MKTHODS <)!•' CASTINO TKST HAMS I-'O K THK A. K. A. 557 that, as a rule, those castiii the former moulds were weaker than in the latter. One hundred tests of different green sand bars, averaging closely alike in size, gave an average strength of 33,700 pounds, whereas i oo tests in dry sand bars gave an average strength of 31,751 pounds, showing a difference of 1,949 pounds or 6 per cent, greater strength for the bars in green sand than those in dry sand. The gray iron showed the greatest and most uniform difference. There were a few casts, in both the ehilled and gray iron, in w h i c h the dry sand bars aver-a g e d the g r e a t e s t strength. One of these varieties is shown in the unfinished dry sand bars of Table l\ i j.j., page 568. It is natural to expect the green sand bars to show the greatest strength on account of the chilling influence of a damp mould. The ivsults of the original tables shown in the A. F. A. Journal also show that tests of green sand bars are more erratic than those of dry sand, although, as a rule, the difference is not sufficient to cause the dry sand bar to be given the preference in general practice; but where the greatest delicacy in testing is desired, by the use of unfinished bars, then the dry sand bar would be preferable. The author selected the bars from green sand for the Tables 115 to i .»0 for the reason that such are almost entirely used in general practice, and hence will permit of a better 144. separated, these so as to give the strength per square inch of the tensile tests in the independent Table 126, to be above the chemical analyses of the different specialties shown in Table 127, both seen on page 570. The actual load at which tensile bars broke is shown in the last column, of easts.A, H, ("and (5 to L. The form of bars as turned for the tensile tests is seen in Fig. i.-pS, page 583.