]\gt ; [ 579 ] XXIV .
On the Values of the Integral being LAPLACE'S Coefficients of the Orders , with an application to the Theory of Radiation .
By the Eon .
J. W. STRUTT , Fellow of Trinity College , Cambridge .
Communicated by W. SPOTTISWOODE , F.R. Received May 17 , \mdash ; Read June 16 , 1870 .
IN the course of am investigation concerning the potential function which is subject to conditions at the surface of a sphere which vary discontinuously in passing from one hemisphere to the other , it became necessary to know the values of the integral , being LAPLACE'S coefficients of the orders respectively .
The expression for in terms of is ; but the multiplication of two such series together and subsequent integration with respect to would be very laborious even for moderate values of and By the following method the values of the integrals in question may be obtained without much trouble .
According to the definition of the functions so that which shows that is the coefficient of in the expansion of the integral on the left in powers of and On effecting the integration and reducing , we obtain as the quantity to be expanded , *TIIOXSON and TAIT'S Natural Philosophy , p. 624 .
lIDCCCLXX .
580 TEE HON .
J. W. STRUTT ON TlfE VALUES OF A DEFINITE INTEGRAL Since we know a priori that the expansion will only whole positive powers of , we may leave out all the terms in ae ( no negative powers of occur)* .
We thus obtain fn this all terms containing negative powers may be thrown away , as they must finally disappear even if retained .
The terms on the left the first contain only even powers of , and those on the righ , only odd .
It appears , too , that with the even powers of go the odd of , and conversely .
Hence if be both even or both odd there is no part of the coefficient of to be found after first line , and none in the first line unless .
Thus if both odd or both even , unless they are the same , in which case *Professo has remarked that the finite expression itself may moffled so as to get rid of these For so far as the terms containing fractional powers of are conoerned , are identical .
INVOLVJNG PRODUCT OF TWlD LAPLACE'S COEFFICIBNTS .
These results*are immediate consequences of what is known with respect to the values of the integrals in which the integration extends over the whole sphere ; for if are both odd or both even , , is an even function of , and so The peculiar character of the integrals over the hemisphere only shows itself when one of the quantities is even and the other odd .
The coefficient of in the expansion is ; coefficient of ; coefficient of ; coefficient of ; coefficient of The law of formation of these series is obvious , and the coefficient of could , if necesssary , be written down .
From the symmetry of the original expression in and we know that the coefficient of must be the same as that of ; so that , in order to obtain all the integrals required , it is not absolutely necessary to consider the coefficients of odd powers of Nevertheless , in the calculation for instance , it would be much easier to obtain it as the coefficient of in the series which multipIies , than as the coefficient of in the series which multiplies The coefficient of coefficient of ; *They would , of course , be more simply obtained by taking the integration in the first instance from to .
1870 .
coefficient of ; coefficient of ; and so on .
these series the coefficients of for moderate values of and may be calculated with facility .
It is desirable to know the limit of the integral when becomes very large , remaining finite .
A distinction is necessary according as it is the even or the odd suffix which is supposed to increase without limit .
The whole coefficient of is a sum of terms of the form , where is zero , or any positive integer , each term multiplied by a numerical factor , which may be regarded as a function of and .
The general term in the expansion of in powers of is irrespective of sign .
If we put , it becomes The coefficient of is thus a series of terms of the form each term multiplied by a factor depending on and but independent of .
The question is which term has the predominance when increases without limit ?
It appears that it is the one corresponding to ; for the ratio of this to the general term is a fraction which increases without limit with : INVOLVING TIIE PRODUCT OF TWO LAPLACE'S COEFFICIBNTS .
5S$ The value , when is indefinitely great , is therefore identical with the coefficient of in the expansion of 1.3.5. .
Now ; and by a known theorem , when is indefinitely great , Finally , therefore , , when increases without limit , takes ultimately the form If be very great though infinitely small , perhaps , compared this becomes In a similar manner it may be proved that the limit , when increases indefinitely , is If , now , increases without limit , we obtain There is no inconsistency in the non-agreement of the values found when and are indefinitely great , for the limiting circumstances contemplated in the two cases are in reality quite different .
It may be convenient for the sake of comparison to repeat here the equation which is true whether be great or small .
The annexed Table contains the exact numerical values of the integrals for which neither suffix is greater than eleven .
If we fix our attention on a given value of ( say 6 ) , while varies , we see that the integrals , or , rather , those of them which do not vanish , begin by being alternately opposite in sign , and increase in value up to ; that when a change of is missed , and that for greater values of the regular alternation of sign is reestablished in conjunction with a steady diminution in numerical value .
This is in accordance with what might have been expected from the general character of the functions Q. They have their maximum ( arithmetical as well as algebraical ) value , namely unity , when , an even function , , vanishing times , and an odd function , , vanishing times for values of ranging from to 1 inclusive .
THE IfON .
J. W. ON VALUE8 OF A RAL lNVOLVrNG THE PRODUCT LAPLACE'S COEFJICIENT8 .
When , therefore , one of the quantities is la..rge and the other not nearly equal to it , , is affected with a rapidly alternating , and consequently the value of the is comparatively very small .
But if and are nearly equal , the functions , in spite of the rapid alternation , keep together as it were in sign for a considerable fraction of the range of integration , and so the value integral is largely increased .
, for all cases included in the Table it will be found that a relation which is , evidently general , although not very easily proved to be so .
After a good deal of trouble I arrived at the following demonstration:\mdash ; If in the expression whose expansion gives the integrals undel consideration , we put we obtain In consequence of the symmetry of this in respect to and , it may be expanded in a series of positive and negative powers of of the form .
, , . .
being functions of The terms that we are engaged in examining are those in or ' so that the question reduces itself to the determination of as a function of , or at least an examination of its nature .
Now is the term independent of after differentiation of the series with respect to .
Hence term independent in On differentiation and reduction we arrive at the expression from which the art independent of has to be selected .
; 586 Tffl HON .
J. W. STRUTT ON TllE OF A DEFINITE mTEGBAh From the first term we have simply .
As for the second , The term in ) independent of is 1 .
The term in is , and generally , if odd , the part independent of in is Thus the term in independent of an even function of even function of even function of Finally , therefore , where are unknown coefficients , of which we will only determine by areference to the expression from which was obtained , which shows that so that Multiplying by and replacing and INVOLVING THE PBODUCT OF TWO LAPLACE'S COEFFICIENTS .
These are the only terms in the expansion of in which the index of is one higher than that of Having regard , now , to the symmetry in and , we see that generally As an application of some of the results of this investigation I will take the following physical problem .
A spherical ball of uniform material is exposed to the radiation ffom infinitely distant surrounding objects .
It is required to find the stationary condition .
For the sake of simplicity , the surface of the sphere will be supposed to be perfectly black , that is , to absorb all the radiant heat that falls upon it , and NEWTON'S law of cooling will be employed , at least provisionally .
If V denote the temperature , it is to be determined by the equations , ( A ) , ( B ) where is a function of the position of the point on the surface , and denotes the heat received per.unit area at that point , is the conductivity , and the coefficient of radiation .
Equation ( A ) is to be satisfied throughout the interior and ( B ) over the surface of the sphere .
If V be expanded in LAPLACE'S series , ; ; and if be the expansion of in a similar series of surface harmonics , we obtain , on substituting in ( B ) and equating to zero the terms of any order , .
( C ) OCCLXX .
588 TRUTT ON VALUESINTEGRAL of the The mean temperature is seen to be independent of the conductivity and of the size of the sphere .
The case where the heat which falls on the sphere proceeds from a single radiantpoint is not only important in itself , but may be made the foundation of the general solution in virtue of the principle of superposition .
Taking the axis in the direction of the radiant-point , we have over the positive hemisphere , that is , from to while over the negative hemisphere It is required to expand in a series of spherical harmonics .
Let , then is a function of , which is equal to over the positive hemisphere and to - over the negative .
The problem therefore reduces itself to the expression of over the positive hemisphere in a series of functions of even order .
The same series will then give - over the negative hemisphere .
Assume Multiplying by , and integrating with respect to from to all the other terms on the right vanishing .
Now coefficient of in the expansion of or Accordingly When is great the coefficient of approximates to This completes the solution for a sphere exposed to the radiation an infinitely distant source of heat situated over the point If its coordinates are , it is only necessary to replace in by .
INVOLVING THE PRODUCT OF TWO LAPLACE'S COEFFICIENTS .
Hence if denote the intensity of the radiation which comes in direction , the general value of is the integration going all round the sphere .
Now is the same as , where is the harmonic element of of order ; so that It is remarkable that the odd terms in ( except ) are altogether without influence .
The reason is simply that they do not affect the total heat falling on any point of the surface .
For this is expressed by the point considered being taken as pole of , which involves no loss of generality .
Now ( THOMSON and TAIT , p. 149 ) where is a function of not containing When the integration with respect to is effected , all the terms will vanish except that whose coefficient is .
For this purpose , therefore , we may take or and we know that vanishes if be odd and different from unity* .
* The proof given is sufficient for the object in view ; but it may be well to notice that the essential thing is that the two surface harmonics which are multiplied together are either both odd or both even .
A harmonic 590 THE HON . .
W. STRUTT .
ON THE VALUES OF A DfflUNIlB The same thing is true for an ellipsoid or body of any figu1e which lies altogether on one side of every tangent plane , namely , that the terms of odd order in ( except are wholly without influence on it , and for the same reason .
We saw that in the case of a sphere the mean temperature was independent of the conductivity , and also of the size of the sphere ; but this depends on NEWrON'S law of cooling .
A comparison , however , may be made which shall hold good whatever be the law of variation of radiation with temperature : for if the conducting-power of any uniform body ( which need not be oval ) be increased in the same proportion as its linear dimensions , a corresponding distribution of temperature will all the conditions .
Conclusions of interest from a physical point of view may be deduced from the foregoing considerations , but I refrain from pursuing the subject at present , as the physical problem was only brought forward in illustration of the mathematical results developed in this paper .
of even er\mdash ; has identical values at opposite points of the sphere , and one of odd order has contrary values .
The product of two harmonics which are either both even or both odd has therefore the same value when integrated over any portion of the sphere , or over what may be called the opposite portion , or as a particular case over two opposite hemispheres .
The last two integrals are the halves of the integral over the whole sphere , which vanishes by a -known property of these functions .

