See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303279655

A Characterization of the Parallelogram

Article · May 2016

CITATIONS READS

0 207

1 author:

Paris Pamfilos
University of Crete
84 PUBLICATIONS 99 CITATIONS

SEE PROFILE

All content following this page was uploaded by Paris Pamfilos on 17 May 2016.

The user has requested enhancement of the downloaded file.

 A Characterization of the Parallelogram

Paris Pamfilos
University of Crete, Greece
[email protected]

Abstract
In this article we discuss a simple characterization of the parallelogram by con-
sidering dissections by lines through a fixed point.

1 The characteristic property

A parallelogram ABCD has a symmetry center coinciding with the intersection point
of its diagonals O. Every line through the point O dissects the parallelogram into two
quadrilaterals or triangles, which lie symmetrically with respect to O, and, consequently,
have the same area (see Figure 1).

D C

Ο
A B

Figure 1: Symmetry of the parallelogram

One can, inversely, ask if there is another kind of convex quadrilateral with the same
property. The answer is no, and this is the subject of the following theorem.

Theorem 1. A convex quadrilateral ABCD is a parallelogram, if and only if, there is a point O,
such that every line through O divides the quadrilateral into two equiareal polygons.

Before to enter into the proof of the theorem, we discuss a simple lemma, which rep-
resents a key point in our arguments.

2 A property of the triangle

Consider a triangle ABC and a point D on base BC. A line through D intersects the other
sides {AB, AC} at corresponding points {B 0 , C 0 }. The triangles {BB 0 D, CC 0 D} cannot
have the same area for many directions of the line through D (see Figure 2). This is made
precise through the following lemma.

Lemma 1. Under the conventions made above, there is no point D on base BC of the triangle,
such that the triangles {DBB 0 , DCC 0 } have the same area for three or more lines through D.

1
 A
α

C'

B'' B β D φ γ C
C''
B'

Figure 2: When the two triangles have the same area;

Proof. Denoting the projections of {B 0 , C 0 } on BC, respectively by {B 00 , C 00 }, the areas of

the triangles are
1 1
(DCC 0 ) = |DC| · |DC 00 | tan(φ) and (DBB 0 ) = |DB| · |DB 00 | tan(φ).
2 2
Without loss of the generality, we assume here that the angle γ is acute and the angle φ
of the line intersecting the triangle is acute towards C. This, implying that C 00 is inside
CD and B is outside/inside BD, depending on whether the angle β is acute/obtuse.
With minor modifications in the formulas below the results are valid for all cases. We
exclude also the case of a right-angled (at B or C) triangle, for which the calculations
below become even simpler.
In fact, calculating the lengths {|B 0 B 00 |, |C 0 C 00 |} in two ways, we see that
|DC| tan(γ)
|C 0 C 00 | = |DC 00 | tan(φ) = (|DC| − |DC 00 |) tan(γ) ⇒ |DC 00 | = ,
tan(φ) + tan(γ)
|DB| tan(β)
|B 0 B 00 | = |DB 00 | tan(φ) = (|DB 00 | − |DB|) tan(β) ⇒ |DB 00 | = .
tan(β) − tan(φ)
Thus, if the areas of the two triangles are equal, the following equation must be valid:
|DC|2 tan(γ) |DB|2 tan(β)
= ⇔
tan(φ) + tan(γ) tan(β) − tan(φ)

tan(φ)(|DB|2 tan(β) + |DC|2 tan(γ)) = (|DC|2 − |DB|2 ) tan(β) tan(γ).

Consider the last equation as a linear equation in x = tan(φ), for a fixed D. Then, if by
assumption, the areas of the two triangles are equal for three different lines through D,
we easily see that the above equation (or a similar one with modified signs) must be valid
for two values of φ. This implies that the corresponding coefficients of the linear equation
must vanish. Investigating the different possibilities, we find that this is impossible.
For example, the vanishing of the right side of the last equation implies |DB| = |DC|,
so point D is the middle of BC, and this, looking at the left side, implies that tan(β) +
tan(γ) = 0, which is impossible. A similar argument applies also to the other possible
cases, regarding the locations of {B 00 , C 00 } relative to the segments {BD, DC}. The details
here are left as an exercise.

3 Proof of the theorem

Assume now that ABCD is a convex quadrilateral and O is a point in its plane, such that
every line through O divides the quadrilateral in two equiareal polygons. This hypothe-
sis trivially implies that point O is inside the quadrilateral. But we can restrict its location

2
 further by thinking as follows. Assume that the intersection point P of the diagonals of

A E'
E

P O
B D

F'
F
C

Figure 3: Location of O

the quadrilateral does not coincide with the middle of one diagonal, say the diagonal
BD, and it is |BP | < |P D| (see Figure 3). Then, we see easily that the ratio of the areas of
the triangles
(ABC) |BP |
= ⇒ (ABC) < (ACD).
(ACD) |P D|
Consequently, we can find a parallel EF to AC, lying between AC and D, such that the
area EF D is half the area of the quadrilateral. It is then easily seen, that the point O with
the property, that every line through it divides the quadrilateral in two equiareal parts,
if it exists, then it must be on EF . But then, drawing lines E 0 F 0 through O, we can find
infinitely many pairs of triangles {OEE 0 , OF F 0 } with equal areas. This contradicts the
previous lemma and shows that EF must coincide with AC, meaning that O must be on
AC. A similar argument shows that O must be also on BD. Hence O must coincide with
the intersection point P of the diagonals, implying the proof of the theorem.

View publication stats