Taking slices across a vast site, as archaeologists do, is a logical way to seek for initial general impressions. However, if I, or anybody else, was starting now the selection would probably be somewhat different. For example, for it to be properly representative the 1339 Dulceti would have been chosen (in place of the Catalan Atlas), and others would have been added, e.g. the 1373-83 Pizzigano and the 1426 Beccari (so as to pick up the 1403 innovations). Instead of Ziroldi and Roselli, an early 15th-century Venetian work would have been used, perhaps the 1421 Cesanis. But that is with the benefit of the hindsight that is described in the extended essay Innovative Names.

In an ideal world the aim would have been to make complete transcriptions of every available work but that would be an immense labour. Instead, two self-limiting decisions were taken: first, to document only the coastal names between France and Morocco on just the ten works listed above and, second, to record the presence or absence of a name - with a tick/check, cross, question-mark, or a symbol to show that the area was not covered - rather than its precise form. The sacrifice of the rich linguistic and dialect information reduced the data collection exercise from decades of part-time work to a few years and enabled the results to be presented in a manageable and manipulatable form. The separate analysis demonstrates the value of that approach (for details see Toponymy Main Menu).

From the transcriptions of the works listed above what were called 'Significant Names' were extracted. These were defined as those first recorded after 1313 and/or which disappeared before 1600. There was also a requirement (occasionally overruled) that the names must appear on the dated work of at least one subsequent chartmaker. The purpose was to remove from consideration rare (or possibly unique) names, since it was transmission that was sought not individuality. [For discussion of 'Rare and Irregular Names' see Innovative Names.]

The exercise was then extended to 24 undated works that were (then at least) considered to date before the mid-15th century. In each case it was noted how many of the innovations first found on the 20 or so dated works were present. This was suggested - in Table 19.3 (pp.416-20) in volume 1 of The History of Cartography - as a pointer (on toponymic grounds alone) to the likely approximate dating of the work in question. While clearly falling far short of scientific proof, and running the very real danger that an unusually innovative undated work would not be given its deserved credit, this did start from the common-sense position that the practice of chart-making was largely one of unimaginative copying from a model, with bursts of innovation attributable probably to just a few individuals over the two centuries being studied.

Back in the 1980s, unless years were to be spent travelling around Europe (as I had no budget for high quality photographs), the only comprehensive source was the monumental work of Youssouf Kamal, most of whose almost life-size illustrations are legible. A set of that rare work was conveniently found in the British Library. However, the Kamal illustrations stop at about 1430. That meant that, apart from a pair of representative Catalan and Italian works of 1467-8, I had no record between the 1420s and 1512. Today, almost everything produced before 1470 is available in the scans accompanying Pujades, Les cartes portolanes (2007), and of course we have the Internet for several others.

In 1987 it was possible only to publish name totals, not name incidences. Returning to the subject in late 2008, after a gap of more than 20 years, inspired and enabled by the work of Ramon Pujades and his corpus of electronic scans that has revolutionised portolan chart research, I realised that the computer and the Internet would allow publication of the earlier data. Access to works of the period 1430-69 on the Pujades DVD, and a selection of later works now available online, meant it was possible to extend and refine the original analysis.

The details of the original toponymic findings are summarised in Volume 1 of The History of Cartography (University of Chicago Press, 1987), pp.415-20, with its methodology explained in Appendix 19.5 (p.461). This is now available online in pdf format. The new analysis is set out in the form of the online Table of 'Significant Names' (a Microsoft Word document). Its 1,800 names includes many more than those summarised in Table 19.3 in the 1987 volume. It now incorporates all the standard names as well, to help those searching for a toponym that might not actually be present on the chart in question.

The new table deals with all the pre-1470 works brought into the public domain via the Pujades DVD, and then extends through the 16th century up to (very selectively) 1677. Its documentation of a name's first and/or last appearance, its comments on lineage and rarity, and the statistical summaries that can now be derived from it, all add up to the first systematic attempt to codify the development of portolan chart toponymy over three centuries.

The goal was to record all the repeated or regularly recurring names (about 770 of which have been identified) that were added to the portolan chart corpus after 1313, or which disappeared from it before 1600. The analysis considers the names written inside the coastal outline, along the continuous shoreline from Dunkirk (Dunkerque) in France to Mogador (Essaouira) in Morocco, i.e. extending down the Atlantic coast of S.W. Europe, all the way round the Mediterranean and Black Seas, and finally along the coast of North and north-west Africa. Islands, big and small, have been omitted, as have names written in the sea (usually the other way up).

The period starts with what is considered to be the earliest survivor, the Carte Pisane (but also taking note of two 13th-century portolani: the 'Liber de existencia riveriarum' and the 'Compasso de navegare') and then ranges over the next four centuries. Charts and atlases were analysed (at this stage) only if they are reliably dated, since the intention is to provide each place-name with the date of its first certain appearance on a surviving chart. This cross-section across the whole body of work - rather than the highly detailed excavation of selected areas, for example those studied by Pujades (2007) - provides a general context against which the often different regional pictures can be compared.

The problem with any cross-section is that it can miss irregular features. This had certainly happened here, for example with some evidently influential but undated Italian works [see Pujades, 2007 pp.350-97]. However, the fact that those names had not otherwise appeared across my wide-ranging sample indicates that they did not have lasting influence.

A concentration, in the first phase, on reliably dated works has avoided dependence on estimated dating, and re-dating. Ramon Pujades's 2007 chart and atlas listing (pp. 63-70) and his toponymic analyses of Catalonia & Valencia and the northern Adriatic (pp. 350-97) are both arranged in a single chronological sequence, intermingling those with reliable and with estimated dates. His expertise has certainly sharpened considerably the quality of early portolan chart dating but some future chronological adjustments are almost inevitable. Since it seems generally accepted that dating on stylistic or palaeographic grounds can rarely claim an accuracy greater than 20 to 30 years at best, a concentration on data derived from charts with a clearly stated date - and most signed works are also dated - provides a useful (though not infallible) foundation for future studies.

It is certainly possible, indeed likely, that some of the undated works are ahead of their time and incorporate toponymic innovations. These will need to be examined in a later phase of the analysis (**), in the light of the suggested dating provided by Pujades and, where appropriate, given due acknowledgement for any novelty they seem to incorporate.

The definition of 'significant' was always an arbitrary one, but necessary to keep the exercise within achievable limits. The opportunity was taken during the new investigation to add a number of other names, accidentally observed when searching for those already identified as 'significant', where I had noticed their subsequent repetition. They included some from a group of anonymous Venetian works, such as the Pinelli-Walckenaer and Corbitis atlases, of whose importance I had not originally been aware.

Since nothing, apart from nuances of orthography, would be learnt from the inevitable repetition of names like Barcelona and Alexandria, the analysis focused instead on names that were either first introduced after Pietro Vesconte's earliest complete work - his chart of 1311 or, for the western sections omitted from that, his atlas of 1313 - or that had disappeared before the arbitrary date of 1600, on the basis of a study that extended well into the 17th century. [The comprehensive analysis of names on charts and portolani before say 1318 is still to be carried out (**); perhaps somebody else will do this.]

Some names are apparently unique, others are very rare, uncommon or intermittent (perhaps reappearing after many decades or even a century or more). Some are found only in the work of a single chartmaker or for a short period. Generally, such names have been excluded from this exercise. [But for comments on those, see Rare names, toponymic gaps and reintroductions (a Microsoft Word document).] Had the unique or even rare names been included, the list would have become unmanageably large and the task considerably more complex. The inclusion of such names would also have undermined the validity and purpose of the exercise. This was to seek out toponyms, which, through being copied by other chartmakers, can be considered to have become part of a shared, even common, corpus. [A few exceptions were made for the northern Adriatic, where the Pujades analysis (2007) provides a comprehensive picture for the period up to 1470.]

Nor does the analysis makes any pretence to be geographically comprehensive. It omits the British Isles, as well as all the islands in the Mediterranean and Black Seas, large and small. However, this was not intended to convey any message about their lesser importance. Rather, it was to keep the exercise within the manageable limits of a very large but not complete sample.

Limitations affecting the name selection
Only the names written inland along the continuous continental coastlines from northern France to Morocco were considered. Names written in the sea were omitted. Almost invariably, these are written the other way round, necessitating continuous turning of the image - assuming that rotation was possible. Had all these names referred to islands - as might have been expected - this factor would not be significant.

Unfortunately, the clear-cut distinction between features along the coast and those offshore was sometimes blurred. The names of islands were frequently included in the coastal sequence - in the case of No.1037a thiasso (in the northern Aegean) the 1570 Sideri chart showed it in both ways - whether or not they are identified as such. Inland names may be treated as if on the coast, for example Seville. Conversely, headland names might be placed in the sea, particularly if space within the coastal outline was limited. The fact that the same name might be written in black along the coast on one chart and the other way round, in red, on another, has led to confusion in the preparation of toponymic lists. Some of the names in the Pujades (2007) listing for the eastern Adriatic are actually written in the sea; that is usually, though not always, noted in his listings. I have not made a systematic attempt to document the 'upside down' toponyms that might prove to be duplicates of those already listed in the main sequence.

The only way to resolve that problem would have been to do a second sweep of all the charts on the Pujades DVD and the several dozen later productions considered. Matching up the two lists would have been complicated and time-consuming. Understandably perhaps, it was not attempted.

It seems most likely that, just as the chartmakers must have had models ('patterns') for the direct copying of the coastal outlines, so they would also have had patterns as a guide to the correct spelling, placing and colour of each name. A simple list of place-names would not have been sufficient for their purposes. It is possible that the variant placing of a name inland or offshore may betray the use of a different pattern. To convey that would require the systematic use of a new convention by the recorder.

For various reasons - for example, lack of space or carelessness - a name was sometimes omitted by the scribe, perhaps to reappear on a later chart. The addition of a new name, or the substitution of a very different form of an existing name, might have been achieved by annotating the pattern (whose neatness would not have been important in a workshop). Probably a new toponymic pattern was drawn up only when the number of annotations began to cause confusion or, more simply, because frequent handling made it hard to read.

There are obvious advantages in knowing the current identification of an earlier place-name, for example if searching for its location on a modern map. But that is not the major purpose of this toponymic exercise, which is concerned instead with the cartographic reality at the time of the chart's construction. What matters here is what actually appears on the chart, not what we might conjecture the chartmaker was trying to represent. Only once we have a clear picture of what names were shown, and when, can we start to assess the significance of additions, the development of variant name forms, abandonment, etc.

It is tempting to assume, when faced with a spelling which through time moves towards the modern or recent form, that this is an 'improvement'. And, conversely, identify 'corruption' in forms that tend in the other direction. But even for those names that can be paired with a modern form - and a number cannot with any confidence - we need to consider the likelihood that some variant spellings represent changes to the way that, over the centuries, locals spoke the name, or changes to the forms used by the chartmakers' sailor informants, who would often have been relaying names that were foreign to them.

The most recent systematic attempt at the contemporary identification of portolan chart names was carried out by Ramon Pujades in 2009 for his work on Vallseca's 1439 chart (see pp.154-181). His identifications have been preferred for those names that were included on that particular chart. Konrad Kretschmer (1909) attempted a comprehensive listing of old and new name forms, both from selected charts and from portolani, up to 1490. Of course, some of his 'modern' versions of place-names have themselves changed over the last century, and some of his identifications are not convincing, but this remains a valuable starting-point for names not present on the 1439 chart.

Any definitive answers to questions of identification must be provided by local historians. Information about any one of the 1,800 names in this listing may be readily available in, say, the archive of a small Italian town, but it will not be generally accessible. So, besides the many names for which no modern equivalent is suggested here, other proposed identifications - particularly for small harbours - need to be treated with caution.

In 2016, a historian of Mediterranean France, Jacques Mille, produced the first systematic analysis of a region's toponymy [for details see the Reference page]. Covering Nos 400-457 in the Excel listing, namely from Port Vendres to Menton, this account deftly merged information from local archives with that from the charts and portolani. That coastline includes the Rhône delta, not only one of the largest of the shifting estuaries featured on the portlan charts but, because of its proximity to the chartmakers themselves, one with an extensive toponymy documenting the changing channels. It would be marvellous if this model could be re-applied elsewhere.

Kretschmer, in 1909, arranged his geographical listing into sections for each of the then nation states. I have done the same. Some of the subsequent boundary changes affect the coastlines covered on the charts. That, combined with the not-infrequent cases where no modern equivalent could be suggested, as well as the addition of names not listed by him, makes it difficult sometimes to be sure how the name sequence and today's boundaries coincide. If, as seems likely, I have made mistakes, I apologise for those and ask to be notified so I can make the necessary correction. I appreciate that this can be a sensitive subject.

Four periods, and different phases in the data gathering, need to be distinguished:

Periods
Period 1 - from the beginning to c. 1430 [the subject of the 1980s analysis]
Transcriptions were made, or existing ones utilised, from the ten works listed Above

Period 2 - 1430-1469 [obtained by examining the Pujades DVD]

Period 3 - 1470-1500 [this incorporates the available scans, reproductions and published transcriptions of dated works]. For details of those used see the listing of Post-1469 charts

Period 4 - 1501 to the late 17th century [based on the available images and transcriptions of securely dated charts, i.e. only a small selection of the many hundreds of surviving works]. For details of those used see the listing of Post-1469 charts

Phases
A succession of sweeps have worked systematically round the continuous coastline from north France to west Morocco (Dunkirk to Mogador).

Completed

• 1st Phase. Period 1 (up to 1430, plus 1468 [Benincasa and Roselli]). Done in the 1980s. Some gaps were later filled in.
• 2nd Phase. Period 4 (after 1500). A first sweep was carried out. More work needs to be done, and particularly on the first half of the 16th century, if more online scans or facsimiles can be identified.
• 3rd Phase. Period 2 (1430-69) & Period 3 (1470-1500). This has been completed (though further scans or transcriptions for the period 1470-1500 are still needed).

• A separate table will be created for the undated works (or at least the supposedly earlier ones). This will provide the evidence that could only be summarised via totals in the 1987 chapter (pp.415-20). The presence or absence of each name, presented in a chronological sequence of its first dated appearance, will be noted

Certain general limitations affecting the selection of charts and atlases must be recognized:

• Only a minute proportion of the original output has survived
• Some originals, particularly separate charts, have suffered damage from use and exposure. This tends to affect the east and west ends, most noticeably the Black Sea, limiting legibility
• From the late 16th century onwards there was an increasing tendency to curtail the charts' extent at either side, excluding both the Atlantic coasts and the Black Sea

After 1500, the rate of production increases noticeably, see Pflederer (2009, p.xxii). However, his large totals for the 17th (and even 18th) century are likely to refer to America and the East or more detailed charts of European coasts. [For a note on the last of the traditional Mediterranean portolan charts see Innovative Names.]

Faced with such vast numbers of surviving productions any researcher would have to be highly selective. But the great majority of the works listed by Pflederer have to be ruled out anyway on the grounds of inaccessibility. Instead, for the two hundred years between the termination of the 2007 Pujades DVD collection (1469) and the latest scans I have identified (1670s) the selection has been determined by availability. Working from home, I have depended on the facsimiles in my collection, on the published transcriptions made by others [always to be used warily; I prefer to make my own mistakes] and, most of all, on the available online scans.

Where there was a choice, I looked for examples of the work of the different major chartmakers or families (e.g. Agnese, Maggiolo, Martines, Oliva/Olives, Prunes, Volcio), choosing the earliest and/or latest available in each case. For Freducci and Sideri, perpetuators of the outmoded Benincasan toponymy, I consulted all I could find. See the Microsoft Word table, listing those Post-1469 charts for which transcriptions or legible scans exist.

For this purpose, of course, online images needed to be as legible as possible. In cases where the quality was insufficient for transcription it was sometimes possible to use the image to note the presence or absence of a particular name, even if its spelling could not be read with confidence.

Sheets in an atlas are generally much better preserved than separate charts that spent their life rolled up. Unfortunately, the majority of available scans are of loose charts, not atlas sheets.

For the period up to 1469, the DVD issued with Ramon Pujades's 2007 work provides coverage of the great majority of charts and atlas sheets. The exceptions are works from the Topkapi Saray Museum in Istanbul, which did not grant permission, and a few unobtainable scans. The Pujades images can be rotated to any of the four cardinal points for ease of reading and enlarged to high resolution. In most cases that achieves adequate legibility, though the quality of some of the scans provided, or damage to the surface of the original, sometimes makes reading difficult.

My initial work, carried out in the 1980s used the large and excellent, black & white reproductions from Youssouf Kamal, for the period up to about 1430, supplemented by originals from the British Library.

For notes on the availability of online scans for the period from 1470 onwards see the Explanatory notes to the 'Table of Later Charts'.

The Pujades listings for Catalonia & Valencia and the northern Adriatic are able to display, at a single view, about 70 names, from as many as eighteen works. This is achieved by using the double-page spread of a large format volume (50 x 30 cm). To cover the 94 works he analysed meant that six such spreads have to be examined for the full history of a particular name. While it may be technically feasible to reproduce such a spreadsheet on a computer screen, it would be so cumbersome as to be effectively incomprehensible. And the sample sections recorded by Pujades covered just 10% of the overall continental coastlines.

The present analysis differs from previous toponymic studies in that it does not reproduce transcriptions of names found on one or a small number of charts, or from a single selected area. Instead, it concentrates on a sub-set of the whole (the list of 'Significant names') and notes a few salient features about them - primarily the date of their first and/or last recorded appearance on a dated chart. Later, it is hoped to match that information against the undated works supposedly produced up to 1469 (**). That will be done by noting, for the individual Significant names, those that are present or absent on each of the undated atlases or charts.

The actual form of those names is not being included in the published web table. Nevertheless it is sometimes necessary to read an inscription carefully to be sure whether a particular name is being referred to, or perhaps whether it is the (a) or (b) variant of that name, for example maomeson or malbuso for Pertuis de Maumusson (Geographical Sort Nos 101-02 in the Table of Significant Names).

For those intending to carry out a toponymic analysis the following points may be worth noting. Reading a name is sometimes difficult or confusing. The surface, particularly of a loose chart, may be rubbed or dirty. Letter strokes which descend below the line may run into ascenders from the name below. Names from one coast may be confused with those from the opposite coast, written the other way round (for example in southern Italy). Some names are obscured by rhumb line intersections or decorative detail. Occasionally, red names seem to have been placed over an earlier black name.

Letter confusion presents many problems, e.g. distinguishing between s/f, a/n/u, r/c, etc. It would seem that chart copyists sometimes had the same problem themselves, leading to changes that may be repeated by others. Initial letters frequently mutate (because of differences of dialect or accent, or misreading), e.g. an L might be interchangeable with an N or T, a C with S or T. Thus No.1603 jungi farie can be found starting variously with J, L, N or V.

The way names are placed can affect, or distort, any listing of transcribed names. It is not unusual for the sequence of names to be confused, on a single chart or a group of them. The convention in those cases is to write them in their usual sequence but number them in the order they actually appear. Names will sometimes be found repeated, one instance perhaps being a mistake for another name.

Red was used to highlight the more important names, on which see Innovative Names. Some places are (almost) invariably in red but some toponyms were up- or down-graded, either temporarily or permanently. Only when the change was noted as having been effected swiftly and permanently has it been recorded as a dating aid.