280 STRABO. CASAUB. 536. parts, the projections in one correspond so exactly with the hollows in the other that they might even be fitted together, so here I have seen the rocks at the distance of two or three plethra, overhanging the river on each side, and nearly reach- ing to the summit of the mountain, with hollows on one side answering to projections on the other. The bed between (the mountains) is entirely rock ; it has a deep and very narrow fissure through the middle, so that a dog and a hare might leap across it. This is the channel of the river ; it is full to the margin, and in breadth resembles a canal. 1 But on ac- count of the winding of its course, the great contraction of the stream, and the depth of the ravine, a noise, like that of thunder, strikes at a distance on the ears of those who ap- proach it. In passing out through the mountains, it brings down from Cataonia, and from the Cilician plains, so great a quantity of alluvial soil to the sea, that an oracle to the follow- ing effect is reported to have been uttered respecting it : " The time will come, when Pyramus, with its deep whirlpools, by ad- vancing on the sea-shore, will reach the sacred Cyprus." Something similar to this takes place in Egypt. The Nile is continually converting the sea into continent by an accu- mulation of earth ; accordingly Herodotus calls Egypt a gift of the river, and Homer says, that the Pharos was formerly out at sea, not as it is at present connected with the main- land of Egypt. 5. [The third 2 in rank is the Dacian priesthood of Jupiter, inferior to this, but still of importance.] There is at this place a body of salt water, having the circumference of a con- siderable lake. It is shut in by lofty and perpendicular hills, so that the descent is by steps. The water it is said does not increase in quantity, nor has it anywhere an apparent outlet. 6. Neither the plain of the Cataonians nor Melitene have any city, but strongholds upon the mountains, as Azamora, and Dastarcum, round which runs the river CarmalasA There is also the temple of the Cataonian Apollo, which is vener- ' l The reading is doubtful. 8 The passage is corrupt. Groskurd proposes Asbamean in place of Dacian, mention being made of a temple of Asbamean Jove in Amm. Marcell. xxiii. 6. Kramer also suggests the transposition of this sentence to the end of 6. 3 Probably the Kermel-su, a branch of the Pyramus.