Projective plane
In mathematics, a projective plane is an incidence structure consisting of a set of points and a set of lines such that any two distinct points lie on a unique line, any two distinct lines intersect in a unique point, and there exist at least four points with no three collinear.[1] This axiomatic definition distinguishes it from higher-dimensional projective spaces by enforcing universal intersection of lines, eliminating parallels and creating a "closed" geometry.[1] Projective planes arise in both synthetic geometry and algebraic settings, serving as a foundational model for studying transformations, conics, and higher-degree curves.[2] Algebraically, a projective plane over a field F is constructed as \mathbb{P}^2(F), the set of equivalence classes of nonzero vectors in F^3 under scalar multiplication, where points are represented as [x : y : z] with not all coordinates zero.[3] Lines in this space are defined by homogeneous linear equations ax + by + cz = 0, ensuring that every pair of points determines a line and every pair of lines intersects.[3] Geometrically, over the real numbers, the real projective plane \mathbb{RP}^2 can be viewed as the set of all lines through the origin in \mathbb{R}^3, or equivalently as a sphere with antipodal points identified, which embeds the Euclidean plane as an affine patch while adding a line at infinity where parallel lines meet.[2][4] This construction resolves affine exceptions, such as parallel non-intersections, and underpins projective transformations that preserve incidence relations.[2] In the finite case, a finite projective plane of order n (where n is a prime power) has exactly n^2 + n + 1 points and the same number of lines, with each line containing n + 1 points and each point incident to n + 1 lines.[5][6] The smallest such plane, the Fano plane of order 2, features 7 points and 7 lines, realizable over the field \mathbb{F}_2.[1] Topologically, \mathbb{RP}^2 is a compact, non-orientable surface that contains a Möbius strip as a subspace, distinguishing it from orientable surfaces like the torus or sphere.[7][4] Projective planes play a central role in unifying classical theorems, such as Desargues' theorem on perspective triangles and Bézout's theorem on curve intersections, where two curves of degrees m and n meet at mn points counting multiplicity.[3] They also appear in design theory as symmetric balanced incomplete block designs and in computer vision for modeling image projections.[5]Definition and Basic Concepts
Axiomatic Definition
A projective plane is an abstract incidence geometry consisting of a set of points, a set of lines, and a binary incidence relation indicating which points lie on which lines.[8] Points and lines are primitive elements without further structure, and the incidence relation forms the basis for all geometric properties.[9] This structure satisfies three fundamental axioms:- Any two distinct points are incident with exactly one common line.
- Any two distinct lines are incident with exactly one common point.
- There exist four distinct points such that no three are incident with the same line.[8]
Incidence and Properties
In a projective plane, the incidence axioms imply that there are no parallel lines, as any two distinct lines intersect at exactly one point.[12] This property arises directly from the second incidence axiom, which guarantees a unique point of intersection for every pair of lines, distinguishing projective planes from affine geometries where parallels may exist.[12] Further consequences of the axioms ensure that each line contains at least three points and each point lies on at least three lines.[13] To see this, consider a line \ell; if it contained fewer than three points, say only two, then selecting a third point not on \ell (possible by the existence of non-collinear points) would contradict the uniqueness of lines through pairs of points. Dually, the same holds for points and lines.[13] For finite projective planes of order n (where n \geq 2), the axioms yield precise counts: there are exactly n^2 + n + 1 points and the same number of lines.[14] Each line contains n + 1 points, and each point is incident to n + 1 lines.[14] The incidence structure is symmetric, meaning the roles of points and lines can be interchanged while preserving the axioms, a principle known as duality.[12] This symmetry manifests in the plane forming a symmetric balanced incomplete block design (BIBD) with parameters v = b = n^2 + n + 1, k = r = n + 1, and \lambda = 1, where every pair of distinct points is contained in exactly one line, and the design parameters satisfy the BIBD relations.[14] Basic configurations illustrate these properties. A complete quadrangle consists of four distinct points, no three collinear, together with the six lines joining them pairwise.[13] Dually, a complete quadrilateral comprises four distinct lines, no three concurrent, along with the six points of intersection.[13] The existence of such configurations follows from the third incidence axiom, which requires four points with no three collinear, ensuring the plane's richness beyond degenerate cases.[12]Examples
Real and Classical Projective Planes
The real projective plane, denoted \mathbb{RP}^2, is constructed as the set of all lines through the origin in \mathbb{R}^3, where each point in \mathbb{RP}^2 corresponds to an equivalence class of vectors under nonzero scalar multiplication.[15] Specifically, two vectors \mathbf{v} = (x, y, z) and \mathbf{w} = (x', y', z') represent the same point if there exists a nonzero scalar k \in \mathbb{R} such that \mathbf{w} = k \mathbf{v}. This formulation captures the projective structure by identifying directions rather than positions, providing a compactification of the plane that includes points at infinity.[16] Points in \mathbb{RP}^2 are represented using homogeneous coordinates [x : y : z], where (x, y, z) \neq (0, 0, 0) and scaling by nonzero reals yields the same point. Lines in \mathbb{RP}^2 are defined by linear equations of the form a x + b y + c z = 0, corresponding to planes through the origin in \mathbb{R}^3. A point [x : y : z] lies on the line [a : b : c] if and only if a x + b y + c z = 0. This coordinate system facilitates computations in projective geometry, enabling the representation of both finite and infinite elements uniformly.[17][16] An intuitive realization of \mathbb{RP}^2 arises by extending the Euclidean plane \mathbb{R}^2 with a line at infinity, which compactifies the space by adding ideal points where parallel lines meet. In homogeneous coordinates, the affine plane corresponds to points with z \neq 0, normalized as (x/z, y/z), while the line at infinity is given by z = 0. This construction unifies Euclidean geometry with projective properties, such as the intersection of all parallels.[15] Projective transformations on \mathbb{RP}^2 are induced by invertible linear maps on \mathbb{R}^3, preserving incidence relations between points and lines. A fundamental result is that any two nondegenerate conics in \mathbb{RP}^2 are projectively equivalent, meaning there exists a projective transformation mapping one to the other; this underscores the uniformity of quadratic forms under projective changes.[18] Visualizations of \mathbb{RP}^2 in \mathbb{R}^3 include the disk model, where the interior represents the affine plane and antipodal points on the boundary disk are identified to form the line at infinity. Another prominent immersion is Boy's surface, a smooth immersion with a single triple point that realizes \mathbb{RP}^2 without boundaries, discovered in 1901.[19][20]Finite Projective Planes
A finite projective plane of order n is an incidence structure consisting of n^2 + n + 1 points and the same number of lines, where each line contains exactly n + 1 points, each point lies on exactly n + 1 lines, any two distinct points determine a unique line, and any two distinct lines intersect in a unique point.[9] Such planes are combinatorial objects that satisfy the axioms of a projective plane but are discrete and finite.[13] The smallest finite projective plane is the Fano plane, which has order n = 2 and is denoted \mathrm{PG}(2,2). It features 7 points and 7 lines, with each line comprising 3 points. This plane is often visualized as a central point surrounded by a triangle, where the points are the three vertices, three midpoints of the sides, and the center; the lines consist of the three sides of the triangle, the three medians connecting vertices to midpoints, and a curved line (or circle) passing through the three midpoints.[13] The Fano plane is unique up to isomorphism as the projective plane of order 2.[13] More generally, the Desarguesian finite projective plane \mathrm{[PG](/page/PG)}(2,q) is constructed over a finite field \mathbb{F}_q, where q is a prime power. The points of \mathrm{[PG](/page/PG)}(2,q) are the one-dimensional subspaces of the three-dimensional vector space \mathbb{F}_q^3, which number q^2 + q + 1; equivalently, these are the equivalence classes of nonzero vectors in \mathbb{F}_q^3 under scalar multiplication by nonzero elements of \mathbb{F}_q. The lines are the two-dimensional subspaces, each containing q + 1 points, and the structure satisfies the projective plane axioms with order n = q.[21] For instance, when q = 2, \mathbb{F}_2^3 has 7 nonzero vectors, yielding the 7 points of the Fano plane.[9] Finite projective planes of order n are known to exist whenever n is a prime power, constructed as the Desarguesian plane \mathrm{PG}(2,n) from the corresponding finite field. However, it remains an open question whether projective planes exist for orders n that are not prime powers, with no such examples verified despite extensive searches.[22]Non-Desarguesian Planes
A non-Desarguesian projective plane is a projective plane that fails to satisfy Desargues' theorem and is therefore not isomorphic to the projective plane arising from a three-dimensional vector space over a division ring.[23] Such planes provide counterexamples to the assumption that all projective planes are coordinatizable by division rings, highlighting the independence of Desargues' theorem from the basic incidence axioms. The historical discovery of non-Desarguesian planes began in the early 1900s, with Forest Ray Moulton providing the first explicit example of a non-Desarguesian affine plane in 1902, whose projective completion yields a non-Desarguesian projective plane.[24] Oswald Veblen and J.H.M. Wedderburn established in 1907 that projective planes satisfying Desargues' theorem are precisely those coordinatizable by division rings, implicitly confirming the existence of non-Desarguesian planes through the possibility of other coordinatizing structures.[23] Key explicit examples proliferated in the 1950s, including finite cases that further demonstrated the diversity of projective geometries beyond vector space models. The Moulton plane exemplifies a synthetic construction of a non-Desarguesian affine plane, obtained by modifying the Euclidean plane such that lines with negative slopes are bent at the y-axis: for x ≤ 0, slope m (m < 0), and for x ≥ 0, slope 2m, while lines with non-negative slopes and vertical lines remain Euclidean straight lines.[25] This bending ensures that Desargues' theorem fails, as perspective triangles with bases on opposite sides of the y-axis have corresponding sides whose joins do not concur due to the change in line slopes.[24] The projective completion of the Moulton plane inherits this failure, serving as an early infinite non-Desarguesian projective plane. Translation planes derived from semifields provide another class of non-Desarguesian examples, where the coordinatizing structure is a semifield rather than a division ring. The Hughes plane, constructed in 1957, is a finite translation plane of order 9 arising from a non-associative semifield of order 9, violating Desargues' theorem because the semifield multiplication lacks the required distributivity over addition. Other translation planes from semifields, such as those of orders p^2 for prime p, similarly fail Desargues' theorem unless the semifield is a field, offering systematic synthetic constructions distinct from vector space models.[26] A near-pencil represents a degenerate synthetic example of a non-Desarguesian structure, consisting of n+1 points where one line contains n points and the remaining lines each contain exactly two points (the off-line point and one on the main line), satisfying the basic incidence axioms of any two points on a unique line and any two lines intersecting at a unique point but failing non-degeneracy conditions.[27] This configuration violates Desargues' theorem, as the near-pencil's lopsided incidence prevents the collinearity required for perspective triangles, underscoring boundary cases in projective plane theory.[27]Constructions
Vector Space Construction
One standard algebraic construction of a projective plane arises from a three-dimensional vector space over a division ring K, yielding the Desarguesian projective plane \mathrm{PG}(2, K). In this setup, the points of \mathrm{PG}(2, K) are the one-dimensional subspaces of K^3, while the lines are the two-dimensional subspaces of K^3. A point lies on a line if the corresponding one-dimensional subspace is contained in the two-dimensional subspace.[28][29] Points in \mathrm{PG}(2, K) can be represented using homogeneous coordinates: a point is an equivalence class [x : y : z], where x, y, z \in K are not all zero, and [x : y : z] \sim [\lambda x : \lambda y : \lambda z] for any \lambda \in K \setminus \{0\}. This equivalence captures the one-dimensional subspaces, as scaling by nonzero elements of K preserves the span. Lines can similarly be represented by the set of points satisfying a homogeneous linear equation, such as a x + b y + c z = 0 with coefficients in K.[28] Classical examples include the real projective plane \mathbb{RP}^2 = \mathrm{PG}(2, \mathbb{R}), constructed from \mathbb{R}^3, which compactifies the real affine plane by adding points at infinity. Another is the complex projective plane \mathbb{CP}^2 = \mathrm{PG}(2, \mathbb{C}), from \mathbb{C}^3, a fundamental object in complex geometry with homogeneous coordinates over \mathbb{C}.[2] For finite cases, when K = \mathbb{F}_q is the finite field with q = p^k elements (p prime, k \geq 1), \mathrm{PG}(2, q) is a finite projective plane of order q, with q^2 + q + 1 points and the same number of lines. These planes are central to the theory of finite geometries and block designs.[29] This construction satisfies the axioms of a projective plane. First, any two distinct points (one-dimensional subspaces) span a unique two-dimensional subspace, hence lie on a unique line. Second, any two distinct lines (two-dimensional subspaces) intersect in a one-dimensional subspace, since their dimensions sum to four, exceeding the ambient dimension three, and the intersection is exactly one-dimensional by linear algebra over K. Third, there exist four points with no three collinear, such as the points corresponding to the spans of (1,0,0), (0,1,0), (0,0,1), and (1,1,1). These planes are Desarguesian, meaning Desargues' theorem holds, due to the linear structure over the division ring.[29][28]From Affine Planes
An affine plane can be viewed as a projective plane with one line removed, specifically the line at infinity, which results in the introduction of parallel lines as those that would have intersected on this removed line.[2] This perspective highlights how the Euclidean plane, for instance, emerges as the real projective plane minus its line at infinity, where parallel lines in the affine structure correspond to lines meeting at infinity in the projective one.[2] To construct a projective plane from a given affine plane of order n, one adds points at infinity to resolve parallelism: for each parallel class of lines in the affine plane (there are n+1 such classes, each partitioning the n^2 points into n lines of n points each), introduce a single new point at infinity incident to all lines in that class.[30] These n+1 points at infinity form a new line, called the line at infinity, ensuring that every pair of lines now intersects exactly once—either at an affine point or at infinity—while preserving the incidence structure and yielding a projective plane of order n with n^2 + n + 1 points and lines, each containing n+1 points.[30] This completion process is reversible: removing the line at infinity from any projective plane produces an affine plane.[31] In coordinate terms, points in the affine plane are represented as pairs (x, y), while the projective closure uses homogeneous coordinates [x : y : z], where affine points correspond to those with z = 1, so [x : y : 1] identifies with (x, y).[2] Points at infinity have z = 0, forming the line [x : y : 0], with lines in the projective plane defined by equations a x + b y + c z = 0.[2] For non-Desarguesian affine planes, this coordinatization generalizes beyond fields to structures like quasifields, but when the resulting projective plane satisfies Desargues' theorem, the coordinates arise from a division ring.[32] The Veblen-Wedderburn theorem establishes that every projective plane coordinatized by a division ring (Desarguesian planes) can be obtained from an affine plane over the same ring by adding the line at infinity, and conversely, any such affine plane derives from removing a line from the projective one.[32] This coordinatization links the geometric structure directly to algebraic properties of the division ring, with finite examples arising from finite division rings of order n^2.[32]Synthetic and Other Constructions
Synthetic constructions of projective planes emphasize axiomatic approaches, building upon incidence relations and geometric configurations without recourse to coordinate systems. These methods often involve completing partial planes—sets of points and lines with specified incidences—into full projective planes that satisfy the standard axioms, such as the existence of unique lines through distinct points and unique points on intersecting lines. One foundational technique is the free completion, pioneered by von Staudt, which generates a projective plane by freely adding points and lines to a given partial plane while enforcing the projective axioms minimally.[33][34] In this framework, the free projective plane generated by an initial configuration is the universal object embedding that configuration, meaning any projective plane containing the configuration as a substructure factors through it uniquely up to isomorphism. Such free planes exist for any finite or countable partial plane satisfying basic incidence conditions, like having no two lines sharing more than one point. Von Staudt's method ensures that the resulting structure is a projective plane where the original configuration is densely embedded, and every closed subconfiguration (one satisfying the full axioms) arises naturally from the completion process.[35][36] A notable example of synthetic alteration arises in the Moulton plane, constructed by modifying the Euclidean plane's axioms to produce a non-Arguesian geometry. Specifically, lines with negative slope are "bent" at the x-axis, altering their incidence relations while preserving most Euclidean ordering axioms except those implying Desargues' theorem. The projective completion of this affine structure yields a non-Desarguesian projective plane, demonstrating how subtle axiomatic changes can evade classical theorems. This construction highlights the flexibility of synthetic methods in exploring plane geometries beyond Desarguesian ones.[37] Hall planes provide another synthetic perspective, derived from translation planes through alternative coordinatizations that redefine multiplication on the underlying quasifield while maintaining translation properties. Marshall Hall Jr. introduced these by constructing quasifields where addition is vector-like but multiplication twists the field operations, leading to planes that embed Desarguesian subplanes but fail full Desarguesian structure globally. Synthetically, this manifests as adjusting incidence in parallel classes to produce non-isomorphic planes of the same order, offering a way to generate diverse projective structures from a base translation framework. The existence of free projective planes underscores the universality of synthetic constructions: starting from any set of points and lines forming a partial plane (with at most one line per pair of points and no three lines concurrent unless specified), one can always adjoin elements to satisfy the projective plane axioms without contradictions, yielding a countably infinite plane if the generator is finite. These planes are characterized by their openness—no finite subconfiguration forces closure under the axioms—and finitely generated free planes coincide with open ones. Such constructions embed arbitrary configurations faithfully, serving as universal models for synthetic projective geometry.[35][36] In classical settings, polarities offer a duality-based construction, where a correlation of order two maps points to lines and vice versa, preserving incidence. A conic in a Desarguesian projective plane induces a polarity via its tangent lines: the polar of a point is the line joining the contact points of tangents from that point to the conic. Synthetically, von Staudt constructed such polarities using complete quadrangles and quadrilaterals, defining conics as the loci of points whose polars pass through a fixed point or intersect a fixed line in a harmonic set. This approach builds conic-based planes by embedding polarity structures into the axiomatic framework, revealing conics as self-dual figures central to projective transformations.[38]Key Properties and Theorems
Desargues' Theorem
Desargues' theorem states that in a projective plane, two triangles are perspective from a point if and only if they are perspective from a line. Two triangles \triangle ABC and \triangle A'B'C' are perspective from a point O if the lines AA', BB', and CC' are concurrent at O. They are perspective from a line \ell if the points AB \cap A'B', BC \cap B'C', and CA \cap C'A' are collinear on \ell. This if-and-only-if condition forms the core of the theorem, establishing a symmetry between point and line perspectivities.[39] The theorem originates from the work of French mathematician Girard Desargues (1591–1661), who proposed it in a 1639 manuscript on perspective drawing for conic sections, though it was not published until 1648 by Abraham Bosse in Brouillon project d'une atteinte aux événements des rencontres d'un cône avec un plan. Desargues' ideas laid foundational groundwork for projective geometry, emphasizing properties invariant under projection, but the theorem was largely overlooked until its rediscovery in the 19th century by mathematicians like Jean-Victor Poncelet and August Ferdinand Möbius, who integrated it into axiomatic frameworks.[40][41] The Desargues configuration realizes the theorem geometrically, comprising 10 points and 10 lines such that each line passes through exactly 3 points and each point lies on exactly 3 lines. The points include the vertices A, B, C, A', B', C' of the two triangles, the perspectivity point O, and the three intersection points on the axis of perspectivity \ell; the lines are the sides of the triangles, the connecting lines AA', BB', CC', and three additional lines completing the incidences. This (10_3) configuration is self-dual and embeds the theorem's conditions, serving as a minimal incidence structure verifying the perspectivity equivalence.[42] The dual form of Desargues' theorem, obtained via the duality principle of projective planes (interchanging points and lines while preserving incidence), states that two triangles perspective from a line are perspective from a point. This self-duality underscores the theorem's symmetry and follows directly from the original statement under duality. A synthetic proof of Desargues' theorem can be given by embedding the projective plane in projective 3-space, where the two triangles lie in distinct planes intersecting along a line. Assume \triangle ABC and \triangle A'B'C' are perspective from point O not on either plane. The lines AA', BB', and CC' concur at O. To show perspectivity from a line, consider the plane determined by O and side AB; it intersects the second plane along a line through A' and B', and similarly for the other sides. The intersections of these planes with the line of intersection of the two triangle planes yield collinear points on the axis \ell, using the incidence axioms of 3-space. This argument relies on the Pasch axiom and plane separation properties but avoids coordinates. For the converse, a similar embedding reverses the roles. In a coordinatized projective plane over a division ring K, Desargues' theorem can be proved using Ceva's theorem in the affine patch. Place the perspectivity point O at the origin in vector space terms, with \triangle ABC having vertices as basis vectors or affine points. Assume coordinates such that A = (1,0,0), B = (0,1,0), C = (0,0,1) in homogeneous coordinates, and \triangle A'B'C' scaled by a homothety centered at O, so A' = \lambda A, etc. The connecting lines concur by construction. To verify axial perspectivity, dehomogenize to the affine plane by setting the third coordinate to 1, yielding points where the side intersections satisfy Ceva's condition (\frac{BA' \cap A'B}{B A'} ) \cdot (\frac{CB' \cap B'C}{C B'} ) \cdot (\frac{AC' \cap C'A}{A C'} ) = 1 via signed ratios in the division ring K, confirming collinearity. The converse follows by applying Menelaus' theorem to transversals on the sides. This proof highlights how the theorem enforces multiplicative structure in the coordinate ring.[44] Desargues' theorem holds if and only if the projective plane is Desarguesian, meaning it is isomorphic to the projective plane \mathrm{PG}(2, K) constructed from a 3-dimensional vector space over a division ring K. This characterization, established through coordinatization theorems, shows that the theorem's validity implies the existence of a division ring coordinatizing the plane's points and lines via homogeneous coordinates, with multiplication and addition satisfying the perspectivity conditions. Non-Desarguesian planes exist where the theorem fails, but all classical planes over fields are Desarguesian. In the vector space construction over K, the theorem holds due to the linearity of intersections and collinearities preserved under scalar multiplication.[45]Pappus' Theorem
Pappus' theorem, named after the Greek mathematician Pappus of Alexandria (c. 290–c. 350 AD), originally appeared in his Collection as a result in Euclidean geometry involving points on a conic section.[46] The projective version, stripped of metric assumptions and applicable to any projective plane, was developed by Karl Georg Christian von Staudt in his 1847 work Geometrie der Lage, where it serves as a key axiom for synthetic projective geometry.[47] In a projective plane, Pappus' theorem states that if points A, B, C lie on one line \ell and points X, Y, Z lie on another line m, then the intersection points P = (AY \cap BX), Q = (BZ \cap CY), and R = (AZ \cap CX) are collinear.[48] This configuration can be viewed as the intersections of opposite sides of a hexagon A, Y, C, X, B, Z with alternate vertices on \ell and m. In a projective plane over a field F, the theorem holds because the underlying vector space structure ensures the collinearity via linear algebra over F.[9] A coordinate proof in the projective plane \mathrm{PG}(2, F) can be given by assigning homogeneous coordinates to the points on \ell and m, and verifying that the determinant of the matrix formed by the coordinates of P, Q, R is zero, which follows from identities in the field F.[48][9] A projective plane satisfies Pappus' theorem if and only if it is isomorphic to \mathrm{PG}(2, F) for some commutative field F, making it a Pappian plane coordinatizable over a field (equivalently, a commutative division ring).[9] This equivalence arises because Pappus' theorem implies Desargues' theorem in the plane, allowing coordinatization over a division ring via von Staudt's construction, and further ensures commutativity of the ring's multiplication through the theorem's configuration.[9][49] Von Staudt's algebra provides a synthetic method to construct field operations purely from projective incidence: addition and multiplication are defined using harmonic conjugates and complete quadrilaterals, with Pappus' theorem guaranteeing the commutativity needed for a field structure rather than a general skew field.[49] This geometric algebra bridges incidence geometry and algebra, showing how Pappus elevates Desarguesian planes (over division rings) to those over fields.[47]Duality Principle
The principle of duality in projective geometry asserts that in a projective plane, interchanging the roles of points and lines in the axioms yields a dual structure that satisfies the same axioms, resulting in another projective plane (isomorphic to the original in many cases, such as Desarguesian planes). This symmetry arises because the incidence relation—where a point lies on a line—is symmetric, allowing points to correspond bijectively to lines while preserving the fundamental properties of any two distinct points determining a unique line and any two distinct lines intersecting at a unique point. Formulated in its full generality by Joseph Diez Gergonne in 1826, the principle highlights the foundational balance between points and lines in projective spaces.[50] A key consequence is that every theorem in projective plane geometry has a dual statement obtained by swapping "point" and "line," with the dual theorem holding true if and only if the original does. For instance, Desargues' theorem, which states that if two triangles are perspective from a point, then their corresponding sides intersect on a line, dualizes to the statement that if two triangles are perspective from a line, then their corresponding vertices lie on a point—yielding the same theorem due to its self-dual nature. This duality extends to all axioms and propositions, ensuring that valid geometric statements transform into equally valid duals without altering the plane's structure.[51] Corollaries of this principle include the symmetry of the incidence matrix representing the plane, where rows and columns (corresponding to points and lines) are interchangeable, reflecting the bijection. Self-dual configurations, such as the complete quadrangle—formed by four points with no three collinear and their six joining lines—exemplify this, as it is isomorphic to its dual, the complete quadrilateral of four lines with no three concurrent and their six intersection points. These configurations underscore the principle's role in generating reciprocal geometric figures.[52] In coordinatized projective planes, duality manifests through the adjoint or polarity, a correlation that maps points to lines and vice versa using the dual vector space. For a point represented by homogeneous coordinates [x : y : z], its dual line is given by the equation xX + yY + zZ = 0, where (X, Y, Z) are variables for points on the line, effectively interchanging the roles via the transpose of the coordinate matrix. This construction preserves incidence and isomorphism in Desarguesian planes derived from vector spaces.[53][2] However, in non-Desarguesian projective planes, such as Hall planes, duality may not fully preserve the structure, as the dual plane is not necessarily isomorphic to the original, lacking the self-duality required for complete reciprocity. These planes, constructed without satisfying Desargues' theorem, often fail to admit a correlation that induces an isomorphism under point-line interchange, limiting the principle's applicability to the axiomatic theory rather than individual models.[54]Transformations
Collineations
A collineation of a projective plane is a bijective mapping from the set of points to itself and from the set of lines to itself that preserves the incidence relation: a point lies on a line if and only if its image lies on the image of the line.[55][56] This ensures that collinear points map to collinear points and that the geometric structure remains intact under the transformation.[57] The collection of all collineations of a projective plane forms a group under composition, called the collineation group, which acts faithfully on the points and lines of the plane.[58] For Desarguesian projective planes over a field K, arising from the projectivization of a 3-dimensional vector space over K, the full collineation group is isomorphic to the projective semilinear group \mathrm{P\Gamma L}(3, K), consisting of invertible semilinear transformations modulo scalar multiples.[59] This group captures all structure-preserving automorphisms in such planes.[60] Collineations are classified based on fixed elements, with perspectivities forming a key subclass: these fix a line (the axis) pointwise and fix the pencil of lines through a point (the center).[61] Elations are special perspectivities where the center lies on the axis, fixing both the point and the line pointwise while permuting other elements accordingly.[61] More general collineations can be composed from these basic types, generating the full group action.[58] The fundamental theorem of projective geometry asserts that, for Desarguesian planes, every collineation is induced by a semilinear map on the underlying vector space, providing a complete algebraic description of the group.[62] In many cases, the isomorphism type of the collineation group uniquely determines the plane up to isomorphism, distinguishing Desarguesian planes from non-Desarguesian ones via their automorphism structures.[58][59] In classical projective planes over the real numbers, using homogeneous coordinates, Euclidean rotations and translations of the affine part extend naturally to projective collineations by acting linearly on the vector space.[2] These transformations preserve the projective structure, mapping infinite lines appropriately and unifying affine and projective symmetries.[2]Homographies and Correlations
In the vector space construction of a projective plane, a homography is a collineation induced by a nonsingular linear transformation on the underlying vector space.[63] Such transformations are represented by invertible 3×3 matrices defined up to scalar multiplication, acting on homogeneous coordinates of points.[64] The action of a homography H on a point with homogeneous coordinates \begin{pmatrix} x \\ y \\ z \end{pmatrix} yields new coordinates \begin{pmatrix} x' \\ y' \\ z' \end{pmatrix} = H \begin{pmatrix} x \\ y \\ z \end{pmatrix}, where the result is also taken up to scalar multiple.[64] Homographies preserve the cross-ratio of four collinear points, a fundamental projective invariant that distinguishes them within the broader class of collineations.[65] A correlation is an incidence-reversing bijection in the projective plane, mapping points to lines and lines to points while preserving the incidence structure in the dual sense.[66] In the coordinate formulation, correlations correspond to nonsymmetric bilinear forms, whereas special cases known as polarities arise from symmetric or Hermitian forms, often with respect to a conic section.[67] In classical real or complex projective planes, correlations such as polarities with respect to the absolute conic define notions of orthogonality: two lines are orthogonal if each passes through the pole of the other with respect to the conic, and absolute elements are the self-conjugate points or lines lying on the conic itself.[68] These absolute elements encode the metric structure underlying Euclidean geometry within the projective framework.[69] The composition of two correlations is always a collineation, as the reversing actions cancel to preserve incidence directly; this property allows correlations to generate the full collineation group when combined appropriately.[70]Substructures
Subplanes
A subplane of a projective plane is a subset of its points and lines that forms a projective plane with the induced incidence relation.[71] In finite projective planes of order n, the order k of any proper subplane satisfies k \leq \sqrt{n}.[71] In Desarguesian planes, which arise from finite fields, subplanes correspond to subfields, so their orders divide n.[72] Fano subplanes are subplanes of order 2, isomorphic to the Fano plane, the unique projective plane of that order.[73] They appear in all Desarguesian planes of even order, due to the subfield \mathbb{F}_2. In non-Desarguesian planes, Neumann's conjecture posits that every such plane contains a Fano subplane, a result verified for many classes including planes admitting orthogonal polarities of even order.[74] [75] Subplanes play a key role in classifying finite projective planes, as their presence or absence distinguishes types; for instance, only certain non-Desarguesian planes of order 25, such as Hughes planes, contain subplanes of order 3, while all known planes of that order have subplanes of order 5.[72] Maximal subplanes, those not properly contained in larger subplanes, further aid in identifying plane structures and testing conjectures on existence.[76] Baer subplanes provide a prominent example, occurring in finite projective planes of order n = m^2 as subplanes of order m = \sqrt{n}, achieving the maximum possible order for proper subplanes.[71] In Desarguesian planes like \mathrm{PG}(2, q^2), Baer subplanes correspond to the subfield \mathbb{F}_q.[72] They are fixed pointwise by certain collineations of order 2, and every quadrangle in the plane lies in exactly one Baer subplane.[77][78]Degenerate Planes
Degenerate projective planes are incidence structures in combinatorial geometry that satisfy the incidence axioms requiring any two distinct points to determine a unique line and any two distinct lines to intersect in a unique point, but fail the non-degeneracy axiom mandating the existence of four points with no three collinear. These structures highlight the boundary conditions of the projective plane axioms and are excluded from the standard definition to ensure a sufficiently rich geometry.[79] A simple degenerate case occurs when all points are collinear on a single line, with no other lines present; here, the intersection axiom cannot be meaningfully tested for multiple lines, and the non-degeneracy fails trivially as any three points are collinear. More substantially, the near-pencil provides a key example: it consists of a distinguished point O not on a base line \ell containing n-1 other points p_1, \dots, p_{n-1}, where the lines are \ell and the n-1 lines each joining O to a distinct p_i. The point-line incidences ensure unique lines through pairs of points and unique intersections between lines—\ell meets each O p_i at p_i, while distinct lines O p_i and O p_j meet at O—yet no four points avoid having three collinear, since any such set must include at least three points from \ell.[5][13] Near-pencils and similar degeneracies relate closely to broader classes of partial planes or linear spaces, which enforce only the unique line through two points but relax the unique intersection condition, allowing structures that are projective-like yet incomplete. In extremal terms, near-pencils realize the minimal number of lines b = v (where v is the number of points) among finite linear spaces, as established by the de Bruijn–Erdős theorem, contrasting with the b = v equality also achieved by proper projective planes.[80] In axiomatic developments, degenerate planes have historically served to test and refine the boundaries of projective geometry, identifying trivial or insufficient configurations that must be ruled out to capture essential properties like the existence of non-collinear points. For instance, early combinatorial studies used them to derive incidence bounds and embeddability results for non-degenerate spaces.[80][79]Finite and Higher-Dimensional Projective Planes
Finite Planes and Orders
Finite projective planes exist for every order n that is a power of a prime, constructed as Desarguesian planes coordinatized by the finite field \mathbb{F}_n.[29] These are the only known orders for which such planes have been established, with the conjecture that no finite projective planes exist for other orders remaining unresolved.[13] For example, no projective plane of order 6 exists, as demonstrated by the Bruck-Ryser theorem, which shows that certain arithmetic conditions must hold for existence.[81] The Bruck-Ryser-Chowla theorem provides necessary conditions for the existence of a projective plane of order n: if n \equiv 1 or $2 \pmod{4}, then n must be expressible as the sum of two integer squares.[82] This theorem, originally proved for projective planes by Bruck and Ryser in 1949 and extended to symmetric designs by Chowla and Ryser in 1950, rules out existence for many non-prime-power orders, such as 6 (since $6 \not= x^2 + y^2 for integers x, y) and 14.[81] Despite these obstructions, the sufficiency of the conditions remains open for most cases. The number of non-isomorphic finite projective planes of order n varies with n. For prime orders, there is exactly one up to isomorphism, the Desarguesian plane over \mathbb{F}_n.[13] For some composite prime powers, multiple non-isomorphic planes exist; for instance, there are four non-isomorphic planes of order 9, including the Desarguesian plane and three non-Desarguesian ones.[83] The following table enumerates the known counts for small orders:| Order n | Number of non-isomorphic planes | Notes |
|---|---|---|
| 2 | 1 | Fano plane, Desarguesian |
| 3 | 1 | Desarguesian |
| 4 | 1 | Desarguesian |
| 7 | 1 | Desarguesian |
| 8 | 1 | Desarguesian, unique by exhaustive search |
| 9 | 4 | One Desarguesian, three non-Desarguesian |