Savary also served as librarian at the Bureau des Longitudes from 1823 to 1829. Then on 24 December 1832, in recognition of his achievements, he was elected to the Académie des Sciences.
He worked on electromagnetism and electrodynamics, some work being done jointly with Ampère. In particular, on this topic, he wrote Mémoire sur l'application du calcul aux phenomènes électro-dynamique Ⓣ (1823).
Savary also developed a theorem (named after him) on the curvature of a roulette, the curve traced out by a point on a fixed curve which rolls on a second curve.
He wrote on the rotation of magnets, studied the intensity of magnetism through an electrical discharge (1827), and applied the laws of gravity to determine the orbits of double stars in close orbit round each other (1827). In fact, on the topic of double stars, he published Mémoire sur les orbites des étoiles doubles Ⓣ (1827), and Sur la détermination des orbites que décrivent autour de leur centre de gravité deux étoiles très rapprochées l'une de l'autre Ⓣ (1827) in Connaissance des Temps. The star x Ursae Majoris is a double star and Savary demonstrated that the two stars move in elliptical orbits with the centre of gravity at the focus of the ellipses. Although this might appear to be a fairly simple consequence of Newton's law of gravitation, nevertheless it was important for it was the first verification of the laws for objects outside the solar system. This was the first verification of the universal nature of Newton's laws. We look at this episode in a little more detail following .
It was on 2 May 1780 that William Herschel discovered that the star x Ursae Majoris was a visual double star. The relative positions of the two stars were first accurately measured in 1826 by Friedrich Georg Wilhelm von Struve, who was at that time director of Russia's Dorpat Observatory. William Herschel's son, John Herschel :-
... was one of the first people to compute orbits for double stars, and x Ursae Majoris was the system that he selected for the initial demonstration of his method, in 1831. Priority in that field, however, must go to Savary, who gave, in Connaissance des Temps for 1830 (which according to its title page was published in 1827), a description of a method of calculating orbits for double stars. In a supplementary paper in the same volume, he opens by saying (in French of course: the following is my translation), "It has occurred to me that it could be useful to add to [the formulae for computing orbits] their numerical application to a specific example. So I will suppose the following data, without attributing any reality to them" - and he lists four dates with associated separations and position angles, all highly reminiscent of, but not identical with, actual observations of x Ursae Majoris. Five pages later, at the conclusion of some stiff mathematics, he ingenuously remarks, "The data from which we started in the preceding calculations are very close to four observations of the double star x of the great bear"; he then goes on to tabulate all 11 of the observations available at the time, and to show that the computed orbit satisfies them very well. In fact, Sir John Herschel, whose orbit was, on his own admission, not nearly such a good match, complained with good reason that Savary had cheated - not Sir John's word: that is my précis of the gentlemanly circumlocution that occupies nearly a whole page (his paragraph 64) - by choosing to solve data that were only approximately the observed ones!
Article by: J J O'Connor and E F Robertson