[Select Column R in the Excel spreadsheet to see the transcribed toponyms (for example in geographical order) or Column 5 if the content is to be sorted in other ways.]

This extends from No.247 Cape St Vincent, the major turning-point in south-west Portugal, as far as 1829 Safi in west Morocco.

The standard authority had long been the 1947 edition by Bacchisio Raimondo Motzo, though the Gautier Dalché 'Liber' edition of 1995 included some corrections to Motzo's readings (pp.229-37). An entirely new edition has now appeared: Alessandra Debanne, Lo Compasso de navegare. Edizione del codice Hamilton 396 con commento linguistico e glossario (Brussels, etc.: Peter Lang for the Gruppo degli italianisti delle Università francofone del Belgio, 2011). Preserved in the Staatsbibliothek, Berlin, MS Hamilton 396.

This text, and its 'Indice toponomastico' (pp.293-342), have been used here for the transcriptions. For comments see the notes at the beginning of Corrections and additions to the index by Alessandra Debanne to her edition of 'Lo compasso de navegare' (a Microsoft Word table). The editor clearly did not have access to a geographically-sequenced name list and left as 'unidentified' a number of names whose general position is already known from charts or other portolani. This applies particularly to the Black Sea. A few names are missing from Debanne's index and in some cases a single entry combined two different places. When investigated, some names proved to refer to islands rather than the mainland, and so were left out. 'Lo compasso', like the 'Liber', includes numerous routes across the Mediterranean (peleios or pelagios), as well as additional details about navigational dangers, water depths, etc.

The surviving manuscript of 'Lo compasso de navegare' (Staatsbibliothek, Berlin, MS Hamilton 396) has the stated date of 1296, but the confusing suggestions have been made that the extant text may be both a revision of an earlier version and at the same time a 14th-century copy, perhaps from the 1320s (Pujades 2011, p.267, see note 9). Motzo (1947) had proposed that what we have is an updated version of a work composed in the mid-13th century (perhaps before the foundation of Manfredonia in 1256 [corrected June 2019]), which dating is not disputed by the most recent editor, Alessandra Debanne (2011). Pujades (2013(b), p.18) pointed out that it includes Palamos, founded by Peter II of Aragon in 1279. However, that might have been added subsequently to an earlier model before production of the surviving 1296 text, as was internally stated to be the case with the Black Sea. Debanne notes, for example, that the description for Brindisi does not mention significant changes to the harbour that took place in 1276 (p.29).

In a review article, Andrea Bocchi (2011, p. 275), suggests that the stated date should probably be read as January 1295 not 1296, which would have bearing on the noted absence of Villefranche, founded in 1295 on the site of Olivoli. He also found evidence of scribal confusion that could indicate later interpolations. Debanne concludes a lengthy linguistic analysis by suggesting Ancona as the most likely birth-place, at least of the surviving exemplar (p.246).

What 'Lo compasso' has the space to offer, which a chart does not, are short evaluative and descriptive comments about the safety of a harbour, nearby navigational hazards, the appearance of the coastline, or occasionally (as on the first page) what to take note of, for example, when entering the Guadalquivir River on the way up to Seville. Some of this information could have been obtained by readings from a chart, especially the distances for the open-sea passages, but the charts could not have been the source for the portolano's evaluative comments. The 'chicken and egg' question of their relationship still remains to be resolved.

For the details of other pre-sorted sequences - such as names not noted earlier, or found only here or in the 'Liber', or, conversely, those added only after 1400 to other portolani and charts - see the yellow-headed, numbered columns in the Explanatory Notes to the Excel spreadsheet.

Pujades (2007 - DVD) C 1

[Select Column S in the Excel spreadsheet to see the transcribed toponyms (for example in geographical order) or Column 12 if the content is to be sorted in other ways. Since it is the main focus of this essay, many of the other analytical columns deal with the Carte Pisane, considering, for example, those of its names not found in the earlier texts, those that pre-empt what otherwise appear first on the work of Vesconte and Dulceti (if this is confirmed as being very early), its unusual names, and its overall toponymy with the sparse Atlantic and Black Sea names stripped out. For the details of these pre-sorted sequences see the yellow-headed, numbered columns in the Explanatory Notes to the Excel spreadsheet.]

This includes southern England and the Atlantic coasts (both rudimentarily) and extends round to No.1819 Azamor in west Morocco, with much of the Black Sea now missing.

The Carte Pisane was discovered in Pisa and acquired by the Bibliothèque Nationale, Paris in 1839 (Rés. G.B 1118). Armignacco (1957, pp. 193-222) offered transcriptions (with the exception of the Black Sea) alongside those from the Cortona chart she was describing. She was herself using photographic reproductions to decipher the many semi-legible, or wholly illegible names. Today, we have the advantage of the relatively high resolution scan on the Pujades DVD, with the ability to rotate the image.

I was lucky to have access to two high resolution images: first, a copy of the chart as it appeared in 1852 and, second, as it is today. This is of considerable significance because, while the Carte Pisane had clearly suffered damage before it reached the Paris library, particularly in the Black Sea area, further degradation occurred subsequently. The 1852 'facsimile' was prepared, full size, by the lithographer Eugeniusz Rembielinski for Edme François Jomard's Les monuments de la géographie ou recueil d'anciennes cartes européenes et orientales publiées en fac-simile de la grandeur des originaux (Paris, 1842-62 - No.IX, 1852). The copyist's scribal accuracy and his evident refusal to include anything, however obvious, that he could not see, makes it far more dependable than most hand-drawn copies, even if the match between its toponymy and the compass line network was not closely observed. As can be seen in the detailed discussion of the most damaged area, the Black Sea, this has allowed new insights into this chart.

Pujades (2007 - DVD) C 2

[Select Column T in the Excel spreadsheet to see the transcribed toponyms (for example in geographical order) or Column 17 if its content is to be sorted in other ways. Column 18 isolates those of its names for which there are no identified antecedents.]

This includes the Atlantic but the British Isles, had they been present, would have been on the section subsequently trimmed away. The chart extends as far as No.1738 C. Caxine (Algeria) but has lost, as a result of the trimming, the large stretch between No. 9 Boulogne in northern France and 410 Sète in southern France.

The only significant study into the Cortona chart was carried out by Vera Armignacco, 'Una carta nautica della Biblioteca dell'Accademia Etrusca di Cortona', Rivista Geografica Italiana 64 (1957): 193-222. This included a detailed comparison of its toponymy with that of the Carte Pisane. Where possible Armignacco's readings have been checked. However, having access to the original, which might possibly have been in better condition half a century ago, led her to propose some readings that cannot be corroborated from the Pujades DVD.

In addition, the Cortona chart's names were considered by Alberto Capacci in his 'Osservazioni sulla carta anonima detta di Cortona', communication to the XXXVI Congresso Geografico Italiano (Genoa, 4-9 May 1992). On p.XIV of his comprehensive 1994 listing, Capacci says he used the Armignacco transcription, but had verified those readings, following a recent restoration ('integrata da una verifica resa possibile dal recente restauro del documento').

The Cortona chart has been generally accepted as a work of the 14th century (and recently confirmed as such by Debanne (2011, p.18)). In his 1995 study on the 'Liber', Gautier Dalché suggested the Cortona chart was drawn in the middle years of that century but based on a model close to the time of the Carte Pisane [still at that time considered by him to be very early] (p.26).

Not on the Pujades 2007 DVD

This extends from southern England and the Atlantic coasts round as far as No.1829 Safi.

[The Lucca chart's toponymy has not yet been transcribed but Column U in the Excel spreadsheet indicates the presence of a name, or the likelihood of that according to a three-point legibility scale. Select that column in the Excel spreadsheet to see the incidence (not transcription) of the toponyms (for example if the default geographical order is required) and Column 19 if its confirmed content is to be sorted in other ways.]

Discovered in 2000 in the Lucca city archives, this was first formally announced by Philipp Billion, in 'A newly discovered fragment from the Lucca Archives, Italy', Imago Mundi: the International Journal for the History of Cartography 63: 1 (2011): 1-21 & coloured plates 1-2. This intriguing chart has many unique features but, while clearly early, its dating is controversial. Billion concluded that its 'toponyms, geography and palaeography point to an earlier date [than 1311] but other evidence placed it during the time of Vesconte'. In his opinion 'not one of the dating techniques used here leads to a date after 1327' (p.13).

The Imago Mundi article is essential reading. For my own general comments see the relevant note in the Census additions. Since the chart was unknown at the time of his major study (2007), Ramon Pujades could not discuss it. The 2012 Paris conference paper that supplied the impetus for this extended essay provides Pujades's arguments for a much later dating of the Lucca chart and, because of their close relationship, the Carte Pisane as well.

The comments that follow concentrate on the toponymy. Remarks relating to other aspects are inserted in the appropriate place in the accompanying essay.

Billion (p.8) highlighted the undisciplined hand as evidence that the author was not a well-trained scribe. Sometimes the letters are not parallel and are different in size, which can make it difficult to work out which part belongs with what toponym, which way round it is, and whether an indistinct line is part of the coast or a toponym. As a result it can be difficult to tell even how many illegible names are involved.

The lettering is larger than necessary and its creator's preference for prefixes leads to some unusually long names - for example, Excel listing No.190a cauo sca maria de finibus terra, 613a lena de lequia (though that is also the Carte Pisane form), 1689 cauo [] portu raxa raxagibele. As a result these sometimes run into toponyms from the opposite coastline. His handling of mistakes was clumsy; nor was he much concerned about the appearance of his work. Sometimes insufficient room was left for a red name. As with other chartmakers, scraping off unwanted lettering seems not to have been an option. 824 tragur, in red, is placed over a (now illegible) black name; he wrote the original 1089 varna incorrectly and so simply rewrote it alongside. He started to write the red name [luco?] to the west of 1556 Alexandria, putting in just three ascenders. Then he seems to have gone over it and crossed through with hatching in black ink. As a result, the black name [torre] de larabo had to start well inland.

A number of the expected North African names (i.e. those found on the 1311/13 Vesconte works) are absent and, while the overall impression is of a relatively sparse toponymy, this is not always the case. For example in the stretch between 864 Durazzo and 902 Larta, covering Albania and northwest Greece, the Lucca chart has 18 toponyms as compared to the 10 or 11 found uniformly on the Carte Pisane, Cortona, 1311 Vesconte and Riccardiana charts.

Rather than labouring for weeks over every letter in every semi-legible name, it was decided to check first if a name was recognisably one of those already recorded. Factors taken into account for hard-to-read toponyms were the overall length, the initial and terminal letters, and the pattern of ascenders and descenders. The majority proved to be present on the work of Vesconte and, in the case of the Lucca, on the Carte Pisane and Cortona chart. Those successful readings were given a mark indicating recognition and legibility.

Other names, which from their form and position showed that they seem to have been previously unrecorded were noted in Column D, if they were sufficiently legible for that to make sense. For the Lucca chart those can be found by sorting on Column X 'Comments' and scrolling down to "Lucca chart only" (40 entries). Some of the unidentified toponyms may turn out to be variants or corruptions of those already known. Because so many names were illegible, a digit was placed at that point in the geographical sequence to indicate the number of unidentified names involved. This is coloured blue and placed inside curly brackets. Even if known toponyms were expected at that point no assumptions were made without visual evidence.

Future scans might allow other names to be read but the serious fading that has affected the red toponyms (sometimes to the point of total invisibility) makes the task far harder, by removing the normally reliable reference-points amidst the littoral sequence.

For the Riccardiana chart, which appeared to have few names not found elsewhere, the semi- and fully illegible names were coded 3 and 4 respectively. Just three "noted only on the Riccardiana" instances can be found by sorting the 'Comments' column (X).

For both the Lucca and Riccardiana charts only the names classed as '1 Definite' and '2 Probable' have been included in the statistical analyses.

Pujades (2007 - DVD) C 4

See 'Why have the Lucca and Riccardiana charts not been fully transcribed?' under Lucca chart above.

[The Riccardiana chart's toponymy has not been transcribed but Column V in the Excel spreadsheet indicates the presence of a name, or the likelihood of that according to a four-point legibility scale. Select that column in the Excel spreadsheet to see the incidence (not transcription) of the toponyms (for example if the default geographical order is required) and Column 20 if the content is to be sorted in other ways.]

The coverage of this chart is complete so far as this investigation is concerned.

Until 2007, when Ramon Pujades discussed this in some detail and assigned it to the first quarter of the 14th century - in other words during the Vescontian period if not even earlier - the chart in the Biblioteca Riccardiana, Florence had been considered simply as a work of the '15th century'. My 1986 Census (no.80) treated it thus and could cite only the Uzielli listing of 1882 in evidence. To my knowledge it had never been reproduced before. Having obtained a high resolution image in the course of his investigation (in case it might qualify for the pre-1470 cut-off in his listing) Pujades immediately realised its importance, as the earliest surviving Genoese chart. It now sits on his chronologically organised DVD after the Carte Pisane, Cortona chart and the first production of Vesconte in 1311. As he noted, its 'coastal toponyms amount to fewer than 1,300 as opposed to the over 1,700 in the 1327 chart of Perrino Vesconte or the over 1,600 in that of Angelino de Dulceto, from 1330' (2007, p.447, note 4).

The Riccardiana chart has been included in the present investigation as a 'control' since I fully support the suggested early dating of Pujades and nobody, as far as I know, has challenged that. Along with the dated productions of Pietro and Perrino Vesconte it can therefore act as another uncontested witness to the state of portolan-chartmaking in that formative period. Until their proposed re-dating by Pujades, the Carte Pisane and Cortona chart had been considered as the sole surviving precursors of Vesconte. No longer witnesses, they have now been placed under scrutiny themselves and cannot therefore be used any longer as fixed comparative points. The Riccardiana chart has had to take their place (along, partly, with the Carignano map, but see the comments below).

Since its own dating remains imprecise, the Riccardiana chart has been tested alongside the Carte Pisane, Cortona and Lucca charts, and its results appear alongside theirs in the various tables accompanying this essay, as well as in the commentary on those. Despite the Riccardiana's ambiguous relationship with the Vescontian production - sometimes seeming to imitate that (or its source), at other times clearly more archaic - nothing has emerged to throw doubt on its dating by Pujades to the period before 1325. [For detailed comments about the relationship between the Riccardiana chart and Vesconte, see Notes on unsigned charts and atlases.]

While the Riccardiana chart may come to be seen as a near contemporary of the Cortona or Lucca charts, and although it does share a few 'early' features with them, its dissimilarity from those is noticeable. The Riccardiana's confident, even hand is in marked contrast to the Lucca's carelessness. Furthermore it seems clear that the Riccardiana chart's creator was a professional scribe. Evidence for this can be seen from some of the scribal abbreviations or 'suspensions' he uses, e.g. No.448 insula sca (followed by a single letter denoting margalita), 617 where trebisace is apparently conveyed by ezac, 1059 gan for longano, and 1492 where gloriata is conveyed by Glata (using the standard contraction for gloria, namely gla). Lack of available space was not involved in any of those cases. That such scribal abbreviations were employed on a portolan chart must presumably imply either that it was intended for a learned customer [we have no information on the chart's early provenance], or that the level of literacy attained by the expected mariner audience was sufficient to decode it, or perhaps no more than that the name in question would have been sufficiently well known to have been expected anyway.

Pujades (2007 - DVD) C 6

The toponymy has not been transcribed but Column W in the Excel spreadsheet indicates the (probable) presence of each name by #. See also Column 33 where a code identifies names first seen on dated charts from 1318 onwards, names not seen after 1330, and those apparently unique to the Carignano map.

The coverage of this map/chart is complete insofar as this investigation is concerned, though some of its detail is obscured and in places its toponymy is illegible. Despite its importance for this survey of the early portolan charts, because it was produced no later than 1330, and probably no earlier than 1325, it has proved relevant in only a few areas of this essay.

The map produced by the Genoese priest, Giovanni da Carignano must have been created in or shortly before 1330 since a record of that year notes that he was no longer alive. [For a biographical note see: Graziella Galliano, 'Mauro, Giovanni (Giovanni da Carignano)', in Dizionario biografico degli Italiani, LXXII (Rome, 2008), 399–401]. This means that it can be confidently used as a commentary on the general context of maritime knowledge in the period leading up to that year. Its dating had long been a matter for conjecture, with little or no evidence to back that up (see Campbell, 1987 pp. 404-6). However, as discussed elsewhere in these pages, its depiction of the British Isles reflects the stage reached in Vesconte's steadily growing understanding during the mid-1320s. Since there is no indication that mapmakers in Genoa had access to the same or better sources than those available to Vesconte in Venice, the Carignano map must therefore have been created in the last decade of the priest's life, and probably no earlier than 1325.

Formerly in the Archivio di Stato in Florence, the map was destroyed in 1943. Access must therefore be via pre-War monochrome photographs, none of which is sufficiently legible to read many of the names. The best record is that in the monumental work by Youssouf Kamal, Monumenta cartographica Africae et Aegypti (4:1:1138-9). The 2007 Pujades DVD can sometimes be used to confirm, from its shape, if an expected name is present but not to read an unusual one, although the zoomable scan in the DVD accompanying Lepore, et al. (2011, No.4) is of slightly higher quality.

[Update, January 2017: the Florence Archives has put online an improved image, to be seen Here. Further update, April 2023: Alberto Quartapelle, 'Giovanni da Carignano: Fourteenth-Century Cartographic Innovator', Terrae Incognitae: The Journal of the Society for the History of Discoveries, March 2023. (Announcing important new findings, derived from the author's now freely available remastered photograph in the Media Chart collection, and arguing for a date of 1329)].

The Carignano map has clearly been meticulously copied from a portolan chart, including a careful reproduction of the time-consuming in-shore navigational hazards. Though there is a peculiarity in its twin-circle compass line network east of Sicily, this, and the scale bars, also conform to the normal conventions found on early portolan charts. Nevertheless, despite its author's recorded nautical connections - he was admonished in 1314 for allowing merchants to store marine equipment in his church - the Carignano 'map' is actually a hybrid between a map and a chart, with most of the coastal names written in the sea rather than inland. Where it differs most from the other early charts is in the dense amount of inland information, specifically the prominence given to the distinguishing flags of numerous towns.

In other respects, the inland detail is reminiscent of the production of the first chartmaker working in nearby Majorca, Angelino Dalorto/Dulceti, the date of whose earliest surviving work (still owned by the Corsini family in Florence) has been read by Pujades as 1330, the last year during which Carignano can have been alive. Catalan charts, whose recorded history starts with that same work of 1330, can be easily recognised by the prominence given to green-coloured mountain ranges. The interior of the Carignano map is reduced, on the monochrome Kamal reproduction, to various shades of grey discolouration, sometimes with indeterminate outlines faintly behind. However, the treatment of the Atlas Mountains, even if obscured, seems similar to that of the 1330 chart. Likewise, the depiction of off-shore dangers uses a simple convention of dots (presumably in red), like that on the 1330 Daloro chart but in contrast to the mixture of dots and black crosses introduced by Vesconte in the mid-1320s. All of this makes it likely that a lost earlier work of Dulceti (originally himself Genoese), or conceivably a further re-dating of the Corsini chart back to 1325 (the former rendering of its stated year, replaced by Pujades's proposed re-reading of its date as 1330), provided Carignano with his model. The maturity of the Corsini chart and the brief known career of Dulceti (1330 to perhaps a little after 1339) make it highly likely that he had been producing charts prior to 1330. The Carignano map's closer imitation of the Dulcetian name-set than that of Vesconte as (described below) further reinforces this suggested link.

Catalan charts also had a larger geographical coverage than Vescontian ones, taking in the Baltic to the north, extending a little further east than the Black Sea, and south beyond the Atlas Mountains. That the Carignano map's limits broadly match those of the 1339 Dulceti chart rather than the slightly narrower scope of the 1330 chart presumably indicates that the full 1339 Dulcetian extent could have been seen at least ten years earlier. If that proves to be the case, close examination of the obvious differences between the Carignano map and the Dulcetian charts, for example with respect to the Baltic, the British Isles and north-west Africa, might reveal the outlines of the earliest Catalan charts, now lost. It is of course possible that Carignano made some cartographic contributions himself in those areas. This is just one area of possible future research.

Toponymy

I am not aware of any published transcription of the Carignano map's toponymy, although Vera Armignacco (1957, pp. 202-207) did include most of its Black Sea names (with a break between Excel No.1168 vospro and 1204 matrega corresponding to the gap in the Cortona chart that was her concern). The missing section extended from east Crimea, round the northern shore of the Sea of Azov and back to the opposite point. I am particularly grateful to Ramon Pujades for supplying me with a draft of his full transcription of the coastal names. This remains unpublished and so is conveyed in the Excel spreadsheet's Column W just by an indication of where the name is (apparently) present rather than by its actual form.

Carignano's treatment of toponymy, an essential element of any portolan chart and the only one that underwent continual modification, was quite different to that on a mariner's chart. Its reduced selection of names was split into two sequences, with the significant ones (including, for example, many of those that would have been in red on a normal chart) placed in the interior of the map, within a circle bearing the city's emblem. The names of other prominent places or provinces were located inside rectangular white labels. This device provides a further link to the work of Dalorto/Dulceti, who seems to have introduced it into the world of cartography (see Name Labels). The remaining coastal names were written not, as usual, inland, but in the sea - traditionally kept clear so that the all-important coastal information would be clearly visible - and in the opposite direction to standard portolan chart practice.

The full 'Foundation' coverage on the Vesconte works of 1311 and 1313 runs to 1004 names; the Carignano map (produced a decade or so later) has just 431. This might suggest that Carignano had intentionally omitted slightly over half the names likely to have been available to him. Was this because of limitations of space? This does not seem to have been generally the case, although there are contradictory indications.

A few instances of low or high toponymic density can be given (with the numbers referring to those in the comprehensive Excel listing). The percentages indicate the proportion of each Carignano sub-total to that on the earliest Vescontian coverage (1311-13):

It might have been anticipated that the area either side of Genoa would have been well provided with names, perhaps some unusual ones. In fact, the Carignano map includes just under half of the early Vescontian toponyms for that stretch and none of the names Vesconte added from 1318 onwards. Other questions remain. Why is the density high for sections of the Atlantic coasts and yet significantly lower for the intervening Mediterranean Spain? Why is much of the route to Constantinople lightly provided with names in contrast to the west and south coasts of the Black Seas beyond that?

A number of the Carignano map's less usual names provide no more than the expected corroboration with a roughly contemporary Genoese production, the Riccardiana chart of c.1320. However, of the 29 names that have been identified on the Carignano map as being otherwise first seen on dated works from 1318 onwards (Excel listing, Column W, then 28), all but four can also be identified on one or more of the Carte Pisane, Cortona, Lucca and Riccardiana charts - an indication that those toponyms, at least, were known in maritime circles before 1330.

The following five Carignano names can be specifically mentioned since they are all absent from the Riccardiana chart:

• 442       Frezur (on the Carte Pisane, Cortona and Lucca charts; otherwise first noted on the 1330 Dalorto chart)
• 488       Levania (on the Carte Pisane, Cortona and Lucca charts; otherwise first noted on the 1330 Dalorto chart)
• 530       Ostia (on the Lucca chart; otherwise first noted on the 1330 Dalorto chart)
• 1106a   Fedonixe (on the Cortona chart; otherwise first noted on the 1330 Dalorto chart)
• 1802a   Cibo (otherwise seen only on the Carte Pisane and Lucca chart - though differently placed by Carignano in the toponymic sequence)

Table A in 'Totals of the black and red names, first seen on dated works after 1313 but included on the four supposedly very early anonymous charts' itemises the number of potential anachronisms (as well as the percentage of the possible totals they represent for each dated group of additional names). The Carignano figures are given separately in an end-note because the much reduced figures would provide misleading comparisons. Even so, the Carignano's inclusion of four names that were otherwise first noted in 1339 is the same total as on the Riccardiana chart, whereas the Carte Pisane, Cortona and Lucca charts have two or three each.

In relating the Carignano map to the work of his contemporaries, Vesconte in Venice and Dulceti in Majorca, it needs to be emphasised that Carignano's debt to Vesconte was partial and indirect, whereas his apparent anticipation of the Dulcetian toponymy is much more marked. Of 246 names added by Vesconte after his initial productions (i.e. during the period 1313-27) Carignano includes only 20, while he reveals knowledge of 13 out of the 149 whose first dated appearances are on the 1330 or 1339 Dulceti charts (produced after his own death). [Use Excel Column 28, then W, and again with the order reversed.] Carignano's mixed inheritance underlines his physical and intellectual distance from Venice on the one hand and closer proximity on both counts to Majorca on the other. But it also reaffirms the degree of partial isolation among those making charts in Genoa. It was not until Francesco Beccari, around 1400, that we first notice much renewed toponymic innovation in that city.

Carignano's priorities

Although the map/chart was certainly not intended for use on board ship, Carignano does seem to have respected the significance of navigational information by carefully placing the toponyms so that they would not obscure the symbols for coastal dangers. He was also reluctant to squeeze in closely-written sequences of toponyms, which meant that numerous names had to be jettisoned. The main reason for the toponymic sparseness of the Carignano map, though, is that his decision to place the names in the sea along a coast with numerous islands left little room for names. The Dalmatian shoreline is a good example, Morea (Peloponnese) another.

On the basis that Carignano was copying a chart by, or similar to, one by Dulceti, the alterations he made to its design must indicate his own rather different priorities. One of those was clearly a desire to pair as many of the major towns and ports as possible with their identifying banner or flag. The names of other prominent towns, both coastal and inland, were emphasised by being placed within large white name labels. In other words, he was primarily interested in the political and urban dimension of contemporary geographical knowledge. Although Vesconte and the recently-discovered Lucca chart include similar information in different ways, neither approach the density of the Carignano map, which surely deserves a detailed vexillological analysis. Did he gather those himself, or are there known earlier precedents? Around the Iberian coast, for example, he includes no fewer than 13 of those, with perhaps another ten inland, where Vesconte had shown just four and the 1330 Dalorto/Dulceti chart six.

Even though Carignano's design decision meant he was forced to reduce his toponymic complement, he did include a handful of additional names (perhaps unique to the portolan charts), mostly of river mouths, in which he clearly had a particular interest. He also seems to have made a systematic attempt to document all the provinces, which are usually shown more sparingly on the portolan charts and often as if they were just another coastal name. These are further reflections of his geo-political rather than merely hydrographical focus.

Making a direct copy of a portolan chart by one untrained in that skill - which seems likely to have applied to Carignano, a priest - would have entailed a considerable amount of painstaking work. The decision to alter the placement of the toponyms would have doubled the effort, since it involved a range of cartographic decisions: resolving which town to provide with its civic arms (though that might have depended on what armorials were available to him), which to place inside a label for added prominence, and which to write into the space available to him along the coast. Since he was clearly averse to squeezing in as many names as possible, preferring instead a neat arrangement of regularly spaced toponyms, his options were still further reduced.

The net result of the adaptations to his portolan chart model was to change its purpose from a hydrographic chart based on the Mediterranean (with a comprehensive toponymic documentation designed for use by a navigator or a landsman monitoring such voyages) to an instructional manual of modern geography. [For a good summary of the map's significance, as reflecting a merging of the 'pragmatic and erudite' traditions, see Pujades (2007) p.271a.]

For comments on the Carignano map's treatment of the British Isles see under Hydrography

Pujades (2007 - DVD) A 11 & A 12

The coverage of these atlases is complete so far as this investigation is concerned.

Named after previous owners - 'Corbitis' being an improved reading by Piero Falchetta of the previous 'Combitis' (see Census update [No.117]) - the two works are clearly in the same, unknown hand and very similar in both appearance and content. Their most likely dating is the first decade of the 15th century. Pujades considers the inclusion of Livorno to indicate a date no earlier than 1406, but for my comment on that see [Livorno].

At the time of my first toponymic analysis in the 1980s I was unaware of the originality and importance of those two works and did not give them the attention they deserved. Recently, their place-names have been systematically checked, both visually from the Pujades DVD and via the Kretschmer listing of 1909 (which includes the 'Atlas Pinelli' (i.e. the Pinelli-Walckenaer Atlas) as one of the works he had examined. As a result, a number of unusual or even unique toponyms have emerged. These can be isolated by sorting the Excel listing on Column 33 and scrolling down to the 'CPW' entries. Those have, in turn, been sub-divided into groups of names: first, those that were revived on those atlases (CPWr) - usually having been seen previously on Vescontian charts - and, second, those found uniquely on one or both of those works (CPWu). 123 of the names have been identified as unique or rare occurrences on the Corbitis or Pinelli-Walckenaer atlases. Pujades focused on those works in his 2013 article because of their supposed toponymic similarities with the Carte Pisane and Lucca chart. However, no more than six of the Carte Pisane's names (5%) are in that group [sort on Column 12, then 33] and just two feature in the list of 14 'revivals' on those atlases.

[For a more detailed discussion of these two works, see The Corbitis and Pinelli-Walckenaer atlases.]