]\gt ; , ( ii ) That ositive electricity within tmplitude an condition , but is distributed fairly uniformly through the atom .
III .
Experiments are described on the absorption of the homogeneous -rays .
It is shown that the first stage in the absorption of a pencil of homogeneous -rays consists in the scattering of the rays .
The absorption of a completely scattered beam of homogeneous -rays is shown to take place according to an exponential law .
In conclusion , I have much pleasure in once more recording my best thanks to Prof. Sir J. J. Thomson for much inspiration and advice the course of these experiments .
Aerial Plane Waves of Finite Amplitude .
By LORD RAYLEIGH , O.M. , F.R.S. ( Received July 8 , 1910 .
) Waves of Finite plitudc without In the investigations which follow , we are concerned with the motion of an elastic fluid in one dimension , say , parallel to .
It is implied not only that there are no component velocities perpendicular to , but that the motion is the same in any perpendicular plane , so that it is a function of and of the time ( t ) only .
If be the velocity at any point the pressure , the density , X an impressed force , the dynamical equation for an inviscid fluid is .
( 1 ) At the same time the " " equation of continuity\ldquo ; takes the form .
( 2 ) The first step , and it was a very important one , in the treatment of waves of finite amplitude is due to Poisson .
* Under the assumption of Boyle 's law , , he proved that for waves travelling in one direction ( positive ) the circumstances of the propagation are expressed by , ( 3 ) " " Memoir sir la orie d Son 'Journ .
de l'Ecole Polytechnique , 1808 , vol. 7 p. 319 .
1910 .
] Plane Waves of Finite Amplitude .
249 neighbouring points on any part of the curve which slopes downwards in the positive direction , and inquire after what time this part of the curve becomes vertical .
If the difference of abscissae be , the hinder point will overtake the forward point in the time .
Thus the motion , as determined by Poisson 's integral , becomes discontinuous after a time equal to the reciprocal , taken positively , of the greatest negative value of For example , let us suppose that , ( 4 ) where is the greatest initial velocity .
When , the greatest negative value of is , so discontinuity will commence at the time The only kind of wave travelling in the positive direction which can escape ultimate discontinuity is one which has no forward slope .
This is the case of a wave forming the transition between a larger constant value of when exceeds a certain value , and a smaller constant value when falls short of a certain value .
As time passes , the slope everywhere becomes easier .
We shall see presently that this wave is a wave of rarefaction , in the sense that during its passage the gas passes from a greater to a less density .
It is worthy of remark that , although we may of course conceive a wave of finite disturbance to exist at aIly moment , there is in general a limit to the duration of its previous independent existence .
By drawing lines in the negative instead of in the positive direction we may trace the history of the velocity-curve ; and we see that as we push our inquiry further and there into past time the forward slopes become easier and the backward slopes steeper .
At a time equal to the greatest positive value of , antecedent to that at which the curve is first contemplated , the velocity would be discontinuous .
The exception is now a wave of condensation , involving a passage always from a less to a greater density .
When discontinuity sets in , a state of things exists to which the usual al equations are inapplicable ; and the subsequent progress of the motion has not been determined .
It is probable , as suggested by Stokes , that some sort of reflection would ensue .
In regard to this matter we must be careful to keep purely mathematical questions distinct from physical ones .
We shall see later how the tendency to discontinuity may be held in check by forces of a dissipative character .
But this has nothing directly to do with the mathematical problem of determining what would happen to waves of finite amplitude in a medium , free from viscosity , whose pressure is under all circumstances proportional to the density .
* To suppose that the problem has 'Theory of Sound , ' 1878 , S251 .
equations .
* A formula equivalent to ( 5 ) was given explicitly , so far as I know , for the first time by Airy who attributes it to De Morgan .
The assumption that in a progressive wave there is a definite relation between and forms the basis of Earnshaw 's investigation .
That such a relation is to be expected may be shown by a line of argument analogous to that already employed in connection with Poisson 's integral .
Whatever may be the law of pressure as a function of density , the velocity of propagation of small disturbances is according to the usual theory equal to , and in a positive progressive wave the relation between velocity and condensation ( s ) is , ( b ) where .
If this relation be violated at any point , a wave will emerge , travelling in the negative direction .
Let us now picture to ourselves the case of a positive progressive wave in which the changes of velocity and density are very gradual but become important by accumulation , and let us inquire what condition must be satisfied in order to prevent the formation of a negative wave .
It is clear that the answer to the question whether or not a negative wave will be generated at any point will depend upon the state of things in the immediate neighbourhood of the point , and not upon the state of things at a distance from , and will therefore be determined by the criterion applicable to small disturbances .
In applying this criterion we are to consider the velocities and condensations , not absolutely , but relatively to quations , p. 364 , and ( b ) , p. 367 .
'Phil .
Mag 1849 , vol. 34 , p. 402 .
The corresponding formula for long tidal waves of finite amplitude was also given .
'Roy .
Soc. Proc January 6 , 1859 ; 'Phil .
Trans 1860 , p. 133 .
Su : .
1910 ] Aerial Plane oWaves of Finite Amplitude at the for 251 those prevailing in the neighbouring parts of the medium , so that the form of ( 6 ) proper for the present purpose is , ( 7 ) whence , ( 8 ) which is the relation between and generally necessary for a positive progressive wave , as laid down by Earnshaw .
* Earnshaw worked with the so-called rangian form of the equations , in which the motions of particular particles are followed , and he obtained complete solutions for a wave progressive in one direction .
In the case of Boyle 's law the relation between velocity and density is that already given ( 5 ) , and in the case of the adiabatic law , where ( 9 ) Earnshaw finds , ( 10 ) where is the velocity of infinitesimal disturbances under the condition represented by , viz. .
In ( 9 ) denotes as usual the ratio of the two specific heats ; and in ( 10 ) , applicable to a positive progressive wave , the constant of integration has been so chosen that corresponds to The generalised form of Poisson 's , appropriate when is given function of , does not appear quite explicitly in Earnshaw 's memoir .
The line of argument already used shows that it must be .
( 11 ) In the case of a gas obeying Boyle 's law , const .
, ( 12 ) and ( 11 ) reduces to Poisson 's form .
In the case of the adiabatic law , we have from ( 9 ) , ( 10 ) , , ( 13 ) so that ( Earnshaw .
( 14 ) Thus ( 11 ) assumes the form , ( 15 ) and this with ( 10 ) may be considered to constitute the solution of the 'Theory of Sound , ' 1878 , S251 .
Lord Rayleigh .
[ July 8 , problem up to the point where discontinuity sets in .
We may fall back upon Boyle 's law by putting in ( 15 ) It appears that whether the relation of pressule to density be isothermal or adiabatic , there is a change of type as the wave advances .
There can be no escape from such a change unless be constant .
Using ( 8 ) , we may deduce in this case being a constant , whence ( 16 ) expresses the only law of pressure under which waves of finite amplitude can be propagated without undergoing a change of type ( Earnshaw ) .
simpler derivation of ( 16 ) will be given presently .
Earnshaw further considers the genesis of disturbance in a gas originally at rest by the motion of a piston ( supposed to be contained in a tube ) , but some of his conclusions appear to need revision .
All that is required in these problems is virtually contained in ( 8 ) , ( 11 ) .
If X denote the position of the piston at time , the velocity of its motion is , and this velocity is shared by the gas in contact with it .
On the positive side the velocity of propagation of ( equal to u ) is , by ( 11 ) , ; so that if at time ( greater than T ) , is to be found at , we must have .
( 17 ) Amon the problems which naturally gest themselves would be to determine what happens when the piston originally at rest at begins to move at time with a constant velocity .
Buf , if this velocity be positiye , the discontinuity , which immediately ensues , causes the failure of our equations .
On the other hand , if the constant velocity be negative , say , the initial discontinuity disappears forthwith , and the subsequent motion may be traced .
Take , for example , the case of Boyle 's law .
We have ; so that 4 , and we have to consider where the velocities and are to be found at time .
Now corresponds to the range of from to , so that ranges from to .
Again , corresponds to the range of from to , so that ranges from to .
The whole range of on the positive side of the piston is now accounted for , except the interval .
from to This is occupied by the transition of velocity 1910 .
] Plane Waves of Finite Amplitude .
253 from to , and we infer , from what was said in the discussion of , Poisson 's integral , that this transition must take place linearly .
Under Boyle 's law the relation between velocity and density ( 5 ) is such that , however fast the piston may recede ( negative ) , a complete vacuum can never be formed behind it .
It is otherwise under the adiabatic law ( 10 ) , where corresponds to ( Earnshaw ) .
It may be of interest to consider further a few examples of ( 17 ) .
Still Boyle 's law , let us suppose that the piston is at rest until , and then moves with uniform acceleration ( g ) , so that The use of these in ( 17 ) gives , ( 18 ) showing that the relation between and is parabolic .
In ( 18 ) we see that vanishes , when .
Thus , if be positive , i.e. if the wave be one of condensation , discontinuity sets in at the front after an interval , reckoned from the beginning of the motion , equal to .
But if be negative , there is no discontinuity , and ( 18 ) relnains valid for an indefinite time .
In general , as in the last example , the discontinuity sets in locally at one point of the velocity-curve , while other parts are temporarily exempt .
It is of interest to inquire under what law the piston must advance so as to generate a linear velocity-curve .
For then , under the adiabatic law , which includes Boyle 's , a velocity-curve , once linear , remains linear , and if discontinuity enters , it must affect the whole curve simultaneously .
It may be worth while to pause here for a moment to inquire what law of pressure is implied in the permanence of the linear character of the velocity-curve .
By ( 7 ) so that if is a linear function of must also be a linear function of .
This requires that where and are constants , and the most general relation between and consistent with the requirements is , ( 19 ) where , are constants .
The relation ( 19 ) may be regarded as a kind which completely expresses the motion of the piston .
The corresponding velocity-curve at time may be verified by means of ( 20 ) , ( 21 ) .
It is expressed by at\mdash ; x ' ( 24 ) exhibiting the linear character of the slope of velocity .
Evidently the slope becomes vertical throughout when In the above example it is not necessary to suppose the law of motion of the piston continued up to .
On the contrary , we may imagine to increase up to some prescribed finite value and then to remain constant .
In this case the slope expressed by ( 24 ) forms the transition between for values of greater than and the finite value at which the acceleration of the piston stops .
this cogue oshows tncreases inplitude.arefaction w numerical value but does not become infinite in any finite time .
The analogue of ( 24 ) is , ' ( 25 ) slope which ever grows easier as time passes .
The problem also admits of solution when the gas follows the adiabatic law ( 9 ) .
As in ( 15 ) , ( 20 ) , ; ( 26 ) and ( 26 ) is to satisfy the condition of making constant .
In this so that const .
On integration we obtain , under the condition that and vanish ether , , ( 27 ) being a constant which , if positive , corresponds to , thereby determining as a function of T. For the value of , we have .
( 2S ) It may be observed that , although is infinite when remains finite .
Using this in ( 26 ) , we get finally , on reduction , , ( 29 ) or .
( 30 ) If is a linear function of , and ( being positive ) the slope of the velocity curve increases until it becomes vertical when If is negative , the wave is one of rarefaction , and ( 30 ) applies however great may be .
By putting we fall back from ( 30 ) to ( 24 ) , and less simply from ( 28 ) to ( 23 ) .
* I have since found that this problem was successfully treated by Hugoniot .
whence and also .
The velocity with which travels in such a wave is accordingly , of which fact ( 11 ) is merely another form of statement .
Riemann 's equations are more general than anything previously given , as not limited to a single progressive wave .
Since Riemann 's equations do not seem to have been applied in any example of continuous motion , I have thought it worth while to inquire whether they can be satisfied when and both linear functions of Boyle 's law being assumed , so that in ( 32 ) , ( 33 ) If we suppose , ( 34 ) we obtain , on substitution in ( 31 ) , equations for the determination of , as functions of the time .
In the first instance we find A-C , ( 35 ) in which to a constant may be added , and further , ( 36 ) 'Gottingen Abhandlungen , ' 1860 , vol. 8 .
1910 .
] Plane Waves oj Finite Amplitude .
when is an arbitrary constant .
Also - , ( 37 ) being another arbitrary constant , and thence , ( 38 ) , ( 39 ) with .
( 40 ) If we allow the origin of to be arbitrary , as well as that of , we may write ( 41 ) , ( 42 ) , ( 43 ) .
( 44 ) If , the logarithmic term disappears , and either or is constant .
In these cases we fall back upon single progressive waves .
If in ( 43 ) , and ( 44 ) gives proportional to .
The density is thus uniform with respect to , but the volume of a iven mass grows ) ortionally with .
The uniform expansion occurs in such a manner that the gas remains unmoved at the origin of co-ordinates .
Since in this case , we see that every part of the gas moves with unaccelerated velocity .
of Permanent When waves are ated in one dimension without change of type , the circumstances are dynamically same as in steady motion , as appears at once by impressing on the system a velocity equal and opposite to that of wave-propagation .
The problem may conveniently be considered under this form .
From the general equation of continuity ( 2 ) , by making equal to zero , or independently , we have , ( 45 ) where is a constant which Rankine called the mass-velocity .
The dynamical equation ( 1 ) reduces to .
( 46 ) 1910 .
] Aerial Plane Waves of Finite Amplitude .
259 The actual force operative upon the element of mass is .
Thus , on integration , , ( 52 ) if the of integration extend from the place where the velocity is to the place where it becomes equal to .
The integral applied force vanishes if the terminal velocities are such that their geometric mean is .
We may apply this to the case of a velocity-curve giving a simple gradual transition from one constant velocity to another constant velocity ? .
Under the above condition the integral force vanishes , but finite forces are required at all points of the slope , except the particular point where It is of some importance to notice that although , under the condition , the applied forces contribute on the whole no momeJltu , yet they do contribute energ ?
f , positive or ative .
To find the work done in unit of time by the forces we have .
( 53 ) The better to interpret this let us suppose that and are positive , and in the first instance that , so that the fluid passes from a less to a greater velocity , or by from a greater to a less density .
In the case of a wave of arefaction we have therefore to consider the sign of , ( 54 ) when .
It is not difficult to prove that this sign is always positive .
When is small , the approximate value of ( 54 ) is , and is therefore positive when .
Again , if we lemove the positive factor from ( 54 ) and then differentiate , we obtain , which is positive .
Hence , , ( 54 ) is necessarily positive .
The propagation of the wave of rarefaction without of type requires that the impressed forces , contributing on the whole no momentum , should nevertheless do work upon , i.e. communicate energy to , the gas .
In like manner , if the wave be one of condensation , i.e. , if the gas passes from a less to a greater density , the operation of the impressed forces is to remove energy from the gas forming the wave .
It follows that although dissipative forces , such as those arising from viscosity , may possibly constitute a machinery capable of maintaining the type of a wave of condeusation , in no case can they maintain the type of a wave of rarefaction .
It is desirable to extend this argument to waves propagated under the adiabatic law .
In general , from ( 45 ) , ( 46 ) , ; so that could be satisfied .
Results equivalent to his may be deduced from ( 45 ) in connection with the condition already found from ( 52 ) to express that there is no change of momentum on the whole .
Thus , .
( 57 ) Similar conclusions were put forward by Riemann in 1860 ( loc. cit Commenting on these results in the " " Theory of Sound\ldquo ; ( 1878 ) , I pointed out that although the conditions of mass and momentum were satisfied , the condition of energy was violated , and that therefore the motion was not possible ; and in republishing this Stokes admitted the criticism , which had indeed already been made privately by Kelvin .
On the other hand , and H. WeberS maintain , at least to some extent , the original view .
*Compare Lamb 's 'Hydrodynamics , ' S280 .
'Collected Works , ' vol. 2 , p. 65 .
'Phil .
Mag 1893 , vol. 35 , p. 316 .
S 'Die Partiellen Differentialgleichungen der Mathematischen Physik , ' Braunschweig , 1901 , vol. 2 , p. 496 .
1910 .
] Aerial Plane of Finite Amplitude .
261 Inasmuch as they ignored the question of energy , it was natural that S Stokes and Riemann made no distinction between the cases where energy is gained or lost .
As I understand , Weber abandons Riemann 's solution for the discontinuous wave ( or bore , as it is sometimes called for brevity ) of rarefaction , but still maintains it for the case of the bore of condensation .
No 4 !
doubt there is an important distinction between the two cases ; nevertheless , I fail to understand how a loss of energy can be admitted in a motion which is supposed to be subject to the isothermal or adiabatic laws , in which no dissipative action is contemplated .
In the present paper the discussion proceeds upon the supposition of a gradual transition between the two velocities or densities .
It does not appear how a solution which violates mechanical principles , however rapid the transition , can become valid when the transition is supposed to become absolutely abrupt .
All that I am able to admit is that under these circumstances dissipative forces ( such as viscosity ) that are infinitely small may be competent to produce a finite effect .
If we suppose that under the influence of small diSSipatiye forces the bore of Stokes and Riemann can be propagated , at least approximately , we naturally inquire whether it can be regarded as the complete outcome of the simple progressive wave with a straight velocity slope which , as we have found , tends after a definite interval of time to assume the character of a bore .
It would seem that the answer must be in the negative .
Boyle 's law , we recognise from ( 5 ) that in the progressive wave , just before the formation of the bore , the relation between the velocities and densities is while in ( 57 ) the relation is .
The two functions of on the right , which are independent of any common addition to and , cannot be identified ( unless ) , as we have found already in ( 54 ) .
This incompatibility may be regarded as a confirmation of Stokes ' opinion that something of the nature of reflection must ensue .
Permanent Begime under the fluence of Dissipative Forccs .
The first ation to be considered under this head is a very remarkable one by Rankine " " On the Thermodynamic Theory of Waves of Finite Longitudinal Disturbance *which ( except a limited part expounded 'Phil .
Trams 1870 , vol. 160 , Part II , p. 277 .
VOL. LXXXIV.\mdash ; A. .
( 58 ) In order to effect the changes specified by and , it is in general necessary to communioate heat to the gas .
Calling the necessary quantity of heat we may write .
( 59 ) Suppose now that .
Equations , ( 59 ) , give ( constant ) where ( constant ) expresses the specific heat of the gas under constant pressure .
Denoting this by , we have notices toisson , Stokes , Airy , shawwithout acquaintance writings oredecessors ; ement h\ldquo ; emoir seems their o 1910 .
] Aerial Plane Waves of Finite Amphtude .
263 Again , suppose ( b ) that .
We find in a similar manner that if denote the specific heat under constant volume , Thus , in general , .
( 60 ) If between ( 58 ) and ( 60 ) we eliminate , there results .
( 61 ) In ( 61 ) corresponds to adiabatic expansion , when according to Mayer 's principle the cooling effect is equal to the external work done by the gas .
Hence , ( 62 ) and therefore , ( 63 ) a relation discovered by Rankine himself in 1850 .
Rankine then applies ( 60 ) to find what heat must be communicated in order that the gas may follow Earnshaw 's law making .
With regard to ( 62 ) , it appears that .
( 64 ) It will be understood that under the condition now imposed of Earnshaw 's relation , as well as of the ordinary gas law , there remains but one independent variable , and that the state of the gas may be expressed in terms of any one of the three quantities We have next to consider how far the necessary supply of heat defined by ( 64 ) can be effected by conduction .
If the initial state ( distinguished by suffix 1 ) and final state ( with suffix 2 ) be of uniformity with respect to , the total quantity of heat received by the gas during its passage must be zero , or .
Hence from ( 64 ) Rankine finds .
( 65 ) This is a necessary condition ; but of course there is nothing so far to show that it is sufficient .
In ( 65 ) are any corresponding values of and within the wave , and we may identify them with Thus .
( 66 ) * We may , of course , also identify with 1910 .
Aerial Plane Waves of Finite Amplitude .
265 .
If this be negative , the transformation is not possible .
From ( 64 ) with use of .
( 70 ) $i In ( 68 ) we may give any corresponding values found in the wave .
Thus lies between and , and is positive .
Accordingly takes the same as .
( 71 ) The integral ( 71 ) is evaluated without difficulty .
Dropping the factor , and writing , we get .
( 72 ) It is evident that ( 72 ) when we substitute for If we expand ( 72 ) in powers of we find ( 72 ) the terms in , , disappearing .
Thus , when is positive , ( 72 ) positive .
Differentiating ( 72 ) with respect to , we get .
( 73 ) When ( 73 ) is reduced to a single fraction , the denominator is positive , and the numerator is We infer that if , ( 72 ) is always positive , and that if , ( 72 ) is always negative .
Hence if , i.e. if the wave be one of condensation , the communications of heat required are such as may arise from conduction ; but if the wave be one of rarefaction , its permanency can in no wise be attained as the result of conduction .
A wave of condensation here means a wave such that during its progress the gas passes always from a less dense to a more dense state , and the most important case is when the limits are finite , so that the passage constitutes the transition om one uniform density to a greater uniform density .
Rankine proceeds to examine more particularly under what conditions a wave can be permanent .
' In order that a particular type of disturbance may be capable of permanence during its propagation , a relation must exist between the temperatures of the particles and their relative positions , such 1910 .
] Aerial Plane Waves of Finite Amplitude .
267 : that the numerator is also positive for that part of the wave where is nearly gS equal to .
Thus , for this part of the wave at any rate , and increase together ; or , since is positive , the gas passes to a condition of greater density\mdash ; the wave must be one of condensation .
This consideration , as we have seen , Rankine overlooked .
And a further limitation presents itself : since there cannot be two pressures in one place , it is evident that must not change sign .
The numerator in ( 79 ) must be positive over its whole from to , and this will nob be the case if exceed , al for common gases to .
The conclusion is that the only kind of wave , involving a transition from one uniform pressure to another , which can be maintained with the aid of conduction is a wave of condensation , and then only when the ratio of pressures does not exceed a moderate value .
The next contribution to the subject upon which I have to comment is contained in a and ably written memoir by Hugoniot .
* This author , though he covers to a great extent the same ground , makes no reference to Stokes , Earnshaw , Riemann , or Rankine , and but a very slight one to Poisson\mdash ; a circumstance which increases the difficulty of comparison .
Since Hugoniot uses the rangian form of equation , his ation runs naturally on the same lines as Earnshaw 's , whose general solution for a single progressive wave is reproduced .
I have already alluded to the solution of special problems relating to the propagation of a wave of variable type .
The most original part of Hugoniot 's work has been supposed to be his treatment of discontinuous waves involving a sudden change of pressure , with respect to which he formulated a law often called after his name by French writers .
But a little examination reveals that this law is precisely the sam as that given 15 years earlier by ankine , a fact which is the more surprising inasmuch as the two authors start from quite different points of view .
Rankine 's ation , as we have seen , is expressly based upon conduction of heat in the gas , but Hugoniot supposes his gas to be non-conducting .
A question of some delicacy is here involved , which will repay careful examination .
It will be convenient to give a paraphrase of 's argument .
This argument depends upon an application of the principle of energy to a region bounded by two fixed planes , the place of discontinuity .
The work done by the fluid as it emerges with volume against the pressure is de Polytechnique , ' 188 , 1889 .
Compare Lamb 's ' Hydrodynamics , ' 1906 , S 280 .
nternal incoming gxpressedl 9plitude .
in If and lie upon the same adiabatic , the work corresponding to the passage from the one state to the other is given without uity by ( 84 ) .
But in the present case the two states do not lie upon the same adiabatic , and the work required is deduced upon the assumption that nothing is involved in the passage at from one adiabatic to the other .
What is actually there required is the communication ( positive or negative ) of an infinitesimal quantity of heat .
From the point of view of the first law the infinitesimal quantity of heat may be ected , but not so from the point of view of the second law , since the transfer is supposed to take place at the zero of temperature .
When heat and work are distinguished , infinitesimal heat at zero may have a finite value .
The imaginary passage to infinity has the advantage of leading rapidly to the required conclusion , but it rather tends to obscure the real nature of the process .
While all the other items of the account are mechanical work , the passage from one adiabatic to the other ( which may take place at constant finite volume ) is a question of heat as distinguished from work .
If during a complete cycle work would be lost and corresponding heat gained , the operation is dissipative and there need be no contradiction if viscosity or heat-conduction enter , but the opposite contingency of a gain of work at the expense of heat is excluded in all cases .
The conclusion is the same as before .
While a wave of condensation nay , perhaps , maintain a permanent regime as the result of dissipative agencies , a permanent wave of rarefaction is excluded .
It is remarked by Hugoniot that even when the ratio is infinite , does not exceed , which for common gases is equal to about 6 .
A similar remark is made by Duhem , discusses the whole question with great generality .
With regard to perfect gases " " lorsqu'une quasi-onde de choc se propage au sein d'un gas parfait , le fluide le plus condense est toujours en amont de l'onde et le fluide le moins condense en aval But , so far as I see , neither of these authors proves that the propagation is possible in any case .
It is a question of great interest to inquire what is the influence of viscosity and especially whether alone , or in co-operation with heat-conduction , it allows a wave of condensation to acquire a permanent .
We proceed to consider this question on the basis of the usual equations , although it must be admitted that their application to conditions which are somewhat extreme raises points of uncertainty .
Reverting to our.original equations , we recognise that ( 45 ) is unaffected ' Zeitschrift .
Physikal .
Chem 1909 , vol. 69 , p. 169 .
, ( 89 ) the origin of being chosen The transition of volumes from to occupies , mathematica ] lie speaking , the whole range from to , but practically it may be very sudden .
Since in ( 88 ) never changes sign , the condition of permanency for a condensational wave can always be satisfied , whatever may be the value of the ratio or , contrasting in this respect with the limitation found to be necessary on Rankine 's conclusion relative to heatconduction .
pressures witation pointed o ugoniot still obtains ; otherwise , one of the 1910 .
] Aerial Plane of Finite nplitude .
271 As regards the velocity of wave propagation into the rarer medium , we have for its square .
( 90 ) Returning to the case where heat development and viscosity are both regarded , we see that in virtue of ( 81 ) Hugoniot 's is still applicable without change , and it leads to the same final relation ( 85 ) as was found by Rankine when heat-conduction is alone considered .
In endeavouring to apply Rankine 's method to the more general case where viscosity is retained , we shall find it more convenient to treat , or , rather than , as independent variable .
If , as before , denotes the total quantity of heat received by unit mass of the gas , we have from ( 60 ) , ( 62 ) , ; or , on elimination of by means of ( 87 ) , .
( 91 ) In ( 91 ) consists of two parts , the first , with which alone Rankine dealt , the heat received by conduction , and the second the heat developed internally under viscosity .
As regards the latter , the heat developed in volume and time is which we are to replace by , and by , so that .
( 92 ) Multiplying this by and subtracting it from ( 91 ) , we ( 93 ) As in Rankine 's ation , the whole heat received by conduction in passing from one uniform state to another uniform state must vanish .
Hence , on integrating between these limits , and dividing out the factor , we have , or , as in ( 86 ) , the same relation as was found by Rankine .
Introducing it into ( 93 ) , we get .
( 94 ) * Lamb 's 'Hydrodynamics , ' S341 .
is constant , the solution is nearly the same as in ( 89 ) , and , in fact , reduces to it when , which represents Boyle 's law .
This case of no conduction is thus satisfactorily disposed of .
Whatever be the ratio of pressures , a wave of condensation is always possible .
lt should be remarked , however , that the supposition of constant does consist with the facts as known for actual gases when differs from unity .
For such gases viscosity , though independent of density , varies with ternperature , so that will not be constant in ( 95 ) .
But since is always positive , this complication merely affects the particular form of the integral and not the general conclusion as to the possibility of a permanent wave .
In general , from ( 94 ) , if we reckon from the terminal state .
( 96 ) The equation of conduction is the same ( 74 ) as before .
And for , from ( 87 ) , ; so that with regard to ( 62 ) the equation of conduction becomes .
( 97 ) By omitting the terms containing we may of course fall back on Rankine 's problem .
Equation ( 97 ) , in its general form , is much more complicated than when either viscosity or heat-conduction is alone regarded , in consequence of the occurrence of the differential coefficient of the second order .
In general , both and are functions of temperature , and therefore of ; but , according to Maxwell 's theory , which assumes a molecular repulsion inversely as the fifth power of the distance , is independent of temperature ( as well as of density ) , and takes the value .
And it would seem that this independence of temperature and density is general , seeing that the ratio is of no dimensions , at least so long as the repulsive force can be represented by an invel.se power of the distance .
* We shall write for the above ratio and * Compare ' Roy .
Soc. Proc ) , vol. 66 , p. 68 ; 'Scientific Papers , ' vol. 4 , p. 463 .
1910 .
] Aerial Plane of Finite Amplitude .
273 .
assume that for a iven gas it is an absolute constant .
Thus , being written for , ( 97 ) takes the form\mdash ; ; ( 98 ) in which may be regarded as the dependent variable .
For we shall write , and if , ( 99 ) our equation , since it contains only through , may be reduced to one of the first order in and , i.e. , , ( 100 ) where , ( 101 ) and .
( 102 ) If can be found as a function of from ( 100 ) , follows by simple integration of ( 99 ) .
In considering equation ( 100 ) we may conveniently as the linear co-ordinate of a material particle of unit mass moving in a straight line with velocity U. The first term , , then represents the acceleration of the particle ; the second , , may be regarded as a resistamce , proportional to the velocity , and at the same time variable with the position ( ) ; and the third term on the right hand represents a force , which is also a function of position .
If be the time in this subsidiary problem , and ( 100 ) may be written while by ( 99 ) .
( 104 ) If be constant , the substitution of for in ( 98 ) is obvious .
If we take , ( 101 ) , ( 102 ) , become , ( 105 ) .
( 106 ) It will be observed that , over the range from to is positive , but that the sign of is doubtful .
If has the greater terminal value is negative ; but , when it has the smaller terminal value , the sign depends upon the ratio .
If this ratio is negative ; otherwise it is positive .
105 ) we reverse the sign of We consider , then , the motion of the particle to be in the positive direction , from to , with zero velocity at both limits , the motion between and being aided by the force which itself vanishes at these limits , and being also subject to a force of the nature of resistance , proportional to velocity .
When does not exceed , the force is a resistance in the sense , i.e. it opposes the motion , and , in any case , it has this character near ( and beyond ) the arrival end But when exceeds , the force becomes what we may call a counterresistance , and aids the motion near the initial end .
As regards , in the neighbourhood of each limit it becomes a force proportional to distance therefrom , repulsive near , and attractive near .
Thus , when is nearly equal to ; ( 107 ) and when is nearly equal to .
( 108 ) The particle , starting from , is bound to go through to .
If it arrives at with zero velocity , we shall have , presumably , a solution of our problem .
It is possible , however , that on first arrival at , it may pass through , and only settle down after a number of oscillations .
To this there does not appear to be any objection ; but if on the return from it overshoots , it can never again return to , since on the left of the sign of is negative ; and then our problem has no solution .
On the other hand , from the nature of , it is not possible for the particle passing through in the positive direction to escape returning .
The character of the start from can be investigated with the aid of approximate equations .
Thus , in ( 100 ) we may treat as constant , say nemay bither positive onegative , mplitude , as in , where is positive , so that .
( 109 ) If in ( 109 ) we assume a , ( 110 ) we find that the equation is satisfied provided that , ( 111 ) one value being positive and one ative .
In the present case , where must be positive when is to be chosen .
The differential equation ( 109 ) can be made homogeneous , and its general solution , *when is 1eal , can be put into the form , ( 112 ) when is an arbitrary constant .
This solution , of course , covers the cases where the particle starts from with a finite velocity , and it appears that .
We conclude from this that when , then , but the conclusion is not safe .
If , however , and are of the same order of magnitude , we may write , where r is not infinite .
this in ( 112 ) , we get When vanishes , the third factor on the left is zero , and , since the first and second factors are not infinite , the conclusion follows that .
This takes us back to ( 110 ) , ( 111 ) , the second solution ( involving ) relating to the case where is negative , the motion one of approach from the positive side to These conclusions may be arrived at more easily from ( 103 ) , of which the general solution in the present case is , ( 114 ) giving From these we may deduce 115 ) A ; whence ( 112 ) follows by elimination of .
For our present purpose , is to vanish when , so that ; and ( 110 ) follows with .
It will be remarked that ( 110 ) makes , as determined by ( 99 ) , infinite when * See , for example , Boole 's ' Differential , Equations , ' p. 33 .
1910 .
] Plrme of Finite Amplitude .
277 If has the same as , so that an increment of velocity communicatsd at any point remains throughout of the same .
Again , if bs throughout of one , as dependent upon it , has the same For example , if be positive over the considered , positive , and negative , then is certainly negative .
Ths increments may be vanishing ovel any part of the The application.to the present is obvious .
If particle passing any point between and , with velocity , arrives at for the first time without velocity , it will still al.rive at without velocity ( it must in any case arrive , since is positive ) , if be diminished , or if be incleased , or if be diminished , or if all these changes occur together .
And in the limit , when is closely approaohed , the ratio of to is in general the same .
By use of this principle we may assure ) selves as to the possibility of a solution in certain cases where does not greatly exceed unity .
We imagine a simplified problem which admits of analytical solution and is derived from the actual one by alteratio1Js which eyerywlJere ( over the from to ) increase and diminish .
If this modified problem admits of the solution required , a fortiori will the problem do so .
If we consider the curve which according to ( 106 ) epresents F as a function of , we see that it is concave downwards , amd that will everywhere be increased if we substitute for the curve the two terminal tangents at and , whose equations are given in ( 107 ) , ( 108 ) .
The abscissa of the point of intersection , is .
FIG. 1 .
FIG. 2 .
As regards its value is given by ( 105 ) with sign reversed , and is diminished when is diminished .
The changes will therefore be in the required direcoion if we represent over by its value at , viz. , , ( 122 ) and from to by its value at , ( 123 ) if for brevity we write for , so that is the ratio of terminal densities in the original LXXXIV , 1910 .
] Plane Waves of Finite Amplitude .
279 S by the broken line formed by three of its tangents , of : which two are the same terminal tangents as before , while the third , , may be taken to be the horizontal tangent parallel to .
( Fig. 2 .
) By ( 106 ) ; the point where is a maximum is determined by , and the corresponding value of is ' 0 .
The abscissae of the points of intersection of the horizontal tangent with the terminal tangents are given by ( 107 ) , ( 108 ) .
For , and for .
As for we are to take the value at MN the value at NI ; and along the value at , as iven by ( 105 ) , with sign changed .
For the two terminal portions the solutions are of the same form as before , but for the middle portion a new form is required .
Making constant in and for , we find on integration , ( 128 ) where is an arbitrary constant , to be determined so as to suit the velocity with which the particle arrives at M. As before , the limiting value of is determined by the consideration that , for our purpose , the arrival at must not be oscillatory .
We get about , and with this value of it is not difficult to show that the velocity of arrival at is below the value prescribed by the solution for .
The necessary conditions are thus fulfilled and we infer that the wave of uniform regime is possible so long as .
Although this ratio is moderate , it exceeds that found admissible in Rankine 's problem , which leaves viscosity out of account .
We there found that the greatest admissible ratio of pressures was , which by ( 68 ) corresponds to for the ratio of sities .
Of course , in Bankine 's problem the solution definitely fails at this point , while in the present problem all that we have so far proved is that the limit exceeds From the low limiting values of found necessary in these two cases in order to secure the reality of the roots of ( 118 ) , it might be inferred that the reality would fail for the limiting motion in the neighbourhood of , when had a considerable value , but such is not the case .
If we use the value of from ( 105 ) appropriate to the terminal point itself , we have ; and .
making .
Hence the correction to , is equal to 1910 .
] Waves of Finite Amplitude .
281 , and .
Thus , corresponding to , we get .
The following are the values of obtained successive ] in this The correction to is everywhere subordinate .
In the last step from to 36 no correction to is applied .
There is a little difficulty in tracing by this method and with full accuracy the final progress to zero when .
If any doubt be left , it may be removed by applying the former method to the course from 35 to 36 , using the value of appropriate to 35 and the terminal tangent as representative of the curve for F. From this it appears that eyen if at were as great as , the moving particle could not pass The conclusion is that even in this extreme case of the solution exists , and that a wave of permanent regime is possible .
Further , from ( 99 ) we see that since and are both positive , is positive throughout , and the transition from the one density to the other takes place without alternation .
After what has been proved little doubt remain but that a solution is possible when has any value lower than 6 .
I have , however , lought it desirable to add a rough calculation ( rough on account of the relative magnitude of the steps ) for the of :\mdash ; It is a question of some importance to consider what is the thickness of the transitional layer in the waves of uniform regime which have been proved to be possible .
Mathematically speaking , the transition occupies an infinite space ; but if we understand the expression to refer to a transition approximately complete , the thickness involved is finite , and indeed extremely small .
Reference to ( 98 ) shows that is of the order , or or , where is tho velocity of the wave .
For the present purpose we 1910 .
] Aerial Plane Waves of Finite .
283 related , not according to the adiabatic law , but to Rankine 's law already discussed .
The which occur may be separated into two stages .
The first is the sudden one in which the fluid passes from the !
atmospheric condition , with velocity ?
to the condition denoted by .
After passing the bore the fluid changes gradually according to the adiabatic law already stated until at the nose of the bullet the condition is represented by , with We are now in a position to calculate the final pressure .
For the first stage we have Rankine 's formula ( 67 ) , making or , if , ( 135 ) determining the pressure just inside the bore in terms of ( the velocity of the bullet quiescent air ) and , the ordinary velocity of sound .
When .
For values of less than , the first does not exist and we may suppose In the second stage we use ( 134 ) with .
Thus , ( 136 ) in which , by a formula analogous to ( 66 ) , Hence .
( 137 ) When is to be calculated from ( 135 ) .
When the resulting value is substituted in ( 137 ) , is determined .
When , we have simply .
( 138 ) If be small , ( 138 ) reduces to or , ( 139 ) as for an incompressible fluid .
When , both give .
( 140 ) From this point onwards the approximate formula ( 142 ) may suffice .
The values found are in good agreement with Mr. Mallock 's curve .
The question as to the linear interval between the bore and the nose of the bullet cannot be answered from the results of the present paper .
In strictly one-dimensional motion , the bore and the plane wall constituting the obstacle could not move at the same speed .

