RECORD OF TESTS TAKEN FROM IRON SEEN IN FIG. 37. No. of Bars. Size of Bars. Contraction. Chill. i %*i% .293 Nearly white. 2 %xilA -.266 Y4 deep. 3 %xi^ .242 K deep. 4 ^xi^ .220 3-16 deep. 5 f$xi^. .200 3-16 deep. 6 %XI& .182 l/s deep. 7 %xi^ .165 YB deep. 8 I XI& .150 3-32 deep. 9 i# x i# .148 3-32 deep. TABLE 25. RECORD OF TESTS TAKEN FROM IRON SEEN )N FIG. 38. No. of Bars. Size of Bai s. Contraction. Chill. i HXI# .178 .03 deep. 2 &XI^ .163 .02 deep. 3 3/3XI^ .150 .01 deep. 4 ^XIJ^ -137 Hardly perceptible. 5 V8xi^ .125 No chill. 6 -KXI^ .112 No chill. 7 H x i^ .IOI No chill. 8 i x iy2 .92 No chill. 9 ij<xij< .88 No chill. This chill was obtained by causing- the end of the test bars farthest from the gate to be formed by a wrought iron bar three-fourths by two inches wide. The twelve test bars of each set were molded in green sand and poured from one gate. The same " temper " of sand was used for both flasks, and the iron was alike in fluidity at the time of pouring. Only nine tests out of each of the twelve bars seen in Figs. 37 and 38 are given. To further demonstrate the deceptive appearance of fractures in pig iron, analyses of three pieces of pig thick. In fact, we were foolish to try to touch it with a drill, as the metal was nearly all chilled or white in color. It is also to be said that all the other test pieces ranging from Nos. 2 to 12 that were made from the pig, sample A, were also much harder than those made from sample B. In measuring the depth of the chill, pieces were broken off one end of the test bars as seen at P, Fig. 37.ith chemical analysis, or the effect of one metalloid upon another, that they are in error, the writer melted down about one hundred pounds of each of the grades A and B in his twin-shaft cupola, seenng the advance of the new, see page 179.n is simply an average of the whole, generally taken from the two ends andith the uncertainty of furnace workings when in urgent need of ten hundred jon of iron; and Sir........................ 2,720 "