]\gt ; [ 493 ] IX .
On the of Vortex Ring , the Action upon each other ofTwo Vortices in Perfect Fluid .
J. J. THOMSON , B.A. , Fellow of Trinity College , Cambridge .
by Lord RAYLEIGH , F.R.S. Reoeived November 16 , \mdash ; Read December 8 , 1881.* following paper contains ( 1 ) a discussion of the vibrations which take place in the axis of the core of a vortex ring whose section is very small in comparison with its aperture when the axis is made to deviate slightly the circular form ; and ( 2 ) liscussion of the action upon each other of two vortex rings which move in such a way that they never approach nearer than a large multiple of the diameter of either .
The fluid in which these vortices exist is supposed to be frictionless and incompressible .
The method which I have employed is the same in both cases , and is purely kinematical .
It is merely the application of the fact that if be any equation to a surface which always consists of the same particles then where are the velocities of the particle at along the axis of respectively , and where the differential coefficients are partial .
The surface of a vortex ring is evidently a surface of this kind , and the equation written is the condition that should be the equation to the surface of a vortex ring .
I have found that this condition , joined to the ordinary expressions for the velocity due to a vortex element , is sufficient to the problems discussed in this paper .
This is an instance of the large number of problems in vortex motion which are capable of purely kinematical solution ; indeed , a vortex theory of gases would be entirely kinematical so long as we only considered the molecules of gas themselves and not their effects upon the containing vessel , \amp ; c. For example , in this theory when two atoms clash , the problem of finding their subsequent motion must be capable of solution by purely kinematical considerations , but in the ordinary theory of * Since the paper was sent into the Society it has been copied by the author with changes in the notation , introduced chiefly to facilitate the printing , but no change of any importance has been introduced into the substance of the paper .
matter .
Problem I. To find the vibrations of the circular axis of a vortex ring .
Usin cylindrical coordinates let the equations to the axis of the core be where and are small compared with the radius of the core when undisturbed ; the summation over all integer values of between zero and infinity .
The axis of is perpendicular to the plane of the vortex , and is measured ffom the axis of as initial line .
The velocity ue to a distribution of vortices is proportional to the magnetic force produced by a system of currents arranged in exactly the same way as the vortices and of the same strength .
Now the vortex filaments we are considering are distributed uniformly ( or very approximately so ) in a ring the radius of whose transverse section is very small in comparison with the radius of the aperture .
Now if electric currents flow uniformly through a conductor of such a shape the magnetic action at a point outside or on the surface , of the conductor is the same as if all the ctlrrents were condensed into one flowing along the axis .
ffence when finding the velocities outside the vortex ring we may suppose the vortices condensed into one at the axis of the core .
If be the angular velocity of molecular rotation , the radius of the transverse section of the core , then is the strength of the vortex which we must suppose placed at the axis of the core .
We shall for brevity denote by The components of the velocity at the point are given by where is the distance of the point from the point a point on vortex whose polar coordinates are is an arc of the vortex rin * See a note by Sir W. THONSON at the end of HELMHOLTZ'S paper on ' Vortex Motion PhiL N 1867 .
' ecbricity and Magnetism , ' 2nd edition , S. 683 .
AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
495 Now from the equation to the axis of the vortex we have neglecting and and If be the cylindrical coordinates of the say Let where the 's are functions of , and Since and are functions of , . .
will be functions of , but since only enters into and in the form , the terms in the 's which involve will be multiplied by or and so will be small .
If then terms of higher dimensions in and We shall only require those expressions for a point nearly in the plane of the vortex where is very small , so that in this case up into several steps .
To determine the parts of ) independent of and , in which we shall not suppose smail , ( 1 ) ( 2 ) . . . .
( 3 ) These are the velocities due to the undisturbed vortex , and in using them in the second half of the paper we require determined without supposing to be small .
2nd .
The values of arising from small terms in As far as now concerned , because they involve AND THE ACTION OF TWO VORTTCES IN A PERFECT FLUID .
3rd .
Small terms in ma { \mdash ; 4th .
Small terms in MR. J. J. THOMSON ON THE yIBRATIONS OF A VORTEX RING , 5th .
Small terms arising from containing .
These are . .
Collecting the terms we find ( 4 ) - ( 5 ) ( 6 ) Fig. 1 .
Let the figure represent a section of the vortex ring by a plane through its straight axis .
Let be the angle which the radius vector drawn from the centre of the section of the core to any point on the surface of the ring makes with the straight axis of the ring .
Let Then the equations to the surface of the core are ' AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
499 Since the vortex rings always consists of the same particles if be an equation to its surface , we must have when the differential coefficients are partial .
is the velocity in the direction in is measured , the angular velocity round the axis of , and ths angular velocity of round a normal to the plane containing the axis of and C. Applying this equation to the first of the equations to the core , we get or in a similar way we find where is the velocity of a point on the surface of the core parallel to the axis of Now is zero when and are both zero , and it will be small in this case since and are both small , hence neglecting the squares of small quantities , these equations become . . . . . . . . .
( 7 ) . . . .
( 8 ) But Substituting for and the values given in equations ( 4 ) and ( 5 ) , we get Since the 's are multiplied by the small quantities we may suppose since we neglect quantities of the order that the 's are found on the supposition that and are zero , or that the 's are the same as if the vortex was undisturbed .
Let us denote the value of the for the undisturbed vortex by German letters .
Equating the two expressions for and putting we get equating the coefficients of and we get . . . . . . .
9 ) ( 10 ) If we we get Since the term is not multiplied by any small quantity we cannot suppose the 's to have the same value as for the undisturbed vortex , we must substitute for Since the other terms are multiplied by small quantities , we may substitute for the 's their undisturbed values .
Making these substitutions we get Equating constant terms and the coefficients of and , we geb . . .
( 11 ) . . . .
( 12 ) The first of these equations gives the velocity of translation of an undisturbed circular vortex ring , the second is the same as the one we previously obtained for We must now proceed to find the values of the supposing the transverse section of the vortex core to be small compared with its aperture .
Since THt ACI : ION OF TWO VORTICES A PERFECT FLUID .
501 except when when Now where Now in the case we are considering is very nearly equal to and is very small , hence is very nearly equal to unity where and is very small Therefore Now since is very small will be very large , and the large part will arise from very small values of , or from values of very nearly equal to , the parts arising from small values of and from values near to will evidently be equal ; hence the integral will be approximately where is large compared with , this integral or , since is large compared with , this is approximately or retaining only the more important terms hence except when and ( 14 ) If we substitute these values for the 's the equation gives or , since is approximately , we get if we substitue this value for and for . . . . .
( 15 ) The second term on the left-hand side of this equation being small compared with the first , we get as a rougher approximation . . . . .
( 16 ) The equation gives on substitution ( 17 ) Substituting for in equation ( 11 ) we find ( 18 ) This agrees to the degree of approximation we are working to with the value for the velocity of translation of a circular vortex found by Sir W. THOMSON and given in Professor TAIT'S translation of HELMHOLTZ'S paper on ' Vortex Motion\ldquo ; ( Phil. Mag. , June , 1867 ) .
The value of given by equation ( 16 ) is also the same as that obtained by Sir W. THOMSON .
Substituting for the in equation we find AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
503 or neglecting terms on the left-hand side which are not multiplied by the large quantity we find or ' ( 19 ) Differentiating equation ( 15 ) with respect to the time and substituting for from ( ] 6 ) we get . . . .
( 20 ) or ( 21 ) ( 22 ) where A and are arbitrary constants .
These equations show that the circular vortex ring of indefinitely smadl section is stable for all displacements of its circular axis , and that the time of vibration for a displacement expressed by is ( 23 ) If be the velocity of translation of the vortex , viz. : ' the time of vibration is Sir W. has proved that the circular vortex ring is stable for all alterations in the shape of the cross section .
If we combine this with the result just obtained we see that the circular vortex rin is stable for all possible displacements .
Sir W. THOMSON has also proved that for a displacement of the order in the shape of the cross section of the vortex arc the time of vibration ; hence these vibrations begin by bein much quicker than those we have been considering , but since for large values of the latter are proportional to whilst the former are only proportional to , the vibrations of a higher order will be quicker for the circular axis than for the core .
When is very great , thus the amplitude of is equal to the amplitude of and a constant quantity ; thus each point on the arc describes a circle about its mean position with an angular velocity ' MR. J. J. THOMSON ON THE yIBRATIONS OF A VORTEX RING , Problem II .
To find the action upon each other of two vortex rings which move * as never to approach nearer than a large multiple of the diameter of either .
For the sake of simplicity we shall suppose that the normals to the planes of vortices intersect .
Fig. 2 .
Let the plane of tlje paper contain and the normals to the two vorticee , let A be the vortex moving along the vortex moving along Let the figure of the circular axis of the vortex A be given by where is measured along and perpendicular to .
Since the vortices never approach near to one another and will be .
small compared with ; they will be functions of the time which we shall have to find .
Let the figure of the circular axis of be given by where and are lred respectively along and perpendicular to .
For the same reason as before and will be small compared with .
To find how the vortex is affected by the vortex A we shall have to find the velocities of the fluid along and due to the vortex A ; in doing this we may as a first approximation assume that the axis of A is circular and in one plane , i.e. , we may calculate the velocities as if and were both zero .
Let denote the angle between and .
Let be taken as the axis of the perpendicular to drawn upwards through the centre of the vortex A , being the axis of Let be the direction cosines referred to these axes of a radius vector in the plane of the vortex ring , drawn from the centre of the vortex ring and making an with the intersection of the plane of the vortex ring with the plane of the paper .
To find ough the centre of a sphere draw planes parallel to the two vortex rings and let these be , the former being parallel to the ring A and latter to D. Let A be the poles of these great circles .
ACTION OF VORTfCES IN A RFECT I'LUID .
50 Fig. 3 .
Then A are parallel to our of , respectively .
The angle A or is , and if is parallel to the radius vector bove r to , is equal to Then The velocity along the due to vortex A is by formula 3 given by where is the strength of the vortex A Now since the vortex rings never approach nearer than a large multiple of their diameter , ' will be small compared with , if we neglect small quantities of a higher order than , we find Let the coordinates of the centre of the vortex ring be , then for a point on the vortex ring Although for reference we give the complete values of and to the order of approximation we are working to , yet when we have in the expressions for the velocities a coefficient consisting of terms of differerent orders , we shall only retain the largest term .
If we do this we find From the formulae ( 1 ) and ( 3 ) we find the velocity along Hence , the velocity along at the vortex due to the vortex A AND THIB ACTION OF TWO rICES IN A PERFECT FLUID .
.507 In this expression for we have only retained the highest terms in each coefficient .
If be the velocity parallel to the axis of , we have The velocity perpendicular to the plane of the vortex The velocity along the radius vector of the vortex ring due to the vortex A or substituting for their values .
( 25 ) These expressions will enable us to find the effect of one vortex on .
We have , for example , expressed the velocity perpendicular to the plalle of due to the vortex ring A in the form .
, in Problem I. we expressed the velocity in the same direction due to the vortex itself in bhe same form , hence the total velocity perpendicular to can be expressed this form , but by formula ( 8 ) the velocity perpendicular to the plane of is plane of the vortex D. To find we must therefore express the velocity due to the vortex A as a function of the time .
In order to make the work as simple as possible we shall suppose that the vortices in their undisturbed states had equal strengths and radii .
In the small terms which express the velocity at the vortex due to the vortex A we may as a approximation calculate the quantities on the supposition that the motion is undisturbed .
In order to make the expressions simple as possible , let us ur the time from the instant when the distance between the centres of the vortices has its least value ( it is easy to see that this will be when the line joining the centres of the vortices is parallel to the line bisecting the angle between their directions of motion ) , then the square of the distance between their centres will be expressible in the form when is the least distance between the centres , the time that has elapsed since the centres were this distance apart ; let be the velocity of translation of either vortex when undisturbed , then Therefore Making these substitutions we find that the velocity perpendicular to the plane of the vortex where .
( 26 ) ( 27 ) AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
509 The velocity along the radius vector where ( 28 ) ( 29 ) We can now write down the differential equations giving and We shall begin with that giving , as the simplest , as a reference to equation ( 19 ) will show that the vortex ring contributes nothing to this term , so that integrating an(1 writing for brevity instead of 4 ) we find ' where the arbitrary constant arising from the integration has been determined so as to make when Substituting for their values we find ( 30 ) The last terms are the only ones that do not vanish when .
Putting find Now is the angle through which the plane of the vortex is turned , and since the vortex moves at right angles to its plane this will be the angle through which the direction of motion of the vortex ring is turned .
Since , is negative , the part of the vortex ring where is positive is tilted backwards ; now as we have taken it , is positive for the upper part of the vortex , hence this part of the vortex is tilted backwards , and the normal to its plane , which is the direction in which the vortex ring moves , is bent towards , the direction of motion of the vortex ring A through an angle whose circular measure Thus the deflection , other things being the same , varies inversely as the cube of the least distance between the vortices .
Let us now consider the effect of the vortex ring on the vortex A B. Let us take the ) endicular to the plane of as the new axis of , the perpendicular to this drawn upwards in the plane of the paper as the new axis of .
The work we went through before consisted in finding expressions for the velocities along the axes of coordinates due to one vortex at a point on the other in terms of , and then finding the velocities perpendicular to the plane of the vortex and along its radius vector in terms of the time by substituting from the equations velocity perpendicular to the plane of the vortex sin , velocity along the radius vector Now in finding the effect of the vortex ring on the vortex A , the general expressions giving the velocities will be the same asbefore , as we have taken S AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
511 ponding axes of coordinates .
The difference in the work will come in when we substitute for the quantities involved their expressions in terms of the time ; if we denote corresponding quantities in this case by affixing dashes to the symbols , denoting them the previous one , we easily find velocity perpendicular to the plane of the vortex A velocity along the radius vector It will be seen that we can get the expressions for the quantities denoted by the accented letters from those for the quantities denoted by the unaccented letters by writing instead of , hence the value of will be got by writing in the expression for ( equation ( 30 ) ) instead of , and interchanging and Hence the value of when Now this being negative shows that the parts of the vortex ring A where is positive are tilted backwards , now is positive in the upper half of the vortex ring A , therefore the direction of motion of the vortex A , which is perpendicular to the plane of the vortex , is turned away from the direction of motion of the vortex through an angle whose circular measure is ; but since in the case we are considering , this angle is the same as that through which the path of the vortex is turned towards the path of the vortex A B. We may express the results we have obtained by saying that the direction of motion of the vortex which is in front when the vortices are nearest together , is bent towards the direction of motion of the one which is behind , that the direction of motion of the latter is bent through an equal amount in the same direction , and that the amount of this bending is Let us now consider the effect the collision has on the size of the vortices .
The equation giving the increase in radius is part independent of in the expression for the velocity along the radius vector of D. NDCCCLXXXII .
3 A reference to equation ( 17 ) will show that the vortex ring itself contribu4 nothing to this term , therefore part independent of in the expression for the velocity due to vortex ring A along the radius vector of D. Hence from equation ( 28 ) we have If we integrate this equation , substitute for their values as given in equation ( 28 ) and determine the arbitrary constant introduced by the integration , that when , we find This expression vanishes when ; it begins by being negative , so that the radius of is diminished at first when , the sign of depends upon the value of , if be less than it is certainly positive when ; when is positive , and its value is As this is positive , the vortex rin is bigger after the collision .
The effect of the vortex ring on the vortex A can be got as we saw before by writing for in the formula given above .
We have thus the ultimate increase in the radius of A given by or since Hence the radius of the vortex ring A is diminished by the collision .
The effect of the collision on the size of the vortices is thus to increase the radius of the one which is in front when the vortex rings are nearest together , and decrease that of the one in the rear by .
Hence the alteration in the radius jberis pibus , inversely to the cube of the shortest distance between the vortices .
THE ACTION OF TWO VORTICES IN A We can now find the force resultant of the impulse after collision .
The impulse for a vortex ring with a fine core equals the strength multiplied by the area .
Let be the small angle through which the direction of motion of the vortices is deflected ; let be the alteration in the radius of either , for the vortex ring in front @a will be positive , for the one in the rear it will be negative but of equal numerical value .
The resolved part of the force resultant of the impulse along the line bisecting their original direction of motion after collision the same as before collision .
The component perpendicular to the bisector of the angle between their directions of motion Substituting for @ and the values nd respectively , we find that the component of the impulse perpendicular to the bisector of the angle between the directions of motion vanishes , as it did before the ; hence we see that the force resultant of the impulse is not altered by the collision , result which we know is true .
We pass on to consider the terms and .
We know that coefficient of in the expression for the velocity the radius vector of the vortex ring D. Now the vortex ring itself , as we see from equation ( 17 ) , contributes to the expression for the velocity along its radius vector the term The vortex ring A contributes as we see from equation ( 29 ) the say Thus Now the coefficient of in the expression for the velocity perpendicular to the plane of the vortex D. The vortex itself contributes to this coefficient the term The vortex A butes the term say Thus Elitninating we find say or writing for the equation takes the form The solution of this differential equation is or choosing the arbitrary constants so that and both vanish when find nt'dt ' \mdash ; nt'dt ' The complete value of is given by the equation .
AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
515 Thus the coefficients of and in this expression for will involve integrals of the type I have not succeeded in evaluating this integral ; it is evident however that the more important part of these integrals will be produced during the time the vortices are nearly at their minimum distance apart .
During the time they are far apart they will not contribute anything appreciable to this integral , so that soon after the vortex rings have passed their minimum distance the equation may without sensible error be written where and are constants and There will be a similar } ) ression for .
Thus the vortex rings are thrown by the collision into a state of vibration about their circular form .
We can find the action of two unequal vortices on each other by means of work of a very similar character to that just given .
The only difference is that instead of the former values for and we must substitute the values where is the velocity of the vortex which is in front when they are nearest together , the velocity of the one in the rear ; is the angle between the line joining their centres when they are nearest together , and the direction of motion of the vortex in the rear , is the angle between this line and the direction of motion of the vortex in front , is the angle between the direction of motion of the vortices ; and are given by the equations I shall not trouble the reader with the expressions for the velocities perpendicular to the plane of either vortex and along the radius vector , but confine myself to quoting the most important consequences to be got from these expressions .
I find that after the collision the direction of motion of ( the vortex which is in R. .
J. SON ON IBRATIO ON A RINd , front when they are nearest together ) is deflected towards the direction of motion the other vortex A through an angle whose circular measure is where are the strengths of the vortices A and respectively , and and their radii , is the relative velocity of the two vortices , viz. : The direction of motion of A is deflected from that of through an angle whose circular measure is The radius of the vortex is increased by The radius of the vortex A is diminished by The velocity of the vortex is dinunished by The velocity of the vortex A is increased by where are the radii of the cross sections of the vortices , A respectively .
The kinetic energy of the vortex is by The kinetic energy of A is ninished by the same amount .
With the help of these results we may find the way in which two vortices affect each other in all cases .
AND THE ACTION OF TWO VORTICES IN A PERFECT FLUID .
5 17 Case 1.\mdash ; Vortices moving in the same direction .
Fig 4 .
1 .
When the positions of the vortices when nearest together are as represented in fig. 4 .
The way in which their paths are deflected is indicated by the dotted lines .
The vortex increases in radius and energy and its velocity is decreased .
Case II.\mdash ; Vortices in opposite directions .
The position of the vortices when nearest together is represented in fig. 5 .
The way their paths are deflected is indicated by the dotted lines in the figure .
The vortex decreases in radius and energy and its velocity is increased .
The vortex A increases in radius and energy and its velocity is decreased .
These results may be summed up in the following rule .
The vortex which first passes through the point of intersection of the directions of motion of the vortices is deflected towards the direction of motion of the other : increases in radius and energy and its velocity is decreased ; the other vortex is deflected in the same direction : it decreases in radius and energy its velocity is increased .
Note added March Sir WILLIAM THOMSON has pointed out to me that when is very great the time of vibration of the single vortex which for this case is ( equation 23 ) when is the wave-length , does not agree infinitely nearly as it ought with the value obtained by him for the rapidity of the transverse of a straight columnar vortex , which by formula 61 of his paper " " On the Vibrations of a Columnl Vortex is The results would agree approximately if were indefinitely great compared with , but it obliges us to neglect the factor , and thus the agreement instead of getting better as it ought gets worse as increases .
The way I determined is only suitable when , is small , as it is only allowable to expand as when remains small within the limits of integration .
I have therefore endeavoured to determine in a way which shall not be open to these objections .
With the notation of the paper since we may easily prove that if be small compared with , that and If we make these substitutions we find from equations ( 10 ) and ( 12 ) Hence we have only to find an expression for Now as is very small this integral will be very large , and as the large part arises when is small or nearly , we may write as the approximate value of AND THE ACIION OF TWO VORTICES A PERFECT FLUI Let us consider the more general integral Considered as a function of it is easy to prove that the integral satisfies the differential equation or if The solution of this differential equation may be expressed in terms of BESSEL'S functions of the first and second kinds ; we shall , however , only solve here the special case which we require .
It is easily proved that Since for it satisfies the differential equation This is the equation solved by Professor STOKES in his paper ' On the effect of Internal Friction on the Motion of Pendulums Camb .
Phil. Trans. , 1850 , and quoted by Sir WILLIAM THOMSON , Phil. Mag. , September , 1880 .
The solution is there shown to be where . .
and if as for when is infinite is shown to be LXXXII .
3 X so that Therefore Now since putting we find Therefore .
Now approximately hence is found , and if we substitute these values in the expressions for we shall be able to find the time of vibration in any particular case .
If we suppose is small , then approximately or if we neglect 1.23 in comparison with 2 Now AND THE ACTION OF TWO VORTICHiS IN A PERFECT FLUID .
521 Substituting we find say say Thus or the time of vibration Now if be not very large and the time of vibration If be large and the time of vibration or if be the wave length which agrees approximately with Sir WILLIAM result .

