]\gt ; On the Velocity of Fall of Spherical Particles th By E. UNNINGHAlI , , Fellow of St. John 's College , mbridge , Assistant Professor of Applied Mathcmatics , University , London .
( Communicated by Sir Joseph Larmor , Sec. R.S. fficeived Jannary18 , \mdash ; Read January 27 , 1910 .
) 1 .
Recent have called attention to the question as to how closely the rate of fall of particles a fluid when very small mates to that iven by the law obtained by Stokes from ] reasoning .
This paper deals theoretically with two of the ain ) divergence from that law .
Firstly , taking the particles to be snlooth spheres movi ] it is shown that the deviation to be expected on account of the dianleter of the particle being small , compared with the lnolecllar free path , is extremely small for such particles as have been experimcnted with ; and a odified formula is given which may ) taken to hold nately for lower pressures or particles of smaller dilnensions .
Secondly , an approximate treatment is made of the effect of the sinrultaneous presence of number of particles moying with the same velocity through the fluid , and it is found that the force required to maintain the motion of one of depends not only on the diameter , the ratio of diameter to the distance between the particles , and in such a way as to increase rapidly this ratio increases beyond the value With the figures ined by J. J. Thomson it is shown that these corrections together would imply an increase of not more than in the charge on the aseous ion .
* See the preliminary experiments of Zeleuy with spores British ) , and of Buller April , ; also the later concordant results of eleny with spherical particles Nature , , 1909 ) ; also Millikan Physical Review , ) Since this paper was prepaled , .
Larmor has informed the writer that eleny is pursuing subject of fall in gases at diminished , with a view to light on the nature of a molecular collision with a solid body , such as is required the theory of the radiometer and other purposes Maxwell , Phil. Tranh .
For convenience of reference , a short calculation on this point has been added .
The formula obtained in the first of this paper agrees with the general usion that of this kind cannot lead to a determination of molecular magnitudes , as at first sight eemed not impossible .
J. J. Thomson , ' Phil. Mag vol. , 1903 .
Mr. E. Cunningham .
Velocity of Steady On the Rate of Diffusion of Cloud of Partides through a Gas .
2 .
Suppose that a cloud of particles of mass at a concentration of } particles per unit volume is through a gas , the mean velocity of } the particles being , the mass of the gas molecule being , and the number of molecules per unit volume being .
It is desired to find the mean force per particle required to maintain velocity It will be assumed in the first instance that the collisions are of the nature of impulses between smooth elastic spheres , and that a collision takes place between a molecule and a particle when the distance apart of their centres is Consider the circumstances of collisions of this type with a single particle , the centre of the molecule at collision being within a small area siu of the sphere of radius surrounding the particle , being spherical polar co-ordinates having the direction of as axis .
Let the velocity of the molecule be and that of the particle The probability of molecular velocity lying within a range dudvdw about is where ; and that of the velocity of the particle lying within a range is where In order that a collision may take place within the iven limits in a given interval of time , the centre of molecule must at the beginning of it lie within a cylindric.al volume of base and height , with the restriction that must be positive .
The magnitude of the impulse I is .
Thus the probable impulse in the direction of on the particle per unit time is I the limits for being unrestricted , but not so for .
This expression is equal to Multiplying by the probability of the iven velocity and i$ , the mean impulse per particle per unit time becomes the limits for and and all values for which The evaluation of ) last is only possible on the assumption that is small compared with the mean relative velocity .
Making this -assumption , and integral , we have where .
Thus tiIl .
If the lower limit for were , the value of this expression would be The error in this value for is rand is therefore of the order times is the mean alne of , so that to the first order the above valne may be taken for K. Hence the mean inlpulse per ) article per tim is This , then , is the force per unit particle required to maintain a drift of the particles relative to the .
Tt represents the rate of transfer of momentum from a particle to the surrounding gas by reason of collisions .
It may be expected that , in the case where the particles are few in nulnber ompared with , and of the same order of 1nagnitude as , the moleeules , the mass velocity of the gas in the neighbourhood of a particle will be very htly affected by the motion of the particle ; but if the diameter of the article becomes comparable with the mean free path of the molecule , will become * This expression is also obtained as a particular case of by ( ' Annals de Chimie et de Physique , ' vol. 5 , 1905 , p. 4266 ) , and is applied by to the consideration of the mobility of ious .
Mr. E. Cunninghaln .
Velocity of Steady impossible to neglect the mass velocity produced in the gas by the motion of bove expressionfor transfer oomentum , ifference between telocity o 1 the particle and the mass velocity of the gas .
It may be remarked that though a cloud of particles has been spoken of above , the result will hold equally well for the mean velocity of a single particle taken over a time sufficiently long to include a very large number of in an infinite fluid .
machi bwhich transfer oomentum farti ]isions .
This ionsidered itoke , which ionstrated oered irdinary hestigation the nbourhood ossume , thatInstead , inemati that tasse m the mass velocity is .
Then the rate of dissipation of momentum on the hydrodynamical theory is being used now for the effective radius .
of the particle , inasmuch as it is assulned to be compared with the molecule .
In the steady motion this diffusion of momentum must just balance the momentum communicated by the particle to the gas .
large compared with , this is , from the above formula , .
Hence , or where is the mean free path of the molecules in the air .
Hence the force required to maintain the velocity is ; that is , the limiting velocity of falling particle will exceed that given by the ordinary formula in the ratio of to unity .
This ratio for a iven gas should therefore differ from unity by an amount which varies inversely as the density .
For air at ordinary pressure and temperature may be taken to be about , so that the limiting velocity would , on the present calculation , be double the usual value for a particle of radius about .
For particles is the mean free path obtained from the viscosity by the formula , not the value corrected for persistence of velocity ( vtde Jeans , ' Kinetic Theory of Gases , ' p. 1910 .
] Fall of Spherical Particles , etc. of the of the spores in Zeleny 's earlier experiments the ratio would be about The nature of the COqOll molecule and particle , so far , been assumed to be that which between smooth elastic spheres .
If this is replaced by hat each molecule ) came e ) in the palCicle , so that on -emission there was no correlation its velocity and that with which it collided , a similar calculation can be npnlses by which the molecules thrown out the particle no , if all directions of re-emission are supposed equally ) ) collision in the direction of the is ) If the integration be effected as before , the nean inlpulse is found to bear to that previously found the ratio of to that is , it is exactl .
double .
If , now , it is assumed , lIax , that a fraction " " \ldquo ; of the molecules experience collisions of the earlier type , and the remainder of the latter type , the equation the velocity becomes Maxwell loc. deduces from of Kundt and for impacts of air molecules ( glass the fraction .
is ) half .
This would reduce the ence frolu Stokes ' Law to of that obtained by putting .
If no experience perfect rellexion , wo.uld be zero , and ence old leduced to one-half .
If it ) asstlmed that for the majority of material particles .
lies the and , the result is that Stokes ' formula will yield velocity equal to half the actual for particles whose radii lie belween for particles of diameter ) , such as those used in Zeleny 's eriments , it will yield a velocity ljetween 1 ) and too ] .
eriments at lower ] ) should sho greater throw on the value of 4 .
A possible sotllce of iation from } a cloud of ) icles of dimensions lor which the hydrod ) ndition of is satisfied , is the mutual influence of the rticles , and the ctfect of the boundaries .
The exact solution of this is beyond Ivsis .
But an approximate 1nethod of attack ests itself , itsed on the fact that if a enclosed volume of fluid and are moving with equal parallel velocities , drift of the fluid with each particle in the direction of its motion in the immediate vicinity of the particle must be compensated by a drift in opposite direction between the particles .
In fact , if the distribution , uniform the mean flux of fluid across any area is zero if the fluid is contained within fixed boundaries .
In Stokes ' problem this condition is absent .
limitedtomovingwithinaconcentricsphericalmassoffluidwhoseboundaryinfluenceofthisconsiderationistoimaginethateachparticleiseffectivelyThemethodemployedtoobtainanideaofthemagnitudeofthe ixed.nstance , posing tadius aandb isuch these 1rounded boncentric spheres oadius brrangement oesarrangement , luid ipaces between teing supposed occupied .
by fluid at rest , the conditions of the problem are to some extent realised ; but the velocity of each particle in its own sphere will be somewhat less than the actual limiting velocity owing to the restriction imposed on the fluid between the spheres .
We may therefore expect by method to obtain a lower limit for the velocity sought in terms of the radius of each particle and the number of particles per unit volume .
The problem is therefore to be solved of the drag due to viscosity on a particle moving with velocity at the centre of a spherical envelope .
This can be effected exactly , the method adopted being that given by Lamb* for the solution of Stokes ' problem .
Taking the axis of in the direction of motion , the velocity at a point can be expressed in terms of spherical harmonics of the first order .
Putting and using the formula of type we obtain ' 3rd edit .
, S324-5 .
Now at the surface of the particle where and at the sphere These conditions the equations The resultant traction on the sphere is found to be , exactly as in the simpler problem , * The value of A obtained from the four equations ives a result which reduces at once to Stokes ' law when is Expanded in powers of , the result is .
It was said above that the value so found is a lower limit for the actual velocity under a given force , .
to the conditions imposed possibly more rigid than the problem demanded .
5 .
If now round each particle be described a sphere of radius , such that the new spheres overlap , but only by so much that they just include whole of the fluid , each particle again supposed to move in its own sphere , a new value for the velocity will be obtained .
The arrangement of particles being as before , will be the distance from a corner to tlJe centroid of a tetrahedron whose edge is 2 , so that .
It is not clear , however , that in this way an upper limit to the velocity of the particles is obtained , the following considerations point in that direction .
In either case the argument is that if the velocity system for the fluid ) obtained when fluid within one of the spheres is set in motion 1 ) enclosed particle , the external fluid being at rest ; and if the velocity systems thus obtained for all the particles are superposed , the resultant velocity system will approximate to the actual in the sense in the neighbourhood of each particle the conditions imposed by the continuity of .
cit. , p. ) , equations ( 5 ) and 13 ) .
the spheres supposed at distance 2 apart in a horizontal line ; ( ii ) along a 1 lnedian of an equilateral triangle of side 2 , there being supposed three spheres with their centres at the corners .
The continuous line refers to velocity found by the use of the larger spheres , the broken line to iven by the smaller .
Comparison of these curves and of others not given here indicates fairly clearly that the former is more likely to be near .
to the true facts .
ccordingly the values are tabulated below of the ratio of the forces 4 usinerthelhere , and , accorditoStokes'Iaw , ifferent values ofrequired taintain avelocity Vaccording tethod adopted a ; the second row of the table gives the corresponding value of ff 1910 .
] Fall of Spherical Particles , etc. 6 .
Assuming , now , that the modified laws obtained above are correct , we may see what correction they imply as necessary to J. J. Thomson 's calculation of the on the aseous ion .
The ures given by him fall into four groups to values of the on the ion\mdash ; respectively .
the ernres coiven for the number of drops per cubic centimetre , we are able to calculate the values of , and thence the correction to be made as in The result is that must be increased by , 6 , per cent. in the four cases respectively .
For the calculated values of the adii of drops , however , the correction on of the first cause ( S 3 ) aulotlllts to about 4 per cent. and is in the opposite sense , and therefore corresponds to a diminution of 4 per cent. in On account of both causes we may say that is to be increased by about 2 per cent. , that is by about IG is clear , therefore , neither of the causes discussed accounts for more than a small fraction of the discrepancy between the older and the more recent estimates .
* Rutherford , Address to British Association , ection A , 1909 ; ) Millikan , ' Physical Review , ' Dec. , 1909 .
VOL LXXXIII .

