Another synthetic proof of Lester’s theorem and some properties of

Fermat points
Ngo Quang Duong

Hanoi, Vietnam

August 1, 2022. Version 1.1

Abstract
This paper introduces another synthetic proof of Lester’s theorem by showing that the two Fermat points
are also the isodynamic points of the orthocentroidal triangle.

1 Introduction
In 1996, Professor June A. Lester discovered the following unexpected result:
Theorem 1.1 (Lester’s theorem, [1]). In every scalene triangle, the two Fermat points, the circumcenter and
the nine-point center are concyclic.

Most proofs of this theorem are computer-based or coordinate-based, including Lester’s, which used complex
coordinates. So far, there is only one published and synthetic proof for this theorem, which is given by Nikolai
Ivanov Beluhov [2].
Besides the lemma, the key idea in the proof given by N. I. Beluhov is that two Fermat points are inverses with
respect to the “orthocentroidal circle” − The circle whose diameter is the line segment joining the orthocenter
and centroid. To prove that, I will use a different approach. I will prove a stronger result: the two Fermat
points are the isodynamic points of the orthocentroidal triangle − The triangle whose vertices are the orthogonal
projections of centroid onto three altitudes.
But at first, we need to establish some properties of the Apollonius circles of the orthocentroidal triangle.
Besides, for consistency, throughout this paper, we name the points as follow:

• Let ABC be a scalene triangle and F1 , F2 be its first Fermat point, second Fermat point.
• H, G, O, N are the orthocenter, centroid, circumcenter, nine-point center of triangle ABC, respectively.

• D, E, F are the orthogonal projections of H on AG, BG, CG, respectively.

• X, Y , Z are the orthogonal projections of G on AH, BH, CH, respectively.
• A1 BC, A2 BC are equilateral triangles such that: A and A2 are on the same side of BC, A and A1 are
on opposite sides of BC.

• B1 CA, B2 CA are equilateral triangles such that: B and B2 are on the same side of CA, B and B1 are
on opposite sides of CA.
• C1 AB, C2 AB are equilateral triangles such that: C and C2 are on the same side of AB, C and C1 are on
opposite sides of AB.
• X1 , X2 , Y1 , Y2 , Z1 , Z2 are the centroids of triangles A1 BC, A2 BC, AB1 C, AB2 C, ABC1 , ABC2 respec-
tively.

Points’ names used in proofs are also preserved.

1
 A
Hb A
Hc D
H
Y Z Y Z

X G G
X

B Ha Ma C B H a Ma C

Figure 1

2 Fermat points as Isodynamic points

Theorem 2.1. A, D lie on the X-Apollonius circle of triangle XY Z.

Proof. Since X, Y , Z, H, G are concyclic

(Y Z, Y X) ≡ (GZ, GX) ≡ (AB, BC) ≡ −(BC, BA) (mod π),

(Y Z, XZ) ≡ (GY, GX) ≡ (CA, BC) ≡ −(BC, AC) (mod π),

then two triangles ABC and XY Z are inversely similar.

Apply Pythagoras’s theorem to triangles Hc AZ and Hb AY :
1 1
AY 2 = AHb 2 + Hb Y 2 = AB 2 cos2 A + BHb 2 = AB 2 cos2 A + AB 2 sin2 A,
9 9
1 1
2 2 2 2
AZ = AHc + Hc Z = AC cos A + CHc = AC cos A + AC 2 sin2 A.
2 2 2 2
9 9
From the above formulas, we deduce that:
AY AB XY
= = .
AZ AC XZ
This implies that A lies on the X-Apollonius circle of triangle XY Z.
If ∠BAC is right angle, then A coincides with D, and A is the reflection of X in Y Z. Therefore, A, D lies
on the X-Apollonius circle of triangle XY Z.
Otherwise, ∠BAC is non-right angle, then A, D, X are pairwise distinct.
Let Ma be the midpoint of BC. Let AHa , BHb , CHc be the altitudes of triangle ABC.
According to the intersecting chords/secants theorem:

AD · AMa = AH · AHa = AB · AHc = AC · AHb .

Therefore, B, Ma , Hc , D are concyclic and C, Ma , Hb , D are concyclic.

(DB, DMa ) ≡ (Hc B, Hc Ma ) ≡ (BMa , BHc ) ≡ (BMa , BA) (mod π),

(DC, DMa ) ≡ (Hb C, Hb Ma ) ≡ (CMa , CHb ) ≡ (CMa , CA) (mod π),

then triangles Ma BD and Ma AB are inversely similar, triangles Ma CD and Ma AC are inversely similar
(angle-angle). From these results and the concyclicity of X, Y , Z, G, H, D, we can do angle chasing.

(Y Z, Y D) ≡ (GZ, GD) ≡ (AB, AMa ) ≡ (BMa , BD) ≡ (BC, BD) (mod π),
(Y Z, DZ) ≡ (GY, GD) ≡ (AC, AMa ) ≡ (CMa , CD) ≡ (BC, DC) (mod π).

Therefore, triangles DY Z and DBC are directly similar (angle-angle). Due to the pairs of similar triangles:

DY DB DB AB AC BMa AB AMa AB XY
≡ ≡ · · = · · = = .
DZ DC AB AC DC AMa AC CMa AC XZ
So D lies on the X-Apollonius circle of triangle XY Z.
In conclusion, A and D lie on the X-Apollonius circle of triangle XY Z.

2
 A
A2 A2

F2

A≡D

G X2 X2
X
U2
D X G
H
B Ha Ma C B Ha Ma C

F2
U2

Figure 2

If ∠BAC is non-right angle, we can construct the X-Apollonius circle of triangle XY Z as the circumcircle
of triangle ADX since these three points are pairwise distinct. In the edge case, ∠BAC is right angle, the
X-Apollonius circle of triangle XY Z is the circle that passes through A, X and tangent to AMa at A. We will
use these remarks to prove the following result.
Theorem 2.2. F1 and F2 lies on the X-Apollonius circle of triangle XY Z.

Proof. Let U2 be point on Ma F2 such that Ma U2 · Ma F2 = 41 BC 2 .

It is well-known that B, C, A2 , F2 are concyclic and A, F2 , A2 are collinear. The inversion with center Ma
and radius 12 BC maps B, C, A2 , F2 to B, C, X2 , U2 , respectively. Therefore, B, C, X2 , U2 are concyclic.
Ma U2 · Ma F2 = Ma X2 · Ma A2 = 14 BC 2 , then A2 , F2 , X2 , U2 are concyclic.
BC is the radical axis of the circumcircles of triangles X2 BC and A2 BC; AA2 is the radical axis of the
circumcircles of triangles A2 BC and A2 F2 X2 ; U2 X2 is the radical axis of the circumcircles of triangles X2 BC
and A2 F2 X2 . So that BC, AA2 , X2 U2 are concurrent or pairwise parallel.
On the one hand, G, X2 are the centroids of triangles ABC, A2 BC, respectively, and according to Thales’s
theorem:
AX AG 2 A2 X2
= = = .
AHa AMa 3 A2 Ma
On the other hand, AHa is parallel to A2 Ma . So that BC, AA2 , XX2 are concurrent or pairwise parallel.
Therefore, X, X2 , U2 are collinear.
(AX, AF2 ) ≡ (A2 X2 , A2 F2 ) ≡ (U2 X2 , U2 F2 ) ≡ (U2 X, U2 F2 ) (mod π).
Then A, X, F2 , U2 are concyclic.
Let’s consider two cases.

Case 1. ∠BAC is non-right angle.

∠BAC is non-right angle, then three points A, X, D are pairwise distinct.
According to the proof of Theorem 2.1, triangles Ma BD and Ma AB are inversely similar, then Ma A·Ma D =
Ma B 2 = 14 BC 2 .
Since Ma A · Ma D = 14 BC 2 = M F2 · M U2 , then A, D, U2 , F2 are concyclic. Therefore, A, D, X, F2 are
concyclic. So that F2 lies on the X-Apollonius circle of triangle XY Z.

Case 2. ∠BAC is right angle.

∠BAC is right angle, then A, H, D are coincident.
The inversion with center Ma and radius 21 BC maps A, A2 , F2 to A, X2 , U2 , respectively. Together with
the collinearity of A, F2 , A2 , we obtain that A, X2 , U2 , Ma are concyclic.
(AMa , AX) ≡ (AMa , X2 Ma ) ≡ (U2 A, U2 X2 ) ≡ (U2 A, U2 X) (mod π).
This means AMa is tangent to the circle through A, X, F2 , U2 , which means the circle coincides with the
X-Apollonius circle of triangle XY Z. Then F2 lies on the X-Apollonius circle of triangle XY Z.
In both cases, F2 lies on the X-Apollonius circle of triangle XY Z. By analogous constructions and argu-
ments, we obtain that F1 also lies on this circle.

3
 Back to Theorem 1.1.
A

H
Y
M Z
F1 N
X G

F2 O

B C

Figure 3

Proof of Theorem 1.1. From Theorem 2.1 and Theorem 2.2, we obtain that F1 , F2 lie on three Apollonius circles
of triangle XY Z. It follows that F1 , F2 are the isodynamic points of triangle XY Z. Since the orthocentroidal
circle is also the circumcircle of triangle XY Z, F1 , F2 are inverses with respect to the orthocentroidal circle.
Let M be the midpoint of HG. Then:
1
M F1 · M F2 = M G 2 = M G · 2M G = M N · M O.
2
From this, we conclude that F1 , F2 , N , O are concyclic.

3 Fermat triangle
When the measure of ∠BAC is other than π3 , 2π 3 , then three points A, F1 , F2 are pairwise distinct. Let’s call
triangle AF1 F2 the A-Fermat triangle (or Fermat triangle).
Previously, we proved that X-Apollonius circle of triangle XY Z passes through D (which is also known as
the A-Humpty point) and two Fermat points. Equivalently, D, X lies on the circumcircle of triangle AF1 F2 .
In this section, I introduce and re-introduce some other properties of this triangle.
Theorem 3.1 (N. I. Beluhov [3]). Euler line of triangle AF1 F2 passes through the centroid of triangle ABC.

In [3], the original theorem was: “Euler lines of 10 triangles with vertices chosen from A, B, C, F1 , F2 are
concurrent at the centroid of triangle ABC”. Beluhov divided ten triangles into three types:

Type I. ABC,
Type II. F1 BC, F2 BC, AF1 C, AF2 C, ABF1 , ABF2 ,
Type III. AF1 F2 , BF1 F2 , CF1 F2 ,

where Type III seems to be the hardest. I will re-produce and improve (with directed angles) Beluhov proof.
The main idea is showing G, the centroid and the circumcenter of triangle AF1 F2 are collinear by Playfair’s
axiom.

Proof. Let Oa , Ga be the circumcenter, centroid of triangle AF1 F2 , respectively.

Let F0 , Ma be the midpoints of line segments F1 F2 , BC, respectively.
Let F1′ be the reflection of F1 in Ma .
We will show that Oa G and GGa are parallel/coincident with F2 F1′ .

Step 1. Classic results.

4
 A

A2
Y1

Oa
Z1
Ga
Z2 F2
F0
G
Y2
X2
F1

B Ma C

X1 F1′

A1

Figure 4

Result 1. The circumcircles of triangles A1 BC, AB1 C, ABC1 pass through F1 .

Triangles ABC1 and AB1 C are equilateral, and each of these triangles has opposite orientation to
triangle ABC. Therefore, triangles ABC1 and AB1 C are directly similar. It follows that triangles
ABB1 and AC1 C are directly similar (side-angle-side).

(BB1 , CC1 ) ≡ (AB, AC1 )

≡ (A1 B, A1 C) (mod π)
(A1 BC and ABC1 are directly similar equilateral triangles).

The congruence implies that the intersection of BB1 and CC1 lies on the circumcircle of triangle
A1 BC, which means F1 , A1 , B, C are concyclic.
Analogously, F1 also lies on the circumcircles of triangles AB1 C and ABC1 .
Result 2. The circumcircles of triangles A2 BC, AB2 C, ABC2 pass through F2 .
Similar to Result 1.
Result 3. X1 Y1 Z1 and X2 Y2 Z2 are equilateral triangles. G is the centroid of these two triangles.
Since A1 BC, AB1 C, ABC1 are equilateral triangles, then X1 , Y1 , Z1 are their circumcenters, respec-
tively.
According to Result 1

(X1 Y1 , X1 Z1 ) ≡ (F1 C, F1 B) ≡ (A1 C, A1 B) (mod π),

(Y1 Z1 , Y1 X1 ) ≡ (F1 A, F1 C) ≡ (B1 A, B1 C) ≡ (BA1 , BC) (mod π).

So triangles X1 Y1 Z1 and A1 CB are directly similar, which means triangle X1 Y1 Z1 is equilateral.

On the one hand
Ma G Ma X 1 1
= = .
Ma A M a A1 3
According to Thales’s theorem, GX1 , AA1 are parallel, and
GX1 1
= .
AA1 3

5
 Similarly
GX1 GY1 GZ1 1
= = = .
AA1 BB1 CC1 3
On the other hand, the following pairs of triangles are directly congruent (side-angle-side)
• ABB1 and AC1 C.
• BCC1 and BA1 A.
• CAA1 and CB1 B.
Therefore, AA1 = BB1 = CC1 .
Hence, GX1 = GY1 = GZ1 , so G is the circumcenter of triangle X1 Y1 Z1 .
Analogously, triangles X2 Y2 Z2 and A2 CB are directly similar, then X2 Y2 Z2 is an equilateral triangle.
Furthermore, G is also the circumcenter of triangle X2 Y2 Z2 .
Result 4. X1 Y1 Z1 and X2 Y2 Z2 have opposite orientations. According to the proof of Result 3, triangles
X1 Y1 Z1 and A1 CB are directly similar, triangles X2 Y2 Z2 and A2 CB are directly similar.
On the other hand, triangle A1 CB and A2 CB are inversely similar.
Therefore, triangles X1 Y1 Z1 and X2 Y2 Z2 are inversely similar, which implies that they have opposite
orientations.

Step 2. Oa , G, Y1 , Z2 are concyclic.

Y1 Z1 , Y2 Z2 are the perpendicular bisectors of line segments AF1 , AF2 , respectively. So Oa is the inter-
section of Y1 Z1 and Y2 Z2 .
Since X1 Y1 Z1 and X2 Y2 Z2 are inversely similar equilateral triangles, and they share the same circumcenter
G, then triangles GY1 Z1 and GY2 Z2 are directly similar.

(Y1 Oa , Y1 G) ≡ (Y1 Z1 , Y1 G)
≡ −(Y2 Z2 , Y2 G)
≡ (Y2 Z2 , GZ2 )
(GY2 Z2 is an isosceles triangle)
≡ (Z2 Oa , Z2 G) (mod π).

Hence, Oa , G, Y1 , Z2 are concyclic.

Step 3. GGa and F2 F1′ are parallel/coincident.
Remind that F0 is the midpoint of F1 F2 and Ga is the centroid of triangle AF1 F2 .
AG AGa 2
Since = = . According to Thales’s theorem, GGa is parallel/coincident with Ma F0 .
AMa AF0 3
F1 M a F1 F0 1
On the other hand, = = , then according to Thales’s theorem, Ma F0 is parallel/coincident
′
F1 F1 F1 F2 2
with F2 F1′ .
Hence GGa and F2 F1′ are parallel/coincident.
Step 4. GOa , F2 F1′ are parallel/coincident.
Notice that A2 , B, C, F2 , F1′ are concyclic.

(GOa , F2 F1′ ) ≡ (GOa , GZ2 ) + (GZ2 , F2 F1′ ) (mod π)

Since Oa , G, Y1 , Z2 are concyclic and GZ2 is parallel/coincident with F2 C, then

(GOa , F2 F1′ ) ≡ (Y1 Oa , Y1 Z2 ) + (F2 C, F2 F1′ ) (mod π)

Since B, C, F2 , F1′ are concyclic, then

(GOa , F2 F1′ ) ≡ (Y1 Z1 , Y1 Z2 ) + (BC, BF1′ ) (mod π)

Since BF1′ and CF1 are parallel/coincident, then

(GOa , F2 F1′ ) ≡ (Y1 Z1 , Y1 Z2 ) + (BC, CF1 ) (mod π).

6
 Two isoceles triangles Z2 AB and Y1 AC are directly similar (side-angle-side). Therefore, triangles ABC
and AZ2 Y1 are directly similar. It follows that

(GOa , F2 F1′ ) ≡ (Y1 Z1 , BC) + (BC, Z2 Y1 ) + (BC, CF1 )

≡ (Y1 Z1 , AF1 ) + (AF1 , BC) + (AB, AZ2 ) + (BC, CF1 ) (mod π).

Since Y1 Z1 is perpendicular to AF1 , then

π
(GOa , F2 F1′ ) ≡ + (AF1 , CF1 ) + (AB, AZ2 ) (mod π).
2
Since F1 , A, B1 , C are concyclic, and AZ2 is perpendicular to BC2 , then

(GOa , F2 F1′ ) ≡ (AZ2 , BC2 ) + (AB1 , CB1 ) + (AB, AZ2 )

≡ (AB1 , CB1 ) + (AB, C2 B) (mod π).

Two equilateral triangles AB1 C and ABC2 have opposite orientations, therefore

(AB1 , CB1 ) + (AB, C2 B) ≡ 0 (mod π).

Hence (GOa , F2 F1′ ) ≡ 0 (mod π), which implies that GOa and F2 F1′ are parallel/coincident.

According to Playfair’s axiom, G, Ga , Oa are collinear. Equivalently, the Euler line of triangle AF1 F2 passes
through G.
I used Theorem 3.1 as another approach to the following.
Theorem 3.2 (Telv Cohl [4]). X is the Euler reflection point of triangle AF1 F2 . The Parry reflection point of
triangle AF1 F2 lies on BC.
The idea is a classic property of Wallace-Simson/Steiner line1 .
Lemma 3.3. Let O be the circumcenter of triangle ABC. Denote the Wallace-Simson/Steiner line of P with
respect to triangle ABC by s(ABC, P ). Then
1
(s(ABC, P ), s(ABC, Q)) ≡ (OP, OQ) (mod π).2
2
Proof of Lemma 3.3. I will prove that
1
(s(ABC, P ), s(ABC, A)) ≡ − (OP, OA) (mod π)
2
to derive the desire result.

Case 1. P coincides with A.

1
(s(ABC, P ), s(ABC, A)) ≡ 0 ≡ − (OP, OA) (mod π).
2

Case 2. P coincides with B.

If P coincides with a vertice of triangle ABC, then the Wallace-Simson line is also the altitude corre-
sponding to that vertice. Therefore, s(ABC, P ), s(ABC, A) are perpendicular to CA, BC, respectively.

(s(ABC, P ), s(ABC, A)) ≡ (CA, CB) (mod π).

According to inscribed angle theorem

1 1 1
(CA, CB) ≡ (OA, OB) ≡ (OA, OP) ≡ − (OP, OA) (mod π).
2 2 2
1 Orthogonal projections of a point on the circumcircle of a triangle to the sidelines of the triangle lie on a straight line. This

line is called the Wallace line (or Wallace-Simson line) of that point with respect to the triangle. Whilst the Steiner line passes
through the reflections of that point in the sidelines of the triangle.
2 For those who aren’t familiar with bold notation: AB is the notation of vector AB, and (OP, OQ) (mod 2π) is directed

angle (more precisely, trigonometric angle) between vectors.

7
 Case 3. P coincides with C. Similar to the previous case
(s(ABC, P ), s(ABC, C)) ≡ (AB, CB)
1
≡ (OA, OC)
2
1
≡ (OA, OP)
2
1
≡ − (OP, OA) (mod π).
2
Case 4. P does not coincide with A, B, or C. Let D, E, F be the orthogonal projections of P on BC, CA,
AB, respectively.
(s(ABC, P ), s(ABC, A)) ≡ (DE, P D)
≡ (CE, CP ) (C, P , D, E are concyclic)
≡ (CA, CP )
1
≡ (OA, OP)
2
1
≡ − (OP, OA) (mod π).
2
Hence
(s(ABC, P ), s(ABC, Q)) ≡ (s(ABC, P ), s(ABC, A)) + (s(ABC, A), s(ABC, Q))
1 1
≡ − (OP, OA) − (OA, OQ)
2 2
1
≡ − (OP, OQ) (mod π).
2

A2
Y1

Oa
Z1
Ga F2
Z2

G
X Y
2
X2
F1

B Ma C

X1 F1′

A1

Figure 5

Proof. (i) To prove X is the Euler reflection point is to prove GOa is the Steiner line of X with respect to
triangle AF1 F2 .

8
 According to Lemma 3.3 and inscribed angle theorem

(s(AF1 F2 , D), s(AF1 F2 , F1 )) ≡ (AF1 , AD)

≡ (A1 F1 , A1 X2 ) (mod π)

On the other hand, F1 F1′ and A1 A2 share the same midpoint. Therefore, A1 F1 and A2 F1′ are paral-
lel/coincident, then

(A1 F1 , A1 X2 ) ≡ (A2 F1′ , A2 X1 )

≡ (F2 F1′ , F2 X1 ) (mod π).

Meanwhile, according to Theorem 3.1, F2 F1′ and GOa are parallel/coincident; F2 X1 and Y2 Z2 are paral-
lel/coincident since they are perpendicular to AF2 .
Therefore

(s(AF1 F2 , X), s(AF1 F2 , F1 )) ≡ (GOa , Y2 Z2 ) (mod π).

Since s(AF1 F2 , F1 ) and Y2 Z2 are parallel/coincident, then s(AF1 F2 , X) and GOa are parallel/coincident.
GOa is the Euler line of triangle AF1 F2 , then X is the Euler reflection point of triangle AF1 F2 .

(ii) The circumcenter of triangle AF1 F2 lies on the perpendicular bisector of AX.

Oa
Ga
F2

X G

F1

P B Ha Ma C

Figure 6

Let P be the Parry reflection point of triangle AF1 F2 .

The Parry reflection point of a triangle is the reflection of the circumcenter in the Euler reflection point.
So X is the midpoint of line segment P Oa .
AX
Since AH = 23 , then the perpendicular bisector of line segment AX is the reflection of the straight line
a
BC in X. Moreover, Oa lies on the perpendicular bisector of line segment AX. Therefore, P lies on BC.
Thus, the Parry reflection point of triangle AF1 F2 lies on BC.

Inspired by Theorem 3.2, I wonder if D is also a triangle center of triangle AF1 F2 . I don’t have an answer
yet. However, I found and managed to prove the following property of D.
Theorem 3.4. Steiner line of D with respect to triangle AF1 F2 passes through the orthocenter of triangle ABC.

Similar to Theorem 3.2, I also use Lemma 3.3.

Proof. Let Ka be the orthocenter of triangle AF1 F2 .
Our goal is proving (HKa , GKa ) ≡ (AX, AD) (mod π), from which we can apply Lemma 3.3.
According to Lemma 3.3

9
 A

Oa
Ga
F2
Ka
X G

F1
D
H

B Ma C

Figure 7

1
(Ka A, Ka G) ≡ (s(AF1 F2 , A), s(AF1 F2 , X)) ≡ − (Oa A, Oa X) (mod π).
2
By inscribed angle theorem

(Ka A, Ka G) ≡ (DX, DA) ≡ (DX, DG) (mod π).

Besides, H, G, X, D lie on the circle whose diameter is HG, so

(DX, DG) ≡ (HX, HG) ≡ (HA, HG) (mod π).

Hence, (Ka A, Ka G) ≡ (HA, HG) (mod π). This implies that A, H, G, Ka are concyclic.
According to Lemma 3.3

1
(s(AF1 F2 , D), s(AF1 F2 , X)) ≡ − (Oa D, Oa X)
2
≡ (AX, AD)
≡ (AH, AG)
≡ (HKa , GKa ) (mod π).

Since s(AF1 F2 , X) and GKa are parallel/coincident, then s(AF1 F2 , D) and HKa are parallel/coincident.
Thus, the Steiner line of D with respect to triangle AF1 F2 passes through H.

4 Conclusion
So, in this paper, I provide another proof for the Lester’s theorem by some properties of the so-called Fermat
triangle and digging a little more into it. There are other properties of this Fermat triangle, for example: the
isodynamic points of triangle AF1 F2 lie on AB, AC. However, I pause here for now. Readers might want to check
out more properties in this post: https://artofproblemsolving.com/community/c6h1329709p7205788.

References
[1] J. A. Lester, Triangles III: Complex triangle functions. Aequationes Mathematicae 53 (1997), 4–35.

[2] N. I. Beluhov, An elementary proof of Lester’s theorem. Journal of Classical Geometry 1 (2012), 53–56.
[3] N. I. Beluhov, Ten concurrent Euler lines. Forum Geometricorum 9 (2009), 271–274.
[4] TelvCohl, Small problem on Parry reflection point, https://artofproblemsolving.com/community/c6h1
396744p7801781

10