Strabo Geography 1.4.5

[5] Eratosthenes being mistaken as to the breadth [of the habitable earth], is necessarily wrong as to its length. The most accurate observers, both ancient and modern, agree that the known length of the habitable earth is more than twice its breadth. Its length I take to be from the [eastern] extremity of India ¹ to the [westernmost] point of Spain; ² and its breadth from [the south of] Ethiopia to the latitude of Ierne. Eratosthenes, as we have said, reckoning its breadth from the extremity of Ethiopia to Thule, was forced to extend its length beyond the true limits, that he might make it more than twice as long as the breadth he had assigned to it. He says that India, measured where it is narrowest, ³ is 16,000 stadia to the river Indus. If measured from its most prominent capes it extends 3000 more. ⁴ Thence to the Caspian Gates, 14,000. From the Caspian Gates to the Euphrates, ⁵ 10,000. From the Euphrates to the Nile, 5000. ⁶ Thence to the Canopic ⁷ mouth, 1300. From the Canopic mouth to Carthage, 13,500. From thence to the Pillars at least 8000. Which make in all 70,800 stadia. To these [he says] should be added the curvature of Europe beyond the Pillars of Hercules, fronting the Iberians, and inclining west, not less than 3000 stadia, and the headlands, including that of the Ostimii, named Cabaeum, ⁸ and the adjoining islands, the last of which, named Uxisama, ⁹ is distant, according to Pytheas, a three days’ sail. But he added nothing to its length by enumerating these last, viz. the headlands, including that of the Ostimii, the island of Uxisama, and the rest; they are not situated so as affect the length of the earth, for they all lie to the north, and belong to Keltica, not to Iberia; indeed it seems but an invention of Pytheas. Lastly, to fall in with the general opinion that the breadth ought not ¹⁰ to exceed half the length, he adds to the stated measure of its length 2000 stadia west, and as many east.