Suppose a case of two rails abutting which are to be welded together. It is a railway crossing where heavy trains pound. The weld must be at least as strong as the rail. It must be so made as not to interfere with the travel of the wheels by coming up over the head of the rail. First of all, the rail ends must be cleaned of oxid and grease with a sand blast or emery-paper or hydrochloric acid. Next, the rail ends are heated to a dull red heat with a kerosene or, preferably, a gasoline torch. This merely assists the hot thermit metal and prevents a premature chilling of the thermit when it is poured in the mold. Two clay molds are next clamped on either side of the junction. The shape of the interior of these molds is, of course, determined by the shape of the collar which is intended to be cast. In this case, as shown in figure 64, the collar should extend 2 inches over each rail end. It shall be twice as thick as the shank of the rail and also the base of the rail. It shall stop short of the rail heads, which shall remain free. The mold is constructed so as to allow the molten metal to be introduced from the bottom, as shown in the figure. After coming in from the runner, at the bottom, the slag and excess steel overflow into the riser.

Fig. 63. - Rails before welding.

Fig. 64. - Welded rail, showing thermit-steel shoulder.

We have then two rails enclosed in a mold whose capacity, over the rails themselves, is known. To produce enough thermit steel to well fill this mold and to allow as much more to fill the runner and the riser, take eighteen times as many ounces of thermit powder as there are cubic inches of surplus space in the mold.

The above amount is arrived at as follows: One cubic inch of steel weighs 4 1/2 ounces. Four and one-half ounces steel is produced by twice as much thermit powder by weight, or nine times. And as the runner and riser take as much fluid as the inside of the mold, we multiply again by 2 and get eighteen.

When a wax collar is first built on the joint, the amount of thermit should be thirty-two times the weight of the wax used. The weight of the wax used is found by subtracting the weight of the piece of wax remaining from the total weight of the original wax lump.

The proper amount of thermit powder is poured into the cone crucible (Fig. 65), and a spoonful of barium hydroxid is heaped upon the thermit. The crucible is placed with its tap hole about 4 inches above and directly over the hole of the riser in the mold. Set off the barium powder with a storm match and get away as soon as the barium is caught. The burning quickly spreads from the barium fuse to the thermit, and in a fraction of a minute the entire contents of the crucible are boiling at a temperature of about 2500 deg. Cent. White smoke, flames, and drops of white hot slag are ejected during the combustion, which is most spectacular and reminds one of the blowing of a Bessemer converter. In working with thermit is is well to wear smoked glasses, as the glare of the reaction and the hot fluid is troublesome. In about thirty seconds the reaction is completed, but the crucible should be allowed to stand for a half-minute longer to enable the slag to rise to the surface. It is probably for the reason that the slag does not have time to rise before the workman taps his crucible that the joints sometimes show blow holes and faulty structure. About a minute after lighting the fuse, the workman knocks the stopper out of the bottom of the crucible, and the white-hot metal pours out into the mould. As the stream enters the mold from below (Fig. 71) it heats the ends of the rails and passes on up and out into the riser. The last of the metal stream remains in the mold, and as it is very much hotter than the melting point of steel, it eats into the sides of the rails and knits fast on cooling. The joint should remain undisturbed for at least five minutes to allow the metal to harden. It may then be treated in a number of ways - either allowed to cool slowly in the mold, in which case the joint will be composed of soft, tough steel, or quenched in oil from a red heat, in which case the joint will be very hard, and perhaps brittle.

While this description does not give all of the steps of rail welding, it will give the reader a fair idea of how all thermit welds are made. The apparatus used is as follows: