WWW-Virtual Library: History
Map History / History of Cartography: THE Gateway to the Subject

WWW-VL Main Catalogue WWW-VL History Central Catalogue
(main menu)


What the
site is
Map History

Explanatory notes and wider implications


'The colours and shapes used to denote some of the smaller islands
and the major estuaries on portolan charts up to 1500'

Mounted on the web 7 March 2011 - additions and corrections are noted in the appropriate place with a dated statement between { }, which can be searched for

Colour Feature List   |  Summary table of the features analysed (a Microsoft Word document)

Colour & Shape Analysis Menu   |  Portolan Charts Main Menu

For the full details of the works mentioned below see the Bibliography



This page and its attendant tables describes the 'Colour & Shape Analysis' (C&SA). Its over 100 features were carefully checked on virtually all the portolan charts up to 1469.
After introductory comments, the features are examined individually (by type), concluding with a section on colour consistency.
The final part, Wider Implications, seeks to extract meaning from the varied evidence and then build a new interlocking theory about chart production and use

This prints out to about 38 pages



Conventions used on this page

The sampled features
The charts examined

COLOUR: The questions
'Reading' the colours
How the colour was used and applied
Colour as an indicator of authorship and lineage

SHAPE: The questions
The different approaches to conveying shape

(Introduction and summary) Majorca
Smaller Aegean islands

Crosses (shape)
Other forms

The questions
Name labels

Other colour conventions

Majorca's stripes
Crosses (painted)
Buttons and other island centres

Detached initial letters
Cautionary note

Consistency conclusions

Benincasa's consistency

His use of brown, mauve, pink, etc.
His other apparent inconsistencies




    The overall value of the findings from the Colour & Shape Analysis

Are 'lollipops' navigational symbols?

Island shapes as a mnemonic device

A deeper understanding of the function of portolan charts

What is the difference between a plain and a de luxe chart?

Conservatism and workshop practice
     Geometric consistency
     Where is the expected conformity?

Expectations contradicted by the C&SA findings

Sixteenth-century continuations



A square-bracketed figure indicates the number of that feature in the Colour & Shape Analysis (C&SA). For further details on those see:


Throughout these pages the terms draftsman, scribe and colourist are used to differentiate three distinct aspects of chart production: the inking in of the coastal outlines, the writing of the place-names and (sometimes) the legends, and the colouring of the islands [on which see Stages in the construction of a chart]. This convenient shorthand is not intended to mean that three people with different specialised training were necessarily involved. Indeed it seems likely that most chartmakers would have been proficient at all three activities and many (perhaps most) charts would have been the work of a single individual. However, the embellishments associated with Catalan-style productions might well have been carried out by a specialist artist.

References in the descriptive notes:

For references to the Colour & Shape Analysis see the C&SA Menu - from which you can access the 13 Microsoft Word tables

See also the summary Conclusions page

For the full bibliogaphical references see the Bibliography



This study, to which I have given the title Colour & Shape Analysis (C&SA) was prompted by an article by Paul Harvey (2009) on the use of colour on medieval maps. Towards the end he mentions portolan charts:

Harvey mentioned several features relating to the work of Benincasa. I realised that, despite having already spent considerable time examining the Benincasa works on the Pujades DVD (see The style and content of Grazioso Benincasa's charts: imitation, innovation and repetition), and having observed his use of colour to distinguish the British islands, I had not thought to look further into that (in retrospect, very obvious) aspect of Benincasa's charts. This study freely acknowledges, with thanks, Paul Harvey's stimulus. As it turned out, Benincasa's 'visual signature', which until then I thought was restricted almost entirely to the British Isles, could now be amplified by a 'chromatic signature', related to several Mediterranean and Black Sea features. In addition, several drafting conventions could now be shown to be part of his personal signature.

Benincasa proved to be so consistent in his use of colour over the smaller islands and major estuaries, not to mention the shapes he gave to those features, that some works tentatively attributed to him by previous scholars could be confidently confirmed as his work, and others more tentatively linked with him shown to be the work of (presumably unauthorised) copyists. For a ready demonstration of Benincasa's regularity in his treatment of the sampled features, in an overall pattern quite unlike that of any other earlier or contemporary chartmaker, see Colour and shape on the work of Benincasa, and Colour Consistency Tables (both are Microsoft Word documents). Because of the conservatism of his successors, those Benincasan conventions (though not always his colour choices) would continue for almost a further century.

It needs to be underlined that such a broad study could not, realistically, have been considered before the comprehensive Pujades visual corpus became available on DVD in 2007. So, further gratitude to him, to the libraries that gave their permission, and to the far-sighted institutions that financed his ambitious project.

These pages comprise the 'Colour & Shape Analysis' (C&SA), a series of comparative Microsoft Word tables with accompanying commentary. [For access to those, see the Colour & Shape Analysis Menu page.] In the past, the visual focus of historians has tended to be on 'macro' features, such as flags, town vignettes, compass roses and wind heads, scales and border designs, tented rulers and other prominent illustrative elements, often found only on Catalan, or 'Catalan-style' charts. The intention here is to look, instead, beneath the radar to a range of sampled elements, present on almost every chart, which while individually and historically of little significance add up, collectively, to an important diagnostic tool. Some of the features analysed are so small they cannot be seen in a typical book illustration and require magnification on an electronic scan. This helps to explain why they have hitherto been overlooked.

The colouring of islands forms as fundamental a part of the portolan chart construction process as any other. To apply the correct colour with the required care needed special training, whether as part of a general apprenticeship or for specialist workers. If there was a colour pattern it would have had to be followed; alternatively some or all of the colour choices might have been memorised. Because those colours swiftly came to be thought essential, firstly, for distinguishing visually the coastline from offshore features, and, secondly, one island from another, no work (after the Carte Pisane and Cortona chart) is without coloured islands, right up into the 18th century. Previously, on the mappaemundi and Ptolemaic maps for example, it was generally the sea that was coloured.

It seemed likely, as a result of the long and meticulous training of apprenticed chartmakers, that they would have learnt, and retained afterwards, specific ways of handling a range of features. If my surmise was correct, this would comprise a 'constructional fingerprint', much of which the future chartmaker was likely to carry with him into his own work and teaching. So what the C&SA sought to test is whether there is indeed evidence to support the hypothesis that there are 'visual', 'constructional' and 'chromatic signatures', whether of a single chartmaker, a chartmaking family or a production centre.

In some ways this project parallels my earlier investigation into toponymy (in the 1987 chapter in Volume 1 of The History of Cartography). That aspect had been the focus of research for a century or more, but usually in terms of a specific chart, chartmaker or small area. There had been no attempt to seek out general patterns across the entire portolan chart region and across time, or to identify innovations. When the patterns in its data came to be analysed one of the results was to flag up unsuspected similarities between works which proved to have shared authorship ['Chapter' pp. 402-3]. The C&SA results should be seen as complementing those earlier toponymic findings. Might any chromatic signatures help in a similar way with the attributions of unsigned works?

Much portolan chart description has been just that: description rather than evaluation. Usually the commentator ranges across the chart noting the more obvious visual features. There was no handy reference guide available to indicate - as it might, say, in a handbook to birds or trees - which elements were usual, which rare and which unique. Nor, without a checklist, was it likely that mention would be made of the absence of an expected feature. The aim of the C&SA is to provide such a guide and checklist, though necessarily on a selective basis. Only armed with such a large body of systematically analysed data, drawn from almost all surviving works of the period up to 1469, is it possible to offer credible conclusions.

If the charts' toponymy can be considered their lifeblood, then these small, sometimes minute features are equivalent to their genetic make-up. And, just as the elements of a DNA sample are judged on their ability to pinpoint a common origin (even if the strands used for identification are otherwise insignificant) so all that matters here is that a clearly defined (non-accidental) feature can be shown to be present in some, but not all, of the charts examined. Human DNA is passed via breeding; portolan chart DNA is spread through copying. So locating these elements, most of which have fallen beneath the notice of previous commentators, can help to track the spread of these small, and in themselves thoroughly unimportant, mutations.

Only by studying everything (or as much as is available) is it possible to spot the intermittent occurrences, that is, those elements that contradict any neat notion of steady development. The C&SA throws up several instances where a feature reappears after an absence of many decades or even a century, whether because the later chartmaker had seen a much earlier work or because the convention was re-invented.

The following statement is an obvious one but nevertheless needs to be said. All conclusions in these interconnected pages must necessarily be tentative because the surviving record represents a tiny fraction of what was produced. In addition, it is probably unrepresentative, and it certainly leaves large gaps in our knowledge. When I, or anybody else, says that a certain feature first appeared on a particular chart this must always be a shorthand way of saying, ponderously: this feature was first seen on chart X by chartmaker Y but it is quite possible (indeed highly likely) that it had appeared on an earlier lost work of that practitioner, and perfectly possible that he should not have the credit for the innovation anyway because it might have been introduced by somebody quite different, of whose work and name we are ignorant.

Any who write about early portolan charts must be careful not to over-personalise achievements in case a resurrected chart forces a radical reassessment in the future. Although this did not involve a newly discovered production, but rather a reinterpretation of what was already known to have existed, the main plank of Grazioso Benincasa's fame - that he was personally responsible for bringing the Portuguese west African discoveries to the wider world - must now (it is argued in the Benincasa essay) be transferred to (or shared with) his Catalan contemporary Petrus Roselli.

However, providing that whatever evidence we have is approached with an open mind, we need not be wholly inhibited from putting forward conclusions that can advance the understanding of our subject. The Colour & Shape Analysis (C&SA) data does provide concrete evidence, of particular value when dealing with those practitioners who have left us more than a single work. Where it has revealed consistent patterns (or even the opposite) this justifies some new ways I am suggesting of looking at the early (and even later) charts. It is my contention that that exercise helps to reaffirm the importance of the portolan charts themselves, in the general absence of supporting documentation, as the primary source of evidence about their own history.

Top of page


Much thorough work has already been carried out into the hydrographic development and the more prominent visual elements typical of Catalan-style charts. The features considered in the C&SA are not, generally, important. Indeed, some are apparently so trivial that it would have been surprising if anybody had examined them before.

The analysis initially looked just at the colour used to distinguish islands and estuaries but, as it proceeded, it became clear that the changing, or alternative, shape given to some of these features was also of diagnostic value, and so those were included as well. Those outlines, once established, remained relatively unchanged, sometimes for centuries, through the forces of conservatism inherent in a process of continuous, faithful copying. The spotlight was therefore directed on those early practitioners - mostly during no more, perhaps, than the first thirty years of the 14th century - who developed most of the forms that would then become unthinking conventions.

The larger islands were usually left uncoloured because they carried internal place-names which would have been obscured. So the selected features were either medium-sized islands or the largest estuaries (Rhône, Danube, Dnipro and Nile). Since the estuaries comprised multiple elements (which were given formalised island shapes) the total number of recorded features (not all of which were present in any single work) came to about 100, far more than the 51 numbered entries.

Features were selected because of their visual prominence and in order to provide sufficient geographical range so as to ensure that incomplete works might still be considered. There are certainly other islands and other conventions that could usefully be studied in a similar way.


An attempt was made to include all the works featured on the Pujades DVD, i.e. almost all those definitely or supposedly produced before 1470, and any of those dated works (1470-1500) for which reproductions could be found. Finally, a selection of 16th and 17th century works was sampled to gain an idea of how long some conventions persisted. Those dating from up to around 1550 that were sufficiently legible are analysed in full (see the Colour & Shape Analysis Menu page).

Top of page



The exercise set out to seek answers to the following questions about island and estuary colouring:


This exercise worked at two removes from the originals, examining scanned images on a DVD, created from photographs taken in a large number of different institutions. This must certainly have introduced minor distortions. But there is a bigger problem.

"The difficulties are real; we can have total confidence only in the examination of original maps with colour standards and with knowledge of the way pigments may have changed in the course of time" (Harvey, 2009 p.52).

This is endorsed by what Cheryl Porter wrote in 1992. '... using visual analysis alone, there are very few who can say with any degree of certainty exactly what a pigment actually is' (p.111) and, elsewhere, 'the colors that we see now when we open a manuscript may not be the same as those that were originally applied ... it is known that vermillion, red lead, lead white, and silver are all prone to blackening' (p.114). See Cheryl A Porter, 'You can't tell a pigment by its color', in. Linda L. Brownrigg (ed.), Making the medieval book: techniques of production. Proceedings of the Fourth Conference of the Seminar in the History of the Book to 1500, Oxford, July 1992 (Anderson-Lovelace; The Red Gull Press, 1995), pp. 111-16. [For references to medieval manuals on illumination see Woodward (1987), p.324.] {This paragraph added 28 April 2011 & amended 12 September 2011}

The most immediately obvious transformations were of green to a muddy brown, and, it would seem, silver to black. Other apparent mutations are of blue towards black, or of gold (paint not leaf) turning brown.

Some colours tend to remain clearly distinct - red, green (with the effect of verdigris?), blue, white, black. Despite that, blue and green can occasionally become hard to separate and I may have mis-identified some of those. Other, less definite tones, can be difficult to describe or distinguish, for example pink, mauve, lavender, purple. These are not primary colours and would have been mixed up specially in the workshop, leading to likely variation. While a trained colour historian could no doubt assign, with some confidence, a precise name to all these variants it is arguable whether this would be of much value to the map historian. Astengo (2007(a)) p.188 noted that examination of the Behaim globe of 1492 found no evidence of 'coloring agents formed by the mixture of different ingredients'; perhaps this points to the chartmakers' need for a wider range of colours. {The previous sentence added 12 September 2011}

The differences we are seeing in these less certain tints may reflect the effects of time or of the reproduction process. No doubt in some cases others would see a different colour to the one I record, and some readings could be improved by examining the original. In attempting to assess the consistency of medieval chartmakers I will almost certainly have added further confusion by sometimes describing as the same tone some colours that should have been kept distinct, and vice versa. With no ability to download samples from the Pujades DVD, I could only work from memory.

But, if we interpret these colours as the result of a conscious selection by the chartmaker, what perhaps matters more is not that the choice was between, say, pink or mauve, but that it was definitely not one of the clearly distinct colours mentioned above. In other words, we can claim some justification for combining some of the indeterminate tints (e.g. mauve and purple) as being in their own portmanteau category.

Atlases - offering protection from light and atmosphere - are more likely to preserve their original tonal quality. For examples of clear, vibrant colours see A 34 on the Pujades DVD (the 1465 Benincasa atlas in Vicenza) or his C 67 (the 1466 Roselli chart in Minneapolis). Surface abrasion can often be seen on charts, with some or most of a patch of colour removed. Under-colour was evidently used (always?). Sometimes that shows through confusingly, or may even be all that remains visible.

Besides definite colours that could be noted, some coastlines have an indeterminate outlining wash, a muddy colour that could be read as yellow, green or brown. Occasionally, islands were treated in a similar fashion, at times so faintly as to raise doubt whether a tint had been applied or not. Those instances were recorded as uncoloured.

On a single occasion, on sheet 2 of the Ziroldi-style atlas in the British Library, Add. MS. 18665 (Pujades A 27), I noticed two different blue tones, with a light version for Hierro and a darker hue, almost purple, for two of the other islands. Otherwise, it seems that a single version of a colour was always employed.

Rather than speculate about the original colour, I have generally noted what can now be seen. As already mentioned, the ageing process has sometimes brought the blue and green closer together. A clue can be found in the colours of the rhumb lines. The eight main winds would usually be drawn in black or brown, the next eight half-winds in green, and finally the sixteen quarter-winds in red. Ramon Pujades has pointed out to me (in a private communication, 14 May 2010) that the rhumbs would have been drawn in ink with a pen, whereas the island colour would have been brushed on with paint, whose manufacture and consistency would have been different from ink. Nevertheless, there are occasions where the green for the rhumb lines seems to match that for the islands.

This may be supported by what Cheryl Porter observed in 1992 about 'pen flourishing'. 'It is logical to expect to find the same sorts of pigments used for these as for the illuminations, but little work has been published on this question ... any coloring substance which would flow easily from a pen could be used'. See Cheryl A Porter, 'You can't tell a pigment by its color', in. Linda L. Brownrigg (ed.), Making the medieval book: techniques of production . Proceedings of the Fourth Conference of the Seminar in the History of the Book to 1500, Oxford, July 1992 (Anderson-Lovelace; The Red Gull Press, 1995), pp. 111-16, especially 115. {This paragraph added 28 April 2011}

In time, non-invasive ways of sampling the chemical composition of portolan chart colours will no doubt be used to recreate their original appearance [on which see the abstract of a paper by Fenella France, 'Scientific and image analysis of portolan charts: Preliminary results and methods', delivered at the Library of Congress on 21 May 2010]. In the meantime, this analysis has had to rely on a pair of untrained eyes.

Top of page


Alberto Magnaghi, in his entry 'Nautiche, carte', in the Enciclopedia italiana di scienze, lettere ed arti (Rome, 1929-39), vol. 24 (1934), p.324b, described how the larger islands are always in green, azure or wine-red, and the smaller ones in gold or green so as to distinguish the islands from the mainland ('Le isole, invece, sono sempre, le maggiori in verde o in celeste chiaro o in rosso-vino, le più piccole in oro or in verde: sempre in modo che, grandi o piccole, risultino di colore diverso da quello col quale son segnate le coste della terraferma').

Magnaghi's suggestion of such a limited range of colours is not generally borne out by the 'Colour & Shape Analysis' (C&SA), though it is a true comment on some of the very early charts (such as the restriction to just red and blue on the recently discovered Lucca chart), and on a few later exceptions represented by Italian (presumably Venetian) atlases from the first half of the 15th century. From what could be distinguished on the Pujades scan (A 26), the anonymous 15th-century Venetian atlas in the Marciana Library, Venice, It. IV, 493, appeared to use gold almost exclusively, and the Bodleian's atlas (A 13) is restricted to just blue and green. The atlas in Lyons, MS 179 (A 28), used just red and green, as does the Arabic copy of a western chart, the so-called 'Maghreb' chart.. But, in general, to distinguish one island from its neighbours - not just the islands from the adjacent coastline as Magnaghi suggests - a much wider range of colours was used. That is summarised in the ColoursUsedTable (a Microsoft Word document).

Even though artists have long distinguished as the three primary colours, red, yellow and blue (plus black and white), only red seems to have been used invariably on portolan charts, once the effectively uncoloured Carte Pisane and Cortona chart are excluded. The absence of blue from three early works - the 1311 Vesconte chart, the earliest surviving Genoese chart (Riccardiana, Florence) and the Dalorto/Dulceti chart of 1330 - is noteworthy, and has possible significance for dating purposes, as is discussed on 'Anonymous works and the question of their attribution to individual chartmakers or to supposed workshops' under entries [1-5]. Blue was an essential part of the palette of later charts.

As demonstrated by the ColoursUsedTable (a Microsoft Word document), most charts used a total of between at least five and eight colours, including, in some cases, the application of black or white paint, for example to form the cross habitually placed over the red-coloured Rhodes. The biggest challenge, for those who depicted the Danube delta as a mass of islands, was to find enough colours with which to differentiate them.

Silver is difficult to detect and is likely to have tarnished. It may indeed lie beneath some of those instances which now appear as black. This would be a plausible explanation for the alternate white and black recorded for Lanzarote and Rhodes on the usually highly consistent Benincasa (Table C: a Microsoft Word document). The white cross of Rhodes could be more simply (and cheaply) achieved by just letting the vellum show through, since, heraldically, white and silver are the same. Isolated instances of the possible use of silver have been detected on Ibiza (R. Soler), Majorca (Roselli), Malta (B. Beccari) and Rhodes (Pizzigani, Vallseca and Pareto) and it seems to appear occasionally in various places on Vesconte's work.

Because of its cost and time-consuming application, we might assume that the use of gold implied a higher status for the island concerned. This does not seem to have been the case. Where gold leaf can be clearly seen - for example, on the Catalan Atlas - it was applied for decorative not political emphasis. This seems to have been the case also with what appears to be the more usual gold paint, for which yellow may have been a substitute, particularly, in the case of Benincasa, when he came to draw a chart rather than an atlas (unless this was a misreading by me). Such gold paint, formed by grinding the mineral into a powder and then mixed into paint form, tends to have a grainy appearance. Gold leaf, by contrast, would usually be burnished. It is presumably coincidental, in relation to the occasions when Benincasa substituted red for gold, that he seems to have followed the usual artistic device of placing the gold over red underpaint. This can be seen for example in three of the Aegean islands on his 1465 atlas (Pujades A 34, sheet 5). {Two preceding sentences added 19 December 2011}

What needs to be acknowledged is the skill and care with which the colour was applied, particularly by those chartmakers who signed their work. The match of colour to the inked outline is far more exact than that of the later colourists of printed maps. Given the ability by the Pujades DVD to enlarge an image to several times its original size, it is possible to see, for instance, the meticulous shaping of the ends of the cross used to denote Paxos in the Ionian Islands [31]. The smallest islands might be entirely formed of colour, without any outlining, just as shoals were denoted by means of pointillist patterns in paint. Pujades discussed the artistic quality of some Catalan work, under the heading 'Mastery of the Paintbrush' (2009, p.352).

Top of page


The detailed results of the C&SA, which considered almost all the surviving works produced up to 1469, are displayed on a sequence of Microsoft Word tables [accessible via the Colour & Shape Analysis Menu page]. Had the choice of colour been left to the whim of the chartmaker, or even to his apprentice, the random results would be immediately obvious, because the tables use coloured type, so that, for example, the notation 'Be' is always itself coloured blue. In most cases, closer examination often reveals considerable consistency when reading along a row for a specific feature, but there is also a certain amount of variation.

That first, cursory, examination would be enough to remove the possibility of habitual randomness within the work of a single chartmaker. We are definitely dealing with conscious, rational decisions. A slightly more careful look would confirm that there were colour conventions shared by different practitioners working in the same place and, sometimes, with those elsewhere as well. Even if an experienced chart colourist might have worked from memory, it is evident that some (probably a majority) of the colour choices were those he had learnt during his training, presumably from a special colour pattern.

There are nearly always one or two exceptions to any generalisation but the following comments on the conventions habitually found on Catalan charts gives an idea of the level of consistency in work created in the Majorcan capital, Palma. The numbers are those in the Colour Feature list (from which this information is extracted):

18. Formentera - red (except Roselli)
19. Ibiza - usually blue
20. Majorca - red and gold stripes NW/SE (never found on Italian work)
21. Minorca - red (green is found only on Benincasa and one other Italian work)
29. Gozo - red (most Venetian charts did not use red)

Ionian Islands:
31. Paxos - always red
34. Kefallonia - mostly blue or uncoloured
35. Zakynthos - red

Aegean islands:
36. Limnos - blue
39. Skyros - always red (also Benincasa, while most other Italians used blue)

Italian productions do not lend themselves to such neat generalisations. There are relatively few Genoese works and no clear colour connection between those working in that city. Surviving Venetian works, half of which are unsigned, tend to display different colours from the Catalans and, although there are some widely-shared colour conventions among Venetian practitioners, there are always exceptions as well.

Top of page



The exercise set out to seek answers to the following questions:

  • What could be learnt about variations in the shapes of individual islands and estuaries?
  • Why were so many islands given imaginative forms?


It has been possible, using the Pujades DVD, to examine on screen, and usually at a greater magnification than that of the original, the great majority of the island and estuary shapes featured in this analysis. What was not possible, however, was to download any details from those scans for purposes of comparison. On occasions, I made freehand copies from the screen, but such work is both time-consuming and imprecise. Nor is my visual memory sufficiently reliable.

The comments and conclusions that follow, therefore, must be partly tentative. Should it become possible in the future for details to be extracted from such a corpus for side-by-side comparison, what is written below, even if not a full and accurate account, could provide a valuable starting-point.

I am not aware of detailed studies of the changing outlines given on portolan charts to the islands or estuaries sampled below, but I should be glad to be informed of any.

We need to distinguish three types of drafting for islands and river deltas. The larger (and possibly medium-sized) islands would have been directly copied from the workshop pattern, apparently by filling in the gaps between headlands fixed by means of 'pounced' points pricked through the model. This would probably also have been the case with the deltas, since they logically formed part of the continuous coastline. Because the smaller islands seem to preserve their own distinctive shape (though not with total precision), they would presumably have been inked in freehand, copying from the model, and left for the colourist. Finally, the smallest islet shapes, and the shoals and those other navigational features drawn in red, would have been created by freehand copying at the colouring stage. These various operations could have been done by one or several people (see Stages in the construction of a chart).

Top of page


Introduction and summary

The shapes of a few islands, such as the Isle of Man, Paxos and the estuarine island on the east coast of the Black Sea (by fassio/faxio) are included in the Colour & Shape Analysis and summarised in the Colour Feature List, since their distinct forms are amenable to classification. The notes that follow relate to other islands that were given distinctive shapes, but ones that are difficult to describe.

Because of the remarkably faithful outlines for most of the Mediterranean and Black Sea coasts from the time of the earliest surviving charts, it might have been assumed that chartmakers (or at least the creator(s) of the lost prototype(s)) shared our own belief in the value of geometric accuracy. If the direction and distance from one point to another was not at least broadly correct what use would these charts have been to a mariner navigating with a compass when out of sight of land?

The shapes of the larger islands - Majorca, Corsica, Sardinia, Sicily, Euboea, Crete, Rhodes and Cyprus - help to confirm that initial impression, since each is a reasonable approximation to its true shape and must owe its outlines to the same process of observation that produced the general outlines. Why then, when we zoom in to look at some of the medium and small-sized islands, and the larger river estuaries, is the picture very different?

Since the depiction of some of these features changes through time, and not necessarily in the direction of greater realism, they provide us with other pointers to authorship and lineage. And, because some of the island shapes defy easy explanation in terms of our preconceptions about accuracy, we are forced to look elsewhere for the answers. [On this see below, Island shapes as a mnemonic device.]

Island shapes can easily be illustrated but they are difficult to describe. And comments such as that the shape on Y is 'similar' to that on X, is usually not very helpful. The notes that follow try to avoid such vagueness, concentrating instead on instances where the shapes being compared are distinct from one another.

For convenience, the coastal knowledge on charts before about 1450 can be crudely divided thus. First, what fell within the direct experience of chartmarkers and/or their clients: in other words the Mediterranean and its basins and, to a slightly lesser extent, the Black Sea. Elsewhere, and this largely applied to what was known or suspected to exist beyond the Pillars of Hercules, much of the cartography was based on largely imaginative geography or the imperfectly understood experience of recent voyagers and their accounts. Until the time that real islands were charted, these Atlantic chimeras were supplied with names based on a jumbled mixture of myth and legend, drawn from a variety of sources.

Some early charts stopped at Gibraltar. Those that ventured into the Atlantic were reasonably confident about the continental coast up to perhaps Denmark, and also south-west along the coast of Africa to about Cape Bojador. The Baltic was largely a mystery to Mediterranean chartmakers but the general outlines of the British Isles emerged speedily during the time of Vesconte (after a halting beginning on the Carte Pisane).

The first appearance of the Canary Islands is on the 1339 Dulceti chart, though involving just two of the larger islands, Lanzarote and Fuerteventura, which are depicted as if in mirror-image (Cortesão, 1969-71, II, pp.56-). After the long gap, perhaps caused by the Black Death, the full group appears on the Pizzigano chart of 1367, with the two eastern islands portrayed with surprising realism and the others roughly sketched in. Shortly afterwards, and certainly by the time of the Catalan Atlas of c.1375, the remainder of the group (excluding La Palma) are represented plausibly as to size and position. This realistic treatment, pointing to some kind of a survey (albeit a rudimentary one) is in marked contrast to all the other Atlantic islands, including the Cape Verde Islands, which appeared a century later in a number of confused variations (see The style and content of Grazioso Benincasa's charts: imitation, innovation and repetition). These recognisable outlines are also strikingly different from the presentation of the Ionian and Aegean islands (discussed under those headings below).

However one peculiarity should be noted. Paralleling the Catalan treatment of Majorca, two of the Canary islands, Tenerife (inferno) and Gran Canaria, were given 'lollipop'-shaped protuberances from their east coasts (like a small circle at the end of a slender 'stick'). First seen on the Catalan Atlas, or perhaps on the undated chart by Guillem Soler, these are very visible on the F. Beccari chart of 1403 (which possibly includes two on Gran Canaria). They continue afterwards and on into the work of Benincasa and, via his successors, well into the 16th century. The purpose of the device that was introduced somewhat earlier by Dalorto/Dulceti in 1330 for Limnos remains elusive (see below ).

What was a chartmaker to do when faced with islands that no-one (apparently) had seen but for which there was (supposedly) textual or oral authority? Unless any of the early voyagers or saints described landing on a perfectly circular island, or on one with the perfect symmetry of a three-leafed clover or a half-moon, it seems clear that the chartmaker had two choices: draw freehand a vast variety of irregular shapes where islands were required, placing a name by each; or assign to each a specific, repeatable shape - either a standard geometric form, a shape borrowed from the natural world, or an implausible, but distinctive, imaginary outline. In other words, these were recognisable symbols not an attempt at geographic representation.

The earliest instance of the formulaic approach to the shaping of islands is found on the 1330 Dalorto/Dulceti chart. A large island (here broadly circular, later entirely so) has been added off the east coast of northern Scotland and another (oval this time) off its south-west coast [C&SA feature numbers 1 & 2]. I leave to others the misdirected speculation as to what these shapes might have been meant to represent. Each is given a name but even if that can be tied to Scotland's offshore geography it does not alter the fact that these forms were intended to be read as loosely indicative and not as geographic reality. That those shapes, and others, can still be seen on the work of Benincasa in the second half of the 15th century, and on that of his successors until the second half of the 16th, is presumably testimony to the continued commercial irrelevance of northern Scotland to Mediterranean merchants.

The mythical Atlantic islands would be gradually swept away, as Madeira and other real islands were encountered and charted. Others would see their initial symbolic shape replaced by a more realistic, post-encounter outline. A contrary example is provided by the Isle of Man (the real island between England and Ireland, rather than the mythical one in the Atlantic) [3]. As successive Vesconte charts sketched in more of the British Isles, about 1321 he added a short-armed diagonal cross in red to represent the Isle of Man. Since this was approximately the right size and in the right place, it can be presumed that Vesconte was either responding to a verbal report from a returning sailor or, even if he knew its shape, he considered the mnemonic device of a cross would serve perfectly well to draw attention to what might have been a navigational hazard, if left unmarked. Dulceti and his Catalan successors replaced the cross with a regular rectangle (or a waisted form of one, somewhat like an hourglass) that gave a better idea of the island's shape. However, the superiority (to our eyes) of the Catalan form was not sufficient to stop the return of the cross version on all charts of the second half of the 15th century.

Scottish islands and the Isle of Man would have had little interest for southern sailors but our final category of symbolic island shapes is apparently more bizarre, since it concerns the Ionian Islands and two of the larger islands in the Aegean (both of which were on the route to Constantinople and the Black Sea). All the islands concerned were well known to the Venetians: indeed, one (Skyros) was occupied by them during the 14th century. The answer must lie in the peculiar complexity of these two archipelagos.

The confusing line of islands, many of them very small, running along the coast of modern-day Croatia would tax any hydrographer and, from the earliest times, no attempt was made to convey the detailed island configuration. Instead, a generalised idea of their size and location was given. For this purpose a range of geometric shapes was used, often symmetrical ones.

Moving south to the Ionian Islands, Corfu and Kefallonia are both well represented with specific shapes. Paxos [31], on the other hand, was given a variety of purely symbolic forms: a cross, and what looks, variously, like a capital letter 'I' or 'K'. But it is Zakynthos [35], at the end of the chain, that it the most surprising.

The island shape instances will now be considered in detail, in the order of their C&SA feature number (moving from west to east and from north to south).

Top of page

Majorca [20]

This is not the place for a detailed study of the island's changing shape over two centuries. That has no doubt been carried out already by the island's energetic historians. Here I just offer some general comments. Starting with the Carte Pisane there was a reasonable approximation, which was to be significantly improved by Vesconte, although he did not appreciate the southernmost extension. By 1330 Dalorto/Dulceti was able to convey the overall shape of the two big bays scooped out of the south-west (Palma) and north-east (Alcudia) coasts. As might be expected, there was further improvement on later Catalan charts. By contrast, Benincasa's outline was, in some respects, inferior to Vesconte's.

One strange aspect of the portrayal of Majorca deserves attention by Catalan historians, if it has not already received this. On some charts, whose overall outline is realistic, two lollipop-shaped projections appear in the middle of the east coast and another pair out of the south-west promontory [see illustration: The protuberances on Majorca and Minorca]. In form, these are reminiscent of those found on Limnos (see No. 36). The intriguing aspect is that these imaginary additions to an otherwise remarkably accurate outline should, apparently, have been made by those same Catalan chartmakers who were based on the island, in Palma, close to the south-westerly examples.

Since Catalan work consists of charts, which tend to be worn - rather than the better preserved Italian atlases - it is not easy to determine when the two pairs of 'lollipops' appeared but the Catalan Atlas certainly includes the two on the east coast, as does one unsigned work attributed to the Cresques atelier (Pujades C 22), and both pairs can be seen on another (Pujades C 19).

The clearest expression of these invented features is on the Cornaro Atlas's copy of a chart by Francesco Beccari, otherwise respected as a great innovator. He must have absorbed the local outline when he worked in Majorca at the end of the 14th century. His son, Battista showed three lollipops off the south-west coast. The Beccari treatment of Majorca contrasts with Pujades's comment that Francesco had 'corrected the location of Sardinia island and substantially enriched the list of nautical toponyms in some sections of the coast' (2007, p.493b). [Neither island was included in my own toponymical analysis.]

By the time of Roselli those extensions have turned into more believable, although still non-existent, projecting headlands, while Benincasa breaks the broadly smooth east coast with four large bays. Rosselló i Verger (1995, p.34) reproduces comparative outlines for the Balearic Islands, starting in 1439. Since he includes only one early Catalan chart, whose re-drawing blurs the lollipop-shaped promontories to the south-west and east, these features are not mentioned in his text (p.356 in the English version). Have Catalan historians discussed these protuberances, and offered an explanation for them?

Examination of the western part of the island's south coast does not reveal any obvious features to which the lollipops might refer. There are several headlands between Palma Bay and Dragonera Island at the point where the coast turns at right angles to the north-east, most notably C.Cala Figuera marking the westerly entrance to Palma Bay. But neither lollipop is placed there and the other headlands are not prominent.

Alternatively, the lollipops might perhaps be considered as navigational symbols. For ships coming out of Palma and heading to the west there were two ranges of rocks protruding across their potential course: first, one at Sas Barbinas leading to Toro Island and, second, one at the entrance to Santa Ponta Bay, leading to Malgrats Island. But those dangers are in the first third of that stretch of coastline, whereas the two lollipops are placed equidistantly between Palma Bay and the south-west extremity.

Neither of the above attempts at a logical explanation can overcome the insuperable difficulty that such, non-intuitive representations would have been understood only by those who already knew what they represented, and who would therefore have no need of such features that distorted a broadly realistic coastal outline.

For a further discussion that brings together the various instances of these intriguing elements, see below and for a note on the coloured stripes applied to the island, see under Other colour conventions: Majorca.

Minorca [21]

At the time, near the end of this project, when I first noticed the protuberances on Majorca I saw a similar feature on Minorca. This has a clear lollipop shape and extends from the island's north coast, perhaps in the vicinity of Fornells and possibly connected with what Rosselló i Verger (1995, p.356) notes is the habitual magnification of that port. Alternatively, given its position roughly in the middle of the north coast, it might be thought to represent the long promontory leading to C. Caballeria. For some reason, that obviously significant navigational feature was not otherwise shown on the charts. The late-14th-century Catalan charts that, by analogy with the Majorcan instances, might be suspected of including it are not clear enough to test that point. It is most readily seen on the 1403 Beccari chart, conveniently (because of a freely available online scan) the one illustrated, see The protuberances on Majorca and Minorca .

What look like two other lollipops extending south-east are clearly intended, if in much exaggerated form, to represent the way into Mahón, a convention that would be long-lasting, see Rosselló i Verger (1995, p.34). For a general comment on the lollipops see below.

Top of page


(see the illustration)

Corfu [30]

The earliest Genoese efforts were more realistic than the formulaic outline given by Benincasa a century and a half later. Little can be gained by descriptions of shape but one feature can usefully be highlighted, a thin promonotory emerging out of the middle of the east coast, representing the citadel in much exaggerated form. First noticed on the 1330 Dalorto/Dulceti chart, this was later turned into another of those 'lollipops', apparently by the time of the Catalan Atlas and certainly on the 1403 Beccari chart. Benincasa - at least on the two copies of his work in the Cornaro Atlas - turned the end of that unconvincing promontory into an arrow-head [see A note on the Cornaro Atlas]. Further work needs to be done here.

Zakynthos [35]

Zakynthos is unrecognisable on the Carte Pisane and Cortona chart. The island has a relatively uncomplicated outline and the twin promontories enclosing the large bay at the southern end were understood by the early Genoese. However, by at least 1313 (the 1311 chart is unclear) Vesconte had added an imaginary northern cape. This was imitated on the earliest of the Genoese charts, in the Riccardiana in Florence (assuming that it does not even pre-date Vesconte), by Dalorto/Dulceti and then by Francesco Beccari in 1403. The 14th-century shape introduced by Vesconte is reminiscent of a bird, perhaps a chicken, but by the time of Benincasa it looks, with its rounded bays at north and south, like a piece from a modern jigsaw puzzle.

Not all chartmakers followed the Vesconte prototype. The 1373-83 atlas by the Venetian Pizzigani gave it the appearance of a standing horse but the 1409 Virga chart is the earliest dated work to make an attempt at reality. Other Venetian charts, for example the anonymous Corbitis Atlas, also reflected a modified shape, as did the 1430 Briaticho atlas. However these 'improvements' were not permanent, since they were ignored by non-Venetian chartmakers. Even Benincasa, when he was working in Venice, continued with the wholly unrealistic earlier form. This can still be seen on the Caloiro e Oliva chart of 1626, and perhaps later still.

Top of page


(see the illustration)

Limnos [36]

Without a doubt, the portrayal of Limnos (stalimene on the charts) was to lead to the most extreme and grotesque forms given by portolan chartmakers to any island in the 'known world'. The island's true shape is too complex to describe and it would have required a long and arduous survey to capture all its contortions. It has the most complicated shape of any of the medium-sized islands on the portolan chart. So, in examining the successive shapes it was given we should not be surprised to find, first, that later efforts were even less realistic than Vesconte in 1318 and, second, that the process of distorting the island's shape so as to remove all pretence at plausibility seems to have been entered into with enthusiasm.

Neither the different, simplified outlines found on the Carte Pisane, Cortona chart or the anonymous Ricciardiana chart, nor the elaborated capital 'M' favoured by Vesconte, would have morphed into what became the 'standard' outline. This appears first on the Dalorto/Dulceti chart of 1330, although there are hints of it in the anonymous Genoese atlas in Paris (MS.Lat. 4850). Initially reminiscent of a cow's udder, this already has the beginning of the strange protuberances that would be developed further later [see the illustration]. With one to the north, a pair each projected west and south, these have the appearance of a lollipop (or toffee-apple), a circle on the end of a thin stick. Sometimes a sixth instance can be made out in the south-west corner. Alternatively, one or more of the projections might be given a slim rectangular shape. Determining how many of the protuberances should be considered lollipops rather than promontories is sometimes difficult from the available scans. This means that the differences in the total number of lollipops stated on the various C&SA tables may not always be significant.

The device was picked up again on the 1385 Soler chart (if not before) and followed thereafter by all Catalan chartmakers and many Italians. The lollipops can still be clearly seen on the Newberry Library's Volcius chart from the 1590s and, in the form of non-existent projecting promontories, until the late 17th century.

Only a few of the Italians resisted, offering instead a complicated, multi-indented coastline that represents a clear attempt to reproduce the island's contorted shape. The 1421 Cesanis is a good example but perhaps the most impressive is on one of the later sheets in the Medici Atlas.

This strategically placed island was to pass (briefly) to Genoa in 1453, but, not surprisingly perhaps, the chart produced in that city by Pareto two years later holds true to the usual form. So too does the 1489 chart by another Genoese, Canepa.

It is hard to see how Limnos's real shape could have prompted these excrescences (on which see below), but evidently its realistic outline was considered less important than the aide-memoire offered by a distinctive, clearly implausible, but easily memorised outline designed to signal the dangers of the island's contorted coastline (see further under Mnemonic device).

Top of page

Skyros [39]

Skyros, an island in the Northern Sporades, was occupied by Venice from 1204 - and has a Venetian castle as evidence of that, although it seems that the Ottoman Turks displaced them at the end of that century. Again the Carte Pisane and Cortona chart give the island unrecognisable outlines. Interestingly, the Carte Pisane's form, a rectangle with a groove out of one of the long sides, is similar to that on the early anonymous Genoese chart (Ricciardiana, Florence - Pujades C 4). From at least 1311, however, Vesconte had introduced what would become the standard outline, and one that cannot be reconciled with the known shape of Skyros. Seen more clearly in his first atlas of 1313, the most distinctive feature of this thin, heavily indented island, arranged precisely north/south, is the symmetrical pair of bays at its southern end, matched with three, equally symmetrical, small islands placed, like full-stops, beneath each of the peninsulas (see the illustration).

Dalorto/Dulceti continued with a variant of this form, introducing an almost detached diamond-shaped peninsula at the north end. This northward-pointing 'neck' would, if anything, elongate later, sometimes with a kink in it, at other times with the end section bending to the left or right. And so the inventions continued, until, with Benincasa, the entire shape had become even more regular and the number of identical bays at the south had been increased to three. The overall effect (at least at the base) might be likened to that of a space rocket.

In this case I am not aware of any subsequent improvement during the 15th century and this form can be seen at least as late as the Newberry Library's Volcius chart from the 1590s. This can perhaps be explained by the lack of sustained Venetian involvement with the island in that period, and the same lack of incentive for change that was mentioned above under Limnos.

Smaller Aegean Islands

Besides the four sample Aegean islands considered in the C&SA, there are approximately 100 others included on the portolan charts, smaller than those or even just islets. So how were they handled?

A good test is to examine the line of about a dozen islands running southwest from Euboea (negroponte) towards Rhodes [best seen on the 1465 Benincasa atlas in Vicenza, Pujades A 34, sheet 4 but also visible on hazier scans here]. By comparing the Aegean on different atlases of Benincasa we can immediately see that, instead of being treated randomly, the conventionalised forms for those islands were strictly adhered to. Tinos is presented as a circular arrangement of three arrow-heads, Ikaria (nicharia) is sausage-shaped, two others are reminiscent of modern jigsaw-puzzle pieces, and another is trifoliate. With minor variations, the colouring was just as consistent.

Studying such island forms requires high magnification on the Pujades DVD. Since there is no way to download images or split the computer screen, any comparison must be done via drawings or memory. A further handicap is that details as small as those can be difficult to make out even on the Pujades scans. While the Catalans may have had a partly different sequence of shapes [further research would be needed there], it seems that those visible on Benincasa's work reflect a long Italian tradition.

Neither of the two supposedly earliest charts, the Carte Pisane and Cortona chart, gives their Aegean islands any recognisable shape and it seems likely (apart from attempting to express the island's size) that the simple, block-like shapes were drawn freehand, without following a model. By the time of the 1311 Vesconte chart, however, there are clear signs of the policy that would inform Aegean nautical cartography for centuries, namely that each island would be given an arbitrary 'signature shape'. This was a visual mnemonic device that would both allow instant recognition of a single island and distinguish it from the others. With a few exceptions, there was no attempt to represent the true outlines, though each island was placed in what was thought to be its correct position and with its true size. On this see Island shapes as a mnemonic device and the illustration of the The line of islands between Euboea and Rhodes, each with its distinctive shape.

It is hard to read the Vesconte island shapes but it appears that Tinos already has its three arms, that Ikaria has its plain shape [the apparent extension is actually a smudge] and Tilos (piscopia) may already have the crown-shape seen on Benincasa. There would certainly be refinement after Vesconte but by the time of the 1403 Beccari, at least, most of the signature shapes seen on Benincasa are in place. Two centuries later still, crude distortions of those same inventions can still be seen on the 1678 Cavallini chart.

Pujades includes ninety named Aegean islands in his listing of the full toponymy of the Vallseca chart (2009, pp. 178-80), but there are a number of other unnamed islets in addition to that. Comparison between a 1463 and 1469 Benincasa atlas demonstrates that even the details of the smallest islands were faithfully reproduced. Nearly always in red, those do not seem to have ink outlines, in which case they must have been inserted at the colouring stage. Some have distinct shapes of their own. Until he had memorised every one of hundreds of such details spread over the chart, the chartmaker would have followed his model - perhaps a special colourist's pattern - with great care.

Such details might be below the interest of any historian, but the care and attention devoted to them by the chartmakers - nearly all of whom should more appropriately be described as chart copyists - showed that they did not accord any hierarchy of importance to the various elements that make up a chart. It is pointed out elsewhere (in the essay on Benincasa) that he showed a cavalier disregard for the west African discoveries he was apparently the first to document on a portolan chart. For today's researchers, that is his main claim to fame, but, in his eyes, failing to make enough space for all the Portuguese names at his disposal, was just a matter of convenience. On the other hand, it would never have occurred to him to take a shortcut when copying the basics elements of his charts, down to the smallest island detail.

Top of page


Small islands were generally given formalised 'signature' shapes. This applies also to a few of those included in the C&SA, which meant that their shapes (or some of their variant shapes) could be codified in a table. [See further below under Mnemonic devices?]

Crosses (shape)

The most obvious recognisable island shapes are two variants of an elaborated Greek cross (with all four arms of the same length). Both are arranged as a saltire, that is diagonal not upright, in relation to north: the Isle of Man [3] between Ireland and England, and Paxos [31] just south of Corfu in the Ionian group. The other crosses discussed in these pages, for Lanzarote [17], Chios [38] and Rhodes [40], were created when the colourist applied a painted cross to the island's different background colour (see below under Crosses (painted)). Here we are dealing with an earlier stage of production during which the island outline was inserted in ink by the draftsman. It is not clear if the smaller islands would have been transferred as part of the direct copying of the general coastline, or if they were drawn in freehand. As mentioned elsewhere, by whatever means the outline was copied, it was always done in a faithful manner.

For details about the successive portrayal of the Isle of Man see the Colour Feature List. For Paxos, see the Microsoft Word 'General table to the colours and shapes'. In each case, the alternative shapes, which are more or less describable, are associated with different groups of chartmakers. Occasional attempts at a more realistic outline for the two islands failed, though, to dent the popularity of the symbolic cross form.

As an indication of the conscientiousness, or inventiveness, of at least some chartmakers, the shape of the Man and Paxos crosses might be varied. For Benincasa, although both his crosses had large arrow-heads at the end of the arms, the Isle of Man version had a chunky form, close to a regular jigsaw-puzzle piece, while his Paxos example had noticeably thin arms. This is perhaps the contrast between the Occitan cross for the Isle of Man and, for Paxos, a 'cross with the ends of the arms bottony (or botonny), i.e. shaped like an architectural trefoil' (Wikipedia under 'Cross'). It is not clear what the significance was in the carefully preserved distinction between those two forms. What may prove to point to a specific visual signature of Benincasa's successor Angelo Freducci is the latter's use of the 'cross crosslet' form, in which each arm is itself a cross. This can be seen on the 1555 Greenwich atlas. The printed Barentsz. chart of 1595 shows Paxos as a much enlarged distortion of the earlier cross shape. The convention continued on manuscript charts until at least 1678.

In Benincasa's case the coloured centre-piece he introduced into Tenerife [14] - created out of colour alone - has a different shape again, which can be variously interpreted as a cross or a four-leaf clover [five-leaf in one instance].

Other forms

Other recognisable forms can be seen in the non-cross shapes for Paxos [31], the Isles of Scilly [4] and fassio/faxio [47]. Vesconte, and perhaps the earliest of the Genoese charts (Pujades C 4 - Florence, Riccardiana 3827), gave Paxos a shape like a capital 'K'; Dalorto/Dulceti, the Catalan Atlas and at least one of the charts attributed to a Cresques atelier, favoured a shape closer to an 'I'; otherwise on all later charts it takes the form of a cross.

The Isles of Scilly [4], which guard the perilous western entrance to the English Channel, are a complex group of islands. Portolan chartmakers tended to show one main island, with a distinctive shape, and then a number of undefined smaller ones to its west. Shapes reminiscent of a banana or an old-fashioned spanner head are seen on early charts but the variety of forms led to the abandonment of the attempt at C&SA classification.

A similar difficulty applied to the notional island, usually near to a place called fassio or faxio [47]. Although no river is usually shown, this was meant to denote the estuary of the Enguri River in the middle of the Black Sea's east coast. Since this represents the extreme east point on the chart, the area is often trimmed away, damaged or illegible. Nevertheless, at least five shapes for that island can be recognised, of which the spanner-head form was much the commonest.

Top of page


Besides a range of islands, it was decided to include four river estuaries in the Colour & Shape Analysis. Two are in the Mediterranean, the Rhône [22-4] and Nile [48-50], and two in the Black Sea, the Danube [41-2] and the Dnipro [43-6]. In addition, the various shapes given to the 'island' at the mouth of the Enguri River on the east coast of the Black Sea beside faxio/fasso are discussed in the Colour Feature List, but see also under Vallseca in Attributions. Because of the complex nature of those deltas, and the fact that the various river courses were constantly changing, it was neither possible nor meaningful for a chartmaker to attempt to convey a true, rather than impressionistic picture. Nevertheless, in two instances - the Rhône and Dnipro - recognisable features can be made out. For the Rhône it is the long sandbank roughly between Sète and Aigues-Mortes, which can already be discerned in the 1313 Vesconte atlas (see the Illustration). For the Dnipro delta the two small, thin, islands running west-east [presumably Tendrovskaya Kosa island and the long sandbank next to it] can be clearly seen on Vesconte's earliest surviving work, the chart of 1311.

Otherwise, in different ways and to varying extents, chartmakers used shape and colour to highlight what were considered to be important navigational estuary features (because of the inland ports they led to) since those simultaneously posed considerable danger from shifting channels. It is unlikely that any hydrographical significance was intended (or understood) by the varying arrangement of simple wedge-shaped 'islands' by which the chartmakers denoted river estuaries. But they do form an important, and often very prominent, component of each practitioner's visual signature.

Of the deltas considered, treatment of the Danube [41-2] was the most erratic. Its most flamboyant presentation can be seen on the earliest surviving dated chart (Vesconte's of 1311), and about fifteen separate 'islands' are shown on both the 1330 Dalorto/Dulceti chart and an unsigned Genoese chart of that general period (Pujades C 9bis). These brought into play a precursor of the 'four-colour theory', since numerous entities in close proximity requiring to be distinguished from one another called upon the use of up to five colours. After that, some 15th-century Venetian works offered a patchwork of islands, but most chartmakers reduced the Danube delta to at most one large and one small island, or else ignored it altogether.

The changing morphology of the Rhône and Danube deltas was considered by James E. Kelley, Jr., 'The oldest portolan chart in the New World', Terrae Incognitae 9 (1977): 22-48, see figs 2-4.

The shapes of the formulaic islands denoting the major river deltas would probably have formed part of the outlines copied from the workshop model in the initial ink drafting process. As such, they are less susceptible to the kind of evolutionary development that can, at times, be seen in the smaller islands.

Comparative tracings for the Dnipro estuary and neighbouring Crimea, and the Nile delta, were offered by Nordenskiöld in his Periplus (p.23 - for the period 1311-1596) and Caraci (Imago Mundi 10, 1953, p.41 - for the period 1435-1556). What is noticeable is how little the outlines change in the early period, if allowance is made for the fact that one of Caraci's sample, Borgia V in the Vatican, is now considered to be a 16th-century work. The 1602 Huntington Library's Juan Oliva chart still shows all the deltas in their 14th-century forms. The main change, introduced in the 17th century, is the reshaping of the Rhône delta to show multiple channels instead of the earlier single course (although the later form can be seen on three charts attributed to the workshop of Joan Oliva c.1585 (Rosselló Verger, 1995, pp. 174-6)).

Top of page



The exercise set out to seek answers to the following questions:

How significant for diagnostic purposes are such conventions as:

  • the uncoloured rectangle set into the middle of an island where its name would be written, dubbed here the 'name label'
  • the number and direction of coloured stripes across Majorca
  • the cross for Lanzarote, Rhodes, and occasionally Chios
  • the treatment of formalised islands, for example presenting circular islands as buttons or with a serrated edge
  • the portrayal of Tenerife's volcano
  • the detached first letter of Aigues Mortes and Damiata


The convention described in these pages as a 'name label' refers to the uncoloured band sometimes left across an island, and into which the name would be placed. The label usually took the form of a rectangle but might, for example in the case of one of the mythical circular islands, be a strip dividing the island in two horizontally. Labels placed within an island would follow its longer axis.

The labels were most commonly applied to certain islands in the Aegean, to Majorca (as an alternative to the Catalan stripes), and less frequently to the mythical islands of the north Atlantic, the Ionian Islands, and the island shapes in the Dnipro delta. There are certainly a few other examples besides those on the islands studied in this analysis, for instance in the Adriatic and Atlantic. When softer colours were used for the background it is sometimes difficult to determine if a label had been intentionally created.

Who invented this device? It may, of course, have been borrowed from a pre-existing tradition within or outside cartography, since it is such an obvious solution to what must have been a common problem. Placing the name within the island removed any possible ambiguity and, particularly for the Aegean, freed up valuable space. The now-destroyed chart produced by Giovanni da Carignano (who died no later than 1330) has numerous name labels but these were used for provinces rather than islands. Since it appears to be an extension of the differential colouring of islands, which itself has yet to be traced earlier in other contexts, this may well have been an independent innovation of the chartmakers.

The earliest dated instances of name labels can be found on the 1330 Dalorto/Dulceti chart, where they were used for Chios and Rhodes. However, they are also seen on two of the anonymous Genoese works, assigned by Pujades to the period 1325-50, both of which are in the BnF Paris (his A 9 and C 11). Each work applies the label to Majorca but the atlas also gives one to Lesbos, in addition to the two islands thus treated by Dalorto. The 1339 Dulceti chart and the Paris atlas both use name labels to let the various red place-names on Corsica, Sardinia and Sicily stand out from the surrounding wash colour. Thereafter, labels are sparse on Catalan works of the second half of the 14th century and were used only by Roselli in the 15th. A few labels - those for the mythical islands montorio/brazil and man, as well as Skyros - are not found on Catalan work at all.

Given Dalorto's likely Genoese training if not origin (see Pujades, 2007 pp. 490-1) and the presence of name labels on two early Genoese works, it would not be surprising if the convention started there. Vesconte, possibly working exclusively in Venice, never used them. However, labels would subsequently become a device most associated with Venetian work, even if their use was intermittent. [The details are set out in Summary table of the features analysed (a Microsoft Word document).]

A striking Venetian extension of the single name label is seen on three closely related anonymous atlases, assigned by Pujades to the period 1425-50 (his A 29, 30 & 31). Not only do one or more in that group add labels to islands otherwise free of them - Ibiza, Minorca, Malta, Corfu, Keffalonia, as well as the Nile delta - but they inserted multiple labels, especially on Majorca, Lesbos, Chios and Rhodes, in one case managing to squeeze seven labels onto Majorca. The addition of labels to other Aegean islands beyond the few considered in the C&SA (as seen here on Pujades A 29 & 30, though not A 31) would also be a feature of at least one atlas from the early 16th century, Paris BnF, Ge AA 567, considered here in the supplementary Census listing as E.1(A).

Name labels, in themselves insignificant, are another useful diagnostic tool for help in determining lineage. Some of the labels set into the mythical Atlantic islands were either introduced by Benincasa or, by the 1460s, are found only on his charts and those of his successors.

Whether the scribe wrote the name first and the colourist added the surrounding tint afterwards, or the other way round, this convention generally involved collaboration between at least two people, or two production stages. Even if the operations were performed by the same person (though presumably at different times) in carrying out the first procedure he needed to be aware that the second would follow. In the cases of islands tinted in soft tones, such as mauve, it might be possible for the name to show through pre-applied colouring, but not, for example, for the deep blue often used for the Aegean island of Limnos or the strong red for Skyros. If no label was provided, those names had to be written alongside.

Occasionally, and examples are Benincasa's treatment of Chios and the mythical Atlantic islands of till and montorio/brazil, the label would be outlined in ink. That is in the same ink as that used for drafting the coastal outlines and was evidently done at that initial stage. This tiny piece of evidence demonstrates cooperation between three individuals - the draftsman, colourist and scribe - or the careful coordination of three successive stages of the chart's construction (see Stages in the construction of a chart).

Top of page


Majorca's stripes [20]

The most immediately visible portolan chart convention is the placing of the Catalan stripes of red and gold across Majorca [20], from the time of the Catalan Atlas (c.1375) onwards. These, varying in number and direction, are habitually seen on 15th-century Catalan charts, but the device was adopted by few Italians before Benincasa. Italians would often leave the island uncoloured or place a name label over it.

Crosses (painted)

Two of the studied islands, the Isle of Man and Paxos, as well as other smaller ones, were given the shape of a cross (see above Shape: Crosses). This note is concerned with the hand-coloured cross added to three islands, habitually in the case of Lanzarote [17] and Rhodes [40], and occasionally for Chios [38]. The cross might be white or the background might be left uncoloured but often two colours were involved. When that appears black today it may well have been silver originally.

Buttons and other island centres

Occasionally, the centre of a mythical circular island would be embellished with other colours. This can be seen on till [1] and montorio/brazil [9] on the Catalan Atlas and rarely thereafter and, only on a single Italian atlas of the period 1425-50 (Pujades C 50), as the treatment for the imaginary man [10].

inferno [14], which, by the 16th century, was known with its modern name Tenerife, was often given a white device to denote the snow-covered volcanoes in its centre. The space for the symbol would either be left blank for the natural vellum colour to show through or the device would be conveyed by the careful application of white. The island first appears on the 1367 Pizzigano chart and, by 1375 with the Catalan Atlas, it already has the device. Whether shaped like a cog-wheel, a circle, or a four- or even five-leaf clover, this would have been the work of the colourist alone. It was presumably that individual, or at that stage of the work of a single person, that occasional later chartmakers made the mistake of placing the symbol on Gran Canaria instead. This has been noticed on a 1508 Russo chart and a 1563 Sideri atlas.

Detached initial letters

The final device noted in this analysis - and not apparently commented on previously - concerns two names whose first letter (invariably in red) was placed separately from the rest of the word: Aigues Mortes [23] and Damiata [51] (for illustrations see respectively here and here). Both those conventions are first seen on the 1330 Dalorto/Dulceti chart. Since each of these instances was widely, but not universally imitated thereafter, they provide useful diagnostic evidence, particularly as a few followed Dalorto in detaching the first two letters of Damiata rather than just the usual 'D'. There is also a unique instance, on the signed Rafel Soler chart, of a similar treatment of the first two letters of Aigues Mortes. The reason for these conventions is not clear but in each case it continued well into the 16th century. It was proposed by Catherine Delano Smith that the detached initial for these two fortified towns might reflect their locations, in river deltas (the Rhône and Nile) surrounded by marsh or water that suggested a degree of insularity. {This sentence added October 2013}.

On his chart of 1403 Francesco Beccari placed the detached 'a' for Aigues Mortes on the island but wrote the name out in full. Some later chartmakers were clearly confused about the Damiata instance and wrote amiata as usual, without the separated initial letter or, as on the 1545 Agnese atlas, transferred the whole word to the island.

If the colourist carried out his work before the scribe, he would have to remember to leave the space required by the 'A' and 'D' respectively - again requiring coordination.

Top of page


Cautionary note

Leaving aside the works considered by Pujades to have been drawn by, or in the hypothetical workshop of, a named chartmaker, only the following have left three or more signed (or confidently attributable) works: Pietro/Perrino Vesconte, Dalorto/Dulceti (3 works only), Vallseca, Ziroldi, Roselli and Benincasa. There are three works signed by Beccari and Viladesters, respectively, but in each case that involves two individuals. Looking for signs of individual or family consistency is seriously hampered by those small samples.

In addition, instances of illegibility, ambiguity in the colour, or the frequent omission of the diagnostically rich Atlantic region, reduce the statistical credibility of the data set out on the detailed Colour & Shape Analysis tables, accessible via the Menu page. A different approach was therefore taken with the Consistency Tables: two of them general and two devoted to Benincasa (see Colour Consistency Tables, a Microsoft Word document):

    A. The chartmakers compared
    B. Consistency totals for eight chartmakers
    C. Apparently inconsistent features on the work of Benincasa
    D. The use of brown, mauve and pink on the work of Benincasa


Table A, at the risk of causing further visual confusion, attempts to summarise the information in two ways. By selecting one typical (?) work by each of fourteen named chartmakers and two sample anonymous works, it offers the chance, at a single view, to compare the colour and shape conventions across a century and a half, and from the main centres, Venice (blue), Genoa (red), unspecified Italian (green) and Palma, Mallorca (yellow). Besides that, by means of the yellow highlighting, it draws attention to conventions that were usual, or even invariable, in the work of the practitioner concerned.

In eight cases, the vertical reading, by chartmaker, can be related to the consistency totals set out in Table B. Note though, that the middle ('100% consistency') column in Table B means precisely that, whereas the yellow highlighting in Table A gives credit for usual patterns, to which there might be occasional exceptions.

Taking the horizontal reading by feature first, there are some instances of great variation, both within and between the centres of production. It is hard, for example, to see any clear patterns in the colouring of the Canary Islands, except for Tenerife [14] and Lanzarote [17]. However, in other cases there are strong regional or universal associations with a specific colour [often red], for example the Isles of Scilly [4], the red and blue combination for Formentera [18] and Ibiza [19], the gold for Majorca [20], the red or associated pink or mauve for Minorca [21] and Zakynthos [35], the blue for Limnos [36], the red for Skyros [39] and as the underlying colour on Rhodes [40]. Most striking of all is the habitual red for Gozo [29] - apart from its treatment by the Venetians, whose lack of named practitioners has reduced their inclusion in Table A.

A number of these conventions continued almost universally until at least the mid-16th century. That perhaps indicates a merging into a common chromatic lingua franca what had formerly been separate regional traditions, a convergence already apparent in the work of Roselli and Benincasa, who shared about 60% of colour and shape conventions.

This interpretation of the development of island and delta colouring as various ideas coalescing into a single concept is supported by examination of the analytical tables for 14th-century Italian work (a Microsoft Word document). The Ionian Islands provide useful evidence because they were sometimes repeated on the sheets for the Adriatic and the Aegean. The number of occasions on which the islands were given different colours on the two sheets concerned, on the work of Vesconte and others, confirms that chromatic conventions were still being established. That process is most apparent in the establishment by Vesconte of clear, if not universally observed patterns, after his two earliest surviving works of 1311 and 1313. Combined with the lack of island colour on the Carte Pisane and Cortona chart, and the small range of colours on other apparently very early charts, this shows us a practice in its early stages of development, one that would only gradually, and partially, become solidified into consistent patterns.

Since almost all charts used at least four different colours, and most more than that, the numerous instances of clear patterns in the conventions used demonstrates that on relatively few occasions we are faced with what might have been random choices by the colourist, rather than the borrowing and perpetuating of local or universal traditions. It must also be remembered that, if the purpose was to distinguish neighbouring elements with different tones, any change could necessitate one or more further ones. In some cases, for example lines of islands like the Canaries or Ionian group, distinct sequences of colour can be distinguished.

The association of a particular colour with a well-known Mediterranean island was logical. Even if the Catalans and Venetians chose a different colour, they tended to persevere with it. In the same way, it is not surprising that inconsistent colouring should be used by many chartmakers (though not Roselli or Benincasa) for the imaginary Atlantic islands or the generalised island shapes used to denote the deltas.

Viewed vertically, and taken in conjunction with Table B 'Consistency totals for eight chartmakers', some idea can be given of the extent to which the various chartmakers followed whatever colouring pattern they might have had in-house. Comparing the red figure, which expresses as a percentage the proportion of features for which no clear logic can be determined, with the equivalent green figure denoting the number of occasions in which no variation was discernible, it is clear that the least consistent of the eight who could be considered for this purpose was Ziroldi. But even he gave a consistent treatment to seven features. Interestingly, in the light of the previous comment, several of those relate to those islands that would have been well-known to any Mediterranean sailor. The only other column to achieve a consistency figure of less than 50% is that for the four charts assigned by Pujades to a Cresques atelier.

At the other end of the scale are the late, and nearly contemporary, Roselli and Benincasa. Since the high total of ten and sixteen works respectively was analysed for general consistency, a single exception was enough to mark them down. If the rare exceptions in the work of those two practitioners were ignored, including the use of brown, pink and mauve, which they themselves clearly treated as alternatives [on which see Colour Consistency Tables (a Microsoft Word document): Table D. 'The use of brown, mauve and pink on the work of Benincasa'], then their consistency totals would have been about 94 and 97 per cent respectively.

Although the colour analysis is inevitably incomplete and inexact, it does provide, first, an indication of the level of colour consistency in a particular chartmaker's work and, second, it emphasises the extent of separate Venetian traditions. The C&SA's main conclusion is that none of those chartmakers who have left us sufficient material for meaningful comparison treated colour as a matter of random choice. Then again, none, until the middle of the 15th century, established a chromatic signature for the entire chart and faithfully observed it throughout. The best explanation for what seems an innate contradiction is that a handful of strong colour associations with specific, well-known Mediterranean islands were widely observed, even by chartmakers who felt free to apply whatever colour they liked elsewhere.

There are indications that sometimes a chartmaker whose earliest (surviving) work shows erratic colour choices came to appreciate the advantage of settling on a definite routine. This, of course, is measurable in the case only of those for whom we have sufficient extant work. Examples would be Vesconte (from about 1318), Roselli (after his first chart) and Benincasa (after his two earliest charts).

Top of page


Benincasa's use of brown, mauve, pink, etc.

For this section see Colour Consistency Tables (a Microsoft Word document): Table D, 'The use of brown, grey, mauve, orange, pink and purple on the work of Benincasa'.

The Consistency analysis (Table A in that same Microsoft Word document) highlighted the need to understand the reasons behind Benincasa's superficially erratic use of what seem to be the following specific colours: brown, grey, mauve, orange, pink and purple (read horizontally across Colour and shape on the work of Benincasa, also a Microsoft Word document). Again, it is necessary to explain that the colours (some of which might have changed slightly with time or have been distorted in photography) could not be directly compared with one another, either within an atlas or between them. Colours, of course, form a continuum with no definite boundaries. However, I re-checked all the entries in this table in one session, in the hope of achieving consistent readings.

Closer examination showed that those ambiguous colours were applied to just seven features. There are no instances of their use elsewhere on Benincasa's charts, after his first, untypical works of 1461 [with the single exception of the smaller island in the Dnipro estuary, on the 1468 atlas, British Library, Add. 6390]. These colours would have been among those that had to be made up by the chartmaker rather than bought from a supplier, as might have applied to red, yellow, green and blue. What appears to be a similar alternating of mauve and pink (if those really are different colours) can be seen on the work of Roselli, i.e. on Ireland [6]), montorio [9] and porto bo in the Dnipro delta [45] - see Colour Roselli (a Microsoft Word document).

One supposition was that the choice between those six tones might have depended on what was available for mixing on a given day, or the results of what could be an unpredictable process. However, a vertical reading of the table soon showed this not to be the case. On most of the atlases three or four of those variant colours were deployed, if my readings of these uncertain tones is correct. Benincasa must therefore have specifically prepared several colours so as to give extra variety to the regular red, gold, green and blue.

Some clues to his motivation come from the seven features involved. Three of them are the colour edging of landmasses - Scotland, England and Sardinia. Two were imaginary, the mythical island of montorio/brazil [9] off the south-west of Ireland and the large 'island' in the delta of the Black Sea's Dnipro river [45]. The final two instances are islands, Gran Canaria [15] - in an archipelago for which few chartmakers developed consistent colouring and which was omitted from many charts - and Lesbos [37] in the Aegean. Unlike Limnos and Skyros, Lesbos did not have a strongly associated colour and was sometimes left untinted or with a faint wash. The suggested explanation for the variant colouring of Lesbos, chromatic variety, is likely to apply to the other six instances. In no case was Benincasa flouting an accepted convention.

Despite the fact that for these seven features one colour sometimes predominates - for instance, the brown edge to Scotland - Benincasa clearly considered the alternatives to be acceptable. It is interesting that what I take to be a salmon pink (or perhaps orange) is used to edge Sardinia on four of his atlases, but nowhere else. In the light of this interpretation of what might fairly be termed consistent inconsistency, the analysis would have been badly skewed if his Lesbos or montorio/brazil had been recorded as 'generally inconsistent'. Equally, to treat that group of colours as identical, in the way that yellow and gold have been conflated, would introduce a questionable bias. For those reasons, those seven features were omitted from the statistical totals in his case.

Benincasa's other apparent inconsistencies

For this section see Colour Consistency Tables (a Microsoft Word document): Table C, 'Apparently inconsistent features on the work of Benincasa'.

Leaving aside those features already treated in the foregoing discussion, there are 17 features where Benincasa certainly or apparently broke from his usual conventions, on at least one occasion. None of these involved the island shapes, which would have formed part of the ink drafting process, nor the scribal peculiarities relating to Aigues Mortes and Damiata. On all those, Benincasa was 100% consistent throughout his career. The inconsistencies therefore concerned the colouring stage only. What patterns can be discerned in these divergences from the norm?

The first, obvious conclusion is that the two earliest charts (one dated 1461 and the other plausibly assigned to that same year) diverge on a few occasions from the rules that would be established by Benincasa by 1463. However in two cases their use of red instead of gold would recur later, in works of 1470 and 1473 (both coincidentally in the British Library) and in two confidently attributed charts. That five of the six works involved there are charts rather than the more usual atlas may have bearing on this. Gold might have been avoided for charts, either because they were cheaper or because the rolling and unrolling of a chart could have caused the gold to break off. Inconsistency, perhaps; but certainly a calculated one.

The other immediately striking feature is the way that the 1469 (Milan) atlas substitutes green with yellow or gold on six occasions. However, green was not omitted altogether, being used five times. Again, while there is no obvious reason for this, it must have been the result of a careful and internally consistent decision and not a lack of paint, since that atlas has only one other colouring peculiarity.

Of the other instances where Benincasa seemed to be erratic, two involved the same substitution of red for gold (for Corfu [30] and for the two small islands in the Rhône delta [24]). These were clearly not random decisions, even if the possible explanation that he had no gold or wished not to use it does not fit with the use of gold elsewhere on some of those same works.

Further apparent inconsistencies affect the crosses placed over two islands: the background of Lanzarote [17] and the cross on Rhodes [40]. What was usually white (or perhaps bare vellum) now appears as an irregular and mottled black. This could be the result of silver being used instead and subsequently tarnishing.

Once those instances are removed, little genuine inconsistency remains for the period from 1463 onwards across all those signed Benincasa works that could be analysed. In all, just seven apparently random instances remain from analysis of the 51 features on 14 works.

Finally, there are the two attributed charts (see at the right-hand end of Colour and shape on the work of Benincasa, a Microsoft Word document). The neck has been cut off in each case, taking with it any signature there might have been. Besides the substitutions of red for gold mentioned above, the two incomplete charts have between them just three unique instances. None of those match the untypical colouring of the pair of earliest charts and so they cannot be readily explained as part of the settling down period before 1463. However, the one or two unusual colour choices should be considered in the context of the otherwise uniform treatment of the other 50 or so features considered.

Top of page





Are 'lollipops' navigational symbols?

Island shapes as a mnemonic device

A deeper understanding of the function of portolan charts

What is the difference between a plain and a de luxe chart?

Conservatism and workshop practice
     Geometric consistency
     Where is the expected conformity?

Expectations contradicted by the C&SA findings

Sixteenth-century continuations

Are 'lollipops' navigational symbols?

Two contradictory suggestions have been made to me privately to the effect that the excrescences found at various points on the early portolan charts, to which I have given the disrespectful name 'lollipops', should instead be interpreted as a navigational symbol. In one explanation this was to indicate a safe harbour (perhaps suggesting beacons), in the other the opposite, namely a line of offshore rocks coming out from land and representing a major hazard. For the reasons below, I find it hard to support either of these ingenious interpretations, nor the more obvious explanation that they denote headlands.

First an attempt to describe what a lollipop is. The portolan charts did not portray the individual characteristic of each headland, any more than they depicted the true nature of the coastline between them. Rather the headlands were given simple, conventional shapes, which varied from one chartmaker to the next. Benincasa, for example, often favoured an arrow-head projection, while others might use more rounded forms. In some cases, the headland would be attached to the mainland by a short line (like a round-headed pin), perhaps intended, sometimes, to represent a promontory, though most were surely not. As a further development, the black line might be elongated until it looks like a stick with a ball on the end, in other words like a lollipop.

As has already been noted, the clearest examples are seen on Limnos [above; and its illustration: Limnos and Skyros with their usual outlines] and Majorca and Minorca [above; with their joint illustration: The protuberances on Majorca and Minorca]. A number of other, less pronounced instances can be seen, for instance on the Benincasa atlases, among them one on valdagosta (off Segna, Istria) and a cluster of at least three out of the north Istrian coast, one out of the north coast of Sicily by cifalu, another near rasamabes on the Libyan coast, others on the Canary islands of Tenerife ( inferno) and Gran Canaria [Islands], and one on the isola de sal in the Cape Verde Islands. A further example, noted only on the jointly-authored Corbitis and Pinelli-Walckenaer atlases (Pujades A 11 & A 12), comprises three lollipops projecting from mele dandole (east of Ventimiglia). A mela, identified by Debanne (2011, p.270) as a Ligurian term, denoted a shoal just off the coast. {The two preceding sentences added 10 February 2017}.

Since the charts carry no explanation for any of the conventions used within them, even for their scale, one is entitled to ask how, if the lollipops had a navigational meaning, that would have been conveyed to the users. The following observations surely put paid to the idea that those features could have carried any type of coded message:

  • No surviving chart before the 1330 (Dalorto/Dulceti) includes them. It was there used just for Limnos
  • The various instances were introduced onto the charts at different times; apart from Limnos (by 1330), most of the others are first seen on Catalan charts of the last quarter of the 14th century
  • The convention was never universally adopted and in some cases disappeared relatively quickly, e.g. those on Majorca and Minorca, whereas the Limnos and Canary Islands examples can still be seen in the 16th century
  • On the most extreme example, Limnos, all the promontories could be given the lollipop shape, but all might instead have been modified into finger-shapes, or some might be one type and some the other
  • While most, for instance those on Limnos and Corfu [see A note on the Cornaro Atlas], are presented as elongated headlands, others, like those on Majorca and Minorca, seem to emerge out of bays. That on Minorca could be interpreted as referring either to the harbour of Fornells or to Cape Caballeria, see above
  • Since rounded headlands are commonplace, sometimes with a very thin strip between those and the mainland, with various lengths of intervening neck, it is not clear - to us or presumably to the original users - which are indeed representations of reality or, alternatively, of any intended symbol
So a convention that, it is suggested, may have contained vital navigational information, whether for the safe entering of a harbour or warning of the danger of offshore rocks, was introduced later than other hydrographical symbols (and subsequently abandoned in some cases) and, on Limnos, was portrayed, erratically, in two different ways. Although Francesco Beccari, through his Majorcan connections, introduced some of the lollipop conventions into Italian chartmaking, it seems to have remained largely a Catalan convention. How, in the face of such partial and spasmodic use, could contemporary users possibly have known what this convention was supposed to mean? From the time of the earliest surviving charts there were unambiguous and universally recognised ways of denoting dangers. Afterwards, the printed sea atlases of the later 16th century (Waghenaer and Barentsz.) introduced a self-explanatory anchor symbol. On the basis of the above I can see no valid reason to add the lollipops as another category of essential navigational information.

Top of page

Island shapes as a mnemonic device

Imaginary islands, or to be fair to the chartmakers, those which had been reported without a precise geographical description, were, not unnaturally, given imaginary shapes. These shapes were obviously fictitious, and intended to be read as such. But they were also standardised. Each has its own accepted shape. Take for example the chain of islands, running north-south, that were thought to exist to the west of Portugal. Compare their first appearance in the later 14th century with the equivalent line on Benincasa's charts a century later and you can recognise the islands by their shape. 'Brazil', one of several islands with that name, is a little larger than the others and shaped as a circle with two or three scoops out of it. Just as we can 'read' those islands and immediately identify a particular one so it must have been easier and quicker for the chartmakers when drawing those freehand.

The discovery - if such it be - that it was common for portolan charts to give many of the medium and small islands of the known world a distinctive but obviously imaginary shape, so that they could be instantly recognised, is perhaps the major finding of this exercise. The islands affected are those with a size similar to that of Rhodes or smaller, down to mere islets. To present the islands in the Atlantic, which were 'known' from hearsay rather than direct experience, as circles, semi-circles, trefoils, and other geometric forms, is hardly surprising. But finding the smallest Aegean islands, and indeed those elsewhere in the Mediterranean, treated in the same way was unexpected (see above, and for an illustration The line of islands between Euboea and Rhodes, each with its distinctive shape). Observing the development in the early 14th century of elaborate, carefully repeated fictions to represent some of the better-known islands in the Ionian and Aegean archipelagos was even more surprising.

The island with the most complicated shape, Limnos, is unusual in that its outlines demonstrate parallel developments, towards greater realism but also towards imaginative excess. The 'lollipop' excrescences first seen on the 1330 chart were to be elaborated by most chartmakers, not just up to 1500 but, in the form of non-existent projecting promontories, until the late 17th century. On the other hand, some Venetian chartmakers, it seems from the early 15th century, grappled with the island's contorted coastline to produce a commendably realistic outline. This is not just a question of expertise, but rather one of attitude. How much did it matter that an island, at the head of the Aegean and on the route to the Black Sea, was given a totally imaginary outline, when it is shown in the correct location and at roughly the right size?

Why was this done? The larger Mediterranean islands - Majorca, Corsica, Sardinia, Sicily, Crete and Cyprus, and to a lesser extent Euboea (negroponte) - were given impressively plausible outlines. They probably owe those to the same original survey(s) that had produced the continental coastlines. Why not some, at least, of the smaller ones as well? It is hard to see any hydrographical reason for apparent absurdities like the lollipop-fringed Limnos [36], the space-rocket shape for Skyros [39], or the other forms borrowed from geometry and architecture, sometimes with hints of the natural world. But these were not passing fancies. Many of those forms became a shared 'signature' among, and sometimes between, groups of chartmakers. What seems to us a grotesque Limnos, first seen on the 1330 Dalorto/Dulceti chart, could not possibly exist in the real world. Yet its fantastic protuberances were certainly still there at the end of the 16th century and traces of the sequence of false promontories into which they developed can still be seen in the late 17th century.

So, if those island outlines can have had no navigational purpose (as discussed above) - the answer has to be sought elsewhere. The most plausible explanation is that these shapes were intended as visual mnemonic devices. The densely-islanded eastern Adriatic is also treated with conventionalised shapes but the fanciful outlines are even more evident in the Aegean. Examination of a modern map shows the problem; a glance at a portolan chart reveals one solution. If, instead of the convoluted, unmemorable shape of a real island, each is given a 'signature' or 'logo' shape different to that of its neighbours, armchair navigation becomes far simpler. If, additionally, the major Aegean islands, for example, Limnos, Lesbos and Skyros, are kitted out with highly individual outlines, the chart's original user (and indeed the researcher today) could immediately find his bearings, aided in the case of the portolan chart by the additional distinguishing feature of colour.

The early chartmaker's aim, it seems, was to place each island and islet correctly as to its position and size. But there was evidently little pressure from users for the island's actual outlines to be represented. The tortuous coastline of Limnos would be a challenge for any hydrographic surveyor, which makes the attempt shown by Vesconte so impressive, with two large bays in the south coast and the hint of one in the north. Venetian chartmakers, evidently in the early 15th century, elaborated that into an outline which reflected several of the island's more important peninsulas and bays. However, many later chartmakers, among them Benincasa, who worked for some years in Venice, ignored that attempt at realism and persevered with the 'lollipops'. So, it seems, did most chartmakers in the first half of the 16th century. Benincasa's persistence with those fictions is instructive. He had been a ship-owner for some decades before he lost his ship to pirates in 1460. In the period 1435-45 he had compiled a written portolan of the Adriatic. For someone who had been so closely involved in chart use and the codification of navigational instructions to have continued to replicate outlines he knew to have no connection with reality, underlines the need for this new interpretation.

The Isle of Man [3], lying between Ireland and England, provides an instructive example. The mnemonic cross form, used on early Italian charts, contrasts with the more realistic outline from their Catalan contemporaries. However, that was abandoned when, in 1439, Vallseca reverted to the symbolic cross instead, to be apparently followed on all Catalan charts into the 16th century. This may be the only identified form of such a repudiation of reality in favour of symbolism but it is surely significant. Even if he was imitating an Italian (say Beccari) Vallseca must have been aware of the semi-realistic Catalan forms, but consciously rejected those in favour of the mnemonic device. The symbolic form was not just acceptable but here, it seems, preferable.

The case of Majorca is one of the most intriguing to emerge from this analysis, with the addition of fictitious coastal features on the island that was at the heart of Catalan chartmaking. These were evidently first placed there by those same Catalan chartmakers, and seen most clearly on the1403 chart of Francesco Beccari, who had earlier worked in Majorca. Nearby Minorca had its own example. Nor does it stop with the Mediterranean or with 1500. For example, a chart by the Portuguese Fernandes Vaz Dourado of West Africa shows the Cape Verde group reverting to angular forms and, more surprisingly (since they had earlier been given realistic outlines), the Canary Islands. In each case, one island is a neat triangle. Another Dieppe-School chart (from the c.1538 'Dauphin Atlas') gives a treatment of the West Indies (including a triangular island) and the eastern seaboard of North America (with one island as a Greek cross) reminiscent of charts two centuries earlier [see Robert Putman, Early Sea Charts (New York: Abbeville, 1983) Plates 17 & 46].

Time-saving could have been another motive. A number of the mnemonic shapes for small islands are simple and could easily have been memorised by a chartmaker. Inserting the twin pairs of 'headphones' denoting Palagruza in the Adriatic (level with Dubrovnik/Ragusa) would have been the work of a few seconds, and likewise the cross of Paxos. Though the 'headphones' (possibly on the 1385 G. Soler chart but definitely a feature of Benincasa's work and borrowed from him by Roselli) are more likely to represent spectacles, which 'were probably invented at the end of the thirteenth century by a craftsman living near Pisa. The evidence originates from a passage by Friar Giordano da Pisa who recounts having met the anonymous craftsman in 1286' [see 'Early Printed Book Contains Rare Evidence of Medieval Spectacles' ('Cultural Compass', the blog of the Harry Ransom Center, the University of Texas at Austin, 17 April 2012)] {The two previous sentences added 31 October 2013}. The convenience of the device continued to be appreciated for centuries, see, for example, the Huntington Library's 1677 Cavallini atlas. {This sentence added 23 August 2015}.

Other shapes, even if more complex, could have been easily transferred to the chart from memory or with a glance at the workshop pattern, because of the usually symmetrical forms, whether made up of straight edges and right-angles or curving patterns. This made it easy to avoid what would otherwise have been the inevitable corruption of shape over many copyings.

An earlier commentator whom I had overlooked, Julio F. Guillén y Tato (in Imago Mundi 12 (1955), p. 111 & 114, fig.11), wrote: "we note a tendency to give many of the islands the configurations of capital letters, to the point that it is possible to select almost all the letters of the alphabet". He was writing about, and illustrating, a much later work, the J. Martinez atlas of 1591. Even if the shapes he illustrates were really intended as letters - and there is scant evidence for that in the early period - there is no apparent connection with the initial letter of the respective island's name. {This paragraph added 20 October 2013}.

The earliest outlines for the Aegean islands, on the Carte Pisane and the Cortona chart, already betray the beginnings of individual characterisation, though some are little more than rectangles and none is distinctive. Clearly, quickly-drawn boxes would have been the simplest of all, so the move to more complex but equally imaginary outlines, visible for the first time on Vesconte's chart of 1311, must have had a reason beyond that of the chartmaker's convenience. Memorability and differentiation are the reasons suggested here.

As discussed by, for instance, Mary Carruthers in The Craft of Thought (1998) and Giorgio Mangani in Cartografia morale (2006, pp.102-06), mnemonic devices were commonplace in a medieval monastic environment. Nor is it hard to find examples from the classical period, with Strabo stating that it was sufficient to note an island's length and breadth and then give it a memorable geometric form - for example Sardinia as a human foot and Sicily as a triangle - so that, by providing territories with familiar shapes, they would be readily memorised (Federico Borca, 'Facies locorum: morfologia, onomastica e percezione delle isole nella cultura romana', Quaderni di Storia 51 (2000), pp. 189-201, especially p.191). [I owe these points to several of those thanked at the foot of the introductory page.]

Thus the concept of mnemonic shapes within a cartographic context was not new. However Strabo's Geographia was not circulated in Latin in western Europe until the 15th century. Nor, evidently did such a device form part of the original conception of the portolan charts. The first hints of mnemonic shapes are not seen until Vesconte's chart of 1311 and a full commitment to the device is not apparent before the Dalorto/Dulceti chart of 1330. It was then applied in a quite different way to that of any classical model, to the smaller not the larger islands. Nor, as far as I could see, were most of the shapes used describable in words. However distinctive they might be, the shapes for Zakynthos, Limnos or Skyros, do not look like representations of well-known everyday objects.

So, while it is perfectly logical to see the portolan charts' mnemonic island shapes as imports from the monastic or other milieus with which the early chartmakers might have come into contact - particularly that of the clerics who are credited by Gautier Dalché and Pujades with the creation of the prototype chart in the early 13th century - we should not rule out the possibility that the idea for the shapes that seem so strange to us may have come from within the world of portolan chartmaking itself. Certainly their inclusion in what was a functional navigational aid, where a premium is placed on spatial accuracy, marks out the portolan chart examples from comparable early learning constructs concerned with non-geographic themes.

Colour, described separately above, could have had a mnemonic value as well, especially when persistently associated with a particular island shape. Giorgio Mangani ( Cartografia morale, 2006, pp.104-06), noted that colour was an integral part of the art of memory - referencing the 12th-century Ugo di San Vittore.

It may be necessary to dispose of any parallels with the so-called stick charts of the Marshall Islanders of Oceania. These were used for teaching purposes but not taken to sea as a navigation aid [Ben Finney in David Woodward & G. Malcolm Lewis, The History of Cartography, Volume 2, Book 3 (1998), p.485, citing William Davenport, 'Marshall Islands navigational charts', Imago Mundi 15 (1960) pp. 21-2.]

On the mnemonic shapes see two illustrated articles by Tony Campbell:
'Why the artificial shapes for the smaller islands on the portolan charts (1330-1600) help to clarify their navigational use', Cartes et géomatique, 216 (June 2013): 47-65 [to be released freely online in summer 2014] (paper delivered at the international conference, 'D'une technique à une culture: les cartes marines du XIIIe au XVIIIe siècle', Paris 3 December 2012).

'Why were medieval sea charts still being produced four centuries later? Does the answer lie in the Aegean?', text and podcast of the paper given to the international conference, "Cyprus on the crossroads of travellers and map-makers from the 15th to 20th century", organized by the Sylvia Ioannou Foundation, Athens, 18-20 October 2012, at the Museum of Cycladic Art.

Top of page

A deeper understanding of the function of portolan charts

That it was some of the larger Aegean islands, and one or two of those in the Ionian chain as well, that were given the most imaginative mnemonic treatment, suggests that, rather than being an admission of defeat in the attempt to reproduce their true outlines, this could have been a way of underlining their geographical importance. Thus what at first sight appears a strange contradiction between these hydrographical fictions, on the one hand, and the remarkably accurate overall outlines of the continental coasts and the larger Mediterranean islands, on the other, can be shown to form part of one coherent whole.

If that is so, what are the wider implications for our understanding of portolan chart function?

The placement of the Aegean islands, however detailed, is far too erratic to have been of use to a helmsman trying to make his way through or between them. But the charts offer what is probably a more or less complete geographical catalogue of the islands, each named and with a readily memorised outline of roughly the right size and in the approximately correct position. For a visiting ship this would be an invaluable aide-mémoire when proceeding through the watery labyrinth, even if a local pilot or past experience was needed to thread a safe way through the dense grouping of islands, particularly at the southern end.

For a helmsman to have depended entirely on documentary aids he would have required a far more accurate chart at a much larger scale. In addition, he would have needed other accessories considered essential today, for example, recognition views (or silhouettes) that convey the appearance of passing islands (as found in 16th-century printed works), enlarged harbour approach plans combined with written guidance, or traced-out marine routes (like those in a recent British Admiralty Pilot).

We know that portolan charts were regularly used at sea, and by sailors who owned personal copies (often along with a compass), as has been amply confirmed by Pujades (2007, pp.458-63). What is less clear is exactly how they were employed by those responsible for the ship's navigation. It is likely when in the Cyclades, for example, where one or more islands would have been in view at any one time, that sailors would use the visual association of the easily memorised shapes (each linked to its name) as a convenient way to monitor their progress and plan their route. The central importance of toponymy to the charts' purpose is thus re-emphasised.

When sailing along a coastline and consulting a chart which presented the important features, principally headlands, in their correct spatial relationship (even if their shape was non-specific and the space in between was wholly unrealistic) the seaman would follow the sequence of names. Within an archipelago, where only a small part of an island would be glimpsed in passing, its overall shape was irrelevant. What mattered was its position and name, which is why the Aegean toponymy is so cluttered, with 100 or so names documenting every feature down to the smallest islet. Whatever navigational use might have been made of a portolan chart for longer voyages, when within sight of land - rather than being used for way-finding - it would have served as a means of recognising the ship's position in relation to a visible feature, or of sequencing a route by means of its place-names.

However, since portolan charts are also rightly termed 'compass charts' - created with the compass and designed to be used with one - we should not ignore the role the charts played when the ship was out of sight of land, which would not normally be for more than two to three days (Astengo, 2007(a) p.174). Before setting out, the distance to the destination and its compass bearing would be observed, and subsequent progress noted on the chart according to a mental estimate of what had actually been achieved when tacking. The 15th-century Benedetto Cotrugli said that the estimated position would then be marked on the chart with a spot of wax (Falchetta, 2008 p.272).

In reading a portolan chart, therefore, we need to be prepared to approach its information in very different ways, reflecting the two levels involved. On the one hand there was the attempted overall accuracy of the continental coastlines and those of the larger islands, which were understood by both producer and user to comprise a sequence of correctly positioned headlands, more important river mouths and settlements, but certainly not a continuously surveyed outline. This would have provided a trusted background, albeit an 'indicative' rather than 'exhaustive' one, against which to assess a planned voyage (Pujades 2007 p.460a). For sailing in an archipelago, on the other hand, the user was presented with the helpful mnemonic 'logo' forms given to the medium and small islands. The difference in these functions is that between navigational and pilotage use.

A similar two-tier accuracy can be seen in the placement of the toponyms. Names will usually be carefully positioned alongside the more obvious coastal features, for example headlands and bays. But the intervening toponyms will usually be evenly spaced, so that there is no precise fit with lesser headlands or the short paired lines indicating a minor estuary. In other words, it was the sequence of names that mattered, not their precise location. {Preceding paragraph added 25 April 2011}

In that sense, the portolan charts offered a generally reliable account of the region's hydrography, and one more detailed than any sailor could carry in his head. The charts were also free from the computational or copying errors of distance (and sometimes even direction) found in the written portolani. If the portolan chart's value in ship navigation was less precise and far less essential than the use of a chart in recent times, the role that Cotrugli describes for it in a sailor's apprenticeship demonstrates that it was not of negligible importance. 'After the ship-boy had served for some time in his apprenticeship, and had become an expert in life on board, in manoeuvres and navigation, he was to learn the use of the portolan chart and become conversant with calculating the ship route' (Falchetta, 2008, p.275).

We need to avoid either of two extreme positions: on the one hand, that the charts were essential for navigation, or, on the other, that, if indeed taken to sea, they had some other, non-navigational purpose. For Patrick Gautier Dalché, in 'Cartes marines, représentation du littoral et perception de l'espace au Moyen Âge' (2001), and Piero Falchetta, in 'The use of portolan charts in European navigation during the Middle Ages' (2008), the portolan charts, while certainly used at sea, were too inexact for precise navigation. Falchetta concluded 'that neither portolans nor charts were really employed in navigation for the purpose of calculating routes. Or at the very least, we can say that they were only one of the elements considered in navigation, and probably not the primary one' (p.274).

Falchetta's statement leaves unclear what other external aid, besides the umbilically-linked chart and compass, could have been involved, since there is little or no evidence of the shipboard use of either the portolani or martelogio. Falchetta's position is also undermined by his own account of Cotrugli's writing, cited above. We should perhaps see the charts' navigational function in the open sea as providing a constant background to the relatively imprecise, and no doubt sometimes conflicting, observations of the crew.

Both the positional accuracy of the coastlines' nodal points and the integrity of the mnemonic island forms depended on a commitment to painstaking care, over many generations, in the copying of even the smallest detail from workshop models. That the potential corruption of the content of documents that are concerned roughly equally with text and shape was staved off until at least the 16th century was due to the fact that the portolan charts' purpose depended on the accuracy with which they were copied. A 'Chinese whispers' approach to freehand copying would have rendered them unfit for their practical use within a very short period. [See further below 'Conservatism and workshop practice'.]

The reinterpretations above, implying that in some respects the portolan charts employed impressionistic devices does not mean that we should interpret them as any less 'scientific' than has been previously thought. For a practical tool what matters is its fitness for purpose. I am merely proposing that the charts served three purposes - when out at sea, when running along a coast, or when working through an archipelago - and that their creators, or perhaps more accurately their later adaptors, chose quite different ways to achieve those complementary requirements.

Top of page

What is the difference between a plain and a de luxe chart?

We know that large quantities of cheap charts were produced (see Pujades, 2007, pp.498-501, and the table of prices on pp.279-80). We might expect that such charts would omit inessential flourishes and that they might be written in a less careful, hasty hand. The aim of mass production is to reduce the amount of time spent, and the precise inscribing of perhaps 1,800 small place-names on each chart, and the colouring of hundreds of islands and minute hydrographical details, would have been major components of the labour involved. For a description of the details required for even the most basic chart and the elements that might be added on charts of varying degrees of adornment, see Stages in the construction of a chart.

Because Catalan charts tend to be ornate, any unillustrated ones (of the type assumed to be referred to in the Vallseca contract of 1433) stand out. The listing of charts given by Pujades (the Catalans did not produce atlases [the so-called 'Catalan Atlas' is a series of panels]) includes a column about the level of decoration [this is only in the Catalan part of his book, 2007, pp. 63-5]. Three surviving examples (with the Pujades 'mínim' level, albeit including a few city vignettes) may be instructive. Two are assigned by Pujades to the 'Cresques atelier' (his C 15 & C 18) and the third is the Soler chart of 1385, from the same general period. We can see from those examples that whatever shortcuts might have been taken, their creators did not reduce the volume of place-names, except perhaps where there was insufficient space round a crowded peninsula, nor did they write in a less legible hand. Nor, as one might have expected, did they omit the island colouring - at least there are no signs of that on any surviving fragment of a completed chart.

An earlier fragment, assigned by Pujades to a Catalan author in the mid-14th century, and certainly of a very basic design, unfortunately covers just the eastern Mediterranean and Black Sea and so was not included in his toponymic analyses (his C 12). Since it is also in poor condition it would be hard to test whether or not it displayed a reduced selection of place-names. [For further comments on this work, see here.] See also the description of an unusually plain Guillem Soler chart (announced in 2011).

So, even if there was no need for a chartmaker to have artistic skills, all the other basic functions would have had to be carried out in the production of a 'no frills' chart. Although it seems probable that they developed workshop procedures to speed up production, which were then passed on to future generations, there is no evidence that chartmakers tried to save time by omitting elements or reducing quality. They were servants of their market and a chart had to be fit for purpose. It seems, therefore, that the only real difference between a plain and an adorned chart relates to the latter's adornments, added as the final stage - unless, perhaps, a better quality vellum might have been selected for de luxe works, along with more expensive colours and gold, and, perhaps, better workmanship.

However, examination of the two Cresques atelier charts just mentioned, shows that the colour was applied with just as much care as on expensive works. Looking at the C&SA table, for example for 14th-century Catalan works, it is not possible from the profiles to distinguish those that are plain from the de luxe versions. In other words, the artistic embellishment that gives some Catalan work its distinctive attractiveness is often no more than icing on the standard cake. Because of that it seems perfectly possible that standard charts might have been kept in stock, to be passed to the artist when a special order came in. It would certainly have reduced the delivery time.

The distinction between plain and ornate is of less relevance for Italian work where much of the production was strictly functional anyway. Parts of some charts survive as recycled fragments, usually in a binding. This is where one might have expected to find samples of charts that were even more rudimentary but none has yet been identified. However, four very plain productions, assigned by Pujades to unknown Genoese practitioners in the first half of the 14th century, some of which are now incomplete (his C 9bis, 10 & 11 and A9), should have been mentioned here {This sentence added 20 October 2013}.

Astengo (2007(a), e.g. pp. 177a, 208-9, 211) discussed the question of plain, functional charts in the later period, designed purely for shipboard use. He concluded that, from the second half of the 16th century, the balance between the practical and the ornate changed in favour of the latter. He of course accepted, as for the earlier period, that charts made for sailors had far less chance of surviving, or reason to do so. Nevertheless, he proposed that a large gap in the extant charts of the Prunes family can be explained in terms of the production of low-cost, and hence effectively ephemeral, works. The officially sanctioned Maggiolo workshop in Genoa was also producing functional works, as corroborated in 1607, in the recommendation for a family member to obtain the post of official cartographer, so as to produce works 'used by all Genoese [and other] sailors'. When he obtained the post four years later reference was made to 'navigation' charts.

So, even if there was a clear move in the later period towards ornamental display and away from practical needs, we can see the traces left by those everyday objects in the archival record and indeed in the increased toponymic dynamism of the second half of the 16th century. {These two paragraphs added 12 September 2011}

Top of page

Conservatism and workshop practice

A process of relentless copying of a document over a period of centuries would normally lead to such cumulative corruption that the manuscript in question became effectively unusable. However, this did not happen with portolan charts, at least until very late. Why not?

Geometric consistency
In his recent doctoral thesis (2010(a) pp.25-6) Joaquim Alves Gaspar summarises one of the conclusions reached in the 1987 dissertation by Scott Allen Loomer (which I have not seen): 'first there is no evidence of any improvement or degradation in the accuracy of the charts with time. The author concludes that the methodology for the chart construction was fixed early in their history and was followed afterwards without alterations' (re Loomer p.168). Gaspar goes on to point out that in his own comparison of the charts of Dulceti (1339) and Aguiar (1492), 'no significant differences in the main geometric features of the two charts were found ... and that the orientation of the Mediterranean was copied from older prototypes and remained more or less constant until 1600 ...' (pp.25-6).

This pattern of conservatism is borne out by Francesco Beccari, noted as apparently the first to correct the mismatch of scales between the Atlantic coasts and the Mediterranean. He either failed to detect a further mismatch between the Mediterranean and the Black Sea or chose to ignore it (see here). This contrasts with the way that he made alterations, albeit small ones, to both the toponymy and outline of the Black Sea, when copies of two of his charts preserved in a later Venetian collection are considered (see Cornaro Atlas). The reluctance to alter the alignment of the Mediterranean so as to match the changes in magnetic variation may have a similar explanation. Even if sailors did not understand the progressive nature of that change, they are likely to have observed its effect in practice, and mentally adjusted directions, in the same way that they must have compensated for distance when in the Black Sea.

That the expected corruption of the coastal outlines did not occur to the portolan charts of the 14th and 15th centuries, must therefore be due to two special circumstances.

In the first place, the all-important nodal points of the coastal outlines were preserved from distortion because some means was found (and we still do not have the confirmed details of this) to transfer directly from a single workshop model (or pattern) to a new chart. That would have been the only way to avoid errors becoming permanent, which would have happened inevitably if each copy became the model for the next. Those patterns, along with the expertise of the individual(s) concerned, represented most of a chartmaker's business capital. They must have been looked after very carefully, copied meticulously themselves when this became necessary, and bequeathed on death or retirement, as the post-mortem inventories of Rafael Monells (1468) [Pujades 2007, p.437, No.121] and Vicenç Pruners (1625) [Pujades, 2007, p.479b] make clear. A freed apprentice, setting out on his own, must surely have taken with him, even if illicitly, a copy of the workshop model. Like printing, direct copying from patterns militated against improvement, but it also guarded against corruption. That the outlines of some late charts have become so distorted - and the 17th-century work of G.B. Cavallini is a good example - is presumably because they were then being copied freehand.

The other essential mechanism for ensuring faithful reproductions of the workshop's master chart was the training that chartmakers had undergone. With so little documentary evidence about the apprenticing of medieval chartmakers, and no details of workshop copying procedures before the 16th century, we are left with whatever evidence can be extracted from the charts themselves. Whether the boy was formally apprenticed to an unrelated master, or was being trained up by an older family member, he was presumably expected, by the end of his seven or so years, to have mastered all aspects of the craft, enabling him to make an entire chart on his own.

There must have been rules governing every stage of production and it would have made sense for the master to instil precise ways of carrying out each task (on which see Stages in the construction of a chart). It is hard to conceive of a workshop where the apprentice was given a range of choices rather than being instructed in the single 'correct' method. When dealing with a young teenager that would be essential. The result would be an unthinking obedience to the strict workshop conventions. In that way, nothing would be left out, added in the wrong sequence or done in the wrong way, and the work of different individuals in a single workshop could blend invisibly.

There cannot have been much use for unskilled or careless labourers in a chartmaking workshop. A moment's inattention could destroy an entire chart, since it was hard to erase a major mistake. The stains from the spilt ink-well in 1838 can still be clearly seen on the 1439 Vallseca chart. So we must assume that any assistants were taken on as permanent employees with the necessary amount of training being invested in them. Whatever their level of skill, their job must have consisted almost entirely of copying from exemplars (models or patterns). Such patterns could be modified to accommodate new information but the replacing of one model with an amended one would not have altered the working methods. If minute comparison was made between the multiple works of one of the more prolific chartmakers, it might reveal just how closely the model was followed: which parts were directly traced, and which copied freehand. Providing any freehand work was always based on the workshop model, there would be no danger of progressive distortion.

Given the above, the work of an apprentice - possibly up to the point when he could complete an entire chart for the master to sign - should not be visible, except in the handwriting. Though even here it is likely that he would have been trained to enter the place-names in a formal hand that tried to imitate his master's.

With very few exceptions, those we term chart makers were no more than chart copyists. A copyist would reproduce unthinkingly what he was taught or what he saw in his models. He did not select, he rarely modified, though he might omit, for example a place-name (perhaps by mistake). To us, the minor details documented in the C&SA are no more than trivial parts of the background, indeed some are barely visible. But the C&SA findings make it clear that the copyists did not make such distinctions themselves. The hierarchy of value and importance that historians have imposed on portolan chart content is ahistorical. Every feature on the chart has been reproduced as the result of a conscious extraction from a model or as an act of memory, and each was placed in the new chart with the same care. Indeed, the smallest of the hydrographical features, barely discernible in the original, might have demanded more close attention than some of the larger elements (on which see 'Small hydrographical details'). It seems that their respect for the authority of the master chart from which they copied was no less than that of a monk in a scriptorium for the text in front of him.

So, if the charts' users were expecting hydrographic improvement or the speedy introduction of new place-names, they would have been disappointed. There was, of course, regular toponymic innovation but much of it was neither timely nor of much significance. Like any other craft requiring extensive training, the portolan chart business was inherently conservative.

Where is the expected conformity?
The C&SA throws up a problem. If - and this is Pujades's contention - a highly efficient workshop (at least in the 15th century) was a necessary response to the 'time and motion' problem, without which it would have been impossible to produce charts fast enough to satisfy the market, would it not have made sense to produce exact replicas of the model each time? As far as colour was concerned, would that not have led to total consistency with, say, Minorca always red, Malta blue, and so on? Why stop to think, and perhaps end up with neighbouring islands in the same colour, or abandon an almost universal convention (such as red for Formentera), when it would have been quicker and easier just to follow the model?

But, if that had happened all unadorned works by a given chartmaker would display consistent island and estuary colour. Not only is that not always the case, but there is not a single chartmaker in the period up to about 1450 whose consistent use of colour pointed to the invariable use of a workshop model containing those elements. Benincasa and Roselli come close to it but some of the others even vary the colours on overlapping charts within a single atlas. Vallseca himself, or his atelier, was not entirely predictable.

Applying modern assumptions about business efficiency would here be clearly anachronistic. Some chartmakers tended to use certain colours, even in a majority of instances, but there are always exceptions, which cannot be explained in terms of a change in policy at a certain date. Nevertheless, their use of colour was never entirely random and, both as individuals, and as part of a local tradition, when considered overall, most had a visual signature, which might help in identifying fragments or unsigned works.

A chartmaker working alone may do as he likes. But one who trains up an assistant/apprentice must surely provide a fixed model that had to be followed very precisely. Any other way would waste time and lead to chaos. Is it possible to envisage an undisciplined workshop? But where do we see the evidence for such anonymous workshop productions, drawn by a single assistant, or perhaps by a team of specialists under the supervision of the master? If division of labour had been employed, this would have made it more likely that the results would have been identical. But can any of the survivors be seen as one of these cloned charts?

Can we, then, learn anything from the C&SA tables about the assistance an individual master chartmaker might have been receiving? To measure consistency we need, of course, at least three surviving works. The consistency analysis (see Colour Consistency Tables, a Microsoft Word document) could therefore consider just seven named chartmakers and the hypothetical Cresques atelier. The fewest consistent colour conventions can be seen in the work of Ziroldi and among the charts ascribed to a Cresques atelier. In the latter case this might, perhaps, be explained in terms of the licence accorded to mature collaborating chartmakers, though the Vallseca productions, assumed to be the product of a sizeable operation, are much more consistent.

Ziroldi, however, is more difficult to understand. His signed works and those attributed to him seem to be in the same hand, and the unsigned works fit neatly into a single overall pattern. But that includes a surprisingly different middle period. Perhaps further research will provide an explanation for that anomaly.

The charts of Roselli and the productions by Benincasa (mostly atlases) demonstrate an obvious adherence to colour conventions - in the case of Benincasa, one of almost total conformity. However, nothing in the respective C&SA tables for their charts argues against the possibility that each worked alone, and they provide us with no evidence about the work of assistants.

Top of page

Expectations contradicted by the C&SA findings

In considering those who have left us three or more works it is now clear that there is no evidence of the following:


  • any chartmaker being wholly random in their colour choices
  • any chartmaker being wholly consistent with those choices, although Benincasa comes very close
  • colour choices being determined by the availability of pigment
  • red being used to indicate importance
  • universal colour conventions, although certain islands tended to be coloured consistently within one of the three main centres of production
  • a 'cloned' chart, identical in all respects to another - as might have been expected from quasi-factory-style production [though the absence may just mean that none has survived]
  • evidence of the work of any assistant(s) within the signed work of a master [though that might still be detected if full comparisons could be made], i.e. it may be impossible to distinguish the work of a practitioner working alone from that of one operating an atelier
  • a known former apprentice (e.g. Macià de Viladesters or Arnau Domenech) reproducing precisely the colour choices of their master
  • differences between the cheaper type of chart [unless none has survived] and ornate versions, other than in their final artwork flourishes
  • realistic outlines (with a few exceptions) being provided for islands smaller than Rhodes
  • a strictly realistic approach being taken even to Majorca, and even by the Catalans
  • general improvement in the shape of medium-sized islands; instead, the fictitious outlines tended to be made more extraordinary
  • the abandonment of imaginary island shapes until well into the 16th century or beyond
Top of page

Sixteenth-century continuations

Though there is some discussion of 16th- and even 17th-century works in these pages, the selective nature of the available evidence reduces the authority behind any suggested conclusions. The main purpose of continuing the analysis beyond the Pujades cut-off date of 1469 was to look for the continuation of earlier conventions or the introduction of new ones on the Old World regions of charts whose makers might by then be more concerned with the newly discovered lands.

Even concentrating on the period between 1469 and 1550, particularly after 1500 when portolan chart production was increasing significantly [Pflederer, 2009, p.xxii], there is a lack of adequate printed reproductions or online scans. The images of dated works from that period accessible to me are listed on the LaterChartsTable (a Microsoft Word document) and I would be very grateful for additions to that list. Certain C&SA features were traced to their apparent conclusion, in some cases, in the late 16th century or even beyond. But the full analysis was carried out for just a few selected works, with the intention of documenting the earliest accessible work by each of the prominent later practitioners.

Because Benincasa's successors preserved so faithfully most of the conventions he had established in the 1460s (if not always his colours), giving them a life of exactly a century, it seemed sensible to reflect this almost entirely separate 'school' by producing different C&SA tables for that 80-year period:

No detailed commentary on these later continuations has been provided but the chronological sequence of non-Benincasan works, when taken in conjunction with the 'last dated appearance' notes to the individual C&SA features - see Summary table of the features analysed (a Microsoft Word document), and the Colour Feature List - does offer pointers for any historian who wishes to dig deeper. Perhaps with the aid of this provisional analysis it will be possible to suggest ways of differentiating works currently assigned imprecisely to either the later 15th or early 16th century.

Top of page

For the full details of the works mentioned above see the Bibliography

Colour Feature List   |  Summary table of the features analysed (a Microsoft Word document)

Colour & Shape Analysis Menu   |  Portolan Charts Main Menu

HOME (main menu)  |  Tony Campbell articles  |  Portolan charts menu  |  Index   |   Sitemap   |   What this site is about   |   What's New

eXTReMe Tracker