A critical review of the hypothesis of a medieval origin for portolan charts

Abstract

Portolan charts are nautical charts, which appear suddenly in the thirteenth century Mediterranean world, without any discernable development path or predecessors. These charts are remarkable for the high degree of realism and accuracy with which they render the coastlines of their core coverage area: the Mediterranean and Black Seas and the European and North African Atlantic coasts. The shape of the coastlines agrees surprisingly well with a modern map on the Mercator map projection. Until now, scholarly consensus considered portolan charts to be original products of medieval cartographers, based on estimates of course direction and distance measured by seamen during trading voyages. Modern scholars have assumed that the charts were drawn as if the earth were flat. They have attributed the high accuracy of the charts to averaging of multiple measurements on the same routes prior to the mapping process. Thus they have tended to regard the close agreement of the charts with the Mercator projection accidental. This study tests the described consensus hypothesis by evaluating its constituent elements. By using geodetic analysis it is shown in this thesis that the shape of the Mediterranean coastline, drawn from a network of distances and directions between coastal points as if the earth were flat, would differ to a statistically significant degree from portolan charts. The map projection of portolan charts is therefore not an accidental by-product of the drawing method, but an intentionally applied design element of the charts. By means of further geodetic and statistical analysis of five portolan charts this thesis also shows that portolan charts of the Mediterranean are mosaics of a limited number of sub-charts, which have accuracies of the order of 12 km. These sub-charts have different scales and orientations; some have significant overlaps. A realistic quantitative model of medieval navigation developed in this thesis shows that that medieval navigation accuracy was insufficient to create charts as accurate as portolan charts. Accuracy improvement by averaging a series of measurements was not known in the Middle Ages. It is also highly unlikely that the compass, in a form that would have allowed quantitative measurement of course direction, was in general use early enough to have played a role in the assumed data collection process. Arabic-Islamic science was considerably more advanced than contemporary European medieval science. However, no early portolan charts from this culture are known; the few extant charts appear to be copies of earlier European charts. Although astronomic positioning was steadily improved in Arabic-Islamic culture, it never achieved the accuracy of portolan charts demonstrated in this thesis. Despite its substantial scientific capabilities and achievements, Arabic-Islamic civilisation proves a highly unlikely origin for portolan charts. The key conclusion from this study is that portolan charts are sophisticated geodetic-cartographic products, of which the creation was well beyond the capabilities of medieval European cartographers. Thus the scholarly consensus on the medieval European origin of portolan charts is shown to be incorrect. As the charts were not created in Arabic-Islamic civilisation either, their origin must lay further back in time.