]\gt ; 5 51 By AlIB , Received November Read ovember 3 1 .
The chief question discussed in this paper ( SS6\mdash ; 13 ) is that of the free oscillations of an atmosphere whose temperature varies with the altitude ; and in particular the case of a uniform vertical temperature-gradient is studied in some detail .
For consistenc it is assumed that the expansions and con- tractions follow the adiabatic law .
The problem is treated as a twodimensional one , the space co-ordinates olved b horizontal vertical ; and the more definite conclusions arl.ived at relate to the case where the ( horizontal ) is somewhat large in comparison with the height of the atmosphere .
The results are most easily interpreted when the -gradient does not much that chal.actel.isbic of a state of convective equilibrium .
The normal modes of oscillation then fall into well-defined types .
In the most important type , the motion of the air-particles is mainly horizontal , and } ) endent of the altitude , and the aves may thelefole be described as ' : longitndinal The velocity of ation of progressive waves is found to be equal to ) , where denotes what be called the virtual atmosphere , of a " " holnogeneous atmosphere\ldquo ; corresponding to the tenlperature of the .
That the result should colle out intermediate in value between the velocity of souud in the lowest stl.atum , iz .
, $/ /(r/ H ) , and the zero velocity corresponding to zero temperature which is postulated in the higher regions was to be anticipated ; but that it should be identical in form with that obtained on the hypothesis of an atmosphere whose expansions are subject to Boyle 's the effect of the vard ( lecrease of temperature exactly compensated by the greater elasticity implied in the adiabatic law , is somewhat relnarkab ] When the radiant falls distinctly below the " " convective\ldquo ; value , the character of the oscillation is ] simple .
The wave-velocity is somewhat increase must vays r value , which is the velocity of sound in the lowest stratum .
2 .
A second type of oscillations depends on the of stability of the atmosphere .
* Rayleigh , ' Phil. Mag. ' ( 4 ) , 1890 , vol. 29 , p. 173 ; 'Scientific Papers , ' vol. 3 , p. 33 VOL. LXXXIV.\mdash ; A. 2 Prof. H. Lamb .
[ Nov. 14 , The work required to bring unit mass of air from the density to the density under the adiabatic condition is .
( 1 ) Hence if we imagine two thin strata of equal mass , whose densities are and pressures , to be interchanged , the work required to effect this will be , per unit mass , ( 2 ) If we avail ourselves of the notion of " " potential temperature , \ldquo ; *i.e .
the temperature which any particular portion of air would assume if adiabatically to some standard density , we have , ( 3 ) where is the constant of the formula $ , ( 4 ) denoting the absolute temperature .
Hence ( 2 ) becomes .
( 5 ) Hence if , we must for stability have ; i.e. the temperature must increase upwards .
Now , if denote depth below a standard level , we have , in equilibrium , ; ( 6 ) and combining this with ( 3 ) and ( 4 ) , we find .
( 7 ) In convective equilibrium , where , and consequently 3 , is the same at all altitudes , we have .
( 8 ) This equilibrium , though stable for some types of disturbance ( S 8 ) , is in other respects neutral .
For complete stability , must be negative , and therefore .
( 9 ) When this condition is fulfilled , we have a series of possible modes of .
Bezold , ' Berl .
Sitzb 1888 , vol. 46 .
1910 .
] On oscillation whose periods .
as they do on the extent which the temperature-gradient differs from the convective value ( S ) , are comparatively .
Oscillations of this character , by local conditions , must undoubtedly occur in the atnlosphere , and may conceivably for some of the minor tions of the barometer .
There remains a third type of oscillations which , when the wave-length is moderately great , approximate to the chal'acter of waves ated vertically in the atmosphere .
These have been discussed in a previous paper by the author .
* From a met , eorological standpoint they can hardly be of importance .
3 .
The theory of the " " longitudinal\ldquo ; waves is of interest in relation to the large-scale oscillations of the earth 's atmosphere as a whole .
This subject was treated by is of some importance in connection with the suggestion put forward by Lord as to the origin of the variation of the baro meter .
's investigation was based on the hypotheses of a uniform equilibrium temperature and an isothermal law of expansion , and on the further the ertical motion of the air-particles may be ected .
S Since the circumstances are then practically those of sound-waves } ) ated horizontally , his results naturally involve the " " Newtoniau\ldquo ; velocity of sound , , where is the of the homogeneous atmospbere to the ssunled uniform temperature , viz. , The hypotheses erred to were , of course , adopted only for nlathernatical convenience .
As representations of actual conditions they imperfect and there is , moreover , great uncertainty as to the mosb suitable value to be attributed to .
It } ) to the writer that firmer for tative conclusions would ) gaiued if it were possible to calculate the wave-velocity ( for ( waves ) , even in the problem , on more natural suppositions as to the constitution of the atmosphere and the law of } ) ansion .
In the actual the temperature , as a rule , diminishes } ) although ( as we have seen ) it is necessary for stability that the vradient should nowhere the convective value .
The special ] ) of a uniform gradient , which is here adopted as a bnHis uf calculation , is itself an artificial one ; but in spite of fact that it inlplies upper to the atmosphere , it ltay to give , on the whole , a better representation of the ' Lond. Math. Soc. Proc. ' ) , 1908 , vol. p. 122 .
' Mecanique Livre Chap. .
See also .
cit. 'Roy .
Soc. 1882 , vol. 11 ; 'Math .
and apers , .
p. 341 .
S Some such assumption is necessary tu make the prol ) , in the absence of a prescribed condition to ) fullilIed , or to , in the upper ions of the atmosphere .
Prof. H. Lamb .
true conditions than the isothermal view , on which , indeed , the earth 's atmosphere is merely a local concentration of a medium diffused through space .
As ards the law of expansion , since permanent inequalities of temperature are postulated in the equilibrium condition , it is proper to ignore the transfer of heat between adjacent portions of the air during the oscillations .
In any case , theory shows that the effect of conduction on such long waves as we have here in view may safely be neglected .
* The main conclusion of Laplace was that the free and forced oscillations of an atmosphere a globe , whether this be at rest or in uniform rotation , are identical with those of a liquid ocean of uniform depth ; but .
in view of the nature of his premises , and of the uncertainty as to the temperature to be adopted in imating the value of , considerable doubt has been felt as to how far this analogy can be relied upon for quantitative results .
The present investigation tends , I think , to show that inferences of this kind will not be very far from the truth , provided the temperature dopted be the mean temperature of the lower strata of the earth 's atmosphere , so far as this can be ascertained .
The formal adaptation of the theory of waves to the case of an atmosphere of relatively small depth covering a globe would follow the same course as in Laplace 's investigation .
4 .
As regards the semi-diurnal variation of the barometer , the passage of Kelvin 's paper already referred to runs as follows:\mdash ; " " The cause of the semi-diurnal variation of barometric pressure cannot be the gravitational tide-generating influence of the sun , because , if it were , there would be a much larger lunar influence of the same kind , while in reality the lunar barometric tide is insensible or nearly so .
It seems , therefore , certain that the semi-diurnal variation of the barometer is due to temperature .
Now , the diurnal term , in the harmonic analysis of the variation of temp jrature , is undoubtedly much in all , or nearly all , places than the semi-diurnal .
It is then very remarkable that the semidiurnal term of the barometric effect of the variation of temperature should be greater , and so much greater as it is , than the diurnal .
The explanation probably is to be found by considering the oscillations of the atmosphere , as a whole , in the light of the very formulae which Laplace gave in his ' Me'canique Ce'leste ' for the ocean , and which he showed to be also applicable to the atmosphere .
When thermal influence is substituted for gravitational , in the nerati e force reckoned for , and when the modes of oscillation * This follows from the equations ( due substantially to Kirchhoff and Rayleigh ) given in the ; ' Hydrodynamics , ' 3rd edit .
, S343 .
Radiation has a different tendency in this respect .
1910 .
] On Atrnospheric corresponding respectively to the diurnal and semi-diurnal terms of the thermal fluence are ated , it will probably be that the period free oscillation of the former rees much less nearly with 24 hours than does that latter with 12 hours ; and that , theref.ore , with comparatively small magnitudes of the tide-generating force , the resulting tide is greater in the semi-diurnal term than in the diurnal The first question which here arises , , whether as a matter of fact the earth 's atmosphere has a mode of oscillation of the requisite type , with a period of about 12 mean solar hours , can at the present time be examined more closely than was possible at the date ( 1882 ) of the above extract .
The free oscillations of an ocean of water of uniform depth coveri of the size of the , rotating with the same angular velocity , have been very fully ated by Hough* in the course of his classical work on tidal theory .
He finds , in particular , that in the case of the most important free oscillation the same eneral character as a semi-diurnal tide wavs ( i.e. its most salient spherical harmonic constituent is the sectorial harmonic of the second order ) , the depth for which the peried is exactly 12 sidereal hours is given by where is the earth 's radius , and its velocity of rotation .
This is evaluated at 29,182 feet .
It is to be remarked , however , that the calculation the mutual attraction of the disturbed fluid has been takeu into account , whereas in the rial o this influence must be quite insensible .
If the disturbance were accurately of the type of a spherical harmonic of the second order the requisite modification would consist merely in multiplying the previous result by the factor where the decimal fraction in the second meml ) is the ratio of the density of the water to the mean density of the globe , as adopted in 's computation .
This would make As the result of a more direct culation , 's algorithm , her with such of bis numerical results as are applicable , I find the last figure being somewhat doubtful .
If put , this ivcs feet .
'Phil .
Tralls , 1897 , vol. 191 , p. 139 .
See pp. 164 , Prof. H. Lamb .
[ Nov. 14 , The substitution of a mean solar for a sidereal half-day as the period involves a further slight diminution , which can be estimated pretty closely from another of Hough 's results .
He finds that for , the speed of the free oscillation in question is given by .
Comparing this with the former result , we infer that for a period of 12 mean solar hours we must have , about .
Assuming that when mutual attraction is ignored this rure is to be reduced in the same ratio as the former one , we have , finally , or , with the previous numerical data , feet .
It must be remembered , of course , that these numerical results can claim no greater accuracy than the theory on which they rest , in which , in particular , the ellipticity of the earth , which is of the order 1/ 300 , is neglected .
On the other hand , the value of for air at C. is about 26,200 feet , with an increase of about 96 feet for every above this temperature .
The mean temperature of the air the earth 's surface is usually estimated at C. This would make feet ; but a somewhat lower value for the mean temperature of the lower strata , away from the immediate influence of the ground , would perhaps be more appropriate .
Without pressing too far conclusions based on the hypothesis of an atmosphere uniform over the earth , and approximately in convective equilibrium , we may , I think , at least assert the existence of a free oscillation of the earth 's atmosphere , of\ldquo ; semi-diurnal\ldquo ; type , with a period not very different from , but probably somewhat less than , 12 mean solar hours .
At the same time , the reason for rejecting the explanation of the semidiurnal barometric oscillation due to a gravitational solar tide seems to call for a little further examination .
The amplitude of this variation at places on the equator is given by Kelvin as inch .
The amplitude given by the " " equilibrium\ldquo ; theory of the tides is about inch .
* Some numerical results given by Hough in illustration of the kinetic theory of oceanic tides indicate that in order that this amplitude should be increased by dy1lamical action some seventy-fold , the free period must suffer from the imposed period of 12 solar hours by not more than 2 or 3 minutes .
Since the difference between the lunar and solar semi-diurnal periods amounts to 26 minutes , it * The numerical values given on p. 520 of the author 's 'Hydrodynamics ' relate to the lunar tide , and are , moreover , by an oversight , stated as ' amplitudes instead of as ranges 1910 .
] On Atmospheric Oscillations .
is quite conceivable that the solar influence in this way be rendered much more effective than the lunar .
The real difficulty , so far as this point is concerned , is the a priori improbability of so very close an }reement between the two periods .
The most decisive evidence , however , appears to be furnished by the phase of the observed semi-diurnal equality , which is accelerated instead of retarded ( as it would be by tidal friction ) relatively to the sun 's 5 .
The conclu.ding part of the paper ( SS13 , 14 ) is an attempt to examine more closely than has hitherto been done the theory of waves on a surface of discontinuity in the atmosphere .
That such waves may play a part in phenomena has been pointed out independently by Helmand Lord Kelvin , but both writers have fined themselves to analogies drawn from the case of superposed homogeneous liquids .
It is to be observed that even on this view the disturbance extends , upwards and downwards from the plane of discontinuity , ] a cqpace which is appreciable fraction of the ; hence , apart altogether from the influence of compressibility , the conditions of the question will be modified when the wave-length is such that the ordinary variation of density within this space becomes sensible .
It seemed worth while to investigate the matter ; but it must be acknowledged that when there are no currents , the one of temperature antl density only , the analogy proves to be adequate , under such conditions as are likely to occur in the atmosphere , for a considerable range of .
For very long waves it would break down , the disturbance to be even approximately concentrated in neighbourhood of the plane of discontinuity .
The discontinuity then beconcs , in fact , an unimportant incident in the eneral upward diminution of density .
When there is a discontinuity of relocity , the upper fluid in steady horizontal motion relative to the lower , the question , when compressibility is taken into account , is more difficult , and 1 have not been able to arrive at any very simple results .
There can be no doubt , , that the aforesaid analogy is sufficient in this case also for wave-lengths less than a certain .
In particular , the dynanlical instability pointed out by KelvinS will remain .
'Brit .
' 1908 , p. 606 .
The forced tides duc to diurnal and -diulnal wa es of temperature have been studiod by bfargules , 'Wien .
Sitzb , vol. 99 , p. 204 .
' Berl .
Sitzb ) ; 'Wiss .
Abh vol. 3 , p. .300 . .
Assoc. Rep ; 'Math .
and Phys. Papors , ' vol. 4 , p. ) .
S 'Math .
and pers , , p. 76 .
Prof. H. Lamb .
[ Nov. 14 , Theory of Long ospheric Waves .
6 .
We consider an atmosphere arranged in horizontal layers of uniform density .
The motions contemplated are restricted to two dimensions , , of which is horizontal and vertical , the positive direction of being downwards .
The equilibrium values of the pressure , density , and temperature are denoted by ; these are functions of only , and a.re subject to the hydrostatic condition , ( 10 ) as well as to the general relation .
( 11 ) The equations of small motion are , in the usual notation , .
( 12 ) , ( 13 ) where .
( 14 ) The expansions being supposed subject to the adiabatic law , we have also , ( 15 ) where , ( 16 ) i.e. is the velocity of sound corresponding to the equilibrium temperature at the point considered .
It is accordingly in general a function of .
If we put , ( 17 ) and continue to neglect small terms of the second order , we have ( 18 ) .
( 19 ) Also , from ( 15 ) , ( 13 ) , and , .
( 20 ) Hence , eliminating and , we find* ( 21 ) * It may be noticed , parenthetically , that in the case of an isothermal atmosphere where is constant , these equations are satisfied by 1910 .
] On If we now write , ( 22 ) we deduce from ( 21 ) , by differentiation , , ( 23 ) , ( 24 ) where ( 25 ) The latter equation shows that an irrotational motion is not possible unless , or , ( 26 ) which we have seen to be the case of equilibrium .
We note also that , ( 27 ) by ( 20 ) .
Eliminating between and ( 24 ) , we obtain .
( 28 ) If we assume that and occur only through a factor , the equations ( 21 ) take the whence ( 30 ) From these , or from ( 28 ) , have 7 .
So far , the vertical distribution of } ) is .
In the case of temperature diminishing upwards with gradient , to which we now proceed , there is an upper linlit to the .
If we take the origin of at this level , W have ( 32 ) Prof. H. Lamb .
[ Nov. 14 , where is the gradient in question .
It easily follows from ( 10 ) and ( 11 ) that where AIso .
Hence and The meaning of the factor , which appears in one of these terms , is to be noticed ; viz. we have where is the temperature-gradient in a state of convective equilif ) rium , as given by ( 8 ) .
To solve ( 37 ) we put obtain , ( 40 ) where This is integrable by series , the solution which is finite for being ; or , in the notation of Dr. E. W. Barnes , * .
( 43 ) The remaining solution of ( 40 ) is of the form where stands for the series in ( 42 ) .
Tbis is not admissible in the present * See , for example , 'Camb .
Trans vol. 20 , p. 253 , where references to other papers are given .
If we had assumed place , we should have found .
The comparison verifies a well-known identity ; see Barnes , .
cit. 1910 .
] On A tmospheric question , since it becomes infinite as for infinitesimal values of whereas the condition to be satisfied at the upper boundary is , or ; see equations ( 27 ) , ( 33 ) .
The formulae ( 36 ) now become ( 45 ) the factor being omitted here , as elsewhere , for brevity .
The condition that at the lower boundary , where , say , taken in conjunction with ( 41 ) , determines the values of and , the being supposed given .
* 8 .
In the case of oscillations about convective equilibrium we have .
( 46 ) It follows from ( 24 ) that ; hence either the motion is irrotational , or the period is infinitely The conditions to which the rotational motions thus indicated are subject follow most directly from ( 21 ) .
These equations are now equivalent to ( 47 ) by ( 26 ) .
Hence const .
( 48 ) The choice of two functions , , to satisfy this equation , together with the two boundary conditions , can be made in an infinite variety of ways .
The types of disturbance are periodic in charactel .
The formula ( 42 ) and ( 45 ) apply , with ' ( 49 ) in place of ( 41 ) .
Since ( 42 ) makes ( 50 ) * The case of apparent failure , where , does not arise .
This would require , by ( 36 ) , , or which violates the condition at the upper boundary .
Prof. H. Lamb .
[ Nov. 14 , the condition that for may be written A complete discussion of the equations ( 49 ) and ( 51 ) is out of the question , but the limiting form to which the results tend as the wave-length increases is easily ascertained .
In the first place , it appears that when is small we have , ( 52 ) approximately , since this ensures , by ( 49 ) , that is also small .
If denote the virtual height of the atmosphere , as defined above ( S1 ) , we have .
( 63 ) The limiting value of the wave-velocity V is accordingly given by The bearing of this result has been discussed in the introduction .
The formulae ( 45 ) , , and ( 39 ) now lead to ( 55 ) the factor being understood .
These values fulfil , as they ought , the irrotational condition .
( 56 ) Since the ratio of the amplitude of to that of is of the order , the motion is mainly horizontal , and the present type of waves may accordingly be characterised as " " longitudinal The remaining solutions of ( 49 ) and ( 51 ) , when is small , involve finite as distinguished from infinitely small values of .
As will be seen presently ( S11 ) , they approximate to the character of waves propagated vertically in the atmosphere .
9 .
In the general case , where ?
is not restricted to the precise value , the elation between and is as in ( 41 ) .
When is smal ] we have still a longitudinal wave for which is of the order , subject to a certain condition .
The equation ( 51 ) leads again to the result expressed by ( 52 ) or ( 54 ) , and substituting in ( 41 ) we find that the implied assumption that is also small will be justified provided be small , i.e. provided the temperature-gradient falls only a little short of the convective value 1910 .
] On Atmospheric 563 .
The limiting form of ( 42 ) , when no assumption is made as to the order of magnitude of , is ( 57 ) or in Dr. Barnes ' notation , , ( 58 ) whilst ( 50 ) becomes .
It appears from ( 41 ) , without making as yet any special assumption to the smallness of , that when is finite , whilst is small , ratio will be very small or very great .
In the former case we have ultimately , and the condition ( 51 ) becoines : .
( 61 ) Since , ( 62 ) in the notatiori of Bessel 's ) tions , this may be written , ( 63 ) provided .
( 64 ) If be a root of ( 63 ) , the corresponding frequency of oscillation is given by and the wave-velocity by , ( 66 ) being defined as before by ) .
This result again is accurate a limiting form for increasing wave-length .
10 .
The equation ( 63 ) might be discussed , when or ) is , with the help of the tal , les of Bessel 's functions , but it may bc sufficient to consider the case where the ratio is small .
It may be noticed that the formula embraces all the modes of the present class , the waves already discussed corresp onding to the of infinitesimal .
The roots of ( 63 ) which relate to the remaining modes are now given by , ( 67 ) Prof. H. Lamb .
[ Nov. 14 .
, approximately ; and in particular the first of these slightly exceeds the first finite root of ( 67 ) .
In convective equilibrium we have , if .
The first finite root of is , very nearly .
Hence for oscillations about a state of very nearly neutral equilibrium we have .
In the case of , which makes , a first approximation , given by ( 67 ) , is , and a second is easily found to be This leads to which is about one-fifth the velocity of the longitudinal type of waves .
As to the character of these slow rotational* modes , we find from ( 24 ) , ( 68 ) or , by ( 65 ) , .
( 69 ) Having to the kinematical meaning of the functions , as defined by ( 22 ) , we see that the rotational quality in the relative motion of a fluid element predominates over the dilatational .
We learn also from ( 45 ) that when the amplitude of is to that of in the ratio , which is small .
Since vanishes at the lower boundary , we infer that the vertical component of the velocity is in eneral relatively small .
The distribution of horizontal velocity depends ultimately on the function which varies as if be the relevant root of ( 63 ) , or less accurately of ( 67 ) .
In the case of the first root , after the small one , this expression changes sign once , and once only , as increases from to .
For , the change of sign occurs for , or The general character of the types of disturbance at present under consideration is most easily apprehended in the case of a " " standing\ldquo ; oscillation .
If on the preceding expressions we superpose others which The rotational character is , of course , present also in the longitudinal waves , unless exactly , though to relatively slight extent .
1910 .
] On A trnospheric Oscillations .
differ only in the of , and reject the imaginary parts , we find , on discarding all but the most important terms , that ( 70 ) The differential equation of the lines of ( oscillatory ) motion , is satisfied by const .
, ( 72 ) or .
( 73 ) If we put we get the lines , but the former of these is only an approximation .
The annexed figure indicates , without any attempt at minute accuracy , the general arrangement of the lines in the case of the lowest finite root of .
In the modes the higher roots there are horizontal nodal planes , in addition to ) oundary .
Returning for a ltoment to the more important " " \ldquo ; of motion first considered , we note that the formulae ( 52 ) , ( 54 ) , cease to be accurate , even as limiting , when the ratio diflers appreciably from unity .
The and ( 66 ) will , however , still apply , the lowest root of ( 63 ) .
As a numerical example , take the case where the temperature-gradient has only one-half the convective value , so that I find that the lowest root of is , approximately , whence .
The result must , of course , in any case ] ) less than , or .
The of wave-velocity is accompanied by a in the character of oscillation , the variation of .horizontal velocity with altitude now sensible .
The preceding formulae also be used to estimate the rapidity of Prof. H. Lamb .
falling away from the state of unstable equilibrium which prevails when , the value of given by ( 65 ) being then negative .
11 .
The modes for which is large are easily accounted for .
We have from ( 41 ) and from ( 45 ) these being approximations which gain indefinitely in accuracy with increase of .
On the supposition that is finite , standing the smallness of , ( 75 ) reduces to , as in ( 64 ) .
If be a root of this equation , the corresponding frequency is by These nlodes are in the limit identical with the waves of vertical displacement discussed in a paper already cited in S2 .
The formulae ( 45 ) show , in fact , that the ratio of amplitude of to that of is for the most part of the order .
If we put the equation ( 76 ) takes the , which is identical with equation ( 88 ) of the paper referred to .
12 .
It may be worth while , for the sake of the contrast , to give the theory of the oscillations of a heterogeneous but incompressible fluid , whose equilibrium density has a similar distribution .
We have now .
( 80 ) If we put as before , we have From ( 80 ) we have 1910 .
] On Oscillations .
the stream-function .
in ( 82 ) , ( 83 ) , and eliminating and , we find* .
( 85 ) If and occur only in the form , we have .
( 86 ) Also , ( 87 ) and .
( 8S ) If we assulue that , ( S9 ) we have ( 90 ) or Wl'iting , ( 91 ) .
( 92 ) The solution which is finite is , ( 93 ) or , ( 94 ) if .
( 95 ) The second solution becoules infinite as for , and is therefore excluded , in virtue of ( 88 ) , by the condition that ] must vanish at the upper boundary .
Since , by ( S7 ) , for , we have This determines , and ) value of follows from ( 95 ) .
It is obvious that , when is small , ( 96 ) is not satisfied by linite values of .
If be large , but so that remains finite , the equation ( 96 ) rends to the form , ( 97 ) or , ( 98 ) provided If be a root of ) , we have , ( 100 ) *Cf .
Love , ' Lond. Math. Soc. Proc. ' ( 1 ) , 1891 , vol. 22 , p. 30 .
LXXXIV.\mdash ; A. 2 Prof H. Lamb .
[ Nov. 14 , and therefore , for the wave-velocity , .
( 101 ) Thus , if , for the sake of comparison with S7 , we put , we have , whence .
( 102 ) That the frequency should be increased by the incompressibility was to be expected ; that the effect is so considerable is due to the great modification which is caused in the character of the fundamental modes .
The modes corresponding to the roots of ( 98 ) have horizontal nodal planes , and the frequencies form , by a descending series , in the case of ( 65 ) .
Waves at a Surface of Discontinuity .
13 .
When we proceed to examine the case of waves ated along a horizontal plane where the equilibrium temperature is discontinuous , it may be sufficient to suppose the temperature uniform throughout each of the regions , above and below this plane , to which the influence of the waves extends .
The plane in question is taken as the plane , and the dependent variables relating to the upper region will be distinguished by accents .
The formulae of S6 will therefore apply to the lower region , with the simplification that is a constant ; so that ( 31 ) becomes .
( 103 ) This is satisfied by , ( 104 ) provided .
( 105 ) We are for a type of motion analogous to that of waves on the interface of two liquids of different densities , in which case the values of are .
We assume , provisionally , that in our case also the roots of ( 105 ) are real and of opposite signs ; moreover , since the disturbance is to vanish for , the negative sign is to be taken .
For the upper region we shall have , ( 106 ) with a similar determination of ; but the positive root is now the appropriate one .
*Cf .
Rayleigh , ' Lond. Math. Soc. Proc. ' ( 1 ) , 1883 , vol. 14 , p. 170 ; 'Scientific Papers , ' vol. 2 , p. 200 .
1910 .
On .
569 If denote the ordinate of the surface of separation , as affected by the waves , we have ( 107 ) for ; and the pressure in either fluid at the point is to be found by putting in the corresponding value of the expression .
( 108 ) Differentiating with respect to , we see that must be continuous at the interface .
* This involves , by ( 27 ) , the continuity of , so that the constants in ( 104 ) and ( 106 ) must be equal .
Again , it follows from 107 ) that must be continuous , whence , by ( 30 ) , .
( 109 ) This , ether with the two equations of the type ( 105 ) , determines the values of , and To obtain a solution , we denote the equal members in ( 109 ) by ; thus .
( 110 ) Substituting in , we have , ( 111 ) with a similar equation in which is replaced by these two equations in the form , ( 112 ) and eliminating , we have ( P\mdash ; P ' ) .
( 113 ) Now Hence ( 115 ) is of the fourth ttree in , but one root only is relevant to the present question .
The common root of ( 112 ) is , ( 116 ) * This might almost have been assumed at once ; but it is to be observed that it would not give the correct condition to be satisfied at ths comulon boundary of two curreis .
Prof H. Lamb .
[ Nov. 14 , whence ( 117 ) If we now write , ( 118 ) , ( 119 ) the equation ( 115 ) becomes ; ( 120 ) whilst .
( 121 ) It is to be noticed that , ( 122 ) where , are the equilibrium densities at the plane , on the two sides .
For sufficiently small wave-lengths , and are very small , and the root of ( 120 ) with which we are concerned is , approximately , whence , ( 123 ) as in the case of superposed incompressible fluids .
* To examine the matter further , the simplest procedure is to tabulate the function ( 124 ) for a series of suitable values of .
The only case of real interest is where the discontinuity of temperature is very slight , so that is a small fraction .
The following table gives a few results calculated on the supposition that , with .
The abrupt step in temperature then amounts to of the mean of the temperatures ( absolute ) above and below .
* Stokes , 'Camb .
Trans 1847 , vol. 8 , p. 441 ; 'Math .
and Phys. Papers , ' vol. 1 , p. 197 .
1910 .
] On Atmospheric The fourth column gives the ratio of the frequency to that of waves of the same length on the surface of separation of two homogeneous liquids with the same discontinuity of density , as given by ( 123 ) , , the ratio is on the present suppositions .
The seventh column is calculated from , taking metres per second ; and the last from .
It is seen that , with increasing wave-length , the wavevelocity tends more and more to exceed the value estimated on the assumption of the eneity and incompressibility of the two fluids .
At the same time , the disturbance tends to become , relatively as well as absolutely , less and less concentrated in the neiohbourhoodo of the plane of discontinuity .
14 .
In this question , again , it is of some interest to compare the case of waves on the common boundary of two liquids , each of which , though incompressible , has a similar gradation of density .
We therefore write , in ( 86 ) , , a constant .
If we assume that , ( 126 ) we derive .
( 127 ) These formulae may be supposed to relate to the lower ; for the upper region we write ( for ( 1 , , respectively .
The continuity of involves , by ( 87 ) , that of , so that .
Also , in virtue of the continuity of ) , we from ( 88 ) where , are the densities just below and just above the plane , If the two fluids had been portions of the same at different temperatures we should have ( 129 ) and therefore Now from ( 127 ) we have .
( 131 ) Hence , if we adopt the elation ( for the sake of the comparison , we must have , or , taking account of the .
This leads to and .
( 133 ) On Atrnospheric Oscillations .
The positive root of this quadratic in is to be taken , since it is the only one which gives the proper signs to , it being assumed that and therefore .
For infinitesimal values of , we reproduce the relations ( 123 ) .
In order to make the variations of density follow exactly the same law as in the atmospheric problem of S13 we must give to , the values ( 129 ) .
In the notation of ( 118 ) , ( 119 ) , we have then .
( 134 ) The following table , like the former one , refers to the case of .
In order that the comparison may be for the same series of wave-lengths , those values of are chosen which were obtained in the previous numerical work .
The significance of the column headed the same as on p. 570 .
The comparison shows the usual effect of a constraint in increasing the frequency .

