]\gt ; Prof. W. B. Morton .
necessarily connict with the fundamental assumptions of H. A. Lorentz and Larmor 's theory .
They do , however , make it probable that the application of this theory to magnetic bodies has not yet been worked out in a satisfactory manner .
added , 1913.\mdash ; The specific in capacity and meability of any material medium are values over volumes large compared with the structural units ( molecules or bodies ) up the medium .
The medium employed has definite values of these quantities for volumes large compared with the olume of one of the spheres used in building it up .
It appears , therefore , to be allowable to apply any theoretical results expressed in terms of and to the medium used .
] Our thanks are due to the Government Grant Committee of the Society for a grant with a large part of the apparatus used in this investigation was originally purchased , and also to the Trustees of the Rice Institute for the facilities for experimental work which they have placed at our disposal .
On the cements of the rticles their Paths in some Ca.ses of Two-dimensional Motion of Frictionless Liquid .
By W. B. , Professor of Physics , Queen 's University of Belfast .
( Communicated by Sir J. Larmor , Sec. R.S. Received June 2 , \mdash ; Read June 19 , 1913 .
) The paths described by the individual particles of a liquid have been inyestigated only in a few cases , those in which the motion is steady , so that the particles ] the stream-lines .
Clerk Maxwell , , pnblished for the paths in an unbounded liquid disturbed by the passage of a circular cylinder .
The curves for particles in contact with the cylinder were plotted by calculation ; the other paths were drawn by eye from a knowledge of their terminal points and curvature .
From these curves were derived others , showing the successive in the deformation of a row of particles which , before the approach of the cylinder , lay in a straight line perpendicular to its motion .
In 1885 , Lord Kelvinl- investigated the paths of particles of a liquid ' Lond. Math. Soc. Proc vol. 3 ; 'Collected Papers , ' vol. 2 , p. 208 .
'Collected Papers , ' vol. 4 , p. 193 .
Two-dimensional ' Motion of Liquid .
enclosed in a rotating ellipsoidal shell .
He showed that they moved , relatively to the shell , along a set of similar ellipses in parallel planes , the period of this motion being the same for all the particles , so that after this period the configuration comes back to the initial one rotated through an Later ( 1889 ) , Riecke*gave diagrams for the cases of translation of a sphele and of a vortex-pair .
His method of procedure was the reverse of Maxwell 's .
He began with a straight row of particles , and , by giving them displacements proportional to the instantaneous values of their velocities , he obtained the approximate forms of the curves on which the particles lay from second to second .
From these the curves described by individual particles were deduced .
ReCently Havelock{ has discussed some points of difficulty in the case of translation of a circular cylinder , specially with regard to the conditions at infinity .
He has explained the uniform displacement forwards of the mass of the liquid .
From a different point of view , that of problem of the same nature has been ated by W. N. Shaw and in tracing the paths of individual particles of air in the course of actual atmospheric movements , supposed to be practically laminar , while the same problem , for vertical movement of the air , has been more recently treated by .
Bjerknes .
In the present paper I have rated the equations of motion and plotted curves for the paths of particles in the well-known simple cases of twodimensional motion , , for liquid contained in a rotating elliptic cylindel .
and in a equilateral ular prism , and for liquid to infinity and disturbed by the translation or rotation of an elliptic cylinder .
The use of the word " " particle\ldquo ; in this connection is convenient , but nires some explanation .
What is really ated is the motion of a mathematical point which moves at each instant with the velocity to its position in tlJe liquid .
If at any a small sphere be put round this point and the motion of points on its surface be followed in the same way , it will often be found that the volume is continuously as time goes on , being pulled out indefinitely in one direction .
This happens , for example , in the simple case of irrotational circulation in a " " free vortex It occurs in liquid contained in a .
triangular and in liquid outside a elliptic cylinder .
For liquid inside 'Wied .
Ann vol. 36 , p. 322 .
'Univ .
Durham Phil. Soc. Proc 1911 , vol. 4 .
' Life-History of Surface Air-Curreuts , ' Publications of Meteorol .
Office , 1906 , No. 174 , 1906 .
VOL. LXXXIX.\mdash ; A. Prof. W. B. Morton .
an elliptic cylinder , on the other hand , or more generally in Kelvin 's case of an ellipsoidal shell , the ' particles\ldquo ; are not continuously deformed , but come back periodically to their initial shapes .
In this case the term particle can fitly be applied to a volume element throughout the motion ; it is obviously inappropriate to the pulled-out filament .
This feature of the motion brings out the limitations inhelent in the conception of the liquid as a continuum in the hydrodynamical treatment , and enforces our instinctive ideas of a molecular structure in actual liquids .
This is pointed out by Havelock in connection with the deformation of volume ] ements in the immediate neighbourhood of a cylinder .
The same point is illustrated in a still more striking way by the occurrence in many cases of points on the solid boundary where the relative velocity of liquid to boundary is zero .
The " " particle\ldquo ; at this point moves with the boundary , particles on either side approach it or recede from it continuously , remaining in contact with the wall .
When there is approach from both sides toward this limiting position , the volume elements bounded on one side by the wall become extended indefinitely into filaments in the direction perpendicular to the wall .
General of Results .
( i ) Elliptic Rotating , Liquid \mdash ; Particles on the surface slide backwards along it , so that their excentric angles decrease by amounts proportional to the angle of rotation the cylinder .
Particles in the interior do the same thing similar ellipses , so that the same of particles constantly forms a radius vector of the ellipse .
The paths in space are thus a family of similar looped curves ( fig. 1 ) .
The apsidal angle and the width of the loops are found , and it is shown that consecutive loops intersect for ellipses less excentric than the form .
( ii ) Equilateral Triangular Prism Rotating , Liquid Inside.\mdash ; Particles on the walls move backward from one angular point to the other , and do not pass the corners .
Their path has a single loop and has the circum-circle of the triangle as asymptote .
Particles in the interiol describe looped paths , the loops becoming closer and smallel towards the centre .
Apsidal and width of loops are obtained ( fig. 4 ) .
( iii ) Elliptic linder with Motion of mslation ough Liquid at Rest at Infinity.\mdash ; The two particles on the surface which are carried in one piece with the cylinder , are those situated where the excentric angle is equal to the angle between the direction of motion and the major axis , and at the diametrically opposite point .
These are limiting positions between which the surface particles move in opposite directions round the two sides Two-dimensional Motion of Frictionless Liquid .
of the cylinder .
The paths of these surface particles in space are looped curves extending from to .
Between the limiting points on each side of the ellipse there are two points at which the velocity of the particle in space has no component in the direction of motion of the cylinder , and one point at which the perpendicular component vanishes .
Simple geometrical relations are found to connect these " " stationary points\ldquo ; and the limiting points on the surface .
For particles not in contact with the cylinder the paths are looped at the initial position occupied by the particle before the approach of the cylinder and the final position where it is left when the cylinder has passed off to infinite distance .
The line these end points is always parallel to the direction of motion of the cylinder and decreases rapidly in length at increasing distance from the line of motion .
The direction of the path at its ends makes with the direction of motion of the cylinder an angle which is the same for all particles , and depends on the shape of the ellipse as well as on its direction of motion .
The angle vanishes only for translation along a principal axis .
In this case it is shown that for a given breadth of cylinder , and for particles lying beyond a certain distance from the central line , there is a definite shape of ellipse which is most effective in displacing the partic ] forward .
Diagrams are given for direct motion of four different forms of cylinder ( figs. 5-8 ) and for oblique motion of a plane lamina ( fig. 9 ) .
Also , for the deformation of a row of particles originally on a straight line ( figs. 10-14 ) .
( iv ) Rotating Elliptr.uder , Liquid Outside.\mdash ; As regards the configuration of the stream-lines relative to the ellipse , it is shown that there are two different cases according as the ellipse is more or less excentric than the form .
For the less excentric forms the relative stream-lines completely surround the ellipse in all cases .
In the other class there is a mound of liquid carried round by the cylinder at each end of the minor axis ( fig. 15 Particles inside these regions describe closed paths relative to the cylinder and never escape from the neighbourhood of the minor axis .
There is one particle in the centre of each region which moves in one piece with the cylinder , and four particles on the surface of the cylinder , at the boundaries of the " " mounds which do the same thing .
These four points are limiting points at which the direction of motion of particles .
along the surface is reyersed , and the circle passing through them is approached asymptotically by threedifl.erent curves in space , viz. , the paths of particles coming in opposite directions along the surface together with the path to the outer boundary of the .
mound The other paths in this case , and all the paths in the less excentric case , show recurring loops .
rams are given for one ellipse of the less excentric kind , an1 Prof. W. B. Morton .
for the plane lamina ( fig. 17 ) .
Expressions are found for the apsidal angle of the paths which are examined in detail for the plane lamina ( fig. 18 ) .
It is shown that for certain paths in the latter case there are three different apsidal distances .
Method of Calculation .
If is the stream function for the motion of the liquid determined by the motion of a cylinder then the stream function for the motion relative to the cylinder is The curves const .
are the relative stream-lines .
The paths of the particles in space are got by combining a motion along these with the motion of the curves themselves as they are carried by the moving cylinder .
Tbus it is necessary to connect the position of a particle on its relative stream-line with the time or with the displacement of the cylinder from an initial position .
Let be expressed in terms of the appropriate orthogonal co-ordinates with length-elements , and let be chosen to fix the position on a stream-line .
The expression for the -component of velocity gives the equation When is removed from the right-hand side by aid of the equation const .
, the integral can , in the cases considered , be expressed by elliptic functions .
will contain or as a factor .
It is convenient to replace as independent variable by or The paths in space are got by to each point on a stream line the corresponding linear displacement , or displacement Liquid , Rotating Elliptic Cylinder.\mdash ; Here .
The relative stream-lines are similar ellipses .
Using the excentric angle to fix position on one of them we find .
Thus in miform rotation of the cylinder the particles slip back along their ellipses so that their excentric decrease at a uniform rate .
This is a particular case of Kelvin 's result for the ellipsoid .
When the cylinder has turned through the angle the particles are back in their original imensional Motion of Frictionless Liquid .
relative positions .
Each element of liquid has received a body rotation through this angle in consequence of a motion which is differentially irrotational at each instant .
Particles which at any instant lie on a radius of the ellipse continue to do so throughout the motion .
Thus , it is easy to follow the progressive deformation and restoration of form of a small volume-element bounded by two radii and two elliptic arcs .
FIG. 1 .
FIG. 2 .
shows the path of on the boundary of the ellipse in which .
For particles inside , the paths are similar curves .
apses of the path , of course , correspond to the ends of the axes of the ellipse ; the motion is " " direct\ldquo ; across the end of the major axis , " " retrograde\ldquo ; the minor axis .
The stationary points occur when the particle is on the radius vector bisecting the between the axes .
The following results are easily obbained : apsidal angle of path is , so that the mean ular motion of the particles in space bears to the motion of the cylinder the ratio The angular distance between the stationary points , i.e. ular width of the loops on the path , is In the apsidal and half-width of loop are plotted for different values of the axis ratio .
The curves intersect near the value showing that for an ellipse of this shape successive loops are in contact .
As the straight line limit is approached the width of a loop approaches a value of about , while the apsidal angle becomes infinite .
( ii)Iiquid Contained in Rotati , Here where the side of the triangle is The equation of the family of relative stream-lines For the curve breaks up into the sides of the Prof. W. B. Morton .
Using as the coordinate-fixing position on the stream-line , we have giving Let be the roots of the cubic in , then the solution of this equation is and are the apsidal distances on the stream-line , being the radius drawn in direction towards an angular point of the triangle , that drawn towards the middle point of a side .
The third root lies outside the limits of the triangle .
The expression for vanishes when .
It is convenient to identify a stream-line by the distance For particles in contact with the wall , then , the elliptic functions become hyperbolic and the inteo r takes the form , where is the distance from the middle point of a side .
For , i.e. the particles move backwards along the wall between one angular point and another , and do not pass the corners .
The path of a particle in space then approaches asymptotically to the circle of the triangle . .
FIG. 4 .
In fig. 3 the curves are drawn for , 1 .
A section of the relative stream-line is shown in each case , continued by the path in space .
The apsidal angle of the path is found to be The mean ular motion of the particle bears to tlJe motion of the prism the ratio of this angle to Two'- Motion of a rictiontess Liquid .
The stationary points are determined by , which leads to , etc. So here , as in the last case , the radius to the stationary position bisects the angle between the apsidal distances .
The ular width of the loops as seen from the centre is Fig. 4 shows how the apsidal angle and half-width of loop vary for streamlines at different distances from the centre .
It appears that successive loops come into contact when is about ( iii ) Translation of Elliptic linder.\mdash ; Let the velocity make angle with the major axis , then , in terms of the usual elliptic co-ordinates , , where on the surface of the cylinder .
Thus .
For the present purpose it is enient to replace the co-ordinate ) , say , and to write the equation of the family of stream-lines in the Then is the argument and the modular angle of the elliptic functions which occur in ) integral of the equation of motion .
The forms of these stream-lines are given in the text-books .
It is easy to show that the line whose parameter is , at points very distant from the ellipse , runs parallel to the line of motion of the centre at a distance from this line of : Particles which follow this particular stream-line will therefore have this for their initial and final distance from the line of motion of the centre .
The passage of the cylinder results in cases in a displacement of particles in the direction of its motion .
* The motion the relative stream-line is from when the cylinder is at infinite It seems clear that this will be the case no matter what is the form of the cylinder .
The statement is equivalent to saying that in the steady flow of a liquid past a cylindrical obstacle each straight stream-line , fter swerving round the obstacle , comes back to the prolongation of the same straight line .
This appears physically obvious if we think of two of the stream-lines so far from the obstacle on opposite sides as to be practically straight throughout their length .
The space between them , at great distances on the two sides of the obstacle , is filled across with straight intermediate lines of flow , so each of these inep , after deflection by the obstacle , must resume its former position .
The same argument does not apply to in three dimensions .
If we consider a tube of lines of flow so far removed from the obstacle on all sides as to be straight , and take sections of it at great distances on both sides of the obstacle , the same bundle of tubes of flow comes out as goes in ; but there is now the possibility of the tubes having been twisted together in passing the obstacle , e.g. , if it possessed a helical structure .
Prof W. B. Morton .
distance behind to when it has passed off to infinite distance in front .
The equation of motion in the original co-ordinate is , where is the distance the cylinder has moved from some fixed position .
Having obtained from the integral of this as a function of , the co-ordinates of successive positions occupied in space by the particle are found by applying the displacement to the corresponding points on the stream-line .
Let the origin be taken at the undisturbed initial position of the particle ; let the axis of X be perpendicular to the direction of motion of the cylinder , and the axis of parallel to it .
Then ( X , Y ) , the sideways and forward displacements of the particle , are found to be given by , .
The special stream-line includes the surface of the cylinder and the hyperbola .
The point on the ellipse and the diametrically opposite point are limiting points for the motion of part-cles the surface , i.e. points of zero relative velocity , as explained above .
The paths of surface particles have , as asymptotes in opposite directions , the lines described by these points on the ellipse .
The expressions for the co-ordinates X , become in this case , the igin now being at the poin corresponding to For the hyperbolic stream-line the co-ordinate must be used .
The values are cos2 which vanish for , and therefore give displacements from the initial position .
The following consequences can be deduced from the formulae:\mdash ; ( 1 ) The distance between the initial and final positions of a particle is got by putting in , which gives Two-dimensional Motion of Frictionless Liquid .
( 2 ) The form of the paths at a great distance from the line of motion .
When is small , X , , are small quantities of the same order , Elimination of gives the equation The paths , therefore , approximate to small circles , radius ( 3 ) Initial direction of motion of a particle .
On the expressions for X , , and putting , we get , which gives the tangent of the angle between the direction of motion of the cylinder and the direction in which a particle begins to move .
It is the same for all the particles , and vanishes only when motion is along one of the principal axes .
The greatest value is when .
For the case of a plane lamina , the in question is the complement of .
So , if the lamina is moving at to its plane , the direction of initial motion is parallel to the plane of the lamina .
This is shown in fig. 9 .
( 4 ) The critical points on the surface of the cylinder .
These are\mdash ; The lin]iting points , and , where the stream-lines diyide ; The point , where the -component of the velocity vanishes and the particle has its greatest sideways displacement ; The points , on opposite sides of , where -velocity vanishes .
The equation with ives and gives as roots of so that tau These equations show that and are pairs of conjugate points in the sense of Fagnano 's theorem on the ification of elliptic arcs .
They have the same length of intercept on the tangent between the point of contact and the foot of the central perpendicular .
Further , it is easy to show that if be the length of this intercept for the pair and for , the relation exists between these , i.e. the sum of the squares of the interoepts is equal to Prof. W. B. Morton .
the square of the maximum intercept at " " Fagnano 's point The symmetry of this shows that the of the two pairs of points can be interchanged .
( 5 ) Influence of the shape of the ellipse on the displacements of a particle , the motion along a principal axis .
Let the motion be along , keep constant , and let increase , beginning with , for the transverse motion of a plane lamina .
If undisturbed distance of a particle from the central hne of motion , maximum displacement sideways , reached when the particle is on the -axis , final displacement in direction of motion , then , in terms of the parameter , .
The first and second equations show that , as is increased , the same value of is found at distances from the central line proportional to .
So there is a continual increase of the sideways displacement of particles at given distance .
For the forward displacement we have If the function of bc plotted , it shows a maximum value about , and closer calculation places this at .
Therefore , to obtain the greatest forward displacement at distance we must have This gives a positive value for if .
For particles closer to the line , the plane disc is the most effective form for causing forward displacement .
Description of Diagrams.\mdash ; In fig. 5 I have drawn the paths of particles in liquid disturbed by the passage of a plane lamina .
The motion must be supposed very slow and the edges of the lamina rounded , so as to avoid the occurrence of discontinuity .
The direction of motion is perpendicular to its plane ; the line with the arrow shows the path of the edge , the dotted line at the bottom gives the half-breadth of the lamina .
The particles which are originally along a line parallel to the plane of the lamina have positions relative to it defined by the values of , so that , if the half-breadth be taken as unity , their initial and final distances from the line of motion of the centre are given by the ents of these angles , say , , and Two-dimensional Motion of Frictionless Liquid .
For comparison wit , this the following figures show the paths of -he same particles when the plane lamina is replaced by an elliptic cylinder FIG. 7 .
FIG. 8 .
of the same breadth , with axis in the direction of motion of , and respectively , as shown by the dotted quadrants .
The values of which would give maximum forward displacement to the five particles are , and .
In agreement with it will be seen Prof. W. B. Morton .
the first and second palticles are displaced farthest in fig. 5 , the third iu fig. 7 , and the fourth and fifth in fig. 8 .
For particles in contact with the cylinder the paths extend to infinity in both directions , and have loops of the same type as the other curves shown .
In the case of the plane lamina the loop contracts to a cusp .
It did not FIG. 9 .
Fig. 9 shows the paths for a plane lamina moving at with its plane .
I have drawn a path , lying to the left of the central one , to show the symmetrical relation between corresponding paths on opposite sides of the line of motion of the centre of the lamina .
It will be seen that one is got from the other by a double reflection .
It has already been pointed out that this is the only case in which the directions of the paths at their ends are parallel to the moving plane .
FIG. 10 .
Two-dimensional Motion of a Frictionless Liquid .
Figs. 10 , 11 , 12 , show the successive forms assumed by a row of particles which , before the approach of the cylinder , lie on a line perpendicular to the direction of its motion .
This initial position of the row is shown as a FIG. 11 .
Prof. W. B. Morton .
datum level in each figure ; the position of the moving solid is shown dotted , the final configuration when the cylinder has moved off to infinite distance is the curve drawn through the upper ends of the paths shown on figs. 5 , 6 , and 9 .
and 14 give the forms of a row of particles lying originally on lines parallel to the motion of a plane lamina .
The paths by which the FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 13 .
FIG. 14 .
particles have moved from their initial to their present positions are shown by dotted curves .
Diagrams like these can be constlucted by sliding a drawing of the relative stl'eam-lines over another , iving the paths in space .
This was the plan used by Maxwell .
It has the practical drawback that the required points are given as the intersections of curves which are often nearly parallel to each other , so that a error in the setting , or in the forms of the curves , causes a large displacement of the intersections .
I found it better to plot separately the co-ordinates X and against , the distance travelled by the cylinder , and then to lead off from these curves pairs of values for each ( iv ) Outsid a Elliptic lin The relatiye stream-lines const .
show peculiarities which I have not seen any reference to , although the above expressions are yely ] known .
The configuration of these curves is quite different for ellipses more excentric and less excentric than the critical form defined by , or Writing the equation of family as dimensional Motion of Frictionless Liquid .
the value gives the surface of the ellipse .
In the less excentric class this is the whole of the stream-line in question , but for the other class there is another part .
Solving the equation as a quadratic in with or The first value corresponds to the surface of the cylinder .
The second is also admissible , provided the point defined by it lies outside the cylinder , .
if or -cos2 If sinh2 , this gives a range of values of extending downwards from the value on both sides .
So here is an additional piece of the streamline over the end of the minor axis and meeting the ellipse at the points defined by .
These are evidently points of no relative motion .
The space between this additional piece of the stream-line and the end of the minor axis is filled by closed stream-lines to smaller values of A. These contract to a point ; the position of this point on the axis is given by .
Here also the particle of liquid remains in a fixed position relative to the rotating ellipse .
All the liquid in the region occupied by the closed stream-lines is carried round with the cylinder , and is kept distinct from the of the liquid .
Stream-lines for which encircle the ellipse .
For the less excentric forms all the lines of this type .
The arrangement of stream-lines above described is shown on fig. 15 for .
Also , for the limiting form of the plane lamina , in one quadrant of fig. 17 .
The arrows show the direction of motion , as seen by an observer moving with the cylinder .
The less complicated forms for the less excentric class are shown on fig. 16 for Prof. W. B. Morton .
To find the paths in space the co-ordinate is again used .
Its connection with , the of rotation of the cylinder , is given by , where is written for .
The reduction of the integral , when the right-hand side has been expressed as a function of , is rather tedious .
The result is In this A is the stream-line constant , as above .
, , with .
These formulae hold when .
When the of the radical is changed in the definition of the constant .
When ( in the more excentric class ) becomes greater than unity and Che reciprocal modulus is used .
On the surface of the cylinder reduces to This for the class .
For replaces A good deal of labour is needed in order to numerical values from the general formula .
* The elliptic functions were taken from 's tables .
I took the amplitudes of modular angles to the nearest minute so that double interpolation was necessary .
The functions were expressed by functions and these were calculated from the expansions in ascending powers of " " The values of were got from the tables of Jahnke and Enlde .
Apses on the Paths.\mdash ; When the particles right round the cylinder the apses are on the major and minor axes .
The apsidal in space comes out The particles which keep near the end of the minor axis in the more excentric class , have apses in the two positions where they cross this axis .
The apsidal angle is got from the expression just given by omitting the I am indebted to Miss Mary Beck for much help in the a1ithmetical work in connection with this case and the preceding one .
Motion of a Frictionless Liquid .
subtracted and replacing the denominator of the first term inside the bracket by Fig. 18 shows the variation of the apsidal angle for different values of the stream-line parameter , in the case of a plane lamina .
For a certain range of the closed stream-lines there is a third apse intermediate between the two which correspond to the transits of the particle across the minor axis .
This is where the particle is at maximum distance from the centre .
We have .
Differentiating with respect to and taking account of the equation of the stream-line which vanishes for .
For the maximum occurs at the point on the cylinder where the external part of the stream-line meets it .
As A decreases the position of maximum moves up the outer part of the stream-line until , for certain value of , it coincides with the more distant point on the axis .
This happens when the value of iven a is equal to For the plane lamina this gives .
So the third apse is found on the paths between and .
From to , when the stleam-line shrinks to a point , there are only the two apses corresponding to the positions on the minor axis .
FIG. 16 .
FIG. 17 .
Description of Diagrams.\mdash ; Fig. 16 shows the forms of the paths for an ellipse of the less excentric class .
As in preceding diagrams the first quadrant is occupied by the relative stream-lines with the VOL. LXXXIX.\mdash ; A. Two-dimensional Motion of Frictionless Liquid .
surface of the cylinder itself .
The two external lines are and .
These stream-lines are continued into the paths in space .
In fig. 17 the plane lamina is taken as the extreme case of the more excentric class .
Here , and the surface of the lamina with the external FIG. 18 .
loop corresponds to .
The other stream-lines drawn are for the values .
The paths are again shown as continuations of the relative stream-lines , except in the case of the outer portion of the lamina itself between the critical point and the edge .
The particles in contact with this move round the edge from the critical point on one side to that on the other side , the motion becoming infinitely slow near the limiting points .
The path has a cusp at the edge of the lamina and approaches the asymptotic circle from the outside .
It has been drawn , for the sake of clearness , from its cusp at the left-hand side of the diagram .
It will be seen that the same asymptotic is approached from the inside by the path corresponding to the external loop and also by the particles which slide along the central part of the lamina between one critical position and the other .

