]\gt ; On the Deflection of the Vertical by of the Earth 's By HORACE LA1B , F.R.S. ( Received January 25 , 1917 .
) The earliest calculations on the above subject in which the flexure of the earth 's surface is taken into account , as well as the direct attraction of the water , are contained in a well known paper by G. H. Dalwi It was there shown , in a particular case , that the effect would conspire with the attlaction , and would certain assumptions be directly proportional to it .
The latter remark stated as due originally ( in a more general form ) to Sir W. Thomson .
On numerical values it was found that the apparent deflection due to would considerably exceed that due to attraction The whole subject has of late excited attention , to its bearing on observations of lunar deflection of and in a recent paper by TerazawaS the matter is specially considered from the latter point of view .
The present paper , after discussing a few typical problems , goes on ( in S4 ) to examine the effect of one or two considerations which have hitherto been disregarded in such calculations .
It is true that the corrections involved are under some conditions negligible , but they are of theoretic interest , and it is found that at great distances from a load , and therefore in of widely distributed load , they may attain considerable relative importance .
In the first place , owing to the deformatiou of the earth 's surface and the altered distribution of density an additional horizontal component of force on the plumb-line is introduced ; this tends to counteract the attraction of the water .
A more important point is that the influence of on the deformation has been ignored .
In attempting to estimate the effect of ravity it has been found convenient , in order to avoid difficulties not altogether of a mathematical kind , to limit the investigation to the case of incompressibility .
I have also ecCed , in the first instance , the disturbance in the field of gravity itself , due to the load and the tion , so far as this affects the strains .
When subsequcntly the alteration of the 'Brit .
Ass .
Reports , ' 1882 ; 'Scientific Papers , ' vol. 1 , p. 444 .
The relation was pointed out independently by Chree , 'Phil .
Mag. ' ( 5 ) , vol. 43 , p. 177 ( 1896 ) .
Love , ' Some Problems of Geodynamics , ' p. 88 ( 1911 ) .
' On Periodic Disturbance of Level Arising from the Load of Neighbouring Oceanic Tides ' Phil. Trans , vol. 217 , p. 35 .
Prof. H. Lamb .
On the Deflection of the field is taken into account a curious point arises .
For mathematical simplicity the ' earth\ldquo ; has been treated , as is usual in such investig.ations , as flat and infinitely extended .
It appears that if this were the case the surface would be unstable , whatever the degree of rigidity , for disturbances exceeding a certain critical wave-length .
This wave-length is , however , enormous , and reason is given for the view that inferences can still legitimately be from our results as to the character of the effects actually produced .
1 .
For the present purpose , the earth is regarded as a semi-infinite isotropic elastic solid , extending indefinitely on the lower , positive , side of the plane .
If a load of surface density ( ) , ( 1 ) where , denotes distance from the , be placed upon it , the usual theory , in which elasticity only is taken into account , gives a surface depression , ( 2 ) and a dilatation ( ) , ( 3 ) where denotes the rigidity , and Poisson 's ratio .
* The gravitation potential due to the load ( 1 ) is , ( 4 ) where is the gravitaGion constant , and the upper or lower in the index is to be taken according as is positive or negative .
For the two values of satisfy Laplace 's equation ; they are continuous at the surface ; and they give the proper discontinuity in the value of .
The surface potential is therefore .
( 5 ) The apparent deflection of the plumb-line due to tilting is , and the deflection due to attraction of the load is .
The ratio of the latter to the former is , ( 6 ) where denotes the radius , and the mean density , of the earth .
Since this ratio is constant and independent of , we infer that it is the same for ' Proc. Lond. Math. Soc vol. 34 , p. 278 ( 1902 ) .
The formula are by the use of , in place of Vertical by Loading of the Earth 's all symmetrical distributions of load , and therefore for a concentrated load , and therefore for all distributions whatever .
This verifies Thomson 's remark already referred to .
If we take for numerical illustration* in C.G.S. units , we find .
( 7 ) Iu the case of incompressibility we have , with the above value of .
( 8 ) The method of from ( 1 ) to the case of any symmetrical distribution of load is explained in the paper referred to .
To deduce the of a concentrated load at the origin we multiply the various expressions by , and integrate from to .
Since , however , the deflection due to the attraction of is we have once , ( 9 ) in virtue of the } ) referred to .
2 .
In a case specially discussed by Terazawa the load is supposed to be distributed uniformly over a circular area .
This can be treated very simply with the help of a formula given by errers .
He proves ( in effect ) that the potential of a circular disc of radius and surface density , ( 10 ) at external points in its plane is , ( 11 ) the lower limit of integration being In the present case we have Putting , we find , ( 12 ) * It is very uncertain what are the best representative values to take for and ( the density ) , on account of the actual variation with depth .
lf we put the above value of makes the velocity of propagation of transverse vibl.ationsagl.ee with the value found by seismologists for the upper strata , viz. , 4 km .
per second .
' On the Potentials of Ellipsoids . . .
of Variable Densities ' Qunrt .
Journ. Math vol. 14 , p. 1 .
the case , also treated in a different manner by Terazawa , the integral reduces ; thus Prof. H. Lamb .
On the Deflection of the in the notation of elliptic integrals , the modulus being . .
For the attraction we derive the simple expression .
Tables of and , abridged from Legendre 's , are included in several modern collections .
The object of Terazawa 's paper was to form some estimate of the possible effect of the pel.iodic tidal loading and unloading of the Atlantic area on Michelson 's observations of lunar deflection of gravity at .
The radius of the circle was taken as 2000 km .
, and the distance of the point at which effect is to be estimated as 3000 km .
If we put we find from ( 14 ) on reference to the Tables .
( 15 ) To find the deflection of the pendulum by attraction we must multiply this by , where is the density of the water , and the tidal elevation .
Assuming , we find - , ( 16 ) which is Terazawa 's result , obtained by a different process .
The tilting effect , obtained by multiplying by , is Since tidal elevation in one part of the ocean is compensated by depression elsewhere , it may be worth while to notice the case where the tidal elevation in a circular basin is supposed to vary as the distance from a nodal line .
It follows from another formula of Ferrers that the potential , at points in its plane , of a disc whose surface density is , is , ( 17 ) the lower limit being .
If we put , this is equivalent to where as before .
From this we find Vertical by Tidal of the 's Surface .
If height of the water be , the radial deviation of the plumb-line due to attraction is found by multiplication by .
If 1 , , I find - The tilting effect is therefore , on the previous assumptions , .
( 21 ) 3 .
Two-dimensional problems of the above kinds can be treated ] by a simpler analysis , but we may convenienlly use the result contained in ( 9 ) .
The loads are now supposed applied on infinitely bands of the surface , parallel ( say ) to the axis of .
Unless the total load ative ) is zero , the integrals which give the surface depression and the potential are ; but we may still draw inferences to the produced by bands of great length at distances small compared with the length .
Thus , for a linear load of amount per mit length , we have , ( 22 ) by the theory of the attraction of an infinite straight line , and therefore .
( 23 ) If we rine a band of breadth to carry a load per unit area , we have for whence the value of follows as before .
A more interesting case is where the load varies as the distance , positive or negative , from the medial line .
For the sake of comparison with a problem discussed by Darwin , depression within the band , as well as at external points , is considered .
The load per unit area being , for , we find from ( 23 ) If this gives .
( 26 ) For we have , taking the " " principal value\ldquo ; of the integral in ( 25 ) , .
Prof. H. Lamb .
On the Deflection of the Hence , for ( 28 ) and , for .
( 29 ) The arbitrary constants of integration have been adjusted so as to make vanish for , and to establish continuity at the edge .
Since must be an odd function , the above expressions will hold also for ative .
The graph of is shown in the figure .
The tilting effect at external points is given by ( 26 ) ; from this the direct attraction of the load can be 4 .
Besides the direct attraction of the load on the pendulum there will be a horizontal force due to the configuration of the solid earth .
If we assume the formulae ( 2 ) and ( 3 ) for the deformation , the potential is found to be , ( 30 ) which gives in the case of a concentrated load a deflection ( 31 ) The effect on the plumb-line is therefore equivalent to a repulsion varying as the inverse distance , whereas the attraction of the load varies as the inverse square .
It is not worth while to give the proof of the formulae , for the effect could only become at very great distances from the load , and in these regions the deformation , which has been assumed to be iven by ( 2 ) , is seriously affected by the action of gravity , which is next to be considered .
The effect here referred to is included implicitly in the amended results .
5 .
When we proceed to examine the effect of gravity in modifying the deformations , we are met by difficulties relating to the state of initial stress of the earth .
To avoid these the material is here assumed to be incompressible , and the initial stress to consist of a uniform hydrostatic pressure* .
( 32 ) *Cf .
Rayleigh , " " On the Dilatational Stability of the Earth 'Roy .
Soc. Proc vol. 77 , p. 486 ( 1906 ) ; 'Scientific Papers , ' vol. 5 , p. 300 .
by Tidal Loading of the Earth 's Surface .
It is also assumed for the present that in calculating the strains we may neglect the alteration in the distribution of gravity due to them and to the load .
It is convenient to begin with the two-dimensional case .
On the above hypothesis the elastic equations are , in the usual notation , ( 33 ) where denotes the excess of pressure over its initial value Assuming for a typical solution , , ( 34 ) we have .
( 35 ) The condition of , ( 36 ) gives A-B , ( 37 ) whilst the condition of vanishing tangential stress , , ( 38 ) at the surface , gives Hence , so that .
( 40 ) It may be noticed that vanishes at the surface .
If be the surface-density of the load , the surface condition is , ( 41 ) VOL. XCIII .
Prof H. Lamb .
On the Deflection of the to be satisfied for .
Since , this becomes to be satisfied for .
Hence if we have .
( 44 Thus where To deduce the effect of a concentrated load of density M we , and integrate from to .
Thus , ( 47 ) or ?
in the ordinary notation of the sine-and rals , , Ci .
( 49 We may notice that the value of is now finite except for .
Fror ( 48 ) we deduce Ci , ( 50 which now takes the place of ( 23 ) .
As regards the numerical value of we have , taking km .
( 51 In any practical application we are therefore concerned only with moderat values of The following table ives in Column I values of the function , ( 52 based on Glaisher 's of the standard .
In Column Il As to the equality of the two integrals , see the author 's 'Hydrodynamics ' ( 1916 ) Art .
243 , where other 1eferences are given .
Phil. Trans p. 367 1870 ) .
An abstract is given in Dale 's 'Five-Figure of Mathematical Functions , ' London , 1903 .
Vertical by Tidal Loading of the Earth 's Surface .
this function is iplied by , and the result subtracted from unity .
This shows the ratio in which the tilting effect is diminisbed from its value as given by the formula ( 23 ) , where gravity was neglected .
At small distances the change is slight , but at a distance of 1000 km .
, for example , the tilting is reduced by the action ravity to the extent of 30 per cent. It is to be remarked that on the present suppositions the residual horizontal attraction on the pendulum , when allowance is made for the altered attraction of the solid , bears the sa1ne constant ratio to the effect as in Art .
1 above .
Becurring to the distribution ( 43 ) , and assuming the potential due to the load and to the change of configuration of the solid ( 53 ) we have , on account of the discontinuity in , ( 54 ) whence ( 55 ) Thus .
( 56 ) Since this does not involve , the ratio will hold for all two-dimensional distributions of load .
With the former numerical data its value has been found to be 6 .
A similar investigation can be made for the case of a load concentrated at a point .
The elastic equations are , ( 57 ) involvin also .
( 58 ) If denote radial displacement in a horizontal plane , so that ( 59 ) Prof H. Lamb .
On the Deflection of the the equations become , in the case of symmetry about the origin , with .
( 61 ) Assuming ( mr ) , ( 62 ) we find The condition of incompressibility , leads to A-B \mdash ; whilst the condition tangential stress at the surface , .
( 66 ) gives Hence , so that .
( 68 ) The surface-condition has the form ( 42 ) .
If the load be ( ) ( 69 ) we find , ( 70 ) where has the value ( 46 ) .
To pass to the case of a concentrated load at the origin , we put Mm , and integrate from to .
Thus , ( 71 ) * A proof of the equality of the two integrals last written is given in the Appendix to this paper .
Vertical by Tidal Loading of the 's Surface .
303 An asymptotic expansion of the latter integral is easily obtained , but has hardly any application from the present of view .
For moderate values of we have recourse to the formula* S where is the Bessel 's function " " of the second kind\ldquo ; defined by .
( 73 ) Hence .
The function has been tabulated by B. A. under the designation , where stands for or being Euler 's constant .
With this assistance the following Table has been constructed .
Column I the values of the function ( 74 ) ; in Column II these are multiplied by and subtracted from unity .
The result gives the ratio in which the surface depression is diminished by gravity:\mdash ; The tilting effect is From ( 74 ) we have , ( 76 ) where KI is written for .
The function has also been tabulated by Smith under the designation .
The following Table gives in *Cf .
'Hydrodynamics , ' Art .
194 .
This formula occurs in Nielsen 's 'Handbuch der Cylinderfunktionen , ' Leipzig , 1904 , p. 238 , but the derivation is different .
The ascending series occurs in another connection in Rayleigh 's 'Theory of Sound , ' Art .
302 , where it is denoted by Nielsen denotes it by , and writes for the function on the -hand side of ( 73 ) .
The notation in the text is employed by H. Weber .
'Phil .
Mag. ' ( 5 ) , vol. 46 , p. 122 .
Prof. H. Lamb .
On the Deflection of the Column I the values of the function in ( 76 ) ; in Column Il these are multiplied by and subtracted from unity .
The results show the ratio in which the tilting effect is reduced by gravity:\mdash ; It is easily seen that the residual horizontal attraction will bear to the tilting elfect the same constant ratio as in the two-dimensional case of the preceding section .
7 .
It remains to take into account the fact that the gravitational field is altered by the presence of the load and the deformation of the surface , and the strains are thereby , to some extent , modified .
Taking first the two-dimensional case , the equations ( 33 ) are replaced by , ( 77 ) where is the potential due to the load and to the change of configuration of the surface .
If we assume ( 78 ) as before , and mx , ( 79 ) we have , writing for in ( 40 ) , .
( 80 ) The load being , ( 81 ) the surface-condition ( 42 ) gives .
( 82 ) Since the value of just above the surface must exceed he value just below by , we have also .
( 83 ) Vertical by Tidal Loading of the Earth 's Hence , ( 84 ) and H. ( 86 ) Hence , putting , ( 86 ) we find , ( 87 ) ?
( 88 ) It appears that for the equilibrium would be neutral ; i.e. deformation wave-length could subsist independently of any external load .
For greater wave-lengths the equilibrium would be unstable .
* This gravitational instability of am infinitely extended plane surface may be unexpected , but it is easily verified in the case of a plane sheet of water , where .
It is evident that in the absence of the earth 's attraction and of rigidity the mass would lend to cunlulate .
The introduction of ordinary gravity , and even of rigidity , merely impose a limit to the at which instability sets in .
This limit is however very .
From ( 86 ) we find .
( 89 ) With the previous numerical of the constants I find km .
, , ( 90 ) whence km .
, or nearly twice the circumference of the earth .
This indicates that the paradox has arisen through our ynoring the curvature and restricted dimensions of the earth .
Waiving this difficulty for the moment we have , ( 91 ) .
( 92 ) * This is seen most by directly investigating the effect of a surface essure P instead of a material load .
We should find , which has the same sign as , or the opposite , according as Pror .
H. Lamb .
On the Deflection of the Hence , for a concentrated linear load , * ( 93 ) .
( 94 ) The second definite integral in these expressions is indeterminate to the extent of additive terms of the form A Its " " principal value\ldquo ; is .
( 95 ) The adoption of the principal value of an indeterminate integral may appear somewhat arbitrary , and it may be questioned whether results obtained in this way have any physical significance .
The consideration of the following modified problem may perhaps serve to justify the procedure .
Suppose that we have an endless series of alternately positive and negative loads at the points These may be represented as the limit when , of ' ( 96 ) For , summing the series , we have , ( 97 ) the limit of which is zero , except when is zero or a multiple of .
Also , if we have , ( 98 ) the limit of which is , however small is taken .
Similarly we should find , ( 99 ) * The term arises See the Appendix .
by Loading of the Earth 's ce .
and so on .
Now , to ( 91 ) we have ' ( 100 ) where ) , , Here , ( 101 ) the limit of which is .
In the terms we may put at once .
It is shown in the Appendix that for values of ) ' between and , ( 102 ) .
( 103 ) It is not difficult to see that if tely lar less than , whilst is small , the result tends to the form which is equivalent to that above obtained .
If we impose periodicity there is in the present no indetcrminatemess , and no instability so as .
If a earth , periodicity is essential in the solution , and there is no mathematical absurdity in regarding the points , for instance , as being the same place .
The curvature would , of course , involve some modification of the elastic equations .
I have made some calculations illustrative of the formulae ( 93 ) , ( 94 ) .
To simplify the numerical work I have assumed , which is not quite consistent with ( 90 ) , but is equivalent to diminishing the value of by about one-tenth .
In Column I below the tilting effect is compared with that given by the formula ( 23 ) where gravity was altogether neglected .
Column lf gives the ratio of the total deflection to the value given by ( 23 ) .
It is remarkable that the mubers in Column II only diffel by about 1 per cent. from those which we should obtain by Prof. H. Lamb .
On the Deflection of the I. II .
I. II .
I. II .
I. II .
I. II .
ax .
I. 0$ Column II of the Table on p. 301 by , which is now the proper ratio equired ( instead of ) to convert the tilting effects there considered into total defle jtions .
This result might have been foreseen .
to the typical and ( 92 ) we note that .
( 105 ) Hence we find after a few reductions mx. ( 106 ) If the last term in the bracket is small , and the expression is almost in a constant ratio to that which occurs in ( 50 ) proyided be written for The deflections at given are , however , diminished from the values based on Art .
5 above , where undisturbed gravity was alone regarded in calculating the and at great distances the effect is considerable .
Since on the present reckoning , ' any line in the above Table relates to a distance four-fifths as great as that implied the corresponding line of the Table p. 301 .
To take a numerical example , let , which means a distance of about 2300 km .
The total deflection is then only 0507 of that given by a purely elastic theory ; on the basis of Art .
5 the ratio would be 8 .
The investigation for the of a load concentrated at a point of the surface follows a parallel course .
We should find , ( 107 ) .
( 108 ) Vertical by Tidal Loading of the Earth 's Surface .
The principal value of the indeterminate integral is , as shown in the Appendix , .
( 109 ) From ( 107 ) we derive - , ( 110 ) with a corresponding formula for .
The former of these integrals is evaluated in ( 74 ) .
As regards the latter we have by differentiation of ( 109 ) . .
( 111 ) The following Table , calculated on the numerical assumptions adopted in the preceding section , gives in Column I the ratio which the surface depression ) bears to ' the value given by the first term in ( 107 ) .
Column II gives the ratio of tilting effect to that given by the first term of ( 110 ) , whilst Colu , mn III gives the ratio of the total deflection to the same term .
It is assumed that the use of the " " principal values\ldquo ; of the indeterminate integrals can be justified by considerations similar to those adduced in the two-dimensional case .
The comments made at the end of the preceding section apply equally well to these results .
Thus , if , the total deflection is of the tilting effect given by a purely elastic theory , whereas on the assumptions of Art .
6 the ratio would be APPENDIX .
The following are proofs of some of the formula emplo , yed in the paper .
I do not expect that they are altogether new , but the literature of Bessel 's functions ( for instance is so diffuse , and the proofs so entangled , that it is Prof. H. Lamb .
On the Deflection of the sometimes difficult to find a concise demonstration , even of an important formula .
( A ) The equation .
( 112 ) is satisfied by ( 113 ) and also by .
, ( 114 ) being of course assumed to be positive .
For ( 113 ) makes , ( 115 ) by partial integrations , and in virtue of Bessel 's equation .
makes . .
( 116 ) The difference between the integrals in ( 113 ) and ( 114 ) must therefore be of the form , where is defined by ( 73 ) .
Now the leading terms in the asymptotic expansions of and for large values of are* and , respectively , whereas either definite integral is evidently at most of the order .
It follows that , and therefore , ( 117 ) as stated in ( 71 ) .
Hydrodynamics , ' Art .
194 .
Vertical by Loading of the Earth 's Surface .
( B ) Consider the integral , ( 118 ) where .
Let this be taken round the contour of the infinite region above the axis of , the point being excluded by a small semicircle described about it .
The infinitely distant parts of the contour contribute to the final result , and we find , ( 119 ) or Hence , separating the real and imaginary parts , .
( 122 ) The llatter formula has been assumed in ( 109 ) .
( C ) If in ( 121 ) we write for and integrate with respect to between the limits and , we obtain , ( 123 ) since ) .
( 124 ) Again , writing / ' for , in ( 122 ) , and from to , we have , ( 125 ) in virtue of , and of Mehler 's formula .
( 136 ) By combining the formulae ( 123 ) and ( 125 ) , we obtain another proof of ( 74 ) .
( D ) To sum the series in ( 102 ) we take the integral round the contour of the infinite region lying to the of the axis of Deflection of the Vertical by Tidal the poles of the rand being excluded as usual by small circles drawn round them .
These poles are at the points , 3 , 5 , . .
( 128 ) and the value of the integral taken in the negative direction round any one of them is .
( 129 ) lf be positive , and , as we will suppose , the part of the integral due to the infinitely distant portions of the boundary will vanish .
The sum of the series of terms of the type ( 129 ) is therefore equal to .
( 130 ) This leads to the formula ( 102 ) .
In the case of the integral , ( 131 ) there is an additional pole , which gives a term The formula ( 103 ) easily follows .

