becomes in its turn a base capable of being employed as known sides of other triangles. For instance, the angles of the triangles ACG and B C F being known by observation, and their sides A C and B C, we can thence calculate the lengths A G, C G, and B F, CF. Again, C G and C F being known, and the included angle G C F, G F may be calculated, and so on. Thus may all the stations be accurately determined and laid down, and as this process may be carried on to any extent, a map of the whole country may be thus constructed, and filled in to any degree of detail we please.

(275.) Now, on this process there are two important remarks to be made. The first is, that it is necessary to be careful in the selection of stations, so as to form triangles free from any very great inequality in their angles. For instance, the triangle K B F would be a very improper one to determine the situation of F from observations at B and K, because the angle F being very acute, a small error in the angle K would produce a great one in the place of F upon the line B F. Such ill-conditioned triangles, therefore, must be avoided. But if this be attended to, the accuracy of the determination of the calculated sides will not be much short of that which would be obtained by actual measurement (were it practicable); and, therefore, as we recede from the base on all sides as a centre, it will speedily become practicable to use as bases, the sides of much larger triangles, such as G F, G H, H K, &c.; by which means the next step of the operation will come to be carried on on a much larger scale, and embrace far greater intervals, than it would have been safe to do (for the above reason) in the immediate neighbourhood of the base. Thus it becomes easy to divide the whole face of a country into great triangles of from 30 to 100 miles in their sides (according to the nature of the ground), which, being once well determined, may be afterwards, by a second series of subordinate operations, broken up into smaller ones, and these again into others of a still minuter order, till the final filling in is brought within the limits of personal survey and draftsmanship, and till a map is constructed, with any required degree of detail.

(276.) The next remark we have to make is, that all the triangles in question are not, rigorously speaking, plane, but spherical existing on the surface of a sphere, or rather, to speak correctly, of an ellipsoid. In very small triangles, of six or seven miles in the side, this may be neglected, as the difference is imperceptible; but in the larger ones it must be taken into consideration. It is evident that, as every object used for pointing the telescope of a theodolite has some certain elevation, not only above the soil, but above the level of the sea, and as, moreover, these elevations differ in every instance, a reduction to the horizon of all the measured angles would appear to be required. But, in fact, by the construction of the theodolite (art. 192.), which is nothing more than an altitude and azimuth instrument, this reduction is made in the very act of reading off the horizontal angles. Let E be the centre of the earth; A, B, C, the places on its spherical surface, to which three stations, A, P, Q, in a country are referred by radii E A, E BP, ECQ. If a theodolite be stationed at A, the axis of its horizontal circle will point to E when truly adjusted, and its plane will be a tangent to the

E

P

M

B

C

sphere at A, intersecting the radii E BP, EC Q, at M and N, above the spherical surface. The telescope of the theodolite, it is true, is pointed in succession to P, and Q; but the readings off of its azimuth circle give not the angle PAQ between the directions of the telescope, or between the objects P, Q, as seen from A; but the azimuthal angle M A N, which is the measure of the angle A of the spherical triangle BAC. Hence arises this remarkable circumstance, that the sum of the three observed angles of any of the great triangles in geodesical operations is always found to be rather more than 180°. Were the earth's surface a plane, it ought to be exactly 180°; and this excess, which is called the spherical excess, is so far from being a proof of incorrectness in the work, that it is essential to its accuracy, and offers at the same time another palpable proof of the earth's sphericity.

(277.) The true way, then, of conceiving the subject of a trigonometrical survey, when the spherical form of the earth is taken into consideration, is to regard the network of triangles with which the country is covered, as the bases of an assemblage of pyramids converging to the centre of the earth. The theodolite gives us the true measures of the angles included by the planes of these pyramids; and the surface of an imaginary sphere on the level of the sea intersects them in an assemblage of spherical triangles, above whose angles, in the radii prolonged, the real stations of observation are raised, by the superficial inequalities of mountain and valley. The operose calculations of spherical trigonometry which this consideration would seem to render necessary for the reductions of a survey, are dispensed with in practice by a very simple and easy rule, called the rule for the spherical excess, which is to be found in most works on trigonometry. If we would take into account the ellipticity of the earth, it may also be done by appropriate processes of calculation, which, however, are too abstruse to dwell upon in a work like the present.

(278.) Whatever process of calculation we adopt, the result will be a reduction to the level of the sea, of all the triangles, and the consequent determination of the geogra phical latitude and longitude of every station observed. Thus we are at length enabled to construct maps of countries; to lay down the outlines of continents and islands; the courses of rivers; the places of cities, towns and villages; the direction of mountain ridges, and the places of their principal summits; and all those details which, as they belong to physical and statistical, rather than to astronomical geography, we need not here dilate on. A few words, however, will be necessary respecting maps, which are used as well in astronomy as in geography.

(279.) A map is nothing more than a representation, upon a plane, of some portion of the surface of a sphere, on which are traced the particulars intended to be expressed, whether they be continuous outlines or points. Now, as a spherical surface

We here neglect the ellipticity of the earth, which, for such a purpose as map-making, is too trifling to have any material influence.

can by no contrivance be extended or projected into a plane, without undue enlargement or contraction of some parts in proportion to others; and as the system adopted in so extending or projecting it will decide what parts shall be enlarged or relatively contracted, and in what proportions; it follows, that when large portions of the sphere are to be mapped down, a great difference in their representations may subsist, according to the system of projection adopted.

(280.) The projections chiefly used in maps, are the orthographic, stereographic, and Mercator's. In the orthographic projection, every point of the hemisphere is referred to its diametral plane or base, by a perpendicular let fall on it, so that the representation of the hemisphere thus mapped on its base, is such as would actually appear to an eye placed at an infinite

distance from it. It is obvious, from the annexed figure, that in this projection only the central portions are represented of their true forms, while all the exterior is more and more distorted and crowded together as we approach the edges of the map. Owing to this cause, the orthographic projection, though very good for small portions of the globe, is of little service for large ones.

perpendicular to E B by the visual line P M E. The stereographic projection of a sphere, then, is a true perspec

tive representation of its concavity on a diametral plane; and, as such, it possesses some singularly elegant geometrical properties, of which we shall state one or two of the principal.

(282.) And first, then, all circles on the sphere are represented by circles in the projection. Thus the circle X is projected into x. Only great circles passing through the vertex B are projected into straight lines traversing the centre C: thus, BP A is projected into C A.

2dly. Every very small triangle, G H K, on the sphere, is represented by a similar triangle, g h k, in the projection. This is a very valuable property, as it insures a general similarity of appearance in the map to the reality in all its parts, and enables us to project at least a hemisphere in a single map, without any violent distortion of the configurations on the surface from their real forms. As in the orthographic projection, the borders of the hemisphere are unduly crowded together; in the stereographic, their projected dimensions are, on the contrary, somewhat enlarged in receding from the centre.

(283.) Both these projections may be considered natural ones, inasmuch as they are really perspective representations of the surface on a plane. Mercator's is entirely an artificial one, representing the sphere as it cannot be seen from any one point, but as it might be seen by an eye carried successively over every part of it. In it, the degrees of longitude,

and those of latitude, bear always to each other their due proportion: the equator is conceived to be extended out into a straight line, and the meridians are straight lines at right angles to it, as in the figure. Altogether, the general cha