The origin of nautical cartography: certitudes, doubts, and perplexities

ABSTRACTEuropean geographical maps and nautical charts have distinct geneses and underwent separate evolutions. When scientific cartography was reborn in the beginning of the fifteenth century, following the translation and dissemination of Ptolemy’s Geography, the portolan chart had already been established as an effective navigational tool for over one hundred years. And while new geographical maps started to be constructed by the erudite, using the coordinates and prescriptions given by Ptolemy some thirteen centuries before, charts continued to be made by artisans, on the basis of navigational data collected by pilots. Despite being a unique achievement in the history of cartography, if not in the history of civilization, the creation of the medieval portolan chart was still poorly understood at the turn of the twentieth century. With the introduction of new research tools and a deeper study of the extant sources, we have now reached a situation where the initial state of perplexity concerning the very emergence of the portolan charts has given way to a broad consensus about some central questions. The purpose of this article is to appraise the present state of knowledge regarding the so-called ‘origins problem’.RÉSUMÉLes cartes géographiques et les portulans européens ont une genèse et des évolutions distinctes. Lorsque la cartographie scientifique renait au début du XVe siècle, après la traduction et la diffusion de la Géographie de Ptolémée, les portulans étaient déjà un outil de navigation efficace depuis plus de cent ans. Alors que des nouvelles cartes géographiques commençaient à être construites par les érudits, en utilisant les coordonnées et conseils donnés par Ptolémée quelques treize siècles plus tôt, les cartes marines continuaient à être faites par des artisans, à partir de données de navigation collectées par les pilotes. Bien qu'il s'agisse d'une réalisation unique dans l'histoire de la cartographie, si ce n'est dans l'histoire de la civilisation, la création de la carte portulan médiévale était encore mal comprise au début du XXe siècle. Avec l'introduction des nouveaux outils de recherche et une étude plus approfondie des sources existantes, nous avons maintenant atteint une situation où l'état initial de perplexité concernant l'émergence même des cartes portulans a laissé place à un large consensus sur certaines questions centrales. L'objectif de ce papier est d'évaluer l'état actuel des connaissances sur ce que l'on appelle, 'le problème des origines'.KEYWORDS: History of cartographyhistory of navigationportolan chartnautical chartcartometric analysis AcknowledgmentsSome of the historiographic developments of the last few years, concerning the genesis of nautical cartography, were triggered by an informal discussion, held by email, between collaborators of the Medea-Chart project: Corradino Astengo †, Dick Pflederer, Gregory McIntosh, Michael Barrit, Tony Campbell, and Wolfgang Koeberer, to name just a few. The arguments exchanged during the debate, which lasted for several months, drove us to revise some of our interpretations, opening new paths for research and eventually crystallizing in new publications. To all colleagues involved in that enriching debate I express my deep appreciation and gratitude some of recent historiographic developments concerning the genesis of nautical cartography were triggered by an informal discussion, conducted by email, between collaborators of the Medea-Chart project. These included, among others, Corradino Astengo †, Dick Pflederer, Gregory McIntosh, Michael Barrit, Tony Campbell, and Wolfgang Koeberer. The arguments put forth during this extended conversation drove us to revise some of our interpretations, opening up novel avenues for research and crystallizing in new publications. I wish to express my deep appreciation and gratitude to all the colleagues who participated in this enriching exchange.Disclosure statementNo potential conflict of interest was reported by the author.Notes1 The hyperspectral analysis of the manuscript was made in 2019, in Paris, following a proposal of project Medea-Chart. The results, which haven’t revealed any surprises, were made available by the Laboratoire de restauration des musées de France, together with those of the radiocarbon dating and pigment analysis: https://merovingio.c2rmf.cnrs.fr/pisane/ (Retrieved January 2023).2 Some recent bibliography about these three charts is mentioned in Gaspar (Citation2019, note 19). For details and images of these early charts, see the Medea-Chart database: https://medea.fc.ul.pt/view/chart/50, https://medea.fc.ul.pt/view/chart/54, https://medea.fc.ul.pt/view/chart/358.3 In fact, there is a slight negative trend in the tilt of charts between ca. 1310 and before 1600. Its likely explanation is the progressive correction of a northward displacement of the eastern Mediterranean and Black Seas, which mostly affects the older charts. See, below, ‘The Sub-chart Theory’.4 In fact, this procedure would probably be unfeasible using the portolan charts that survived to our days, due to their small scale. Cotrugli wrote his De Navigatione for the Venetian Senate, with the purpose of composing a treatise on navigation. The original was lost but two copies are extant: Beinecke Rare Book and Manuscript Library, MS 557 and Lawrence J. Schoenberg Collection MS LJS 473 (Falchetta, Citation2009). A commented excerpt of the parts dealing with the navigational chart is in Almeida (Citation2022, pp. 53–60).5 The process aims to solve, using a table or diagram, the right triangles associated with the track of a ship at sea and the deviations from the planned course. Solutions are provided for the eight classical rhumbs, separated by one point (11 ¼ °).6 As noted by Tony Campbell, by the end of the fourteenth century several Mediterranean maritime authorities had decreed that every ship must carry at least two charts on board (Campbell, Citation1987, pp. 439–440).7 This subject will be addressed below, when discussing how the charts were made.8 That is the signed and dated chart of the Portuguese cartographer Jorge de Aguiar (1492). See Gaspar and Krtalić (2023, Ch. 11) and the Medea-Chart database: https://medea.fc.ul.pt/view/chart/4.9 That is not the opinion of Tony Campbell (personal communication), who considers that the reason Neckam referred to those devices was because they were novel. On this point, we may argue that their use would only have come to the knowledge of Neckam, an English cleric working far from Italy, if they were relatively ubiquitous by the time. In fact, he may have known about the floating compass while he was teaching in Paris, before 1186 (Beazley, Citation1911).10 The Compasso de Navegare is preserved in the Staatsbibliothek, Berlin (MS Hamilton 396) and bears the date of 1296. Based in its content, Motzo (Citation1947) considers the manuscript to be an updated copy of an older original, made before 1256.11 Because of the contamination caused by northward displacement of the eastern Mediterranean, especially in the older charts, it was decided not to use the line connecting Gibraltar to Antioch, as in Figure 1.12 This feature also agrees with the results of the analysis made by Roel Nicolai of the data contained in the Compasso de Navegare. See Nicolai (Citation2018).13 This anomaly was previously noted by Gautier-Dalché (Citation1995, p. 80) and by Kelley (Citation1995), who justly asked “Could it be that portolani owe more to portolan charts than vice versa?”. In a contribution focused on the analysis of the Compasso de Navegare, Nicolai (Citation2018, pp. 3, 5, 18) found numerous other instances of “impossible course legs”, and echoed Kelley’s suggestion that portolans were likely based on data scaled from charts.14 That is, by observing the Sun and stars (especially the Pole Star) at the moments when their geographic directions had been previously tabulated by the astronomers, such as culminations, risings and settings. See Gaspar (Citation2019, pp. 10–11).15 This is a difficult question, whose clarification should take two related facts into account: the major differences between the toponymic sequences in the Liber, Compasso de Navegare and Carte Pisane (Campbell, Citation2011, ‘Toponymic Transmission Before 1311 ['Precursor Names']’); and the compelling evidence that the earliest known portolan charts (or, more precisely, their prototypes) may have been made as piecewise assemblages of regional charts. See below, ‘The Sub-chart Theory’).16 Loomer (Citation1987, p. 167) suggested that the tilt of the charts was a side effect of the triangulation scheme, while Nicolai (Citation2014) considered that the sub-charts used to construct the representation of the Mediterranean and Black Seas were made using geodetical data and the Mercator projection.17 The idea is easy to understand if we imagine that one wants to determine the relative positions of a number of cities in a country, knowing only the distances between them. This is what is called multidimensional scaling in social sciences, a process equivalent to the least square adjustment used in Geodesy and many other disciplines.18 These parameters were then fine-tuned in successive runs of the model until the resulting grid of meridians and parallels became as close as possible to the ones implicit in the old nautical charts. Establishing a maximum distance is justified by the fact that few trips would have been made across exceptionally large distances in the Mediterranean, without a scale. The inclusion of a weighting factor aimed to better identify the actual data used to construct the charts. I concluded that a large preference (80%) was given to directions over distances (20%), which is a reasonable and expected result, considering that mariners must have trusted more the courses measured by the magnetic compasses than the estimated distances.19 This grid can be estimated from a set of points of known present day latitudes and longitudes, positively identified in the old chart and a modern map. The numerical process was facilitated by the use of MapAnalyst, a computer application designed for the purpose of comparing maps, freely available on the internet: https://mapanalyst.org/.20 It is important to remark that the slight convergence of the meridians in the old chart is explained by the usage of distances in its construction. Such convergence would be absent if only directions had been used.21 Indeed, even a medieval navigator would have given much more weight to compass directions than to estimated distances, due to the uncertainly associated with the latter. That is the reason why, following the introduction of latitude measurements in navigation, in the last decades of the fifteenth century, positions at sea started to be determined using observed latitudes and compass directions, rather than latitudes and distances.22 See Nicolai (Citation2016a, pp. 8–11). As previously noted, the author’s explanation for the tilt affecting all portolan charts up to ca. 1600 was that it was introduced by the same map makers to make the resulting chart agree with the compass directions.23 Although the chart was not obviously constructed on geographical coordinates, an arbitrary grid was created on the equirectangular projection centered at the 36° N parallel, from which latitudes and longitudes on the chart could be read. The grid was fixed at Punta Tarifa (36.0° N, 5.6° W), where both the latitude and longitude errors are zero.24 As previously noticed by other researchers for a few charts, less obvious subdivisions can be identified using a finer analysis, but such an exercise would be beyond the point I want to make here. See, for example, Loomer (Citation1987, p. 159) and Nicolai (Citation2014, p. 269).25 These two irregularities were also noticed by Nicolai (Citation2014, Ch. 7) on his cartometric analysis of five charts, but the corresponding control points were rejected as outliers.26 See Gaspar (Citation2010, p. 81, Citation2011, p. 237). It is noteworthy the fact that the Mercator projection was first proposed by Gerard Mercator in his planisphere of 1569 but was only fully adopted in nautical cartography well into the 18th century, after the longitude problem was solved. Its application to mapping in the Middle Ages would necessarily imply that geographical coordinates of some places could be determined by astronomical methods. Even assuming that such an undertaking was within the reach of the hypothetical older civilization postulated by Nicolai, why choose the Mercator projection for mapping small regions in the Mediterranean?27 This was more recently also proposed by Tony Campbell (Citation2021–Citation2023, ‘I.4c. The rectification procedure’).28 See Nicolai (Citation2021). The fact that the Portuguese charts of the late fifteenth and early sixteenth century were based on various sources of data was first noted and discussed by Gaspar (Citation2010), and then revisited in Gaspar (Citation2012, pp. 191–194) and McIntosh & Gaspar (Citation2021, pp. 167–174). The detailed analysis and conclusions made in these works about the origin of those data, and how they shaped the geometry of the charts, were apparently disregarded by Nicolai.29 Muhammad al-Idrisi’s Tabula Rogeriana (or Book of Roger) was commissioned by the Norman King of Sicily, Roger II, and completed around 1154. The book is written in Arabic and contains seventy maps covering the world known at the time. Al-Idrisi was certainly influenced by Ptolemy’s Geography, which had been translated into Arabic by al-Khwarizmi (ca. 780 – ca. 850) and al-Battani (ca. 858-929). Such an influence seems clear in the small round representation of the ecumene which accompanied later editions of the book. See Medea-Chart database: https://medea.fc.ul.pt/view/atlas/660, https://medea.fc.ul.pt/view/chart/238.30 See Campbell (Citation2021–Citation2023, ‘B.3b.1. London’s Black Cab drivers’). The situation of the taxi drivers cannot, strictly speaking, be taken as equivalent to the one of our medieval pilot because, as Campbell noted, those drivers learned much of their memorized information from maps. The necessity of keeping that information in their minds resulted from the simple fact that a paper map could not be safely consulted with the car in motion. As we know, the limitation has been resolved with the present-day digital maps and GPS systems.31 Campbell (Citation2021–Citation2023). The fact that most pilots were probably illiterate is used many times throughout the text to argue against textual support for navigation. See, for example, ‘L.3. Literacy or Illiteracy?’.32 Even if the ancient periploi were not used in navigation, as asserted by Campbell (2023,’ C1. Navigation in the Ancient World’), the similarities of organization and content with the medieval portolan suggest that some continuity did exist.33 As noted above, there is a difficulty here, related to the possible connection between the alleged proto-portolan chart, based on astronomical directions and representing the whole Mediterranean Sea, and the later regional portolan charts, based on compass directions. Rather than considering that the latter evolved naturally from the first, it could well be that they were mostly unrelated processes.Additional informationFundingThis project received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement 714033-Medea-Chart / ERC-2016-STG).Notes on contributorsJoaquim Filipe Figueiredo Alves GasparJoaquim Filipe Figueiredo Alves Gaspar is a Captain of the Portuguese Navy (retired), specialist in navigation, mathematical cartography, and Geographical Information Systems, presently an investigator in the Interuniversity Center for the History of Science and Technology, in the Faculty of Sciences of the University of Lisbon, Portugal. In the last six years, he was the Principal Investigator of project Medea-Chart, supported by the European Research Council, whose main scientific goal was to study the birth, technical evolution, and use of nautical charts during the Middle Ages and Early Modern periods.