DKSK.N OF TKST MAR AI'1'1,1 ANCJKS, KTC. 513 ijginvs these \vould measure one and one-eighth inches diameter at the large end, and one and three-tliirty-seeonds of an inch at the small end, and of the same length seen in Pig. 122. By having a ring; at the large end, as seen at II, Pigs. 122 and 128, the smaller end will always be the down one in moulding, and in ramming the mould, do so to such a degree of hardness as to permit suilieient straining, due to head press-ure, to have the castings eome out closely alike as to size at the bottom and top. It is well to mention at this point that should any desire to make their test bars in a " dry-sand " mould, they can readily do so, as there is no wood whatsoever connected with the flasks, thus making it practical, to place the mould in an oven to be dried. For malleable and steel testing and some speeial purposes in iron, a i4 dry-sand " mould might often be found a very good method to adopt. Referring to the question of " chilling/' it cannot but be readily seen that as arranged by this system, the test bar and the chill must remain in close contact until removed by hand, hence truly recording the full chilling qualities of the iron. At V V, Pig. 126, page 522, can be seen the chill used in this system. It is simply two half-circles three inches long by three-eighths of an inch thick, having a hole drilled in them to fit over the pattern tips W \V, Pig. 122, These chills are set on over the pattern before starting to fill the nowel with sand, and in shaking out, must, of course, be picked up and used as long as they last. They are made of a soft steel shaft, which, after being drilled or bored out, are then split as seen. See page 502. In the case of very hard grades of iron, such asld ask such, after having I''