EFFECT OF EXPANSION ON SHRINKAGE, ETC. 387 The fact of this expansion was first practically demon-stratedby Mr. John R. Whitney, of Philadelphia, Pa., whose experiments are recorded in the National Car and Locomotive Builder of May, 1889, and cited in Chapter LIIL, page 382. Experiments carefully made by the writer indicate that there is a constant relation between this expansion and the preceding shrinkage and forcibly demonstrate the necessity of "feeding" a casting to make its interior solid. This is a matter with which all makers and users of castings have experienced difficulty. The founder being heretofore unable to define correctly the principles involving the urgent necessity of c' feeding,' * has failed to impress the moulder with its importance in making sound castings. Heavy-work founders and moulders know that hard grades of iron shrink much more than soft grades, a fact for which no satisfactory explanation has heretofore been given. By recent expansion experiments I have discovered that hard grades of iron expand more at the moment of solidification than soft ones. Fig. 74, page 389, is a diagram recording four such experiments. The manner in which the automatic records were obtained will be described further on. It is sufficient to say at present that the scale of inches in the diagram measures the length of travel of the pencils on the long recording-arms of the apparatus employed, not the actual length of expansion. The end of the short arm of each lever, following actual expansion, travels ^ inch for i inch traveled by the pencil, and the length of the test bars being 48 inches, i inch of the expansion or contraction record represents an actual expansion or contraction of 3 in 1536, or 0.195 ries: Prof. Edward Turner, in his "Elements of Chemistry, ''published in Philadelphia in 1835, by Desilver, Thomas & Co., says, page 20: "Water is not the only liquid which expands under the reduction of temperature, as the same effect has been observed in a few others which assume a highly crystalline structure in becoming solid; fused iron, antimony, zinc and bismuth arc examples of it." Prof. Thomas Graham, also, in his " Elements of Chem-. 5 o^.