METALLURGY OF CAST IRON. of manganese in steel castings showing a contrast between round and square fractures. A study of Figs. 105 and 106 impresses one with the importance of arranging for the greatest possible uniformity in providing for the radiation of heat from a test specimen, and also to afford it the most favorable condition to arrange its crystals uniformly throughout its body. It requires no great stretch of the imagination to conceive what a great influence the simple matter of slight differences in the " temper " of sand in a mould may have in causing non-uniformity in the even texture of a square bar compared to the even structure possible in a round bar. Mr. John E. Fry, in a paper before the Eastern Association, May 2, 1894, condemning one-half inch square test bars, clearly illustrates the effect of a little variation in the " temper " or dampness of sand, often making small bars wholly unreliable as a test for the relative strength of any kind of cast iron. Before leaving Figs. io5and-io6,let me call attention to their clear exemplification of the necessity of casting test bars on end, in order to insure uniform cooling off. The heavy-work founder knows that metal first solidifies at the bottom of a mould, and if he is '"feeding" a heavy casting, the metal, by solidifying at the bottom first, will gradually force his " feeding rod" upward, thus demonstrating that the greatest FIG. 105. FIG. 106. I Sifif, iH.j J,ould from head pressure are liable to make the area of the bars vary at different heights.on produced.