Square root approximations in Old Babylonian mathematics: YBC 7289 in context (Q1282317)

The tablet referred to in the title shows a square with the two diagonals. Along one side is written ``30'', and along one diagonal the numbers ``1, 24, 51, 10'' and ``42, 25, 35''. The tablet was first published by \textit{O. Neugebauer} and \textit{A. Sachs} in Mathematical cuneiform texts [American Oriental Series. 29, New Haven, Conn. (1945; Zbl 0060.00401); henceforth MCT], who already noticed that 1, 24, 51, 10, interpreted as 1; 24, 51, 10 is a very good approximation to \(\sqrt 2\) (and 42, 25, 35 thus the length of the diagonal in an order of sexagesimal magnitude corresponding to the choice for the side). They showed that the value can be derived by ``alternating approximation of \(\sqrt a\) by arithmetic and harmonic means of previously found approximations''. The present article makes a number of points with regard to this. (1) It shows that the small round tablet under examination is of a type (im.šu, ``hand tablet'') currently used for rough work by students when solving school problems, which is hence the likely character of the tablet -- the value 1; 24, 51, 10 being probably taken from a table of constants. (2) It discusses, with examples, the normal Old Babylonian first approximation to irrational square roots, \(\sqrt{n^2+d}=n+{d\over 2n}\) \((d\) positive or negative), and how this formula follows from elementary geometric considerations (familiar in medieval algebra, but not referred to in MCT). Starting from \(n=3/2\), this gives the approximation \(\sqrt 2=1;25 (=17/12)\), which appears elsewhere in the Old Babylonian record. (3) It considers the outcome of an iteration of this procedure, which is indeed algebraically identical with the outcome of the alternation of arithmetic and harmonic means of MCT, and thus yields the value of the tablet. (4) It shows that this is not likely to be the way the Babylonians derived their value, since the computation of the second-order correction involves division by the ``irregular'' number 1;25, and that any reasonable regular approximation to 1;25 gives a wrong result. It is pointed out that the first approximation is also irregular in the other instances considered, which provides a good explanation that the Babylonians go no further. (The reviewer may add that the tablet VAT 6598 does contain what looks as an attempted second-order approximation -- but since the unpleasant division is forgotten, it errs much more than the first approximation). (5) It quotes Hero's rule for approximating square roots (Metrica I.8 [reviewer's observation: probably an interpolation]), which speaks explicitly of the possibility of iteration, and points out that the procedure, though algebraically equivalent with the Old Babylonian rule, is computationally much simpler. Whereas a second-order calculation using a regular approximation to the first-order result is prohibitively cumbersome if we use the Old Babylonian rule, Hero's (?) rule would have made it possible.