stroke is precisely equal to that evaporated during the exhaust stroke, and consequently no condensed steam can leave the engine as water.

Let us suppose for the sake of argument, however, that an engine using 20 lbs. of 100 lbs. steam per horse per hour discharges 2 lbs. of water per horse per hour. As each of these brought, in round numbers, 1185 thermal units into the engine and takes away only 212 units, it is clear that each pound must leave behind it 973 units; consequently the cylinder will be hotter at the end of each revolution than it was at the beginning, and the process would go on until condensation must entirely cease. It will be urged, however, that a steam jacket certainly does discharge water, and that in considerable quantity, which it did not receive; and as this is apparently indisputable, we are here face to face with one of the puzzles to which we have referred. The fact, however, is in no wise inconsistent with that advanced. If an engine with an unjacketed cylinder regularly receives water from the boiler, that engine will discharge precisely an equal weight of water. The liquid will pass away in suspension in the exhaust steam. The engine has no power whatever of converting it into steam. The case of a jacketed engine is different. Such an engine will evaporate in the cylinder water received with the steam, but it can only do so at the expense

of the steam contained in the jacket. For every I lb. of water boiled away in the cylinder 1 lb. of steam is condensed in the jacket; and the corollary is that if an engine was supplied with perfectly dry steam there would be no steam condensed in the jacket, save that required to meet the loss due to radiation and the conversion of heat into work. The effect of the jacket will be to boil a portion of the water during the close of the stroke, and so to keep up the toe of the diagram, and so get more work out of the steam. If, however, the steam was delivered wet to the engine, it is very doubtful if the jacket could be productive of much economy. The water would be converted into steam during the exhaust stroke, and no equivalent would be obtained for the steam lost in the jacket.

In a good condensing engine, about 3 lbs. of steam per horse per hour are condensed in the jacket. The cylinder will use, say, 15 lbs. of steam, so the total consumption is 18 lbs. per horse per hour. It is none the less a fact, although it is not generally known, that the average marine boiler sends over about 8 per cent. of water in the form of insensible priming with the steam. Now, 8 per cent. of 18 lbs. is 1.44 lbs., so that in this way we have nearly one-half the jacket condensation accounted for as just explained. One horse-power represents 2562 thermal units expended per hour, or say

2.6 lbs. of steam of 100 lbs. pressure condensed to less than atinospheric pressure; and 1.44+2.60 =4.04 lbs. per horse per hour, as the necessary jacket condensation if no water is to be found in the working cylinder at the end of each stroke. That this quantity is not condensed only proves that the water received from the boiler, or resulting from the performance of work, is not all re-evaporated.

Something still remains to be written about the true action of the steam jacket, but this we must reserve for another chapter. We have said enough, we think, to show that, as we have stated, the jacket has more to do than to keep the cylinder hot. With jacketed engines, more than any other, it is essential that the the steam should be dry. In the case of an unjacketed engine, water supplied from the boiler will pass through the engine as water, and do little harm; but if the engine is jacketed, then the whole, or a part of this water, will be converted into steam, especially during the period of exhaust, when it can do no more good than if it were boiled away in a pot in the engine room. This is the principal reason why such conflicting opinions are expressed concerning the value of jackets. That depends principally on the merits of the boiler.

COMPOUND, TRIPLE EXPANSION AND QUAD

RUPLE EXPANSION ENGINES.

Cylinder Condensation and Re-evaporation.— The method of variation of this waste was qualitatively determined by Clarke about 1850; was roughly gauged by him, both as to magnitude and as to its effect in limiting the ratio of expansion; was quantitatively investigated by Hirn and Isherwood afterward, and was finally made the subject of an investigation by Messrs. Gately and Kletsch, in which it was endeavored to ascertain with some degree of accuracy the method of the variation of the waste with variation of each of the essential conditions affecting and determining it. The result of the research in brief was to show that the waste varied, in the cases studied, sensibly as the square root of the expansion and as the time of exposure, and was subject to a very slow decrease as the pressure adopted increased, the engine being worked condensing; decreasing about twice as rapidly, the condenser being thrown off.

Variation with ratio of expansion was also capable of being expressed with great accuracy by an hyperbolic expression, the product of areas of surface exposed up to the point of cut-off and the percentage of condensation being found sensibly constant. Under ordinary working conditions, the steam pressure being about sixty pounds per square inch, by gauge, the cut off at

one-third, and the speed of piston 554 feet per minute, and of rotation sixty-eight revolutions, the condensation was about one-third, or the equivalent of fifty per cent. of the total consumption of a similar engine having a non-conducting cylinder, and thus free from this waste. Reduced to quantity of steam and of heat wasted, per square foot of surface exposed to point of cut off, per minute of exposure, and per degree of range of temperature between prime and exhaust steam, Professor Marks finds the co-efficient to be .02 pounds, or 18 B. T. U., nearly, a result closely confirmed by the investigations of the same author, taking Hill's experiments for comparison, and also corroborated by the later work of other investigators.

These facts and laws being established, it becomes possible to determine the behavior of steam entering any given cylinder, and its method of working and of waste in any engine. Common experience, as well as theoretical considerations based upon the investigations already made, proves that it is impossible to expand steam in the ordinary single-cylinder engine with satisfactory gain of efficiency, beyond a point variable with the conditions assumed, but which may be roughly taken as not far from that giving a ratio of expansion equal to about one-half the square root of the steam pressure, measured from vacuum for the condensing engine