b

Forum Geometricorum
Volume 7 (2007) 1117. b b
FORUM GEOM
ISSN 1534-1178

On a Porism Associated with the Euler

and Droz-Farny Lines

Christopher J. Bradley, David Monk, and Geoff C. Smith

Abstract. The envelope of the Droz-Farny lines of a triangle is determined to

be the inconic with foci at the circumcenter and orthocenter by using purely
Euclidean means. The poristic triangles sharing this inconic and circumcircle
have a common circumcenter, centroid and orthocenter.

1. Introduction
The triangle ABC has orthocenter H and circumcircle . Suppose that a pair
of perpendicular lines through H are drawn, then they meet the sides BC, CA,
AB in pairs of points. The midpoints X, Y , Z of these pairs of points are known
to be collinear on the Droz-Farny line [2]. The envelope of the Droz-Farny line
is the inconic with foci at O and H, known recently as the Macbeath inconic, but
once known as the Euler inconic [6]. We support the latter terminology because of
its strong connection with the Euler line [3]. According to Goormaghtigh writing
in [6] this envelope was first determined by Neuberg, and Goormaghtigh gives an
extensive list of early articles related to the Droz-Farny line problem. We will not
repeat the details since [6] is widely available through the archive service JSTOR.
We give a short determination of the Droz-Farny envelope using purely Eu-
clidean means. Taken in conjunction with Aymes recent proof [1] of the existence
of the Droz-Farny line, this yields a completely Euclidean derivation of the enve-
lope.
This envelope is the inconic of a porism consisting of triangles with a common
Euler line and circumcircle. The sides of triangles in this porism arise as Droz-
Farny lines of any one of the triangles in the porism. Conversely, if the orthocenter
is interior to , all Droz-Farny lines will arise as triangle sides.

2. The Droz-Farny envelope

Theorem. Each Droz-Farny line of triangle ABC is the perpendicular bisector of
a line segment joining the orthocenter H to a point on the circumcircle.

Proof. Figure 1 may be useful. Let perpendicular lines l and l
through H meet
BC, CA, AB at P and P
, Q and Q
, R and R
respectively and let X, Y , Z be
the midpoints of P P
, QQ
, RR
.
The collinearity of X, Y , Z is the Droz-Farny theorem. Let K be the foot of
the perpendicular from H to XY Z and produce HK to L with HK = KL. Now
the circle HP P
has center X and XH = XL so L lies on this circle. Let M , M

be the feet of the perpendiculars from L to l, l
. Note that LM HM
is a rectangle
Publication Date: January 16, 2007. Communicating Editor: Paul Yiu.
12 C. J. Bradley, D. Monk and G. C. Smith

Q

O
R N Y
H
T

M
B
P
P X C
M
K

Q
Z
L

R

Figure 1. The Droz-Farny envelope

so K is on M M
. Then the foot of the perpendicular from L to the line P P
(i.e.

BC) lies on M M
by the Wallace-Simson line property applied to the circumcircle
of P P
H. Equally well, both perpendiculars dropped from L to AB and CA have
feet on M M
. Hence L lies on circle ABC with M M
as its Wallace-Simson line.
Therefore XY Z is a perpendicular bisector of a line segment joining H to a point
on the circumcircle.

Note that K lies on the nine-point circle of ABC. An expert in the theory of
conics will recognize that the nine-point circle is the auxiliary circle of the Euler
inconic of ABC with foci at the circumcenter and orthocenter, and for such a
reader this article is substantially complete. The points X, Y , Z are collinear
and the line XY Z is tangent to the conic inscribed in triangle ABC and having
O, H as foci. The direction of the Droz-Farny line is a continuous function of
the direction of the mutual perpendiculars; the argument of the Droz-Farny line
against a reference axis increases monotonically as the perpendiculars rotate (say)
anticlockwise through , with the position of the Droz-Farny line repeating itself
as increases by 2 . By the intermediate value theorem, the envelope of the Droz-
Farny lines is the whole Euler inconic.
We present a detailed discussion of this situation in 3 for the lay reader.
On a porism associated with the Euler and Droz-Farny lines 13

Incidentally, the fact that XY is a variable tangent to a conic of which BC, CA

are fixed tangents mean that the correspondence X Y is a projectivity between
the two lines. There is a neat way of setting up this map: let the perpendicular
bisectors of AH, BH meet AB at S and T respectively. Then SY and T X are
parallel. With a change of notation denote the lines BC, CA, AB, XY Z by a,
b, c, d respectively; let e, f be the perpendicular bisectors of AH, BH. All these
lines are tangents to the conic in question. Consider the Brianchon hexagon of
lines a, b, c, d, e, f . The intersections ae, f b are at infinity so their join is the line
at infinity. We have ec = S, bd = Y , cf = T , da = X. By Brianchons theorem
SY is parallel to XT .

3. The porism

O
N
Z

H Ot
U

B C

J V

Figure 2. A porism associated with the Euler line

In a triangle with side lengths a, b and c, circumradius R and circumcenter O,

the orthocenter H always lies in the interior of a circle center O and radius 3R
since, as Euler showed, OH 2 = 9R2 (a2 + b2 + c2 ).
We begin afresh. Suppose that we draw a circle with center O and radius R in
which is inscribed a non-right angled triangle ABC which has an orthocenter H,
so OH < 3R and H is not on .
This H will serve as the orthocenter of infinitely many other triangles XY Z
inscribed in the circle and a porism is obtained. We construct these triangles by
choosing a point J on the circle. Next we draw the perpendicular bisector of HJ,
and need this line to meet again at Y and Z with XY Z anticlockwise. We can
certainly arrange that the line and meet by choosing X sufficiently close to A,
14 C. J. Bradley, D. Monk and G. C. Smith

B or C. When this happens it follows from elementary considerations that triangle

XY Z has orthocenter H, and is the only such triangle with circumcircle and
vertex X. In the event that H is inside the circumcircle (which happens precisely
when triangle ABC is acute), then every point X on the circumcircle arises as a
vertex of a triangle XY Z in the porism.
The construction may be repeated to create as many triangles ABC, T U V ,
P QR as we please, all inscribed in the circle and all having orthocenter H, as
illustrated in Figure 2. Notice that the triangles in this porism have the same cir-
cumradius, circumcenter and orthocenter, so the sum of the squares of the side
lengths of each triangle in the porism is the same.
We will show that all these triangles circumscribe a conic, with one axis of
length R directed along the common Euler line, and with eccentricity OH R . It fol-
lows that this inconic is an ellipse if H is chosen inside the circle, but a hyperbola
if H is chosen outside.
Thus a porism arises which we call an Euler line porism since each triangle in
the porism has the same circumcenter, centroid, nine-point center, orthocentroidal
center, orthocenter etc. A triangle circumscribing a conic gives rise to a Brianchon
point at the meet of the three Cevians which join each vertex to its opposite contact
point.
We will show that the Brianchon point of a triangle in this porism is the isotomic
conjugate Ot of the common circumcenter O.
In Figure 2 we pinpoint Ot for the triangle XY Z. The computer graphics system
CABRI gives strong evidence for the conjecture that the locus of Ot , as one runs
through the triangles of the porism, is a a subset of a conic.
It is possible to choose a point H at distance greater than 3R from O so there
is no triangle inscribed in the circle which has orthocenter H and then there is no
point J on the circle such that the perpendicular bisector of HJ cuts the circle.

The acute triangle case. See Figure 3. The construction is as follows. Draw AH,
BH and CH to meet at D, E and F . Draw DO, EO and F O to meet the sides
at L, M , N . Let AO meet at D and BC at L . Also let DO meet at A . The
points M , N , E , F , B and C are not shown but are similarly defined. Here
A
is the midpoint of BC and the line through A
perpendicular to BC is shown.

3.1. Proof of the porism. Consider the ellipse defined as the locus of points P
such that HP + OP = R, where R the circumradius of . The triangle HLD is
isosceles, so HL + OL = LD + OL = R; therefore L lies on the ellipse.
Now OLB = CLD = CLH, because the line segment HD is bisected by
the side BC. Therefore the ellipse is tangent to BC at L, and similarly at M and
N . It follows that AL, BM , CN are concurrent at a point which will be identified
shortly.
This ellipse depends only on O, H and R. It follows that if T U V is any triangle
inscribed in with center O, radius R and orthocenter H, then the ellipse touches
the sides of T U V . The porism is established.
On a porism associated with the Euler and Droz-Farny lines 15

A A

M
O

B L A
L C

D D

Figure 3. The inconic of the Euler line porism

Identification of the Brianchon point. This is the point of concurrence of AL, BM ,

CN . Since O and H are isogonal conjugates, it follows that D and A are reflec-
tions of D and A in the line which is the perpendicular bisector of BC. The same
applies to B , C , E and F with respect to other perpendicular bisectors. Thus
A D and AD are reflections of each other in the perpendicular bisector. Thus L
is the reflection of L and thus A
L = A
L . Thus since AL , BM , CN are
concurrent at O, the lines AL, BM and CN are concurrent at Ot , the isotomic
conjugate of O.

The obtuse triangle case. Refer to Figure 4. Using the same notation as before,
now consider the hyperbola defined as the locus of points P such that |HP
OP | = R. We now have HL OL = LD OL = R so that L lies on the
hyperbola.
Also A LB = CLD = HLC, so the hyperbola touches BC at L, and the
argument proceeds as before.
It is a routine matter to obtain the Cartesian equation of this inconic. Scaling so
that R = 1 we may assume that O is at (0, 0) and H at (c, 0) where 0 c < 3 but
c = 1.
16 C. J. Bradley, D. Monk and G. C. Smith

M
N

A E A

B L C
O

D B

Figure 4. The Euler inconic can be a hyperbola

The inconic then has equation

4y 2 + (1 c2 )(2x c)2 = (1 c2 ). (1)
When c < 1, so H is internal to , this represents an ellipse, but when c > 1 it
represents a hyperbola. In all cases the center is at (2c , 0), which is the nine-point
center.
One of the axes of the ellipse is the Euler line itself, whose equation is y = 0.
We see from Equation (1) that the eccentricity of the inconic is c = OH R and of
course its foci are at O and H. Not every tangent line to the inconic arises as a side
of a triangle in the porism if H is outside .

Areal analysis. One can also perform the geometric analysis of the envelope us-
ing areal co-ordinates, and we briefly report relevant equations for the reader in-
terested in further areal work. Take ABC as triangle of reference and define
u = cot B cot C, v = cot C cot A, w = cot A cot B so that H(u, v, w) and
O(v + w, w + u, u + v). This means that the isotomic conjugate Ot of O has
co-ordinates


1 1 1
Ot , , .
v+w w+u u+v
The altitudes are AH, BH, CH with equations wy = vz, uz = wx, vx = uy
respectively.
On a porism associated with the Euler and Droz-Farny lines 17

The equation of the inconic is

(v + w)2 x2 + (w + u)2 y 2 + (u + v)2 z 2 2(w + u)(u + v)yz
2(u + v)(v + w)zx 2(v + w)(w + u)xy = 0. (2)
This curve can be parameterized by the formulas:
(1 + q)2 1 q2
x= ,y = ,z = , (3)
v+w w+u u+v
where q has any real value (including infinity). The perpendicular lines l and l

through H may be taken to pass through the points at infinity with co-ordinates
((1 + t), t, 1) and ((1 + s), s, 1) and then the Droz-Farny line has equation
(sw + tw 2v)(2stw sv tv)x (sw + tw + 2(u + w))(2stw sv tv)y
+(sw + tw 2v)(2st(u + v) + sv + tv)z = 0. (4)
In Equation (4) for the midpoints X, Y , Z to be collinear we must take
v(tw + u + w)
s= . (5)
w(t(u + v) + v)
If we now substitute Equation (3) into Equation (4) and use Equation (5), a discrim-
inant test on the resulting quadratic equation with the help of DERIVE confirms
the tangency for all values of t.
Incidentally, nowhere in this areal analysis do we use the precise values of
u, v, w in terms of the angles A, B, C. Therefore we have a bonus theorem: if
H is replaced by another point K, then given a line through K, there is always a
second line through K (but not generally at right angles to it) so that XY Z is a
straight line. As the line l rotates, l
also rotates (but not at the same rate). How-
ever the rotations of these lines is such that the variable points X, Y , Z remain
collinear and the line XY Z also envelops a conic. This affine generalization of the
Droz-Farny theorem was discovered independently by Charles Thas [5] in a paper
published after the original submission of this article. We happily cede priority.

References
[1] J.-L. Ayme, A Purely Synthetic Proof of the Droz-Farny Line Theorem, Forum Geom., 4 (2004)
219224.
[2] A. Droz-Farny, Question 14111, Educational Times, 71 (1899) 8990.
[3] L. Euler, Solutio facili problematum quorundam geometricorum difficillimorum, Novi Comm.
Acad. Scie. Petropolitanae, 11 (1765); reprinted in Opera omnia, serie prima, Vol. 26 (ed. by
A. Speiser), (n.325) 139-157.
[4] A. M. Macbeath, A Compactness Theorem for Affine Equivalence-Classes of Convex Regions,
Canad. J. Math., 3 (1951) 5461.
[5] C. Thas, The Droz-Farny theorem and related topics, Forum Geom., 6 (2006) 235240.
[6] V. Thebault, J. R. Musselman and R. Goormaghtigh, Solution to problem 3758. Amer. Math.
Monthly, 44 (1937) 668672.

Christopher J. Bradley and David Monk, c/o Geoff C. Smith: Department of Mathematical Sci-
ences, University of Bath, Bath BA2 7AY, England
E-mail address: [email protected]