Japanese theorem: a little known theorem with many proofs (Q2566888)

If a cyclic \(n\)-gon is partitioned into \((n-2)\) triangles by diagonals through a vertex, then the sum of the inradii of these triangles does not depend on the particular choice of the vertex. According to page 193 of \textit{R. A. Johnson}'s [Advanced Euclidean geometry (Dover Publications, New York) (1960; Zbl 0090.37301)], this is a Japanese temple geometry theorem that dates back to 1800, and it can be proved by induction once it is established for cyclic quadrilaterals. The quadrilateral case appears as Example 3.5(1) (pages 125--126) in \textit{H. Fukagawa} and \textit{D. Pedoe} [Japanese temple geometry problems (The Charles Babbage Research Centre, Winnipeg, Canada) (1989)], where it is also added that the four incenters involved form a rectangle. Configurations related to this rectangle are studied by \textit{J.-L. Ayme} [Forum Geom. 3, 225--229 (2003; Zbl 1073.51007)]. An elegant proof of the aforementioned beautiful theorem is given on pages 24--26 of \textit{R. Honsberger}'s [Mathematical Gems III (Dolciani Math. Expositions 9, MAA, Washington, D. C.) (1985; Zbl 0568.00001)], who proves that the sum of the inradii is in fact the same for every triangulation by diagonals even if the diagonals used in the triangulation do not emanate from the same vertex. The proof is based on a theorem of Carnot stating that the sum of the signed distances from the circumcenter of a triangle to the sides is equal to the sum of the circumradius and inradius of the triangle. The paper under review presents five proofs of the quadrilateral case that are based on \textit{T. Hayashi}'s paper in [Mathesis III (6), 257--260 (1906; JFM 37.0523.02)]. The authors give a brief history of each proof, and show two different ways for proceeding by induction. They freely use facts proved in what they refer to as Part I, a reference that is not cited, at least not clearly, in the references.