The diameter and the traversale: in Girard Desargues' studio (Q2420597)

This paper deals with some aspects of Desargues' \textit{Brouillon project}. In particular, the authors focus on the middle part of the \textit{Brouillon}. The article under review proposes a mathematical-historiographical thesis: the fundamental concept of involution allowed Desargues to construct the notion of polarity with respect to a conic and Desargues fully guessed the global character of a polarity, namely that a polarity is an involutory reciprocity between straight lines and points of the plane where the conic lies. Furthermore, the entire Apollonius' theory of diameters for the conics enter within Desargues' new theory where the notion of ``traversale'' plays a fundamental role. This thesis is interesting, but, in my opinion, the attentive and careful mathematical analyses proposed by the authors are even more interesting than the historiographic thesis in itself. For, they offer new and clear interpretations of several complex passages by Desargues. Content: In the introduction (of which the authors offer a French and an English version), they claim their thesis that the notion of ``traversale'' associated to a conic is a core-notion for Desargues, that it has been interpreted as equivalent to the notion of polarity, but that a careful examination is still missing. Their intention is to provide such examination (p. 388 English version). After having offered a brief picture of Desargues' \textit{Brouillon}, they enter a very refined and important mathematical question which they will develop in the course of the paper, a question that, if not clear, might induce strong misunderstandings of Desargues' thought. In their words: ``To define the traversale, Desargues used his theorem on involutions induced by a pencil of conics through four base points. The involution thus defined is (almost) never equal to the one defined by polarity of a single conic in the pencil, and Desargues switched from the first to the second in a very discreet way, but, as the rest of the text shows, he had a very clear view of those matters'' (p. 389). In the first section (pp. 390--395), the authors explain Desargues' vocabulary, which was so strange and unusual that it was one of the main causes of the scarce interest towards Desargues' ideas until Poncelet's rediscovery. This vocabulary is due to the fact that Desargues was aware that he was introducing objects of a new kind. For example, the projective line, is not anymore the line of the Euclidean geometry. Thence, he invented new names, which, however, also in the ``golden age'' of projective geometry (namely the 19th century), when Desargues' thought was appreciated, were not used. The authors expound then the way in which Desargues introduced the point at infinity and explain in a clear manner the importance of the complete quadrangle in Desargues' work as well as the origin of the concept of involution. They also remark that, though the concept of harmonic ratio plays an important role within the \textit{Brouillon}, it is never explicitly named (p. 392). The remarkable way in which Desargues introduced the notion of cone and conic (also including the degenerate cases) is also clarified (pp. 394--395). In the second section (pp. 396--400), the authors enter the problem of the traversals and argue that, at the beginning, Desargues defined such a concept in respect to two incident straight lines and not in respect to a general conic, as Taton is inclined to think (p. 396). This granted, the authors face Desargues' definition of ``traversale'' (p. 397) of an external point \(F\) in respect to a configuration of two straight lines \(NB\), \(NC\) (\(F\) does not belong to the two lines), as the line on which a particular involution is defined. The straight lines passing through \(F\) are called \textit{ordonnées} and Desargues also distinguishes the case in which the \textit{ordonnées} cut the configuration \(NB-NC\) and the case in which they do not (p. 397). The authors also explain how the concepts of \textit{traversale} and \textit{ordonnées}, through the introduction of the point at infinity, when applied to the conics allow Desargues to extend Apollonius' notion of a conic's diameter (p. 397). In this context, considering the traversal of the centre of a conic, the notion of straight line at infinity is also introduced (pp. 397--398). Afterwards, the authors explain how Desargues used a complete quadrangle inscribed in a conic to define an involution connected to the concept of traversal and, in this circumstance, they distinguish the two involutions which might be confused if the reader is not careful enough (p. 399). However, they will clarify the whole situation in the third section (pp. 400--417) which is divided into eight subsections and which is the very conceptual core of this paper. The authors begin their explanations analysing Desargues' definition of \textit{traversale} as a straight line \(t\) such that, given a conic and a point \(F\) external to the conic, each line passing through \(F\) and cutting the conic in two points, which can indicated by \(L\) and \(M\), cuts \(t\) in a point \(H\) so that \(L,M,F,H\) are in involution (p. 400). There are two problems: to prove that such a traversal exists and that it is unique. The authors offer an amazingly and detailed explanation of Desargues' proof concerning the existence of the straight line \(t\). They have to work with the involutions and with their projective character (Section 3.1., pp. 400--404). Therefore, Section 3.1. clarifies Desargues' proof. Section 3.2. (p. 404) is brief, but it is important because the authors expound the nature of the two above mentioned involutions. To summarize: the first involution is constructed using four points which belong to a complete quadrilateral inscribed in a conic. Thence, it -- properly speaking -- does not connote a sole conic, but a pencil of conics because, as well known, a conic is univocally identified by five and not by four points. This involution has two double points which do not belong to the conic and the points where the line through \(F\) cuts the conics are correspondent. However, if another involution is considered where the points of the conic are double points, while \(F\) and the point of the traversal are correspondent, then this involution characterises univocally the conic. The authors call the first involution ``l'involution de Desargues''. It is useful to construct the traversal. Whereas the second involution is named ``l'involution de polarité'', which is necessary to deduce all the properties of the polar transformations (p. 404). In Section 3.3. (pp. 404--406), the authors prove a fundamental lemma of incidence. The situation is like this: given a conic and a point \(F\) external to the conic, draw the traversal \(t\) of \(F\) in respect to the conic. Afterwards, from a point \(N\) belonging to \(t\) and external to the conic draw a straight line cutting the conic in the points \(D\) and \(C\). Join \(D\) with \(F\). Such a straight line will cut the conic in \(E\). Join \(C\) with \(F\). The straight line \(CF\) will cut the straight line \(NE\) in a point \(B\). The lemma states that \(B\) lies on the conic. In practice, this lemma claims that the construction does not depend on the specific quadrangle used to develop it. Desargues' argument in incomplete and the authors fulfil this gap with a clear reasoning following a method which might be surely available to Desargues. The next Desargues' step (Section 3.4., p. 407) consists in the determination of \(t\) when the quadrangle is given. In particular: if \(F\) is external to the conic and two straight lines cutting the conic in four points are drawn from \(F\) (so to determine an inscribed quadrangle), then the traversal is the line joining the two diagonal points of the quadrangle. Desargues does not prove such statement, but the authors show that it is easily deducible from the previous lemma. In this context, Desargues also reaches a proposition which is the dual of that already mentioned: if a traversal with respect to a conic is given, then it is also given its \textit{but} (to use Desargues' term, the pole to use our term). What is happening is that Desargues is constructing the whole theory of polarity with respect to a conic, fully guessing the dual character of any polar transformation. In Section 3.5. (pp. 408--409), the authors summarise Desargues' results in two dual theorems of which the former states the existence, uniqueness and way of construct the traversal of a given point in respect to a given conic; whereas the latter states the existence, uniqueness and way of construct the \textit{but} of a given traversal in respect to a given conic. Section 3.6. (pp. 409--412) is dedicated to the reciprocal position of a traversal and its \textit{but} in respect to a conic: it is easy to prove that if the \textit{but} is external, the traversal cuts the conic in two points and vice versa; if the \textit{but} is internal, the traversal is external to the conic and vice versa. The difficult question arises when the \textit{but} is on the conic or the traversal is tangent to the conic. After the introduction of a further theorem (p. 410) concerning an involution constructed on the traversal and through which it is possible to determine the tangent of a conic through a point \(A\), the authors show that if the \textit{but} \(F\) lies and the conic, its traversal is the tangent to the conic at \(F\) and vice versa. Now Desargues claims that in an involution of four points, when one of them is at infinity, and \(A\) and \(B\) are the two double points of the involution, then the point corresponding to the point at infinity is the middle point of the segment \(AB\). Through this theorem and the concept of point at infinity, Desargues is able to include Apollonius' theory of the conics' diameters within his theory of involution. In particular, if the \textit{but} \(F\) is at infinity, its traversal is a diameter of the conic. On the other hand, if the point \(F\) coincides with the centre \(O\) of the conic, its traversal is the line at infinity of the conic's plane and vice versa. In Section 3.7. (pp. 412--415), the authors come back to the distinction between Desargues involution and involution of polarity. At the beginning of his treatise Desargues had taught how to construct the involution of polarity for a straight line cutting a conic, now, at p. 22, he teaches how to construct such an involution for lines external to the conic, being, thus, excluded only the case in which a line is tangent to the conic. The authors explain clearly all the constructive steps and they claim that this general construction shows that, in Desargues' mind, a conic with the associated involutions induces a new geometrical (projective) structure in the whole plane. This means that Desargues had guessed the global action on the whole plane of a planar transformation. Later on, Desargues also includes the case of a straight line tangent to the conic, so completing the whole picture. In this way Desargues' doctrine of polarity appears as a real theory. In Section 3.8. (pp. 415--417), the authors explain how Desargues made his ideas wider, connecting his theory of traversals (or, as we say today, his theory of polarity) with considerations connected to spatial perspective. For, since polarity concerns geometrical properties of incidence, the polar configuration in respect to a conic produced by a plane cutting a cone can be projected on another plane cutting the cone if the vertex of the cone is considered the centre of projection. The authors explain Desargues' reasoning. Finally, the way in which the theory of the conjugate diameters of a conic as well as those of the asymptotes of a hyperbola enter Desargues' theory of polarity is also clarified. In the conclusion (pp. 417--420), the authors point out that Poncelet in the \textit{Traité des propriétés projectives des figures} (1822) offers a foundation to projective geometry, which from the end of the 18th-beginning of the 19th century had become a branch of mathematics whose importance was increasing. Poncelet's aims was to offer a new kind of synthetic geometry (in general then called ``modern geometry'') which was as broad (or even broader) and reliable as analytic geometry. Within this picture the principle of continuity and the point at infinity played, as well known, a fundamental role. Through quotations drawn from Poncelet's work, the authors stress how Poncelet ascribed to Desargues the merit for this two fundamental concepts. Poncelet also highlighted that the main method of the new geometry could not be the theory of proportion but the theory of transversals, which is nothing but a corollary of the method of projections. Poncelet fully recognised Desargues' merits. The authors conclude that their research has shown that Desargues is also the inventor of the theory of polarity. In the appendix (pp. 420--426), the theory of polarity within the picture of contemporary projective geometry is explained. Commentary: This paper is very good and recommendable for any reader who wishes to enter Desargues' way of thinking as well as some of the most important aspects of the doctrine of polarities. For, the explanations given by the authors are so clear and detailed that they offer both a clear historical picture of the way in which Desargues worked and develop a mathematical work, which, in the successful attempt to fulfil some gaps in Desargues' proves, offers a panorama interesting also for professional mathematicians. Therefore, both under the historical and mathematical point of view this paper is extremely good and represents an excellent prototype of a serious research in the history of mathematics. Some absolutely minor critical observations: I am perfectly aware that the bibliography of a paper must include only the necessary references; however, in my opinion the bibliographical references are too restricted because there are some papers which face subjects similar to that addressed by the authors and which should have been, at least, mentioned. Only to give an example: the paper by \textit{F. Ghione} and \textit{L. Catastini} entitled ``Nella mente di Desargues tra involuzioni e geometria dinamica'' [Boll. Unione Mat. Ital., Sez. A, Mat. Soc. Cult. (8) 8, No. 1, 123--147 (2005; Zbl 1202.01051)], which is also freely available on the internet and which faces problems close to those addressed by the authors might have been mentioned. The profound paper ``De L'irruption ou invention de l'infini actuel, de l'espace actuellement infini et de l'involution comme invariant numérique, dans l'œuvre de Desargues'', by \textit{J.-P. Le Goff} [in: Prospettiva e geometria dello spazio. Sarzana (La Spezia): Agorà edizioni. 177--270 (2005)] is a further work having deep intersections with that of the authors and which would have been appropriately mentioned. Finally, it is false that Brianchon in 1817 (p. 419) introduced the terms ``pole'' and ``polar''. The term ``pole'' was introduced by \textit{M. Servois} [``Solution du premier des deux problèmes proposés à la page 259 de ce volume, et du problème proposé à la page 126 du même volume'', Annales de Mathématiques Pures et Appliquées 1, 337--341 (1810--1811)]; whereas the term ``polar'' was introduced by \textit{J.-D. Gergonne} [``Géométrie analytiques. Théorie analytique des pôles des lignes et des surfaces du second ordre'', ibid. 3, 293--302 (1813)]. These minor critics notwithstanding, this research is an outstanding contribution in history of mathematics, reach from a conceptual point of view, and accurate from a historical perspective.