]\gt ; On Waves .
In each case the time of was two minutes .
The results are given in terms of the total charge in electromagnetic units contained in the specimen when the original difference of potential between the surfaces was unity .
In the first column times measured from the instant at which both surfaces are connected to the earth are given .
These results are plotted as figs. 20-22 .
For these spec imens the values of are ( after two minutes charging ) Glass ( test-tube ) .
Paraffin .
Ebonite .
The author 's thanks are due to the Managers of the Royal Institution for placing the resources of the Davy Faraday Laboratory at his disposal , and to the Director , Prof. Sir James Dewar , for much helpful kindness and consideration .
On Water Waves by a Local on or beneath the By K. TERAZAWA , Ri-Gakushi in the Imperial University of Tokyo .
( Communicated by Sir Joseph Larmor , F.R.S. Received June 28 , 1915 .
) The classical problem of waves produced in deep-sea water by a local disturbance of the free surface has been investigated by Prof. H. Lamb* in a very able manner .
He completed the theory of wave propagation in one dimension and in two horizontal dimensions when the initial disturbance is concentrated in the immediate neighbourhood of a line or a point , assuming that Fourier 's double integral theorem can be applied in ) a case .
In this paper I venture to discuss the same problem , especially the oscillations at the centre of the disturbance , following Prof. Lamb 's method , in cases where the initial disturbance is spread over a certain ext , ent of the free surface ; and , as an application of the general solution , I propose to treat the case in which the initial disturbing source is situated at a finite depth from the free surface , where the surface wave is produced by an explosion like that of a mine under water .
S1 .
Supposing the water to extend to infinity , horizontally and downwards , and taking the axes of and on the undisturbed free surface and that of ' Lond. Math. Soc. Proc vol. 2 ( 2 ) , p. 371 ( 1904 ) , or ' Hydrodynamics , ' third edition , SS236-239 , 251-262 .
VOL. XCIL\mdash ; A. Mr. K. Terazawa .
On Water vertically downwards , we have , on the assumption that the motion is infinitely small and irrotational , , ( 1 ) where denotes the pressure , the density of water , the constant of gravity and the velocity-potential which satisfies the equation .
( 2 ) If ' denotes the depression of water-level at the point and at time below undisturbed surface , then the pressure condition to be satisfied at the free surface , to be small , is , ( 3 ) and the kinematical condition is .
( 4 ) Hence , for , we must have or , if the time factor be .
( 5 ) The solution of equation ( 2 ) , which makes when , is , in cylindrical co-ordinates , , ( 6 ) where is Bessel 's of th order and is any positive number .
From equations ( 5 ) and ( 6 ) we get .
( 7 ) The modes of irrotational motion of water by surface disturbance present themselves , as Prof. Lamb points out , in two forms : ( 1 ) by an initial displacement of the surface , without initial velocity ; ( 2 ) by an initial impulse applied on the surface , without initial surface displacement .
The most general case is a combination of these two .
I. S2 .
As the typical solution for the case of initial rest we take .
( 9 ) caused by Local Disturbance .
The water is supposed to be unlimited in extent , and hence there is no limitation to the length of waves , but the motion will be the result of the superposition of waves of infinite variety of length .
Therefore to obtain the general expression which embraces the superposition of all such solutions we must make use of the double integral theorem of Fourier 's type , ( 10 ) with some conditions concerning the function .
Thus , corresponding to the initial conditions , , ( 11 ) we have , ( 12 ) , ( 13 ) where ; ( 14 ) for these expressions clearly satisfy the conditions specified above .
As to the disturbing function , we can snme its form in several ways , under the condition that the theorem ( 10 ) must rernain valid .
Now we consider the case where the initial elevation of the surface is spread over the whole surface according to the law .
( 15 ) This is the simplest form which satisfies the ouble integral theorem .
By taking the value of A sufficiently small , we may consider that it approaches , when , to the case where the initial displacement of reat amount is confined to the neighbourhood of the origin .
* It been kindly pointed out to me by a referee that this statement requires correction and precision .
However small A may be taken , the initial elevation is spread niformly along the radius , but is not concentrated in the sense of Cauchy and Poisson , though it is to be expected that only the central part when the height is considerable can ntribute sensibly to the amplitudes of the waves .
It might appear that the total initial elevated volume , which is expressed by the integral when should be finite .
But it seems to me that this condition is not a necessary one for the validity of the integral theorsm ( 10 ) .
This remark applies to both the initial distribution and that is treated in S4 .
But for the initial distribution that is treated on p. 69 there can be no doubt , for it is of finite total amount ; Mr. K. Terazawa .
On Deep-Sea Put the expression ( 15 ) into ( 14 ) , then it appears , by a well known formula , that ( ) ( 16 ) Whence the eneralised expressions for and will become \ldquo ; ( 17 ) Acos ' ( 18 ) By expanding and into series , we , ( 19 ) Acos ; here is eliminated by the relation The definite integral contained in ( 19 ) and ( 20 ) can be evaluated as follows:\mdash ; By using Hansen 's form of Bessel 's function and the integral formula , we hnd that If we put then we have and when b vanishes it is concentrated at a point , the result )eing then the well known form of Cauchy and Poisson .
A main object of this paper is , as stated , to trace the effect on the waves propagated to a distance , and especially on the course of the disturbance at the centre , of a broadening of initial disturbance from a mere point to a wider region .
This has also the advantage that we do not have to proceed to a Iimit .
] * Gray and Mathews , ' Treatise on Bessel Functions , ' p. 79 .
caused by Now , define the function by being the zonal harmonic of nth order , then the integral expression oi , which corresponds to Laplace 's integral for , will be Therefore the required integral becomes .
( 22 ) The result of this , in the case of , was found by Prof. .
W. Hobson .
* In this case , if we remember that , where is the associated function defined by we have ( 23 ) in agreement with his result .
Hereafter we will use ( 22 ) for and ( 23 ) for Hence , putting these values ( 22 ) and ( 23 ) into equations ( 19 ) and ( 20 ) , we get and .
( 25 ) 'Lond .
Math. Soc. Proc vol. 25 , p. Mr. K. Terazawa .
On Water The values of and are as follows:\mdash ; ' ( 26 ) , For the case of symmetry about the origin , i.e. , the first summations in ( 24 ) and ( 25 ) disappear , and therefore we have simply , ( 27 ) , ( 28 ) similar expressions to those which were found by Prof. Lamb for the case of complete concentration of the source .
The next simplest case is that of , though it may not be very important in practice .
In this case we obtain , ( 29 ) ( 30 ) These solutions do not give any information as to what takes place at the immediate neighbourhood of the origin .
Not only the initial data but the displacement at any time will be infirlite at the origin ; such a point is excluded in our fundamental assumption .
The appropriate initial data and the solutions to illustrate this point will be given presently ( S4 ) .
S3 .
The series ( 25 ) is not convenient when we deal with the case in which has large values , since it converges rather slowly in such a case .
The suitable expressions in this case , at least when and , can be obtained by a similar method to that employed by Prof. Lamb for the case of complete concentration of the source .
If we put , ( 31 ) then ning to ( 18 ) we have -A .
( 32 ) Waves caused by Local Disturbance .
By using the integral expression for and putting with ; we get .
( 33 ) Now expanding into a s.eries and using the formula ( 34 ) where , and ; we find corresponds to .
Hence where The functions and a , re closely connected with Fresnel 's integral as is well known .
If we put ( 36 ) Mr. K. Terazawa .
, On then we shall have* sirl If is a large quantity , then is plso large , and the functions and can be expressed as semi-convergeuC series of the form ( 38 ) If we neglect and higher powers of it.in these series , we have , from ( 36 ) , approximately Putting , or , ( 39 ) and , ( 40 ) can be written iu the form .
( 41 ) From ( 40 ) it that when is an odd number , and when is an even number .
If we find the values of and V in proper form , we can obtain the expression for by ( 41 ) and ( 32 ) .
The special case was discussed by Prof. Lamb and it was found approximately that , ( 42 ) Lamh , ' Hydrodynamios , ' p. 366 .
Lord Rayleigb , 'Scientific Papers , ' III , p. 129 .
Waves caused by Disturbance .
and therefore we get .
( 43 ) For the case and ( 44 ) The value of the integral contained in the expression of cannot he found very easily .
Since we see that and similarly ; and if we remember that then we can evaluaGe these integrals , following the method which is fully explained by N. Nielsen in his book on Cylinder-Functions .
* I calculated these rals after this method , and found that For large values of we can use the asymptotic expansions of Bessel 's functions , and , to the degree of approximation above mentioned , it is sufficient to retain only the first term of them .
Thus we may take .
Putting these values in the above formulae ( 45 ) , then by ( 44 ) , after simplifying , we have .
( 46 ) Substituting this for in ( 32 ) , we obtain .
( 47 ) 'Handbuch .
Cylinderfunktionen , ' SS 77 , 81 , 82 .
Mr. K. Terazawa .
On Though we have been considering in the above only the terms involving , it must be understood that the same analysis could be applied to the terms containing , if the initial data be so given .
a proper assumption to illustrate what occurs at the origin , take the disturbing function of such a form as instead of ( 15 ) .
This function does not violate the double integral theorem ( 10 ) .
Supposing the value of to be sufficiently small , this approaches , at a distance from the origin , to the function assigned in S2 , and yet it is finite at the origin .
Therefore we might take , even in this modified assumption , the solutions in S2 as those for the point at which has a moderately large value .
For the case of symmetry the function defined by ( 14 ) becomes This integral can be effected by using the double integral theorem ( 10 ) in a special case , and is found easily to be ( 49 ) supposing to be positive .
Hence we have the expressions for and from ( 17 ) , ( 18 ) , ( 50 ) .
( 51 ) Making use of the integral formula ( 23 ) , we get the following solutions , in place of ( 27 ) and ( 28 ) , ( 52 ) with and with , At the origin , we have The is valid for only when Or , as a special case of the integral of N. Sonine ( ' Math. Ann vol. 16 , p. 50 ) , we get the above result , putting in his formula Waves caused by Disturbance .
and since ) , the equation ( 53 ) gives .
( 54 ) If we put , for the present , then the above expression can , referring to , ( 36 ) , be put in the form and , therefore , using the integral expressions ( 37 ) for the functions and we have Or , putting again , for the sake of simplicity , , ( 55 ) it follows that .
( 56 ) The expressions ( 54 ) and ( 56 ) for the displacement of the point which was initially situated at the origin are equally inconvenient to discuss its subsequent motion , but it is not very difficult to obtain the general feature of it from the latter form .
For the small value of the value of the function can be obtained by expanding the exponential function in the integral sign and integrating it term by term .
For the large value of the asymptotic expression for the function is ( 57 ) Therefore the course of the ordinate at the origin tends approximately to . .
the ( 5S ) when is great .
* E. T. Whittaker , 'Modern Analysis , ' ; A. E. H. Love , 'Phil .
Trans vol. 207 ( 1907 ) , p. 196 .
Waves caused by a Local Disturbance .
The highest point in the curve corresponds to the initial prescribed tion at the origin .
As increases from zero the elevation decreases and reaches once to the value nil ; and then begins the depression of the surface at the point under consideration .
After that there is only one maximum of the depression , the amount of which is smaller than that of the initial elevation .
Then it decreases more and more , very slowly , until , after an infinitely long time , it takes the limiting value zero .
The number of zero-point and maximum-point can be determined without using the above We can assi several assumptions to the disturbing function to illustrate the movement at the origin .
For example , if we take the form ( 59 ) instead of ( 48 ) , then it appears that , ( 60 ) which can be found in a similar way as before .
In this case and , performing the integration , we have ( 61 ) where and are defined in ( 53 ) .
At the origin this becomes ( 62 ) Concerning this series and also as to the motion at the origin , we can make a similar interpretation to the former case .
II .
S5 .
Quite analogous treatments can be employed in the case where the initial condition is an impulse applied on the free surface .
The typical solution is , ( 63 ) .
( 64 ) This formula gives , in fact , the Cauchy-Poisson series as a lirst pproximation when is taken very small ( Lamb 's 'Hydrodynamics , ' p. 408 , equation 20 ) .
See the footnote on p. 69 .
] Mr. K. Terazawa .
On Water Corresponding to the initial conditions at the free surface , the general solution is ( 66 ) ; ( 67 ) where .
( 68 ) At we assume the initial impulse is given by , ( 69 ) then the value of becomes corresponding to ( 16 ) in the former case .
The explicit form for the solution can be obtained by performing the operation upon that of the former case .
Thus from and ( 25 ) we get , ( 70 ) where , ; and .
( 71 ) Specially for , from ( 27 ) and ( 28 ) , , ( 72 ) .
( 73 ) and for , from ( 29 ) and ( 30 ) , Waves caused by a Local Disturbance .
Now , when is a large quantity , we can make use of the function defined .
Thus A .
Referring to the expressions for and found in ( 42 ) and ( 46 ) , we get ( 76 ) for ; and ( 77 ) for S6 .
As in S4 , we take the function expressing the initial impulse of the form ( 78 ) to illustrate the history of the point which was situated initially at the origin .
For the case of symmetry about the origin , the function becomes Thus we have , ( 79 ) .
( 80 ) Or , performing the operation upon ( 52 ) and ( 53 ) , we obtain ' ( 81 ) , ( 82 ) .
in which as before .
At the origin , becomes .
( 83 ) .
K. erazawa .
On This series call be put in a similar form to that ill S4 ; or rather directly from ( 56 ) it follows that , ( S4 ) , for simplicity , it is ) .
( S5 ) large values of , using ( 57 ) , we can take for its asymptotic expansion of the form ( 86 ) From these expressions we can make out the general feature of the movement of the point under consideration .
By the aid of the Table I we can draw the approximate diagram which represents the isplacement at the origin , units of scales being modified by ( 85 ) .
For , we have and as soon the impulse is applied on the free surface the point at the origin gains suddenly a finite velocity and to move downwards .
It will be seen , from the above figure , that at about the depression becomes nlaximum , at about it takes the value zero and then begins the elevation of the surface , and it takes the maximum value at about , after that it decreases slowly until it tends to its linliting value zero , and the point comes to If we assign another assumption to the initial impulse like that expressed Waves caused by a Local Disturbance .
in , it will probably take place by a similar mode of motion , but in a somewhat complicated form .
III .
S7 .
Now we come to the problem of surface waves caused by an explosion which takes place at a finite depth from the free surface .
* In this case , we might take the value of the impulse , i.e. that of , given by the explosion as .
the initial datum , but the mathematics would be complicated and it might not give any concrete results .
The following consideration , however , may serve to treat the problem without touching on the difficulties .
Suppose that an explosion under water sends out a pressural pulse of the of a wave of sound , propagated by compression at speed , and leaving unruffled water in its wake .
As each ray reaches the free surface it is reflected totally , as required by the condition that the impulse is null at the surface , while a surface normal velocity caused by the pulse remains and determines a hydrodynamic flow throughout the water .
We may take the velocity of the propagation of the pressural pulse to be infinite , then the ima , method in will be applichle as follows .
Let the point which the explosion occurs be at the depth from the free surface and on the axis of .
Suppose the initial value of due to the source at .
to be , where is the distance of any point in the water from , then we must consider a fictitious negative source of equal strength at the image of with regard to the surface , to fulfil the condition that the initial impulse is zero at the free surface .
The initial value of should then be , where This value of gives the initial normal elocity ( 87 ) over the free surface and it dies away of itself .
Thus the problem is , transformed to one of the surface motion , being given the initial surface normal velocity ( 87 ) .
If we take * This problem and the next following were undertaken mainly to illustrate in a general way the nature of the tidal waves produced by a bmarine earthquake , with its source at a point or on a long straight fault in the strata .
The source of the explosion is supposed to be so deep or its force so gentle that the surface of the water is not broken by the ejection of a column of water .
VOL. XCII.\mdash ; A. Mr. K. Terazawa .
On Water as the typical solution , then the general solution corresponding to the initial data ( 87 ) will be ( 88 ) with .
Since , by ( 60 ) , we have ( 89 ) and accordingly .
( 90 ) These are the same essions as ( 79 ) and ( SO ) up to the constant factor .
Therefore putting we get , ( 91 ) and .
( 92 ) At the point just above the explosion , that is at the origin , , ( 93 ) we have ) ( 94 ) Comparing ( 94 ) with ( 84 ) , we see that the curve representing the displacement at the is quite the same as in .
( 2 ) , only inverted in form and the unit of ordinate being different .
If we take as the values of which give the maximum , zero and the minimum of the displacement , and , then the times at which they occur will be given by respectively , in which , and t3 are expressed in seconds and in feet .
Conversely , ?
we can measure any of them , then the depth can be calculated approximately by using the above relations .
Waves by a Local Disturbance .
S8 .
We will conclude this paper by discussing the explosion problem in a horizontal canal of infinite depth and infinite length .
Take the axis of along the canal and on the undisturbed free surface of water , and that of vertical downwards as before .
If we suppose that all the circumstances are independent of , we must consider the initial value of iven by the explosion to be of the form .
Therefore , considering the image as before , we have the effective initial normal velocity at the free surface ( 95 ) in place of ( 87 ) .
Corresponding to this initial velocity , the general solution wiIl be in which .
By a well known formula we get and consequently kx .
( 96 ) Expanding and then making use of the integral formulae , it may be , ( 97 ) where , .
At the point just above the explosion , i.e. along the axis of This series can be put in the form , ( 98 ) in which is defined by ( 93 ) .
* Since this work was finished my attention has been called to a paper by Prof. Lamb in the 'Annali di Matematica , ' vol. 21 , p. 237 1913 ) , where the surface waves that accompany a cylinder travelling uniformly transverse to its length under water are investigated .
H. Weber , ' Die part .
Diff.-Gleich .
, ' vol. 1 , p. 43 .
Mr. K. Terazawa .
On Deep-Sea Water For large values of , if we use the asymptotic expansion ( 57 ) , then Making use of Table I we can trace the curve for In this figure the units of length and time are modified by From this figure it will be seen that the line which was initially situated .
the axis of is suddenly put in motion and there begins an elevation of the free surface , and that after a certain time ( about ) there is the one only maximum of the elevation , and then it decreases more and more until the surface takes the original plane form .
If we take for the value of which ives the maximum of the elevation , then the time at which it occurs is given by , in which is reckoned from the time of explosion in seconds and is measured by feet .
Lastly , at a point very far from the source of disturbance , if we neglect compared with , we may put and in .
Then we ilave ( 99 ) This formula and those that follow are applicable only when is moderately small , this being the only range of values which matters in the physical problem , each phase of the physical disturbance travelling , as is well known , with constant acceleration .
When is large , the most essential terms of the series ( 99 ) are situated far along it , and in deducing them from ( 97 ) we must not replace by unity .
Thus the impossible physical feature of persistence of the amplitude for all time with period diminishing randly without limit , which is indicated in ( 100 ) and in fig. 4 , would not occur in an exact solution : it can be cut away while the earlier undulations of the curve remain applicable .
caused by Local Disturbance .
or referring to ( 35 ) and ( 36 ) this can be put in the form with For large values of , from ( 38 ) , we get ( 100 ) Further , if we put , then .
( 101 ) Again , if we put then we have .
( 102 ) The function has been tabulated by E. Lommel in his memoir on the diffraction problem , and the other functions needed are deduced in Table II , so that the forms of the few waves can be traced without difficulty .
* The first term corresponds to the known Cauchy-Poisson expression for the disturbance at a distance due to a point source , and the second is the modiIication in it that is produced by the finite extent of the source : it represents a gradual fall of the surface to the general level .
' Abh . .
Bayer .
Akad . .
Wiss .
, Cl vol. 15 ( 1886 ) , Tafel Waves caused by a Local Disturbance .
The figure 4 , which belongs to equation ( 101 ) , shows the rise and fall of the surface at various times and at a particular place of a great distance from the , the unit of time being modified by .
For moderately small values of , which alone are applicable , the curve is unsymmetrical with regard to the axis of time and as the time increases the rise and fall becomes more and more symmetrical and , on the other hand , the tions follow .one another with ever decreasing period .
For larger values of , the course of the curve ceases to represent the motion , which really dies away , for reasons explained already ( foot note supra ) .
As a practical application , the earlier part of this curve serve to determine the distance of the point under observation from the source of the explosion , if we measure the time of any zero-point of the displacement or the interval ( i.e. period ) between any two adjacent ones .
For example , if we take , as the values of of the zero-points , and if we measure the :actual time between them : then the distance will be given by in the foot-second-unit system .
It will be noticed that the period in this formula would be a large quantity of the order The figure 5 , which belongs to the formula ( 102 ) , shows the wave profile at a particular time at various distances from the origin , the unit of horizontal scale being modified by .
The second part of the diagram is condensed along- the axis of to 1/ 10 scale of the first part .
As we advance farther and farther towards the infinite distant point from the origin , the essential groups of the waves are found to increase continually in length and to dimlnish continually in , in such a manner that remains constant in each phase .
This figure applies only at great distances from the origin , i.e. when is large , so that has to for the earlier part to be applicable ; e.g. , gives in the foot-second-unit system .
In the region some distance away where equation ( 102 ) holds , and therefore the fig. 5 is applicable , the wave-length can be determined by the . .
values of of the earlier in this diagram .
In conclusion I wish to express my sincere thanks to Prof. ciir J. Larmor his kind advice during the progress of this work .

