J. M. Ordway (Am. M. E.,' 1883, vol. v. p. 95, 1884, vol. vi.) gives relative heat transmission through 1 inch of 50 substances, but, as he changes his standard of comparison, it is difficult to analyse his results. 'C. E.,' 1891, vol. cviii. states that the loss of heat through covered steam pipes through which super-heated steam was passing is at the rate of 1° F. per 3 square ft. of surface. Comparing these values with those obtained for the transmission of heat through tube surfaces, it will be noticed that the external surface of a boiler would seem to be about ten times more efficient as a heat dissipater than the very much thinner tubes are as heat absorbers (see p. 125). This is of importance, because it may happen that a boiler is made so long that the last foot of tube length supplies less heat than is given away by the last foot of shell. Generally the entire circumference of the boiler shell is about one-fourth to onesixth of the sum of the circumferences of the boiler tubes, and if this boiler is unlagged there would be no advantage in allowing the gases to escape at any temperature less than about 400° F. above that of the water in the boiler, or say 750° F. If properly lagged the gases may be cooled much lower. In the above-mentioned experiments heat was supplied to the pipes or plates by steam, and the cooling took place in air. With the exception of Miss Bryant's experiments, no direct measurements have yet been carried out to ascertain whether the chief resistance is encountered on the hot or cold side. This could best be done by repeating A. F. Yarrow's experiments on the curvature of heated plates and estimating the mean temperature of the plate by means of its lineal expansion. Some light might be thrown on the subject by comparing the previous experiments with the following (W. S. Hutton, 1887, p. 253). The units of heat which a 4-in. plate will transmit per square foot per hour, if supplied with an unlimited amount of water on one side and steam on the other, are there given : An interesting feature of W. S. Hutton's figures is that they show that the transmission of heat is less dependent on the conductivity of the material than on its other properties. Thus tiles and glass plates Notes to Table on p. 130 Isherwood, Experimental Researches, vol. ii. 2 Franklin Inst., 1878, iii. vol. lxxv. p. 153. 3 W. Meunier, Rev. Ind., 1884. Ibid., Soc. I. Mul., 1879, p. 730. 5 E. Deny, ibid. 1883, vol. liii. p. 575, and 1884, vol. Iv. N. N., C. E., 1895, vol. cxxi. p. 300. p. 15. are nearly as good as cast iron, while copper and tin plates are the least efficient. Smoke-box Radiation.-Marine boiler smoke-boxes are usually fitted with baffle plates, one inside and one out. The only advantage they offer is cheapness combined with a reduction of radiated heat; but heat lost by convection must be greater than with a plain plate, for a natural and very powerful current of air or hot gases is induced between the plates. To reduce the loss of heat from smoke-boxes the space between the baffle plates should be filled with non-conducting material, or if this is too heavy or too costly, the edges of the baffle plates should be flanged so as to press against the smoke-box plates. Only one baffle plate will then be needed-an inside one. If two are fitted, they should both be flanged and both fitted on the hottest side, so as to protect the casing and door from heat and thus prevent its warping. With water-tube boilers loss of heat, warping of the casing, and then leakage of air make themselves seriously felt. From the foregoing it is evident that our knowledge of the transmission of heat is very limited, and possibly incorrect; nevertheless the collection of previous experiments into one chapter has the advantage of showing in what direction further information should be sought, and the following experiments readily suggest themselves : : 1st. Experiments on the flexure of plates which are being heated on one side either by water, air, or radiant heat, and are being cooled on the other side by any of these methods. 2nd. A repetition of M. Geoffroy's experiments on a subdivided boiler, with accurate measurements of the temperatures of the waste products of combustion at various points, and also analyses of the gases, and calorific determinations of the fuels. Temperature of Fire Bars. The parts of the boiler which suffer most from the effects of slowness of transmission of heat are the fire bars. They are exposed to an intense heat at their upper surface, to radiation on part of their side surface, and the only available means of cooling them is by the air which passes over their sides. When the fires are thick, when much air is admitted above the bars, and when the draught is forced, the bars are naturally exposed to the very serious danger of overheating. Assuming that the heat received by the upper surface of the bars is at the rate of 50,000 thermal units per hour per square foot, and assuming that this heat is being transmitted to the passing air at the high rate of ten units for every degree of difference of temperature, then if a is the width of the air space, h the depth of the fire bar, b its breadth, and t its excess temperature over that of air, we have In order to keep the temperature t below 1,000° F. it would be necessary to make h = 2. b + 2 a, which is perhaps fairly near the truth; but this leaves out of account the quantity of air, which is certainly an important factor. With ordinary draught the air supply is at the rate of about 400 lbs. per hour per square foot of grate, or about 7 lbs. per minute, which is certainly not a large quantity. Comparing the temperatures t, and t2 in two different sets of fire bars in which the various dimensions are a1 and a, b, and b1⁄2, h, and h2, and the air supply Q, and Q2, we have This would show that the more work a grate has got to do the deeper ought the bars to be made, and the greater should be their number. But as it is at first not so much the danger of burning the bars as of their being bent which has to be guarded against (see p. 10), making them very thin without reducing their length would increase this trouble. Until their temperatures have been taken under varying conditions, which is not difficult, it is idle to speculate any further on their behaviour, and at any rate the above formula should be looked upon more as an indication as to how much information is yet wanted than as a practical guide for new departures. 134 CHAPTER VI STRENGTH OF MATERIALS In this country, especially during the last twenty years, steel has almost entirely supplanted wrought iron for ship and boiler construction. No further excuse is therefore needed for discussing its peculiarities more exhaustively than those of the older metal. Besides, the two materials are so intimately related, that the large amount of research expended on one must assist in explaining the other. Wrought Iron. It is usually stated that steel is a homogeneous material, while wrought iron is a conglomerate of crystals, granules, or fibres cemented together by films of slag. Layers and threads of slag undoubtedly exist in iron, but it is unreasonable to suppose that they are as uniformly distributed as the above theory would make it appear. The manner of the production of wrought iron in the puddling furnace seems to be, that drops of molten pig are slowly converted into plastic wrought iron, their centres always retaining a slight excess of carbon and other impurities, while the flame and slag acting on their outer surfaces are converting them into almost pure iron, possessing, amongst other qualities, that of welding with the greatest ease. That these numerous drops of white-hot metal, which may now be called granules, or even lumps, should stick together is but natural, and the trouble to be expected-and it does exist is that numerous cavities filled with slag will come into existence. When drawn out into bars or plates wrought iron would therefore consist of numerous fibres whose individual outer surfaces are very pure and soft, and whose cores contain the small percentage of carbon which had been allowed to remain. Professor Wedding has shown how by the microscope we can distinguish between the hard and soft parts, for he found that the greater the percentage of carbon, the darker the colour if the iron or steel is raised to a blue heat; and by carefully polishing and etching samples of iron, and then heating them sufficiently to make them appear of a uniform dark straw colour, the microscope will show that this uniformity is an illusion, and that the metal really consists of innumerable cells of soft iron surrounding hard cores. Influence of Producing Temperatures.-Had the temperature of the puddling furnace been as great as that in any of the steel furnaces, each granule would have been melted, and the product would have been mild steel of the same tenacity as the puddled iron, but it is well known that the first of these metals requires quite twice as much horse-power for machining it as iron of the same tenacity, and there does not seem to be a better explanation than that this is due to the somewhat higher temperature at which it has been produced. Steels from various makers, but of the same tenacity, are said to show great differences as regards the power required to chip them. An explanation for this might be sought for in variations in the casting temperatures, and these again would depend on the firebrick lining used. J. W. Cabat ( American Inst. Mining Engs.,' vol. xiv. p. 85) deals with the influence of casting temperatures, and remarks that coldblown charges of rail steel work better than hot-blown ones, and that open-hearth spring steel is similarly affected by the furnace temperature. The Basic Bessemer Steel Process, using Phosphoric Pig Iron, and usually called the Thomas Gilchrist Process.-The pig is run into the converter in a molten state, and subjected from below to the action of a strong blast, which first removes the carbon and silicon, and then attacks the phosphorus. It was at one time believed that the development of heat in all the Bessemer processes was due to the burning of the carbon contained in the pig, but this has been disproved. Dr. F. C. G. Müller (Deut. Ing.,' 1878, vol. vi. p. 387) mentions that silicon burns away first, and that for every per cent. consumed the temperature of the molten metal is raised 540° F.; that the carbon is not consumed until a temperature of 2,550° F. has been reached, and that its burning does not raise it. Phosphorus, like silicon, adds much heat to the bath, but will not burn until practically all the silicon and carbon have been consumed. It has also been found that the phosphorus cannot be consumed unless lime is present in the converter, and as this would attack and melt the ganister lining, which is nearly pure silicic acid, it is necessary to give the converters a basic lining (dolomite), from which the steel made by this process derives. its name of basic steel.' Only very little silicon may be tolerated in the pig intended for this process, as it attacks the dolomite, and in order to obtain a sufficiently high temperature, which cannot be done by burning carbon, phosphorus must be present in large proportions, viz. from 2 to 3%. In some pigs phosphorus is not sufficiently plentiful, and then, in order to obtain the right heat, it has to be melted in Siemens-Martin furnaces instead of cupolas. During the process of manufacture the blast is kept up until all the phosphorus is consumed, and the metal in the converter is then almost absolutely pure iron. Spiegel and ferro-manganese are now added in the right proportions, and the charge is ready for casting. At present the only available means for judging of the purity of the molten steel is to count the number of revolutions of the blowing engines from the time when the carbon lines in the spectrum have disappeared, but, in spite of assertions to the contrary, it does not appear that sufficient reliance can be placed on this proceeding, which is as follows: The composition of the pig iron is ascertained from samples if bought, or from daily returns if run direct from the blast furnace. The number of revolutions of the blowing engine required for supplying all the air necessary to consume all the carbon and silicon is estimated, and also how many extra revolutions will be required to remove the |