 |
(main menu) |
CAMPBELL articles |
CHARTS menu |
|
SITEMAP |
site is ABOUT |
NEW |
 |
Acquired in 1948 by the Danish Maritime Museum, the Helsingør Atlas had eluded portolan-chart historians until its re-discovery in April 2024. With its addition to the MEDEA-Chart Database of high-resolution images of its five sheets, the atlas became freely available worldwide.
The three aims of this essay were to identify the atlas's author, where he worked, and what sources he had used. Little progress had been made until an accidental re-sighting of the Luxoro Atlas. Although unsigned and undated, this was clearly in the same hand as a chart produced by Francesco Cesanis in 1421. It was immediately evident that the Helsingør Atlas's creator had closely copied aspects of the Luxoro Atlas, namely its miniature format, its sheet arrangement and most of the coastal outlines. That answered many of the questions about the source used by the Helsingør Atlas, but not about its toponymy or a number of its unusual features.
The portolan charts' place-names have always been one of the best guides to a work's dating and lineage, and at first glance this appeared to place the Helsingør Atlas some 40 years later than the Luxoro Atlas. For example, among the Helsingør Atlas's toponyms, and found for the first time on dated charts, were a number seen on the 1461 chart by Grazioso Benincasa and the 1462 chart from Nicolò Fiorino. However, a check among the less-studied charts and atlases of the middle years of the 15th century found that several of those toponyms had appeared on undated works thought to have been drawn before 1460. Indeed, the number of names for which there was no earlier claim was whittled down from twenty to four. Furthermore, since most of those 1461-2 names were on Benincasa's maiden chart, the likelihood of his being responsible for adding those names himself can be questioned. Until 1460, Benincasa had been a sea captain until he lost his ship to pirates. The next year we find him issuing his first chart. Is it likely that he could have copied from one or more models and have adjusted those to add the 18 toponyms that his chart seemed to have introduced? His subsequent career (up to 1482) shows no signs of originality. If that line of reasoning, that his supposed toponymic innovations were actually borrowed from others, is joined with the strong possibility that the four crucial '1461-62' names might well be found in charts not yet available in legible reproduction - or that may be discovered in future - then the necessity for the Helsingør Atlas to be dated after 1462 disappears.
Much of the essay focuses on a number of unusual, if not rare, features seen on the Helsingør Atlas, checking for those on the charts and atlases of the period, listed in the final Table (12). It was hoped that the repetition of those might point to one or more chartmakers who could have provided a source for the non-Luxoro Atlas elements. The results were unexpected: most of the eight main features being compared appeared on just a few of those listed works, and not a single one of them included more than three of the instances.
Two of what might be termed 'signature features' had special significance in relation to the two chartmakers who book-end this investigation, Batista Beccari and Benincasa. The 'Lacus Fortunatus', an island-dotted bay set into the west coast of Ireland, and associated with Saint Beatus, was provided with a formulaic note that would be repeated afterwards, even if in a reduced form. The first instance, on Beccari's chart of 1435, named 368 as the number of the islands. That figure was repeated in the Helsingør Atlas but Benincasa, and his many imitators, cited 367. This provides a handy differentiator between the followers of Beccari and Benincasa.
Another significant feature was the turning of the charts' corners into triangles whose longest side was a scale bar. The two shorter sides were either equal (isosceles) or unequal. The Helsingør Atlas and Benincasa are examples of the first form, for which Benincasa had been considered the originator on his first atlas of 1463. However, two of the atlases thought to date to the second quarter of the 15th century include that feature, one in each of the styles. This, again, leaves Benincasa's primacy in doubt. Moreover, there is no evidence that the Helsingør Atlas borrowed anything from Benincasa, nor are there any other indications that the atlas is later than 1460, which is thus proposed as a tentative date for the atlas.
There were a number of other observations, among them these:
Although this investigation into the very small, and what turns out to be highly unusual, Helsingør Atlas, found dissimilarity rather than shared patterns, and individuality instead of repetition, it does give us a glimpse into the largely unstudied world of Venetian ateliers, presumably small, operating independently, and producing anonymous, undated works, some of which remain un-reproduced.
The Helsingør Atlas's five charts are the oldest preserved in Denmark by well over a century and it shares with the Luxoro Atlas the distinction of being the smallest surviving portolan atlas, at least for the period up to 1500.
When the creator of the MEDEA-CHART Database, Joaquim Alves Gaspar, sent out an appeal to European archives for information about any unrecorded portolan charts they might hold he was informed of an atlas in the Danish Maritime Museum in Helsingør (M/S Museet for Søfart, 1948:0182), acquired in 1948 but hitherto unknown to researchers. As shown by the 'The Pflederer Census of Portolan Charts & Atlases' (privately published and regularly updated), its five charts are the oldest preserved in Denmark by well over a century. Thorbjørn Thaarup, a curator at the Museum, kindly supplied high resolution images, which have been added to the Medea Database (and see [PP 4]) .
At the time of its acquisition, the Atlas had been tentatively miss-attributed to Battista Agnese (c.1540) [on which see the Appendix].
Gaspar recognised a connection with Benincasa (fl. 1461-82) and alerted others working in this field. Not only do I owe my awareness of this intriguing portolan atlas to both Thaarup and Gaspar, but they have most helpfully provided information and suggestions.
The Helsingør Atlas is a handsome production and very small in size (on which see B.3. 'Miniature atlases').
It has an ornate binding that is apparently original, and with the top-edge stamped by gauffering into the gilded surface. This becomes visible when the upper edges of the vellum sheets are pressed together. It is likely that one binder was responsible for both elements [see PP 3]. Adding to the impression of a luxury production, some of the island colouring on the charts themselves uses gold leaf.
The investigation described in this essay was prompted by an unsuccessful first attempt to locate the Helsingør Atlas, in terms of its likely date and place of production. It was soon realised that the atlas's originality merited a deeper analysis.
But first, for those not already au fait with what many might consider a backwater of medieval cartography, what is a portolan chart? In brief, it is a hand-drawn sea chart, originally devised for the use of sea-going mariners, which emerged, perhaps at the end of the 13th century, on the west coast of Italy. Drawn on a single animal skin or, as here, on a series of sheets in an atlas [PP 4], they have some invariable features. The most obvious of those are the coastal outlines of the whole of the Mediterranean Sea along with its islands, and extending to the Black Sea. A few thousand toponyms are placed inland perpendicularly to the coastline, so as to leave the sea clear. Underlying everything is a network of compass directions which will be discussed below (B.2. 'The compass-lines network'). There is of course far more to a portolan chart or atlas than that, for example the detailed warnings of dangers to shipping, caused by rocks and shoals, conveyed by black crosses and red dots.
1
Charts could also be highly ornate and used for many purposes besides navigation. I hope that is adequate as an introduction to the 'mystery' of the portolan chart.
It was immediately obvious that the Helsingør Atlas was older than the Museum's attribution to the prolific mid-16th century chartmaker, Battista Agnese. So a quick assessment was made, which found elements that were attributable to two of the better-known 15th-century chartmakers, Batista Beccari (fl.1426-39) and Grazioso Benincasa (fl.1461-82) [PP 6]. An open-ended search was set in motion for the source or sources of that atlas. The time-band that was thought to be relevant, 1430 to the later 1460s, reflected the combined activity of those two chartmakers. However, the Helsingør Atlas has been the main focus of attention throughout, with its five charts reproduced first as a composite image in the Medea-Chart Database (No. 1) and then in eight images (2-9). [It is the Medea sheet numbers - of, variously, whole and half-sheets - that will be cited throughout rather than the five original sheets.]
If the hand-writing had been recognised in one of the reproduced works whose date was assured, we would have been left to do no more than conjecture whether the Helsingør Atlas was earlier or later than the other. But if such a matching had been with an anonymous, undated work, that would probably have left the dating question unresolved. Furthermore, as a scribe would presumably have mimicked the hand of the model only if they were working in the same atelier - and perhaps not even then - the search for specific handwriting and the type of atlas being copied could well be two separate operations [PP 10].
The exercise to be described below is formed of two distinct parts: first, the range of different kinds of feature, which might point to an individual chartmaker, a place of production or school; and, second, the charts' extensive, and steadily developing, toponymy. The decision to concentrate broadly on atlases rather than loose maps was partly one of convenience. The place-names on single-skin charts, which would have been exposed to the elements, are usually more difficult to read than those on the sheets sheltering in the protection of a bound volume. Indeed, some are effectively illegible. In addition, vastly more single-sheet charts survive than atlases, and the inclusion of all of those would have required a disproportionate amount of time. For instance, unsigned works that are thought to belong to the second half of the 15th century comprise around 50 charts compared to fewer than 10 atlases. A comprehensive examination that included all the separate charts would of course be valuable but unlikely to do more than match one or two works with those that are signed and dated.
This is a good place to pose a question that I do not think has been asked before. Why is it assumed that the elements which make an atlas by Benincasa, for example, immediately recognisable, were necessarily devised by him? The newly available Helsingør Atlas, with some similarities to Benincasa's work, leaves us faced with two diametrically opposed possibilities: first, that the Helsingør Atlas must post-date the 1461 Benincasa chart - or indeed his 1463 atlas, because it includes some of the features supposedly introduced there - or, second, that Benincasa had done no more than revive features from a Beccari chart. Resolving that dichotomy would decide whether the Helsingør Atlas (or at least its contents) should be confidently dated pre- or post-1461.
In the attempt to resolve those issues, three separate investigations were involved: the identification of 'signature' elements in the Helsingør Atlas and those being compared with it; the drafting process and handwriting; and the first instance of certain toponyms. Each of those benefitted from a number of detailed finding aids, as follows:
and various works by Ramon Pujades, dealing with the period before 1470:
[The DVD attached to that includes almost all the pre-1470 charts and atlases, however access may now be difficult.]
Reversing the actual chronology of this investigation, it was only after several months spent searching for clues as to where the Helsingør Atlas might fit into the broader history of the portolan chart, that there was an accidental breakthrough. The eureka moment was the realisation that the Helsingør Atlas was closely modelled on the Luxoro Atlas. Given that the chronological window for the Helsingør Atlas had originally been set at 1435-1460 or even later, the Luxoro Atlas - in the distinctive handwriting of Francesco Cesanis, and evidently earlier than his dated 1421 chart - had not been considered. 2 But, once the Luxoro Atlas appeared in the discussion about miniature portolan atlases (discussed below) 3 it was immediately clear that the Luxoro Atlas had to be the close model for the Helsingør Atlas. There can be no doubt that the Helsingør's creator saw the Luxoro atlas, or, more likely, a lost intermediary, and imitated not only the simplified compass-line structure (on which see the following section) and most of the sheet divisions, but also the placement of the toponyms. But the heart of a chart is the coastal outlines and they are so similar in the two works that they must have been copied by some direct replication method, i.e. not by eye [PP 7 & 8]. 4
No attempt was made to compare the coastal configurations of the Luxoro and Helsingør Atlases in any detail. However, such an exercise would probably have been superfluous since, once the portolan chart hydrography had arrived at its mature form around 1340, no significant alterations were made afterwards (to my knowledge). 5
The strange, and perhaps unique, way of arranging the network of compass lines looks identical at first sight on the two works but close inspection shows minor variations with regard, for example, to the relationship between a toponym and the nearby compass lines. This is not surprising because there is no necessary connection between the hydrography and the compass network during the drafting process. The arrangement of the lines is purely mathematical and hence forms a rigid network, whereas the coastal outlines - themselves part of a fixed system - could be moved en bloc by altering the sheet boundaries in relation to the chart content. The main difference in the arrangement of the two atlases is that the Luxoro combined the Atlantic coasts into a single sheet whereas the Helsingør Atlas split that coverage into two separate sheets at northern Spain. That explains why the Luxoro has four sheets against the Helsingør's five. Overall, while the sheet boundaries are broadly the same on the two atlases, there are slight differences in the placements of the outer borders.
In other words, the relationship between the compass lines and the hydrography is the result of nothing more than drafting decisions. When it comes to the toponymy, on the other hand, the differences between the two atlases are both noticeable and highly significant. The place-names are one of the most important elements of a chart, at least for dating purposes, and the place where most of the significant changes on portolan charts occurred. Toponymy, across the full 1430-70 period, will be discussed in some detail later because it offers pointers to dating but, at this stage, a test was carried out to see if there was any repetition of unusual Luxoro names in the Helsingør Atlas.
Ten randomly selected place-names, widely-spaced geographically, and previously recorded as being unusually present on the Luxoro Atlas and/or the signed Cesanis chart of 1421, were looked for on the Helsingør Atlas. Eight were certainly absent and two, if present, were in an unrecognisable form. Furthermore, since none of a cluster of six 'Cesanis' names in Apulia appeared in the Helsingør Atlas either, it can be safe to assume that its chartmaker obtained his toponyms from one or more alternative sources. This was only to be expected because, as will be demonstrated later, the Helsingør Atlas incorporated a considerable number of the place-names that were first noted on dated charts in the period 1430-62. 6
The link to the Luxoro Atlas is a major leap forward but that did not mean that the investigation had been concluded. On the contrary, the revelation of what had inspired the Helsingør Atlas, almost certainly several decades afterwards, leaves unknown the identity of the draftsman, where he worked on the atlas, and when it was made - precisely the information we are seeking. Furthermore, since the Luxoro Atlas restricted itself entirely to information focused on navigation, it leaves us to look for the incidences of various additional features found on the Helsingør Atlas but not generally present on other charts. It was hoped that a search across all the accessible portolan works of the relevant period might throw up sufficient repetitions to highlight one or more individual productions. That is, as they say, after digressions on the compass-lines network and miniature atlases.
The most obvious characteristic of a portolan atlas or chart is the network of lines radiating out from a central point to the edges of the page (for an atlas) or the animal skin (a loose chart). This can be easily recognised by two sensed (rather than actually drawn) circles: first, a wider one, joining up the sixteen intersection points, and then another closer in. There were variations in the treatment but none of the atlases that it is was possible to examine for this exercise has the arrangement seen on the Luxoro and Helsingør Atlases, where, instead of there being a single central point, lines radiate from the middle of two of the sides. Note that, because most of the images of this atlas in the Medea-Chart Database are of half sheets, the other half needs to be envisaged to appreciate the full system. [PP 9] contrasts the networks in the Helsingør Atlas and Benincasa's first atlas (1463). Simple compass networks radiating out from the centre can be seen on two other miniature atlases: one by Fiorino (1462) and the other an anonymous atlas, c.1460-70. 7
Perhaps the most remarkable feature of the Helsingør Atlas is its diminutive size. Just how small is it? Well [PP 2] shows that the volume is no larger than a human hand. For comparisons between that and other smaller atlases, both Ramon Pujades and the Medea Chart Database include the dimensions of most but not all the atlases. Pujades also provides comparative measurements for the length between a few paired places from which the charts' scale, as distinct from its size, could be deduced. 8
Using those aids, a concerted effort was made to gather measurement data for comparison with other smaller atlases. In assessing the figures that follow, a number of provisos need to be taken into account. In some instances there were contradictions, in others, ambiguous statements or none at all. Some of the figures were probably supplied by the holding institutions, not necessarily working in the same way. Furthermore, some numbers might refer either to the volume's binding or to the sheets within it, and it is not always clear whether the measurement was of a folded or opened-out chart.
A particular problem concerned the Luxoro Atlas, whose size Pujades (DVD: A.14) had given as 16 x 23 cm [roughly 6¼ x 9 in]. However, those proved to be the measurements of a facsimile which had been enlarged by 44%. The true figure was given as 112 x 157 mm [roughly 4 ¼ x 6 in]. 9 The measurements for the Helsingør Atlas (kindly provided by Thorbjørn Thaarup at the Helsingør Maritime Museum), are 109 (width) x 154 (height) mm [roughly 4 ¼ x 6 in], and, for the charts once opened, 214 mm [roughly 8 ½ in]. The very small differences between those figures (just 3 mm) is not enough to weaken the contention that the Helsingør Atlas is a direct copy of a Luxoxo-style atlas and drawn at the same scale. 10
The miniature portolan atlas, signed by Nicolò Fiorino in 1462, comprises eight charts spread across two sheets. At first viewing it is likely to be incomprehensible, even though, the separate charts are all oriented to the north (or perhaps south) with one exception (Spain, which points east). However the images are not arranged in the usual geographical order. To assist anyone wishing to consult it on the Medea Chart Database (/atlas/41) working from left to right on the two numbered sheets, the sequence is:
The Fiorino atlas does not have obvious similarities with the Helsingør Atlas (indeed as shown in Table 4) it included only one of the eight 'unusual thematic features' found on the latter. But it is undoubtedly one of the smallest surviving atlases, even if it is not clear what the figure of 14 cm, cited below, refers to. Because of its unusual arrangement it would not be easy to compare the sheet divisions with those of the Helsingør Atlas, while the treatment of the compass-line network is as rudimentary as possible, comprising no more than a 32-line rose radiating from a central point. This atlas deserves more attention, alongside the ex-Kraus 11 and Luxoro atlases, with which it has been compared.
The best available figures for the other undersized atlases are set out in Table 1. Miniature atlases (with the smaller figure given first even if that might not be the volume's width). The rounded-up figures in centimetres come from the DVD accompanying Pujades (2007) which were repeated in the Medea Chart Database. So, with the Luxoro and Helsingør figures restated to ease comparison, this seems to be the overall picture.
(Leading to the separate Microsoft Word page)
On the basis of the above, there seems to be no reason that the Luxoro and Helsingør Atlases should not be jointly considered the smallest surviving portolan atlases in the period up to 1500, and that by a large margin. However, the size information came from the DVD that accompanied Ramon Pujades's 2007 work, and that stops at 1469. I am not aware of similar data for the period after that, and there might well be later examples of miniature atlases. 12 Miniaturisation must have presented major problems when the toponyms came to be added. It is therefore surprising that the great majority of the habitual names do seem to be present on the Helsingør Atlas, though that was an impression rather than the result of a systematic analysis. That the single most challenging area on the portolan chart, Croatia's Istrian Peninsula, proved impossible to populate with all of its usual toponyms is hardly surprising, though the fact that the pared-down selection of names on the Luxoro and Helsingør Atlases is almost identical reinforces their closeness.
Because such small portolan atlases seem to be bijou objects, presumably designed for non-maritime use and fitting neatly on a normal bookshelf, they might have been expected to survive in greater numbers. That they did not, makes it likely that very few were produced.
By the 15th century, there was a range of conventions in use on portolan charts. Some were very obvious and widely repeated, others, which might be considered trifling, could still be valuable indicators of authorship if they were not regularly included. Taking those features in turn, the following comments start with elements that affect the whole portolan chart or atlas, then moving to specific ones.
What immediately catches the eye - in atlases but not in single charts - are the occasional triangles in one or more corners. In a few cases, for instance the atlases of Pietro Vesconte (1318) and Giacomo Ziroldi/Giroldi (1426), and including the Helsingør Atlas, the corners have been used for carefully-drawn colourful designs. However, a pair of diagonal lines on the inward side of the triangle was later used to enclose a scale bar instead, as again on the Helsingør Atlas. [See PP 12]. The triangular scale bar has been seen on seven other instances. But those divide between, first, those triangles formed of three lines of different lengths and, second, where the lines meeting at the corner are of equal length (isosceles). My attention was drawn to this important distinction by Richard Pflederer. Some caveats need to be stated. At least one instance was noted where two paired triangles were of clearly different sizes. Since only one triangle was measured for each work, it is possible that an atlas could have harboured triangles of the other type as well. Nor were the pair of lines along the outer border, when they have been described as being of equal length, necessarily precisely so. They may well have been drawn by eye, not measured. Another diagnostic tool is the distinction between those chartmakers who used dots for the mile divisions and those who preferred a capital 'I' with serifs. Only numbers 101 and 161 in the 'Sort date' column of Table 2. The mathematical structure of the corner triangles chose the second form. In relation to the search for the likely date of the Helsingør Atlas, Table 2 shows that only two instances are supposedly prior to Benincasa's first atlas of 1463, whose treatment of that feature came to be a readily recognised hallmark. In 2011, I had stated "that it was Benincasa who introduced the diagonal scales".
13
That may still be true because there are few, if any, datable features on portolan charts supposedly from the mid-15th century (at least known to me). The expression, '1425-50' (or more broadly 'second quarter of the 15th century'), was assigned by Ramon Pujades in his outstanding work of 2007, as a broad-brush estimate. I can see no compelling reason that that precludes some of those in that category actually belonging to the Benincasa period. However, there is a further uncertainty because, while the isosceles triangles on the atlas now in Venice align with Benincasa and the Helsingør Atlas, the Vatican example does not, as Table demonstrates. It is unsurprising that that no Catalan chartmakers feature in this list, given that they concentrated almost entirely on separate charts rather than atlases. Not only are the small number of works that include this feature Italian but most, if not all, are Venetian.
(Leading to the separate Microsoft Word page)
Most portolan-atlas sheets enclosed the chart itself within ruled black-ink lines. This resulted in a range of patterns, whether a set number of equidistant lines, two pairs (divided by empty space or scale bars), or perhaps with added colour or ornamentation. The Helsingør Atlas uses the simplest form, just three equidistant lines (though in places there appear to be four, which might be caused by visual confusion from the following sheet). The Luxoro Atlas, on whose structure the Helsingør Atlas was based, had simple border outlines as well but used the vertical edges between the half sheets to house the scale bars that would instead be placed in the Helsingør's corner triangles. As was usual, both of those atlases gave the Adriatic and Black Sea their own separate sheets - or in this case - half-sheets. That meant that in the Luxoro Atlas the bars measuring out the scales for those respective regions - significantly enlarged for the Adriatic and slightly reduced for the Black Sea - run down alongside the bars marking the atlas's general scale - thus making those differences of scale obvious.
The routine 'house-keeping' nature of this feature means that drawing the lines would almost certainly have been a standard procedure. The variations the in border style has not been analysed in detail here but should certainly be considered as part of a work's signature.
Another feature of general relevance is the manner in which a few large rivers were denoted by means of a thick line, nearly always in blue, extending inland. [ PP 13] Two relevant points can be made. First, although it is not a particularly significant feature, this formulaic addition must have been present on the source for the Helsingør Atlas and, second, because, as can be seen in Table 4. 'Significant and unusual thematic features on the Helsingør Atlas found on the portolan atlases and charts of the mid-15th century' chose the second form.
Where some toponyms appear to run up carelessly against the outer margin and were thus truncated (for example along the Egyptian coast and the north shore of the Black Sea) [PP 15] this should perhaps be interpreted instead as the addition of a thin band of tape.
From close observation, Thorbjørn Thaarup, a curator at the Maritime Museum, observed that "The border of the chart pages are definitely made of an inner line (drawn directly on the chart) and two outer lines (drawn on a thin band of tape). There seem, however, to be instances of a second drawn line added directly on the chart, where the tape was too thin". Besides covering up parts of some toponyms, this also caused the origin-point of the compass-line system to be obscured, e.g., south of Barcelona (on Sheet 3). Turning to the lines bordering the corner scales (black for the first two sheets, and red for the last three) those cross the black lines and are then blocked by the tape, which cuts off everything. But then, in a few places, the lettering, evidently in the same hand, extends over the tape.
If that sequence has been read correctly we seem to be left with contradictory evidence: either the tape was applied by the person who created the atlas and, on realising the truncations, tried to remedy them, or the occasional extension of a toponym over the tape was done by somebody who had managed to obtain the necessary details. Neither seems very plausible.
This is the most significant of the specific elements that have been noticed on a limited number of charts and atlases, all of them Italian. It concerns the treatment of Ireland's holy lake, 'Lacus Fortunatus'. This 'lake' (i.e., a large bay, as it is clearly shown) - perhaps Donegal or Galway - is let into the west coast of Ireland, and colourfully dotted with islands. It had been supplied with an accompanying legend on portolan charts since the time that Ireland itself was first included there in the early 14th century. [ PP 14] Pietro Vesconte's original legend recurs on a small number of 15th-century works, usually with additional wording. The crucial element was the number of islands credited to the mythical archipelago that fills the bay: whether the short, cryptic form of the note with 358 Irish islands, used by Pietro Vesconte from about 1321 onwards and also favoured by some of the unsigned atlases from the 15th century, or a longer version with either 367 or 368 islands. As shown in Table 3. 'The Lacus Fortunatus inscription', most of the charts that included the 'Lacus Fortunatus' label are undated and there is no obvious way to place them in date order. It is worth noting that the legend was not seen on any dated works between Beccari (1435) and Benincasa (1461).
The wording on the Helsingør Atlas is similar to both the Beccari and Benincasa versions but has variant phrasing of its own. That none of the longer statements are precisely the same may be explained by the fact that the chart scribes, who sometimes had to fit the statement around the pre-entered coastal names - and who were competent to handle contractions and use a siglum for, e.g. 'que' or 'ubi' - might have altered the wording in order to fit the space. While it is disappointing that no common 'Lacus Fortunatus' wording has been found, apart from in the work of Benincasa and his followers, the lack of a match for the Helsingør Atlas's version is perhaps unsurprising. The crucial point is that the Helsingør Atlas, by following the generally accepted 368 total, was not copying Benincasa's 367, which became the norm on his numerous productions and those of his followers and imitators. 14
(Leading to the separate Microsoft Word page)
Dundalk and Dublin bays are the most prominent along Ireland's east coast, with Dundalk being significantly larger. Unlike Beccari (1435) and Benincasa (1461), the Helsingør Atlas enlarges Dundalk Bay noticeably, as can be seen in PP 16. Although, in general, the coastal outlines on the Helsingør Atlas mirror those on Francesco Cesanis's unsigned Luxoro Atlas, this did not apply to the east coast of Ireland, where the Cesanis version of Dundalk Bay had been more open and even larger. 15 As can be seen in Table 4. Significant and unusual thematic features on the Helsingør Atlas found on other portolan charts and atlases, only six other instances were noted (ignoring subsequent works by any single chartmaker).
The Isle of Man was usually portrayed as a simple X. PP 17 contrasts the expected shape with the Helsingør Atlas's distorted three-pointed outline. Perhaps this is a mistake he could not readily correct. No other example of that form was observed although the 1421 Cesanis chart has a design with three, not four, semi-circular incisions, while the Corbitis Atlas (c.1400: Medea /atlas/144) has a very irregular shape.
As the chronological window within which the Helsingør Atlas might have been made included the period of Grazioso Benincasa's known activity (1461-82) it will be necessary, later on, to consider whether his work provided, in some way, a model for the Helsingør draftsman. But, for now, it is appropriate to consider his treatment of some of the islands around the British Isles.
At first glance, the Helsingør Atlas's version looks very similar to that seen on the Benincasa atlases. That impression is strengthened by the heavy outlining of the coast, distinguishing England & Wales, Scotland and Ireland in different colours. This addition presumably followed the procedure for some of the larger islands in the Mediterranean but it makes the British Isles stand out visually on their own. Benincasa was consistent in his atlases in tinting Ireland green, as does the Helsingør Atlas, but Benincasa varied the outline colouring for England and Scotland, sometimes making the same choice as the Helsingør Atlas.
Around the British coasts can be seen prominent stylised islands, imaginary or misunderstood, named Till, Scurce, Brazil, and Man (south-west of Ireland). These were included in the 'Colour & Shapes Analysis' (numbers 1, 2, 9 and 10), a new version of which, comparing the Helsingør Atlas with Batista Beccari and Benincasa, is included below, see Table 7. Comparing styles seen on the Helsingør Atlas and Benincasa's early productions, with those on previous Italian charts . 16 In the case of Till and Scurce the name is written across the island on Benincasa's charts but the Helsingør Atlas omits their names. [PP 17 ] So, on the basis of their respective handling of the British Isles, the Helsingør Atlas, while being possibly coeval with the Benincasa period, was not a slavish copy of his work.
Large, simplified city views or diagrams are sometimes included on portolan charts. Genoa and Venice are the most frequent main subjects, though neither appears on the Helsingør Atlas. However, there are clear signs that a Venice vignette was probably intended. The reasoning is as follows. There is a natural gap in the toponymy where Venice would have been named (but wasn't), one caused by the abrupt change of direction of the coast at that point. At the top of P 21 there is a comparison between the treatment of the vicinity of Venice on the Luxoro and Helsingør atlases. Note how the Luxoro Atlas, which does not have a vignette, leaves a modest gap at that point, whereas the angle in the case of the Helsingør Atlas is one of well above 90°. The four scans underneath in PP 21, with their tributes to Venice, show how a similar image could have fitted neatly in that vacant space. This suggests that a representation of Venice would almost certainly have been present on the source that was copied for the Helsingør Atlas. Furthermore no equivalent gap on the Helsingør Atlas was left for Genoa, which is merely named in the toponymic sequence in the usual way. Which seems to dispose of the possibility that the Helsingør Atlas could have been copied from a Genoese model.
It is hard to see how the draftsman involved could have intentionally omitted what was arguably the most important toponym on the chart. There would seem to be a number of possible explanations for the absence of both Venice's toponym and its vignette: carelessness; a lack of artistic competence; the atlas's incompleteness; or antipathy towards Venice. To take those possibilities in turn. Instances of his carelessness are attested later and would seem to be the most likely explanation.
17
As to the possibility that, as a draftsman, he did not feel qualified to draw an image of Venice, there is no reason for that needing to be as elaborate as some of those illustrated in PP 21, since very simple versions were also available. The possible explanation that the atlas was incomplete can be readily disposed of. Since the atlas's binding is both expensive and evidently original, who would have knowingly paid for an unfinished product? The final suggestion, that a vignette was intentionally omitted would surely mean that, not only was the Helsingør Atlas not a Venetian production, but also that its compiler was from a rival port, perhaps Genoa, or perhaps, that the omission was requested by the original purchaser.
18
Table 4. Significant and unusual thematic features identified on the Helsingør Atlas seen in other portolan charts and atlases
considers various styles of Venice vignettes. Out of nine examples (ignoring repeats by a single chartmaker), only two of them featured Venice and nowhere else, and two others went to the other extreme with multiple vignettes. Most of the instances pairing Venice with Genoa were on Catalan charts. Since it was unusual for Venetian charts to include illustrated details, the absence of an image of "La Serenissima" from a Venetian chart is not unexpected. However, since only one of the Venetian vignettes (on a chart now in Barcelona) is likely to be Venetian, raises the remote possibility that the Helsingør Atlas might be a Catalan production.
19
A striking feature of the Helsingør Atlas is the way that it treats the outline of the coast between Ravenna and Trieste by straightening the curve. Indeed, the hockey-stick effect, with its bend at Venice looks as if the two lines had almost been drawn with a ruler. PP 22. As shown in Table 4. Significant and unusual thematic features on the Helsingør Atlas found on other portolan charts and atlases,
just three instances besides the Helsingør Atlas were noted, broadly similar but not so mathematically precise. Not surprisingly, one of those was the Luxoro Atlas, which has already been identified as the model for the Helsingør Atlas's coastal outlines. The other two are undated but considered to belong to the second quarter of the 15th century. Rather than being a hydrographical improvement this simplification runs counter to the fundamental purpose of a portolan chart, which was to document the coastal outlines as accurately as the relatively small scale allowed, and preserve those by precise copying. C.2f. The North coastline of the Adriatic Sea
C.2g. The Black Sea moved west
Approximately one-third of the Black Sea extends further to the east than the Levant coastline. In order to gain some space, while also allowing the Mediterranean scale to be increased, some chartmakers detached the Black Sea and moved it west so as to align its east coast with the Levant coastline. Because most atlases gave the Black Sea its own separate sheet anyway, it might have been expected that this space-saving adjustment would have applied to charts and not atlases. However, Table 4. Significant and unusual thematic features on the Helsingør Atlas found on other portolan charts and atlases contradicts that with nine of the ten instances being atlases. Nevertheless, the earliest noted instance of this device can be seen in a chart by Albertin de Virga (1409 - Medea /chart/438, see Campbell 1987, p. 444) and another exception is the chart made by the Luxoro Atlas's creator, Cesanis in 1421. Besides the Luxoro Atlas itself, three of the other atlases are undated and placed by Ramon Pujades in the period 1425-50. One is thought to belong to the 1460s, and the rest are reliably dated to that decade. The Helsingør Atlas was one of those that moved the Black Sea but, because the Medea version presents the two halves of the relevant sheet separately it may not be immediately apparent (see PP 23)
Until the mid-1460s, portolan charts covered no more of West Africa than the coastline south as far as Cabo Bojador (buyetder, 26.08 North, in Western Sahara), as well as the Canary Islands. However there was one exception, namely a chart focused on the Atlantic, arranged in portrait rather than the habitual landscape shape, drawn by Andrea Bianco in 1448 (Medea /chart/89). This format allowed the coverage to be extended south as far as Cape Vert (14.43 N). Since this seems not to have been repeated - and Bojador was as far south as the space on a typical skin allowed - that remained the normal terminus. However, whereas a chart was constrained by space, there was no obvious reason for the long delay in adding one or more sheets to a portolan atlas. This occurred first in 1465, when Grazioso Benincasa bolted on an extra sheet, thus extending European knowledge as far as Cabo Roxo (12.2 N). Since the Portuguese did not publicise these developments themselves, Benincasa would become the cartographic chronicler of that steady southwards push. With further sections of the West African coast and the nearby islands having being revealed, Benincasa was able to add a further sheet to his atlases in 1468.
20
The above has direct relevance for the Helsingør Atlas because the extension beyond Bojador is the hall-mark of the atlases that imitated Benincasa, post-1465. However, for some reason Helsingør's creator arranged his first and second sheets - containing the North Atlantic and West Africa - in such a way that his most southerly African toponym falls perhaps as much as ten names short even of Bojador, the charts' standard terminus since the 14th-century. [PP 24 ] That is directly related to the strange lowering of the British Isles on the first sheet of the Helsingør Atlas, taking with it the most northerly name on mainland Europe, which, as usual, aligned with Scotland's north coast. It is hard to see any rational reason for this sacrifice of valuable space. To remove the possibility that one or two further sheets extending down West Africa might have existed originally but later been removed, Thorbjørn Thaarup at the Danish Maritime Museum examined the atlas and could find no evidence that it is not complete. Perhaps the curtailment of West Africa can best be explained by the lack of concern among Mediterranean sailors about coasts that only Portuguese ships were likely to visit. As late as 1462, Fiorino included the African coast no further than just short of Salé (34.02 N) and hence well above Bojador.
21
Certainly Benincasa himself was casual about the West Africa toponymy, often stopping a few names before his ultimate terminus, Cauo de Sancta Maria, and, with respect to the Cape Verde Islands, sometimes failing to make space for all the islands.
22 So, with the Helsingør Atlas not even reaching Bojador, let alone continuing beyond that, we have been deprived of a potential dating aid. However, a comparison between the Cesanis chart of 1421, his Luxoro Atlas, and the Helsingør Atlas reveals that the Luxoro Atlas stops at Salé (34.02 N ), the Helsingør Atlas continues two names beyond samotamat (itself 12 names after Essaouira = Mogador, 31-51) whereas the 1421 chart has a different sequence of five toponyms after samotamat. This demonstrates, that the Helsingør Atlas could not have taken from the Luxoro Atlas the toponyms and related coastal outlines beyond Salé.
(Leading to the separate Microsoft Word page)
All chartmaking involved close copying of the coastal outlines at least, and that imitation would often extend to other elements as well. Beyond that, and leaving to one side the charts' toponymy, there could be a wide range of additional features, treated in a variety of ways. That previous section of the investigation looked to see which of those, out of a list of 13 potentially unusual features found on the Helsingør Atlas, could be seen elsewhere.
The eight relevant features which emerged during a wide-ranging comparison between the Helsingør Atlas and other portolan charts, fall into three categories: those concerning the coastal outlines (Dundalk and the north Adriatic), a space-saving device (moving the Black Sea), and what might be described as stylistic choices (the remaining five). No doubt other relevant aspects may have been overlooked. Table 4. Significant and unusual thematic features on the Helsingør Atlas found on other portolan charts and atlases is a subset of the much longer list of the portolan charts and atlases that have been consulted for this essay. It comprises only those that had included one or more of the eight features, and it runs to 26 works.
If there was an expectation that at least one portolan chart or atlas would replicate a majority of those elements, and hence provide a plausible date and place of origin for that atlas, it was a resounding failure. Reading down the right-hand column of Table 4, with its totals of the number of times the eight aspects seen on the Helsingør Atlas were matched elsewhere, none of the other works displayed more than three. Turning to the foot of the table, across that list of 26 charts and atlases, one feature was noted no more than three times, and the strange shape for the Isle of Man was not seen anywhere else.
So, while the search continued for the birth details of the Helsingør Atlas, that table can be read in very different way: rather than revealing an expected conformity it unearthed widespread disparity. This pattern will recur when other elements seen on the Helsingør Atlas are investigated later.
Moving on, the focus turns to aspects of the drafting of the Helsingør Atlas - handwriting, spelling, and the colours used - before considering the last major element, the toponymy.
When trying to determine the likely date and place of a portolan atlas or chart, handwriting is one of the obvious features to consider, visible, mostly but not exclusively, in the place-names. Calligraphy is, or could be, the only truly personal element in a portolan chart since everything else would have been copied, whereas the form and position of each of the toponyms were likely to have been committed to memory. Apart from a few examples of a rigid 'house' style, such as that of the Vesconte or Benincasa families, it seems that other portolan-chart scribes were allowed some latitude, within the tight space constraints imposed by the portolan chart. Furthermore, as the possible total of 2,000 names on a single chart might mean well over 10,000 individual letters, it is unlikely that anybody would have wanted the distraction of having to write in an alien hand.
Literacy would not have been required for using a portolan chart but it would certainly be a prerequisite for someone transcribing the toponyms. Although it would have taken time to memorise the sequence and placement of many hundreds of place-names, chartmaking novices would presumable have been made to memorise the full sequence, whether by eye or silent recitation. It is also likely that some would have learnt the scribal side of chartmaking, with its specific house-style lettering, as apprentices.
Because all the portolan works consulted in this exercise contain contraction signs indicating omitted letters - whether for reasons of space, contemporary usage, workshop practice or simply whim [the need is not always clear] - those involved must have had at least some kind of general scribal training. Since knowledge of the conventions was sometimes insufficient to reassemble a much curtailed word, there was presumably an assumption that the user would recognise the name by its position. The handling of those symbols by the Helsingør Atlas's author shows him to have been well-trained.
We are left to guess, though, how portolan-chart draftsmen might have obtained the special skills involved when working on charts as small as those in the Helsingør Atlas. The names on a portolan chart usually needed to be written in a smaller than normal size, which might mean minuscule lettering in the case of a miniature atlas. Maintaining a legible hand in places where the toponyms for one coastline backed up against another must have been particularly demanding. An extreme example of that situation, Croatia's Istrian peninsula, has already been mentioned, where the number of toponyms had to be drastically reduced. Somebody had to decide which names to include and, although in this case most of those were written in red, the choice of black names are not exactly the same on the Luxoro and Helsingør Atlases.
In addition, not only would the geographical sequence of the toponyms have to be memorised, but, when entering the ordinary names in black ink, it is likely that the necessary spaces would have had to be left so that the less frequent, significant red names could be added later.
Once fully trained, portolan scribes did not always hold rigidly to one particular spelling, nor to precisely the same letter forms, although they would doubtless be following their version of the toponym's pronunciation. The repetitive work involved in entering perhaps the same 2000 names, seemingly endlessly, probably explains, for example, why a short 's' might be used in the middle of a word one time and the long form on the next occasion. Furthermore, some charts would have been produced, or perhaps just lettered, by freelance journeymen, who might have been allowed some licence to add the lettering in their own style.
It is not hard to imagine the need for self-expression lying behind the occasional flourish, like the 'q' in the Guadalquivir river (uardaq(u)ibir) next to Seville on sheet 3 of the Helsingør Atlas [see insert in PP 10] - surprising, because the word was not identified as referring to the river.
It is natural to expect that the form of particular letters in the toponyms might help in arranging anonymous (and hence undated) portolan works chronologically. However, in this instance that has not proved to be of much help because in the first half of the 15th century there was a major turning away from the archaic medieval letter forms in favour of more simple, regular rounded styles, as the Renaissance looked back to Classical lettering. This is at least partially true for portolan charts as well, as can be seen by comparing the writing on 14th-century work with that in the period being dealt with here. Indeed, the new letter shapes seem to have found their way onto the charts surprisingly early. As a result, six out of the nine unusual or rare letter forms found in the Helsingør Atlas, and considered here [PP 10] and in Table 5. Distinctive letter forms seen on the minority of charts and atlases which used a capital 'short' (modern) 'S' in their toponyms were already present on the 1403 Francesco Beccari chart (Medea /chart/81). This negates the dating value of this approach, since the Helsingør Atlas must have been drawn several decades after that.
The hand of the draughtsman who drew the Helsingør Atlas has a number of distinctive characteristics. The most striking of those was his treatment of the capital 'S'. The form he used is broadly recognisable to us today but appears strange because of the way the letter has been turned to the left by about 90°. This Helsingør version had not seen anywhere else and so it was decided to group together other instances of left-leaning forms.
'd'. The ascender curls up to the left and then to the right
'e'. This is formed of two separate elements
'G'. This was used in its enlarged capital form when it was the initial letter, and sometimes within the word. It is the only other letter besides the 'S' to be thus treated.
'i'. The dot [or 'tittle'] was originally a larger mark but was reduced to a small dot in the late 15th century. Apart from the Helsingør Atlas, this was noted only on a single, post-1469 atlas.
'r'. The Helsingør Atlas has two different versions, both illustrated in [PP 10]. The most used outline is formed of two short strokes (sometimes separated); the other is close to a modern form with the addition of a horizontal line beneath. Perhaps significantly, the 1439 Catalan chart by Gabriel de Vallseca, which has a broadly similar range of letter forms, uses alternate outlines for the 'r' that nearly match those on the Helsingør Atlas (see for the Vallseca versions the red names either side of Seuta).
'S'. [See also PP 11]. The most noticeable letter is the treatment of the letter 'S', whose traditional long version (like an 'f' without the crossbar) was beginning to be replaced by the 'short' or 'round' form' used today. The use or absence of the modern form is clearly of some significance in itself but the way the letter was drawn and how it was treated provide further tests of similarity or difference. These variations can be divided into three: (1) whether a short capital S is used at all, (2) if it is, how it is aligned and (3) where in the toponym the place-name is seen. To take those in turn:
The difference between the two 'S' styles is so great that there is no possibility of confusion. But the second factor, the orientation of the modern S, is what first suggested that the version in the Helsingør Atlas could be unique. The shape is normal but it is, as it were, oriented to the west not north, looking as if it had fallen over. Once the available charts and atlases had been checked it was clear that a minority of works used the new S form', and, further, that a small number of those had swung the letter a bit to the left. But the first impression proved to be correct: no other portolan production was seen with its S habitually turned to the left by 90° degrees. [PP 11] provides examples of the three variants but it needs to be understood that the angle of lean, between 0°, 45° and 90° , is not precise but on a continuum. Moreover, there can sometimes be slight variations within a single portolan work.
The third way that the chartmaker could make a choice is related to where the round S was placed in the name. In addition, while most were enlarged to provide an initial letter, a smaller version sometimes appears within or at the end of the toponym. As a further variation, the 1421 Cesanis chart and his Luxoro atlas, neither of which generally used the round 'S' form, included at least one example of an un-enlarged S, demonstrating that he had that alternative form available.
That the angle of lean might have increased over time is contradicted by the chronological arrangement of Table 5. Distinctive letter forms seen on the minority of charts and atlases which used a capital 'short' (modern) 'S' in their toponyms
'x'. The lower left arm is extended [this and the following letter form are illustrated as an insert on PP 10]
'z'. This looks like an opened-out number 3
(Leading to the separate Microsoft Word page)
If at this point it had been possible to match the Helsingør Atlas's penmanship with at least one other surviving work, there would have been no need for a lengthy discussion about the likely place and period of its creation. However, that exercise had a limited success. None of the charts and atlases accessible through reproduction consistently matched the Helsingør Atlas letter forms. Indeed, one variant was not found elsewhere at all (the 'e') and another (the dotted 'i') was seen in just a single late atlas. Nor does the fact that the three works with slightly more Helsingør letter forms were Beccari (1403 in Savona, near Genoa), Pareto (1455 in Genoa) and Roselli (1462 in Majorca) provide help with the dating aspect.
Furthermore, that four of the chartmakers cited were working in Palma, Majorca rather than Venice - the assumed birth-place of the Helsingør Atlas - adds to the growing evidence that the interchanges of information between chartmakers must have been informal, personal, piecemeal and frequent. Instead of conformity and shared styles, what shines through is the heterogeneity. This will be corroborated when the attention turns to the charts' toponymy.
From letter forms to spelling is not a huge jump, and a suitable subject for comparison presented itself, appearing relatively frequently and in a number of different variations.
In seeking for clues to the place or date of a portolan chart or atlas, the shape of the individual letters is just one of the methods available. Another approach is to look for spelling that points to a particular dialect. The interchangeable letters, x, s, and ç, are one example of that. If those dialect variations in the portolan-chart toponyms have not yet been codified, that would be a useful aid. In the meantime, a different approach may be helpful. The portolan charts' toponymy is rich in saints' names, but it is unlikely that any were venerated in more places than St George. Martyred in 303, he was chosen as patron saint by many countries and cities, including Genoa.
23
Nineteen instances have been found in my Excel listing but, unsurprisingly, only two responded to a search for 'georg.' Instead, a range of variant forms had to be looked for, preceded by 'san' or more usually 'S'. The purpose of this exercise was to use those differences in an attempt to home in on the likely place or school where the Helsingør Atlas might have been created. The instances noted in the geographically-organised Excel listing can be found under the following numbers. The eight in bold appear on the Helsingør Atlas, while italics identifies four others that were looked for on 19 charts or atlases, selected to include all the known chartmakers of the relevant period. The seven numbers left in Roman were not checked: 373, 648, 797, 831, 852f, 923c, 996d, 1002, 1018, 1057, 1104a, 1150a, 1176, 1198, 1241, 1302b, 1373, 1433a, 1607a The four main dialect forms are sufficiently distinct (if the minor spelling differences are set aside) to show a clear pattern. The occasional choice of an alternative form is probably not significant (perhaps the aberrant version was not even recognised as referring to St George) and it is intriguing that the Helsingør Atlas, having used giorgio consistently, chose zorz. for the instance between landermiti and foya in west Asia Minor.
24
Of more interest perhaps is the general change by Cesanis from the ziorzi of the Luxoro atlas to the çiorçi of his 1421 chart.
(Leading to the separate Microsoft Word page)
So what can be learnt from Table 6? First, that the (B) zorzi form was the most used, variously on charts produced in Venice, Genoa and the Catalan headquarters in Parma, Majorca, from which the four (D) Jordi instances came as well, thus marking out that form as exclusive to Catalan works. The (C) çorçi examples are Italian, possibly all from Venice. Which leaves the (A) giorgio form used by the Helsingør Atlas's draftsman, in concert with Benincasa and two late Italian atlases. Their precise provenance is yet to be determined but they are considered to be 'perhaps Venetian' (see the BnF reports on these). 25
When seeking St George's name in Apulia, between Monopoli and Bari, it became apparent that St George was sometimes superseded by St Vito. Thinking that this might point to the place of production, this information was included as well (see the right-hand column). However, san vito's absence from the Helsingør Atlas did not prove to be significant since the presence or absence of that name occurred equally with Venetian, other Italian and Catalan works - further evidence that when such information was being shared, locality was not necessarily relevant. Once more, the attempt to locate the Helsingør Atlas in a particular school was unsuccessful.
What was discussed previously included some of the more obvious features on the Helsingør Atlas. However, small details that could easily be overlooked can be just as valuable components in a particular chartmaker's 'signature'. In 2011, a detailed inventory was made of 51 features, the 'Colour and Shape Analysis'. This was as extensive as the availability of readable reproductions then allowed. Extracts from three of those separate tables have been combined here in Table 7 as a single spreadsheet populated by the atlases and charts most relevant for the present purpose:
N.B. For the three Microsoft Word tables that follow you need to click on 'Enable Editing' to enlarge the image:
(Leading to the separate Microsoft Word page)
Prompted by similarities between the Helsingør Atlas and Benincasa's 1461 chart, the central purpose of Table 7 was to see how the handling of the selected features on those agreed or differed. To help with the somewhat dense and codified information, the instances where the Helsingør Atlas and Benincasa concur are highlighted in blue in the left column. In three instances, red is used to indicate further that no other instance of that particular form had been found elsewhere. Although only seventeen of the 51 rows showed a full match, several others were similar, perhaps having the same shape but giving it a different colour. The clear impression is that the Helsingør Atlas and the 1461 Benincasa chart are closer to one another than to any of the others, although there is some overlap with the 1435 Beccari chart (whose column was moved next to the Helsingør Atlas to ease comparison). 26
At this point it may be helpful to stand back from the detailed analyses of the distinctive characteristics of the Helsingør Atlas when compared to the portolan charts and atlases of the period 1435-1480, by summing up the sequence of previous Tables and the commentary on them. We start with a review of the methodology pursued.
Keeping the Helsingør Atlas as the focus at all times, the process has involved three stages. First, obtaining reproductions of as many portolan works as possible from the relevant period (those are listed in
Table 12. CHARTS AND ATLASES RELIABLY DATED FROM 1430 UP TO THE 1460S OR POSSIBLY MADE DURING THAT PERIOD .
The second phase involved identifying the Helsingør Atlas's idiosyncrasies, some of which were immediately obvious, whereas others emerged only later. In the third phase, each of the works in that long list of charts and atlases was examined for instances of the features that had been identified for their potential rarity. As a result, the majority of the charts and atlases involved have not been referred to in the body of this essay because they did not throw light on the Helsingør Atlas's hallmark characteristics. The overall aim was to record all the matchings of those 'signature' features found on the Helsingør Atlas, from either direction: whether by that atlas's author or by others copying him.
In an attempt to sum up and make sense of the findings so far, another table has been devised. Table 8 is a precis of Tables 2-6, with their information whittled down so as to isolate - from the full list of the unusual conventions found on the Helsingør Atlas - those instances where that particular feature had been noted no more than five times. This rarity test removed some relatively common features, such as the blue rivers and the move to the west of the Black Sea. On the basis that if there was a model hiding among the portolan works examined, or, alternately, a work that had clearly been copying the Helsingør Atlas's noteworthy aspects, that would naturally rise to the top in this exercise. In particular there was a hope that it might help to bring us nearer to the identity of the Helsingør Atlas's author, and the place where it could have been made.
(Leading to the separate Microsoft Word page)
In reality, as Table 8 demonstrates, the result was the opposite of what might have been expected. Fourteen charts or atlases did indeed match some of the Helsingør Atlas's idiosyncrasies. However, reading down the right-hand column, it is immediately evident that all bar two of the works involved included no more than a pair of the Helsingør Atlas's ten 'signature' features. The exceptions, each with five matches, are two late works: Benincasa, in his first atlas (1463), and a 'pseudo-Benincasa' atlas of 1469+ in the Bibliotheque nationale de France (Medea /atlas/185). (See note 25). For various reasons, neither could have been used as a model by the Helsingør Atlas's draftsman.
The significance of Table 8 lies, not in its failure to identify any charts or atlases that shared a significant number of the Helsingør Atlas's unusual features, but rather for revealing that those idiosyncrasies were indeed distributed relatively widely, perhaps among as many as 25 chartmakers. But only one or two of the features each time.
A few introductory comments about the charts' toponymy might be helpful since a major focus of portolan-chart studies has been the sequence of place-names running around the edge of the entire Mediterranean, along with its islands, and extending to the Black Sea. In contrast to the broadly unchanging coastal outlines, the toponymy was dynamic. Names were being continually introduced or abandoned, with others confined to the chartmakers in a single port or workshop, some recurring erratically, and so on.
It is likely that most, if not all, non-standard or novel place-names will have owed their inclusion in a portolan chart to the result of a private discussion between a pilot and a chartmaker. What other route can be proposed? I am not aware of any written text - which would inevitably be in manuscript in the period we are dealing with - that might have prompted a toponym's inclusion. That does not necessarily mean that all additional names must have had navigational or mercantile relevance, but political significance does not seem to have been important. 27
Even if the chartmaker was (or had been) a mariner or sea-going merchant himself- and that was certainly true of some Venetian practitioners 28 they could hardly have gathered all that information themselves. So it is not surprising that out of around 120 names seemingly first appearing on charts dated between 1430 and 1460, the 43 of those assigned to Andrea Bianco are the only ones reliably attributable to a sea-going chartmaker.
Once a name had been incorporated by a chartmaker it could be copied, as an addition or alternative, or ignored. How quickly this happened or how widely it spread was left to chance. Some toponyms were speedily and generally adopted; others, for no apparent reason, were broadly, or even totally, ignored, leaving us with few, or no, further sightings. Nor, given the random way that toponyms migrated, does the absence of a name necessarily have any significance.
So, when seeking to date a chart or atlas, or to identify its place of production on the basis of its toponymy alone, the first noted appearance on a dated work will often be a useful guide to its first inclusion in a portolan chart. But not always. In this context the fact that the Helsingør Atlas includes 15 out of 30 of the names first noted on works dated 1461 and 1462 does not entirely preclude it having been produced earlier than that. When talking, for example, about "1462 Fiorino" names, it should not be taken to mean that he must have been responsible for introducing the toponym in question to a portolan chart, and on that date. The charts we have access to must represent a minute proportion of what was produced. Furthermore, of the 15 names with attested dates of 1461 and 1462 included in the Helsingør Atlas it transpires that only four have not been seen on unsigned works possibly produced earlier. 29
The tool that has been deployed here is my Excel spreadsheet, a comprehensive geographically-arranged listing (of mainland names only) whose numbers in the list are cited throughout for identification purposes. 30
(Leading to the separate Microsoft Word page)
Two of the columns (F & G) record the first time that the toponym had been noted as appearing in a datedwork, paired with the name of the chartmaker involved. A selection of those was extracted for the purposes of this analysis and can be seen in Table 9.
The information in that table can be sorted to display those introduced names in different ways: in a geographical sequence (Columns A & B), by the date of introduction (F), or in an alphabetical list of the chartmakers responsible for first using that name (G). The purpose of that exercise was to identify, first, which of those introduced toponyms can be found in the Helsingør Atlas, second, which chartmaker was responsible, and, third, when that name had first appeared in a reliably-dated work. Had there been a full match between the Helsingør Atlas and a specific chartmaker's toponymic introductions in that 45-year period, it would have identified a probable source. However, that did not happen, as can be seen from the irregular distribution of the blue 'yes' in the Helsingør Atlas column of Table 9 (whose significance is explained below).
Having tested a small sample of the 150+ names first recorded in the period 1435-80, from the dozen chartmakers involved, it became evident that the Helsingør Atlas was including small clusters of names first seen on the work of Benincasa (1461-80) and Nicolò Fiorino (1462). The full list of the introductions by those two chartmakers is included in Table 9, along with up to three randomly-chosen names from other chartmakers.
An important general caveat is needed. The main purpose of the Excel spreadsheet, which has evolved out of a more restricted investigation in the 1980s, was to pair each toponym with the date of its first appearance on a dated chart. While not entirely ignored, unsigned charts (which nearly always omitted the date as well) could not be included in a chronologically-organised survey. However, over time, notes were added when the occurrence of a name on an undated work evidently pre-dated its earliest confirmed appearance. The notes in the 'Comment' column of Tables 9 & 10 refer to those instances along with others that emerged as a result of the present exercise. This is of considerable relevance because there are a number of undated charts or atlases, nearly all of apparently Venetian origin, which, like the Helsingør Atlas, might be dated either before or after Benincasa's 1461 chart.
What factual evidence can we learn from the Table 9 before attempting to use that information for unravelling the relationship between the Helsingør Atlas and those portolan charts included for comparison in this investigation? In attempting to interpret that table, two things need to be kept in mind. First, that little can be learned from the occasional outlier instances, which are far from unusual in portolan-chart toponymy, and, second, that while two of the names that had been first noticed on the 1435 Beccari chart appear on the Helsingør Atlas, it is probably more significant that the other seven were omitted. In addition, the Helsingør Atlas includes one name each from Bianco (1448) and Pareto (1455), but none of the eleven introductions on Antonio Pelechan's fragmentary 1459 chart of the Adriatic (Medea /chart/388). None of names picked up by the Helsingør Atlas was from a Catalan chartmaker.
This leaves the main focus on Benincasa and the less well-known Venetian, Nicolò Fiorino. Despite the cluster of names whose introduction has been confidently dated to Benincasa's first work (the 1461 chart) half are absent from the Helsingør Atlas: nine names are included but 10 are not. For Fiorino, the figures are six inclusions and four omissions. However, what is more significant is that a number of the supposedly 'Benincasa' or 'Fiorino' names had actually appeared on undated Italian charts that probably or certainly pre-date 1461-62 (as noted in the 'Comment' column on Table 9). This reduced the number of names in the Helsingør Atlas which might indeed have appeared for the first time in 1461 or 1462 from 15 to eight.
In case any of those eight instances might be found in unsigned Venetian atlases, whose toponyms had not previously been examined in detail, a further trawl was undertaken. As a result, two similar atlases considered to belong to the second quarter of the 15th century [Parma (II, 32, 1624), Medea /atlas/32 and Vatican (Vat. Lat. 9615), Medea /atlas/24], proved to include four of the names in that group of eight. That left a mere four names on the Helsingør Atlas which might have genuinely appeared on a portolan chart for the first time in 1461 or 1462. [For those, see the blue 'yes' in the right column of Table 9]. Since a number of the potentially relevant atlases are not yet available in legible reproduction, earlier instances for those four places - pietra santa (Excel no. 499), pig(n)ea (723), castel nouo (848) and montanea (1338) - may well emerge in time.
When toponyms are treated in the simplified manner of Table 9 - i.e. that the name is, in some recognisable form, either present or not - the data lends itself to statistical analysis. However, if the Helsingør Atlas had different versions of those toponyms central to the dating issue, that should probably rule out direct copying from the dated work in question.
To provide clarity on that subordinate point - the precise form of the name - Table 10 compares the spelling of the 19 'yes' names on the Helsingør Atlas with that of their first noted instance on a dated work.
(Leading to the separate Microsoft Word page)
Considering the wide range of variations in the names on portolan charts in general, sometimes morphing into two or more forms - or even, when deviations were too great, having to be treated as two distinct places - the close matches shown on Table 10 between the forms in the Helsingør Atlas and those in the work of the chartmaker first found to have used that name, is striking. Part of the explanation for that convergence was presumably that the five chartmakers involved were all Italian, since there was a large and sustained gulf in the selection and form of the toponyms between the Italian chartmakers and their Catalan rivals in Majorca. Of the names first recorded for the period 1435-80 only 19% were on Catalan charts, and none of the introduced names seen on the Helsingør Atlas had a Catalan origin.
(Leading to the separate Microsoft Word page)
The first, and most important observation is that, out of the list of names that were first noted on dated charts between 1463 and 1480, not a single one appears in the Helsingør Atlas (See Table 9. 'Names first noted on dated charts or atlases in the period 1435-80'). The overall similarity between the toponyms on the Helsingør Atlas and those that were seemingly first encountered in the period 1435-62 on the charts and atlases of Italian chartmakers working in different ports, points to closer interconnections than had been previously thought. Taking into account the small number of charts emerging from Genoa [which at one time in the mid-15th century had only a single active chartmaker] 31 that turns the spotlight on Venice. Since Venetian work was often left unsigned and undated, and was sometimes of uncertain origin, major problems arise when assigning primacy. Which of the unknown chartmakers was copying from whom? The unexpected finding that the Helsingør Atlas was sharing parts (but not all) of different chartmakers' toponymic selections suggests that individual choice may have been involved rather than just unthinking imitation. The appropriation of toponyms from a wide variety of sources by the Helsingør Atlas's draughtsman (or his model) reinforces the findings throughout this essay of unexpected heterogeneity.
Furthermore, the manner that the Helsingør Atlas incorporated toponyms that were first noted on dated works during the relevant period, 1435-80, was very selective. Only a small proportion of the names from a few different chartmakers was used at all.
When seeking to determine who might have been responsible for a portolan production and where they were based, a small detail can sometimes be crucial. For the most part, chartmaking ateliers were in the business of replicating their in-house model as closely as possible. The unquestioning imitation that could be involved as a result meant that what might seem to be an utterly trivial detail can be significant. It is doubtful, for example, that the apprentice knew why he was detaching the first letter of Aigues-Mortes and putting in on a nearby island. But that, like many other, easily overlooked, conventions can help to narrow down a possible provenance for a portolan work. The 'Colour and Shape Analysis' (Table 7) provides a wide range of very small comparison points, which reveal the house-style of an individual chartmaker or the school involved. But no overall pattern matching that of the Helsingør Atlas was revealed. Indeed, the closest pairing was with the work of Benincasa, who can otherwise be ruled out as a major source for that atlas.
Before attempting to pull things together, including suggesting a date for the Helsingør Atlas, a few necessary provisos need to be taken into account. First, it has to be appreciated that portolan-chart historians have their hands tied behind their backs. The theoretical tools we construct to track the interrelationships between chartmakers depends on scrutinising the miniscule - and probably unrepresentative - proportion of their work that has survived. Even though a full-time, lone practitioner could possibly make 20 or more charts a year, amounting perhaps to several hundred over a working lifetime, some have left us no more than their name.
32
Furthermore, as far as the present exercise is concerned, a sizeable number of the relevant extant works are unsigned and thus also undated. This means they are less likely to have been chosen for reproduction, and, as a result, could not be examined. F.1b. Other potential dating evidence
The question arises: are there other aids for the dating of unsigned charts than the toponymic and visual evidence that has already been considered here? Well, there are some, but their value is limited. Very occasionally a feature on a portolan chart is independently datable by reference to a historically-verified event, thereby providing a proven terminus post quem for that work. The flags placed alongside prominent ports are one obvious example of such a potentially datable feature, and newly-named places another. Although some Italian charts of the mid-15th century do include flags, that does not apply to the Helsingør Atlas, and anyway such evidence is often out-of-date or misleading. As for newly founded or (re)-named places, recognition tended to be very belated. The first appearance on a portolan chart of the small number of independently datable toponyms includes two within the period covered by this analysis: Ville Franche (founded in 1295, but not noted on a chart until 1455) and Castel Nuovo = Herceg Novi (created in 1382 and seen first on a dated work by Nicolò Fiorino in 1462). The Helsingør Atlas is the only one, among all the charts and atlases that have been investigated here, that includes both those places; Benincasa never named Castel Nuovo.
33 Another possible aid to dating is the handwriting. Although palaeologists can often date a hand to a particular period, this is more difficult for portolan charts, whose clustered toponyms demanded lettering that was both unusually small and also clearly legible. Furthermore, the way that handwriting seems to have often been included in chartmaking apprenticeships tended to preserve and standardise a particular style. It had been hoped that idiosyncrasies in letter forms would help in dating the Helsingør Atlas but
Table 5. 'Distinctive letter forms seen on the minority of charts and atlases which used a capital 'short' (modern) 's' in their toponyms'
roundly disproved any link between letter-forms and chronology, at least for the period involved. Lacking external references that could have helped date the Helsingør Atlas, we might have turned instead to visual comparison of the coastal outlines, perhaps for guidance on whether that or the 1461 Benincasa chart preceded the other. However, the hydrography of the portolan charts remained broadly unchanged for centuries, after evolving up to around 1340. That is because copying the outlines was not done freehand but by tracing, or by some other direct transfer method which would ensure faithful replication.
34
So, not only is it difficult to describe hydrographic variations in words but those would probably have been no more than minor and very gradual.
Comparisons of one undated work with others may have to focus on the presence or absence of a single feature - and perhaps on its specific treatment - in an attempt to place two or more works in a chronological sequence. The natural assumption that any work which included an apparently unusual item would therefore be later than the one that lacked it might well be true, but omissions and revivals (sometimes decades later) are too frequent for a dating estimate to be based on that alone.
It is not disputed that a large majority of the portolan charts and atlases in the second and third quarters of the 15th century are likely to have been drawn in Venice, rather than in Genoa or Palma, Majorca. Indeed, that origin is often corroborated by the Venetian spellings of the toponyms. Venice's history is richly documented in their extensive archives and, as a result, the biographies of individual chartmakers have been carefully extracted from those records.
35 However, since the kitting out of the regular trading convoys was closely supervised, and given the primary importance of shipping and navigation to Venice, it might have been expected that practical assistance would have been provided to chartmakers. But that does not seem to have happened.
36<>
Perhaps there was no need for a helping hand, since the significant number of charts and atlases assumed to have been produced in Venice, nearly all unsigned and in different hands, testifies to a thriving market of independent chartmakers. Even if some unsigned works have been correctly described as Italian and then tacitly assumed to be Venetian there is always a possibility that they were actually produced elsewhere than the three main centres: Venice, Genoa and Majorca. We know of a few examples: Francesco Beccari signed his 1403 chart from Savona (west along the coast from Genoa), Bertran & Ripoll's 1456 chart was made in Barcelona, and the 1459 Pelechan chart in Crete. So, though unlikely, it is possible that some of the unsigned portolan charts might have been produced in some other port. In the light of that, how confident can we be that the Helsingør Atlas was drawn in Venice? The dialect forms in the toponyms - even assuming that those were his own rather than those of the model he was using - do no more than point, in general, to shared Italian conventions, as borne out in
Table 10. 'Comparing the spelling of the toponyms introduced in the period 1435-62 and their forms on those included in the Helsingør Atlas'. Although there must have been a vignette of that city on the exemplar being used
37
it is possible that the reason for the omission has been misinterpreted. Might the Helsingør Atlas's draftsman have had an antipathy towards Venice? Of one thing we can be sure. The Helsingør Atlas is not a Catalan production, even if their chartmakers used some letter forms that are close to those seen on the Helsingør Atlas..
See Table 5. 'Distinctive letter forms seen on the minority of charts and atlases which used a capital 'short' (modern) 'S' in their toponyms'.
Five Catalan chartmakers were active in our period but none of the toponyms they seem to have introduced are noted on the Helsingør Atlas. Nor were they responsible for adding any of the eight unusual features.
38 F.1c. Venetian origin?
Batista Beccari in 1435 and Grazioso Benincasa from 1461 act as a pair of book-ends in terms of influencing the content and style of 15th century portolan-chartmaking in Italy. However, neither could have acted as a direct source for the Helsingør Atlas.
Benincasa was a native of Ancona who first came to attention as the author of a detailed written portolano of the Adriatic, Aegean and Black seas (1435-45), compiled when he was a ship's captain. He turned to chartmaking in 1461 and, thereafter, left us many works, mostly atlases, in the period up to 1482.
When the Medea images of the Helsingør Atlas were first announced, Benincasa's name was suggested as that work's inspirer because of two very visible features: first, the triangular scale bars [PP 12] and (C. 1a) and, second, the 'Lacus Fortunatus' note in Ireland [PP 14] and (C. 2a). Since an unusual number of Benincasa's works have survived, and he was to be succeeded by a sequence of disciples who kept his distinctive style alive for over a century, it was not unreasonable to assume that those signature elements had been introduced by him. However, after the closer examination documented in Table 4 it became clear that neither of those features, nor indeed any of the other unusual elements noticed on the Helsingør Atlas, had been introduced by Benincasa.
The British Isles has more noticeable conventions than any other area on a portolan chart: the Lacus Fortunatus and its legend, the striking calligraphy for the names of the component countries, different shading for their coastlines, and disks for hypothetical island groupings. PP 17 makes clear that Benincasa's handling of those features was closely copied from a Beccari chart like that of 1435. Theoretically, the Beccari outlines and trade-mark elements could have continued to circulate after his death and the closure of his atelier by 1438 39 even if no surviving chart estimated to have been produced in that 25-year period reproduced those. Hence a crucial question for placing the Helsingør Atlas chronologically, is how and when its compiler might have obtained the Beccari version of the British Isles and other less obvious features, given that those would not have come via Benincasa. The triangular corner-pieces, for example, can be seen (with their isosceles lines) on an unsigned chart, supposedly from the second quarter of the 15th century (Venice, Bib. Naz. Marciana, It.IV, 493 (5077), Medea /atlas/31). [This feature was restricted to atlases and hence could not have appeared on the Beccari chart.]
On the other hand, when it comes to toponymy, Benincasa does have to be taken into account. Table 9. 'Names first noted on dated charts or atlases in the period 1435-80' demonstrated that his first chart of 1461 has slightly more of the nine unusual names that are considered there than any of the other charts and atlases consulted. Furthermore, a significant number of the names in the 1461 chart, which had seemingly not been recorded previously, were included in the Helsingør Atlas. Those supposed toponymic innovations are set out in Table 9, but the subsequent commentary explained that the majority were probably not new but rather copied from undated works, likely to precede 1461. 40 The crucial point to consider is that Benincasa is assumed to have come new to chartmaking, aged about fifty, only after losing his ship to pirates in 1460. 41 He then produced his first surviving chart the following year in Genoa, and would sign the first of his many extant atlases in Venice in 1463.
Benincasa would obviously have been well versed in using a portolan chart, but learning the various skills required to draw one would have been another matter, since there was no manual. 42 The most likely person in Genoa to have tutored Benincasa would seem to have been the priest-cartographer, Bartolomeo Pareto, known for a chart of 1455. Pareto died in or before 1464 so he could theoretically have assisted Benincasa in 1461 with the drafting of that first surviving chart. 43 However, even though Benincasa included the two names first observed on Pareto's chart of 1455 (Table 9), there is nothing in the statistical data in the other Tables to show that he was imitating Pareto. So, another dead end.
Much use has been made in this essay of toponymic evidence as a potential aid to dating. But common-sense needs to be considered as well. How could Benincasa not only produce a professional-looking chart just a year after abandoning a life at sea, and simultaneously manage to include eighteen previously unknown toponyms, spread around the eastern half of the Mediterranean? From his own experience alone? Can the addition of those, covering 13 different areas (see Table 9), be put down to him being an unusually observant mariner? How would he have acquired that information? Contrast that with what followed later, when he would add to successive atlases no more than fifteen toponyms on eight separate occasions over twenty years. Both before and after 1461, those names were introduced in twos and threes, and spread widely. That is a pattern which fits in with what we see elsewhere, when a returning mariner would pass on information in person to a chartmaker. Nor is it likely that the post-1461 names were from Benincasa's own observations because, apart from his many well-documented moves between Italian ports, it is hard to see how he could have returned to being a mariner, collecting toponymic data.
Nor is there evidence for the portolani or any hypothetical written source supplying place-names for use on the charts. 44 Since, perhaps half of the toponyms not recorded on a dated work until 1461 can still be seen on undated charts and atlases considered by Ramon Pujades to belong to the second quarter of the 15th century, it is more likely that Benincasa would have found most, if not all of them, on lost chart(s) from which he must have been copying. If this interpretation were to be generally accepted, that the 18 'new' toponyms credited to Benincasa were actually in circulation before 1461, and if the same logic were applied to the eight names first observed on Fiorino's charts in 1462, of which half can be seen on prior undated works, that would lessen or remove any necessity to date the Helsingør Atlas after 1462.
The clarity of Benincasa's works shows that he had thought carefully about their professional appearance. He would also master the mathematics of issuing atlases at four different scales, at least. Yet the strict discipline he observed in his drafting means that the overall appearance of the 1463 atlas was very similar to his final one dated 1480 (Medea /atlas/157). Leaving aside his willingness to embrace the West Africa discoveries, with separate updates to both the coastal outlines and the Cape Verde Islands, it doesn't seem that any new features can be assigned to him. Thus, Benincasa should be characterised as less an innovator than a successful imitator.
There is no doubt that the compiler must have been an experienced scribe. This is clear in the way he used contractions, for example for Saint and -que in the 'Lacus Fortunatus' label (C.2a). However, he did have shortcomings.
Some significant problems were caused by overlaps and, even more so, gaps, which meant the outlines were incomplete and some toponyms omitted. These will be looked at in turn, with their Excel numbers cited for geographical reference.
No doubt, larger than necessary overlaps between successive sheets can be seen on other portolan atlases, but there are a couple on the Helsingør Atlas which deserve notice.
186 cormes - 215 bogno. Northwest Iberia (Sheets 1 & 2). Overlap of 15 names. [PP 19]
It was usual for there to be a small amount of overlap between adjacent sheets - indeed it would have been strange and unhelpful if there hadn't been - but the overlap for the northwest coast of Iberia between the first two sheets is excessive. Nevertheless, it provides an opportunity to see the Helsingør Atlas's compiler at work by comparing the two versions. We would hardly have expected to find differences on the relatively short stretch between Cape Finisterre and Porto(gallo). But leaving aside the interchanges between 'y' and 'i' - a trifling example of spelling that would typically range freely so long as pronunciation was respected - it is surprising to see three omissions (a different category of error) among just 15 or so names. One of them, Pontevedra, had been in the red ink that marked its prominence.
873 lavelona - 942 coron. Southeast Adriatic and west Greece (Sheets 4 & 6). Overlap of approximately 30 expected names. PP 20.
The overlap here was somewhat different. Although involving 30 toponyms, which would seem to iindicate some unnecessary work, the only difference between the respective name sequences was that Sheet 4 adds a single toponym not seen on Sheet 6. Even if this is to be considered an omission - and the only one - neither that name, nor the shared one above, is included in my Excel listing. Neither toponym is fully legible but this suggests that the Helsingør Atlas's names might be worth transcribing in full. Furthermore, the consistency between those two sheets suggest that the errors of the Northwest Iberia overlap were not typical.
1383 lesmire (Sheet 6) - 1392 altologo (Sheet 7). West Asia Minor. Perhaps nine names missing, none in red - this is the gap that shows up on the montage as the white strip. This gap is not present on the Luxoro Atlas, confirming that it should be put down to an error on the Helsingør Atlas or its immediate model.
1354 paradiso (Sheet 6) - 1359 dardanello (Sheet 8). Sea of Marmara. Perhaps five names missing. Again, this error is not seen on the Luxoro Atlas.
Compared to overlaps, gaps have a very different effect even if they share a similar origin, namely the faulty selection of material for neighbouring sheets. Overlaps are the result of generosity, gaps of miserliness. But both effects come about when a chartmaker - unless he is closely following a model - chooses from a total chart what would serve as the outer edges of his selections. In this case the Helsingør Atlas's author copied very closely the sheet contents of the Luxoro Atlas in some cases but not all. [PP 7 & 8] illustrate that point for the Aegean and East Mediterranean sheets respectively but, as just described, making your own sheet selection can lead to problems. Gaps are more serious because they introduce breaks in the portolan charts' otherwise continuous coastline. To add to the confusion, the points where the breaks start or end are disguised on the Helsingør Atlas, whereas others, including Benincasa, signal an interruption in the coastline by showing a clear separation.
In other instances the Helsingør Atlas's creator opened himself to criticism for a range of mistakes. The following examples of his inattention were noted, not ruling out others that might appear with a more thorough investigation.
It has long been appreciated that certain features on a portolan chart or atlas were formal conventions, widely used and for a long period. Others were restricted to a particular chartmaker or perhaps a wider group/school. It was those in the latter category that have diagnostic value, as they can sometimes help with authorship and dating. Examples that were not used in this investigation - because they do not appear on the Helsingør Atlas - are the manner in which the largest estuaries were sometimes shown as a group of multi-coloured islands, or a feature, perhaps restricted to Giacomo G/Ziroldi, of rivers snaking in from the sea to run into an ornamental terminus. Conventions of that kind represent the most visible part of a chart's 'signature'. We are grateful, therefore that the Helsingør Atlas seems to include more of those than any other portolan chart or atlas from this period.
As it is those conventions - along with the toponymy - that has provided this essay with most of its evidence, it may be useful to have a quick summary.
To recap, this is a list of the Helsingør Atlas's unusual features (so far identified):
Table 4. ('Significant and unusual thematic features on the Helsingør Atlas found on other portolan charts and atlases') deals with eight abnormal or rare features in the Helsingør Atlas and their occurrence in the work of about 25 practitioners. As already explained, only four of those chartmakers included as many as three of the eight characteristics that have been focused on. Furthermore, those 25 chartmakers represent just one third of those on the overall list of charts and atlases being investigated (Table 12 ('Charts and atlases reliably dated from 1430 up to the 1460s or possibly made during that period') as being the only ones to include any of the identified features at all. It would be natural to expect that, when two or three of the characteristics are seen on an individual work, the same pattern would be repeated several times. Had that happened it could at least have indicated partially-shared origins. There are hints of this in Table 4 but not enough to overshadow the fact that not one of the 75 or so charts and atlases considered in this exercise included even half of those eight idiosyncrasies found on the Helsingør Atlas.
It is not surprising that the Catalan chartmakers, working closely together in Palma, Majorca, tended to share conventions of the kind we are dealing with here and that is partially borne out in this investigation. But the number of their productions - all of them charts, not atlases - is far less than that for their Italian counterparts, and they adopted fewer of the Helsingør hallmarks. It would seem that Italian chartmakers, and more specifically the numerous Venetian ones, must have been accorded choice about the style and content of their charts. Instead of offering the restaurant equivalent of a set meal, the Italian chartmakers provided a varied a la carte, although the self-disciplined Benincasa is a notable contradiction to that.
Indeed, this in-depth unveiling of the Helsingør Atlas cautions against over-reliance on the concept of schools or ateliers working as isolated entities. There are certainly several examples of that. But the Helsingør instance highlights a very different pattern in which borrowing of one chartmaker from another - most likely from seeing a rival's production - was both widespread and highly selective, sometimes reviving features or toponyms from an earlier period. Considered together, these tendencies can be characterised as those of individualistic plagiarists. In the case of 15th-century Venetian practitioners, this contradicts any assumption of a steady development that could help with dating.
We have come to the point where placing an estimated date on the Helsingør Atlas cannot be avoided any longer. However, when trying to arrive at a date in this case there is little hard evidence available. No reference to a relevant event from the outside world has been noted and little can be learned from the dated works in the period leading up to the 1460s, which are mostly Catalan productions. Two revivals distort the picture: the first involves the Helsingør Atlas itself with its author's borrowing of the building blocks - that is the miniature format, coastlines and compass network - from the Luxoro Atlas of perhaps 40 years earlier. The second throwback was Benincasa's disinterment of Beccari's 'Lacus Fortunatus' note 25 years after it had, apparently, been last seen. But the biggest, and most imponderable, question remains: how did the Helsingør Atlas's draftsman gather up as many as eight rare conventions (as described in Table 4. ('Significant and unusual thematic features on the Helsingør Atlas found on other portolan charts and atlases') and yet no other chartmaker included more than three of them? And none of those peculiarities were new since all were present by 1450 (apart from the uniquely distorted symbol for the Isle of Man [PP 17]).
The Helsingør Atlas overlaps with Benincasa's work in some respects, and more so than with other chartmakers. But several dissimilarities rule out the possibility that either could have provided a model for the other or, probably seen the other's work. Which leaves the question of where each of them might have obtained the features and toponyms which lie at the heart of this essay, and are crucial for the Helsingør Atlas's dating.
If, as has been assumed here, Benincasa was a novice chartmaker, working in Genoa at the time of his first portolan chart, he must have had a model to work from (or perhaps more than one). The Helsingør Atlas's creator, on the other hand, had a ready-made model in the Luxoro atlas, as far as the format and coastal outlines were concerned, but he must have had to seek out one or more patterns for those unusual features discussed earlier, along with toponymy from many sources. Put together, the various Tables do not reveal what model Benincasa might have found in Genoa, neither do they point to a source for the Helsingør Atlas's peculiarities.
In a nutshell, who was copying whom? Was the Helsingør Atlas's author responsible for gathering up those conventions, from a variety of Italian sources, for other chartmakers to then copy them, even if no more than one or two at a time? That could have pushed back the Helsingør Atlas's date. However, since none of the studied works could have copied the Helsingør Atlas, it seems more appropriate to place that close in date to, but unconnected with, Benincasa's entry on the chartmaking stage in 1461.
So, as regards the need for a plausible date for the Helsingør Atlas, it can first be confidently stated that there are no indications that it contains anything datable after 1462 [unless that should be pushed a year later by its inclusion of the corner triangles that Benincasa may or may not have introduced]. Likewise there is no evidence to suggest that it should be pushed back into an earlier period. Indeed, the chronologically-ordered Tables show more matches between the Helsingør Atlas and the charts and atlases being compared with it in the later period.
For a year or a date-range to be proposed requires a choice between two competing scenarios. The first - easier than the other and in line with traditional opinion - would simply cite the unchallenged fact that the Helsingør Atlas includes four place-names first noted on the dated works of 1461 (Benincasa) and 1462 (Fiorino). On that basis the Helsingør Atlas would have to be dated after 1462. The alternative interpretation, which I favour, refutes that argument with one that moves attention away from the four toponyms for which there is no recorded antecedent, focusing instead on the fact that, of the twenty names first confirmed on the work of Benincasa and Fiorino, exactly half have documented appearances on evidently earlier works lacking a stated date (Table 9. 'Names first noted on dated charts or atlases in the period 1435-80' (). The crux of the matter is that if those four Helsingør names are to be held to their attested dates (1461 & 1462) then the atlas has to be dated 1462+, whereas an interpretation that considers it likely that the novice chartmaker, Benincasa, must have obtained the toponymic information from one of the numerous, now lost, charts and atlases, permits a slightly earlier dating.
Clamping a definite year onto the Helsingør Atlas is both difficult and reckless but, in line with the second of those two alternatives, a suggested date of c.1460 seems a reasonable compromise.
This essay started out with the aim of dating the Helsingør Atlas and establishing who might have drawn it, or at least the likely chartmaking centre involved. That the scope has spread in many directions was a response to the inconclusive results of a succession of tightly-defined investigations. It was fortuitous that at the half-way point an accidental sighting of the Luxoro Atlas proved to be a eureka moment. Produced by Francesco Cesanis around 1420, this was clearly the example of a model from which the Helsingør Atlas's creator had traced the basic information. The reason that the Luxoro Atlas had not been considered at that point was because the period considered relevant for this exercise had started in 1430, with the main focus about 30 years later. That revelation provided a full answer to a number of questions. The miniature format of the Luxoro Atlas, the coastal outlines and the way they related to the individual sheets, as well as the evidently unique compass-line system, all were imitated, closely or broadly, by the Helsingør Atlas. Had that connection been missed, this essay would have been an embarrassment. Instead, the remaining challenge was (only?) to identify the author (perhaps via his handwriting), the atlas's date, by comparing its toponymy with other charts and atlases of the period, and finding other instances of about a dozen unusual features.
As each line of investigation - however interesting - failed to throw clear light on the two major concerns, the atlas's authorship and date, others methods had to be devised. Even if some of those seems trivial, for example the spelling of St George's name or the angle of lean of the capital letter S, each provided useful insights. In many cases what stood out from the various comparisons was not the expected patterns and similarities from the borrowings of one chartmaker from another, but rather the randomness of individual choices.
What we now know of the Helsingør Atlas can be boiled down to little more than this. Its creator, or one or more intermediaries over the period c.1420-c.1460, had access to the Luxoro Atlas or, more likely, a copy of it. They then updated its toponymy, made minor adjustments to the coastlines, and added the unusual features discussed in this essay. That those conventions are found only sparingly on other portolan charts and atlases, and then only a few at a time, remains the most unexpected and interesting aspect of the Helsingør Atlas's content.
Some questions remain unresolved. Why, as an evidently experienced chartmaker, was the Helsingør Atlas's draftsman sufficiently careless as to omit a stretch of the coastline (which shows up clearly in the composite Medea image: atlas/712), as just one example of a number of mistakes. How should we interpret his inclusion of those rarely-seen features, since all of them can be seen on unsigned works tentatively datable prior to 1450? If there was any originality in what he was doing it rested on no more than the fact that there is no known precedent for the way he collected together those signature features. We can however state with confidence that no single chartmaker or school whose work has survived could, on the one hand, have acted as a model that could have supplied all the Helsingør Atlas's unexpected features nor, on the other hand, have themselves been modelled on that.
The thwarted attempt to fit the Helsingør Atlas into the context of the middle years of the 15th century by comparing it with reliably dated charts and atlases meant that the less-studied unsigned works had to be examined as well. Although this exercise in turn failed to find more than a few of the sought-after connections, there was a positive side-effect: a better understanding of the way that Italian chartmaking operated, particularly in Venice. Working in the same city, and probably close to one another near the harbour close to the area frequented by sailors, mutual borrowing of information between the chartmakers would have been expected. However, the evidence points to something rather different. Instead of physical proximity leading to shared conventions, and perhaps even some collaboration, like that of the Catalans in Majorca, the Venetian chartmakers seem to have made their own choices of features and toponyms, by borrowing (or plagiarising) from other portolan charts rather than through personal contact with their fellow practitioners.
Selling their works unsigned and undated, these chartmakers evidently worked alone rather than in a group atelier following a common pattern. The 2007 DVD accompanying Ramon Pujades's comprehensive survey of the period up to 1469 notes the few occasions when two works were evidently in the same hand [see the blue 'cf' notes in Table 12]. Most of those Venetian practitioners would have been supplying their unadorned charts and atlases to working pilots. Not unnaturally, they would have had little or no interest in the Helsingør Atlas's unusual features, since those were irrelevant for navigation. Only one of those, the enlargement of Dundalk Bay which made it the most prominent feature along Ireland's east coast, could have been of any aid to pilots.
The Luxoro Atlas was the smallest portolan atlas that has been identified [at least for the period up to 1500] and, since the Helsingør Atlas's imitation was clearly traced from that model and thus must have the same dimensions, it should share that accolade. Whatever the Helsingør Atlas's intended function, the lavish, and presumably expensive, binding makes it improbable that it was designed to be used at sea.
Despite many hours spent trying to understand it, this atlas retains some secrets, the biggest one being the sources its creator used. Apparently neither influential, nor itself seemingly influenced, the Helsingør Atlas has managed to cram into its miniature frame an unprecedented number of unanswered questions. Having studied and written about medieval portolan charts for over forty years I have never before attempted an in-depth analysis of a specific chart or atlas. I hope others will agree that such an effort was justifiable in this case.
With happy timing, as this essay was nearing completion new evidence emerged about the route the Helsingør Atlas had taken before it arrived at the Maritime Museum in 1948. Thorbjørn Thaarup found a series of letters, in Danish, which revealed that the antiquarian booksellers in Copenhagen, Branners Bibliofile Antikvariat, and specifically their representative, Hans Götz, had handled the Atlas's sale to the Museum. He had negotiated a lower price from "the bookshop abroad" that was offering it, down to 4,500 dkr. A generous supporter, C.K. Hansen, paid that sum and then donated the Atlas to the Museum. As was not unusual, the foreign booksellers did not want to divulge their identity or details about the atlas's provenance.
That simple account leaves two questions: who might that non-Danish bookseller have been and how did the Atlas manage to be attributed to the 16th-century chartmaker Battista Agnese.
It is no doubt mere coincidence but the first of those four letters in the Museum's archive (September-November 1948) was dated 20 September, which happened to be the same month that saw the foundation conference of the International League of Antiquarian Booksellers [ILAB or LILA], held, moreover, in Copenhagen. 45 A contemporary, chatty, account of that meeting by the wife of one of the English attendees mentioned that Hans Götz "who had haunted the Amsterdam conference [there had been two meetings prior to 1948] was still being cold-shouldered by the Danish association", but had been "suave and smiling" at the conference soirée hosted by Branners. 46 In addition we learn, perhaps unexpectedly, that Götz was a member of the ABA ([UK] Antiquarian Booksellers Association) but had played that down so as to avoid embarrassing the three other British members.
Götz seems also to have taken care not to reveal anything about the bookseller who then owned the Atlas. But it is likely that it would have been based in Western Europe, if not in Italy. Götz's ABA membership points to a close connection with the British antiquarian booksellers. And in the first letter (20 September) there is a reference to him consulting an English bookseller's catalogue and having acquired a 17th century portolan chart for the Museum a few years previously (Medea /chart/6700). Might the foreign bookseller have been English?
Götz also plays a central role in the false Agnese identification. [It needs to be pointed out that although the Museum mentioned Agnese it was with a question-mark]. 47 The bookseller with whom Götz was working seems to have been the first to make a definite attribution to Battista Agnese, a prolific Venetian producer of relatively small atlases in the middle years of the 16th century. In his own words, Götz was the Branners' map expert, who had worked with maps and charts for 28 years, and he was adamant the Atlas was by Agnese. He make that point in the first of those letters and elaborated on it later. By 8 October he was citing various not particularly relevant references. But it was only a month later (11 November) that he mentioned the essential work on Agnese, Henry R. Wagner's, 'The Manuscript Atlases of Battista Agnese', in the Papers of the American Bibliographical Society, vol. 25 (1931). Part of that letter is in English and is an extract from a letter to Götz from Wagner. Having admitted, "I cannot see anything from which to date it as my system of dating Agnese atlases is based on configuration and the names of the West Coast of North America", Wagner nevertheless judged it to have been produced about 1540 or 1541 - or much earlier or much later. He ends: "Like you, I also think it was done by Agnese. It certainly looks like his atlas". It is hard to understand how the specialist in Agnese could have overlooked the many differences between the Helsingør Atlas and the distinctive style of that chartmaker. He says he sawthe atlas but that was probably via photographs, perhaps of uncertain quality, and 17 years had elapsed since his publication. He had no doubt moved on to other projects. Having been born in 1862, he was 86 in 1948. [I owe much of the above to Thorbjørn Thaarup at the Danish Maritime Museum.]
Owners, along with booksellers and auction houses, when faced with a portolan chart or atlas without a signature tend to want to push back the suggested date. The older it is, the greater the commercial value. It is therefore unusual that the Helsingør Atlas was placed in the middle of the 16th century, a hundred years too late. This can perhaps be explained by another issue, that of authorship. If a chartmaker is known or asserted, that can validate its authenticity, while simultaneously carrying with it an approximate date. Agnese's atlases, often relatively small, were highly popular. The Pflederer Census records over 100 surviving atlases, albeit three quarters of those are unsigned they can be readily identified because of Agnese's distinctive style.
It was not unnatural when faced with a very small, elegantly presented, unsigned atlas to have immediately thought of Agnese. We should not criticise them for that mistake, since they had none of those aids that are now freely available to us in electronic form and that attribution was endorsed by the accepted Agnese expert.
1. On which see Campbell & Barritt, 'The Representation of Navigational Hazards: the Development of Toponymy and Symbology on Portolan Charts from the 13th Century onwards'. Half a dozen names of dangers were added to charts in the relevant period but none appears on the Helsingør Atlas (as can be seen in the Navigation Dangers Summary Table.
2. On Cesanis see 'Anonymous works and the question of their attribution to individual chartmakers or to their supposed workshops'; and Tony Campbell, 'Portolan Charts from the Late Thirteenth Century to 1500', in: J.B. Harley & David Woodward (eds), The History of Cartography, Volume 1 (University of Chicago Press, 1987), pp. 371-463, (see 402-3) [available online Here].
4. On which see Sima Krtalic, 'Anchoring the Image of the Sea: Copying Coastlines on Manuscript Nautical Charts from the Late Middle Ages and the Early Modern Period', Imago Mundi, 74:1 (2022): 1-30.) The Luxoro images are used courtesy of the Medea Chart Database (/atlas/168). Those lack clarity compared with the reproductions in Pujades (2007) pp. 216-7, and the accompanying DVD (A.14). However those cannot be reproduced here.
5. That said, the treatment on the Helsingør Atlas of Dundalk Bay, the Isle of Man and the north Adriatic coast [PP 16, 17 & 22], do show unusual coastal shapes, but those were not widely imitated.
6. The word 'dated' is significant because the majority of charts and atlases in that period are neither signed nor dated. Guessing the year or period is sometimes unavoidable but dependence on such estimates should be avoided. When it comes to the summing up of a range of different elements in an attempt to arrive at a plausible dating for the Helsingør Atlas it was only ever likely to point to a period rather than a year.
8. Ramon Pujades, Les cartes portolanes: la representació medieval d'una mar solcada. [In Catalan and Spanish, with English text 'Portolan charts: the medieval representation of a ploughed sea', pp. 401-526]. (Barcelona: Institut Cartogràfic de Catalunya; Institut d'Estudis Catalans; Institut Europeu de la Mediterrània; Lunwerg, 2007), see pp. 204-9. , along with the English version of the explanatory text, pp. 477-8. Now available online: Here.
9. See Portolan Census, No. 81.
10. See B.1. 'The Luxoro Atlas as a pro forma for the Helsingør Atlas'.
12. Indeed, it seems that some of the mid-15th century atlases by Battista Agnese may be of comparable size, as a small sampling of the 100+ on the Medea Chart Database threw up one instance of 14 x 20 cm (/atlas/59).
13. On this in general see 'Diagonal corner scales'.
14. On this see also 'Lacus Fortunatus', and Barbara Freitag. Hy Brasil: the metamorphosis of an island from cartographic error to Celtic Elysium. Textxet. Studies in Comparative Literature 69 (Amsterdam & New York: Rodopi, 2013, pp. 7-8, 36.
15. This feature can be made out on the 1421 Cesanis chart in the DVD in Pujades 2007, C. 32, but not the blurred Medea version (/chart/471). The Luxoro Atlas is damaged at that point.
16. 'The features chosen to illustrate the developing portrayal of islands and estuaries on portolan charts up to 1469' (with links to comments), illustrated in the (1st scan).
17. See F.2b. 'What can we learn about the draftsman of the Helsingør Atlas and his methods?'.
18. On doubts about the initial assumption that the Helsingør Atlas was a Venetian production - like its model, the Luxoro Atlas - see F.1c. 'Venetian origin?'.
19. This thought will recur later with the handwriting observations, see Table 5. 'Distinctive letter forms seen on the minority of charts and atlases using a capital 'short' (modern) S in their toponyms'.
20. On this see West African coast.
21. That this curtailment occurs with the Helsingør Atlas as well is presumably no more than a coincidence that they both ran out of space at the same point. However, Battista Agnese, in the middle years of the 16th century, consistently chose a terminal point on the relevant Old World sheet, similar to that on the Helsingør Atlas.
22. See Cape Verde Islands.
23. Alberto Capacci. La toponomastica nella cartografia nautica di tipo medieval (Genoa: Università degli Studi di Genova, Centro Interdipartimentale di Studi Geografici Colombiani, 1994), VII-XII lists about 60 examples from 15 portolan charts and atlases across the long history of the charts.
24. No. 1373 in the Excel listing - the name is truncated by the outer border.
25. Italien 1698 and Italien 1710 .
26. The Beccari chart's omission of everything east of the entrance to the Aegean partially reduces its relevance to this exercise although most of the diagnostic features were included.
27.. On this, in the context of the Carte Pisane, see D.3. Historical evidence
28. Piero Falchetta, 'Marinai, mercanti, cartografi, pittori. Ricerche sulla cartografia nautica a Venezia (sec. XIV-XV)', Ateneo Veneto 182 (1995):7-109.
29. Unfortunately for our purposes, Ramon Pujades's monolithic 2024 publication - which provides full transcription of all the names from 19 mostly 14th century works, but extending to Beccari (1403) and the Catalan Estense mappamundi (1463), pp.552-726 - does not cover the period with which we are concerned. However, for the comprehensive toponymy of two regions, the coastlines of Catalonia and the northern half of the Adriatic up to 1469, see Ramon Pujades, (2007), pp.356-73 & 392-3.
30. 'Listing and analysis of portolan chart toponyms along the continuous coastline from Dunkirk to Mogador (early 14th to late 17th century) including the transcribed names from the 'Liber de existencia riveriarum' and 'Lo compasso de navegare' as well as the Carte Pisane and Cortona chart'
31. See Pujades, 2024 p. 365 / 387.
32. See Pujades, 2007 pp.486-7.
33. Carte Pisane Specific Names Tables. B. Toponymic time-lag (from physical creation to recognition by mariners). [NB. For the Microsoft Word tables you need to click on 'Enable Editing' to enlarge the image.]
34. On which see Sima Krtalic, 'Anchoring the Image of the Sea: Copying Coastlines on Manuscript Nautical Charts from the Late Middle Ages and the Early Modern Period', Imago Mundi, 74:1 (2022): 1-30.
35. Piero Falchetta, 'Marinai, mercanti, cartografi, pittori. Ricerche sulla cartografia nautica a Venezia (sec. XIV-XV)', Ateneo Veneto 182 (1995):7-109.
36 See Corradino Astengo,'The Renaissance chart tradition in the Mediterranean', in: David Woodward (ed.) The History of Cartography. Volume 3. Cartography in the European Renaissance (Chicago & London: University of Chicago Press, 2007), pp.212-3. "From the middle of the fifteenth century onward, the Republic had instituted a number of magistratures or public offices whose authorities made wide use of maps in the exercise of territorial control. However, there never seems to have been a state workshop for the production of nautical charts, nor did Venice adopt the Genoese system of relying on a private workshop subject to state control."
37. See C.2e. 'The (missing) Venice vignette'.
38. On Venetian chartmaking see further: 'Venetian Practice'
39. Pujades 2024 p. 361 / 383.
40. Some of those earlier sightings only came about through a trawl of the less-studied undated works. Since a number of those were not available for study, there is a reasonable likelihood that further precursory instances will surface in the future.
41. Piero Falchetta, 'Marinai, mercanti, cartografi, pittori. Ricerche sulla cartografia nautica a Venezia (sec. XIV-XV)', Ateneo Veneto 182 (1995):7-109 at pp. 54-60.
42. See for a check-list of the competencies required: Workshops, and for technical reconstructions Kevin Sheehan, 'The Functions of Portolan Maps: an evaluation of the utility of manuscript nautical cartography from the thirteenth through sixteenth centuries'. (PhD, 2014].
43. Pujades 2024, p.361 / 383.
44. [See Origins Essay, K.3e.'Updating the names' and the Francesco Beccari 1403 statement about informants - Campbell, (1987, pp. 427-8. The 'Statement to the Reader' is transcribed (but not translated) in Pujades, 2007 p. 461.
45. International League of Antiquarian Booksellers: 'ILAB History'.
47. See the Catalogue description and the Museum's comments.