5 21 PDF
5 21 PDF
5 21 PDF
CHARACTERIZATIONS OF CYCLIC
QUADRILATERALS
MARTIN JOSEFSSON
1. Introduction
If a quadrilateral is inscribed into a circle so that all four vertices lie
on the circle, it is most often referred to as a cyclic quadrilateral and the
vertices are said to be concyclic. This circle is called the circumcircle and
its center, which is equidistant from the vertices, is called the circumcenter.
Other names that have been used for this class of quadrilaterals are inscribed
quadrilateral, concyclic quadrilateral and chordal quadrilateral.
In textbooks covering Euclidean geometry and on mathematical problem
solving websites, necessary conditions of cyclic quadrilaterals are often given
in theorems and exercises. Then it is assumed that the quadrilateral is
cyclic and a relation between angles or lines or distances shall be proved.
In contrast to what many textbook authors and problem creators seem to
realize or care about, it is common also for the converse theorem to be
true. This means that if the relation we just referred to holds in a convex
quadrilateral, then it can be concluded that it is a cyclic quadrilateral.
In this paper we shall prove 19 such necessary and sufficient conditions.
Many of these are not so well known, and several are probably previously
unknown characterizations. We have references for all of the direct theorems
except for three; it is the converses for about half of the conditions that we
consider to be new. In fact, it’s very difficult to find any textbook, paper,
or website covering more than just three necessary and sufficient conditions
for cyclic quadrilaterals. Thus collecting a large number of characterizations
of cyclic quadrilaterals with proofs is the primary goal for this paper. ——
——————————–
Keywords and phrases: Cyclic quadrilateral, Convex quadrilateral,
Characterization, Necessary and sufficient condition, Converse
(2010)Mathematics Subject Classification: 51M04
Received: 10.01.2019. In revised form: 20.02.2019. Accepted: 18.02.2019.
6 Martin Josefsson
The second theorem is a simple corollary to the first of the five basic
characterization we reviewed in the previous section.
Theorem 3.2. In a convex quadrilateral ABCD,
∠A − ∠B = ∠BAC − ∠ABD
if and only if ABCD is a cyclic quadrilateral.
Proof. Since ∠A = ∠BAC + ∠DAC and ∠B = ∠ABD + ∠DBC (see
Figure 2), we have that ABCD is cyclic if and only if
∠DAC = ∠DBC ⇔ ∠A − ∠BAC = ∠B − ∠ABD
and the relation follows.
The direct part of the next theorem is almost the same as Problem 574
on the wonderful website [7] by Antonio Gutierrez.
8 Martin Josefsson
Figure 3. ABCD is cyclic iff the angle bisectors at E and P are parallel
The direct part of the following theorem was a problem in [3, p. 134].
Theorem 3.4. In a convex quadrilateral ABCD, let the perpendicular to
AB at A meet CD in A′ , and the perpendicular to CD at C meet AB in
C ′ . The line A′ C ′ is parallel to BD if and only if ABCD is cyclic.
Proof. The line A′ C ′ is parallel to diagonal BD if and only if ∠CDB =
∠CA′ C ′ (see Figure 4). Quadrilateral AC ′ CA′ is cyclic by construction since
it has two opposite right angles. Thus ∠CAC ′ = ∠CA′ C ′ . Quadrilateral
ABCD is cyclic if and only if ∠CDB = ∠CAB = ∠CAC ′ = ∠CA′ C ′ .
diagonals are called the bimedians of the quadrilateral. To prove this char-
acterization of cyclic quadrilaterals was Problem 5 on the third round of the
2001 Brazilian Mathematical Olympiad [1]. Another proof than ours was
given in [12, p. 19].
Theorem 4.1. The four maltitudes of a convex quadrilateral are concurrent
if and only if it is a cyclic quadrilateral.1
Proof. (⇒) If the quadrilateral is cyclic, we shall prove that the maltitudes
are concurrent. We label the midpoints of the sides as Ma , Mb , Mc and Md .
Let the bimedians Ma Mc and Mb Md intersect at G, the circumcircle have
center O, and let Q be a point on the extension of OG such that OG = GQ
(see Figure 6). Then Ma G = GMc since G is the intersection of the diagonals
in a parallelogram. We also have ∠OGMc = ∠QGMa , so triangles OGMc
and QGMa are congruent (SAS). Thus ∠GMc O = ∠GMa Q and so the lines
OMc and Ma Q are parallel. But OMc is perpendicular to the side CD,
hence so is the extension of Ma Q. In the same way the extensions of Mb Q,
Mc Q and Md Q are perpendicular to DA, AB and BC respectively, so all
four maltitudes are concurrent at Q.
1
Their point of intersection is called the anticenter of the cyclic quadrilateral since it is
the reflexion of the circumcenter in the centroid (the point where the bimedians intersect).
Characterizations of cyclic quadrilaterals 11
The direct part of the following theorem was Problem 334 at [7].
Theorem 5.2. In a convex quadrilateral ABCD where the diagonals in-
tersect at P , let A′ and C ′ be the projections of A and C respectively on
diagonal BD, and let B ′ and D′ be the projections of B and D respectively
on diagonal AC. Then
A′ P · P C ′ = B ′ P · P D′
if and only if ABCD is a cyclic quadrilateral.
12 Martin Josefsson
(⇒) Now assume ABCD is cyclic. Then ∠ADB = ∠ACB and ∠CAD =
∠CBD, so from equations (2) and (3) we get
AA′ BB ′ BC BB ′ B′P
(4) = ⇒ = = ,
DA BC DA AA′ A′ P
DD ′ CC ′ BC CC ′ P C′
(5) = ⇒ = = .
DA BC DA DD′ P D′
Hence
B′P P C′
= ⇒ A′ P · P C ′ = B ′ P · P D′ .
A′ P P D′
(⇐) Conversely, if A′ P · P C ′ = B ′ P · P D′ , we will prove that ABCD is
cyclic. Using (3), we get
B′P P C′ BB ′ CC ′
= ⇒ = ⇒ AA′ · CC ′ = BB ′ · DD′ .
A′ P P D′ AA′ DD′
Using (2), this implies
(6) DA sin ∠ADB · BC sin ∠CBD = BC sin ∠ACB · DA sin ∠CAD.
From the sum of angles in triangles ADP and BCP , we have ∠ADB =
π − (∠P + ∠CAD) and ∠ACB = π − (∠P + ∠CBD) where ∠P = ∠AP D =
∠BP C. Thus equation (6) can be rewritten as
sin (P + ∠CAD) sin (P + ∠CBD)
= .
sin ∠CAD sin ∠CBD
Characterizations of cyclic quadrilaterals 13
The direct part of the next theorem was a problem in [9, p. 255], Problem
573 at [7], Problem 483 (a) in [5], and the first geometry problem ever to
appear in the American Mathematical Monthly (in 1894) according to [5].
It seems that the converse was considered for the first time in [5].
Theorem 6.2. If the extensions of the sides of a convex quadrilateral in-
tersect at E and F , then the angle bisectors at E and F are perpendicular
if and only if it is a cyclic quadrilateral.
2And thus not a parallelogram, rhombus, rectangle or square, which are special cases
of trapezoids when using inclusive definitions.
14 Martin Josefsson
In [8, p. 109] it was given as a part of Exercise 106 to prove the following
characterization of cyclic quadrilaterals (formulated differently).
Theorem 6.4. Let the extensions of opposite sides in a convex quadrilateral
ABCD intersect at E and F . The Miquel point of the complete quadrilateral
ABECF D lies on the line EF if and only if ABCD is a cyclic quadrilateral.
16 Martin Josefsson
The direct part of the last theorem was Problem 4 at the competition
Tournament of the Towns in the fall of 2015 for senior A-level [2].
Theorem 6.6. Let the extensions of opposite sides in a convex quadrilateral
intersect at E and F , and the midpoints of the diagonals be M and N . Then
∠EM F + ∠EN F = π
if and only if it is a cyclic quadrilateral.
Proof. (⇒) In a cyclic quadrilateral ABCD, the pairs of triangles ECA
and EBD, and also F AC and F BD are similar (AA), so ∠EM A = ∠EN D
and ∠F M C = ∠F N D (see Figure 14). Hence
π − ∠EM C = ∠EN F + ∠F N D ⇒ π = ∠EN F + ∠EM C + ∠F M C
and we get π = ∠EN F + ∠EM F .
(⇐) We do an indirect proof of the converse. If the vertex A in a con-
vex quadrilateral ABCD lies inside the circumcircle to triangle BCD, then
∠EM A < ∠EN D and ∠F M C > ∠F N D. Hence
π − ∠EM C < ∠EN F + ∠F N D < ∠EN F + ∠F M C
and we get that π < ∠EN F + ∠EM F . In the case when the vertex A is
outside the circle, this inequality is reversed, completing the proof.
18 Martin Josefsson
7. Concluding remark
Was this the end? What about Ptolemy’s theorem some readers might
object. Yes, it was deliberately excluded, but not forgotten. There will be a
sequel to this paper covering for instance necessary and sufficient conditions
of cyclic quadrilaterals concerning the diagonals.
Theorem A.3. Any two opposite angles of a convex quadrilateral are sup-
plementary if and only if it is a cyclic quadrilateral.
Proof. (⇒) If quadrilateral ABCD is cyclic with circumcenter O and if
the angles at the center are B and D, then by the inscribed angle theorem
B = 2∠B and D = 2∠D. Thus 2(∠B+∠D) = B+D = 2π so ∠B+∠D = π.
The other equality ∠A+∠C = π can be derived in the same way, or it follows
from the sum of angles in a quadrilateral.
(⇐) If ∠B + ∠D = π we can prove that the quadrilateral is cyclic by
an indirect proof. Assume that it is not cyclic, but triangle ABC has a
circumcircle. The vertex D lies either outside this circle or inside it. Assume
this time it is inside. Extend CD to D′ on the circle (see Figure 17). Then
ABCD′ is a cyclic quadrilateral and by the direct part of the theorem, we
get ∠B + ∠D′ = π. That means ∠D′ = ∠D, but according to the exterior
angle theorem ∠D = ∠D′ + ∠DAD′ , which is a contradiction. In the same
20 Martin Josefsson
way it can be proved that D is not outside the circle, hence it is on it and
ABCD is cyclic.
Proof. (⇒) We consider only the case where the chords intersect outside
the quadrilateral; the other case is similar. In a cyclic quadrilateral ABCD,
∠CAB = ∠CDB according to Theorem A.1, so triangles CAP and BDP
are similar (AA) since the angle at P is common to these triangles (see
Figure 18). Then
AP PC
(9) =
DP PB
and it follows that AP · P B = DP · P C.
(⇐) If AP ·P B = DP ·P C in a convex quadrilateral, then (9) holds. Since
the angle at P is common to triangles CAP and BDP , they are similar
(SAS). Thus ∠CAB = ∠CDB, and so ABCD is a cyclic quadrilateral
according to the converse in Theorem A.1.
Characterizations of cyclic quadrilaterals 21
References
[1] 23-rd Brazilian Mathematical Olympiad 2001, The IMO Compendium Group, avail-
able at https://imomath.com/othercomp/Bra/BraMO01.pdf
[2] 37th International Mathematics Tournament of the Towns, Senior A-Level Paper,
Fall 2015, available at
http://www.math.toronto.edu/oz/turgor/archives/TT2015F_SAproblems.pdf
[3] N. Altshiller-Court, College Geometry, Barnes & Nobel (1952). New edition by Dover
Publications (2007).
[4] X. Bin and L. P. Yee (editors), Mathematical Olympiad in China, East China Normal
University Press (2007).
[5] S. Collings, D. Sokolowsky and N. Tan, Problem 483, Crux Math., 6 (1980) 226–230.
[6] C. V. Durell and A. Robson, Advanced Trigonometry, G. Bell and Sons, London
(1930). New edition by Dover Publications, Mineola (2003).
[7] A. Gutierrez, Geometry from the Land of the Incas, accessed 27 December 2018,
http://gogeometry.com/problem/index.html
[8] J. Hadamard, Lessons in geometry: I. Plane geometry, American Mathematical So-
ciety (2008).
[9] J. L. Heilbron, Geometry Civilized. History, Culture, and Technique, Oxford Univer-
sity press (1998).
[10] R. Honsberger, Episodes in Nineteenth and Twentieth Century Euclidean Geometry,
MAA (1995).
[11] R. Honsberger, Mathematical Morsels, MAA (1978).
[12] B. Micale and M. Pennisi, On the altitudes of quadrilaterals, Int. J. Math. Educ. Sci.
Technol., 36 (2005) 15–24.
[13] D. Monk, New Problems in Euclidean Geometry, UKMT (2009).
[14] A. Ostermann and G. Wanner, Geometry by its History, Springer (2012).
[15] R. Villemoes, Baltic Way 2002–2006 Problems and Solutions, Department of Math-
ematical Sciences. University of Aarhus, 2007, available at
http://www.balticway07.dk/data/ekstra/bw02-06-online-a5.pdf
[16] C. Vincze and L. Kozma, College Geometry, 2014, available at
http://math.unideb.hu/media/vincze-csaba/collegegeom-main-1.pdf