12. Eratosthenes, when he is speaking of the lakes near Arabia, says, that the water, when it cannot find an outlet, opens passages under-ground, and is conveyed through these as far as the Cœle-Syrians,[495] it is also compressed and forced into the parts near Rhinocolura[496] and Mount Casius,[497] and there forms lakes and deep pits.[498] But I know not whether this is probable. For the overflowings of the water of the Euphrates, which form the lakes and marshes near Arabia, are near the Persian Sea. But the isthmus which separates them is neither large nor rocky, so that it was more probable that the water forced its way in this direction into the sea, either under the ground, or across the surface, than that it traversed so dry and parched a soil for more than 6000 stadia; particularly, when we observe, situated midway in this course, Libanus, Antilibanus, and Mount Casius.[499] [CAS. 742] Such, then, are the accounts of Eratosthenes and Aristobulus. 13. But Polycleitus says, that the Euphrates does not overflow its banks, because its course is through large plains; that of the mountains (from which it is supplied), some are distant 2000, and the Cossæan mountains scarcely 1000 stadia, that they are not very high, nor covered with snow to a great depth, and therefore do not occasion the snow to melt in great masses, for the most elevated mountains are in the northern parts above Ecbatana; towards the south they are divided, spread out, and are much lower; the Tigris also receives the greater part of the water [which comes down from them], and thus overflows its banks.[500] The last assertion is evidently absurd, because the Tigris descends into the same plains (as the Euphrates); and the above-mentioned mountains are not of the same height, the northern being more elevated, the southern extending in breadth, but are of a lower altitude. The quantity of snow is not, however, to be estimated by altitude only, but by aspect. The same mountain has more snow on the northern than on the southern side, and the snow continues longer on the former than on the latter. As the Tigris therefore receives from the most southern parts of Armenia, which are near Babylon, the water of the melted snow, of which there is no great quantity, since it comes from the southern side, it should overflow in a less degree than the Euphrates, which receives the water from both parts (northern and southern); and not from a single mountain only, but from many, as I have mentioned in the description of Armenia. To this we must add the length of the river, the large tract of country which it traverses in the Greater and in the Lesser Armenia, the large space it takes in its course in passing out of the Lesser Armenia and Cappadocia, after issuing out of the Taurus in its way to Thapsacus (forming the boundary between Syria below and Mesopotamia), and the large remaining portion of country as far as Babylon and to its mouth, a course in all of 36,000 stadia.