disk is very considerable ; but M. Meyer ( Crelle , 59 ; Pogg .
cxiii .
76 ) seems to have arrived at the conclusion that the additional effect of the air at the edge is proportional to the thickness of the disk .
If the additional force near the edge is underestimated , the resulting value of the viscosity will be in excess .
7 .
Each of the forty experiments on dry air was calculated from the concluded values of the viscosity of the air and of the wire , and the result compared with the observed result .
In this way the error of mean square of each observation was determined , and from this the " probable error " of u was found to be *036 per cent. of its value .
These experiments , it must be remembered , were made with five different arrangements of the disks , at pressures ranging from 0.5 inch to 30 inches , and at temperatures from 51 ?
to 74 ?
F. ; so that their agreement does not arise from a mere repetition of the same conditions , but from an agreement between the properties of air and the theory made use of in the calculations .

