In the interval there were one or two writers from whose work something is to be gathered of Greek geographical knowledge and theory about the middle of the fourth century. The philosopher Plato (427–347 B.C.) may be referred to here in connection with his story, based on an Egyptian tradition, of the great island of Atlantis, that land which plays so important a part in later mythical geography. In the Egyptian story it lay just beyond the Pillars of Hercules, and adjacent to it was an archipelago. This would aid its later identification with the Canaries, though it came also to be connected with America and other known lands besides, as well as giving name to an island in the Atlantic Ocean, the disproof of whose existence may almost be called modern. From Plato’s account of Atlantis as the home of a powerful people who in early times invaded the Mediterranean lands, it has also been sought to associate the tradition with Crete at the period of the Ægean civilization mentioned in the first chapter. Some fragments exist of the writings of the historian Ephorus of Cyme in Æolis (_c._ 400–330 B.C.). He seems to have endeavoured to cover the whole field of the world as known to the Greeks, and conceived the four most distant regions of the earth to be occupied on the east by Indians, on the south by Ethiopians, on the north by Scythians, and on the west by Celts. The last he considered as occupying all Spain, as well as Gaul. Strabo (p. 24) commended his geographical work and his skill in separating myth from history. A document of this period is the Periplus, already referred to as known under the name of Scylax. This class of work became more and more common as navigation developed, and corresponded in some measure to the modern Admiralty guide or pilot. The Periplus is confined mostly to the regions known to the Greeks bordering the Mediterranean. From the Pillars of Hercules the writer follows the north coast eastwards, including the Adriatic and the Euxine as far as the mouth of the Tanais, which he regards as the continental boundary. He then follows the Levantine coast, the north African coast westward, and the west African coast as far as the island of Cerne. He incidentally makes what is regarded as the earliest extant mention of Rome; but his notions of rivers and other features away from the coast are generally erroneous.