The portolan chart emerged, probably at the end of the 12th century, to match the strictly pragmatic needs of illiterate Mediterranean navigators. It has no cartographic or textual antecedent. Instead, the charts’ source was sailors’ mental maps which pilots had used for millennia – built up out of shared memories of open-sea (pelagic) voyages. Paired with that was their sophisticated direction-finding abilities, manifested in the recently decoded mental wind compass. The origin story is one of orally transmitted memories, with no writing involved. Because there is no evidence or likelihood of financial or organisational assistance from others, we must assign full responsibility for the first, highly innovative, marine chart to the pilots themselves.
Setting the scene
Nobody recorded the charts’ birth. Neither the charts themselves nor any extant text throw any direct light on their
antecedents. This leaves the charts as crucial witnesses to their own history.
In the absence of ‘hard’ evidence we are forced to turn to hypothesis, logic, common sense, intuition. We receive help, though, from the fact that the oldest survivor, the Carte Pisane of c.1270, shares many practical features (not seen before in cartography) with the charts that followed over the next four centuries. Since the portolan chart was clearly not in existence in 1154, that gives no more than about a century for any previous development.
The likely perceptions of the charts’ creators and early users are considered, on the grounds that they had probably never seen a map before. Their largest single innovation is undoubtedly an idea: the concept of the portolan chart.
[Setting the scene.]
The contrasting milieus of sea and terra firma in the Middle Ages
The land divides, whereas the sea unites. The political boundaries of terrestrial space are matched by an indivisible
marine environment. To measure land, incompatible local measures were used, but agreeing on a single marine unit (the
portolan mile), a shared toponymy, and a common mental wind compass, must have been a precondition for the piloting of
merchant vessels in the open sea.
[The contrasting milieus of sea and terra firma in the Middle Ages.]
Alternative origin hypotheses
suggested in the past for the gathering of the data shown in the portolan charts are unconvincing. Joining up the
statements of the distance and direction from one place to the next in a portolano would have been too imprecise, and
the idea of a full coastal survey is anachronistic. The ‘sub-charts’ theory is critiqued later.
Instead, the source for the portolan chart suggested here is the spatial memories of countless longer voyages that must have been held in the head of every widely-travelled sailor, At some point those must have been set down in graphic form.
The charts represent a seaman’s view, and to hope to understand the process of creation we have to place ourselves mentally on the rolling deck of a boat.
[Alternative origin hypotheses.]
There are no identifiable features on the charts that can be traced back to any antecedent. On the contrary, the portolan charts include a suite of unprecedented conventions, probably, in both number and originality, without parallel in the history of cartography.
[Possible cartographic antecedents for the portolan charts.]
Assertion: Most sailors around 1200 were illiterate so why would a textual origin for the charts be likely?
There is no evidence that the portolan charts were based on pre-existing textual sources. Why would we expect one, since the essential elements of the chart are graphic and based on memorised experience?
[Why is there a widespread assumption in favour of a textual origin for the portolan chart?]
Because the 70 sheets of al-Idrīsī’s wall map of 1154 (called here the Charta Rogeriana) are dispersed throughout the volume, combined with the barrier of its Arabic text, it is highly unlikely that any Christian sailor would have seen it. Furthermore, the Charta’s purpose and content are so unlike those of the portolan charts that there can have been no borrowing either way. Since al-Idrīsī’s 15 years of research would certainly have unearthed a portolan chart had it existed then, its creation must have taken place after 1154. For the same reason, that denies any Islamic contribution to the origin of the portolan chart.
[Why the portolan charts made no use of the Charta .]
Indirect support for the mental-map thesis comes from the latest research into the brain’s capacity for spatial cognition. Feats of memorisation (for instance London’s Black Cab drivers, whose hippocampus continues to enlarge with experience), and the documented navigational skills of single-handed circumnavigators and indigenous groups around the world operating without instruments, are examples. Based on experience and orally transmitted instructions this knowledge was vital for livelihoods, even survival.
Indigenous cartography provides numerous instances of the memorisation of non-literate navigation systems that are more complex than anything experienced in the Mediterranean.
[Spatial cognition and mental maps.]
“Greek and Roman seafarers, employing no instruments whatsoever, practiced both coastal and open-sea navigation as matters of routine…” (Danny Lee Davis) and “Regarding ancient knowledge of navigation, extant evidence points towards a practical knowledge of seaborne travel and orientation which developed through ‘mental-maps’ based on experience and practice, and was likely predominantly transmitted orally” (Carmen Obied).
[Navigation in the Ancient World.]
Nobody would have routinely left port without knowing the direction to take, any islands or dangers likely to be met with on the way, and the approximate distance and time involved to reach the planned destination.
Nor, in the period before the portolan chart and the magnetic compass, would they have done so unless they were confident of being able to determine at least one direction when in the open sea.
For centuries, indeed millennia, ships had set sail across the Mediterranean, without instrumental aid besides a log line. They used ‘dead reckoning’: the ability through long experience to estimate direction, distance and elapsed time, and hold that in their head. Such navigational lore would have come from their own detailed memories or, orally, from other sailors. After all, for land to be recognised it had to have been anticipated.
[Pre-knowledge of the route.]
The broad skill set of sea signs required for non-instrumental sailing is described, with specific Mediterranean factors highlighted.
Assertion: Details of pelagic courses in the Mediterranean must have been held in sailors’ memories for millennia, hence a 12th-century sailor would have been using some kind of mental map, in the absence of any textual or graphical guide.
The island-rich Mediterranean would have assisted the creation of mental maps. Examples are given of routine feats of memory more challenging than anything required of a medieval pilot.
[The mental map in the Mediterranean context.]
Whenever information has been vital, there would have been a strong incentive to commit it to memory.
Whoever first imagined the idea of a portolan chart, and anyone using the prototypes, would have already been skilled in navigating without one.
Other researchers assume that pilots used mental maps, although they have not taken the further step of seeing those as the source for the portolan chart.
[Was sailors’ knowledge the source for the portolan chart?]
Assertion: No mariners would have set out on pelagic voyages without knowing what lay beyond the horizon, and neither would they have done that unless they were confident that they could recognise at least one direction once they were out on the open sea.
Determining direction is arguably the most important single aspect of open-sea navigation. Put simply, without a mental compass there could have been no pelagic sailing.
There must have been a level of direction-finding ability sufficient to make pelagic voyaging viable in terms of both safety and the time expended. Furthermore, such proficiency must surely have been widely spread among pilots, not the preserve of a select few. [On this see the separate essay The mental wind compass of the medieval Mediterranean: the rediscovery of its structure and interpretations of its use.]
[The mental wind compass.]
The introduction and use of the magnetic needle / compass is related to the hypothetical forerunner of the portolan chart and the oldest pilot guide, the Liber, all of which seem to have emerged in the early 13th century.
It would be wrong to assume that the introduction of the magnetic compass led immediately to an abandonment of refined dead-reckoning skills and the associated use of the mental wind compass.
We cannot be sure of this but it is likely that the charts’ forerunner was based exclusively on astronomical observations, whereas by the time of the Carte Pisane (c.1270) the directions had become entirely magnetic.
[The magnetic compass.]
Contrasting the likely content of any lost text with what was required for navigation makes it improbable such a document could have existed or been disseminated among pilots.
[Could there have been textual antecedents for the pilot books as opposed to mental ones?]
There are 196 open-sea (transfretus) routes in the Liber against 1,287 in Lo compasso, yet only one-third of the 196 were repeated in Lo compasso. This confirms their different sources.
[The pelagic statements listed in the Liber.]
The statements in both the Liber and Lo compasso which describe themselves as referring to ‘pelagic’, i.e. open-sea courses, can be shown instead to refer to straight lines drawn between pairs of pelagic termini. Understanding the true meaning of that evidence is essential for any coherent explanation of the portolan charts’ origin.
[How do we know that the statements could not describe actual pelagic courses?]
Given Joaquim Alves Gaspar’s discovery that the Liber’s statements were broadly realistic in terms of their positions relative to one another, it is hard to deny that that information had been scaled off a cartographic document of some kind. That, in turn fixes the date of the charts’ geometric structure as being no later than the estimated date of the Liber (c.1210). Since the 196 statements copied out by Liber’s author provide a relatively comprehensive coverage; plotting their positions might effectively re-create that lost urdocument.
[Why the Liber’s pelagic statements must have been copied from a cartographic document.]
The gathering of information about distance and direction would have been far from straightforward, even assuming it was copied from a cartographic document.
[How were the Liber’s pelagic statements constructed? .]
The surprisingly large variation in some of the longer distances in the two early portolani indicate both that they used different sources and that – at the time of the Liber at least – there were no agreed mileage figures for those courses.
[The discrepancies between the stated pelagic distances in the Liber and Lo compasso .]
The longer mileage totals in the Liber could not have derived from any mental process but must rather have been extracted, through measurement, from a rectified diagram. Whereas miles were used for short coastal stretches, pelagic voyages would have been measured in terms of elapsed time.
[Measuring distance in terms of time or miles.]
Distance statements in terms of miles rather than time would have only become possible at all once there was a realistic outline for the Mediterranean and its islands, based on pelagic termini, to provide the necessary authority. It is likely that it was the portolan chart itself that introduced a common measure, the ‘portolan mile’, at least to the open areas of the Mediterranean, as a standard rather than just a mental concept.
[The conceptual issues involved.]
Prompted by Joaquim Alves Gaspar’s visibility chart of the Mediterranean, it is suggested that land sightings at particular points, or their absence, might have led to three distinct methods of data-gathering, all of which could have fed into the creation of the portolan chart.
[The importance of landward visibility.]
There is broad support for the theory, based on cartometric analysis, that the portolan chart was created by merging pre-existing charts of the separate basins (such as the Adriatic and Aegean) but also others that followed the hypothetical outlines of divisions across the open sea. This essay argues against that theory for which there is no historical evidence, whether textual or artefactual, nor any logical support - noting instead that the proposed sub-divisions would have been invisible to sailors and irrelevant for navigation.
[The ‘sub-charts’ hypotheses.]
The Adriatic and Aegean are clearly defined sub-basins whose distortions point to separate information-gathering processes than that carried out for the open Mediterranean.
[The Adriatic and Aegean.]
Neither the portolan chart nor the portolano could have been created from the other: the chart does not provide accurate distances between coastal features and the portolano could not have supplied details of pelagic courses or offshore dangers.
Like humans, the portolan charts had two parents: pelagic pilots for the overall geometry and in-shore navigators for the coastal outlines and toponymy.
[The parenting of the portolan charts.]
Because the direction and distance statements in the early portolani for the open-sea courses could have been derived only from a marine chart, this confirms that the chart was earlier and that part of the information contained in the portolani could not have existed prior to the chart.
[Settling the priority question.]
There are three relevant considerations.
On the basis of those indications it is proposed that the Liber should be provisionally
dated no later than 1210, and that the marine chart used for the information in the Liber would have ended at
Constantinople. This would indicate that a usable chart, covering only the Mediterranean, would have been in
existence before 1204.
[Establishing a terminus ante quem for the portolan chart.]
Working from the other direction, al-Idrīsī’s famous map of 1154 provides convincing evidence that the portolan chart’s origin could have been no earlier than that date since the Arabic map shows no hint of a marine chart, despite 15 years of information-gathering. This point seems not to have been made before. Put together, those place the likely date of the portolan charts’ origins in the half-century between 1154 and 1204.
[The Charta Rogeriana establishes the portolan chart’s terminus post quem.]
It is argued – against suggestions that Genoa or Pisa would be the most likely birth-places for the portolan chart – that solid evidence is needed before settling on any alternative. If the portolan chart was not born in a cathedral precinct or a scholar’s study but in a portside tavern – and likely through the collaboration of pilots with different origins – the hunt for a specific location for the urchart may be a chimaera.
[Where might the portolan chart have originated?]
Once it was realised that the interconnections between pelagic courses, particularly those starting or finishing at the same place, created ‘inadvertent triangles’, the mental chart could have started to turn into a simplified proto-cartographic diagram.
The charts’ geometry could plausibly have come from one source only: pelagic courses memorised by pilots.
The numerous courses sailed by an experienced navigator would probably have been memorised as an interconnected network, instead of or in addition to being stored in the brain as separate routes.
For the Liber’s author to be able to cite distances from all around the Mediterranean, at least the rudimentary outlines of the portolan chart we recognise today must have been completed. Because that pelagic data excludes the Adriatic and Aegean seas, the first attempt at an overall chart of the Mediterranean might have omitted those.
[What might such a pelagic diagram have looked like?]
The way that the putative mariner(s) who devised the portolan chart arrived at that aerial view, extrapolated from the profiles they had observed, is a major perceptual jump.
[The sailors’ viewpoint.]
There is no evidence for a textual source for the portolan chart but the Liber’s puzzling descriptions of the Tuscan island of Pianosa as quadrangular and Cyprus as a double quadrangle may point to its author’s use of a simple pelagic diagram that lacked the infilling of the coastlines.
[The missing link.]
The ensuing pelagic diagram would become more accurate in its placements than any single mariner, or indeed group, however experienced, would have been able to match from their own mental knowledge.
The possible stages of that transmission are described. Once the mental map was in graphic form, knowledge, restricted for millennia to what individuals could absorb in a lifetime, could now be potentially available to anybody, including novices. Where development had been stymied by pilots retiring just as their knowledge reached its peak, incremental improvement was not only possible now but inevitable. Once provided with a portolan chart, a ship’s boy might have started with considerably more information about the hydrography of the Mediterranean than a 60-year-old mariner would previously have had locked in his head.
[The downloading process.]
The inevitable discrepancies in the placement of those pelagic termini that featured in multiple incoming and outgoing transits could, by averaging out and with subsequent checking, have provided the means to rectify the geometric diagram that underlies the portolan charts.
[The rectification procedure.]
Once the demand for the charts grew sufficiently, trained draftsmen and scribes would have been needed to replicate them for sale. The fine and varied detail on the charts called on special skills for what was essentially several different documents in one. To preserve the charts’ main purpose, namely their geometric accuracy, only strict workshop discipline could have avoided progressive distortion. This was the single main key to the charts’ longevity.
[The role of the scribe.]
The inability to locate documentary evidence for the charting of the Black Sea during the period between the sketchy account given in the Liber (c.1210) and the Carte Pisane (c.1270), which already has a full outline, leaves us to conjecture how or when that survey was achieved.
[The shape of the Black Sea.]
Alternative suggestions for the source of the coastal outlines are considered, concluding that they were inserted later as formulaic infilling to a pelagic diagram.
It has recently become apparent that there is a typology of coastline demarcation, including some variations that may hint at how the shore would appear to a mariner.
[How were the coasts demarcated?]
The Carte Pisane’s focus was on seaborne rather than riverine trade, with rivers shown with no more than a pair of short parallel lines denoting the estuary. Yet that would have been important for the merchant users, since many of the larger ports were well upriver. Apart from the main deltas and larger rivers, there was no attempt to differentiate their size or importance.
The seven largest islands in the Mediterranean were treated with the same precision as the continental coastlines, which they functionally resemble; so the position of their defining headlands were probably copied directly. By contrast, for the smaller islands only their position, size and name were necessary for navigation. As a result, a growing number of those were given artificial shapes as a memory aid.
The Aegean arguably represents the portolan charts’ greatest achievement. With no long pelagic courses to provide a geometric framework, and faced with an unusually complex surrounding coastline, the approximate fixing of the positions of the 100 or so islands was probably achieved by a collaborative pooling of knowledge about small groups of them. Unlike the process followed for the rest of the Mediterranean, in this case it seems that the survey must have started with the detail rather than an overall outline.
[Charting the Aegean archipelago.]
It seems likely that the marine hazards of rocks and shoals were of at least equal importance, for both the charts’ origin and use, as the coastal outlines and toponymy. Only a chart could indicate their nature and location with sufficient precision. A few of those dangers are already shown, simply but prominently, on the Carte Pisane. A visual language, with a different symbol for each type of danger, continued to develop up to the end of the 14th century.
The earliest clear reference to a portolan chart, from Egidio Colonna in 1279, noted that sailors “seeing the dangers of the sea and wanting to avoid wrecking their own ships, draw a map of the sea on which sea ports, hazards and other such things are described in proportion …”. The Skerki Bank, probably the single most treacherous feature in the Mediterranean, is named and depicted on the Carte Pisane, as well as being referred to in both the Liber and Lo compasso.
A description of a navigational hazard in a pilot book could never match its graphic depiction or precise location on a chart. During the drafting of a chart a significant amount of time was taken up by the intricate symbology of the dangers, which was faithfully replicated for centuries, underlying the importance of that feature for the charts’ users.
[The treatment of dangers in the portolani and chart compared.]
The later additions of hazards represents, along with the toponymy, the most significant signs of updating. How did the depictions of dangers find their way onto the charts? It is suggested the most likely way would been for a sailor to draw them onto his own chart, in situ, and then have a chartmaker copy off the details when he returned home.
[How were the dangers recorded and how did those get onto the charts?]
A new suggestion is made that fishermen, with their unequalled local knowledge, might have been, at least partially, the source of the information about navigational hazards.
The portolan charts’ innovation in omitting the normal positional dot for the toponyms granted flexibility to the draftsmen about their placement – which sometimes meant the names were spaced out evenly, although still preserving their sequence.
The three earliest extant listings of Mediterranean coastal names – on the Liber, Lo compasso and the Carte Pisane – are largely independent of one another, indicating that at least three data-gathering exercises must have been carried out in or before the 13th century. The portolani could not have derived their coastal itinerary from the charts because the imprecise placement of the toponyms on the latter meant that distances could not have been calculated from them.
[The Carte Pisane’s toponymy compared to that of the portolani.]
Only by combining a number of coast-hugging journeys and using multiple sources, would it have been possible to compile a toponymic itinerary covering so systematically the 22,000 km involved.
[How were the toponymic lists in the early portolani compiled?]
The uncertainty about the meaning of those toponyms that lack a qualifying term – is it a gulf, headland, port or offshore danger, for example? – may be interpreted as constructive ambiguity. It left the user to apply which (or all) the possible meanings, while, by reducing the potential number of toponyms, it lightened the copyist’s load. Portolan chart toponymy has to be seen as often indicative rather than precise, both as to its potentially dual or multiple meanings and its geographical placement – more of a label to jog memory than a unique identifier.
About 90% of the red names on a portolan chart refer to ports, whereas the geometric structure of the pelagic courses had focused on the capes. This underlines the separate development of the charts’ graphic and textual elements.
The combination of orality and memorisation is the key to understanding medieval navigation. Nor can the portolan chart itself be understood without accepting the hugely important role of orality in its origin and use.
[Orality and the portolan chart.]
Whoever purchased an early chart could never have encountered anything like it before.
[What needed to be explained orally for the charts’ effective use?]
There would need to have been oral explanations of several of the features on a portolan chart before it could have been effectively used for navigational purposes.
[Aspects of the charts’ information that required explanation.]
The omission of a number of potentially relevant features underlines how the charts were both indispensable and yet, at the same time, no more than one navigational tool among many oral and memorised ones.
[Information not provided by the charts but necessary for navigation.]
Information held in the head of a sailor would have been far more secure than any simple diagram or scribbled note. Hence it was hardly likely that the oral tradition would have been rendered obsolete by either the magnetic compass or the portolan chart.
[Orally transmitted pelagic knowledge.]
Several parallels are identified with the London taxi drivers, particularly in relation to the need for continuous updating to the respective mental maps.
[London’s Black Cab drivers.]
The similarities of the portolan chart and the London Underground Map are striking: both of them mixed reality with simplified generalisation to provide an indispensable, long-lasting, practical tool for travellers. Each required additional memorised information for their effective use, and each displays the startling originality of exceptional minds.
[The London Underground Map.]
Why does there appear to be an assumption that the prominent place occupied by the charts’ toponymy indicated that the sailor users must have been literate? In any event, most artisanal literature dates from no earlier than the 15th century.
[Would medieval sailors have been literate?]
Assertion: Even if some medieval sailors were literate, the inexperienced among them could not have found in the portolani the knowledge needed for navigation.
[In what way could the written word have contributed to medieval navigation?]
Besides toponymy there is almost no writing on the Carte Pisane.
[Information aimed at literate users.]
It is argued that Italian charts, at least, reduced the textual content to just the toponyms, which would probably have already been committed to memory in many cases.
It seems that there must have been a tacit agreement that linguistic differences should be submerged in a ‘lingua franca’, which allowed broadly standard toponymic forms to be widely shared.
Oral instruction in place of the missing explanation for the charts’ use – and a device such as the mnemonic island shapes – would have avoided the illiterate sailor being disadvantaged.
[Information that could be understood by non-literate users.]
The Carte Pisane’s almost total lack of text besides the toponyms makes it reasonable to suggest that the charts’ creators intentionally produced a ‘nonlinguistic’ navigation aid for illiterate users.
[Were the charts initially designed with illiterate users in mind?]
Being able to envisage the Mediterranean as a whole was crucial to the creation of the charts and the use made of them. Only pilots could have endorsed the charts’ veracity and alerted chartmakers to deficiencies such as the mis-sized Adriatic.
We should not treat the nascent portolan chart as a cartographical production, but rather as a graphical diagram or a scaled outline plan.
[The necessary graphical skills.]
The charts were only devised because they were better than pilots’ memorised knowledge.
There was no intention to create a comprehensive and fully accurate survey but instead a functional tool for navigators to supplement their experience, not make it redundant.
Because there is no sign of any theoretical dimension to the functional portolan chart, we can safely assume that each of its features was included for a specific reason. As a test, try to find anything that could be removed from the Carte Pisane without compromising its purposes.
[The varied purposes of the portolan charts and how they were used.]
One of the main functions of the portolan charts must have been to aid the selection and definition of a route, while at the same time placing that into the wider context of the overall coastal and insular configuration, along with any likely hazards. Local informants could have been an important source of pelagic information. Indeed, harbour-masters and other port-based individuals may thereby have unwittingly contributed to the pelagic diagram that underpinned the original chart.
[Planning a voyage.]
For practical purposes there was little of value to a pelagic pilot in a portolano; however the detailed navigational comments and the stated distances between neighbouring places would have been useful for a coastal mariner.
[The advantages to a navigator of a chart over a portolano.]
Not all areas were suitable for a portolan chart, for example the extensive shoals of the Black Sea and the tidal Atlantic.
[The portolan charts’ limitations.]
Assertion: There can be no discussion of the portolan charts’ origin without considering their purpose(s); or their purpose without looking at usage; or their usage without considering them as objects of commerce.
The cost of a single chart has been estimated in today’s money at around 1,000 euros.
[The portolan chart as an object of commerce.]
Once the charts had become useful for actual navigation, sufficient copies would need to have been produced not only to create but also to maintain a market. There is no evidence that the portolan chart benefited from any external financial or organisational help.
[How might the market have been established?]
The fundamental dissimilarities between the seven surviving charts produced before or during the Vescontian period (1311-30) contradict any argument in favour of a single linear development. Indeed, it is likely that there had been a number of local centres in the pre-Carte Pisane period.
[Not a single lineage.]
The portolan chart must have been reproduced (sometimes with intentional modifications) in greater numbers, not just than any other map but probably any other document besides religious texts.
The tight discipline enforced on a trainee copyist is likely to reveal itself through repetitions of distinctive features in his own later productions. While that presumably applied to Perrino Vesconte, there is no sign of apprenticeship in the other early charts.
A richly ornamented chart is no different in its basic content than a purely functional one.
No convincing evidence has been found for any direct involvement of the trading community in the charts’ creation or subsequent development.
[The possible contribution of merchants to the charts’ origin and development.]
The charts’ content would not have satisfied mercantile needs. However, merchants clearly bought them and must have found them useful for showing the relationships between ports and islands.
[How important was trade in the charts’ origin?]
Portolan charts might be classified as either ‘Functional’ or ‘Augmented’, on the grounds that the former were designed for practising navigators and the ‘Augmented’, with their sometimes lavish adornment, for use, enjoyment or display away from the sea.
[Should we divide the portolan charts into two types?.]
The wholly unrepresentative survival rate of the ‘Augmented’ examples compared to the ‘Functional’ ones means that the former are often – understandably though unhelpfully – the focus of historians, to the detriment of the ‘genuine’ workaday examples, which, for centuries, helped save sailors’ lives. Once plentiful, those are now very rare.
In some respects the charts of the Age of Discoveries were noticeably inferior in precision to the outlines on the earliest portolan chart two centuries earlier. The formulaic treatment of the smaller islands beyond Europe, whose true shape was of little navigational significance, is (unsurprisingly) reminiscent of the Mediterranean conventions that were being perpetuated by the very same chartmakers. This comparison also provides a firm rebuttal of the notion that the medieval seamen would have lacked the capacity to create the portolan chart.