But the celebrated Catalan map of 1375[100] above mentioned introduced a significant novelty, converting the disc into an annulus of land—of course, still circular—surrounding a circular body of water dotted with islets ([Fig. 5]). The preferred explanation thus far advanced connects these islets with the Seven Cities of Portuguese and Spanish legend.[101] But there seem to be nine islands, not seven, and it is not clear what necessary relation exists between isles and cities nor whence the idea is derived of the central lake or sea as a background. Moreover, the Island of the Seven Cities was most often identified with Antillia far to the south, and there seems no warrant for identification with Brazil. All considered, this explanation seems arbitrary, inadequate, and unconvincing.

Fig. 5—Section of the Catalan map of 1375 showing the islands of Mayda and Brazil. (After Nordenskiöld’s photographic facsimile.)

The same ring form with inclosed water and islets is repeated by a map of the next century copied by Kretschmer.[102] It varies only by showing just seven islets, if we may rely for this detail on his handmade copy.

Possible Identification with the Gulf of St. Lawrence Region

Now, in all the Atlantic Ocean and its shores there is one region, and one only, which thus incloses a sheet of water having islands in its expanse, and this region lies in the very direction indicated on the old maps for Brazil. I allude to the projecting elbow of northeastern North America, which most nearly approaches Europe and has Cape Race for its apex. Its front is made up of Newfoundland and Cape Breton Island. The remainder of the circuit is made up of what we now call southern Labrador, a portion of eastern Quebec province, New Brunswick, and Nova Scotia. This irregular ring of territory incloses the great Gulf of St. Lawrence, which has within it the Magdalens, Brion’s Island, and some smaller islets, not to include the relatively large Anticosti and Prince Edward. It has two rather narrow channels of communication with the ocean, which might readily fail to impress greatly an observer whose chief mental picture would be the great land-surrounded, island-dotted expanse of water. The surrounding land would itself almost certainly be regarded as insular, for there was a strong tendency to picture everything west of Europe in that way, even long after the time when most of these maps were made. Even when Cartier[103] in 1535 ascended the St. Lawrence River it was in the hope of coming out again on the open sea—a hope that implies the very conception of an insular mass inclosing the gulf, not differing essentially from the showing of the Catalan map of 1375. The number of the islands is immaterial. We may picture the Catalan map-maker dotting them in from vague report as impartially as the far better known Lake Corrib is besprinkled with islands in most of the old maps—far more plentifully than the facts give warrant.

But it would seem that other observers were more impressed by the separation of Newfoundland, due to the Straits of Belle Isle and Cabot and the waterway (of the gulf) connecting them behind the great island. As a rule the maps presenting Brazil in this divided way adhere to the accepted latitude, which does not differ appreciably from that of the St. Lawrence Gulf region. The dividing passage, mainly from north to south but slightly curved at the ends which join the ocean, corresponds fairly well with the facts. The maps of Prunes, 1553[104] ([Fig. 12]), and Olives, 1568,[105] may be cited as instances of this divided form of Brazil. No explanation seems yet to have been offered except Nansen’s,[106] that the dividing channel represents “the river of death (Styx),” and Westropp’s,[107] that it may be owing to mistaken copying of a name space or label on some older map. But the former lacks any better basis than conjectured fancy and the latter is refuted by the position of the channel on most maps and by the general aspect of the delineation. As a matter of fact, the showing of most of the maps differs in little more than proportions from that of Gastaldi illustrating Ramusio in 1550,[108] when the Gulf of St. Lawrence was fairly well known to many, but appears as a rather narrow channel behind a broken-up Newfoundland, extending from the Strait of Belle Isle to the Strait of Cabot. As in the much older map referred to, the delineation of Gastaldi is perhaps to be explained by concentration of attention on the waterway and the ignoring of the wider parts of the expanse. Absolute demonstration of the causes of the divided Brazil of some maps and the ring of land inclosing an island-dotted body of water in others is, of course, impossible; but we can show that in the designated direction there is a region presenting both of these unusual features, so that one of the visitors might well be especially taken up with one set of characteristics, another with the other set, and might depict the region accordingly. This is the more probable because the region was peculiarly exposed to accidental or intentional discovery from the west of the British islands and is known, in fact, to have been the first to be reached therefrom of all North America in times of historic record.

It must not be supposed that Brazil was always thought of as relatively near Europe. Nicolay in 1560[109] ([Fig. 6]) and Zaltieri in 1566[110] prepared maps which show a Brazil Island in distinctly American waters, practically forming part of the archipelago into which Newfoundland was supposed to be divided, or at least lying between it and the Grand Banks. These presentations no doubt may have been suggested by American discoveries and later theories, especially as no navigator had been able to find Brazil at any point nearer Europe; but again they may be at least partly due to surviving early traditions of the great distance westward at which this island lay. The Brazil of Nicolay and Zaltieri is, to be sure, a very small affair; but their maps were made about two and a half centuries after the earliest one which shows this island—ample time for many misconceptions to creep in. Their only value is in their illustration of locality.

The Catalan Map of about 1480

More important in every way is a Catalan map ([Fig. 7]) preserved in Milan and reproduced by Nordenskiöld in 1892,[111] but since copied partly by Nansen, by Westropp, and by others. It belongs to the fifteenth century—perhaps about 1480—and deserves clearly to rank as the only map before Columbus, thus far reported, which shows a part of North America other than Greenland. The latter had long before appeared in the well-known map of Claudius Clavus, 1427[112] ([Fig. 16]), no doubt on the faith of the early Norse narratives and subsequent commercial intercourse, for the Norse Greenland colony is known to have existed in 1410 and probably did not die out entirely until much later. The Catalan map of about 1480 shows Greenland also as a great northwestern land mass beyond Iceland, identifying it by name as Illa Verde (Green Island). But just south, or west of south, of this Greenland at a slight interval and southwest of Iceland is drawn and named a large Brazil of the conventional circular disc form. Its position is that of Labrador, or perhaps Newfoundland, as it would naturally have been understood and reported by the Norse explorers. It can be nothing but one or both of these regions of America with perhaps neighboring lands.