]\gt ; [ 367 ] XI .
On a new Form of Tangential Equation .
JOHN CASEY , .D .
, F.R.S. , M.R.I.A. , Professor of Mathematics in the Catholic University of Ireland .
Received January 24 , \mdash ; Read February 22 , 1877 .
INTRODUCTION .
ART .
1 .
The tangential equation of a curve is , as is well known , a lation among the coefficients in the equation of a variable line , which being fulfilled , the line must be a tangent to the curve .
Let be the origin , OX , OY the axes ; and let a variable line MN in any of its positions make an intercept on OX and an angle with it ; then the equation of the line is and and , the quantities which determine the position of the line , may be called its coordinates .
From this it follows that any relation between and , such as , . . . . .
( 1 ) will be the tangential equation of a curve which is the envelope of the line .
This form of equation will be the special subject of this paper .
Occasionally our investigations will embrace collateral subjects , when their importance will be such as to justify the digression .
It will be seen that our form of equation admits of easy transformation into all the known forms of equation ; that it adapts itself with great facility to the various problems of the Integral Calculus relating to curves , such as Rectification , Curvature , Involutes , \amp ; c. , and gives its results in very simple forms .
In most of the methods of Modern Geometry , such as Pedals , Parallel Curves , Reciprocation , \amp ; c. , it solves in a very simple manner problems that are very difficult by any other method .
I have illustrated it throughout by numerous examples , most of which are of historical interest .
Some of the problems discussed are , I believe , now solved for the first time , among which I may mention the rectffication of Bicircular Quartics by Elliptic Functions .
To this outline of the subject of this paper I may add that the form of equation is suggested by Kinematics .
For if we differentiate the equation with resI ) to a variable ( the time ) we get . . . . . .
( 2 ) Now if we suppose a rigid body to move so that a fixed point in it , say the centre of gravity , describes a right line , then will be the linear velocity of the centre of gravity , and will be the angular velocity with which the body revolves lound the same point .
Then the equation ( 2 ) will be the most general equation of the motion of such a body .
It gives linear velocity divided by velocity as a function of the angle through which the body has rotated .
From this it will be seen that some of our results will have a physical as well as a purely mathematical interest .
With these remarks we proceed to the subject of the paper .
CHAPTER I. @ECT10N I.\mdash ; Transformation of Cartesian into Tangential Equations .
2 .
Definition.\mdash ; We shall find it convenient to call the line OX , on which the variable line makes the intercept , and with which it makes the angle , the director line .
3 .
If the Cartesian equation of a curve be , we can by the usual process find the condition that the line touches it ; this condition will be our tangential equation .
For this purpose the equation of the line may be written in the form , . . . . .
( 3 ) where denotes tano ; and eliminating between this and the equation , we shall have an equation in of the form . . . . . .
( 4 ) The discriminant of this will be tangential equation required .
It can be transformed into the usual form of tangential equation by changing into and into .
This is evident by comparing the equations Cor. The usual form of tangential equation can be transformed into our form as follows:\mdash ; Let . . . . .
( 5 ) be the tangential equation , say of the nth degree ; divide by , and change into and into FORM OF TANGENTIAL EQUATION .
4 .
The coefficients in equation ( 4 ) are deserving of notice .
Equated to zero , they are the tangential equations of curves which possess interesting geometrical relations to the curve U. For the purpose of examining their properties , let the curve be given by the equation .
( 6 ) then substituting in this the value from equation ( 3 ) , and equating the result with equation ( 4 ) , we get the following system of identities:\mdash ; ( 7 ) 5 .
The system of identities ( 7 ) are remarkable for their symmetry , the equation being independent of all but the coefficients of the highest powels of of all the homogeneous terms lower than the in and , \amp ; c. Transformed into the usual form of tangential coordinates , they become ( 8 ) 6 .
We shall now examine the geometrical interpretation of the equations ( 8 ) , first , for the sake of illustration , in special cases , and then we shall give the general results .
We may remark in passing that all the contravariants of curves can be expressed in terms of these tangential curves ; for instance , if be a cubic , the envelope of the line which cuts it in three points , whose distances are in arithmetical progression , is the curve ; ( 9 ) and if be a quartic , the envelope of the line which it cuts harmonically is the determinant ' .
( 10 ) 7 .
Let the culve be a conic , then the equation ( 4 ) becomes Now if , it is evident the line will cut the curve in two points , which are equally distant from the axis of ; but when becomes ; . . . .
( 11 ) that is , a conic section .
Hence we have the following theorems , the second of which is the projection of the first , and follows from the equation in ?
, as the first does from the corresponding one in and 1st .
If a variable line intersect a conic section , and if the locus of its middle point be a right line , its envelope is a conic section .
2nd .
lf a variable line be cut harmonically by a conic section and a pair of lines , its envelope is a conic section touching the pair of lines .
8 .
Let be the cubic , ( 12 ) and the curve will be . . . . . .
( 13 ) This equation is the condition that the locus of the mean centre of the points where the line meets the curve is the axis of ; and since the axis of may be any line , we have the following theorem :a variable line intersect a cubic in such a manner that the locus of the mean centre of the points where it meets the cubic is a right line , its envelope is a curve of the third class .
9 .
The equation ( 13 ) , expressed in the usual notation of tangential coordinates , is . . . . . .
( 14 ) FORM OF TANGENTIAL EQUATION .
This is the analytical statement of the theorem we get by projecting that of the last article ; and since in projection the line at infinity becomes a finite line , it may be expressed as foUows:\mdash ; Being a curve of the third degree , , and two fixed hnes and ; then if be two variable points on and respectively , such that the polar line of with respect to passes through , the envelope of the OO ' is a curve of the third class .
10 .
Since the highest power of contained in equation ( 14 ) is the first , the tangential cubic which it represents has one double tangent , namely the line joining the points and , which we may call the line .
Similarly the line is a single tangent .
The same thing can be shown geomeFig .
2 .
trically , as follows : the lines and intersect in , then is the point whose equation is .
Now since the polar line of passes through , then the polar conic of passes through ; but this conic intersects the line in two points , and the line joining to each of them is a tangent to .
Hence from any point of the line CN can in general two tangents be drawn to ; and we shall see immediately that CN itself is a tangent .
This agrees with the fact of the curve being of the third class .
Let the polar conic of intersect CL in the points , then the lines are tangents to ; in words , CL is a double tangent , and it is plain that are its points of contact .
, let the polar line of intersect CN in , then is a point of contact , so that CN is a 11 .
Since the point moves on , its polar conic will pass through four fixed points , namely , the four poles of CN with respect to U. Hence any line will be cut in involution by the polar conics of the points ; and we have the following theorem:\mdash ; If from any three points in CN three pairs of tangents be drawn to , these will meet its double tangent in six points in involution , and the two points of contact of the double tangent belong to the involution .
12 .
We find the limiting points of the involution as follows : the pole conic of the line CL with respect to intersect CN in the points ; then since the poleconic is the locus of points whose polar conics touch , the polar conics of the points will touch .
Let the points where they touch it be denoted by , then will the double points of the involution .
Or thus , the double points will be the points of contact of the two conics , which can be drawn through the four poles of CN to touch 13 .
From the last article , it is plain that each of the lines is a of coincident tangents to the curve ; and since CN is itself a tangent we see that from each of the points can be drawn only two tangents to ; but the curVe is of the third class , therefore it must pass through and .
Hence we have the followirg theorem:\mdash ; The curve jrsects the line CN in the points where the pole-conic of CL meets it , and it touches in the point whose polar conic passes through C. 14 .
The polal line of the point with respect to will cut CN at its point of contact with .
The same polar line will be a tangent to the pole-conic of , and will be the polar of the point with respect to the polar conic of C. Hence it will with divide the segment of CL included between the points of contact with 15 .
We can get the equation of the line of which is the envelope as follows :Since we have ; and substituting in equation ( 13 ) we get ( 15 ) which is the required line , and the discriminant with respect to will be Cartesian equation of .
This discriminant is .
( 16 ) This equation is of the fourth degree , as it ought , since the curve has a double tangent .
16 .
If we denote the equation ( 15 ) by , and since a cusp is a point at which three consecutive tangents intersect , the conditions that there shall be a cusp are that ; and eliminating and from these equations , we get the following detel minant : ( 17 ) This determinant is a cubic in , showing that there are three cusps .
The values of , got from this equation , if substituted in equation ( 15 ) , will give us the three cuspidal tangents .
17 .
If we denote the singularities by the following notation\mdash ; Class Double tangents Cusps Degree Double points Points of inflection we have the singularities of the curve as follows:\mdash ; ( 18 ) FORM OF TANGENTIAL EQUATION .
Observation.\mdash ; The curve A2 for a cubic has properties similar to .
They differ only in that the lines CL and CN are interchanged , for CL is a single and CN a double tangent to 18 .
If be the general curve of the nth degree , gives th following theorem:\mdash ; a curve of the degree , and two lines CL and , then if be two movable points on these lines , such that the polar line of with respect to may pass through , the envelope of OO ' will be a curve of the class , to which the line CL will be a multiple tangent of the order .
19 .
If in the equation for given in art .
4 we substitute for its value , as in art .
( 15 ) , we shall find the equation of OO ' in the form .
( 19 ) Hence ( see SALMON'S ' Higher Curves , ' second edition , p. 66 ) we ve , ( 20 ) All this will also follow from the propositions of the following articles , of which this and the preceding are special cases .
20 .
We will now examine the general case The equation gives us the following theorem :be a curve of the th degree , and , CN two given lines , then ?
be two points taken on these lines , such that the th polar of with respect to passes through , then the envelope of OO ' is the curve of the nth class 21 .
The curve touches the line CL in points and CN in points .
Demonstration.\mdash ; Since the polar of passes through , the polar of passes through O. Hence we have two ways of the curve .
Now let the point move along CN until it becomes consecutive to , and it is evident that the points in which its polar intersects CL will be points of contact of CL with .
In like manner the points in which the polal of a point consecutive to on the line CL intersects the line CN will be points of contact .
Hence the proposition is proved .
Cor. The number of double tangents which has For the line CL is equivalent to double tangents , and the line CN to double tangents ; we have .
( 21 ) PROFESSOR J. CASEY ON A NEW 22 .
The curve is of the degree .
Demonstraf , ion.\mdash ; If be any point on the line , then since the polar of cuts CL in points , the lines drawn from to these points will make up tangents , and the line CL itself counts for tangents .
Hence the tangents which can be drawn from are accounted for .
Now if the point itself be on the curve , only tangents can be drawn from it , and two of the points in which the line CL is intersected by the polar curve of the degree must coincide , that is the polar curve must touch .
Hence we have to find the points on CN whose polar curves of the degree , with respect to , will touch .
In order to find the number of solutions of this problem , we will use trilinear coordinates .
Let be the coordinates of two fixed points on , then the coordinates of any variable point on it are , and the polar curve of this point of the degree , with respect to , is .
( 22 ) Now this equation contains the variables in the degree , and its coefficients contain in the mth degree .
Hence the condition that it will touch any given line will contain in the degree ; and this is the number of points in which the curve intersects the line , but it touches CN in points ; the total number of points in which the curve meets CN is .
Hence the proposition is proved .
23 .
The following are the singularities for the curve ( 23 ) Cor. , and is therefore the same for the curves \amp ; c. ; that is it is independent of Cor. 2 .
The curves have the same singularities .
Examples .
( 1 ) Fin the tangential equation of the cuspidal cubic Eliminating between this and the equation , we get ( 24 ) The discriminant of this is , ( 25 ) which is the required tangential equation .
FORM OF TANGENTIAL EQUATION .
In the usual notation this is . . . . . .
( 26 ) The equation ( 24 ) shows that the sum of the 's of the points where any line cuts is to the square of the of the angle which the line makes with the axis of .
and the sum of their reciprocals is proportional to the reciprocal of the intercept which the same line makes on the same axis .
Cor. It is evident that similar theorems hold for the curve ( 2 ) Let the curve be The tangential equations are , . . . . .
( 27 ) . . . . . .
( 28 ) ( 3 ) Find the tangential equations of the cissoid .
They are ( 29 ) . . . . . .
( 30 ) ( 4 ) Find the tangential equation for a cubic in its canonical form\mdash ; that is , referred to its three chords of inflection as axes .
This question is solved by supposing the coefficients in the equation ( 12 ) to vanish , except ; then equation ( 14 ) becomes the conic , . . . . .
( 31 ) and the cnrve for the Hessian of the cubic is , . . . . .
( 32 ) a curve which has double contact with the former .
( 5 ) Find the equations of the curves for the trinodal quartic , . . . . .
( 33 ) , . . . . . . . . .
( 34 ) . . . . . . .
( 35 ) ( 6 ) The points where the curve intersects the line CN may be found as follows:\mdash ; If a vanable point moves along the line CL the envelope of its polar curve of the degree with respect to will be a curve of the degree which will cut CN in the required points .
Similarly the points where it cuts CL may be found .
SECTION II.\mdash ; Transformation of Polar into Tangential Equations .
24 .
The polar equation of a curve being given , to find its tangential equation .
Let the polar equation be , then Fig. 8 . . . .
( 1 ) Also we have , that is , we have . . .
( 2 ) and . . . .
( 3 ) Then eliminating and between equations ( 1 ) , ( 2 ) , ( 3 ) .
The result will be the tangential equation .
Ex. Let the polar equation be .
( 36 ) We find , by taking logarithmic differentials , ; and Hence the tangential equation is or , putting in place of ( 37 ) 25 .
The family of curves represented by equation ( 36 ) includes several important species .
The following Table contains the principal , with their corresponding tanequations .
FORM OF TIAL EQUATION .
.377 CHAPTER II .
SECTION I.\mdash ; Transformation of Tangential into Cartesian 26 .
The tangential equations of a curve being given , to find its Cartesian equations .
Consider two consecutive positions of the line , such as intersecting in ; then is the point of contact Fig. 4 .
of LP with its envelope , and the diameter of the circle described about the infinitesimal triangle LL'P is evidently equal ; Hence and ; the Cartesian coordinates of the point are ( 46 ) . . . .
( 47 ) These values can be got also by the analytical method of finding envelopes .
For differentiating the equation with respect to , we get ; and substituting this in the equation , we get 27 .
From the results of the last article we get the subtangent . . . . . .
( 48 ) subnormal . . . . .
( 49 ) 28 .
If the movable line be a double tangent , it is evident that for the same values of and we must have two different values for , one value corresponding to each point of contact .
Hence , since , we must have two different values for .
This will happen when is given in the form of a fraction whose numerator and denominator each vanish .
Thus , suppose the equation of the curve to be given in the form ; then we have ; and therefore the conditions for a double tangent are , ( 50 ) which correspond with the conditions in Cartesian coordinates for a double point . .
Professor AYLEY considers a " " curve as described ( see SALMON'S ' Higher Curves , ' second edition , p. 33 ) by a point which moves along a line at the same time that the line revolves round the point .
There is , then , this peculiarity at a point of inflection , the line first becomes stationary and then reverses the sense of its motion From this it follows that the line will cut off a maximum or minimum intercept on the director line when it passes through a point of inflection , and also it will make in the same case with the same line a maximum or minimum angle .
Hence when is an inflectional tangent , maximum or minimum and maximum or minimum .
Examples .
( 1 ) If a line of constant length slide along two rectangular lines , to find its envelope .
In this case we have evidently ; Hence from equations ( 46 ) , ( 47 ) we get ; ( 51 ) ( 2 ) If from any point in an ellipse perpendiculars be let fall on the axes , find the envelope of the line joining their feet .
In this case , and the required equation is . . . . . .
( 52 ) ( 3 ) Let ; then if we put , we get the Cartesian equation . . . . . . . . .
( 53 ) ( 4 ) If , the Cartesian equation is ( 54 ) Compare equation ( 51 ) .
Cor. If in this example we substitute for , we get ( 55 ) FORM OF TANGENTIAL EQUATION .
as the rtesian equation of the curve ( 5 ) Let , where .
This curve is the evolute of the ellipse .
The Cartesian equation is . . . .
( 56 ) ( 6 ) To find a curve in which the subnormal is constant , let the constant be ; then from equation ( 49 ) we have ; ( 57 ) which is the common parabola ( see art .
25 ) .
Cor. In like manner the curve in which the subtangent is constant is ( 58 ) or ( 59 ) ( 7 ) If in fig. 4 ( art .
26 ) PL be produced to meet OS in , required to find the curve in which : LT in a given ratio , say .
Here we have evidently , if ; the required curve is ( 60 ) or .
( 61 ) PION II.\mdash ; Transformation of the Tangential into the Intrinsic Eguation .
30 .
If we differentiate the value of given in art .
26 , we get ; but , ( 62 ) Hence if be the tangential equation of a curve , its intrinsic equation is . . . . .
( 63 ) The result in equation ( 62 ) may be written in a form which in practice we shall find more useful .
Thus . . . . . . . . .
( 64 ) 31 .
Equation ( 63 ) may be established geometrically as follows:\mdash ; Let be two consecutive positions of the movable line , ; Fig. 5 .
their points of contact with the envelope , and their point of intersection .
Let be a perpendicular on .
Now is an element of the curye , and denoting it by , we have ; Which is the same resuI as before .
Cor. ( see fig. art .
26 ) ; . . . . . .
( 65 ) ( PL ) .
( 66 ) From this it will be seen that the triangle LPQ is an important one in this theory .
Observation.\mdash ; The geometrical method of proof shows that this theol.em holds even when the director line OL ( see def .
art .
2 ) is any plane curve ; and we shall further on have to make use of this generalization .
32 .
Before giving examples of the process of this section we will give the following integral reduction .
To reduce to the normal form of elliptic integrals .
Let , and .
Then after some easy reductions , we get Now and FORM OF TANGENTIAL EQUATION .
where .
( See , ' Theory der elliptischen Functionen Hence .
( 67 ) Examples .
( 1 ) Let the tangential equation be , to find the intrinsic equation .
Here we have ; Hence from equation ( 64 ) se ; . . . . . .
( 68 ) We can get a formula of reduction for this integral as follows : then by differentiation and reduction , ; , ( 69 ) which is the required formula ; and the integral will ultimately depend on known forms .
( 2 ) Let the tangential equation be that of the evolute of an ellipse , We have Hence , from equation ( 64 ) , where ; ; ( 70 ) and this is the intrinsic equation of the evolute of an ellipse .
( 3 ) To find a curve in which the radius of curvature bears a constant ratio to the normal , the given condition is expressed by the equation ; .
( 71 ) If be any even integer the integration on the right-hand side can be performed .
See WILLIAMSON'S ' Integral Calculus .
' ( 4 ) To find a culve whose tangential equation is the same as its intrinsic equation .
Here we have , or ; , ( 72 ) where stands ( 5 ) If the tangential equation of a curve be , and the intrinsic equation , find the curve .
We have ; Hence ; ( 73 ) ( 6 ) To find the intrinsic equation of the curve This is the curve whose ordinary tangential equation is or the curve whose trilinear equation is We have ; ; FORM OF TANGENTIAL EQUATION .
from equation ( 64 ) we have .
( 74 ) We reduce the firsi of these integrals to the normal form of elliptic integrals as follows:\mdash ; Let , and we find where and are connected by the equation .
( 76 ) The second integral in equation ( 74 ) may be delived from the first by changing the sign and putting for .
Hence we have at once Jco , ( 77 ) where is given by the equation ; ( 78 ) and substituting from equations ( 75 ) ( 77 ) in ( 74 ) , we get the required intrinsic equation ( 79 ) where , and are given by the equations ( 76 ) , ( 78 ) .
SECnON III.\mdash ; Transforination of the Intnnsic into the Tangential Equation .
33 .
We shall have much use to make of the intrinsic equation of the catenary in thi and in subsequent sections ; for this reason , and also on account of extremely elementary character , we give here an investigation of the leading properties of that curve .
Let be the lowest point of a uniform string AOB , suspended at the points A and , and let the tension at be denoted by , and at any other point by : Then if we consider the equilibrium of the portion Fig. 6 .
we find that the forces acting on it are , and its own weight ; and these are parallel resI ) ectively to the sides of the triangle E. Hence , by the property of the triangle of forces , is the angle which the tangent at makes with the tangent at O. Now if be the length of and the length of a portion whose weight equal to , we have , since the string is uniform , ; .
( 80 ) 34 .
The equation , which we have just obtained , is the intrinsic equation of the catenary ; we get the Cartesian equation from it as follows :Make O , and draw FX parallel to D. Then we shall take these lines as axes .
Now let the coordinates of the point be denoted by and , and we have ; but , equation ( 80 ) , Again , we have and , Hence ; ( 81 ) and this is the Cartesian equation of the catenary .
35 .
From the value FORM OF TANGENTIAL EQUATION .
we get Hence , from equation , ( 82 ) 36 .
If from the point we let fall the perpendicular on the tangent at , we have evidently ; but .
( 83 ) Again we have . . . . . .
( 84 ) Hence the locus of the point is the involute of the catenary .
37 .
From the diagram we have and cosec .
; is the length of a pertion of the string whose weight is is the length of a portion whose weight is T. 38 .
The Intrinsic Equation of a curve being given , to find its Tangential Equation .
This problem is the converse of the one solved in art .
30 , Section II .
Let be the given intrinsic equation , .
Hence from equation ( 64 ) we have ; Hence the required tangential equation is . . . . . .
Examples .
( 1 ) Find the tangential equation of the catenary .
Here .
See equation ( 80 ) .
Hence ( 86 ) ( 2 ) Find the tangential equation of the involute of the catenary ; that is , the tangential equation of the locus of the point ( see art .
33 ) .
The intrinsic equation of the involute of the catenary is Hence we have .
Hence ; and integrating the first integral by parts , we find it equal Hence , where is the constant of integration , which is evidently equal to the required tangential equation is . . . . .
( 87 ) ( 3 ) Let the intrinsic equation be .
Then we find ( 88 ) This formula fails when ; but in that case we have , and we find .
( 89 ) ( 4 ) Find the equation of a curve , being given Here we have , if ; .
See equation ( 64 ) .
FORM OF TANGENTIAL EQUATION .
or , since is an arbitrary constant , .
( 90 ) ( 5 ) It is required to find the equation of two curves A and , which are so related that the Tangential Equation of A is the same as the Intrinsic of the Tangential of the same as the Intrinsic of A. Let be curve , , B. Then by the first condition we have and by the second condition ; And by differentiating and some easy reduction we get or : Hence the multiple 4 being put to the arbitrary constant in order to avoid fractions ; This may be written ( ) being an arbitrary constant ; This is the tangential equation of the curve B. To find the equation of A we have , from equation ( ) , and subtracting this from equation ( o ) we get ; This is the intrinsic equation of , and therefore the tangential equations of A and respectively are . . . . .
( 91 ) CHAPTER SECTION I.\mdash ; Evolutes .
39 .
If the tangential equation of a curve be , we have proved , in art .
30 , .
Hence if 9 denote the radius of curvature , we have ; if be the tangential equation of a curve , the intrinsic equation of its evolute is ( 93 ) 40 .
If our movable line had been given by equation , we get in the usual manner Hence ; FORM OF TANGENTIAL EQUATION .
and , as in the last art .
, we find .
( 94 ) 41 .
The Tangential Equation of a Curve being given , to find the Tangential Equation of its Evolute .
This problem is solved by articles 38 and 39 .
For if be the tangential equation , the intrinsic equation of the evolute is .
( Art .
39 .
) Let this by , and , by art .
38 , the tangential equation is Now we have , from the value of , ; and integrating by parts we easily get .
Multiplying by , and integrating again , we get Hence the tangential equation required , . . . . .
( 95 ) 42 .
The foregoing result may be obtained very simply from geometrical considerations as follows .
In fig. 4 ( art .
26 ) the line PS is a tangent to the evolute , and the angle OSP ; then we have OS Hence if OS be taken as the directing-line , the tangential equation of the envelope of SP is ; but the envelope of SP is the evolute , and therefore we have the same result as before .
43 .
The right-hand side of equation ( 95 ) may be written PROFESSOR J. CASEY ON A NEW Hence if \amp ; c. represents for the successive evolutes what we have denoted by for the curve itself , we have ; similarly , .
Hence in general .
, ( 96 ) 44 .
Since ; and denoting this by and the corresponding functions for the evolutes by , \amp ; c. , we have , from equation ( 96 ) , ( 97 ) 45 .
In art .
26 we have found the coordinates of a point ou the curve :\mdash ; These assume , if we substitute from art .
44 for the value , the symmetrical form , ( 98 ) ; ( 99 ) and hence , from art .
44 , if we denote by the coordinates of a point on the nth evolute , , ( 100 ) .
( 101 ) 46 .
By using LEIBNITZ'S theorem , we find , from equation ( 98 ) , Hence , by equations ( 100 ) , ( 101 ) , we get .
( 10.2 ) FORM OF TANGENTIAL EQUATION .
Similarly , from equation ( 99 ) , we get .
( 103 ) 47 .
The Intrinsic Equation of a Curve given , we can find the Tangential Equation of the Evolute thus : Let be the given intrinsic equation , then the intrinsic equation of the evolute is .
( see WHEWELL , " " On the Intrinsic Equation of Curves Phil. Trans. vol. viii .
p. 659 ) ; and therefore , by art .
38 , equation ( 85 ) , . . .
( 104 ) Cor. The tangential equation of the second evolute is , ( 105 ) and , in general , of the th evolute .
( 106 ) Examples .
( 1 ) Find the tangential equation of the evolute of the catenary .
Here we have ; .
( 107 ) which is the required equation .
The following three examples are illustrations of art .
39 .
( 2 ) To find the intrinsic equation of the evolute of the curve ; ( see example 6 , Section II .
, Chapter II .
( 108 ) ( 3 ) If the curve be the lemniscate , ( see art .
25 ) , and ( see art .
39 ) .
( 109 ) is the intrinsic equation of the evolute .
( 4 ) Let the given be the equilateral hyperbola , we have ; the evolute is . . . . . .
( 110 ) next five examples are illustrations of art .
43 .
( 5 ) Let the curve be , its evolute will be .
( 111 ) The tangential equations of the successive evolutes of the curve are ; and in general , . . .
( 112 ) , . . . . .
( 113 ) where the sign to be used according as is even or odd .
( 7 ) Find the evolute of the logarithmic curve .
Cartesian equation of the curve is and the tangential is ; and therefore the tangential equation of its evolute is . . . . . .
( 114 ) ( 8 ) Let the curve be the polar one , The tangential equation is and the evolute is .
( 115 ) This result could be easily obtained geometrically .
( 9 ) tangential equation of the evolute of the curve is . .
( 116 ) Hence the tangential equation of the evolute of the common parabola is .
( I17 ) FORM OF TANGENTIAL SECTION II.\mdash ; Involutes .
48 .
From the equations in art .
43 for the successive evolutes of a curve , we can conversely infer the equations of the successive involutes : thus , let the gential equation of a curve be , the tangential equations of the successive involutes are or , as it may be written , ; and in general , for the nth involute , . . . . . . .
( 118 ) Mathematicians have recognized it as legitimate to interpret the symbol of differentiation with a negative index , as denoting integration ; therefore we may write the equation ( 118 ) as follows:\mdash ; . . . . . .
( 119 ) Hence the equation ( 96 ) includes the formulae both for evolutes and involutes , according as is regarded as positive or negative .
By an extension of the notation of art .
44 , the last equation may be written .
( 120 ) 49 .
If denote the coordinates of a point on the first involute , those of a point on the second involute , \amp ; c. , we have , . . .
( 121 ) ; ( 122 ) and , in general , , ( 123 ) .
( 124 ) 60 .
The Tangential Equation of a Curve being given , to find the Intrinsic Equation of its Involute .
This problem is solved by articles 30 and 48 .
Thus , if be the tangential equation of the involute , ( see equation ( 64 ) ) ; but by article 48 we have ; : that is , .
( 125 ) , . . . . .
( 126 ) or , as it may be written , .
Hence we have the following theorem:\mdash ; If be the tagential equation of a curve , the intrinsic equation of its involute is .
Cor. 1 .
Since is the length of the involute , is the length of the given curve .
Hence from equation ( 125 ) we have the following theorem : be the tanequation of a curve , the length of the curve is given by the equation .
( 128 ) Cor. 2 .
The equation ( 127 ) is equivalent to the following:\mdash ; ; . . . .
( 129 ) for we have proved , art .
30 , that if be the tangential equation , the intrinsic equation is and we get the intrinsic equation of the involute from this by integration .
51 .
From the intrinsic equation to find the tangential of the involute .
Let be the given equation , then the intrinsic equation of the involute is Hence from equation ( 85 ) , art .
38 , the tangential equation of the involute is .
( 130 ) FORM OF TANGENTIAD EQUATION .
Observation.\mdash ; Under each of the heads Evolute and Involute it will be observed we have solved three problems , which may be stated briefly as follows:\mdash ; Given To find Tangential equation of a curve , Tangential equation of its evolute , involute .
Tangential , , , , Intrinsic Intrinsic , , , , Tangential We have omitted the problems given the intrinsic equation of a curve to find the intrinsic equation of its evolute and involute , because these had been previously solved by WHEWELL ( see ' Cambridge Philosophical Transactions , ' nlready cited ) .
Examples .
Examples 1-3 are illustrations of art .
48 , 4 and 5 of , and 6-8 of art .
51 .
( 1 ) Let be the equation , it is required to find the involute .
From equation ( 118 ) , art .
48 , we have We can get a formula of reduction for this integral as follows:\mdash ; Put ; , ( 131 ) which is the required formula .
Cor. If be the equation of a curve , the equation of its involute is .
( 132 ) Compare equation ( 116 ) .
( 2 ) Find the involute of the curve ( that is , of the logarithmic curve ) , we have .
( 133 ) Cor. \mdash ; alog .
( 134 ) ( 3 ) Let the curve be then the tangential equation of its involute is sin . . . . .
( 135 ) Of these four integlals , the first and fourth are elementary , and the is derived from the second by in place of and changing signs .
Hence question will be solved integrate To reduce this to elliptic integrals , let , and we easily find sin Now Hence , . . . . .
( 136 ) and the question is completely solved .
( See art .
32 .
) ( 4 ) Let be the tangential equation of a curve , then ( see art .
50 ) the intrinsic equation of its involute is .
( 137 ) ( 5 ) Let ; then we find for the involute .
( 138 ) ( 6 ) Let the intrinsic equation of a curve be ; then ( see art .
51 ) the tangential equation of its involute is . . . . .
( 139 ) Similarly , if co .
( 140 ) is its involute .
FORM OF TANGENTIAL EQUATION .
( 7 ) Let .
then ; , . . . .
( 141 ) ( 8 ) Find the tangential equation of the involute of a circle .
The intrinsic equation of the circle is ; , ; the tangential equation of the involute of the circle is .
( 142 ) We can verify the foregoing result geometrically as Fig. 7 .
follows : in the annexed diagram , which represents a circle and its involute , we have , and ; and since QB is parallel to PO , we have at once OB ; that is , which proves the proposition .
OHAPTER IV .
SECTION I.\mdash ; Positive Pedals .
52 .
If we make the perpendicular to our director line the initial line , it is evident that the polar equation of the first positive pedal of the curve .
( 143 ) Hence the tangential equation of any curve is at once transformed into the polar equation of its first positive pedal by changing into , and multiplying the function on the right-hand side by Thus the tangential equation of the parabola is ( see art .
25 ) ; hence its first positive pedal is a ( 144 ) which equation repl.esents , as is well known , the cissoid .
Again , the pedal of the logarithmic curve is , ( 145 ) and of the ellipse This curve is a bicircular quartic .
53 .
The tangential equation of the evolute of the curve is .
Hence the polar equation of the first positive pedal of the evolute is . . . . . .
( 147 ) 54 .
The foregoing result can be shown geometrically as follows ( see fig. art .
) The perpendicular OT on PO is the radius vector of the pedal of the evolute ; but Cor. 1 .
The equation ( 147 ) may be written .
( 148 ) This also appears from art .
43 ; and from the same article we see that the first positive pedal of the nth evolute is . . . . . .
( 149 ) Cor. 2 .
If in the last equation be taken as negative , we have the first positive pedal of the involute .
Cor. 3 .
If and denote the vectores of the first positive pedals of the evolute and involute of , we have ; . . . . .
( 150 ) Hence we have the following theorem:\mdash ; The length of a curve is equal to the sum of the radii vectores of the first positive pedals of its evolute and involute .
FORM OF TANGENTIAL EQUATION .
Cor. 4 .
If denote the radius vector of the first positive pedal of a curve , .
( 181 ) SECTION II.\mdash ; Negative Pedals .
55 .
We have seen in art .
52 that the polar equation of the first positive pedal of a curve is obtained from its tangential equation by changing into , and multiplying the function on the right-hand side by .
Hence , conversely , we have the following theorem : be the polar equation of a curve , the tangential equation of its first negative pedal is . . . . .
( 152 ) Thus the polar equation of a parabola is ; the tangential equation of its first negative pedal is , . . . . .
( 153 ) or , in Cartesian coordinates , , . . . . .
( 154 ) showing that it is in the : parabola .
56 .
The equation of the line whose envelope is the negative pedal is Hence the points where this line meets its envelope are given by the equations , . . .
( 155 ) ; . . . .
( 156 ) and by eliminating between these equations , we get the equation of the pedal .
Cor. or . . . . . . . .
( 157 ) Hence the distance from the extremity of to where the perpendiculal to it meets its envelope is .
Examples .
( 1 ) Find the first negative pedal of the cardioide .
polar equation of this curve is , taking the perpendicular to the cuspidal tangent as the initial line , ; ; ( 158 ) Therefore the pedal is a circle .
( 2 ) Find the negative pedal of Then put the result of eliminating will be .
( 159 ) By differentiating the values of and given in equations ( 155 ) , ( 15C ) , then squaring and \amp ; c. , we get the length of the first negative pedal , ( 160 ) an equation hich agrees with equation ( 151 ) , but expressed in a different notation : 57 .
If in art .
39 we substitute for , we get , from equation ( 93 ) , .
Hence we have the follow theorem:\mdash ; If be the polar equation of a curve , the intrinsic equation of the of itsJirst negatire pedat is . . . . . .
( 161 ) In like manner , from art .
50 , the intrinsic equation of the of the first ative pedal is .
( 162 ) Cor. Thoe intrinsic equation of the nth evolute is , ( 163 ) and of nth involute is .
( 164 ) FORM OF TANGENTIAL EQUATION .
Examples .
Find the intrinsic equation of the first negative pedal of an ellipse .
polar equation of the ellipse , where Hence by art .
56 , equation ( 160 ) , the equation of its first negative pedal is or ( 165 ) and the intrinsic equation of the evolute of the pedal is .
( 166 ) 58 .
The convel.se of the problem in art .
56 is , being given the intrinsic equation of a curve to find the polar equation of its first positive pedal .
Let be the given intrinsic equation , then we bare , from equation ( 151 ) , ' .
( 167 ) Cor. 1 .
The polar equation of the positive pedal of the evolute is , ( 168 ) and of the involute .
( 169 ) Cor. The equation ( lti7 ) may be written Jsin . . . . . .
( 170 ) See BOOLE'S ' Differential Equations whele the render will find illustrations of the cases in which the symbol on the right-hand side of equation ( 167 ) may be usefully expanded in ascending powers of , and thus integration on the right-hand side rendered unnecessary , for the process then will be performed by differentiation .
Cor. 3 .
The equation ( 169 ) may be written . . . . . .
( 171 ) 59 .
Since , if we put for the symbol , we get ( 172 ) if the right-hand side of equation ( 169 ) be differentiated twice with respect to , we get the right-hand side of equation ( 168 ) .
Hence equation ( 168 ) may be written Jcos . . . . .
( 173 ) 60 .
From equation ( 160 ) art .
56 we have at once the foJlowing theorems .
If we have three polar curves given by the equations ; then , , if the colresponding lengths of their negative pedals be denoted by we shall have . . . . . .
( 174 ) .
If the corresponding lengths of the nth evolutes of their first negative be , and 2 , then .
( 175 ) Cl. To find the curve whose length bears a constant ratio to the radius vector of its first positive pedal .
The given condition is expressed by the equation ; Hence , ( 176 ) where ; This curve is the equiangular spiral ; and we infer from the form of its equation that FORM OF TANGENTIAL EQUATION .
its reciprocal with respect to a circle whose radius is is another equiangular spiral whose equation in polar coordinates is . . . . . .
( 177 ) Cor. The positive and negative pedals of equiangular spirals are also equiangular spirals , and so is the inverse .
So that every geometrical transformation of this curve is another curve of the same species .
SECTION III.\mdash ; Reciprocal Curves .
62 .
We have seen that the polar equation of the first positive pedal of the curve is ; and the reciprocal of a curve being the inverse of its first positive pedal , then the polar equation of the reciprocal of is .
( 178 ) Thus the reciprocal of the parabola is , ( 179 ) or , in Cartesian coordinates , , ( 180 ) which is another parabola , as it ought , since the centre of reciprocation is a point on the curve .
63 .
Since the value of derived from art .
178 is we infer , from art .
55 , that the equation of the first negative pedal of the curve is Hence we have the following theorem : ) be the tangential equation of a curve the procal of itsjirst positive pedal or the first negative pedal of its reciprocal is ( 181 ) 64 .
If the intrinsic equation be given , say , then we have , from equation ( 167 ) , the polar equation of its reciprocal , ( 182 ) Cor. The polar equation of the reciprocal of the th evolute is ( 183 ) 65 .
If in equation ( 182 ) we put , we find ; , ( 184 ) an equation which gives the intrinsic equation of a curve in terms of the polar equation of its reciprocal .
66 .
If in equation ( 178 ) we put , we get Hence if be the polar equation of a curve , ( 185 ) is the tangential equation of its reciprocal .
Obs.\mdash ; The problems we have solved in this section may be briefly stated thus : Given Tangential equation of a Polar , Intrinsic , Polar \ldquo ; , To find Polar equation of its reciprocal .
Tangential equation of its reciprocal .
Polar Intrinsic , ( 1 ) Let it be required to find the eciprocal of the catenary .
The intrinsic equation is ; , and , substituting in equation ( 182 ) , Hence , from equation ( 170 ) , we have Jsec ; then performing the integrations , and determining the constants by the condition FORM OF TANGENIIAL EQUATION .
: 9 must be infinite when , we have and , and the required cquation is .
( 186 ) ( 2 ) Find the reciprocal of the curve ; the tangential equation of its reciprocal is ( 3 ) Find the reciprocal of the cycloid .
The intrinsic equation is Hence , from equations ( 182 ) and ( 170 ) , siri Now it is evident that must be infinite when vanishes , and that must be equal to when .
Hence , and therefore the required curve is ( 188 ) ( 4 ) The reciprocal of the logarithmic curve is or , in Cartesian coordinates , ( 1 S9 ) ( 5 ) Find the reciprocal of the curve Here we have , from equation ( 178 ) , ; CHAPTER V. SECTION I.\mdash ; The Cycloid .
67 .
There is one curve which , though we have very seldom mentioned hitherto in our memoir , was the one which led to the discovery of its methods .
This curve is the cycloid ; and the reason it has not been more frequently used in our illustrations is that we consider its importance demands a chapter to itself .
The novelty of the methods and of most of the results is our apology for devoting so much space to its tion .
68 .
In the figure , art .
26 , it is evident that the point is the centre of instantaneous rotation for the line , because the motion of the points and are respectively at right angles to the lines LQ and PQ respectively , and since the coordinates of the point are OL and .
Hence the locus of the centres of instantaneous rotation of the line , whose position is given at any time by the quantities and , where , is the curve obtained by eliminating between the equations ( 191 ) 69 .
In the same fig. , if be at right angles to , and since , the values of will be ; and therefore the motion of the line LP will be given by supposing a curve whose equation is the system ( 192 ) to roll on the curve whose equation is the system ( 191 ) , and the line LP will be the axis of with respect to this rolling curve .
70 .
Let , then ; let be the origin , , the angle XLP ; then if be the point of contact of LP with its envelope we have , by art .
26 , the diameter Fig. 8 .
of the circle touching OX at and passing through the point .
Hence if we erect LB at right angles to OX , and PB to , the diameter LB of the circle LPB will be constant and equal to , and the arc LP of the same circle will be equal 2 ; the arc line line AB .
Hence if we make , the arc'PB FORM OF TANGENTIAL EQUATION .
will be equal to the line ; and the point may be considered as fixed in the circle LPB , and the locus of will be the curve described by a fixed point in the circle LPB rolling on the Iine AC .
In other words the locus of is a cycloid .
71 .
Since , the equations ( 191 ) denote a right line , and the equations ( 192 ) the circle .
Hence the cycloid is the envelope of a fixed diameter of the circle , which rolls along the line .
Therefore we have two methods of generating the same cycloid , either as a locus or an envelope .
72 .
The coordinates of the point are , from equations ( 46 ) , , the system From equation ( 62 ) we have the intrinsic equation .
, ( 195 ) and from ( 61 ) .
( 196 ) If we differentiate the equation we have the differential equation of the cycloid IV .
constant .
( 197 ) 73 .
From equation ( 93 ) , art .
39 , the intrinsic equation of the evolute is .
( 198 ) Hence the evolute is another cycloid .
We can show the same thing geometrically ; for we have seen that the arc line .
Hence denoting BC by , and the angle PBC by , we have , and therefore the envelope of PB is a cycloid .
Cor. If the line PB be produced to , making , then is the centre of curvature .
74 .
From ] fall the perpendicular LQ on the diameter of the revolving circle , Fig. 9 .
then it is evident that the angle XLQ ; and denoting this angle by , we have DCCCLXXVII .
or Hence the envelope of is a cycloid , and it is evident that is the point of contact .
This is the cycloid that would be described by a fixed point in the circumference of the circle , whose diameter is the line IL , rolling on the line Z. Cor. From let fall the perpendicular BW on the diameter of the rolling circle , then we have ; and therefore the locus of the point is the envelope of and it is the evolute of the cycloid described by Q. 75 .
The circle whose centre is , and which touches the line , also touches the locus of Q. Hence we have the following theorem:\mdash ; If a variable circle has its centre on a given cycloid , and if it touches the tangent at its vertex , its envelope is another cycloid .
76 .
If a variable circle touch a given cycloid , and also touch the tangent at the vertex , the locus of its centre is a cycloid .
Fig. 10 .
Or we may give a direct proof of this last theorem : let the angle XLQ be denoted by ; now we have ; hence the envelope of LQ is a cycloid .
Again , , but ; and therefore is the point of contact of LQ with its envelope , and the proposition is proved .
77 .
If be three fixed tangents to a variable } cloid , we have Hence constant .
Hence the tangent at the vertex of the cycloid is divided in a given anharmonic ratio by the three given tangents and the line infinity .
Hence we have the following theorem:\mdash ; Being given three fixed tangents to a variable cyeloid , the envelope of the tangent at the vertex is a parabola .
78 .
If four fixed tangents to a cycloid be given , the tangent at the vertex is a comm.on tangent to two parabolas .
Now being given two parabolas they have , in addition to the FORM OF TANGENTIAL EQUATION .
common tangent at infinity , three finite common tangents .
Hence we have the following theorem : \mdash ; Four lines being .
given , three cycloids can be described to touch them .
79 .
If two variable tangents to a cycloid intersect at a constant angle , and a circle be described about the triangle formed Fig. 11 .
by and the tangent at the vertex of the cycloid , then the envelope of the diameter of this circle passing through the points is a cycloid ; the envelopes of the chords passing through the same point , and through the highest and lowest points of the circle , are cloids .
Let the tangents , intersect in and let be the centre of the circle APA ' ; then since the angle APA ' is constant , is constant , AA ' is constant .
Hence the base and the vertical angle of the triangle APA ' is constant ; the diameter of the circumscribing circle is constant , and it is evident that the loci of the points are right lines parallel to AA .
Again , since is the middle point of AA ' , ( OA-OA ' ) Hence , constant ; therefore ( see art 72 , equation ( 197 ) ) the envelope of CP is a cycloid .
( 2 ) Since , we have , for finding the envelope of ; therefore it is a cycloid .
( 3 ) The angle FEP , and the envelope of FP is a cycloid .
80 .
If be the centres of the circles APA ' and TPT ' , then is endicular the tangent at the vertex of the cycloid , and equal to radius of its generating aircle .
Demonstration.\mdash ; Since , and , we have . .
PA . .
Hence the radical axis of the two circles is parallel to AA ' , and therefore CC ' is perpendicular to AA ' .
Again , the radical axis of the two circles Fig. 12 .
passes through ; hence , by a known Property of coaxal circles , the rectangle PA . .
PD ; that is , ; but .
Q. E. D. 81 .
If the angle TPT ' be constant , the locus of , the centre of circle described about the triangle formed by the two tangents and the chord of contact , is a right line .
This is evident , since CC ' and perpendicular to AA ' , and the locus of is a right line .
( See art .
79 .
) .\mdash ; Intern and Ertern Cydoids .
82 .
nition.hen the extremity of the revolving ladius of the generating circle describes a cycloid , a fixed point in the radius describes a , which , according as the point is inside or outside the circle , I shall intern or ) cycloid .
These curves are usually called the prolate and the curtate cycloid ; but the names I have adopted are more gestive .
83 .
To find the intrinsic equation of an extel.n cycloid .
Let BPL be the genel.ating circle of the cycloid , the point which describes it , and the point which describes the tern cycloid ; then denoting by , and the angles as in the diagram , we have , since , the coordinates of the point given by the equations ( 199 ) If we differentiate these equations with respect to , then square and add , we get ; FORM OF TIAL EQUATION .
where . . .
( 201 ) Hence the arc of an extern cycloid is equal in length to an elliptic arc .
If we make , we find from equation ( 201 ) ; but from the triangle we have Hence Hence we can apply formula of transformation to the function on the of equation ( 200 ) , and we get Elliptic Functions , ; Now since is the centre of instantaneous rotation , the locus of the po will be at right angles to ; that is , the tangent at will be perpendicular to and the tangent at the highest point will be perpendicular to ; hence the angle between these tangents will be equal to and herefore the transcendental equation ( 202 ) is the intrinsic equation of the extern cycloid .
84 .
If the point be inside the circle , that is , if the curve bc an intern cycloid , the formula ( 202 ) will still hold , and be the intrinsic equation of the curve ; but the modulus of the functions will be greater than unity .
A simple transformation of that rmula will give one in which is changed into its reciprocal .
If we interchange the quantities and in equation ( 200 ) , the value of remains unaltered ; hence when the point is inside , and less than , if we wish that the modulus of the functions and should beless than unity , instead of formula ( 202 ) we shall have the \mdash ; .
( 203 ) In this formula , and the angle is the angle IP'B . .
If we differentiate the equation ) we get , after some slight leduction , Hence if denote the radius of curvature of an extern cycloid , we have ( 204 ) In the figure , art .
82 , we have evidently ; ( 20o ) Hence we have the following elegant construction for the centre of curvature of an intern or extern cycloid:\mdash ; the cenfre I of the circle let fall theperpendicular IK on the normal to thoe curve , and then the third proportional to and will be the radius of curvature .
86 .
Since and , we have ( 206 ) Again , if the line , that is , the normal to the curve , meet the polar of the point , with respect to the generating circle , in the point , then the line BN is divided harmonically , and we have or .
( 207 ) Hence we have the following theorem :portion of the normal to an intern or extern ?
cloid at any point of the curve included between the polar of the point with respect to the centre of the generating circle and the corresponding centre of curvature is bisected by the centre of instantaneous 87 .
By art .
30 , equation ( 64 ) , if be the tangential equation of a curve .
; but by art .
85 , Hence ; FORM OF : TANGENTIAL EQUATION .
Hence the tangential equation of an inter or extern cycloid is .
( 208 ) 88 .
In the tnangle , art .
83 , we have : : : : ; and from the equation of the cycloid described by the point we have ; eliminating , we have for the envelope of the line ; . . . .
( 209 ) and this is the tangential equation of the evolute of an intern or extern cycloid .
89 .
The tangential equation ( 208 ) can be expressed very simply as follows .
For if we take the conic , . . . . .
( 210 ) we easily find its tangential equation to be Hence we have the following theorem:\mdash ; lf be the tangential equation of the evolute ( see equation ( 209 ) ) , and the tangential equation of the ellipse ( 210 ) , then the equation of the intern or extern cycloid is . . . . . .
( 212 ) Cor. 1 .
The intrinsic equation of the evolute of an intern or extern cycloid is .
( 213 ) Cor. 2 .
If denote the lengths of an extern or intern cycloid , its evolute , and its auxiliary conic ( see equation ( 210 ) ) , taken on the three curves from points whose tangents are parallel to other three points whose tangents also are parallel , then . . . . . .
( 214 ) .
If be the radii of curvature of the same three curves at points whose tangents are parallel , . . . . .
( 215 ) Cor. 4 .
be infinite when either or is infinite ; but will be infinite when the auxiliary conic is a hyperbola and the point of contact at infinity .
No if is less than , we hare the following theorem:\mdash ; An intern cycloid has two points of inflection , the tangents at evhich are parallel to ptotes of the auxiliary conic .
SECTION III .
90 .
If two fangenfs to a cycloid intersect at a constant angle , the locus of their point of intersech.on is an extern cycloid .
nstration.\mdash ; Since angle APA ' ( see fig. art .
79 ) between the tangents is constant , is constant , AA ' is cons.tant ; but the diameter of the circle about the triangle APA ' ; Hence CP is constant .
Again we have ( see art .
constant .
Hence if CP wele equal to , the locus of would be a cycloid ; but since is always greater than , CP is greater than , and therefore the locus of is an extern cycloid .
Lemma .
If two tangents , PT , , to any given curve be inclined at a constant angle , the aircle described about the triangle formed by the two tangents and the chord of contact touches the locus of P. Demonstration.\mdash ; Let be a consecutive point on the locus , then the tangents from touch the curve in the points .
Hence , since the angle , the quadrilateral is inscribed in a circle , and the line joining the consecutive points is a tangent to the circle .
Hence the proposition is proved .
91 .
If two tangents to a given make a given angle , the centre of the circle described about the triangle formed by the two tangents and the chord of contact is the centre of instantaneous rotation for the extern cycloid , which is the locus of the intersection of the tangents .
Demonstration.\mdash ; Since the angle is given , the locus of is an extern cycloid , and therefore , by the preceding lemma ( see fig. art .
80 ) , is normal to the locus of P. Again , since is a point in the revolving radius of a circle whose centre is and radius , and we have proved CC ' , the circle rolls on the locus of .
Hence the proposition is proved .
92 .
If the angle TPT ' be constant , the radius of the circle TPT ' is a mean FORM OF TANGENTIAL EQUATION .
tional between the radius of the circle APA ' and half the chord of curvature at passing through the centre of APA ' .
nonstration.\mdash ; Let the radii of the circles be , and the angle CPC ' be ; then , by art .
86 , we have PK : PC ' : : : .
But : : ; .
Q.E.D. CHAPTER VL SECTION I.\mdash ; Epicyeloids .
93 .
The form of tangentialequation employed in the previous portion of this memoir may be usefully generalized as follows :Thus , instead of taking a directing line OX ( see art .
1 ) and a variable line LP , making an angle with OX at the distance from the origin , let us take a directing curve OX , and a variable line , making an angle with the curve at , and denoting the arc OL by ; then any relation between and , such as , may be called tangential equation of the curve which the line LP envelopes .
Let us take a consecutive position , of , then is the point where LP touches its envelope , and Let the insic equation of the curve be , then the angle LPL ' is easily seen to be ; and if denote the diameter of the circle described about the infinitesimal , we have Hence if 9 denotes the radius of curvature of the directing curve at , we have ; Hence If denotes the length of the curve which is the envelope of from some fixed point in it up to , then ( see art .
31 ) .
Hence ; IIDCCCLXXVIL 3 ; that is , ( 216 ) If we denote the angle which LP makes with OR by , we have evidently ( 217 ) and we have also .
( 218 ) Hence eliminating and between three last equations , there will be a resulting equation between and , say .
( 219 ) and this be the intrinsic equation of the envelope of 94 .
Let the directing curve be the catenary , and let the functional symbols be the same ; then , since , we hay ; Now we have , from the intrinsic equation of the catenary , ; Hence , making these substitutions in equation ( 216 ) , and putting , we have the required intrinsic equation ( 220 ) 95 .
Let the curve be the cycloid , and let , as before , be the same , then we get the intrinsic equation . . . . .
( 221 ) a curve which we shall find to be a parallel to the cycloid .
96 .
The most application of our general equation is where the directing curve is a circle , and the relation between and is linear ; that is , , where is a constant .
It will be seen that in this simple case the envelope belongs to the class of curves known as epicycloids and hypocycloids , and it will belong to one or the other according as is positive or negative , or , what comes to the same thing , according as is positive or negative .
97 .
Let , and let the radius of the directing circle ALC be , then , from art .
93 , ; ( 222 ) FORM OF TANGENTIAL EQUATION .
but ; that is , since , the arc ; and making the arc semicircle LPD , we Fig. 16 .
have the arc .
Hence the locus of is an epicycloid described by the rolling of the circle DPL on the circle EDX .
98 .
Def. We shall call the circle DPL the generating circle of the epicycloid , and the circle EDX , on which it rolls , the base .
It is evident that the motion of the circle DPL with respect to the director circle ALC is that of pure sliding , and its motion with respect to EDX is that of pure rolling .
99 .
Since is the centre of similitude of the circles and ALC , we have ; but , from equation , ( 222 ) Again , since and arc we have- .
Hence LP : : : arc AL : arc AC ( 223 ) Hence , if a variable arc , , has one , fixed , and be divided in a given ratio in the point , the envelope of the chord LC is an epicycloid , which it touches in a point , determined by the proportion : : arc : arc AC .
100 .
If A be a fixed point , and variable points , and if .
the ratios be given arc AB : arc : arc , then , from the last article , the envelope of each side of the triangle BCD is an epicycloid touching the circle in the point A. Hence we have the following theorem :a variable polygon be inscribed in a , and if the envelopes of all the sides but one be epicycloids which have a point of contact with the circte , then envelope of the naining side is another cloid , havin.the same point of contact with the circle .
Cor. If the points of contact of the sides of the triangle BCD , with their respective envelopes , be , the three lines BB ' , CC ' , , are concurrent .
This is evident from art .
99 .
101 .
In the figure ( art .
97 ) , since the arc , then denoting XI ) by and the angle which PD makes with the circle XDE by , we have ; .
( 224 ) therefore the envelope of PD is an epicycloid , whose directing circle is the circle XDE .
Hence , the evolute of an epicycloid is another epicycloid , and the director circle of one is the base of the other .
102 .
If denote the diameter of the generating circle of the evolute , we have , as in art .
97 , .
( 225 ) But and denote respectively the diameters of the generating circles of an epicycloid and its involute .
Hence , the difference between the reciprocals of the diameters of the generating circles of an epicycloid and its involute equals reciprocal of radius of directing circle of the epicycloid .
Cor. In the equation , the constant is the diameter of the generating circle of the involute .
This follows from the present article combined with equation ( 222 ) .
It was on this account that the negative suffix was put to 103 .
In the figure ( art .
97 ) , if PD meet its envelope in , then is the centre , and the radius of curvature at ; but , and : ( 226 ) That is , the base of an epicycloid divides its radii of curvature in the constant ratio of the diameters of the generating circles of the and its evolute .
104 .
Let ( see fig. , art .
97 ) be the point where LC meets the epicycloid , which is FORM OF TANGENTIAL EQUATION .
the involute of the locus of , then from the last article we have from art .
97 ; from art .
99 .
Hence the points are harmonic conjugates to the points and C. Cor. 1 .
Every radius of curvature of an epicycloid is divided harmonically by the base of the epicycloid .
Cor. 2 .
If in the figure ( art .
100 ) be points on the involute of the three epicycloids which tou , ch the sides of the triangle BCD , corresponding to the points of contact of these epicycloids , then the are collinear .
105 .
From art .
99 we see that the arc : arc AC : : .
Hence , letting fall the perpendicular O , and denoting the angle AOE by the angle LO will be ; if , and if we denote the radius of the circle by , we have .
Hence the equation of the tangent to the epicycloid is . . .
( 227 ) For examples of the case in which the envelope of this line is an algebraic curve , see SALMON'S ' Higher Curves , ' p. 270 .
106 .
The equation ( 227 ) may be written in the form Hence if the line OX be taken as the director line , the tangential equation of the epicycloid is .
( 228 ) 107 .
In order to find the intrinsic equation we have Hence from equation ( 64 ) , art .
30 , we find : , ( 229 ) which is the required intrinsic equation .
The same result may be obtained from art .
93 , equation ( 216 ) .
Cor. If we substitute for its value we find , from art .
97 , equation ( 222 ) combined with ( 229 ) , that . . . . . .
( 230 ) Hence putting and doubling , we have the whole length of the epicycloid from cusp to cusptwice the sum of the diameters of the generating circles of the curve and its involute .
SECTION II.\mdash ; The Hypocycloil .
108 .
Having discussed at considerable length the properties of the epicycloid , we shall treat very briefly those of the hypocycloid .
In fact , analytically , the latter curve differs from the former only in the sign of a parameter ; hence the properties of one curve are with slight modifications true of the other .
The most interesting are those which are found by considering the curves in combination .
109 .
In the equation of art .
Fig 18 denote the which LP makes extelnally with the tangent to the director circle ( which comes to the same thing as to consider negative .
Now if chance its sign , since we must regard as positive , must change sign ; in other words has changed direction ) .
Also let denote the angle which LP makes with the tangent to the director circle at the origin ; then we have ; Hence , if denote the diameter of the circle LPE described about the infinitesimal triangle LL'P , we have , . . . .
( 231 ) and we find , as in art .
97 , that is , the arc ; and making the arc semicircle L will be a fixed point , and we shall have the arc EX .
Hence the locus of is the hypocycloid generated by the rolling of the circle .
EPL on the 110 .
Since , see equation ( 231 ) , and , see equation ( 222 ) , .
( 232 ) Ilence if an epicycloid and hypocycloid have the same director circle , and if the generating circles of their involutes be equal to one another , the diameter of the generating circles of their involutes is a harmonic mean between the diameters of the generating circles of the curves themselves .
111 .
From equation ( 231 ) we get and from equation ( 222 ) FORM OF TANGENTIAL EQUATION .
Hence are in harmonical progression .
Hence , with the same hypothesis as in the last article , the radius of the common director circle is a harmonic mean between the ralii of their bases .
112 .
Several propositions proved for the cycloid may with scarcely any modification of the demonstration be extended to and hypocycloids .
Thus : .
Iffrom the point where the generating circle of an epi- or hypoc.ycloid touches the base a perpendicular be ll on the revolving radius , the envelope of the perpendicular is an epi- or hypo .
cycloid . .
If the perpendicular be let fall from the point where the generating circle touches the director circle , the envelope is an epi- or hypocycloid . .
The envelof ) of the revolving radius is an epi- or hypocyctoid . .
If two mgents , PT , PT ' , to an epi- or hypocycloid meeting the director circle in the points make a constant angle , the locus of the centre of the circle about the triangle APA ' is a circle . .
The envelope of the diameter of this circle which passes through an epi- or hypocycloid . .
The envelopes of the chords passing through and through the highest or lowest points are epi- or hypocycloids .
SECTION III.\mdash ; Extern cloids .
113 .
In the same manner as we have called the curve described by a fixed point in the revolving radius of the generating circle of a cycloid an in or extern cycloid , we shall call the curve described by a fixed point in the plane of the generating circle of an epicycloid an in- or extern epicycloid according as the point is inside or outside the circumference of the circle .
Similarly we shall have an in- or extern hypocycloid ; so that the curve embraces four distinct species ; but as they differ only in the magnitude or sign of a parameter , their properties are virtually the same ; hence we shall discuss only the extern hypocycloid .
114 .
Let be the point in the radius IP ; then , since is the centre of instantaneous rotation , BP ' will be a normal to the curve , and perpendicular to will be a tangent .
The curve will have points of inflection .
This follows at once from a beautiful theorem of Professor BALL'S , Astronomer Royal of Ireland:\mdash ; " " That if a plane figure is moving in a plane according to any law , there is always a circle of points rigidly connected with it , such that three consecutive positions of each point are in a right line\ldquo ; * .
( See ' Proceedings of the Royal Irish Academy , ' December 11 , 1871 .
) Another proof be given in the course of our investigations .
Let the equation of the curve described by the point be ; * This circle is called the " " circle of infl ) ctions.\ldquo ; The theorem in the text was rinally given by sAYAItr in his ' Legons des Machines.'\mdash ; November 1877 .
then if the diameter of the generating circle be , we have , from art .
97 , Now let the angle IBP ' be denoted by and by , and we hare : BP ' But , if hence Again , we have ; sin . . . . . .
( 233 ) And this is the equation of the evolute of the extern epicycloid .
115 .
From the diagram we have ; Now let a consecutive position of the line intersect in the point , and the arc OB in the point ; then if denote the diameter of the circle described about the infinitesimal triangle BB'Q , we have and from equation ( 233 ) we get ; .
Hence .
( 234 ) If the polar of the point with lespect to the generating circle meet in , we have , see art .
86 , equation ( 207 ) , Hence , from equation ( 234 ) , if be perpendicular to , we have .
( 235 ) FORM OF TANGENTIAL EQUATION .
Therefore BQ is half the harmonic mean between and .
; and this gives a geometric construction for the centre of curvature at the point 116 .
From equation ( 235 ) we get Now if we make , the point would be the centre of curvature of an extern cycloid , see art .
86 .
Hence by subtraction .
( 236 ) That is , the distance between the centres of curvature of an extern cyctoid and epicycloid is a third proportional to the lines and .
This vanishes , as it ought , when is at inity .
117 .
From equation ( 234 ) we get the value of , that is , of ; thus and BP ' Hence , remembering that , we get ; that is , the radius of curvature at .
( 237 ) 118 .
The following geometrical expression for the radius is remarkable for its simplicity and symmetry .
From equation ( 235 ) , BK ' .
BN ' and from art .
86 , ; .
BN ' .
( 238 ) Cor. 1 .
The anharmonic ratio of the four points , .
( 239 ) Cor. 2 . .
( 240 ) 119 .
If be a point of inflection , the radius of curvature at will be infinite , and thel.efore the denominator of the fraction on the right of equation ( 237 ) will vanish ; and therefore .
( 241 ) Hence must be negative .
But is negative : or the point must lie between and , and therefore is less than ; that is , the curve must be an inter epicycloid ; and , from what we have proved , we see that the angle must be obtuse .
Again , we hare and ; and therefore the following conditions for determining a point of inflection:\mdash ; 1st .
The angle must be obtuse .
2nd .
BK : : : from equation ( 241 ) ; that is , : : CI : .
Hence the triangle can bc constructed ; and it follows that there are two points of inflection in each revolution of the generating circle .
120 .
Since ( see art .
114 ) , and also we have ; but .
( 242 ) Again radius of curvatule , ; .
( 243 ) Now , if denote the corresponding arc of the extel.n cycloid , we have , from art .
85 , ; ( 244 ) This formuia , which connects the extern epicycloid with the extern cycloid , may be also obtained as follows , for we have evidently FORM OF TANGENTIAL EQUATION .
; But , ; ( 245 ) but ( see art .
83 ) , .
( 246 ) The same result as before .
CHAPTER VII .
SECTION I.\mdash ; Parallel Curves .
121 .
The intercept which a parallel at the distance from the movable line at either side makes on the directing line is , the choice of sign depending on the position of the parallel with respect to the origin .
Hence we have the following theorem:\mdash ; If be the tangential equation of a curve , the tangential equation of a parallel curve at the distance is . . . . . . . .
( 247 ) Thus the parallel to the parabola is , . . . . .
( 248 ) and the parallel to the cissoid is .
( 249 ) 122 .
By the method of art .
26 we get the coordinates of a point on the parallel curve to be , . . . . .
; ( 251 ) ESSOR J. CASEy ON A NEW and eliminating between these equations , we have the Cartesian equation of the parallel curve .
123 .
Since the ordinary tangential equation of a curve is the envelope of the line , the tangential equation of the parallel curve is the envelope of .
Hence we have the following theorem:\mdash ; If the tangential equation of a curve is the equation of the parallel curve is . . . . . .
( 252 ) 124 .
If the result of the last article be expanded by TAYLOR'S theorem , it may be written in the form ; or , cleared of radicals , . . . . . .
( 253 ) Hence the class of the parallel curve is twice the class of the original , and is independent of the sign of .
This shows the fgures got by taking plus and minus are both included in the equation of the parallel curve .
125 .
As in art .
30 we get for the intrinsic equation of the parallel curve* . . . . . .
( 254 ) Cor. From the value of we see that the radius of curvature of the parallel curve differs from the radius of curvature of the original curve by the quantity , as is otherwise evident .
Examples .
( 1 ) Find the intrinsic equation of the parallel to the curve the function on the right being the elliptic integral of the species .
* The following is an elegant focal property of parallel curves :single focus of the original curve is a double focus of the parallel curve .
Dcmonstration .
Let a tangent from a point I meet the original curvs in two consecutive points ; then if bo the centres of two circles , of which passes through , the line IP will be a normal to each , and therefore a normal to any curve of which these circles are generators .
Now let the point I be one of the circular points at infinity ; and since the parallel to any curve is the envelope of a circle of constant radirs whose centre moves along the given curve , the line IP will be a normal to the parallel curve at I , and therefore a tangent at I ; hence if two tangents be drawn to the original curve from the circular ints at infinity , these tangents will touch the parallel curve at the circular points .
Hence the theorem is proved.\mdash ; November 1877 .
FORM OF TANGENTIAL EQUATION .
Here we have ; .
( 255 ) ( 2 ) Find the intrinsic equation of the parallel to the rurve or .
In this case ; and we find , as before , .
( 256 ) ( 3 ) The Cartesian equation of the curve parallel to the parabola is , by the equation of art .
122 , the result of eliminating between the equations This problem may be solved more easily by finding the envelope of the line or , what is the same thing , the line Writing this in the form we get , by a known method , the required envelope to be .
( 257 ) Cor. The characteristics of this curve are .
( 258 ) We shall find the reciprocal of this curve in a future articJe .
( 4 ) The envelope of a fixed tangent to the generating circle of a cloid is a parallel to another cycloid whose generating circle has a diameter equal to the diameter of the former .
Fig. 20 .
Let PQ be the tangent at the point fixed in the revolving circle ; then , since is the centre of instantaneous rotation , , the foot of the perpendicular from on PQ , will be the point of contact of with its envelope .
From I let fall the perpendicular IE on ; then it is easy to see that the locus of the point will be a cycloid whose generating circle will have IB for diameter , and EB will be the normal at to the cycloid .
Now ; since ladius of generating circle , we see that the locus of is a parallel to the cycloid .
The same result may be shown thus:\mdash ; It is evident that ian ; the equation of the curve which is the envelope of LP is and the intrinsic equation of this is , . . . . .
( 259 ) which is a parallel to a cycloid .
Cor. 1 .
The envelope of any line in rigid connexion with the generating circle of a is a parallel to another cycloid . .
In like manner the envelope of afixed tangent to , or of any line in rigid connexion with , the erating circle of an cycloid is a parallel to another epicycloid .
126 .
From the equation ( 254 ) we infer that if be the intrinsic equation of a curve , the equation of the parallel to it is ; and hence ( see art .
64 , equation ( 182 ) ) the polar equation of the reciprocal of the parallel curve is , . . . . .
( 260 ) or Jsin .
( 261 ) In this equation we have used as the numerator to on the left side of the instead of of recent articles in order to avoid confusion of notation .
FORM OF TANGENTIAL EQUATION .
127 .
the reciprocal of the parallel curve is found at once from the tangential equation .
For it is evident that the polar equation of the reciprocal of the parallel to the curve is , ( 262 ) where is the distance between the curve and its parallel , and is the radius of the circle of reciprocation .
Examples .
( 1 ) To find the reciprocal of the parallel to a parabola .
We have is the required equation .
This , expressed in Cartesian coordinates , is .
( 263 ) This curve has three double points , namely , the origin and the points where the conic meets the line at infinity .
Again , the curve is evidently the envelope of the conic . . . . . .
( 264 ) The discriminant of this conic is This shows that there are two values of , for which the conic breaks up into a pair of lines ; hence the curve has four double tangents .
Therefore the characteristics of the curve are ( 2 ) Find the reciprocal of the parallel to the curve From equation ( 37 ) , art .
24 we have at once the polar equation of the reciprocal .
( 265 ) ( 3 ) Find the intrinsic equation of the lemniscate .
The tangential equation is ( see art .
25 ) and , by the method of art .
30 , we find the intrinsic equation to be .
( 266 ) If we put , this equation becomes or .
( 267 ) ( 4 ) Find the reciprocal of the parallel to an ellipse .
Here is the required equation .
This curve , in Cartesian coordinates , is This curve can by linear transformation be changed into a bicircular quartic .
For writing the equation in full by putting in place of , and then into , we get .
( 269 ) 128 .
To find the reciprocal of a bicircular quartic , with respect to one of its circles of inversion .
The following method of generating these curves is given in my memoir on " " Bicir cular Quartics\ldquo ; Transactions Royal Irish Academy , xxiv .
Let be a conic , called the focal conic , a circle ; then if from the centre of the circle we let fall a perpendicular OT on any tangent to , and take two points , , in opposite directions from on OT , such that the locus of the points is a bicircular quartic .
Now , denoting OT by , and OP by , we get from this construction or .
( 270 ) FORM OF TANGENTIAL EQUATION .
Again , since OP .
OP ' , the points are inverse points with respect to the circle and the perpendicular through to the line OP will be the polar of ; therefore the envelope of this perpendicular will be the leciprocal of the bicircular qualtic .
Now , let be the angle which the perpendicular makes with the axis of , or the directing line , and intercept , then we have , and Therefore the tangential equation of the reciprocal of the bicircular quartic is . . .
( 271 ) Cor. The equation ( 271 ) is also the first negative pedal of a bicircular quartic .
129 .
If we divide the equation ( 271 ) by , and then change into and into , we get , after a slight reduction , which is in the ordinary form of tangential equations .
130 .
From the equation ( 271 ) it is plain that to each value of there are two values of .
This is otherwise evident ; for erecting the PK and to OP Fig. 21 .
and OP these perpendiculars will make intercepts on the director line , which will be the required values of .
Let be the centre of the generating circle , then will be a point on the focal conic , and , CP ' will be normals to the quartic , and will be tangents .
Now if be the point of contact of with its envelope , then the angle , and therefore OK ' is parallel to , and OK to CP .
Hence , drawing from the point two parallels to the normals at , they will meet in the points of contact of these lines vrith their envelopes , and they will intersect the tangents perpendicularly in the points Cor. 1 .
The locus of the points is evidently the first positive pedal of the bicircular quartic .
Cor. 2 .
The first positive pedal of a bicircular quartic is the inverse of its first negative pedal ; for evidently .
OK ' .
OP ' NDCOCLXXVII .
Cor. 3 .
OK'-OKdiameter of generating circle of the .
quartic ; for , denoting the angle by , we have OK ' ; OK'\mdash ; diameter of generating circle .
SECTION II.\mdash ; Rectification of Bicircular Quartics .
131 .
through the point ( see diagram , art .
130 ) we draw a consecutive line OQQ ' , then the perpendiculars to this line at the points will pass through the points , and we have eyidently Hence , denoting the elements , PQ of the quartic by and , we have ( 273 ) if denotes radius CP of the generating circle ( see Cor. 3 , art .
130 ) .
132 .
Mr. .
ROBERTS showed , in Journal , vol. .
p. 194 , " " Sir les arcs des Lignes Aplanetiques that the difference between two arcs of a Cartesian oval is expressed by an arc of an ellipse ; and Professor showed some time afterwards , in Annali , ' that the arc of a Cartesian oval is the sum of three elliptic arcs .
We propose in this section to extend these theorems to bicirculars in general .
We will show that GENOCCHI'S theorem is an immediate inference from ROBERTS'S , and that each is only a particular case of a more general theorem which holds for all bicirculars , and which can be expressed in terms of the radii of the generating circles of these curves .
In order that we may not have to be referring to other wlitings , we shall investigate briefly the leading properties of these curves , referring for a fuller discussion to the author 's memoir on Bicirculars .
133 .
In art .
128 , equation ( 270 ) , we have the polar equation of a bicircular quartic .
This , expressed in Cartesian coordinates , is . . . . . .
This equation is the envelope of the conic , . . . .
where represents the expression , and the circle Now the discriminant of the equation ( 275 ) is , ( 276 ) a biquadratic equation showing that there are four pairs of double tangents .
If the four values of be denoted by , we have the equations of the four pairs of double tangents to the bicircular ( see my memoir " " On Bicirculars art .
47 ) .
These pairs of lines are , \amp ; c. ; FORM OF TANGENTIAL EQUATION .
and , from the same article , the double points of these pairs of lines are the four centres of inversion of the quartic .
Since one value of is obviously in the foregoing equation , we see that the pair of double tangents drawn from the centre of the circle of inversion will , when that centre is taken as origin , be Hence , if the other centres be taken respectively as origin , the equations of the other pairs of double tangents will be , ( 277 ) , ( 278 ) . . . .
( 279 ) Now since the pair of lines are the asymptotes of the reciprocal of the couic , we infer that the pairs of lines ( 277 ) , ( 278 ) , ( 279 ) are the asymptotes of the reciprocals of the other focal conics .
Hence we have the following system as the equations of these conics:\mdash ; , ( 280 ) .
( 282 ) Hence the four focal conics of a bicircular quartic are confocal .
134 .
Since the equation ( 276 ) may be written in the form and this is the discriminant of ( where and have the values in art .
; see SALMON'S 'Conics , ' p. 324 ) , we infer that the same values of which will make a pair of lines will make a pair of lines ; the two pairs of lines will have the same double point , their equations referred to that point as origin being , ( 283 ) . . . . . . . . .
( 284 ) Hence we have the following theorem:\mdash ; If and be a corresponding focal conic and circle of inversion of a bicircular quartic , and if be the roots of the cubic which is the discriminant of , then if be given in its canonical form , the equations of the other three focal conics are got from this by changing respectively into ; and 136 .
When represents a pair of lines , the coordinates of the double point are , by the usual process , if referred to the centre of as origin , and if referred to the centre of as origin .
Hence we have the following theorem:\mdash ; If and be a corresponding focal conic and circle of inversion of a bicircular quartic , and if be the three roots of the equation which is the discriminant of , then the coordinates of the centres of the three other circles of inversion ( 285 ) ' ( 286 ) ' ( 287 ) 136 .
Being given , and , the equation of the quartic is .
( 288 ) Again , being iven the equation of the same quartic is .
( 289 ) In order to compare the equations ( 288 ) and ( 289 ) , which represent the same curve , they must be referred to the same origin ; we will therefore transform ( 288 ) to the same origin as ( 289 ) , and we get .
( 290 ) FORM OF TANGENTIAL EQUATION .
Since the equations ( 289 ) and ( 290 ) represent the same curve , their absolute terms must be the same .
Now if the absolute term in the latter equation be reduced by means of the relation , it becomes and the absolute term in equation ( 289 ) is ; hence we get ( 291 ) That is , the sum of the squares of the radii of and equals square of the distance between their centres .
Hence and cut eaoh other orthogonally .
137 .
The propositions established in articles 133-136 are those which we shall require for the present investigation .
They are proved in the memoir already cited , but by another method .
It is useful to recapitulate them here:\mdash ; ( 1 ) A bicircular quartic is the envelope of a variable circle whose centre moves on a given conic , called the focal conic , and which cuts a given circle orthogonally .
( 2 ) The circle is a circle of inversion of the quartic .
( 3 ) There are four circles of inversion and four focal conics .
( 4 ) The four focal comics are confocal .
( 5 ) The four circles of inversion are mutually orthogonal .
( 6 ) The centres of the circles of inversion are such that any three will form a selfconjugate triangle with respect to the circle which has the fourth for centre ; in other words , the four centres form the angular points and the point of intersection of perpendiculars of a plane triangle .
138 .
The proposition proved in art .
131 is our fundamental one for the rectification of bicirculars ; it will be seen that it is a generalization of Mr. theorem already referred to .
On Fig. 22 .
account of its importance we will here give an elementary proof of it , but under a slightly different enunciation .
If OPP ' be any line cutting a circle in the points , then if two circles passing through touch in the points respectively , the difference between their diameters is equal to the diameter of J. Demonstration.\mdash ; Let be the centre of J. Join , and produce them to meet the line LOM drawn perpendicular to OP .
Join .
Now , evidently , ; P'M\mdash ; PLP'Kdiameter of J. Hence the proposition is proved .
Cor. If the point be within the circle , we shall have the sum of the diameters equal the diameter of J. 139 .
If we denote the diameter of by , and if a line ( see last diagram ) infinitely near OP make an angle with OP , then and Hence by art .
138 we have .
( 292 ) .
If the point be inside we have . . . . .
( 293 ) 140 .
If four circles be mutually orthogonal , and if any figure be inverted with respect to each of the four circles in succession , the fourth inversion will coincide with the original figure* .
Fig. 23 .
Demonstration.\mdash ; It will plainly be sufficient to prove the proposition for a single point , for the general proposition will then follow .
Since the four circles are mutually orthogonal , their four centres will form the angular points and the intersection of the perpendiculars of a plane triangle .
Let them be the points ; CO produced will intersect AB perpendicularly in , and the squares of the radii of the four circles will be equal to the four rectangles AB .
AD , BA .
BD , -CO OD , .
OD , one of the circles being imaginary , namely the one the square of whose radius is -CO OD .
Now let be the point we operate on , and let be its inverse with respect to the circle , and the inverse of with respect to the circle B. Juin and CP meeting in P Now since is the inverse of with respect to the circle the square of whose radius is AB AD , we have the rectangle AB .
The triangle ADP is similar to the triangle AP'B , therefore the angle ADPangle AP'B ; in like manner , the angle angle AP'B , therefore the triangles ADP and are equiangular , and the rectangle AD rectangle P. DP Again , because is the intersection of the perpendiculars of the triangle ABC , the rectangle AD .
OD ; hence CD DP and the angles CDP and being the com* An important extension of this theorem can be got by combining it with the folowing proposition , which is proved in art .
95 of my memoir on ' Cyclides and Sphero-Quartics\ldquo ; :\mdash ; If a sphero-quartic be projected on one of the planes of circular section of any quadric passing through it by lines parallel to the greatest or least axis of the quadric , the projection will be a bicircular quartic whose centres of inversion will be the projection of the centres of inversion of the sphero-quartic .
The extension is as follows .
There exists in quartics a series of inscribed quadrilaterals ABCD , whose sides AB , , DA , taken in order , pass through the es of the four cones of the sphero-quartic .
FORM OF TANGENTIAL EQUATION .
plements of equal angles are equal : therefore the triangles and CDP are equiangular , and ) angle ; hence the four points are concyclic , and therefore the point is the inverse of with respect to the imaginary circle the square of whose radius is -CO OD , and whose centre is the point O. Again , the angle ODP ; therefore the points are concyclic , and is the inverse of with respect to the circle whose centre is , and the square of whose radius is the rectangle CD OD .
Hence the proposition is proved .
Cor. 1 .
If the point be at infinity , the point will coincide with , and the point will be the reflection of with respect to the line AB . .
If the points be the of inversion of a bicircular quartic , and if the point be on the curve , the points will also be on the curve .
141 .
Let the radii of the generating circles of the bicircular quartic which touch it at the four pairs of points , , , be denoted by respectively .
Let the angle which the line APP ' makes with any fixed line in the plane , say the axis of , be.denoted by , and the angles which the lines , CPP make with the same line by Now if the points describe infinitesimal arcs , we have ( see art .
139 ) , denoting these arcs by \amp ; c. , Hence ; . . . . .
( 294 ) ( 295 ) Hence the arc of a bicircular quartic is the sum of four similar integrals .
We shall find that each of them is expressed in terms of elliptic integrals .
This theorem is our generalization of ROBERTS'S and GENOCCHJ'S theorems* .
* The following proof of the theorem art .
138 will apply equally to sphero-quartics , and will lead to an important extension of the theorem of this article:\mdash ; Let be two oonsecutive tangents to the focal conic of the bicircular quartic ( or , in the case of a sphero-quartic , to the focal sphero-comics ) , and OPP ' , OQQ ' two perpendiculars to ( see fig. art .
130 ) .
If intersect the generating circle in the points , it is evident , from geometrical considerations , that Rn ' .
But for cular quartics and for sphero-quartics ; hence , remembering the theorem in the footnote to art .
140 , we have , for ephero-quartics , and the rectification of sphero-quartics is reduced to elliptic integrals .
November 1877 .
142 .
If the bicircular becomes a Cartesian oval , the point will be at infinity , and we shall have ( see cor .
1 , art .
140 ) , Hence and ; . . . . .
( 296 ) . . . . . . .
( 297 ) This is GENOCCHI'S theorem .
Cor. By integrating the equation we get , ( 298 ) which is ROBERTS'S theorem .
143 .
To reduce the integral to the normal form of elliptic integlals .
Let the focal conic of the bicircular be , and the corresponding circle of inversion .
The equation of a to the conic is or say ; therefore if be the point of contact , we have Now if be the eccentric angle , ; ; Hence ( 299 ) Again , since S is the radius of the generating circle whose centre is the point on the focal conic , and which cuts the circle orthogonally , we have FORM OF ]iNTIAL EQUATION .
Put ; then , and .
Making these substitutions , we get , ( 301 ) where stands for the quartic . . . .
144 .
In order to reduce still furthel the expression ( 301 ) , we must decompose into simpler fractions , or say the fraction into simpler fractions .
Let us , for shortness , put for the expression king this substitution , and denoting the eccentlicity of the conic by , we find , by dividing \amp ; c. , Then decomposing the fraction still remaining , and substituting , we get , after some reduction , ( 303 ) Since is a quartic function of the variable , each of these integrals belongs to the domain of the elliptic integrals .
( See ' Elliptic Functions XDCCCLXXVn .
3 145 .
If the bicircular quartic be a Cartesian oval , the focal conic is a circle ; and taking the line joining the centres of and as the axis of , we may write their equations in the forms and we get ; ( S04 ) and this represents an arc of an ellipse .
Hence ROBERTS'S and GENOCCHI'S theorems are proved .

