The next day Columbus left San Salvador and sailed westward along the coast. That was the direction in which, according to the natives of Guanahani, the mainland and the capital of the King or the Grand Khan were to be found. That, too, was the direction in which Mangi and Cathay were to be found according to the map of Toscanelli, assuming Cuba to be Cipango: which Columbus at this stage of his enterprise confidently believed. Of the researches of the great voyager along the Cuban coast we have a detailed account in his journal. Unfortunately, there is no certain means of identifying the points at which he landed. They are described as being so many leagues from his starting point, San Salvador; wherefore it is obvious that all depends upon the identity of the latter. Yet it seems to me that his account of his coastwise explorations strongly confirms the theory that his San Salvador was Port Naranjo and not Nuevitas. For we are told that six leagues westward he found a cape or point of land extending toward the northwest; ten leagues further another point, extending toward the east; one league further a small river, which he called the Rio de la Luna; and beyond it another much larger river, which he called the Rio de Mares. This latter river had for its estuary a broad basin resembling a lake, and its entrance was marked by two round mountains on the one side and a lofty promontory on the other. Now, making reasonable allowance for lack of accuracy in measurements and for discrepancies in descriptions, this account may readily be applied to the coast westward from Port Naranjo to Nuevitas, while it is altogether inapplicable to the coast westward from Nuevitas. For a score of leagues westward from Naranjo there are capes and mountains and rivers, and there is more than one river with precisely such a lagoon-like estuary as that which Columbus found at his Rio de Mares. Indeed, Port Padre, with its extensive lagoon into which several rivers flow, or Port Manati, with the Cramal and Yarigua rivers, might either of them be identified, in approximate distance and in topography, with the Rio de Mares. On the other hand, if we were to assume Nuevitas to have been the starting point, what should we find? Either he must have been skirting the outer side of the Sabinal and Romano keys, and Guajaba Island, which do not at all coincide with the description given, or he must have been navigating the great littoral lagoon between those keys and the mainland of Cuba; in which latter case it is to be observed that that part of the Cuban coast does not correspond with his description, and that it is certainly extraordinary that he made no mention of his voyage having been in what is practically an inland sea. That he could have passed in through the Nuevitas Channel, or the Carebelas Channel, or the Guajaba Channel, without observing and remarking upon Sabinal Key, Guajaba Island, or Romano Key, is simply not supposable. Such a