Analysis of the Roots of Equations .
By the Rev. R. MURPHY , MA .
, Fellow of a aus College , Honorary Member of various Philosophical Societies .
Communicated by J. W. LUBBOCK , Esq. FR .
S. Received April 6 , -Read April 27 , 1837 .
1 .
THE object of this memoir is to show how the constituent parts of the roots of algebraical equations may be determined , by considering the conditions under which they vanish , and conversely to show the signification of each such constituent part .
2 .
In equations of degrees higher than the second the same constituent part of the root is found in several places governed by the same radical sign , but affected with the different corresponding roots of unity as multipliers .
3 .
The root of every equation , of which the coefficients are rational , contains a rational part , for the sum of the roots could not otherwise be rational .
This rational part , as such , is insusceptible of change in the different roots of the same equation , consequently its value is the coefficient of the second term ( with a changed sign ) divided by the number of roots , or index of the first term .
4 .
The supposed evanescence of any of the other constituent parts implies that a relation exists between the roots ; if such a relation be expressed by equating a function of the roots to zero , that constituent part will be the product of all such functions , and a numerical factor .
5 .
The joint evanescence of various constituent parts implies the co-existence of various relations between the roots , and thus an interpretation may be given to each of the constituent parts , riveting the expression of the root in the memory , and beautifully converting the solution of a problem into a condensed enunciation of various theorems .
For simplicity these principles are first applied to equations of lower degrees .
6 .
Let us take for example the general quadratic equation x2 + ax + b , the two roots .
of which are represented by XI ) x2 in the formule , 1i--g + Va aa=24 To find a suppose 06 0 , a relation is then established between the quantities xl and a2 , viz. to^= 0 , MnCccxxxvII .
xx2 vanishing with M is a factor of it ; and since the roots must be symmetricaQy involved in a. , the other factor is x2x1 , whence k ( xI2)2where k is simply a number .
a Put for its expression in terms of the roots , and we thus have XI =2 X+ xi 24A xk5m Xl + X2 XlA XiX2 _ V4k , whence x1i-X2 =(V1 X2 ) V4 k , or k-= then _2 ( XI4 ( 1 2 which is a symmetrical function , and therefore easily expressed by the coefficientsse 7 .
From this it follows that a=0 is the condition that the equation may have two equal roots ; but if the proposed quadratic be represented by p= 0 , and its derived equation by p ' Q)9 which is the same as 2x+a= 0 , the condition for two equal roots is obtained by eliminating x between these equations , which by the theory of elimination gives as the sought condition P1 ( xI ) .
( X2 ) =0 we have thus a ka ' lp ' ( XI ) P ' ( X2 ) k ' being numerical .
Now p ( x ) = ( x x ) ( x -X2 ) , therefore I ' ( v ) = ( x )+ ( X x2 ) , whence 1 ' ( X1 ) = X1-X e ( X2)=X2X1 consequently k ' - ; and converting the sum and product of XI , x2 into the coefficients , we obtain TV=-g +74 -b 2=e ( 4 b8 .
The constituent parts in the roots which have been the objects of investigatiout a were and o , and with respect to their evanescence we have the following theorem .
The vanishing of that part of the root of a quadratic which is under the radicalJ sign implies the existence of two equal roots .
The vanishing of the other part which is unaffected by that sign signifies that the roots are equal , but with contrary signs .
By the aid of this theorem we shall be able to find two of the three constituent parts of the roots of a cubic equation .
Let us now extend the same views to equations of the third degree , and let X 2x3bethethreerootsofthecubicx3+a 2+bx+c=O .
Put -a + +07 X ' = -+V+Y3 a e= + 02/ + Air- .
The numbers 4 , 4 are the imaginary cube roots of unity , and we may observe that the formula for x3 , rv differ only from that for x2 in having 42 , 43 respectively instead of 4 .
10 .
The quantities a , j3 , which are obviously similarly involved , are the roots of a quadratic , and of the forms M=d+ Val ' al = a/ all ; the three quantities 4 , a ' and a !
l are the constituent parts of the roots of the cubic 3~~~~~X in the sense in which those words have been used ; the first is the same as 3 and the other two can be found from the conditions of their evanescence as follows .
1 1 .
Suppose a !
" = 0 , the theorem of art .
8 .
gives us a= 3 , whence we find x2 = X3 now x2 x3 vanishing with a !
' is a factor of it ; and the other symmetrical factors are x3 x2 , X1 -x3 , x3 xI , x x2 , x2 -a , k being a number , we must therefore have a't =k ( xI X2)2 ( X1 X3)2 ( X2 -3)2* 12 .
To find ac ' in like manner , suppose a !
' A the theorem of art .
8 before referred to , in this case makes a &amp; =-( and the three roots of the cubic accordingly are changed to the following : a X1 =a 02 ) whence we readily find that 2 XI = sx + x3 , therefore 2X 2x3 is a factor of a !
, and the other symmetrical factors are 2 x2 a- , and 2 x3 -2 hence if k ' be a numerical factor M =(2 x , xx3 ) ( 2 x2 x3 ) ( 2 X3 -^ ) ; these symmetrical functions can be expressed by the given coefficients of the equation .
The constants k , k ' may be easily found in various ways , perhaps the simplest means is to sueppose a ' 0O to find k , and " =0 to find k ' .
If we put x3= 0 and x , =2 x1 , we have a ' O. , and ac " =4 kX16 , henceo =2 X13 V/ * , = -2 x3 V/ k , and substituting in the formula for x2 , we have ( since a XI ) 2xi=xl+x,0V12 V/ k-x , 024/ 2 Vk 1 =@(1.-0 ) / 2 V k , therefore 2Vk ( 0)3 -3 ( 02 0 ) and 4k ( = 9 ( O -2 + 0 ) but 1+G+ 02 = 0 ; therefore 14 .
In like manner to find k ' suppose X2 == 0 , which makes ce " = 0 .
and cz ' +2 k'.x13 , andc =a o ' , therefore =y+2 x1 i/ 2 k ' or -V=21c ' , whence 15 .
The preceding analysis furnishes the following formula , 3 x1= ( X1 + X2 + X3 ) + / { + ( 2 Z- , x3 ) ( 2 x2 x3 ) ( 23 x 2 ) ++V 3A 3 ( Xi .
2)2 , ( x-3)2 ( X2X3)2 } +1 { +(2xi1-X2-X3 ) ( 2X2-xi-x3 ) ( 2x3-XI-X2 ) -~ 3 ( x1 -x)2 ( XI x)2 ( 22 2 / -(1 X2)2(I X3)(X2 X3)2 the corresponding formule for 3 x2 , 3 x3 being obtained by writing 0 and 02 before the cubic radical signs .
In consequence of the negative multiplier 3 under the sign of square root it is visible that this formula is not arithmetically applicable when the three roots are real and unequal , which is usually termed the irreducible case .
16 .
The cubic surds in the formula above given are actually extractible , which ver ifies the solution .
To this end let 2 xi X23 = A , and x2 -X=3 B , then 12 x2 1-XI-x3 = Gz(3 A ) 2 x3 = XI(23 B A ) therefore 2 ( 2x1 -x2-x3 ) ( 2 X2-XI 3 ) ( 2 --x X2 ) =8 ( A3-9 A B2 ) .
Again Xi -2 =4 .
( A B ) xi =3 + ( A + B ) therefore 2* ( x1X2 ) ( X1 X3 ) ( X2 X3 ) = 8-(A2 -B3 ) the total quantity under the first cubic surd thus becomes 4 .
{ A3 +3 A2 .
( B V 3 ) +3 A. ( B V 3)2 + ( B V -3)3 the cube root of which is I { A +BV 3 } , and the actual root of the second surd is similarly 4 { A BV 3 } .
But 2.(A +B -3 ) X1-x2 2-4/ -3 e= X1 + x2 02 + X3 0 .
And ( A-B V-3 ) =xj+x2d+X3d ; the formulav of this solution become then those of VAUDERMONDE , viz. 3 x1 = ( X1 + X2 + X3 ) + ( X1 + X2 02 + X3 0 ) + ( X1 + X2 0+ x3 02 ) 3 x2 = ( X1 + X2 + X3 ) + ( X1 D+ X2 + X3 02 ) + ( X 02 + x2 + X3 0 ) 3 x3 = ( X1 + x2 + X3 ) + ( X1 02 + X2 0+ X3 ) + ( X1 0+ X2 02 + X3 ) , which contain a complete verification .
17 .
The two constituent parts a ' a &amp; " of the roots of a cubic equation have been rek solved into factors by observing the relations established between the roots by their evanescence ; another mode exists for forming the same quantities by elimination between the proposed equation and its first and second derived equations .
Let the given equation x3 +a X2 +bx+c=0 be represented by p ( x ) = 0 , and the first derived , viz. 3 X2 +2ax+b=0 by I ' ( x ) = 0 , the second derived 6x + 2a =0 by@ " ( x ) 0 .
When d ' =0 we have seen that the equation p ( x ) =0 has two equal roots But when p ( x ) =( has equal roots , is expressed by making the result of eliminating x between p ( x ) =0 and q ' ( x ) =0 to vanish , and by the theory of elimination this result is p ' ( 'x1 ) .
p ' ( x2 ) .
f ' ( x3 ) hence we have ( h being a number ) , all =h p't(xl ) p# ( 2 ) * pt ( 3 ) .
19 .
Suppose now the joint evanescence of cg and d ' , the equation has then three equal roots xl = x23= 3 , the system of equations a !
=0 a ' 0 , the first of three dimensions relative to the roots , the second of six , are therefore the two conditions necessary for the existence of three equal roots .
The results of the elimination of x between p ( x ) =0 p " ( x ) =0 and P ( x ) O if ( )0 , give also the two conditions for three equal roots of the same dimensions as the above , with which this system is identical , we have thus , h ' being numerical ast =t At p"(XI ) .
Art ( x2 ) .
t(X3 ) .
20 .
Since P &gt; ( X = ( X XI ) ( X2 ) ( X 3 ) ' ( x ) = ( X X2 ) ( XX3 ) + ( x -x ) ( x -x3 ) + ( X X ) ( X X2 ) p"(x ) =2 ( x-x3 ) +2 ( x2 ) +2 ( x-x1 )I therefore P ' ( XI ) .
Pt ( X2 ) Pt ( X3 ) = ( XX2)2 ( X1 X3)2 ( X2 X3)2 $ " ( X1 ) .
it ( X2 ) .
p " ( X3 ) =8 ( 2 XIX2x3 ) ( 2 X2 -X3 ) ( 2 x3 2 ) the values of a ' a ' are therefore conformable to those before found , and t__ rBt ' _L ___ 2 .
3 ' 8 = .
33 21 .
In the elimination of a quantity between two equations into which that quantity enters rationally , it is in general indifferent which of the two equations is selected that its roots may be substituted for such quantity , for the dimensions of the product is the same whether we substitute the n roots of an equation of n dimensions in one of m , or the m roots of the latter in the former of n dimensions , and take the products ; these products are not only of the same dimensions but imply the coexistence of the same system of equations , and can only differ from each other by numerical multipliers , when the numerical coefficient in the one differs from that of the other .
In the present instance , if t MS be the roots of the equation qV = 0 , and X that of t= 0 , the values of d all may also be expressed in the following form , all -H v ( it ) .
p ( it ) allK .
p ( X ) the factors H , K being numbers .
22 .
Now if we observe that by the equation pt ( t ) = 0 , t : 22 a. b 1=as a:.nd therefore 03 2a 02b 4 01(Ib )fi 2ab 3.~ O* 9 ).+ 9 ?
a it will follow that o ( 0 ) or 0+a 12 +b1+ c , is the same as At+ B , pu1tting ab Bus~2 Qb H~ence 9 ( ~i ) ) When the quantity ( ) in the expression for the root which is unaffected by surds vanishes , one root is equal to the sum of the other two with changed signs .
When-and or " vanish conjointly , two of the roots are each one half of the third with a changed sign .
a When and a ' vanish jointly , the equation has two roots equal , but with con , trary signs , and the third root is zero .
Whena ' , all vanish simultaneously , the three roots are equal to each other and to zero .
24 .
Biquadratic Equations.-Let x1 X2 , TX31 xbe the four roots of the equation x4 + ax3 + bx2 + cx + d , the number 4 being composite allows the subdivision of the sumns of the roots taken two and two into three pairs , which have rational suns , thus XI+t o,2=-(3+ ++ X ) t 3+c ) ( X2 + X4t + ) XI +X4+ 2 -(2 ?
x3 ?
+ ) therefore the equation , of which the roots are aaX 4+a l 4+a X1 + 2+ 1+X 3+ + X , 1+ X4 + -t ( 3+ + &amp; c ) is one of six dimensions , but without terms involving the odd powers of the unknOWr ' quantity , and therefore these quantities are the square roots of the roots of a cuhicY equation .
Put therefore +a xl +X3+ =2 24 +a ?
+x -=2 Vr XI+X4 +=2 V7 therefore 3a 2*1i + ( XI + X2 +X3 +4 ) +_ -2 ( A a+ v0+ v / ) o=4 +V M+ V+V Y , the quantities oa , 3 , y being the roots of a cubic equation , are of the form = ct +0 / ( ' + 4/ 1 " 7_ o ' + 62 d ( 3 +0/ ( 3 " , when 0 , 02 are the imaginary cube roots of unity , and I ' , ( " being the roots of a quadratic are of the forms I -O !
1 + V/ Mit M3 v ' !
' The three other roots of the biquadratic are 2= 2 A/ a -t1+ e)= 4+ Va/ t/ X3 =2 UP -(XI + A/ ) =/ Vey+ x4 =2 V y-(xI +-2 ) = yr7 The constituent or essentially different parts of the roots are a al/ , cIi , x ' , which we proceed to analyse by the conditions of their evanescence .
25 .
Suppose oi " O. , then by art .
23 two of the roots of the cubic are equal , or ( 3= ry from whence we have X3 = x4 , therefore x3 x4 is a factor of a " ' , and forming all the other symmetrical factors , we have a " ' =k ( x1 -x2)2 ( X1 X3)2 ( XI X4)2 ( X2 x3)2 ( X2 X4)2 ( X3 -x4)2 k being a numerical multiplier .
26 .
Next suppose c ' = 0 , then by the properties of the roots of the cubic already demonstrated we have 2o= ( 3 + r , or o(3y v , whence ( V/ o+ VO3 ) ( V V ) ( y+ VCM ( 'yVa ) therefore ( X X4 ) ( X2 -3 ) ( XI Xx3 ) ( X4 X2 ) one factor of a " ' is found thus to be ( xI x4 ) ( X2 X3 ) + ( XI x3 ) ( X2 X4 ) The two remaining symmetrical factors are ( xI X2 ) ( X3 X4 ) + ( XI X4 ) ( X3 X2 ) ( X1 -X3 ) ( X4 -X2 ) + ( X -X2 ) ( X4 x3 ) and ac " is the product of all three multiplied by a numerical factor k ' .
27 , Again , suppose C ' = 0 , then by the article above referred to a+(+y 0 ; but ( 3+ r ( X X2)2 + ( Xa3 )2 += ( X1 X3)2 + ( X2-X4)2 += ( X _ X4)2 + ( x2 X)2 the sum of which being per se symmetrical , shows that ca ' has no other but a numerical factor ; therefore Mc ' = ic " { ( x1 _2)j ( X( .
: ) ( .
, p r ( X'2-x4)2 + X_42 The numbers k , k ' , k " may be found by the values already given for the roots of a cubic , by which we have = GI +4 ( of +v ( jU ) + ( all Vil ) and a+ +y therefore K"ash , 3 *(2 X --)(2 0 --)(2 ya1 .
and the factors into which A ' is multiplied are the same as 2 .
( 2 MM y ) .
( 2 r ) .
( 2 r 3 ) therefore Lastly =-92.33 .
( I-03)2 ( a y)2 ( a 7)2 .
But since x+ r(x1 v ) + ( X2 -4 ) and ( + ?
= ( XI X4)2 + ( x29 ) therefore =2 ( xI x4 X1 x3 + X2 x3 2 X4 ) -2 ( x1 X2 ) ( X4 X3 ) Similarly a oy and ( -y are expressed , and comparing the expression for a " ' thus arising with that found before , we have MIII 24 -22 .
33 a &amp; 't ' 26 .
O ' or k -33 .
29 .
Let us next seek the same quantities tx ' , ox " , tx"I , by the theory of elimination. .
When v , " ' = 0 .
the proposed equation which it will be convenient to express by p ( x ) =0 has two equal roots , the condition for which is also obtained by eliminating x between p ( x ) =0 and the first derived A ' ( x ) = 0 , the function of the coefficients arising from this elimination is of the same dimensions , and expresses the same condition as the constituent quantity v , " ' , and therefore only differs from it by a numerical multiplier .
This quantity in a symmetrical form relative to the roots is therefore MM = h. PI ( a"I 3 .
@ ( X2 ) o PI ( X3 ) e P ) ( X4 ) and since PI ( XI ) = ( X1 X2 ) ( X1 -X3 ) ( X1 X4 ) we have the same result as by the former method and h= k. 30 .
When I ' " vanishes jointly with sc/ " ' , then since tx= ( 3y we have also n-T'2_X3= X4 , therefore the equation p ( x ) =0 has three equal roots , and since a " ' =0 denotes the existence of two equal roots , therefore ac " =0 is the additional condition for a third , or the system of equations ae " = 0 , al"l =0 are equivalent to the system ( x ) = 0 , I ' ( x ) = 0 , P " ( x ) = 0 .
Now =( )= 6 x2 +3 ax +6=z+b for abridgment , we shall next form an equation in z indicative of two equal roots , and eliminating z by the equation z+b0 we shall obtain &amp; a " .
When x1=2 ' then 6 XI2 +3 ax1 6x2-3x , ( 2x1x3 +x4 ) 3 ( x1 x3 + X1 x4 ) 3 ( X1I X3 + X2 X4 ) the equation z+3 ( X1 X3 + X2 X4 ) =0 expresses that x1= and forming a cubic equation by taking all the symmetrical simple equations , a condition in % for the existence of any pair of equal roots is obtained , viz. F(z)={z+3(xl3+x2x4)}.{z+3(X1 X2 + X3X4)}.{z+3(14+2x3 ) } 0F ( z ) = Z3 +3b z2 +9 ( a c4 d ) z+ 27 { d ( a2 4 b ) + e2 } = 0 ; and if z be eliminated between this equation , and z+b= 0 , the result multiplied by a constant will be ad or cl-l h'F ( -b ) .
Now one factor of F(b ) =2 ( X x3 + x2x4 ) ( x x2+xx4+x23+x3x4 ) = ( X1 X2 ) ( X3 X4 ) + ( X1 X4 ) ( X3 X2 ) ; the other two being symmetrical with it gives the same value of a " as before , and h'= k ' .
31 .
If ol =0= a"0 , ' d0 simultaneously , then x2 = X2 X3 =4A Nowa ' 'is a function of two dimensions , as is that arising by eliminating x between p " ( x ) = 07 ( x ) = 0 , and gives the same condition , but 'p " ' ( x ) =6 ( 4 x+ a ) , therefore a " ( -i ) is the result of this elimination , therefore a h ' = h " p " ( --4- ) .
Now P " ( x ) =2 ( 6 X2 +3ax+ b ) therefore p " ( a ) 2b3 a , 4 4~~~ a and our former value of a ' is k " { 3 x122 x1 x2 } = i " ( 3 a2 8 b ) hence It " = -4 ic " .
z 2 Collecting the results of the last three articles they give us o=-3 * ( 4 ) =4 33 IF ( b ) Q4 =ttt--,3 I ( X ) IO ( X2 ) IO ( X3 ) ( VA ) , all of which may be expressed easily in terms of the coefficients .
33 .
Theorems deduced.-The root of the biquadratic X4 +a x3 +b X2 +cx+d= being expressed by to= -ii+ V/ { os ' + 4/ @2 ' + V/ r " ' ) + i/ ( os " V/ c2 " ) } + V/ { c ?
1 +4V ( M2 ' ?
V &amp; " ~ ) + 42 J ( &amp; s " V vcat ) } +v { 6 % ' +4 ( cl +)+ V(a " V c2 ) } where 4 , 42 are the imaginary cube roots of unity , then the condition a " ' =0 denotes the existence of two equal roots in the proposed equation .
The condition '"= 0 denotes the following relation of the roots ( x1 X4 ) ( X2 T3 ) + ( X1 X3 ) ( x2 x4 ) = 0 .
The system of coexisting conditions a " = 0 , a " ' 0 are necessary and sufficient for the existence of three equal roots .
The condition oa ' =0 denotes the following relation of the roots , I ( X1 -x2)2 = 0 .
The simultaneous system of conditions os ' = 0 , osf = 0 , a"= 0 essentially and suffix ciently express the coexistence of four equal roots .
a The rational part of the root -4 only vanishes with the sum of the roots .
34 .
We now proceed to determine the constituent parts of the roots of equations of the fifth degree by the conditions of their evanescence .
4 , 42 represent the imaginary cube roots of unity .
&amp; 2 W3 , w4 the imaginary fifth roots of unity .
X1 , X2 , X3 , x4 , X5 are the five roots of the proposed equation of the fifth degree , viz. x5 +a x4 + bx3 +CxX2+ dx +e 0 .
X1 =5+ &lt; 2 v++v3+r a2 5 +Z + 2 , / +W3 Vi + W4v a i.2 a5 + 1,4 +d W.1,.+ s. .
1v .
5 =+ A V/ a+WVP+ Q2 Vr+V a5 5 44 a/ a ?
+0 813 + 4S2 , + W , , &gt; &gt; the formulae for x3 , X4 , x5 , xl are derived frord the formula for x2 , by writing succesSively W2 , cW35 r4 , c5 instead of w. 3 a ' + P ' V 'V 'y a ' V f+'P+ VI ' V= o ' Vh[i ' -/ V 7 ' +V b ' , such being the forms of the roots of a biquadratic .
Again , the expressions for [ I ' , y ' , Y ' as roots of a cubic , are [ a " it +V pt " +V 7 " y _al + O , &gt; / V tP + O2 Vr"l = a " + O2 ft3 F+dV r " .
Lastly , ( " , y " as roots of a quadratic , are expressed by the following formulae It _ =at +/ WIV it fit _ IV The quantitiesj , ae ' , al " , t Iv are the constituent or essentially distinct parts of the roots xDx2 , x3 , x4 , x5 , and the analysis of their formation is to be sought by observing all the conditions under which each may vanish .
35 .
If for a we put X1 +,2 +3+4+ X5 it is obvious that the system of five equations for the roots is equivalent to one of only four , viz. 5Va =X1 + 6)4 x2 + cW X3 + 0r2 X4 +O X5 5VP= X1 +3 X2 + 6)X3 + W4 X4 +2 x5 5V7 xI + W2 X2 +W 4X3 + WX4 + W3 55 V/ 8 =I+ WX2 + W2X3 + 3xV4 + C4 X5 the right hand members of which equations differ from each other only by the particular fifth root of unity in use , which considered as X in the first , will be W2 , g3 ah respectively in the second , third , and fourth .
36 .
Suppose aIv = 0 .
then by art .
33 two roots of the biquadratic must be equal .
First let y= 6 , which can only happen under the five following relations , V/ 7t =VnVr=V2V7/ VMVr= W4 V , V r ; V , whfich furnish five factors , linear functions of the differences of the roots , and since these differences may be taken either way , we have also from the same equations five other factors equal to the former and with contrary signs .
In this manner ten factors of a=v may be found by equating any pair of the four quantities a , P , ry ~ , and the number of pairs being six , the whole number of factors of al is sixty , these factors are very easily formed , and here we present the first thirty factors , the remaining thirty being formed merely by changing the signs of these , or , which is the same , inverting the order of the differences of the roots in each factor .
37 .
For greater clearness we shall subdivide these thirty factors into five groups , from the first of which xi is excluded , from the second x2 , and so on ; each subdivi .
sion contains six factors , four of which are of one form , and two of a different form they are as follow : ( x2 X4 ) +~ ( X3 -X4 ) +0 ( X3-X5 ) o ( 1oV ( X3 -X2 ) +w ( X5-X2 ) + 2(5_4 ) **** ( 2 .
) ( X4 X5 ) + ( X22-X5 ) +2 ( X2-X3 ) * ?
* .
( 3 .
) ( X5-X3 ) +O ( X4 X3 ) + W2 ( X4.X2 ) ?
Q ( X2 XB5 ) + ( 0j + 0j ) ( X3 4 ) 5o ( X3 ~X4)+(2+ 3 ) ( X5 5X2 ) ( 6*)j ( XI1 X4 ) + GV ( X5 4 ) + 0j2 ( X5 X ) oo o(* ( X3 -x5 ) + ( X4 -X5 ) +2 ( X4 XI ) *r 2 ) ( X4 &lt; X3 ) +W ( XI X3 ) + c2 ( xi x5 ) * .
( 3 ) ( X5 1 XI ) +(31)~ ( X3X4 )e^woeo ( 4 .
) ( X( ))+ ( ( 2+ 3 ) ( ? ?
5( 4 ) , , , ,5 ( 4 ) ( x-X)4 ) + x2x2 + ( 63 2 +X XV ) ( 3 .
) ( XI X.2 ) +6 ( X4 )Xj2 ) +O ( X4 Xj ) ( ( X2 5 ) + *j ( XI _X5).+ 02 ( XI X4 ) ***..(2 .
) ( X4 X ) + OJ ( X5 Xr + O2 ( X5 X2 ) ... '..(3- ) 1(C .
( X5 4 ) + &amp; J(X2 X &gt; 4 ) + k2 ( X2S XI ) ?
s5 e ( 4 o ) ( Xi X &gt; 5 ) + ( OJ + aJ3 ) ( X4i X|2 ) ?
eo ( 5 * ) ( X &gt; 4 2 ) + OJ2 + &amp; J3 ) ( X5 XI ) * o*o(6.)S ( XiX5 ) + *j ( X3 _ X5 ) + W2 ( X3 _ 2 ) 2*9 @*(1* ( 2 3 ) +( &amp; )+ k ( X5 1 )3 e XI)o e( 2.* ) ( 8t3 -Z + V(X+ ap2 ( X12 X5 ) * ** ( 3 .
) i( ( X5 2 ) +W ( X1 X2 ) +2( -3 ) se9w ( 4 .
) ( XI3X &gt; ) -t ( &amp; J2 + *J3 ) ( X32 X5 ) + &amp; J ( x2 X3 ) +2 ( *2 -X4 ) * ( 1 ( X2 xi ) + ij ( t4-xi ) + w2 ( X4 -x3 ) e ( 2 .
) !
( X3 Xi )+ W(X 4)+ W2 ( XI 2 ) e e(3.e ) ( E. ( X4 X2 ) +j ( X3X2 ) + c2 ( X3 -X ) ( 4 .
( XI } X4 ) + ( *J1 +{ &amp; 53 ) ( X2 X73 )e eee(5e ) ( X2 X3 ) + ( 2 + 3 ) ( X4 X ) =2X )hew , A5 = AV lie ( VIVr ) B5 = -i ( / y ) W4 ( a(4/ 5 2Ar ) B _ .
, -j1)Q(fM-isV ) C6- , = , , 11 ( V/ SG V/ ( 1 ( V -v 4 ( _ 0E6=(a(o -4(/ GCa 0 Therefore 7-a-h.A1 B2C3D4E , 1 f3-d =-h.A2B3C4D E2 23412~~1 tx ty= A3 B4 C , D2 Es _ 1.5 a= h. A4B , 2D3E4 95 3-7h .
A5 B5 C5 D5 E. Kx8-h.A.B6 .
C6 .
D6 .
E6 39 .
Now the conditions necessary for the evanescence of as " ' are by art .
33 .
( ) ( ( 3 -r ) + ( a r ) ( ( 3= 0( r ) ( b -(3 ) + ( a ( ) ( O t ) 0 ( M-d)(y-t)+ ( V- ) ( y= o Substitute the values of a ( -y , &amp;c. above found , and including h2 in the nune ical multiplier k of the whole , we shall have f= k ' ( A5 A6.B5B6 C5 C6 .
D5D6 .
E5E6 -A2A3.B3B4 .
C4 C1 .
D1 D2 .
E2E3 ) x ( A5A6 .
B5B6.CS C6 DS D6 .
1 EE6 + AA14 .
B1 B2 .
C2 C3.-D3D4 .
E4EI )X ( A2A3 *B3B4.C4 C1 .
D1 D2 .
E2E3 + A4A1 .
B1 B2 .
C2 C3 .
D3D4 .
E4E1 ) .
40 .
The condition that a " may vanish is Y(P )2 =0 by art .
33 .
Helce 'V ( A12 B22 C32 D42 E12 + A2B2 C42 D12 E22 + A52 B52 C52D2 E52 + A32 B42 C12 D22 1+ A4 2 C22 D32 E42 + A62 B62 C62 D62 E62 ) , k " being a numerical quantity .
41 .
The terms which compose oc ' are indicated below by the indices which enter them placed between brackets .
[ 51 =4 1x5 [ 4,1 ] r_ 5 &gt; X14X X2 [ 3 , 2 ] -1 0 0xX13 X22 [ 3 , 1 , 1 -20X13 ( X2 X3 + X2 X4 +x X5 + X4 X5-4X2 X5-x 4 X3 X4 ) 20x23 ( x1 x4 + xtx5 + x3x4 + X3X5 4xlx 34x4x5 ) -20x3 ( X1X2 + X1X4 + x2X5 +x4 X5 4xx5 -4x2 X4 ) 20x43 ( xV x3 + oh x5 + x2x3 + 2x5 4x1 x2 34 " 35 ) -20x53 ( X1X2 + X1X3 + X2X4 + X3X4 4x1x4 4x2x3 ) [ 2 2 , 1 ] -30 { x12 x22 ( x3+x5 -4X4 ) + 12 2(X4 + X5 4 X2 ) +x 2x 42 ( X2 + x3 4 x5 ) + X12 x52 ( X2 + x4 4 x3 ) + X22 X32 ( X1 + x4-4 x5 ) + X22X42 ( x1 + 54x3 ) + X2 X52 ( X3 + X4 4 x1 ) + X32x42 ( X +2 4x ) +X32X52(X1 + X24x4 ) + X42X52(X1 +X34X2 ) } [ 2 11 11 ] 60 xi22x3X4 [ 1 , 1,1 , 1 , 1 ] 480Xx2x34 4X5 .
The quantity ca ' is the sum of all these multiplied by a constant kill .
42 .
It remains to give the value of the constants k , k VI , k " ' , which may be easily found by comparison of the above values with the constituent parts of the roots of a biquadratic ; they are as follows : k{c(o 1 60 Q4 k= -V -J\(w 5 ) / Q24 .
33 k " ' 1 X1 X2 + X3 + X4 + XS and the term uninfluenced by surds is X ' 43 .
All the constituent parts except the last-mentioned vanish when all the roots ale equal , but for their separate evanescence the condition for the equality of roots is insufficient , the requisite conditions are easily seen from the factors of such parts already given ; they are of two classes , arising from the different relations of the couples { ( a , W2 ) , ( om , a3 ) , ( ,3 , G4 ) , ( w3 , w4 ) } , and of the two couples { ( W2 , W3 ) , ( S , W4 ) } , each member of which contains the same imaginary part with the other , which is not the case with the former couples .
44 .
1 cannot at present , from thepressure of other engagements , pursue these investigations further ; if any additional light to the analyst is furnished by the preceding imperfect reflections on a subject so often treated , the author 's object will in a great measure be attained ; they at least tend to show the imperfection of our MDcccxxxVHT 2 A knowledge with respect to the conditions resulting from elimination , where more than two equations are concerned , and they exhibit in the higher powers relations between pairs of roots which have not been as yet expressed , even by the differential calculus .
London , April 1 , 1837 .

