ROUND VS. SQUARE TEST BARS. 485 line for radiation or line for heat to escape is upward, or through the " cope " of a mould0 For this reason, if we would break a casting a foot square into halves down the center of its vertical position, as when cast, we would find the last spot to solidify would generally be about three inches from the top, or one-fourth its height below the cope surface. It makes no difference how small a body of metal may be, the same principle is applicable to it as to the large body, and goes to fully demonstrate the irregularity for a central point of latest solidification which must exist in a test bar cast flat. Then again, uneven cooling is bound to cause more -or less internal contraction strain in a test bar. It must be evident that a test bar cast on end will ^ have an even radiation from all portions of its surface at any height, and thus give to the bar the best uniform grain throughout any section and also the best opportunity to lessen strains so far as cooling off has any effect. More information on the necessity of casting test bars on end will be found in the next Chapter, page 488. The nature of all cast iron is such that any elements in a mould possessing heat-conducting powers, that will either chill or make closer the grain of the metal in the skin or surface, are very effective in changing results in the strength and contraction of iron, especially in light castings or small test bars. There is a great difference in iron in its susceptibility to elements tending to chill. Some iron, if poured into a dry sand mould, would show a gray fracture, but if poured into an iron or green sand mould, would show at the surface a • white or chilled iron, the depth of which depends upon the character of the iron, the thickness of castings, either in a letter discussing the testing of cast iron and attacked the usual formulae for loaded beams asound burs..........................*fi$7 "