Al'I'KMMX TO OH U'TKK I. IV. TK KATS SH K INK A< IK. 407 the castings came to the hands of a moulder who understood the cause of shrink holes and could tell such cavities from blow or dirt holes. After this moulder had made one mould and observed the proportion of thicknesses in the easting there was no more trouble. The. difficulty had lain in not providing means to convey hot metal to supply the shrinkage of the heavy part. This was d<»ne by attaching a feeder, as at II, having a connection with the casting, as at J, both bodies of which were so much larger in area than the section of the casting at. < J that assurance was afforded that the metal would solidify in the heaviest section of the casting at (» before it. would do so in the feeders II and ), thus jn'ivintf a head <»f molten metal which could settle down from tin* feeder to make a solid casting. Pouring these castings on end, instead of on their flat, could do no in>od, as the metal would solidify first in, the thin part of I/ lonjj; before it would do so in the heavy section of (i. If a heavy feeder as at the dotted line M, madr of the same pr«»portions •^ \ and II, had 1 ecu carried down from t In* top of Jhe upended mould to the heavy section, sound castings would have boon produced, but otherwise th«*v were as well made on their Hat as on their cud. Another Illustration of this principle of fooclin^1 is found in not obtaining"™" "•««.«JWUHAU.' Ih*, H?, i 'Vl.lMH-'.U liIIOWINti I'OSI- souwl ilan^rs, as at N, Fit;. UON <»i- SUKIISK nou-.s.