ment, such as the photographic recorder introduced and used with such brilliant results by the late Sir W. Roberts-Austen.1

Volume Changes due to Temperature.-To Mr Keep we are also indebted for much valuable information in relation to the changes which take place during the cooling of cast iron from the moment of solidification until it reaches the ordinary temperature of the foundry. In connection with these researches a special form of apparatus was employed for automatically recording the expansion or contraction which takes place during the cooling of a bar of cast iron after it has commenced to solidify. For this purpose the casting is made in sand, in a special box (see fig. 42), and two steel pins are placed in the mould at a fixed distance apart. As the metal cools these pins move inwards or outwards, and their motion, when

magnified, is communicated to the point of a pencil which moves on the surface of a sheet of paper fixed on a rotating cylinder. In this way autographic records can be readily obtained. In examining the record obtained from metals such as tin, lead, or copper, it will be found that the curve is a uniform one, due to the regular contraction of the metal from the temperature of solidification to that of the air (see fig. 43). With cast iron, however, the shape of the curve is quite different. White cast iron behaves more or less like the metals previously mentioned, the deviations from the regular form of the curve being relatively small; but with grey cast iron there are definite points of arrest in the shrinkage, and at these points actual expansion takes place. The character and amount of the expansion varies with the composition of the iron to be tested. Speaking

1 Introduction to Metallurgy, 5th edit., p. 194.

FIG. 44.-Cooling Curves of Cast Iron with different Proportions of Silicon.

generally, the greyer the iron, and the richer in silicon, the more pronounced is the expansion. In very grey iron there are usually three separate expansions noted. This is illustrated in fig. 44, in which are given a number of autographic records of iron with varying proportions of silicon. These expansions no doubt account for the well-known fact that soft iron does not contract during cooling as much as white or harder metal.1

A point about these experiments which is particularly interesting is that Mr Keep has definitely proved that the separation of graphite is not completed when the metal first assumes the solid state; but that such separation continues while the solid metal is cooling.

FIG. 45.-Cooling Curves from Grey Cast Iron. Showing three expansions, and diminution in combined carbon, on cooling from the point of solidification.

This was proved by casting a suitable number of samples side by side, and quenching them, one after the other, in cold water, at regular intervals of time, so as to ascertain the condition at varying stages during the cooling process. It was then found that the percentage of combined carbon in the series diminished from 0-60 before the first expansion, to 0.06 when the third expansion had taken place. This is illustrated in the special diagram given in fig. 45. The temperatures corresponding to these expansions were not determined, but the bar after the third, or last, expansion was still quite visibly red hot.

1 For a summary of these experiments, see Keep's Cast Iron (Chapman and Hall), a book which can be recommended to the careful perusal of every foundryman.

It may be mentioned that Mr O. F. Hudson has, at my suggestion, recently conducted some experiments with cast iron, and has observed that the cooling curve for No. 1 Northampton grey pig iron shows three distinct points of retardation at temperatures of about 1115, 955°, and 860° C. These curves were, by the kind

FIG. 46.-Pure Iron. Magnification, 75 diameters. Ferrite. The dark areas are due to reflection; they appear light on rotating the specimen.

permission of Dr T. K. Rose, taken in the laboratory of the Royal Mint.

The Micro-Structure of Cast Iron.

The application of the microscope to the testing of iron and steel is due to Sorby; later workers, such as Arnold, Osmond, Roberts-Austen, Stead, and Sauveur, have done much to extend and improve our knowledge of this important branch of metallurgy. The samples which are to be examined are first very carefully prepared by smoothing the surface with a fine file or emery wheel, polishing with very fine emery or rouge, and etching with nitric acid of 1-2 density,