]\gt ; [ 289 ] XVI .
On the Contact of Conics with Surfaces .
By , F.R.S. Received February lG , \mdash ; Read March 10 , 1870 .
IT is well known that at every point of a surface two ents , called principal tangents , may be drawn having three-pointic contact with the surface , .
having an intimacy exceeding by one degree that generally enj ed by a straight line and a surface .
The object of the paper is to establish the corresponding theorem respecting tangent conics , viz. that " " at every point of a surface ten conics may be drawn having six-pointic contact with the surface ; \ldquo ; these may be called Principal Tangent Conics .
In this investigation I have adopted a method analogous to that employed in my paper " " On the Sextactic Points of a Plane Curve\ldquo ; ( Philosophical Transactions , vol. clv .
p. 653 ) ; and as I there , in the case of three introduced a set of three arbitrary constants in order to comprise a group of expressions in a single formula , so here , in the case of four variables , I introduce with the same view two sets of four constants .
If these constants be lepresented by , I consider the conic of fivepointic contact of a section of the surface made by the plane , where , and , and is indeterminate ; and then I proceed to determine , and thereby the azimuth of the plane about the lin , so that the contact may be six-pointic .
The thence arisin turn out to be strictly analogous to those to the case of three variables , except that the arbitrary quantities cannot in general be divided out from the final expression .
In fact , it is the presence of these quantities which enables us to determine the position of the plane of section , and the equation whereby this is effected to be of the in : besides this of the in the coordinates giving rise to the theorem above stated .
Beyond the question of the principal tangents , it has been shown by CLEBSCH and SALMON that on every surface a curve may be drawn , at every point of which one of the principal tangents will a four.pointic contact .
And if be the of that of the surface intersecting I7 in the curve in question will be .
Further , it has been shown that at a finite number of points the contact will be five-pointic .
The number of these points has not yet been completely determined ; but } has shown ( Crelle , vol. lviii .
p. 93 ) that it does not exceed .
Similarly it appears that on every surface a curve may be at every point of which one of the principal tangent conics has a seven.pointic contact , and that at a finite number of poig ' : the contact will become ) ointic .
But into the discussion of these latter problems do not propose to enter in the present communication .
IlJCCCLXX . .
Conditions for a Sextactic Point .
Let be the equation to the giyen surface , and that to the surface whose section by the plane is to have a six-pointic contact with the corresponding section of at the point .
Also , following the method of Professor CAYLEY Philosophical Transactions , cxlix .
, let the coordinates of a point of be considered as functions of a single parameter ; then for the present purpose the coordinates of a point consecutive to may be taken to be . .
; ( 1 ) and these values when substituted in must satisfy the equation .
Then writing for shortness ( 2 ) substituting the values ( 1 ) in U. , expanding as far as terms of the fifth degree , and arranging the l.esult in lines of the degrees , 1 , . .
5 , respectively , we shall have ( 3 ) each line of which , being of an order different from the rest , must separately vanish .
Let us write , as usual , ( 4 ) Then combining the equation with the corresponding expression in V , viz. , we obtain the usual expressions for two-pointic contact , ; . .
( 5 ) which , since and V are both homogeneous in , are equivalent to only two independent conditions .
These conditions may be comprised in the single formula ME .
W. SPOTTISWOODE ON THE CONTACT OF CONICS WITH SURFACES .
291 ' , ( 6 ) where are arbitrary quantities to which any values may be given , provided only that all the determinants of the matrix do not simultaneously vanish , since in that case the equation ( 6 ) would became nuga tory .
Comp n with the equation , and observing that the differentials may be replaced by the determinants ' ( 7 ) ( the usual lule of signs adopted , viz. the columns being taken in the cyclic order , the determinants are to have the signs , - , , - , lespectively ) , it is clear that the equation , when combined with the corresponding equation in , is equi- valent to .
Similarly , the equation , combined with the corresponding equation in , is equivalent to ; and so on .
Hence the series of conditions comprised in ( 3 ) may be expressed as follows : , ( 8 ) which correspond to equations ( 3 ) of my paper on Sextactic Points above quoted .
Preliminary Transformation .
The next step is to effect a transformation of the first three equations of the system ( 8 ) , corresponding to that given in of the same paper .
As was stated in the introduction , the transformation does not throw out as factors any lineo-linear functions of the arbitrary quantities and the variables ; but it reduces the expressions transformed to functions of such lineo-linear functions , viz. , and Taking the columns of the matrix ( 7 ) two and two in the usual cyclic order , viz. , and calling the determinants so formed ; .
writing ' ' ' ; ( 9 ) and writing , further , ( 10 ) and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , and writing , there , the quantities , . .
, . .
will be found to satisfy the following relations , useful in subsequent transformations , viz. :\mdash ; .
( 11 ) And in terms of this notation the developed form of will be iven by .
( 12 ) This being premised , our first object is to investigate , as was done in the case of plane curves , an expl.ession for , which in virtue of ( 12 ) will consist of two parts : first , terms of the form , . .
; and secondly , terms of the form . .
Referring to , we have MR. W. SPOTTISWOODE ON THE CONTACT OF CONICS WITH SURFACES .
293 . . .
, 7 and vriting . . . . .
( 13 ) we may deduce ( 14 ) ' In the same way or Similarly , it will be found that .
A A ( 16 ) ; so that , being the degree of , we shall have , on collecting all the terms of the forms ( 15 ) and ( 16 ) , the following expression:\mdash ; . .
And if to each side of this equation we add in which the operations , . .
affect the quantities , only and not , . .
; then we shall have the full expression for , viz. Now referring to ( 5 ) , and calling the ratios therein contained , and substituting for for , . .
, we have ( wh\mdash ; wg ) .
A A .
W. SPOTTISWOODE ON THE CONTACT OF CONICS WITH SURFACES .
295 Further , if we agree upon the proper mode of development we may write .
A C A A B suppose . . . .
( 17 ) which expression , by giving obvious values to , . .
, .
, may be written thus : .
( 18 ) Lastly , putting ( 19 ) A B .
' the expression for finally becomes and consequently the system may be replaced by .
( 20 ) If in the foregoing expressions we put , we shall have the case of plane curves , and as the last suppositions give , the expression ( 20 ) then reduces itself , as it should , to that given in equation ( 16 ) of the memoir above quoted .
S 3 .
Elimination of the Constants of the Quadric Before proceeding to the application of the formula ( 20 ) to the present problem , it will be convenient to premise that if be any two rational integral and homogeneous functions of , the nature of the operation is such that ; ( 21 ) . . . . . . . . . . . . . . . . . . . .
from which it follows that ( 24 ) the outstanding which occur in 23 ) having cancelled one another .
Furthermore , applying the mula ( to the following products , and bear ng mind that , are all linear in , we have , by means of ( 24 ) , , ( 25 ) : : : so that ( 26 ) the coefficients of in ( 25 ) having cancelled one another .
This being premised , let us return to the equations ( 20 ) .
These equations being per fectly general , we may replace by any other function we please .
Hence if we ace by , the system resulting will be an equivalent for the system , viz. the second , third , and fourth of the system ( 8 ) .
Making this substitut on in ( 20 ) , and remembering that is of the degree , so that the numencal factor , we have MR. W. SPOTTISWOODE ON THE CONTACT OF CONICS WITH SURFACES .
297 ; or the first and last of these expressions , and substituting from ( 26 ) , ; or substituting for , . .
, or . . . . .
; and if , . . . . .
( 28 ) the expression ( 27 ) is equivalent to which contains only the four coefficients , 1 .
In this expression the coefficient of 7 , 7 ' , , us If , therefore , ( 30 ) will take the form , .
to which be added the analogous equations in Another , and for some purposes a more convenient form may be given to these equations by the transformation : , suppose ; If , therefore , ( 30 ) will take the form , .
to be added the analogous equations in Another , and for some purposes a more convenient form may be given to these equations by the .
transformation : , suppose ; If , therefore , ( 30 ) will take the form , .
to be added the analogous equations in Another , and for some purposes a more convenient form may be given to these equations by the .
transformation : , suppose ; H. If , efore , henceforward esent the differential coefficients of upon the supposition that are re arded as constants , we shall have so that ux ; and if represent the detcrminants from the first three together with the fourth , the fifth , the sixtl ] , and the seventh respectively of the following matlix : A MR. W. WOOD ON THE CONTACT OF CONICS WITH SURFACES .
299 and if , as was indicated above , in the transformed expression is supposed to be differentiated without reference to ; then and the equations , of which is one , take the form Representing any one of these equations , say , the first , by , the equations may be replaced by a system of the form ( 20 ) ; and writing them in the form , . .
, where is indeterminate , we may from the five equations so written eliminate the five quantities , 1 , ; and the resulting equation takes the form ( 32 ) where is a numerical factor .
S 4 .
Determination of the extraneous factors .
The degree of the equation ( 32 ) in its present form is ; but it admits of reduction , in the first place , as follows .
Since the equation is identically satisfied , we have by diffelentiation , and by ) stitution of the values of , . .
from ( 11 ) , , suppose .
From this we may deduce the following system : . .
( 33 ) . .
: : , since , it follows that But since . . . . .
( 34 ) the expression above written reduces itself to .
( 36 ) Hence , if we multiply the first three columns of ( 32 ) by respectively , and add to the fourth multiplied by , the whole equation will be divisible by ; and on the division made the equation in question will take the form ( 36 ) which is of the degree MR. W. SPOTTISWOODE ON THE CONTACT OF CONICS WITH SURFACES .
Next it may be shown that is a factor of this expression .
For when the following equations subsist , whence by elimination .
( 37 ) so that , omitting a numerical factor , will be replaced by , and ( 36 ) will then become But by ( line ) ( line ) ( line 3 ) to ( line 4 ) , the whole of line 4 will be divisible by and a numerical factor ; so that the expression becomes Again , line 4 from line 1 line 4 from line 2 line 4 from line 3 line 4 from line 5 , and hout by 2 , the expression is reduced to lastly , subtracting ( column 1 ) ( column 2 ) ( column 3 ) from ( column 4 ) , and ( column 1 ) ( column 2 ) ( column 3 ) from ( column 5 ) , we have ( remembering that ) Dt ; which vanishes identically .
Hence is a factor of ( 36 ) ; and on dividing it out the degree of the expression ( 006 ) is reduced to It remains now to be shown that is likewise a factor of ( 36 ) .
Putting , that equation becomes .
X . .
; and following a process similar to that adopted in the former transformation , this may be reduced to . . . . . . .
and thence to . . . . . .
Dt .
MR. W. SPOTTISWOODE ON THE CONTACI ] OF CONICS WITH SURFACES .
303 and being two functions , with the exact forms of which we are not concerned .
The expression then takes the form of the product cf two factors , viz. , and Hence is a factor of ( 36 ) .
NolV that equation as it stands is of the third degree in ; so that being divided out it is reduced to the second degree , say , to the form .
Now .fo ( 31 ) , and forming the equations in , . .
; 1 , . .
, the equations to ( 36 ) may by a similar process be shown to be divisible by ) , ; and those divisions been effected , the equations in question will be reduced to the forms , .
, in which it is to be observed , from the symmetry of the expressions , that the coefficients of , . .
are the same .
But as these equations those all lead to the same result , namely the determination of the sextactic points , they can differ from one another only by Hence , as in my memoir before quoted , equation ( 50 ) , we must have identically .
, .
( 38 ) which can hold good in general , only in virtue of divisible by , and by being diyisible by and by , and so on .
Hence ( 36 ) is divisible not only by but by ; and that division ?
been effected , the degree of ( 36 ) will be reduced to This completes the enumeration of the extraneous factors ; but although the of the equation ) ) cannot in general be depressed below l , we have yet to show that , as stated in the introduction , the variables enter to the degree 10 in the form of lineo-linear functions of the arbitral.y quantities .
From what has gone before , it is clear that the tities X in ( 36 ) , are all functions of only , that is to say , not of , . .
, . .
, except in so far as they are included in , D. It remains to be proved that this is still the case after the differentiations and operation involved in ( 36 ) have been .
For this purpose let represent any function of , and le , . .
indicate differentiation respect to , . .
so far as they appear explicitly in , irrespectively of , then ; ; and if be of the degree in , and .
, it follows that . . .
( 39 ) : : : : Again , whence . . . .
( 40 ) : : : whence , again , ; : : : : and combining these expressions , Hence . . . . . . . . . .
which , by a method precisely similar to that followed above in investigating the extraneous factors , may be reduced to NIR .
W. SPOTTISWOODE ON THE CONTACT OF CONICS tVITH SURFACES .
305 which is the same as if had not been affected by the by the operation .
The expression is therefore a function of , and not of , . .
, , . .
, excepting so far as they appear in , D. If , therefore , we put , we conclude that ( 36 ) , when divested of its extraneous factors , is a function explicitly of the degree in and of the degree 10 in , D. But are themselves linear homogeneous functions of ; so that the expression in question may be as explicitly of the degree in , and of the degree 10 in .
'I'his equation , solved for , will consequently give ten positions of the cutting plane all passing through the point , for which the curve of section is sextactic at the point .
Hence the theorem , " " If be any line through any point of a surface , ten conics may be drawn in planes passing through having six-pointic contact with the surface at the point It is to be observed that the plane of section is not necessarily normal ; but if the line whose six coordinates are be made to coincide with the normal at point whose rectangular coordinates are , it will follow that with other relations which would abbreviate , although not essentially simplify , some previous expressions .
The results of placing the line in the tangent plane are noticed below .
on of more than two-pointic contact .
If , instead of being as hitherto quadric , be linear , we shall have the case of ts to the curve of section of with the plane .
And if the contact be threepointic , each of the ratios , . .
will be , in virtue of ( 20 ) of , equal to : H. But since is linear , and consequently the condition for a of three-pointic contact is .
Now ; so that the condition may be written , ( 41 ) which will determine two values of , and consequently two positions of the plane for which the tangent line has three-pointic contact .
It may be noticed that in the solution of the equation above written there occurs the following expression : IIDCCCLXX .
2 But if represent the Hessian of , then , . .
; and the expression in question will be equal to the product of the Hessian into where , . .
, . .
are the six coordinates of the line about which the cutting plane is supposed to revolve , as given in equations ( 9 ) .
The two positions of the cutting plane will coincide , if either the Hessian or this expression vanishes ; the latter , regarded as a l.elation between the coordinates of the line , expresses the condition that the plane of section may contain both principal tangents , .
may be the tangent plane , as may be verified by the following considerations .
If the plane of section coincide with the tangent plane , then , . .
, which , being substituted in the expression in question , will cause it to vanish identically ; or we may proceed otherwise , thus : regarding as the constants of the cutting plane , the equation is ; . . . . .
( 42 ) the equation of the plane itself may be written ; ( 43 ) and since this plane passes through the point , we have also ; . . .
( 44 ) by means of which three equations the ratios may be determined .
By a process quoted by Professor CAYLEY Quarterly Journal Mathematics , , the solution of these equations depends upon the square root of the quantity , ae , . . .
( 45 ) , ffl , , ff , $ , ffl , which is in fact identical with the expression ( 41 ) .
The locus of the points for which one of the principal ents meets a given line , say , the line , will be found by eliminating ffom the equations ( 42 ) , ( 43 ) , ( 44 ) , combined with the following condition : . . . .
( 46 ) But since the principal tangent is the intersection of the plane with the tangent plane , we have ( 47 ) MR. W. SPOTTISWOODE ON THE CONTACT OF CONIC8 WITH SURFACES .
307 Substituting these values , the condition above written becomes and the ratios will be expressed by the determinants of the matrix ' ' But if by analogy with ( 11 ) we write it will be found that , putting the ratios are equal to ; and consequently when these are substituted in the terms having for a coefficient vanish , and the equation of the locus resulting will be . . . . .
( 48 ) To find the locus of points at which one of the principal tangents has a four-pointic contact with the surface , we must add to the equation the following , viz. which , as has been shown in a former part of this paper , may be replaced by any one of the group ' . . . . . .
( 49 ) But , remembering that represent the differential coefficients of with respect to , on the supposition that are constant , and writing it is easy to deduce the following system : : : But is equivalent to the system , . . .
to or Substituting these values of , in the equation , the terms having for a coefficient will vanish , and the equation will take the form . . . . . . . .
( 50 ) But from what has gone before it is clear that when resolved into its factors is of the form , where represents the expression ( 45 ) ; hence putting ; and consequently since , . .
But , the upper or lower sign being taken according as one or other principal tangent is the subject of consideration .
Substituting then in the equation ( 50 ) , we have But since we are seeking the condition under which either one or the other principal tangent may have four-pointic contact , the terms of ambiguous sign must disappear ; and the condition required will take the form which is of the degree , and may be compared with CLEBSCH'S form , viz. ; but the companson of the terms in and appears difficult .
The additional condition for a five-pointic contact on the part of one of the principal tangents will be ; or , having reference to ( 20 ) and to the consideration that , the condition will be .
But the further discussion of this question I postpone to another occasion .
' IV I

