We

therefore, will not represent the earth as we see it, or as we can possibly see it; and to construct such a globe with all the details of its surface clearly manifest, while at the same time we see the convexity and have to look up to some parts of the surface and down upon others, really introduces fresh misconceptions while getting rid of old ones. cannot reproduce in a model all the characteristics of the globe we live on, and must therefore be content with that mode of representation which will offer the greater number of advantages and be, on the whole, the most instructive and the most generally useful. This, I believe, is undoubtedly the hollow globe, in which, however, the outer surface would be utilised to give a general representation of the earth as proposed by M. Reclus, and which would no doubt be itself a very interesting and attractive object.

Advantages of a Concave Globe.

I will now proceed to show, in some detail, how the concave surface of a hollow globe is adapted to fulfil all the purposes and uses which M. Reclus desires.

We should, in the first place, be able to see the most distant regions in their true relative proportions with a facility of comparison unattainable in any other way. We could, for instance, take in at one glance Scandinavia and Britain, or Greenland and Florida, and by a mere turn of the head could compare any two areas in a whole hemisphere. Both the relative shape and the relative size of any two countries or islands could be readily and accurately compared, and no illusion as to the comparative magnitude of our own land would be possible. In the next place, the relief of the surface would be represented exactly as if the surface were convex, but facilities for bringing out all the details of the relief by suitable illumination would be immensely greater in the hollow globe. Instead of being obliged to have the source of illumination only fifty feet from the surface, it could be placed either at the pole or opposite the equator at a distance of 200 or 300 feet, and be easily changed in order to illuminate a particular region at any angle desired, so as to bring out

the gentlest undulations by their shadows. Of course, electric lighting would be employed, which by passing through slightly tinted media might be made to represent morning, noon, or evening illumination.

It is, however, when we come to the chief scientific and educational use of such a globe-the supply of maps of any portion of the earth on any scale by means of photography-that the superiority of the concave model is so overwhelming as to render all theoretical objections to it entirely valueless. We have seen that on the convex surface of a globe such as M. Reclus has proposed, photographic reproductions of small portions only would be possible, while in areas of the size of any important European State, the errors due to the greater distance and the oblique view of the lateral portions would cause the maps thus produced to be of no scientific value. But, in the case of the concave inner surface of a sphere, the reverse is the case, the curvature itself being an essential condition of the very close accuracy of the photographic reproduction. A photograph taken from anywhere near the centre of the sphere would have every portion of the surface at right angles to the line of sight, and also at an equal distance from the camera. Hence there would be no distortion due to obliquity of the lateral portions, or errors of proportion owing to varying distances from the lens. We have, in fact, in a hollow sphere with the camera placed in the centre, the ideal conditions which alone render it possible to reproduce detailed maps on the surface of a sphere with accuracy of scale over the whole area. For producing maps of countries of considerable extent the camera would, therefore, be placed near the centre, but for maps of smaller areas on a larger scale, it might be brought much nearer without any perceptible error being introduced, while even at the smallest distances and the largest scale the distortion would always be less than if taken from a convex surface. It follows that only on a concave globular surface would it be worth the expense of modelling the earth in relief with the greatest attainable accuracy, and keeping it always abreast of the knowledge of the day, since only in this way could accurate photographic reproductions of any portions of it be readily obtained. For absolute accuracy of reduction the sensitive surface would have to be correspondingly concave, and this condition could probably be attained.

I will now point out how much more easily access can be provided to every part of the surface of a concave than to that of a convex globe. Of course, there must be a tower in the position of the polar axis. This would be as small in diameter as possible consistent with stability, and with affording space for a central lift; and it would be provided with a series of outside galleries supported on slender columns, at regular intervals, for affording views of the whole surface of the globe. This general inspection might be supplemented by binocular glasses with large fields of view and of varying powers, by means of which all the details of particular districts could be examined. For most visitors this would be sufficient; but access to the surface itself would be required, both for purposes of work upon it, for photographing limited areas at moderate distances, and for close study of details for special purposes. This might be provided without any permanent occupation of the space between the central tower and the modelled surface, in the following manner.

Outside the tower and close to the galleries will be fixed, at equal distances apart, a series of three or four circular rails, on which will rest by means of suitable projections and rollers, two vertical steel cylinders, exactly opposite to each other and reaching to within about ten feet of the top and bottom of the globe, with suitable means of causing them slowly to revolve. Attached to these will be two light drawbridges, which can be raised or depressed at will, and which also, when extended, will have a vertical sliding motion from the bottom to the top of the upright supports. The main body of this drawbridge would reach somewhat beyond the middle point from the tower to the globular surface, the remaining distance being spanned by a lighter extension sliding out from beneath the main bridge and supported by separate stays from the top of the tower. When not in use, the outer half would be drawn back and the whole construction raised up vertically against the galleries of the tower. The two bridges being opposite each other, and always being extended together, would exert no lateral strain upon the tower.

By means of this arrangement, which when not in use would leave the whole surface of the globe open to view, access could be had to every square foot of the surface, whether for purposes of work upon it or for close examination of its details; and, in comparison with the elaborate and costly system of access to the outer surface of a globe of equal size, involving about five miles of spirally ascending platform and more than a mile of stairs, besides the rotation of the huge globe itself, is so simple that its cost would certainly not be one-twentieth part that of the other system. At the same time, it would give access to any part of the surface far more rapidly, and even when in use would only obstruct the view of a very small fraction of the whole globe.

A Suggested Mode of Construction.

A few words may be added as to a mode of construction of the globe different from that suggested in the project of M. Reclus. It seems to me that simplicity and economy would be ensured by forming the globe of equal hexagonal cells of cast steel of such dimensions and form that when bolted together they would build up a perfect oblate spheroid of the size required. As the weight and strain upon the material would decrease from the bottom to the top, the thickness of the walls of the cells and of the requisite cross struts might diminish in due proportion, while the outside dimensions of all the cells were exactly alike. At the equator, and perhaps at one or two points below it, the globe might be encircled by broad steel belts to resist any deformation from the weight above. A very important matter, not mentioned by M. Reclus, would be the maintenance of a nearly uniform temperature, so as to avoid injury to the modelling of the interior by expansion and contraction. This might be secured by enclosing the globe in a thick outer covering of silicate or asbestos packing, or other non-conducting material, over which might be formed a smooth surface of some suitable cement, or papier-mache, on which the broad geographical features of the earth might be permanently delineated. With a sufficiency of hot-water pipes in and around the central tower, and efficient arrangements for ventilation, the whole structure might be kept at a nearly uniform temperature at all seasons.

It has now, I think, been shown that the only form of globe worth erecting on a large scale is one of which the inner surface is utilised for the detailed representation and accurate modelling of the geographical features of the earth's surface, while on the outside, either by painting or modelling or the two combined, all the grander features could be so represented as to be effectively seen at considerable distances. But as to the dimensions of such a globe there is room for much difference of opinion. I am myself disposed to think that the scale of 1이이이이이, proposed by M. Reclus, is much too large, and that for every scientific and educational purpose, and even as a popular exhibition, half that scale would be ample. The representation of minute details of topography due to human agency, and therefore both liable to change and of no scientific importance—such as roads, paths, houses, and enclosures-would be out of place on such a globe, except that towns and villages and main lines of communication might be unobtrusively indicated. And for adequately exhibiting every important physiographical feature-the varied undulations of the surface in all their modifications of character, rivers and streams with their cascades and rapids, their gorges and alluvial plains, lakes and tarns, swamps and peat-bogs, woods, forests, and scattered woodlands, pastures, sand dunes and deserts, and every other feature which characterises the earth's surface, a scale of 2이이이이이th, or even one of 비베이이이이th, would be quite sufficient. And when we consider the difficulty and expense of constructing any such globe, and the certainty that the experience gained during the first attempt would lead to improved methods should a larger one be deemed