if there are any powerful chemicals in the boiler water, either acids or alkalies, or even neutral salts. Unless copper pipes are fitted inside the boiler, the only sources of supply are the condenser tubes, the feed pipes, and the pumps. Here again it is the use of vegetable and animal lubricants which causes the mischief, for both readily attack copper or brass, as can be seen by their greenish colour if either has been in contact with these metals. Distilled water also seems to be a solvent; it certainly attacks lead. It is also believed that the minute particles which are worn off the working parts of the pumps are carried bodily into the boiler, and J. MacFarlane Gray (N. A.,' 1861, vol. ii. p. 157) has detected specks at the bottom of pit holes, and attributes pitting to that cause; and Professor Lewes (N. A.,' 1889, vol. xxx. p. 340) also believes that the presence of copper causes pitting. The author's observations do not support either view, for although he has been very careful to observe green scale patches, he never could detect signs of pitting near them, nor did the corrosion seem in any way to be increased at these points. These views are supported by Mr. H. W. Hirman's statement, in answer to Mr. MacFarlane Gray (see above), that land boilers using town water pitted, although here there could be no question of the presence of copper. Additional remarks on galvanic action will be found further on. Zinc Salts. There remains one set of salts which possibly play an important part in boilers of the present day, but of which no mention is made in any books or papers, and the following remarks are therefore purely speculative. It is a well-known fact that if boiler water is often renewed the reduction of the zinc slabs is more rapid than if the same water is used over and over again. A very obvious explanation would be, that the more zinc salts are dissolved in the water, the less corrosive it is. If this view is correct, the painting of the insides of boilers with zinc oxide, and the addition of some zinc salts-the chloride of course excepted-to the boiler water ought to have a beneficial effect, and recent limited experiences show this to be the case. Neutral Salts. The following experiments throw a little light on the part played by what would appear to be perfectly harmless salts in increasing the corrosive power of acids (Journal of the Camera Club,' London, 1892, vol. vi. p. 52). M. Gourdon's experiments in 1873 show that exceedingly diluted sulphuric acid could still be made to attack zinc metal by adding to it various salts. On the same page L. Warnerke gives a table showing that by adding various salts to a one per cent. solution of sulphuric acid, the speed with which it attacks zinc metal can be varied considerably. The amount of corrosion is given in decimals of millimeters of depth per hour. The interesting point about these results is, that both lead and copper salts stand very low down in the two lists, showing that the accepted notion, that they are very injurious, is, in one sense at least, a wrong one; and the bare fact that these various salts can influence the behaviour of sulphuric acid makes it appear probable that zinc salts also have an influence on the action of corrosive substances found in a boiler-apparently a beneficial one. Possibly experiments carried out on the above lines might show that manganese salts, which enter the boiler water as the steel plates corrode, act injuriously, and this might explain why the presence of this metal in steel has been looked upon as increasing its corrodibility. Alkalies. Sea water contains much free carbonic acid, and this has the power of making the lime salts soluble, but when treated with weak acids the carbonic acid disappears, and the alkalies which are thereby liberated constitute the so-called alkalinity of sea water. As long, however, as these alkalies are accompanied by carbonic acid they may be considered as non-existent, at least as far as their protective influence goes. III. Air in Hot Water. The question of air absorption by water is fully dealt with on p. 60. It is here only necessary to recall that, contrary to the generally accepted view, water on being boiled does not part with all its air; if, therefore, air is admitted into the boiler it will be found in the water. Of course, as the amount of air in the steam space will be small the effective air pressure will also be relatively small. The action of the air, which generally accompanies feed water, is discussed in the papers which have been already mentioned. A. Wagner found that pure water will not attack iron, except in the presence of air; that the action is more severe if carbonic acid is present, and still more so if the chlorides of magnesium, ammonium, sodium, potassium, barium, or calcium are present, the magnesic chloride being the most powerful. These views are corroborated by most of the other writers, but do not seem to be quite correct. The Absorption of Air by a fluid is stated to take place as follows: The volume of gas which a definite quantity of fluid can absorb is indeMercury salts are known to be efficient protectors against rust. pendent of the pressure, and decreases with rising temperature. Therefore, as 1,000 cubic ins. of water of 0° C. will absorb 48.9 cubic ins. of oxygen at atmospheric pressure, they will also absorb an equal volume at two atmospheres, and so on. Of course, as the pressure is increased the density of the oxygen is increased, so that the weight of absorbed gas is proportional to the pressure. This law also holds good for a mixture of gases. As an example, estimate the quantity of air which pure cold water will absorb. The coefficiencies are 0489 for oxygen, 0235 for nitrogen. Their respective densities are as 16 to 14. And as there are 20% of oxygen and 80% of nitrogen in the air, the densities of these two gases are as and respectively of what they would be if pure. 1,000 cubic ins. of water will therefore absorb 48.9 cubic ins. of oxygen, having a density of 16, and 23.5 cubic ins. of nitrogen, having a density of × 14. The relative weights of the absorbed gases would therefore be as 48.9 × } × 16=1565 to 23.5 × ×14=263·0, or as 60 to 100. In the atmosphere their relative weights are as 286 to 1000. This shows that by compressing air in presence of water, and then allowing it to escape again, a mixture would be obtained in which there is about 100 per cent. more oxygen than in the atmosphere. Practical difficulties stand in the way of utilising this action commercially. Air contained in Hot Boiler Water.-A similar calculation will show how much oxygen is contained in boiler water, but taking into account that the temperature is high, that the coefficient of absorption is smaller than when cold, that the air pressure in the steam space is small, and that the density is also less, there can be little doubt that only traces of oxygen are to be found in the hot water of a boiler. In the same way that silicon and aluminium expel air out of molten steel, some salts may be expected to drive it out of water, but little is known on this subject (see p. 140, also M. Lodin, ' Comp. Rend.,' 1880, vol. xci. p. 217). Air in Feed Water.-The conditions under which air and water meet in the feed pumps and their air chambers are totally different from the above. There the temperature is low, and the pressure of the steam vapour so slight that it can be neglected. During each delivery stroke the pressure on the water is raised to 150 lbs., which would cause 8% volume of oxygen and 16% of nitrogen to be absorbed, i.e. 3-2 lbs. of oxygen per ton of cold feed water. That there is nothing improbable about this is clearly demonstrated by the fact that Sir W. Thomson's (Lord Kelvin) sounding apparatus cannot be used for great depths, because at 200 fathoms (= 40 atmospheres) all the air in the glass tube will be absorbed by the water which has entered it. Oxygen and Carbonic Acid.-It has already been mentioned that oxygen alone, or even when in company of neutral salts, will not attack iron. Carbonic acid, like all acids, will attack iron, forming salts. But most of these salts are active absorbers of oxygen, and thereby change into ferric salts, but these require an extra equipment of acid like the sulphate, or else are too weak to hold any acid. For instance, there is a ferrous carbonate, but no ferric carbonate. In both cases the result is rust, consisting chiefly of ferric hydrate. But ferric hydrate (brown rust) easily parts with some of its oxygen to any iron with which it may be in contact, whereby it is reduced to ferrous hydrate (black rust). When both carbonic acid and oxygen are present in boiler water the carbonic acid first dissolves iron and liberates hydrogen; the oxygen then reduces the ferrous salt to ferric hydrate, and liberates the carbonic acid, which again attacks the iron. At the same time, the red rust gives up part of its oxygen to adjoining iron, and both rusts are then black. Thus every pound of carbonic acid present in the water will corrode 1:35 pounds of iron, and if in addition oxygen be present each pound will corrode about 3-7 pounds. This latter quantity is independent of the amount of carbonic acid present, except in so far as that when there is only a trace of this gas the process of chemical interchange will be so slow that a large proportion of the oxygen may escape with the steam without doing harm. Red or Black Mud.-In new boilers one frequently notices much black or dark red mud, and the plates may appear to be rough or even covered with pit holes leading one to believe that the boiler is corroding fast. The true explanation will generally be found to be that as in modern boilers the pressure and temperature are high, water acquires the power of dissolving the silica in the mill scale on the plates; this slag contains ferric and ferrous oxide of iron in a finely divided state which sink to the bottom of the boiler. The roughness of the exposed iron plate is therefore a natural condition brought about in the rolling mill and only shown up after the glossy scale has disappeared. Position of Feed Discharge. Having admitted the saturated feed into the boiler, there are three ways of dealing with it : 1. Either lead it so that it enters the bottom of the boiler or easily falls there. 2. Admit it at some point in the boiler- say, over the back end of the tubes, so that it gets thoroughly mixed with the hot water and loses its air. 3. Lead it through pipes which are fixed inside the boiler, in order that it may become sufficiently heated to part with all its air. All these plans are in use. In the case of No. 3 the internal pipes, if made of iron, suffer very seriously from pitting. If made of copper they also suffer; and in such cases much green scale will be found in the boiler. No. 2 is the general practice. Few engineers have had the courage to discharge into the steam space, but where tried the plan works satisfactorily, and ought to assist the water circulation, on account of 20% more steam being generated and condensed. Boilers liable to prime could not be worked in this way, as was mentioned by J. H. Hallett (M. E.,' 1884, p. 350); but at any rate the feed should then be introduced at a point where it will be heated as quickly as possible. No plan could be worse than to discharge the feed at the bottom of a boiler, for it is clear that, in comparison with other parts, the spaces under and between the furnaces can have only a very restricted circulation. The defect is increased if they are filled with cold, and therefore heavy, water (see fig. 93: the dark spaces at the bottom represent cold water); then, as the fire bars, the ashes, and the inrushing cold air all prevent heat from reaching the bottoms of the furnaces, the water has absolutely no chance of rising except by the very slow process of being fed from below. This speed is about 5 ft. per hour, or 1 in. per minute. There is, therefore, no difficulty in heating all the feed slowly and along one particular zone, viz. along the line of the fire bars. Under these conditions the straining to which both the shell and furnaces, but particularly the latter, are subjected must be excessive, for not only is there a sudden jump from cold air below the grate to white heat above them, but on the water side there is also a sudden rise from about 100° F. to over 350° F. Pitting. It is, however, not with stresses and circulation that we are at present dealing, but with corrosion, due to air absorbed by the feed water. What takes place with this air is shown in fig. 94. of the boiler, and if it is satu- COLD FIG. 94 cubic inches of air per hour. This is equal to about one bubble in. diameter per second. The excessive differences of temperature along this line of grate, and the consequent excessive straining of |