]\gt ; INDEX SLIP .
of over of nn Solid .
Phil. , 1904 , pp. 1-42 .
rthquakes , Relations of ]heory to ] ) of .
, Horace .
Phil. Trane .
, , vol. 208 , 1904 , pp. 1-42 .
Waves , Surface , ) Tremors , in nn Iastlo Solid .
LAMB , Horace , , vol. , 1904 , pp. 1-42 .
PHILOSOPHICAL TRANSACTIONS .
I. On the Propagation of Tremors over ffie face of HORACE LAMB , F.R. Received June 11 , \mdash ; Read ll , \mdash ; Revised Octol ) , 1903 .
INTRODUCTION .
1 .
THIS paper treats of the propagation of vibrations over the surface of " " semiinfinite\ldquo ; isotropic elastic solid , i.e. , a solid bounded only by a plane .
For purposes of description this plane may be conceived as horizontal , and the solid as lyin below it , although gravity is not specially taken into account .
* The vibrations are supposed due to an arbitrary application of force at a point .
In the problem most fully discussed this force consists of an impulse applied vertically to the surface ; but some other cases , including that of an internal source of disturbance , are also ( more briefly ) considered .
Owing to the complexity of the problem , it has been thought best to collcentl'ate attention on the vibrations as they manifest themselves at the free surface .
The modifications which the latter oduces into the character of the waves propagated into the interior of the solid not examined minutely .
The investigation may perhaps claim some interest on theoretical grounds , and also in relation to the phenomena of earthquakes .
] nology ] naturally endeavoured from time to time to interpret the phenomena , all in their broader features , by the light of elastic theory .
Most of these ttempts have been based on the laws of wave-propagation in an unlimited medium , developed by and STOKES ; but Lord RAYLEIGH'S ) very special of surface- waves has made it evident that the influence of the in nlodifying the character of the vibrations is more definite and more serious than been suslJected .
The present memoir seeks to take a further step in the of tbeory to actual conditions , by investigating cases of by abandoning ( ultimately ) the restriction to simple-harmonic vibrations .
Although stances of actual earthquakes must greatly the highly idealized state of * Professor BROMWICH has shown ( Proc. Lond. Math. Soc vol. 30 , p. 98 ( 1898 ) ) in snch problems as are here considered the effect of gravity is , from a of , Ulnll ) 'Proc .
Lond. Math. Soc vol. 17 , p. 4(1885 ) ; 'Scientific Papers , ' vol. 2 , p. 441 . .
CCIII .
A 359 .
ON OF things which we are obliged to assume as a basis of calculation , it is hoped that the solution of the problems here considered may not be ] It is found surface disturbance ) oduced by impulse of short duration be ysed roughly into two parts , which we may distinguish as the ' minor tremol\ldquo ; ) the " " lnain shock respectively .
The minor tremor sets in at any place , with some .abruptness , after an interval equal to the time which a wave of displacement would take to traverse the distance from the source , Except for certain marked features at the inception , and again ( to a lesser extent ) at an epoch corresponding to that of direct arrival of transversal waves , it may be described , in general terms , as consisting of a long undulation leading up to the main shock , out gradually after this has passed .
Its time-scale is more and more protracted , its amplitude is more and more diminished , the greater the distance from the source .
The main shock , on the other hand , is propagated a solitary wave ( with one maximum and one minimum , in botl ] the horizontal and vertical displacements ) ; its time-scale is constant ; thd its alnplitude diminishes only in accordance with the usual law of annular divergence , so that its total energy is maintained undiminished .
Its velocity is that of free Rayleigh waves , and is accordingly somewhat less than that of waves of transversal displacement in an unlimited mediul ' The paper includes a number of subsidiary results .
It is conyenient to begin by the problems in their two-dimensional , calculating ( for instance ) the effect of a applied uniformly along a line of the surface .
The interpretation of the results is then comparatively and it is found that a good deal of the analysis can be utilized afterwards for the three-dimensional cases .
Again , the investigation of the effects of a simple-harmonic source of disturbance is a natural preliminary to that of a source varying accol.ding to an arbitrary law .
Incidentally , new solutions are given of the well-known problems where a periodic force acts transversely to a line , or at a poin , in an unlimited solid .
These serve , to some extent , as tests of the analytical method , which presents some features of intricacy .
2 .
A few preliminary and conventions as to notation may be put in evidence at the outset , for convenience of reference .
The usual notation of BESSEL'S Functions " " of the first kin \ldquo ; is naturally adhered to ; thus we write : ( 1 ) .
* Compare the passage of Lord paper : ' ' It is not improl that the } aves herc play an important part in rthquakes , and in the collision of sfic solids .
Diverging in two dimensions only , they must cquire f great from the sonrce a increasing preponderance The lcnlations indicate that the preponderance is mnch greater than would be inferred from mere of the ntVS of two-dimensional and divcrgencc .
EMORS OVER THE OF AN ELASTIC SOLID .
.3 By a known theorem we have also , be and positive .
For .
present purpose it is convenient to follow H. WEBER* in adopting as the standard function " " of the second kind\ldquo ; .
( 3 ) .
It is further necessary to have a special symbol for that of the two functions ( 2 ) and ( 3 ) which is appropriate to the representation of a ging wavesystem ; we write , after ( 4 ) , so that . . . . . .
( 5 ) .
We shall also write , in accordance with the usual conventions , , ( ) , ( 6 ) .
For large values of we have the asymptotic expansion ( 7 ) .
In the two-dimensional problems of this paper we shall have to deal with a numbel of solutions of the equation . . . . .
( 8 ) , constructed from the type . . . . .
( 9 ) where is , and , or . . . . .
( 10 ) , Diffmgen d. math .
Physik , ' , 1899-1901 , vol. 1 , 1 ) . .
HEINE ( ' tioneIl , , 1878-1881 , vol. 1 , p. 185 ) omits the factor .
fn terms of the more where is EULER'S constant .
The function been tabuIated ( see J. H. MICHELL , 'Phil .
' Jan. , 1898 ) .
'Phil .
Mag vol. 43 , p. 259 ( 1897 ) ; ific P } ) ' vol. 4 , p. 283 .
I have introduced the iactor reversed the sign .
SSOR l LAMB ON PROPAGATION OF ccordi n the , taken positively .
In particular , we shall lneet with the tion ( 11 ) ; and it is important to recognize that this is identical with ( hr ) , where .
To see this , we remark that , as given by ( 11 ) , is an even function of , and that for it assumes the form ( 12 ) , by the method of contour-integration .
* This is obviously equal to ( hy ) .
Again , the mean value of any function which satisfies ( 8 ) .
taken ound the circumference of circle of radius which does not enclose singularities , is known to be equal to , where is value at the centre .
We can therefore adapt an gument of THOMSON and to that a solution of ( 8 ) which no ities in the , and is symmetrical with respect to the axis of , is mined by its values at points of this axis .
We have , accordingly , ( 13 ) .
Again , in some three-dimensional problems where there is symmetry about the axis of we to do with solutions of ( 14 ) , based on the type ( 15 ) , where and is defined as in ( 10 ) .
In pal.ticular , we have the solution . . .
( 16 ) , uction .
At points on the axis of symmetry it takes the form ( 17 ) .
If we equate severally the real and imaginary parts in the second and third members of ( 12 ) , we reproduce known resuIts .
H. , ' Math. Ann vol. 1 ( 1868 ) .
Philosophy , ' S498 .
AN LASTIC SOLII ) .
Sinccl the mean of a function which satisfies , taken .
th snrf ce of a sphere of dius r not enclosing any singularities , is equal to .
Tere is the value at the centre , nent aowed f and TAIT enables us to assert that ( 18 ) , where Finally , shall require FOURILR'S Theorem in the form .
nd the analogous formul As particular cases , if in ( 19 ) we have thd for , then ( 21 ) ; and , if in ( 20 ) for , and for then ( 22 ) .
These are of course well-known results * H. , ' Crelle , ' vol. 69 ( 1868 ) .
If in ( 18 ) we put , and then separately the , we deducc These are resnlls .
cieIttific P ' vol. 3 , pp. 46 , 98 ( 1888 ) ; ON , ' ' vol. 25 , p. 71 ( 1893 ) ; and , ' Math. vol. 16 ) .
H. WEBER , ' Part .
Diff.-Gl .
etc vol. 2 , p. 190 .
Since occurs herc and in ( 20 ) only mediate variable , no confusion is likely to be caused by its to denote elastic constant .
S H. WEBFR , ' .
Diff.-Gl .
etc vol. 1 , } It may ) noticed that if in ] ( 20 ) we pnt formulae give1l in the foot-uote above .
PART I. TWO-DIMHiNSIONAL PROBLEMS .
3 .
The equations of motion of an isotropic elastic solid in two dimensions are ( 23 ) where are the component displacements , is the density , are the elastic constants of LAlI , and . . . . . . . .
( 24 ) .
These equations are satisfied by ' ( 25 ) provided ( 26 ) .
In the case of simple-harmonic motion , the time-factor being , the latter equations take the forms . . . . .
( 27 ) , where ( 28 ) , the symbols denoting ( as generally in this paper ) the wave-slownesses , i.e. , the reciprocals of the wave-velocities , corresponding to the irrotational and equivoluminal types of disturbance respectively .
The formulae ( 25 ) now give , for the component stresses , ( 29 ) .
, ' Camb .
Trans vol. 6 ( 1838 ) ; 'Math .
Papers , ' p. 261 .
The introduction of special symbols for wave-slownesses rathsr than for wave-velocities is prompted by analytical considerations .
The term " " wave-slowness\ldquo ; is accredited in Optics by Sir W. R HAMILTON .
OF AN ASTIC ) .
In the applications which we in view , the tions of the solid are supposed due to prescribed forces acting ) the plane . ? .
We thel.efore assume as a typical solution of ( 27 ) , ) plical ) to egion r , where is real , and are the positive real , or positive quantities determined by ( 31 ) .
For the region , the corresponding assumption would be .
The time-factor is here ( and often in the sequel ) temporarily omitted .
The expressions ( 30 ) , when substituted in ( 25 ) and ( 29 ) , give for the displacements and stresses at the plane ( 33 ) , and ( S4 ) .
The forms corres o to would be obtained by affixing accents to A and B. and reversing the signs of 4 .
In order to illustrate , and the same time test , our method , it is convenient to begin with the solution of a known problem , , where a periodic force transyersally on a line of matter , in an unlimited elastic solid .
Let us imagine , in the first instance , that an extraneous force of amount per unit area acts parallel to on a thin stratum coincident with the .
The normal stress will then be discontinuous this plane , \mdash ; . . . . . .
whilst the tangential stress is continuous .
These conditions give , by ( 34 ) , ( 36 ) .
Again , the continuity of and requires ( 37 ) .
* This conventioIu should be carefully ttended to ; it throughout the paper .
RAYLEIGH , ' Theory of Sound , ' 2nd ed. , S376 .
Hence We have , then , for ( 38 ) .
( 39 ) .
To pass to the case of an extraneous force concentrated on the line we make use of ( 19 ) .
Assuming that the of this formula vanishes for all but infinitesimal values of , fo which it becomes infinite in such a way that we write , in ( 39 ) , , and integrate with respect to from to We thus obtain , for or , on reference to ( 13 ) , ( 40 ) , ( 41 ) , where .
If we put , we find from ( 25 ) , on inserting the time-factol that for large values of the radial and transverse displacements are ( 42 ) , respectively .
Use has here been made of ( 7 ) .
A simple expression can be obtained for the rate at which the extraneous * The indeterminateness of the formula ( 19 ) in this case may be evaded by supposing , in the first stance , that the force , instead of being concentrated ou the line , is uniformly distributed over the portion of the plane lying between .
It appears from we sbould then have If we finally make we obtain the rssults ( 40 ) .
The second of these ults is equivalent to that given by RAYLEIGH , lo .
ci for the case of incompressibility .
TREMORS OVFR SURFACE OF AN TIC force does work in generating the cylindrical waves which travel outwards from the source of disturbance .
The formulae ( 40 ) give , for the value of at the ( 43 ) .
This expression is really infinite , bnt we are only concerned with the part of it in the same phase with the force , which is finite .
Taking this alone , we have , ( 44 ) .
Discarding imaginary parts , we find that the mean rate , ' unit length of the axis of , at which a force does is .
5 .
We may proceed to the case of a " " \ldquo ; elastic solid , bounded ( say ) by the plane , and lying on the positive side of this plane .
We exalnine , in the first place , the effect of given periodic forces applied to boundary .
As a typical distribution of normal , we take . . . . .
( 4 ) , the factor being as usual understood .
The constants in ( 30 ) are determined means of ( 34 ) , .
: ( 47 ) .
Hence ( 48 ) , where , for shortness , . . . . .
( 49 ) .
We shall find it collyenient , presently , to write also . . . . . . .
* The awkwardness is evaded if ( as in a previous instance ) we tril ) the force uniformly of the axis of .
This will introduce a factor under the ] sigllS in the sccond member of ( 44 ) .
VOL. CCIII.\mdash ; A. The surface-values of the displacements are now given by ( 33 ) , viz. : ( 51 ) .
The effect of a concentrated forcs acting parallel at points of the line is deduced , as by writing , and integrating from - to ; thus ( 52 ) .
In similar , corresponding to the tangential surface forces : ( 53 ) , we should find .
And , for the effect of a concentrated force acting parallel to at the origin , ( 55 ) .
The comparison of in ( 52 ) with in ( 55 ) gives an example of the general principle of reciprocity .
We may also consider the case of an internul source of disturbance , resident ( say ) in the line , the boundary being now entirely free .
The simplest type of source is one which would produce symmetrical radial motion ( in two dimensions ) in an unlimited solid , say , . . . . .
( 56 ) , where , de1lotes distance from the source .
If we superpose on this an equal source in the line , we obtain , . . . . . . .
( 57 ) , * RAYLEIGH , ' Theory of Sound , ' vol. 1 , S108 .
TREMORS OVER THE SURFAC OF AN ELASTIC SOLID .
where .
It is evident , without calculation , that the condition of zero tangential stress at the plane is already satisfied ; the normal stress , however , does not vanish .
It appears from ( 13 ) that the hood of the plane .
the preceding value of is ( 58 ) .
Substituting in ( 29 ) we find that this Inakes ( 59 ) .
Comparin with ( 46 ) , we see that the desired condition of zero stress on the boundary will be fulfilled , provided we superpose on ( 57 ) the solution obtained from ( 30 ) and ( 48 ) by putting and afterwards integrating with respect to from to .
The surfacedisplacements qponding to this auxiliary solution are obtained from ( 51 ) , and if we incorporate the part of due to ( 58 ) , we find , after a slight reduction , .
These calculations might be greatly extended .
For example , it would be easy , with the help of Art .
4 , to work out the case where a vertical or a horizontal periodic force acts on an internal line parallel to .
And , by means of the reciprocal theorem already adverted to , we could infer the horizontal or vertical displacement at an internal point due to a given localized surface force .
6 .
It remains to interpret , as far as possible , the definite integrals which occur in the expressions we have obtained .
It is to be remarked , in the first place , that the integrals , as they stand , to a certain extent indeterminate , owing to the vanishing of the function for real values of .
It is otherwise evident a that on a particular solution of any of our problems we can superpose a system of free surface waves having the wavelength proper to the imposed P .
The theory of such waves has been by Lord and is moreover necessarily contained implicitly in our analysis .
Thus , if we put in ( 47 ) , we find that the conditions of zero surface-stress are tisfied , provided ( 61 ) , is a root of , and , denote the corresponding values of Now , in the notation of ( 49 ) and ( 50 ) , \mdash ; 6 .
( 62 ) .
luating this to zero , we have a cubic in , and since , it is plain that there is a real root between 1 and .
It may also be shown without much difficulty that the remaining roots , when real , lie between and .
The former root makes real and positive , and therefore cannot make .
The latter roots make positive imaginaries , and therefore cannot make latter equation has only two real roots , where Thus , in the case of incompressibility it is found that , and that the remaining roots of ( 62 ) are complex .
On POISSON'S hypothesis as to the relation between the elastic constants , the roots of ( 62 ) are all real , , they are , so that .
; this will usually be taken as the standard case for purposes of numerical illustration .
In analogy with ( 28 ) , it will be convenient to write . . . . . . . .
( 63 ) , where denotes the -slowness of the )yleigh w. The corresponding wave-velocity is According as we suppose , or , this is times , or 9194 times , the of ation of plane transverse waves in an unlimited solid .
The further properties of free waves are contained in the formulae 'Proc .
Lond. .
Soc vol. 17 , 'Scientific Papcrs , ' vol. 2 , p. 441 .
AYLEIGH ( [ or .
, where it is also shown that they roots of , not of , if to their real ) osive .
TREIsIORS OVER THE SURFACE OF AN ELASTIC SOLID .
and ( 30 ) .
We merely note , for purposes of reference , that if in ( 33 ) we put and accordingly , from ( G1 ) , . . . . .
( 64 ) , we obtain by superposition a system of standing waves in which ( 65 ) .
The theory here recapitulated indicates the method to be pursued in ting the definite rals of Art .
5 .
We fix our attention , in the first instance , on their ' ' principal values in CAUCHY'S sense , and afterwards superpose such a system of free ayleigh waVes as will make the final result consist solely of waves outwards from the origin of disturbance .
It may be remarked that an alternative procedure is possible , in which even temporary indeterminateness is avoided .
This consists in inserting in the equations of motion ( 23 ) frictional terms proportional to the velocities , and finally making the coefficients of these terms vanish .
This method has some advantages , especially as regards the positions of the " " singular points\ldquo ; t6 be referred to .
The chief problem of this paper was , in fact , first worked through in this mallner ; but as the method seemed rather troublesome to expound as regards some points of detail , it was abandoned in favour of that explained above .
7 .
The most important case , and the one here chiefly considered , is that concentrated vertical force applied to the surface , to which the formulae ( 52 ) relate .
The case of a force , expressed by the formulae ( , could be treated in an exactly similar manner .
Since is evidently an odd , and an even , function of , it will be sufficient to take the case of positive .
As regards the displacement , we consider the ( C6 ) , taken round a suitable contour in the plane of the complex variable If this contour does not include either " " poles\ldquo ; , or " " branch-points\ldquo ; , of the function to be integrated , the result will be zero .
A convenient contour for our purpose is a rectangle , one side of which consists of the axis of except for small semicircular indentations the singular points specified , whilst the remaining sides are at an infinite distance on the side It is easily seen that parts of the integral due to these infinitely distant sides will vanish of themselves .
If we adopt for the radicals , * The sense in which the terms horizontal\ldquo ; and vertical\ldquo ; are used is in the sccond sentence of the Introdnction .
Fig. 1 .
at points of the axis of , the consistent system of values indicated in fig. we find , for the various parts of the first-mentioned side , where the terms with and in the denominator are due to the small semicircles about the points .
Equating the sum of these expressions to zero , we find , since , $ ( 67 ) , * The function under the integral sign in ( 66 ) is uniquely determined ( by continuity ) within and on the contour when once the vnlues of the radicals and at some one point are assigned .
The convention implied in the text is that the radicaIs are both positive at the point .
It will be noticed that over the portion of the axis of between and the function in ( 66 ) diffcrs from that involved in the value of as given by ( 52 ) .
This is allowed for in the second member of ( 67 ) .
Corrections , or rather adjustments , of this occur repeatedly in the transformations of this paper .
The symbol is used to distingnish the ( incipal value\ldquo ; of an integral ( with respect to a real variable ) to which it is prefixed .
TREMORS OVER THE FACE OF AN ELASTIC SOLID .
1.5 where ( 68 ) , a numerical quantity depending only on the ratio To examine the value of we take the integral round the same contour .
Integrating the axis of we find I I and thence by addition , since the terms due to the infinitely distant parts of the contour vanish as before , ( 70 ) , where .
( 71 ) .
Hence if to the principal values of the expressions in ( 52 ) we add the system of free Rayleigh waves , ( 72 ) , is eyidently of the type , we obtain , on inserting the time-factor , ( 73 ) , ( 74 ) .
This is for positive ; the responding results for negative would be obtained by changing the sign of in the exponentials , and reversing the sign of The solution thus found is made up of waves travelling outwards , right and left , from the origin , and so satisfies all the conditions of the.question .
The first term in gives , on each side , a train of waves travelling unchanged with the velocity .
The second term gives aggregate of waves travelling with yelocities rangin from to .
As is increased , this term diminishes indefinitely , owing to the more and more rapid fluctuations in the value of On the other hand , the rt of which corresponds to the first term of remains embedded in the first definite integral in ( 74 ) .
To disentangle it we must have recourse to another treatment of the integral .
One way of doing this is to take the integral round the pair of contours shown in fig. 2 , where a consistent scheme of Fig. 2 .
values to be attributed to the radicals and indicated .
For the only parts of the left-band contour which need be taken into account we find TREMORS OVER THE SURFACE OF AN ELASTIC SOLID .
' Similarly , .in the right-hand contour , We infer , by addition , .
If we multiply this by , and add in the term due to the free Rayleigh waves represented by ( 72 ) , we obtain , as an equivalent form of ( 74 ) , ( 76 ) It is evident that all terms after the first diminish indefinitely as is increased .
From this we can deduce , by the same method as in Art .
4 , an expression for the meitn rate at which a vertical pressure does work in generating waves , SK VOL. CCIII.\mdash ; A. If in ( 73 ) and ( 76 ) we regard only the terms which are sensible at a great distance from the origin , we have , for positive , \mdash ; -- ( 77 ) ; and similarly for ative we should find ' ( 78 ) .
These formulae represent a system of free Rayleigh waves , except for the discontinuity at the origin , where the extraneous force is applied .
The vibrations are elliptic , with horizontal and vertical axes in the ratio of the two numbers and , which are defined by ( 68 ) and ( 71 ) , respectively .
To calculate these , we have , since and therefore . . . . .
( 79 ) , where , by differentiation of ( 62 ) , ( 80). .
In the case of incompressibility I find ; whilst on POISSON'S hypothesis so that the amplitudes are , for the same value of and for the same applied force , about ] what they are in the case of incompressibility .
A similar treatment applies to the formulae ( 55 ) , which represent the eflect of a concentrated horizontal force .
Taking account only of the more important terms , I find , for positive , . . . . .
( 81 ) , and , for negative , . . . . .
( 82 ) , where . . . .
( 83 ) .
TREMORS OYER THE SURFACE OF AN ELASTIC SOLID .
The ratio of to is , of course , equal to that of to is , moreover , identical with , in conformity with the of reciprocity already to .
It appears , therefore , from the numerical values of above given , that for ; and for Again , in the case of the internal source ( 56 ) I find , for large positive values of . . . . .
( 84 ) , and , for large negative values , . . . . .
( 85 ) .
The factor indicates how the surface effect ( at a sufficient distance ) varies with the depth of the source .
8 .
If in any of the preceding cases we wish to examine more closely the nature and magnitude of the residual disturbance , so far as it is manifested at the , it is more convenient to use the system of contours shown in fig. S. With this system we Fig. 3 .
can so adjust matters that the radicals and shall assume in all arts of the axis of exactly the values with which we are concerned in formulae such as ( 52 ) .
It is convenient , for brevity , to denote by ) values AHsumed by the same radicals on the two sides of the lines , and by their values on the two sides of the line , these values being supposed determined in xnce with the quirements of continuity .
Thus , with the allocation the figure , we shall have , for small values of ( 86 ) , approximately .
Taking the integral ( 66 ) round the several contours , in the directions shown } ) the arrows , we find .
( 87 ) , where , in the first integral , , and , in the second , The integral ( 69 ) , taken round the same contours , gives ( 88 ) , on the same understanding .
The definite integrals in these results can all be expanded in asymptotic forms by means of the formula ( 89 ) ; and when , and therefore also , is sufficiently large , the first terms in the expansions will give an adequate approximation .
OVER THE SURFACE OF AN ELASTIC SOLID .
Thus , taking account of ( 86 ) , the last ) of ( 87 ) and are equivalent to and respectively .
Substituting in ( 52 ) , and adding in the system ( 72 ) as before , we have , for large positive values of , .
( 90 ) , .
( 91 ) .
The first terms in these expressions have already been interpreted .
The residual disturbance constitutes a sort of fringe to the cylindrical elastic waves which are propagated into the interior of the solid , and consists of two parts .
In one of these the wave-velocity , or , is that of equivoluminal waves ; the vibrations ( at the surface ) are elliptic , the ratio of the vertical to the horizontal diameter of the orbit being , or for .
The remaining part has the wave-velocity , or , of irrotational waves ; the surface vibrations which it represents are rectilinear , the ratio of the vertical to the horizontal amplitude being , or 3535 for .
With increasing distance the amplitude of part diminishes as , whereas in an unlimited solid the law is , as appears from ( 42 ) .
SiInilar results will obviously hold in the case of the other problems considered in Art .
5 .
9 .
It has been assumed , up to this stage , that the primary disturbance varies as a simple-harmonic function of the time .
It is proposed now to generalize the law of variation , and in particular to examine the effect of a single impulse of short duration .
From this the general case can be inferred by superposition .
It is to be noticed , in all our formulae , that if we write the symbol which determines the frequency will disappear , except in the ; this greatly facilitates the desired generalization by means of FounILl'S theorem .
Thus , in the case of a concentrated vertical pressure acting oI1 the surface , the formulae ( 73 ) and ( 74 ) lead to ( 92 ) , ( 93 ) .
The definite represent aggregates of waves , of the same general type , travelling with slownesses anging from to , and from to , respectively .
If we suppose that vanishes for all but small values of , it appears from the horizontal disturbance at a distance begins ( as we should expect ) after a ax , is the time a wave of expansion would take to travel the distance ; it lasts till a time , which is the time distortional waves would take to travel the distance ; and then , for a while , ceases .
* Finally , about the time , comes a solitary wave of short duration ( the same as that of the primary impulse ) represented by the first term of ( 92 ) .
This wave is of unchangin type , whereas the duration of the preliminary disturbance yaries directly as , and its amplitude ( as will be seen immediately ) varies inversely as If we put . . . . . . . . . .
( 94 ) , the integration extending over the short range for which is sensible , the preliminary horizontal disturbance will be given by . . .
( 95 ) , provided ( 96 ) , where .
The following table gives the values of for a series of values of , on the hypothesis of , or * This tcmporitry cessation of the horizontal motion is special to the case of a normal impulse .
If the impnlsc be genti , the co1ltrttst between the horizontal and vertictd motions , in this respect , is ovorsed .
TREMORS OVER THE SURFACE OF AN ELASTIC SOLID .
The function has a maximum value when ; it cflanges s when ; and it has a minlimmum value when The graph of this function is shown in the upper part of fig. 4 .
If the scales be Fig. 4 .
properly chosen , the curve will represent the variation of with the " " preliminary\ldquo ; disturbance , at any assigned point .
For ) purpose the horizontal scale must vary directly , and the vertical scale inversely , as The calculations were made almost entirely by Mr. H. J. WOODALL , to whom I am much indebted .
The interpretation of the expression ( 93 ) for the vertical displacement is not quite so simple .
For a given value of , the most important part is that corresponding to , or , nearly , when the inte.grand in the second term changes sign by passing through infinity .
This is the epoch of the main shock ; the minor disturbance which sets in when leads up continuously to this , and only dies out gradually after it .
As a first step we may tabulate the function V defined by V for ' ( 97 ) .
The function has a minimum value when , and a zero when ; it changes from to when or Its is shown in the lower part of fig. 4 , and also ( on a smaller scale , so as to bring in a greater range of ) in fig. 5 .
It is postulated that the function is sensible only for values of lying within a short range on each side of ; the function will therefore be sensible only for values of in the neighbourhood of .
We will suppose that for iven values of and its ( as a function of ) has some such form as that of the * As the case of , the tions are due chiefly to Mr. TREMORS OVER THE SURFACE OF AN ELASTIC SOLID .
dotted curve in fig. 5 .
If be constant , the effect of increasing will be to cause this graph to travel uniformly from left to right ; and if we imagine that in each of Fig. 5 .
its positions the integral of the product of the ordinates of the two curves is taken , we get a mental picture of the variation of as a function of , on a certain scale .
For the greater part of the range of , the integral will be approximately proportional to the ordinates of the curve V , , we shall have ( 98 ) , in analogy with ( 95 ) .
But for a short range of , in the , the statement must be modified , the dotted curve being then in the hbourhood of the vertical asymptote of the function V ( ) .
Since the the is to be taken , it is evident that as approaches the critical epoch passes it , will sink to a relatively low minimum , and then passing through zero will attain a ) correspondingly high maximum , after which it will decrease asymptotically to zero , the later stages coming again under the formula ( 98 ) .
Although the above argument gives perhaps the best view of the whole course of the disturbance , we are not dependent upon it for a knowledge of what takes ) lace VOL. CCIN .
\mdash ; A. about the critical epoch .
We may proceed , instead , by generalizing the expressions ( 77 ) .
introduces , in addition to the giyen function , whose Fourier expression is . . .
( 99 ) , the related function ( 100 ) ; , we have .
( 101 ) .
It does not appear that the connection between the functions and has been specially studied , although it presents itself in more than one department of hysics .
The cases may be noted as of interest from our present point of view : --\mdash ; ( 102 ) ; ' ' ( 103 ) ; ' ( 104 ) .
It is evident , generally , that if be an odd function , will be an even function , and The of , as given by ( 101 ) , are represented graphically in fig. 6 , tor the case have the forms given in ( 102 ) .
* Moreover , writing we . . . . .
( 105 ) ; the of a article is theFefore an ellipse with horizontal and vertical semiaxes and .
And if from the equilibrium position we project any other position of the particle to vertical straight line , the law of 's motion is that the projection ( R ) ibes this line with constant velocity .
See fig. 7 , where the direction is supposed to be downwards .
* The ) ctwecn the of in the grnphs of and ) ends upon the ratio of stic c. The figures constructed on the hypothesis of TREMORS OVER THE SURFACE OF AN ELASTIC SOLID .
Fig. 6 .
Fig. 7 .
A similar treatment would apply to the formulae ( 81 ) , and ( with some modification ) to ( 84 ) .
It remains to justify these approximations by showing that the residual disturbance tends with increasing to the limit .
For this purpose we have recourse to the of Art .
8 .
As a sufficient example , take the second terln in the last member of ( 88 ) .
If we multiply by , take the real part , and substitute , the corresponding term in the value of , as given by , assumes the form* where the functions and , which do not involve , are of the der when is large .
If we generalize this expression by Theorem ( see equation ( 99 ) ) , we obtain , in the case of an impulse of short duration , ( 106 ) .
*The symbols are here used temporarily in new sensos , SSOR H LAMB ON PROPAGATTON OF For any particular phase of the motion , varies as , and the expression ( 106 ) therefore varies inversely as .
This confirms , so far , our previous results ( 95 ) and ( 98 ) .
Hence with increasing distance from the origin the disturbance tends to the limiting form represented by ( 101 ) .
Before leaving this part of the subject , it is to be remarked that the peculiar protracted character of the minor which we have found to precede and follow the main shock is to some extent special to the two-dimensional form of the question .
It is connected with the fact , dwelt upon by the author in a recent paper , *that even in an unlimited medium a solitary cylindrical wave , whether of the irrotational or equivoluminal kind , is not sharply defined in the rear , as it is in front , but is prolonged in the form of a " " tail In the -dimensional problems , to which we are about to proceed , this cause operates in another way .
The internal waves are now spherical instead of cylindrical , and so far there is no reason to expect a protraction of a disturbance which in its was of finite duration .
But at the surface they manifest ) emselves as annular waves , and accordingly we shall find clear indications of the peculiarity of two-dimensional ) to which eference has been made .
On the whole , however , it ) that the epochs of arrival of irrotational and equivoluminal waves are relatively more clearly marked and isolated than in the twodimensional cases , PART II .
PROBLEMS .
10 .
Assuming symmetry about the of , we write , so that denotes displacement perpendicular to that axis .
A typical solution of the elastic equations , convenient for our , is derived once from , if an infinite number of two-dimensional , of the kind specified by to be arranged uniformly in all azimuths the and take the .
In this way we obt in ( 33 ) , with the of llotation , .
( 108 ) .
* Cited on p. 37 TREE IORS OVER { OF SOLH ) .
2 !
) Also , , for the corresponding stresses the phne , we have Although the above tion is pose , it be ) ) to give the direct investigation , ) ting from ] the tions , where ( 111 ) .
In the case of } ) -harmonic motion these tisfied by , provided . . . . .
, and , where are defined as befors by ( 28 ) .
A ) articular solution of ( 114 ) is ( 115 ) .
provided . . . .
On the hypothesis of symmetry about O- , we , and the formulae ( 112 ) , ( 115 ) are equivalent to ' , ( 8 ) .
* Lond. Math. , for the statical vestigation .
LAMMB ON PROPAGATION OF If we take , as the typical solution of ( 11S ) and ( 116 ) , , . . .
( 119 ) , rhere have the same meanings and are subject to the same convention as in Art .
3 , lave , from ( 118 ) , ( 120 ) ; and thence the stresses in the plane ( 121 ) .
The formulae differ and ( 109 ) only in the substitutio ] ] of for B. The notation of ( 119 ) is adopted as the basis of the subsequent calculations .
If we are to assume , in place of ( 119 ) , , . . . . .
( 122 ) , the corresponding forms of ( 120 ) and ( 121 ) would be obtained by affixing accents A and , and changing the of and where they occur explicitly .
11 .
As in Art .
4 , we begin by applying the preceding formulae to the solution of a known problem , , where a given periodic force acts at a point in an unlimited solid .
Let us suppose , in the first place , that an extraneous force of amount Z. per unit area , acts parallel to on an infinitely thin stratum coincident with the plane .
The formulae ( 119 ) will then apply , and for .
The normal will be discontinuous , viz. : . . . . . .
( 123 ) , whilst is continuous .
( 124 ) .
Also , the continuity of requires ( 125 ) .
We ' ( 126 ) , TREMORS SURFACE OF AN ELASTIC SOLID .
and therefore , , .
( 127 ) .
To pass to the case of a concentrated force , acting parallel to at the origin , we have recourse to the formula ( 20 ) , where we suppose to vanish for all but infinitesimal values of , and to become infinite for these in such a way that .
We therefore write , and integrate with respect to from to We thus find , for which are equivalent , by ( 18 ) , to .
( 128 ) , .
This will be found to agree with the known solution of the problem .
If we retain only the terms which are most important at a great distance , we find , from ( 118 ) , ( 130 ) .
Inserting the time-factor , the radial displacement is ( 131 ) , and the transverse displacement in the meridian plane is .
( 132 ) .
Returning to the exact formulae ( 128 ) , the expression for the velocity ) arallel to at the plane is found to be ( 133 ) , * A more rigorous procedure would ) to suppose in the first instnnce the force is distributed over a circular area of radius , using the formula ( 22 ) .
If in the we , we obtain the results in the text .
STOKES , 'Camb .
Trans vol. 9 ( 1849 ) ; Ylathematical and Physical Papers , ' vol. 2 , p. or , taking the real part , pt terms in pt . . . . .
( 134 ) .
The terms in pt remain finite when we put and the mean rate at which a does work in generating waves is thus found to be ( 135 ) , and denoting as before the two elastic wave-slownesses .
The result ( 135 ) can be deduced , as a particular case , from formulae given by Lord KELVIN .
12 .
Proceeding to the case of a semi-infinite solid occupying ( say ) the region we begin with the special distribution of surface-stress : , . . . . . .
( 136 ) .
The coefficients in ( 119 ) are now determined by ( 137 ) , whence ( 138 ) , the function having the same meaning as in Art .
5 .
The corresponding surface-displacements are . . .
( 139 ) .
This result might have been deduced immediately from ( 51 ) in the manner indicated at the beginning of Art .
10 .
* The terms in become infinite .
If the force ) distributed a circular area , the vardness is avoided .
A ctor is thus introduced under the mtegral signs in the first line of ( 135 ) , where denotes ( for the moment ) the radius of the circle .
, we make infinitely small .
' Phil. Mag Aug. 1899 , pp. 234 , 230- .
MORS O AN STIC S 3.3 It we put in we system of free surface-waves , in which ( 140 ) , where is the positive of , and are the corresl ) onding values of .
These are of the nature of " " standing\ldquo ; waves .
To pass to the case of a concentrated vertical pressure ) at put in accordance with ( 20 ) , , and integrate from to The formulae ( 139 ) become ( 141 ) .
Again , the case of an internal source of the type ( 142 ) , where denotes distance the point , can ) solved by a process similar to of Art .
5 .
First , superposing an equal source at , distance from which is denoted by we have ( 143 ) ; and therefore , } ) , in the ) ourhood of the ) lane . . .
( 144 ) .
This makes ( 145 ) , This may be regarded as the kinetic analoguc of 's well-known statical It might appsar at first sight that a simplcr procedure would be possible , and that the effect of pressure concentrated at a point might be inferred by snpcrposing lines of pressure ( through ) uniformly in all azimuths , and using the results of S7 .
It is easily seen , however , that such a dist1ibution of lines of pressure is equivalent to a pressure-intensity varying inversely as the distance from O. ' This is not an adequate representation of a localized pressure , since it makes the total pressure on a area having its centre at ease indefinitely with the radius of the circle . .
CCTIT .
' LAMB ON THE ATION OF .
( 146 ) .
The dditions to ( 14.3 ) which are required in order to annul esses the found by writing in , and then integrating with respect to to .
In this way we obtain , finally , ( 147 ) .
In a similar manner , with the help of Art .
11 , we might calculate the effect of a periodic force , acting at an internal point .
13 .
the sake of comparison with our previous two-dimensional formulae , it is convenient to write , from ( 2 ) and ( 6 ) , cosb ( 148 ) .
The thus equivalent to coslt .
( 149 ) .
results are closely comparable with ( 52 ) , and our previous methods of treatment will apply .
It is , however , unnecessary to go through all the details of the work , since the definite rals with respect to which appear in ( 149 ) can be derived those in ( 52 ) by } ) , the operation upon the lattel , and then replacing by cosb , from ( 67 ) and ( 70 ) we derive ( 150 ) , OVEP , THE OF where and are the numerical quantities defined by ( 68 ) and ( 71 ) .
Substituting in ( 149 ) we have . .
( 152 ) , .
K. .
( 153 ) , where the notation of the various Fullctions is as in Superposing the system of free waves in which .
, ? .
( 154 ) , we obtain , finally , on inserting the time-factor , .
H. .
( 155 ) , ( 156 ) .
Since these expressions are made up entirely of diverging ) waves , they constitute the solution of the problem where a periodic normal force is applied to the surface at the origin .
alternative form of ( 156 ) , which puts in evidence that part of the vertical disturbance which is most important a great from the , is obtained from ( 75 ) .
Attending only to the " " singular\ldquo ; term , we find .
( 157 ) , and , from ( 149 ) , . .
( 158 ) .
Adding in the system ( 154 ) we have altogether . . . . .
LAMB ON OF , by ( 7 ) , a distance . . .
( 160 ) .
ared with ( 77 ) .
ations Q elliptic , with the same ratio of vertical in case of two dimensions ; but the amplitude diminishes to the usual , of annular 1 the same manner we , in of an internal source of the type ( 142 ) , \amp ; .
, ( 161 ) , the the effect of the depth of the source .
night be derived from the formulae Art .
8 by the iiice .
attempting to give the complete results , would be ) , it be suflicient to ascertain their general form , order of nitude , when .
To take , example , the due to the distortional , if we perform the operation on the second terms of the unnumbered expressions which occur between equations ( 89 ) bove , and then replace by , the more important part of the result in case is by ctor .
result is to be substituted nite int ' with to occur in ; the ) .
terms in of the } ) respectively .
By lethod by totic expansion ( 7 ) of the function 1 is iued , it be shown , these ter1ns are ultimately coml ) with the been restored .
the same , the terms in and to the ultimately compal.able with in case the TREMORS OVER THE OF AN ELASTIC SOLID .
.37 annular Rayleigh waves , so that the latter ultimately predominate .
* lt is also much more apid than in the case of elastic waves divergin from a centre in an unlimited medium , where the amplitude varies inversely as the distance .
14 .
The genel.alization of the results , so as to to an -variation of the source , follows much the course as in Art .
9 .
The full interpretation is however more difficult , so least egards the linor tremors .
The main part of the disturbance , in the case of a local vertical pressure applied to the surface , is obtained by generalizing the formulae ( 159 ) .
These may be written .
( lb2 ) .
Hence , corresponding to an arbitrary pressure , we have .
( 163 ) , where , in analogy with ( 100 ) , ( 164 ) .
The chal.actel .
of the function of represented by the first definite integral in ( 163 ) has been examined by the author various simple forms of , and a similar treatment applies to the second integl.al .
For example , if we take ( 165 ) , ' it is found , on puttmg values of large , and for values of .
( 1 ( 7 ) , ately .
in ( 163 ) , we , ignorin the esidual t , .
the footnote on p. 2 " " gation in Two DimensioIts 'Proc .
Lond. Math. .
quation ( of the cited .
It muy noticed the functious the of ( lli6 ) are intelchanged , with a change of sign , when we sig of S The bymbol , no equired in the sense of , \amp ; c ,38 ON TH PROPA ATION OF wl .
following numerical table is derived from one given ou 155 of the rred to :\mdash ; xtrenleb . . . . . . .
The of and as functions of , in the neighbourhood of the critical epoch , are in fig. which may be ) with fig. 6 .
The corresponding orbit of a surface ticle is traced in .
where the positive direction of is nwards ; it llay be derived by geneous s portion of the curve whose polar equation is $$ .
amplitude ) .
this part of the distul.bance diminishes , with distance from the urce , according to the law plete essions for the obtained by generalizing ( 155 ) and ( 156 ) .
They may be written .
( 169 ) , ( 170 ) , where nd V are the functions tabulated in Art .
9 .
the footnote on TREMORS AN SOLII ) .
Fig. 8 .
Fig. 9 .
The method applied in that Article to obtain a general view of the ] progress of the vertical displacement at any point might be employed again here , the upper lower curves in fig. 4 being combined with auxiliary movable graphs of and considered as functions of .
In the case of a primary impulse of the type , both graphs would have somewhat the form of the lower curve in fig. 8 , the functions practically ( except for a constant factor ) of the type , where in the more important part of the .
Both grapbs , if drawn the scale of fig. 4 or 5 , would be excessively contracted izontally when we are concerned values of large compared with .
Owing to the compensation between positive and negative ordinates in the auxiliary graphs , it is plain that the disturbance expressed by the tegrals in ( 169 ) and ( 170 ) will be relatively very small except has values for which the gl.adient of or is considel'able .
As SSOR HON TION OF regards the displacement the minor tremor will consist of a single to-and-fro oscillation about the epoch , followed after an interval by a somewhat similar oscillation about the epoch , with almost complete quiescence between .
As regards the vertical displacement , there will be a to-and-fro oscillation about the epoch , then a period of comparative quiescence , and finally a gradually increasing negative displacement ( with a slight irregularity at the epoch ) leading up to the main shock , after which there is a gradually decreasing positive displacement .
The expression for the horizontal displacement may be treated in a different manner .
Transforming ( 169 ) we have ( 171 ) .
A rough sketch of the graph of is easily made , and the function is , in such a case as ( 165 ) , one-signed , but its integral with respect to does not converge when the lower limit is large and negative .
The method therefore fails to OVER THE OF AN SOLID .
give us a convenient view of the pro ress of as a function of difficulty is due to the ities of ) to eference IIas already been In attempt , based on the former method , is made 10 represent ( very the whole ress of the horizontal and vertical displacements due to a impulse of the type ( 165 ) at a distance large compared with SUMMARY .
We may now briefly review the principal results of the foregoing , so far they may be expected to throw on the propagation of seismic trelnors over the surface of the earth .
It has been necessary to idealize this problem in yarious ways in order to render it amenable to calculation .
In the first place , the material is taken to be compact and homogeneous , to have , in fact , the properties of the " " isotropic elastic solid\ldquo ; of theory .
Moreover , the of surface is neglected .
Again , instead of a disturbance originating at an internal point , we study chiefly the an impulse applied yertically to the surface .
Under these conditions the disturbance spreads over the surface in the form of a annular wave-system .
The initial form this system will depend on the history of the primitive impulse , but if this be of limited duration , the system actually deyelops a characteristic form , marked by three salient features travelling with the velocities proper to irrotational , equivoluminal , and Rayleigh , respectively .
As the wave-system , thus established , passes any point of the surface , the horizontal displacement shows first of all a single well-marked oscillation followed by a period of coml ) arative quiescence , and then another oscillation corresponding to the epoch of arrival of equivoluminal The whole of this stage constitutes what we have called the ' minor tremor\ldquo ; ; it is , of course , more and more acted the greater the distance from the source , and its amplitude continually diminishes , not only absolutely but also relatively to that of the " " main shock which we identify with the arrival of the Rayleigh wave .
It may be remarked that the history of the minor tremor depends chiefly on integral of the primitive impulse ; the main shock , on the other han follows the time-scale of the primitive impulse , and is affected by every feature of the latter .
* Similar statements apply to the ertical displacement , except that tho minor tremor leads up more gradually to the main shock , being interrupted , howeyer , by a sort ofjerk at the epoch of of equivoluminal ways .
The history of the horizontal and vertical cements , about the epoch of the shock , in the case of a typical impuJse of the type ( 165 ) , is shown in fig. 8 ; * Observational evidence in favour of the existenec of the three in rlisturbance has been collected discussed by R. D. DHAM , ' On the PJ of Motion to Great Distances ' Phil. Trans , 1900 , vol. 194 , p. 135 .
VOL. CCIII .
\mdash ; A. whilst fig. 9 shows the orbit of a surface-particle .
In fig. 10 a sketch is attempted of whole press of the disturbance .
These results are of a fairly definite , but they are based , as has been said , on purely assumptions , and it remains to inquire how far they are likely to be modified by the actual conditions of the eartl ] .
The substitution of an source for a impulse will clearly not aHect the general character of the results at a distance great compared with the depth of the source , although differences of detail in the waye- at the critical epochs will occur , and we can no longer assume that the disturbance is the sane all vertical ) lanes through the source .
Again , the chief qualitative difference introduced by the of the earth ] be that the minor tremor , whose main features evidently associated the outerop of spherical elastic waves at the surface , will be ated directly rough the earth , so that the first two epochs will ( at distances comparable with the radius ) be accelerated relatively to the main shock , which being due to the waves will travel , with the velocity opel to these , ovel the It is a more difficult matter to estimate the nature and extent of the modifications produced by eneity .
It is , ) erhaps , to exaggerate these , for the qualitative effect of ) of elastic would not be serious , and even considerable discontinuities would have little influence if their scale were small compared with the of the primitive ) ulse .
A coyel.ing of loose material over the solid causes only local , though irregular , tions , yith some ) ation of energy .
It must be acknowledged that our theoretical curves differ widely in two respects from the recol.ds of raphs .
In the first place , they nothing corresponding to the successions of ations which are characteristic of the latter .
It would appear that such indications , so far as they are and not instrumental , are to be ascribed to a ccession of primitive hocks , in itself probable enough .
Again , theory gives vertical and horizontal lnovements of the same order of magnitude , and in the case of the ayleigh waves , at all events , where a definite comparison can be made , the vertical amplitude is distinctly the greater .
The obseryations , on the other hand , make out the vertical lnotion to be relatively small .
The difFiculty must occur on almost any conceiyable theory , and appears indeed to be clearly nised by ists , who are thelnselves disposed question the competence of their in this respect .
Cf .
D. The of force ylcig waves on a surface is .
see Professor BROIWICH , This term is used in the same al sense in which in of the " " length\ldquo ; of a wave ] .
Therc is no } the connection , of anything analogo to oscillatory

