]\gt ; XVI .
The Small Free and of a A. E. H. of St. John 's College , } } icated Professor G. H. DARWIN , F.R. S 1 .
Historical KIBCHHOFF'S first theory of ) second ; BOUSSINESQ ; . . . . .
491 S 2 .
rfheory of paper for thin she . . . . .
406 S.3 .
Internal strain in an element of the sIlcll . . . . .
499 S 4 .
Geometrical of ) doformation of extensibIe snrf . . .
50 S 5 .
Equations of motion and boundary-conditions . . . . .
512 S 6 .
Possibility of certain modes of vibration . . . . . ?
S 7 .
Vibrations of spherlcal shell . . . . .
S 8 .
Vibrations of cylindrical shell . . . . .
5:38 S 9 .
Summary of results . . . . .
543 I PROPOSE , in the first place , to give , brief account of the principal theories of the vibrations flexure of a thin elastic plate hitherto put forward , and afterwards to apply the method of one of them to the case when the plate in its natural has finite curvature .
over the early attempts of Mdlle .
, the first who succeeded in obtaining a theory of the flexure of a thin plane plate was POISSON .
In his memoir ae he obtains the ential etion for the deflection of plate , which is generally admitted , and certain boundary-conditions , which ) met less general acceptance .
The idea of method be sim ply stated .
plate being very thin , ) may ] the functions which occur in the equations of and boundary-conditions in of the riable expressing the distance of a particle the middle-surfice the , taking only the up to the third order , we obtain the erential etions for the deterlnination of the displacements which are generally admitted .
The .
of POI , ssoy 's -conditions is as ] the to form of ) * " " moire sir et le vcment des ) iqneh , \ldquo ; ' Paris ' 1829 .
Cf .
, ' ophy , , pp. lIIt .
A. .
H. ON TIIE SMALL FREE TIONS to be held in its actual position , partly by the forces directly applied to its mass , and I ) ) by the action of the emainder of the plate exerted across the ; if the plate be now cut out , it will be necessary , in order to hold it in the sanne tion , to apply at ever point of its edge a distribution of force and with that exerted by the remainder before the plate was cut out .
Now , it has been shown by KIRCHHOFF * that these ions express too much , and that it is not generally possible to tisfy them ; but bhe method proposed by fSON and 1T gives a rational explanation of KIRCHHOFF'S union of two of POTSSON'S boundary-conditions in one , and renders his complete .
However , the oction raised by DE ST .
VENANT the tal assumption that the in all element can be expanded in integral powers of the distance from the middle-surface , seems to require a different theory .
The next epoch in the theory of plates is marked by memoir just referred to .
The method rests on two assumptions , .
: ( 1 ) Every straight line of the plate which was originally perpendicular to the plane bounding surfaces remains straight after the deformation , and perpendicular to the surfaces which were originally arallel to the plane bounding surfaces ; ( 2 ) all the elements of the middle-surface ( i.e. , the which in the natural state was midway between the plane parallel surfaces ) remain unstretched .
Both these assumptions may be shown to be approximately true in the cases of flexure and transverse vibration , but , as assumptions , they ) .
In memoir of KIRCHHOFF'S the union of two of POISSON'S boundary-conditions in one was first effected , the method employed to tain the equations being that of virtual work .
The theory of this memoir will be referred to as KIRCHHOFF'S " " first theory.\ldquo ; has given a general method for the treatment of elastic bodies , some of whose dimensions are indefinitely small in comparison with others .
In this method we consider , in the first place , the equilibrium of an element of the body all whose dimensions are of same order as the indefinitely small dimensions .
When we know the potential due to the internal strain of such an element , we obtain by .
the remaining dimensions the whole potential energy due to the elastic strain of the body .
Then , taking into account all the forces which act on the body , we can form the equation of virtual work , which will lead directly to the differential equations and boundary-conditions of our problem .
In KIRCHHOFF'S method it appears that , to a first ) proximation , the bodily forces produce displacements which are negligible compared with those produced by the surface-tractions exerted upon the by contiguous elements , and that , to the * ' ' Uober das GleiclJgewicht und die Bewegung einer elastischen eibe , \ldquo ; ' CmILLE , .Tourn .
Math vol. 40 .
, pp. 190-1 .
Translation of ' Elnsticitat , ' Note on S 73 , p. 725 .
S ube ] ematische Physik , ' 406 et AND TION OF Slfl LL. same of ) roximation , the placements , when divided by quantities of one dimension in , are ) ' ible c with the s The tion of this method to the theory of plates to have been first made by GEHRING , a pupil of , at the latter 's and the results will be found thirtieth lectul .
C , and in 'Theorie der ster Korper , ' .
We shall call the thus deduced " " second theory.\ldquo ; POISSON and had both al.rived at the which express that the ction excrted on elenl e of a sn normal to the middle-surface of plate is ) to the middle .
These eqnations are fundamental iu KIRC , second eory .
Tbis appears to lie at the root of the objection raised by DE to this theory , it is stated by him and .
, if they exist , may produce Hects , especially when the material of the plate is not isotropic .
It seems unnecessary to explain in detail and TAIT'S tment of the problem .
We need only note here that he equations are a for this theory also .
[ Added July , 1888.\mdash ; An ortant inference from the method is that line of particles to the middle-surface is } ) oximat ely normal to this surface after strain .
This is expressed by the nishinr ) ' of the and , as given by equations ( 11 ) .
This conclusion is tely bound up with the conclusion that and vanish .
At the of the plate and given values which do not vanish , and the approximate ndiculari t of line-elements originally dicular to the middle-surface will down .
The transition from a state of things in which and exist at the edge to one in they , on a surface parallel to the edge and very to it , is by he sion in and TAIT'S 'Natural Philosophy , ' .
The conclusion seems to be that KIRCHHOFF'S general method for the tment of bodies , some of whose dimensions are indefinitely small !
comparison with others , cannot be ) lied to the elements situated very near to the of a plate , as the strain is not ) uced in these by the action of conti uous elements .
We , nevertheless , it as giving correctly , not the potential energy due to { he strain of elenlent at distance from , but also the whole potential arising the strain in all the elements .
It will thus lead us to the of lnotion or equilibrium and diti o ] The theory of the flexure of an elastic plate ) placed much c light by the rehes of BOUSSINESQ , who has treated the subject in manner in two memoirs .
In the first of these lie proved that is an approximation to the actual state of within element of the *I use TAI notation for the elastic Tranhlation of ' ' p. C91 . .
A. .
H. ON SlIAL I VIBRATIONS thd ] that to a higher degree of approximqtion than and T. for ) kness of the plate , and the plane of for the middle-surface in natural , we , on inte ration , with reference to the bodily forces are such that if applied to a body of finite size they would produce deformations indefinibely great compared with those produced in the plate , and that do not vary rapidly from one element to another , we see are small tred ith , U. BoUSS 1 proceeds to express three of the strains in of the rest by means of the tions S , as was done in second theory ; then , by means of these approximate values , he finds to a higher order , and on substituting in the general equations of equilithe well-known equation for the deHection of the plate .
The method of aring the union of of POISSON'S boundary-conditions in one is the same as that previously by TAIT .
BOUSSrNESQ returned to the subject in 1 .
in a second memoir published in ' ' Apparently dissatisfied with the assumptions , he oposed to consider the subject in the manner .
Let the plate be divided into elementary rectangular prisms , whereof the linear dimensions are all coml)arable , these prisms bounded by the plane surfaces of the plate , and by pairs of planes at right to these surfaces .
neighbouring pl.isms must be in nearly the same condition as ards strain , in the case of the edge .
Hence , generally , the component stresses will be approximately the same at all points on the same surface parallel to the middlesurface , and not finitely near the edge of the plate .
Hence , in this kind of eqnilibrinm , the stresses will be approximately independent of the position on the middle-surface of the centre of the element .
This is precisely KIRCHHOFF'S result* deduced from the tics of the system , it appears certainly true when the plate is very thin .
BOUSSINESQ wishes his theory to to of small finite thickness , and he proposes to replace the equ tions just found by the ; * ' p. 45.3 , on equationls ( .
AND DEFO OF A TITIN ELASTIC ] these suppositions are more general than those of the former paper , itnd author to take account better of the effects of lotropy of the of the plate .
DE ST .
VENANT obtains the equations for flexure on the assum p , ( 2 ) that the of bent without , so that the extension of any line-element through a point distant the middle smface and parallel thereto is , where is the radius of curvature of the normal section of the middle-surface a line parallel to the element , From these suppositions , of which the first is justified in the manner of BoussINhSQ'S memoirs , ordinary equations deduced and extended to the case of ) .
From the inapplicability of the second of these suppositions to the case when the plate is curved we may be justified .
it the to be for the theory .
The question between the methods of second and BOUSSINESQ'S memoirs may to be of the of ) by the former .
It seems to be estabIished that the terms occur in CLFBSCH'S equations are correct to the order of approximation adopted ; but the question whether , if it were desirable to obtain a degree of in equations , this could be eflected by means of second theory ; it appears so long as the lationS S are retained with ) for the ) of giving three of the strains in terms of the rest , this question must be in the egative .
It must be observed that only uses these equations for this purpose , just as BOUSSINESQ does in his first memoir , ] the and conditions are found by ) ] ) the principle ) .
virtual work .
In a recent paper I have proposed a modification of theory , with view of showing how his equations , ) accuracy has been disputed by BOUSSINESQ , can be made exact .
The to are those unmumbered on of the ' In these certain rential coefFicients are introduced , and vards I ; and BOUSSINESQ has contended that they should be ined .
the referred to I have to show that these differential coefficients have no meaning so as we are treating the equilibrium of an elementary ) ortion of the plate , lsions are of the same order as the , so the equations be exact ply omitting these differential coelhcients .
As will fter a , KInCHHOFF'S process plics directly to the theory cf a thin elastic shell , and the modification ) oposed in the theory of has place cqnally in that of shells .
be fully explained in the sequel ( Art .
2 ) .
*Translation of nsCH .
Note to S73 .
This will be proved in the sequel .
' Elasticitat , ' pp. 306 , 3,07 , equations and ) .
S " " Note on of the ) ] of Elastic ' il . .
Proc vol. 6 , 1887 . .
Theory of Shells .
In this paper ) potential energy of deformation of an isotropic elastic shell is by the same method as that employed by KIRCHHOFF in his discussion of the tions of a plane plate .
* The shell is supposed to be bounded by surfaces parallel to its middle-surface , and is deformed in any arbitrary manner .
The expression given by KIRCHHOFF energy of the plate per unit area of its middle- surface is where is the thickness of the plate , the rigidity , and , being the ratio of ]inear lateral contraction to linear longitudinal extension of ) material ; are the extensions of two line-elements of the middle-surface initially at right angles , complement of the angle between them after strain ; are quantities defining the curvature of ) middle-surface after strain , .:\mdash ; snni of principal lrvatures , measure of curvature ; so that , if be the of curvature fter strain , the first term of the above reduces to A expression to that given by KIBCHHOFF is obtained below in the case of the shell initially curved ; but here the qual tities q are replaced by the rence of their values in the strained and unstrained states , a result which might have been anticipated from the remarks made by ( Vorlesungen , ' p. 413 ) on the strain of a rod initially curved , since the strain of an element is a linear function of these qnantities .
We wish to obtain equations of motion and boundary-conditions in terms of the displacements of point on the middle-surface of the shell , these being reckoned parallel to the lines of curvature and perpendicular to the tangent plane at the point .
For this purpose it is necessary to express all the quantities which occur in the potential-energy-function in ternls of these displacements .
As the geometrical theory of the deformation of extensible surfaces appears not to have been hitherto made out , * Called above " " KIRCHHO } second theol v\ldquo ; ANY ) I)EFORMATION OF A THIN ELASTIC it was necessary to give the elements of such a theory for small deformations .
The expressions obtained for the principal radii of curvature show that the potential energy due to bending is never the same quadratic function of the changes of principal curvature as for a plane ] ) , except in the single case where the middle-surface is a sphere and unstretched .
The general variational equation of motion is developed in the of surface and line integrals , and the equations reduce to those of CLEBSCH* in the case of a plane plate .
The terms herein which depend on externally applied forces are obtained directly , without the use of the arbitrary multipliers which render the calculations of CLEBSCH so tedious , and without the necessity which he finds for correcting an " " error\ldquo ; as regards the distribution of force at the edge , thus avoiding some of the criticisms of DE ST .
VENANT .
We know that when a plane plate vibrates the transverse displacement is independent of the displacements parallel to the plane of the plate ; and when the transverse vibrations alone are taking place no line on the -surface is altered in I discuss the question whether vibrations of the shell are possible in which this last condition holds good , and show that it leads to three partial differential equations giving the displacements as functions of the position of a point on the middle-surface , and that these equations are not in general of a sufficiently order to admit of solutions which shall also satisfy the conditions which hold at a free .
This result is quite independent of the theory adopted , as the equations of inextensibility are in the most general case a system of he third order , while the boundary-conditions are four in number .
It would , of course , be possible to find a system of forces applied to the boundary which could artificially maintain this kind of vibration .
It appeors , then , that the term of the potential enelgy which depends on the bending , which is multiplied by , is small compared with the term depending on the stretching , which is multiplied by ; and , in order to obtain the limiting form of the theory when we may form approximate equations of equilibrium and motion and boundary-conditions by omitting the term in .
Having formed these equations , I proceed to discuss the question whether the shell can execute vibrations in which there shal ] be no tangential displacement , and it is shown that this requires both the principal radii of curvature of the middle-surface to be constant at every .
The frequencies of the purely radial of a sphere and an itely long circular cylinder are given ; the displacement is a simple harmonic function of the time , and is the same at all points of the sphere or cylinder .
The formula for the frequency admits of independent verification .
Another general result deduced from the approximate equations is that any shell whose middle-surface is a surface of revolution * ' pp. .306 , 307 ; Equations ( 105 ) , ( 106 ) .
Ibid. , p. 284 .
Translation of CLEBSCIT , p. 691 .
The method of CLEBSCH is styled " " ) scure , indirecte , fort compliquee can execute purely tangential vibrations such that every point moves perpendicularly to the mel.idian through it , and the displacement is symmetrical about the axis of revolution .
The special problem of the vibrations of a spherical shell has been discussed by RAYLEIGH .
* In his paper it is assumed that no line on the middle-surface is altered in length ; the boundary-conditions are not considered .
The form of the potential energy taken is a quadratic function of the changes of principal curvature the middle-surface , and this I have proved to be in this case the true form in Art .
7 .
The assumption of inextensibility does in this case lead to expressions for the dis- placements which cannot satisfy the boundary-conditions which hold at a free edge .
The method developed in this paper is ) lied to the problem , and the approximate equations integrated .
The solution comes out in tesseral harmonics with fractional or inlaginar indices , and the frequency is given by a transcendental equation ; in case the shell be hemispherical this equation is simplified , and to express the symmetrical vibrations only the ordinary zonal harmonics with real integral indices are equired , and the frequency equation can be solved .
As a further example of the application of the method to small vibrations I have discussed the vibrations of a drical s ] .
The displacement of a point on the middle-surface is expressed by simple harmonic functions of the cylindrical coordinates of the point .
In the case of the symmetrical vibrations the frequency equation is easily solved .
ARON has applied the method of CLEBSCH to the problem of shells .
In his memoir a point on the middle-surface of the shell is considered defined by two parameters , as in GAUSS'S theory of the curvature of surfaces ; the displacements are referred to an arbitrary system of fixed axes ; and the expressions found for are the same as those in Art .
4 of this paper , but the work contains a small error ( see note to Art .
4 ) .
Free use is made of arbitrary multipliers in order to obtain the equations of equilibrium referred to the fixed axes .
As these are in a very unmanageable shape , a method of forming equations referred to moving axes is indicated ; the equations are first obtained with reference to fixed axes , and it is proposed to transform these .
The transformation is not effected , but some reductions are made with a view to it ( pp. 169 et seq In these reductions all effects due to the change of direction of the axes as we go from point to point on the middle-surface are neglected , so that the results are erroneous ( see note to Art .
6 ) .
A theory of the vibrations of a shell whose middle-surface is a surface of revolution has been given by MATHIEU .
The method is similar to that employed by POISSON for the plate , viz. , taking for the middle-surface , all the quantities which occur * " " On the lntinitesimal Bending of Surfaces of Revolution ' London Math. Soc. Proc vol. 17 , 1882 .
" " Das Gleichgewicht und die Bewegung einer unendlich dunnen beliebig gekrummten elastischen Schale 'CRELLE , Journ. Matb vol. 78 , 1874 , p. 138 .
' Memoir surle Mouvement vibratoire des Cloches 'Journ .
de l'Ecole Polytechn cahier .
AND DEFORMATION OF A THIN ELASTIC SHELL .
are expanded in powers of , and approximate equations taken .
These tions a included in those given in the present paper for shells whose middle-surface is any whatever .
MATHIEU gives for the special case some of the theorems on purely normal and purely tangential vibrations here proved ( see notes to Art .
13 ) .
The solution for spherical shells is giyen in his paper .
The introduction of the generalised tesseral harmonic into this solution enables us to recognise that a certain type of vibration given by MATHIEU cannot exist ( see note to Art .
18 ) .
The objections raised by DE ST .
VENANT to POISSON'S method for plates seem to lie equally against its extension to shells .
S3 .
Internal in Element of the Shell .
1 .
Suppose the lines of curvature on the middle-surface of the shell to be drawn ; let these be the curves const .
, const .
; then ) point on the middle-surface is given by its .
At each intersection of a curve with a curve suppose the normal to the middle-surface drawn and lengths marked off upon it qrds and outwards from the surface , the loci of the extremities of these lines are two surfaces parallel to the middle-surface .
If we suppose the space between these surfaces filled with isotropic elastic material we obtain the elastic solid shell which we wish to treat .
Let the middle-surface be covered with a network of the lines const .
, const .
at tances from each other comparable with the thickness of the shell , and suppose the normals drawn as above described at all the points of these curves .
The shell will thus be divided into a great number of elementary prisms ; and , to KIRCHHOFF'S general method , we must first discuss the equilibrium ofone of these elementary prisms .
Let be the parameters of the centre of one of these elementary prisms before strain .
Imagine three line-elements of the shell to proceed from , the elements ( 1 ) and ( 2 ) being along the lines through , and ( 3 ) along the normal at to the middle-surface .
Then after strain these lines are not in general coorthogonal , but by means of them we can construct a system of rectangular axes to which we can refer points in the prism whose centre is P. Thus , is to be the origin , the axis of is to lie along the line-element ( 1 ) , and the plane of is to contain the line-elements ( 1 ) cvond ( 2 ) ; then the line-element ( 2 ) will make an indefinitely small angle with the axis , and the line-element ( 3 ) will make an indefinitely small angle with the axis By means of the lines of vature and the middle-surface we can construct a system of orthogonal surfaces , so that we may use the formulae of orthogonal coordinates with reference to We write for the distance between two near points\mdash ; .3 2 .
The point is defined before the strain by its , and lies on a certain surface ( the middle-surface ) .
The prism whose centre is is held in equilibrium by the action of adjacent prisms , and its parts are not in the same oonfigura- tion as that in which they would be found if this prism were separated from the rest of the shell and left to itself .
Now , if this portion were isolated from the action of neighbouring portions , any point of it ( Q ) would take a certain position defined by the intersection of three surfaces of the mily ( , which we may take to be .
Hence , when this prism is subject to the action of neighbouring prisms the position of will be given with reference to the axes at by , and after the strain is effected it will be given by referred to the axes of defined in Art .
1 .
The component displacements of are Consider a system of rectangular axes fixed in space , and after strain let , be the coordinates of referred to this system , and let the directions of the axes connected with those of the fixed axes by the scheme \mdash ; Then , after strain the coordinates of are ( 1 ) , .
These expressions are functions of ; and , hence , for each of them we have and .
In foxming these differential coefficients it is important to observe that have no differential coefficients with respect to Throughout the space within which exist , viz. , the range of values of which correspond to points within the elementary prism treated , do not vary .
In his theory KIRCHHOFF first introduces the differential coefficients analogous to *This remark was made by ARON , in his memoir in BORCBARDT'S Joul'nal , ' vol. 78 , p. 138 .
AND DEFORMATION OF A THIN ELASTIC . .
, and afterwards neglects them as small .
So that the ) and ( 7 ) to be obtained below are unaffected by the modification of the theory here proposed .
Equating the differential coefficients of ( 1 ) with respect to and , we get ' ; and , similarly , by differentiating with respect to and 3 .
Now , taking the set of three equations above written , multiply them by and add , then by and add , then by and add , and repeat the process on the second set ; the six resulting equations may be written , ( 2 ) and ( 3 ) In these are the extensions of the line-elements ( 1 ) , ( 2 ) , and is the sine of the angle the axis makes with the line-element ( 2 ) after strain , so that , if be the direction cosines of the ] -element ( 2 ) after strain referred to the fixed axes of , ' ( 4 ) Also , ( 5 ) .
According to the general principles of KIRCHHOFF'S method , we may for a first approximation omit the which occur in equations ( 2 ) and ( 3 ) , thus re-writing : \mdash ; ( 6 ) ( 7 ) ince we must have ' we find .
( 8 ) Let be the values of before strain , and let , and similarly for the others , then AND DEFORMATION OF A THIN ELASTIC SHELL .
In ( 6 ) and ( 7 ) suppose ; then , with five other equations ; subtracting these from ( 6 ) and ( 7 ) , we find ( 9 ) 4 .
These are simply the conditions of continuity of the mass of the shell when deformed .
To obtain the forms of from them we shall have to introduce stress-conditions .
As the quantit , ies in ( 9 ) are small , it will be sufficient to omit products , and so form equations of equilibrium of the element referred to the coordinates a , s if we were referring to fixed axes at P. If be three functions of to be deoermined , we have ' Hence , for the six components of strain , and for the cubical dilatation S , ' ' , ( 10 ) @ To determine , we have the stress-equaticns S2 ' being the modulus of compression , and ihat of rigidity .
Hence , and Thus , If there be no surface-tractions on the surfaces initially parallel to the middlesurface , viz. , , then , and , and also at the surfaces S2/ so that Thus , ( 11 ) Hence , *Expressions equivalent to these have been given by ARON , but his work contains an error .
His equations , , p. 145 , are strictly analogous to equations ( 6 ) and ( 7 ) above , but the terms in . . .
are all omitted .
The test ' is not applied ; if it had been , there would have resulted equations which in my notation are , but the values of are calculated subsequently by the method of Art .
7 , and ) the same as those given in equations 8 ) .
AND DEFORMATION OF A THIN LASTIC SHELL .
; ' and the potential energy per unit volume is term in where is written for Multiplying this expression by , and integrating from to , the term in disappears , and we find for the potential unit area or ( 12 ) The term containing is the term depending on the bending , and the term contaiping is the term depending on the stretching of the middle-surface .
We shall hereafter denote by the expressions S4 .
Theory of Small of Extensible Surfaces .
5 .
We have now , by means of equations ( 4 ) and ( 5 ) , to express the energy in terms of the displacement of a point on the middle-surface .
Let the displacements of the point on the middle-surface , being parallel to const .
, to const .
, and along the normal outwards , and let denote differentiation along the normal .
The square of the length of a line joining two near points of the surface before strain is and the square of the length of the same two points after strain is neglecting small quantities of a higher order .
But this same square of the new of the line is where S , are the increments of produced by strain , so that @ , and similarly for , also , and so for .
In the two expressions for the square of the new length of the line we may equate coefficients of , and ) , and omit powers of , or their differential coefficients above the first ; thus ( 18 ) where we ] ) written , iq accordance with LAME'S result Leqons Coordonnees Curvilignes , , viz , are the principal radii of curvature of the middle-surface before strain , reckoned inwards .
AND DEFORMATION OF A THIN ELASTIC SHELL .
6 .
To express in the way , we of fixed axes whose directions coincide with those of the axes at before strain .
The coordinates referred to of a point near on the surface are @ where , are the elementary rotations of the axes about themselves , when the origin is changed from to .
:\mdash ; ( 14 ) Hence , by equations ( 4 ) , ; substituting for , and neglecting small quantities of the second order , we find , ( 15 ) , whence , These are the .
, referred to axes at each point determined by three certain line-elements at the point , if denote the change in the value of these as we go from any point to a near point on the middle-surface , then referred to the fixed axes at , we have , and so on ; so that ' S ; in the same way SM We may form the ' from these , for are small quantities of the second order , and are also of the second order ; hence , to the first order , using equations ( 5 ) we obtain AND DEFORMATION OF A THIN ELASTIC SHELL .
( 16 ) The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to The relation reduces to , ( 17 ) and each of these expressions vanishes ( LAME , p. 80 ) ; thus , this condition is fulfilled identically .
Using these relations , we find ( 18 ) 7 .
The quantities defined by equations ( 5 ) have been calculated directly ; we wish to obtain an interpretation in terms of quantities defining the curvature of the middle-surface after strain .
*This may be taken as a verification in some degree of the preceding work .
In endeavouring to form equations referred to the above set of moving axes , ARON neglects the and deduces values of ( my notation ) , which do not satisfy the relation ( see the memoir above quoted , pp. 169 et seq In consequence , he is obliged to make an assumption that 8 is a small quantity of the second order .
If the relations ( 17 ) had not been known , the theory of deformation would them .
First , suppose we are ' with an inextensible surface , then By equations ( 5 ) , since , and Hence , taking the notation of SALMON'S ' Geometry of Three Dimensions , ' chapter 12 , section 4 , we have ; and the equation for the principal radii of curvature so that , if be the roots of this equation , ' .
Also *SALNON , p. 346 .
I have changed the sign of so that the roots shall be the and of Art .
6 .
AND DEFORMATION OF A THIN ELASTIC SHELL .
In the case of the sphere , this is - , where is the radius , or -@ @ ; for any other inextensible surface this will not be the case .
Now , suppose the surface extensible , and consider as two parameters defining a point on the deformed surface ; in this view they will not be orthogonal parameters , and we find or to the first order ; so Again , ' with similar expressions for To find from the definitions in equations , we notice that the terrns in will be multiplied by terms of the form Now and similarly for ; thus , the differential coefficients of , will be multiplied by factors which vanish identically , viz. , they are of the form .
Hence , equation for the radii of curyature being where nearly , we find that to the first order We have already found expressions for in terms of the displacements ; ence , we have found expressions for the new principal curvatures and the position the new principal planes , in terms of the displacements , for the position of these planes depends only on or on ; we have ] found the interpretation of the in terms of the quantities defining the curvature and the extension .
In the case of an inextensible sphere , the potential energy due to bending is @ @ any other surface , whether extensible or not , ) will not be the case .
If the middle-surface were unextellded , the above would be right to small quantities of the first order , but we always require the potential energy correct to small quantities of the second order .
S 5 .
of Motion and Boundary-Conditions .
8 .
Following KIRCHHOFF'S method , we are going to ) the principle of virtual work to obtain the differential equations of motion and equilibrium , and the boundaryconditions .
Let be the components of the force per unit mass parallel to the lines of curvature const .
, const .
, and perpendicular to the tangent plane to the middle-surface , acting at any point of the shell .
Let QP be perpendicular to the middle-surface before strain , and let be the direction cosines- of QP after strain referred to axes at , as in Artt .
1 , 6 ; if be the displacements of and the distance PQ , then , when a small variation in the configuration is made , the displacements of will be found from equations ( 1 ) , dropping the , to be AND DEFORMATION OF A THIN ELASTIC SIIELh .
@u @v @w Let be the components of the system of forces per unit area applied to the edge of the shell , and holding it in its actual configuration .
The systems of forces acting at all points of a line through perpendicular to the middlesurface , and the similar system , will each reduce to a resultant force and couple .
The resultant of the system is a force at whose components are per unit area , and a couple whose components are per unit area .
The resultant of the syslem is a force at the point in which the middle-surface cuts the edge , whose components are per unit length of the curve in which the middle-surface cuts the edge , and a coupIe whose components are per unit length of the same curve .
The general variational equation of motion is SV SW SS SS SS SWZ where is an element of area of the middle-surface and an element of arc of the edge .
Observing that by equations ( 15 ) , and integrating with respect to ffom to , we get the equation which contains in itself all the equations and conditions of the All the double integrals which occur in this equation can be expressed , partly as surface-integrals over the , and partly as line-integrals round the edge , by means of the theorem , ( 20 ) where the first tegration extends to all values of which correspond to points cn a surface for an edge , and are the cosines of the angles which the normal to the edge drawn on the surface and produced outwards makes with the directions of the lines const .
, const .
at the edge .
To prove this theorem , a line of curvature const .
meet the edge in an number of points , and let , . .
be the values of X at these points , then .
; so The partial integrations win be effected by means of the.relations @ , @ , ( 21 ) @ ( S ) .
MAXWELL , ' Electricity and ' Arb .
21 .
This theorem is otherwise proved by ARQN .
-AND DEFORMATION OF A THIN ELASTIC SHELL .
In evaluating the line-integrals we shall use the formula\mdash ; , ( 22 ) in which is the element of the normal to the edge drawn as above stated .
9 .
From equations we have\mdash ; , ( 23 ) so that @u\mdash ; Su Sw .
( 24 ) Again , Sv @v\mdash ; Sw , ( 25 ) by integration by parts .
Again , Thus , Sw Sv @ S @w @ @v Su Su Sv .
( 26 ) AND DEFORMATION OF A THIN LASTIC SHELL .
Using ( 22 ) , we find for the line-integral part , Sw Sw ( 27 ) where , by integration by parts , the second term becomes Sw Again , dSu dSv Sv Sv , ( 28 ) MR. A. E. H. LOVE ON THE SMALL FREE VIBRATIONS where the line-integral part is Sv .
Again , @u Sv Sw . . . .
( 29 ) 10 .
Collecting the terms , we have the differential equations of motion , ( 30 ) AND DEFORMATION OF A THIN ELASTIC SHELL .
The first terms in these equations reduce to those in CLEBSCH'S equations ' Elasticitat , ' pp. 306-307 ) in case the shell becomes a plane plate .
The second terms in arise from the " " rotatory inertia The third terms in arise from the term in the potential energy , and depend the bending ; tlje fourth terms ( in 2 ) arise from the term , and depend on ) 11 .
The boundary-conditions are ( 33 ) , ( 34 ) .
( 35 ) The first terms in each of these equations are the san1e as those in CLEBSCH'S equations , pp. 306 , 307 .
The couple is that called by DE ST .
VENANT the moment of torsion ; the couple is that called by him the moment of flexure , and their axes are the normal and tangent to the edge respectively .
The former of these may be considered as arising from a distribution of force in lines normal to the middle-surface and in the edge ; the difference of the forces in consecutive elements gives rise to a resultant force normal to the middle-surface which coalesces with C. This is the explanation of the union of two of the boundary-conditions given by POISSON in one .
We are going to apply the equations just developed to determine the small free vibrations of the shell .
The terms depending on the rotatory will be neglected .
S6 .
Possibility of Certain Jfodes of Vibration .
12 .
Now let us suppose , if possible , that the shell vibrates in such a manner that no line on the middle-surface is altered in length .
requires that be all zero .
Thus , from equations ( 13 ) we derive O. AND DEFORMATION OF A THIN ELASTIC SHELL .
These are three partial rential equations to determine the forms of ; and , if either and or and be eliminated , they will in general lead to an equation the third order to determine or .
When one of these is determined , the rest are known .
But at the edge we have to satisfy four boundary-conditions , and this not be generally possible with solutions of a of equations such as the above .
13 .
Since may not in general be regarded as of a higher order of small quantities than , it follows that the term in in the potential energy which contains as a factor is very great compared with the term in which contains , and we may form approximate equations of vibration and boundaryconditions by omitting the latter term .
The equations of motion thus formed are-(36 ) And the boundary-conditions are\mdash ; ( 37 ) ( 1 .
) Let us examine the possibility*of purely normal vibrations .
Since , the equations of motion become simply , ( 38 ) where is the ratio of linear lateral contraction tolinear longitudinal extension of the material of the shell .
*MATHlEII convinces limself of the impossibility by genelal reasoning .
In order that all parts of the system may be in the same phase , it is necessary that const .
an over the surface .
Again , in the equations we must pick out the terms containing , and , observing that is independent of , we may write them\mdash ; Thus , and , .
But , by equations ( 17 ) , Substituting , we get ( 39 ) So that const .
all over the surface .
The two conditions of possibility of normal vibrations show that the middle-surface must have both its principal radii of curvature constant at every point .
These conditions are satisfied by the sphere , the circular cylinder , and the plane .
Again , if the surface be bounded by an edge , we have , since ; these can coexist for all values of if ] , and To make positive , or the material resist distortion , we must have positive , so that cannot be ; the equation makes , so that or the material of the shell would offer no resistance to compression ; thus , the equations above written cannot coexist for all values of , and hence one of the two must be zero , and one of the two equations and must hold at the edge .
These conditions cannot be satisfied on a sphere or cylinder .
AND DEFORMATION OF A THIN ELASTIC SHELL .
The complete spherical shell may execute purely radial vibrations , and the frequency is where is the radius .
* The indefinitely long circular cylinder may also execute purely radial vibrations with a frequency being the radius .
Observing that the more accurate equations of motion and boundary-conditions , which contain the terms in , will in all such terms have only differential coefficients of with respect to or , the above theory is seen to hold also if these more accurate equations be considered .
( 2 .
) Again , consider the possibility purely tangential vibrations , the edge being a lin of curvature .
Since , the third of equations ( 36 ) gives at all points of the surface .
Now , the boundary-conditions at const .
are and two functions it will not generally be possible to satisfy these con- ditions .
If , however , the surface be of revolution , and be the longitude , then , and all the conditions can be tisfied by taking at all points of the surface , ( 3 ) at the edge ; * [ In the paper as read , this result was verified by reference to a tion scb in the ticnl Tripos , part III .
, 1885 .
It has since been pointed out to me that it coincides with the formula given by LAMB in ' London Math. Soc. Proc vol. 14 , p. 50.\mdash ; July , 1888 , ] MATHIEU deduced the possibility of some tangential vibrations from his diffel.ential equations .
and the equation of motion is .
( 40 ) Hence a general theorem :\mdash ; For any surface of revolution there exists a system symmetrical vibrations , in which every element moves to the meridian plane through a distance which is the same for all points on a parallel of latitude , and the frequency of such vibrations depends only on the rigidity of the material , while the ratios of the intervals are independent of the material .
These are the only purely tangential vibrations of which the shell is capable .
14 .
Let us examine more minutely the question whether a spherical shell can vibrate in such a manner that no line on the middle-surface is altered in length .
Taking , the colatitude and of a point on the middle-surface , and the radius , , thus , ( 41 ) ; ( 42 ) Suppose all zero , then and , These are the conditions given by Lord RAYLEIGH , and they show that and are conjugate solutions of the equation AND DEFORMATION OF A THIN ELASTIC SHELL .
and is given by the equation Substituting from , we find ' .
Now , .
Thus , Hence , The boundary-conditions from the terms in Su , Sv in Art .
11 are now ; since 1 , and is positive , the only way of satisfying these equations is to take at the edge .
Now , And , as shown by Lord RAYLEIGH , ( 44 ) * This might have been written down at once by the aid of GAUSS'S deformation theorem .
MR. A. E. H. LOVE ON THE SMALL FREE VIBRATIONS From ( 43 ) ; put then , and ; so that ; hence , Again , ; hence , so that and cannot both vanish all along any curve drawn on the middle-surface , unless the A vanish , which gives no displacement .
We have shown explicitly in this particular case that the assumption that no line on the middle-surface is altered in length does lead to expressions for the displacements which cannot satisfy the boundary-conditions which hold at afree ed.ge .
AND DEFORMATION A THIN ELASTIC SHELL .
S7 .
of Spherical Shell .
15 .
Let us now apply the equations of Art .
( 13 ) to the discussion of the of a spherical shell ; we have , ( 45 ) .
, .
, .
, .
, .
, .
, .
, .
, .
In these we are to substitute for their values and for their values from .
Let us take as functions of to be proportional to , then the period is ; also take , where is a number , then we have the three equations\mdash ; c.ot , ( 46 ) , ( 47 ) The first two of these are\mdash ; , ( 49 ) .
( 50 ) Substituting from ( 48 ) , these are\mdash ; , ( 51 ) and , writing ( 53 ) ( 48 ) becomes .
( 54 ) stlcu c for , we find , ( 55 ) .
( 56 ) must be the same for as for , we may put where is an integer .
AND DEFORMATION OF A THIN ELASTIC SHELL .
Then , for as functions of it is convenient to take equations ( 51 ) , ( 54 ) , and ( 56 ) , which become , ( 57 ) ' , ( 58 ) .
( 59 ) Differentiating ( 59 ) with respect to , dividing by and subtracting from ( 57 ) , Write , thus , ; ( 60 ) and ( 59 ) becomes .
( 61 ) We are going to stitute from ( 60 ) in ( 58 ) and ( 61 ) ; the result will enable us to eliminate V , and obtain an equation for We have therefore Substituting in ( 61 ) , we have , on multiplying by .
( 62 ) Now , substituting for from ( 61 ) and for from ( 60 ) , ( 58 ) becomes or , multiplying by Multiply this by and add it to ( 62 ) , thus , \mdash ; or .
( 63 ) Also , between ( 60 ) and ( 61 ) , eliminate V , then , \mdash ; .
( 64 ) The equations we have to satisfy are ( 61 ) , ( 63 ) , and ( 64 ) .
Writing instead of these become AND DEFORMATION OF A THIN ELASTIC SHELL .
, ( 66 ) .
67 Of these ( 67 ) gives V when and are known .
The solution of ( 66 ) consists of two parts\mdash ; one , the complementary function which satisfies ( 66 ) when w ; the other , the particu ] integral which satisfies ( 66 ) when is a solution of ( 6.5 ) .
We may show first that this particular integral is proportional to ; take it to be .
For , writing ( 65 ) in the form and differentiating , we have and the left-hand side is found by using ( 66 ) to be So that is a particular integral of ( 66 ) , if which are both satisfied by .
( 68 ) Thus , is a particular integral of ( 66 ) .
.3 MR. A. E. H. LOVE ON THE SMALL FREE VIBRATIONS 16 .
We have Ylow to consider the complementary functions .
In equations ( 65 ) , ( 66 ) , write , then these will be the equations of tesseral haxmonics of orders respectively .
Calling ) ( ) the solution which does nob become infinite for , we have , To find V we have so that .
Hence , ( 69 ) 17 .
Properties of .
The ltial equation is , ( 70 ) .
and for any value of , real or imaginary , this is satisfied by the integral Also , if we put , ' Handbuch der lnctionen , ' pp. 225 et seq. AND DEFORMATION OF A THIN ASTIC SHELL .
is a solution of the differential equation of zonal harmonics , ( 71 ) and this is satisfied by the integral This form would nob be adapted for arithmetical work if were imaginary .
If be imaginary , then will , where is real .
If be integral , ( 71 ) is the equation of MEHLER'S 'Kegelfunctionen ' ; and it is shown by NEUMANN* that this equation is satisfied by the integral ' and this is finite when , but infinite when ; the form of demonstration adopted holds equally when is not integral .
In general , writing , and changing the independent variable to , the equation for becomes so that one solution for is the hypergeometric series where ; and this is finite for or , so that which converges for all real values of between and \mdash ; 1 , but diverges for In our equations the quantity is always real ; the quantity may be complex of the form \mdash ; ; in any case we have always a solution of our equations in series 01 deh'nite integrals .
18 .
Supposing , known , we shall be able to down the values of ; and then , supposing the surface bounded by a small circle const .
, we have for the boundarv , -conditions * ' Ueber die Mehler'behenKegelfunctiouell themat .
Aunalen , ' vol. 18 , 1881 . . . . .
( 72 ) for some fixed value of Returning to our equations ( 69 ) , we have , omitting the factors , ; hence , ; or ( 73 ) omitting and factors , and writing and for , .
Substituting in the boundary-equations ( 72 ) , we have , on elimination of the ratio , the frequency-equation ; ( 74 ) AND DEFORMATION OF A THIN LASTIC SHELL .
and , if at the edge , or the shell be hemispherical , this is .
( 75 ) In the case of the symmetrical vibrations and the expressions found involve indeterminates .
In any other case the above expressions show that it will not be possible for the motion to be purely tangential , since , for this , , and we should have to make for some value of 19 .
In the case of the symmetrical vibrations we have to put independent of in ( 54 ) , ( 55 ) , ( 56 ) ; this gives ( 76 ) From which ( 77 ) From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which From which where have the same meaning as before .
Hence , ( 78 ) The boundary-equations ( 72 ) become ( 79 ) * Using only the differential equations , MATHIEU supposed that there ) be unsymmetrical tangential vibrations .
MR. A. E. H. ON THE SMALL FREE VIBRATIONS which can be satisfied either by , and , ( 80 ) or , and ( 81 ) This gives two types of motion .
In the first the motion is partly tangential and partly radial .
Since cannot have equal roots , and cannot vanish together , or there are no lines of no displacement .
The displacement is purely radial along the lines , and purely tangential along the lines .
The ratios of the frequencies of the component vibrations of this type depend on , i.e. , on the material of the shell .
To the second type the motion is purely tangential , every point moving through a distance a , long the parallel to the edge through it , which is the same at all points of the parallel .
The lines are nodal .
The ratios of the frequencies of the component vibrations are independent of the material of the shell .
20 .
For a hemispherical bowl at the edge .
( 1 .
) In the motions of the first type is to vanish with ; hence , is an odd integer , or , being any integer , we have say , where .
This gives ; . . .
( 82 ) this equation has always real roots .
If be the roots , and the corresponding values of , according to the formula , then ( 83 ) To get arithmetical results , let us choose the equation for becomes and are given by the table:\mdash ; AND DEFORMATION OF A THIN ELASTIC SHELL . .
9.676 .
1.414 .
The tones of the second series are all near together ; those of the first are separated by intervals rather less than for a harmonic scale .
( 2 .
) In the motions of the second type , and vanishes with ; hence is an odd number , and . . . . . .
( 84 ) If be the value of corresponding to , we have and ( 85 ) and is given by the table:\mdash ; These intervals are nearly fifths .
S8 .
Vibrations of Shell .
21 .
a fUrther example , suppose the middle-surface of the shell cylindrical ; and , to fix , suppose there is a rigid disc at one end , and at a distance from it a free bounded by a circle .
Let be the radius of the circular section of the cylinder ' .
and cylindrical coordinates of a point on the middle-surface , the origin being at the centre of the rigid disc .
In the equations of motion of Art .
13 , we have to put Taking proportional to , and , these equations become , ( 86 ) , ( 87 ) .
( 88 ) Put , then ( 88 ) is ; ( 89 ) and ( 86 ) , ( 87 ) give 1 \amp ; 3 a or , ( 90 ) .
( 91 ) AND ORMATION OF A THIN ELASTIC SHELL .
Let ; then for and as functions of we have the equationls ( 92 ) where .
( 93 ) , To satisfy these , take ( 94 ) Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence where , To satisfy these , take Then whence This is a quadratic in .
: ; and we have .
( 96 ) Let be the roots of the quadratic ( 95 ) , then ( 7 ) so that ( 98 ) , from these , we find , dropping the time-factor , , ( 99 ) sirl If there is a rigid disc at , then and vanish with , so that ; ( 100 ) The first of these is , by ( 96 ) , so that ( 100 ) can only be satisfied by , both zero , and consequently both zero , unless we take and If , we have and , so that the terms in which contain all vanish identically .
It follows that to satisfy the conditions at we must drop out the terms .
The boundary-conditions at are\mdash ; ( 101 ) where we have to take only the part in and to write ' Hence , we have and AND FORMATlON OF A THIN ELASTIC SHELL .
Eli ting , we get or From ( 95 ) , ; substituting and re-arranging , we find CD ( 102 ) this equation gives the frequency .
22 .
In the case of the vibrations , , and we have , but is finite .
Thus , the equationjust written involves some indeterminates .
We take the solutions The conditions reduce to ; hence , either , and or , and This giyes two types of motion .
In the first , the motion is partly tangential and partly radial ; the expressions for the displacements are , . . . . .
( 103 ) where the eqnation for is , or ; ( 104 ) and being any integer .
The displacement is , for each normal type of vibration , wholly tangential along the circles , and wholly .
radial along the circles ; there are no points or lines of no displacement .
The frequency depends on the length and radius of the shell , and the ratios of the intervals for consecutive tones depends on i.e. , on the material of the shell .
In the motions of the second type the displacement is purely tangential , and is expressed by ( 105 ) where the equation for the frequency is or . . . . . .
( 106 ) In this case the circles ) , are nodal lines .
The frequency varies inversely as the length of the cylinder , and the intervals between consecutive tones are independent of the material of the shell .
Note .
, 1888.\mdash ; In the paper as read some examples were next given of the pplication of the method to problems of equilibriurn .
These are now withdrawn , as of little physical interest , and not directly relevant to the subjed of the paper ( see Summary ) .
AND DEFORMATION OF A ELASTIC @HELL .
S9 .
This paper is really an attempt to construct a theory of the vibrations of bells .
In any actual bell complications will arise , which have been omitted in this discussion , partly from variations of the thickness in different parts , and partly from the want of isotropy in the material .
We can hardly expect a metal which has been subjected to the process of bell-manufacture to be other than very aeolotropic , while it is notorious that ] are usually thickest at the rim .
The difficulty of the problem in its general form seems to make it advisable to begin with the case of an indefinitely thin perfectly isotropic shell , whose thickness everywhere constant , and so small compared with its linear dimensions , that powers of it above the first may neglected in mathematical expressions , which contain the first and higher powers multiplied by quantities of the same order of magnitude .
Of previous theoretical work we have examples in Lord RAYLEIGH'S Theory of Sound , ' and in his paper on the " " Bending of Surfaces of Revolution in ARON'S and MATHIEU'S memoirs , and in IBBETSON'S treatise on Mathematical Theory Elasticity .
In the 'Theory of Soun Lord RAYLH : IGH treats the vibrations of a thin ring or infinite cylinder of matter , supposed to be deformed in such a that the motion is in one plane and the elements remain unextended , and remarks that at the time of publication this was the nearest nation to a theoretical treatment of bells .
He afterwards applies his theory of the bending of surfaces to obtain a more exact analytical method of treating the problem , but his disregard of the boundaryconditions which hold at a free eclge appears to vitiate this theory .
can be said to have attained a theory of bells , and the interest of his memoir is mainly mathematical ; his inaccuracies have been already referred to .
I have also viously referred to the objection which lies against MATHIEU'S method of treatment ; this and the complexity and difficulty of some of his analysis seem to render a new method .
I shall have to refer to IBBETSON later .
The theory here put forward rests on the form of the funotion expressing the potential-energy of deformation per unit of tlJe m of the shell .
Supposing that the surface is stretched and has its curvature changed , we find that the energy consists of two terms .
One of these contains only the functions defining the stretch , while the other contains also those the bending of the middlesurface .
The modulus of stretching is proportional to the thickness , while the modulus of bending is proportional to its cube .
Unless , therefore , the functions expressing the stretching , , the extensions and shear of rectangular line-elements of the middle-surface , are of a higher order of small quantities than those defining the bending , viz. , the changes of the principal curvatures and of the directions of the principal planes , the vibrations depend on the term which involves the and not on that which involves the bending .
Now , it seems to have been universadly *Partly rewritten , July , 1888 .
assumed by English writers that the reverse of this is the case , viz. , that the vibrations take place in such a way that no line on the middle-surface is altered in length .
This will be borne out by a reference to Lord RAYLEIGH and IBBETSON .
The theory of the present paper rests on the fact that the functions expressing the stretching and those expressing the changes in magnitude and direction of curvature are of the same order of nall quantities .
This is proved in the following potential energy consists of two parts ; one , , proportional to the thickness ; and the other , , proportional to .
The first is expressed in terms of the stretching , and the second in terms of the bending of the middle-surface .
Some previous theories have proceeded as if only occurred .
If this were the case , we ought to get an approxi- mation by supposing that .
This is equivalent to assuming that there is no stretching of the middle-surface .
We should therefore get an approximation by supposing the surface inextensible to the first order .
The stretching and the bending are expressed , to the first order , by linear functions of certain differential coefficients of the displacements .
Our supposed method of getting an approximation is then to make the functions expressing the stretching vanish .
Now , I have shown ) the functions expressing the displacement are thus , to a certain extent , determined , and that in such a way that the boundary-conditions cannot be satisfied .
The boundaryconditions referred to are the exact conditions found by retaining the complete expression for the potential .
Tt is inferred that the functions expressing the stretching cannot be taken equal to zero for an approximation ; or , in other words , small compared with those expressing the bending ; and , thus , and , are of the same order of magnitude .
The conclusion that is small compared with seems inevitable .
The argument breaks down for a plane plate through the vanishing of the curvatures ; is then alone of importance .
In the case of an open shell or bowl whose linear dimension is small compared with its radius of curvature , and large compared with its thickness , both terms are important .
When this is so , we get a class of cases for which the linear dimensions concerned are of three different orders of magnitude , and this case will not come under the method of the present paper .
It may be compared with the problem of the watch-spring mentioned in THOMSON and TAIT'S ' Natural Philosophy , ' Part 2 , p. 264 , which stands between a bar and a plate .
The very open shell or bowl stands in the same way between a plate and what I have called a shell .
The theory of this paper proceeds as if alone occurred .
It is to be regarded as the limiting form for indefinitely thin shells .
A complete theory of bells , even when regarded as uniformly thick and isotropic , could only be obtained by using the exact equations formed by retaining both terms of the potential energy .
Agaiu , English writers have assumed that the potential energy , which they suppose to depend only on the bending , will be the same quadratic function of the changes of principal curvature as it is for a plane plate .
The same authorities as before may be quoted , and we may also refer to a question set in the Mathematical Tripos , AND ORbl AfION OF A THIN LASTIC SHELL .
January 18th , morn .
, 1878 , question .
To test ass ) involved the investigation of Artt .
7 , 8 , and the result is that it is only in the case of spbere supposed unstretched that the potential energy this form .
This is the case treated by Lord RAYLEIGH , but his method still fails , a complete ) cannot be bent without stretching , while , if the sphere be incomplete , the condi tions which hold at a free edge cannot be satisfied ; this is explicitly proved in Art .
14 .
A general result is derived from the consideration of the functions expressing the kinetic and potential energies , only being ined .
Both these functions are proportional to the thickness of the shell , and thus the periods of vibration are independent of the thickness .
That this result holds for a complete thin spherical shell vibrating in any manner has been demonstrated by LAMB ( ' London Math. Soc. Proc vol. 14 , 1882 , p. 52 ) .
His equations ( 7 ) and ( 9 ) when reduced are independent of the thickness .
Two general results are obtained without solution from the equations of motion .
The first is , that involving displacement along the normal only are impossible except in the cases of the plane , complete sphere , and infinitely long circular cylinder .
IBBETSON'S treatment of the problem appears to assume ( 1 ) in extensibility , ( 2 ) the incorrect formula for the energy , ( 3 ) normal ( Jisplacements .
The other result is that any surface of revolution can execute purely tangential vibrations which are symmetrical with respect to the axis of revolution , and in which the motion is purely torsional , or perpendicular to the planes through the axis .
These must not be confounded with the familiar vlbrations of finger-bowls , which are most probably a type with two nodal meridian The theory of the vibrations of a thin spherical shell bounded by a small circle is an interesting example of general theory of vibrations of an elastic solid .
In an infinite solid there are two types of vibratory motion , the longitudinal and the distortional , both of which are propagated as waves .
In a bounded solid this state of things is modified by reflexions at the bounding-surfaces , so that the purely longitudinal and purely tangential waves do not in general exist separately .
Again , in all cases of displacement in one direction only , as in the vibrations of strings , bars , and plates , there may be displacements in different directions which are independent of each other , with their corresponding nodal lines or points .
also is modified in the general solid .
The types of vibration , for example , of a portion of a ] ) shell bounded by a small circle are partially made out in this essay .
One ediate result is that there are in general no nodal lines , properly so called .
Tn any type the displacement along the parallels vanishes at one set of meridians ; the other displacements vanish ether at another set of sets equal intervals round the sphere .
There appears to good eason to suppose erally that the corresponding proposition will not obtain with reference to nodal Jllels .
The establishment of the fact would require a of the frequency * RAYL , ' Sound , ' vol. 1 , Art .
234 .
tion , ud this I have not been able to effect .
One case , however , is readily solved , and that is where the displacement is symmetrical with respect to the pole of the sphere .
It appears here that the vibrations divide themselves into two types , one purely tangential with displacement along the parallels , the other partly radial and partly consisting of displacements along the meridians .
There are no nodal meridians .
In the purely vibrations there exists a series of nodal paralle]s , whose number corresponds to the type of vibration .
The intervals for the various tones are each of them nearly a fifth .
In the partly radial vibrations the radial displacement vanishes at one set of small circles , and the tangential displacement at another set .
The number and position of the nodal circles for the purely tangential vibration coincide exactly with the number and position of the circles along which the tangential displacement vanishes in the corresponding partly radial mode .
The vibrations of the two types belong to different normal modes of vibration , and have different frequencies .
If we like to extend the meaning of " " nodal lines so as to include the small circles just referred to , then we may state another result in the form that for radial vibrations there are two periods and modes of vibration which have the same set of " " nodal lines The tones of one of these sets are all very ; those of the other set are separated by intervals nearly the same as for a harmonic scale .
A discussion of the vibrations of an elastic shell in the form of a circular cylinder closed at one end by a rigid disc perpendicular to its axis leads to similar conclusions to types of vibration and their definition by nodal lines .
It is unfortunate that solutions of the frequency equation for the case of two ' ' nodal\ldquo ; meridians dividing the shell into four equal portions could not be obtained , as these probably include the gravest mode of vibration of which the shell is capable .
The tones of the symmetrical vibrations discussed are very high , and the theory in its present state cannot easily be tested by experiment .
There is , however , one result which would seem to admit of practical verification , viz. , it is found that , for similar thin shells , the frequency is independent of the thickness , and varies inversely as the linear dimension .

