In addition to this, the perpendicular[684] is not correct, supposing it true that Narbonne lies under almost the same parallel as Marseilles, and that this latter place is under the same parallel as Byzantium; which is the opinion of Hipparchus. Now the line drawn across the sea lies under the same parallel as the Strait [of the Pillars] and Rhodes; and the distance from Rhodes to Byzantium, which both lie under the same meridian, is estimated at about 5000 stadia; to which the above-mentioned perpendicular ought to be equal. But since they say that from the recess of the Galatic Gulf, the greatest distance across the sea from Europe to Libya is 5000 stadia, it seems to me that either there is some error in this statement, or that at this point Libya must incline very much to the north, and so come under the same parallel as the Pillars. Polybius is likewise mistaken in telling us that this said perpendicular terminates close to Sardinia; for instead of being close to Sardinia, it is far west thereof, having almost the whole of the sea of Liguria[685] between it and that island. Besides this he makes the length of the sea-coast too great; but this [error] is not so considerable [as the two preceding]. 4. After this Polybius proceeds to set right the mistakes of Eratosthenes. In this he is sometimes successful; at others his corrections are for the worse. For example, Eratosthenes gives 300 stadia from Ithaca to Corcyra; Polybius makes it above 900. From Epidamnus to Thessalonica Eratosthenes allows 900 stadia; Polybius says above 2000. In these instances he is correct. But where Eratosthenes states that from Marseilles to the Pillars there are 7000 stadia, and from the Pyrenees [to the same place] 6000, and Polybius alters this to more than 9000 from Marseilles, and little less than 8000 from the Pyrenees,[686] he is quite mistaken, and not so near to the truth as Eratosthenes. For all are now agreed that, barring the indirectness of the roads, the whole length of Iberia is not more than 6000 stadia[687] from the Pyrenees to its western limits; notwithstanding Polybius gives 8000 stadia for the length of the river Tagus, from its source to its outlets, and this in a straight line without any reference to its sinuosities, which in fact never enter into the geographical estimate, although the sources of the Tagus are above 1000 stadia from the Pyrenees. His remark is quite correct, that Eratosthenes knew little about Iberia, and on this account sometimes makes conflicting statements concerning it. He tells us, for example, that the portion of this country situated on the sea-coast as far as Gades is inhabited by Galatæ,[688] who possess western Europe as far as Gades; nevertheless, in his account of Iberia he seems quite to have forgotten this, and makes no mention of these Galatæ whatever.