]\gt ; INDEX SLIP .
HOUGII , S. S.\mdash ; On the Application of Harmonic Analysis to the Dynamical Theory of the Tidos.\mdash ; Part .
I. On Laplace 's " " Oscillations of the First Species and on the Dynamics of Ocean Currents .
Phil. Trans. , , 1897 , vol. 189 , pp. Harmonic Analysis applied to Tidal Theory .
Hough , S. 8 .
Phil. rans .
, 1897 , vol. 189 , pp. Ocean , Free Oscillations of the .
Hough , S. S. .
Trans. , , 1897 , vol. 189 , pp. 201-257 .
Ocean Currents , Dynamics of , as affected by the Rotation of the arth .
Hough , S. 8 .
Phil. Trans. , , 1897 , vol. 189 , pp. ) Tides , Dynamical Theory of ; Laplaoe 's Theory of .
Hough , S. S. Phil. Trans. .
) , vol. 189 , pp. 201-257 .
- 201 IX .
On the Application of Harmon Analysis to the of Tides.\mdash ; Part I. On " " lations of the First Species and Dynamics of Currents .
S. S. HOWGH , of St. John 's College , and Student the University of mbridge .
Communicated by Professor G. H. DARWIN , F.R. Received March \mdash ; Read Apri18 , 1897 .
THE earliest attempt to subject the Theory of the Tides to a rigorous dynamical treatment was given by LAPLACE in the first and fourth books of the ' Ce'leste .
' The sub.iect has since been treated by AIRY , *KELVIN DABWIN , LAMB , S and other writers , but with the exception of the extension of LAPLACE'S results to include the theory of the long-period tides , but little practical advance has been made with the subject , in spite of the enormous increase in the power of the mathemat , ical resources at our disposal , and the problem has remained in very much the condition as it was left by LAPLACE .
This arises no doubt partly from the difficulties inherent to the subject , but partly from the form in which the theory as origioallvr presented by LAPLACE in the ' Mecanique Celeste , ' which has been described by AIRY as " " perhaps on the whole more obscure than any other part of the same extent in that work The obscurity complained of does not however to have been removed by LAPLACE'S successors , and it was the fact that every presentment of the theory with which I acquain ted some points of difficulty , that in the first instance led me to take up the problem ab , partly with the pose of allaying the doubts which had in my own mind as to the validity of certain approximations ernployed by LAPLACE and adopted by his successors , and partly in the hope that I might be able to extend the results of LAPLACE to meet more tnlly the case presented by the circumstances actually existent in Nature .
Up to the present I have been unable to free the problem to any extent from the limitations which have been imposed by previous writers , and consequently would be futile to claim that the snlts I am now able to put forward materially advance ncyc .
Metropolitana ' ; Art .
, " " Tides and Waves Section III .
'Phil .
, vol. 50 .
'Encyc .
Britannica ' ( 9th edition ; Art .
, " " Tides S Hydrodynamics , ' chapter viii . .
@ .
S. HOTiGH ON THE APPLICATION OF HARMONTC our knowledge of the tides as they actually exisb ; but I venture to hope that these results , as applied to the oscillations of an ideal ocean .
considerably simpler in character than the actual ocean , may prove of some interest from the point of view of pure hydrodynamical theory .
In SS1-4 I have devoted co1lsiderable space to the formation of the dynamical equations .
The equations obtained agree with those used by LAPLACE , and con sequently ib may be thought that I have .
been unnecessarily diffuse over this part of the subject .
My apology is that the questionable , if not erroneous , reasoning which has often been assigned for the various approximatiors introduced seemed to me to warrant a very minute examination of the formation of these equations .
An analytical treabment , such as that I have used , seems to me to be the only safe method of procedure to ensure that the approximations do not involve the neglect of any terms which may be of equal importance with those retained , many of which are extremely small .
The method adopted follows LAPLACE in so far as it consists of a transformation of the general equations of oscillation of a rotating fluid .
I trust however that these general equations in the form I have used , which seems to be the simplest form to which they can be reduced , may be found less " " repulsive\ldquo ; than those employed by LAPLACE .
S5 deals with the integration of these equations .
The forms of solution discussed in the present paper are those which are symmetrical with respect to the axis of rotation .
The types of oscillation represented by these solutions have been named by LAPLACE , " " Oscillations of the .
First Species but he omitted to discuss them in detail on the grounds that the oscillations of such character , which might be expected to exist in Nature , would be modified to such an extent by friction that they would be far better represented by the old ' equilibrium theory than by a dynamical theory which failed to take due account of the action of friction .
The tides in question will be of long period , the shortest of the periods half a lunar month in duration , but Professor DARWIN was , I believe , the first to caII attention to the fact that this length of period will hardly be sufficiently great to render the of friction of such paramount importance , and hence he added to the work of LAPLACE a discussion of the long-period tides when not subject , to frictional influences .
I have recently attempted to estimate the effects of ffiction on the tidal oscillations of the ocean , the results at which I have arrived fully confirm the view of Professor DARWIN .
as to the small influence of friction on the lunarfortnightly tides , and render it highly probable that the effects will be almost equally slight on the solar long-period tides .
The method of integration I have employed differs from that used by DARWIK , my aim having been to express the results by means of series of zonal harmonics instead of by the power-series obtained by him .
The advantages of this are two-fold ; firstly , * In a paper read before the London bfath .
Soo .
, Deoember , 1896 .
ANALYSIS TO THE DYNAMICAL THEORY THE TIDES .
it allows of our including in our analysis the effects dne to the gravitationd attracof the water ; and secondly , the convergence of the series obtained will be much more rapid .
The latter circumstance is of particuIar value , as it has enabled me to treat with considerable success the of the free tions of the ocean .
The general problem of the small oscillations of a rotating system possessing a finite number of degrees of freedom has been discussed b.y THOMSON and TAIT ; the extension to meet the case where the number of degrees of freedom is infinite involves analytical considerations of some delicacy .
As a rule , the transition from the case of a system with finite freedom to that of a system with infinite freedom is effected by the employment of " " norlnal coordinates , \ldquo ; and the chief difficulty in the solution of problems relating to the vibrations of the latter class of system consists in the discovery of these coordinates .
The researches of TBOMSON and TAIT just mentioned show however that in a rotating system these normal coordinates do not exist , and hence that the methods arily employed to deal with the oscillations of a system about a state of equilibrium will no suffice for the of our problem .
In most " " gyrostatic\ldquo ; problems which have been solved hitherto , solution has been obtained by means of a system of quasi-normal coordinates .
When such coordinates exist , only a finite number of oscillations of certain particular types are possible , and , by constraining the systern to vibrate in one of these types , we may treat it in the same manner as a system with a finite number of degrees of freedom .
The period-equation for the free oscillations of an assumed type will then only possess a finite number of roots , and will consequently be an algebraic equation usually most readily obtained in a determinantal form .
It is shown at the end of S5 that the coordinates we have used possess this property when the depth of the ocean follows certain restricted laws ; but in general no such quasi-normal coordinates exist , and whatever coordinates be employed , the displacements in any of the fundamental modes of vibration can only be expressed by means of an infinite number of such coordinates .
The most advantageous choice of coordinates will then be that which leads to most converging series .
As however the oscillations of an assumed type can only be expressed by an infinite series of coordinates , it follows that an infinite number of oscillations of any assumed type must be possible , and that consequently the period-equation for oscillations of this type will have an infinite number of roots and will therefore be transcendental instead of algebraic in character .
Ic is possible that the transition from systems with finite freedom to systems with infinite freedom may be treated with advantage by the employment of determinants of infinite order ( as a means of sing the transcendental period-equation ) , after the manner introduced into analysis by * Natural Philosophy , ' Part I. , S345 .
RAYLEIGH , ' Theory of Sound , ' vol. 1 , S87 .
Cf .
, ' Acta Matbematica , ' vol. 7 ; BRYAN , ' Phil. ' 1889 .
G. W. HILL , in his 'Researches on the Lunar Theory in the present instbnoe are able to avoid the difficulties involved in the use of these infinite determinants , in that the forms of determinant which occur are those which are associated with continued ctions .
S6 deals with the analytical discussion of the deduction of the period-equation .
The method is based on a paper by Lord KELVIN , in which the author defends the procedure of LAPLACE against certain allegations to which it had been subjected b.y AIRY , but have endeavoured to present the arguments in a somewhat different light , so as to bring out more clearly the analogy between our problem and the general problem of vibrating systems with finite freedom .
In I have given illustrations of the method of solving the period-equation numerically , and of the subsequent determination of the type of motion for the different fundamental modes .
As the ground in these sections is almost entirely new , I have devoted considerable time and ]abour to the arithmetical determination of the periods and types of the principal oscillatiollS for a system comparable with the earth in magnitude .
The results are tabulated in these sections .
S11 deals briefly with the forced tides of long period due to the moon in an ocean of uniform depth .
The results agree with those previously obtained by other methods , but differ from them in analytical form .
In S12 I have given illustrations of a means of extending the method of numerioal computation to cases where the law of depth is less restricted in character .
The consideration of forced tides of very long period , dealt with in S13 , points to the existence of free oscillations of infinitely long period .
This , I believe , was first noted by Professor LAMB , but the appJication of the dynamical equations for the tides to the treatment of these free oscillations has not been previously carried out .
The types of motion in question appear to be of considerable importance , as they throw light on a phenomenon which in the past has been the subject of considerable controversy .
The difficulties which have been met with in attempts to account for the existence of ocean currents all seem to me to arise from an overestimate of the effects of viscosity on the motion of the sea .
The large-scale ocean currents have been attributed by Sir JOHN HERSCHELS and others entirely to the influence of the { trade\ldquo ; and other prevailing winds , which give rise to slow steady lnotions hich , in the absence of friction , would remain permanent even were.the originating cause entirely to cease .
The difficulty in accepting this view arises from the assumption that such currents wouid succumb to the influence of frictional * Acta Mathematica , ' vol. 8 .
'Phil .
Mag 1875 , p. 227 .
' Hydrodynamics , ' S S 'Physical Geography , ' SS67-60 .
ANALYSIS TO THE DYNAMICAL THEORY OF THE ] resistances in the course a few days .
* If this assumption be correct it will of ooure be necessary to invoke some more constant cause than the fickle winds in anation of ocean currents , but unfortunately the causes put forward by the chief opponents of the wind theory , namely , the differerlces of density arising from differences of temperature , salinity , \amp ; c. , though no doubt satisfying the criterion of being more constant their action , to be equally ineffective in maintaining the currents against such large resistances as would be required to destroy the currents due to the winds in a few days .
If , on the other hand , the period of subsidence of the free current-lnotions is to be reckoned rather by years , these motions could not fail to be excited and mainfained by such causes as the winds even against the action of friction .
In S14 I have dealt with the dynamics of ocean currents on the supposition that they are of the nature 01 ' free steady motionls ( probably maintained by a variety of causes ) , and that the influences of viscosity axe extremely small .
A remarkable result is the extremely restricted character of the possible forms of steady motion as contrasted with the case where the ocean covers a non-rotating globe , in which latter case the possible forms of steady motion are large extent arbitrary .
It is found that if the density of the water is uniform , the only forms of steady motion possible when the depth depends on the latitude alone are those in which the water moves along parallels of latitude , while in general the paths of the fluid particles coincide with certain lines depending only on the distribution of land and water and on the configuration of the ocean bed .
The equation by which these lines are defined is of an extremely simple character , and from it we could at once trace out the forms of the stream-lines on a chart if we had a sufficient knowledge of the configulatiou of the ocean bed .
The equator will always be one of these stream-lines , and herein we seem to have the explanation of the fact that the ocean-currents always tend to set along the equator , but in other respects it is shown that the effects of variations of density will seriously interfere with the simple laws which must hold so long as the density is uniform .
The importance of the earbh 's rotation in influencing ocean-currenfs has long been recognised by physicists , but I am not aware that previous attelnpt has been made to investigate this influence mathematically .
The numerical results obtained in S 15 are interesting , as showing how a cause , which on a non-rotating globe could not give rise to any a , ppreciable currents , may be rendered highly effective in main- taining currents as a consequence of the rotation of the earth .
In attempting to account for ocean currents , the real question at issue is : How far are the suggested causes capable of maintaining currents against the , action of friction ?
To answer this question an investigation , either mathematical or experimental , as to the effects tion is essential .
Such an investigation I have endeavoured MAURY , ' Physical raphy of the Sea S 93 .
to supply in another pape mathematical difficulties have compelled me , in treating of friction , to omit from consideration the influences due to the rotation .
We have already called attention to fact that the free steady motions on a non-rotating globe are far less restricted in character than those on a rotating globe , while , in that the latter essentially violate what appears t , o be a necessary condition when the water is viscous , namely , that there can be no slipping at the bottom , it seems to me to be probable that even the limited forms of steady motion here dealt with would be no longer possible if the water were viscous , but that , if they were started by any means , they would at once give place to periodic of comparatively short period .
This conclusion has been forced on me by the apparent impossibility of satisfying the equations of motion of a viscous ocean on a rotating globe ) means of slowly declining current-motions .
If such should be the case , it follows that no stable currents can exist without variations in the density of the water .
As however I have not as yet been able to support this view by anything approaching a rigorous natical treatment , the question must for the present remain open .
for the of a Mass Iiquid .
Suppose we are dealing with the small ions o a mass of liquid about a sf , ate of steady motion consisting of a rotation as a rigid body with angular velocity about a certain axis .
Take this axis as axis of , and refer to a set of rectangular axes rotating about it with uniform angular velocity .
Then , in the steady motion supposed , the will have no motion relatively to these axes .
Let denote the relative velocity-components at the point due to the small osci]lations .
The actual velocity-components parallel to the instantaneous positions of the moving axes will then be , therefore , if we suppose the amplitude of the vibrations sufficiently small to allow of our neglecting squares and products of the small quantities , the differential equations of motion of the liquid may be written in the .
cit. , ante .
The condition that there can be no slipping the bottom will reduce the number of degrees of freedom of the system , henee we may anticipate that certain of motion which were possible before this condition was imposed will no longer exist afterwards .
, ' odynamics , ' vol. 1 , p. 22 .
ANALYSIS TO THE DYNAMICAL THEORY OF TH TIDES .
, , where V ' denotes the potential of the bodily forces acting on the liquid , the fluid pressure : and the density .
If now we put const . . . . . .
( 1 ) , the above equations reduce to ( 2 ) , while the incompressibility of the is expressed by the additional equation ( 3 ) .
These equations , originally given by POINCARE , suffice , in conjunction with certain conditions which must hold at the boundary , for the determination of the four functions .
They are perbaps the simplest equations of which to make use when dealing with the oscillations of a mass of liquid of finite extent in three dimensions , and , for this purpose , they were first solved by PUINCARE in a adapted for satisfying boundary-conditions at an ellipsoidal surface , while additional applications have been considered by BRYAN , LOVE , and myselfS The possibility of solution in each of these cases however turns on the fact that it only required to satisfy boundary-conditions at a single ellipsoidal or spheroidal surface , whereas , in the problem presented by the tides , it is to satisfy conditions at two surfaces , namely , the ocean bed and the free surface of the ocean .
There is , however , a feature attached to this problem which bless us to surmount * ' Acba Mathematica , ' vol. 7 , p. 356 .
'Phil .
Trans 1889 .
'Proc .
Lon .
Math. Soc vol. 19 .
S 'Phil .
Trans 1895 .
MR. S. S. HOtTGH ON THE APPLICATION OF hARUorIC the difficulties arising from this .
the fluid which constitutes the ooean may be regarded as a thin layer distributed over an approximately .
spherical surface , circumstance which enables us to reduce the number of our independent variables and to treat the problem as two-dimensional one .
Before proceeding to the mation of our equations , let us examine t.hem in the form in which they are given above .
If we suppose the system is executing a simple harmonic vibration in period , we may put each proportional to , and therefore replace \amp ; c. , by \amp ; c. Tflus the equations ( 2 ) give Now in important class of oscillations , viz. , the tides of long period , the value cf will be small compared with that of ; while for another class of motions , viz. , the steady ocean-currents , we must suppose absolutely zero .
In these cases , if we retain only the most important terms , the equations of motion take the approximate form Hence , applying the operators to the first two , and making use of the third , we find Likewise also from the equation of continuity , From this we see in the case of tides of very long period the velocity of the fluid particles is approxima , tely the same at all points in the same line parallel to the polar axis , .
while in the case of the ocean-currents this is rigorously the case .
Now in order to effect the transformation of the equations of motion , it has been assumed by LAPLACE and bis ] owers that , on the analogy of " " long waves\ldquo ; when there is no otation , all fluid particles which are at one instant in a vertical line will remain in such a line .
This assumption appears to require some modification in the case of our rotating system .
We shall see hereafoer however that the assumption in question will not lead to appreciable error , provided that the depth of the ANALYSIS TO. .
THE DYNAMICAL THEORY OF THE TIDES .
water is small in comparison with the radius of the solid globe on which it resides , hypothesis which will certainly be applicable in the case of the earth .
S2 .
The Boundary-conditions .
Before proceeding with the approximations which we propose to employ hereafter , let us examine the boundary-conditions to which the functions are subject in the general case .
Suppose the fluid resides on the surface of a solid nucleus which is constrained to rotate with uniform angular velocity about the axis of .
We in trod.uce this constraint so as to avoid the complications resulting from the reactions of the fluid motion on that of the nucleus .
Since in the case of the earth the mass of the ocean is exceedingly small compared with that of the solid parts , such reactions would be very minute , while for most of the more important types of oscillation they would not exist at all .
In such types the problem is not affected by the introduction of the constraints .
The boundaries of the ocean where it is in contact with the- solid nucleus may then be regarded as fixed atively to the moving axes Ox , Oy , and the condition to be satisfied at these boundaries is that there is no flow of fluid across them .
Denoting by the direction-cosines of the normal to the surface , this condition is expressed analytically by the equation ( 4 ) .
Next , , let us examine the boundary-conditions at the free surface .
Let denote the direction-cosines of the normal to this surface in its undisturbed position , and let be the distance between the displaced surface and the mean surface measured along the normal to the latter .
Then we may equate the velocity at the mean surface in the direction of this normal to the rate at which increases ; thus at the undisturbed surface we have ( 5 ) .
Lastly , we must express the condition that the pressure at the actual free surface is zero ( or constant ) .
Now if denote an element of the normal to the undisturbed surface , and denote the values at this surface of the pressure and its rate of increase along the normal , the pressure at the actual surface is ( 6 ) , and this we have seen is to be equated to a constant .
MR. S. S. HOUGH ON THE APPLICATION OF HARMONIC But by definition of we have const const ) .
( 7 ) , where denotes the potential at in the steady motion , the potential due to the attraction of the layer of fluid contained between the actual free surface and its mean position , and the disturbing potential , which may be regarded as due to some external attracting system .
Since in ( 6 ) is already associated with the small factor , in calculating we may all small quantities of the order , and thus replace this expression by its value in the steady motion .
But from ( 7 ) we have in this case const whence Now , since the free surface of the ocean must be an equipotential surface , the resultant of gravitation , including centrifugal force , must be perpendicular to this surface .
Denoting by its value , we have and therefore ( 8 ) .
Introducing the values of from ( 7 ) , ( 8 ) into the expression ( 6 ) , and equating the latter to a constant , we find const , or , on equating periodic parts to zero , ( 9 ) , where the bars used to denote at the undistilrbed free surfaoe .
ANALYSIS TO THE DYNAMICAL THEORY OF TIDES . .
Transformation of the Equation of Continuity .
We proceed now to the transformation of our equations into a form analogous to that used by LAPLACE , dealing first with the equation of continuity .
refer to a system of orthogonal curvilinear coordinates , and suppose that the undisturbed surface of the ocean coincides with one of the surfaces const , say On the surface take a small parallelogram PQRS , bounded by curves of the systems const , const .
Through the sides of this parallelogram draw the surfaces const , COIlst , to meet the inner of the ocean in the quadrilateral and the distorted free surface in The surfaces const , const , are , of course , supposed to be in rotation with angular velocity , in common with the axes Ox , , Oz .
Let denote the relative velocity-components at the point parallel to the norrnals to the surfaces of reference and in the directions in which respectively increase , and let , where may be regarded as a function of , be the equation to the surface of the solid earth .
Let be parameters associated with our orthogonal system of coordinates , such that the line-element is given by Then the volume of liquid which flows in a unit of time across the face const of an elementary parallelopiped whose adjacent ed ges are is But if in the above figure we suppose that are the coordinates of , and that , PS , the total flow of liquid across the face PP'Q'Q will be MR. S. S. HOUGH ON THE APPLICATION OF HARuONIC found by tegrating the above expression with respect to between the limits Thus , the rate at which liquid is entering the elementary volume across the face PQ ' expressed by If in this expression- we change into , we shall obtain the rate at whick fluid is flowing in the positive direction across the face ; therefore , the rate at which fluid is leaving the element across the face is expressed by In like manner , the rate at which fluid enters across the face PS ' is and that at which it escapes across the face is Lastly , in virtue of the boundary-equation ( 4 ) , which holds at ) surface , the rate at which fluid enters over this surface is zero , while in virtue of ( 5 ) , which holds at the surface PQRS , the rate at which fluid escapes over the latter surface is expressed by Now the total amouut of liquid contained within the elementary volume under consideration is constant , and therefore , if we equate to zero the sum of the rates at which fluid is entering over all the six faces , we obtain the equation of continuity in the form \mdash ; Sa or ( 10 ) .
So far no approximation has been made other than that involved in supposing the ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
brations small .
If we now suppose that the depth of the water is small , so that is a small quantity , the above equation admits of considerable simplification .
Using square brackets to denote values at the mean surface , we have , by Theorem , .
, whence , . .
Now , if bo small as supposed above , though is finite , we may omit all the terms on the right except the first .
This amounts to sing that the horizontal velocity is sensibly uniform throughout the depth , not on account of the small value of its rate of variation , but on account of the small distance through which this variation can take effect , a tion which is not inconsistent with the results of S1 .
Hence , on neglecting small terms of the order we have , and , in like manner , .
Let denote the depth of the ocean at the point .
Then , provided be small in comparison with the radii of curvature of normal sections of the surfaces const , const , const , we may put with errors of the order of the square of the ratio of to these radii of curvature ; and therefore Substituting these values in ( 10 ) , we find ( 11 ) , where we have now used bars to denote surface values .
The standard of comparison is considered in the next section .
If we suppose ihat the ffee surface of the ocean is a spheroid of revolution about the axis Oz , it will be convenient to refer to a system of spheroidal coordinates lelated to by the equations The line-element for this system of coordinates is given ; whence , if we identify with respectively , we have and , supposing that is the equation to the free surface , the equation ( 11 ) becomes We have already neglected on the right small terms of the order compared with those retained ; we now propose to make the further hypothesis that the spheroidal of the ocean is of small ellipticity .
In this case will be small and large , in such a manner however that is finite and equal to the polar radius ; further will be approximately equal to .
Hence we find ( 12 ) , where the terms omitted on the right are of order and of order compared with those retained .
S4 .
Transformation of the Equations .
Let denote the inclination axis of of the normal to the surface const .
through any point ; then the direction-cosines of this normal will be ; ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
and the direction-cosines of the normals to the surfaces const , coust , will be respectively .
Hence we have from which we obtain Again and therefore from ( 2 ) , we find Similarly , and siu siu Allowing for the differences in the notation , the three equationsjust obtained agree with those given by Professor LAMB .
* If we suppose , each proportional to , they may be written *odynamics , ' p. 344 .
MR. S. S. ON APPLICATION OF HABMQNIC .
The equations in the form we have just written them will hold good whatever be the depth of the ocean or the ellipticity of its surface .
We now proceed to introduce approximations similar to those of the last section .
In the first place we have , as in S2 , ' and this by ( 12 ) we seeis of the order compared with or .
Hence , omitting terms of the order , and of order compared with those retained , the equations 13 ) take at the surface the approximate forms .
, ( 14 ) where , in conformity with the notation of .
S2 , we have denoted by the rate of variation of in the direction of the normal to the surface of the ocean .
From the equations ( 14 ) it appea.rs that is a quantity of the same order of magnitude as ; also if we the operator to each of the equations ( 13 ) , we shall obtain equations which enable us to express in terms of , and the surface-values of and its differential coefficients .
A little-consideration will show that in general must be of the order Now the approximations introduced in the last section will hold good provided that we may neglect and co rison with ; or , in our present notation that we may neglect and ANALYSIS THE DyNAMlCAbrHEORy OF TH TIDES .
in comparison But we have seen that are of the order , while are of the order Hence re both of the order The approximations will therefore be hnissible , provided is a..small quantity , that is- , provided that the depth of the ocean is small in comparison with the radius of the solid , a hypothesis as to the validity of which there can be no doubt .
now to equations , and solving for we find ' But we have rigorously and therefore , with errors of the order of the we may replace by Hence , finally , we obtain the values of , 5 with errors of the order compared with their true values in the form , ( 15 ) .
NDCCCXCVII , ' Substituting these values in the right-hand member of ( 12 ) , we Qbtsin ' ( 16 ) .
Hence , provided that be not equal to zero , we have ( 17 ) This equation , in conjunction with the pressure equation ( 9 ) of S2 , serves to determine in terms of .
It is equivalent to the well-known equation used by LAPLACE in the ' Mecanique Cklest omitting from consideration for the present the types of motion defined by , we propose in the present paper to discuss only those solutions which are symmetrical with respect to the axis of rotation .
In order that such solutions may exist , we must suppose that are independent of ; the equation ( 17 ) will then reduce to ( 18 ) , where for brevity we have put S5 .
Integration by Means of Zonal Harmonics. .
Suppose that is expressible as a series of zonal harmonics of the form .
Neglecting the ellipticity of the surface , we may at once write down the value at the surface of the potential due to this distribution ; we have , namely , , Part I. , Book IV .
, S 8 .
P OF THE TIDES .
where the density is expressed in gravitational units .
if denote the mean density of the earth as a whole , including the ocean , we have , with the same degree of approximation , and therefore .
As this only involves the ratio , it is independent of the unit of mass employed .
Next suppose that the surface-value of the disturbing potential can be expressed by means of the series .
Then equation- ( 9 ) of S 2 gives , or , if we write for brevity , , we have ( 20 ) .
Now , ffom the equation which defines the zonal harmonics , we find Hence , if we integrate ( 18 ) with respect to , we obtain where A is an arbitrary constant , which may be seen to be zero by putting Therefore S. HOUGH ON THE APPLICATION OF HABUON0 ( 21 ) .
But , by well known properties of the zonal harmonics , we have ; no arbitrary constant being necessaly since both sides vanis efore whence This relation will hold good when , provided we replace by zero .
Thus the right-hand member of ( 21 ) is equal to The lett-hand member may in virtue of ( 20 ) be written in the form . .-. .
Equating the two members and dividing.by , we obtain ANALYSIS TO THE DYNANICAL THEORY OF THE TIDES .
Hence , provided be constant , the two members will be identical , if all positive integral values of we have ( 22 ) , it being understood that This is on ' hypothesis that the depth is constant ; a more general hypothesis would be to suppose that is of the form where and are constants Assuming this form for , the left-hand member of ( 21 ) becomes which , by the properties proved above , is equal to Identifying this with the right-hand member , we obtain .
On introducing the values of from ( 19 ) , equations ( 22 ) , ( 22A ) may be written ( 28 ) , Z22 MR. S. S. HOUGH ON THE APPLICATION OF HABKOmO The law of variable depth which we have assumed appears to be ffie most general law which will lead to a difference-relation connecting the ceessive C 's of order higher than the second .
We shall confine ourselves chiefly to the case where the , depth is uniform , but the following remark with respect to the more general case seems worthy of attention .
If we put for brevity and suppose all the 's zero , so that thel 6 is no disturbing force , equations ( 23A ) be written ( 24 ) .
Now , suppose in the expression ) for the depth is of the form , where is an integer , and for greater definiteness let us suppose that is even .
Then and , and therefore the equations ( 24 ) will all be satisfied if . . .
and all the 's with odd suffixes vanish .
Eur. there ,.these will be sati if , . . .
are all aero , provided is a root the equation ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
' ' . . . . .
It follows that there exist certain types of free oscillation for which all the values of with suffixes greater than are zero .
For these types the height of the surfac , ewaves will be expressible by a finite series of terms terminating with a term involving In like manner , if the disturbing force be derivab]e from a potential function of the form of a second order harmonic , the equations which the force.d-oscillations are .
' If we suppose , . .
all zero , the first of tbsse equations will serve to determine , . . .
' terms of , while the remaining equations will be satisfied identically .
Thus the forced tides for the law of depth in question will be expressible by a finite series of terms terminating with a term involving This genel.al law does not hold when , owing to the presence of a term in the right-hand member of the second equation .
The fact that for these laws of depth the tide-heights could be expressed by finite series , instead of by the infinite series required , was original]y proved by LAPLACE in the ' Mecanique lest .
*Part I. , Book , S 5 .
MR. S. S. HOUGH ON THE APPLICATION or HARMONICS .
The Period-.Equation for the Free Oscillations .
Returning to the case of uniform depth , we see that the equations ( 23 ) divide themselves into two groups , in one of which only even suffixes and in the other of which only odd involved .
We therefore conclude that the types of oscillation divide themselves into two classes , in the former of which the of the surface-waves will be expressible entirely by harmonics of even order , and in the latter by harmonics of odd order alone .
An exactly similar treatment is applicable to each of these classes ; we shall therefore select for discussion the former set , ontenting ourselves as regards the latter wit .
merely stating results .
Denote by the expression ( 25 ) .
Then , putting all the 's equal to zero , the types of free oscillation will be determined by the equations . . . .
( 26 ) .
At first sight it might appear that whatever be the value of these equations will serve to determine , . .
in succession in terms of , whereas we know that this should only be possible for certain determinate values of corresponding to the different periods of free oscillation .
The manner in which these values of are to be determined involves arguments similar to those by KELVIN* in justification of the procedure of LACE with reference to the forced oscillations after it had been attacked by and From equations ( 26 ) , we obtain by actual solution 7.9 . .
* 'Phil .
Mag 1875 .
Cf .
also an analogous problem treated by .
' Phil. Trans 1880 , Part - .
p. 133 et " " Tides and Waves S3 .
" " Tidal Researches " " U. S. Coast Survey 1873 ) .
ADYSIS TO THE DYNAMICAL TIIEORY THE TIDES .
\ldquo ; \ldquo ; \ldquo ; \ldquo ; and in general or , the Cleterminant which multipiies in the last equation written down by Now the determinant is an algebraic polynomi .
ofldegree in . .
If therefore we equate it to- zero , we should obtain an algebraic equation of degree .
If has as its value any of the roots of this equation the first of equations ( 26 ) will be consistent , while will vanish .
By increasing the value of we shall approximate more and more closely to the case where an infinite number of such equations are satisfied while we shall impose an additional condition on the viz. \mdash ; that at some stage one of them must vanish .
Though in general in the actual motion none of the quantities , are zero , they are however subject to an im portant restriction , namely , that the series must form a converging series , and therefore we must.satisfy the equation The latter , equation may be regarded as the period-equation for.the free oscillations .
It follows that as is increased the roots of the equation , which make ; must approach closer and closer to certain definite limiting values , which correspond to the different periods of free .
oscillation , and that the series calculated in succession from equations ( 26 ) can only form a convergent series when has one of these values .
Now we see that is the equation obtained by eliminating the 's from .
et , of , quations XCVII. . . .
( 27 ) .
From these we obtain and therefore by successive applications 'rhus the eliminant of equations ( 27 ) can be expressed in the form \mdash ; . . . . .
We therefore see that the roots of the equation are the roots of This form for the equation has au advantage over the determinantal form , in that it enables us at once to proceed to the limit when is infinitely great , and thus to express the period-equation for the oscillations by means of the transcendental equation . . .
( 28 ) .
ANALTSIS TO THE DYNAMICAL THEORY OF THE TIDES .
We may however obtain a number of alternative forms for our period-equation .
From ( 27 ) we find and therefore by successive ations Thus , we have This is an alternative form for the equation ; by making infinite , we obtain as an alternative form of the period-equation where is any even integer .
S7 .
Numerical Solution of .
the Period-Equation .
The method I have used to solve the above equation will perhaps be best explained by giving a- rical example in detail .
Taking for the ratio of the mean density of the earth to that of water the value given by we deduce * Boy \amp ; 0 .
' vol. 56 , 1894 , p. 132 . .
MR. S. S. HOUGH ON THE APPkICAEIQN Or , : whence : the values of approximating closer and closer to unity as increases .
Next take , which corresponds to a depth of of the eart .
radius , or about 7260 feet .
With this value of the depth we find \mdash ; 02I2$7 , \mdash ; : , .
Introduce for brevity the \mdash ; . . .
The period-equation ( 29 ) may then be written : . . . .
( 30 ) .
pp se that has-a found bequatingA5 zero one of the ntiWes for .
putting , we btain w. th , theical- values . . . .
: ANALTSIS TO THE DYNAMICAL THEORY OF TIDESand with this value for we find The value of is for for 8 , , , With these values we find for the successive convergents to the continued-fract.ion , while the successive convergents to the continued fraction are .
It will be observed that these continued fractions converge with great rapidity ; so long as the depth of the ocean is not less than that we are here using , I find that when has a value in the neighbourhood of a root of the equation , the continued fractions are represented without sensible error by their fourth convergents , while in many cases the second convergents will form a sufficiently accurate approximation to their values ; this rapid convergence of course greatly facilitates the numerical computation .
In practice , the simplest method of evaluating the continued fractions is to assume that , for a sufficiently large value of and then to compute \amp ; c. , in succession from the formula Thus , in the present instance , we may put , and deduce whence , above , . .
S. S. HOUGH ON TflE APPLICATION OF ' In like manner , order to evaluate , we assuni that , \amp ; r some uffi small value of , and then compute , . .
, in turu by means of the formula In the present case we find whence and This being a small quantity we conclude that there is a root of the period-equ.ation differing but slightly from the value assumed for .
A closer approximation will be found by using this value in , and again equating to zero ; in other words , by putting The second approximation to the root is - , th erefore gi by Taking this value , and proceeding as before , we find We have now found that , when , and when whence , by interpolation , we conclude that when In general we shall at this stage obtain a sufficiently close approximation to the root sought , as may be verified by actual substitution .
Should however great accuracy be desired , we may re-start the computation , the value already found as a first approximation , and so continue until the desired degree of accuracy is atta , ined .
The number of cases in whi.ch I have ound a repetition of the process necessary is however extremely limited .
'ANALYSIS TO THE DYNAMICAL THEOIIY OF THB TIDES . .
By the method here sketched I have calculated the first six roots of the periodequation for four different depths of the ocean COlTesponding to the values 7260 , 14520 , 29040 , 58080 feet respectively , and the results are embodied in the , respectively , for .
These depths are equivalent to about following tables:\mdash ; TABLE I. feet ( Correoted vueof period whereis notation .
Depth 14,520 feet Depth 29,040 feet nnnnnn By a comparison of the 2nd and 3rd the roots of the frequency-equation are by simply solving the equations , and that this approximation improves the reater n becomes .
In the column I have given the periods oscillation f an ocean of the same depth when the rotation is annulled , calculated.by means of the formula where now denotes a constant such that hours .
It will be seen that the approximation obtained by omitting the rotation continually improves with increasing values of , but in no case will it lead to as accurate a result as the formula For instance , taking the case , the error introduced by using the first formula amounts to about 14 per cent. , whereas the second fornl gives the frequency with an error less than one per cent. of its true value .
S8 .
Unsymmetrical Types .
An exactly similar method of treatment is applicable to the types which are represented by a series of harmonics of odd order ; the period-equation for these types is given by where now denotes an odd and denote respectively the continued fractions \mdash ; . . .
inf .
ting this case in the same manner as the last , I have found the first six roots and the corresponding periods of oscillation for the four depths employed as follows:\mdash ; ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
-2a3 TABLE II .
; . .
feet ( S9 .
Numerical Computation of the Height of the Surface-Waves .
We have next to evaluate the quantities . .
; when once the periods have been determined thiS will present no difficulty .
Suppose we are dealing with the MR. S. S. HOUGH ON THE APPLICATION OF BARnONIC type whose frequency approximates to the root of the equation ' we have seen in the preceding sections how to evaluate , . . .
and , . . .
Also we have and therefore .
Thus the height of the surface-waves is given by where is the root of the frequency-equation in question , and is an arbitrary constant .
Continuing with the particular numerical example dealt with in S7 , we take , and deduce Lg 281944 , Neglecting , we find and in like manner ; ANALYSIS TO THE DYNAMICAL THEORY OF THE lrom these we deduce whence , final ; Combining with our solution a second , obtained by changing the sign of whereever it occurs , we obtain a solution in the real form . .
where are arbitrary constants .
This determines the type of oscillation for that particular mode which is in question .
It will be seen that the coefficient of -predominates , and that consequently the deformation of the surface will be similar in character to that which takes place when there is no rotation , in which case the height of the surface-waves is expressed by a single harmonic term .
The nodal circles will however be displaced from their positions when the rotation is annulled .
S10 .
Numericat Expressionsfor the Height of the SurfaceBy the method illustrated in the preceding section I have computed the series which indicate the types of oscillation for each of the forty-eight cases for which the periods are tabulated in SS7 , 8 ; these series are given in the following tables .
To obtain the height of the surface-waves , the series here given must be multiplied by a simple harmonic function of the time of arbitrary amplitude and phase , but whose period is found from the corresponding entry in the preceding tables .
'-.:.j TABLE .
\mdash ; Heights of Surface-Waves for Symmetrical ; ( 7,260 feet ) .
; ( 14,520 feet ) .
2 2 2 . . . . . .
; ( 29,040 feet ) .
; ( 58,080 feet ) , ANALYSIS TO THE DYNAMICAL OF TIDES .
TABLE IV.\mdash ; Heights of Surface-Waves for Unsymmetrical Types .
( 7,260 feet ) .
9 9 9 . . . . . .
; ( 14,520 feet ) .
feet , \mdash ; . . . . . . . . . . . .
; . . . . . . . . . . .
S11 .
Forced Tides .
Leaving now the problem of the free oscillations , le .
us leturn bo the equations of S5 , when we retain the .
It is obvious in the first place that a disturbing force whose potential at the surface is expressible by surface-harmonics of even order alone , or of odd order alone , will give rise to a forced oscillation of like character .
Further we may consider separately the effects of the different terms in the disturbing potential and superpose the results .
Suppose for example that the surface-value of the disturbing potential is expressible by the single harmonic term where we will suppose even .
The equations ( 23 ) which determine the type may be written C. , / 5 .
with the condition that ; whence we obtain and therefore or Thus the height of the tide is given by The expressions all depend on the frequency of the disturbing force .
It is obvious from the above that the tides become very large when approaches a root of the equation ; ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
and this equation , as we have already seen , is the equation which determines the periods of free oscillation .
It is usual in Tidal Theory to express the height of the forced tides in terms of the height of the corresponding " " equilibrium-tides.\ldquo ; If we denote the height of the equilibrium-tide arising from the disturbing potential in question by we see , on omitting all the terms on the left of ( 23 ) which depend on inertia , and replacing by that ; and therefore If then denote the height of the equilibrium-tide , we have . .
The most important practical ] ication of the above theory is the case where the disturbing potential involves only a single term of the second order , and the period of the disturbance is long compared with the period of rotation .
Thus taking , which corresponds to the case of lunarfortnightly tides in an ocean of depth 7260 feet , we find whence , neglecting , we obtain in succession Thus \mdash ; and therefore In the same manner , when ; when ; and when The lunar-fortnightly declinational tides have been ed by Professor DARWIN* for depths which correspond with the first and third cases given above , the results being expressed in series proceeding according to ascending powers of the variable If we replace the various powers of by their values in terms of the .
zonal harmonics , we may deduce the following series from those giveu by Professor DARWIN ; when , we find . .
, while , when .
The difference between these expansions and those we have given above , is to be explained by the fact that we have included in our analysis the effects due to the attraction of the water on itself .
I have re-computed the lunar-fortnightly tides , starting with the assumption that obtained practically identical .
results by the two methods .
We see then that the effect of the gravitational attraction of the water is to difoinish the tides , as compared with the equilibrium tides , in the first case by about 8 per cent. , and in the nd by about 5 per cent. 'Encyc .
Brit Art : " " Tides S18 .
FERRERS : 'Spherical Harmonics , ' p. 27 .
ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
S12 .
Lunar-Fortnightly Tides in Ocean of Variable Depth : A similar method of treatment may be employed when the depth follows the less restricted law , of the form made use of in S5 .
The numerical computation is greatly facilitated in this case when takes the form \mdash ; where is a small integer , since , as have seen , the series which express the tide-heights will then rapidly terminate .
For example , taking so that which makes the value of for the earth about 15,454 feet , the values of given by the equations whilst all the remaining 's vanish .
The oyed is that introduced at the end of S5 .
Taking , we find in the case of the htly tides whence we obtain at once Thus , if , where denotes the co-latitude , which for a system of the dimensions of the earth makes the law of depth feet , we find for tide-height the expression NDCCOXCVII .
Similarly , when - . . .
; : when . . .
; and when , Again , if we take . . . .
the series for will terminate with a term involving , and the values of must be computed from the equations ' ; from which we find , wheu , - . .
Thus for the law of depth ; . . .
' .
in like manner , when . .
' ; ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
when : ; and : when : For other values of we must employ a method similar to that of the last section .
The general formulae for the computation of the forced tides due to a disturbing potential of the second order are ( 31 ) , where Let us introduce the notation . . . . . .
Then it may be shown , as in the last sections , that for values of greater than 2 , ( 32 ) , and therefore the first two of equations ( 81 ) may be written .
S. S. HOUGH ON THE APPLICATION OF RABWomc .
On solving these equations we obtain .
, ( 33 ) , after which we can deduce in succession by means of For example , taking which makes the law of depth for the earth feet , we find for the lunar-fortnight .
lie tide and whence , if we suppose , we obtain in succession , from the formula Also ; whence from ( 33 ) we find Again ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
whence finally ; and the expression for the tide-height becomes . .
As a further illustration , I have computed the series for the tide-height for the case where , that is , where the depth is 14,520 feet at the poles and shallows to 4840 feet at the equator .
The value of in this case is as follows:\mdash ; When is positive , that is , when the depth at the equator exceeds that at the poles , the series appear to converge more rapidly than when he depth is uniform , but the opposite is the case when the water is deeper at the poles than at the equator .
S13 .
Forced Oscillations of Infinitely Long Period .
* If we suppose so small that we may neglect , we find , on putting the height of the forced tides the following four series in place of those given in S11 :\mdash ; . .
The lunar-fortnightly tides therefore differ only very slightly from tides whose period .
is infinitely long .
The difference between these latter and the solar semi- Several of the conclusions of the present section have been previously arrived at by Professor LAMB ( ' Hydrodynamics , ' chapter viii .
; but , on account of the important light which they throw on the later sections , I have thought it desirable to treat the questions in some detail , even at the risk repeating is already well known .
MR. S. S. HOUGH ON THE fION OHBRKOWlG annual tides will be quite inappreciable , and we may take the above series as a good representation of the solar period tides , unless effects of friction become important for such tides .
The fact that when the period of the disturbing.force iS increased without limit the free surface does not tend to approach its equilibrium form appears at first sight to be at variance ith the general law of oscillalmg syste The explanation of this apparent anomaly may perhaps be made clear by considering a simple form of\ldquo ; rostatic \ldquo ; system which sses only .
two degrees of freedom .
In the absence of frictional forces , the general equations of motion of such a system may , by a propex choice of coordinates , be expressed in the form Here denote the generalized coordinates of the system .
Of the terms on the , the are due to inertia , the terms are described by rfHONSON .and TAIT as " " motional\ldquo ; forces , and the terms as " " positional\ldquo ; forces ; X , are the generafized components of the external disturbing force .
If now be supposed proportional to ; we find from the above equations whence we may obtain in terms of X , Y. When the period of vibration is indefinitely prolonged , will approach zero ils a limit , and the limiting form of the tion w in general be implies that the nts will in general tend to acquire their equilibriumvalues as the period of the disturbing force is lengthened .
There will , however , be an exception to this law if one or both of the positional forces vanish .
Let us first examine the nature of the free oscillations in such cases ; omitting X , , and supposing that while remains finite , we have for the..determinntion of the free motions or , if we denote by the generalized velocity-oomponents so that * TROMSON and TAIT , ' Natnral Philosophy , ' ; ol : 1 , p. ( 1886 edition ) .
ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
hese equations wiil be satisfied if const .
Ib follows that the 'system is capable of a small free steady motion relative to the rotating axes , defined const , If both are zero , the equations for the free motions become ; both of which are satisfied by supposing that are small arbitrary constants , and -therefore In the latter case the equilibrium-state defined by is not the only condition of ] ative equilibrium , but any other configuration of the system in the neighbourhood of this one will also form a configuration of relative equilibrium .
Let us now consider the nature of the iimiting forms of the forced oscillations when the period is indefinitely .
In the former case we must suppose that the disturbing forces are such that they do no work when the coordinate is varied , so that , as otherwise the stability of the system will be destroyed ; the -equations-of motion for the forced oscillations then beoome .
whence .
and in the limit The velocity-component will therefore always remain of the same as the disturbing force , while the amplitude of vibration of the Coordinat will tend -increase without limit .
In the latter case the lions of disturbed motion may be . . .
S. a HOUGH ON THE APPLICATION OF .
whence in the limiting case Hence both the velocity-components tend to finite limits while the amplitudes of vibration of both coordinates crease without limit .
The essential characteristics of both cases are ( i ) that one or more of the generaliA coordinates does not appear explicitly in the equations of motion but only corresponding velocity-component , and hence ( ii ) that .
is one root of the frequency-equation the free modes of vibration , from which it follows that ( a ) either free steady motions relatively to the rotating system are possible , -or that the configuration of relative equilibrium defined by is not isolated .
The two conditions , may both be expressed by stating that the steady motion defined by is not the only form of steady motion of which the system is capable .
The two cases both illustrated by our problem .
For if we suppose the waters of the ocean displaced horizontally in such a manner that the form of the surface is unaltered , we shall evidently obtain a new configuration of relative equilibrium , while , as we shall see in the next section , if we suppose the fluid is in relative motion in such a manner that the fluid particles are moving along parallels of latitude , it is possible by a proper adjustment of the free surface to ensure that such a motion should be permanent .
The coordinates which depend on the horizontal displacements alone are analogous the coordinate in the former of the illustrations we have given above , and to the coordinates in the latter .
They do not appear explicitly in the equations of motion , but only through the corresponding velocity-components .
We conclude , then , that the horizontal velocities will be of the order of the disturbing forces , whereas the horizontal excursions of the fluid particles will tend to increase without limit as the period is prolonged .
By way of explaining how these circumstances may arise physically , let us suppose for the moment that is actually zero , and consequently that the disturbing force is constant .
In the case of a system oscillating about a position of equilibrium , the introduction of a constant disturbing force will have the effect of slightly changing the configuration about which oscillation 's corresponding with t , he free modes of vibration take place .
Suppose now that a disturbing force , such as that which gives rise to the riod tides , tending to increase the surface-elliptioity of the .
ooean , is suddenly applied to our rotating system when in a configuration of relative equi- librium .
It will immediately set up oscillations , the initial motion such each particle will tend towards the position in which it would be in equilibrium under the new disturbing influence .
The new position of is such that in it there will be more water in equatorial regions , and less water in polar regions , than in the old .
Thus the initial motion involves a flow of water directed from the poles towards ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
the equator .
The water however coming from higher latitudes into lower will reach these lower latitudes with an amount of rotation less thau that which is appropriate for these latitudes if the whole were in a state of steady motion as a rigid body .
here are no forces acting which tend to modify the angular momentum about the polar axis of an elementary ring of water , which coincides with a parallel and consequently currents will be started , in virtue of which each particle of fluid will move along a parallel of latitude from east to west .
The effect of the disturbing force is therefore to modify the state of steady motion about which the free oscillations take place from a uniform rotation of the whole system as a rigid body to a state in which there exist horizontal westerly currents .
If , as is usual in dealing with forced oscillations , we suppose the free oscillations to be annulled , we see that the " " forced oscillation\ldquo ; arising from such a constant disturbance as we have been considering will be of the nature of a steady motion relatively to the rotating earth , consisting of a westerly flow of the whole ocean , the oCity however varying with the latitude .
In the case of a periodic disturbance of very long period , the motion set up at any instant will be of like character , provided that the viscosity of the fluid is not sufficient to sensibly affect the currents in question in the course of a single period .
An equilibrium-theory will only be applicable when the rate of dissipation of such motions is so rapid that they practically disappear in a time which is short compared with the period of the disturbing force .
Now in an ocean whose depth is equal to the mean depth of the actual ocean , it seems highly improbable that such currents would be appreciably affected by viscosity in the course of a few months .
Hence it appears that the present theory in which the effects of viscosity are totally disregarded will almost certainly give a far better representation of the lunar longperiod tides than the equilibrium theory , and most probably also of the long-period solar tides .
S14 .
Free Motions .
In the last section we have called attention to the fact free oscillations of infinite period are possible , or that the system with which we are concerned is capable of free steady motions .
We proceed in the present section to examine the nature of these steady motions .
Referring back to SS2-4 , we see that the general equations of motion of the ocean when free from external disturbing influence , and at the same time supposed steady , so that , can be expressed in the form where , as t , here is no longer any ambiguity , we have omitted the bars from the symbols Since when the motion is steady , we see that the first equation .
is identically satisfied if we suppose both independent of .
In this case we may take , where the constants are arbitrary , and deduce ; whence .
( 36 ) .
The last equation gives the velocity which must be imposed on the partioles of water in latitude in order that the ffee surface may be maintained in the form defined by 85 ) without any external force .
We see that it is theoretically possible to maintain an arbitrary surface-form by correctly distributing the longitudinal velocities of the fluid particles .
If however the series ( 35 ) involves harmonics of odd order the value of V given by ( 36 ) becomes infinite at the equator , and to prevent a flow of liquid across the equator it would be necessary to impose an infinite velocity on the particles of water there .
Hence , if the water extend either wholly or partially over both hemispheres , the distribution of velocity and the form of free sul.face must be symmetrical with respect to the equator , at least so far as concerns that part of the ocean which communicates across the equator .
Conversely , any arbitrary initial distribution of longitudinal velocity symmetrical with respect to the equator may be rendered permanent by an appropriate meant of the free surface .
These results hold good whatsoever be the law of depth , provided it be a function of the latitude alone .
If be not supposed independent of , we see from the second and thi1d of equations ( 34 ) that the velocity of flow across any element inclined at an angle to the meridian iS TO THE DYNAMICAL THEORY OF THE TIDES .
Hence , if , there will be no flow across the element .
It follows that the function may be regarded as a stream-function , the paths of the particles of water always coinciding with the lines const .
But from the equation of continuity we have , on putting , and replacing by their values in terms of or the general solution of which is ( 37 ) , where denotes an arbitrary function of It follows that the stream-lines const coincide in direction with the lines const ( 38 ) .
Thus , if the depth be a function of the latitude alone , the stream-lines must necessarily coincide with the parallels of latitude , and the only forms of steady motion possible are those in which the water has no latitudinal velocity .
In the more general case , the stream-lines of the possible steady motions are given by the equation ( 38 ) , and ffom this equation they might at once be traced out on a cbart if we had a sufficient knowledge of the depth of the ocean in different parts .
In particular , whatever be the law of depth , the equator will be one of the free stream-lines corresponding to an infinite value of , while the shores will also be stream-lines corresponding to zero values of this expression .
An infinite number of stream-lines will converge towards those points where the coast-line intersects the equator , and it is only by passing through one of these points that a particle of water could pass from the northern to the southern hemisphere , or vice versed .
As however the velocities at these particular points tend to become infinite , the equations which we have used which involve the neglect of the squares of the velocities will not be applicable to the region surrounding the points in question .
It seems most probable that the inclusion of the terms involving the squares of velocities would have the effect of diverting the streamlines , so as to cause them to follow the coast-lines even in the immediate neighbourhood of the equator .
An attempt to trace out the lines const for the North Atlantic Ocean from data obtainable from the Admiralty charts quickly showed that the theory here put forward is inadequate in itself to explain even the more salient features of the circulation in the region in question .
Observation however indicates the direction in which we must .
look for the defects of this theory .
The excessively low temperature of the water beneath the surface*in equatorial regions can only be explained by supposing that this water has travelled thither from higher latitudes , whereas we know that the currents at the surface , for the most part , set from the equator towards the poles .
We conclude that the under-currents of the actual ocean differ materially from the surface-currents , and in this respect the actual circulation differs from the types of circulation with which we have hitherto been concerned and which are the only possible types of circulation in our ideal ocean in which the density is uniform .
It has been urged by some authorities that the variations in the density of the water arising from differences of temperature , salinity , \amp ; c. , are the sole causes which maintain ocean currents , but in that we have seen that currents could exist even without such variations , it seems to me to be highly improbable that such is the case , though there can be no doubt , in the light of our present analysis , that these variations are largely effective in determining the course which the currents pursue .
If we suppose that the ocean consists of a number of horizontal layefs of different densities , but that the density throughout each stratum is uniform , then for each of the strata a function will exist defined by \mdash ; const , and the horizontal velooities for any stratum will be connected with the corresponding function by the equations ( 34 ) .
The equation of continuity for any stratum may be formed as in S3 , provided we replace by the depth of the stratum in question instead of the depth of the whole ocean .
The stream-lines for any stratum will therefore still be given by the equation ( 38 ) , with this modification in the meaning of the symbol .
It follows that the equator will still be one of the free streamlines , but the motion elsewhere may be totally different from what it be if the density were the same throughout .
It seems probable then that.the result we have obtained with reference to the tendency of the currents to set along the equator will still hold good even when the density is variable , and this conclusion is borne out by The principal facts at present known in relation to the distribution of temperature in the ooean will bo found in the ' Report of the ChaUenger Soientific Reanlts , ' ' Chemistry and ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
observation , it being noticeable at a glance at the Admiralty current charts that there is no.tendency to cross the equator except in the immediate neighbourhood of the coast .
The rigorous treatment of the problem of ocean currents , as affected by variations in the density of the water , appears to be hopelessly beyond the powers of mathematical analysis , and I will therefore leave the subject with the brief indications already given in this section , and will conclude the paper with an example illustrating another means by which possibly ocean currents are in part maintained , and which is instructive in showing the very important part played by the rotation of the earth in rendering ctive a cause which otherwise could give rise to no sensible currents .
S15 .
On Currents due to poration A cause which has been advocated* in explanation of ocean currents is the fact that in equatorial regions the amount of water evaporated into the atmosphere largely exceeds that precipitated in the form of rain in these regions .
The excess of water in the atmosphere is carried away to be precipitated in temperate and polar regions , thereby giving rise to an excess of precipitation over evaporation in the latter regions .
It has been urged with some reason that , as the actual amount of water in equatorial regions does not diminish nor that in polar regions increase from year to year , there must be a continual flow of water from the poles towards the equator .
The fact that this flow of water is in the opposite direction to that observed at the surface , which for the most part sets from the equator towards the poles , is explained by attributing the counterflow to undercurrents .
If however we subject the question to the test of mathematical analysis , we shall find that though such a flow towards the equator must necessarily exist , it is so slow as to be completely masked by larger currents due to other causes .
The flow in question will however give rise indirectly , in consequence of the rotation , to currents in the absence of dissipative forces would tend to increase without limit .
The explanation of this fact will be obvious after the discussions of S11 .
The effects of evaporation and precipitation may be conveniently represented mathematically by an appropriate distribution of sources and sinks over the free surface .
This will modify the surface-conditions at the free surface but will not interfere with the dynamical equations .
Instead of equating to we must replace it by a certain function of the position on the surface , independent of the time , but depending on the rate of evaporation and precipitation at the place , See an Article by PROCTOR in ' St. Paul 's Magazine , ' Sept. , 1869 , reprinted in 'Light Soience ' ( 1st series ) , p. 114 .
which over sufficiently long intervals of time we may regard uniform .
A simple law which will serve for purposes of illustration may be chosen as follows:\mdash ; .
The equations with which we have to deal will then be ( 39 ) , where , if we neglect the attraction due to the surface-inequalities , we may take To obtain a particular solution of these equations suppose where \amp ; c. , are all independent of Substituting these expressions in the equations ( 39 ) , and equating coefficients of we find If be a function of alone , these will besatisfied by provided be also a function of alone .
Next , if we equate the terms independent of in the two members of ( $9 , we obtain ANALYSIS TO THE DYNAMICAL THEORY OF THE TIDES .
Suppose the system starts from rest in its position of relative equilibrium , so that when t. Then whence , and therefore The last equation gives ' no arbitrary constant being added , since both sides vanish when .
Thus , or Suppose for example that is constant ; we shall then obtain const const .
Choosing the constant so that bhe mean value of over the surface is zero , we obtain finally Hence the particular solutions of the differential equations which represent the " " forced\ldquo ; motion due to the disturbing influence in question are MR. S. S. HOUGH ON THE APPhICATION OF HARrONlC , We see then that the effects of evaporation and precipitation will be to cause a steady flow of water , not by means of undercurrents only , but by currents sensibly uniform throughout the depth , towards the equator ; in addition to these currents the cause in question will give rise to longitudinal currents , not of a steady character , but increasing uniformly with the time , and these will be accompanied by an appropriate continuous deformation of the free surface .
Were it not for viscosity these currents would increase without limit and ultimately endanger the stability of the system , but under the action of dissipative forces a steady state must ultimately be attained , in which the rate at which the currents are generated exactly balances that at which they are destroyed .
Thus , suppose the type of motion set up is such that if left to itself it would be reduced in the ratio 1 : in a period .
If denote the velocity of any particle , the law of variation of under the influence of viscous forces is then whereas , if there be no viscosity and the system is subjected to such a disturbance as we have been dealing with , the velocity varies according to the law where is constant .
Equating the rate of increase of the velocity without viscosity to the rate of decrease under the influence of dissipative force , we find that the ultimate state is defined by , or Thus , if the disturbing influence tends to set up one of the possible types of motion of which the system is capable under viscosity , the ultimate velocity of any particle will be that which it would acquire in a period equal to the modulus of decay of the type of motion in question .
By way of numerical illustration , take a year as the unit of time and an inch as the unit of length , and suppose .
This will imply an annual rainfall at the poles which exceeds evaporation by 40 inches , and an annual rainfall at the equator which is less than evaporation by 20 inches .
Further , suppose .
Then .
will be numerically greatest when , or in latitude , and the greatest value of is 28,900 .
The maximum latitudinal velocity will therefore ANALYSIS TO THE DYNAMICAL THEORY OF TIDES .
amount to 28,900 inches , or about half-a-mile , per annum .
This velocity will be inappreciable to observation .
The amount of longitudioal velocity generated in the course of a year in any latitude is given by the formula .
This will be greatest when , that is , in latitude nearly , and its greatest value corresponds to a velocity of about of a mile per hour .
The maximum current velocity due to this cause would therefore amount to about four miles per hour if the modulus of decay is as long as 20 years .
The most crucial test to which we can subject the , theory of ocean currents here put forward will consist in the evaluation of the moduli of decay for the types ol concerned .
This I have endeavoured to do , but as the work involves analytical considerations of a somewhat different character from those which occur in the present work , I have deemed it advisable to present the results in a separate paper .
These results , so far as they are applicable , seem to point to a modulus of decay far in excess of the 20 years here required , but the mathematical difficulties have compelled me , in dealing with friction , to leave the rotation entirely out of account .
It appears that in the simpler system so treated , the types of free current motion are far more arbitrary in character than those at which we have arrived by including the rotation .
This arises from the fact that our rotating system wil ] be capable of a large number of free oscillatory motions besides those which we have examined in which the periods of oscillation always bear a finite ratio to the period of rotation .
As the period of rotation is lengthened , the period of each of these types of ation is prolonged , and the possible forms of steady motion where there is no rotation must include the limiting forms of each of these types of oscillatory motion .
The moduli of decay of the free current motions when there is rotation may therefore be very different in value from those obtained in the paper referred to , but it does not seem to me that they could be much less in order of magnioude than the moduli of decay of the principal types of free oscillation .
If this prove to be the case , the estimate of 20 years , which we have taken for the modulus of decay , will not be so excessive as might at first sight appear .
Of course , if the water does not cover the whole earth , or if the depth be not uniform , the currents due to the rotation will follow the free stream-lines defined by equation ( 3s ) instead of following the parallels of latitude .
The rotation will , no doubt , produce its maximum effect when the stream-lines in question coincide with the parallels of latitude , but this circumstance does not alter our main to the adequacy of evaporation and other such causes to generate currents comparable with those known to exist in the ocean .
* Read before the London Mathematioal Society , December 10th , ] CXCVII .

