of the cooling of a railway axle between the temperatures of 32° and -40° F., mentioned in the discussion on the same paper (M. E.,' 1891, p. 590, plate 123). There the rate of heat transmission has dropped to about 0018 evaporative unit. Information on this subject, and particularly on the radiating power of luminous and non-luminous flames, could be obtained by measuring their temperatures, their sizes, and the heat generated or fuel consumed. Hot flames will be found smaller than cold ones, and for the same temperatures their exposed surfaces should bear some relation to the rate of combustion. That the size of the flame affects the evaporation from the furnace plates is shown by Geoffroy's experiments. It did not increase in proportion to the fuel consumed, showing that there are limiting conditions; and it would appear that these are, that the heating power of a coke fire depends, not so much on the exposed heating surface as 'upon the grate surface, and that after a certain point has been reached, the heating power of a grate covered with coke is independent of the amount of fuel consumed. Secondly, block fuel, which gives off a luminous flame, radiates nearly 50% more heat in the fire box than coke does, indicating that the size of the flame, or perhaps its nearness to the plates, is an important factor. It will also be found, that with this fuel increased consumption causes increased evaporation in the fire box, showing that the flame had either increased in size or, on account of its higher temperature and density, had grown more effective. In these trials the consumption varied from 50 lbs. to over 100 lbs. of fuel per square foot of grate per hour, and the evaporation ranged from 20 lbs. to 40 lbs. of water per foot of fire-box heating surface per hour. It will be of interest to compare these values with those obtained by other experimenters. M. Hirsh (Soc. d'Enc.,' 1890, vol. v. p. 30) carried out two sets of experiments for determining the evaporation; in the first he bolted a small cylinder to the water side of the heating surface, directly over the fire bridge. It was sufficiently high to reach well above the water level, and was connected to a gauge glass by means of a small pipe which was also used for feeding purposes. When the boiler was at work this cylinder was constantly replenished with water, but while readings of the gauge glass were taken the feed was cut off. The coal consumption per square foot of grate varied from 17 lbs. to 53 lbs. per hour, and the evaporation varied from 26 lbs. to 62 lbs. per square foot per hour from and at 212° F. The best results were obtained when the grate was burning about 40 lbs. per hour, which shows that the experiments were not as reliable as could be wished; but if better arrangements were to be made for ensuring circulation, similar to those employed in Field tubes, it is possible that the absolute value of the various parts of boiler heating surfaces could be directly obtained. In Graham's experiments the evaporation did not exceed 20 lbs. per square foot per hour, and Stephenson found it to be 16 lbs. The experiments detailed by A. F. Yarrow (N. A.,' 1891, vol. xxxii. p. 108) can be used for determining the amount of heat transmitted through a plate. He measured the curvature of a tube plate which was covered with water and heated over a smith's fire. Its acquired 530 of its radius was 550 ins., i.e. one of its surfaces had expanded length more than the other side (assuming the plate to have been 1 in. thick). But this can only have been brought about by the fire side of the plate being 330° F. hotter than the other, and as experiments made on the flow of heat in iron bars (see table, p. 118) show that this difference of heat per inch of distance can only exist if sufficient heat is being transmitted to evaporate 142 lbs. from and at 212° F., this must have been the evaporation under these conditions. A. F. Yarrow also made similar comparative experiments on iron and copper plates placed over gas jets, and found that 57 lbs. per square foot were evaporated over the iron plates, and only 32 lbs. over the copper plates, which is in accordance with other experiments. (See pp. 120 and 131.) Mean Temperatures of Heating Surfaces. Another set of interesting experiments were carried out by J. Hirsh (Soc. d'Enc.,' 1890, vol. v. p. 302), and are mentioned in the second part of his paper. He constructed a small kettle about 10 ins. in diameter, which had an iron bottom about in. thick. Arrangements were made for keeping the water level at a constant height, while a strong gas and air blast was directed against the bottom, of which a small area of 4 ins. in diameter was exposed. Twenty-four holes were drilled into the bottom of this plate in. in diameter and in. deep, and these were filled with lead and tin alloys. In all 38 experiments were carried out, of which the results of a few are given in the following table. Sir John Durston ('N. A.,' 1893, vol. xxxiv. p. 133) has carried out similar experiments in open and closed vessels. See following tables : In the above experiments the fusible alloys were soldered to the fire sides of the plates; in the following experiments they were round or cubic plugs driven or placed in cavities in the plates. On account of the insufficient contact the recorded temperatures are probably too low. See Miss Bryant's paper (C. E.,' 1897, vol. cxxxii. p. 274) for remarks on the reliability of measuring temperatures by means of alloys. It is also well known that fusible plugs in boilers are not as reliable as they should be. 1500 430 363 2000 430 344.5 2000 510 358 2000 550 351 In order to compare these values with the estimated transmission in A. F. Yarrow's experiments, the excess temperatures above 212° must be divided by g, that being the ratio of the thicknesses of plates, and it will then be seen that in this case the heat transmitted is less, but grave doubts are entertained whether the indications of the alloys can be relied upon. Their outer surface is not iron, and they are not soldered to the bottoms of the holes, as they should be, and according to some supplementary experiments in which two thicknesses of plate were bolted together, the resistance to the passage of heat across the boundary of the two metals will have been very great, and the indicated temperatures must therefore be looked upon as excessive. This is particularly the case where the plate has been covered with scale, and it is of interest to note that when this is in. thick the temperature of the fire side of the iron plate has to be increased an extra 460° when evaporating 55 lbs. of water per hour. According to this experiment, scale offers about five times as much resistance to the passage of heat as iron does, whereas laboratory experiments show the ratio between iron and plaster of Paris to be as 1 to 100. This may be due to the thermal conductivity of iron having been measured along the fibre, while in these experiments the heat travelled across the plate and across the various layers of fine slag. Then, too, the boiler scale is in a saturated condition. A very important point to be noted is, that even a 3-in. plate can be heated to above the melting temperature of lead, if coated with a little scale on the water side, provided the fire is so hot that 55 lbs. of water are evaporated per square foot per hour, and it is therefore not unreasonable to suppose that most of the late troubles with the tube plates of Navy boilers are due to over-heating, particularly when it is remembered that it is probably not water, but moist steam which is in contact with these plates. The last set of experiments also show that an injurious effect is obtained by allowing grease to settle on heating surfaces. Somewhat similar experiments have recently been made by the late Dr. Kirk (Enging.,' 1892, vol. liv. p. 333). As in the above case, plugs of alloys were fitted into the bottom of a plate for determining its temperature. It was originally 23 ins. thick, and was gradually reduced to 12 in. As no measurements were taken of the water evaporated, the results are of less value than the above. Wye Williams ('N. A.,' 1894, vol. xxxv. p. 284) fitted thousands of studs into some furnace plates in order to increase the heating surface. They projected 3 ins. on both sides, but on the fire side they burnt away to 2 ins. As oxidation (blue) commences at about 500 F., the ends of these studs were doubtless heated to 1,000° F. or more and must have been effective as heat transmitters. Some patent boilers consist of tubes partly filled with water, sealed at both ends, and fitted like these studs, half in the fire, half in the water; the heat transmission is a more rapid one than by conductivity, but these boilers are not satisfactory. Transmission of Heat. The table on p. 130 contains the results of some interesting experiments, in which one side of the plate or tube was exposed to hot steam and the other side to the atmosphere. Other experiments will be found in the following publications:M. Burnat, C. E.,' 1874, vol. xli. Straw and similar coverings. W. J. Bird, N. Engl. T.,' 1879, vol. xxix. p. 7; 1881, vol. xxxi. p. 77; 1882, vol. xxxii. p. 35; 1885, vol. xxxv. p. 159; 1886, vol. xxxvii. p. 13. Steam losses in pipes exposed to atmospheric influences. J. J. Coleman, Phil. Soc.,' Glasgow, 1883, p. 73. H. Collins, Brit. Assoc.,' 1891, p. 780. K |