And it is remarkable. Because, yes, we can accept with Mcintosh that it was within the competence of navigators of the thirteenth century to have produced the excellent portolan outline of the Mediterranean that was to require so little improvement over the next 500 years – in other words, we can accept that it could have been done. We can even accept that it might have been done. But it is much harder to agree that this is what actually was done, since neither Mcintosh nor any other scholar favouring the gradual ‘evolutionary’ explanation for the very early perfection of the portolan genre has yet been able to provide us with even a single example of charts that illustrate even a single aspect of this proposed ‘gradual evolution’.
In my opinion, therefore, Peter Whitfield is right to evaluate the Carta Pisane, the oldest surviving portolan in the world, as ‘one of the most enigmatic charts in the history of mapmaking’.73 In his 1996 study, Charting of the Oceans, he elaborates on this theme:
The appearance of this chart (and of the others which survive from the following century) is one of the most mysterious events in the history of mapmaking. A glance at the Pisan Chart immediately reveals two outstanding features: the coastlines of the Mediterranean are drawn with striking accuracy; and the map is covered with a network of lines radiating from two central points, which clearly impose the form of the compass over the whole map. How did this highly accurate map suddenly appear in medieval Italy, and how exactly was it linked to the compass? Was it the original work of a single individual, or was it descended from a line of much older charts which had been developing for centuries? The former is difficult to believe, but the latter cannot explain why there is no shred of evidence for the existence of such maps before 1270.74
Whitfield outlines the orthodox scholarly response that the evolution of the portolans must have taken place within the oral lore of mariners and within the textual tradition – going back to the Greek periploi – of books of sailing directions:
One famous example entitled Lo Compasso da Navigare was current among Italian mariners and it would be tempting to suppose that the contents of a text such as this had been transformed with the aid of compass bearings into the Pisan Chart. Unfortunately, the places named in Lo Compasso differ sharply from those named on the map, even the names in Italy itself. Moreover, the transition from a list of names and bearings to an accurate map is an enormous one, requiring not only a high degree of geometric and drafting skill, but also an imaginative leap to create a graphic form for which there was no parallel. Even if the Pisan chart was based on some now-lost portolano, we have no real idea how it was done. Nor can we really answer the most fundamental question of all about the chart – how was it used? We have no independent description of its use, although we do know, from examination of the chart itself, that the compass lines were plotted before the map itself was drawn …75
Concerning how it was done, Whitfield notes:
Later practice was to make a running survey, in which coastal features – capes, bays or islands – were sighted from two, three or four positions as the ship sailed by. Starting from the ship’s course, the distances run and the angles of sight were used to build up a profile of the coast. This method was in use by the later sixteenth century, but we can only conjecture whether it was known at the time the Pisan Chart was drawn. If it was not, it is extremely difficult to account for the accuracy of some of the coastlines, which would scarcely be improved on this scale until the eighteenth century.76
But even if we admit that running-survey and compass techniques were somehow being used on ships to produce sea-charts as early as the thirteenth century (which most historians of science would rule out) we still come against the unexplained enigma of the miraculous and fully formed de novo appearance of the Carta Pisane. As we’ve seen, not a single chart pre-dates it that demonstrates in any way the gradual build-up of coastal profiles across the whole extent of the Mediterranean that must have occurred before a likeness as perfect as this could have been resolved.
It is possible, of course, through the vicissitudes of history, that all the evidence for the prior evolution of portolans before the Carta Pisane has simply been lost. If that were the case, however – in other words if the Carta Pisane is a snapshot of a certain moment in the development of an evolving genre of maps, and if we accept that all earlier ‘snap-shots’ have been lost, wouldn’t we nevertheless expect that such an ‘evolving genre’ would have continued to evolve after the date of the earliest surviving example?
Whether we set the date of the Pisane between 1270 and 1290 (as Whitfield suggests)77 or a little later – between 1295 and 1300 – as other scholars have argued, we’ve seen that that there was no significant evolution afterwards.78
Now kept in the Bibliothèque Nationale in Paris, the enigmatic Pisane is an unsigned chart and scholars have no idea who the cartographer might have been.79
Next comes what Whitfield rightly describes as the ‘startling and precocious’ work of the earliest chartmakers known to us by name in the first half of the fourteenth century. These include Vesconte and Pizzagano in Venice, and Dulcert and Valseca in Majorca. None of them seems to have copied the Carta Pisane directly, but neither do they add significant cartographical detail in the central Mediterranean/Black Sea area covered by the Pisane. On the contrary, what we see in their more lavish maps are only the effects of very minor tinkering and stylistic improvements. The basic template inherited from the thirteenth century remains unaltered and stays that way for the rest of the life of the genre.
So the hypothesis of a gradual evolution of portolan charts out of books of sailing directions does not withstand close scrutiny. Convinced of this, A. E. Nordenskiold sought a more satisfactory explanation and came, after many years of study, to a radical conclusion – that the original model for all the portolan charts, a hypothetical common ancestor that he refers to as the ‘normal portolano’ is most likely to have been derived from the long lost sea-charts of the Phoenician geographer Marinus of Tyre.80
In other words, the Carta Pisane and the other early portolan charts that started the genre were not a ‘development’ of anything. They were a legacy.
The Sea-fish of Tyre
Nordenskiold points out that the same legends and place names, presented in the same way, appear on all portolan charts. He makes a special illustration of this with reference to the Catalan Atlas of the fourteenth century, Giroldis’ portolan of the fifteenth century, and one by Volontius of the end of the sixteenth century, but argues that it is true for all portolans:
When to this is added
(1) that the Mediterranean and the Black Sea have exactly the same shape on all these maps; (2) that a distance scale with the same unit of length … occurs on all these maps, independently of the land of their origin; (3) that the distances across the Mediterranean and the Black Sea measured with this scale agree perfectly on different maps; (4) that the conventional shape given to a number of smaller islands and capes included in the maps remained almost unaltered on portolanos from the 14th century to the end of the 16th; then it may be held as completely proved that all these portolanos are only slightly altered and emended ‘codices’ of the same original which I designate by the name normal portolano.81
In his quest ‘to determine when and where the normal portolano was composed’,82 Nordenskiold uncovered a previously overlooked passage in a work written in AD 955 by the important Arab geographer Masudi who states that he had: ‘seen the maps of Marinus, and that these by far surpassed those of Ptolemy’.83
The portolans are the only maps drawn in ancient times or in the Middle Ages that are better than the maps of Ptolemy.84 We cannot say for sure how ancient their origins are. But they must have a background somewhere. It is Nordenskiold’s hypothesis that ‘the first origin of the portolanos is to be derived from the Tyrian charts described by Ptolemy under the name of Marinus’85 and that the world map of Marinus could have been ‘a real portolano, provided with a text’.86 Moreover,
/> If Ptolemy himself had not always spoken of Marinus as a definite personality, it could have been conjectured that the name Marinus of Tyre, or the Tyrian sea-fish, had only been a collective name for a certain category of nautical maps … The numerous editions mentioned by Ptolemy mean that the Tyrian charts were made for a practical purpose, and the improvements, introduced according to Ptolemy in every new edition, constituted the germ of the future masterpiece …87
This is an interesting speculation, for indeed there is no mention of Marinus outside of Ptolemy which independently confirms the Phoenician geographer’s existence. Nor is it too much to ask of the facts to suggest that the famous seafaring city of Tyre to which Marinus supposedly belonged might have originated a special category of charts that came to be known, colloquially, by a name something like the ‘Tyrian sea-fish’. Perhaps, despite the personalization, it was an atlas of ‘Tyrian sea-fish’ regional charts and a ‘Tyrian sea-fish’ world map that Ptolemy ‘corrected’ and ‘improved’ in the second century AD, and not the work of any individual geographer?
And I’ve already noted that we only have Ptolemy’s word for it that he actually did improve on Marinus. Maybe he thought he was doing that – while all the time his ‘improvements’ were only making the Phoenician charts worse. That would explain why Arab mariners of the tenth century still treasured the original Marinus maps that they had somehow managed to preserve and declared them to be so much better than the Ptolemaic ones.
Arabia without maps
Just three Arab portolans, all classic ‘normal portolanos’ of the Mediterranean and Black Sea area, have ever been found. The earliest dates from 1300, very close to the date of the Carta Pisane, and the other two from 1413 and 1461 respectively.88 This suggests at least two things to me: first, like the Europeans, the Arabs made no attempt to develop the inherited normal portolano (other than putting modern names and legends on their copies of it); secondly, although the Marinus ‘normal portolano’ had been preserved by the Arabs, as Masudi testifies, and although there was clearly some demand for it, the survival of only three Arab copies suggests that its use never became anything like as widespread in Arab seafaring as it did in the seafaring of the Europeans.
In his discussion of Arab cartography, A. E. Nordenskiold has this to say:
Various admirable descriptions of distant lands and of extensive voyages written by Arabian scholars and far surpassing the geographical productions of the same period among the Christians, are still extant. But similar perfection was never attained by the Arabian maps, which, if they were original drawings and not, as the planisphere of Idrisi, mere copies or reproductions from Ptolemy, are not only far inferior to the maps of the Alexandrian geographer, but not even comparable to the Esquimau-sketches brought home by English and Danish polar travellers from the icy deserts of the polar regions.89
This may seem an over-harsh judgement, since there is no doubt that the Arabs were brave and adventurous explorers. For example, the same Idrisi mentioned in the passage above also indicates that in the tenth century Arab sailors crossed or attempted to cross the Atlantic.90 But it is true that Idrisi, geographer to King Roger II of Sicily at the end of the twelfth century, did base his beautiful maps on Ptolemy.91 And it is true, with the exception of the three rare Arab portolans that have survived (one of them being notably early), that the quality of the rest of Arab cartography in this period was not high.
Al-Tunisi Arabic portolan, western section, AD 1413.
Regardless of whether the Arabs themselves were good or bad at making maps, however, as Nordenskiold points out:
It is shown by the passage referred to in Masudi, that the maps of Marinus of Tyre were still extant in the middle of the 10th century, that is to say, shortly before the time when the first portolan maps were drawn. Since that time they have completely disappeared. It might be legitimately concluded from this that the portolanos may have arisen as a modernization of the Tyrian sea-fish undertaken during the Crusades, and that they stood in the same relation to the maps of Marinus as the tabulae modernae in the printed editions of the Geography of Ptolemy stood to the Alexandrine geographer’s own work.92
On marvellous things
By documenting the presence among the Arabs at so late a date of good ancient maps that were (a) attributed to Marinus of Tyre and (b) recognized as belonging to a distinct tradition superior to Ptolemy, Nordenskiold provides at least the beginnings of a plausible answer to the riddle of the ‘lost evolution’ of the portolans prior to the Carta Pisane. Here is the scenario in brief: ‘sea-fish’ maps, unadulterated by Ptolemy, that had been carried to perfection by the second century AD were preserved by Arab culture until the thirteenth century AD. Then at least part of the legacy – a chart of the Mediterranean and Black Sea region, Nordenskiold’s ‘normal portolano’ – fell into European hands, providing the model, with the necessary modernization of place names, etc., for the Carta Pisane and the entire portolan genre.
In my opinion this is a more rational and more parsimonious way to account for the highly developed state of the normal portolano than to ask us, as most historians do, to accept that such striking and precocious cartography somehow ‘evolved’ out of books of sailing directions. And Nordenskiold’s hypothesis, though it leaves unanswered all questions about the roots and antiquity of the Marinus tradition before the second century AD, is also on sound logical ground by reminding us of the role of the Phoenicians in all this.
Known to have circumnavigated Africa by 595 BC,93 2000 years before the Portuguese, the Phoenicians maintained fleets throughout the Red Sea, the Indian Ocean and the Mediterranean (at powerful naval and mercantile cities like Tyre, Sidon and Carthage), planted major colonies on the Atlantic coasts of Europe and North Africa, and crossed the Atlantic at least as far as the Azores and the Canary Islands.94 They were, without contest, the greatest mariners of the ancient world. Indeed, between the time of Ptolemy and the time of the Portuguese one looks in vain for any other seafaring culture of the Mediterranean/Black Sea region that would have had both the capacity and the inclination to devise a map like the normal portolano.
Moreover, if the normal portolano is indeed derived from the lost atlas of Marinus of Tyre, then it follows that other high-quality maps of regions much further afield than the Mediterranean and the Black Sea, and indeed a world map, might also have been preserved by the Arabs – for we know from Ptolemy’s testimony that other Marinus maps, including a world map, did once exist. It will therefore do no harm to keep an open mind to the possibility that the portolan world maps that began to appear during the century after the Carta Pisane,95 might also have been influenced by earlier ‘Tyrian sea-fish’ maps of Phoenician origin. Christopher Columbus, whose passionate belief in lands across the Atlantic led to his ‘discovery’ of the New World, seems to hint at a Phoenician connection when he describes one of the inspirations for his journey:
Aristotle in his book On Marvellous Things reports a story that some Carthaginian merchants sailed over the Ocean Sea to a very fertile island … this island some Portuguese showed me on their charts under the name Antilia.96
Antilia first appears on a portolan chart of 1424. It is a mysterious presence there, a riddle, to which we will return.
What Guzarate showed da Gama
The suggestion has been made that ‘world’ portolans – indeed, any that show regions outside the normal portolano area – could have been based on the lost world map of Marinus. And if the normal portolano reached Europe after being preserved among the Arabs for many centuries, it could be the case that the Arabs preserved the world map too. We’ve seen that some Arab portolans of the Mediterranean/Black Sea area do exist – although they are very few in number. So it makes sense to look for traces among the Arabs of a portolan world map as well.
Nordenskiold believed he had identified such a trace. Combing through geographical works from the Age of Discovery, he found a passage in J. De Barros’ Asia (first Portuguese edition
printed 1552) which states that the Arabs in the Indian Ocean possessed sailing charts with degree-lines, ‘perhaps comparable in their finish to the portolanos’:97
When Vasco da Gama during his first voyage, in April 1498, arrived at Malindi on the east coast of Africa, he there procured a pilot named Guzarate to sail his ship to India. Da Gama was much pleased with him, especially since the pilot showed him a map made in the Arabian (Moorish) manner of the whole Indian coast, without compass lines but divided by meridians and parallels into small squares. The pilot also showed him some nautical instruments intended for determining latitude, different to those which da Gama had brought with him.98
There are a number of points of great interest in this report:
The name that De Barros gives for the pilot is quite different from the name of ‘Ahmed-bin-Majid’ provided by other sources. In fact, Guzarate doesn’t sound much like a name at all. What it does sound like is a nickname or familiar term – ‘Gujerati’ – that may still be heard on Kenya’s Swahili coast today in reference to natives of the Indian state of Gujerat. Is it possible that da Gama’s ‘Arab pilot’ was in fact an Indian pilot – a Gujerati?
The map is said to show the ‘whole Indian coast’.
The map is said to be ‘without compass lines’ – which takes it far from the standard European presentation of a portolan.
The map is said to possess meridians and parallels – again far from the normal portolano, which has no meridians and parallels. However, these meridians and parallels are also said to divide Guzarate’s map into ‘small squares’. It is of note in this respect, though they do not result from intersecting meridians and parallels, that the Carta Pisane has four areas divided up into small squares and two other areas divided into slightly larger squares. Such divisions occur on no other portolan chart known in the west.99