During the past few years, the encroachment of steel castings upon the sphere of pig iron has been very extensive, and perhaps this statement may also forcibly apply itself to the forge and smithy. In order that castings may be considered first-class, it is requisite that they should be quite compact and sound. This is of greater importance in specifying castings in foundry practice, and is more valuable in determining their strength than the quality of the mixings from which they are run; therefore in all cases castings are designed with a view to free themselves from blowholes and dirt, heavy sections being cast uppermost, so that there is not any shrinkage due to lack of feeding. Fillets are well attended to, because, instead of being a factor of safety, they may become the reverse. They should be curved, because there is then less liability to crack than when made straight; and frequently light sections of a complicated casting are strengthened, in order that internal strains may be equalised, as well as to avoid draw, caused by unequal contraction. At the same time due consideration should be given to the crystallisation of the metal, the contour of all castings being as far as possible free from abrupt changes.

The difficulties met with in the iron foundry, as compared with those in the steel foundry, are comparatively nothing. The mechanical properties and constructive uses of iron castings, set them entirely apart from steel, but generally speaking, the one element, carbon, characterises the quality of each. The greater the content of combined carbon, the harder and less fusible the pig; the greater the content of graphitic carbon, the softer and more fusible, and cases are known in which silicon up to 3« per cent. has been added to white irons, resulting in good strong soft costings. Phosphorus of course, renders the metal more fluid, but the resultant casting brittle, Remelting iron up to a certain extent improves its tenacity and transverse strength, but beyond, these properties suffer, and its resistance to crushing and hardness increases.

The tensile test is the most trustworthy for iron castings, as its results are much more favourable to a good working foundry mixture than the resistance to crushing, because frequently the best crushing tests are quite unsuitable for foundry use- Commercial testing is generally made upon bars 3 feet by 2 inches by 1 inch without being machined, specifications requiring from 25 to 32 cwts. as a central breaking load, allowance being generally made for blown and scabby samples in such terms as, "Should any of the tests he lower than 1 per cent, of the above strains they will be rejected." Some American specifications state that the bars shall be cast horizontally, the pouring to be done at two points simultaneously, the gates being one-third the distance from either end. When east otherwise the method must be accurately reported to the purchaser. Dry sand moulds are used, the tensile and compression test bars being taken from the transverse bars. Bottom-poured bars are objected to, because they chill readily, through the metal constantly coining in contact with a higher and colder portion of the mould, and everything points, as far as the author is aware, to top pouring as being the most beneficial, especially with a goad head of metal, for then the tensile result may be improved from 5 to 8 per cent. Generally the tensile tests do not exceed 8 or 9 tons per square inch, but are always specified to exceed 7 tons per square inch, and the author has tested samples of general iron up to 11« and 12« tons per square inch; cylinder iron to 13« and 14 tons per square inch; each having very good fractures and being good workable metal.

The tensile tests are always made with shackles having spherical seats, as the ultimate strength is very much affected by the non-axial position of the sample. It is frequently asserted that American castings are far ahead of English, and probably taking them weight for weight and sectional area, they are stronger; but at the same time, the strongest and best castings can be obtained from the English foundries, provided the purchaser is prepared to pay the price. Hence, the skill of a foundry manager is taxed to a great extent in blending certain brands of pig of which he has practical and analytical knowledge, to produce the class of casting required, embodying the requisite hardness or softness, with strength and toughness. This is carried out to such an extent in some foundries, that the selected brands are melted together, cast into pigs, broken, the best fractures reserved, and at the same time test bars are cast, in order to gain information for this special work. Should a contract be let at a very low figure, stringent means must be taken to keep the founder up to specification, and then, with all care, an inspector is not absolutely certain that his transverse bars are of the same quality as the castings, or even if they are from the same heat, therefore such samples should form part of the casting, and the inspector should see them broken from that casting, or, as in the case of girders and similar work, from 2 to 4 per cent, extra should be cast, and then selections made from the bulk, tested to destruction, and should the selection break below the specified load, the bulk should be condemned. It is often advisable, in these deflection tests, to note the ratio of the first permanent set to the ultimate load, which is perhaps sometimes a fine accomplishment, and rewarded by curious results.

In moulding, many of the operations are common to all castings. Patterns should have good draw, and be provided with draw screws or irons; the rapping holes and the general arrangements for loosening should also be well attended to. Core prints are not absolutely necessary in all cases; where they are large and rest upon the bottom of the mould, they can be set in position by measurement, and their weight, or the combined weight of them and the top part of the mould, will keep them in position, but usually cores are fixed by means of prints, and justly so, because they will save time and the risk is infinitely less. The cylinder patterns and large pipes are lagged up, which is the correct way, jointed longitudinally and accurately dowelled. A large piece of timber, well seasoned and dry, is not always available, and there is a greater amount of warping and shrinkage than when lagged up with 1«-inch timber. In the end it is cheaper to form the shape between the two cylinders, by a shaping core, because if this was formed by the pattern, it would make the pattern very weak where it requires to be strongest, not to mention the use of gaggers and irons.