Chapter III The Piri Re’is Map in Detail
In undertaking a detailed examination of the Piri Re’is Map of 1513, I shall break down the map into sections representing originally separate source maps of smaller areas, which appear to have been combined in a general map by the Greek geographers of the School of Alexandria. 
With regard to each of the source maps, which I shall refer to as “component maps,” since they are the parts of the whole, I will identify such geographical points as are evident in themselves, or are rendered plausible by their position on the trigonometric grid, and will find their errors of location.
Since in some cases the component maps were not correctly placed on the general map, we have two sorts of errors: those due to mistakes in compilation of the local maps into the general map and those due to mistakes in the original component maps. These can be distinguished because if a component map is misplaced, all the features of that map will be misplaced in the same direction and by the same amount. If the general error is discovered and corrected, then the remaining errors will be errors of the original local maps. We have discovered that in most cases the errors on the Piri Re’is Map are due to mistakes in the compilation of the world map, presumably in Alexandrian times, since it appears, as we shall see, that Piri Re’is could not have put them together at all. The component maps, coming from a far greater antiquity, were far more accurate. The Piri Re’is Map appears, therefore, to be evidence of a decline of science from remote antiquity to classical times.
1. The western coasts of Africa and Europe, from Cape Palmas to Brest, including the North Atlantic islands (Cape Verdes, Canaries, Azores, and Madeira) and some islands of the South Atlantic.
Longitudes, as well as latitudes, along the coasts are seen to be remarkably accurate (see Table 1). The accuracy extends also to the North Atlantic island groups as a whole, with an exception in the case of Madeira.
The accuracy of longitude along the coast of Africa, where it is greatest, might be attributed ·simply to our assumptions as to the center and radius of the projection, but for two considerations. First, the assumption regarding the length of the radius (that is, the length of the degree) was not reached with reference to the coast of Africa, but with reference to the width of the Atlantic and the longitude of the coast of South America. It will be seen from our map (Fig. 18) and from Table 1 that both these coasts, separated by the width of the Atlantic, are in approximately correct relative longitude with reference to the center of the projection on the meridian of Alexandria. This seems to mean that the original mapmaker must have found correct relative longitude across Africa and across the Atlantic from the meridian of Alexandria to Brazil.
It is also important that most of the islands are in equally correct longitude. The picture that seems to emerge, therefore, is one of a scientific achievement far beyond the capacities of the navigators and mapmakers of the Renaissance, of any period of the Middle Ages, of the Arab geographers, or of the known geographers of ancient times. It appears to demonstrate the survival of a cartographic tradition that could hardly have come to us except through some such people as the Phoenicians or the Minoans, the great sea peoples who long preceded the Greeks but passed down to them their maritime lore.
The accuracy of placement of the islands suggests that they may have been found on the ancient source map used by Piri Re’is . The “discoveries” and mapping of these islands by the Arabs and Portuguese in the 15th Century may not, then, have been genuine discoveries. It is possible that the 15th Century sailors really found these islands as the result of accidental circumstances (being blown off course, etc.). On the other hand, nothing excludes the possibility that source maps used by Piri Re’is, dating from ancient times, were known in some form to people in Europe. Possibly some of the early voyages to some of these islands, particularly the Azores, were undertaken to confirm the accuracy of the old maps. It is hardly, if at all, possible that these 15th Century navigators could have found correct longitude for the islands. All they had to go by were rough guesses of courses run, based on the direction and force of the wind, and the estimated speed of their ships. Such estimates were apt to be thrown off by the action of ocean currents and by lateral drift when the ship was trying to make to windward.
A good description of the problem of finding position at sea is given by a 16th Century writer quoted by Admiral Morison in his Admiral of the Ocean Sea:
“O how God in His omnipotence can have placed this subtle and so important art of navigation in wits so dull and hands so clumsy as those of these pilots! And to see them inquire, one of the other, ‘how many degrees hath your honor found?’ One says ‘sixteen,’ another ‘a scant twenty’ and another ‘thirteen and a half.’ Presently they ask, ‘How doth your honor find himself with respect to the land?’ One says, ‘I find myself forty leagues from land,’ another ‘I say 150,’ another says ‘I find myself this morning 92 leagues away.’ And be it three or three hundred nobody agrees with anybody else, or with the truth.” (140:321-322)
In the days of Piri Re’is no instruments existed by which the navigator at sea could find his longitude. Such an instrument did not appear for another 250 years, when the chronometer was developed in the reign of George III. It does not seem possible to explain the accuracy of longitude on the Piri Re’is Map in terms of navigational science in the time of Piri Re’is.
The case for latitude is somewhat different. Latitude could be determined in the 15th and 16th Centuries by astronomical observations. However, observations taken by trained people with proper equipment were one thing, and observations taken by explorers were quite another. Morison says that Columbus made serious mistakes in finding latitude. Speaking of the First Voyage he says: ” … We have only three latitudes (all wrong) and no longitude for the entire voyage” (140:157). He describes one of Columbus’ attempts to find his latitude as follows:
On the night of Nov. 2 (1492) two days before the full moon, he endeavored to establish his position by taking the altitude of the North Star with his wooden quadrant. After applying the slight correction he decided that Puerto Gibara, actually in Lat. 21° 06′ N, was in 42° N, the Latitude of Cape Cod (140:258).
For a long time after the four voyages of Columbus we find the latitudes of Cuba and Haiti wrong on the maps of the time. Almost all mapmakers put the islands above rather than below the Tropic of Cancer. 
To return to the problem of longitude, Morison remarks that the only method of finding longitude known in the 16th Century was by the timing of eclipses, but that nobody was successful in applying it. He says:
The only known method of ascertaining longitude in Columbus’ day was by timing an eclipse. Regiomontanus’s Ephemerides and Zacuto’s Almanach Perpetuum gave the predicted hours of total eclipse at Nuremberg and Salamanca respectively, and if you compared those with the observed hour of the eclipse, wherever you were, and multiplied by 15 to convert time into arc (1 hour of time = 15° of Longitude) there was your longitude west of the Almanach maker’s meridian. Sounds simple enough, but Columbus, with two opportunities (1494 and 1503) muffed both, as did almost everyone else for a century. (140:185-186)
Morison describes in an interesting manner the failure of an attempt to find the longitude of Mexico City in 1541 (twenty-eight years after Piri Re’is drew his map):
At Mexico City in 1541 a mighty effort was made by the intelligentsia to de termine the longitude of the place by timing two eclipses of the moon. The imposing result was 8h 2m 32s (= 120° 38′ west of Toledo) but the correct difference of longitude between the two places is 95° 12′, so the Mexican savants made an error of 25½°, or 1450 miles! Even in the 18th Century Pere Labat, the earliest writer (to my knowledge) who gives the position of Hispaniola correctly, adds this caveat: “I only report the longitude to warn the reader that nothing is more uncertain, and that no method used up to the present to find longitude has produced anything fixed and certain” (140:186).
With this backwardness of the 16th Century science of navigation, I cannot see how the accuracy of the Piri Re’is Map can be explained, either as to latitude or longitude.  Figures 19-24 illustrate the poor qualities of the maps that were drawn at this time.
With regard to latitude there are several complications in the Piri Re’is Map. Its history—that is, the history of the source map used by him for the Atlantic coasts—must have been a long one, for several different stages of mapmaking are reflected in it.
We thought at first that the horizontal line running through Point III represented the equator of the projection. This would involve a design as in Fig. 16, with the line from the center to the point of intersection of this horizontal line with the perimeter of the circle as the determining line; in reference to this line all the other radii were laid out at angles of 22½° to the north and south. Our results indicate that something like this was done; that is, that the source map we are discussing (embracing Africa, Europe, and some of the islands) was at some time during the history of the map placed on the projection in this way. It might have been done visually, or empirically, so to speak, simply by placing the African coast of Guinea at the correct distance north of this central line, which was taken for the equator. This was an error, with reference to the mathematical projection, the equator of which in fact lies nearly five degrees north of this line, as shown in Fig. 17.
We concluded that an error was made here because some of the Piri Re’is Map (to be discussed below) in fact is in line with the equator of the trigonometric projection. It seems probable that we are dealing here with the work of different people who redrew the map at different times with different ideas. The large wind roses in the North and South Atlantic Oceans, apparently identifying the Tropics, may have been superimposed on the map by the geographers who made the error in placing their source map. Figure 17 shows that according to the trigonometric projection the northern projection point lies on the Tropic of Cancer, while the southern one does not coincide closely with the Tropic of Capricorn; the altered projection shows the northern projection point several degrees off the Tropic of Cancer while the southern projection point lies fairly close to the Tropic of Capricorn. The geometry of the projection makes it impossible for both projection points to coincide with the respective Tropics by either arrangement. It may be noted, however, that whoever converted the latitude from the trigonometric projection to make the equator coincide with the central projection point on the perimeter of the circle did not alter the longitudes of the coasts, which are close to correct by either system. We may conclude that the whole source map embracing Europe and Africa was simply shoved due North about five degrees. This would have produced some minor errors in longitude, but of too small a scale to be noticeable to us. Piri Re’is himself may have made the change.
Note: Figures 19-24: This selection of maps drawn in the Age of Discovery illustrates the weaknesses of the cartographic science of the period. So far as relative distances, land shapes, and particularly longitude are concerned these maps are much inferior to the Piri Re’is Map. None of these maps suggests the use of trigonometry.
Some of the islands included on this component map may have been added by Piri Re’is on the basis of accounts of recent explorations. Those islands which appear to be seriously misplaced in latitude or longitude may have been such additions. On the other hand, it may be the case with some (as perhaps with Madeira and Fernando da Naronha) that their errors in latitude may be due to having been ignored when the component map was shifted northward.
All things considered, I am inclined to believe that Piri Re’is himself shifted the component map northward to make it agree with his view that the line through the central projection point (Point III) must represent the equator. If he did this, it indicates a good knowledge on his part of the latitudes of the Guinea coast. As we shall see, this knowledge was available as the result of the Portuguese exploration of that coast in the 15th Century. This exploration resulted in some careful scientific observations of latitude such as were not available from accounts of the American explorations.
To sum up, then, this part of the Piri Re’is Map suggests that Piri Re’is had a source map of Africa, Europe, and the Atlantic islands, based on maps probably drawn originally on some sort of trigonometric projection adjusted to the curvature of the earth. By default of any alternative, we seem forced to ascribe the origin of this part of the map to a pre-Hellenic people—not to Renaissance or Medieval cartographers, and not to the Arabs, who were just as badly off as everybody else with respect to longitude, and not to the Greeks either. The trigonometry of the projection (or rather its information on the size of the earth) suggests the work of Alexandrian geographers, but the evident knowledge of longitude implies a people unknown to us, a nation of seafarers, with instruments for finding longitude undreamed of by the Greeks, and, so far as we know, not possessed by the Phoenicians, either.
2. A special projection in the Caribbean, including part of the coast of South America.
The Caribbean part of the Piri Re’is Map offered us the greatest difficulties. It seemed entirely out of line. The coast appeared to trend the wrong way. It looked at first like some of the very worst mapmaking imaginable. From our studies of the portolan projection, however, I was prepared to accept the possibility of there being more than one North on this map. Estes had pointed out that the portolan design permitted a change of North from one part of a map to another, if and when it became desirable to move from one square, or grid, to one of the others that the design made possible.
I was looking at the map one day when I suddenly found that by twisting my head to one side, I could make some sense of the Caribbean section. I saw that there was indeed another North in this area. I assumed to start with that it might be integrated with the mathematics of the world projection. It had already become evident to us that it was theoretically possible to take any one of the map’s projection points, whose positions were now known, and repeat the portolan design by drawing a circle with this point as a center, and then constructing a grid within it exactly as with the world projection. This would be a satellite grid, and any North line could be chosen to suit the mapmaker’s convenience.
To solve this problem it was necessary to locate a North line, that is, a prime meridian. By identifying on the map a number of geographical localities which lay at the same latitude on the modern map of the Caribbean, I drew a rough parallel. I then looked for—and found—a line on the Piri Re’is projection at right angles to this. The line I found came down from Projection Point I at the top of the map and bisected what looked like the Peninsula of Yucatan. The angle of this line to the meridians of the main part of the map was 78¾°; this meant that it lacked one compass point (11¼°) of being at a right angle to the north of the rest of the map.
Gradually it became possible to extend the mathematical system of the whole projection to this part of the map. The common point was Projection Point I, which we had located at 51.4° North Latitude and 36.9° West Longitude. We assumed this point to be at the same latitude in both parts of the map. Since the length of the degree was (by assumption) the same, we could lay out parallels of latitude at five-degree intervals down to zero, which was, then, the equator of this special projection. Latitude was thus integrated mathematically with the world projection.
We found, after a number of tests, that the Ptolemaic spacing of parallels had also been applied in this component map.
The longitude problem presented much greater difficulty. Our first solutions were largely guesswork. Finally, the problem was solved by dropping a line, from the intersection of the prime meridian of our Caribbean section with the equator of that section, to the bottom of the map, where it intersected the register of longitude of the main grid extended westward. The longitude of the point of intersection at the bottom of the map became the longitude of our local prime meridian, and thus both the latitudes and longitudes of the Caribbean section were determined (See Fig. 18).
Now, if the reader will visualize the entire Caribbean grid as suspended from Point I, hanging down with no place to put its feet, and then swung through an arc of 78¾°, he should get the idea. Since the swing of the projection is so exact, and since, as the tables show (Table 1), the latitudes and longitudes of the identifiable places around the Caribbean are remarkably accurate, we are sure that the accuracy of this special projection is not a coincidence.
Perhaps the reader may wonder at the mapmaker’s reason for resorting to this device. The only answer I am able to suggest—and it is only a guess—is that he may have had ancient maps (maps ancient then) of the Caribbean area, with ample notations of latitude and longitude, but drawn, like a modern map, on some sort of spherical projection. Perhaps because he was unfamiliar with spherical trigonometry, he may have been forced to treat the round surface of the earth as a series of flat planes. He therefore had to have different norths in areas that were too far removed from each other in longitude. He was clever enough to work out a scheme by which he could preserve the accuracy of latitude and longitude in the Caribbean. He had to find just the right angle for North that would achieve this purpose, and he did so. But it is probable that he did not achieve the full accuracy of his ancient sources.
Strong support for this hypothesis is provided by a comparison of the Piri Re’is Map with a modern map of the world drawn on a polar equidistant projection (see Figs. 25, 26, 27). This map was drawn for the use of the Air Force during World War II. It was centered at Cairo, Egypt, because an important U.S. air base was located there. Since Cairo is not far from the center of the Piri Re’is world projection, this modern map gives us a good idea of what the world would look like on a projection of this kind centered on Egypt. If we look at Cuba on this equidistant map, we notice that it appears to run at right angles to a latitude line drawn through Cairo. In other words, if we regard the map as representing a flat surface, then Cuba runs north and south, just as it seems to run with reference to the main projection of the Piri Re’is Map. Furthermore, in both cases we see Cuba much too far north.
How is this to be explained? What else can we conclude but that the mapmaker, confronted by a spherical projection he did not understand, had to translate his geographical data (latitudes and longitudes of places in the Caribbean) into terms of a flat surface? This contains the implication, of course, that spherical trigonometry must have been known ages before its supposed invention by Hipparchus in the second century B.C. It also raises another question: How did it happen that a world map, apparently drawn ages before Hipparchus, was centered on Egypt? Can we ascribe such advanced knowledge to the early Egyptians? If not, to whom? I do not apologize for raising such thorny questions. They are at present unanswerable. Perhaps they will be answered some day.
To sum up, then, our mapmaker was faced with the problem of indicating True North both for the Atlantic and for the Caribbean area, which extends much farther west. Since the portolan projection is a rectangular projection and the earth is round, it is evident that you cannot extend it through many degrees of longitude without getting to a place where the meridians will not point north at all. The geometrical scheme of the portolan projection, with several possible Norths, was the only way to solve this problem. But there had to be mathematical calculations. Only by trigonometry could the correct angle for the Caribbean prime meridian be found.
The peculiar projection for the Caribbean area permits some conclusions as to the probable history of the map as a whole. In the first place it is clear that Piri Re’is could not have constructed this part of his world map. Such a thing as two Norths on the same map was unheard of in the Renaissance. To Piri Re’is, the idea of changing the direction of north in the middle of the ocean would be lunacy, and all the mapmakers of the age would have looked at the matter the same way. But even if he had the idea, even if he knew some trigonometry (of which there is no evidence)  he still could not have drawn the map, because neither he nor, as far as is known, anyone else at that time had any information as to the longitudes of places in the Caribbean.
What applies to Piri Re’is applies also to Columbus. Columbus could not have drawn any part of the map included in the special grid because for him, as for Piri Re’is, there could be only one North on a map. It is possible, however, that this special grid may provide a solution to one of the problems of Columbus’ first voyage.
Let us suppose that Columbus had a copy of this map of the Caribbean, as it appears on the Piri Re’is Map. (Piri Re’is himself believed this was the case.) Perhaps the map showed the Azores, or even some part of the European coast, so that by simple measurement Columbus was able to get an idea of the scale of the map and the distance across the ocean to the Caribbean islands.
We know he had some sort of map and that he had an idea of how soon he would find land. But we also know that he did not find land where he had expected to find it. Instead, he had to sail about one thousand miles farther and was faced with a threatened mutiny of his crew. Finally he made a landfall at the island of San Salvador (Watling Island) or some other island nearby.
Now, if you look at San Salvador on our map (Fig. 39) and note its longitude on the main grid of the map, you will see that it lies just west of the 60th meridian on that grid instead of at 74½° West Longitude where it actually should be. But if you swing the map around and find the longitude of the island on the special Caribbean grid, it turns out to be at 80.5° West. The trouble that Columbus ran into may now be understood. His error in not understanding the map he had may have led to a mistake of about 14° or about 840 miles in his estimate of the distance across the Atlantic, and thus nearly caused the failure of his expedition.
Let us consider the probabilities of Columbus’ having carried with him from Spain a copy of this component map of the Caribbean. He need not have had with him the entire source map used by Piri Re’is, including South America . The evidence is that he did not suspect that a continent lay to the south of the Caribbean until he ran into the fresh water of the Orinoco out at sea.
We have seen that Piri Re’is, in all probability, had ancient maps at his disposal in Constantinople. It is quite possible that copies of some of these had reached the West long before his day. Greek scholars fleeing from the Turks brought thousands of Greek manuscripts to Italy before the fall of Constantinople in 1453. Much earlier still, in the year 1204, a Venetian fleet, supposedly intended to carry a crusade to the Holy Land, attacked and captured Constantinople. For about sixty years afterward Italian merchants had access to map collections in Constantinople.
We have reason to believe that good maps of the St. Lawrence River were available in Europe before Columbus sailed in 1492. In Fig. 34 we see a map of the river and the islands near its mouth that the mapmaker Martin Behaim placed on a globe he made and completed before Columbus returned from his first voyage. Columbus was not an ignorant mariner, as some people seem to imagine. He was quite at home in Latin, which indicated some education, and he was a cartographer by trade. It is known that he traveled widely in Europe, always on the lookout for maps. His voyage was not a sudden inspiration; it was a deeply settled objective, one followed with perseverance for many years, and it required, above all, maps. The historian Las Casas said that Columbus had a world map, which he showed to King Ferdinand and Queen Isabella, and which, apparently, convinced them that they should back Columbus.
Many have thought that this map may have been the map said to have been sent to Columbus by the Italian scholar Toscanelli (see Fig. 35). But a Soviet scientist has presented a strong argument against this, including evidence that the Toscanelli letter to Columbus, accompanying the map, was a forgery (209). In any case, the Toscanelli Map, whether Columbus had it or not, is a very poor map.
Cuba on the Piri Re’is Map presents some very interesting problems.
In the first place, Cuba was wrongly labeled Espaniola (Hispaniola, the island now comprising Haiti and the Dominican Republic) by Piri Re’is. This error was accepted by Philip Kahle who studied the map in the 1930’s (106). Nothing could better illustrate how ignorant Piri Re’is was of his own map. The mislabeling of Cuba also clearly shows that all he did was to get some information verbally from a sailor captured by his uncle, or from some other source, and then try to fit the information to a map already in his possession, a map he may have found in the Turkish Naval Archives, which possibly inherited it from the Byzantine Empire. In Figures 36, 37 I have compared the island I have identified on the Piri Re’is Map as Cuba with a modern map of that island.
This comparison shows that what we have in this island on the Piri Re’is Map is a map of Cuba, but a map only of its eastern half. We can identify a number of points around the coasts and in the interior. The western half is missing, but, as if to compensate for this, the island is shown at twice the scale of the rest of the map, so that it subtends about the correct amount of longitude for the whole island. Oddly enough, there is a complete western shoreline where the island is cut off, as if, when the map was drawn, all of western Cuba was still beneath sea level. We observe that some islands are shown in the west in the area now occupied by western Cuba.
There is good evidence that a map of a thus truncated Cuba was well known in Europe before the first voyage of Columbus. In Figure 38 I have compared the Cuba of the Piri Re’is Map with the island labeled “Cipango” on the Behaim Globe (completed before Columbus’ return from his first voyage), on the Toscanelli Map, and on the Bordone Map of 1528.
It seems quite clear that Bordone’s island, which of the three most closely resembles the Piri Re’is island, was not inspired by the current information on Cuba. Cuba on the maps made by the 16th Century explorers in no way resembles the island on the Piri Re’is Map. (See, for example, Fig. 39, the Piri Re’is Map of 1528. Here Piri Re’is represents Cuba in a form typical of the other maps of the day. He had evidently abandoned his ancient maps.)
In view of the possibility that an ancient map of the eastern half of Cuba may have been circulating in Europe before Columbus’ first voyage, it becomes increasingly easy to accept the idea that Columbus may have found a good map, at least the Caribbean section of the Piri Re’is Map, and that this may actually have led him to America. In view of these facts one of my students, Lee Spencer, revised the old verse:
In Fourteen Hundred Ninety Two
Columbus sailed the ocean blue.
With maps in hand drawn long before
He headed straight for Cuba’s shore.
Much fame he gained, so I am told,
For he proved true the maps of old.
The Piri Re’is representation of Cuba suggests that the Caribbean section of his map was itself a compilation of originally separate local maps. One of these may be identified in the map of Hispaniola.
Here we have still another North. The arrow on our map indicates the direction of north for Hispaniola and some adjacent islands. It does not agree with the Norths either of the main grid or of the Caribbean grid; it is not, so far as we can see, integrated with the trigonometric projection. Columbus could not have placed it on the map (assuming he had it) because, if there was one thing Columbus could determine, it was north, and he would therefore have aligned Hispaniola with the rest of the Caribbean islands on the main grid of the map.
3. A map of the Atlantic coast of South America, from Cape Frio northward to the Amazon, with an error in scale.
On the Piri Re’is Map, South America consists of a compilation of various local maps differing in scale and in orientation. This particular component map is on too small a scale, as shown by the inset grid, Fig. 18, but is in correct longitude. It is possible that we can partially reconstruct the story of this map.
It was, in the first place, an accurate map of the coast. But it seems that the mapmaker may have been operating under the impression that Point IV of the world projection pattern lay on the Tropic of Capricorn, and he placed this component map so that its southern end lay on this assumed tropic. This left its northern end too far south, because of an error in the scale. The mapmaker, however, may have been unaware of this because of a failure to identify the river shown there as the Para River, one of the mouths of the Amazon. According to my interpretation, the map does show the course of the Amazon coming down to its Para River mouth, but it does not show the Island of Marajo. The map would suggest that it may come from a time when the Para River was the main or only mouth of the Amazon, and when the Island of Marajo may have been part of the mainland on the northern side of the river. If the mapmaker knew of the Island of Marajo as existing in his time he might not identify the river on his source map with the Amazon. We shall see evidence shortly that he did know of the existence of the Island.
The evidence for my interpretation of this part of the map is in the agreement of the inset grid on our tracing with the topography, as shown in Table 1.
4. A map of the Amazon and the Island of Marajo, correctly placed on the equator of the trigonometric projection; some other component maps.
One part of the Piri Re’is Map that seems to date without modification from the time when a trigonometric projection was used to compile a world map from various local maps is a map of the Amazon with a very good representation of the Island of Marajo. Here both mouths of the Amazon are shown. The upper one, the mouth of the Amazon proper, is shown about 10° north of the river just suggested as the Para River, on the inset grid of our map. It lies about 5° north of the line used as the equator for Africa and Europe. Interestingly, both the duplications of the Amazon suggest the actual course of the river, while all the representations of it in the later maps of the 16th Century bear no resemblance to its real course. Moreover, the excellent representation of the Island of Marajo is quite unique. Nothing like it can be found on any map of the 16th Century until after the official discovery of the island in 1543. Where could Piri Re’is have got his accurate conception of this island? If he had somehow obtained the information as to its shape, how could he have placed it correctly both in latitude and longitude, with reference to a mathematical projection of which he was almost certainly ignorant?
This Island of Marajo did quite a bit of drifting after Piri Re’is’ day. It turned up on Mercator’s 1569 Map of South America, but here we find it placed at the mouth of the Orinoco! (See Fig. 41.)
To the east of the South American coast the Piri Re’is Map shows a large island where no such island now exists. One might suppose that this island was imagined by Piri Re’is in the same way that many mythical islands were placed on other maps of the Renaissance. Piri Re’is actually does have such an island, which he names “Antillia” (No. 94 on our map). It looks artificial. But the island we have now to deal with does not have this artificial look. It has the appearance of a real island, with harbors, and islands off the coast. Some of the photographs show highlands around the coasts (indicated by deeper color) and a large central plain.
The fact that this island has more behind it than the imagination of one cartographer is suggested by another map, presented to the Paris Academy of Sciences in 1737, and associated with the name of the French geographer Philippe Buache, a member of the Academy. On this map we see an island, very roughly the shape and size of the island on the Piri Re’is Map, placed directly on the equator! Between this island and the coast of Africa we see another island, where none now exists. The map has indications that these islands were even then former, not present, islands. The coasts are hatched, suggesting approximations . Inside these coasts smaller islands are indicated, as if they were remnants left by the submergence of the larger islands. Indications are given that the Cape Verde and Canary Islands were once connected with the mainland of Africa, and other island groups are shown in the North Atlantic where none now exist (Fig. 42).
What is this map? Does it illustrate a legend of submerged islands in the Atlantic? If it does, then certain facts about the locations of these islands are relevant. One is that the big island on the Piri Re’is Map is located right over the Mid-Atlantic Ridge (formerly called the Dolphin Ridge) at the spot where two tiny islands, the Rocks of St. Peter and Paul, jut up above the sea, just north of the equator and about 700 miles east of the coast of Brazil. (See Fig. 43 .) Another fact is that the island to the east of the corresponding island on the Buache Map is located just over the Sierra Leone Rise, a mountain range on the ocean bottom. Finally, as the reader can see, a cross-section of the equatorial Atlantic, from South America to Africa, shows both the Mid-Atlantic Ridge and the Sierra Leone Rise, although rather crudely.
Many will dismiss these facts as mere coincidence. Admittedly, there is no proof; but I feel strongly that something more than coincidence is involved. Figure 43 shows the present ocean bottom.
Another component map, which may be briefly dealt with here, shows the mountainous area on the western side of Piri Re’is’ South America . This component map was added to the general map, but it was not integrated with the trigonometry of the projection. There are errors both in scale and in orientation, as shown in Figure 18.
It seems at first glance that the mountains shown here must have been intended for the Andes. However, Kahle, one of the earlier students of the map, rejected this on the ground that the Andes were not yet discovered when Piri Re’is drew his map. On this controversial point the following considerations may be urged:
First, what is the probability that a cartographer, by pure invention, would place an enormous range of mountains on the western side of South America, where one actually exists?
Second, the various rivers, including both Amazons on Piri Re’is’ map of South America, are shown flowing from these mountains, which is correct.
Third, the drawing of the mountains indicates that they were observed from the sea—from coastwise shipping—and not imagined.
Fourth, the general shape of the coast on the map agrees well with the South American coast from about 4° South Latitude down to about 40° S. It is between these latitudes that the Pacific cordillera of the Andes closely parallel the coast. There is even a suggestion on this coast of the Peninsula of Paracas.
Kahle adds an observation that apparently contradicts his own conclusion. He positively identifies one of the animals shown in the high mountains as the llama, on the ground that it shows the protruding tongue peculiar to that beast. (106) Llamas are found in the Andes; but it is doubtful they could have been known to anyone in Europe in 1513.
5. A map on the main grid, from Equator II to the Peninsula of Paria.
An interesting problem on this component map is the identification of the Orinoco. The river itself is not shown, and neither is the modern delta. Instead, two estuaries extending far inland (for a distance of about 100 miles) are shown close to the site of the present river. The longitude on the grid would be correct for the Orinoco, and the latitude is also quite accurate (see Table 1). Is it possible that these estuaries have been filled in, and the delta extended this much since the source maps were made? If so, this is comparable to the extension of the delta of the Tigris-Euphrates in Mesopotamia in the last 3,500 years, since Ur of the Chaldees lay on the seacoast.
6. A map from the Gulf of Venezuela to Yucatan, omitting about 7° of coastline between the Gulf of Venezuela and the Peninsula of Paria.
A point of considerable importance here is the shape of the Atrato River. According to our grid, the river is shown for a distance of 300 miles from the sea, and its eastward bend at about 5° North Latitude corresponds to the geographical facts. This implies that somebody explored the river to its headwaters in the Western Cordillera of the Andes sometime before 1513. I have found no record of such an early exploration. Yucatan supposedly had not been discovered in 1513, either.
7. The Caribbean Islands: The Leeward and Windward Groups, the Virgin Islands, Puerto Rico on the main grid of the map; more questions about Columbus.
These islands are more accurately placed on this map, in reference to latitude and longitude, than they are on any other map of the period.
Piri Re’is wrote, in his long inscription about Columbus, that this part of the map was based on a map Columbus drew. Here the two different grids overlap to some extent: Some islands are on the special grid already discussed, and some are on the main grid. I have pointed out that one of Columbus’ errors may have been due to not understanding the special grid. The Leeward and Windward Islands, which Columbus discovered, are on the main grid on this component map. Nevertheless, it is hardly possible that he could have added them to the map, as Piri Re’is supposed. For we see them in remarkably correct latitude and longitude on the trigonometric grid of our hypothetical Alexandrian compilation. Not understanding the grid, not even dreaming of its existence, and not being able to find either correct latitude or correct longitude, how could Columbus have correctly located the islands? Piri Re’is gives names to these islands, and says that they are the names given by Columbus, yet the names are wrong! (140:408-409) It looks as if Piri Re’is here depended upon hearsay information and did not really see a map drawn by Columbus.
One group of islands on the Caribbean part of the map, the Virgin Islands, are so far out of position, so badly drawn, and so far out of scale that they might well have been added to the map by Columbus or interpolated by Piri Re’is on the basis of some contemporary report.
One of the most unusual features of this part of the map is that some features can be interpreted as two different localities, according to the grid one uses.
8. The lower east coast of South America from Bahia Blanca to Cape Horn (or Cape San Diego) and certain Atlantic islands on the main grid of the map.
Two of my students, Lee Spencer and Ruth Baraw, discovered that about 900 miles of the east coast of South America were simply missing from the Piri Re’is Map, two different source maps having apparently been erroneously put together on the general compilation. Earlier students of the map—Kahle, Goodwin, Mallery —had all assumed that the map was continuous and complete as far as it went.
Kahle’s assumption of an unbroken coast required a rather forced interpretation of the map. On this assumption it was necessary to conclude that the mapmaking here was very bad. However, it seems that someone before Kahle had had the same idea. Figure 44 shows how that interpretation actually fits the oblong grid of the map. The equator is different from that of the main projection, but the length of the degree of latitude has been increased in the same way . This detail serves to support our impression of the long and complex history of this map. There is no way of knowing how many peoples of how many epochs had their fingers in the pie.
The method used by Spencer and Baraw to verify their observation of the omission of the coastline was to try identifying localities by comparison with the modern map, first from one end of the coast, and then from the other. They started first with Recife and went all the way down the coast from point to point. Everything went well as far as Cape Frio, but south of Cape Frio they thought the Piri Re’is Map ceased to correspond with the modern map at all. Then they started from the bottom, from what we assumed to be Cape Horn, or Cape San Diego (No. 74, Figure 18), and went northward identifying localities. Here again everything seemed to agree very well with the modern map until they came to a point just below Cape Frio. Farther than this they could not go. The missing coast lay in between . Our grid assisted us very much in the final verification of the break, for it gave us its value in degrees.
The omission of the coast between Cape Frio and Bahia Blanca apparently resulted in a loss of about 16° of South Latitude and about 20° of West Longitude. Therefore, in Table 1, I have added these amounts of latitude and longitude to the ones found by our grid. When this is done, the positions of the identified localities are correct to an average error of less than a degree. More important is the fact that they are correct relative to each other.
It appears significant that Piri Re’is, who stuck names taken from explorers’ accounts on much of his map (making numerous errors), did not attempt to place any names on the southern part of this coast of South America. The reason offers itself: There were no explorers’ accounts. That coast had not been explored by 1513.
The Falkland Islands appear in this section of the map at the correct latitude relative to this lower east coast, but there is an error of about 5° in longitude. The Falklands are supposed to have been discovered by John Davis in 1592, nearly eighty years after Piri Re’is made his map (68:869). 
South of Cape Horn, or Cape San Diego, the coast on the Piri Re’is Map appears to continue unbroken, but here we have been able to identify another break, or rather omission.
9. The Antarctic.
Proceeding as in the case of the break in the east coast of South America, we first identified localities down to the vicinity of Cape Horn (including specifically Cape San Diego), then jumped to the next cape to the eastward, assuming as a working hypothesis that it was the Palmer or Antarctic Peninsula as claimed by Mallery. This assumption would require that the sea between the Horn and the Antarctic Peninsula had been omitted by the mapmaker. This assumption appeared to be supported by our identification of the Shetland Islands . These islands are not far off the Antarctic coast. The omission of the sea between (Drake Passage) automatically would put the South Shetlands too far north by the width of the strait, which happens to be about 9°. If the reader will compare the positions of the Falklands and the South Shetlands on a globe with their positions on the Piri Re’is Map, as we have identified them, he will see how the Antarctic coast seems to have been simply pushed northward, and Drake Passage omitted.
Interestingly enough, we find that the same mistake was made on all maps of the Renaissance showing the Antarctic. When we come, in the next chapter, to the examination of the map of Oronteus Finaeus, we shall discover the probable reason for this error.
The extraordinary implications of Captain Mallery’s claim that part of the Antarctic Continent is shown on the Piri Re’is Map demand unusually thorough verification, considering that the continent was supposedly discovered only in 1818. This is no slight matter. Important questions, for geology as well as for history, depend upon it. We may begin with a brief survey of the historical background.
A good many world maps of the 16th Century show an antarctic continent.  As we shall see, Gerard Mercator believed in its existence. A comparison of all the versions suggests that there may have been one or two original versions, drawn according to different projections, which were copied and recopied with emendations according to the ideas of different cartographers.
The belief in the existence of the continent lasted until the time of Captain Cook, whose voyages into the South Seas demonstrated the non-existence of a southern continent at least in the latitudes where one appears on these maps (112). The idea of an antarctic continent was then given up, and geographers began to explain the maps as the work of geographers who had felt the need to have a land mass at the South Pole to balance off the concentration of land in the northern hemisphere. This seemed to be the only reasonable explanation, for in the first place there apparently was no such continent, and in the second place there was no reason to suppose that anyone in earlier times (Romans, Greeks, Phoenicians) could have explored those distant regions.
When we began our study of the southern sector of the Piri Re’is Map our first step was to compare it carefully, not with a flat map of the Antarctic, but with a globe. Figure 45, traced from a photograph of a globe,  shows a striking similarity between the Queen Maud Land coast and the coastline on the Piri Re’is Map. It should be especially noted that on the modern globe the Queen Maud Land coast lies due south of the Guinea coast of Africa, just as the coastline referred to by Mallery does on the Piri Re’is Map. 
This was an encouraging beginning. We went on to make a thorough examination. We asked ourselves, first, how does the coast in question on the Piri Re’is Map compare in its extent, character, and position, with the coasts of Queen Maud Land? (These coasts are named the Princess Martha and Princess Astrid Coasts.) With the gradual development of the mathematical grid we could answer two of these questions.
In the first place, we found that the Piri Re’is coast, according to our grid, extends through 27° of Longitude as compared with 24° on the modern map, a very remarkable degree of agreement. At the latitude of the coasts (about 70° S) a degree of longitude is only about 20 miles, so that the error is not great. The grid also shows the coast in good position; it is about 10°, or 200 miles, too far west.
With regard to latitude, we must take account of the omissions we have noted above—part of the South American coast and Drake Passage. Together these omissions account for about 25 ° of South Latitude. When these degrees are added to those found by the grid for the Queen Maud Land coast, the coast appears in correct latitude (see Fig. 18).
We have noted that the omission of the South American coastline resulted in a loss of about 16° of West Longitude. The omission of Drake Passage resulted, we found, in adding about 4° to this, making 20° be accounted for. This, with the 10° westward error of the Queen Maud Land coast, creates a deficit of some 30° between that coast and the Antarctic Peninsula. This appears to be made up for by the fact that the Weddell Sea, as we have identified it on the map, extends through only 10° of longitude, instead of 40°, as would be correct.
Now it might be argued that this result is artificial, and that we have deliberately twisted the evidence to support the conclusion, but this is not the case. My students, Lee Spencer and Ruth Baraw, had already established the omission of 900 miles of the South American coast without any thought of Antarctica. They were not interested in the bearing of their discovery on the question of the Queen Maud Land coast. We did not even see the connection until long afterwards, when the grid was worked out, and the same is true of the omission of Drake Passage. The omission is obvious from the map itself : the strait simply isn’t there. In the case of both omissions we were able to measure approximately the amounts of latitude and longitude involved.
There is in addition the comparison of the character of the Queen Maud Land coast, as shown on the ancient and on the modern map. It is plain, from the modern map, that this coast is a rugged one. Numerous mountain ranges and individual peaks show up above the present levels of the ice. The Piri Re’is Map shows the same type of coast, though without any ice. The numerous mountains are clearly indicated. By a convention of 16th Century mapmaking heavy shading of some of the islands indicates a mountainous terrain (Frontispiece).
Coming to greater detail, Mallery’s chief argument was the striking agreement of the map with the seismic profile across Queen Maud Land (see Figs. 46, 47, 48, and Note 10). The reader will note that the profile shows a rugged terrain, a coastline with mountains behind the coast and high islands in front. The points of the profile below sea level coincide very well with the bays between the islands on the Piri Re’is Map. This amounts to additional confirmation. The identification of specific features of the coast, as shown in Table 1, appears further to strengthen the argument.
If the Piri Re’is Map stood alone, it would perhaps be insufficient to carry conviction. But it does not stand alone. We shall shortly see that the testimony of this map ·regarding the Antarctic can be supported by that of several others.