]\gt ; XXII .
A Memoir on Prepotentials .
By Professor CAYLEY , F.R.S. Apri18 , \mdash ; Read June 10 , 1875 .
THE present Memoir relates to multiple integrals expressed in terms of the ultimately disappearing variables , and the same number of parameters , and being of the form rhere g and depend only on the variables .
Such an integral , in regard to the , is said to be " " prepotential and in the particular case to be ' : potential I use throughout the language of hyper-tridimensional and are regarded as coordinates of points in dimensional space , the former of them determining the position of an element of attracting matter , the latter being the attracted point ; viz. we have a mass of matter distributed in such manner that , being the element of lower-dimensional volume at the point , the corresponding density is , a given function of , and that the element of mass exerts on the attl.acted point a force inversely proportional to the power of the distance integration is extended so as to include the whole attracting mass ; and the integral is then said to represent the Prepotential of the mass in regard to the point .
In the particular case , the force is as the invelse square of the distance , .
and the integlal represents the Potential in the ordinary sense of the word .
The element of volume is usually either the element of solid ( spatial or dimensional ) volume . .
, or else the element of superficial ( -dimensional ) volume .
In particular , when the surface ( -dimensional locus ) is the ( -dimensional ) plane , the superficial element .
The cases of a -dimensional volume are in the present memoir consideled only incidentally .
It is scarcely necessary to remark that the notion of density is dependent on the dimensionality of the element of volume : in passing from a spatial distribution , , to a supelficial distribution , , we alter the signification of .
In fact if , in order to connect the two , we cvine the spatial distribution as made over an indefinitely thin layer or stratum bounded by the surface , so that at any element of the surface the normal thickness is , where is a function of the coordinates of the element the spatial element is , and the element of mass is ; and then changing the signification of , so as to denote by it the product , the expression for the element of mass becomes 9 , which is the formula in the case of the superficial distribution .
The space or surface over which the distribution extends may be spoken of as the material space or surface ; so that the density 9 is not for any finite portion of the material space or surface ; and if the distribution be such that the density becomes for any point or locus of the material space or surface , then such point or locus , considered as an infinitesimal portion of space or surface , may be excluded from and regarded as not belonging to the material space or surface .
It is allowable , and frequently convenient , to regard as a discontinuous.function , having its proper value within the material space or surface , and having its value beyond these limits ; and this being so , the integrations may be regarded as extending as far as we please beyond the material space or surface ( but so always as to include the whole of the material space or for instance , in the case of a spatial distribution , over the whole dimensional space ; and in the case of a superficial distribution , over the whole of the -dimensional surface of which the material surface is a part .
In all cases of surface-integrals it is , unless the contrary is expressly stated , assumed that the attracted point does not lie on the material surface ; to make it do so is , in fact , a particular supposition .
As to solid integrals , the cases where the attracted point is not , and is , in the material space may be regarded as cases of coordinate generality ; or we may regard the latter one as the general case , deducing the former one from it by supposing the density at the attracted point to become The present memoir has chiefly reference to three principal cases , which I call , and a special case , , included both under A and : viz. these A. The prepotential-plane case ; general , but the surface is here the plane so that the integral is B. The potential-plane case ; , and the surface the plane , so that the integral is C. The potential-surface case ; , the surface arbitrary , so that the integral is D. The potential-solid case ; , and the integral is PROFESSOR CAYLEY ON PREPOTENTIALS .
It is , in fact , only the prepotential-plane case which is connected with the partial differential equation considered in GREEN'S memoir ' On the Attractions of Ellipsoids ' ( 1835 ) , and called here\ldquo ; the prepotential equati For this equation is satisfied by the function and therefore also by and consequently by the integral ( A ) that the prepotential-plane integral ; but the equation is not satisfied by the value nor , therefore , by the prepotential-solid , or general superficial , integral .
But if , then , instead of the prepotential equation , we have " " the potential equation\ldquo ; ; and this is satisfied by and therefore also by Hence it is satisfied by ( D ) the potential-solid integral , provided that the point does not lie within the material space : I would rather say that the integral does not satisfy the equation , but of this more hereafter ; and it is satisfied by ( C ) the potential-surface integral .
The potential-plane integral ( B ) , as a particular ccse of , of course also satisfies the equation .
Each of the four cases give rise to what may be called a distribution.theorem ; viz. given V a function of satisfying certain prescribed conditions , but otherwise arbitrary , then the form of the theorem is that there exists and that we can find an expression for , the density or distribution of matter over the space or surfaoe to whi\amp ; the theorem relates , such that the corresponding integral V has its given value , viz. in A and there exists such a distribution over the plane , in such a distribution over a given sir , face , and in such a distribution in space .
The establishment , and exhibition .
in connexion with each other , of these four distribution-theorems is the principal object of the present memoir ; but the memoir contains other investigations which have presented themselves to me in treating the question .
It is to be noticed that the theorem A belongs to GREEN , being in fact the fundamental theorem of his memoir of , already referred to .
Theorem , in the particular case of tridimensional space , belongs also to him , being given in his 'Essay on the Application of Mathematical Analysis to the theories ofElectricity and Magnetism ' ( Nottingham , 1828 ) , being partially rediscovered by GAUSS in the year 1840 ; and theorem , in the same case of tridimensional space , to -DIBICHLET : see his memoir " " Sir un moyen de verifier l'expression du potentiel relatif a un mass quelconque en ou heterogene Crelle , .
xxxii .
pp. 80-84 ( 1840 ) .
I refer more particularly to these and other researches by GAUSS , JACOBI , and others in an Annex to the present memoir .
On the Prepotential Surface-inteqral.\mdash ; Art .
Nos. 1 to 18 .
1 .
In what immediately follows we require limiting condition , the prepotential of a uniform ( -coordinal ) circular disk* , radius , in regard to a point on the axis ; and in particular the value is required in the case where the distance ( taken to be always positive ) is indefinitely small in regard to the radius R. Writing , where the new variables are such that the integral becomes where is the element of surface of the -dimensional unit-sphere ; the integral denotes the entire surface of this sphere , which ( see Annex I. ) is rfhe other factor , is the -integral of Annex II .
* It is to be throughout borne in mind that denotes a set of coordinates , a set of coordinates ; the adjective coordinal refers to the number of coordinates which enter into the equation ; thns , is an coordinal sphere ( observe that the surface of such a sphere is -dimensional ) ; , according as we tacitly associate with it the condition , or arbitrary , is an -coordinal circle , or cylinder , the surface of such circle or cylinder being -dimensional , but the circumference of the circle dimensional ; or if we attend only to the -dimensional space constituted by the plane , the locus may be considered as an -coordinal sphere , its surface being dimensional .
$ PROFESSOR CAYLEY ON PREPOTENTIALS .
679 2 .
We now consider the prepotential-surface integral As already mentioned , it is only a particular case of this , the prepotential-plane integral , which is specially discussed ; ) ) at present I consider the general case , for the purpose of establishing a theorem in relation thereto .
The surface ( -dimensional surface ) is any given surface whatever .
Let the attracted point be situate indefinitely near to the surface , on the normal thereto at a point , say the normal distance NP is ; and let this point be taken as the centre of an indefinitely small circular ( -dimensional ) disk or segment ( of the surface ) , the radius of which , although indefinitely small , is indefinitely large in com- parison with the normal distance .
I proceed to determine the prepotential of the disk ; for this purpose , transforming to new axes , the origin being at and the axes of in the tangent-plane at , then the coordinates of the attracted point will be , and the expression for the prepotential of the disk will be where the limits are given by Suppose for a moment that the density at the point is , then the density throughout the disk may be taken , and the integral becomes where instead of I write 9 ; now denotes the density at the point N. this change , then ( by what precedes ) the value is Positive .
to 7 .
3 .
I consider first the case where is positive .
The value is here or since is indefinitely small , the -integral may be neglected , and the value is Observe that this value is independent of , and that the expression is thus the as if ( instead of the disk ) we had taken the whole of the infinite tangent-plane , the is positive ; in afterwards writing , we mean by the limit of an indefinitely small positive quantity density at every point thereof being .
It is proper to rark Shrt the ected terms are of the orders : so that the complete value multiplied by is equal to the constant terms of the orders \amp ; c. 4 .
Let us now consider the prepotential of the remaining portion of the surfaoe ; every part thereof is at a distance fromd exceeding , in fact far exceeding , ; so that imagining the whole mass to be collected at the distance , the prepotential of the remaining portion of the surface is less than ; viz. we have thus , in the case where the mass is finite , a superior limit to the prepotential of the remaining portion of the surface .
This will be indefinitely small.in comparison with the prepotential of the disk , provided only is indefinitely small compared with , that is indefinitely small.in comparison with .
The proof assumes that the mass is finite ; but considering the very rough manner in which the limit was obtained , it can scarcely be doubted that , if not universally , at least for very general laws of distribution , even when is infinite , the same thing is true ; viz. that by taking sufficiently small in regard to , we can make the prepotential of the remaining portion of the surface vanish in comparison with that of the disk .
But without entering into the question I assume that the prepotential of the lemaining portion does thus vanish ; the prepotential of the whole surface in regard to the indefinitely near point is thus equal to the prepotential of the disk ; viz. its value is which , observe , is infinite for a point on the surface .
5 .
Consideling the prepotential V of an arbitrary point as a given function of the coordinates of this point , and taking for the coordinates of the point , which is , in fact , an arbitrary point on the surface , then the value of V at the point indefinitely near to will be , if denote the same function of that V is of .
The result just obtained is therefore , or , what is the same thing , AyLEY ON PBEPOTENTIAIS .
As to this , remark that V is not an arbitrary function of : non constat that there is any distribution of matter , and still less that there is any distribution of matter on the surface , which will produce at the point , that is at every point whatever , a prepotential the value of which shall be a function assumed at pleasure of the coordinates .
But suppose that V , the given function of , is such that there does exist a distribution of matter on the surface ( viz. that V. satisfies the conditions , whatever they are , required in order that this may be the case ) , then the foregoing formula determines the distribution , viz. it gives the expression of 9 , that is , the density at any point of the surface .
6 .
The theorem may be presented in a somewhat different form ; regarding the prepotential as a function of the normal distance , its derived function in regard to is , that is ; and we thus have , or , what is the same thing , where , however , given as a function of , the notation requires explanation .
Taking to be the inclinations of the normal at , in the direction NP in which the distance is measured , to the positive parts of the axes of , viz. these cosines denote the values of . .
each taken with the same sign , and divided by the square root of the sum of the squares of the last-mentioned quantities , then the meaning is cos 7 .
7 .
The surface may be the plane , viz. we have then the prepotential-plane integral , . . . . .
( A ) where ( like ) is positive .
In afterwards writing , we mean by the limit of an indefinitely small positive quantity .
The foregoing distribution-formulae then become , ( A ) 682 .PBOFESSOR CAYLEY ON PBEPOTENTIAL@ .
and which will be used in the sequel .
It will be remembered that in the preceding investigation it has .
been assumed that is positive , the limiting case being excluded to 13 .
8 .
I pass to the case .
we here have the potential-surface integral : ( C ) it will be seen that the results themselves under a remarkably different form .
The potential of the disk is , as before , where here denotes the density at the point ; and the value of the -integral terms Observe that this is indefinitely small , and remains so for a point on the surface ; the potential of the remaining portion of the surface ( for a point near to or on the surface ) is finite , that is , neither indefinitely large nor indefinitely small , and it varies continuously as the attracted point passes through the disk ( or aperture in the material surface now under consideration ) ; hence the potential of the whole surface is finite for an attracted point on the surface , and it varies continuously as passes through the surface .
It will be noticed that there is in this case a term in Vindependent of ; and it is on this account necessary , instead of the potential , to consider its derived function in regard to ; viz. neglecting the indefinitely small terms which contain powers of , I write The corresponding term arising from the potential of the other portion of the surface , viz. the derived function of the potential in regard to 8 , is not indefinitely small ; and calling it , the formula for the whole surface becomes This is , as regards , the caso throughout ; a limiting value , if not expressly stated to be induded , is always excluded .
PROFESSOR CAYLEY ON PREPOTENTIALS .
9 .
I consider positions of the point on the two opposite sides of the point , say at the normal distances these being positive distances measured in opposite directions ffom the point N. The function V , which represents the potential of the surface in regard to the point , is or may be a different function of the coordinates of the point , according as the point is situate on the one side or the other of the surface ( as to this more ) .
I represent it in the one case by V ' , and in the other case by ; and in further explanation state that is measured into the space to which V ' refers , into that to which efers ; and I say that the formulae belonging to the positions of the point are , instead of V ' , V I have written to denote that the coordinates , as well of as of are taken to be the values which belong to the point N. The symbols denote where and are the cosine inclinations of the normal distances to the positive parts of the axes of ; since these distances are measured in opposite directions , we have .
If we imagine a curve through cutting the surface at right angles , or , what is the same thing , an element of the curve coinciding in direction with the normal element and if denote the distance of from a fixed point of the curve , and for the point becomes , while for the point it becomes , or , what is the same thing , if increase in the direction of NP ' and decrease in that of NP then if any function of the coordinates of be regarded as a function of , we have 10 .
In particular , let denote the potential of the remaining portion of the surface , that is , of the whole surface exclusive of the disk ; the curve last spoken of is a curve which does not pass through the material surface , viz. the portion to which has reference , and there is no discontinuity in the value of as we pass along this curve through the point N. We have value of at the point , and value of at the point ; and the two points coming to coincide together at the point 684 PROFESSOR CAYLEY ON PBfflOTENlIAla .
, we have then We have in like manner and the equations obtained above may be written in which form they show that as the attracted point passes through the surface Rom the position on the one side to on the other , there is an abrupt change in the value of , or say of ' the first derived function of the potential in regard to the orthotomic arc , that is in the rate of increase of V in the passage of the attracted point normally to the surface .
It is obvious that if the attracted point traverses the surface obliquely instead of normally , viz. if the arc cuts the surface obliquely , there is the like abrupt change in the value of Reverting to the original form of the two equations , and attending to the relation , we obtain or , what is the same thing , ( C ) 11 .
I recall the signification of the the potentials , it may bedifferent functions of the coordinates of the attracted point , for positions of this point on the two sides of the surface ( as to this more presently ) , and are what respectively become when the coordinates are replaced by , the coordinates of a point on the surface .
The explanation of the symbols is given a little above ; denotes the density at the point .
12 .
The like remarks arise as with regard to the former distribution theorem ( A ) ; the functions cannot be assumed at asure ; non constat that there is any distribution in space , and still less any distribution on the surface , which would give such values to the potential of a point on the two sides of the surface respectively ; .
but assuming that the functions are such that they do arise from .
distribution on the surface , or say that they satisfy all the conditions , whatever they are , required in $ CAYLEY ON PREPOTFNTIALS .
order that this may be so , then the formula determines the distribution , viz. it gives the value of 9 , the density at a point .
of the surface .
13 .
In the case where the surface is the plane , viz. in the case of the potentialplane integral , V ( assumed to be positive ) ; then , since every thing is symmetrical on the two sides of the plane , V ' and are the same functions of , say they are each are each of them the same function , say they are each , of that V is of , and the distribution-formula becomes ( B ) viz. this is also what one of the prepotential-plane formulae becomes on writing thereiu , or Negative.\mdash ; Nos. 14 to 18 .
14 .
Consider the case .
The prepotential of the disk is ; and to get rid of the constant term we must consider the derived function in regard to , viz. this is and we have thus for the whole surface where , which relates to the remaining portion of the surface , is finite ; we have thence , writing , as before , in place of V , or say 15 .
Consider the case negative , but .
The prepotential of the disk is here ; and to get rid of the first term we must consider the derived function in regard to viz. this is ; whence for the potential of the whole surface where , the part relating to the remaining portion of the surface , is finite .
Multiplyin by ( where the index is positive ) , the term in disappears ; and writing , as before , in place of V , this is or say ; viz. we thus see that the formula originally obtained for the case positive extends to the case , and , but ; but , as already seen , it does not extend to the limiting case 16 .
If be negative and between and , we have in like manner a formula ; but here being negatiye , the term does not disappear : the formula has to be treated in the same way as for , and we arrive at ; viz. the formula is of the same form as for the potential case .
Observe that the formula does not hold good in the limiting case 17 .
We have , in fact , here the potential of the disk ; whence , since in the complete differential coefficient the term 8 vanishes in comparison with ; and then , proceeding as before , we find ; but I have not particularly examined this formula .
18 .
If be negative and that , then the prepotential for the disk is ; and it would seem that in order to obtain a result it would be necessary to proceed to a derived function higher than the first ; but I have not examined the case .
CAYLEY ON PBbPOTENTIALS .
Continuity of the -surface Integral.\mdash ; Art .
Nos. 19 to 25 .
19 .
I again consider the prepotential-surface integral in regard to a point not on the surface ; is either positive or negative , as afterwards mentioned .
The integral or prepotential and all its derived functions , first , second , \amp ; c. infinitum , in regard to each or all or any of the coordinates are all finite .
This is certainly the case when the mass is finite , and possibly in other cases also ; but to fix the ideas we may assume that the mass is finite .
And the prepotential and its derived functions vary continuously with the position of the attracted point , so long as this point in its course does not traverse the material surface .
For greater clearness we may consider the point as moving along a continuous curve ( one-dimensional locus ) , which curve , or the part of it under consideration .
, does not meet the surface ; and the is that the prepotential and each of its derived functions varies continuously as the point passes continuously along the 20 .
Consider a " " region that is , a portion of space any point of which can be by a continuous curve not meeting the material surface connected with any other point of the region .
It is a legitimate inference , from what just precedes , that the prepotential is , for any point whatever within the , one and the same function of the coordinates , viz. the theorem , rightly understood , is true ; but the theorem gives rise to a difficulty , and needs explanation .
Considel , for instance , a closed surface made up of two segments , the matter being distributed in any manner over the whole surface ( as a particular case , a uniform spherical shell made up of two hemispheres ) ; then , as regards the first segment ( now taken as the material surface ) , there is no division into regions , but the whole of the dimensional space is one region ; wherefore the prepotential of the first segment is one and the same function of the coordinates of the attracted point for any position whatever of this point .
But in like manner the prepotential of the second segment is one and the same function of the coordinates for any position whatever of the attracted point .
And the prepotential of the whole surface , being the sum of the prepotentials of the two segments , is consequently one and the same function of the coordinates of the attracted point for any position whatever of this point ; viz. it is the same function for a point in the region inside the closed surface and for a point in the outside region .
That this is not general the case we know from the particular case , , of auniform spherical shell referred to above .
21 .
Consider in general an unclosed surface or segment , with matter distributed over it in any manner ; and imagine a closed curve or circuit cutting the segment once ; and let the attracted point move continuously along the circuit .
We may consider the circuit as corresponding to ( in ordinary tridimensional space ) a plane curve of equal periphery , the corresponding points on the cirenit the a1rve being points at equal distances along the curves from fixed points on the Wo curves reupectively ; and then treating the plane curve as the base of a cylinder , we may represent the potential as a length or ordinate , , measured upwards from the point on the plane curve along the generating line of the cylinder , in such wise that the upper extremity of the length or ordinate traces out on the cylinder a curve , say the prepotential curve , which represents the march of the prepotential .
The attracted poin may , for greater convenience , be represented as a point on the prepotential curve , viz. by the upper instead of the lower extremity of the length or ordinate ; and the orclimate , or height of this point above the base of the cylinder , then represents the value of the prepotential .
The before-mentioned continuity-theorem that the prepotential curve corresponding to any portion ( of the circuit ) which does not meet the matenal surface is a continuous curve , .
that there is no abrupt change of value either in the ordinate of the prepotential curve , or in the first or any other of the derived functions \amp ; c. We have thus ( in each of the two figures ) a continuous curve as we pass ( in the direction of the arrow ) from a point on one side of the segment to a point the other side of the segment ; but this continuity does not exist in regard to the remaining part , from to , of the prepotential curve corresponding to the portion ( 'of the circuit ) which traverses the material surface .
22 .
I consider first the case ( see the left-hand figure ) : the prepotential is here a potential .
At the point , which corresponds to the passage through the material surface , then , as was seen , the ordinate ( Potential V ) remains finite and continuous ; but there is an abrupt change in the value of , that is , in the direction of the curve : the point is really a node with two branches crossing at this point , as shown in the figure ; but the dotted continuations have only an analytical existence , and do not represent values of the potential .
And by means of this brznchto-branch discontinuity at the point , we escape from the foregoing conclusion as to the continuity of the potential on the passage of the attracted point through a dosed surface .
23 .
To show how this is I will for greater clearness examine the case in ordinary tridimensional space , of the uniform spherical shell attracting ding to the inverse square of the distance ; instead of dividin the shell into divide it by a plane into any two segments the wherein remt CAYLET ON pREpOTENTIAls .
centres of the two segments respectively , and where for graphical convenience the segment A is taken to be small .
We may consider the attracted point as moving along the axis , viz. the two extremities may be regarded as meeting at infinity , or we may outside the sphere bend the line round , so as to produce a closed circuit .
We are only concerned with what happens at the intersections with the spherical surface .
The ordinates represent the potentials , viz. the curves are for the segments , and the whole spherical surface respectively .
Practically , we construct the curves , and deduce the curve by taking for its ordinate the difference of the other two ordinates .
The curve is , as we know , a discontinuous curve , composed of a horizontal line and two hyperbolic branches ; the curve can be laid down approximately by .
the segment A as a plane circular disk ; it is of the form shown in the figure , having a node at the point corre sponding to A. [ In the case where the ment A is actually a plane disk , the curve is made up of portions of branches of two hyperbolas ; but taking the segment A as being what it is , the segment of a spherical surface , the curve is a single curve , having a node as mentioned above .
] And from the curves and , deducing the curve , we see that this is a curve without any discontinuity corresponding to the passage of the attracted point through ( but with an abl.upt change of direction or node corresponding to the passage through B ) .
And conversely , using the to detelmine the curve , we see how , on the passage of the attracted point at A into the interior of the sphere , in consequence of the branch-to-branch discontinuity of the curve , the curve , obtained by combination of the two curves , undergoes a change of law , passing abruptly from a hyperbolic to a rectilinear form , and how similarly on the passage of the attracted point at from the interior to the exterior of the sphere , in consequence of the branch-to-branch discontinuity of the curve , the curve again undergoes a change of law , abruptly reverting to the hyperbolic form .
24 .
In the case positive the prepotential curve is as shown by the right.hand figure in p. 688 , viz. the ordinate is here infinite at the point corresponding to the passage through the surface ; the value of the derived function changes between infinity and \mdash ; infinity ; and there is thus a discontinuity of value in the derived function .
It would seem that when is fractional this oCcasions a change of law on passage through the surface , but that there is no change of law when is integral .
In illustration , consider the closed surface as made uP of an infinitesimal circular disk , as before , and of a residual portion ; the potential of the disk on an indefinitely near point is found as before , and the prepotential of the whole surface is where , the prepotential of the remaining portion of the surface , is a function which varies ( and its derived functions vary ) continuously as the attracted point traverses the disk .
To fix the ideas we may take the origin at the centre of the disk , and the axis of as coinciding with the normal , so that , which is always positive , is ; and the expression for the prepotential at a point on the normal through the centre of the disk is viz. when is fractional there is the discontinuity of law , inasmuch as the changes from to ; but when is integral this discontinuity disappe.ars .
The like considerations , using of course the proper formula for the attraction of the disk , would .
apply to the case or negative .
25 .
Or again , we might use the formulae which belong to the case of a uniform coordinal spherical shell ( see Annex No. III viz. we decompose the surface as follows , surface disk residue of surface ; and then , considering a spherical shell touching the surface at the point in question ( so that the disk is in fact an element common to the surface and the spherical shell ) , and being of a uniform density equal to that of the disk , we have diskspherical shell\mdash ; residue of spherical shell ; and consequently surfacespherical shell-residue of spherical residue of surface ; and then , considering the attracted point as passing through the disk , it does not through either of the two residues , and there is not any discontinuity , as regards the prepotentials of these residues respectively ; there is consequently , as regards the prepotential of the surface , the same discontinuity that there is as regards the of the spherical shell .
But I do not further consider the question from this point of view .
ESSOR CAYLEY ON PREPOTENTIALS .
The Potential Solid Integrat.\mdash ; Art .
No. 26 .
26 .
We have further to consider the prepotential ( and in particular the potential ) of a material space ; to fix the ideas , consider for the moment the case of a distribution over the space included within a closed surface , the exterior density being zero , and the interior density being , supposed for the moment , constant ; we consider the discontinuity which takes place as the acting point passes from the exterior space through the bounding surface into the interior material space .
We may imagine the interior space divided into indefinitely thin shells by a series of closed surfaces similar , if we please , to the bounding surface .
; and we may conceive the matter included between any two consecutive surfaces as concentrated on the exterior of the two surfaces , so as to give rise to a series of consecutive material surfaces ; the quantity of such matter is infinitesimal , and the density of each of the material surfaces is therefore also infinitesimal .
As the attracted point comes from the external space to pass through the first of the material surfaces\mdash ; suppose , to fix the ideas , it moves continuously along a curve the arc of which measured from a fixed point is \mdash ; there is in the value of V ( or , as the case may be , in the values of its derived functions \amp ; c. ) the discontinuity due to the passage through the material surface ; and the like as the attracted point passes through the different material surfaces respectively .
Take the case of a potential , ; then , if the surface-density were finite , there would be no finite change in the value of V , but there would be a finite change in the value of as it is , the changes are to be multiplied by the infinitesimal density , say , of the material surface ; there is consequently no finite change in the value of the first derived function ; but there is , or may be , a finite change in the value of and the higher derived functions .
But there is in V an infinitesimal change corresponding to the passage through the successive material surfaces respectively ; that is , as the attracted point enters into the matenal space thel is a change in the law of V considered as a function of the coordinates of the attracted point ; but by what precedes this change of law takes place without any abrupt change of value either of V or of its first derived function ; which derived function may be considered as representing the derived function in regard any one of the coordinates .
The suppositions that the density outside the bounding surface was zero and inside it constant , were made for simplicity only , and were not essential ; it is enough if the density , changing abruptly at the bounding surface , varies continuously in the material space within the bounding surface* .
* It is , indeed , enough if the density varies continuously within the bounding surface in the neighbourhood of the point of passage through the surface ; but the condition without loss of generality be stated as in the text , it being understood that for each abrupt change of density within the bounding surface we must consider the attracted point as passing through a new bounding surface , and have regard to the resulting discontinuity .
NDCCCLXXV .
4 conclusion is that being the values at points within and without the bounding surface , V ' and are in general different functions of the coordinates of the attracting point ; but that the surface we have not only , but that the first derived functions are also equal , viz. that we have , .
27 .
In the general case of a Potential , if does not vanish at the attracted point , but has there a value different from zero , we may consider the attracting dimensional mass as made up of an indefinitely small sphere , radius and density , which includes within it the attracted point , and of a remaining portion external to the attracted point .
Writing to , then , as regards the potential of the sphere , we have ( see Annex III .
No. 67 ) , and as regards the remaining portion ; hence , as regards the whole attracting mass , has the first-mentioned value , that is we have where is the same function of the coordinates that 9 is of ; .
the potential of an attracting mass distributed not on a surface , but over a portion of space , does not satisfy the potential equation but it satisfies the foregoing equation , which only agrees with the potential equation in regard to a point outside the material space , and for which , therefore , is The equation may be written ; or , considering as a given function of , in general a discontinuous function but subject to certain conditions as afterwards mentioned , and taking the same function of that V is , then we have .
( D ) PEOFESSOE CAYLEY ON PRBPOTENTIALS .
69$ viz. this equation determines as a metion , in general a discontinuous function , of such that the corresponding integral may be the given function of be coordinates .
The equation is , in fact , the distribution-theorem D. 28 .
It is to be observed that the given function of must satisfy certain conditions as to value at infinity and continuity , but it is not ( as in the distributiontheorems , and it is ) required to satisfy a partial differential equation ; the function , except as regards the conditions as to value at infinity and continuity , is absolutely arbitrary .
The potential ( assuming that the matter which gives rise to it lies wholly within a finite closed surface ) must vanish for points at an infinite distance , or more accurately it must for indefinitely large values of be of the form , Constant .
It may be a discontinuous function ; for instance outside a given closed surface it may be one function , and inside the same surface a different function of the coordinates ; viz. this may happen in consequence of an abrupt change of the density of the attracting matter on the one and the other side of the given closed surface , but not in any other manner ; and , happening in this manner , then V ' , being the values for points within and without the surface respectively , it has been seen.to be necessary that , at the surface , not only V ' , but also Subject to these conditions as to value at infinity and continuity , Vmay be any function whatever of the coordinates ; and then taking , the same function of , the foregoing equation determines , viz. determines it to be for those parts of space which do not belong to the material space , and to have its propel value as a function of for the remaining or material space .
The Prepotential Plane Theorem A.\mdash ; Art .
Nos. 29 to 36 .
29 .
We have seen that if there exists on the plane a distribution of matter producing at the point a given prepotential ( viz. is to be regarded as a given function of ( a then that the distribution or density 9 is given by a determinate formula ; but it was remarked that the prepotential V cannot be a function assumed at pleasure ; it must be a function satisfying certain conditions .
One of these is the condition of continuity ; the function V and all its derived functions must vary continuously as we pass , without traversing the material plane , from any given point to any other given point .
But it is sufficient to attend to points on one side of the plane , say the upperside , or that for which is positive ; and since any such point is accessible from any other such point by a path which does not meet the plane , it is sufficient to say that the function must vary continuously for a passage by such path from any such point to any such point ; the function V must therefore be one and the same function ( and that a continuous one in value ) for all values of the coordinates and positive values of the coordinate If , moreover , we assume that the distribution which corresponds to the given potential V is a distribution of a finite mass . .
over a finite portion of the plane viz. over a portion or area such that the distance of a point within the area from a fixed point , or say from the origin , is always finite ; this being so , we have the further condition that the prepotential V must for indefinitely large values of all or any of the coordinates reduce itself to the form The assumptions upon which this last condition is obtained are perhaps unnecessary ; instead of the condition in the foregoing form we , in fact , use only the condition that the prepotential vanishes for a point at infinity , that is when all or any one or more of the coordinates are or is infinite .
Again , as we have seen , the prepotential V must satisfy the prepotential equation . .
These conditions satisfied , to the given prepotential V , there corresponds on the plane , a distribution given by the foregoing formula , and which will be a distribution over a finite portion of the plane , as already mentioned .
30 .
The proof depends upon properties of the prepotential equation , or , what is the same thing , . .
say , for shortness , Consider , in general , the integral . . .
taken over a closed surface lying altogether on the positive side of the plane the function being in the first instance arbitrary .
Writing the integral under the form .
, we reduce the several terms by an integration by parts as follows:\mdash ; PROFESSOR CAYLEY ON PREPOTENTIALS .
term in is de de , : is , is . . .
Write to denote an element of surface at the point ; and taking .
7 , to denote the inclinations of the interior normal at that point to the positive axes of coordinates , we have . .
; and first terms are together here denoting the value at the surface , and the integration being extended over the whole of the closed surface : this may also be written where denotes an element of the internal normal .
The second terms are together . .
de . .
dell ' W. We have consequently W. 31 .
The second term vanishes if satisfies the prepotential equation ; and this being so , if also for all points of the closed surface , then the first term also vanishes , and we therefore have where the integration extends over the whole space included within the closed ; whence , being a real function , , . . .
for all points within the closed surface ; consequently , since vanishes at the surface , for all points within the closed surface .
32 .
Considering as satisfying the equation , we may imagine the closed surface to become larger and larger , and ultimately infinite , at the same time flattening itself out into coincidence with the plane , so that it comes to include the whole space above the plane , say the breaks up into the surface positive infinity and the infinite plane The integral separates itself into two parts , the first relating to the surface positive infinity , and which vanishes if at infinity ( that is , if all or any of the coordinates are infinite ) ; the second relating to the plane is here denoting its value at the plane , that is when and the integral being extended over the whole plane .
The theorem thus becomes .
Hence also if at all points of the plane , the right-hand side vanishes , and we have Consequently . .
, for all pointswhatever of positive space ; and therefore also for all points whatever of positive space .
33 .
Take next , each of them a function of , and consider the integral , taken over the space within a closed surface this in a similar manner , we find it to be where the integration extends over the whole of the closed surface ; and by of reasoning it is also with the same limits of integration ; that is , we have de .
de .
CAYLEX ON which , if each satisfy the prepotential equation , becomes And if we now take the closed surface to be the surface positive infinity , together with the plane , then , provided only and V vanish at infinity , for each integral the portion belonging to the positive infinity vanishes , and there remains only the portion belonging to the plane ; we have therefore where the functions have each of them the value belonging to the plane viz : in considered as iven functions of we regard as a positive quantity ultimatelyput , and where the integrations extend each of them over the whole plane .
34 .
Assume an expression which , regarded as a function of , satisfies the prepotential equation in regard to these variables , and which vanishes at infinity when all or any of these coordinates are infinite .
We have ; and we have consequently where it will be recollected that is ultimately ; to mark this we may for write Attend to the left-hand side ; take the same function of , that is of ; then first writing the expression in the form write , the expression becomes where the integral is to be taken from to for each of the new variables Writing , where , we have also , and the integral is where denotes the surface of the -coordinal unit sphere , and integral is to be taken from to ; the values of the two factors thus are , and Hence the expression in question is and we have or , what is the same thing , 35 .
Take now V a function of satisfying the prepotential equation in regard to these variables , always finite , and vanishing at infinity , and let be the same function of , satisfying the prepotential equation in regard to the last-mentioned variables , and consider the function where the integral is taken over the infinite plane ; then this function ( V \mdash ; the integral ) satisfies the prepotential equation ( for each term separately satisfies it ) , is always finite , and it vanishes at infinity .
It also , as has just been seen , vanishes for any point whatever of the plane .
Consequently it vanishes for all points whatever of positive space .
Or , what is the same thin , if we write ( A ) where 9 is a function of , and the integral is taken over the whole infinite plane , then if is a function of satisfying the above conditions , there exists a corresponding value of 9 ; viz. taking the same function of which V is of , the value of 9 is .
( A ) where is to be put in the function .
This is the prepotential-plane theorem ; viz. taking for the prepotential in regard to a given point function of satisfying the prescribed conditions , but otherwise arbitrary , there exists on the plane a distribution given by the last-mentioned formula .
PROFESSOR CAYLEY ON PREPOTENTIALS .
36 .
It is assumed in the proof that is positive or zero ; viz. is positive , or if negative then } ; the limiting case is included .
It is to be remarked that by what precedes , if be positive ( but excluding the case the density is given by the equivalent more simple formula The foregoing proof is substantially that given in GREEN'S memoir on the Attraction of Ellipsoids ; it will be observed that the proof only imposes upon V the condition of vanishing at infinity , without obliging it to assume for large values of the form The Potential-surface Theorem C.\mdash ; Art .
Nos. 37 to 42 .
37 .
In the case , writing here , we have precisely , as in the case , ; and if the functions satisfy the equations , then ( subject to the exception presently referred to ) the second terms on the two sides lespectively each of them vanish .
But , instead of taking the surface to be the surface positive infinity togetherwith the plane , we now leave it an arbitrary closed surface , and for greater symmetry of notation write in place of ; and we suppose that the functions and , or one of them , may become infinite at points within the closed surface ; on this last account the second terms do not in every case vanish .
38 .
Suppose , for instance , that at a point indefinitely near the point within the surface becomes ; then if V be the value of at the point , we have ; and since , except at the point in question , the integral may be taken over any portion of space surrounding this point , for instance , oyer the space included within the sphere , radius , having the point for its centre ; or taking the origin at this point , we have to find .
, where and the integration extends over the space within the sphere 39 .
This may be accomplished most easily by means of a particular of the stmentioned theorem ; viz. writing , we have , .
or the required value is over the surface of the last-mentioned sphere .
We have , if for a moment that is , ; and hence where is the whole surface of the sphere . .
, viz. it is into the surface of the unit-sphere . . .
This spherical surface , say is and we have thus , and consequently 40 .
Treating in like manner the case where at a point indefinitely near the point within the surface becomes and writing to denote the same function of that is of , we have , instead of the foregoing , the more general theorem . .
where in the two solid integrals respectively we exclude from consideration the space in the immediate neighbourhood of the two critical points and respectively .
Suppose that is always finite within the surface , and that is finite except at the point , and moreover that are such that , then the equation becomes PROFESSOR CAYLEY ON PREPOTENTIALS . .
In particuIar this equation holds good if is 41 .
Imagine now on the surface a distribution producing at a point within the surface a potential , and at a point without the surface a potential V where , by what precedes , is in general not the same function of ( that V ' is of It is further assumed that at a point on the surface we have V ' : that V ' , or any of its derived functions , are not infinite for any point within the surface : that V or any of its derived functions , are not infinite for any point without the surface : and that for any point at infinity .
Consider V ' as a given function of ; and take the same function of .
Then if , as before , then Similarly , considering as a given function of and take the same function of .
Then , by considering the space outside the surface , or say between this surface and infinity , and observing that does not become infinite for any point in this space , we have ; and adding these two equations , we have But in this equation the functions and each of them belong to a point on the surface , and we have at the surface suppose ; the term on the right-hand side thus is , which vanishes in virtue of ; and the equation thus becomes : that is , the point interior , we have V ' . .
In exactly the same way if be an exterior point , then we have ; and adding , and omitting the terms which vanish , that is , , 42 .
Comparing the two results with we see that V ' , satisfying the foregoing conditions , there exists a distribution on the surface , producing the potentials V ' and at an .
interior point and an exterior point respectively ; the value of 9 in fact being , ( C ) where are respectively the same functions of that V ' , are of .
The Potential-solid Theorem D.\mdash ; Art .
No. 43 .
43 .
We have as before ( No. 40 ) , where , assuming first that is not infinite for any point whatever , we have no term in ; and taking next before , we have ; the equation thus becomes where may be a discontinuous function of the coordinates , provided only there is no abrupt change in the value either of or of any of its first derived functions , .
it may be any function which can represent the potential of a solid mass on an attracted point ; the resulting value of is of course disconPROFESSOR CAYLEY ON PREPOTENTIALS .
tinuous .
Taking , then , for closed surface 8 the boundary of infinite space , and each vanish at this boundary , and the equation becomes ; viz. substituting for its value , and comparing with where the integral in the first instance extends to the whole of infinite space , but the limits may be ultimately restricted by being , we see that the value of 9 is being the same function of that V is of , which is the theorem D. Examples of the Theorems.\mdash ; Art .
Nos. 44 to 49 .
44 .
It will be remarked , as regards all the theorems , that we do not start with known limits ; we start with V a function of , the coordinates of the attracted point , satisfying certain prescribed conditions , and we thence find 9 , a function of the coordinates or , as the case may be , which function is found to be for values of or lying beyond certain limits , and to have a determinate non-evanescent value for values of or lying within these limits ; and we thus , as a result , obtain these limits for the limits of the multiple integral V. 45 .
Thus in theorem , in the example where the limiting equation is ultimately found to be , we start with a certain function of ( suppose ) and , viz. V is a function of these quantities through , which denotes the positive root of the equation the vaIue in fact being , and the resulting value of is found to be for values of for which .
Hence V denotes an integral the limiting equation being , say this is the -coordinal sphere .
And similarly , in the examples where the limiting equation is ultimately found to be , we start with V a certain function of , . . .
( or directly and through where denotes the positive root of the equation and the resulting value of is found to be for values of for whi\amp ; : .
Hence V denotes an integral , the limiting equation being , say his is the -coordinal ellipsoid .
It is clear that this includes the before-mentioned case of the coor inal sphere ; but it is , on account of the more simple form of the -equation , worth while to work out directly an example for the sphere .
46 .
Three examples are worked out in Annex IV .
; the results are as follows:\mdash ; First , defined for the sphere as above ; positive ; over the sphere This is included in the next-mentioned example for the ellipsoid .
Secondly , defined for the ellipsoid as above ; positive ; over the ellipsoid This result is included in the next-mentioned example ; but the proof for the general .
value of directly applicable to the value for the case in question .
Thirdly , and the ellipsoid as above ; positive ; or positive , and apparentiy in other cases , over the ellipsoid as above , And we have in Annex V. a fourth example ; here and the ellipsoid are as above : the result involves the Greenian functions .
$ PROFESSOR CAYLEY ON PREPOTENPIALS .
47 .
We may in the foregoing results write ; the results , writing therein for , and in the new forms taking and for the two sets of coordinates respectively , also writing for , would give integrals of the form for the coordinal sphere and ellipsoid and ; say these are prepotential solid integrals ; and then , writing , we should obtain potential solid integrals , such as are also given by the theorem D. The change can be made if necessary ; but it is more convenient to retain the results their original forms , as relating to the -coordinal sphere and ellipsoid .
There are two cases , according as the attracted point is external or internal .
For the sphere : an external point ; writing , the equation has apositive root , .
this is ; and will have , or it may be replaced by , this value : for an internal point ; as approaches zero , the positive root of the original equation gradually diminishes and becomes ultimately , viz. in the formulae is to be replaced by this value For the ellipsoid : an external point ; writing , the equation has a positive root , and will denote this positive root : for an internal point ; as approaches zero the positive root of the original equation gradually diminishes and becomes ultimately , viz. in the formulae is to be replaced by this value The resulting formulae for the sphere may be compared with formulae for the spherical shell , Annex VI .
, and each set with formulae obtained by direct integration in Annex III .
We may in any of the formulae write , and so obtain examples of theorem B. 48 : As regards theorem , we might in like manner obtain examples of potentials relating to the surfaces of the coordinal sphere , and ellipsoid , or say to spherical and ellipsoidal shells ; but I have confined myself to the sphere .
We have to assume values V ' and belonging to the cases of an internal and an external point respectively , and thence to obtain a value , or distribution over the spherical surface , which shall produce these potentials respectively .
The result ( see Annex VI .
) is over the surface of the coordinal sphere . .
for exterior point and for interior point where .
Observe that for the interior point the potential is a mere constant multiple of The same Annex VI .
contains the case of the -coordinal cylinder , which is peculiar in that the cylinder is not a finite closed surface , but the theorem is found to extend to it .
49 .
As regards theorem , we might in like manner obtain potentials relating to the coordinal sphere and ellipsoid ; but I confine myself to the case of the sphere ( see Annex We here assume values V ' and belonging to an internal and an external point respectively , and thence obtain a value 9 , or distribution over the whole dimensional space , which density is found to be for points outside the sphere .
The result obtained is over coordinal sphere for exterior poin for interior point where The remaining Annexes VIII .
and have no immediate reference to the theorems , which are the principal objects of the memoir .
The subjects to which they relate will be seen from the headings and introductory paragraphs .
ANNEX I. Surface and Volume of Sphere \amp ; 51 .
We lequire in dimensional space , , the volume of the sphere , and , the surface of the same sphere . .
, .
Writing , we have with the limiting condition ; but in order to take account as of the negative as the positive values of , we must multiply by .
The value is therefore PROFESSOR CAYLEY ON PREPOTENTIALS .
707 extended to all positive values of , such that ; and we obtain this by a known theorem , viz. Volume of dimensional sphere Writing , we obtain , where is the element of surface of the unit-sphere ; we have element of volume , where is to be taken ' from to 1 , and thence that is , ; surface of dimensional sphere 52 .
Writing for , we have Volume of dimensional sphere Surface of do .
which forms are sometimes convenient .
Writing in the first forms , or in the second forms , we find in ordinary space Volume of sphere and Surface of sphere as they should be .
ANNEX II .
The Integral to 63 .
53 .
The integral in question ( which occurs ante , No. 2 ) may also be considered as arising from a prepotential integral in tridimensional space ; the prepotential of an element of mass is taken to , where is the distance of the element from the attracted point P. Hence if the element of mass be an element of the plane , coordinates , being the density , and if the attracted point be situate in the axis of at a distance from the origin , the prepotential is For convenience it is assumed throughout that is positive .
IIDCCCLXXV .
5 .
PBOFESSOR CAYLEY O PBEPOTENTIAIi8 .
Suppose that the attracting body is a circular disk radius , having the ongin for its centre ( viz. that bounded by the curve ) ; then writin we have which , if is a function of only , is ; and in particular if , then the value is the integral in regard to being taken from to .
It is assumed that is not negative , viz. it is positive or ( it may be ) zero .
1 consider the integral which I call the -integral , more particularly in the case where is small in comparison with R. It is to be observed that not being , and being finite , the integral con- tains no infinite element , and is therefore finite , whether is positive , negative , or zero .
54 .
Writing , the integral is the limits being and In the case where is positive this is ; viz. the first term of this is and the second term is a term expansible in a series containing the powers \amp ; c. of the small quantity , as appears by effecting therein the substitution the value of the entire integral is by this means found to be 55 .
In the case where is , or negative , the formula fails by that the ele meant of the integrals becomes infinite for indefinitely large values of Recurling to the original , it is to be observed that the integral has a PROFESSOR CAYLEY ON POTENTIALS .
709 finite value when ; and it might therefore at first sight be imagined that the factor might be expanded in ascending powers of , and the value of the integral consequently obtained as a series of positive powers of .
But the series thus obtained is of the form , where being positive , the exponent is for a sufficiently small at first positive , or if negative less than , and the yalue of the integral is finite ; but as increases the exponent becomes negative , and equal or greater than - and the value of the integral is then infinite .
The inference is that the series commences in the form , but that we come at last wheh is fractional to a term of the form , and when is , or integral , to a term of the form , the process giving the coefficients , C. . .
, so long as the exponent of the corresponding term is less than ( in particular , there is a term , and the expansion-process does not give any term of the result ) , and the failure of the series after this point being indicated by the values of the subsequent eoefficients coming out 56 .
In illustration , we may consider any of the cases in which the integral can be obtained in finite terms .
For instance , Integral is , from to viz. expanding in ascending powers of this is or we have here aterm in .
And so , Integral is , from to ; viz. expanding in ascending powers of this is or we have here a term in 57 .
Returning to the form and writing herein , or , what is the same thing , , and for * Term is , which , } large , is reduced to , the value is where observe that is or negative , but X being a positive quantity less than 1 , the function is finite for the whole extent of the integration .
58 .
If , this is where observe that in virtue of the change made from to function which becomes infinite , which does become infinite , , it has become allowable in place of to write When is small , the integral which is the third term of the foregoing expression is obviously a quantity of the order ; the first term is , which , neglecting terms in , is , and hence the approximate value of the -integral is or , what is the same thing , it is where the integral in this expression is a mere numerical constant , which , when is a positive integer , has the value ; and neglecting this in parison with the logarithmic term , the approximate value is PROFESSOB CAYLEY ON POTENTIALS .
59 .
I consider also the case the integral is here ; and the first term of this being , this is consequently As regards the second term of this we have or taking each term between the limits 1 , ; viz. this integral has the value ; and the value of the -integral is consequently which is of the form terms in ; say the approximate value is where the first telm is the term , given by the expansion in ascending powers of ; the second term is the term in .
And observe that term is the value of calculated by means of the ordinary formula for a Eulerian integral ( which formula , on account of the negative exponent - , is not really applicable , the value of the integral being ) on the assumption that the of a negative is interpreted in accordance with the equation ; viz. the value thus calculated is on the assumption ; and this agrees with the foregoing value .
60 .
It is now easy to see in general how the foregoing , where is and fractional,.gives at once the value of the term in .
Observe that in the integral is always between 1 and X positive quantity less than 1 ) ; the function to be integrated never becomes infinite , Imagine for a moment an , where is positive or negative .
We conventionally write this , understanding the first symbol to mean and the second to mean ; they of course properly mean and ; but the terms in , whether zero or infinite , destroy each other , the original fornl , in fact , showing that no such terms can appear in the result .
In accordance with the convention we write ; and it follows that the term in is this last expression wherein will remembered , negative fraction ) being understood according to the convention ; and so understanding it the value of the term is where the of the negative is to be interpreted in accordance with the equation .
we have , \amp ; c. , so as to make the argument of the positive .
Observe that under this convention we have , or the term is 61 .
An example in which is integral will make the process clearer , and will serve instead of a general proof .
Suppose , the expression is used to denote the value PROFESSOB CAYLEY ON PREPOTENTIALS .
But we have agreeing with the former value .
62 .
The case of a is more simple ; to find the logarithmic term of we have only to expand the factor so as to obtain the term involving ; we have thus the term where , so that neglecting the terms in .
this is , and the term in question is The general conclusion is that being negative , the -integral has for its value a series proceeding in powers of , and which up to a certain point is equal to the series obtained by expanding in ascending powers of and integrating each term separately ; viz. the series to the point in question is continued so long as the exponent of is less than ; together with a term when is fractional , and when is integral ; viz. fractional this term is and integral , it is 63 .
It has been tacitly assumed that is positive ; but the formulae hold good if or negative .
Suppose is or a negative integer , then and the special term involving or vanishes ; in fact in this case the -integral is 714 PROFESSOR ON PREPOTENTIAL\amp ; where has for its value a finite series , and the integral is therefore equal to a finite series .
If be fractional , then the of the negative quantity must be understood as above , or , what is the same thing , we , instead of , write ; thus , being integral , the exceptional term is ; for instance , , the term is ; or , since , and , the term is , agreeing with a preceding result .
ANNEX III .
Prepotentials of Uniform Spherical Sheu and Solid Sphere.\mdash ; Nos. 64 to 92 .
64 .
The prepotentials in question depend ultimately upon two integrals , which also arise , as will presently appear , from prepotential problems in two-dimensional space , and which are for convenience termed the ring-integral and the disk-integral respectively .
The analytical investigation in regard to these , depending as it does on a transformation of a function allied with the hypergeometric series , is I think interesting .
65 .
Consider first the prepotential of a uniform dimensional spherical shelL is the equation of the surface bein ; and there are the two cases of an internal point , , and an external point , The value is a function of say this is ; and taking the axes so that the coordinates of the attracted point are , the integral is where the equation of the surface is still .
Writing , where , we have , or the integlal is Assume , . . .
, where ; then .
Moreover , , where is the element of surface of the -dimensional unit-sphere ; or for , substituting its value , we have ; and thence PROFESSOR CAYLEY ON PREPOTENTIALS .
a .
The integral as regards is from to , or as regards from 1 to -1 ; whence reversing the sign the integral will be from to ; and the lequired integral is thus where is the surface the -dimensional unit-sphere ( see Annex I ; and for greater convenience transforming the second factor by writing therein the required integral is into which last expression including factor without factor is the ringintegral discussed in the present Annex .
It may be remarked that the value can be at once obtained in the particular case , which belongs to tridimensional space , viz. we then have which agrees with a result given , 'Mecanique Celeste , ' Book XII .
Chap. II .
66 .
Consider next the prepotential of the uniform solid dimensional sphere , equation of surface , and the two cases of an internal point and an external point ( as before ) .
1ransforming so that the coordinates of the attracted point are , the integral is where the equation is still .
Writin here , where , we have , where is an element of surface of the -dimensional unit-sphere ; the integral is therefore where , as regards and , the integration extends over the circle .
The value NDCCCLXXY CAYLEY ON pBEPmBNEIALs .
of the first factor ( see Annex I. ) is ; and writing in place of respectively , the integral is into over the circle ; viz. this last expression without the factor is the diskintegral discussed in the present Annex .
67 .
We find for the value in regard to an internal point which in the particular case is ; viz. the integral in is here , or we have .
It may be added that in regard to an external point , the value is which in the same case is where the -integral is ; and the value of V is therefore Recurring to the case of the internal point ; then writing , and observing that , we have particular ordinary space value which right ) .
PROFESSOR CAYLEY ON PREPOTENTIALS .
717 68 .
The integralsreferred to as the ring-integral and the disk-integral arise also from the following integrals in two-dimensional space , viz. these are in the first of which an element of arc of the circle , the integration being extend.ed over the whole circumference , and in the second the integration extends over the circle is written for shortness instead of , viz. this is considered as always positive , whether is positive or negative ; it is moreover assumed that is zero or positive .
Writing in the first integral , the value is ; viz. this represents the prepotential of the circumference of the circle , density varying as , in regard to a point in the plane of the circle ; and similarly the second integral represents the prepotential of the circular disk , density of the element at the point , in regard to the same point , it being in each case assumed that the prepotential of an element of mass upon a point at distance is 69 .
In the case of the circumference , it is assumed that the attracted point is not on the circumference , not ; and the function under the integral sign , and therefore the integral itself , is in every case finite .
In the case of the circle , if be an interior point , then if be or positive , the element at the attracted point becomes infinite ; but to avoid this we consider not the potential of the whole circle , but the potential of the circle less an indefinitely small circle radius the attracted point for its centre ; which being so , the element under the integral sign , and consequently the integral itself , remains finite .
It is to be remarked that the two integrals are connected with each other ; viz. the circle of the second integral being divided in rings by means of a system of circles concentric with the bounding circle , then the prepotential of each ring or annulus is determined by an integral such as the first integral ; or , analytically , writing in the second integral , and therefore , the second integral is viz. the integral in regard to is here the same function of that the first integral is of ; and the integration in regard to is of course to be taken from to But the -integral is not in its original form such a function of as to render possible the integration in regard to ; and , in fact , obtain the second integral by a different and in some respects a better process .
70 .
Consider first the ring-integral , which writing therein as above , and multiplying by 2 in order that the integral , instead of being taken from to , may be taken from to , becomes Write ; then ; gives gives , and the integral is if for shortness ( obviously ) .
The integral in is here an integral to the general form viz. we have Ring-integral .
71 .
The general function , or , what is the same thing , , and consequently transformable by means of various theorems for the transformation of the hypergeometric series ; in particular the theorems , ; and if , or , what is the same thing , , then ; in verification observe that if then also , and that with these values , calculating each side by means of the formula the resulting equation , , becomes $ PROFESSOR CAYLEY ON PREPOTENTIALS .
that is which is true , in virtue of the relation 72 .
The foregoing , .
and in particular the formula which I have written , are taken from KUMMER'S Memoir , " " Ueber die hypergeometrische Reihe Crelle , , viz. the formula in question is under a slightly different form , his formula ( 41 ) p. 76 ; the formula ( 43 ) , p. 77 , is intended to be equivalent thereto ; but there is an error of transcription , , in place of , which makes the formula ( 43 ) erroneous .
It may be remarked as to the formulae generally , that although very probably may denote a proper function of , whatever be the values of the indices , and the various transformation-theorems hold good accordingly ( the -function of a negative argument being interpreted in the usual manner by means of the equation \amp ; c. , yet that the function , used as denoting the definite integral , has no meaning except in the case where and are each of them positive .
In what follows we obtain for the ring-integral and the disk-integral various expressions in terms of -functions , which are afterwards transformed into -integrals with a superior limit and inferior limit , or ; but for values of the variable index , lying beyond certain limits , the indices and , or one of them , of the -function will become negative , viz. the integral represented by the -fimction , or , what is the same thing , the , will cease to have a determinate value , and at the same time , or usually so , the argument or alguments of one or more of the -functions will become negative .
It is quite possible that in such cases the results are not without meaning , and that an interpretation for them might be found ; but they have not any obvious interpretation , and we must in the first instance consider them as inapplicable .
73 .
We require further properties of the -functions .
Starting with the equation , , each side may be expressed in a fourfold form : , where , instead of , it is proper to write ; affi then to each form applying the ansformation , we have I select on the left-hand the second form , and equating it successively to the four right-hand forms , attending to the relation , we find .
Putting herein , the formula become .
( I. ) ( II .
) . .
( IIL ) , .
PBOFESSOE AYLEY ON PBEPOTBNTIALS .
whele observe that on the right-hand side the -functions of I. and IV .
only differ by the sign of , and so also the -functions of II .
and III .
only differ by the sign of We hence have ; and comparing with ( IV .
) , .
74 .
The foregoing mula , Ring-integral , where , gives , as well in the case of an exterior as an interior point , a convergent series for the integral ; but this series proceeds according to the verse of We may obtain more convenient formulae applying to the cases of an internal and an external point respectively .
75 .
Internal point , and therefore , where the -functions on the right-hand side are respectively the -forms being obtained by means of the transformation ; viz. this gives whence the results just written down .
We hence have Ring-integral As a verification write , the four integrals are and hence from each of them Ring-integral which is in fact the value obtained from Ring-integral on putting therein ; viz. the value is 76 .
External point , and therefore , PROFESSOR CAYLEY ON PREPOTENTIALS .
where the -functions on the right hand are respectively ; we have then Ring-integral Observe that in II .
and III .
the integrals , except as to the limits , are the same as in the corresponding formulae for the interior point .
If in the -integrals we put in place of , and ultimately suppose indefinitely large in comparison with , they severally become , ; and they all four give Ring-integral which agrees with the value when is indefinitely large .
NDCCCLXXY .
77 .
We come now to the disk-integral , over the circle .
Writing , we have and the integral therefore is where the integration in regard to is performed at once , viz. the integral is or multiplying by 2 , in order that the integration may be taken only over the semicircle , positive , this is the term being taken between the proper limits .
78 .
Consider first an intenor point .
As already mentioned , we exclude an indefinitely small circle radius , and the limits for are from to value at the circumference ; viz. if here , then we have , and consequently and the integral therefore is As regards the second term , this is to , or , whatis the same thing , we may multiply by 2 and take the integral from to .
Writing then , and consequently , the term is ' and the value of the disk-integral is PROFESSOR CAYLEY ON PREPOTENTIALS .
But we have and thence ; that is or , what is the same thing , ; and the expression for the disk-integral is therefore 79 .
Writing as before , \amp ; c. , and , this is As a verification observe that if , each of the -functions becomes ; hence the whole first term .
this ' and the complete value is vanishing , as it should do , if 80 .
In the case of an exteriol point the process is somewhat different , but the result is of a like form .
We have Disk-integral referring to the point and 9 to the point M. Attendin first to the integral , and writing as before , this is the inferior and superior limits being here the values of which correspond to the points , A respectively , say , and ; hence , reversing the sign and interchanging the two limits , the value of is the above integral taken from to .
But similarly the value of is the same integral taken from to ; and for the two terms together the value is the same integral from to ; viz. we thus find Disk-integral ; viz. writing as before \amp ; c. , and , this is 81 .
As a verification , suppose that is indefinitely large : we must recur to the last preceding formula ; the value is thus ; viz. this is where the integral of the first term vanishes ; the value is thus where we may multiply by 2 and take the integral from to .
Writing then , the value is where the integral and hence the value is ; viz. this , over the circle , as is easily verified .
82 .
Reverting to the interior point PROFESSOR CAYLEY ON PREPOTENTIALS .
Disk-integral ; then reducing the expression in by the transformations for and the like tlansforma for , the term in may be expressed in the four forms:\mdash ; into II into into into 83 .
The first and fourth of these are susceptible of areduction which does notappear to be applicable to the second and third .
Consider in general the function ; the second -function is here ; viz. this is or , since the first term vanishes between the limits , this is Hence the two -functions together are that is .
We have therefore ; and from the same equation written in the form , we obtain .
84 .
Hence the terms in are , , respectively , and the corresponding values of the disk-integral are which we may again verify by writing therein , viz. the -functions thus become and and consequently the integral is PBOFESSOR CAYLEY ON phEPOTENTIALS .
85 .
But the forms nevertheless belong to a system of four ; from the formulae writing therein , we deduce ; and the last-mentioned values of the disk-integral may thus be written in the four forms : -term in , , ' ; and since the last of these is in fact the second of the original forms , it is clear that if instead of the first we had taken the second of the original forms , we should have obtained again the same system of four forms .
86 .
Writing as before \amp ; c. , the forms are \mdash ; term in ' ' ' ' 87 .
The third of these possesses a remarkable property : write instead of , and at the same time change into , the integral becomes \mdash ; term in ; and hence writing or , and therefore , the value is \mdash ; term in Hence the term in is into expression where the factor which multiplies is , as it should be , the ring-integral ; it in fact agrees with one of the expressions previously obtained for this integral .
88 .
Similarly for an exterior point ; starting in like manner from , Disk-integral and reducing in like manner , the term in may be expressed in the four forms into into into into PROFESSOR CAYLEY ON PREPOTENTIALS .
89 .
For the reduction of the first and fourth of these we have to consider ; viz. this is ; that is , .
[ I repeat for comparison the foregoing equation , ; by adding and subtracting these we obtain two new formulae ] ; for.reduction of the fourth formula the equation may be written .
90 .
But it is sufficient to consider the first formula ; the term in is , and the corresponding value of the is , which we may again verify by taking therein indefinitely large ; viz. the value is then , as above .
It is the first of a system of four forms , the others of which are , , .
And hence , writing as before , the four values are NDCCCLXXY .
where we may in the integrals write in place of , making the limits ; but the actual form is preferable .
91 .
In the third form for write , at the same time changing into mt ; the new value of the disk-integral is Writing here , that is , and observin that if be positive , the factor vanishes for the value at the lower limit , we see that on this supposition , positive , the value is ; viz. the term in is intp the expression that is into which is in fact into the value of the ring-integral .
92 .
Comparing for the cases of an interior point and an exterior point , the four expressions for the disk-integral , it will be noticed that only the third expressions correspond precisely to each other ; viz. these are : interior point , ; the value is where , if be positive ( which is in fact a necessary condition in order to the applicability of the formula ) , the term in vanishes , and may therefore be omitted : and CAYLBY ON PRfflPOTENTIALS .
exterior point , ; the value is differing only from the ceding one in the inferior limit in place of of the integral .
We have positive , and also positive ; viz. may have any value dimimshing from to - extreme values not admissible .
ANNEX IV .
Examples of Theorem A.\mdash ; Nos. 93 to 112 .
93 .
It is remarked in the text that in the examples which relate to the -coordinal sphere and ellipsoid respectively , we have a quantity , a function of the coordinates of the attracted point ; viz. in the case of the sphere , writing we have and in the case of the ellipsoid the equation having in each case a positive root which is called .
The properties of the equation are the same in each case ; but for the sphere , the equation being a quadric one , can be solved .
The equation in fact is and the positive root is therefore Suppose to gradually diminish and become ; for an exterior point , , the value of the radical is , and we have ; for an interior poirlt , , the value of the radical , supposing only indefinitely small , is , and we have ' or , what is the same thing , .
the positive root of the equation continually diminishes with , and becomes ultimately If or be indefinitely large , then the radical may be taken , and we have indefinitely large , 94 .
Every thing is the same with the general equation ; the left-hand side is for , and ( as decreases ) continually increases , becoming infinite for ; there is consequently a single positive value of for which the value is ; viz. the equation has a single positive root , and is taken to denote this root .
In the last-mentioned equation , let gradually diminish and become ; then for an exterior point , viz. if , the equation has ( as is at once seen ) a single positive root , and becomes equal to the positive root of this equation ; but for an interior point , or .
the equation just written down has no positive root , and becomes , that is the positive root of the original equation continually diminishes with , and for becomes ultimately ; its value for small is in fact given by .
Also or any of them indefinitely large , is indefinitely large , 95 .
We have an interesting geometrical illustration in the case is here determined by the equation ; viz. is the squared -semiaxis of the ellipsoid , confocal with the conic , which passes through the poin .
Taking , the point in question , , is a point in the plane of , outside the ellipse , and we have through the point a proper confocal ellipsoid , whose squared -semiaxis does not vanish ; but , then the point is within the ellipse , and the only confocal ellipsoid through the point is the indefinitely thin ellipsoid , squared semiaxes , which in fact coincides with the ellipse .
96 .
The positive root of the equation has certain properties which connect themselves with the function We have ( the accents denoting differentiations in regard to ) , or where and we have the like formulae for . . .
' We deduce ; PROFESSOR AYLEY ON and to this we may join , being arbitrary , Again , defining as immediately appears , we have ; and passing to the second differential coefficients , we have where with the like formulae for. . .
Joining to these , we obtain , where the last two tel.ms destroy each other ; and observing that we have , the result is 97 .
First example .
, and is the positive root of ] V is assumed , where is positive .
I do not work the example out ; it corresponds step by step with , and is hardly more simple than , the next example , which relates to the ellipsoid .
The result is ' if , if ; hence the integral taken over the sphere 98 .
Second example .
the positive root positive .
Consider here the function ; this satisfies the prepotential equation .
We have in fact with the like expressions for ; also Hence or , for .
and their values , this is Moreover V does not become infinite for any values of , not ; and it vanishes for points at ; and not only so , but for indefinitely large values of any of the coordinates it reduces itself to a numerical multiple of ; in fact in this case is indefinitely large , : consequently throughout the integral is indefinitely large , and we may therefore write that is The conditions of the theorem are thus satisfied , and we have for either of the formula : in the former of them must be positive ; in the latter it is sufficient if be positive ) .
CAYLEY ON REPOTENTIALS .
99 .
We have the same function of that V is of ; viz. writing for the positive root of the value of is Considering the formula which involves first , if , then when is the value of ?
is not ; the integral is therefore finite ( not indefinitely large ) , and we have , consequently But , then when is indefinitely small , is also indefinitely small : viz. we then have ; the value of is and hence 100 .
Again , using the formula which involves ; we have here or substituting for and their values and multiplying by , we find and therefore Hence , writing , first for an exterior point or is not , and the expression vanishes in virtue of the factor ; whence also ; next for interior point or is , hence also is infinite ; and neglecting in comparison with it the terms \amp ; c. , the value is and we have as before , .
Hence in the formula has the value just found , or , what is the same thing , we have over ellipsoid 102 .
We may in this result write .
There are two cases , according as the attracted point is exterior or interior : if it is will denote the positive root of the equation ; if it be interior , will ; and we thus have , for exterior , for interior point ; but as regards the value for an interior point it is to be observed that unless be negative between since positive hypothesis ) the two sides of the equation will be each of them infinite . .
Third example .
We assume here where PROFESSOB CAYLEY ON and , as before , is the positive root of the equation is positive in order that the integral may be finite ; also is positive .
104 .
In order to show that V satisfies the prepotential equation , I shall , in the first place , consider more general expression , where is a constant positive quantity which will be ultimately put .
The functions previously called and will be written and , and will now denote ; whence also , subtracting from the evanescent function , we have , say this is ; and we have thence , by former equations and in the present notation , In virtue of the equation which determines , we have ; and thence ; and thence ; and thence ; and thence ; and thence ; and thence ; and thence with like expressions for. . .
Also ; and hence - .
105 .
Writing I ' , for the first derived coefficients of I , in regard to , we have I ' , and the integral is therefore , , ; vanishing for , this is Hence , writing instead of , we have ; viz. this is \copy ; .
PROFESSOE CAYLEY ON SBEPOTENTIALS .
or , instead of , writing , this is .
We have here , suppose .
Also contains the factor , is suppose .
106 .
Substituting for , and their values , and , the whole result contains the factor , viz. we have ; and if here , except in the term , we write , we have and the formula becomes ; or instead using now in their original significations , , and this is or , what is the same thing , ; viz. the expression in is We thus see that being infinitesimal is infinitesimal of the order ; and hence being , we have ; viz. the prepotential equation is satisfied- by the value where is positive .
107 .
We ve consequently a value of corresponding to the foregoing value of V ; and this value is where , writing for the positive root of have we thence obtain ; or multiplying by , and substituting for its value we have where the second term , although containing the evanescent factor is for the present retained .
108 .
I attend to the second term .
, Suppose , then as diminishes and becomes does not become zero , but it becomes the positive root of the equation ; CAYLEY ON PRfflPOTENTIALS .
hence the.term , containing as well the evanescent factor as the other evanescent factor , is .
Suppose , then as diminishes to zero , A tends to become , but is finite and , whence is indefinitelylarge ; and since becomes , which is finite , the denominator may be reduced to , and the term therefore is which , the other factor being finite , vanishes in virtue of the evanescent factor Hence the second term always vanishes , and we have ( being ) 109 .
Considering first the case , then as diminishes to zero , does not become ; the integral contains no infinite element , and it consequently vanishes in virtue of the factor But if , then introducing instead of the new variable , that is , and writing for shortness , the term becomes where , as regards the limits corresponding to , we have , and corresponding to we have the positive root of .
But is indefinitely small ; except indefinitely small values of , we have , and ; and if be indefinitely small , then whether we take the accurate or the reduced expressions , the elements are finite , and the corresponding portion of the integral is indefinitely small .
We may consequently reduce as above ; viz. writing now the formula is ; or writing , the integral becomes , which is ; that is , we have and consequently that is viz. has this value for values of such that , but 110 .
Multiplying by a constant factor so as to reduce to the value , the final result is the limits being given by the equation is where is the positive root of In particular if , or PROFESSOR CAYLEY ON PREPOTENTIALS .
there are two cases , exterior , is positive root of \mdash ; interior , vanishes , viz. the limits in the integral are ; must be negative , ositive as before , in order that the -integral may not be infinite in regard to the element It is assumed in the proof that and are each of them positive ; but , as appears by the second example , the theorem is true for the extreme value ; it does not , however , appear that the proof can be extended to include the extreme value The formula seems , however , to hold good for values of beyond the foregoing limits ; and it would seem that the only necessary conditions are , and each of them positive .
The theorem is in fact a particular case of the following one , proved , Annex X. No. 162 , viz. over the ellipsoid where denotes : assuming , we have and the theorem is thus proved .
111 .
Particular cases : ; Cor. In a somewhat similar manner it may be shown that Multiply the first by and subtract the second , we have ; or writing for , this is and we have similar formulae with ( instead of ) in the numerator .
112 .
If , we have .
which , erentiated in respect to , gives the formula ; hence conversely , assuming the , . . .
formulae , we obtain by integration the last preceding formula to a constant pres , viz. we thereby obtain the multiple integral -hand function , where is independent of ; and by taking these all infinite , observing that then , the two integrals each vanish , and we obtain In particular , then which , putting therein , gives the potential of an ellipsoid for the cases of an exterior point and an interior point respectively .
V. GREEN'S Integrahon of the Prepotential Equation to 128 .
113 .
In the present Annex I in partreproduce GREiEN'S process for the integration of this equation by means of a series of functions analogous to LAPLACE'S Functions , and which may be termed " " Greenians\ldquo ; ( see his Memoir on the Attraction of Ellipsoids , referred to above ) ; each such function gives rise to a Prepotential Integral .
GREEN shows , by a complicated and difficult piece of general reasoning , that there exist solutions of the form ( see post , No. 116 ) , where is a function of the new variables without , such that being a function of only ; these functions of the variables are in fact the Greenian Functions in question .
The function of the order is ; those of the order 1 are ; those of the order 2 are , \amp ; c. , and -functions each of the form The existence of the functions just referred to other than the -functions involving the squares of the variables is obvious enough ; the difficulty first arises in regard to these -functions ; and the actual development of them appears to me important by reason of the light which is thereby thrown upon the general theory .
This I accomplish in the present Annex ; and I determine by GREnN'S process the corresponding prepotential integrals .
I do not go into the question of the Greenian Functions of orders superior to the second .
114. .
I write for greater clearness instead of to denote the series of variables ; viz. will denote a series of variables ; corresponding to these we have the semiaxes , and the new variables ; PROFESSOR CAYLEY ON IALS .
747 these last , with the before-mentioned function , are the new variables of the problem ; and for convenience there is introduced also a quantity ; viz. we have : where That is , we have a function of determined by ; and then are given as functions of the same quantities by the equations ; also , considered as a function of the same quantities , is 115 .
Introducing instead of the new variables , the transformed differential equation is where for shortness : -\amp ; c .
Also .
116 .
To integrate the equation for we assume where is a function of only , and a function of ( without ) , such that being a function of only .
Assuming that this is possible , the remaining equation to be satisfied is obviously Solutions of the form in question are \ldquo ; : ; and it can be shown next that there is a solution of the form 117 .
In fact , assuming that this satisfies , we must hare identically : ; PROFESSOR OATLEY ON so that from the term in we have ; or , what is the same thing , with the like equations from ; and from the constant term we have 118 .
Multiplying this last by , and adding it to the first , we obtain ; viz. putting for shortness , this is and similarly and to these we join the foregoing equation Eliminating , B. . .
we have an equation which determines as a function of ; and the equations then determine the l.atios of , B. . .
, so that these quantities will be given as determinate multiples of an arbitrary quantity M. The equation for is iu fact and the values of , B. .
are then , . . .
valueswhich seem to be dependent on : if they were so , it would be fatal to the success of the process ; but they are really independent of 119 .
That they are independent of depends on the theorem that we have where / is a quantity independent of determined by the equation ( is in fact the value of on writing ) , and that .
omitting the .
arbitrary multiplier , the values of , B. . .
then are , . . .
; or , what is the same thing , the value of is 120 .
[ To explain the ground of the assumption observe that , assuming \ldquo ; then multiplying out and reducing , we obtain ; viz. the equation divides out by the factor , thereby becoming that is , it gives for the foregoing value : hence clearly , having this value , we obtain by symmetry , , . . .
, proportional to ; .
the ratios , not only of , but of : B. . .
: will be independent of ] 121 .
To complete the transformation , starting with the foregoing value of , we have \amp ; c. ; so that we have : and t PROFESSOR CAYLEY ON PREPOTENTIALS .
Substituting for , their values , this last becomes ; viz. this is ; or substituting for its value , and diyiding out by , we have the equation for the determination of 122 .
The equation for is of the order ; there are consequently functions of the form in question , and each of the terms , can be expressed as a linear function of these .
It thus appears that any quadric function of , . . .
can be expressed as a sum of Greenian functions ; viz. the form is ( lines ) , viz. the terms multiplied by \amp ; c. respectively are those answering to the roots . . .
of the equation in The general conclusion is that any rational and function , . . .
7 can be expressed as a sum of Greenian functions .
123 .
We have next to integrate the equation Suppose , a particular solution is ; , a solution is ; in fact , omitting the constant denominator , writing , and therefore the equation to be verified is , which is right .
Again , suppose value belonging a paTticular solution is \mdash ; ; in fact the constant factor , or writing and therefore the equation to be verified is ; or putting for shortness \mdash ; ' this is which is true .
And generally the particular solution is deduced ffom the yalue of by writing therein in place of , . . .
7 respectively : say the value thus obtained is , where is what becomes by the above substitution .
124 .
Represent for a moment the equation in by PROFESSOR CAYLEY ON PREPOTENTIALS .
and assume that this is satisfied by , then we have ; and therefore ; viz. , multiplying by , this is or ; viz. substituting for its value , this is Hence , integrating , an arbitrary constant , and arbitrary , where the constants of integration are ; or , what is the same thing , taking the same function of that is of ( viz. is what becomes on writing therein in place of , . . .
respectively ) , then where X may be taken : we thus have Recollecting that so that for we have , the assumption comes to making V vanish for infinite values of 125 .
We have to find the value of corresponding to the foregoing value of .
the value of V , on writing therein in place of , then ( theorem A ) Take the same function of that is of , viz. the positive root of and corresponding to , viz. so that is the same function of that V is of say this is then we have for 9 the value , where is to be put 126 .
Suppose is , then if is not , but is the positive root of is , and we have is for all points outside the ellipsoid But if , then on vriting , we have where term in is PROFESSOR CAYLEY ON PBEPOTENTIALS .
Hence where are what become on writing therein .
It will be remembered that is becomes on changing therein into ; hence is what becomes on writing therein Moreover is what becomes on changing therein into lvl'iting , we have ; hence is what becomes on thelein into .
And is propel in to restore original bless by writing in place of 127 .
where , since for the value of V about to be mentioned vanishes for points outside the ellipsoid , the integral is to be taken over the ellipsoid and then ( transferring a constant factor ) if V the colresponding value of 9 is where is what becomes on vriting therein , and is what becomes on in place of 128 .
Thus putting for shortness , we have in the three several cases respectively , , ' , , 756 PROFESSOR CAYLEY ON PREPOTENTIAL\amp ; and for the case last considered same function with for ' where is the root of the equation ANNEX VI .
Examples of Theorem C.\mdash ; Nos. 129 to 132 .
129 .
First example relating to the coordinal sphere Assume , , ( a constant ) , these values each satisfy the potential equation .
V ' is not infinite for any point outside the surfaces , and for indefinitely large distances it is of the proper form .
is not infinite for any point inside the surface ; and at the surface V ' The conditions of the theorem are therefore satisfied ; and writing we have , where ; hence PROFESSOR CAYLEY ON PREPOTENTIALS .
which at the surface is Hence ( viz. is constant ) .
130 .
Writing for convenience ( a constant which may be put ) , also , 1ve have , and consequently for exterior point for interior point By making all indefinitely large we find viz. the expression on the right-hand side is here the mass of the shell thickness Taking we have the ordinary formulae the Potential of a uniform sphelical shell .
131 .
Suppose , but let the surface be the infinite cylinder .
Take ' each satisfying the potential equation ; but V ' , instead of vanishing , is infinite at infinity , and the conditions of the theorem are not satisfied ; the Potential of the cylinder is in fact infinite .
But the failure is a mere consequence of the special value of , viz. this is such that , instead of being positive , is .
Reverting to the general case of dimensional space , let the surface be the infinite cylinder ; and assume V ' ( a constant ) , these satisfy the potential equation ; viz. as regards V ' , we have , that is V ' is not infinite at any point outside the cylinder , and it vanishes at infinity , except indeed when only the coordinate is infinite , and its form at infinity is not is not infinite for any point within the cylinder ; and at the surface we hare We have , where at the surface ; and therefore ( viz. is constant ) ; or , what is the same , writing , whence , and writing also , we have for an exterior point for interior point 132 .
This is right ; but we can without difficulty bring it to coincide with the result obtained for the dimensional sphere with only in place of ; we may in fact , by a single integration , pass from the cylinder to the -dimensional sphere or circle , which is the base of this cylinder .
Writing first where refers to the variables and the sphere ; or usin now in this sense , then in place of the we have : and the limits of being , , then in place of we may write simply .
This being so , and putting for shortness , the integral is and we have without difficulty [ To it write , then the integral is in the first place converted into , which , putting and therefore , becomes which has the value in question .
] Hence replacing A by its value we have or ; PROFESSOR CAYLEY ON that is or or ; viz. this is the formula for the sphele with instead of ANNEX VII .
Exannple of Theorem D.\mdash ; Nos. \amp ; 133 .
The example relates to the dimensional sphele Instead of at once assuming for V a form satisfying the proper conditions as to continuity , we assume a form with indeterminate coefficients , and make it satisfy the conditions in question .
Write for ; for : In order that the two values may be equal at the surface , we must have and in order that the derived functions \amp ; c. may be equal , we must hayG \amp ; c. ; viz. these are all satisfied if only We have thus the values of A and , or the exterior potential .
as above value of the interior potential must be The corresponding values of are of course and and we thence find if if Assuming for the value , the last value becomes ; and writing for shortness , we have over dimensional sphere , for exterior point , for interior point 134 .
The case of the ellipsoid for dimensional space may be worked out by the theorem ; this is in fact what is done in tridimensional space ' by DIRICHLET in his Memoir of 1846 above referred to .
ANNEX VIII .
Prepotentials of the Homatoids.\mdash ; Nos. to 137 .
135 .
We have in tridimensional space the series of figures\mdash ; the plane , the line , the point ; and there is in like manner in dimensional space a corresponding series of terms ; the coordinal plane\mdash ; the line , the point : say these are the homaloids or homaloidal figures .
And ( taking the density as uniform , or , what is the same thing , ) we may consider the prepotentials of these several figures in regard to an attracted point , which , for greater icity , is taken not to be on the figure .
136 .
The integral may be written which still relates to dimensional space : the coordinates of the attracted point instead of being are ; viz. we have the coordinates , the coordinates , and the coordinate : and the integration is extended over the figure to , . . .
to And it is also assumed that is positive .
It is at once clear that we may reduce the integral to say for shortness ' where , is a constant as legards the integration , and where the limits in regard to each of the variables We may for these variables write , where ; and we then have PROFESSOR CAYLEY ON , where is the element of surface of the coordinal unit-sphere .
We thus obtain where the integral in regard to is taken from to , and the integral over the sulface of the unit-sphere ; hence by Annex I. the value of this last factor is The integral represented by the first factor will be finite , provided only be positive ; which is the case for any value whatever of if only be positive .
The first factor is an integral such as is considered in Annex II .
; to find its value we have only to write , and we thus find it to be , viz. and we thus have 137 .
As a verification observe that the prepotential equation , that is ; for a function V which contains only the variables becomes which is satisfied by V a constant multiple of ANNEX IX .
The of ispheric 1 138 .
The formula obtained ( Annex IV .
No. 110 ) is proved only for positive values of ; but writing therein , it becomes a formula which is obtainable as a particular case of a more general one ( notation to be presently explained ) , being a result obtained by JACOBI by a process which is in fact the extension to any number of variables of that made use of by GAUSS in his Memoir 'Determinatio attractionis quam .
exerceret planeta , \amp ; c. ' ( 1818 ) .
I to develop this theory .
139 .
JACOBI'S process has reference to a class of -tuple integrals ( including some of those here previously considered ) which may be termed ' ' epispheric\ldquo ; : considerin the variables connected by the equation , or say they the coordinates of a point on tuple unit-sphere , then the form is , where is the element of the surface of the unit-sphere , and is any function of the cooldinates ; the integral is taken to be of the form , and we then obtain the general result above leferred to .
Before going further it is conyenient to remark that taking as independent variables the coordinates , we have where stands for ; we must in obtaining the integral take account of the two values of , and finally extend the integral to the values of which satisfy If , as is ultimately done , in place of we write respectively , then the value of is , where now stands for ; we must in finding the value of the integral take account of the two values of , and finally extend the integral to the values of which satisfy 140 .
The determination of the integral depends upon formulae for the transformation of the spherical element , and of the quadric function First , as regards the spherical element ; let the variables which satisfy be regarded as functions of the independent variables , . . .
; then we have , for shortness .
' ' Suppose we effect on the variables a transformation PROFESSOR CAYLEY ON YTIALS .
thus introducing for the moment variables X , , . . .
, which satisfy identically , then considering these as functions of the foregoing independent variables , . . .
, we have X , : 141 .
Considering next the variables X , , . .
as linear functions ( with constant terms ) of the new variables ) , or say as linear functions of the quantities , 1 , which implies between them a linear relation ; assuming that we have so that in consequence of the left-hand side , the -hand side is also ; viz. are connected by : let represent the spherical element belonging to the coordinates , . . . .
Considerin these as functions of the foregoing independent variables , . . .
, we have . . . . .
' . . .
' ' . . .
' 142 .
We have in this expression ) , each of them a linear function of the quantities X , , . . .
; the determinant is consequently a linear function of like determinants obtained by substituting for the variables any out of the variables X , ; but in virtue of the equation ) NDCCCLXXY .
51 these determinants are proportional to the quantities X , respec tively , and the determinant thus assumes the form where is the like determinant with , and where the coefficients , . . .
are precisely those of the linear relation ; the Iast-mentioned expression is thus , or , substituting for its value , we have ; viz. comparing with the foregoing expression for we have which is the requisite formula for the transformation of 143 .
Consider the integral which , from its containing a single quadric function , may be called ' ' one-quadric.\ldquo ; Then effecting the foregoing transformation , and observing that the integral becomes where X , denote given linear functions ( with constant terms ) of the variables , or , what is the same thing , given linear functions of the quantities , 1 , such that identically We have then , and as the corresponding spherical element .
144 .
We may have X , such linear functions of , 1 that not only as above , but also ; CAYLEY ON REPOTENTIAL8 .
and this being so , the integral becomes where the coefficients , B. . .
are given by means of the identity ; viz. equating the discriminant to zero , we have an equation in , the roots whereof are -B -L .
The integral is ' which is of the form where I }isionally assume that are all positive .
145 .
To transform this , in place of the variables connected by , we introduce the variables such that , . . .
where and consequently Hence writing to denote the spherical element corresponding to the point , we have by a former formula or , what is the same thing , Hence integrating each side , and observing that , taken over the whole spherical surface , is , we have 146 .
For write herein respectively , and multiylying each side by , where is any positive integer or fractional number less than , integrate from to .
On the left-hand side , attending to the relation , the integral in regard to is where , as before is independent of ; the value of the definite integral is which , replacing by its value and multiplying by , and prefixing the integral sign , gives the left-hand side ; hence forming the equation and dividing by a numerical factor , we have and in particular if , then or , if for we restore the values , then ; viz. we thus have where is in fact a given rational and integral function of ; .
it is -Disct .
147 .
Consider in particular the integral ; here afXT ekWT ; PROFESSOR CAYLEY ON PREPOTBNTIALS .
viz. the discriminant taken negatively is which is , and consequently -A . . .
al.e the roots of the equation 148 .
The roots are all real ; moreover there is one and only one positive root .
Hence taking to be the positive root , we have A. .
all positive ; and therefore fortiori , . . .
, E-L all positive , which agrees with a egoing provisional assumption .
Or , writing gl.eater convenience to denote the positive quantity that is taking to be the positive l.oot of the equation we have or , what is the same thing , we have where on the left-hand side now denotes , and the limiting tion is . .
149 .
Suppose , then if the equation has a positive root differing from zero , which may be represented by the same letter ; but if then the positive root of the original equation becomes ; viz. as gradually diminishes to zero the positive root also diminishes , and becomes ultimately zero .
Hence writing , we have or , what is the same thing , now denoting the positive root of the equation or else denoting , according as or In the case , the inferior limit being then , this is in fact JACOBI'S theorem Crelle , ; but JACOBI does not consider the general case where is not , nor does he give explicitly the formula in the other case 150 .
Suppose being in the first instance not , then the former alternative holds good ; and observing , in regard to the form which contains in the denominator , that we can now take account of the two values by simply multiplying by 2 , we have ' on the light-hand side denoting , and the limiting equation being , each PROFESSOB CAYLEY PBEPOEENTIALS .
where is here the positive root of the equation , which is the formula referred to at the beginning of the present Annex .
We may in the formula write , thus obtaining the theorem under two different forms for the cases and respectively .
ANNEX X. Methods of and BOOLE.\mdash ; Nos. 151 to 162 .
151 .
The notion that the density is a discontinuous function vanishing for points outside the attracting mass has been made use of in a different manner by LEJEUNEDIRICHLET ( 1839 ) and BOOLE ( 1857 ) : viz. supposing that 9 has a given value within a given closed surface and is outside the surface , these geometers in the expression of a potential or prepotential integral replace by a definite integral which possesses the discontinuity in question , viz. it is for points inside the surface and for points outside the surface ; and then in the potential or prepotential integral they extend the integration over the whole of infinite space , thus getting rid of the equation of the surface as a limiting equation for the multiple integral .
152 .
LEJEUNE-DIBICHLET'S paper " " Sir un nouvelle methode pour la determination des integrales multiples\ldquo ; is published in ' Comptes Rendus , ' .
viii .
pp. 155-160 ( 1839 ) , and Liouv . .
pp. 164-168 ( same year ) .
The process is applied to the form over the ellipsoid ; but it would be equally applicable to the triple inte- gral itself , or say to the -tuple integral or , indeed , to over the ellipsoid ; but it may be as well to attend to the first form , as more resembling that considered by the author .
153 .
Since is or , according as is or , it follows that the is equal to the real part of the following expression , 770 PROFESSOR CAlLBY ON PREPOTENTIALS .
where the integrations in regard to arenow to be extended to for each variable , A further transformation is necessary : since positive and positive and writing herein for , and for , we have and the value is thus where the integral in regard to the variables is and here the -integral is and the like for the other integrals up to the z-integlal .
The resulting value is thus which , putting therein , is 154 .
But we have to consider only the real part of this expression ; viz. lvliting for shortness , we require the real part of iting here for its exponential , and using the formula ( positive ) , and the like one ( negative ) ( in which formula must be positive and less than 1 ) , we see that the real part in question is , or is ' according as or PROFESSOR CAYLEY ON 155 .
If the point is interior , , and consequently also , and the value , writing instead of , is But if the point be exterior , , and hence , writing for the positive root of the equation , ; viz. is the positive root of the ; then is greater than 1 , and continues so as increases , until .
for becomes and for larger values of we have ; and the expression thus is ; .
the two expressions in the cases of an interior point and an exterior point respectively give the value of the integral This is in fact the formula of Annex IV .
No. 110 , writing therein and 156 .
BOOLE'S researches are contained in two memoirs dated 1846 , " " On the Analysis of Discontinuous Functions Trans. Royal Irish Academy , vol. .
( 1848 ) , pp. 124-139 , and ' ' On a certain Multiple Definite Integral do .
pp. 140-150 ( the particular theorem about to be referred to is stated in the postscript of this memoir ) , and in the memoir " " On the Comparison of Transcendents , with certain applications to the theory of Definite Integrals Phil. Trans. vol. 147 , for 1857 , pp. 745-803 , the theorem being the third example , p. 794 .
The method is similar to that of , and was in fact suggested by , LEJEUNE .
DIRICHLET ; the auxiliary theorem made use of in the memoir of 1857 for the representation of the discontinuity being , which is a deduction from FOURIER'S theorem .
Changing the notation ( and in particular writing and for his and i ) the method is here applied to the determination of the -tuple integral ( where is an arbitrary function ) oyer the ellipsoid 157 .
The process is as follows : we have ; .
viz. the right-hand side is here equai to the left-hand side or is cording as or is consequently obtained by multiplying the right-hand side by and integrating from to for each variable .
Hence , changing the order of the integration , where Now , . . .
if , . . .
158 .
Substituting , and integrating with respect to between imits - , we have ; or , what is the same thing , writing in place of , this is ; that is , writing we have ; or , writing , this is 159 .
BOOLE writes ; viz. starting from FOURIER'S theorem , ( where is regarded as vanishing except when is between the limits , and the limits of are taken to be accordingly ) , then , according to an admissible theory of PROFESSOR CAYLEY ON general erentiation , we have the result in question .
He has in the formula instead of my ; and he proceeds , " " Here increases continually with .
As varies from to also varies from to .
To any positive limits of will correspond positive limits of : and these , as will hereafter appear [ refers to his note ] , will in certain cases replace the limits and in the expression for V 160 .
It seems better to deal with the result in the following manner , as in part shown p. 803 of BOOLE'S memoir .
Writing the integral in the form effect the integration in regard to ; viz. according as is greater or less than , then , or ' or ; and consequently , writing for its value , or as above 161 .
To further explain this , consider as an -coordinate and as a -coordinate ; then tracing the curve for positive values of , this is a mere hyperbolic branch , as shown in the figure , viz. ; and as continually increases to continually decreases to zero .
The limits are originally taken to be from to and to , viz. over the infinite strip bounded by the lines , Ol , 11 ; but within these limits the function under the integral sign is to be replaced by zero whenever the values are such that is less than , viz. when the values belong to a point in the shaded portion of the strip ; the integral is therefore to be extended only over the unshaded portion of the strip ; viz. the value is the double integral being taken over the unshaded portion of the strip ; or , what is the same thing , the integral in regard to is to be taken from ( say from ) to , and then the integral in regard to is to be taken from to , as before , is the positive root of the equation , that is of 162 .
Write , and therefore and ; then the limits of correspond to the limits of , and the formula becomes where is retained in place of its value .
This is in fact a form ( deduced from BOOLE'S result in the memoir of 1846 ) given by me , Cambridge and Dublin Mathematical Journal , vol. .
( 1847 ) , p. 219 .
If in particular , then , and thence ; and thence restorin for its value , we have as the value of the integral over the ellipsoid .
This is in fact the theorem of Annex I .
No. 110 in its general form ; but the proof assumes that is positive .

