II .
" On the Integrability of certain Partial Differential Equations proposed by Mr. Airy .
" By R. MOON , M.A. , late Fellow of Queen 's College , Cambridge .
Communicated by Professor J. J. SYLVEaSTER .
Received April 30 , 1867 .
( Abstract .
) The equation -= , +-4-y , ... ( 1 c1y " d/ .v die where a , f. , y are functions of x , includes two equations recently proposed for solution by Mr. Airy , and affords a good illustration of the ordinary incapacity of partial differential equations of the second order for solutions involving arbitrary functions .
If the above equation admit of an integral solution containing one or more arbitrary functions , it must be capable of being derived from an equation of the form F(:)v / ( ) =ef ( .
, , ) } , e h ... ... .
( 2 ) 486 where F and f are definite , and q is arbitrary .
From this last we get F'(X ) + F'()_p=q , '(f){f'( ) +f '(z)p } , F'(y ) +F ( z)q='(f){f(y ) +f'(z)q } ; and eliminating o ' between these , we shall arrive at an equation between xyzpq which is the only partial differential equation of the first order free from arbitrary functions which is obtainable from ( 2 ) without assigning to pa definite form .
Hence ( 1 ) must be derivable from an equation of the form f(xysp ) = 0 , or of the form 0=q +f(y ) , ... ... .
( 3 ) wheref is definite .
Differentiating , we have d2z d2z O-4-f'(y)+f'(z)q+f'(p ) d d ?
+ f ' ( +f'(p)+ f(Z ) d+ '( dx die +/ '( +/ ) Multiplying the second equation by A ( where A is any function of xyzpq ) and adding , we get ( 1o2 , d2dz + f(p ) +A } ddy +Af'(p ) d+f'(y)+Af'( ) ( 4 ) + ( q+ Ap)f'( ) .
If ( 1 ) admits of a solution of the form ( 2 ) , ( 1 ) must be identical with ( 4 ) , or be capable of becoming so by virtue of ( 3 ) .
Ience 0=A+f'(p ) , -a2=Af'(p ) , Sp + yT=f'(y ) + Af'( ) +(Ap-f )f'(z ) ... ( 5 ) From the two first we get f'(p)= ?
+ , A=-T .
Hencef must satisfy both the equations f'Cp)h ? ?
, }. .
( 6 ) 0 =f !
( y ) ) ?
~f'( ) -(f ?
+ ap)f ' ( )(p + J* ) .
The first gives us f= F(xyz ) + p , where F is arbitrary .
Substituting this value in ( 6 ) , observing that f '(x ) = ''( ) p ( y ) ( f'(z ) we get 0=F'(y ) aF'( -(F + 2)p ) '(z(- " +.+ dO .
( 7 ) But F contains only xyz ; hence , in order that this last equation may hold , the coefficient of p must =0 , i. e. we must have dx(8 ) o=_+2cF'( , )+(3 + ?
... ... ( 8 ) in which case ( 7 ) reduces to 0= F'(y ) aF'(x)--F .
F'()-yz ... . .
( 9 ) F must satisfy both ( 8 ) and ( 9 ) .
By integration of ( 8 ) we get F=F , ( xy)T + z. Substituting this in ( 9 ) , observing that F'(x)F ' +,2( ) lT d+ F'(y)= ( F ' ) , we get 0 =F'(y ) ?
'F ( x)+ .
10 + 11 ( 23 + ^VpF 2i+ d. y But , since F1 contains x and y only , in order that this may hold we must have the coefficient in it of z= 0 , i. e. 0 d^ .
I da 1 " p T , 0d2 ( xa dx ) 2\ ) in which case ( 10 ) becomes =F Fl(y ) a'(x ) +.F , ; a\+ dx FC .
whence by integration we get I ipda+ -1-+ida\ ^F2 ( a2 LD ) ( y dx ) where 9 is arbitrary .
Hence we get for the first integral of ( 1 ) zOen ( 10 ) is satisfied , I+ -PD da J dv-dx in which h case ( 0 b ecoms2a2 o=q+aP 2 ( )+_ Z+ / .
d whea e9 is a whence by ordinary integration we obtain the complete integral of the form , eJ 2u.)-.dx dx We have above a very simple example of a general principle , viz. that in order that a partial differential equation of the second order , or a pair of simultaneous partial differential equations of the first order , may admit of a solution containing arbitrary functions , the coefficients must satisfy a certain equation of condition ; from which it follows that , except in the simplest instances ( in which the terms of the equation of condition vanish ) , there is a moral certainty that such a differential equation or pair of equations which have not been specially selected for the purpose , and whose coefficients do not involve a disposable quantity by which the equation of condition may be satisfied , will not admit of a solution involving arbitrary functions .
The equations applicable to the motion of an elastic fluid along the axis of a tube afford a remarkable illustration of the scope of these remarks .
Those equations consist of a pair of partial differential equations of the first order involving five variables , viz. y , t , p , v , p ; and it may be shown Opriori , that when derived upon a true theory they must be capable of a solution containing two arbitrary functions ; from which it follows that a third equation will require to be satisfied .
For this purpose we have p , the pressure , ready to our hands .
From the fact of the existence of the equation of condition not having been suspected by the founders of the theory of fluid-motion , at the same time that it was absolutely necessary for them to assign a form to p , they had recourse for that purpose to an empirical method ; thus , on the one hand , depriving us of the power of satisfying the requirements of the problem , and on the other , abandoning the means for the determination of p which the analysis furnishes .
It cannot be matter of surprise that the law of pressure suggested under these circumstances should be entirely erroneous , as ( by two other independent methods , one founded upon purely physical , the other upon purely analytical considerations ) I have elsewhere shown .

