By imperfect mechanical composition is meant imperfect union of the parts of rails. Steel heads are welded to the rest of the rail in a variety of ways, and this welding is necessarily imperfect. A number of sections of rails etched with acid plainly showed this want of homogeneity, as did likewise prints taken from the etched surfaces. Before such rails have lost weight appreciably, they are used up by the constant rolling they undergo. The advantage of a steel rail is its homogeneity, but a good iron rail, such as those made under the direction of the speaker, for the Reading Railroad Company, is likely to prove better than one of poor steel. The life of a steel rail is chiefly affected by the temperature at which it is rolled and annealed. It ought not to wear off more than 1 mm. for 20,000,000 tons of traffic, and is usually calculated to wear 10 mm. before it is taken up. In other words, it would last about 20 years on roads doing as much business as the New York Central. It is, however, unlikely that our steel rails will stand more than half this amount of traffic. The effects of chemical composition are but little understood. Some of the purest irons have turned out utterly worthless. Apparently the absolute quantities of carbon, silicon, aluminum, phosphorus, etc., present are not of so much importance as their relative proportion. One specimen containing carbon 0.16, silicon 0.08, and phosphorus 0.012, could be bent double when cold, while another, containing carbon 0.58, silicon 0.56, and phosphorus 0.011 broke at once. The physical tests for tensile and torsional strength, usually made on a portion cut out of the head of the rail, are not sufficient, because the flaws before spoken of exist mostly in the flange of the rail, and fracture usually begins there. The effect of cold rolling and shocks that a rail is exposed to was shown by a piece of rail made by the Campbells, Sheffield, Eng., which had been worn 3 mm. by a traffic of 60,000,000 tons at Spuyten Duyvel. The head had been somewhat flattened, and the flange driven down into the foot to a certain extent. Under such usage an iron rail would have gone to pieces long ago. Sometimes steel rails crumble all at once and pieces fall out of the head. This is probably due to some physical defects or to crystallization from shocks. The cause has not yet been definitely ascertained.