Complete Quadrilaterals

Helia Kalanaki
Contents
1 Introduction 1
1.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Complete Quadrilaterals 4
2.1 Spiral Similarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Newton-Gauss Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Miquel Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3 Relationship with the Projective Geometry 10

3.1 Harmonic Bundles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Pole and Polar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 Brocard Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1 Introduction
1.1 Preliminaries

Glossary
cline A cline (or generalized circle) refers to either a circle or a line.. 2
concurrent Two or more lines are said to be concurrent if they intersect in a single
point.[8]. 4

disjoint Two sets are said to be disjoint if there are no common elements. In other
words, when the intersection of the two sets is empty, then those sets are said to be
disjoint sets.[1]. 10

orthocenter The point of concurrence of the altitudes of a triangle is called the

orthocenter of the triangle.. 6, 13

Definition 1.1 (Power of a Point) We define the power of P with respect

to the circle ω by
P owω (P ) = OP 2 − r2 .
This quantity may be negative. Actually, the sign allows us to detect whether
P lies inside the circle or not. [4]

Definition 1.2 (radical axis) The radical axis of two circles is a line that is
the locus of all points that have equal powers with respect to both circles. [6]

Definition 1.3 (coaxial) If a set of circles have the same radical axes, then
we say they are coaxial. [4]

Definition 1.4 (Homothety) A homothety h is a transformation defined by

a center O and a real number k. It sends a point P to another point h(P ),
multiplying the distance from O by k. The number k is the scale factor. It
is important to note that k can be negative, in which case we have a negative
homothety. [4]

A homothety h with center O acting on 4ABC:

 O
A

h(A)

B C

h(B)
h(C)

A negative homothety with center O:

h(C)

A
O B

h(B) C
h(A)

Definition 1.5 (Inversion) Inversion is useful for turning circles into lines
and for handling tangent figures. The idea is to view every line as a circle with
infinite radius. We add a special point P∞ to the plane, which every ordinary
line passes through (and no circle passes through). This is called the point at
infinity. Therefore, every choice of three distinct points determines a unique
cline—three ordinary points determine a circle, while two ordinary points plus
the point at infinity determine a line. With this said, we can now define an
inversion. Let ω be a circle with center O and radius R. We say an inversion
about ω is a map (that is, a transformation) which does the following.
-The center O of the circle is sent to P∞ .
-The point P∞ is sent to O.
-For any other point A, we send A to the point A∗ lying on ray OA such that
OA × OA∗ = r2 .
Try to apply the third rule to A = O and A = P∞ , and the motivation for the
2
first two rules becomes much clearer. The way to remember it is “ r0 = ∞” and
r2
“∞ = 0”.
At first, this rule seems arbitrary and contrived. What good could it do? First,
we make a few simple observations.
 1. A point A lies on ω if and only if A = A∗ . In other words, the points of ω are
fixed.
2. Inversion swaps pairs of points. In other words, the inverse of A∗ is A itself.
In still other words, (A∗ )∗ = A. [4]

O A A*

Lemma 1.6 (Inversion and Tangents) Let A be a point inside ω, other

than O, and A∗ be its inverse. Then the tangents from A∗ to ω are collinear
with A.

This configuration is shown in the figure above. We just need to check that OA.OA∗ =
r2 . Inversion is not very interesting if we only look at one point at a time
-how about two points A and B?

A*

A
O

B*

This situation is shown in the figure above. Now we have some more structure.
Because OA.OA∗ = OB.OB∗ = r2 , by power of a point we see that quadrilateral
ABB*A* is cyclic. Hence we obtain the following theorem.

Theorem 1.7 (Inversion and Angles) If A∗ and B∗ are the inverses of A and
B under inversion centered at O, then ∠OAB = −∠OB∗A∗ .
2 Complete Quadrilaterals

Definition 2.1 (Complete Quadrilateral) A complete quadrilateral is the

gfigure determined by four lines no threei of which are concurrent.
f The most
common configuration in which a quadrilateral ABCD and one takes the inter-
sections E = AB ∩ CD and F = DA ∩ BC. [2] In addition, the points could be
split into three pairs such that the connecting segments do not belong to any of
the given lines. These three segments are called diagonals of the quadrilateral.
[3] [5] h

D k

C B
F

For instance the diagram above is the complete quadrilateral ABCDEF has three di-
agonals: AC, BD, and EF .(Because, according to the definition above, these segments
are the only ones which do not belong to any of the lines AD, BC, AB,and DC.)

2.1 Spiral Similarity

Definition 2.2 (Spiral Similarity) A spiral similarity centered at a point O

is a composition of a dilation and a rotation with respect to O. [2]

B C A0
B0

A C0

The most commonly occurring case of a spiral similarity is between two segments.
Consider a spiral similarity at O mapping a segment AB to CD. [4]
4OAB is similar to 4OCD. We now determine O in terms of A, B, C, D via complex
numbers. It is easy to check that
c−o d−o
= .
a−o b−o
That implies
ad − bc
o= .
a+d−b−c
So O is uniquely determined by A, B, C, D. That implies in general there is exactly
one spiral similarity taking any segment to any other segment. The exception is if
ABDC is a parallelogram, since then a + d = b + c and the spiral similarity fails to
exist.

Lemma 2.3 (Spiral Centers) Let AB and CD be segments, and suppose X =

AC ∩ BD. If (ABX) and (CDX) intersect again at O, then O is the center of the
unique spiral similarity taking AB into CD. [4]

Proof This is actually just a matter of angle chasing. We have

∠OAB = ∠OXB = ∠OXD = ∠OCD

and similarly
∠OBA = ∠ODC.
That implies 4OAB ∼ 4OCD, which is sufficient.

Whenever all six points in the figure appear, we automatically have a pair of similar
triangles. By now, an observant reader may have realized that there is more than
one set of similar triangles in the figure. We see that in fact, 4OAC ∼ 4OBD as
AO BO
well. After all, ∠AOC = ∠BOD and CO = DO (the ratios arising from the original
spiral similarity). What this means is that spiral similarities occur in pairs. More
precisely, we get the following proposition.

Lemma 2.4 The center of the spiral similarity taking AB to CD is also the
center of the spiral similarity taking AB to CD. [4]

Thus we have a second spiral similarity, but this time we know its center. What
happens if the Spiral Centers Lemma is applied again, this time in the other direc-
tion? Does this really mean that AB ∩ CD lies on (AOC) and (BOD) as well? That
is precisely Miquel’s theorem.

2.2 Newton-Gauss Line

 Definition 2.5 (Newton-Gauss Line) Consider the three diagonals of a
complete quadrilateral, namely AC, BD, PQ. It turns out their midpoints are
collinear. The line through them is called the Gauss line (sometimes also called
the Newton-Gauss line). [4]

Theorem 2.6 (Gauss-Bodenmiller Theorem) The circles with diameters

AC, BD, FQ are coaxial. Their radical axis is a line passing through each of the
four orthocenters of the triangles 4FAB, 4FCD, 4EAD, 4EBC. The radical
axis is sometimes called the Steiner line (or sometimes Aubert line). [7]

B
C
F

Proof We begin with the following lemma:

Lemma 2.7 Let 4ABC be a triangle and let M , N be points on sides CA and
AB respectively.Then the orthocenter H of triangle 4ABC lies on the radical axis
of the circles with diameters BM and CN . [7]
 A

F
E
N
H

B C

Proof Let E, F be the feet of the B, C-altitudes in triangle ABC respectively. Then
∠BEM = ∠CEF = 90◦ so E lies on the circle with diameter BM and F lies on the
circle with diameter CN . Hence, it suffices to show that HB.HC = HC.HF (so that
the power of H with respect to the two circles is equal). But since the reflection of
H over the sidelines of the triangle 4ABC lie of the circumcircle of triangle 4ABC
we have that HB.HE and HC.HF are both equal to half the power of H with respect
to the circumcircle of triangle 4ABC. This completes the proof of the lemma.

Let H1 ,H2 ,H3 ,H4 be the orthocentres of triangles 4ABF , 4BCE, 4CDF , 4DAE
respectively. Returning to the problem, note that segments AC, BD, and FE are
cevians in triangle 4ABF so from the claim we know that H1 the radical centre of
the circles wit diameters AC, BD, EF . Similarly, points H2 , H3 , H4 are also radical
centres of these circles. Hence, either these circles are coaxial or the orthocenters of
triangles 4ABF , 4BCE, 4CDF , 4DAE coincide. But the latter situation is clearly
impossible, so this completes the proof.

So that we can conclude the following:

Theorem 2.8 (newton-line-thm) Let ABCD be a quadrilateral and let E =

AB ∩ CD and F = DA ∩ BC. Then the midpoints of the segments AC, BD, EF are
collinear. [7]

2.3 Miquel Point

Theorem 2.9 (Miquel’s Pivot Theorem) Let ABC be a triangle and let
D,E,F be points lying on sides BC, CA, AB respectively. Then the circumcirles
of triangles 4AEF , 4BFD, 4CDE concur. [7]
Proof Let the circumcircles of the triangles 4BDF and 4CDF intersect again at P .
Then we have that:
∠EPF = 360◦ − ∠FPD − ∠DPE = 360◦ − (180◦ − ∠B) − (180◦ − ∠C) = 180◦ − ∠A
So quadrilateral AEPF is cyclic. This completes the proof.

F E

B D C

Theorem 2.10 (Miquel’s Theorem) Let ABCD be a quadrilteral and let

E =AB ∩ CD and F = DA ∩ BC. Then the circumcircles of triangles 4ABF ,
4BCE, 4CDF , 4DAE concur at a point M , called the Miquel Point of complete
quadrilateral ABCDEF . [7]

Proof Applying Miquel’s Pivot Theorem to triangle 4ABF with points C, D, E we

have that the circumcircles of triangles 4DAE, 4BCE, and 4CDF concur. Applying
Miquel’s pivot Theorem three more times in the same way then yields the desired
result.

E
B

M C

Lemma 2.11 (Centers are Concyclic with the Miquel Point) The four
centers of (PAB), (PDC), (QAD), (QBC) lie on a circle passing through the Miquel
point. [4]
 E
B

C
M

Lemma 2.12 (Altitudes from the Miquel Point) The feet of the perpen-
diculars from M to lines AB, BC, CD, DA are collinear. Furthermore, the line
though these four points is perpendicular to the Gauss line. [4]

E
B

C
M

Theorem 2.13 (Miquel Point of a Cyclic Quadrilateral) Let ABCD be a

cyclic quadrilateral inscribed in circle ω with diagonals meeting at R. Then the
Miquel point of ABCD is the inverse of R with respect to inversion aroundω.

Proof Let O be the circumcenter of ABCD, and let R∗ be the image of R. It suffices
to show R∗ = M . Angle chasing (left as an exercise) lets us establish ∠AR∗ B = ∠APB,
so that R∗ lies on (PAB). Similarly, R∗ lies on (PCD), (QBC), and (QDA). Hence R∗
is indeed the Miquel point.
3 Relationship with the Projective Geometry

Definition 3.1 (Cross Ratios) The cross ratio is an important invariant in

projective geometry. Given four collinear points A, B, X, Y (which may be points
at infinity), we define the cross ratio as
XA YA
(A, B; X, Y ) = ÷
XB YB
Here the ratios are directed with the same convention as Menelaus’s theorem;
in particular, the cross ratio can be negative! If A, B, X, Y lie on a number line
then this can be written as
x−a y−a
(A, B; X, Y ) = ÷
x−b y−b

You can check that (A, B; X, Y ) > 0 precisely when segments AB and XY are
disjoint or one is contained inside the other. We also generally assume A 6= X,
B 6= X, A 6= Y , B 6= Y . [4]

3.1 Harmonic Bundles

Definition 3.2 (Harmonic Bundles) The most important case of our cross
ratio is when (A, B; X, Y ) = −1. We say that (A, B; X, Y ) is a harmonic bundle
in this case, or just harmonic. Furthermore, a cyclic quadrilateral AXBY is a
harmonic quadrilateral if (A, B; X, Y ) = −1. Observe that if (A, B; X, Y ) = −1,
then (A, B; Y , X) = (B, A; X, Y ) = −1. We sometimes also say that Y is the
harmonic conjugate of X with respect to AB; as the name suggests, it is unique,
and the harmonic conjugate of Y is X itself. [4]

Lemma 3.3 ((9.9) (Harmonic Quadrilaterals)) Let ω be a circle and let

P be a point outside it. Let PX and PY be tangents to ω. Take a line through
P intersecting ω again at A and B. Then
(a) AXBY is a harmonic quadrilateral.
(b) If Q = AB ∩ XY , then (A, B; Q, P ) is a harmonic bundle
 X

Q
B
P

Lemma 3.4 (Inversion Induces Harmonic Bundles) Let P be a point on

line AB, and let P ∗ denote the image of P after inverting around the circle with
diameter AB. Then (A, B; P , P ∗ ) is harmonic.

Lemma 3.5 ((9.11)Cevians Induces Harmonic Bundles) Let 4ABC be

a triangle with concurrent cevians AD, BE, CF (possibly on the extensions of the
sides). Line EF meets BC at X (possibly at a point at infinity). Then (X, D; B, C)
is a harmonic bundle. [4]

Lemma 3.6 (Complete Quadrilaterals Induces Harmonic Bundles)

Let ABCD be a quadrilateral whose diagonals meet at K . Lines AD and BC
meet at L, and line KL meets AB and CD at M and N . Then (K , L; M , N ) is a
harmonic bundle. [4]

Proof Let P = AB ∩ CD, and let Q = PK ∩ BC. By Lemma 3.5, (Q, L; B, C) = −1.
Projecting onto the desired line, we derive
P
−1 = (Q, L; B, C) = (K , L; M , N ).

3.2 Pole and Polar

Definition 3.7 (Poles and Polars) Let the image of the point P under in-
version with respect to the circle with center O and radius r be P0 . The polar
of P is the line p perpendicular to the line OP at P0 . In this case, the point P
is called the pole of p. [6]
 g

X
Γ

O A P
h

f2
Theorem 3.8 (La Hire’s Theorem) A point X lies on the polar of a point
Y if and only if Y lies on the polar of X. [4]

La Hire’s theorem demonstrates a concept called duality: one can exchange points
for lines, lines for intersections, collinearity for concurrence. Simply swap every
point with its polar and every line with its pole. We can now state an important
result relating poles and polars to harmonic bundles.

Lemma 3.9 Let AB be a chord of a circle ω and select points P and Q on line
AB. Then (A, B; P , Q) = −1 if and only if P lies on the polar of Q. [4]

Q
B
P

Proof We consider only the case where P is outside ω and Q is inside it. Construct
the tangents PX and PY to ω. Lemma 3.3 gives (A, B; P , XY ∩ AB) = −1, so Q lies on
the polar of P (namely line XY ) if and only if (A, B; P , Q) = −1.

3.3 Brocard Theorem

 Theorem 3.10 (Brocard’s Theorem) Let ABCD be an arbitrary cyclic
quadrilateral inscribed in a circle with center O, and set P = AB ∩ CD,
Q = BC ∩ DA, and R = AC ∩ BD. Then P , Q, R are the poles of QR, RP ,
PQ, respectively. In particular, O is the orthocenter of triangle 4PQR. [4]

A
R
O

Q B
C

We say that triangle 4PQR is self-polar with respect to ω , because each of its sides
is the polar of the opposite vertex.

X
A
R
O

Q B Y C

Proof First, we show that Q is the pole of line PR. Define the points X = AD ∩ PR
and Y = BC ∩ PR. By Lemma 3.5, both (A, D; Q, X) and (B, C; Q, Y ) are harmonic
bundles. Therefore, X and Y both lie on the polar of Q, by Lemma 3.9. Since the
polar of Q is a line, it must be precisely line XY , which is the same as line PR. The
same can be used to show that P is the pole of line QR and R is the pole of line PQ;
projective geometry is immune to configuration issues. (This is part of the reason
we like points at infinity.) This gives that PQR is indeed self-polar. Finally, the
definition of a polar implies that O is the orthocenter of triangle 4PQR, completing
the proof.
References
[1] disjoint-sets. https://www.cuemath.com/algebra/disjoint-sets/.

[2] Jafet Baca. On a special center of spiral similarity. https://www.awesomemath.

org/wp-pdf-files/math-reflections/mr-2019-01/mr_1_2019_spiral_
similarity.pdf, 2019.

[3] Alexander Bogomolny. Theorem of complete quadrilateral what is it? a math-

ematical droodle. https://www.cut-the-knot.org/Curriculum/Geometry/
Quadri.shtml.

[4] Evan Chen. Euclidean Geometry in Mathematical Olympiads. Mathematical

Association of America, 2016.

[5] J. L. Coolidge. A Treatise On the Circle and the Sphere. AMS - Chelsea Pub-
lishing, 1971.

[6] Stefan Lozanovski. A Beautiful Journey Through Olympiad Geometry. 2016.

[7] Cosmin Pohoata Titu Andreescu, Sam Korsky. Lemmas in Olympiad Geometry.
XYZ Press, 2016.

[8] Eric W. Weisstein. Concurrent. https://mathworld.wolfram.com/

Concurrent.html.