Memoir on Cubic Surfaces .
By Professor CAYLEY , F..S .
Received November 12 , 1868 , -Read January 14 , 1869 .
THE present Memoir is based upon , and is in a measure supplementary to that by Professor SCHLXPLI , " On the Distribution of Surfaces of the Third Order into Species , in reference to the presence or absence of Singular Points , and the reality of their Lines , " Phil. Trans. vol. cliii .
( 1863 ) pp. 193-241 .
But the object of the Memoir is different .
I disregard altogethler the ultimate division depending on the reality of the lines , attending only to the division into ( twenty-two , or as I prefer to reckon it ) twenty-three cases depending on the nature of the singularities .
And I attend to the question very much on account of the light to be obtained in reference to the theory of Reciprocal Surfaces .
The memoir referred to furnishes in fact a store of materials for this purpose , inasmuch as it gives ( partially or completely developed ) the equations in plane-coordinates of the several cases of cubic surfaces , or , what is the same thing , the equations in point-coordinates of the several surfaces ( orders 12 to 3 ) reciprocal to these repectively .
I found by examination of the several cases , that an extension was required of Dr. SALMON'S theory of Reciprocal Surfaces in order to make it .
applicable to the present subject ; and the preceding " Memoir on the Theory of Reciprocal Surfaces " was written in connexion with these investigations on Cubic Surfaces .
The latter part of the Memoir is divided into sections headed thus:- " Section 1=12 , equation ( X , Y , Z , W)3=0 " &amp;c. referring to the several cases of the cubic surface ; but the paragraphs are numbered continuously through the Memoir .
The twenty-three Cases of Cubic Sutrfaces-Explanations and Table of Singularities .
Article Nos. I to 13 .
1 .
I designate as follows the twenty-three cases of cubic surfaces , adding to each of them its equation : I =12 , ( X , Y , W)3-0 , II =12-C2 , W(a , b , c , f , g , hXX , Y , Z)2+2kXYZ=O , III =12-B3 , 2W(X+Y+Z)(lX+rY+nZ ) + 2kXYZ=O0 , IV =12-2C , , WXZ+Y2(rZ+tW)+( , b , c , djX , Y)3=0 , V =12-B4 , WXZ +(X+Z)(Y2-aX2-bZ2)=0 , VI =12-B3-C2 , WXZ+Y2Z+(a , b , c , dXCX , Y)3=0 , VII=12-B , , WXZ +Y2Z +YX -Z = 0 , =12--3C , , Y'+Y2(X+Z+W)+4aXZW=O , IX =12-2B3 , WXZ+(a , b , c , dIX , Y ) =0 , X =12-B , --C , WXZ+(X+Z)(Y2-X2)-O , XI =12-B6 , WXZ+Y2Z+XZ3-0 , XII =12-U6 , W(X+Y+Z)2+XYZ=0 , XIII =12-B3-2C2 , , WXZ+Y2(X+Y+Z)=0 , XIV =12-B , C , WXZ+Y2Z+YX2=0 , XV =12-U7 , WX2+XZ2+Y2Z=0 , XVI 124C , , W(XY+XZ+YZ)+XYZ=0 , XVII =12-2B3-C0 , WXZ+XY2+y3 =0 , XVIII =12-B , -2C2 , WXZ+(X+Z)Y2=0 , XIX =12B6-C2 , WXZ+Y +X3 =0 , XX =W12-U8 , WX +XZ2+Y3 =0 , XXI =12-3B3 , WXZ+Y3=O , XXII = 3 , S(1 , 1 ) , WX2+ZY2=0 , XXIII= 3 , S( , 1 ) , X(WX+YZ)+Y3= 0 ; 2 .
WVhere C2 denotes a conic-node diminishing the class by 2 ; B,3 B3 , B , , B , a biplanar node diminishing ( as the case may be ) the class by 3 , 4 , 5 , or 6 ; and U6 , U7 , U8 a uniplanar node diminishing ( as the case may be ) the class by 6 , 7 , or 8 .
The affixed explanation , which I shall usually retain in connexion withi the Roman number , shows therefore in each case what the class is , and also the singularities which cause the reduction : thus XIII =12-B3 , -2C2 indicates that there is a biplanar node , B3 , diminishing the class by 3 , and two conic-nodes , C0 , each diminishing the class by 2 ; and thus that the class is 12-3-2.2 , ==5 .
As regards the cases XXII and XXIII , these are surfaces having a nodal right line , and are consequently scrolls , each of the class 3 , viz. XXII is the scroll S(1 , 1 ) having a simple directrix right line distinct from the nodal line , and XXIII is the scroll S(1 , 1 ) having a simple directrix right line coincident with the nodal line : see as to this my " Second Memoir on Skew Surfaces , otherwise Scrolls , " Phil. Trans. vol. cliv .
( 1864 ) pp. 559-577 .
3 .
The nature of the points C , 33 , B , B , 36 , U7 , U , requires to be explained .
( =C02 ) is a conic-node , where , instead of the tangent plane , we have a proper quadric cone .
B(=B3 , B , B , or B , ) is a biplanar-node , where the quadric cone becomes a plane-pair ( two distinct planes ) : the two planes are called the biplanes , and their line of intersection is the edge : In B3 , the edge is not a line on the surface-in the other cases it is ; this implies that the surface is touched along the edge by a plane , viz. in B4 , B. the edge is torsal , in B , it is oscular : 232 In B4 , the tangent plane is distinct from each of the biplanes : In B , , the tangent plane coincides with one of the biplanes ; we have thus an ordinary biplane , and a torsal biplane : In B6 , the tangent plane coinciding with one of the biplanes becomes oscular ; we have thus an ordinary biplane , and an oscular biplane .
U(=U6 , U7 or U8 ) is a uniplanar-node , where the quadric cone becomes a coincident plane-pair ; say , the plane is the uniplane .
It is to be observed that there is not in this case any edge .
The uniplane meets the cubic surface in three lines , or say " rays , " passing through the uniplanar-node , viz. In U6 , the rays are three distinct lines : In U7 , two of them coincide : In U8 , they all three coincide .
4 .
To connect these singular points with the theory of the preceding Memoir , it is to be observed that they aie respectively equivalent to a certain number of the cnicnodes C(=C2 ) and binodes B(=B , ) , viz. we have C2== C , B3= B , B4=2C , B , -= C+ B , JB6=3C , 1U6= 3C , U=72C+ B , U8= C+2B .
5 .
I take the opportunity of remarking that although the expressions cnicnode and binode properly refer to the simple singularities C and B , yet as C , =C , C2 is properly spoken of as a cnicnode , and we may ( using the term binode as an abbreviation for biplanar-node ) speak of any of the singularities B , , B4 , B , , B6 as a binode .
Thus the surface X-=12-B4--C has a binode B4 and a cnicnode C2 ; although theoretically the binode B4 is equivalent to two cnicnodes , and the surface belongs to those with three cnicnodes , or for which C-3 .
I use also the expression unode for shortness , instead of uniplanar-node , to denote any of the singularities U6 , U7 , Us .
6 .
The foregoing equations ( substantially the same as SCHLXFLI'S ) are canonical forms ; the reduction of the equation of any case of surface to the above form is not always obvious .
It would appear that each equation is from its simplicity in the form best adapted to the separate discussion of the surface to which it belongs ; there is the disadvantage that the equations do not always ( when from the geometrical connexion of the surfaces they ought to do so ) lead the one to the other ; for instance , V=12-B4 incldes VII=12 --B5 , but we cannot from the equation WXZ+ ( X+Z)(Y-aX2X2 bZ2)= of the former pass to the equation WXZ+Y2Z+YXX-X3=0 of the latter .
This would be a serious imperfection if the object were to form a theory of the quaternary function SUirFAES. .
( X , Y , Z , W)3 ; but the equations are in the present Memoir used only as means to an end , the establishment of the .
geometrical theory of the surfaces to which they respectively belong , and the imperfection is not material .
7 .
IE have ; used .
the capital letters .
( X , Y , Z , W ) in place of SC11LXFLI'S ( X , Y , Z , w ) , reserving these in place of his ( p , q , r , s ) for plane-coordinates of the cubic , surfaces , or ( what is the same thing ) point-coordinates of the reciprocal surfaces ; but I have in several cases interchanged the coordinates ( X , Y , Z , W ) so that they do not im this order correspond to SCHLiPrLI'S ( x , y , , w ) : this has been done so as to obtain a greater uniformity in the representation of the surfaces .
To explain this , let .
A , B , C , D be the vertices of the tetrahedron formed by the coordinate ; planes .
A=-YZW , B=ZWX , C=WXY , D=XYZ ; the coordinate planes have been chosen so .
that determinate vertices of the tetrahedron shall correspond to determinate singularities of the surface .
8 .
Consider first thel surfaces which have no nodes B. orU .
It is clear that the nodes C(2 might have been .
taken at any vertices whatever of the ; tetrahedron ; they are taken thus : there is always a node C2 at D ; when there is a , second node C2 , this is .
at C , the , third one is at A , and the fourth at B. 9 .
Consider next the surfaces which have a binode B , , B4 , B5 , or B6 ; this is taken to be at D , and the biplanes to be X=O , Z=0* ( the edge being therefore IDB ) , viz. in B5 or B6 , where the distinction arises , X= 0 is the ordinary biplane , Z-=0 the torsal or ( as the case may be ) oscular biplane .
If there is a second node , this of necessity lies in an ordinary biplane ; it may be and is taken to be in the biplane X=0 , at C. I suppose for a moment that this is a node C2 .
It is only when the binode is B3 or B4 that there can be a third node , for it is only in these cases that there is a second ordinary biplane Z-0 ; but in these cases respectively the third node , a C2 , may be and is taken to be in the biplane Z= 0 , at A. 10. .
The only case of two binodes is when each is a B3 .
Here the first is as above at D , its biplanes being X=0O , Z-=0 ; and the second is as above in the biplane X=O , at C ; the , biplanes thereof are then X0 ( which is thus a biplane common to the two binodes , or say a common biplane ) , and a remaining biplane which may be and is taken to be W= 0 .
If there is a third node , this may be either C2 or B3 , but it will in either case lie in the biplane Z=0 of the first binode , and also in the biplane W=0 of the second binode , that is , in the line BA ; and it may be and is taken to be at A ; if a binode , then its biplanes are of necessity Z==0 , W=0 ; and the plane Y=O will be the plane through the three binodes D , C , A. 11 .
If there is a unode , then this may be and is taken to be at D , and its uniplane may be taken to be X=0 ; in the surface XII= 12-U6 the uniplane is , however , taken ' to be X+Y+Z=O0 .
There is never , besides the unode , any other node .
12 .
The result is that the nodes , in the order of their speciality , are in the equations taken to be at D , C , A , B respectively ; and that ( except in the case III -12 -B3 ) the biplanes of the first binode are X= 0 , Z= 0 ( for a binode B5 or B6 , X= 0 being the ordinary biplane , .
Z=0 .
the ; special biplane ) , those of the second binode X=0 , W=0O , Explanation in regard to the determination of the number of certain singularities .
Article Nos. 14 to 19 .
14 .
In the several cases I to XXI , we have a cubic surface ( n= 3 ) , with singular points C and B but without singular lines .
The section by an arbitrary plane is thus a curve , order n= 3 , that is , a cubic curve , without nodes or cusps , and therefore of the class a'= 6 , having _'=0 double tangents and -'=9 inflexions .
The tangent cone with an arbitrary point as vertex is a cone of the order a=6 , having in the case I=12 , 5=0 nodal lines and z=6 cuspidal lines , but with ( in the several other cases ) C nodal lines and B cuspidal lines ( or rather singular lines tantamount to C double lines and B cuspidal lines ) : the class of the cone , or order of the reciprocal surface , is thus '=6.5-2(0+ C)-3(6+B)=12-2B-3C .
15 .
In the general case I=12 , there are on the cubic surface 27 lines , lying by 3 's in 45 planes ; these 27 lines constitute the node-couple curve of the order -'=27 , and the node-couple torse consists of the pencils of planes through these lines respectively , being thus of the class '= 6'=27 ; the 45 planes are triple tangent planes of the node-couple torse , which has thus t'=45 triple tangent planes .
But in the other cases it is only certain of the 27 lines , say the " facultative lines " ( as will be explained ) , which constitute the node-couple curve of the order e ' : the pencils of planes through these lines constitute the node-couple torse of the class b'= ' ; the t ' planes , each containing three facultative lines , are the triple tangent planes of the node-couple torse .
Or if ( as is somewhat more convenient ) we refer the numbers b ' , t ' to the reciprocal surface , then the lines , reciprocals of the facultative lines , constitute the nodal curve of the order b ' ; and the points t ' , each containing three of these lines , are the triple points of the nodal curve .
Inasmuch as the nodal curve consists of right lines , the number kI of its apparent double points is given by the formula 2k'=b '2-61 ' ; and comparing with the formula '=6'Y2-b'-2k'-3y'-6t ' , we have q'+3y'=0 , that is , '=-0 ( q ' the class of the nodal curve ) , and also y'= 0 .
16 .
In the general case I=12 , the spinode curve is the complete intersection of the cubic surface by the Hessian surface of the order 4 , and it is thus of the order do=12 ; but in the other cases the complete intersection consists of the spinode curve together with certain right lines not belonging to the curve , and the spinode curve is of an order a ' less than 12 : this will be further explained , and the reduction accounted for ( see post , Nos. 24 et seq. ) .
17 .
Again , in the general case 1=12 , each of the 27 lines is a double tangent of the spinode curve , and the tangent planes of the surface at the points of contact are common tangent planes of the spinode torse and the node-couple torse , stationary planes of the spinode torse ; or we have 3P=2'-=54 .
In the other cases , however , instead of the 27 lines we must take only the facultative lines , each of which is or is not a double or a sine tangent of the spinode curve ; and the tangent planes of the surface at the points of contact are the common tangent planes as above-that is , the number of contacts gives 3 ' , not in general=2 ?
' .
236 There are not , except as above , any common tangent planes of the two torses , that is , not only yr=0 as already mentioned , but also i'=0 .
I do not at present account a priori for the values O'=16 , 8 , and 16 , which present themselves in the Table .
The cubic surface cannot have a plane of conic contact , and we have thus in every case C=0 ; but the value of B ' is not in every case = 0 .
19 .
In what precedes we see how a discussion of the equation of the cubic surface should in the several cases respectively lead to the values b ' , t ' , ' .
, Y ' , flj ' , % ' , B ' , and how in the reciprocal surface the nodal curve of the order b ' is known by means of the facultative lines of the original cubic surface .
The cuspidal curve c ' might also be obtained as the reciprocal of the spinode-torse ; but this would in general be a laborious process , and it is the less necessary , inasmuch as the equation of the reciprocal surface is in each case obtained in a form putting in evidence the cuspidal curve .
The Lines and Planes of a Cubic Surface ; Facultative Lines ; Explanation of Diagrams .
Article Nos. 20 to 23 .
20 .
In the general surface 1=12 , we have 27 lines and 45 triple-tangent planes , or say simply , planes : through each line pass 5 planes , in each plane lie 3 lines .
For the surfaces II to XXI ( the present considerations do not of course apply to the Scrolls ) several of the lines come to coincide with each other , and several of the planes also come to coincide with each other ; but the number of the lines is always reckoned as 27 , and that of the planes as 45 .
If we attend to the distinct lines and the distinct planes , each line has a multiplicity , and the sum of these is= 27 ; and so each plane has a multiplicity , and the sum of these is=45 .
Again , attending to a particular line in a particular plane , the line has a frequency 1 , 2 , or 3 , that is , it represents 1 , 2 , or 3 of the 3 lines in the plane ( this is in fact the distinction of a scrolar , torsal , or oscular line ) ; and similarly , the plane has a frequency 1 , 2 , 3 , 4 , or 5 , according to the number which it represents of the 5 planes through the line .
It requires only a little consideration to perceive that the multiplicity of the plane into its frequency in regard to the line is equal to the multiplicity of the line into its frequency in regard to the plane .
Observe , further , that if M be the multiplicity of the plane , then , considering it in regard to the lines contained therein , we get the products ( M , M , M ) , ( 2M , M ) , or 3M , according as the three lines are or are not distinct , but that the sum of the products is always= 3M , and that in regard to all the planes the total sum is 3X 45 , = 135 .
And so if M ' be the multiplicity of the line , then , considering it in regard to the planes which pass through it , we get the products ( M ' , M ' , M ' , M ' , M ' ) , ( 2M ' , M ' , M ' , M ' ) ... ( 5M ' ) , as the case may be , but that the sum of the products is =5M ' , and that in regard to all the lines the sum is 5X 27 , =135 , as before .
21 .
The mode of coincidence of the lines and planes , and the several distinct lines and planes which are situate in or pass through the several distinct planes and lines respectively , are shown in the annexed diagrams I to XXI * : the multiplicity of each line iCAYLEY OiN CUBIC $SURFACES .
appears by the upper marginal line , and that of each plane by the left-hand marginal column ( thus in diagram I , 27 X 1==27 and 45 X 1=45 , 1 is te mutipliciy ofeach line ; and it is also faclethe mltiplici eachpne ) ; the frequencies of a line and plane in egard to each other appear by the dots in the square opposite to the line and plane in question , these being read , for the frequency of the line vertically , and for the frequency of the plane horizontally ; thus * indicates that the frequency of the line is =3 , and the frequency of the plane is =2 .
There should be and are in every line of the diagram 3 dots , and in every column of the diagram 5 dots ( a symbol m being read as just explained , 2 dots in the line , 3 dots in the column ) .
22 .
For the urface 1eI=12 , there is of course no distinction between the lines , but these form only a single class , and the like for the planes ; but for the other surfaces the lines and planes form separate classes , as shown in the diagrams by the lower marginal explanation of the lines , and the right-hand marginal explanation of the planes .
I use here and elsewhere " ray " to denote a line passing through a single node ; " axis " to denote a line joining two nodes edgeas above ) to denote the edge of a binode ; any other line is a " mere line .
" An axis is always torsal or oscular- ; when it is torsal , the plane touching along the axis ontains a third lie which is the " transversal " of such axis ; but a transversal may be a mere line , a ray , or an axis ; in the case XVI= 124C2 , each transversal is a transversal in regard to two axes .
23 .
In the general case I=12 , each of the 27 lines is , as already mentioned , part of the node-couple curve ; and the node-couple curve -is made up of the 27 lines , and is thus a curve of the order 27 .
In fact each plane through a line meets the cubic surface in this line , and in a -conic ; the line and conic meet in two points , and the plane ( that is in any plane ) through the line is thus a double tangent plane touching the surface at the two points in question ; the locus of the points of contact , -that is the line itself , is thus part of the node-couple curve .
But in the other cases , II to XXI , certain of the lines do not belong to the node-couple curve ( this will be examined in detail in the several cases respectively ) ; but I wish to show herehow in a general way a line passing through a node , say a nodal ray , is not part of the node-couple curve .
To fix the ideas , consider the surface II=12-C2 ; there are here through C , six lines , or say rays : attending to any one of these , a plane through the ray meets the surface in the ray itself and in a conic ; the ray and the conic meet as before in two points , one of them being the point 0 : : the plane touches the surface at the other point , but it does not touch the surface at C2 .
( I am not sure , and I leave it an open question , whether we ought to say that at a node C2 there is no tangent plane , or to say that only the tangent planes of the nodal cone are tangent planes of the surface ; but , at any rate , an arbitrary plane through C2 is not a tangent plane .
) The plane through the ray is only a single tangent plane , not a double tangent plane ; and the ray is not part of the node-couple curve .
We say that a line of the surface is or is not " facultative " according as it does or does not form part of the node-couple curve .
238 Ais ; the -ifferent kinds thereof.f Article Nos. , 24 to , 26 .
24 .
A line joining two nodes is anaxis ; such a lineis always a line , and it is a torsal or oscular line , of the surface. .
But some further distinctions are requisite ; ; using the expressions in their strict sense , cnicnode =C , binode =B , an axis is a CC-axis joining two cnicnodes , or it is a CBR-axis joining a cnicnode and a binode , or it is a BB-axis joining two binodes .
A CC-axis is torsal , the transversal being a mere line , not a ray through either of the cnicnodes ; a CB-axis .
is torsal , the transversal being a ray of the binode ; a BB-axis is oscular .
The distinction is of course carried through as regards the higher biplanar nodes B4 , B5 , B6 , and the uniplanar nodes U6 , U7 , , U8 : thus ( B-=B)the edge of a binode B5 is not an axis at all , but ( B4=2C ) the edge of a binode B , is a CC-axis ; ( B5=B+C ) the edge of a binode B 5is a CB-axis ; ( B6=3C ) the edge of a axes-viz .
the examination of the several cases shows that in the complete intersection each CC-axis presents itself twice , each CB-axis 3 times , and each BB-axis 4 times .
We thus see that a CC-axis , or rather the torsal plane along such axis , is the pinch-plane or singularity j ' ; the CB-axis , or rather the torsal plane along such axis , the closeplane or singularity z'=1 ; and the BB-axis , or oscular plane along such axis , the bitrope or singularity B'=l ; for a cubic surfacintersecin e with singular lines the expression of being in fact '=12 2j 3 --413 ' .
There are , however , some cases requiring explanation ; complete intersection is made up of the edge4 times an ofeean octic curve ; the conideration of the examinaton of the s hows , however , thate edge taken once is really part of the spinode curve ( viz that twicehis curve is made up of the edge taken once and of the the spinode curve .
26 .
singula remark in furthe x ithhe x o , ththe torsal plane veral sections , in showing the close complete intersection of th cubic surface with thesingular liHes sian is made up , I have not in referred to the axes in the above precise significations ; thucases requiring explanationhe binode B ca is C+VIII-12-B , and the edge is thus a CB-axis , while the axis B5C2 is a CB-axis omp a CC'axis ( ection+1 i 2 , '=ade u. The complete i ntersection should t herefore consist of the spinode curve , ( viz. edge ( as a curve -axis ) 3 times edge axis ( as a B-axis of a CC-axis ) 2curve3 , =5 times it is in the section stated ( in perfect consistency herewith , but wersechout the full explanation ) that the intersection is made up of the axis 5 times , the edge 4 times , and a cubic curve-which cubic curve together with the edge once constitutes the spinode curve ; and so in other cases:veral this explanation will , I think , rem ove all difficulty .
remove all difficulVty..239 On the Determination of the Beci2rocal Equation .
Article Nos. 27 to 32 .
27 .
Consider in general the cubic surface ( *XX , Y , Z , W)3=0 , and in connexion therewith the equation X +Yy+Zz+Ww= O , which regarding therein X , Y , Z , Was current coordinates , and x , y , z , w as constants , is the equation of a plane .
If from the two equations we eliminate one of the coordinates , for instance W , we obtain ( *Xw , Yw , Zw , -(Xx+Yy+Zz))3=0 , which , ( X , Y , Z ) being current coordinates,7is obviously the equation of the cone , vertex ( X=O , Y-0 , Z=0 ) , which stands on the section of the cubic surface by the plane .
Equating to zero the discriminant of this function in regard to ( X , Y , Z ) , we express that the cone has a nodal line ; that is , that the section has a node , or , what is the same thing , that the plane xX+yY+zZ+wW=O is a tangent plane of the cubic surface ; and we thus by the process in fact obtain the equation of the cubic surface in the reciprocal or plane coordinates ( x , y , z , w ) .
Consider in the same equation x , y , z , w as current coordinates , ( X , Y , Z ) as given parameters , the equation represents a system of three planes , viz. these are the planes xX+yY+zZ+wW'=O , where W ' has the three values given by the equation ( *XX , Y , Z , W)3= 0 , or , what is the same thing , X , Y , Z , W ' are the coordinates of any one of the three points of intersection of the cubic surface xyz by the line =-Y=Z ; ( X , Y , Z , W ' ) belongs to a point on the surface , and xX+yY+zZ+wW'=0 is the polar plane of this point in regard to a quadric surface X2+Y2+Z2+W2+ 0 ; the equation ( *Xw , Yw , Zw , -(Xx+Yy+zz))3=0 is thus the equation of a system of 3 planes , the polar planes of three points of the cubic surface ( which three points lie on an arbitrary line through the point x=O , y=O , z=0 ) .
In equating to zero the discriminant in regard to ( X , Y , Z ) , we find the envelope of the system of three planes , or say of a plane , the polar plane of an arbitrary point on the cubic surface , -or we have the equation of the reciprocal surface , being , as is known , the same thing as the equation of the cubic surface in the reciprocal or plane coordinates ( x , y , z , w ) .
In what precedes we have the explanation of an ordinary process of finding the equation of the reciprocal surface , this equation being thereby given by equating to zero the discriminant of a function ( *-(X , Y , Z)3 , that is , of a ternary cubic function .
28 .
The process , as last explained , is a special one , viz. the position of a point on the surface is determined by means of certain two parameters , the ratios X : Y : Z which fix the position of the line joining this point with the point ( x=0 , y=O , z==O ) .
More generally we may consider the position of the point as determined by means of any two parameters ; the equation of the polar plane then contains the two parameters , and by taking the envelope in regard to the two parameters considered as variable , we have the equation of the reciprocal surface .
29 .
But let the parameters , say 0 , p , be regarded as varying successively ; if p alone 240 vary , we have on the surface a curve 0 , the equation whereof contains the parameter 0 , and when Q varies this curve sweeps over the surface .
The envelope in regard to p of the polar plane of a point of the surface is a torse , the reciprocal of the curve 0 , and the envelope of the torse is the reciprocal surface .
In particular the curve 0 may be the plane section by any plane through a fixed line , say , by the plane PQ=O0 ; the section is a cubic curve , the reciprocal is a sextic cone having its vertex in a fixed line ( the reciprocal of the line P= 0 , Q= 0 ) , and the reciprocal surface is thus obtained as the envelope of this cone ; assuming that the equation of the sextic cone has been obtained , this is an equation of a certain order in the parameter 0 ; or writing a=P : Q , we obtain the equation of the reciprocal surface by equating to zero the discriminant of a binary function of ( P , Q ) .
30 .
With a variation , this process is a convenient one for obtaining the reciprocal of a cubic surface : we take the fixed line to be one of the lines on the cubic surface ; the curve 0 is then a conic , its reciprocal is a quadricone , and the envelope of this quadricone is the required reciprocal surface .
This is really what SCHLFiFLI does ( but the process is not explained ) in the several instances in which he obtains the equation of the reciprocal surface by means of a binary function .
I remark that it would be very instructive , for each case of surface , to take the variable plane successively through the several kinds of lines on the particular surface ; the equation of the reciprocal surface would thus be obtained under different forms , putting in evidence the relation to the reciprocal surface of the fixed line made use of .
But this is an investigation which I do not enter upon : I adopt in each case SCHLAFLI'S process , without explanation , and merely write down the ternary or ( as the case may be ) binary function by means of which the equation of the reciprocal surface is obtained .
31 .
It is to be mentioned that there is a reciprocal process of obtaining the equation of the reciprocal surface ; we may imagine , touching the cubic surface along any curve , a series of planes ; that is , a torse circumscribed about the surface , ad the equation whereof contains a variable parameter d ; the reciprocal figure is a curve , the equations whereof contain the parameter 0 ; the locus of this curve is the reciprocal surface ; that is , the equation of the reciprocal surface is obtained by eliminating d from the equations of the curve .
In particular let the torse be the circumscribed cone having its vertex at any point of a fixed line ; the reciprocal figure is then a plane curve , the plane of which passes through the line which is the reciprocal of the fixed line ; it is moreover clear that if the position of the vertex on the fixed line be determined by the parameter 0 lisnearly ( for instance if the vertex be given as the intersection of the fixed line by a plane PQ=0 ) , then the equation of the plane of the curve will be of the form P'= Q ' , containing the parameter 0 linearly ; the other equation of the plane curve will contain 0 rationally , and the elimination will be at once effected by substituting in this other equation for 0 its value , P ' Q ' .
And observe moreover that if the fixed line be a line on the cubic surface , then the cone is a quadricone having for its reciprocal a conic ; the reciprocal surface is thus given as the locus of a variable conic , the plane of which always passes through a fixed line ; there are thus on the reciprocal surface series of such conics .
It would be very instructive and interesting to carry out the investigation in detail .
32 .
The equation of the reciprocal surface is found by equating to zero the discriminant of a ternary or a binary function* , viz. this is a ternary cubic , or a binary quartic , cubic , or quadric .
The equation as given in the form disct .
=0 , contains a factor which for the adopted forms of equations is always a power or product of powers of w , z , xt known apriori , and which is thrown out without difficulty , the equation being thereby reduced to the proper order .
There is the singular advantage that the , process puts in evidence the cuspidal curve of the resulting reciprocal surface , viz. for a ternary cubic , the form obtained is S3-T= 0 , and for a binary quartic it is the equivalent form I3-27J2=0O ; but for the factor : thrown out as just mentioned , we should have simply ( S=O , T=0 ) , or , as the case may be , ( I=-0 , J-=0 ) for equations of the cuspidal curve ; the existence of the factor occasions however a modification , viz. the intersection of the two surfaces is not an indecomposable curve , and the cuspidal curve is in most cases , not the complete intersection , but a partial intersection of the two surfaces .
In several cases it thus happens that the cuspidal curve is obtained as a curve i p , =0 , without or with further speciality .
Similarly when the equation of the reciprocal surface is obtained by means of a binary cubic ; if the coefficients hereof ( functions of course of the coordinates x , y , z , w ) be A , B , C , D , then the surface is ( AD-BC)24(ACB2)(BDC2 ) = 0 , having the cuspidal curve B , B , D =0 , subject however to modification in the case of a thrown out factor .
Explanation as to the Sections of the Memoir .
Article Nos. 33 &amp; 34 .
33 .
As regards the following Sections I to XXIII , it is to be observed that for the general surface I=12 , I do not attempt to form the equation of the reciprocal surface , and in some of the other cases , II=_12-C2 &amp;c. , the equation of the reciprocal surface is either not obtained in a completely developed form , or it is too complicated to allow of its being dealt with , for instance so as to put in evidence the nodal curve of the surface .
Portions of the theory given in the latter sections are consequently omitted in the earlier ones , and in particular in the Section I there is given only the diagram of the 27 lines and the 45 planes ( with however developments as to notation and otherwise which have no place in the subsequent sections ) , and with the analytical expressions for the several lines and planes , although from the want of the equation of the reciprocal surface ithese -analytical expressions have no present application .
And so in some of the next following sections , no application is made of the analytical expressions of the lines and planes .
34 .
I call to mind that if a line be given as the intersection of the two planes AX+BY+CZ +DW=O0 , A'X+B'Y+C'Z+D'W= 0 then the six coordinates of the line are a , , c , ; g , h =AD'-A'D , BD'-B'D , CD'-C'D , BC'-B'C , CA'-C'A , AB'-A'B , and that in terms of its six coordinates the line is given as the common intersection of the four planes ( .
h7 , -g , a Xx , Y , Z , W)=0 , -h , .
f , 6 g , -f , 'c -I -6 , c , and that ( reciprocating as usual in regard to X2+Y2+Z2+W20 ) the coordinates of the reciprocal line are ( f , g , h , a , b , c ) ; that is , this is the common intersection of the four planes ( .
c -b , - , f , y , z , w)==0 .
-c .
a , g b , a , .
h -f , -g , -h , .
It is in some cases more convenient to consider a line as determined as the intersection of two planes rather than by means of its six coordinates ; thus , for instance , to speak of the line X=O0 , Y=O rather than of the line ( 0 , , 0 0 , 1 , 0 , 0 ) ; and in some of the sections I have preferred not to give the expressions of the six coordinates of the several lines . ?
I=12 , Equation ( X , Y , Z , W)3=0 .
Article Nos. 35 to 46 .
35 .
There is in the system of the 27 lines and the 45 planes a complicated and manysided symmetry which precludes the existence of any unique notation : the notation can only be obtained by starting from some arrangement which is not unique , but one of a system of several like arrangements .
The notation employed in my original paper " On the Simple Tangent Planes of Surfaces of the Third Order , " Camb .
and Dub .
Math. Journ. vol. iv .
1849 , pp. 118-132 , and which is shown in the right hand and lower margins of the diagram , starts from such an arrangement ; but it is so complicated that it can hardly be considered as at all putting in evidence the relations of the lines and planes ; that of Dr. HART ( SALMON " On the Triple Tangent Planes of a Surface of the Third Order , " same volume , pp. 252-260 ) , depending on an arrangement of the 27 lines according to a cube of 3 each way , is a singularly elegant one , and will be presently reproduced .
2 L2 243 But the most convenient one is SCHLAFLI'S , starting from a double-sixer ; viz. we can ( and that in 36 different ways ) select out of the 27 lines two systems each of six lines , such that no two lines of the same system intersect , but that each line of the one system intersects all but the corresponding line of the other system ; or , say , if the lines are 1 , 2 , 3 , 4 , 5 , 6 1 ' , 2 ' , 31 , 4 ' , 51 , 6 ' , then these have the thirty intersections 1 ' , 2 ' , 3 ' , 4 ' , 51 , 6 ' 2 3 .
0 .
0 4 .
5 ?
Any two lines such as 1 , 2 ' lie in a plane which may be called 12 ' ; similarly the lines 1 ' , 2 lie in a plane which may be called 1'2 ; these two planes meet in a line 12 ; and any three lines such as 12 , 34 , 56 meet in pairs , lying in a plane 12 .
34 .
56 .
We have thus the entire system of the 27 lines and 45 planes , as in effect completely explained by what has been stated , but which is exhibited in full in the diagram .
37 .
The diagram of the lines and planes is 244 It has been mentioned that the number of double-sixers was =36 , these are as follows:1 , 2 , 3 , 4 , 5 , 6 Assumed primitive .
1 1 ' , 2 ' , 3 , 41 , 51 , 6 ' 1 , 1 ' , 23 , 24 , 25 , 26 Like arrangements. .
15 2 , 2 ' , 13 , 14 , 15 , 16 1 , 2 , 3 , 56 , 46 , 45 Like arrangements .
20 23 , 13 , 12 , 4 , 5 , 6 36 where , if we take any column , of two lines , we have the complete number 216 of pairs of non-intersecting lines ( each line meets 10 lines , there are therefore 27-1-10 , =16 , which it does not meet , and the number of non-intersecting pairs is thus -27.16=216 ) .
39 .
We can out of the 45 planes select , and that in 120 ways , a trihedral-pair , that is , two triads of planes , such that the planes of the one triad , intersecting those of the other triad , give 9 of the 27 lines .
Analytically if X==0 , Y=O , Z=0 and U=0 , V=0 , W= 0 are the equations of the six planes , then the equation of the cubic surface is XYZ+kUVW=0 .
See as to thispost , No. 44 .
The trihedral plane pairs are12 ' , 23 ' , 31 ' 1'2 , 2'3 , 3'1 No. is =20 12 ' , 34 ' , 14.23.56 2'3 , 4'1 , 12.34.56 =90 14.25.36 , 35.16.24 , 26.34.15 14.35.26 , 25.16.34 , 36.24.15 =10 120 The construction of the last set is most easily effected by the diagram 123 X 456 312 564 231 645 14 25 36 35 16 24 26 34 15 It is immaterial how the two component triads 123 and 456 are arranged , we obtain always the same trihedrai pair .
246 ONc CoUBIC SURFACES .
40 .
Dr. HART arranges the 27 lines , cubically , thus : Al , Bi a1 ( bl ?
i C 'l , 7b A2 B2 02 a,2 b2 2 C2 12 P 72 A3 B3 C3 a , b3 03 C33 73 where letters of the same alphabet denote lines in the , same plane , if only the letters are the same or the suffixes the same ; thus A , , A2 , A3 lie in a plane AA2A3 ; AI , B , C , lie in a plane A1B , C , .
Letters of different alphabets denote lines which meet according to the Table a c1 b2 C3 b1 G C3 l a2 3 A , B,1 C C1 273 1Y 72 C3 7l a2 23 3 c2 a3 b6 a2 b3 cl 2 c3 c 1 , A2 B2 2 A3 Y1 2 73 1 P2 c3 1 72 63 C1 a2 C3 a1 b12 a3 b1 2 A3 B3 C3 72 ?
53 I1 2C 33 1l 2 73 c1 where the letter in the centre of the square denotes a line lying in the same plane with the lines denoted by the letters of each vertical pair in the same square .
Thus Al lies in the planes A , a , a , , AbP,32 , A3 , c37 ( and in the before-mentioned [ two planes A , A2A , , A , B1C , ) .
41 .
I find that one way in which this may be identified with the double-sixer notation is to represent the above arrangement by 1 , 2 ' , 12 3 ' , 4 , 34 13 , 24 , 56 14 , 25 , 36 2 , 6 ' , 26 1 ' , 16 , 6 4 ' , 5 , 45 23 , 46 , 15 3 , 35 , 5 ' and then the identification may apparently be effected in ( 720 X 36= ) 25920 ways , viz , we may first in any way permute the } , , 2 , , , 3 , , 4 , , 6,5 by this means not altering the double-sixer , , 2 , 34 , and then upon the arrangements so obtained make any of the substitutions which permute inter se the 36 double-sixers .
42 .
The equations of the 45 planes are obtained in my paper last referred to , viz. taking the equation of the surface to be W(l , 1 , 1 , 1 , , 1t I+lm , +m n++ + 'X , Y Z , , W)2+XZY=0 , 247 W7e have X=O 0 , 0 , Z=-0 W-=0 for the equations of the planes ( 12.34.56=- ) , ( 42'=y ) , ( 14'=z ) , ( 12'=w ) ; and representing by f= X +-Y+Z+W-W=0 the equation of any other plane ( 41'=f ) , the equation of the cubic surface may be presented in the several forms : 0=U==Wff +k-YZ =5= gg +kIZX -Whh +k~XY =WdO +k^7 -=W11+ kyzx =-Wmm+kzxy =Wnn1 +-kxyz =Wlj + -kyzx =Wmim +kizxy =Wn , n +kxyz -=Wppi +k-yz =Wqq ' +-knzx =Wrr , +kyxy =-Wppl +kyz =Wqql + klzx =Wrr , +k-xy , which are the 16 forms containing W , out of the complete system of 120 trihedral-pair forms .
45 .
The 27 lines are each of them facultative ; we have therefore b'= ' -27 ; t'=45 ; moreover each of the lines is a double tangent of the spinode curve , and therefore P ( =_ 2')=5 4 .
46 .
The equation of the reciprocal surface is not here investigated ; its form is S3 -T2=0 , where S=(x , y , z , w)4 , T=(*X ) , y , z , w)6 ; wherefore n-=12 .
The nodal curve is composed of the lines which are the reciprocals of the original 27 lines ( b'=27 , t'-45 tt su pra ' ) .
It may be remarked that the reciprocal of a doublesixer is a double-sixer .
Hence the 27 lines of the reciprocal surface may be ( and that in 36 different ways ) represented by where 12 is now the line joining the points 12 ' and 1'2 ; and so for the other lines .
T'he lines 12 , 34 , 56 meet in a point 12 .
34 .
56 ; the 30 points 12 ' , 1'2 ... 56 ' , 5'6 , and the fifteen points 12.34.56 make up the 45 points t ' .
The above equation , S3-T2=0 , shows that the cuspidal curve is a complete intersection 6 X4 ; c'= 24 .
Section II =12--2 .
Equation W(a , , c , f , g , hjX , Y , Z)2+2kXYZ0= .
Article Nos. 47 to 59 .
47 .
It may be remarked that the system of lines and planes is at once deduced from that belonging to 1=12 , by supposing that in the double-sixer the corresponding lines 1 and r , &amp;c. severally coincide ; the line 12 , instead of being given as the intersection of the planes 12 ' , 1'2 , is given as the third line in the plane 12 , which in fact represents the coincident planes 12 ' and 1'2 .
25 &amp; 5 In explanation of the discussio of the e reciprocal surface , it is convenient to remark that we have Node C2 , X= 0 , Y=- , Z=0 .
Tangent cone is ( a , 6 , c , f , .
, h)xX , Y , Z)=0 .
Nodal rays are sections of cone by planes X-0 , Y= 0 , Z=0 respectively , viz. equations of the rays are X=O , bY2+2fYZ-+c 2=0 , Y=O , c Z_a+2gZX+C-X2=0 , Z =0 , aX2+ 2hXY+-bY2= 0 .
Reciprocal plane is w=O0 .
Conic of contact is ( A , B , C , F , G , Hx , y , z)2=0 , o t0 .
Lines are tangents of this conic from points y-0 , = 0 ) ( y 0 , z =0 ) , ( =-0 , y=0 ) respectively , viz. , equations are w=0 , cfy--2fyz + 6z2= , tw=0 , az2--2gzx +c z 0 , w=O , 6x'-2hxy+-ayz=O .
57 .
The equation shows that the section by the plane w=0 is made up of the conic ( A , B1 , C , F , G , PI( , y , z)2=0 twice , and of the six lines , tangent to thisconic , viz. the lines wzO , CY-2f-yz +b62= 0 , w=O , az2 -2gzx +cx2=0 , w=O , bx--2hxy +ay-2=0 , each once ; the lines in question ( reciprocals of the nodal rays ) are thus mere scrolar lines on the reciprocal surface .
58 .
I do not attempt to put in evidence the nodal curve of the surface ; by what precedes it is made up of 15 lines , intersecting 3 together in 15 points ; and if we denote the six tangents of the conic just referred to be 1 , 2 , 3 , 4 , 5 , 6 , then the fifteen lines are respectively lines passing through the intersections of each pair of these tangents ; viz. through the intersection of the tangents 1 and 2 , we have a line 12 ; and so in other cases ; that is , the 15 lines are 12 , 13 ... .
56 .
The lines 12 and 34 meet ; and the lines 12 , 34 , 56 meet in a point ; we have thus the 15 points 12 .
34 .
56 , triple points of the nodal curve .
59 .
As regards the cuspidal curve , the equation of the surface may be written ( L'-12kwM)(4M2+ 3LN)-(LM+ 9wN)2 3(L2M2 + LN18kwLMN--1 6kwM327kw22N 2)-0 , and we thus have or , what is the same thing , 4M2+ 3LN =0 , LM + 9kwN=0 , L2 -12kwM-0 , L , 12M , -9N =0 kw , L , M 262 In the discussion of the equation it is convenient to write down the relations of the two surfaces , thus : Cubic surface .
Reciprocal surface .
B3 , X=0 , Y=0 , Z=0 Plane w O , Biplanes X+ Y+ Z=0 Points in w=0 , viz. ZX+mY+nZ-=0 , x=y=z and x : : y = : z= : n : , intersecting in edge .
in line ( mn-n)x+(n--l)y+(1--m)z= , O that is , xx+p-y+-vz= 0 , or -=0 .
Pays in first biplane , Lines in plane w=O , and through first X=O , Y+Z=O ; Y O , Z+X=0 , point , viz. Z=0 , X+Y=0 ; y--=0 , z-x-=0 , x-y-0 ; rays in second biplane , lines through second point , viz. X=-0 , mnY+nZ-=0 ; Y=0 , nZ+ X= ny--= 0 , nZ---Z=0 , xt-my==0 .
Z=0 , IX+mY=0 .
70 .
The equation puts in evidence the section by the plane w=0 , viz. this is the line =0 ( reciprocal of the edge ) three times , and the six lines ( reciprocals of the rays ) each once .
Observe that the edge is not a line on the cubic ; but its reciprocal is a line , and that an oscular line on the reciprocal surface ; the six lines ( reciprocals of the rays ) are mere scrolar lines on the reciprocal surface ; they pass , three of them , through the point x=y=z , and the other three through the point x : y : z=l : in : 9n ; that is , they are six tangents of the point-pair ( reciprocal of the pair of biplanes ) formed by these two points .
71 .
I do not attempt to put in evidence the nodal curve on the surface ; by what precedes it consists of 9 lines , reciprocals of the mere lines .
If we denote by 1 , 2 , 3 and 4 , 5 , 6 the lines which pass through the points x-=0 , y=O , z=0 and through the point xy : z= : m : n respectively , then these intersect in the nine points 14 , 15 , 16 , 24 , 25 , 26 , 34 , 35 , 36 ; and through each of these there passes a nodal line which may be represented by the same symbol ; that is , we have the nodal lines 14 , ... .
36 .
Two lines such as 14 , 25 meet ; and three lines such as 14 , 25 , 36 meet in a point ; we have thus the six points 14 .
25 .
36 &amp;c. triple points on the nodal curve ; as before , b'=9 , t'= 6 .
72 .
The cuspidal curve is given by the equations k2W-2kwt+o-2 , 24(k2w+o- ) , 36 ( 4lmnJc xyzw z2 i=0 .
Jco , k2v22wt + o , 2(/ wz.+ 04 ) Writing down the two equations , ( ko2w 2kwt + o ) ' 24kw(lwo + a+ ) 0 , ( k22 2kwt+ a2)(kw + c ) + 18w( lmnkJ xyzw t 42 ) i 0 , these are respectively of the orders 4 and 5 ; but they intersect in the line w=0 , o=0 taken four times , or say , the cuspidal curve is a partial intersection 405-4 ; c'=16 .
268 Combining with the foregoing the equation of the surface XZW+Y2(yZ+WW)+(a , b , c , dJX , Y)3-0 , it appears that these have along the line X=O , Y=O the common tangent plane X=O , or , what is the same thing , that they meet in the line X=0 , Y=0 ( the axis ) twice , and in a residual curve of the tenth order , which is the spinode curve ; the equations may be presented in the somewhat more simple form XZW+Y2(Z+YW)+C(a , , c , d c , Y)7 , =0 , 4y7Y2ZW--4(yZ+ W)(a , b , c , dXX , Y)3+12yXY2(CaX+bY ) +X4(--12ac 9b2)-3 d(4aXY+ 6 bX2Y2 4cXY + dY4 ) 0 , which , however , still contain the line X=0 , Y=-0 twice .
The spinode curve , as just mentioned , is of the tenth order ; that is , we have -'=10 .
Each of the 6 mere lines is a double tangent to the spinode curve , but the transversal is only a single tangent : to show this , observe that the equations of the transversal are X=O0 , yZ+-W+B dY=0 ; substituting in the equations of the curve the first equation , that of the cubic surface is of course satisfied identically ; for the second equation , writing X-0 , this becomes Y2{ -4rySZWV--4dY(yZ +W ) -3d62Y2 } 0 ; or writing herein dY= -(yZ ?
+iW ) , it becomes Y2(yZ--_W)2=0 .
The value Y2-0 gives X=0 , , 0 yZ+ , W=0 , viz. this is a point on the axis X-0 , Y-=0 not belonging to the spinode curve ; the value ( yZSW)2=0 gives a point of contact X=00 , yZZ+-W+dY==0 , yZ--TW=0 ; and the transversal is thus a single tangent .
Hence the number of contacts is 2.6+1 , =13 ; that is , we have P'=13 .
Reciprocal Surface .
80 .
The equation is found by equating to zero the discriminant of the binary quartic { xX2 +yXY-(Z + yw)Y2 } +4Zw{X(a , b , c , d(X , Y ) 3- ?
74 } ) , or say this is ( *30X , Y)4 , where the coefficients are 6x2 +24Cazw , 3xy +18bzw , y2 ( z + yw)x+2czw , 3(+z + yw)y +6 dzw , 6(z-yw)2 .
81 .
Forming the invariants , these are I= A2+ 24Uzw + 144pz2w2 -J= =A + 36AUzw +216Vz2w2+ 86 4yzw3 272 being , it will be observed , functions of x , y , z+yrw .
The transformed equation is A2(rA2^ AV1 , +U+ ) + w= O , where the term Q may be calculated without difficulty : the first term of this is = { y2+4( z+yw)x}2 .
4ry2b2[ fiy--f(dz+ 7w)]. .
[ X+f4y-f ( z +yW ) ] , the developed expressions of - ; ( A2AV1 + U2 ) and of y282 into the product of the linear factors being in fact each x 4 .
y2 +Xy .
die +x2y2 .
-+X3 .
3 +yt 9 .
-a y +[ d(-d2_-6cyr ) +2y(3cd + 9b3y ) + Xy2( 3d 4ay ) +fy ad](z ++yvw ) + [ x2(9c2-6bd-2ary ) +xy(33ad9bc)-+y2 .
3ac](6z + yw)2 + [ ( 6c92)+y .
3ab](]z+yw)3 Ca24 .
( .z + yw)4 .
The form puts in evidence the section by the plane w= 0 , which is the reciprocal of the node D , viz , this is a conic ( the reciprocal of the tangent cone ) twice , and four lines , the reciprocals of the nodal rays , each once .
And similarly for the section by the plane 83 .
The nodal curve is made up of the lines which are the reciprocals of the six mere lines and the transversal ; viz. we have three pairs of lines and a seventh line , the lines of each pair intersecting at a point of the seventh line , and these three points being the triple points of the nodal curve ; t'=3 as before .
84 .
The equations of the cuspidal curve are at once reduced to the form A2 +A24Uzw +144z2w-2=0 , AU +(18V-12p4A)zw +72z2W2=-0 , which are two quartic surfaces hsaving in common the conics z-=0 , A=0 , and w=0 , A =0 ; or we may say that the cuspidal curve is a curve 4 .
4-2-2 ; that is c'=12 .
274 And the lines are abc f ' Yh equations may be written __ _ .
__ _ ... _0 1/ b J'-/ a 2( Va-'/ Vb ) 001100 -2 2(-/ Va-Vb ) -2 21( V-+V ' ) -2 2(-/ -+-/ +b ) -2 ( 3 ) ( 4 ) ( 1 ) ( 2 ) ( 1 ' ) ( 2 ' ) ( 11 ' ) ( 12 ' ) X-=o , Z=0 X+z=o , W-0 x=0 , Y-ZV/ b=O x=o , Y+Zv=o O=0 , -Xv/ a+Y=O Z=- , X/ a+Y-=0 but for the other lines the coordinate expressions are the more convenient .
( 21 ' ) ( 22 ' ) 88 .
The four mere lines facltaCtive , counting twice ; and the transversal are each facultative ; the edge is also '=-b'=7 , t'=3 .
g That the edge is as stated a facultative line counting twice , I discovered , and accept , a posteriori , from the circumstance that on the reciprocal surface the reciprocal of the edge is ( as will be shown ) a tacnodal line , that is , a double line with coincident tangent planes , counting twice as a nodal line .
Reverting to the cubic surface , I notice that the section by an arbitrary plane through the edge consists of the edge and of a conic touching the edge at the biplanar point ; by what precedes it appears that the arbitrary plane is to be considered , and that twice , as a node-couple plane of the surface : I do not attempt to further explain this .
89 .
Hessian surface .
The equation is ( X+Z)XZW+(Xz)2Y2+ ( X+Z)(3a , -a , - , 3bXX , Z)3 0 .
Combining with the equation XZW7 +(+ Z)(Y2_ 2bZ2 ) O , and observing that from the two equations we deduce -XZY2+ ( X + Z)(X3 + bZ3 ) = 0 , it appears that the complete intersection of the Hessian and the surface is made up of the line X=0 , Z=0 ( the edge ) twice ( that is , the two surfaces touch along the edge ) , and of a curve of the tenth order , which is the spinode curve ; c'=10 .
01000 I , / 6 iI -~b , / b0000001 Vab 1 V/ ab I1I d &gt; 0100001Va1 V'a 11O __c 0000119222 276 This , completely developed , is 6 4w6 .
a(a + b)2{ ( ct + -)y(_ )2 } -32w .
2ab6 3(a+ )[(a-2b)x -+(-2a+b)z1y + ( { _-Z)2[(3a+5 )2+(5a-3b)z ] J -16w4 " 3ab(a27ab+b2)y4 + [ b(9a " 2+26cab2)x226ab(ab)xz + a(-a 2+ 26ab + 9b2)z2]y2 + ( ( -_)2[ _ 12a +b)xl 22abxz +a(12)xz2 ] + 8wf 30a6[2a-h)x+(-a+2 &amp; )z]y4 j ?
-[b(-2a+5b)x3+ b(3a-2b)x2z+a(-2a+ -3b),2+a(5a-2b)z3]y + 2(x--z)2[2bx3+bx2z+az22ay3 ] J+ 4w 2 3ab(a+z)y6 -+ [ b(9a2b)x2 +Sxza(-2+ aZ]4(-2 b+ 2[6 &amp; X4+ &amp; x3z-(a+Cb)X2z2 +acz -6i-aZ4]y9 X2+ 4'2z2( ' z)2 + 2iv 1 2ab(x + z)y -J[3b+2bx2 2z , +2axz+3az3]y4 I +4 2z2(S+z)^y J+ y4(cy2_-2)(y2_Z2)=0 , where we see that the section by the plane w= 0 ( reciprocal of B4 ) is made up of the line tz==0 , y==0 ( reciprocal of the edge ) four times , and of the lines w=0 , ac2-2t=0 ; w=O , by2-2=:0 ( reciprocals of the rays ) each once .
93 .
The surface contains the line y=0 0 , w=0 ( reciprocal of the edge ) ; and if we attend only to the terms of the lowest order in y , w , viz. x , r{16(x-z)2w+ 8(+z)y2w +y } , which terms equated to zero give we see that the line in question ( y=O , w=0 ) is a tacnodal line on the surface , the tacnodal plane being w 0 , a fixed plane for all points of the line : it has already been seen that this plane meets the surface in the line taken 4 times ; every other plane through the line meets the surface in the line taken twice .
We have in what precedes the a posteriori proof that in the cubic surface the edge is a facultative line to be counted twice .
94 .
Cuspidal curve .
The equation of the surface may be written ( L-12w2M)(4M2+ 3LN)(LM+i 9w2N)2= 0 , 278 giving the two sheets { ( 4x+12cz)2-144cxz-144bdz2}w -24dxzy=0 and w=0 ; in regard to the last-mentioned sheet the form in the vicinity thereof is given by w =Ay3 , viz. we have approximately L=y2 , M=2dxy , and thence y4-12z.Ay3 .
2dxy=0 , that is , 11 A=-4dxz or w= 4dxy3 ; the line is thus a flecnodal line on the surface L2-12zwM=O .
-24dxz Or '=24 die Next as regards the surface LM+9zwN=O ; the line y==O , v=0O is a simple line on the surface , the terms of the lowest order being 9zw(-4d2)2)=0 ; that is , we have w =0 , and for a next approximation w=Ay3 , viz. L=-y2 M=-2dxy , N=--4d2 2 , and therefore -2dXy3+9z .
Ay3(-4d222 ) = O0 , that-is , A= --l8dz or w=-y3 ; there is thus a threefold intersection with one sheet and a simple intersection with the other sheet of the surface L2--12zwM=0 .
The surfaces intersect , as has been mentioned in the conic z=0 , y2+4xw=0 ; or we have the line y=O , w==0 four times , the conic once , and a residual cuspidal curve of the order 4 .
4-4-2 , =10 ; that is , c'==10 .
Section VII = 12-B5 .
Equation WXZ+Y2Z+YX-Z3=0 .
Article Nos. 103 to 116 .
103 .
The diagram of lines and planes* is Lines .
VII='12-B .
XXX ii c en o 11 II II 11 1 ?
5t 5 Si 01 ? ?
1x15= 15 00. .
1X20= 50 ' 12 ' 0 2x 5=10 13 ' *I 34 45 * X5 ?
J _O 34 45 4~~~~ Torsal biplane .
Ordinary biplane .
Planes each containing a mere line .
The planes are The lines are Z= 0 , [ 10 ] X-O , Z=0 , ( 0 ) X0 , [ 00 ] Y=0 , Z-=0 ( 1 ) Y+Z=O , [ 12 ' ] X=0 , Y+Z=O , ( 2 ' ) Y-Z=0 , [ 13 ' ] X= , Y-Z=o , ( 3 ' ) XW=0 , Y+Z=0 , ( 12 ' ) X+W=0 , Y-Z=0 , ( 13 ' ) .
105 .
The two mere lines are facultative , and the edge is also facultative ; g'=1'=3 ; ' 0 .
106 .
Hessian surface .
The equation is Z(WXZ +Z +YX2Z3 ) 4X2Z X4+ 44Z4=0 .
The complete intersection with the surface is thus given by the equations WXZ+Y2Z +TYXZ3= 0 , 4X2YZ +X4+4Z4 =0 , which is made up of the line X=0 , Z=0 ( the edge ) four times and a curve of the eighth order .
To see this , observe that the last-mentioned surfaces have in common the line X=0 , Z=0 , which is on the first surface a torsal line ( equation in vicinity being Z =X2 ) , and on the second surface a triple line ( equations in vicinity being Z--X2 and X-YZ3 ) .
But Z-Y X2 touches Z=X2 , and the line counts thus ( 2+2=z ) 4 times .
107 .
I say that the complete intersection is the line ( X=0 , Z=0 ) three times together with a spinode curve made up of this same line once and of the curve of the eighth order ; and that thus d-=9 .
The discussion of the reciprocal surface in fact shows that the reciprocal of the edge is a singular line thereof , counting once as a nodal and twice as a cuspidal line thereof ; the cuspidal tangent planes are the reciprocals of the several points of the edge , and the edge is thus part of the spinode curve .
The reasoning may appear to show that the edge should be counted twice , but it must be counted once only , making the order =9 as mentioned .
108 .
I find that the octic component of the spinode curve is a unicursal curve , the equations of which may be written X : Y : Z : W -162 : 40+1605 : 1603 : -5-84--160s ; the values of 0 at the binode B , are 6=00 , ==o , and we thus obtain in the neighbourhood thereof the two branches T ( / Z)2 Xt3aX Z\ ) Y Z)2 w=5 ' w=2s5w ) and Ww -284 these two lines and of the line y=O , w=0 ( reciprocal of edge ) , as will presently appear ; so that we have b'=3 .
112 .
The equations of the cuspidal curve are L212wM=O , LM + 9w2N=0 , 4M2 + 3LN =0 .
Attending to the two equations L2-12w"2M =y4 + 8y2w+16x2w2+ 24yzw + 48w4= 0 , LM + 9w2N yz + 2y2w " 4xyzw+(8-72 ) -6 4xw3+ 18z2w2= , these surfaces are each of the order 4 , and the order of their intersection is =16 .
But the two surfaces contain in common the line ( y=O0 , w=0 ) 7 times ; in fact on the first surface this is a cusp-nodal line 4xw +y2+Ay2=0 ; and on the second surface it is a nodal line w(4xy+18zw)=0 ; the sheet w=0 is more accurately 4xw+y2+By3 ... =0 ; whence in the intersection with the first surface the line counts 5 times in respect of the first sheet and 2 times in respect of the second sheet ; together ( 5+2=)7 times , and the residual curve is of the order ( 16 -7 =)9 .
113 .
I say that the cuspidal curve is made up of this curve of the 9th order , and of the line yO0 , w=0 ( reciprocal of the edge ) once ; so that c'=10 .
In fact , considering the line in question y==O , w=0 in relation to the surface , the equation of the surface ( attending only to the lowest terms in y , w ) may be written Z2(y2+ 4w)2 + w(( y32 ) + W2(3 6ty3 ) + &amp;c. 0 , giving in the vicinity of the line 4^+y2_ y4xw-+y=Ay , and then z3 XZ 2A+ x ( 2+ _36 )= 0 that is , A2=-22 or 4 xw+-y2=-/ -2 .
y ; wherefore the line is a cusp-nodal line , counting once as a nodal and once as a cuspidal line ; and so giving the foregoing results b'=3 , c'=10 .
114 .
I revert to the equation which exhibits there :nodal lines ( a-w=0 , y+z=0 ) , ( x+-w=O , y-z=O ) for the purpose of showing that they have respectively no pinchpoints ; that is , that in regard to each of them we have j'=O .
In fact for the first of these lines , neglecting the terms which contain --w , y+z conjointly in an order above the second , the equation may be written 64w3(x+w)2 ( X-_W)2 + 8w2(x+w)2(x+3w)(z +y)2 + 8w2(z-y)2(x -3)(x_-w)2 287 The three transversals are each facultative ; '= '= 3 ; t'=0 .
121 .
Hessian surface .
The equation is 4aXZW(3Y+X+Z+W ) +Y2(X2+Z2 W2 -2XZ -2XW-2ZW ) = 0 .
The complete intersection with the cubic surface is made up of the lines ( Y= 0 , X= 0 ) , ( Y=0 , Z-0 ) , ( Y0O , W=0 ) ( the axes ) each twice , and of a sextic curve which is the spinode curve ; a-=6 .
The spinode curve is a complete intersection 2x3 ; the equations may in fact be written Y3+Y2 ( X+Z+W)+4aXZW=O , 3Y2+4Y(X+Z +W ) +4(XZ+XW+ZW)= 0 ; the nodes D , C , A are nodes ( double points ) of the curve , the tangents at each node being the nodal rays .
Each of the transversals is a single tangent of the spinode curve ; in fact for the transversal Y+Z+X0 , W-=0 , these equations of course satisfy the equation of the cubic surface ; and substituting in the equation of the Hessian , we have Y2(X-Z)2'=0 .
But Y+Z+X=X 0 , W=0 , Y==0 is a point on the axis W-=0 , Y=0 , not belonging to the spinode curve ; we have only the point of contact Y+X+Z-0 , W= , 0 X-Z-=0 .
Hence '-3 .
]eciprocal Surface .
122 .
The equation is found by means of the binary cubic , aT(T-yU)2+ ( T-xU)(TzU)(T-wU ) , viz. writing for shortness y=xz+-xw+zw , Y= -zxW -+W , z =XZZ ; this is a binary cubic ( *T , U)3 , the coefficients whereof are 3(a + ) , -2cy - ?
, y2 + a , 3s , and the equation is hence found to be 4a ?
f(y-2 + rY ) + a2{(127~ 2)y4 ( 83y7+36)y3+(30 P+ 872)y2 3c+ 27'2 } +2a{(6y2_-2y_ 9)y2+(12f3 -2^y2g`-18y)y+ 2y3 + 27T2h9Ir } _(P27+ 18P7a-4p3 -4y'27'2)=0 ; or substituting for 3 , y , ; in the first and last lines their values ( = xz+z+ , xz + Oz-+z , txzWv ) , 292 and the lines are X=0 , Y =0 , ( 0 ) f , X-Y=0 , Z =0 , ( 1 ) f2X-Y=0 , Z =0 , ( 2 ) 3X-Y=0 , Z =0 , ( 3 ) f , X-Y=0 , W=0 , ( 4 ) f2X-Y=0 , W=0 , ( 5 ) f3X-Y=0 , W-0 , ( 6 ) 128 .
There is no facultative line ; i'-b'-0 , t'=0 ; and hence also '=0 .
129 .
Hessian surface .
The equation is X{ZW(cX+ dY)3X(ac-b2 , ad-bc , bd-c2.X , Y)2 } =0 , so that the Hessian breaks up into the plane X=0 ( axial or common biplane ) and into a cubic surface .
The complete intersection of the Hessian with the cubic surface is made up of the line X=O0 , Y=O ( the axis ) four times ; and of a system of four conics , which is the spinode curve ; d'=8 .
In fact combining the equations WXZ+(a , 6 , c , d)X , Y)3=0 and ZW(cX + die ) -3X(ac-b2 , ad-be , bd-c2XX , Y)2=O , these intersect in the axis once , and in a curve of the eighth order which breaks up into four conics ; for we can from the two equations deduce ( a , b , c , djX , Y)3(cX-+ die )+ 3X2(ac-b2 , ad-bc , bd-c2XX , Y)2=0 , that is , ( 4ac--3b2 , ad , bd , cd , d2XX , Y)4-0 , a system of four planes each intersecting the cubic XZW +(a , b , c , dXX , Y)3 0 in the axis and a conic ; whence , as above , spinode curve is four conics .
It is easy to see that the tangent planes along any conic on the surface pass through a point , and form therefore a quadric cone ; hence in particular the spinode torse is made up of the quadric cones which touch the surface along the four conics respectively .
Reciprocal Surface .
130 .
The equation is obtained by means of the binary cubic X(xX+yY)2+4zw(a , b , c , d(X , Y)3 , viz. calling this ( *xX , Y)3 the coefficients are ( 3xq2+12azw , 2xy+12bzw , y2+12czw , 12dzz ) .
295 eliminating zw , we obtain ( 3(bd-c2 ) , 2(-ad2-3bed + 4c " ) , 6(bc-ad)b , a2d-~ &amp; 3 , which shows that the cuspidal curve lies in 4 planes , and it hence , consists of four conics ; these are of course the reciprocals of the quadric cones which touch the cubic surface along the four conics which make up the spinode curve .
134 .
The equation of the surface , attending only to the terms of the second order in Y , Z , 1V , is 27d2x4zw -= 0 ; it thus appears that the point y = , z=O , w==O ( reciprocal of the plane X-=O ) ( which is oscular along the axis joining the two binodes , or BB-axis ) is a binode on the reciprocal surface , the biplanes being z= 0 , w= 0 , viz. these are the planes reciprocal to the binodes ( X=0 , Y=0 , W-0 ) and ( X-0 , Y==O , Z=0 ) of the cubic surface ; we have thus B'=1 .
It is proper to remark that the binode y=O , z =0 , w=v 0 is not on the cuspidal curve , as its being so would probably imply a higher singularity .
135 .
A simple case , presenting the same singularities as the general one , is when t-d , b=c=0 : to diminish the numerical coefficients assume a=d=:- , the cubic surface is thus 12XZW+X3+Y3--0 , and the equation of the reciprocal surface , multiplying it by 4 , becomes ~3~3 z w3 + 6x2z2W2 + ( 9x412x3y)zw 4y3(-3 )= ( ) , viz. this is the surface 4y6 -4y3x(x2 + 3w ) +zw(3x2+zw)2= 0 considered in the Memoir " On the Theory of Reciprocal Surfaces .
" The cuspidal curve is , as there shown , composed of the four conics y 0 , 3x2+zw-O and y3--2x'30 , x_--zw=O ; and it is there shown that the two points ( x-O , y=0 , z--0 ) , ( x=0 , -=0 , w=O0 ) , each reckoned 8 times , are to be considered as off-points of the reciprocal surface .
136 .
The like investigation applies to the general surface , and we have thus 0'-16 ; the points in question are still the points ( x=-0 , y-=0 , z-0 ) , ( x=0 , y=0 , w-=0 ) ; viz. these are the points of intersection of the surface by the line ( x=0 , y--0 ) , which points are also the common points of intersection of the four conics which compose the cuspidal curve , that is , they are quadruple points on the cuspidal curve ; it does not appear MIDCCCLXIX .
2s 297 The facultative lines are the edge counting twice , and the mere line ; =-b=_3 ; t'=1 .
140 .
Hessian surface .
The equation is X(X + Z)(ZW + 3X2 -XZ)+ Y2(XZ)2= 0 .
The complete intersection with the surface consists of the line ( X=0 , Y=0 ) , the axis , 4 times ; the line ( X=0 , Z=0 ) , the edge , 2 times ; and a sextic curve , which is the spinode curve ; d= 6 .
Writing the equations of the surface and the Hessian in the form X(ZW+Y2)-X3 +Z(Y2-X2)=0 , X(X+Z)(ZW+Y2)+(Z-3X){-X3+Z(Y2--X2 ) Z0 , we see that the equations of the spinode curve may be written ZW+Y2=0 , -X3+z(Y2 -X2)=0 viz. the curve is a complete intersection , 2X 3 .
There is at B4 a triple point W -W , W ) ; and at C2 a double point , the tangents coinciding with the nodal rays W=-- , Y2-X2=0 .
The edge and the mere line are each of them single tangents of the spinode curve .
But the edge counting twice in the nodal curve , its contact with the spinode curve will also count twice , that is , we have P'=2.1+1 , =3 .
Reciprocal &amp; Srface .
141 .
The equation is obtained by means of the binary cubic 4w2X(X+ Z)2 + 4wZ(X + Z)(xX+zZ)+y2XZ2 ; or calling this ( x*X , Z)3 , the coefficients are ( 12w2 , 8w2+4wx , 4q2+4wx7+4wz+y2 , 12wz ) , and thence the equation is found to be 16V4[y2(x-_ Z)2 ] + 16Gw3[25z)y22((x2z)(x-)2 ] + 8w2[yO4+ ( T2 -xz+ 6z2)y22x2(X_-)2 ] + 4w[(2x+ 3z)y4--2x2(x+)y2 ] , q_ y(y-2 Xa2)=0 , where the section by the plane w= 0 ( reciprocal of binode ) is y4(y2-X)=0O , viz. this is the line w=-0 , y=0 ( reciprocal of the edge ) four times , and the lines w=-0 , y--x=0O ( reciprocals of the biplanar rays ) .
The section by the plane z=0 is found to be ( y2-2)(y2+4w+4w2)2= 0 , viz , this 2 s2 299 viz. the surface consists of the uniplane X+Y+Z=O twice , and of a quadric cone having its vertex at U6 , and touching each of the planes X-=0 , Y=0 , Z=0 .
The complete intersection with the cubic surface is made up of the rays each twice and of a residual sextic , which is the spinode curve ; o'=6 .
The equations of the spinode curve are W(X+Y+Z)2+XYZ-= , X2+Y2+Z2-2YZ2ZX-2XY 0 , viz. the curve is a complete intersection , 2X 3 .
Each of the mere lines is a single tangent ( as at once appears by writing for instance W=0 , X=0 , which gives ( Y-Z)2-0 ) ; that is , P'=3 .
Reciproccl Surface .
154 .
The equation is found by means of the binary cubic 4(T-xU)(T--yU)(T-zU)+wT2U , viz. writing for shortness 3 x+y+ z , =xyz , then the cubic function is ( 12 , w-43 , 4y , -12 XT , U)3 , and the equation of the reciprocal surface is found to be 432$2 + 6473 ( W -4 )3 ) + 72(w-43)yr ; ( w -43)2y2=O ; expanding , this is w3.-+w2.12S -yr2 +8w .
-663p2+Py2++ 1.6(43 -32y-t322Sys+4y3+27T2)=O ; or substituting for 3 , y , 3 in the first and last lines , this is +W .
--yz +tw .
( 12 } , -y2 ) + 8w .
623 -+3py2 +9 y+ 16(y-z)2(_z-)2(xy)2=0 ( where 3 , r , y-x+y+z , yz+-zx+-y , xyz ) .
The section by the plane w=0 ( reciprocal 304 cone having its vertex on the second surface , we see moreover that the spinode curve is a nodal quadriquadric .
Instead of the last equation we may write more simply 4Y(X+Z)+16XZ 3W(3Y+ -4X+4Z)=-0 .
The equations of the transversal are W=-O , X+Y+ZO0 , and stibstituting in the equations of the spinode curve we obtain from each equation ( X-_Z)2-0 , that is , the transversal is a single tangent of the spinode curve ; =3'1 .
Beciprocal Sutrface .
162 .
The equation of the cubic is derived from that belonging to VI-12--B3-C by writing therein a=b-=0 , c=- , d=l .
Making this change in the formule for the reciprocal surface of the case just referred to , we have L =y+4(x+z)w , -M=2x(y+2w ) , N -42 P -16x2(y+w--x ) ; and substituting in the equation L2P + 8zM39zLMN -2 7z2N2= 0 , the equation divides by x2 ; or throwing this out , the equation is ( y2 + 4xw + 4zw)2(y +w-x-z ) -8xz(y + 2w)3 + 9x(y2 +4 4xw + 4zw)(y + 2w ) -27x2Z2w0 ; reducing , this is W3 .
16(xz)2 + w2 y2(x + ) Z+ +2y(x -4xz + ) zI +(x +)(2x-z)(-+2xz ) +w y +(8y(+ z ) -22 ( 4 2+23xz+4 z ) ) +36xyz(x+z ) _-27x2zJ + y(y-x)y-(-z)=O .
The section by the plane w=O0 ( reciprocal of B , ) is w=0 , y=O ( the edge ) 3 times ; and wt-O , y-x= 0 ; w=-O , y-z-O ( reciprocals of the CB-axes ) .
163 .
Nodal curve .
This is the line y=-x=z ; wherefore b'=1 .
To put the line in 2T2 307

