]\gt ; [ 389 ] V. On the Theory of Free Lines .
J. H. MICHELL , Trinity College , mbridge .
by Professor J. J. THOMSON , F.R. Received Jannary 3 , \mdash ; Read January 16 , 1890 .
Introduction .
FHE attention of thematicians was first called to the subject of the present paper by a memoir of HELMHOLTZ'S in 1868 , on " " Discontinuous Fluid Motion.\ldquo ; In discussing the steady motion of liquids past salient edges of fixed obstacles , it found that the assumption of continuity of the motion leads to negative pressures the liquid .
HELMHOLTZ showed , in the paper above-mentioned , that some cases of bhis kind could be solved by assuming a surface of discontinuity , on one side , of which bhe liquid is at rest , and he gave a mathematical solution of one case where the motion is in two dimensions .
The next advance in the subject was made by KIRCHHOFF who , in 1869 , in a paper entitled " " Sir Theory freier Fliissigkeitsstrahlen\ldquo ; in 'CRELLE'S Journal , ' gave a generalization of the method which HELMHOLTZ had used , and obtained thereby the solution of three new interesting cases .
Subsequently in his 'VorIesungen uber mathematische Physik , ' he published another method and worked out the same problems by means of it , but gave no new ones .
RAYLEIGH in the 'Philosophical Magazine , ' December , ] 876 , discussed the solutions of KIRCHHOFF , and gave a drawing of the bounding free stream lines in one case .
As far as I know , these are the only investigations published on the mathematical side respecting a branch of hydrodynamics of great theoretical and practical interest .
In considering the method of transformation of polygons given independently by SCHWARZ and CHRISTOFFEL I have been led to a new transformation , which there with theirs , gives a general solution of the problem of free non-reentrant stream lines with plane rigid boundaries .
A considerable number of the cases of high interest prove to be of a tolerably simple nature , and I have worked out several in detail .
These problems occupy the first part of the paper .
In the second part I have given some extensions of the transformation formulae , which are applicable to problems of condensers and the form of bollow vortices in tain cases .
* ' Berlin Monatsberichte , ' IS68 ; and ' Gesamm .
Abhandl vol. 1 .
28 .
6 .
90 Theory of Let be two conjugate functions with respect to the two variables , so that , and write may be regarded as the rectangular coordinate of a point in a plane which we shall call the plane , and similar]y are the coordinates of a point in the plane .
Consider the functions Since they can be written in the form they both satisfy LAPLACE'S equation .
and we have so that V , are conjugate functions with respect to or The transformations of the present paper will be deduced from the properties of the function V , so that its nature must be considered in detail .
We have as alternative forms of V If the element of arc of constant in the plane be given by then so that MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINFS .
Lnd where is the angle the tangent to the curve makes with the axis of [ Mr. BRILL has used the function as a means of transformation ( ' Cambridge Phil. .
Proc vol. 6 , and Kessenger of bIath August , 1889 ) , and has thus anticipated me in one of the general theorems given in the latter part of this paper as I shall lotice in the proper place .
I was not acquainted with his work when I developed the method here given .
] Let be the curvature of the curve at , We have the well-known which is ( 1 ) Now let the arc of the curve be collnected with by the equation that ' or from ( 1 ) .
( 2 ) If consist of parts of straight lines we have simply .
( .3 ) The formula ( 2 ) suggests a general method for finding a transformation , such that is an arbitrary curve in the plane .
If the region within corresponds , point for point , to the part of the plane above , the problem is reduced to finding a potential function , which is continuous throughout the space bounded by a straight line , and such that over that straight line We have still to discuss the question of the singular points of V. The function V will be finite and continuous for all points except where two branches of a curve ( or ( curve ) cut .
At such a point is either zero or intinite , and in either case V is infinite .
It will be sufficient for our purpose to consider only the simplest singularity , that is , in which we have in the neighbourhood of a point of this nature and therefore here the value of will depend on the ature of the singularity in question , and will be seen from the particular problems to which we proceed .
PROBLEM I. To find the transformation , which makes the area for which is positive in the plane correspond point to the area inside a given rectilinear polygon in the plane .
( The problem of SCHWARZ and CHRISTOFFEL .
) Consider the conditions which the function V must satisfy in the plane .
( a ) There are to be no singular points for positive .
( b ) Along we have ( c ) At certain points of , which correspond to the angular points of the polygon in the plane , we have V infinite .
It is plain from this specification that the function V is ( to a constant ) merely the potential of masses at the singular points .
along ; and , therefore , so that where is the product symbol .
It remains to find the quantities .
Draw a small semicircle of radius around the point * ' ' Ueber einige Abbildungsanfgaben\ldquo ; ( ' CRBLLE , ' vol. 70 , 1869 ) .
" " Sul problema dell temperature stazionarie \amp ; c. ( ' Annali di Matematica , ' vol. 1 , 1867 ) .
MR. J. H. MICHELL ON THE THEORY OF FREE LINES .
and on this semicircle let , so that .
Consequently , as we pass from to ) goes from to The amplitude of dzfdw has , therefore , decreased by But the increase of amplitude is where is the intern al angle of the polygon which corresponds to Therefore , or so that the transformation becomes which is the formula given by SCHWARZ and CHRISTOFFEL .
PROBLEM SPECIAL CASE .
For the study of nolt-reentrant free stream lines we require a special case of this formula .
Suppose th , polygon to consist of a series of ) line sinfinite in on direction , all parallel to , so that the of the polygon are either or , and , therefore , or Let correspond to an angle , and , therefore , to an end of a line within a finite distance of the origin , and let .
correspond to an ange , and , therefore , to the adjacent ends of two lines at an infinite distance from the origin .
SCHWARZ'S formula then becomes It is plain that there cannot be more than one factor more in the numerator than in the denominator , so that we can write itnd , therefore , on integrating we may at once determine the tances between consecutive lines in terms of the or in terms of the For consider the passage of around the small semicircle above described .
We have that increases by in passing the point , the rJce between the parallel lines and is , or , in terms of the is After we have fixed on the angle which is to correspond to , we in general choose the position of two other points , and then the transformationlormula is determined to an additive constant .
For example , take the case of two doubly infinite lin es AB , , with a semiEF between them .
We may take the zero angle to be angles to be .
Then if the angle is , we have . .
J. H. MICHELL OF ] STREAM LINES .
Let the distance between AB and OF be and that veen OF and CD Then inteobrating past the points , we find determine A and , and , therefore , make the formula definite .
Other examples will occur in the physical application .
PROBLEM II .
The second transformation which we need may be stated most simply as an electrical problem .
Let there be any number of infini ely long plane conductors , all in the same plane , and with parallel edges .
It is required to find the potential at any point when these ductors are raised to given potentials .
A T$ AB , , . . .
be the sections of the conductors by the plane ( xy ) .
Everything is symme with regard to the line ABF which we take to be Consider the specifications of the transforrnation-function V where is the potential and are the lines of force .
We must have over the conductors , since they are , and .
therefore , There will be infinite points at the edges , B. of the nductors , and ) at points in the field corresponding to branch points of ( or ) .
These last will be distributed symmetrically with respect to From these conditions it is plain that the solution is that is the potential of masses at the singular points in question , so that we may write and , therefore , It remains to find the antities / For a singular point in the field where branches of meet we have simply .3 For an edge of the conductor we may proceed thus : Let then , so that as goes from goes from to and , therefore , the amplitude of increases by .
Now the amplitude goes from to , therefore , Hence ' where refers to a point in the field and to the edge of a conductor .
If one of the conductors reduce to a line , we have two of the equal , say , and there is a factor in the denominator , and so for any number of line conductors .
There are many other cases for which a formula like the above applies ; it is not always necessary that the conductors should be in the same plane .
It is very easy to perceive such cases by considering whether the equation is satisfied over the conductors where is an element of a normal to a conductor .
By combining this transformation with that of cqCHWABZ and CHRISTOFFEL we get a general solution for the free-stream-line problem , as I shall presently show .
It is necessary first to deduce some special formulae , which will be continually used hereafter .
( a ) Take first the case of one conductor and one line\mdash ; Let the conductor extend from to and the line distribution be at MR. J. H. MICHELL ON THE THEORY OF FREE Then the general formula reduces to where A is real , ) that there is no ularo point in the fieId .
The nature of the multiplier is obtained by that between and integral of this is supposing the potential of th conductor is zero .
( b ) Now let the conductor be two inite plaries , with a gap from to between them , and let there be a line dtstribution at a Then where A is real .
And the potential of the conductor being zero .
( c ) If we pub in the result of , and then make , we get for tho case of a semi-infinite conductor to with a line distribution at Ai the potential of the conductor being These results , ( b ) , , could of course be deduced from the known formulae for elliptical conductors .
: MR. J. H. ON TH THLORY OF STREAM , of Free The presence of sharp salient edges in a moving liquid always implies surfaces of discontinuity , but these may be closed or unclc ) according to circumstances .
It would be difficult to give a rule as to the kind of motion in a given case , for an alteration in the relative sizes of the solids concerned will totally alter the character of the motion .
As an illustration of this take the case of two parallel planes of finite breadth placed symmetrica ] lie one behind the other in a broad stream .
If the second plane be of less than a certain width , free stream lines will proceed from the edges of the first , the second will be in still water .
Now suppose the second plane oadene , until it cuts the stream lines from the first plane .
The character of the motion is changed .
Two vortices will appear behind the first plane , and in addition there will be free stream lines from the edge of the second piano .
No method has yet been discovered which will give solutions of cases where there is motion on both sides of a surface of discontinuity .
In the problems treated in the present paper there is always still water on one side of a free stream line .
In the present section the motion considered is in two dimensions , the boundaries are lane , and the free stream lines are non-reentrant .
Let be the coordinates of a point in the liquid , the potential and stream functions respectively .
The region in the plane corresponding to moving liquid in the plane will be bounded by straight lines , infinite in one direction at least and parallel to The area in the plane is therefore of the nature treated in Problem I. , that is , it is bounded by a polygon whose angles are alternately four right angles and zero .
This al.ea , then , by means of Problem I. , may be transformed into the part of a new plane in which is positive , where In this plane the boundaries of the liquid , both the plane boundaries and the free stream lines , are represented by the line Let , as before , We have seen that V is a potential function , considered as a fnnction of , and , therefore , it is also a potential function as a function of , for are conjugate with respect to \ldquo ; Further , we have seen that along a straight boundary constant we have , and , since all the straight boundaries correspond to portions of , we MR. J. H. ON OF AM must have along ] these portions .
Along a free stream line the pressure is constant , since it must be equal to the pressure on the liquid which rest .
Now , in steady motion , we have BERNOULL 1 pressure equatiou constant .
rherefole , along a free line is constant , and , therefore , is constant .
All the portions of which do not correspond to plane boundaries correspond bo free stream lines , for which V is constant .
Lastly is zero at angles of the dary , at here stream lines branch .
For these points V is finite a positive .
In all the cases we shall consider , these infinite points will be along , so that line ) branches at the boundary .
We have now reduced the problem to findirJg a function V which satisfies ; is fin ite and Iltinuous in that half of the plane for which is )ositive ; is constant along rbs of ; adong the other parts ishes d ; at points along it This problem has an obvious solution .
It is plain that V is merely the potential due to conductors coinciding with those parts of for which V is constant , and having that constant as potential , together ith masses at the points for wLich V is infinite .
The general solution of this has been given in Problem II .
Let be the conjugate of V , so that .
Then , ting Problem II .
into the present notation we have for a point mass two factors of the denominator coincide .
this , Now Cberefore MR. J. H. MICHELL ON THE THEORy OF STREAM LINES .
rular yessel of width has aperture in the botton ] , Here there are only two , each of which , and therefore Let the two angles be , where Then , observing that we get or Where , Between and , that is along the bottom of the vessel from the edge of the aperture ) the angle on the right , we have ; herefor the distance between the two points is MR. J. H. MICHELL ON OF LTNES .
similarly the other piece of the bottom of the vessel is of length When , that is , when Therefore the velocity in the vessel at a distance perture is and therefore breadth of the vessel is ( say ) .
The breadth of the aperture is and breadth of the jet is ultimately The question is now reduced to a matter of the integration of For the general case elliptic tegrals occur , and the expressions for our purpose may as well be left in their present .
If , however , the aperture be in the centre of the vessel , the integra]s will work out , and we get a simple expression for the contraction of thejet .
To this we now proceed .
Sub-Excuuple I.\mdash ; Jet from an } ) erture in the of the bottom of a rectangular vessel .
In this case we may take the of the vessel at The expression for then reduces to We now have Now and therefore WheIl c that is , at , we have Therefore the velocity in the vessel at a distance from the apertut.e is Therefore the breadth of the vessel is ( say ) .
So that the breadth of the aperture is Hence , since the final breadth of thejet is , we have as the ratio of the breadth of the jet to that of the aperture Now therefore and MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINES .
Also Thel.efore the contraction .
Now , getting rid of the special units , Jet be the breadth of the vessel as before , the breadth of the aperture , and that of the jet , then If be very large with get Lim .
This is the result obtained by from KIRCHHOFF'S solution for the of an aperture in an infinite plane bounding wall .
As decreases from infinity the contraction also continually decreases , until when the contraction is zero .
In order to get some idea of how soon finite breadth of the vessel affects the contraction ratio perceptibly , consider the case when or The contraction ratio is then and so that the finiteness of the vessel has very little effect oil the jet if the breadth is more than twice that of the aperture .
The equations to the free stream line * ' Phil. Mag Dec. , 18 The former gives Now where is the arc .
Therefore When we have which is the form given by AYLElGH ( ' Phil. Mag 1876 ) for KIRCHHOFF'S case .
Example II.\mdash ; Tube projecting far into the bottom of a vessel of given breadth .
The tube is supposed to project so far into the vessel that the motion at its bottom may be neglected .
here are here two singular points only , , corresponding to the coalescing gles of the bottom of the vessel .
The appropriate transformation is , therefore , A Since , when , , we have So that and as before ; NIR .
J. H. MlCHELh ON OF LINES .
erefore .
When , that is , at a distance upwards from the tube , .
erefore t breadth of the vessel is .
In passing over increases by , herefore the breadth of the tube is ( say ) .
fence , , herefore ubstituting in the vaJue of we have or On squaring this gives or , getting rid of the special unit of , if be the breadth of the jet , we have or , When is very great we get and the ratio of contraction is case was the first of free stream lines given , and Lord ' has the for the coefficient of contraction when there is a tube projecting inwards in a vessel of finite breadth .
The assumption made is , however , not that of this example , for he has taken the velocity .
the bott( of the sel to be the ) that at a distance upwards the tube .
This is one of the very few in the contraction can be determined accurately from elementary pl.inciples .
It is scarcely worth while to put the proof of this down , but it is remarking the corresponding case in three dimensions , can ] be worked out , viz. , the case of one circular projecting .
into Let be the radii of the cylinders , and the radius of the jet , then This includes the case first noticed , I lieve , by BORDA , of a cylindrical tube in an infinite no wall , the pro being just the same .
Ill.\mdash ; Tube proj e a sel of If we take the case of a tube projecting a vessel of great eadth , we do not get such simple expressions .
The transformation to the plane is the same as before , but there is now a singular point at , corresponding to .
We arrive at the correct result first supposing the vessel of finite breadth , so that there are points at for each of w'bich - , and making We have , therefore , * ' ' The Contracted Vein ' Phil. Mag Dec. 1876 . .
de Acad. .
I owe this reference to the kindness of Lord MR. J. H. MICHELL ON THE THEORy OF FREE STREAM Between and 1 we have , therefore the length of the pipe is Now , an elliptic integral which will be brought to the standard below .
Therefore , length of the pipe is , where .
From to have ( say ) , where is another elliptic integral ) be reduced Summing up , the breadth of jet is ; that of is , so that the contraction ratio is where the length of the pipe is To reduce the elliptic integrals , we have Now ( CAYLEY , ' Elliptic Functions , ' p. 315 ) where and Thus , where and F are the third and first elliptic integrals .
And , where is the second elliptic integral .
Suppose now the length of the pipe is small , so that is nearly unity .
We have If we put , we may put the factor in the integral equal to unity , and so get .
So that the length of pipe becomes to the first order ; MR. J. H. IIICHELL ON THE THEORY OF FREE STREAM while the aperture becomes to the first order , and the breadth ofjet is So that and , therefore , which is the approximate expression required .
Example .-Flow from an aperture in a pipe in which the water is at rest .
By considering the symmet17 of the motion this can be reduced to the case of flow from a rectangular vessel in which the aperture extends from the bottom up , and the bottom is continued into a horizontal plane , as in the figure .
The diagram consists of two doubly infinite straight lines , , as before .
But if in the plane we make correspond to , the arrangement of points is not the same as before and the notation must be altered .
We take , then , for th edge of the aperture , the right angle of the vessel , and then put for We then have and the arrangement of points on is as in the figure .
only nlar point for is at , and , therefore , from Problem II .
, Along the free stream line where lies between and If we have Therefore .
If we put in the first we get and , in the second , putting we have .
J. H. ON OF AM Hence when , i. e. , at the edge of the therefore the ratio of the .
to the aperture is The velocity at is therefore the width of the vessel is therefore From which we obtain .
or This is for the vessel considered .
For the pipe CASE II .
of a stream against ptane .
A stream of given breadth impinges at a given angle against a plane .
The impinging stream is bounded by the stream lines The stream line which branches at the point on the plane is The diagram in the plane consists of two infinite straight lines with a semiinfinite one between them , as in the figure .
In transforming to the plane we suppose that , corresponds to and that are the extremities of the plane .
We must then take , an unknown constant for the point , where the stream line divides , observe that We then have where therefore and Along we have , therefore , the following arrangement of points:\mdash ; free rigid rigid free MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINES .
The only singular point of is at Hence and , therefore , When , we have .
Therefore , if be the inclination of the stream to the plane , we have or Since we have before found this gives This equation merely expresses that the momentum parallel to the plane is unaltered by impact .
Along the stream lin we have therefore where lies between and ; and for we have therefore where lies between If we put in the first and in the second , we get for and for where in cases lies between .
When in both we get therefore , and the equations of the bounding stream li-nes are now completely determined .
CASE III.\mdash ; Flow of broad past plane in which there is an perture .
Let BC be the aperture in a plane wall ABCD , and let the stream flow from left to right .
The left boundary of the jet will be the continuation of strea AB .
The right boundary will be one branch of a stream which divides on the plane CD at some point E. MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINES .
The diagram in the plane will then consist of one infinite straight line and one semi-infinite , as in the figure .
We suppose to be the bounding stream lines .
In transforming to the plane we take the point to be to be , and then use the two arbitrary constants at our disposal by making the edges of the aperture to be , and the branch point of to be , where We then have and , remembering that the bounding stream lines are , we obtain so that There is only one singular point for , viz. , , and at this point , so that no factor appears in The arrangement of points on is as in the figure rigid free igid and hence so that When , we have and , therefore , the velocity of the stream is When , that is at a great distance .
the jet , so that the inclination of the jet to the wall is ultimately the final breadth of the jet being There appear to be two constants here , whereas there only to be one .
The explanation is that we have not yet expressed that the two parts of the boundary are in the same plane .
This will give a relation between and Between and we have Now and ' where ' Thus along , measuring from the edge , we have MR. J. H. MICHELL ON THE HEORY O In like manner between and on , we have , To find and we must put and respectively in the above , and then decrease indefinitely .
If we then renlember that , in passing the point increases by we get and ; In the case we are considering , and , therefore , or so that Using these equalities , we now get for the breadth of the aperture where is the velocity of the stream , the breadth of the jet , and the inclination of the jet to the plane .
To get rid of the special units let be the velocity of the stream , of the jet , the breadth of the aperture , that of the jet .
Then therefore So that and therefore Hence and the inclination of the jet to the plane is An interesting element not yet calculated is the distance of the branching of the stream line from the edge of the aperture .
This distance is where we have used the relation between and to simplify the expression , .
MR. J. H. MICHELL ON THE THEORY OF LINES .
Therefore , in terms of and , this distance is where For example , let the stream have half the velocity of the issuing jet , so that the jet makes an angle of with the plane , and its breadth is approximately , while is about CASE IV .
from pipe liquid is The liquid is flowing along a pipe bounded by the plane walls , DE , and there is an aperture AB in the former .
The left boundary of the jet is , the right boundary is a stream line which branches at a point on BC , and DE is The diagram is as in the figure , consisting of two infinite lines , with a semi-infinite line between them .
In transforming to the plane , we suppose that corresponds to , and that are the edges of the aperture .
The constants of the transformation are then determined , and we take for the branch point , for the jet at an infinite distance , for in the pipe .
Then where and therefore The arrangement of points on is as in the figure and from Problem II .
we have , so that Hence , in the pipe , the velocity at is , and at is The breadth of the pipe is , therefore , Now therefore .
This is the relation between , and .
To find the size of the aperture we must integrate between and .
Within bhese limits we have MR. J. H. ON THE THEORY OF STREAM LINES .
The former gives between and and between and Also in passing the point changes by ; therefore , putting in in , and proceeding to the limit , we get as the breadth of the aperture , The direction of the jet makes an angle with the bounding planes .
The breadth of the is To sum up : Let be the breadth of the pipe , of the aperture , of the jet .
Let , further , be the velocity at that at that of the ( 1 ) ( 2 ) and therefore ( 3 ) ( 4 ) ( 5 ) with MR. J. H. MICHELL ON THEORy OF FREE STREAM LINES .
The equation ( 6 ) may be written .
Now , therefore ; and , by addition and subtraction , , .
From ( 1 ) Therefore or , ; so that And Also from ( 2 ) , therefore MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINES .
so that and These may be reduced to the form .
, Further , Whence , finally , So that we get for An interesting special case is the flow from an aperture in the side of a rectangular vessel , of which the bottom is at a considerable distance from the aperture .
For this case , and , therefore , ' expressing the equality of inflow and outflow .
So that or which is the simplest expression for this case .
The angle the jet makes with the bounding wall is in the general case , while in the particular case it is simply MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINES .
We will now go on to consider problems in which the region of is not simply connected , and consequently SCHWARZ'S transformation does not apply .
First consider the area outside a closed polygon .
We may state the problem electrically thus : PROBLEM III .
To find the potential due to a polygonal prismatic conductor at a given potential which we may take to be zero , the field at infinity being at an infinite potential .
Let be the potential , the lines of force , and let increase by in going round the polygon .
Then the area in the plane is a rectangle bounded by The conditions which the transformation function V satisfy are over ( b ) finite and continuous at all points within a nnnite distance in the rectangle .
( c ) V periodic in so that V .
( d ) V infinite at points along We can determine V from these specifications by means of W. THOMSON'S method of images .
For if we repeat the points at equal distances along , and make V the potential of these points , the conditions will clearly be satisfied .
Hence and MR. J. H. MICHELL ON THE OF STREAM LINEiS .
and exactly as in the treatment of SCHWARZ'S formula we have , where is the exterior ngle of the polygon corresponding to So that , finally As an example , take th case of a rectangle .
The four singular points may be taken to be and we } ) taking for simplicity .
Hence and is an elliptic integral of PROBLEM IV .
Suppose now there are two polygonal pl.ismatic conductors , one at potential , the other at , and at first ( A ) suppose that one of the conductors is within the other .
Let increase by in going round either polygon .
The area in the plane is now a finite rectangle bounded by In order to satisfy the condition over and we must have a double system of images of the singular points , viz. , at .
BRILL has already given this formula ( ' Messenger of Math August , 1889 ) , and I only insert it here for the sake of completeness .
NOTE.\mdash ; Apri129 .
Since the above paper was read , Mr. BBILL has given the next transformation in the same journal .
MR. J. H. MICHELL ON THE THEORy OF FREE STREAM LINES .
Thus each singular point gives a factor or , what is the same thing , where are JACOBI'S functions so indicated , and , therefore , where being the complete elliptic integrals usually so denoted .
If be the internal angle of the figure corresponding to , we have , as before , so that , finally , ( B ) Suppose now one conductor is outside the other , and that the potential at infinity is zero , that of the conductors being - .
We suppose equal and opposite quantities of electricity on the conductors , being the cyclic constant , as before .
The terms corresponding to angles of the polygon will be the same as in ( A ) .
But there is now in addition a singular point in the field which we proceed to determine .
At a great distance from the prisms the potential will be the same as for two lin distributions at the centres of mass , say at So that ultimate]y , and or and , therefore , there is a point of order at the point in the rectangle corresponding to the potential at infinity .
Let this point be .
Then we have as the general expression for the potential of two polygonal prismatic conductors exterior to each other .
Hollow Vortices .
In conclusion I shall show how the methods of this paper may be applied to find the form of hollow vortices .
Inside a vessel bounded by plane walls , let there be a hollow vortex in steady motion .
Let be the free steam line of the vortex , the rigid boundary .
In the plane the area is bounded by being the circulation the vortex .
The function V satisfies the following conditions:\mdash ; over , if the velocity along the free steam line be unity .
over is periodic with respect to , so that V .
These conditions are to be satisfied by taking equal singular points at distances along , and then continually reflecting these points in the two planes , but in reflecting in the image is of opposite sign to the object .
Corresponding , then , to a point at we have positive images at and negative images at Therefore , corresponding to this point , we have a factor in where MR. J. H. MICHELL ON THE THEORY OF FREE STREAM LINES .
It is quite clear that all but the simplest cases will be of quite unmanageable complexity .
One of the very simplest cases may be taken as an example .
Example.\mdash ; Hollow vortex between two parallel planes .
Here the two singular points If for simplicity we put , and , therefore , , the singular points are at Hence This integrates at once , giving tn The equation to the vortex is , therefore , tn

