Pelusium.[598] Worthless, too, is the argument in connexion with this, being the inference from a proposition not admitted; for Eratosthenes never asserts that from Babylon to the meridian of the Caspian Gates is a distance of 4800 stadia. We have shown that Hipparchus deduces this from data not admitted by Eratosthenes; but desirous to controvert every thing advanced by that writer, he assumes that from Babylon to the line drawn from the Caspian Gates to the mountains of Carmania, according to Eratosthenes’ description, there are above 9000 stadia, and from thence draws his conclusions. 37. Eratosthenes[599] cannot, therefore, be found fault with on these grounds; what may be objected against him is as follows. When you wish to give a general outline of size and configuration, you should devise for yourself some rule which may be adhered to more or less. After having laid down that the breadth of the space occupied by the mountains which run in a direction due east, as well as by the sea which reaches to the Pillars of Hercules, is 3000 stadia, would you pretend to estimate different lines, which you may draw within the breadth of that space, as one and the same line? We should be more willing to grant you the power of doing so with respect to the lines which run parallel to that space than with those which fall upon it; and among these latter, rather with respect to those which fall within it than to those which extend without it; and also rather for those which, in regard to the shortness of their extent, would not pass out of the said space than for those which would. And again, rather for lines of some considerable length than for any thing very short, for the inequality of lengths is less perceptible in great extents than the difference of configuration. For example, if you give 3000 stadia for the breadth at the Taurus, as well as for the sea which extends to the Pillars of Hercules, you will form a parallelogram entirely enclosing both the mountains of the Taurus and the sea; if you divide it in its length into several other parallelograms, and draw first the diagonal of the great parallelogram, and next that of each smaller parallelogram, surely the diagonal of the great parallelogram will be regarded as a line more nearly parallel and equal to the side forming the length of that figure than the diagonal of any of the smaller parallelograms: and the more your lesser parallelograms should be multiplied, the more will this become evident. Certainly, it is in great figures that the obliquity of the diagonal and its difference from the side forming the length are the less perceptible, so that you would have but little scruple in taking the diagonal as the length of the figure. But if you draw the diagonal more inclined, so that it falls beyond both sides, or at least beyond one of the sides, then will this no longer be the case; and this is the sense in which we have observed, that when you attempted to draw even in a