INTERNATIONAL JOURNAL OF GEOMETRY

Vol. 13 (2024), No. 2, 86 - 112

CHARACTERIZATIONS OF PARALLELOGRAMS
PART 1

MARTIN JOSEFSSON

Abstract. We collect, categorize and prove 62 necessary and sufficient

conditions for when a quadrilateral is a parallelogram.

1. Introduction
The parallelogram is one of the most fundamental and well-known quadri-
laterals, probably due to its symmetric nature. It has an abundance of in-
teresting and beautiful properties, many of which are less famous than they
merit. Here we will focus on properties that are unique to parallelograms,
that is, those that distinguish parallelograms from other quadrilaterals.
Basic properties and characterizations of parallelograms are studied in
most high school geometry textbooks. In such books, and also online, it’s
easy to find proofs of the following necessary and sufficient conditions for
when a quadrilateral is a parallelogram:
• Both pairs of opposite sides have equal length
• Both pairs of opposite angles have equal measure
• One pair of opposite sides are parallel and have equal length
• The diagonals bisect each other
In addition to the definition, that both pairs of opposite sides are parallel,
how many other characterizations of parallelograms do you know? There are
a few that sometimes are mentioned in textbooks, but it’s quite difficult to
find sources that have compiled more than ten characterizations as theorems
and exercises (the top contender we found is the English Wikipedia page [38],
listing eleven characterizations at the time of writing this paper).
Many authors of articles, books, and Olympiad problems have contributed
to the knowledge on necessary and sufficient condition for a quadrilateral to
be a parallelogram, but they are scattered in many places and it’s common
for new ones to appear one at a time.
————————————–
Keywords and phrases: Parallelogram, sufficient condition, converse,
diagonal, bimedian, congruence, trapezoid
(2020)Mathematics Subject Classification: 51M04, 51M25
Received: 17.01.2024. In revised form: 12.02.2024. Accepted: 27.01.2024.
 Characterizations of parallelograms part 1 87

This is the first of a two part paper on characterizations of parallelograms.

Together they contain 103 such characterizations. A few months before the
writing process started, we knew of only one third of them. The second
third was found online as problems in various mathematics competitions or
in less known articles and books. The last third was discovered while we
were doing the research and thought about what other properties that might
also be sufficient conditions. We have found no references for those as char-
acterizations, but several of them are known properties of parallelograms.
There are no concave or crossed quadrilaterals that can be a parallelo-
gram, so when we write quadrilateral, it means a convex quadrilateral.
The disposition of the two papers are such that all characterizations are
grouped regarding what they mainly are about, but at the end of the second
part we will give a historical overview in chronological order according to
the oldest date of publication that we know of for each of them together
with references.

2. Equal or parallel sides

While several classes of quadrilaterals are defined in various ways in dif-
ferent textbooks, a parallelogram is almost always defined in the following
way. This also makes our first characterization:
Definition 2.1. A quadrilateral is a parallelogram if and only if both pairs
of opposite sides are parallel.
Euclid did not directly define a parallelogram in his Elements [22], which
is surprising considering how thorough he was defining most other concepts,
but it’s nonetheless clear from the context what he meant when using the
word parallelogram. According to Proclus [12, p. 325], it was Euclid who
invented this word (παραλληλóγραµµo in Greek). The English word paral-
lelogram was used for the first time in 1570 in a translation of the Elements
according to [25]. An old synonym is rhomboid [41, p. 4], but it usually
refers to parallelograms that are neither rhombi nor rectangles.
In our first theorem, we prove five characterizations of parallelograms that
deal with equal or parallel sides of a quadrilateral. Condition (c) is from
[41, p. 26], where it was not stated separately as a sufficient condition, but
part of a proof of another sufficient condition, so from this proof it’s clear
that it implies a quadrilateral to be a parallelogram. We found (d) and (e)
in [10, p. 47], which are related to a congruence criterion of triangles rarely
included in textbooks. Condition (e) was further investigated in [8].
Theorem 2.1. A quadrilateral ABCD satisfies any one of:
(a) it has two pairs of opposite sides with equal length
(b) it has one pair of opposite parallel sides that have equal length
(c) it has two pairs of equal alternate angles between a diagonal and the
sides
(d) AB = DC > AC and ∠ACB = ∠CAD
(e) AB = DC < AC and ∠B = ∠D
if and only if it’s a parallelogram.
88 Martin Josefsson

Proof. (a) In a parallelogram ABCD, triangles ABC and CDA are con-
gruent (ASA) due to equal alternate angles and the common diagonal, so
AB = CD and BC = DA (see Figure 1). Conversely, in a quadrilateral
ABCD with AB = CD and BC = DA, triangles ABC and CDA are
congruent (SSS), so opposite sides are parallel due to equal alternate angles.

Figure 1. A parallelogram with diagonal AC

(b) In a parallelogram, opposite sides are parallel (by definition) and equal
according to (a). Conversely, if in a quadrilateral ABCD it holds that
AB = CD and AB ∥ CD, then triangles ABC and CDA are congruent
(SAS), so BC = DA and ABCD is a parallelogram by (a).
(c) That ABCD is a parallelogram if and only if ∠DAC = ∠BCA and
∠DCA = ∠BAC is a direct consequence of a necessary and sufficient con-
dition for parallel lines (equal alternate angles) and Definition 2.1.
(d) In a parallelogram, AB = DC and ∠ACB = ∠CAD. Conversely,
when AB = DC > AC and ∠ACB = ∠CAD, then triangles ABC and
CDA are congruent (SSA), see Figure 1, so ABCD is a parallelogram ac-
cording to (a). Many readers will recognize (SSA) as the ambiguous case,
which in general is not a true congruence case, but it is true when the given
angle is opposite the longer of the two considered sides, see [13].
(e) This proof is left to the reader, as it’s very similar to (d). □

3. Diagonals
Here we prove nine different characterizations of parallelograms that are
about the diagonals. Condition (b) is taken from [41, p. 28], where it was
not stated separately as a sufficient condition, but part of a proof of another
sufficient condition, so from this proof it’s clear that it implies a quadrilateral
to be a parallelogram. (c) and (e) are from the old books [41, p. 26] and
[32] (exercise on page 381), while (d) was found in the paper [1, p. 235], and
(g) is from the recent book [35, p. 188]. To prove that (h) is a property of
parallelograms was Problem 155 in [26]. The last characterization is from
the Mexican Mathematical Olympiad in 1996 according to [40], but the proof
we cite is from [39, p. 34].
Theorem 3.1. A quadrilateral ABCD with diagonal intersection P satisfies
any one of:
(a) it has bisecting diagonals
(b) the two pairs ABP , CDP and BCP , DAP are congruent triangles
 Characterizations of parallelograms part 1 89

(c) each diagonal divide it into two congruent triangles with the same
orientation
(d) each diagonal bisects the perimeter of the quadrilateral
(e) each diagonal bisects the area of the quadrilateral
(f) the sum of any two adjacent diagonal parts is equal to the sum of the
other two diagonal parts
(g) it has equal distances from opposite vertices to the corresponding
diagonals
(h) the line segments from a random point on any diagonal to the opposite
vertices together with this diagonal and the sides divide it into two
pairs of triangles with equal area
(i) E and F are the midpoints of AB and BC respectively, where E =
DM ∩ AB and F = DN ∩ BC, and where M and N trisect AC such
that AM = M N = N C
if and only if it’s a parallelogram.
Proof. (a) In a parallelogram ABCD, AB = DC, ∠BAC = ∠DCA and
∠ABD = ∠CDB, so triangles ABP and CDP are congruent (ASA), see
Figure 2. Then AP = CP and BP = DP .
Conversely, when AP = CP and BP = DP in a quadrilateral ABCD,
then triangles ABP and CDP are congruent (SAS) due to vertically op-
posite angles at P , so AB = DC. In the same way we get BC = DA, so
ABCD is a parallelogram according to Theorem 2.1 (a).

Figure 2. The diagonal intersection P

(b) In a parallelogram ABCD, the pairs of triangles ABP , CDP and

BCP , DAP are congruent due to bisecting diagonals and vertically oppo-
site angles at P (see Figure 2). Conversely, if these opposite triangles are
congruent in pairs, then the diagonals bisect each other and ABCD is a
parallelogram according to (a).
(c) Any diagonal divide a parallelogram into two congruent triangles
(SSS). Conversely, such congruent triangles directly yield that opposite sides
are equal, so the quadrilateral is a parallelogram according to Theorem 2.1
(a).
(d) Each diagonal bisects the perimeter of a quadrilateral if and only if
{ { {
a+b=c+d 2a + b + d = 2c + b + d a=c
⇔ ⇔
a+d=b+c b−d=d−b b=d
(see Figure 2), where the last two equalities are equivalent to the quadrilat-
eral being a parallelogram according to Theorem 2.1 (a).
90 Martin Josefsson

(e) Since any diagonal in a parallelogram divide it into two congruent

triangles by (c), these pairs of triangles clearly have the same area.
Conversely, if both triangles ABC and ABD have an area equal to one-
half the area of quadrilateral ABCD, then they have the same height due
to equal bases. Let H1 and H2 be the foot points of the two heights (see
Figure 3). Then H1 CDH2 is a parallelogram according to Theorem 2.1 (a)
since CH1 and DH2 are parallel and have the same length, so AB and CD
are parallel. In the same way BC and DA are parallel, making ABCD a
parallelogram by definition.

Figure 3. Two parallel heights with equal length

(f) Using the notations w = AP , x = BP , y = CP , z = DP for the

diagonal parts (see Figure 2), we have that
{ { {
w+x=y+z 2w + x + z = 2y + x + z w=y
⇔ ⇔
z+w =x+y x−z =z−x x=z
where the last two equalities are equivalent to that the quadrilateral is a
parallelogram according to (a).
(g) In a parallelogram ABCD, since triangles ABC and CDA have equal
area by (e) and the same base AC, the distances from B and D to AC are
equal, and similarly for triangles ABD and CDB.
Conversely, if the distances from B and D to AC are equal, then triangles
ABC and CDA have equal area, and similarly for triangles ABD and CDB.
Then ABCD is a parallelogram according to (e).

Figure 4. TDAX = TABX and TBCX = TCDX

(h) In these pairs of triangles, the bases (on the diagonal in question) are
always equal (see Figure 4, where TXY Z denote the area of triangle XY Z).
The quadrilateral is a parallelogram if and only if the heights to this diagonal
 Characterizations of parallelograms part 1 91

in all four triangles are equal according to (g), so the pairs of triangles have
equal area if and only if the quadrilateral is a parallelogram.
(i) First we join BM and BN (see Figure 5). When ABCD is a parallel-
ogram, AC and BD bisect each other at P , so BP = P D and AP = P C.
Since AM = N C, we get M P = P N , so BN DM is a parallelogram. Then
BN ∥ EM where M is the midpoint of AN , so E is the midpoint of AB
according to the converse of the midpoint theorem. In the same way, F is
the midpoint of BC.

Figure 5. The interior parallelogram BN DM

Conversely, when E and F are the midpoints of sides AB and BC in a

quadrilateral ABCD, and M and N are the midpoints of AN and CM ,
then EM ∥ BN and F N ∥ BM according to the midpoint theorem. Hence
BN DM is a parallelogram, so M P = P N and BP = P D. But AM = N C,
so AP = P C. Then ABCD is a parallelogram according to (a). □

As a comment to (d), we note that quadrilaterals where one diagonal

bisects the perimeter are called extangential quadrilaterals, see [17].

4. Angles and trigonometry

Next we prove seven characterizations of parallelograms that are about
angles or trigonometric relations. Condition (b) is from [41, p. 26], where
it was not stated separately as a sufficient condition, but part of a proof of
another sufficient condition, so from this proof it’s clear that it implies a
quadrilateral to be a parallelogram. The basic (d) is for some reason not
included in textbooks as a characterization of parallelograms. Conditions
(e) and (g) together with their proofs are from [27] and [31] respectively. To
prove (f) was a problem in [14], where the published proof applied vectors.
The proof we give was found online several years ago and now we cannot
find a reference for it.
Theorem 4.1. A quadrilateral ABCD with sides AB = a, BC = b, CD =
c, DA = d and diagonal intersection P satisfies any one of:
(a) it has two pairs of equal opposite angles
(b) any pair of adjacent angles are supplementary
(c) the sum of any two adjacent angles is a constant
(d) the two angles between the extensions of opposite sides are zero
92 Martin Josefsson

(e) ∠CAD ∠ACB ∠DBA

∠CAB = ∠ACD and ∠DBC = ∠BDA
∠BDC

(f) da cos A + ab cos B + bc cos C + cd cos D = 0

(g) P A sin A = P C sin C and P B sin B = P D sin D
if and only if it’s a parallelogram.
Proof. (a) In a parallelogram ABCD, triangles ABC and CDA are con-
gruent (ASA), so ∠B = ∠D. In the same way, by drawing diagonal BD, we
have ∠A = ∠C. Conversely, if ∠A = ∠C and ∠B = ∠D in a quadrilateral,
then ∠A + ∠D = π by the angle sum of a quadrilateral. Thus ∠A = δ,
where δ is the exterior angle at D (see Figure 6), proving that AB ∥ DC.
In the same way BC ∥ AD, so ABCD is a parallelogram by definition.

Figure 6. The exterior angles at B and D

(b) In a quadrilateral ABCD, let β and δ be the exterior angles at B and

D respectively (see Figure 6). Then β + ∠B = π and δ + ∠D = π. We have
that ABCD is a parallelogram if and only if ∠A = δ and ∠A = β, which is
equivalent to ∠A + ∠B = π and ∠A + ∠D = π. Similar expressions hold at
each vertex.
(c) In a parallelogram, that constant is π by (b). Conversely, if
∠A + ∠B = ∠B + ∠C = ∠C + ∠D = ∠D + ∠A
in a quadrilateral, then it directly follows that ∠A = ∠C and ∠B = ∠D.
Hence it’s a parallelogram according to (a).

Figure 7. Angle ξ between extensions of AB and DC

(d) In quadrilateral ABCD, let the extensions of AB and DC intersect

at J, which we assume without loss of generality to be outside of BC (see
Figure 7). Then we have
AB ∥ DC ⇔ ∠ABC = γ ⇔ γ+ξ =γ ⇔ ξ=0
where γ is the exterior angle at C and we applied the exterior angle theorem
in triangle BCJ. In the same way we have that the other pair of opposite
 Characterizations of parallelograms part 1 93

sides are parallel if and only if the angle between their extensions is zero.
Both conditions must hold for ABCD to be a parallelogram (by definition).
(e) It’s obvious that these equalities are satisfied in a parallelogram due
to equal alternate angles.

Figure 8. The angles α, β, γ, δ

Conversely, suppose we have ∠CAD ∠ACB ∠DBA

∠CAB = ∠ACD and ∠DBC = ∠BDA in a
∠BDC

quadrilateral. We denote these respective quotients by Ā, C̄, B̄, D̄ so that

Ā = C̄ ≡ m and B̄ = D̄ ≡ n. Let α, β, γ, δ denote the angles CAB, DBC,
ACD, BDA respectively (see Figure 8). Then we have ∠CAD = mα,
∠ACB = mγ, ∠DBA = nβ, and ∠BDC = nδ. Applying the angle sum in
the four subtriangles created by the diagonals yields
δ + mα = β + mγ and α + nβ = γ + nδ,
so we get β − δ = m(α − γ) and n(β − δ) = γ − α. Hence by substitution,
nm(α − γ) = γ − α, that is,
(nm + 1)(α − γ) = 0.
It follows that α = γ and in turn β = δ, so both pairs of opposite sides are
parallel. Then ABCD is a parallelogram by definition.
(f) In a parallelogram ABCD with sides a = AB = CD = c and b =
BC = DA = d, angles ∠A = ∠C and ∠B = ∠D, we get
da cos A + ab cos B + bc cos C + cd cos D = 2ab(cos A + cos B) = 0
since ∠A + ∠B = π according to (b).
Conversely, applying the law of cosines in a quadrilateral ABCD with
diagonals p = AC and q = BD (see Figure 6) yields
p2 = a2 + b2 − 2ab cos B, q 2 = b2 + c2 − 2bc cos C,
p2 = c2 + d2 − 2cd cos D, q 2 = d2 + a2 − 2da cos A.
Adding these four equalities, we get
2(p2 + q 2 ) = 2(a2 + b2 + c2 + d2 ) − 2(da cos A + ab cos B + bc cos C + cd cos D),
and since we now assume the expression in the last parenthesis is zero, this
simplifies into
p2 + q 2 = a 2 + b2 + c 2 + d 2 .
94 Martin Josefsson

In any quadrilateral there is the following similar expression, called Euler’s

quadrilateral theorem:
a2 + b2 + c2 + d2 = p2 + q 2 + 4v 2
where v is the distance between the diagonal midpoints. Comparing the
last two equalities, we see that we get 4v 2 = 0, that is v = 0. This means
that the diagonals bisect each other, so the quadrilateral is a parallelogram
according to Theorem 3.1 (a). A proof of Euler’s quadrilateral theorem can
be found for instance in [3, pp. 9–10], a marvelous book that any enthusiast
of quadrilaterals ought to own.
(g) In a parallelogram ABCD, it holds that P A = P C, P B = P D and
sin A = sin B = sin C = sin D since adjacent angles are supplementary, so
P A sin A = P C sin C and P B sin B = P D sin D.

Figure 9. Projections of P onto the sides

Conversely, when these two equalities hold in a quadrilateral, let E, F ,

G, H be the projections of P onto the sides AB, BC, CD, DA respectively
(see Figure 9). Then AEP and AHP are right angles, so AEP H is a cyclic
quadrilateral. Applying the law of sines yields
HE AE AE PA
= = =
sin A sin ∠AHE sin ∠AP E sin π2
so HE = P A sin A. In the same way we get F G = P C sin C; hence
HE = F G. By symmetry, EF = GH and we can thus conclude that
EF GH is a parallelogram. Then EF ∥ GH and HE ∥ F G. Now let the
extensions of HG and EF intersect the extensions of BC and DC at M
and N respectively. Next we denote ∠F N C = ∠CGM = ∠HGD ≡ φ. In
cyclic quadrilateral HP GD we have
(1) ∠DP H = ∠DGH = φ
and by the exterior angle theorem in triangle F N C, it holds that ∠CF N =
∠C − φ. Then
(2) ∠EP B = ∠BF E = ∠C − φ.
 Characterizations of parallelograms part 1 95

Since DB is a straight line segment, we have

(3) ∠DP H + ∠HP E + ∠EP B = π
and in cyclic quadrilateral AHP E, it holds
(4) ∠HP E = π − ∠A.
By substituting (1), (4) and (2) into (3), we get
φ + π − ∠A + ∠C − φ = π.
Hence ∠A = ∠C. In the same way it can be proved that ∠B = ∠D, so
ABCD is a parallelogram according to (a). □

5. Angle bisectors and bimedians

In this section we prove eleven necessary and sufficient conditions for
when a quadrilateral is a parallelogram concerning angle bisectors or bi-
medians. A bimedian is a line segment between the midpoints of opposite
sides (see Figure 11). Conditions (a) and (b) were stated in the book [35,
p. 188], while (c) and (d) were found at [37] and [5] respectively (the latter
concerned a problem cited from an unspecified book that was asked at a
math forum, where no solution was given). The characterizations (f) and
(g) were proved in the recent paper [9], while (i) was stated as a defining
property of parallelograms in connection with a hierarchical classification in
[34] and is one of our favorite characterizations of parallelograms. To prove
(j) was a problem on the Leningrad High School Olympiad in 1980. We
give our proof of it, but other proofs can be found in [3, p. 258] and [23].
The last condition was studied in [6, pp. 140–141], where the given proof
of the converse is attributed to John Webb. Our simpler proof only applies
formulas used to prove (j).
Theorem 5.1. A quadrilateral ABCD satisfies any one of:
(a) it has two pairs of opposite parallel angle bisectors
(b) one angle bisector is perpendicular to two adjacent angle bisectors
(c) all four angle bisectors form a rectangle
(d) the bimedians quadrisect the area of the quadrilateral
(e) the bimedians quadrisect the perimeter of the quadrilateral
(f) each bimedian bisects the area of the quadrilateral
(g) each bimedian bisects the perimeter of the quadrilateral
(h) the bimedians intersect at a diagonal midpoint
(i) the bimedians and the diagonals are concurrent
(j) the sum of the length of the bimedians equals the semiperimeter
(k) the bimedians have constant length for all quadrilaterals with sides
of fixed length
if and only if it’s a parallelogram.
Proof. (a) Let ∠A = 2α, ∠C = 2γ, and the angle bisectors at A and C be
AE and CF respectively, with E ∈ DC and F ∈ AB (see Figure 10). In a
parallelogram, ∠BAE = ∠DEA, and since α = γ, then AE ∥ F C. In the
same way the angle bisectors at B and D are parallel.
96 Martin Josefsson

Conversely, if AE ∥ F C, then ∠F CD = ∠AED = ∠EAB, so ∠C = ∠A.

In the same way ∠D = ∠B, so ABCD is a parallelogram according to
Theorem 4.1 (a).

Figure 10. The four angle bisectors

(b) Let ∠A = 2α, ∠B = 2β and ∠D = 2δ. In a parallelogram, 2α+2δ = π,

so the angle ε between the angle bisectors at A and D is
ε = π − (α + δ) = π − π
2 = π
2
using the angle sum of a triangle (see Figure 10). In the same way, the angle
between the angle bisectors at A and B is a right angle.
Conversely, if in quadrilateral ABCD we have ε = π2 , then α+δ = π− π2 =
2 , so ∠A+∠D = 2α+2δ = π. In the same way we prove that ∠A+∠B = π,
π

so ABCD is a parallelogram by Theorem 4.1 (b).

(c) This is a weaker condition than the one in (b), but the proof is more
or less the same except that here we have more angles to consider. We let
the reader write the full proof.

Figure 11. Bimedians Ma Mc and Mb Md

(d) In a parallelogram, each bimedian is parallel to two opposite sides,

so the two bimedians divide the parallelogram into four congruent smaller
parallelograms with equal area.
Conversely, if the bimedians Ma Mc and Mb Md divide a quadrilateral
ABCD into four regions of equal area, we first consider the Varignon paral-
lelogram Ma Mb Mc Md . Its diagonals are the bimedians in ABCD, and they
divide Ma Mb Mc Md into four triangles with equal area since each pair of adja-
cent triangles have base and height of equal length (see Figure 11). Hence we
 Characterizations of parallelograms part 1 97

only need to prove that if the areas of triangles AMa Md , BMb Ma , CMc Mb ,
DMd Mc are equal, then ABCD is a parallelogram. Since AMd = DMd ,
we get that Mc and Ma are equidistant from AD, so AD and Ma Mc are
parallel. By a similar argument, BC and Ma Mc are parallel, so all three of
AD, Ma Mc , BC are parallel. In the same way AB, Md Mb , DC are parallel,
so ABCD is a parallelogram.
(e) The bimedians quadrisect the perimeter of a quadrilateral ABCD
with sides a = AB, b = BC, c = CD, d = DA if and only if
d a a b b c c d
+ = + = + = +
2 2 2 2 2 2 2 2
which is equivalent to a = c and b = d. This characterizes a parallelogram
according to Theorem 2.1 (a).
(f) It has been proved for instance as Proposition 4 in [18] and Theorem
1 (a) in [9] that a bimedian bisects the area of a quadrilateral if and only
if the two sides it connects are parallel. Hence it holds that each bimedian
bisects the area of a quadrilateral if and only if both pairs of opposite sides
are parallel, which is equivalent to the quadrilateral being a parallelogram
by definition.
(g) Since a bimedian connects the midpoints of a pair of opposite sides, it’s
more or less trivial that a bimedian bisects the perimeter of a quadrilateral if
and only if a pair of opposite sides have equal lengths (see Figure 12). Hence
each bimedian bisects the perimeter if and only if both pairs of opposite
sides have equal lengths, which is equivalent to a parallelogram according
to Theorem 2.1 (a).

Figure 12. Ma Mc bisects the perimeter ⇔ BC = DA

(h) Let M and N be the diagonal midpoints. It’s quite well-known that
the bimedians intersect at the midpoint of M N in all quadrilaterals (for
a proof, see [2, pp. 108–109]). Let Q be the point where the bimedians
intersect. Then we have that Q = M is equivalent to M N = 0, which
according to Theorem 3.1 (a) is equivalent to the quadrilateral being a
parallelogram.
(i) With the same notations as in (h), we have that Q = M = N is
equivalent to that the quadrilateral is a parallelogram (M N = 0).
(j) Let n = Ma Mc and m = Mb Md be the bimedians that connect sides
a, c and b, d respectively in a quadrilateral ABCD with sides AB = a,
BC = b, CD = c, DA = d and diagonals AC = p, BD = q (see Figure 11).
98 Martin Josefsson

In [3, p. 11] it was proved that

4n2 = a2 + c2 − b2 − d2 + p2 + q 2
and according to Theorem 10 in [18], we have the equality
p2 + q 2 = b2 + d2 + 2ac cos ξ
where ξ is the angle between the extensions of opposite sides a and c. Via
substitution, we get after applying a well-known trigonometric half-angle
formula that
√( ) ( )
a+c 2 2 ξ a+c
(5) n= − ac sin ≤
2 2 2
where equality holds if and only if ξ = 0. By symmetry there is the similar
formula for the other bimedian
√( ) ( )
b+d 2 2 ψ b+d
(6) m= − bd sin ≤
2 2 2
where ψ is the angle between the extensions of sides b and d, and equality
holds if and only if ψ = 0. Then we get
(7) m + n ≤ 21 (b + d) + 12 (a + c) = s
where s is the semiperimeter, and equality holds if and only if ξ = ψ = 0,
which is equivalent to the quadrilateral being a parallelogram according to
Theorem 4.1 (d).
(k) In a parallelogram, it’s trivial that each bimedian has the same length
as the two opposite sides it’s parallel to, and when the side lengths are fixed,
the length of the bimedians does not change if we tilt the parallelogram.
Conversely, we have a quadrilateral with sides of fixed length and shall
prove that the only quadrilateral where the length of the bimedians does
not change as we vary the vertex angles is a parallelogram. Rewriting (5),
we get (( )
( ) )2
ξ 1 a + c
sin2 = − n2
2 ac 2
where a, c, n are constant, so the right hand side is constant. Then the left
hand side must also be constant since equality holds, and the only possibility
for this to happen for all ξ is when ξ = 0. In the same way, using (6), we
get that ψ = 0 is the only possibility. There is just one type of quadrilateral
where ξ = ψ = 0, the parallelogram according to Theorem 4.1 (d). □

6. Trapezoids
Next we consider ten properties that make a trapezoid a parallelogram.
Note that a parallelogram is a special case of a trapezoid when using inclu-
sive definitions. This is the preferred way of making definitions nowadays in
mathematics, but unfortunately many high school textbooks still use exclu-
sive definitions for trapezoids (as discussed in [19, pp. 75–78]). A perfectly
acceptable way of defining a parallelogram, which has very rarely been used
in textbooks (according to [36, p. 21]), is that a parallelogram is a trapezoid
where both pairs of opposite sides are parallel. This is condition (a) in the
 Characterizations of parallelograms part 1 99

following theorem, where (b) was found in [33]. Condition (c) was formu-
lated slightly differently and proved in [28, pp. 283–284], while the direct
part of (g) was Exercise 132 in [30, p. 140]. (i) and its proof are cited from
[3, pp. 96–97] and (j) was Problem 6 in the final round for Grade 7 on the
Moldovan Mathematical Olympiad in 2001 [15].
Theorem 6.1. In a trapezoid ABCD, let Ma , Mb , Mc be the midpoints of
AB, BC, CD respectively, and Ta , Tb , Tc , Td be the areas of triangles ABP ,
BCP , CDP , DAP respectively, where P is the diagonal intersection. Then
ABCD satisfies any one of:
(a) it has two pairs of opposite parallel sides
(b) it has a pair of opposite equal angles
(c) one diagonal bisects the perimeter
(d) one diagonal bisects the area
(e) all transversals between the lateral sides are bisected by the bimedian
between the bases
(f) AB ∥ DC and BJ = 31 BD, where CMa intersect BD at J
√
(g) AB ∥ DC and AE = EF · EG, where E, F , G are the points
where a line through A intersects diagonal BD and the sides BC,
CD respectively, or their extensions
(h) AB ∥ DC and Ta = Tc
(i) AB ∥ DC and Ta + Tc = Tb + Td
(j) AD ∥ BC and AO = 4OMc , where AMc intersect DMb at O
if and only if it’s a parallelogram.
Proof. (a) All parallelograms are trapezoids when using inclusive defini-
tions of quadrilaterals. Conversely, if a trapezoid has two pairs of opposite
parallel sides, then it’s a parallelogram by definition.
(b) In a parallelogram, opposite sides are parallel and opposite angles
are equal. Conversely, if AB ∥ CD and ∠A = ∠C in a quadrilateral
ABCD, then, since ∠ABD = ∠CDB, triangles ABD and CDB are con-
gruent (AAS), so AB = CD. Then ABCD is a parallelogram according to
Theorem 2.1 (b).
(c) Any diagonal in a parallelogram bisects the perimeter due to equal
opposite sides. Conversely, suppose we have a trapezoid ABCD with AB ∥
CD and that diagonal AC bisects the perimeter so AB + BC = CD + DA.
Also suppose without loss of generality that AB ≥ CD. Then there is a
point J on AB such that AJ = CD, making AJCD a parallelogram (see
Figure 13). Thus we also have AD = JC and JB = AB − CD. Now we
apply the triangle inequality in triangle JBC to get JB + BC ≥ CJ, which
we by substitution can rewrite as AB − CD + BC ≥ AD. Hence we have
{
AB + BC ≥ CD + DA
AB + BC = CD + DA.
For both to be true it requires triangle JBC to be degenerate, which can
only happen if JB = 0. Then trapezoid ABCD is a parallelogram.
(d) Any diagonal in a parallelogram bisects the area since it divides it into
two congruent triangles. Conversely, suppose we have a trapezoid ABCD
with AB ∥ CD and without loss of generality that AB ≥ CD. Construct J
100 Martin Josefsson

Figure 13. Trapezoid with AJ = DC

as in Figure 13. Then triangles ADC and CJA are congruent, so they have
the same area. Denoting the area of triangle XY Z by TXY Z , we have
{
TADC = TCJA + TCBJ
TADC = TCJA .
Then TCBJ = 0 must hold, and since the height cannot be zero, we get
JB = 0. Hence B = J, which proves that ABCD is a parallelogram.
(e) In parallelogram ABCD, a transversal QR with Q ∈ AD and R ∈ BC
intersects Ma Mc at a point I (see Figure 14). We draw ST ∥ AB through
I with S ∈ AD and T ∈ BC. Then ∠QIS = ∠RIT , ∠QSI = ∠RT I and
SI = AMa = BMa = T I since AMa IS and BMa IT are parallelograms.
Hence triangles QIS and RIT are congruent (ASA), so QI = RI.

Figure 14. An arbitrary transversal QR

Conversely, in a trapezoid where QI = RI and AB ∥ DC, we have SI =

T I since any line segment parallel to the bases is bisected by the bimedian
between these bases. Also, ∠QIS = ∠RIT due to vertically opposite angles,
so triangles QIS and RIT are congruent (SAS). Hence ∠QSI = ∠RT I,
implying that AD ∥ BC, and this proves that ABCD is a parallelogram
due to both pairs of opposite sides being parallel.
(f) In a parallelogram we have AB = DC and AB ∥ DC. Then Ma B =
1
2 DC and triangles Ma BJ and CDJ are similar (AA), so 2BJ = JD. It
follows that BJ = 31 BD.
We prove the converse with a contrapositive proof. Suppose without loss
of generality that AB < DC, so Ma B < 12 DC (see Figure 15). Triangles
 Characterizations of parallelograms part 1 101

Figure 15. Trapezoid with AB ∥ DC

Ma BJ and CDJ are still similar (AA) since we assume that AB ∥ DC, so
2BJ < JD. It follows that BJ < 31 BD.
(g) In a parallelogram, triangles BEF and DEA are similar (AA), as are
AEB and GED (AA), so we get
EF BE AE BE
= , =
AE DE EG DE
and it directly follows that (AE)2 = EF · EG (see Figure 16).

Figure 16. The transversal AG

Conversely, if (AE)2 = EF · EG holds in a trapezoid with AB ∥ DC, so

that triangles ABE and GDE are similar (AA), then we have
EF AE AE BE
= , = .
AE EG EG DE
BE
Hence DE = EF
AE , which together with equal vertical angles at E implies
that triangles BF E and DAE are similar. Then AD ∥ BC and together
with the assumption AB ∥ DC, we have that ABCD is a parallelogram.
(h) The areas of triangles ABP and CDP are in a trapezoid ABCD with
a = AB and c = CD given by
a2 h c2 h
Ta = and Tc =
2(a + c) 2(a + c)
when a ∥ c according to [3, p. 96] (note that opposite sides are denoted a
and b in that book), where h is the height of the trapezoid (the distance
between a and c). Hence Ta = Tc if and only if a = c. Together with a ∥ c,
this characterizes parallelograms according to Theorem 2.1 (b).
(i) In addition to the two formulas for Ta and Tc in the proof of (h), it’s
also proved in [3, p. 96] that
ach
Tb = Td =
2(a + c)
102 Martin Josefsson

so we have that Ta + Tc ≥ Tb + Td holds in all trapezoids where a ∥ c due

to the AM-GM-inequality a2 + c2 ≥ 2ac, with equality if and only if a = c.
This case is equivalent to the trapezoid being a parallelogram according to
Theorem 2.1 (b).

Figure 17. Trapezoid with AD ∥ BC

(j) We use notations as in Figure 17. Applying the law of sines three
times yields
1
AO AD OMc 2 CD sin η sin φ
= , = , = 1
sin η sin θ sin φ sin (π − θ) CD 2 BC
so we get
AO AD sin η 2 sin θ 2AD sin η 4AD
= · = 1 = .
OMc sin θ CD sin φ 2 BC sin η
BC
Then
AO
= 4 ⇔ BC = AD
OMc
which together with BC ∥ AD characterizes a parallelogram according to
Theorem 2.1 (b). □

7. Bisect-diagonal quadrilaterals
In this section we will prove seven necessary and sufficient conditions for
when a bisect-diagonal quadrilateral is a parallelogram. A bisect-diagonal
quadrilateral is a quadrilateral where at least one diagonal is bisected by
the other diagonal. This not so well-known type of quadrilateral has been
studied in [21] and [7]. Characterizations (a), (f) and (g) in the following
theorem were found in [33], where the author discusses student attempts to
find new sufficient conditions for parallelograms, but the proof we give for
(f) is cited from [11, p. 36] (Example 1.4.3). (g) was only stated in [33, p.
210] but no proof was given (it was left as a challenge for the reader), and
we have not been able to find it neither stated nor proved anywhere else.
Condition (e) was proved in another way in [21, p. 217].
Theorem 7.1. A bisect-diagonal quadrilateral ABCD with sides a = AB,
b = BC, c = CD, d = DA and diagonal intersection P satisfies any one of:
(a) it has one pair of opposite parallel sides
 Characterizations of parallelograms part 1 103

(b) P is at equal distance from a pair of opposite sides

(c) two opposite vertices are at equal distance from the bisected diagonal
(d) angles M AP and N CP are equal, where M and N are the midpoints
of DP and BP respectively and AP = P C
(e) a2 + b2 = c2 + d2 and AP = CP
(f) it has one pair of opposite equal angles and the diagonal that joins
the vertices of those angles bisects the other diagonal
(g) it has one pair of opposite equal angles, the diagonal that joins the
vertices of those angles is bisected by the other diagonal, and the
diagonals may not be perpendicular unless all four sides have equal
length
if and only if it’s a parallelogram.
Proof. (a) We know that one diagonal is bisected and one pair of opposite
sides are parallel in a parallelogram, so let’s prove the converse. Suppose
that AP = P C and AD ∥ BC in a quadrilateral ABCD. Then angles DAP
and BCP are equal, and together with equal vertical angles at P we have
that triangles AP D and CP B are congruent (ASA), so BP = P D. Hence
ABCD is a parallelogram according to Theorem 3.1 (a).

Figure 18. AP = P C and IP = P J

(b) In a parallelogram, P is at equal distance from any pair of opposite

sides since opposite subtriangles created by the diagonals are congruent.
Conversely, if AP = P C and IP = P J in a quadrilateral where I and J are
the projections of P on AB and DC respectively, then triangles AIP and
CJP are congruent (RHS), see Figure 18. This implies that angles IAP
and JCP are equal, so AB ∥ DC making ABCD a parallelogram according
to (a).

Figure 19. AP = P C and BK = DL

104 Martin Josefsson

(c) In a parallelogram, any pair of opposite vertices are at equal distance

to the diagonal not containing those two vertices due to congruent trian-
gles. Conversely, suppose AP = P C and BK = DL in a quadrilateral (see
Figure 19). Then triangles AP D and CP B have equal area, so we get
1
2 AP · DP sin ∠AP D = 12 CP · BP sin ∠CP B
implying that DP = BP due to equal vertical angles at P . Hence ABCD
is a parallelogram according to Theorem 3.1 (a).

Figure 20. AP = P C and ∠M AP = ∠N CP

(d) In a parallelogram, triangles AP M and CP N are congruent (SAS)

since M P = N P , so ∠M AP = ∠N CP . Conversely, suppose angles M AP
and N CP are equal in a quadrilateral where AP = P C (see Figure 20).
Then triangles AP M and CP N are congruent (ASA), so M P = N P . This
implies that DP = BP since M and N are midpoints on DP and BP ,
making ABCD a parallelogram according to Theorem 3.1 (a).

Figure 21. Two triangle medians must be equal

(e) A quadrilateral where AP = CP is a parallelogram if and only if also

BP = DP according to Theorem 3.1 (a). Applying the median formula in
triangles ABC and ACD (see Figure 21) where AC = p, we get that the
quadrilateral is a parallelogram if and only if
√ √
1
2 2(a2 + b2 ) − p2 = 12 2(c2 + d2 ) − p2
which is equivalent to a2 + b2 = c2 + d2 .
 Characterizations of parallelograms part 1 105

(f) The direct theorem is trivially true, so let’s prove the converse. Sup-
pose we have a quadrilateral where ∠B = ∠D and AP = CP , but let for
the sake of contradiction the other diagonal not be bisected (see Figure 22).
First we assume that BP < DP . Then we can extend BD to a point H such
that DP = HP , making AHCD a parallelogram by Theorem 3.1 (a), so
∠H = ∠D according to Theorem 4.1 (a). But we also have that ∠B = ∠D,
so ∠B = ∠H must hold. Next we apply the exterior angle theorem twice to
get
∠B = ∠ABD + ∠CBD
= ∠BAH + ∠BHA + ∠BHC + ∠BCH
= ∠BAH + ∠H + ∠BCH > ∠H.
This is a contradiction to ∠B = ∠H, so the assumption that BP < DP was
wrong. In the same way it can be proved that BP > DP is wrong, so we
must have BP = DP . This means that the diagonals bisect each other, so
ABCD is a parallelogram according to Theorem 3.1 (a).

Figure 22. ∠B = ∠D and AP = CP

(g) The direct theorem is obviously satisfied in all parallelograms. The

exception that the diagonals may not be perpendicular unless all four sides
have equal lengths is necessary to exclude general kites but to include rhombi
and squares, which are simultaneously parallelograms and kites.
For the converse, suppose without loss of generality that ∠ABC = ∠ADC,
BP = DP , and BD ̸⊥ AC, and for the sake of contradiction, also that
AP < CP . Extend diagonal AC beyond A to E such that EP = CP .
Then EBCD is a parallelogram according to Theorem 3.1 (a), so ∠EBC =
∠EDC. By subtracting equal angles, we get ∠EBA = ∠EDA. Now let
D′ be the reflection of D in the line EC. Next we draw the circumcircle to
triangle AED′ . Then, with notations as in Figure 23, we get
∠EDA = ∠ED′ A = ∠EF A = ∠F EB + ∠EBA > ∠EBA
where we used that angles subtended by the same chord on the same seg-
ment of a circle are equal, and the exterior angle theorem. We have reached a
contradiction: ∠EBA = ∠EDA and ∠EBA < ∠EDA. Hence the assump-
tion that AP < CP was wrong, and in the same way we can prove that
106 Martin Josefsson

Figure 23. ∠B = ∠D, BP = DP and BD ̸⊥ AC

AP > CP is also wrong. Thus AP = CP , which together with BP = DP

implies that ABCD is a parallelogram according to Theorem 3.1 (a). □

8. Two-dimensional metric relations

In the following theorem we prove twelve characterizations of parallelo-
grams that are two-dimensional metric relations. Several of these were dis-
covered while conducting the research before and during the writing of this
paper. In a way, (b) is the most fundamental of all necessary and sufficient
conditions of parallelograms since it’s expressed as only one equality and in
terms of the sides alone. (d) is quite famous and one of the oldest character-
izations, due to Euler according to [29, p. 35], and to prove the inequality
case in (f) was a problem on the Mediterranean Mathematics Olympiad in
the year 2000 [4]. Condition (g) is from the book [35, p. 188] and (h) was
proved in [24] as a part of the proof of the converse of our Theorem 3.1 (e).
The authors of [9] remember seeing (i) as a problem in some journal, but
they cannot recall a reference, and we have not been able to track it down
either. Characterization (j) is from the Croatian Mathematical Olympiad
in 1996 [16], while (l) and its proof are cited from [28, p. 187].
Theorem 8.1. In a quadrilateral ABCD with sides a = AB, b = BC,
c = CD, d = DA; diagonals p, q; bimedians m, n; semiperimeter s, and
area K, let Ma , Mb , Mc , Md be the midpoints of a, b, c, d respectively,
mA = AMb , mB = BMc , mC = CMd , mD = DMa be the medians in
triangles ABC, BCD, CDA, DAB respectively, and Ta , Tb , Tc , Td be the
areas of triangles ABP , BCP , CDP , DAP respectively, where P is the
diagonal intersection. Then ABCD satisfies any one of:
(a) ab = cd and da = bc
(b) (a − c)2 + (b − d)2 = 0
(c) (a + c)2 + (b + d)2 = 2(p2 + q 2 )
 Characterizations of parallelograms part 1 107

(d) a2 + b2 + c2 + d2 = p2 + q 2
(e) a2 + b2 + c2 + d2 = 2(m2 + n2 )
(f) a2 + b2 + c2 + d2 = 54 (m2A + m2B + m2C + m2D )
(g) Ta = Tb = Tc = Td
(h) Ta = Tc and Tb = Td
(i) Tb = Td = 14 K
(j) Tb2 = T√a Tc and 2Td = Ta + Tc
(k) K = 2 (s2 − p2 )(s2 − q 2 )
1

(l) TABC ≤ TCDA ≤ TBCD ≤ TDAB

if and only if it’s a parallelogram.
Proof. (a) We solve the system of equations
{ { {
ab = cd a2 bd = bc2 d a=c
⇔ ab cd ⇔
da = bc da = bc b=d
which proves that the quadrilateral is a parallelogram if and only if these
two equations are satisfied according to Theorem 2.1 (a).
(b) It’s trivial that (a − c)2 + (b − d)2 = 0 holds in a parallelogram.
Conversely, since algebraic squares are non-negative for real numbers, there
is just one solution to the equation (a − c)2 + (b − d)2 = 0, and that is for
both terms to be zero, yielding a = c and b = d. This implies that the
quadrilateral is a parallelogram according to Theorem 2.1 (a).
(c) Theorem 10 in [18] states that in a convex quadrilateral with sides a,
b, c, d and diagonals p, q, it holds that
p2 + q 2 = b2 + d2 + 2ac cos ξ
where ξ is the angle between the extensions of sides a and c. By symmetry
we have
p2 + q 2 = a2 + c2 + 2bd cos ψ
where ψ is the angle between the extensions of sides b and d. Adding these
two equations yields
2(p2 + q 2 ) = a2 + c2 + b2 + d2 + 2ac cos ξ + 2bd cos ψ
and using the assumption (a + c)2 + (b + d)2 = 2(p2 + q 2 ), we get after
expansion and simplification
2ac + 2bd = 2ac cos ξ + 2bd cos ψ
which is equivalent to
2ac(cos ξ − 1) = 2bd(1 − cos ψ)
and, in turn, holds if and only if
( ) ( )
2 ξ 2 ψ
4ac sin + 4bd sin = 0.
2 2
This equation only has one possible solution, that both terms on the left
hand side are zero, which is equivalent to ξ = ψ = 0. Hence the only type
of quadrilateral where the equality (a + c)2 + (b + d)2 = 2(p2 + q 2 ) holds is
a parallelogram according to Theorem 4.1 (d).
108 Martin Josefsson

(d) This characterization is a direct consequence of Euler’s quadrilateral

theorem (for a proof, see [3, pp. 9–10])
(8) a2 + b2 + c2 + d2 = p2 + q 2 + 4v 2 ≥ p2 + q 2
with equality if and only if the distance v between the diagonal midpoints is
zero, which is equivalent to a parallelogram according to Theorem 3.1 (a).
(e) It’s well known that the midpoints of the sides of a quadrilateral are
the vertices of a parallelogram (discovered by Varignon), whose diagonals
are the bimedians of the quadrilateral and that its sides are half the length
of the diagonals of the quadrilateral. Applying the equality from (d) in this
Varignon parallelogram, we get
( p )2 ( q )2 ( p )2 ( q )2
+ + + = m2 + n2 ,
2 2 2 2
that is,
(9) p2 + q 2 = 2(m2 + n2 ).
Inserting this into (8) yields
a2 + b2 + c2 + d2 = 2(m2 + n2 ) + 4v 2 ≥ 2(m2 + n2 )
with equality if and only if the quadrilateral is a parallelogram (v = 0).
(f) Applying the triangle median formula (see Figure 24), we get that
4(m2A + m2B + m2C + m2D )
= 2(a2 + p2 ) − b2 + 2(b2 + q 2 ) − c2 + 2(c2 + p2 ) − d2 + 2(d2 + q 2 ) − a2
= a2 + b2 + c2 + d2 + 4(p2 + q 2 )
= a2 + b2 + c2 + d2 + 4(a2 + b2 + c2 + d2 − 4v 2 )
= 5(a2 + b2 + c2 + d2 ) − 16v 2 ≤ 5(a2 + b2 + c2 + d2 )
where we inserted (8) in the third step. We have equality if and only if the
quadrilateral is a parallelogram (v = 0, according to Theorem 3.1 (a)).

Figure 24. The medians mA , mB , mC , mD

(g) Since the diagonals bisect each other in a parallelogram, adjacent

subtriangles have equal area due to equal heights, so Ta = Tb = Tc = Td .
 Characterizations of parallelograms part 1 109

Conversely, we have
{ {
Ta = Tb w=y
⇒
Ta = Td x=z
where w = AP , x = BP , y = CP , z = DP are the diagonal parts and
we used that adjacent subtriangles have equal heights (see Figure 25). This
proves that in fact only three of the four subtriangles need to have equal area
in order for us to be able to conclude that the quadrilateral is a parallelogram
according to Theorem 3.1 (a).

Figure 25. The diagonal parts and the four subtriangle areas

(h) Applying a well-known formula for the area of a quadrilateral (see

Corollary 1.5.2 in [3, p. 15]), we have that
{ {
1 1
Ta = Tc 2 wx sin θ = 2 yz sin θ
⇔
2 xy sin (π − θ) = 2 zw sin (π − θ)
1 1
Tb = Td
where w, x, y, z are the diagonal parts and θ is one of the angles between
the diagonals (see Figure 25). This is equivalent to
{ { {
wx = yz wx2 y = wyz 2 x=z
⇔ wx yz ⇔
xy = zw xy = zw y=w
which holds if and only if the quadrilateral is a parallelogram according to
Theorem 3.1 (a).
(i) It’s trivial that Tb = Td = 14 K are satisfied in a parallelogram according
to (g). For the converse, we rewrite these equalities as
1
2 wz sin θ = 21 xy sin θ = 1
4 · 12 (w + y)(x + z) sin θ
(see Figure 25), which is equivalent to
4wz = 4xy = wx + wz + yx + yz.
Hence by simplification, we have the system of equations
{
wz = xy
(10)
2wz = wx + yz
110 Martin Josefsson

which we solve by writing 2wz as wz + wz and use wz = xy for one of

the terms. Then the second equation yields wz + xy = wx + yz, which is
equivalent to
(z − x)(w − y) = 0
with solutions z = x and then y = w, or w = y which implies that z = x.
These identical cases imply the quadrilateral is a parallelogram according to
Theorem 3.1 (a).
(j) In a parallelogram, Ta = Tb = Tc = Td according to (h), so both
2
Tb = Ta Tc and 2Td = Ta + Tc are satisfied.
Conversely, rewriting the equality Tb2 = Ta Tc , we get (see Figure 25)
(1 )2 1
2 xy sin (π − θ) = 2 wx sin θ · 21 yz sin θ
which is equivalent to xy(xy − wz) = 0. Hence xy = wz since xy ̸= 0.
The second equality 2Td = Ta + Tc can in the same way be rewritten as
2wz = wx + yz, so again we have the system of equations (10). In the proof
of (j) we showed that it implies a parallelogram.
(k) In [20] it was proved that the area of a convex quadrilateral can be
calculated with the formula
√
K = 12 [(m + n)2 − p2 ][p2 − (m − n)2 ].
This can, with the help of (9), be rewritten as
√ √
K = 12 [(m + n)2 − p2 ][(m + n)2 − q 2 ] ≤ 21 (s2 − p2 )(s2 − q 2 )
where equality holds if and only if the quadrilateral is a parallelogram ac-
cording to (7).
(l) In a parallelogram we obviously have equality throughout. Conversely,
from TABC ≤ TBCD we get TABP ≤ TCDP by subtracting the common
area of triangle BCP (see Figure 25). In the same way TCDA ≤ TDAB
yields TCDP ≤ TABP , so TABP ≤ TCDP ≤ TABP and thus we must have
TCDP = TABP . It follows that TCDA = TDAB , so C and B have the same
distance to DA, that is, CB ∥ DA. In the same way it’s proved that
AB ∥ DC, so ABCD is a parallelogram by definition. □

9. Part 2
In the second part of this compilation of characterizations of parallelo-
grams we will study characterizations that are about one-dimensional metric
relations, symmetry, vectors, and coincidences.

References
[1] Al-Sharif, A. I.; Hajja, M.; Krasopoulos, P. T., Coincidences of Centers of Plane
Quadrilaterals, Results Math., 55(3) (Nov. 2009) 231–247.
[2] Alsina, C. and Nelsen, R. B., Charming Proofs. A Journey into Elegant Mathematics,
MAA Press, 2010.
[3] Alsina, C. and Nelsen, R. B., A Cornucopia of Quadrilaterals, MAA Press, 2020.
[4] Andreescu, T.; Feng, Z.; Lee Jr., G., Mathematical Olympiads 2000–2001. Problems
and Solutions From Around the World, MAA, 2003.
 Characterizations of parallelograms part 1 111

[5] Cola, C., Given a quadrilateral with 4 equal areas, prove that it is a parallelogram,
Mathematics Stack Exchange, 2021,
https://math.stackexchange.com/questions/4057138/
[6] De Villiers, M., Some Adventures in Euclidean Geometry, Lulu Press, 2009.
[7] De Villiers, M., Some more properties of the bisect-diagonal quadrilateral, Math.
Gaz., 105 (Nov. 2021) 474–480.
[8] Gant, P., When is a Parallelogram Not a Parallelogram?, Math. Gaz., 39 (Sep. 1955)
191–195.
[9] Hajja, M. and Krasopoulos, P. T., More characterisations of parallelograms, Math.
Gaz., 107 (March 2023) 76–83.
[10] Halsted, G. B., Elementary Synthetic Geometry, John Wiley & Sons, New York,
1896.
[11] Hang, K. H. and Wang, H., Solving Problems in Geometry. Insights and Strategies
for Mathematical Olympiad, World Scientific, Singapore, 2018.
[12] Heath, T. L., The Thirteen Books of The Elements, Vol. 1 (Books I and II), Dover
Publications, New York, 1956.
[13] Hirschhorn, D. B., Why Is the SsA Triangle-Congruence Theorem Not Included in
Textbooks?, Math. Teacher, 83(5) (May 1990) 358–361.
[14] Hoehn, L., Problem 515, The College Mathematics Journal, 24(5) (Nov. 1993) 473;
Solution ibid, 25(5) (Nov. 1994) 467–468.
[15] IMOmath, 45-th Moldova Mathematical Olympiad 2001, n.d., available at:
https://imomath.com/othercomp/Mda/MdaMO01.pdf
[16] jasperE3 (username), area chasing iff parallelogram, AoPS Online, 2021,
https://artofproblemsolving.com/community/c6h2728683p23760535
[17] Josefsson, M., Similar metric characterizations of tangential and extangential quadri-
laterals, Forum Geom., 12 (2012) 63–77.
[18] Josefsson, M., Characterizations of trapezoids, Forum Geom., 13 (2013) 23–35.
[19] Josefsson, M., On the classification of convex quadrilaterals, Math. Gaz., 100 (March
2016) 68–85.
[20] Josefsson, M., Heron-like formulas for quadrilaterals, Math. Gaz., 100 (Nov. 2016)
505–508.
[21] Josefsson, M., Properties of bisect-diagonal quadrilaterals, Math. Gaz., 101 (July
2017) 214–226.
[22] Joyce, D. E., Euclid’s Elements. Book I, 1996,
https://mathcs.clarku.edu/~djoyce/java/elements/bookI/bookI.html
[23] Klamkin, M. S., The Olympiad Corner, Crux Math., 10(9) (Nov. 1984) 289.
[24] Klamkin, M. S., The Olympiad Corner, Crux Math., 11(1) (Jan. 1985) 12–13.
[25] MacTutor, Earliest Known Uses of Some of the Words of Mathematics (P), University
of St Andrews, Scotland, n.d.,
https://mathshistory.st-andrews.ac.uk/Miller/mathword/p/
[26] Mead, G. F., The Mathematical Visitor, 1(4) (Jan. 1880) 165.
[27] Okumura, H., A Characterisation of Parallelograms, Math. Gaz., 72 (June 1988)
117–118.
[28] Pop, O. T.; Minculete, N.; Bencze, M., An introduction to quadrilateral geometry,
Editura Didactică şi Pedagogică, Bucharest, Romania, 2013.
[29] Sandifer, C. E., How Euler Did It, MAA, 2007.
[30] Saul, M., Hadamard’s Plane Geometry. A Reader’s Companion, AMS, 2010.
[31] Seimiya, T., Problem 1841, Crux Math., 20(4) (April 1994) 111–112.
[32] Todhunter, I., The elements of Euclid for the use of schools and colleges, Macmillan
& Co., London, 1871.
[33] Toumasis, C., When is a Quadrilateral a Parallelogram, Math. Teacher, 87(3) (March
1994) 208–211.
[34] Tydd, M., Flying Two Kites: Part 2, AMESA KZN Mathematics Journal, 9 (2005)
51–54, available at:
http://dynamicmathematicslearning.com/matthew-tydd-two-kites.pdf
[35] Ufnarovski, V.; Wikström, F.; Madjarova, J., Mathematical Buffet. Problem solving
at the Olympiad level, Studentlitteratur, Lund, Sweden, 2016.
112 Martin Josefsson

[36] Usiskin, Z. and Griffin, J., The Classification of Quadrilaterals. A Study of Definition,
Information Age Publishing, 2008.
[37] vinod19 (username), If the bisectors of angles of a quadrilateral enclose a rectangle,
then show it is a parallelogram, brainly, 2016,
https://brainly.in/question/1003966
[38] Wikipedia, accessed July 2023, https://en.wikipedia.org/wiki/Parallelogram
[39] Wong Y. L., MA2219 An Introduction to Geometry, 2009, available at:
https://bpb-us-w2.wpmucdn.com/blog.nus.edu.sg/dist/2/12659/files/2021/
08/Notes_MA2219-1.pdf
[40] Woodrow, R. E., 10th Mexican Mathematics Olympiad National Contest November
1996, Crux Math., 26(2) (2000) 71.
[41] Young, J. R., Elements of Geometry, with Notes, Carey, Lea & Blanchard, 1833.

SECONDARY SCHOOL KCM

MARKARYD, SWEDEN
E-mail address: [email protected]