]\gt ; Notes concerning Oscillations upon Globe .
By LORD BAYLEIGH , O.M. , F.B.S. ( Received May 6 , \mdash ; Read May 27 , 1909 .
) Speculations on tidal questions are much hampered by our ignorance of the peculiar influence of the earth 's rotation in any but the simplest cases .
The importance of this element was first appreciated by Laplace , and he succeeded in obtaining solutions of various problems relating to a globe completely covered with water to a depth either uniform hout , or at any rate variable only with latitude .
His work has been extended by Kelvin , G. Darwin , and Hough .
For an excellent summary , reference may be made to Lamb 's ' Hydrodynamics , ' which includes also important original additions to the theory .
But it must not be overlooked that a theory which supposes the lobe to be completely covered with water has very little relation to our actual tides .
Indeed , in practice , tidal prediction borrows nothing from Laplace 's theory , unless it be to look for tidal periods corresponding with those of the tVenerating forces .
And this correspondence , although perhaps first brought into prominence in connection with Laplace 's theory , is a general mechanical principle , not limited to hydrodynamics .
If the theory of terrestrial tides is to advance , it can only be by discarding the imaginary globe completely covered with water and considering examples more nearly related to the facts , as was done in some by YouIlg and Airy in their treatment of tides in canals .
It is true that we are unlikely to obtain in this way more than very rough indications , but even such are at present lacking .
I am told that opinions differ on so fundamental a as whether the Atlantic tides are generated in the Atlantic or are derived from the Southern Ocean .
Probablv both sources contribute ; but a better judgment , based on some sort of discussion on priori principles , does not appear hopeless .
In this connection , it is interesting to observe that a comparison of spring and neap tides shows that the moon is more effective relatively to the snn than would be expected from the ratio of the generating forces .
This indicates some approach to synchronism with a natural free oscillation .
That the approach is closest in the case of the moon indicates that the free period is longer than those of the actual lunar and solar tides .
Were it not for the complication due to the earth 's rotation rendering all tidal problems vortex problems , as lCelvin put it , questions such as this could be treated ithout great difficulty , and perhaps illustrated by IlJodels .
Notes Concerning Tidal Oscillations npon a Rotating Globe .
449 There is improbable in an oscillation backwards and forwards across the Atlantic having a period somewhat exceeding 12lunar hours , but a treatment at all precise demands the inclusion of the rotation .
This suggests the problem of the oscillations of a ocean bounded ) two meridians .
The present paper does not profess to make more than a modest bution to the subject .
It commences by further the theory of the free vibrations of a plane rectangular sheet of liquid , initiated in a former paper , only under the restriction that the angular velocity is relatively small .
Subsequently , the corresponding problem for an ocean on a , bounded by two meridians , is attempted , but with limited lccess .
Probably a better command of modern mathematical resources would lead to further results .
Plane Recta Sheet .
If be the elevation , the component velocities at any point , the equations of free vibration , when these quantities are proportional to \ldquo ; ( 1 ) and , ( 2 ) in which denotes the angular velocity of rotation , the uniform depth of the water , and the acceleration of gravit The boundary walls will be supposed to be situated at When is evanescent , one of the principal vibrations is represented by ; ( 3 ) and is proportional to , so that .
( 4 ) This determines the frequency when .
And since by synmetry a positive and a negative must influence the frequency alike , we conclude that ( 4 ) still holds in general so ) as can be neglected .
The equation for is at the same time rednced to Taking and as given in ( ) and the corresondin as the ' Phil. Mag vol. 5 , p. 297 , 1903 .
condition is satisfied in the case of terrestrial lakes of moderate especially if they are situated near the equator .
elvin , ' Phil. Mag August , 1880 ; Lamb , ' Hydrodynamics ' ( 3rd ed Lord Rayleigh .
Notes Concerning Tidal [ May 6 , approximation , we add terms , proportional to , whose forms are to be determined from the equations ( 6 ) , ( 8 ) , as well as and , satisfying ( 8 ) , since does so .
They represent in fact a motion that would be possible in the absence of rotation under forces* parallel to and proportional to .
This consideration shows is an odd function of both and , and an even function .
The former investigation proceeded from the assumption for of the form , ( 9 ) which provides for the boundary condition to be satisfied at whatever functions of the coefficients , etc. , may be .
The value of thus obtained was , ( 10 ) where , 2 , 3 , etc. This value of may be employed to obtain a correction to .
If we introduce terms , proportional to , our equations ( 1 ) become , with retention of ; or with regard to the equations satisfied by the terms with zero and unit suffixes , These are the equations that would apply to a rotationless sheet under the action of forces parallel to and proportional to and respectively , and of speed .
In order that the motion thus determined should be , as has been supposed , of the second order in , it is necessary that these forces include no component capable of stimulating the principal motion .
For this purpose the force parallel to Tt will be observed that these forces are not derivable from a potential .
1909 .
] Oscillations upon Rotating G'lobe .
may be omitted from consideration as operating only upon .
Accordingly the condition to be satisfied is , ( 11 ) an equation which may be regarded as giving a correction to In the present case the integration , between the limits for an for , is straightforward and we .
( 12 ) The limiting values of have been supposed for the sake of brevity to be .
If we denote them by , we are to replace in ( 10 ) , ( 12 ) by respectively .
At the same time ( 4 ) becomes .
( 13 ) The method fails if is equal to an odd multiple of .
It would then become necessary to modify the initial assumption , as formerly explained in treating the case of the square , and would rise in magnitude so as to be of the first order in Equation ( 12 ) is not conyenient in its application to the case where is very small .
If we expand the and hyperbolic ents in powers of , we obtain series whose sums are zero for the terms independent of and proportional to , but for higher powers of the series are divergent and no satisfactory conclusion can be drawn .
I have applied ( 12 ) to calculate the value of for the case where .
For the various terms of the series within braces hyperbolic ents , we get ( with reversed signs ) ) , etc. , giving in all about Also .
Herlce The inadequacy of ( 12 ) to deal satisfactorily with the case where / is small led me to seek another solution .
Here we assume in the first instance a form for which satisfies the conditions at , ( 14 ) VOL. LXXXII.\mdash ; A. Lord Rayleigh .
Notes Concerning Tidal [ May 6 , where , 1 , 2 , etc. , making when .
Hence , by ( 7 ) , no arbitrary function of being added , since is odd Further , by ( 6 ) , ( 15 ) which is to be made to vanish when for all values of between .
Now , between these limits , Hence , when Now , since satisfies ( 8 ) , ( 18 ) whence , being an even function of , if , ( 19 ) where .
( 20 ) If , ( 19 ) changes its form for one or more of the values of In ( 19 ) is a constant whose value is to be found from ( 17 ) .
We get , when ; so that finally .
( 21 ) Also from ( 15 ) .
( 22 ) The circumstances are such as to justify the differentiation under the sign of summation .
Collected Papers , 1909 .
] Oscillations upon a Rotating The introduction of ( 21 ) into ( gives the correction to in another form .
We find where , 1 , 2 , 3 , etc. Whatever be the value of becomes larger as increases , and ultimately .
If be small enough , this occurs even for , and we may then omit the in ( 23 ) .
If , further , 1 can be neglected in comparison with , we may take and ( 23 ) becomes , ( 24 ) so that tlje correction is of the third order in , or in , if we replace / '1 by its general value , Comparing ( 12 ) and ( 23 ) , we see that .
( 25 ) If we take , we find from ( 20 ) , so that all the coths on the right of ( 25 ) are nearly equal to unity .
The first term gives and the two following are and , so that the right-hand member of ( 25 ) is , in sufficient agreement with the number previously calculated from the series on the left .
So far we have supposed that the type of vibration is foumded upon .
There is no difficulty in the solution so far as to apply to the type , ( 26 ) where is an integer .
We find , ( 27 ) where the lation relates to , taking in succession the values , 1 , 2 , etc. , and It is uned that ; ( 29 ) otherwise one or more terms corresponding to the lower values of will change their form .
Lord Rayleigh .
Notes Concerning Tidal [ May 6 , A process similar to that already employed when gives for the corrected value of , ( 30 ) which agrees with ( 23 ) when we put .
It is to be observed that the general value of is now given by , ( 31 ) so that ( 30 ) may also be written , ( 32 ) in which does not appear directly .
There is also another class of primary vibrations in which the motion is parallel to and is expressed by , ( 33 ) being an integer .
For this case we find , in the same way , , ( 34 ) , ( 35 ) where takes the values , 1 , 2 , .
( 36 ) It is here assumed that , so that is positive eyen when In this case is an even function of and an odd function of , while is odd in and even in The value of is now given by ; ( 37 ) and for the correction to we have .
( 38 ) It will not be forgotten that in this formula , as well as in ( 32 ) , is neglected .
In the examples hitherto given the primary motion is parallel to of the sides of the rectangle .
I will now take an example from the 1909 .
] Osciltations upon a square , where the primary motion is symmetrical with respect to and md defined by , ( 39 ) sides of the square being the lines .
In harmony with 39 ) we get ( 40 ) since satisfies we see that Also .
( 42 ) The equations of the next approximation , analogous to ( 6 ) , ( 7 ) , are , ( 43 ) ; ( 44 ) they are the same as if implessed forces beted parallel to ?
and respectively and there were no rotation .
From this may infer that is even in and odd in , while is odd in and even The procedure is much the same as before .
We assume , ( 45 ) Nhere , 1 , 2 , etc. , making when .
From ( 43 ) , ( 44 ) deduce ; ( 46 ) that , ( 47 ) ince q when Now , between the limits for .
( 48 ) And thus .
( 49 ) But satisfies , ( 50 ) and hence , being odd in , takes the form where Lord Rayleigh .
Notes Concerning Tidal [ May 6 , This fonm obtains when .
When , pSi7 , and S7 . .
( 53 ) Using ( 49 ) , ( 51 ) to determine , we get , ( 54 ) in which the first term may be deduced from the general term.by putting if we remember that Also from ( 46 ) .
( 55 ) It is evident , however , that there must be another expression for analogous to that given for 11 , and such as would be obtained by rting from ( 56 ) instead of from .
We may , in fact , interchange and if we reverse the sign of .
Thus ( 57 ) In applying these results to find a correction to , we have , much as before , ; and thus .
( 58 ) In accordance with what has been said , if one of these integrals vanishes , so does the other , and we may confine our attention to the former .
In the first place In integrating we have ( 60 ) and ( 61 ) 1909 .
] Oscitlations upon Rotating Globe .
Thus , ( 62 ) being given by ( 52 ) .
For calculation of the first term under the of ntnnation ( the form must be modified .
We find .
( 63 ) The most important term is the next for which ) .
Under the sign of summation we have ( 64 ) The following terms are , and , so that ether we may take as the sum of the terms under the of summation Accordingly ; and this result , being already of the right dimensions , applies whatever be the size of the square .
It may be remarked that the of correction is the opposite of that applicable to the circle , for which approximately* .
( 66 ) These results are , of course , applicable only under the restriction that is small compared with , the latter quantity on the size and depth of the sheet of liquid .
In the case of lakes and seas upon the earth , we have also to remember that depends upon the latitude .
At the equator vanishes .
of liiqui An attempt will now be made to apply sinlilar methods to the free vibrations of an ocean on a rotating globe , the water of uniform depth , and bounded by vertical walls coincident with two meridians , and .
Using a similar notation , we have as the eneral equations , ( 67 ) ( 68 ) with the equation of continuity ( 69 ) * Lamb , .
cit. , p. 306 .
Lamb , .
cit. , p. 314 .
Lord Rayleigh .
Notes Concerning Tidal [ May 6 , Here denotes the colatitude , is the radius of the globe ; its angular velocity of rotation ; the velocities along and perpendicular to the meridian .
As is usual , we shall write for when convenient .
By ( 67 ) , ( 68 ) , and may be expressed in terms of , and substitution in ( 69 ) will then give an equatio1l in only .
When , this equation is the well-known one , .
( 70 ) We will suppose that the primary motion\mdash ; that which would obtain if \mdash ; is represented by , and that , so that the primary motion is wholly in latitude .
And we will begin with the further supposition that .
( 71 ) Substitution in ( 70 ) shows that The motion is that which might obtain equally over the complete sphere , the liquid heaping itself alternately at the two poles .
It is to be observed that , under the circumstances here contemplated , ( 70 ) holds good so lonerb as can be neglected , since all that is required in its formation is the omission of and of .
We write it in the form ( 73 ) and we observe that ( 73 ) is satisfied alsO by and by ( 73A ) if are the correctional terms proportional to .
If we substitute , or , in the hand membe of ( 73 ) , we ; so that ( 74 ) Here vanishes at the limits of .
If we assume , ( 75 ) we may deduce for the left-hand member of ( 74 ) .
( 76 ) For the expansion of the right-hand member we have , between and ; ( 77 ) 1909 .
] Oscillations upon Rotating Gtobe .
so that for the general term , ( 78 ) being an odd integer .
For even values of vanishes .
The complete integral of ( 78 ) as complementary function , two functions of , one odd and one even , each multiplied by an arbitrary constant .
In the present case we have to do only with the odd function and its coefficient is to be determined by the consideration that remains finite at the poles .
A complete treatment presents considerable difficulties .
Reference ay be made to Theory of Sou , S 338 .
In the present case the of spherical harmonics is unity , but is not necessarily integral , still less an integer not exceeding 1 .
When is small , the calculation simplifies , for then the second term on the left of ( 78 ) predominates , and ultimately we have , ( 79 ) which with ( 75 ) determines for the extreme case .
We may pursue the approximation with respect to by substituting from ( 79 ) in the first and third terms of ( 78 ) .
For this purpose we may use .
( 80 ) As was to be expected , the term in contributes nothing .
The result of substituting is thus ; and accordingly the second approximation to is .
( 81 ) When is known , the terms in and may be found from ( 67 ) , ( 68 ) .
We will now } ) to calculate a correction to in the manner already employed for plane sheets .
We have to make , ( 82 ) where ?
, and the integrations extend from to for , and from to for .
The calculation is straightforward , and we find , ( S3 ) Lord Rayleigh .
Notes Concerning Tidal [ May 6 , where ( odd ) has been replaced by its equivalent , continuance of the approximation from ( 81 ) gives another term in ( 83 ) involving ( as well as ) and .
It appears that ( 84 ) when we substitute for , their values , and If it were important , the approximation with respect to could be carried further without much difficulty .
We will now pass on to consider the most important primary mode in which is an even function of , proportional to the zonal harmonic of order 2 , or to .
In agreement with this we take .
( 85 ) Substituting the value of in ( 70 ) , we see that ; ( 86 ) and satisfies , ( 87 ) an equation satisfied also by and } ) .
( 88 ) Substituting ( 88 ) in ( 87 ) , we get , ( 89 ) the analogue of .
Retaining ) and understanding , as before , that is an odd integer , we get , with use of ( 77 ) for the general term , , ( 90 ) where This equation may treated in the same way as was ( 78 ) ; but it may be well to introduce a modification which would be convenient in pursuing the approximation further .
We will divide ( 90 ) by the factor which multiplies the parenthesis on the right , ; and assume* , ( 91 ) Thomson and Tait 's ' Natural Philosophy , ' 2nd ed. , Part , p. 210 .
1909 .
] Oscitlations upon Rotating where is even .
Substitnting in the right-hand member of ( 90 ) , we find as the cient of .
( 92 ) Hence , 1 .
1 , , 3 .
, 7 .
; the -hand members being zero for 8 and higher values of .
It will be seen that one of the coefficients is arbitrary , providing the necessary undetermined element .
The problem would be so to choose it as to satisfy the condition at the pole .
When may be treated as , we may divide the system of equations by it , as the first approximation , ( 93 ) after which , etc. , vanish .
To obtain a second approximation we substitute the result of the first approximation in the smaller terms .
Thus In like manner , ; after which the are zero to this order .
The approximation may be pursued , and at each step another ente .
In this process the difficulty of satisfying the general condition at the pole is evaded .
If we stop at the first approximation ( 93 ) , we have .
( 94 ) Using this value of in ( 82 ) , we find ( 95 ) as the correction applicable for this mode of vibration of a narrow lune .
From some points of view , there is in the use of , rather than , as indel ) endent variable .
In place of ( 90 ) we have* ; ( 96 ) * See ' Theory of Sound , ' Lord Rayleigh .
Notes Concerning Tidal [ May 6 , but for the moment we will take on the right the more general form .
Assuming , ( 97 ) we find on substitution in the left-hand member of ( 96 ) as the coefficient of .
( 98 ) Thus for the term depending on , , etc. ; so that , simply .
( 99 ) For the term in , etc. ; so that .
( 100 ) In like man.ner , for the term in Introducing the numerical values of , we find for the sum of the three contributions , 4 .
6 .
8.13 ( 102 ) and there is also to be added the complementary function , .
( 103 ) in which A is a constant .
The expressions in ( 102 ) , vanish when .
It is further necessary \mdash ; and this is the condition determining A\mdash ; that should vanish at the equator .
Now in which vanishes .
So far as regards ( 99 ) , is finite when so that no further question arises here .
But for ( 100 ) , ( 101 ) , ( 103 ) , is infinite when , and a further scrutiny is called for .
1909 .
] Osciltations As a'first step we may examine ( 103 ) , taking it in the more general form* ( 104 ) In Gauss ' notation for hypergeometric series , , ( 105 ) where Since is finite when , and so also is .
But for we have of.which the first part , being finite , need not be arded .
Thus when .
( 106 ) Now , in which , when nearly , .
( 107 ) In the present case and thus when .
( 108 ) , as at present , , ( 109 ) when .
Thus has a finite value at the equator , as was to be expected .
It may be proved without difficulty that ( 102 ) when and is not an even integer .
Any finite number of terms which may have negative denominators being excluded , the remainder may be expressed as a hypergeometrical series .
But the form is more complicated than before , and the evaluation of would be rather tedious , even if practicable .
'Theory of Sound , ' .
cit. Here Infinite Series , ' Bromwich , p. 171 .
Rev. H. V. Gill .
On New Kind [ May 7 , A question obtrudes itself as to what happens when is an even integer .
When , there is synchronism between the primary and a derived vibration , and the occurrence of the infinite denominator is what have been expected .
But in the case of other even integers no synchronism is apparent , and it would seem that the complication is of an analytical character only .
The solution compounded of ( 102 ) and ( 103 ) changes its form .
It would be of interest to follow out the process , say for the case , which night roughly represent the circumstances of the Atlantic Ocean , but I am not prepared to undertake the task .
On a of Glow lladium in Vacuum Tubes .
By .
H. .
GILL , S.J. , B.A. , Downing , Cambridge .
( Communicated by Prof. Sir J. J. Thomson , F.B.S. Received VIay 7 , \mdash ; Read , 1909 .
( PLATE 1 .
) The phenomenon described in this paper was first noticed when making some preIiminary experiments in an ation suggested by Prof. Sir J. J. Thomson .
The effect of the ases contained in metals on the nature of the discharge obtained when these metals are employed as electrodes in vacuum tubes has long been a question of interest .
It was suggested that by employing palladium one hope to obtain interesting results , to the property possessed by this metal of absorbing large quantities of The result of these preliminary experiments has already been briefly stated , * but as the effect seems to be a new one , it seemed of sufficient interest to call for a fuller investigation .
Apparatus.\mdash ; A tube was fitted up containing an ebonite plug through which two stout brass leads were passed .
To the nities of the leads was attached , by means of screws , a piece of palladium foil about an inch in length and a tenth of an inch wide .
The leads and plugs were fixed in position by sealing-wax .
The tube was sealed by the blow-pipe to a mercury pump and was in connection with a pressure auge , drying-bulb , etc. The tube could be isolated from the rest of the apparatus by means of a barometric mercury seal .
By passing an electric current through the palladium , it could be heated to any required temperature .
( See Plate 1 , fig. 1 .
) ' Camb .
Phil. Soc. Proc vol. 14 , No. 6 , p. 578 , 1908 .

