made use of a large magnet constructed with exceptional care by H. Freiberg, out of "Eibiswalder naturhartem Wolfram-Stahl." Its dimensions (rectangular parallelopipedon) and weight are these: length, 29 centimeters; breadth, 3.6 centimeters; thickness, 1.2 centimeters; weight, 1,030 grams. This magnet was carefully tempered by heating to low redness and sudden cooling in lime-water at 20°. After being magnetized to saturation between the poles of a powerful electro-magnet, it was kept at 100° for 35 hours; thereupon again magnetized and exposed to steam for 10 hours more. The following are the observed magnetic moments (C. G. S.) at 200:

On June 21, 1883, at-200, Wild obtained for the same quantity, 28.8 × 103. The magnet, despite the treatment which it had experienced, therefore, still lost about 1 per cent. of its total moment, during the intervening three months. But this variation is little more than 0.0001 of the total intensity per day, if the loss were proportional to time. Referred to July 1, however, when the actual measurements were commenced, the said decrement cannot be estimated as above 0.00005 per day-particularly so if it be borne in mind that the diminution in question must gradually vanish at a continually decreasing rate through infinite time, or that a final and definite moment is being asymptotically approached. Indeed, the said limit was practically reached in the second third of July, as the following results show. Wild further remarks that final magnetization of 28 C.G.S. units of magnetism per gramme is to be considered as a satisfactory value. The results in question are as follows:

We venture to remark that with so unusually large a magnet even better results could have been obtained by repeated boiling and remag netization. In consideration of the dimensions, the decrement of 1 per cent. is by no means surprising. The tubular magnet discussed in the above was operated upon three successive times by our method, before definite adjustment for absolute work. We are not aware, however, hether it will not take some time, in order that a magnet kept at 1000 thoroughly regain the state of molecular equilibrium for 20°.

CHAPTER VII.

A PHYSICAL DEFINITION OF STEEL BASED ON THE ELECTRIC BEHAVIOR OF IRON WITH GRADUALLY INCREASING DEGREES OF CARBURATION.

INTRODUCTION.

Nature of the problem.-A detailed and thoroughly exhaustive study of the problem in hand, viz, in how far the effect of carburation on the galvanic and thermo-electric properties of iron is available for the general classification of iron-carburets, calls for working facilities at a puddling furnace, for instance, or other technically satisfactory agency for the decarburation or carburation of iron. Possibly, however, similar work might be done on a small scale in the laboratory, if the necessarily complicated apparatus or opportunities for constructing the same were at the observer's disposal. These advantages were not within our

reach.

The problem is of a kind, moreover, which is apt to mislead the investigator into insuperable and almost infinite complications. To avoid these it is absolutely necessary to conduct the experiments with reference to some thoroughly preorganized plan. In the absence of the above-mentioned metallurgical apparatus we were obliged to content ourselves with commercial products, and the main purpose of the present memoir has therefore been restricted to the development of a plan or scheme of operations for the general and tentative study of the electrical behavior of iron-carburets. These efforts have not been unsuccessful; indeed they appear to be of considerable promise. They have already afforded us a method for the physically exact definition of steel which we regard as important. As a whole, the present chapter furnishes an essential and interpretative sequel to our researches on the hardness of steel.

Electrical manifestation of mechanical properties.-The very remarkable effect of rapid and of prolonged cooling from red heat, respectively, on the physical and chemical properties of iron-carburets has always been a subject of great metallurgical interest.147 In the case of steel the contrast between the two states or conditions thus produced is particularly well marked and of the greatest practical value. Experience has shown, however, that the said processes may be applied, with much advantage, to most of the other iron-carbon products. It is thus that

147 On Karsten's theory, relative to the nature of these effects, see Percy's Metallurgy, edited by Wedding and others, Vol. II, p. 167, et seq., Braunschweig, 1864.

the question naturally suggested itself to us, whether the remarkable parallelism discovered in the variation of the degree of hardness of steel and its galvanic and thermo-electric properties was not to be considered as only a special case of the behavior of iron-carburets gener ally, under analogous circumstances. In this respect we believed ourselves justified in predicting that those characteristic mechanical qualities which distinguish steel from other iron-carburets must necessarily be sharply outlined in a general electrical diagram adapted to the classifi cation of iron carburets as a whole. A similar idea, as we subsequently found, seems incidentally to have occurred to Joule, 148 since he remarks, after having given the necessary experimental data: "I believe the excellence of the latter metal (steel) might be tested by ascertaining the amount of change in thermo-electric condition which can be produced by the process of hardening." But neither Joule nor others have given the subject more than this inadequate consideration, and it was not until our investigations on steel had been fully developed that the problem attracted our attention.

Critical operations.-At first sight a comparison of the electrical intervals comprehended between the hard tempered, and soft annealed states, appeared to be rich in promise; but the processes of slow and of most rapid cooling possible, from red heat, are as yet not sufficiently defined, even if we abstract from decarburation, etc., for obtaining iron carburets in two characteristic physical states. In the case of slow cooling the temperature in red heat to which the specimen has been exposed, as well as the time during which exposure takes place are important items, particularly when the cast-irons are under examination. In the case of rapid cooling the temperature to which the redhot rod is suddenly and permanently lowered is additionally to be considered. It would not, however, be difficult to define the two processes in question succinctly. A rod, for instance, suddenly chilled from red heat in water at ordinary temperature and then annealed by long expos ure in ether vapor at 350 might appropriately be termed glass-hard; if annealed in vapor of sulphur (500°) or of cadmium (700°), soft. For the very large and physically important class of iron-carburets, wroughtiron, low-carbureted steel, and steel, these details, fortunately, do not produce any serious distortions; the thermo-electric hardness and the specific resistance of steel, no matter what the process may have been by which a given rod was softened, remain very nearly constant in value-at least when compared with the enormous range of variation of these qualities due to tempering. The same is true for the hard condition of the carburets between iron and steel, where it is only necessary to choose the temperature before sudden cooling sufficiently high to insure the appearance of hardness, and to chill in water at ordinary

148 Joule: Phil. Trans. 1859, I, p. 96.

roo:n-temperature. Decarburation is, however, under all circumstances. to be avoided,149 and the exposures to high temperature must not be prolonged. It follows therefore that it will be expedient to commence the present investigation by a consideration of the electrical properties of the carburets in question, that is such in which the total carbon is less than about 2 per cent. To this may be added that within the in-. terval (0-2 per cent.) those modifications in the mode of occurrence of the carbon in iron, which are the cause of such great diversity in the character of the different species of cast iron, are as yet comparatively without marked influence. Thus it appears that our results for this set of products may be considered as satisfactory and definite. In order to complete the discussion conveniently, however, it is desirable to include certain essential properties of the cast-irons, or in other words to prolong the loci of our diagrams into the region of cast-iron, without going into any details. That this is readily possible will appear in the sequel. A further introductory remark may be added here. Commercial ironcarburets are never pure, but contain in greater or smaller amounts vitiating impurities like phosphorus, sulphur, silicon, and the like. Each of these produces its own electrical effect, as has been seen in the earlier chapter (III) on alloys. The discrepancy thus introduced need not by any means be negligible, and full consideration is given to it in a later paragraph. For the present it will be expedient to suppose this secondary electrical effect to be absent, or that the material in hand is a pure iron-carbon product.

Nomenclature. As a convenient nomenclature to be used throughout, we will designate the process of softening steel, that is cooling from red heat as slowly as necessary by the Roman numeral "I"; the process of sudden cooling from the same temperature (hardening, tempering, glass-hard) by the Roman numeral "II." In like manner all constants which refer to I or II, are to be marked with the subscript 1 or 2, respectively. For instance, h1, 81... h2, 82. In like manner we may, without confusion, consider "soft state" and "glass-hard state," "state I" and "state II," respectively identical, etc.

149 In this place it is well to call to mind an important result of Forquignon's (Ann. de Chim et de Phys. (5) XXIII, p. 538, 1881). He found that steel kept at red heat for seventy-two hours, in an envelope of hematite, lost nearly one-half of its total carbon. This corresponds to the loss of nearly four-fifths of total carbon, due to continued exposure of cast iron to red heat in the preparation of malleable cast iron (Percy op. cit., p. 143.) In many physical experiments with steel, particularly in magnetic work, it is often necessary to soften steel by heating it to redness for some time. It is therefore obvious that the product thus obtained cannot be considered identical, as regards total carburation, with the original carburet. Possibly this method of eliminating carbon may be available for obtaining points in an electric diagram corresponding to ironcarburets lying between iron and steel.

(767)

WROUGHT-IRON.

Electrical data.-Pure iron subjected to the process II is mechanically indistinguishable from the same metal when subjected to I. So also the electrical difference between these two states is practically insignificant. This is true approximately for commercial iron with less than 0.2 per cent. of carbon, and the more nearly true, moreover, the smaller the amount of this element.

For the electrical conductivity of iron Chwolson150 gives the following results: If a hard-drawn wire is ignited at low redness its electrical resistance is found to vary about -0.4 per cent. If the ignition be intense about +5.3 per cent. The process II produces a variation of only 0.7 per eent. in comparison with the hard-drawn state. If, therefore, we compare the rod in the state II with the same rod in the state I, we find a total electrical change about +1.1 per cent. or -4.6 per cent., respectively, according as I was produced by gentle or by intense ignition.

For results of this kind, with reference to the thermo-electric behavior of iron, we searched in vain. But the relation between the variations of thermo-electric and galvanic constants is initially (i. e. for very small amounts of a foreign element alloyed to any given metal) linear, as we proved both in the case of steel and of alloys of silver. Hence results of the same order as Chwolson's may be at once predicted for the thermoelectric behavior of commercial iron. Sir William Thomson151 found that the thermo-electric hardness152 of iron, like that of steel, is increased by the process II. Joule153 finally remarks, "I find that in steel the (thermo-electric) change is in the same direction as in iron, but of enormously greater magnitude."

These small variations amounting to less than 2 per cent. of the total resistance or thermo-electric hardness, are for the present purposes at least, quite negligible; particularly so when contrasted with the corresponding change of the electrical constants of steel (200-300 per cent.). Where great accuracy is sought for, special measurements may be made. For this reason we accept for wrought iron the values for the electrical constants given in the following table (76). Here thermoelectric hardness is represented by h, specific resistance by 8; h1 refers to iron subjected to process I; h2 to the same wire subjected to process II, etc., as has been stated. Furthermore 4h-h2-h1; 48=82-81; Alog

150 Chwolson: Bulet. de St. Petersb., X, p. 379, 1877; also Carl's Rep., XIV, p. 26, 1878. 151 Thomson: Phil. Trans., 1856, III, p. 722.

152 On the definition of thermo-electric hardness, see our paper in Wied. Ann., XI, p. 970,1880, or this memoir, Chapter II, p. 65.

153 Joule: Phil. Trans., 1859, I, p. 95-97.