by tribute, after deducting the wages of the soldiers necessary for guarding the island and exacting the taxes. And the other islands adjacent to this would be still more unproductive.
If, then, to the distance between Rhodes and the Dnieper be added four thousand stadia north of the latter place, the whole would come to 12,700 stadia; and since from Rhodes to the southern limit of the habitable earth there are 16,600 stadia, its total breadth from north to south would be under 30,000 stadia.[728] Its length from west to east is stated at 70,000 stadia, the distance being measured from the extremities of Iberia to those of India, partly over the land and partly across the sea. That this length is contained within the quadrilateral aforesaid, is proved by the proportion borne by these parallels to the equator. Thus the length of the habitable earth is above twice its breadth. It has been compared in figure to a chlamys, or soldier’s cloak, because if every part be carefully examined, it will be found that its breadth is greatly diminished towards the extremities, especially in the west.
We have now been tracing upon a spherical surface the region which we state to be occupied by the habitable earth; and whoever would represent the real earth as near as possible by artificial means, should make a globe like that of Crates, and upon this describe the quadrilateral within which his chart of geography is to be placed. For this purpose, however, a large globe is necessary, since the section mentioned, though but a very small portion of the entire sphere, must be capable of properly containing all the regions of the habitable earth, and presenting an accurate view of them to all those who wish to consult it. Any one who is able will certainly do well to obtain such a globe. But it should have a diameter of not less than ten feet: those who cannot obtain a globe of this size, or one nearly as large, had better draw their chart on a plane-surface, of not less than seven feet. Draw straight lines, some parallel, for the parallels [of latitude], and others at right angles to these; we may easily imagine how the eye can transfer the figure and extent [of these lines] from a plane-surface to one that is spherical. What we have just observed of the circles in general, may be said with equal truth touching the oblique circles. On the globe it is true that the meridians of each country passing the pole have a tendency to unite in a single point, nevertheless on the plane-surface of the map, there would be no advantage if the right lines alone which should represent the meridians were drawn slightly to converge. The necessity for such a proceeding would scarcely ever be really felt. Even on our globe itself[729] the tendency of those meridians (which are transferred to the map as right lines) to converge is not much, nor any thing near so obvious as their circular tendency.