Page:Hector Macpherson - Herschel (1919).djvu/48

42 of the atmospheric condition of Jupiter what has been known as the "trade-wind" theory. "As the principal belts on Jupiter are equatorial, and as we have certain constant winds upon our planet that regularly, for certain periods, blow the same way, it is easily supposed that they may form equatorial belts by gathering together the vapours which swim in our atmosphere and carrying them about in the same direction. This will by analogy account for all the irregularities of Jupiter's revolutions." Herschel devoted no other paper to Jupiter, but in 1797 he communicated to the Royal Society his "Observations of the Satellites of Jupiter, with a Determination of their Rotation". From his determinations of their variable brightness, he concluded that the rotation periods of all four satellites coincided with their periods of revolution round Jupiter—a conclusion confirmed by subsequent research. Herschel also made the first attempt to measure the diameters of the satellites. His conclusion—also confirmed by later observers—was "that the third satellite is considerably larger than any of the rest; that the first is a little larger than the second and nearly of the size of the fourth; and that the second is a little smaller than the first or fourth or the smallest of them all".

"The planet Saturn," Herschel wrote, "is perhaps one of the most engaging objects that astronomy offers to our view. As such it drew my attention as early as the year 1774." And it received more of Herschel's attention than any of the other planets. Six of his papers to the Royal Society dealt with Saturn, the first in 1789 and the last in 1805. He concluded in 1789 that the planet had an atmosphere of considerable density; and from the appearance of the belts he inferred that it "turns upon an axis which is perpendicular to the ring," and this view was confirmed by his detection of a considerable polar flattening. In December, 1793, he stated that the period of rotation "is probably not of a long