soon determined by trial. But it is especially important to regulate the operation of drying according to the quality of the material used; when the casting fails and the projectiles are full of flaws, it may always be attributed to the presence of moisture. Whatever be the quality of the sand, it is prepared as described under the preceding head; that which has been once used cannot be again employed without the addition of fresh sand.

Of Clay for Cores.

Although cores are now made of sand, we shall say a few words on the preparation of them from clay, and consequently on that of the clay itself: our remarks will apply also to the preparation of the nucleus of cores made of sand.

Argillaceous earth retains water with greater force, shrinks more in drying, and has a greater tendency to crack, in proportion, as it contains a greater quantity of alumina. The use of grey earths should be avoided, because they have not sufficient consistence; nearly all the earths which effervesce with acids are of this sort.

The clay is first dried, then pulverized, or rather beaten with a bat, to break the lumps, and sifted for the purpose of separating the pebbles, it is then moistened and well worked, adding at the same time about one third of horse dung. The viscous liquor contained in the dung prevents the clay from cracking, diminishes the shrinking, makes it less compact, less hard after drying, and easier to break when the shell is to be emptied. The clay used for the core of the eye should be passed through a silken sieve, and mixed with a smaller proportion of dung; the same may be said of that used for the last coats of cores made after the old method.

Of the models of Hollow Projectiles.

The

The globe, or model, is generally made of copper: it consists of two hemispheres joined by a tongue and groove, in such a manner that they touch each other only on the exterior circumference; the inner part should be bevelled at a large angle: this is an essential point, for by this form the seam of the projectile is made thinner than it would be if the two hemispheres touched each other on a larger surface. The metal should be from three to four lines* thick, so that it may not yield when the sand is rammed on it. diameter of a model for any calibre is variable, on account of the different degrees of expansion and contraction of different kinds of cast iron: it always expands in passing from a fluid to a solid state, and afterwards contracts in cooling. This increase and diminution of volume varies not only in different foundries, but also in different kinds of metal obtained in the same furnace by working it differently.

In general the grey metal expands more in crystallizing, or contracts less in cooling, than the white metal; the former may give projectiles of too great dimensions, whilst those cast with white metal in the same moulds may be too small. The diameter of the model should therefore be regulated according to the quality of metal which, in the furnace employed, appears to be most fit for the object in view, and most frequently obtained. On the other hand

*The French measures are retained in this article without reduction to the corresponding dimensions in English measures, because the former bear nearly the same proportion to the latter that the corresponding calibres bear to each other in the French service and in ours; besides they are easily reduced if necessary, to English measures; a French foot being equal to 12.79 English inches, very nearly. TRANS.

it advantageous for the service that the diameters of the projectiles should agree as nearly as possible with those of the largest gauge used in the inspection; which diameters, for twelve inch mortars and twenty-four and sixteen pounder guns, differ eighteen points from the calibre of the bore, and for other pieces, one line. It is only by trial, making the globe at first too large, that we can obtain the proper diameter, which should be such that the greater number of projectiles shall not pass through the intermediate gauge. To obtain this result it often happens that, for large calibres, the diameter of the model should exceed, by several points, that of the large gauge.

The addition of the dust of coal, or coke, to the sand facilitates the cleaning of the projectile and makes the surface more even, consequently the diameter smaller, so that a greater number pass through the intermediate gauge, or even through the small one. If water be poured on the iron whilst not, it contracts more. (See casting and finishing Projectiles.) Hence the necessity of being well acquainted with all the circumstances of the manufacture, when by the first trials, the dimension of the model is to be determined; if too small it cannot be corrected by a coating of tin, as has been sometimes attempted; it would be necessary to procure a new one, which occasions a considerable expense.

The model should be turned and finished in all its parts with the greatest precision. It was formerly the custom to flatten the models of all projec tiles very much at the poles: it was thought that the metal contracted more in the horizontal than in the vertical direction. We have for a long time contended against this opinion. Experiments have proved to us that the alleged difference in the contraction of the metal does not exist. If projectiles, of a medium weight, moulded with spherical models, are sometimes elongated, it is to be attributed solely to the unskilfulness of the workmen, who have not sufficiently compressed the sand about the lower pole: it then happens, especially in the case of large projectiles, that the weight of the metal, causing the sand to yield, produces the elongation in question. When the workmen have the requisite degree of skill and intelligence, the models of all hollow projectiles, below the calibre of ten inches, should be perfectly spherical: by flattening them we obtain many flattened, and ill shaped, projectiles. If, on the contrary, the models are spherical, the workman soon learns to ram the sand properly, so that the number of elongated projectiles is very small, whilst nearly all the others are perfectly spherical. Workmen generally prefer flattened models, because they are more afraid of obtaining elongated projectiles which will be rejected, than of producing a quantity of others more or less badly made.

The models of ten inch and twelve inch shells may be flattened from four to six points, as the sand cannot always be rammed sufficiently to prevent it from yielding to the pressure of these heavy castings. This explanation of the cause of the elongation of projectiles overthrows an absurd opinion, which has generally prevailed; there are, however, other causes of elongation which will be explained under the head of moulding.

The upper hemisphere of the model is pierced with a round hole intended to receive an iron spindle; in twelve inch shells the hole is nine lines in diameter; the length of the spindle is seven inches, four lines: it consists of three parts; one part is cylindrical and cut with a screw thread; it serves to fix to the hemisphere of the model a sort of handle by means of which it is managed: the second part is a truncated cone, all the dimensions of which are perfectly similar to another conical part on the spindle of the core, and it serves to prepare for the latter a lodgment in the sand. The third part is VOL. XVIII.-No. 1.-JULY, 1836.

2

nearly cylindrical, having also a diameter equal to that of the spindle of the core. In speaking of the flasks, we shall return to this subject. The upper hemisphere of the model of shells is pierced besides with two rectangular holes for the ears. Their position and form have undergone several variations; the following is the usual manner of tracing them. On a diameter perpendicular to the axis passing through the eye, lay off on each side three inches three lines, draw perpendiculars at these points, and with the radius of the twelve inch shell increased by that of the hole in the ear, cut the two perpendiculars at points which determine the centres of the holes: all shells being similar figures, the centres of the holes will be always found on the same radii. The diameters of these holes are laid down at four lines six points, for the twelve inch shell; three lines nine points, for the ten inch, and three lines three points for the eight inch. In practise it is necessary to make them a little larger, to afford the requisite play to the rings. It is essential that the mortices for the ears should be large enough to admit of their remaining in the sand, with the rings, after the hemisphere has been removed. Each ear is divided into two parts, which are generally joined by a tenon, so that they may be easily withdrawn in succession, without displacing the ring.

The ears placed as we have said, on a great circle, ought to fit accurately on the sphere, and form with it a continuous surface. The rings which are semi-elliptical, should be made with great precision, and finished with the file, so that the brazing, which is on the straight part, may not be visible. They are made of iron wire about two-fifths of an inch thick: it is necessary that they should play freely in the ears, and should fall down entirely on the surface of the shell. The diameter of the hole which they make should therefore be greater than their own: this is effected by enveloping them with a coat of clay, which should be quite round and well dried. When the shell is cast, this clay being removed, the ring has the requisite play.

The lower hemisphere of the model is also pierced with a hole which receives a piece called the false spindle; its dimensions are arbitrary; it is pierced at the inner end, with a mortice which receives a key. The other end is also pierced with a square hole in which is introduced a bar of iron or small ruler. The object of this spindle is to prevent the model from being detached from the mould when the flask containing it is raised; for this purpose a bit of wood is slipped under the rule; this acting like a wedge against the edge of the flask raises the rule and consequently presses the mould against the sand.

Of the Spindle of the Core, and of the Pattern.

The spindle of the core is divided into two parts by a swell several lines in height, in the form of a truncated cone, the base of which nearest to the core, has a diameter only three or four points less than the greatest diameter of the eye. The diameter of the other base is a little smaller than the first; in order that the spindle of the model, which should be perfectly similar to that of the core, may be withdrawn from the sand without causing any derangement of the mould. The dimensions of the part of the spindle opposite to the core are determined by the height of the flask, as we shall see further on. The part which supports the core, added to that which forms the eye of the projectile is equal in length to the distance from the upper circumference of the eye to the bottom of the shell, less a small quantity, and varying with the calibre: it is not important whether it be a little longer or a little shorter; the only essential point is, that the swell and the part of the

spindle opposite to the core, should be perfectly equal to their corresponding parts in the spindle of the model, and that the length of the swell be strictly determined according to the dimensions of the flask. We shall return to this subject.

The spindle may be either solid or hollow. Solid spindles having one or two deep grooves extending through their whole length, are also pierced, at the part which supports the core, with two rectangular holes in which pieces of slate are placed, to support the clay. In these grooves are placed straws to facilitate the disengaging of the gases. Hollow spindles intended principally for cores of sand, are pierced with five or six holes two lines in diameter. I think the latter kind preferable; they are besides easier to make, lighter, and less apt to spring than the solid spindles; they are made of sheet iron fifteen or twenty points thick, cut into pieces of proper size and rolled hot on a mandril. It is not necessary that the edges should be brazed together; it is sufficient that they join. The swell of the spindle is made by a ferrule which is brazed on. It is essential that the ferrule and all that part of the spindle which is to be similar to the spindle of the model, should be turned to the exact dimensions required. Both kinds of spindles should be flattened at the end opposite to the core, in order that it may enter into a crank; in that part there is also a hole to receive a key, when the core is placed in the mould. The other end should have a small conical indentation to receive the point of the screw which serves to fix the spindle in the lathe.

The dimensions of the core are determined by means of wooden patterns, of which there should be three, because the core is not finished at one operation. The radius of the first pattern differs ten lines, of the second four lines, from that of the core when finished; the third should give an exact section of the core, including that of the eye. The pattern is very easily drawn, for all its dimensions are given by those of the shell. A similar profile, made of iron, a gauge, and calibres for the eye, serve to verify the dimensions of the core. Before entering into the details of moulding, we shall describe

the flask.

Of the Flasks.

The flasks are boxes of wood, or of cast iron, without bottoms, divided into two unequal parts, each of which contains the mould of a hemisphere, and which are joined together by dowel pins, wedges, hooks, or small bolts and keys: the connexion by means of screws seems to me very defective.

The thickness of the boards of which wooden flasks are made should be from fifteen to eighteen lines for ten and twelve inch shells, and from ten to twelve lines for other projectiles. Wooden flasks are generally square; three of the angles are partly filled by triangular prisms of wood, to increase their solidity and diminish their capacity. The size of the flasks should be such as to leave a space of about an inch, or an inch and a half, around the model: if this space were greater, the preparation of the mould would require too much time; it would increase the expense, and at the same time impair the result of the operation, because the sand always yields more or less to the expansion of the metal, which is greatest at the points of least resistance, and this effect will be greater where the sand is thicker, the difficulty of ramming it firmly, being then increased. The part of a flask which contains the mould of the hemisphere in which the eye of a shell is placed, we shall call the drag; the other part the cope. The former which is ten inches eight lines deep for twelve inch shells, and to which the slides that receive

the wedges are adapted, contains a cast iron traverse, reinforced in the middle of its length, and pierced with a hole. The depth of this hole, or the thickness of the bar, or traverse, is four inches; its width is arbitrary. The hole, which is nine lines in diameter, receives the spindle of the model, the swell of which should rest exactly against this bar, as should also the swell of the spindle of the core. Accuracy in the position, and consequently in the thickness, of the sides of the projectile, depends therefore on the precision with which the bar is made and fixed in its place, as well as on accuracy and perfect identity in the form of the spindles.

This bar, or traverse, is therefore the most important part of the flask. If it were bent up or down the thickness of metal at the eye would be too great or two small: if its position were deranged laterally, the position and direction of the eye would vary accordingly. In verifying the flasks therefore the principal attention should be directed to the position, dimensions, and solidity of the traverses. They are let in their whole thickness into the sides of the flask, and kept in their places by screws, straps and keys. The depth of the drag is generally determined by the semi-diameter of the model added to the height of the swell of the spindle, and the thickness of the traverse. The sum of these three dimensions is ten inches eight lines for a twelve inch shell. The depth of the cope is equal to the semi-diameter of the model increased by two or three inches allowed for the thickness of the coat of sand: that depth is consequently from eight inches to nine inches in the example we have chosen. It may without inconvenience, be greater; but no variation can be allowed in the depth of the drag, unless corresponding variations are made in the spindles or in the thickness of the

traverse.

When the flask is so arranged that the shell is cast with the eye downward, it is necessary to give the cope a greater depth than it would require if the shell were cast with the eye uppermost; because in the former case the sand in the cope is supported by the board on which the flask rests, but not in the latter.

Cast iron flasks have a round form with a swell or projection at the part where the gate is placed. The sides may be vertical, or may consist of two truncated cones placed base to base, giving a swell in the middle of the height. The two parts are connected together by dowels and ears through which key bolts pass. The traverses should be cast separately, and the holes drilled cold, to secure greater accuracy. In casting them at the same time with the flasks it would be impossible to avoid slight variations in all their dimensions; cast iron flasks are far preferable to wooden ones, because they can be better joined, and are much less subject to derangement, and the traverses can be adjusted in them with greater precision and solidity. We have already said that there is an advantage in having the coat of sand thin; but in that case the wood, affected by the heat and steam, becomes warped, which always causes errors in the dimensions of the projectile: hence another reason in favor of cast iron flasks.

The gate or channel by which the metal is conducted into the moulds is curved and terminates at the extremity of a horizontal diameter of the mould. It is called a heel gate, and it is formed in the sand by two pieces of wood, one of which, placed vertically, has a conical form: the diameter of its greater base is two and a half inches, that of the lower base is nine lines for twelve inch shells, and its height is necessarily equal to that of the upper portion of the flask: the other piece of wood which forms the heel, is