Poncelet's porism: a long story of renewed discoveries. I (Q907761)

\textbf{1. Introduction} The text in issue is the first part of a long history of Poncelet's closure theorem (after Cayley (see the quotation from Mention at p.~74), named \textit{porism} from the Greek πoρiσμoν [sic] which means ``a proposition affirming the possibility of finding the conditions under which a certain problem becomes indeterminate or capable of infinite solutions'' (see p.~2)), its discovery, its various proofs and its developments. This first part also could be entitled ``From Chapple to Halphen'', since the content spans from Chapple to Halphen, i.e., to the end of the 19th century. The paper is a historical one in character, full of quotations, but is also much technical, not an easy-to-read paper. It does not limit itself to stating some beautiful theorem, but it reproduces the principal proofs, analyzing any detail in depth. A historian interested in Poncelet's closure theorem (therein after, simply PCT), must not miss to read this pamphlet with its wealth of information. What is missing, on the contrary, is the human side of this long adventure on the discovery and tuning of PCT. Perhaps, some biographical sketch would be not out of place for two reasons: 1) to anchor some abstract mathematical stuff to real (to speak in philosophical terms), making the results more lively; 2) the reader also could be interested in the specific figure of a mathematician. For example, little is said about Jean-Victor Poncelet, apart from the long Footnote 25 (p.~20), where the author tells that Poncelet worked on \textit{his} PCT for the first time during his captivity (until 1814) in Saratov, after the defeat of Napoleon in the Russian campaign. I take the opportunity to notice that searching for ``porism'' or ``Poncelet's closure theorem'' in the online Encyclopedia of Mathematics [\url{https://www.encyclopediaofmath.org}] leads to no record. Unfortunately, also the Wikipedia section devoted to the PCT is very meager [\url{https://www.wikidata.org/wiki/Q1785610}]. No one, engaged in this field of research can find something nontrivial there. It is also for this purpose that the paper under review is to praise for the tremendous source of information gathered in a unique place. The reader who is not satisfied with the history traced down by the author, can rely on his abundant references. There are only two negative features in it: any quotation is given in the original language. Unfortunately, while the quotations in French, Italian (of course) and English are correct, the German ones are full of typos. The second negative or curious feature is the presence of Figure 30 (p.~100), illustrating one of Darboux's theorems, which is not cited in the text, albeit no one can deny its value in order to grasp visually Darboux's fourth theorem for \(n=5\) (p.~99). This is also the right place to stress the quality of the drawings (36 in total), which are much elegant, easy to understand and without any redundant information. This was said as a sort of introduction about the style of the paper under review. A very informative paper for historians but nontrivial and unsuitable for a layman. \textbf{2. Content} The paper is 122 pages long and is articulated as follows: after a general introduction which covers the goal of the first and second part [Zbl 1345.01007], there is a specific introduction to Part I (the part in issue). Ten sections follow, ended by five pages of references. The sections are: {\parindent=10mm \begin{itemize}\item[1.] The prehistory: from Chapple to Steiner [William Chapple (1718--1781), John Landen (1719--1790), Leonhard Euler (1707--1783), Simone Antoine Jean Lhuillier (1750--1840)]; \item[2.] The theorems and methods of Poncelet; \item[3.] Jacobi and the use of the elliptic functions; \item[4.] Trudi: the forgotten work; \item[5.] Cayley's explicit conditions; \item[6.] An algebraic approach through invariants [George Salmon (1819--1904) and Fran\-ce\-sco Brioschi (1824--1897)]; \item[7.] Other contributions from 1850 to 1875 [Friedrich Julius Richelot (1808--1875), J. Mention (1821--?), Théodore Florentin Moutard (1827--1901), Jakob Rosanes (1842--1922) and Moritz Pasch (1843--1930), Jakob Steiner (1796--1863), Michel Floréal Chasles (1793--1880)]; \item[8.] \((2,2)\)-correspondences and closure problems [Cayley again and Adolf Hurwitz (1859--1919)]; \item[9.] The theorems of Darboux; \item[10.] Poncelet polygons in Halphen's treatise. \end{itemize}} I have inserted in square brackets the authors treated in the respective section. It is curious that, albeit Euler is often cited in the paper, only a very short paragraph is devoted to him (pp.~13--14). The author (Footnote 16, p.~17) remarks that the Swiss mathematician Simone Antoine Jean Lhuilier is sometimes written as ``L'Huilier''. By the way, Enriques reffered to this mathematician always as L'Huilier. This to clarify another thing. Regarding Mention, the author writes (Footnote 74, p.~71): ``very little is known about J. Mention (1821--?), probably a Russian mathematician''. Throughout the article, the author often also refers to Nicola Trudi (1811--1884), since he was neglected during his life and is less known also today, despite his important contributions to the subject. \textbf{3. A short review} Obviously, the core of the paper is Section 2, devoted to the work of Poncelet. The author states Poncelet's closure theorem as follows: \textbf{Theorem PCT}: ``Let \(C\) and \(D\) be two smooth conics in the projective plane, if there exists a polygon of \(n\) sides which is inscribed in \(C\) and circumscribed about \(D\), then for every point \(P\in C\) there is one such polygon having \(P\) as one of its vertices.'' (p.~20) I.e., if there exists an \(n\)-gon inscribed in \(C\) and circumscribed about \(D\), then there is an infinite (\(\aleph_1\) for precision) number of such \(n\)-gons; an \(n\)-gon for any point of \(C\), but since \(C\) has \(\aleph_1\) points (we are in \(\mathbb{R}^2\); see the famous proof by Cantor), then, there are \(\aleph_1\) \(n\)-gons, and so infinitely many. The generalization to \(n\) conics, known as the Poncelet general theorem (PGT for short), is given as follows: \textbf{Theorem PGT}: ``Let \(C,D_1,D_2,\dots,D_{n-1}\) be conics from a pencil \(\mathfrak{F}\). Consider a \(n\)-gon \(P,P_1,\dots,P_{n-1}\) inscribed in \(C\) and having the side \(P P_1\) tangent to \(D_1\), the side \(P_1 P_2\) tangent to \(D_2\) and so on until \(P_{n-2} P_{n-1}\) tangent to \(D_{n-1}\). Then, if \(P\) move along \(C\) in such a way the sides \(P P_1\), \(P_1 P_2\) etc. remain tangent, respectively, to \(D_1\), \(D_2\) etc., the \(n\)th side \(P_{n-1} P\) envelops a conic belonging to \(\mathfrak{F}\).'' (ibid.) In other words, satisfied some premise, the polygonal line \(P,\dots,P_{n-1}\) closes, since the theorem states the existence of a line between \(P\) and \(P_{n-1}\). We could term this line a closed path. The author offers both proofs of the PGT by Poncelet: the \textit{analytical} (pp.~26--30) as well as the \textit{synthetical} one (pp.~30--33). Since it is impossible to take in account all the results offered in this paper, I limit myself to some curiosity. For example, Steiner proved the PCT for circles (pp.~78--80). He supposed that two circles \(c\) and \(C\) are given, the first one inside the latter. Then, one can build a sequence of circles \(C_1,\dots,C_{n-1}\) which is tangent both to \(c\) and \(C\) and for any circle \(C_i\) of the sequence, \(C_i\) is tangent both to \(C_{i+1}\) and \(C_{i-1}\). Then, Steiner claims: either the sequence never closes for any \(n\), or it closes. As is easy to see, this chain of circles is the pendant of the chain of lines in PGT. Of great interest is Section 8 about \((2,2)\)-correspondences, which are a particular case of the more general \((m,n)\)-correspondences. As a matter of fact, let \(f(x,y)=0\) be a polynomial, such that \(\deg x=m\) and \(\deg y=n\). Then there is always a correspondence \(\phi\) which maps any \(x_i\) (for \(i=1,\dots,m\)) to all of the \(y_1,\dots,y_n\) and which maps any \(y_j\) (for \(j=1,\dots,n\)) to all of the \(x_1,\dots,x_m\). Obviously, a \((2,2)\)-correspondence is an \((m,n)\)-correspondence for \(m=n=2\). In this case, given a polynomial \(f(x,y)=0\), \(\phi\) maps \(x_i\) to \(y_1\) and \(y_2\), and \(y_j\) to \(x_1\) and \(x_2\). In other words, any point \(x\) is in correspondence with two points \(y\), and any point \(y\) is in correspondence with two points \(x\). This section, introducing the \((2,2)\)-correspondences, is a turning point in the text, because the author often refers to these correspondences in connection with PCT in the subsequent parts. On page 84, the author notices that any symmetric \((2,2)\)-correspondence is associated with a biquadratic equation of the type \[ f(x,y) = ax^2y^2 + bxy(x+y) + c(x^2 + y^2) + dxy + e(x+y) + f = 0. \] Why are these correspondences such relevant for the PCT? As Cayley wrote [JFM 02.0505.01]: ``The porism of the in-and-circumscribed polygon has its foundation in the theory of the symmetrical \((2,2)\) correspondence of point on a conic; viz. a \((2,2)\) correspondence is such that to any given position of either point there correspond two positions of the other point; in a symmetrical \((2,2)\) correspondence [\(\dots\)] \(x,y\) are connected by an equation of the form \[ (x,1)^2 (y,1)^2 = 0, \leqno{(\ast)} \] which is symmetrical with respect to the parameters \(x,y\).'' (ibid.) (\(\ast\)) is the way Cayley denotes a biquadratic equation. An \textit{in-and-circumscribed} polygon \(P\) is a polygon which is inscribed in a conic \(C\) and circumscribed about a (not necessarily different) conic \(D\). On page 102, we encounter the definition of a \textit{system of conics} in a footnote. A system of conics \(S\) is a parameterized finite summatory, i.e., it is given by the equation \(\sum a_{ij}(\lambda)x_ix_j=0\), the coefficients of which depend on a parameter \(\lambda\). Well, ``Chasles defined \textit{first characteristic} of the system [of conics] the number \(\mu\) of conics in \(S\) which pass through a point, and \textit{second characteristic} of the system the number \(\nu\) of conics in \(S\) which are tangent to a line'' (Footnote 95, p.~102). Last but not least, the paper ends with the observation that ``Halphen brought to light a connection between Poncelet's closure theorem and the development in continued fractions of \(\sqrt{X(x)}\)'' (p.~118).