He had to rely entirely on his own invention, and to earn his own experience. James Short had solved the problem of giving to metallic surfaces a perfect parabolic figure (the only one by which parallel incident rays can be brought to an exact focus); but so jealous was he of his secret, that he caused all his tools to be burnt before his death;[320] nor was anything known of the processes by which Herschel had achieved his astonishing results. Moreover, Lord Oxmantown had no skilled workmen to assist him. His implements, both animate and inanimate, had to be formed by himself. Peasants taken from the plough were educated by him into efficient mechanics and engineers. The delicate and complex machinery needed in operations of such hairbreadth nicety as his enterprise involved, the steam-engine which was to set it in motion, at times the very crucibles in which his specula were cast, issued from his own workshops. In 1827 experiments on the composition of speculum-metal were set on foot, and the first polishing-machine ever driven by steam-power was contrived in 1828. But twelve arduous years of struggle with recurring difficulties passed before success began to dawn. A material less tractable than the alloy selected, of four chemical equivalents of copper to one of tin,[321] can scarcely be conceived. It is harder than steel, yet brittle as glass, crumbling into fragments with the slightest inadvertence of handling or treatment;[322] and the precision of figure requisite to secure good definition is almost beyond the power of language to convey. The quantities involved are so small as not alone to elude sight, but to confound imagination. Sir John Herschel tells us that "the _total_ thickness to be abraded from the edge of a spherical speculum 48 inches in diameter and 40 feet focus, to convert it into a paraboloid, is only 1/21333 of an inch;"[323] yet upon this minute difference of form depends the clearness of the image, and, as a consequence, the entire efficiency of the instrument. "Almost infinite," indeed (in the phrase of the late Dr. Robinson), must be the exactitude of the operation adapted to bring about so delicate a result. At length, in 1839, two specula, each three feet in diameter, were turned out in such perfection as to prompt a still bolder experiment. The various processes needed to insure success were now ascertained and under control; all that was necessary was to repeat them on a larger scale. A gigantic mirror, six feet across and fifty-four in focal length, was accordingly cast on the 13th of April, 1842; in two months it was ground down to figure by abrasion with emery and water, and daintily polished with rouge; and by the month of February, 1845, the "leviathan of Parsonstown" was available for the examination of the heavens.