was supported by the observation that kish, which is a form of graphite in large flakes or plates, separates from grey cast iron when the metal is kept long fluid, as in the hearth of the blast furnace, or in a foundry ladle. But the researches of W. J. Keep, to which further reference will be made (see Chap. V.), have shown that important changes take place in the red hot metal after it has become solid and while it is still cooling in the mould. It will be sufficient here to observe that with ordinary grey cast iron there are three distinct arrests in the rate of contraction or expansion observed while it is cooling from the molten state to the temperature of the air. Similarly with grey cast iron there are also three arrests noted in the cooling curves taken by a recording pyrometer, through the same range of temperature. Though there is room for much further observation in this interesting direction,' sufficient work has already been done to show that there is a considerable separation of graphitic carbon from grey cast iron long after the metal has become solid, and some most important changes take place at or below 900° C. These cooling arrests present some analogy to the effects observed when certain varieties of steel are cooled through the same range of temperature.

In cast iron these changes of volume appear to be connected with the separation of temper carbon, and when well marked are characteristic of strong, soft foundry metal.

The iron-founder may perhaps at first be a little overwhelmed with the apparent complexity which is introduced by further knowledge in reference to the states in which carbon occurs in cast iron, and the conditions under which it separates, but doubtless many of the difficulties which were formerly met with in reconciling the results of chemical analyses with the physical properties of the castings were due to empirical methods; and with improved knowledge, doubtless, such difficulties will gradually disappear.

Silicon in Cast Iron.

All cast iron contains more or less silicon, the usual proportions being from about 0.25 per cent. in specially pure, or in white pig iron, to a maximum about 4.5 per cent. in ordinary blast furnace practice; but pig iron which contains 10 per cent. of silicon is regularly made in the blast furnace for special purposes, and 15 per cent. or 18.per cent. of silicon can be introduced in the blast furnace if required. 1 See Note, p. 131.

By the use of the electric furnace any proportion of silicon which is desired may be obtained; 30 per cent. and 50 per cent. ferro-silicons are now commercial articles, while the element itself can be purchased, of about 98 per cent. purity, for less than 10s. per lb. The rich alloys are, however, like the element itself, too expensive and too infusible to be profitably employed by the iron-founder.

In reference to the mode in which silicon occurs in cast iron, it was at one time believed that silicon was present both in the combined and in the graphitic form. But more careful investigations conducted in recent years1 have shown that silicon always exists in cast iron as a silicide. This silicide, according to Carnot and Goutal,2 usually has a composition corresponding to Fe,Si, though from rich ferro-silicons the compound FeSi has been isolated, while, when much manganese is present, a silicide (FeMn),Si may also occur.

When irons which contain a considerable porportion of silicon are microscopically examined, the silicide is observed in characteristic crystalline forms; but in ordinary foundry iron, in which the silicon does not exceed 5 per cent., the silicide cannot be distinguished by the eye, as it exists in the form of a solidified solution in the iron, and forms part of the structureless matrix from which the graphite is at first thrown out, and from which the phosphorus eutectic, if present, afterwards separates.

It was at one time very generally held that silicon was injurious in all proportions when present in cast iron, though it had been pointed out by Sefström, by Snelus, and still more plainly by Ledebur, that with irons in which much silicon is present the carbon is wholly, or nearly wholly, in the graphitic state. It was also known that certain silicious irons obtained in the United States could be advantageously used as softeners in the foundry; but it may be claimed that there was no accurate or quantitative knowledge of the influence of silicon previous to 1885, when my first paper on "The Influence of Silicon on the Properties of Cast Iron" was published in the Journal of the Chemical Society. The principle adopted in connection with these experiments was extremely simple, though it does not appear to have been previously applied. The practice had hitherto been to endeavour to deduce from the analyses of samples produced in actual practice the influence of the various elements which were present. On account of the number of variables which have to be considered, 1 For an outline of these researches see Metallurgy of Iron, p. 197.

2 Inst. Jour., 1899, vol. i. p. 453. See also P. Lebeau, ibid., 1901, vol. ii. p. 513.

this, which may be called the analytical method, involves a very large number of observations being made before any definite conclusion can be drawn. For the other, or what may be called the synthetical method of investigation, cast iron was first specially prepared so as to contain as little silicon and other impurities as possible. It was then mixed with ferro-silicon, containing about 10 per cent. silicon, so as to give a graduated series of test bars with definite proportions of silicon, while the other elements were kept as nearly as possible constant. By adopting this method more information was obtained from a single series of experiments than could have been gained by many years of practical observation, and the synthetical method of investigation in the hands of Arnold, Hadfield, Keep and other investigators, has since widely extended our knowledge of the metallurgy of iron and steel. As applied to cast iron, the synthetical method, by elucidating the influence of silicon, gave the key to the successful application of scientific knowledge in the iron-foundry. Subsequent experiments following similar lines, and particularly those conducted by Mr W. J. Keep, have determined with great accuracy and completeness the influence of the other elements which are present in foundry iron.

Returning now to the influence of silicon on cast iron, I have a series of samples of the turnings obtained from the test pieces produced in the original experiments, the results of which were, as before mentioned, published in 1885. These samples are arranged in order of silicon content. Those who have had any experience in the working of cast iron will see at once a marked difference in the properties of the various members of the series. Commencing at the one end with pure cast iron, the metal was so hard as to be cut only with the greatest difficulty, and the fragments chipped off in small irregular pieces. At the other end of the series the 10% silicon pig was so brittle that it could be crushed to powder in a steel mortar, though it was too hard to be readily cut in the lathe. The intermediate members of the series exhibited gradations in softness and working qualities, the samples from about 2 to 3.5% being particularly soft, and their turnings are in the shape of those twisted spirals which are so characteristic of metal which can be readily machined.

The series therefore illustrates the fact, which cannot be too often insisted upon, that cast iron with either too little or too much silicon is relatively hard and brittle; and that the proportion of

silicon must be varied according to the particular purpose which is

in view.

Tons per

FIG. 15.-Influence of Silicon on the Properties of Cast Iron.

Tensile.

The pieces from which these turnings were obtained were carefully

tested for tensile and crushing strength by Professor Kennedy.. They were also analysed, the analyses being checked by Mr J. P. Walton, and the results of these tests are embodied in the table given on page 60. These results may be conveniently expressed graphically as in fig. 15,1 from which it will be seen that for crushing, transverse, and tensile strength respectively, the influence exerted by silicon is of a uniform character, which may be represented by a curve rising to a maximum. But it will be observed that the maximum point is different in each case, thus emphasising the necessity for a composition in accordance with the special properties which are desired in the product. In these experiments the proportions of silicon corresponding to the various properties were as follows:

But it must be borne in mind that these values are only true under the circumstances of these particular experiments, and the most suitable amount of silicon varies according to the proportion of sulphur, manganese and other constituents, and also with the size of the casting. It has, however, been amply demonstrated in actual practice that the figures as above given are approximately correct. The hardness in this series of samples was determined by the means of a "sclerometer" (see fig. 39), the results being expressed by the weight in grams, pressing upon a diamond point, required to produce a standard scratch on the polished surface of the metal. The result is graphically shown in fig. 16, from which it will be noted that the maximum hardness was obtained with white iron; that the hardness steadily decreased with the addition of silicon until 2.5 per cent. was reached, when with further addition of silicon the hardness was again increased.

The influence of silicon on cast iron was reinvestigated by Mr W. J. Keep of Detroit, and the results of his investigation have since