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

net/publication/344768595

A SURVEY ON CONICS IN HISTORICAL CONTEXT: an overview of definitions

and their relationships

Preprint · August 2020

CITATIONS READS

0 699

2 authors, including:

Dirk Keppens
KU Leuven
27 PUBLICATIONS 80 CITATIONS

SEE PROFILE

All content following this page was uploaded by Dirk Keppens on 26 March 2021.

The user has requested enhancement of the downloaded file.

 INTERNATIONAL JOURNAL OF GEOMETRY
Vol. 10 (2021), No. 2, 50 - 66

A SURVEY ON CONICS IN HISTORICAL CONTEXT:

an overview of definitions and their relationships.

DIRK KEPPENS and NELE KEPPENS1

Abstract. Conics are undoubtedly one of the most studied objects in ge-
ometry. Throughout history different definitions have been given, depending
on the context in which a conic is seen. Indeed, conics can be defined in
many ways: as conic sections in three-dimensional space, as loci of points in
the euclidean plane, as algebraic curves of the second degree, as geometrical
configurations in (desarguesian or non-desarguesian) projective planes, . . .
In this paper we give an overview of all these definitions and their interre-
lationships (without proofs), starting with conic sections in ancient Greece
and ending with ovals in modern times.

1. Introduction
There is a vast amount of literature on conic sections, the majority of
which is referring to the groundbreaking work of Apollonius. Some influen-
tial books were e.g. [49], [13], [51], [6], [55], [59] and [26]. This paper is an
attempt to bring together the variety of existing definitions for the notion of
a conic and to give a comparative overview of all known results concerning
their connections, scattered in the literature. The proofs of the relationships
can be found in the references and are not repeated in this article. In the
first section we consider the oldest definition for a conic, due to Apollonius,
as the section of a plane with a cone. Thereafter we look at conics as loci of
points in the euclidean plane and as algebraic curves of the second degree
going back to Descartes and some of his contemporaries. Next we enlighten
the connection between conic sections and conics as discovered by Dandelin.
Then we focus on two important projective definitions for conics in pappian
planes, attributed to Steiner and Von Staudt. Closely connected with the
notion of a conic is that of an oval. We formulate the famous theorem of
Segre on ovals in finite planes over Galois fields of odd order and we also
mention a nice characterization theorem due to Buekenhout which has its
origin in a theorem of Pascal concerning inscribed hexagons in a conic.
————————————
Keywords and phrases: Conic section, Conic, Oval, Euclidean plane, Projective
plane, History of Mathematics
(2010)Mathematics Subject Classification: 51A30, 51A35, 51E15, 51M04
Received: 01.08.2020. In revised form: 24.01.2021 Accepted: 19.10.2020
1Part of the research was conducted for the second author’s Master’s thesis.
 A survey on conics in historical context 51

Both the definitions of Steiner and Von Staudt can be generalized to non-
pappian projective planes. This topic is discussed in another section with
several interesting results obtained by Strambach and with an algebraic
description of conics in non-desarguesian planes due to Krüger. Finally we
look at conics and ovals in topological planes where a theorem of Buchanan
on ovals in the complex plane surprisingly interfaces with Segre’s theorem
for ovals in finite desarguesian planes.

2. Apollonius’ definition: conics as conic sections

Conics or conic sections are geometrical objects with a long history, dating
back to the ancient Greeks. It is generally accepted that conics appeared for
the first time in the work of Menaichmos, who lived in the fourth century
BC. He discovered the curves as a by–product of his search for a solution
to the Delian problem of doubling the cube. One of the most prominent
geometers of the Greek tradition was Apollonius from Perga (±262−190
BC). He was born in the city of Bergama in Turkey, in ancient times known
as Pergamon and an important center for hellenistic culture. Apollonius’
Conica, a work consisting of eight books, seven of which are preserved, was
of great importance and its influence on later work was enormous2.
The Conica was the first systematic study of conics in the style of the famous
Elementa of Euclides. The oldest preserved manuscript referring to books I
to IV is “Vaticanus graecus 206” dating from the 12th century and stored in
the Biblioteca Apostolica Vaticana. The content of books V to VII is only
known from Arabic translations, e.g. the ones kept in the Bodleian Library
of the University of Oxford under the name “MS Marsh 667”.
Apollonius defined conics as conic sections, i.e. as the intersection curves of
a (right or oblique) double–napped cone with a plane not passing through
the top of the cone. The type of the conic section (ellipse, parabola or
hyperbola) depends on the inclination angle of the plane. So the setting of
Apollonius’ definition of a plane conic is in fact a spatial one. One needs
the three-dimensional euclidean space (see figure 1).

Figure 1. Conic sections by Apollonius: intersection curves

of a (double) cone and a plane.
2In the Renaissance period there was a renewed interest for the ancient Greeks. Several
translations in Latin of books I to IV were made, the first of which in 1537 by Giovanni
Battista Memmo and another in 1548 by Francesco Maurolico. The most important
translation of the 16th century is due to Federico Commandino. His Apollonii Pergaei
Conicorum libri quattuor had a leading position till the 18th century.
52 Dirk Keppens and Nele Keppens

3. Conics as loci in the euclidean plane and conics as

algebraic curves
In the (real) euclidean plane conics are defined as loci of points. This
definition can be found in any textbook on euclidean geometry. An ellipse is
the locus of points for which the sum of the distances to two given points has
a constant value. So it is the set of points P for which d(P, F )+d(P, F ′ ) = 2a
with 2a > 0 a fixed constant. An hyperbola is the locus of points for which
the absolute value of the difference of the distances to two given points
is a constant: |d(P, F ) − d(P, F ′ )| = 2a. The points F and F ′ are called
“foci” and the line joining them, assuming F ̸= F ′ , is the “major axis” (a
circle is a special case of an ellipse for F = F ′ ). The name “focus” has been
introduced by Kepler. In Ad Vitellionem Paralipomena, quibus Astronomiae
pars Optica Traditur (1604) he studies optical problems but in the fourth
chapter general properties of conics are discussed and there the word focus
appears for the first time. A parabola is the locus of points for which the
distance to a given point F is equal to the distance to a given line ℓ (with
F not on ℓ). The point F and the line ℓ are called “focus” and “directrix”
respectively. A parabola can also be interpreted as the limiting case of a
one–parameter family of ellipses Eλ(F, F ′ , 2aλ) with F a fixed point and F ′
going to infinity along the major axis while 2aλ−d(F, F ′ ) remains fixed. The
three separate definitions are called the “classical definitions of a conic as
point loci ” (or the “two foci definitions” (for ellipse and hyperbola) (figure
2).

Figure 2. Classical definition of conics as point loci in the

euclidean plane.

By a suitable choice of a coordinate system in the plane, a conic defined as

point locus, can be represented algebraically by a standard equation of the
second degree with real coëfficiënts. For a parabola the standard equation
is y 2 = 2px (with p the distance from F to ℓ). For the ellipse and the
2 2
hyperbola the standard equations are xa2 + yb2 = 1 with a2 − c2 = b2 and
 A survey on conics in historical context 53

2
x2
a2
− yb2 = 1 with c2 − a2 = b2 respectively (where 2c is the distance between
the foci F and F ′ ). Conversely, any algebraic equation of one of these three
forms, represents a conic. The proofs for this are elementary.
The interpretation of conics in the euclidean plane as algebraic curves of
the second degree goes back to the pioneering work of Pierre de Fermat
(1607–1665): Ad locos planos et solidos Isagoge from 1629 and published
posthumously in 1679. Also the “new method” of Descartes (1596–1650)
(referring to the use of coordinates in geometry, now called “analytic geom-
etry”) which first appeared in La Géométrie has contributed to the study
of conics by means of their algebraic equations. Descartes’ method was
further disseminated by the Dutch mathematician Frans Van Schooten
(1615–1660). In a second latin translation of Descartes’ work under the title
Geometria a Renato Des Cartes (1659–1661), an appendix written by Johan
De Witt (1625–1672) was added. In the appendix, Elementa Curvarum
Linearum, liber primus et liber secundus, conics are first defined as loci of
points and the geometric theory found in the books of Apollonius was devel-
oped, in the second part they are characterized by means of second degree
equations in two indeterminates. A similar approach is found in Tractatus
De Sectionibus Conicis (1655) by the British mathematician and contempo-
rary John Wallis (1616–1703).

A common definition for the three types of conics as loci of points is also
possible if one introduces a positive constant ε, called excentricity. A conic
is now defined as the locus of points P for which the distance d(P, F ) to a
given point F equals ε · d(P, ℓ) with ℓ a given line, not through P (for ε = 1
we recognize the definition of a parabola). This focus-directrix definition of a
conic can be used as an alternative for the three separate classical definitions
(two foci definitions) formulated above (figure 3).

Figure 3. Focus–directrix definition of a conic.

If one chooses a coordinate sytem with the x–axis through F and perpen-
dicular to ℓ and with the y–axis coinciding with ℓ, then an equation of the
p2
form (1−ε2 )x2 − 2pε x+ ε2 = 0 appears. The parameter p in{this formula satis-
x = x′ + ε(1+ε)
p
fies p = ε·d(F, ℓ). Applying a coordinate transformation
y = y′
finally reduces the equation to y = 2px − (1 − ε )x (the accents by the
2 2 2
54 Dirk Keppens and Nele Keppens

undeterminates are omitted). According to the value of the excentricity ε

one can distinguish three types of conics. For ε < 1 we call them of elliptic
type, for ε = 1 of parabolic type and for ε > 1 of hyperbolic type.
One can prove in an elementary way that a conic defined by the focus-
directrix definition is also a conic classicaly defined as locus of points. More-
over a conic of parabolic, elliptic, hyperbolic type is indeed a parabola, el-
lipse, hyperbola in the sense of the classical definition and the converse is
also true.

By the way we also mention that analogues of the different definitions

given above still make sense in non-euclidean metric geometries (elliptic and
hyperbolic planes), see e.g. [52], [54], [25], [20], [21] [18], [27], [35], [33] and
[11]. We do not mention all results here, but there are several remarkable
differences with the euclidean case. For example in hyperbolic planes at
least six types of conics (or even more according to the used definition) can
occur instead of three and the different definitions (as conic section, as point
locus in the plane and as algebraic curve) are no longer equivalent!

4. Conics are conic sections: Hamilton, Dandelin and Morton

Apollonius’ definition of a conic section in euclidean 3–space has at first
glance nothing to do with the three classical definitions or with the equiv-
alent focus–directrix definition of a conic as point loci in the plane. The
first one who noticed a connection between both concepts was the Irish
mathematician Hugh Hamilton (1729–1805), Bishop of Ossory, in De Sec-
tionibus Conicis: Tractatus Geometricus (1758) translated into English as
“A Geometrical Treatise of the Conic Sections” in 1773.
“He was the first to deduce the properties of the conic section from the
properties of the cone, by demonstrations which were general, unencumbered
by lemmas, and proceeding in a more natural and perspicuous order” [58].
But usually reference is made to two Belgian mathematicians and friends
with each other: Adolphe Quetelet (1796–1874) and Pierre Dandelin
(1794–1847). Quetelet wrote his doctoral thesis on conics De quibusdam
locis geometricis nec non de curva focali and Dandelin described a simple
geometric construction for finding the foci of a conic section. If a conic
section is defined as the intersection curve of a plane π with a cone K then
the point of contact of the plane π with a sphere touching K and π at the
same time and with center lying on the axis of the cone, is a focus of the
conic section. As a consequence it follows that conic sections in the sense of
Apollonius are the same objects as conics classicaly defined as loci of points
in the plane (two foci definition). Dandelin proved this result in [16] in 1822
using what are now called “Dandelin spheres”. This theorem is also known
under the name “Théorème de Dandelin–Quetelet” or “Théorème belge sur
la section conique”.
“Dans ces derniers temps, MM. Quetelet et Dandelin, en considerant les
coniques dans le solide, sont parvenus à de forts beaux résultats nouveaux,
dont le suivant offre, je crois, la première construction qu’on ait donnée des
foyers des coniques dans le cône . . .” [12]
 A survey on conics in historical context 55

Independently of Dandelin the British mathematician Pierce Morton

(1803–1859) described in On the focus of a conic section (1819) [36] a similar
construction and in addition he also could interpret the directrix and the
excentricity. More precisely he showed that the directrix is the secant of the
plane π with the plane containing the tangent circle of the touching sphere
cos β
and that the excentricity ε = cos α with α the half top angle of the cone
and β the angle of the plane π with the axis of the cone. It was the first
complete proof of the equivalence of Apollonius’ spatial definition of conic
section and the focus–directrix definition in the plane.
“A complete determination of the foci and directrices of the sections of the
cone by means of the focal spheres was at length proposed by Pierce Morton”
[55]

5. Two projective definitions: Von Staudt and Steiner

The development of projective geometry with Girard Desargues (1591–
1661) as one of its pioneers, shed a new light on conics. In 1639 Desargues
wrote Brouillon Project d’une atteinte aux événements des rencontres du
cône avec un plan, which can be translated as “draft for a study on the
intersections of a cone with a plane”. It contains a coherent theory of conics
using properties which remain invariant under projection. In this work De-
sargues presents a theory of conic sections in a way that was fundamentally
new in his time. For the first time “points and lines at infinity” were intro-
duced, a radical innovation. The culmination point of projective geometry is
reached in the 19th century with Jean-Victor Poncelet, Michel Chasles,
Jakob Steiner and Karl Von Staudt.

The Swiss mathematician Jakob Steiner (1796–1863) was very impor-

tant for synthetic geometry. In Systematische Entwickelung der Abhängig-
keit geometrischer Gestalten von einander (1832) and in the posthumously
published Vorlesungen über synthetische Geometrie (1867) he lays the foun-
dation of projective geometry and sheds new light on conic sections. In the
original formulation of the definition the underlying projective plane was
the real projective plane, but this can be generalized without any problem
to projective planes over any field F.
His definition sounds as follows: let α be a projective but not perspective
mapping between two pencils in a projective plane PG(2,F) (a pencil is the
set of all lines through a given point). Then the intersection points of cor-
responding lines form a (non–degenerate) Steiner conic [50], [51].
In this definition a projective mapping between pencils is used. One could
also take a projective map between point rows, resulting into dual, tangen-
tial or line conics (instead of punctual or point conics). Tangential Steiner
conics have been considered by Chasles [13] independently from Steiner.
Next we come to another definition for a conic in a pappian projective
plane, which is due to German mathematician Von Staudt (1798–1867).
His definition which first appeared in Geometrie der Lage (1847) [56] is based
on the concept of a polarity. It was originally only for the real projective
plane, but it remains valid in pappian projective planes PG(2,F) over an
arbitrary field F with characteristic not equal 2.
56 Dirk Keppens and Nele Keppens

A polarity π of a projective plane is an incidence preserving but not type

preserving (points are mapped onto lines and lines onto points) automor-
phism of order two. If the image π(P ) of a point P under a polarity π is a
line that goes through P , then P is called an absolute point of π. A polar-
ity does not necessarily have absolute points. If π has no absolute points,
we call it an elliptic polarity, in the other case we use the term hyperbolic
polarity. If char F ̸= 2 then a hyperbolic polarity of PG(2,F) has more than
one absolute point. If F is infinite, there are an infinite number of absolute
points. The action of a polarity of a pappian projective plane on the points
and the lines can be descibed algebraically as follows: if the line [u, v, w] is
the image of the point (x, y, z), then
   θ 
u x
 v  = A ·  yθ 
w zθ
and the image of the line [u, v, w] is the point (x, y, z) with
[ ] [ ]
x y z = uθ v θ wθ · A−1

in which θ is a field automorphism of F, θ2 = 1 and A is a non–singular

3 × 3–matrix over F with (AT )θ = A.
The polarities of PG(2,F) can be classified in three types:
orthogonal polarities: if θ = 1 and AT = A and char F ̸= 2
pseudo–polarities: if char F = 2 and θ = 1 and AT = A and not all of the
diagonal elements of A are zero.
hermitian or unitary polarities: if θ ̸= 1, θ2 = 1 and (AT )θ = A.
Now we are ready to give Von Staudt’s definition of a conic: a Von Staudt
conic in a pappian projective plane PG(2,F) with F a field (char F ̸= 2), is
the set of absolute points of a hyperbolic orthogonal polarity of the plane.
By this definition a Von Staudt conic is never empty.
From the matrix representation of a polarity it is clear that any Von
Staudt conic is also an irreducible algebraic conic, i.e.the set of points with
homogeneous coordinates (x, y, z) which are a solution of the quadratic equa-
tion F (x, y, z) = 0 ⇔ a11 x2 + a22 y 2 + a33 z 2 + a12 xy + a13 xz + a23 yz = 0 with
not alle coefficients aij ∈ F equal to zero. If F (x, y, z) can not be factorized
in linear factors over F we say that the algebraic conic is irreducible. Con-
versely, any irreducible algebraic conic over a field with characteristic ̸= 2 is
equivalent with a Von Staudt conic.
Indeed, the conic a11 x2 + a22 y 2 + a33 z 2 + a12 xy + a13 xz + a23 yz = 0 defines
a11 21 a12 12 a13
the matrix A =  2 a12 a22 21 a23 . Since A is symmetric and non–
1
1 1
2 a13 2 a23 a33
singular, it defines a hyperbolic orthogonal polarity and the set of absolute
points (hence the Von Staudt conic) is the given algebraic conic.
Remark that in the original definition of Von Staudt the term polarity was
not accompanied by the specification “orthogonal” because it was only for
the real projective plane and orthogonal polarities are the only type that
can occur in such a plane.
 A survey on conics in historical context 57

Any Von Staudt conic is also a Steiner conic. If one considers a variable
point of a Von Staudt conic that is connected with two fixed points of that
conic, one obtains a pair of lines which are corresponding lines in a projec-
tivity between the pencils through the fixed points.
Also the converse theorem is valid (though not trivial): any Steiner conic in
a projective plane over a field with characteristic ̸= 2 is also a Von Staudt
conic. In pappian planes (with char ̸= 2) both definitions are thus equivalent
(for a proof see e.g. [17], [15] or [19]).
Also note that Von Staudt’s conics are self-dual objects (by the correspon-
dence between absolute points and absolute lines).

In contrast to conics in the euclidean plane the distinction between the

three types of conics is lost in a projective setting. All conics are projectively
equivalent. So it is no longer possible to distinguish parabolas, ellipses and
hyperbolas. If one considers the affine plane over the reals (which is in
fact the real euclidean plane stripped of metric concepts) non–degenerate
conics defined algebraically by a second degree equation Ax2 + By 2 + Cxy +
Dx + Ey + F = 0 can be classified by looking at the discriminant δ =
C 2 − 4AB. It determines the number of intersection points of the conic with
the line at infinity. If δ = 0 the conic is of parabolic type (one intersection
point), while for δ < 0 it is of elliptic type (no intersection points) and
for δ > 0 of hyperbolic type (two intersection points). This classification
remains (partially) valid in the affine plane AG(2,F) over an arbitrary field.
If δ = 0 the conic is of parabolic type, if δ ̸= 0 it is non–parabolic (the
difference between elliptic and hyperbolic is not always possible). E.g. if F
is the complex field, then a non–parabolic conic in AG(2,C) has always two
intersection points with the line at infinity. If F is a finite field GF(q), then
there is for δ ̸= 0 a hyperbolic and an elliptic type according to δ is or is not
a square in GF(q).

6. Conics and ovals in finite projective planes: Segre’s

theorem
The Italian mathematician Beniamino Segre (1903–1977) was one of
the founders of finite geometry and of projective geometry over a finite
field in particular (today also known as Galois geometry). One of the most
famous theorems in that branch of geometry deals with conics and ovals.
A prominent property of conics is the fact that a line can intersect a
(non–degenerate) conic in at most two points. This fact is the basis of the
definition of an oval. An oval in a (finite or infinite) projective plane is a set
O of points such that any line of the plane is incident with either 0, 1 or 2
points of O and for each point of O there exists a unique line (called tangent
line) which is incident with that point and with no other points of O. In a
finite projective plane of order q, an oval can equivalently be defined as a
set of q + 1 points no three of which are collinear.
Any irreducible (non–empty) conic in a pappian projective plane PG(2,F)
with F an arbitrary field, is an oval. In particular any irreducible conic in
PG(2,q) is an oval.
If F is infinite, it is not hard to find examples of ovals which are not conics,
58 Dirk Keppens and Nele Keppens

e.g. in PG(2,R) the point set O = {(x, y, z) | y 4 = xz} is an oval but not a
conic.
In the finite case, the situation is more intrigating.
First let q be even. By a theorem of Qvist [39] all tangent lines of an
oval intersect in a common point, the nucleus. Adding this point to the
oval gives rise to an hyperoval. If one starts in particular with a conic K,
the hyperoval H which arises by adding the nucleus N is called a regular
hyperoval or hyperconic. Starting from such a hyperconic H = K ∪ {N }
and omitting a point P , distinct from N gives an oval which is also called
a pointed conic. If q = 2h with h > 2 one obtains an oval which is not a
conic, since it has q ≥ 5 points in common with K (two distinct conics can
not have more than four points in common). A classification of (hyper)ovals
of PG(2,q), q even, is not known yet and it seems to be very difficult. For
small values of q Segre has proved the following: if q = 2, 4, 8 all hyperovals
are regular and if q = 2 or 4 they are conics; for q = 8 they are either conics
or pointed conics.
Next, let us consider the case of q odd. In 1949 the Finnish mathemati-
cians Järnefelt and Kustaanheimo formulated the conjecture that any
oval in a finite projective plane PG(2,q) with q odd must be a conic [28].
In a review, Marshall Hall Jr. said that “The reviewer finds this conjecture
implausible.” But in 1955 the conjecture was proved by Beniamino Segre
[47]. Hall was again a reviewer of the paper of Segre, where he then said
“The fact that this conjecture seemed implausible to the reviewer seems to
have been at least a partial incentive to the author to undertake this work.
It would be very gratifying if further expressions of doubt were as fruitful.”
The remarkable theorem had indeed a great influence on the development
of finite geometry. Two mention only two examples in which the statement
plays a crucial role: in circle geometry any finite ovoidal Laguerre plane of
odd order must be miquelian and in the theory of generalized quadrangles
the quadrangle of Tits T2 (O) is always classical for q odd. Both results
follow from Segre’s theorem.
The original proof of Segre is partially of algebraic and partially of geomet-
ric nature. The so-called lemma of tangents (any inscribed triangle of an
oval lies in perspective with the triangle formed by the tangent lines in the
points) forms a crucial element in the proof.

7. Pascalian ovals and a theorem of Buekenhout

Blaise Pascal (1623–1662) which is best known for his contribution to
probability theory, also left an indelible mark in connection with conic sec-
tions. In 1640, at the age of sixteen, he writes Essay pour les coniques. This
one page pamphlet contains a famous theorem named after him and which
is also known as the “hexagrammum mysticum theorem”: if six arbitrary
points are chosen on a conic and joined by lines in any order to form a
hexagon, then the three pairs of opposite sides of the hexagon meet at three
points which lie on a straight line. Only two copies of this essay have been
preserved, one in the Bibliothèque Nationale de Paris and the other as a part
of a set of manuscripts by Leibniz in the Königlichen Öffentlichen Biblio-
thek of Hannover in Germany. The Essay can be viewed as an announcement
 A survey on conics in historical context 59

for a comprehensive work on conics Conicorum opus completum which was

written by Pascal after 1640 but was never published and the manuscript of
which is lost.
It is intrigating that an age-old theorem turned out to be interesting
in modern geometry. In his doctoral thesis entitled “Étude intrinsèque des
ovales” Francis Buekenhout studies among other things ovals in projective
planes. He formulates the property of Pascal for an oval as follows. Let O be
an oval in a projective plane (finite or infinite). An inscribed hexagon of O is
a cyclic ordered set of six (not necessarily distinct) points A0 , A1 , . . . , A5 of O
such that all lines Ai Ai+1 , i = 0, . . . , 5 are distinct and with the assumptions
that Ai and Ai+3 are distinct and that the line Ai Ai+1 is a tangent line to O if
Ai = Ai+1 . Such hexagon has the property of Pascal if the three intersection
points Ai Ai+1 ∩ Ai+3 Ai+4 are collinear. An oval is called pascalian if for all
inscribed hexagons the property of Pascal is valid.
By Pascal’s theorem any conic in a pappian plane is of course also a
pascalian oval. In [10] the following interesting characterization theorem is
proved: if in a projective plane there exists a pascalian oval O, then the
plane must be pappian and O is a conic. The proof makes use of involutions
and is based on a theorem of Tits about transitive permutation groups. A
simpler and shorter proof, using coordinates, was given in [2] and in [41].
Buekenhout’s theorem is a nice illustration of the fact that a rather sim-
ple geometric property (the existence of a pascalian oval in the plane) can
have very strong consequences for that plane (it must be pappian). In
Buekenhout’s approach the theorem of Pascal is expressed entirely within
the language of permutation groups and it was the starting point of further
research about so-called abstract ovals. For a recent survey on that subject,
see [5].

Figure 4. Oval with the Pascal property

8. Idiosyncrasy in non–classical projective planes and some

more definitions: Krüger conics and Ostrom conics
Since the publication of the famous book Grundlagen der Geometrie by
David Hilbert (1862–1943) the development of non–desarguesian geom-
etry has grown steadily. It is well known that the properties valid in non–
desarguesian planes can differ substantially from those in pappian planes
60 Dirk Keppens and Nele Keppens

over a field. Let us first recall some definitions. Given a projective plane
P, a point P and a line ℓ in P, we say that P is (P, ℓ)–desarguesian if for
each pair of triangles ABC and A′ B ′ C ′ which are perspective from P and
for which two pairs of corresponding side lines intersect each other on ℓ, also
the intersection point of the third pair of corresponding side lines is on ℓ.
If P is (P, ℓ)–desarguesian for any choice of P and ℓ, we call P a desarguesian
plane. Any desarguesian plane is isomorphic to a PG(2,F) with F a skewfield
or a field (in the second case the plane is called pappian and additionally
the axiom of Pappus is valid). If the above formulated property (also called
Desargues configuration) is not valid for at least one point–line pair, we call
the plane non–desarguesian. A particular class of non–desarguesian planes
are the Moufang planes. They are (P, ℓ)–desarguesian only for any incident
point–line pair.
Both the definitions of Von Staudt and Steiner for a conic in a pappian
plane can be generalized for desarguesian (but non–pappian) and for non-
desarguesian planes. A very lucid paper in which a comparison between both
definitions is discussed in detail by Strambach in [53]. A Von Staudt conic
in an arbitrary projective plane is defined as the set of absolute points of a
hyperbolic orthogonal polarity. Hyperbolic means that the set of absolute
points is not empty. Since we have no longer a matrix representation for
polarities in non–desarguesian planes, a different definition for “orthogonal”
is required: a polarity is orthogonal if the set of absolute points forms an
oval in the plane. By this definition Von Staudt conics are special ovals (with
the points being absolute points of a polarity), we could say that conics are
polar ovals.
A projective plane does not always possess polarities. In a finite projective
plane any polarity has absolute points, but in an infinite non–desarguesian
plane there may exist polarities with a finite number of absolute points or
even without absolute points. As a consequence the existence of Von Staudt
conics in non–desarguesian planes is not always garantueed. On the other
side there are several examples known of special non–desarguesian planes
with polar ovals, e.g. the Figueroa planes [14], [7], the Coulter-Matthews
planes [30] and the Albert planes [1]. It has also been proved that a transla-
tion plane P admits a Von Staudt conic if and only if P can be coordinatized
by a commutative semifield with char ̸= 2, see [22] and [29]. The first ex-
ample of a Von Staudt conic in a finite projective plane over a semifield is
given in [57] while in [23] examples in finite and infinite planes are given.
Not any oval is a Von Staudt conic and in an infinite plane there are
always examples of ovals that are not conics. Strambach has proved namely
the following theorem: if in a projective plane each oval is a Von Staudt
conic, i.e. each oval is a polar oval, then the plane must be finite [53]. In the
Hughes–planes of order q 2 an example of an oval is constructed by extend-
ing a conic in the Baer subplane PG(2,q) of the Hughes plane. This oval
defines a polarity but the set of absolute points does not coincide with the
oval. Hence it is not a Von Staudt conic, see [43]. Ovals in special classes
of projective planes were also studied in e.g. [31], [32] and [44].
 A survey on conics in historical context 61

The definition of Steiner also can be extended for non-pappian planes. A

Steiner conic is the locus of intersection points of corresponding lines of a
projective mapping α (but not a perspectivity) between two pencils (through
the points S and T ). In pappian planes α(ST ) ̸= ST because α is not a
perspecitivity, but in non-pappian planes it may happen that α(ST ) = ST
even if α is not a perspectivity. The Steiner conics corresponding to an α
for which α(ST ) = ST are called degenerate (see e.g. [40]). We now restrict
ourselves to the non–degenerate case.
The deviant behaviour of Steiner conics in non-pappian planes is immedi-
ately clear in desarguesian non-pappian planes PG(2,K) with K a skewfield.
It is known that any Steiner conic is such a plane can be seen, after a suit-
able coordinatization, as the pointset {(1, X, Y ) | XY = 1} ∪ {(0, 1, 0)} ∪
{(0, 0, 1)}. If P (1, x0 , y0 ) is a point of the conic, then all lines through P ,
except for one, has exactly one other point in common with the conic if and
only if x0 belongs to the center of K. Such a point P is called regular. For a
non–regular point Q the number of intersection points of a line through Q
with the conic depends on the number of solutions of an equation of the kind
xa − ax = b. If one considers for example the plane PG(2,H) over the skew-
field of real quaternions H = {a + bi + cj + dk | i2 = j 2 = k 2 = ijk = −1},
then Q(1, i, −i) is non–regular and the line Y = −X through Q intersects the
conic in at least three other points (1, i, −i), (1, j, −j) and (1, k, −k), even
in an infinite amount of points of the form (1, q, −q) with q = bi + cj + dk
and b2 + c2 + d2 = 1. This illustrates the strange fact that a conic in a desar-
guesian, non–pappian plane, is not always an oval. There are also examples
known of Steiner conics which are not ovals (hence not Von Staudt conics)
in finite non-desarguesian projective planes of order bigger than 4. After all
these observations it is natural to ask under which conditions the class of
Steiner conics and the class of Von Staudt conics in a plane are identical.
This question was answered in [53] and partially in [4]. Both concepts coin-
cide only if the plane is pappian over a field of characteristic distinct from
two. Moreover it is true that a projective plane is pappian if any Steiner
conic is an oval in the plane.
Segre’s theorem (see section 6) can not further be generalized for non-
desarguesian finite planes of odd order. Indeed, if the class of ovals coincides
with the class of Steiner conics then the plane is the pappian plane PG(2,q)
with q odd, see [53]. We will need this result in the next section.

For Steiner conics in desarguesian planes there always is a standard al-

gebraic equation. In order to find an algebraic description for conics in
non-desarguesian planes, Krüger introduced an alternative definition in
[34]. Given four points A, B, C and D, no three of which are collinear, he
defines a Krüger conic in an arbitrary projective plane as the locus of inter-
section points of corresponding lines for a projectivity α between the pencils
with base points A and B which is the composition of the perspectivities
[A, BC, D] and [D, AC, B] (see figure 5). In this we used the notation [S, ℓ, T ]
to indicate a perspectivity with axis ℓ from the pencil of lines through S to
the pencil of lines through T .
62 Dirk Keppens and Nele Keppens

Each Krüger conic is a Steiner conic, but not conversely. The two con-
cepts are the same only if the plane is desarguesian (not necessarily pap-
pian) [53]. Making a specific choice for the coordinatization of the plane, a
Krüger conic can be seen as the point set {(x, y) | x ◦ y = 1} ∪ {(∞), (0)}
with x ◦ y = T (x, y, 0) the multiplication associated with the coordinatising
planar ternary ring (R, T ). Another choice of the quadrangle for coordinati-
zation leads to the equivalent description as the set {(x, y) | y = x2 } ∪ {(∞)}
with x2 = x ◦ x. Krüger investigated those conics especially in Moufang
planes (coordinatized by an alternative division ring).
A Krüger conic is not always a Von Staudt conic. For example {(x, y) | y =
x2 } ∪ {(∞)} in a Moufang plane over the real octonions O is not an oval as
there are lines in the plane that intersect the conic in infinitely many points
[2].

Figure 5. Definition of a Krüger conic

In the literature still another definition of conics in non–pappian projec-

tive planes was introduced by Ostrom [37], based on a generalization of har-
monic sets. An Ostrom conic is a special kind of oval. Examples are known
in infinite non–pappian planes, but it is not known whether there exist ex-
amples in finite non–desarguesian planes. Garner in [24] conjectures that
a finite projective plane admitting an Ostrom conic must be pappian (and
than the concept coincides with that of a Steiner conic and a Von Staudt
conic). He also gives an explicit example of a Von Staudt conic which is
not an Ostrom conic in a finite plane of order 27 over an Albert twisted
semifield. In a pappian plane any conic is always an Ostrom conic. In [42]
an example is given of an oval which is not a Von Staudt conic (hence not
an Ostrom conic) in a projective plane of order 9 over a Dickson nearfield.

9. Ovals and conics in topological planes

When one connects geometry with topology a new field of research arises:
“topological geometry”. The study of topological projective planes as part
of this, was initiated by Kolmogorov, but interest increased in particular
since 1950 by publications of Salzmann [45], [46] and Skornyakov [48].
A topological projective plane is a projective plane in which both the point
 A survey on conics in historical context 63

and the line set are endowed with a non-trivial topology such that “con-
necting distinct points” and “intersecting distinct lines” are continuous
operations. A well–known important classification theorem states that the
only compact, connected topological Moufang planes are the desarguesian
topological planes PG(2,K) with K ∈ {R, C, H} and the non–desarguesian
topological plane PG(2,O) over the alternative octonion division ring. These
are the so–called classical topological planes and they have topological di-
mension 2,4,8 and 16 respectively. The lines in the topological projective
plane over the complex numbers are closed subsets homeomorphic to 2–
spheres. Since any conic in the complex plane is an oval and since the
stereographic projection from a point on a conic upon a line in the plane is
an homeomorphism we see that conics in PG(2,C) are also homeomorphic
to 2–spheres. Hence they are closed ovals. It is remarkable that also the
converse is true. Thomas Buchanan, an American-born mathematician
who lived in Germany and a student of Strambach, proved that any closed
oval in the topological projective plane over the complex numbers is a conic.
The proof of this rather surprising theorem which can be found in [8] makes
use of topological tools as well as of results from complex functions theory
(e.g. the theorem of Casorati–Weierstrass on holomorphic functions).
Now let O be an oval in a topological projective plane and denote by LO
the set of lines intersecting O in at least one point. The map
{
P Q if P ̸= Q
O × O → LO : (P, Q) 7→
tQ if P = Q
induces a bijection ψO : O ∗ O → LO between the (symmetrized) cartesian
product O ∗ O and LO which is continuous in all points (P, Q) of O ∗ O met
−1
P ̸= Q. We put φO = ψO : LO → O ∗ O : ℓ 7→ ℓ ∩ O
An oval in a topological projective plane is by definition a topological oval if
the map φO is continuous. A topological oval in a compact, connected pro-
jective plane is compact and hence also closed. Conversely it was proved that
each closed oval in a compact connected projective plane of finite dimension
is a topological oval. Do there exist topological ovals in any compact plane?
The answer is negative. There do not exist topological ovals in compact
projective planes with topological dimension larger than four [9].
In line with Buchanan’s theorem that closed ovals (in particular topolog-
ical ovals) in PG(2,C) are conics, the question arises whether the complex
plane is the only compact plane (with a non–discrete topology) in which
topological ovals and conics are the same objects. To answer this question
one needs some more topology and algebra (such as completeness, discrete
valuation, non–archimedean local field, Hensel’s lemma). With these auxil-
iary tools the following theorem was proved in [53]: in a compact projective
plane P the class of topological ovals is the same as the class of Steiner conics
if and only if P is the complex plane. In this theorem it is assumed tacitly
that the topological plane is non–discrete (so the plane must be infinite). If
one considers a finite projective plane endowed with the discrete topology,
then one obtains a compact, totally disconnected projective plane.
The result mentioned in section 8 that ovals and Steiner conics are identical
objects only in the finite pappian plane PG(2,q) with q odd can be reformu-
lated in topological terms: in a finite compact projective plane of odd order
64 Dirk Keppens and Nele Keppens

the class of topological ovals coincides with the class of Steiner conics if and
only if the plane is a pappian plane over a Galois field of odd order. Com-
bining this with the preceeding theorem yields another remarkable theorem
which combines Segre’s theorem and the theorem of Buchanan: in a compact
projective plane P (possibly with a discrete topology in case the plane is
finite) the class of topological ovals is the same as the class of Steiner conics
if and only if P is either the complex plane PG(2,C) or a finite projective
plane PG(2,q) over the Galois field GF(q) with q odd.

10. Summary
Throughout history, there has been a wide variety of definitions regard-
ing the term ”conic section”. In this paper we have provided an overview
focusing on some milestones. The theorem of Dandelin unifies the spatial
definition of Apollonius’ conic sections and the focus–directrix definition
of conics in the euclidean plane, the theorem of Segre and the theorem of
Buekenhout puts ovals and conics in a common framework while the theorem
of Buchanan adds a topological aspect. Other results under consideration
deal with the projective definitions by Steiner and Von Staudt for pappian
planes as well as their generalization to non–desarguesian projective planes.
Extensions of the different definitions to non-euclidean metric geometries
are only touched sideways.

References
[1] Abatangelo, V., Enea, M. Korchmaros, G. and Laratoa, B., Ovals and unitals in
commutative twisted field planes, Discr. Math. 208–209 (1999), 3–8.
[2] Artzy, R., Pascal’s theorem on an oval, Amer. Math. Monthly 75 (1968), 143–146.
[3] Artzy, R., The conic y = x2 in Moufang planes, Aeq. Math. 6 (1971), 30–35.
[4] Berz, E., Kegelschnitte in desarguesschen Ebenen, Math. Z. 78 (1962), 55–85.
[5] Bamberg, J., Harris, T., and Penttila, T., On abstract ovals with Pascalian secant
lines, J. Group Theory, 21 (2018), 1051–1064.
[6] Besant, W.H., Conics treated geometrically, Cambridge University Press (1869).
[7] Brown, J., Ovals in infinite Figueroa planes, J. Geom. 43 (1992), 41–52.
[8] Buchanan, T., Ovale und Kegelschnitte in der komplexen projektiven Ebene, Math.-
Phys. Semesterber. 26 (1979), 244–260.
[9] Buchanan, T., Hähl, H. and Löwen, R. Topologische Ovale, Geom. Dedicata 9
(1980), 401–424.
[10] Buekenhout, F., Plans projectifs à ovoı̈des pascaliens, Arch. Math. 17 (1966),
89–93.
[11] Chao, P. and Rosenberg, J., Different definitions of conic sections in hyperbolic
geometry, Involve 11 (2018), 753–768.
[12] Chasles, M., Aperçu historique sur l’origine et le développement des méthodes en
géométrie, J. Gabay, Bruxelles (1837).
[13] Chasles, M., Traité des sections coniques, Gauthier–Villars, Paris (1865).
[14] Cherowitzo, W. Ovals in Figueroa planes, J. Geom. 37 (1990), 84–86.
[15] Coxeter, H.S.M, Projective Geometry, Springer–Verlag, New York (1987).
[16] Dandelin, G., Mémoire sur quelques propriétés remarquables de la focale parabolique,
in Nouveaux mémoires de l’Académie Royale des Sciences et Belles–Lettres de Brux-
elles, T. II. (1822), 171–202.
[17] Enriques, F., Lezioni di geometria proiettiva, Bologna (1898).
[18] Fankhänel, A., On conics in Minkowski planes, Extracta Math. 27 (2012), 13–29.
[19] Faulkner, E., Projective Geometry, Dover Publications (2006).
[20] Fladt, K., Die allgemeine Kegelschnittgleichung in der ebenen hyperbolischen Ge-
ometrie, J. Reine Angew. Math. 197 (1957), 121–139.
 A survey on conics in historical context 65

[21] Fladt, K., Die allgemeine Kegelschnittgleichung in der ebenen hyperbolischen Ge-
ometrie. II, J. Reine Angew. Math. 199 (1958), 203–207.
[22] Ganley, M.J., Polarities in translation planes, Geom. Dedicata 1 (1972), 103–106.
[23] Garner, C., Von Staudt conics in semifield planes, Aeq. Math. 11 (1974), 183–188.
[24] Garner, C., Conics in finite projective planes, J. Geom. 12 (1979), 132–138.
[25] Goodner, D., Conic Sections in the Elliptic Plane, Math. Mag. 34(2) (1960), 81–94.
[26] Heath, T.L., Apollonius of Perga. Treatise on Conic Sections, Cambridge University
Press (1896).
[27] Horváth, A.G. and Martini, H., Conics in normed planes, Extracta Math. 26 (1)
(2011), 29–43.
[28] Järnefelt, G. and Kustaanheimo, P., An observation on finite geometries, 11th
Skand. Math. Kongress Trondheim (1949), 166–182.
[29] Knarr, N. and Stroppel, M., Polarities of shift planes, Adv. Geom. 9 (2009), 577–
603.
[30] Knarr, N. and Stroppel, M., Polarities and unitals in the Coulter–Matthews planes,
Des. Codes Cryptog. 55 (2010), 9–18.
[31] Korchmaros, G., Ovali nei piani di Hall di ordine dispari, Atti Accad. Naz. Lincei
Rend. 56 (1974), 315–317.
[32] Korchmaros, G., Ovali nei piani di Moulton d’ordine dispari, Atti Convegni Lincei
17 (1976), 395–398.
[33] Kurusa, A., Conics in Minkowski geometries, Aeq. Math. 92 (2018), 949–961.
[34] Krüger, W., Kegelschnitte in Moufangebenen, Math. Z. 120 (1971), 41–60.
[35] Molnár, E., Kegelschnitte auf der metrischen Ebene, Acta Math. Acad. Sci. Hungar.
31/34 (1978), 317–343.
[36] Morton, P., On the focus of a conic section, Trans. Cambridge Phil. Soc. (1830).
[37] Ostrom, T.G., Conicoids: conic–like figures in non-pappian planes, in Geometry,
von Staudt’s Point of View, ed. Plaumann, P. en Strambach, K. (1981).
[38] Quételet, A., Mémoire sur une nouvelle théorie des sections coniques considérées
dans le Solide, in Nouveaux mémoires de l’Académie Royale des Sciences et Belles–
Lettres de Bruxelles, T. II. (1822), 123–154.
[39] Qvist, B., Some remarks concerning curves of the second degree in a finite plane,
Ann. Acad. Sci. Fenn. Ser. A 134 (1952), 27 pp.
[40] Riesinger, R., Entartete Steinerkegelschnitte in nichtpapposchen Desarguesebenen,
Monatsh. Math. 89 (1980), 243–251.
[41] Rigby, J.F., Pascal ovals in projective planes, Canad. J. Math 21 (1969), 1462–1475.
[42] Rodriquez, G., Un esempio di ovale che non è una quasi-conica, Boll. Unione Mat.
Ital. Serie 3 14 (1959), 500–503.
[43] Room, T.G., Polarities and ovals in the Hughes plane, J. Austr. Math. Soc. 13
(1972), 196–204.
[44] Salzmann, H., Polaritäten von Moulton–Ebenen, Abh. Math. Sem. Univ. Hamburg
29 (1966), 212–216.
[45] Salzmann, H., Topologische projektive Ebenen, Math. Z. 67 (1957), 436–466.
[46] Salzmann, H., Topological planes, Adv. Math. 2 (1967), 1–60.
[47] Segre, B., Ovals in a finite projective plane, Canad. J. Math. 7 (1955), 414–416.
[48] Skornyakov, L.A., Topological projective planes (in russian), Trudy Moskov Mat.
Obsc. 3 (1954), 347–373.
[49] Steel, R., A Treatise of Conic Sections, Grierson, Dublin (1723).
[50] Steiner, J., Systematische Entwickelung der Abhängigkeit geometrischer Gestalten
von einander, Fincke, Berlin (1832).
[51] Steiner, J., Vorlesungen über synthetische Geometrie: Die Theorie der Kegelschnitte
in elementarer Darstellung, Volume 1, Teubner, Leibzig (1867).
[52] Story, W.E., On non–euclidean properties of conics, Amer. J. Math. 5 (1) (1882),
358–381.
[53] Strambach, K., Projektive Ebenen und Kegelschnitte, Arch. Math. 64 (1995), 170–
184.
[54] Sykes G. S. and Peirce, B., Spherical conics, Proc. Amer. Acad. Arts and Sciences
13 (1877-1878), 375–395.
 66 Dirk Keppens and Nele Keppens

[55] Taylor, C., An introduction to the ancient and modern geometry of conics, Bell and
Company, Cambridge (1881).
[56] Von Staudt, K. Geometrie der Lage, Bauer und Raspe, Nürnberg (1847).
[57] Wagner, A., On perspectivities in finite projective planes, Math. Z. 71 (1959), 113–
123.
[58] Wills, J., Lives of Illustrious and Distinguished Irishmen: From the Earliest Times
to the Present Period, Arranged in Chronological Order, and Embodying a History
of Ireland in the Lives of Irishmen, volume 6, MacGregor, Polson (1845).
[59] Zeuthen, H., Die Lehre von den Kegelschnitten im Altertum, A.F. Höst & Sohn,
Kopenhagen (1886).

FACULTY OF ENGINEERING TECHNOLOGY

KU LEUVEN UNIVERSITY
B-9000 GHENT, BELGIUM
E-mail address: [email protected]

KONINKLIJK ATHENEUM DENDERMONDE

B-9200 DENDERMONDE, BELGIUM
E-mail address: [email protected]

View publication stats