ends of the boiler are covered with plating, and, in order to make the casing quite smoke-tight, the outer wall of tubes is also covered with light plating, but this has not to resist any great heat.

The first water-tube boiler put into a torpedo-boat by the Author's firm afforded a very satisfactory means of comparison between the new boiler and its locomotive rival, which had been placed in a sister vessel.

The result of steaming was eminently satisfactory, and the saving in fuel at equal speeds was sufficiently evident without exact experiment, the boat under natural draught being about 1 knot an hour faster than the other vessel, the full-power trials showing also a difference of 0.67 knot speed in favour of the former. Some evaporative trials were made by the Portsmouth authorities, and the results seemed to indicate that equal duty could be obtained when the proportionate quantity of water evaporated was 2·36 from the water-tube boiler to 1.00 from the locomotive boiler. This boat has been at work for three years, and, with a view to ascertaining the state of the heating-tubes in the boiler, several tubes have been taken out. They afford a sample of tubes under varying conditions. Some were taken from the firebox, where they are exposed to the full intensity of the flame and radiant heat, some were from the flue, and some from the outer layer, which are exposed to more gentle heat through only one-half of their circumference. The condition of these tubes when cut open was found to be very satisfactory.

The small amount of scale in their interior is an important feature, and is a great contrast to the water-tubes taken from the boiler of the “ 'Propontis" where the circulation was not so well provided for. The original thickness of the tubes was 15 B.W.G., and they have suffered no perceptible diminution. The interior surface is for the most part in excellent condition, but pitting does seem to have commenced slightly in the upper part of some of the tubes, though not sufficiently to show any reduction of thickness where a pit has been cut through in dividing the tube. It should be observed, in connection with this pitting, that zinc blocks were not put into the boiler until it had been working for about ten months.

A few of the outer tubes are much corroded on their external surface, but it is worthy of particular note that these are the tubes least exposed to the action of the fire, and that the damage is due to some cause independent of the stress to which the boiler is subjected while at work. It is, in fact, caused by the outside of the tubes having been wet and not to the action of the fire. The

boiler shown in Plate 1, Fig. 8, has been at work for five years, and it continues in excellent condition, no leak having appeared in any of the tubes, which are of steel. In the case, however, of some boilers built much more recently there has been trouble from pitting, and also from corrosion of the outer surface.

The pitting may be due to the presence of globules of lead left in the tubes from the process of bending. The uncertainty which arises from the use of iron or steel seems to make it desirable to employ brass or copper for these tubes, and experiments have been made with both these materials. Brass has given decidedly superior results; the surface of the copper tubes, when exposed to intense heat, had become somewhat rough, showing that the metal had suffered some loss; while the brass tubes retained a surface so smooth that it was not conceivable that any appreciable amount of metal was gone.

It has been suggested that brass tubes would suffer in the part of the boiler above the level of water-mark in the separator; but the most careful examination has failed to discover any sign of injury at this part, although in the experimental boiler, which was not designed for economy of fuel, the products of combustion were at a very high temperature even above the highest part of the generating tubes. The result of the experiments has been to induce the Author's firm to use brass for these tubes in all the boilers now under construction. The highest pressure hitherto used in these boilers has been 250 lbs. per square inch.

Ordinary steam-pipe joints are not suitable for this high pressure, and the Author is indebted to Mr. Perkins for a very concise description of a suitable joint. He said, "It must be metal, and it must be narrow."

In making a steam-tight joint with metal, two courses are open. The one is to use accurately fitted surfaces of considerable extent on flanges of great rigidity, or, if not of great rigidity, so secured that their elasticity has an equal effect throughout their circumference. The accurately fitted joint is necessarily expensive. The other method is to use a narrow metal joint upon which the intensity of pressure can be made so great that a continuous line of contact may be produced independent of any small irregularities. Fig. 2, p. 55, shows the form of joint employed.

With 200 lbs. pressure per square inch used with some boilers previously constructed, no great difficulty was found with the water-glasses, but the addition of another 50 lbs. seems to overtax them. This difficulty was overcome by the substitution of talc for glass, as had already been done by Mr. Perkins.

Experiments on evaporation were made with one of the larger boilers by measuring the feed-water from tanks and weighing the coal consumed, the steam produced being allowed to escape from the safety-valves. The duty obtained appeared to be very high, and clearly indicated that more carefully conducted experiments were necessary to give sufficient weight of evidence to establish the truth of so unusual a performance. Experiments have since been made by Professor Kennedy, which although not quite bearing out the first trial very nearly do so. It is well known that in a torpedo-boat, where engines of considerable Fig. 2.

Copper ring with
ands brazed together

STEAM JOINTS FOR WATER-TUBE BOILER. WORKING PRESSURE 250 LBS.

power are compressed into very little space, there is no room to spare, and the Author was by no means sanguine that it was possible to make measurements. These, however, have been successfully accomplished by Professor Kennedy, enabling him to give a balance showing what was spent and how it was expended. The Author wishes to point out that, if this can be done for powers of over 700 HP. on a torpedo-boat, the difficulties are much less on larger vessels; and that trials of this kind may be made with advantage on board some of the vessels in Her Majesty's Navy.

The measurement of water returned from the engines to the boiler presented the greatest difficulties, the space available for the temporary tanks used for this purpose being so small that they were

filled in a very short time; but they were so ingeniously arranged, with conical tops and bases, that any considerable error in measurement was not possible. In the first experiments, the feed-pumps took their supply directly from the measuring-tanks. This, however, caused inconvenience in the stoke-hold; for the limited water-space in the boiler renders it specially sensitive to failure in water-supply, and this arrangement precluded any proper control. Also, when working at small power, the water was sent into the boiler much too rapidly, causing a fall in the steam-pressure, and then, while the other tank was filling, there was a long pause when no water could be obtained. The difficulty was overcome by putting a temporary tank in the stoke-hold, into which the water was pumped after being measured. A donkey-pump then pumped the water to the boilers as required.

The Author has much pleasure in bringing before the Institution the diagrams, Plate 3, Figs. 1 to 11, which are simply copies of those accompanying Professor Kennedy's report. In the trials referred to in the report, as Professor Kennedy explains, only one boiler was used; at the same time the combined economy of engines and boiler has been calculated. The result is, that when a point is reached in which the engines may be expected to work economically, the one boiler is forced much beyond its most economical rate of evaporation; for it is evident that when two boilers are used double the amount of steam can be supplied to the engines for a given rate of evaporation. Again, the airpressure of 2 inches used in trial "E" was a test beyond the rate of working for the maximum HP. With this air-pressure sufficient steam was supplied by the single boiler for 770 indicated HP., but when both boilers were working together the greatest demand made upon each was limited by the size of the engines to produce steam sufficient for 650 indicated HP. This could be obtained with a wind-pressure of 14 inch.

The Paper is accompanied by several drawings and tracings, from which Plates 1, 2 and 3 and the Figs. in the text have been engraved.

[APPENDIX.

TORPEDO-BOAT BOILER AND ENGINE TRIALS (PLATE 3).

DEAR SIRS, I have now pleasure in giving you a statement of the results of the five trials of the Thornycroft boilers and torpedo-boat engines which I made at your request in November last.

I have made five experiments in all-two with natural draught, one with about a of an inch of air-pressure in the stoke-hold, one with about an inch air-pressure, and one with about 2 inches of air-pressure. I have distinguished these trials by the letters "A" to "E," in the order in which they were carried out, namely trial "A" on 21st November, 1888, trial "B" on 22nd November, trial "C" on 24th November, trial "D" on 26th November, and trial "E" on 29th November. It will, however, be more convenient to take them in an order corresponding to the air-pressure in the stoke-hold, and in what follows I have therefore done this. Before going on to describe any of them, however, I may state briefly the methods adopted.

OBSERVATIONS IN BOILER TRIALS.

The essential matter to be experimented upon was the behaviour of the boiler under different conditions.

I had the coal weighed roughly into sacks of 1 cwt. and cwt. before going on board, and stowed in the stoke-hold in these sacks; each sack was weighed in the stoke-hole by a tested spring-balance before being emptied, and the weight of the sacks themselves was also afterwards determined. The water was measured in the engine-room on its way from the hot-well to the feed-pump. For this purpose two cylindrical tanks were used, each holding about 540 lbs. of water, and three-way cocks were arranged above and below these tanks, so that one of them could always be receiving the delivery from the pump while the other was in communication with the feed-pump and therefore being emptied. The tanks were filled up to a marked height on a glass water-gauge, and when empty were completely drained by the lower cock. The possible error in filling was very small, amounting to only 1 lb. of water for a range of 4 inches on the gauge, 2 inches above and 2 inches below the mark. As it was quite easy to keep the water-level within this range during the trials, the probable error due to over or under filling is practically nil. The feed-pump could always empty one tank in less time than the air-pump could fill the other. In trials "A" and "B" the feed-pump delivered direct into the boiler as usual, in the other trials the feed-pump delivered into a small open tank holding about 50 gallons placed in the stoke-hold, and the actual feeding of the boiler was done by a donkey in