(main menu)
CAMPBELL
articles
CHARTS
menu
SITEMAP
site is
ABOUT
NEW
|
|
(main menu) |
CAMPBELL articles |
CHARTS menu |
|
SITEMAP |
site is ABOUT |
NEW |
|
additions and corrections will be noted in the appropriate place with a dated statement between { }, which can be searched for
The 3.5 metre world map that Muhammad al-Idrīsī completed in or a little after 1154 is preserved only in later copies of his geographical treatise, the Nuzhat al-mushtāq. That work’s Introduction describes the 15 years he had spent gathering the information and gives details of the processes involved in creating the map, in up to four different forms. Using a Processual Approach, the contradictions in that purportedly authorial account have been tested against feasibility and common sense. A number of questions have been asked, perhaps for the first time, and some tentative revisions offered. It is considered unlikely, for example, that the small circular world map fronting the volume (but not mentioned in it) has any direct connection with al-Idrīsī. It is also proposed that the ‘silver disc’ is more likely to have been rectangular, so as to match the format of the wall-map, as was clearly and repeatedly specified.
Another arm of the Processual Approach is concerned with the circulation of cartographic information, which meshes well with this essay’s search for evidence of the subsequent dissemination of the idrisian geography. In seeking for any borrowing between the 1154 map and the portolan charts (broadly contemporary but significantly more realistic), it is clear that the authors of each of those must have been unaware of the other. Had al-Idrīsī seen a portolan chart, he would certainly have taken note of it. On the basis of that, the portolan chart’s origin can be confidently placed in the narrow period 1154-1210 – the latter being the most likely dating for the Liber de existencia riveriarum, whose direction and distance statements can only have come from a marine chart.
As for the dissemination of the large world map, it is unsurprising – given the lack of evidence that its 70 sections were ever joined up – that anyone effectively ‘saw’ it as a whole until the 20th century.
This essay emerged from a need to test suggestions that al-Idrīsī‘s large world map of 1154 might have inspired the creators of the portolan chart. Given al-Idrīsī‘s celebrity, and the column inches devoted to him in the cartographic literature, it was anticipated that the salient features about the man and his work would have been clearly set out by others. It turned out instead that there was a lack of unanimity among Arabists, despite the fact that the volume in which the map is preserved contains detailed descriptions about its creation. Unfortunately, those contemporary statements proved to be contradictory, while the list of ‘well-known’ facts about al-Idrīsī and his map include some that are implausible. In the absence, therefore, of a coherent story that could explain the map’s origin, production or dissemination, it became necessary to look more closely into its history. Only once that investigation had been carried out would it be possible to relate the world map to the issue of portolan-chart origins [on which see Here]. The possible portolan-chart connection will be returned to at various points in what follows.
Approaching these far from straightforward problems as a generalist map historian rather than a specialist in Islamic cartography, I have been emboldened to suggest some new interpretations, deriving from the exercise of a relatively new Processual Approach, which focuses on the materiality of early maps. 1 By investigating the physical properties of the various elements that made up the idrisian project, this essay attempts to draw inferences, from the limited information available, as to the likely procedures involved in each case. The processual approach also concerns itself with the circulation of cartographic information, which meshes well with this essay’s search for evidence of subsequent dissemination of the idrisian cartographic outlines.
Muhammad al-Idrīsī, or, alternatively, al-Sharīf al-Idrīsī, is best known for his Kitāb Nuzhat al-mushtāq fi’khtirāq al-'āfāq [often abbreviated to Nuzhat al-mushtāq], or ‘The book of pleasant journeys into faraway lands’ in Ahmad’s translation. 2 [It is sometimes referred to instead as the Kitāb Rujā (‘The book of Roger’).] This is an illustrated book of geography formed of three elements: a long text and two very different world maps. The first is a small circular map which, it now seems, lacks any direct connection with al-Idrīsī. The second is one of the masterpieces of Islamic cartography, a large rectangular world map – offered to the world in sectioned form, spread throughout the Nuzhat al-mushtāq. Not only does this wall-map have no stated title but, more surprisingly, neither has it been given a generally accepted one. It is sometimes referred to as the Tabula Rogeriana but, for reasons explained below, ‘Charta Rogeriana’ is preferred here.
al-Idrīsī’s world map project comprised two other significant elements, both carrying images of that large map. One of those was evidently on wood, the other engraved on silver, but melted down shortly afterwards. Each is known to us only from perfunctory and ambiguous descriptions. Those four elements will be discussed in turn.
al-Idrīsī, was, it now seems, born in Sicily (rather than Ceuta, Morocco as previously supposed and where he is honoured with a large statue). As the exiled grandson of the ruler of Malaga in Islamic Spain (al-Andalus) he seems to have been brought up in the Sicilian royal household. He completed this, his most celebrated work, for Roger II, the Norman king of Sicily, in Palermo. 3
Sicily occupies a pivotal place in the Mediterranean, and its central geographical position was enhanced politically at that time by the intellectual energy resulting from the fusing of Arabic, Byzantine (Greek) and Western (Latin) thought at Roger’s court. Against that, it seems that the extent of al-Idrīsī‘s own journeys has been exaggerated in the past, and Jean-Charles Ducène notes no more than that his "travels would have taken him to the Iberian Peninsula, present-day Morocco (he crossed the Strait of Gibraltar several times), Constantine (in present-day Algeria) ..." 4
The literature is in general agreement that the oldest surviving exemplar of the Nuzhat al-mushtāq is that preserved in Paris (BnF, MS. Arabe 2221) and clearly dated 1300. However, Annie Vernay-Nouri has pointed out that the number 700 on folio 351v (at the end of the 8th section of the 7th climate) is neither a colophon date (AH 700 = 1300-01) nor is it written in the same ink as the rest of the text. 5 Her close examination of other extant MSS throws further doubts on previously suggested datings, but, as yet, there is insufficient evidence to assign new ones. She also suggests that the incomplete St Petersburg example (cited in Ahmad’s list as a copy from the ‘beginning of the 14th century’), may be from the same period as the Paris MS and possibly even from the same atelier. 6 This imprecision is unfortunate and it is to be hoped it may be resolved. In the meantime the Paris MS will be referred to here, for convenience, as c.1300.
The first of the two cartographic illustrations in the Nuzhat al-mushtāq is a small circular world map (to be discussed in some detail later) but the main focus of interest is the large world map separated out into 70 rectangular sections, which are dispersed unevenly throughout the volume’s 350 folios. Helpfully for today’s researchers, the sections were joined up into a single map in a re-drawing by Konrad Miller in 1928, including transliteration of its Arabic script. 7 The easiest way to view Miller’s re-creation is by means of the Wikipedia scan, with its northward orientation (the ‘upside down’ version).
To distinguish the element that concerns us from the book that contains it, the title that Konrad Miller used for his own re-creation, ‘Charta Rogeriana’, is used here. Because of the fragmented nature of the original, it is only feasible for most researchers to study the map in Miller’s composite version. The use of the term Charta also serves to emphasise that Miller’s readily accessible re-drawing, rather than the dispersed original sections, is the source for most of the comments that follow. However, for detailed research it is essential to consult the Nuzhat al-mushtāq itself. Online scans are available for the two versions on which Miller mainly based his redrawing: the oldest surviving example (c.1300) in the Bibliothèque nationale de France, Paris, MS. Arabe 2221, and that in the Bodleian Library, Oxford, MS.Pococke 375 [a late example from 1456]. 8
The Charta is divided into seven horizontal climate zones, and further split into ten vertical divisions. The sections are distributed throughout the volume, each with its own separate commentary. It is referred to here, and generally elsewhere, as comprising 70 sheets, since that figure reflects Miller’s numeration. However, there is a further horizontal band at the south, whose coverage is akin to that of ten half-sheets (extending to the south of the first climate band). Those trace out, conventionally, a straight continuous coastline (whose names identify this as covering the Red Sea, East Africa and beyond to the south). However, that coastline is shown instead as running confusingly due east to the edge of the map. The distinctive island of Sokotra (gezira sokotra ) marks the point where the Gulf of Aden meets the Indian Ocean. The reason for the Charta’s outline turning sharply to the east rather than continuing southwards can perhaps be best explained by a need to remain within the available space.
In Miller’s version these half-sheets are additionally used to carry his numeration for the columns; while in the original volume the 10 southernmost sections are enlarged to accommodate the extra half sheets as well. The Wikipedia ‘upside-down’ version (already cited) has reversed the original orientation for the convenience of modern users, by placing north at the top. To access the Arabic lettering, use an Arabic version of Miller’s re-drawing, by Muhammad Bahja al-Atharī and Jawād 'Alī, “restored it to its original Arabic in a verified and revised status”, based on five of the Nuzhat al-mushtāq MSS. This was published by the al-Misāha publishing house (Baghdad, 1951 CE / 1370 AH). [I owe this reference to Fateme Savadi.] To read transcriptions of the Arabic into Roman characters, consult Miller’s drawing (via Wikimedia) with its southwards orientation; alternatively, for greater resolution, use the Reddit version.
The following attempt at an overview of the historiography of the Nuzhat al-mushtāq, and particularly its maps, was prompted by the apparently irreconcilable inconsistencies within and between the textual and graphic evidence available to us. As just one of the many difficulties facing those trying to understand the Charta fully, the relationship between text and map is not straightforward, since each includes material missing from the other. One element where this is very apparent is the toponymy. Whereas there are about 2,500 toponyms on the Charta, twice as many can be found in the text. 9
The analysis that follows, based in some places on little more than logic, runs counter to some of the conclusions of previous commentators, all of whom have far more specialist knowledge of Islamic cartography. But I am not aware of any who have faced up fully to the Nuzhat al-mushtāq’s internal contradictions. If I have been able to contribute usefully to this debate it is as a historian of cartography.
The written account of the process that resulted in al-Idrīsī‘s extended geographical and cartographical survey comes from a single source, the Introduction to the Nuzhat al-mushtāq itself. Although this is first encountered in a much later manuscript of c.1300 (see the note above about its disputed dating) it has had to be generally assumed that that would have matched al-Idrīsī’s original text of 1154 (although that also is questioned below). Most of the difficulties lie not with the large planisphere (the Charta), incorporated in the volume in multiple sections, but with the three other manifestations of the world map: a circular map, whose modest size (about 33 cm) was presumably dictated by the format of the Nuzhat al-mushtāq volume; as a large silver plate (or disc), apparently destroyed seven years after it was made; and finally as an outline on a ‘drawing board’.
Jean-Charles Ducène’s recent explanation of the compilation process for the Nuzhat al-mushtāq’s text and its, apparently, single map, as set out in the work’s Introduction, is as follows. “[al-Idrīsī‘s] collation took him fifteen years. He then drew a map and arranged to have it transferred onto a silver disc showing all countries, coastlines, seas, rivers, roads, and distances; this planisphere was destroyed in 555/1161 9a ... Finally, artisans transferred this map into a book, the king added some complementary text, and the work was given its title in … January 1154”. 10 Ahmad’s 1992 translation adds further details about the sequence of events, stating that, “He [the king] wished to make sure of the accuracy of what these people had agreed upon both of longitudes and latitudes [and in measurements between places]. So he had brought to him a drawing board 11 (lauh al-tarsim) and had traced on it with iron instruments item by item what had been mentioned in the aforementioned books, together with the more authentic of the decisions of the scholars.” Once the silver plate had been prepared, “he had engraved on it a map of the seven climates … according to the version appearing on the drawing board, not differing from it at all 12 and thus following what had been decided there without any variation ". 13
Because of the interweaving of the different strands, an attempt will be made to separate out the various issues. Among the
questions that will be looked at in turn are the following:
There is no general agreement about the relationship between the small circular world map and the 70-section rectangular planisphere [the Charta]. When comparing their outlines, reasonable arguments can be made that each is earlier than the other. Does the crudeness of the circular map’s coastlines point to an older, less developed understanding or should it be seen instead as a simplified summary of the Charta, perhaps rendered even less faithful to the original through the progressive degradation of repeated copying? 14 Inevitably, its modest size ruled out any degree of detail. Compared to the 2,500 toponyms on the Charta, for instance, there must be barely 100 on the circular map.
Found in the (supposedly) oldest surviving manuscript of the Nuzhat al-mushtāq, the circular map has almost invariably been chosen to illustrate al-Idrīsī‘s world view, presumably because of its attractiveness and convenience. A very similar example of that same map has emerged, however, in the recently discovered Book of Curiosities, whose text is datable to c.1020-50. Despite that, the unique surviving exemplar is considered to have been copied around 1200, and its circular world map conceivably later still. 15 As a result, there are competing arguments as to whether the earliest appearance of that map is the one found in the Book of Curiosities (which would thus make it irrelevant for idrisian studies) or, alternatively, if it should be confirmed as the work of al-Idrīsī himself, and therefore a later addition to the earliest surviving text of the Book of Curiosities.
Resolving that point is made more difficult by the fact that, in each case, the oldest extant manuscript dates from about a century and a half after the creation of the work in question.
There are several surviving examples of the small circular map. 16 When the oldest version of the Nuzhat al-mushtāq’s circular map (BnF Arabe 2221) is added to a side-by-side comparison 17 – i.e. placed between the c.1200 copy of the Book of Curiosities, and the late Oxford example (Pococke 375, dating from 1456) 18 – there is no clear progression. For instance the Red Sea is missing on the Paris map (apparently, chronologically the middle of the three) but it is clearly indicated on the other two. Evidently, fidelity in the copying of such manuscripts cannot be assumed.
The small map shares many features with the Charta, even if its coastal outlines were handled more schematically and with simplified detail. The cartographic overlap applies particularly to Asia, but the two maps are noticeably dissimilar in some other areas. The shapes of Italy, the British Isles and the eastern Mediterranean are unrecognisably different. Neither does the small map repeat the Charta’s massive enlargement of Cyprus and Euboea, although the author’s home island, Sicily, is similarly exaggerated, very noticeably so when compared to Sardinia and Corsica. [It may or may not be coincidental that Sicily is neatly arranged on a single section in the Charta, whereas Cyprus and Euboea are split between two.]
The dating and provenance of the small circular map might not seem to be a matter of great significance. However, the broad similarities between that and the large rectangular Charta prompt further observations. If it could be established that the small map is an 11th-century compilation, then it could not have been compiled by al-Idrīsī, and its presence in the Nuzhat al-mushtāq would therefore be that of a cuckoo in a strange nest. So, should we date that small map 150 years prior to al-Idrīsī, or alternatively to a period after his death? At issue here is our understanding of what should, or should not, be considered to be idrisian characteristics. Karen Pinto states clearly that, in her view, the circular image is “a version of the Idrīsī world map”. 19
Kahlaoui’s recent publication (2018), with its long section on al-Idrīsī, 20 might have been expected to throw light on these issues. Unfortunately the book lacks citations to recent publications, or indeed any mention of the Book of Curiosities (perhaps because of its origin as a PhD dissertation completed ten years earlier). There are passing references to the small circular world map and a pointer to more in the following chapter. However nothing further was found there. 20a
Two paired alternatives for the dating of the circular map will be considered in turn:
A. That the circular map had been an integral part of the original version of the Book of Curiosities
(c.1020-50); or, alternatively, that it was added to the sole surviving example of that work when it was copied
around 1200
These alternatives are debated by Yossef Rapoport and Emilie Savage-Smith in their recent study of the Book of Curiosities. 21 It is not surprising that no clear verdict emerged, particularly when close comparison of the versions found, respectively, in the Book of Curiosities and the Nuzhat al-mushtāq, reveal mutually exclusive toponyms. Some of those tie in with their own texts, some do not.
Furthermore, given the similarities between the circular map and the Charta, we would have to identify which elements reflected the new information resulting officially from Roger’s (but presumably in practice al-Idrīsī’s) geographical survey of 1139-1154. If there was significant overlap between the two, we would have to treat the Charta, not as reflecting al-Idrīsī’s fresh view of the world, but rather as extending and refining the outlines (particularly in the Mediterranean) seen in the version previously included in the Book of Curiosities, with the addition of detailed textual descriptions, and numerous new toponyms.
As Savage-Smith points out, the stakes are high because, as she freely admits, her support of an early date for the circular map’s outlines logically leads to “a major revision of the history of Islamic cartography”. 22 Rapoport’s contrary view cites physical evidence to assert that the map is a later addition, perhaps not just in relation to the original make-up of the Book of Curiosities but also to the structure of the surviving copy from c.1200. That argument would leave al-Idrīsī’s reputation for originality intact.
Karen Pinto offers support for that view, considering al-Idrīsī’s work to belong to “a unique tradition not representative of the bulk of medieval Islamic mapping”, and outstanding “for inserting a heightened cartographic mimesis into the late medieval Islamic mapping repertoire”. 23 If, taking the opposite view, it is still argued that responsibility for the originality of the Charta’s content should be transferred to somebody else, neither their name or lineage is known, nor have any of their maps apparently survived. [Though see The Charta’s possible antecedents.]
B. That the circular map was created by al-Idrīsī as a summary of his large
rectangular map (the Charta); or by somebody else loosely based on that
Because of their broad similarities, the two early survivors of the circular map – in the c.1200 (Book of Curiosities) and the c.1300 (Nuzhat al-mushtāq) – must derive from a common original. Significantly, there is no reference to a circular world map in the Nuzhat al-mushtāq’s Introduction. The obvious interpretation would see this as a summary of idrisian cartography (though, from the significant differences already mentioned between that and the Charta, surely it would not actually have been drawn by him). Hence, the uncalled-for presence of the circular map in the Paris version of al-Idrīsī’s text might perhaps be explained by the need for an index map (of sorts), which would surely have been appreciated by any reader. One of the many problems is that some of the toponyms on the circular world map do not match anything in the sectional maps.
Has one important point been missed? When we look at early Islamic cartography as a whole, surely the Charta stands out from all the others, and for one obvious reason. Like Ptolemy's Geographia, on which al-Idrīsī’s map was supposedly partly based, there is at least partial concern for realistic coastlines, as evidenced by some recognisable continental and island shapes (on which see Contents of the Charta). There was also some attempt to record the relative positions of places, even if the East African coast, which actually trends south, had to be turned through 90 degrees to point east so as to fit into the first ‘climate’ (as already described). In other words, spatial accuracy (or ‘geographic truthfulness’ in Kahlaoui’s term), 24 must have formed part of al-Idrīsī’s remit. In this he was surely exceptional, as Pinto’s recent general survey confirmed.
Earlier Islamic cartographers clearly found that schematic, diagram-like cartographic images were sufficient for their users’ purposes, once the considerable amount of information that could still be conveyed had been decoded. Since such realism cannot be seen anywhere in the Book of Curiosities, is that not sufficient reason, on its own, to challenge the idea that the circular map could have formed an integral part of that work?
Putting all those elements together, I would argue that the circular world map occupies an undeservedly central position in idrisian studies. It is probably best viewed as a corrupted copy of the Charta’s information and hence has no valid place in al-Idrīsī’s history. Given the presumed rectangularity of both the drawing board and the large extant map (on which see later), we are entitled to ask when, how and in particular why a circular format would have been introduced at all. It is not only probably post-idrisian but also a markedly inferior work to the Charta, which should therefore retain its status as a unique and revolutionary leap forward, and be considered instead as al-Idrīsī’s crowning glory.
What image would have appeared on the silver plate?
There are two alternative formats to be discussed and three possible maps involved
We are told of the silver plate’s weight – 134 kilos (roughly equivalent to that of two average people) – but not, apparently its size, beyond that it was of ‘large extent’. 25 Information about large silver plates was not readily available to me but given the metal’s low structural strength there might be doubts about the feasibility of what the text describes.
One un-footnoted modern source described the plate as ‘probably measuring 3.5 x 1.5 metres’. 26 This of course refers to a rectangular shape and the word ‘tablet’ was used in that case rather than ‘disc’. Since the height-width ratio (1 to 2.3) is identical to that of the Miller re-drawing of the Charta, 27 it seems likely that the Charta’s dimensions were applied by mistake to the plate. When considering size, if it was joined up the Charta would indeed have a width of about 3.5 metres. That compares with two large world maps from the Christian West: the mid-13th century Ebstorf (3 metres square on 30 pieces of parchment) and the late-13th century Hereford (diameter 1.3 metres, drawn on a single skin).
There is no question that the text describes the silver plate as circular, since that is the unequivocal meaning of the Arabic word da’ira. However, this is effectively contradicted in other passages in the Nuzhat al-mushtāq’s Introduction that can only be interpreted as references to a rectangular map, which alone would be a faithful reflection of the Charta. Scholarly opinion remains divided on this. One recent experiment has cut across that divide, transforming the Charta’s content into a circular format (as will be described).
The alternative formats will be looked at in turn.
There are two variants of the circular interpretation: on the one hand, that the disc represented a massively enlarged form of the small world map, or, on the other, that it featured the full content of the rectangular Charta but re-configured into a circular format.
If it is agreed that the small circular world map could not have been drawn by al-Idrīsī then neither could it have served as the source for the silver disc. However, in the light of what appears to be the prevailing opinion, or even assumption, that the small world map does indeed represent an official summary of what resulted from the ambitious project of Roger and al-Idrīsī, and is hence roughly contemporary with that, it has to be considered as a candidate source for the silver disc. It is certainly the case that the horizontal climate divisions appear in the same way in both versions. The difference in the geographical outlines may be partly attributable to the distorting effect of the respective projections, but no doubt the difficulty in converting to a circular format would also have played a part.
If indeed the small circular map did serve as the model, we would have to envisage its simplified details being vastly enlarged. If that was the case it would not, as the Introduction insists, have remained a faithful copy of the drawing-board original. Following that same argument, it would presumably have served as a demonstration of royal splendour rather than being concerned with the scientific rigour that clearly motivated the whole project. Hence, in the light of the dismissive comments above, it seems highly unlikely that the small world map, about which, as has already been explained, there is no mention in al-Idrīsī’s text, would have played any part in the preparation of the silver plate.
A recent exercise, which has produced dramatic, tangible results, worked on the assumption that the plate was indeed circular but, instead of using the small circular map, the entire content of the Charta was transformed from a rectangular shape into a circular one. The Factum Foundation first joined up the 70 sections of the rectangular map, aligning the mis-matches at the section boundaries, so as to produce, for the first time, a single, fully coherent image. What they did next is more speculative: reverse engineering the likely contents of the lost circular disc. 28 In ‘Re-creating the lost silver map of al-Idrisi’, they used digital technology to transform an essentially projectionless world map (i.e. one that takes no apparent notice of the inward-curving lines of longitude) into one on a globular projection. The result was then engraved on a two-metre silver disc. Interestingly, that process (‘neither facsimile nor copy’ but rather ’a re-creation’) produces a map that occupies no more than about two-thirds of the disc. This left a large vacant space, whereas the rectangular format extended to the outer margins on all sides. 29
It was when I had just completed writing up my own radical re-interpretation of the silver plate, that I first became aware of this elaborate project to recreate the lost disc. Without wishing to diminish the achievement of ‘re-creating’ the al-Idrīsī map itself, in a ‘restored’ form – and one whose intermediate stages will, I hope, be made widely available – might this exercise not provide support for what I am about to suggest: namely, that the original plate was rectangular not circular?
Three questions deserve answers. First, how could that transformation from rectangle to circle (or, indeed the other way round) have been practically feasible at that scale, with the technology available at the time? Second, why would it have made sense to produce an image designed to be displayed, with its top (i.e. southern) third left blank? And lastly, how would any such transformation fit in with the instructions in the Introduction (partly repeated from earlier) that, once the disc had been made ready, “he [Roger II] had engraved on it a map of the seven climates … according to the version appearing on the drawing board, not differing from it at all and thus following what had been decided there without any variation; finally, artisans transferred this map into a book”? In other words, it was categorically stated that there was a single rectangular master version (on the drawing board) from which identical copies were taken, to be used, first, for the silver plate and, second, in a sectioned version, for the book. The fact that the latter operation was to be carried out by ‘artisans’ presumably emphasises its non-creative nature.
Ahmad gives the size of the individual sheets in the earliest surviving text of the Nuzhat al-mushtāq as 26 x 21 cm,30 but, because the maps fold out, those have twice the width of the text. Hence each map sheet is about 26 cm (height) and 42 cm (width). Since there are seven horizontal climate divisions, that would give a total height of 1.82 metres, while the ten vertical sections when joined would run out at about 4.2 metres. However, there are sizeable blank margins around all sides of the map sections (which would have been trimmed off for any joined-up version) so – assuming that the surviving copies respect the original scale – the calculated size of the Charta (1.82 x 4.2 m.) might need to be reduced by 15%, if not more. Whatever the precise figure, the Charta would surely have been much too large for it to have been drawn out and viewed as a whole on joined vellum skins. 31 It is of course possible that the original, drawing-board map might have been noticeably smaller than that in the Nuzhat al-mushtāq, since it would have been relatively easy – using the square-grid copying method – to have enlarged it, for instance so that there were 70 sections to distribute among the 350 folios, rather, say, than 35.
Altering the projection, as in the Factum Arte exercise, would have been far more challenging. Early map projections are, broadly, defined by the treatment of the lines of longitude, whether parallel to one another or converging on the poles, either straight or curving. Because the Ptolemaic world map was based on the astronomically-determined positions of hundreds of places, it would have been relatively easy in that case to adjust the outlines for different projections. However, the Charta’s underpinning structure was more theoretical than astronomical and so would not have lent itself readily to such manipulation.
At least two valuable benefits could flow from the Factum Arte exercise if versions of their corrected Charta in both their rectangular and circular forms could be made freely available to researchers. In the first place, it would be easier to see how close a match there is between the coastal outlines of the rectangular Charta and the small circular map, once they were in the same projection, and, second, the rectified version of the Charta, which was produced by them as an intermediate stage, would surely reveal, for the first time since the 12th century, what its original appearance might have been. The loss of the silver plate was tragic, since it could otherwise have made the map available for others to copy in its entirety from the outset, thereby gaining the early use it deserved. All in all, the Factum disc is a fitting tribute to al-Idrīsī.
That still leaves uncertainty as to which of the two maps was featured on the silver disc.
Ahmad’s conclusion: “it is likely that the sectional maps were based on the drawing-board map and the silver map”,
side-steps the issue. 32
Kahlaoui, considers that the original form of the map on the drawing board
(or his preferred ‘tabula of coordinates’) was ‘presumably rectangular’. However his comment on the relationship
of the various elements is unconvincing and contradictory:
If we assume, as is apparent from the text, that the circular world map in silver was identical
to the rectangular world map, notably that the exact detailed depictions were used, according to Roger’s order …
then we can assume that the rectangular world map [a mistake for ’tabula of coordinates’?] was not merely a table of
graphic coordinates but rather a rectangular world map.
33
There are undeniable caveats to either argument: the text’s insistence on the unambiguous word ‘disc’ in favour of a circular silver plate on the one hand, and the precise and repeated statements in the Introduction (emphasised in bold in the extract above) about the fact that it was the ‘drawing-board’ map which provided the common image for both the engraved plate and the Charta, on the other. There is certainly no mention of any GIS-like mathematical transformation.
Might the following, admittedly speculative, scenario help to cut that Gordian Knot?
The oldest surviving text of the Nuzhat al-mushtāq dates from around a century and a half after al-Idrīsī’s time, the equivalent of 1870 for us today. How many times might that text have been copied in the interim? 34 By around 1300, what memories might have remained, particularly about the silver plate? Since the Paris example of the Nuzhat al-mushtāq includes the uncalled-for circular map, might that have suggested the disc shape to the copyist who – not appreciating the Charta’s real format from the 70 small sections he was reproducing – altered the stated format by inserting a different word? By 1300, unless the Nuzhat al-mushtāq had been lost and then re-discovered, there could have been a century and a half of repeated re-copying.
None of the map’s other manifestations has survived, given the loss of both the drawing-board version and the silver plate, and each subsequent exemplar of the Nuzhat al-mushtāq would presumably have been replicated from an existing one, not the autograph version. Even if each copy attempted to be a direct facsimile of its model, there would still be inevitable degradation over the ensuing centuries. Conversely, it is hard to see how there could have been any improvement, since that would require a new source, and none seems to have been reported. On that basis, the most authentic modern reconstruction would be a montage of the map sheets in the Paris volume, with any missing or illegible sections supplied from other early exemplars, in a manner that clearly marked them out from the Paris pattern.
It should be noted here that Miller’s reconstruction is misleading in some minor respects. For example, the Paris exemplar repeats toponyms when the entity involved had spanned two sheets, or they extend into the blank margins, whereas Miller, presenting a composite map, not unnaturally includes just a single instance of each. Similarly there are differences in the presentation of what Emilie Savage-Smith described as ‘a richly decorated work of art’. 35 Miller’s version is undoubtedly colourful but the Paris copyist handled that element with more subtlety, for instance in the two blue tones used for the sea and the delicate white outlining.
Because we can see that the maps were not meticulously replicated on the surviving copies which came after 1300, there is no reason to assume they had been wholly faithful up to that time, a comment that could also have applied to the text. Perhaps ‘squaring the circle’ will be thought acceptable here in the face of the limited and contradictory evidence. After all, this radical re-interpretation of the shape and content of the silver plate comes down to a single short Arabic word on the one hand, and a sequence of clear, twice-repeated instructions about fidelity, on the other.
The ‘drawing board’ presents a processual problem. Since it was described as the place where the historical and incoming information was correlated, this would surely have had to be set up in such a way that new details could be readily accommodated, while what was already there might be easily amended or rejected. How else could the editing have been done? That implies a temporary, intermediate medium, perhaps sections of paper, vellum or cloth that could be readily replaced. Yet it was specifically stated that the individual pieces of information were to be ‘traced’ on the wooden surface ‘with iron instruments’ (see the bold passages in ‘The Maps' section). The idea that the content of the map, as it gradually emerged over the 15 years, would have been permanently engraved into the surface makes no apparent sense, yet that is what it appears to say. Altogether perplexing, unless, as has already been suggested, the literal wording of the text (in its surviving version of 150 years later) can reasonably be challenged.
It is disappointing that we know so little about this drawing board, arguably the most important of the Charta’s four manifestations. It was there that the map was compiled, edited and finalised, in order to provide the model for the other three.
Jean-Charles Ducène examined al-Idrīsī’s Islamic sources in some detail, noting in particular his debt to the 9th-century geographer al-Khwārazmī and an Islamic re-working of Ptolemy. 36 There is, however, a difficulty with al-Khwārazmī’s world map, since we do not know what it looked like. It was lost and has had to be recreated from his astronomical tables, notably by S. Razia Ja’fri (1985). 37 Rapoport & Savage-Smith reproduce a circular form of that reconstruction (i.e. with converging meridians) 38 and there is also a rectangular version, set down over a squared grid. The overall appearance is similar to that of the Charta and some of al-Khwārazmī’s coastal outlines approximate to those of the Charta. But it is natural to wonder – given that the re-creation must have depended on single points rather than irregular coastal outlines – how reliably the modern version reflects the lost original. Might knowledge of the Charta’s appearance have influenced the reconstruction, even if subconsciously? Perhaps that was why the Ahmad chapter did not include it, although al-Khwārazmī’s maps of the Nile and Sea of Azov are much reproduced. There is an important point here for al-Idrīsī’s reputation for originality. If al-Khwārazmī’s world map was truly as shown in the reconstruction, it would seriously weaken claims that the Charta was the first to break away from formalised coastlines (on which see What place does the small circular map have in this discussion?).
So as to understand the geometric relationship between the Ptolemaic outlines and those on the al-Khwārazmī and al-Idrīsī maps, might a comparative cartometric analysis be carried out? In the case of Ptolemy and al-Khwārazmī the astronomically-determined coordinates survive. Those could first be contrasted and the result then compared with the positions of the same named points on the Charta. Would such an analysis confirm or contradict any borrowings, and possibly reveal if such astronomical tables, or even fresh observations, formed part of the Charta’s underlying geometry? [Or has such work already been done?]
As an overall model for the Charta, prior to the full availability of the Ptolemaic outlines, 39
there would have been a choice between formulaic T-O-based Christian maps and their Arabic
equivalents on the one hand, or one of the slightly more realistic mappaemundi, on the other. The Charta’s
depictions of the Aegean and Black Sea are strongly reminiscent of those on the Cotton Map (11th or 12th century)
40
which “probably offers the closest copy we have of a Roman map from which the outlines of many
medieval world maps ultimately derive”. 41
How did al-Idrīsī obtain his information?
al-Idrīsī – or, as the flattering text would have it, Roger himself – found earlier Arab geographies inadequate and so “had turned to contemporary experts in the field. Finding these sources similarly unsatisfactory, he set about collecting travellers’ accounts”. 42 In attempting to understand how Roger and/or al-Idrīsī went about the onerous task of compiling their map, it needs to be remembered that most of the information would have come to them in non-graphic form, as written geographical descriptions or oral testimony. It would not have been equivalent to the way that a later cartographer would have collated existing maps. In this case, the main challenge would have been in translating into graphical form disparate sources: presumably geographical theories, clear statements of perhaps questionable reliability, remembered experience, and so on. al-Idrīsī was evidently receptive to fresh information but he cannot be blamed for those occasions when he was misled by those he interviewed. Nevertheless, the resulting map is a repository of the merged geographical knowledge of the Christian and Islamic worlds. It is hard to imagine a single other individual in that period who might have gathered more.
Even if much of the geographical knowledge was collected as part of the field-work of the royal emissaries, as it was claimed, 43 a number of al-Idrīsī’s informants would have arrived at Palermo by sea. Among those must have been some who had direct navigational experience, since the more recognisable coastal outlines on the Charta could only have been provided by sailors. Indeed, al-Idrīsī describes a maritime itinerary around the Peloponnese, and another covering the northern coasts of the Black Sea – both of which must have been carried out on board a ship. 43a However, as Kahlaoui notes, there are more references to merchants among the acknowledged informants than mariners, and sea captains and owners were mentioned only for the Indian Ocean. 44
Kahlaoui also describes how the accompanying text provided a wide range of information from mariners, among which was “a list of sites, with the distances between them and their neighboring coastal sites”, along with “notes about the kind of goods (food and water) available at each location” as well as “tools to aid in the visualization of the outline of the coastline.” 45 But those types of information, reminiscent of a pilot guide, could have been of little assistance to al-Idrīsī in the preparation of his large map.
When a cartographer who had little or no first-hand knowledge himself went about deciding which was preferable from among the competing versions, how did he choose? It was presumably the uneven and unpredictable process of data-gathering for the Charta that resulted in a map which is fairly realistic in some parts and heavily distorted in others. Straight and uncomplicated coastlines, such as those for the north and west of the Iberian Peninsula are well represented, whereas the shape of Italy (which is less recognisable here than on the Ptolemaic maps), as well as the intricacies of the Adriatic and Aegean seas – and indeed the eastern Mediterranean as a whole – could never have been obtained from any navigators’ reports that al-Idrīsī might have received. For millennia, mariners, working with a mental ‘wind compass’, would have observed and memorised the configurations of the coastlines and the wind bearings from one headland or island to the next. It is therefore inconceivable that the Charta’s contorted Italy could have had a maritime origin. In those cases – as one would expect – al-Idrīsī’s sources must have been either purely verbal ones or based on no more than rough sketches.
Parts of the North African coast opposite Sicily are certainly plausible (for example, from the Golfe de Gabès up to Cape Bon and then west to Tunis and ‘mersa al haraz’ [marsacares on the Carte Pisane]) and it is probable that the information in that case was obtained from open-sea crossings reported back to al-Idrīsī by navigators. 46 This could also explain a few similarities between the Charta and the Carte Pisane (the earliest surviving portolan chart of c.1270, which can be seen in a BnF scan), for instance around Tunis. 47 It is difficult to see how al-Idrīsī’s own reported visits to the Maghrib could have provided those outlines. Moreover, the single most notable feature of the portolan charts – indeed their most distinctive coastal signature – namely, the deep Gulf of Sirte, is only partially conveyed on the Charta and is given a similar depth to that of two other, non-existent gulfs. Where the Charta is visibly more advanced than the Carte Pisane is in its portrayal of the Atlantic coasts. Note in particular Great Britain and Iberia.
A few instances of apparent agreement between the Charta and Carte Pisane might suggest borrowings from one or the other. But there are two separate arguments against those possibilities: first, their different respective purposes and, second, the Charta’s limited circulation (to be discussed under The charta’s legacy).
The Charta’s place-names are at their densest around the coasts but they appear to refer to settlements only, not to natural features, though a few rivers and mountain ranges are identified. This would fit in with Roger’s desire to know about 'seas and gulfs', but despite the Charta’s deserved reputation for innovation and thoroughness, it would have been of limited use to a sailor. It is true that what is considered one of the signature features of the portolan charts, namely the placement of the toponyms inside the coastline, is also found on the Charta, but there is a significant difference. Whereas the portolan-chart toponymy winds its way round the coastline in an unbroken sequence, with the names placed at right angles to that – which meant that the user had to rotate the chart to see them the right way up – the Charta’s toponyms are generally placed horizontally, to be read from a single viewpoint, like normal text. Furthermore, the Charta’s solid blue sea would not have allowed depiction of any offshore dangers (such as the Skerki Bank, to the west of Sicily and hence not far from Palermo), which was one of the portolan charts’ primary functions.
Interestingly, Sicily’s fairly regular shape was conveyed realistically, even if the island was massively enlarged. Although at first glance that would appear to flatter al-Idrīsī’s royal patron specifically, some of the other islands are also re-sized, whether magnified or reduced. For example, while Sicily is shown at about four times the size of Sardinia, that, in turn, is three-times larger than Corsica. This is another indication that geographical truth was not the only consideration. By contrast, the portolan charts might have simplified or distorted some features but never those affecting the geometric precision of the relative positions of the termini for pelagic voyages, which underpins their navigational function.
Both the Charta’s purpose and the way its information was gathered were fundamentally different to those of the portolan chart. The density of the inland detail confirms that al-Idrīsī and his patron were focused more on topography than hydrography, on geography not geodetic accuracy, and certainly not on the needs of pelagic sailors. In short, the Charta reflects a landsman’s view.
Roger and al-Idrīsī’s method of information-gathering, by interviewing multiple informants passing through Palermo, was inevitably piecemeal. Common standards could not have been imposed nor any overall linking structure. Hence the unevenness in the quality of the map’s outlines. No amount of ‘improvement’, even if the process had extended for 30 years, could have provided a single geodetic foundation comparable to the portolan charts’ self-correcting framework formed out of multiple pelagic courses.
In the context of knowledge distribution, it is arguable how much importance Roger or al-Idrīsī accorded to the dissemination of their great work. Was the silver plate more than a vanity project, to be shown only to courtiers and high-ranking visitors? There do not seem to be any first-hand descriptions during the seven years it was apparently displayed (perhaps only privately) in the royal palace. It is undeniable that the introductory text mentioned Roger’s personal need for enlightenment rather than any kind of universal elucidation.
As for the role of the Nuzhat al-mushtāq, the intent would presumably have been to preserve the extensive geographical commentary (built up over 15 years) alongside the cartographic segments about which it was expounding. But lacking a graphic index to show the context of each map section, any sense of the cartographic whole would inevitably have been lost. While ‘virtually’ present, the Charta as a whole was effectively invisible. A modern user’s concerns about the apparent impracticality of such an arrangement might thus be misplaced. It seems that for Roger II – though probably not for al-Idrīsī – the text was valued more highly than the map sections. In that interpretation, there would have been no joined-up version of the Charta because it was not necessary for Roger or his court, who could have referred to the silver plate instead.
If anybody had decided that the map was of sufficient interest to be gathered up and presented as a whole, the result has not survived, nor any description of such. Neither, despite the fact that the Nuzhat al-mushtāq was providing commentary on the map, were those sections invariably treated as an integral part of the work by subsequent copyists. Several of the surviving versions contain less than half of the map sheets and others omit them entirely. 48
Despite the determined information-gathering, no apparent effort was made to broadcast the results. Hence it is appropriate to ask what influence the Nuzhat al-mushtāq and the Charta had on subsequent cartography. The cumbersome arrangement of the text must also have been a deterrent to reader engagement. Why, given its ambition to describe the world, was the text not translated out of Arabic into Greek and Latin so as to serve the Christian regions? Perhaps the death of al-Idrīsī’s patron Roger in 1154, the same year that the work was effectively completed, or his own death, seemingly between 1166 and 1175, 49 explains that missed opportunity.
Ducène describes the Nuzhat al-mushtāq as being “widely circulated until the nineteenth century”, but neither he nor Ahmad found evidence for its influence in Arabic circles before 1286, and that was purely textual. The only noted exception was a related idrisian text of 1192, the Rawd al-Faraj (of which just four examples are known and one lost). 50 The first documented use of idrisian cartography by others was not until centuries later. 51
Any argument there might be that the Charta could have influenced the creators of the portolan chart, runs up immediately against the difficulty – probably an insuperable one – that contemporaries must have faced when trying to access the Charta’s information, even assuming they were aware of it. It is generally agreed that al-Idrīsī’s map had (undeservedly) very little impact, if at all, in the Christian world, because of its awkward structure and the barrier of its Arabic script. Ahmad concluded that “if there was any influence of al-Idrīsī in Western Europe, it was only indirect”. 52 But, besides those hindrances, the Nuzhat al-mushtāq (and hence the Charta) does not seem to have been generally accessible until significantly later, since the oldest surviving example dates supposedly from about 1300. Put simply, it seems that nobody between al-Idrīsī's time and 1928 is likely to have seen a joined-up version of al-Idrīsī’s cartographic image as it applied to the Mediterranean.
Prefiguring the multi-sheet, large-scale topographical maps of today, which are divided evenly and arbitrarily according to geographical co-ordinates, the Charta Rogeriana imposed a grid of 70 equally-sized segments. Thus Italy, for example, is spread over four of Miller’s sections: 32, 33, 42 & 43. 53 Contrast that arrangement with the formula adopted by Pietro Vesconte for the oldest surviving portolan atlas (1313). Its logical division on geographical grounds offered separate charts, each with a coherent identity. The Black Sea and Aegean, for example, have their own dedicated sheets, while some later atlases would treat the Adriatic in the same way.
This lack of early copies of the Nuzhat al-mushtāq belies the simplistic assumption that, because of its quality, originality and size, the Charta must have been highly influential. A superficial search for repetition of idrisian outlines, in Western cartography at least, drew a blank. Is it unfair to conclude that by the time the Charta could have become known to Christian scholars and mariners the world had moved on?
A further uncertainty concerns the relationship between the text and map in the Nuzhat al-mushtāq. Broadly speaking, the commentary is focused on the content of a particular map section, even when the entity being described extends into one or more other sections, placed elsewhere in the volume.
It has been proposed, by Yossef Rapoport among others, that the 70 map sections in the volume had a life of their own; indeed, that they might have been created separately rather than resulting from the dissection of a single large original. Some awkwardness at the Charta’s joins implies that taking separate extracts might possibly have been the method used – at least as demonstrated on Miller’s composite re-drawing from the Paris and Oxford exemplars. After all, we are not entitled to impose our own logic or methodology, particularly when the project in question was almost certainly unprecedented.
Here, as in everything else, we suffer in not having access to a contemporary original. As already mentioned, the Paris copyist seems to have carefully contained the toponyms within the sheet boundaries, whereas in the orignal, some would have spread across the divide. In that case, at least, the 70 sections must have been separately drafted (and perhaps adapted), rather than being slavishly copied off as the small relevant segment of a single large map. Hence, one more uncertainty.
Nevertheless, it is hard to see any way that such a large and complex map could have been constructed in that manner. Logic points instead to the reverse process, that the map was compiled steadily over a number of years as a single entity (the drawing-board master) and then copied individually section by section into the volume at the end. Kahlaoui endorses that interpretation: “The 70 sectional maps were a subsequent dissection, or cutting into 70 pieces, of one original rectangular world map that was based on mathematical cartography”. 54
The lines that divide the map into ten vertical divisions have, not unnaturally, been interpreted as longitude divisions, although there is, as far as I am aware, no direct confirmation of that. However, it is certainly possible that each vertical division was intended to cover 10 degrees of longitude. The Mediterranean ranges over approximately 50 degrees and, since that sea occupies roughly half of the map’s width, that would be consistent with a system divided into 10 units each representing 10 degrees of longitude – starting with a prime meridian at the entrance to that sea, where the then-known world was considered to have started. 55
The seven climates, with the top of the first one representing the equator, conveyed latitudes. They also defined the height of each map image, which is spread across the verso and recto of successive folios, giving a ‘landscape’ appearance. Unless much restoration work has been done, the width of each map section has allowed equivalent blank margins on all sides. This suggest that the vertical lines might have had a non-geographical, more prosaic purpose. Because text is malleable and can be made to fit the space, it would have been natural for the volume’s format to be dictated by the size of the map sections. A division into ten might have been chosen so as to match the width requirement for each two-page map. It may be significant that the circular world map, which shows the latitude climate lines, does not replicate those vertical divisions. Furthermore, in the Paris manuscript the map sections along the top (south) row carry no indication that their width is defined by longitude. In other words, it seems likely that the ten vertical divisions have no geographical significance but were added precisely so as to guide a dissection process that produced sections of the appropriate width. 56
For a complete picture of al-Idrīsī’s production we would have to compare each of the Charta’s map sections with its related textual description. Might some at least of the dissimilarities between text and map result from the assumption that al-Idrīsī was given a free hand after Roger’s death in 1154?
A table that compared the toponyms of each – including the circular map as well – would be valuable. Ducène describes an abridged version of the Nuzhat al-mushtāq, which “contained 2335 toponyms, listed in order from west to east. Each toponym was assigned to one of the seven climates and was briefly described”. 57
The Charta Rogeriana has potential significance for the discussion about portolan chart origins. Two related questions can sensibly be put: does the Charta include information that might be traceable to a marine chart or, conversely, might it have influenced, or even inspired, the creator(s) of the portolan chart? In which direction might any potential borrowing have gone? In other words, which came first? Hence the Charta’s date – 1154 or perhaps 1158, or even a little later still 58 – has to be considered as crucial evidence in our attempt to understand the emergence of the portolan chart.
There is no evidence that al-Idrīsī made use of a portolan chart. He was not a sailor and there is no reason to suppose that he would have been able to appreciate the worth of a revolutionary maritime chart. But, had he been shown one, we can be confident that its user would have vouched for the accuracy of its coastlines and hence relevance for Roger’s project.
It is always dangerous to argue from the absence of information but, in this case it seems justifiable because it is inconceivable that al-Idrīsī’s meticulous and well-documented, 15-year search for first-hand geographical accounts would have failed to discover any nascent portolan chart that might then have existed. My failure to identify any traces of such borrowing, ties in with the majority opinion among researchers today. Kahlaoui, for example, concludes that "it is extremely difficult to argue for the Idrisian use of early portolan maps". 59
The significance of this cartographic silence – which has not apparently been noted before – provides robust support for a portolan chart origin date of no earlier than 1154.
In some areas the Charta represents the best depiction of the Mediterranean coastlines at that time, especially in those parts where sailors can be presumed as the source. But the arrival of the portolan chart no more than a century later (and perhaps less than 50 years afterwards) would have rendered most of that information obsolete, had it ever been generally available. As a result, the Charta became a cul de sac as far as the coastal cartography of the Mediterranean was concerned, though that comment should not be applied to the regions beyond that. Nevertheless, the Charta does provide an overview of the state of geographical knowledge in Islamic circles at a period probably shortly before the charts emerged.
S. Maqbul Ahmad, ‘Cartography of al-Sharif al-Idrīsī’, in J.B. Harley & David Woodward (eds), The History of Cartography, Volume 2, Book One, Cartography in the Traditional Islamic and South Asian Societies (Chicago University Press, 1992), pp.156-74
Jean-Charles Ducène (2018). ‘al-Idrīsī, Abū 'Abdallāh’, in: Kate Fleet (et al., eds) The Encyclopaedia of Islam. Three (Brill: Leiden & Boston, 2018), pp.91-99 [the repeated references are to this work rather than the next]
Jean-Charles Ducène (2018). L'Europe et les géographes Arabes, Paris, 2018 (especially pp.196-230)
Jean-Charles Ducène. ‘Les coordonnées géographiques de la carte manuscrite d’al-Idrīsī’, Der Islam 86 (2009): 271–85
Tarek Kahlaoui. Creating the Mediterranean: Maps and the Islamic Imagination (Brill, 2018)
Karen C. Pinto. Medieval Islamic Maps: an Exploration (University of Chicago Press, 2016)
Yossef Rapoport. Islamic Maps (Oxford: Bodleian Library, 2019)
Yossef Rapoport & Emilie Savage-Smith. Lost maps of the Caliphs: Drawing the world in eleventh-century Cairo (University of Chicago Press, 2018)
Annie Vernay-Nouri, ‘Réexamen de la tradition manuscrite d’al-Idrīsī’, Journal of Islamic Manuscripts, 6 (2015)
Building on the chapter by S. Maqbul Ahmad in The History of Cartography, 2:1 (1992) I have also benefited hugely from Jean-Charles Ducène’s revisionist entry in The Encyclopaedia of Islam (2018) (on which see also its extensive bibliography) and from Yossef Rapoport and Emilie Savage-Smith’s Lost maps of the Caliphs: Drawing the world in eleventh-century Cairo (2018).
I have received significant help for this essay from Yossef (Yossi) Rapoport, Queen Mary University of London, and want to acknowledge his firm guiding hand with issues in the history of Islamic cartography, and the generosity with which he has shared information. I am also very grateful to Professor Jean-Charles Ducène, École Pratique des Hautes Études, Paris, for his authoritative comments.
However, the responsibility for the various conclusions (some, no doubt controversial) remains with me.
1. On the Processual Approach see Matthew Edney’s blog entry.
2. S. Maqbul Ahmad, ‘Cartography of al-Sharif al-Idrīsī’, in J.B Harley & David Woodward (eds), The History of Cartography, Volume 2, Book One, Cartography in the Traditional Islamic and South Asian Societies (Chicago University Press, 1992), pp.156-74, at p.156.
3. What was known of al-Idrīsī and his maps – for example from Ahmad, 1992 – has been significantly revised by, for instance, Alloua Amara and Annlies Nef, ‘al-Idrīsī et les Hammūdides de Sicile: nouvelles données biographiques sur l’auteur du Livre de Roger’, Arabica, vol. 48 (2001), pp.121-7 (a reference I owe to Yossef Rapoport); and Jean-Charles Ducène in his article, ‘al-Idrīsī, Abū 'Abdallāh’, in: Kate Fleet (et al., eds) The Encyclopaedia of Islam. Three (Brill: Leiden & Boston, 2018), pp.91-99. This reference as well came via Yossef Rapoport, who kindly shared with me a pre-publication draft of his own chapter, ‘The grid of al-Sharif al-Idrīsī’, now published in Islamic Maps (Bodleian Library, Oxford, 2019).
4. Ducène, 2018 p.94.
5. Annie Vernay-Nouri, ‘Réexamen de la tradition manuscrite d’al-Idrīsī’, Journal of Islamic Manuscripts, 6 (2015): 335-50 [a reference I owe to Victor de Castro Leon].
6. Annie Vernay-Nouri, 2015 p.347. In an earlier article she had compared the map’s toponymy in four examples of the Nuzhat (those in Paris, Istanbul, and the two in Oxford) - see Annie Vernay-Nouri, ‘Tradition manuscrite et toponymie dans la cartographie du Maghreb: étude de quatre manuscrits du Nuzhat al-mushtāq fi’khtirāq al-'āfāq d’al-Idrisi’, in L’onomastica africana: congresso della Société du Maghreb préhistorique antique et médiéval (Sandhi: Ortacesus, 2012), 171-77. [I owe this reference to Victor de Castro Leon].
7. Miller described his version as ‘Wiederhergestellt’ (Restored) and there are some justified doubts about its veracity.
8. But note that in the BnF copy most of the top border (i.e. the southern half-sheets) has been trimmed away, and the final, north-east corner section, which is entirely sea, is omitted, hence the numeration starts at 2; so add one to the Miller numbers. The Bodleian example has also been digitised by Factum Arte, on which see their commentary.
9. Ducène, 2018 pp.94-5.
9a. A 2021 essay by Alfred Hiatt, attempting “a summation of what is currently known about al-Idrisi and the Nuzhat al-mushtaq”, is sceptical about the information in the Nuzhat’s Preface. For example he considers that “the often-repeated claim that it [silver disc] was destroyed in an uprising of 1161 is in fact no more than a conjecture of the learned nineteenth-century historian, Michele Amari”. ‘Geography at the Crossroads: The Nuzhat al-mushtaq fi ikhtiraq al-afaq of al-Idrisi’, in: Alfred Hiatt (ed.), Cartography between Christian Europe and the Arabic-Islamic World, 1100-1500: Divergent Traditions. Series: Maps, Spaces, Cultures, Volume: 3 (Brill, 2021), pp.113-36, at p.120.
10. Ducène, 2018 p.92, with my emphasis.
11. Tarek Kahlaoui, Creating the Mediterranean: Maps and the Islamic Imagination (Brill, 2018), translates this as ‘tabula of coordinates’, p.153, but Jean-Charles Ducène (personal communication 11 February 2020) confirms that this must mean a drawing board.
12. Kahlaoui has ‘they should follow its form and shape’, p.153.
13. Ahmad, 1992 p.159, again with my emphasis.
14. For the effects of that in later periods, see four versions contrasted by Ahmad, 1992 p.161.
15. Yossef Rapoport & Emilie Savage-Smith. Lost maps of the Caliphs: Drawing the world in eleventh-century Cairo (University of Chicago Press, 2018) pp.19 & 24.
16. For an enlargeable scan, see Wikipedia. Also Rapoport & Savage-Smith, 2018 pp.24-27, Plates 3 & 4 (for direct comparison of the two Oxford manuscripts).
17. See the Gallica scan (folios 3v-4r).
18. For an illustration see Ahmad, 1992, colour plate 11. The dating of the Bodleian volume is not certain, and 1553 is also cited – see Ahmad p.174.
19. Karen C. Pinto. Medieval Islamic Maps: an Exploration (University of Chicago Press, 2016), p.165.
20. Kahlaoui, 2018 pp.142-78.
20a. There is discussion about the source and dating of the circular world map in a 2008 article by Tarek Kahlaoui which I had missed. He concludes there that, “at best, the circular world map would not be part of the original work and would have been included in the thirteenth century copy [of al- Idrisi’s work]. It is also possible that it was added even after the manuscript was finished in the thirteenth century”: ‘Towards reconstructing the Muqaddimah following Ibn Khaldun's reading of the Idrisian text and maps’, Journal of North African Studies (special issue on Ibn Khaldoun), 13, 3 (2008): 293-307 (at pp. 301-2. I do not have his 2018 book to hand but he points out that he had indeed discussed the ‘Book of Curiosities’ in some detail (pp. 119-41). I apologise for those regrettable oversights. {This sentence added 18 April 2022 }.
21. Rapoport & Savage-Smith, 2018 pp.24-7.
22. Rapoport & Savage-Smith, 2018 p.27.
23. Pinto, 2016 p.25.
24. Kahlaoui, 2018 p.147.
25. On this see Tadeusz Lewicki, ’A propos de la genèse du Nuzhat al-mushtaq...’, Studi Magrebini, I (1966): 41-55.
26. Leo Bagrow (revised and enlarged by R.A. Skelton), History of Cartography (1963), p.57.
27. And likewise my own calculation – see the following section: Would the silver plate have been circular or rectangular?
28. ‘Re-creating the lost silver map of al-Idrisi’ (a film by Oscar Parasiego, with Factum-Arte, July 2019).
29. The Factum Foundation re-creation had a central place in the ‘Talking Map’ exhibition at the Bodleian Library, July 2019-March 2020.
30. Ahmad, 1992 p.173.
31. Or paper, which, while it was used for the surviving copies, may not have been readily available in the 12th century.
32. Ahmad, 1992 p.159.
33. Kahlaoui, 2018 pp.153-5.
34. The amount of degradation is clear from comparing the shape and alignment of Sicily and Sardinia on the Paris (c.1300 – ff.203-4) and Oxford (1456? – go to section 33) versions. Evidently tracing was not involved, and much of Factum Arte’s work consisted in undoing mis-matches between the original sections. It would not have been surprising if the copyists treated the maps more as pictures than scientific diagrams.
35. See ‘Cartography’, in: Peregrine Horden & Sharon Kinoshita (eds) A Companion to Mediterranean History, ed. (Oxford: Wiley-Blackwell, 2014), p.191.
36. Ducène, 2018 pp.93-5.
37. Ahmad, 1992 p.163, note 40
38. Rapoport & Savage-Smith, 2018 p.94.
39. Although the Charta’s horizontal division into climate zones and its vertical dividing lines show some knowledge of Ptolemy’s writings, see Ahmad, 1992 pp.156-7.
40. British Library, Cotton MS. Tiberius B.V. f.56v.
41. P.D.A. Harvey, Medieval Maps (British Library, 1991), p.25, Plate 19.
42. Ducène, 2018 p.92; see also Ahmad, 1992 p.159
43. Although Ducène, 2018 p.94, questions whether those were ever sent out.
43a. Jean-Charles Ducène, L'Europe et les géographes Arabes (Paris, 2018), pp.202, 223-4.
44. Kahlaoui, 2018 pp.149-50.
45. Kahlaoui, 2018, between pp.142-67.
46. This paragraph and the following section are repeated from the long essay: ‘Mediterranean portolan charts: their origin in the mental maps of medieval sailors, their function and their early development’ (January 2021).
47. I owe this suggestion to Richard Pflederer.
48. Ahmad, 1992 pp.173-4; even the Paris exemplar lacks three of the map sheets along the bottom, i.e. north side, but that might have happened during the volume's eventful life, as described by Annie Vernay-Nouri, ‘Réexamen de la tradition manuscrite d’al-Idrīsī’, Journal of Islamic Manuscripts, 6 (2015). It is of course possible that, because of their attractiveness, some map sheets might have been removed from the volume.
49. Ducène, 2018 p.92.
50. Jean-Charles Ducène, personal communication 11 February 2020; Ahmad, pp.163-7.
51. First mentions – 15th century: Ahmad, 1992 p.170; 16th century: Ducène, 2018 p.96, noting also that idrisian outlines were used in the portolan atlases of al-Sharafi, 1551 and 1571. The suggestion that there might be idrisian elements in the eastern half of Vesconte’s world map (c.1320) and the Catalan Atlas (1375) are unconvincing.
52. Ahmad, 1992 p.172.
53. Ahmad, 1992 p.162, Fig.7.6.
54. Kahlaoui 2018, p.155; on this see also the Factum Foundation commentary, which gives examples of how the sometimes ill-matching sheets had to be rectified.
55. Jean-Charles Ducène, ‘Les coordonnées géographiques de la carte manuscrite d’al-Idrīsī’, Der Islam 86 (2009): 271–85 – a work I have not yet been able to see.
56. Even if the vertical divisions had carried longitude numbers in versions of the complete map, those could not have been easily marked on the book sections since they would have fallen along the outer edges of each. There are a number of latitude and longitude figures in the Paris MS, written in black or red in standard (eastern) Arabic script, whereas the manuscript itself is in the western Maghribi script. However, those are later additions, not just to al-Idrīsī's original but to that manuscript as well (I owe those observations to Fateme Savadi).
57. Ducène, 2018 p.96. It is possible that one of the 19th-century commentators has already listed the toponyms, for example, La première géographie de l’Occident, translated from the Arabic by Amédée Jaubert, revised by Henri Bresc and Annliese Nef, Paris 1999 – not available in the British Library.
58. Ducène, 2018 pp.95-6.
59. Kahlaoui, 2018 pp.165-6.