where the partial quotients a¡ are integers, positive for / > 0. We write
a = [a0, fli, ... ], and the truncated expression at the ith step is then a
rational number /?//#/ called the /th convergent for a. Although some facts
on such expansions were known to, and used by, the Indian astronomer
Aryabhata (circa 450), Bombelli (1526-1573) and Huygens (1629-1695), it
is clear that Euler was quite unaware of their work. In fact Euler was the
first person to give a general account of the subject, which initially appeared
in his correspondence in connection with Riccati's differential equation. He
soon became interested in them for their own sake, observing that rational
numbers have finite continued fractions obtained by a process identical with
the Euclidean algorithm, that periodic continued fractions represent
quadratic irrationals, and noting also that the expansion of any real number
into a continued fraction supplies the best rational approximations for that
number. Besides deriving the iterative formulae for p¡ and q¿ in terms of the
partial quotients at he also gave 'Euler's rule' for the explicit expressions for
them as sums of certain products formed out of these a¡. It was natural for
him to compare the approximations to ÍÑ from his new-found theory to
those obtained from the solutions to Pell's equation, and he discovered that
the two algorithms are in fact identical. It was inevitable that he also
discovered the 'palindromic' property of the expansion, namely that the
expansion has the form