This subject, of such important interest to all users of steam, has been ably discussed by Rowan, in a paper read before the British Association in 1876.

Crace Calvert's previous experiments showed that iron and steel were rapidly corroded when simply immersed in sea-water. Iron immersed in water containing carbonic acid oxidized rapidly; but when subjected to the action of dry oxygen or dry carbonic acid gas, there was no appreciable action on the surface of the metal. In damp oxygen, there was sufficient oxidation to just call for mention; in damp carbonic acid, there was a formation of white carbonate of iron; a mixture of the two gases in the dry state produced no effect; but rapid oxidation occurred in the presence of damp oxygen and damp carbonic acid. Calvert also found that distilled water, free from air and as pure as could be obtained, exercised little effect upon iron, though it dissolved lead to a small amount; distilled water, aerated, acted, however, in a rapid manner upon lead. Wagner found that the presence of the chlorides of ammonium, barium, calcium, magnesium, potassium, and sodium in the water, more or less largely increased the production of rust, and he found also that the corrosive action of these salts was considerably stimulated if the water held in solution air and carbonic acid.

Chloride of magnesium appears to be the most active agent of corrosion when alone, but combinations of the other chlorides are also very destructive, even when formed of those - e. g. barium, calcium, sodium - which singly have little or no corrosive action on iron at a temperature of 212° F. (100° C). The presence of an alkali, however, such as ordinary commercial potash or soda, appears to protect the iron from the action of the water even in the presence of air and carbonic acid. But the most destructive agent appears to be distilled water in the presence of air and carbonic acid.

Summed up, Rowan's paper amounts bo this: That it is the presence of air or gases in the distilled water, and also in the ordinary "pure " natural waters, which enables them to corrode the metal of the boilers in which they are used; but that when water is distilled free from air, it has practically lost all corroding power. This dictum strikes at the views of those engineers who suggest that the condensed steam should be aerated, and with equal strength at what is known as the aero-steam-engine, Warsop's principle, in which air is forced into the boiler.

Land boilers are affected by corrosion equally with those used on board ship; but in the former case remedies can be applied which in the latter case are impossible. In marine boilers are found a variety of corrosive actions depending to a great extent on the use of fresh water, with more or less (or none at all) sea water. The only marine boilers using fresh water exclusively are Rowan & Horton's and Perkins', though some of the ordinary boilers have been occasionally tried with fresh water, the result being a rapid corrosion. Although fresh water is "exclusively" used in the patent boilers mentioned, it is found practically impossible to prevent a small quantity of sea water finding its way into them; but, in the absence of analyses, it is unknown to what extent this leakage takes place. Perkins seeks to exclude all sea water and all oil or grease of any kind, and if these desiderata can be attained, the corrosion cannot be of any great importance; but still there will, Rowan thinks, be some, for the condensation of the steam takes place in the presence of a small amount of air, and some of it is sure to be mixed with the water forced into the boiler.

Stingl some time back made researches on the action of grease in steam boilers, and came to the conclusion that the fatty acids were potent factors in producing corrosion in steam boilers. Stingl not only demonstrated that the fatty acids were the cause of the rapid corrosion in a certain boiler, but by a process of purification and filtration of the same water he showed the exact amount of injury worked by them. In cases where grease reaches the boiler, it can also carry with it particles of other matters, and these, apparently by galvanic action, do mischief - a statement which, in spite of the incredulity of some engineers, Rowan considers is well established. In connection with some land boilers which used Glasgow water first, and afterwards Loch Katrine water, Rowan furnishes some interesting facts, from which it would seem that while " pure " natural waters act destructively upon boiler plates when used at starting a new boiler, they are innocuous when employed after the boiler has received a scale either from the previous use of a calcareous water or from lime pumped into it for the purpose.

Thus a boiler at Port Dundas, which had been worked with ordinary hard river water, continued to work for years when supplied with Loch Katrine water, without showing signs of corrosion; but a set of new boilers at the same works, which were fed with Loch Katrine water from the commencement, required new tubes in a very short time. Similar results have been experienced elsewhere. The boilers in question were working high-pressure engines, so that no condensed water was fed to them, and neither grease nor other corrosive agent was put into them. Hence the effects observed are attributed solely to the water. Now, in Loch Katrine water there is nothing that can account for the corrosive action, save that it contains 7 to 8 cub. in. of gas to the gallon (about 3 of which are oxygen). It is a remarkably pure natural water, almost perfectly soft, and containing no mineral matter with which the carbonic acid and oxygen can combine. The corrosion in the boilers in question was checked and prevented by feeding them each morning for a time with a wash of lime and water, for the lime, hardening on the plates from the heat, effectually preserved the iron from the effects of the pure water in the presence of carbonic acid and oxygen.