The experiments made by M. Melseuns on oaken blocks exposed to the fumes of _liquid ammonia_ show that the conservating fluids follow the precise course that would be taken by decay. In wood treated with creosote the tar acts on the very parts first exposed to injury, and on the course that would be taken by decay, which is thus rendered impossible. The methods of injection suggested by M. Melseuns in 1845 did not answer equally well with every kind of wood. After trying wooden blocks in every sort of condition, dressed and in the rough, green and dry, sound and decayed, M. Melseuns found that alder, birch, beech, hornbeam, and willow were easily and completely impregnated; deal sometimes resisted the process, the innermost layers remaining white; poplar and oak offered a very great resistance--indeed, with poplar it was found necessary to repeat the process. The decay of sleepers, prepared and unprepared, will often depend on their form. Three forms have been used: 1st, the half-round sleeper, 10 inches by 5 inches; these are now almost universally used; 2nd, the triangular sleeper, about 12 inches wide on each side, used by Mr. Cubitt on the Dover line, but since abandoned; and 3rd, the half square, 14 inches by 7 inches, used by Mr. Brunel and still in use. Mr. G. O. Maun, in reporting on the state of the sleepers of the Pernambuco Railway, states that fair average samples taken out on the 1st December, 1863 (laid in 1857), show that the half-round intermediate sleeper is in the most perfect state of preservation; in fact, nearly as good as on the day it was put down; while the square-sawn or joint sleeper has not withstood the effects of the climate so well. The kind of ballast in which it will be most advisable to lay the sleeper is another important point to be attended to. About 12 miles of the Pernambuco Railway are entirely laid with creosoted sleepers, principally in white sand. In this description of ballast the half-round sleepers have suffered, since the opening of the first section of the line in 1858 up to 1866, a depreciation of not more than 1 per cent., whilst the square-sawn sleepers have experienced a depreciation of not less than 50 per cent. Had the latter been placed in wet cuttings with ballast retentive of moisture, no doubt the whole of them would have required to be renewed. Hence it is evident that fine open sand ballast, which allows a free drainage during the rains, is best adapted for the preservation of sleepers in the tropics: it has also been found to be the best in most countries.