Forum Geometricorum

Volume 10 (2010) 165–173.

FORUM GEOM
ISSN 1534-1178

Characterizations of Bicentric Quadrilaterals

Martin Josefsson

Abstract. We will prove two conditions for a tangential quadrilateral to be

cyclic. According to one of these, a tangential quadrilateral is cyclic if and only
if its Newton line is perpendicular to the Newton line of its contact quadrilateral.

1. Introduction
A bicentric quadrilateral is a convex quadrilateral with both an incircle and a
circumcircle. One characterization of these quadrilaterals is obtained by combining
the most useful characterizations of tangential and cyclic quadrilaterals, that the
consecutive sides a, b, c and d, and angles A, B, C and D satisfy
a + c = b + d,
A + C = B + D = π.
We review a few other characterizations of bicentric quadrilaterals before proving
two possibly new ones.

D
Y
C

Z
X

A B
W

Figure 1. The tangency chords and diagonals

If the incircle in a tangential quadrilateral ABCD is tangent to the sides AB,

BC, CD and DA at W , X, Y and Z respectively, then the segments W Y and XZ
are called the tangency chords in [8, pp.188-189]. See Figure 1. In [4, 9, 13] it is
proved that a tangential quadrilateral is cyclic if and only if the tangency chords
are perpendicular.
Problem 10804 in the M ONTHLY [14] states that a tangential quadrilateral is
cyclic if and only if
AW DY
= .
WB YC
Publication Date: December 21, 2010. Communicating Editor: Paul Yiu.
166 M. Josefsson

Using the same notations, Hajja proved in [11] that a tangential quadrilateral is
cyclic if and only if
AC AW + CY
= .
BD BX + DZ
If E, F , G and H are the midpoints of W X, XY , Y Z and ZW respectively
(see Figure 2), then the tangential quadrilateral ABCD is cyclic if and only if
the quadrilateral EF GH is a rectangle. This characterization was Problem 6 on
China Western Mathematical Olympiad 2003 [5, pp.182-183].

D
Y
C
G
F
Z
X

H
E

A B
W

Figure 2. ABCD is cyclic iff EF GH is a rectangle

2. Two characterizations of right triangles

To prove one of the characterizations of bicentric quadrilaterals we will need the
following characterization of right triangles. The direct part of the theorem is an
easy exercise 1, but we have found no reference of the converse result.
Theorem 1. In a non-isosceles triangle the median and altitude to one of the sides
divide the opposite angle into three parts. This angle is a right one if and only if
the angle between the median and the longer of the sides at the considered vertex
is equal to the angle between the altitude and the shorter side at that vertex.

Proof. We use notations as in Figure 3. If C = π2 , we shall prove that α = β.

Triangle AM C is isosceles 2 with AM = CM , so A = α. Triangles ACB and
CHB are similar, so A = β. Hence α = β.
Conversely, if α = β, we shall prove that C = π2 . By the exterior angle theorem,
angle CM B = A + α, so in triangle M CH we have
π
A+α+γ = . (1)
2
1
A similar problem also including the angle bisector can be found in [1, pp.46-49] and [12, p.32].
2The midpoint of the hypotenuse is the circumcenter.
Characterizations of bicentric quadrilaterals 167

α β
γ
m

A x B
M H

Figure 3. Median and altitude in a triangle

Let x = AM = BM and m = CM . Using the law of sines in triangles CAM

and CM B,
sin α sin A x sin α
= ⇔ =
x m m sin A
and with α = β,
sin (α + γ) sin B x sin ( π2 − A) cos A
= ⇒ = π =
x m m sin ( 2 − α) cos α
π
since B + α = 2 in triangle BCH. Combining the last two equations, we get
sin α cos A
= ⇔ sin 2α = sin 2A.
sin A cos α
This equation has the two solutions 2α = 2A and 2α = π − 2A, hence α = A or
α = π2 − A. The second solution combined with B + α = π2 gives A = B, which
is impossible since the triangle is not isosceles by the assumption in the theorem.
Thus α = A is the only valid solution. Hence
π
C = α+γ +β = A+γ+α =
2
according to (1), completing the proof. 
Corollary 2. Let CM, CD and CH be a median, an angle bisector and an altitude
respectively in triangle ABC. The angle C is a right angle if and only if CD
bisects angle HCM .

α β

A B
M D H

Figure 4. Median, angle bisector and altitude in a triangle

168 M. Josefsson

Proof. Since CD is an angle bisector in triangle ABC, we have (see Figure 4)

α + ∠M CD = ∠HCD + β. (2)
Using Theorem 1 and (2), we get
π
C= ⇔ α=β ⇔ ∠M CD = ∠HCD.
2


3. Corollaries of Pascal’s theorem and Brocard’s theorem

Pascal’s theorem states that if a hexagon is inscribed in a circle and the three
pairs of opposite sides are extended until they meet, then the three points of inter-
section are collinear. A proof is given in [6, pp.74-75]. Pascal’s theorem is also
true in degenerate cases.
In [7, p.15], the following theorem is called Brocard’s theorem: if the extensions
of opposite sides in a cyclic quadrilateral intersect at J and K, and the diagonals
intersect at P , then the circumcenter O of the quadrilateral is also the orthocenter
in triangle JKP (see Figure 5). An elementary proof of this theorem can be found
at [16].
K

C
P

O
A J
B

Figure 5. Brocard’s theorem

To prove our second characterization of bicentric quadrilaterals we will need

two corollaries of these theorems that are quite well known. The first is a special
case of Pascal’s theorem in a quadrilateral. If the incircle in a tangential quadri-
lateral ABCD is tangent to the sides AB, BC, CD and DA at W , X, Y and Z
respectively, then in [9] Yetti 3 calls the quadrilateral W XY Z the contact quadri-
lateral.
3Yetti is the username of an American physicist at the website Art of Problem Solving [3].
Characterizations of bicentric quadrilaterals 169

D L
Y
Z C

A J
W B

Figure 6. Pascal’s theorem in a tangential quadrilateral

Corollary 3. If the extensions of opposite sides in a tangential quadrilateral inter-

sect at J and K, and the extensions of opposite sides in its contact quadrilateral
intersect at L and M , then the four points J, L, K and M are collinear.
Proof. Consider the degenerate cyclic hexagon W W XY Y Z, where W and Y are
double vertices. The extensions of the sides at these vertices are the tangents at W
and Y , see Figure 6. According to Pascal’s theorem, the points J, L and M are
collinear.
Next consider the degenerate cyclic hexagon W XXY ZZ. In the same way the
points M , K and L are collinear. This proves that the four points J, L, K and M
are collinear, since M and L are on both lines, so these lines coincide. 
Corollary 4. If the extensions of opposite sides in a tangential quadrilateral inter-
sect at J and K, and the diagonals intersect at P , then JK is perpendicular to the
extension of IP where I is the incenter.

Proof. The contact quadrilateral W XY Z is a cyclic quadrilateral with circumcen-

ter I, see Figure 7. It is well known that the point of intersection of W Y and
XZ is also the point of intersection of the diagonals in the tangential quadrilat-
eral ABCD, see [10, 15, 17]. If the extensions of opposite sides in the contact
quadrilateral W XY Z intersect at L and M , then by Brocard’s theorem M L⊥IP .
According to Corollary 3, M L and JK are the same line. Hence JK⊥IP . 

4. Two characterizations of bicentric quadrilaterals

Many problems on quadrilaterals in text books and on problem solving web sites
are formulated as implications of the form: if the quadrilateral is a special type
(like a bicentric quadrilateral), then you should prove it has some property. How
170 M. Josefsson

D L
Y
Z C
P

X
I

A J
W B

Figure 7. Perpendicular lines JK and IP

about the converse statement? Sometimes it is concidered, but far from always.
The two characterizations we will prove here was found when considering if the
converse statement of two such problems are also true. The first is a rather easy
one and it would surprise us if it hasn’t been published before; however we have
been unable to find a reference for it. Besides, it will be used in the proof of the
second characterization.
Theorem 5. Let the extensions of opposite sides in a tangential quadrilateral in-
tersect at J and K. If I is the incenter, then the quadrilateral is also cyclic if and
only if JIK is a right angle.

Proof. We use notations as in Figure 8, where G and H are the midpoints of the
tangency chords W Y and XZ respectively and P is the point of intersection of
W Y and XZ. In isosceles triangles W JY and XKZ, IJ⊥W Y and IK⊥XZ.
Hence opposite angles IGP and IHP in quadrilateral GIHP are right angles, so
by the sum of angles in quadrilateral GIHP ,
π
∠JIK = ∠GIH = 2π − 2 · − ∠W P Z.
2
Hence we have
π π
∠JIK = ⇔ ∠W P Z = ⇔ W Y ⊥XZ
2 2
and according to [4, 9, 13] the tangency chords in a tangential quadrilateral are
perpendicular if and only if it is cyclic 4. 
Now we are ready for the main theorem in this paper, our second characteri-
zation of bicentric quadrilaterals. The direct part of the theorem was a problem
4This was also mentioned in the introduction to this paper.
Characterizations of bicentric quadrilaterals 171

D
Y
C
Z
H P
X
I
G

A J
W B

Figure 8. ABCD is cyclic iff JIK is a right angle

studied at [2]. The Newton line 5 of a quadrilateral is the line defined by the mid-
points of the two diagonals.
Theorem 6. A tangential quadrilateral is cyclic if and only if its Newton line is
perpendicular to the Newton line of its contact quadrilateral.

D
Y L
C
Z
H P
F X
N
I G
E

A J
W B

Figure 9. The Newton lines in ABCD and W XY Z

5It is sometimes known as the Newton-Gauss line.

172 M. Josefsson

Proof. We use notations as in Figure 9, where P is the point where both the diag-
onals and the tangency chords intersect (see [10, 15, 17]) and L is the midpoint of
JK. If I is the incenter, then the points E, I, F and L are collinear on the Newton
line, see Newton’s theorem in [7, p.15] (this is proved in two different theorems in
[1, p.42] 6 and [17, p.169]). Let M be the intersection of JK and the extension of
IP . By Corollary 4 IM ⊥JK. In isosceles triangles ZKX and W JY , IK⊥ZX
and IJ⊥W Y .
Since it has two opposite right angles (∠IHP and ∠IGP ), the quadrilateral
GIHP is cyclic, so ∠HGI = ∠HP I. From the sum of angles in a triangle, we
have
∠IN G = π − (∠GIF + ∠HGI) = π − (∠JIL + ∠HP I)
where N is the intersection of EF and GH. Thus
π  π
∠IN G = π − ∠JIL − − ∠HIP = − ∠JIL + ∠KIM.
2 2
So far we have only used properties of tangential quadrilaterals, so
π
∠IN G = − ∠JIL + ∠KIM
2
is valid in all tangential quadrilaterals where no pair of opposite sides are parallel7.
Hence we have
π π
EF ⊥GH ⇔ ∠IN G = ⇔ ∠JIL = ∠KIM ⇔ ∠JIK =
2 2
where the last equivalence is due to Theorem 1 and the fact that IM ⊥JK (Corol-
lary 4). According to Theorem 5, ∠JIK = π2 if and only if the tangential quadri-
lateral is also cyclic.

D Y C

G
Z X
E F
H

A B
W

Figure 10. An isosceles tangential trapezoid

It remains to concider the case when at least one pair of opposite sides are par-
allel. Then the tangential quadrilateral is a trapezoid, so 8
A + D = B + C ⇔ A − B = C − D.
6That the incenter I lies on the Newton line EF is actually a solved problem in this book.
7Otherwise at least one of the points J and K do not exist.
8
We suppose without loss of generality that AB  CD.
Characterizations of bicentric quadrilaterals 173

The trapezoid has a circumcircle if and only if

A + C = B + D ⇔ A − B = D − C.
Hence the quadrilateral is bicentric if and only if
C − D = D − C ⇔ C = D ⇔ A = B,
that is, the quadrilateral is bicentric if and only if it is an isosceles tangential trape-
zoid. In these EF ⊥GH (see Figure 10, where EF  AB and GH⊥AB) com-
pleting the proof. 

References
[1] T. Andreescu and B. Enescu, Mathematical Olympiad Treasures, Birkhäuser, Boston, 2004.
[2] April (username), Bicentric quadrilateral, Art of Problem Solving, 2008,
http://www.artofproblemsolving.com/Forum/viewtopic.php?t=216084
[3] Art of problem Solving, Olympiad geometry forum,
http://www.artofproblemsolving.com/Forum/viewforum.php?f=4&
[4] M. Bataille, A duality for bicentric quadrilaterals, Crux Math. 35 (2009) 310–312.
[5] X. Bin and L. P. Yee (editors), Mathematical Olympiad in China, East China Normal University
Press, 2007.
[6] H. S. M. Coxeter and S. L. Greitzer, Geometry revisited, Math. Assoc. Amer., 1967.
[7] D. Djukić, V. Janković, I. Matić and N. Petrović, The IMO Compendium, Springer, 2006.
[8] H. Dörrie, 100 Great Problems of Elementary Mathematics, Dover Publications, 1965.
[9] Fermat-Euler (username), Prove that cc1 bisects, Art of Problem Solving, 2005,
http://www.artofproblemsolving.com/Forum/viewtopic.php?t=55918
[10] D. Grinberg, Circumscribed quadrilaterals revisited, 2008, available at
http://www.cip.ifi.lmu.de/∼grinberg/CircumRev.pdf
[11] M. Hajja, A condition for a circumscriptible quadrilateral to be cyclic, Forum Geom., 8 (2008)
103–106.
[12] R. Honsberger, In Pólya’s Footsteps, Math. Assoc. Amer., 1997.
[13] M. Josefsson, Calculations concerning the tangent lengths and tangency chords of a tangential
quadrilateral, Forum Geom., 10 (2010) 119–130.
[14] A. Sinefakopoulos and D. Donini, Problem 10804, Amer. Math. Monthly, 107 (2000) 462;
solution, ibid., 108 (2001) 378.
[15] K. Tan, Some proofs of a theorem on quadrilateral, Math. Mag., 35 (1962) 289–294.
[16] K. Vittas, Art of Problem Solving, 2006, post no 8 at
http://www.artofproblemsolving.com/Forum/viewtopic.php?t=110887
[17] P. Yiu, Euclidean Geometry Notes, Florida Atlantic University Lecture Notes, 1998.

Martin Josefsson: Västergatan 25d, 285 37 Markaryd, Sweden

E-mail address: [email protected]