Around the Gauss theorem on the values of Euler's digamma function at rational points (Q6562940)

The purpose of this paper is to introduce a formula for the digamma function \(\Psi\equiv \textup{D}\log \Gamma(x)\) for rational \(x=\frac pq, 0<p<q.\) However, the formula is false, as can be seen by putting \(\frac pq=\frac 13\). Then, \(\Psi(\frac 13)=-\gamma-\frac{\pi}{2\sqrt{3}}-\frac 32\log 3\). Note that the formula on top of page 312 contains the expression \(-\log 2-\log 3+\sqrt{3}\log\sin\frac{\pi}{3}=-\log 2-\log 3+\sqrt{3}\log(\frac{\sqrt{3}}{2})\). In [\textit{A. P. Prudnikov} et al., Integrals and series. Vol. 1: Elementary functions. Vol. 2: Special functions. Transl. from the Russian by N. M. Queen. New York etc.: Gordon \&| Breach Science Publishers (1986; Zbl 0733.00004), Appendix II.2], the same formula is given with a plus sign for the last term instead. In [\textit{A. Erdélyi} et al., Higher transcendental functions. Vol. II. Bateman Manuscript Project. New York-Toronto-London: McGraw-Hill Book Co (1953; Zbl 0052.29502); \textit{I. S. Gradshteyn} and \textit{I. M. Ryzhik}, Table of integrals, series, and products. Transl. from the Russian by Scripta Technica, Inc. 5th ed. Boston, MA: Academic Press, Inc. (1994; Zbl 0918.65002)], yet other formulas are given. Finally, the quoted formula cannot be found neither in [\textit{B. C. Berndt}, Ramanujan's notebooks. Part I. New York etc.: Springer-Verlag (1985; Zbl 0555.10001), page 188], nor in [\textit{G. E. Andrews} et al., Special functions. Cambridge: Cambridge University Press (1999; Zbl 0920.33001), page 30]. The correct formula is found in [\textit{C. Fr. Gauß}, Allgemeine Flächentheorie. (Disquisitiones generales circa superficies curvas. Frankfurt am Main: Harri Deutsch (1921; JFM 48.0008.01), page 157] and in [\textit{J. L. W. V. Jensen}, Ann. Math. (2) 17, 124--166 (1916; JFM 46.0563.02)]. The paper concludes with some interesting trigonometric calculations, which give several expressions for hypergeometric series. In this way, formulas for the following hypergeometric series for rational \(a=p/q\), where \(p,q\in\mathbb{N}\) and \(0<p<q\), are proved:\N\[\N_{3}F_{2}\left[\begin{array}{c} \frac{1-a}{2},\frac{1+a}{2},\frac{1}{2}\\\N\frac{3}{2},1 \end{array}1\right],\ _{3}F_{2}\left[\begin{array}{c} 1-a,1+a,\frac{1}{2}\\\N\frac{3}{2},2 \end{array}1\right],\N\]\Nand\N\[\N_{4}F_{3}\left[\begin{array}{c}1-\frac{a}{2},1+ \frac{a}{2},1,1\\\N\frac{3}{2},\frac{3}{2},2 \end{array}1\right],\ \ _{4}F_{3}\left[\begin{array}{c} 1-a,1+a,1,1\\\N\frac{3}{2},2,2 \end{array}1\right]. \N\]\NFormulas for the last two hypergeometric series hold for arbitrary \(a=p/q,\ 0<p<q\). In the trigonometric formulas, the function argument \((2k-1)\) shall always be multiplied by the next quotient factor.