For the same reason, the cylinder and its casing should be independent of the frame. In many horizontal engines, the cylinders overhang the frame throughout the entire length, by reason of the joining of their front portions with the frames. Although such a construction is attended with no serious consequences in small engines, nevertheless in large engines it is exceedingly harmful. Indeed, in most modern single-acting engines, the pistons are directly connected with the crank-shaft by the piston-rod, without any intermediate connecting-rod or cross-head. The vertical reaction of the motive effort on the piston is, therefore, taken up entirely by the thrust of the cylinder, which is also vertical (Fig. 24). This thrust, acting against an unsupported part, may cause fractures; at any rate, it entails a rapid deterioration of the cylinder joint.

The Frame.—Gas-engines driven as they are, by explosions, giving rise to shocks and blows, should be built with frames, heavy, substantial, and broad-based, so as to rest solidly on the ground. This essential condition is often fulfilled at the cost of the engine's appearance; but appearance will be willingly sacrificed to meet one of the requirements of perfect operation. For engines of more than 8 to 10 horse-power, frames should be employed which can be secured to the masonry foundation without a separate pedestal or base. Some manufacturers, for the purpose of lightening the frame, attach but little importance to the foundation and to strength of construction, and employ the design illustrated in place of the crank-shaft bearing (Fig. 25); others, in order to facilitate the adjusting of the connecting-rod bearings, prefer the second form (Fig. 26). It is evident that, in the first case, a part of the effort produced by the explosion reacts on the upper portion of the connecting-rod bearing, on the cap of the crank-shaft bearing, and consequently on the fastening-bolts. In the second case, if the adjustment be not very carefully made, or if the rubbing surfaces are insufficient, the entire thrust due to the explosion will be received by the meeting parts of the two bushings, thus injuring them and causing a more rapid wear. In the construction of large engines, some manufacturers take the precaution of forming the connecting-rod bearings of four parts, adjustable to take up the wear, so that the effort is exerted against the parts disposed at right angles to each other. A form that seems rational is that shown in Fig. 27, in which the reaction of the thrust is taken up by the lower bearing, rigidly supported by the braced frame, in the direction opposite to that of the explosive effort.

The sum of the projecting surfaces of the two bearings should be so calculated that a maximum explosive pressure of 405 to 425 pounds per square inch will not subject the bearings to a pressure higher than 425 to 550 pounds per square inch.

Fly-Wheels.—In gas-engines particularly, the fly-wheel should be secured to the crank-shaft with the utmost care. It should be mounted as near as possible to the bearings; otherwise the alinement of the shaft will be destroyed and its strength impaired. If the fly-wheel be fastened by means of a key or wedge having a projecting head, it is advisable to cover the end of the shaft by a movable sleeve. The fly-wheel should run absolutely true and straight even if the explosion be premature. In well-built engines the fly-wheels are lined up and shaped to the rim. The periphery is slightly rounded in order the better to guide the belt when applied to the wheel.

Furthermore, fly-wheels should be nicely balanced; those are to be preferred which have no counter-weights cast or fastened to the hub, the spokes, or the rim. The system of balancing the engine by means of two fly-wheels, mounted on opposite sides, is used chiefly for the purpose of equalizing the inertia effects. Special engines, employed for driving dynamos, and even industrial engines of high power, are preferably fitted with but a single fly-wheel, with an outer bearing, since they more readily counteract the cyclic irregularities or variations of speed occurring in a single revolution (Fig. 28). If in this case a pulley be provided, it should be mounted between the engine and the outer bearing. The following advantages may be cited in favor of the single fly-wheel, particularly in the case of dynamo-driving engines:

1. The single fly-wheel permits a more ready access to the parts to be examined.

2. It involves the employment of a third bearing, thus avoiding the overhang caused by two ordinary fly-wheels.

3. It avoids the torsional strain to which the two-wheel crank is subjected when starting, stopping, and changing the load, the peripheral resistance varying in one of the fly-wheels, while the other is subjected to a strain in the opposite direction on account of the inertia.

4. Two fly-wheels, keyed as they are to projecting ends of the shaft, will be so affected at the rims by the explosions that the belts will shake.

The third bearing which characterizes the single-fly-wheel system, is but an independent support, resting solidly on the masonry bed of the engine. The bearing with its independent support is sufficiently rigid, and is not subjected to any stress from the crank at the moment of explosion, the reaction of the crank affecting only the frame bearings. With such fly-wheels, reputable firms guarantee a cyclic regularity which compares favorably with that of the best steam-engines. For a duty varying from a third of the load to the maximum load, these engines, when driving direct-current dynamos for directly supplying an electric-light circuit, will insure perfect steadiness of the light; and the effectually aperiodic measuring instruments will not indicate fluctuations greater than 2 to 3 per cent. of the tension or intensity of the current. The coefficient of the variations in the speed of a single revolution will thus be not far from ¹⁄₆₀.

Straight and Curved Spoke Fly-Wheels.—The spokes of fly-wheels are either straight or curved. In assembling the motor parts it too often occurs that curved spoke fly-wheels are mounted with utter disregard of the direction in which they are to turn. It is important that curved spokes should be subjected to compression and not to traction. Hence the fly-wheels should be so mounted that the concave portions of the spokes travel in the direction of rotation, as shown in the accompanying diagram (Fig. 29). If a single fly-wheel be employed on an engine of the type in which the speed is governed by the "hit-and-miss" system, the fly-wheel should be extra heavy to counteract the irregularities of the motive impulses when the engine is not working at its full load, or in other words, when no explosion takes place at every cycle.

The Crank-Shaft.—The crank-shaft should be made of the best mild steel. Those shafts are to be preferred the cranks of which are not forged on (Fig. 30), but cut out of the mass of metal; furthermore, the brackets or supports should be planed and shaped so that they are square in cross-section.

Such a design involves fine workmanship and speaks well for the construction of the whole engine. Moreover, it enables the bearings to be brought nearer each other, reduces to a minimum that part of the crank-shaft which may be considered the weakest, and permits a rational and exact counterbalancing of the moving parts, such as the crank and the end of the connecting-rod. The best manufacturers have adopted the method of fastening to the cranks balancing weights secured to the brackets, especially for high-speed engines or for engines of high power. The projecting surface of the crank-pin should, as a rule, be calculated for a pressure of 1,400 pounds per square inch.

Cams, Rollers, etc.—The cams, rollers, thrust-bearings, as well as the piston-pin in particular, should be made of good steel, case-hardened to a depth of at least .08 of an inch. Their hardness and the degree of cementation may be tested by means of a file. This is the method followed by the best manufacturers.

Bearings.—All the bearings and all guides should be adjustable to take up the wear. They are usually made of bronze or of the best anti-friction metal.

Steadiness.—The steadiness of engines may be considered from two different standpoints.

1. Variation of the Number of Revolutions at Different Loads.—This depends chiefly on the sensitiveness of the governor, which should be of the "inertia" or of the "ball" (or centrifugal) type. The first form is rarely employed, except in small engines up to 10 horse-power, and is applicable only to engines in which the "hit and miss" system is employed (Fig. 33). The second form is more widely used, and is applicable to engines having "hit-and-miss" or variable admission devices. In the first form, the governor simply displaces a very light member, whatever may be the size of the engine, for which reason the dimensions are very small. In the second form, on the other hand, the governor acts either on a conical sleeve or on some other regulating member offering resistance. Evidently, in order to overcome the reactions to which it is subjected, it must be as heavy and powerful as a steam-engine governor. Sufficient allowance is made in a good engine for variation in the number of revolutions between no load and full load, not greater than two per cent. if the admission be of the "hit-and-miss" type, and five per cent. if it be of the variable type.

2. Cyclic Regularity.—This term means simply that the speed of the engine is constant in a single revolution. In practice this is never attained. Allowance is made in engines used for driving direct-current dynamos for a variation of about ¹⁄₆₀; while in industrial engines a variation of ¹⁄₂₅ is permissible. Cyclic variation depends only on the weight of the fly-wheel; whereas variation in the number of revolutions is determined chiefly by the governor.

Governors.—Diagrams are here presented of the principal types of governors—the inertia governor, the ball or centrifugal governor controlling an admission-valve of the "hit-and-miss" type (Fig. 34), and the ball or centrifugal governor controlling a variable gas-admission valve (Fig. 35).

In distinguishing between the operation of the two last-mentioned types, it may be stated that the former bears the same relation to the hit-and-miss gear as it does, for example, to the valve gear of a Corliss steam-engine. In other words, it is an apparatus that indicates without inducing, admission or cut-off. The second type, on the other hand, operates by means of slides and the like, as in the Ridder type of engine, in which it controls the displacement of the cut-off or distribution slide-valve and is subjected to variable forces, depending on the pressure, lubrication, the condition of the stuffing-boxes, and the like.

In gas as well as in steam engines, designs are to be commended which shield the delicate mechanism from strains and stresses that are likely to destroy its sensitiveness, as is the case in the automatic cut-off of the Corliss steam-engine.

Governors should be provided with means to permit the manual variation of the speed while the engine is in operation.

For small motors, one of the most widely used admission devices is that of the "hit-and-miss" type. As its name indicates, this admission arrangement allows a given quantity of gas to enter the cylinder for a number of consecutive intervals, until the engine is about to exceed its normal speed. Thereupon the governor cuts off the gas entirely. The result is that, in this system, the number of admissions is variable, but that each admitted charge is composed of a constant proportion of gas and air.

The governors employed for the "hit-and-miss" type are either "inertia" or "centrifugal" governors.

Inertia governors (Fig. 33) are less sensitive than those of the centrifugal type. They are generally applied only to industrial engines of small power, in which regularity of operation is a secondary consideration.

Centrifugal governors employed for gas-engines with "hit-and-miss" regulation are, as a general rule, noteworthy for their small size, which is accounted for by the fact that, in most systems, merely a movable member is placed between the admission-controlling means and the valve-stem (Fig. 34). It follows that this method of operation relieves the governor of the necessity of overcoming the resistance of the weight of moving parts, more or less effectually lubricated, and subjected to the reaction of the parts which they control.

In engines equipped with variable admission devices for the gas or the explosive mixture, the governor actuates a sleeve on which the admission-cam is fastened (Fig. 35). Or, the governor may displace a conical cam, the reaction of which, on contact with the lever, destroys the stability of the governor. These conditions justify the employment of powerful governors which, on account of the inertia of their parts, diminish the reactionary forces encountered.

The centrifugal governor should be sufficiently effectual to prevent variations in the number of revolutions within the limits of 2 to 3 per cent. between no load and approximately full load. Under equivalent conditions, the inertia governor can hardly be relied upon for a coefficient of regularity greater than 4 to 5 per cent.

The manner of a governor's operation is necessarily dependent on the admission system adopted. And the admission system varies essentially with the size, the purpose of the engine, and the character of the fuel employed.

Vertical Engines.—For some years past there seems to have been a tendency in Europe to use horizontal instead of vertical engines, especially since engines of more than 10 or 15 horse-power have been extensively used for industrial purposes. The vertical type is used for 1 to 8 horse-power engines, with the cylinder in the lower part of the frame, and the shaft and its fly-wheel in the upper part (Fig. 36). The only merit to be attributed to this arrangement is a great saving of space. It is evident, however, that beyond a certain size and power, such engines are unstable. In America particularly, many manufacturers of high-power engines (50 to 100 horse-power or more) prefer the vertical or "steam-hammer" arrangement, which consists in placing the cylinder in the upper part, and the shaft in the lower part of the frame as close to the ground as possible (Figs. 37 and 38). The problem of saving space, as well as that of insuring stability, is thus solved, so that it is easily possible to run up the speed of the engine. There is also the advantage that the shaft of a dynamo can be directly coupled up with the crank-shaft of the engine, thus dispensing with a belt, which, at the least, absorbs 4 to 6 per cent. of the total power. It should, nevertheless, be borne in mind that the direct coupling of electric generators to engine-shafts implies the use of extremely large and, therefore, of extremely costly dynamos. Furthermore, by reason of this arrangement, groups of electro-generators can be disposed in a comparatively small amount of space. Some English manufacturers are also beginning to adopt the "steam-hammer" type of engine for high powers, the result being a marked saving in material and lowering of the cost of installation.

Power of the Engine.—The first thing to be considered is that the power of a gas-engine is always given in "effective" horse-power, and that the power of a steam-engine is always given in "indicated" horse-power in contracts of sale. In England and in the United States, the expression "nominal" horse-power is still employed. It may be advisable to define these various terms exactly, since unscrupulous dealers, to the buyer's loss, have done much to confuse them.

"Indicated" horse-power is a designation applied to the theoretical power produced by the action of the motive agent on the piston. The work performed is measured on an indicator card, by means of which the average pressure to be considered in the computation of the theoretical power is ascertained.

The "effective" or brake horse-power is equal to the "indicated" horse-power, less the energy absorbed by passive resistance, friction of the moving parts, etc.

The "effective" work is an experimental term applied to the power actually developed at the shaft. This work is of interest solely to the engine user.

In a well-built motor, in which the passive resistance by reason of the correct adjustment and simplicity of the parts, is reduced to a minimum, the "effective" horse-power is about 80 to 87 per cent. of the "indicated" horse-power, when the engine runs under full load. This reduced output is usually called the "mechanical efficiency" of the engine.

"Nominal" horse-power is an arbitrary term in the sense in which it is used in England and America, where it is quite common. The manufacturers themselves do not seem to agree on its absolute value. A "nominal" horse-power, however, is equal to anything from 3 to 4 "effective" horse-power. The uncertainty which ensues from the use of the term should lead to its abandonment.

In installing a motor, the determination of its horse-power is a matter of grave importance, which should not be considered as if the motor were a steam-engine or an engine of some other type. It must not be forgotten that, especially at full load, explosion-engines are most efficient, and that, under these conditions, it will generally be advisable to subordinate the utility of having a reserve power to the economy which follows from the employment of a motor running at a load close to its maximum capacity. On the other hand, the gas-engine user is unwilling to believe that the stipulated horse-power of the motor which is sold to him is the greatest that it is capable of developing under industrial conditions. Business competition has led some firms to sell their engines to meet these conditions. It is probably not stretching the truth too far to declare that 80 per cent. of the engines sold with no exact contract specifications are incapable of maintaining for more than a half hour the power which is attributed to them, and which the buyer expects. It follows that the power at which the engine is sold should be both industrially realized and maintained, if need be, for an entire day, without the engine's showing the slightest perturbation, or faltering in its silent and regular operation. To attain this end, it is essential that the energy developed by the engine in normal or constant operation should not exceed 90 to 95 per cent. of the maximum power which it is able to yield, and which may be termed its "utmost power". As a general rule, especially for installations in which the power fluctuates from the lowest possible to double this, as much attention must be paid to the consumption at half load as at full load; and preference should be given to the engine which, other things being equal, will operate most economically at its lowest load. In this case the consumption per effective horse-power is appreciably higher. Generally, this consumption is greater by 20 to 30 per cent. than that at full load. This is particularly true of the single-acting engines so widely used for horse-powers less than 100 to 150.

In some double or triple-acting engines, according to certain writers, the diminution in the consumption will hardly be proportional to the diminution of the power, or at any rate, the difference between the consumption per B.H.P. at full load and at reduced load will be less than in other engines. It should be observed, however, that this statement is apparently not borne out by experiments which the author has had occasion to make. To a slight degree, this economy is obtained at the cost of simplicity, and consequently, at the cost of the engine. At all events, the engines have the merit of great cyclic regularity, rendering them serviceable for driving electric-light dynamos; but this regularity can also be attained by the use of the extra heavy fly-wheels which English firms, in particular, have introduced.

Automatic Starting.—When the gas-engine was first introduced, starting was effected simply by manually turning the fly-wheel until steady running was assured. This procedure, altogether too crude in its way, is attended with some danger. In a few countries it is prohibited by laws regulating the employment of industrial machinery. If the engine be of rather large size one, moreover, which operates at high pressure—such a method of starting is very troublesome. For these reasons, among others, manufacturers have devised automatic means of setting a gas-engine in motion.

Of such automatic devices, the first that shall be mentioned is a combination of pipes, provided with cocks, by the manipulation of which, a certain amount of gas, drawn from the supply pipe, is introduced into the engine-cylinder. The piston is first placed in a suitable position, and behind it a mixture is formed which is ignited by a naked flame situated near a convenient orifice. When the explosion takes place the ignition-orifice is automatically closed, and the piston is given its motive impulse. The engine thus started continues to run in accordance with the regular recurrence of the cycles. In this system, starting is effected by the explosion of a mixture, without previous compression.

Some designers have devised a system of hand-pumps which compress in the cylinder a mixture of air and gas, ignited at the proper time by allowing it to come into contact with the igniter, through the manipulation of cocks (Fig. 39).

These two methods are not absolutely effective. They require a certain deftness which can be acquired only after some practice. Furthermore, they are objectionable because the starting is effected too violently, and because the instantaneous explosion subjects the stationary piston, crank, and fly-wheel to a shock so sudden that they may be severely strained and may even break. Moreover, the slightest leakage in one of the valves or checks may cause the entire system to fail, and, particularly in the case of the pump, may induce a back explosion exceedingly dangerous to the man in charge of the engine.

These systems are now almost generally supplanted by the compressed-air system, which is simpler, less dangerous, and more certain in its effect.

The elements comprising the system in question include essentially a reservoir of thick sheet iron, capable of resisting a pressure of 180 to 225 pounds and sufficient in capacity to start an engine several times. This reservoir is connected with the engine by piping, which is disposed in one of two ways, depending upon whether the reservoir is charged by the engine itself operatively connected with the compressor, or by an independent compressor, mechanically operated.

In the first case, the pipe is provided with a stop-cock, mounted adjacent to the cylinder, and with a check-valve. When the engine is started and the gas cut off, the air is drawn in at each cycle and driven back into the reservoir during the period of compression. When the engine, running under these conditions by reason of the inertia of the fly-wheel, begins to slow down, the check-valve is closed and the gas-admission valve opened, so as to produce several explosions and to impart a certain speed to the engine in order to continue the charging of the reservoir with compressed air. This done, the valve on the reservoir itself is tightly closed, as well as the check-valve, so as to avoid any leakage likely to cause a fall in the reservoir's pressure.

In the second case, which applies particularly to engines of more than 50 horse-power, the charging pipe connected with the reservoir is necessarily independent of the pipe by means of which the motor is started. The reservoir having been filled and the decompression cam thrown into gear, starting is accomplished:

1. By placing the piston in starting position, which corresponds with a crank inclination of 10 to 20 degrees in the direction of the piston's movement, from the rear dead center, immediately after the period of compression;

2. By opening the reservoir-valve;

3. By allowing the compressed air to enter the cylinder rapidly, through the quick manipulation of the stop-cock, which is closed again when the impulse is given and reopened at the corresponding period of the following cycle, this operation being repeated several times in order to impart sufficient speed to the motor;

4. By opening the gas-valve and finally closing the two valves of the compressed-air pipe.

The pipes and compressed-air reservoirs should be perfectly tight. The reservoirs should have a capacity in inverse ratio to the pressure under which they are placed, i.e., they increase in size as the pressure decreases. If, for example, the reservoirs should be operated normally at a pressure of 105 to 120 pounds per square inch, their capacity should be at least five or six times the volume of the engine-cylinder. If these reservoirs are charged by the engine itself, the pressure will always be less by 15 to 20 per cent. than that of the compression.

CHAPTER III

Location.—The engine should be preferably located in a well-lighted place, accessible for inspection and maintenance, and should be kept entirely free from dust. As a general rule, the engine space should be enclosed. An engine should not be located in a cellar, on a damp floor, or in badly illuminated and ventilated places.

Gas-Pipes.—The pipes by which fuel is conducted to engines, driven by street-gas, and the gas-bags, etc., are rarely altogether free from leakage. For this reason, the engine-room should be as well ventilated as possible in the interest of safety. Long lines of pipe between the meter and the engine should be avoided, for the sake of economy, since the chances for leakage increase with the length of the pipe. It seldom happens that the leakage of a pipe 30 to 50 feet long, supplying a 30 horse-power engine, is much less than 90 cubic feet per hour. The beneficial effect of short supply pipes between meter and engine on the running of the engine is another point to be kept in mind.

An engine should be supplied with gas as cool as possible, which condition is seldom realized if long pipe lines be employed, extending through workshops, the temperature of which is usually higher than that of underground piping. On the other hand, pipes should not be exposed to the freezing temperature of winter, since the frost formed within the pipe, and particularly the crystalline deposition of naphthaline, reduces the cross section and sometimes clogs the passage. Often it happens that water condenses in the pipes; consequently, the piping should be disposed so as to obviate inclines, in which the water can collect in pockets. An accumulation of water is usually manifested by fluctuations in the flame of the burner. In places where water can collect, a drain-cock should be inserted. In places exposed to frost, a cock or a plug should be provided, so that a liquid can be introduced to dissolve the naphthaline. To insure the perfect operation of the engine, as well as to avoid fluctuations in nearby lights, pipes having a large diameter should preferably be employed. The cross-section should not be less than that of the discharge-pipe of the meter, selected in accordance with the prescriptions of the following table:

The records made are exact only when the meters (Fig. 40) are installed and operated under normal conditions. Two chief causes tend to falsify the measurements in wet meters: (1) evaporation of the water, (2) the failure to have the meter level.

Evaporation occurs incessantly, owing to the flowing of the gas through the apparatus, and increases with a rise in the temperature of the atmosphere surrounding the meter. Consequently this temperature must be kept down, for which reason the meter should be placed as near the ground as possible. The evaporation also increases with the volume of gas delivered. Hence the meter should not supply more than the volume for which it was intended. In order to facilitate the return of the water of condensation to the meter and to prevent its accumulation, the pipes should be inclined as far as possible toward the meter. The lowering of the water-level in the meter benefits the consumer at the expense of the gas company.

Inclination from the horizontal has an effect that varies with the direction of inclination. If the meter be inclined forward, or from left to right, the water can flow out by the lateral opening at the level, and incorrect measurements are made to the consumer's cost.

During winter, the meter should be protected from cold. The simplest way to accomplish this, is to wrap substances around the meter which are poor conductors of heat, such as straw, hay, rags, cotton, and the like. Freezing of the water can also be prevented by the addition of alcohol in the proportion of 2 pints per burner. The water is thus enabled to withstand a temperature of about 5 degrees F. below zero. Instead of alcohol, glycerine in the same proportions can be employed, care being taken that the glycerine is neutral, in order that the meter may not be attacked by the acids which the liquid sometimes contains.

Dry Meters.—Dry meters are employed chiefly in cold climates, where wet meters could be protected only with difficulty and where the water is likely to freeze. In the United States the dry meter is the type most widely employed. In Sweden and in Holland it is also generally introduced (Fig. 41).

In the matter of accuracy of measurement there is little, if any, difference between wet and dry meters. The dry meter has the merit of measuring correctly regardless of the fluctuations in the water level. On the other hand, it is open to the objection of absorbing somewhat more pressure than the wet meter, after having been in operation for a certain length of time. This is an objection of no great weight; for there is always enough pressure in the mains and pipes to operate a meter.

In many cases, where the employment of non-freezing liquids is necessary, the dry meter may be used to advantage, since all such liquids have more or less corroding effect on sheet lead and even tin, depending upon the composition of the gas.

The dry meter comprises two bellows, operating in a casing divided into two compartments by a central partition. The gas is distributed on one or the other side of the bellows, by slides B. The slides B are provided with cranks E, controlled by levers M, actuated by transmission shafts O, driven by the bellows. The meter is adjusted by a screw which changes the throw of the cranks E and consequently affects the bellows. The movement of the crank-shaft D is transmitted to the indicating apparatus. In order to obviate any leakage, this shaft passes through a stuffing-box, G. The diagrams (Figs. 42-43) show the construction of a dry meter, the arrows indicating the course taken by the gas.

Care should be taken to provide the gas-pipe with a drain-cock, at a point near the engine. By means of this cock, any air in the pipe can be allowed to escape before starting; otherwise the engine can be set in motion only with difficulty. If the engine be provided with an incandescent tube, the gas-supply pipe of the igniter should be fitted with a small rubber pouch or bag, in order to obviate fluctuations in the burner flame, caused by variations in the pressure (Fig. 44). As a general rule, the supply-pipe should be connected with the main pipe on the forward side of the bags and gas-governors. The main pipe and all other piping near the engine should extend underground, so that free access to the motor from all sides can be obtained, without possibility of injury.

Anti-pulsators, Bags, Pressure-Regulators.—The most commonly employed means of preventing fluctuation of nearby lights, due to the sharp strokes of the engine, consists in providing the gas-supply pipe with rubber bags (Fig. 45), which form reservoirs for the gas and, by reason of their elasticity, counteract the effect produced by the suction of the engine. Nevertheless, in order to insure a supply of gas at a constant pressure, which is necessary for the perfect operation of the engine, there are generally used, in addition to the bags, devices called gas-governors, or anti-pulsators (Fig. 46).

Although these devices are constructed in different ways, the underlying principle is the same in all. They comprise a metallic casing, containing a flexible diaphragm of rubber or of some fabric impermeable to gas. Suction of the engine creates a vacuum in the casing. The diaphragm bends, thereby actuating a valve, which cuts off the gas supply. During the three following periods (compression, explosion, and exhaust) the gas, by reason of its pressure on the diaphragm, opens the valve and fills the casing, ready for the next suction stroke.

Other devices, which are never sold with the engine, but are rendered necessary by reason of the conditions imposed by the gas supply are sold under the name "pressure-regulators" (Fig. 47). They consist of a bell, floating in a reservoir containing water and glycerine (or mercury), and likewise actuate a valve which partially controls the flow of gas. This valve being balanced, its mechanical action is the more certain. Such devices are very effective in maintaining the steadiness of lights. On the other hand, they are often an obstacle to the operation of the engine because they reduce the flow and pressure of the gas too much. In order to obviate this difficulty, a pressure-regulator should be chosen with discrimination, and of sufficiently large size to insure the maintenance of an adequate supply of gas to the engine. Frequent examinations should be made to ascertain if the bell of the regulator is immersed in the liquid. In the case of anti-pulsators, care should be taken that they are not spattered with oil, which has a disastrous effect on rubber. Anti-pulsators are generally mounted about 4 inches from a wall, in order that the diaphragm may be actuated by hand, if need be.