‘Alī al-Sharafīʼs 1551 Atlas: A Construct Full of Riddles

Abstract

This paper examines the practices of translation in the 1551 Atlas of ‘Alī al-Sharafī of Sfax (d. after 1579). Drawing on conceptual frameworks from translation studies, linguistics, and other disciplines of textual and historical analysis, we argue that the iconographic and textual elements of the atlas –such as calendrical tables, qibla charts, knot patterns in frames, and depictions of flags and banners in sectional charts– reflect a quest for ways to culturally adapt map-making practices that were shared across the Mediterranean realm. As such, the product in question cannot be fully understood through the lens of traditional dichotomies such as Islamic vs. Christian or European vs. Middle Eastern or North African, as its cultural references and semiotic repertoire extend beyond subregional and confessional demarcations.

*This research is part of the research project ‘Mediterranean Nautical Cartography in Arabic and Ottoman Turkish: Islands or Gateways of Knowledge in the Sea of Transcultural and Translinguistic Translation Processes?’ directed by Sonja Brentjes at the Max Planck Institute for the History of Science. We would like to thank Sonja Brentjes for her suggestions in the elaboration of this paper.

In the last decades, the study of map-making has been moving away from positivist assumptions about representational accuracy and objectivity. The critical cartography movement spearheaded by John B. Harley sought to integrate elements from post-structuralist textual analysis into the study of maps.¹ While these methodologies have remained rather marginal in the study of non-Western map-making, recently authors have made attempts to apply them to the Islamicate archive.² However, traditional taxonomies like East and West, Europe and Islam continue to be applied to map-making. Such dichotomies cannot adequately account for what is the most salient feature of the material we are concerned with: while Arabic and Ottoman maps might seem at first glance classifiable as “Islamic” products, a careful analysis brings to light the commonalities that link them with the map-making traditions of the northern edge of the Mediterranean. The corpus comprising Italian, Majorcan, French, Portuguese, Spanish, Arabic, Ottoman Turkish, and Greek sources reflects the shared nature of the cultural, economic, political, and military spaces that constituted the late medieval and early modern Mediterranean realm.

To demonstrate this, we will focus on an atlas produced in 1551 by ‘Alī al-Sharafī (d. after 1579) from Sfax. We will show how different visual and textual elements in it can be considered as results of translation. To evaluate the presence of translation practices in these elements, a combination of terms and methods investigating translation as process or result proved helpful: 1) Eugene Nidaʼs linguistic notion of ‘dynamic equivalence’, which means that translation should not just be understood as a substitution of semantically equivalent terms but rather as a quest to find in the target language concepts that are contextually equivalent to those in the source language.³ 2) The concept of ‘oblique translation’ defined by Jean-Paul Vinay and Jean Darbelnet as the search for parallel concepts, rather than simply parallel categories (or words), between the source and target languages, meaning that metalinguistic elements are always involved in translation.⁴ 3) The differentiation between source- and target-oriented translation, appropriated from historical studies, helped us to tackle the complicated question of the atlasʼs functional orientations. The former is understood as a mode of translation where the properties of the source language are preserved even if they violate the properties of the target language. The latter assimilates the linguistic and content features of a source text to the syntax, semantics, and metalinguistic knowledge horizons of the target audiences and their values. 4) On this basis, we examine how, through these processes, the atlas attests to the existence of a semiotic lingua franca, perhaps similar to the well-known case of the lingua franca that was spoken across the western and central Mediterranean in the Early Modern period. As Jocelyne Dakhlia has shown, the latter was primarily a trading language that changed significantly from one period and region to another. While it was shared by diverse groups of interconnected communities, it was neither a creole, nor the mother tongue of a specific group, nor was it the sterile biological hybrid result of mixing languages. It was a transitory, variable, limited, unwritten, and non-heritable contact language used in the Mediterranean Sea and beyond its borders.⁵ Likewise, the semiotic repertoire of our object of study can be considered a sign of the cultural complexity of map-making traditions in the early modern Mediterranean.

1 Al-Sharafī and the 1551 Atlas

Born in Sfax to a family of possibly Andalusi origin, ‘Alī al-Sharafī worked in Qayrawan⁶ in the later years of his life.⁷ Ruled by the Hafsid dynasty (1223–1574) since the early thirteenth century, Ifriqiya (roughly identified with modern-day Tunisia) was then disputed by the Habsburgs, the Portuguese, and the Ottomans, who would finally take control of the territory in 1574.⁸ Ifriqiya was at the crossroads of heavy migration flows, with Moriscos and Jews fleeing into its territory after various waves of expulsion from Spain and Portugal. It also experienced a continuous flow of corsairs, pirates, merchants, and other kinds of people.⁹ Portolan charts in Arabic have been extant since the fifteenth century. The knowledge of such map-making practices and the arrival of charts from the northern Mediterranean are also documented in written Arabic sources of the time. Refugees from the Iberian Peninsula, converts from Majorca, Sicily, Italy, and southern France, together with merchants, diplomats, and captives could have contributed to the dissemination of sea charts and the knowledge related to them.¹⁰

Al-Sharafī’s surviving output consists of two atlases, dated respectively 1551¹¹ and 1571,¹² and a world map dated 1579.¹³ Most of the existing scholarship has treated al-Sharafī’s work as derivative, dismissing it as a copy of European or, at best, Ottoman map-making.¹⁴ Some have even reproached it for excluding the New World at a time when maps from both ends of the Mediterranean already included it.¹⁵ In contrast, Monica Herrera Casais has taken a less positivistic approach to al-Sharafī’s work, analysing his 1571 atlas as a window onto the multi-ethnic composition of North African port towns and of the intellectual exchanges between Moriscos, Jewish, and Christian newcomers and the long-established Berber and Arab Muslim communities of the region.¹⁶ Building on this latter approach, we propose to examine al-Sharafī’s 1551 atlas as a product of cultural translation, broadly defined.

2 Practices of Translation in the 1551 Atlas

One of the most noticeable features of medieval and early modern map-making is the uniqueness of each product. As such, when we speak of translation practices applied to maps, this should not be taken to mean that any given map is simply an identical copy of another, with only its textual elements (toponyms, inscriptions, and calendrical tables) translated literally into another language. Our understanding of translation, as mentioned above, transcends in this case the merely linguistic realm and encompasses a broader range of semiotic elements. However, this does not mean either that any given element in a target map, so to speak, can be said to be the equivalent of a similar element in a source map. Nor can all the elements in a map be said to have been necessarily inspired only by other maps. Our studies rather show that some elements in a translated chart or atlas are better understood as an ‘adaptation’ or a ‘creative reinterpretation’ of specific cultural practices in the production and reproduction of cartographic and geographic knowledge. The material evidence for this claim is found in the results of artistic and symbolic practices and their preferences in the areas of religion, manuscript production, architecture, court culture, and naval and military equipment.

2.1 Textual Elements

Held today at the National Library of France, this atlas contains fourteen folios: a statement of authorship, a calendrical table of lunar mansions, a qibla map,¹⁷ a circular world map, seven sea charts, a circular table with shadow lengths for the daily prayers, and two textual tables for agricultural and climatic prognostication.¹⁸ Its textual elements possess two formal properties. On the one hand, they serve as a second layer of framing (the first one being visual in nature, as will be discussed below), by setting the cartographic elements within a multi-dimensional space of verbal, written knowledge. Al-Sharafī constructed this space by drawing from geographic, astronomical, and meteorological sources. A comparison with other extant atlases from the Mediterranean shows that al-Sharafī clearly made his own choices and did not simply reproduce previously established practices. This means he accepted the format of a naval atlas that combined textual and visual elements but organized its structure and content in an individual manner. In this way, he managed to combine information and a form of layout that crossed cultural and epistemic boundaries: the so-called anwā’ texts (see below), geographical works, naval atlases, timekeeping instructions for religious purposes, and agricultural and prognostic calendars.

The most prominent source explicitly acknowledged by al-Sharafī is the geography text Nuzhat al-mushtāq fī ikhtirāq al-āfāq by al-Sharīf al-Idrīsī (d. 1165). As it was one of the classics of geographic knowledge and map-making in Arabic, it is hardly surprising that al-Sharafī would quote from it as an authority. While more rigorous philological work needs to be done to determine the depth of al-Sharafī’s actual knowledge of Idrīsī, it is clear that he took information from this source. On the upper side of the folio featuring the circular world map, al-Sharafī says:

You [should] know that the Earth is round like a sphere and the Water is glued to the [whole] Earth and naturally stands on it. It is divided into two halves by the equator which (extends) from East to West. This is the longitude of the Earth. The northern quarter of the Earth is inhabited, whereas the rest of it is empty without people, due to the intense cold and the frost. Also, the southern quarter is uninhabited due to the intense heat and the passage of the Sun. The Encompassing Sea encircles half of the Earth all around uninterrupted in a connected enclosure that girdles it like a belt; only half of it is visible like an egg submerged in the water. This inhabited quarter was divided by the scholars (‛ulamā’) into seven climates (aqālīm) with their seas, as was mentioned by the author of the Nuzhat al-mushtāq fī ikhtirāq al-āfāq and by Ibn al-‘Aṭṭār in his [work] Ikhtirāq al-aqṭār [that you should] consult – God is the Wisest.¹⁹

Although this quotation is based on clauses taken from the Nuzhat al-mushtāq, they do not appear in an uninterrupted sequence in al-Idrīsī’s text, but are rather scattered over various sections.²⁰ It is beyond the scope of this study to analyse in detail all the occurrences of the clauses that conform to al-Sharafī’s quotation.²¹ However, it seems clear that al-Sharafī introduces his own ellipses here to make al-Idrīsīʼs quotation fit into the context of the atlas. This suggests that al-Sharafī followed a widespread practice in Islamicate contexts that consisted in fusing different sources to produce a new text or a new version of an older text.

The information in the calendrical, meteorological, and agricultural tables of the atlas draws from the tradition of anwā’ in the Muslim world, as well as Arabic translations of agricultural texts of pre-Islamic origin. Since around the eighth and ninth centuries, this term had been connected to the idea of dividing the heavens into twenty-eight lunar stations.²² While many anwā’ works reproduce a similar set of standardized information, the wording in al-Sharafī’s tables bears enough resemblance to a number of Andalusi sources to suggest that he might have relied on them for his works. These are the Calendar of Córdoba and the Kitāb al-anwā’ by ‘Arīb ibn Sa‘īd (d. 980–81), the anonymous Risāla fī awqāt al-sana (thirteenth century?), and the Risāla fī l-anwā’ attributed to the Maghribi scholar Ibn al-Bannā’ (d. 1320).²³ However, in his tables, al-Sharafī did not limit himself to reproducing the information contained in these sources. He rather adapted it to what seems to be the Maghribi equivalent of Andalusi (and in some cases standard Arabic) terms to create a format of diagrams and tables proper to atlases and charts.²⁴ For example, in a section describing fruits and vegetables, the word for pears, which in Andalusi sources appears as kummathrà, is substituted by ijāṣṣ.²⁵ The same occurs in a section dealing with climatological phenomena: the climate period known in other sources as ayyām/layālī al-‛ajūz and al-Samā’im appears in al-Sharafī’s work as layālī al-ḥusūm. Also, al-Sharafī incorporated other periods and festivities that belong to a Berber or local context such as ayyām al-‛amūd (13–15 July), yawm al-ṣadama (8 October), or yawm ḥarth Ādam (17 October). These references do not appear in other calendrical books or any written sources known to us. Thus, al-Sharafī (or his ancestor) sought to incorporate and adapt, via a search for cultural equivalences, the textual information of Maghribi as well as Andalusi calendars into a monthly structure of three tables and one circular diagram.²⁶

In the cases discussed above, the search for equivalence took place primarily at a lexical level. But the atlas also contains examples in which the equivalence was established at the level of structure and symbolism. This corresponds to the metalinguistic parallelism that is captured in the concept of ‘oblique translation’. One example is the choice of location and layout for the information about the maker(s), title, date, patron(s), or place of production. In Italian, Majorcan, or French atlases this information either appears on the border of a sectional chart or a calendar or is missing. In the former case, it offers a set of data which often includes the name of the map-maker, his origin, the city where the atlas or chart was composed, and the date of its composition.²⁷ But in an atlas there is no fixed place for this kind of information to appear. It can be added to the first folio, the last folio, or some folio between the two. Only after the introduction of title pages did it become more regularly provided on the first or second page. This change is visible in a few atlases produced during the sixteenth and seventeenth centuries. Remarkably, in his statement of authorship al-Sharafī adapts the structure of an early modern title page, fusing it with the conventions used in Islamic manuscripts, including Qur’anic quotations, calligraphy, terminology related to Islamic dating, and extended praises to the divine:

Praise be to God, eternal blessing and peace be upon our Lord Muḥammad, his messenger. This atlas (ṭabla) was made by the servant [of God] who is in need of him and cannot dispense with him, ‘Alī b. Aḥmad al-Sharafī, al-Ṣafāqusī [= of Sfax] – may God favour him. This [atlas] was finished the day of Thursday, the first day of the month of Ramaḍān of the year 958 (2 September 1551).²⁸

The Muslim standard for placing and formatting this kind of information is the colophon at the end of a text, whether newly composed or copied. Al-Sharafī moved it to the front, taking with him the formulas praising God and the Prophet that normally precede the colophon. The phrasing of the passage that corresponds to the standard colophon conforms with what scribes or students usually provide after copying a text. While frontispieces and title pages are also known from Islamic manuscripts, in particular when they were produced for courtly patrons or institutions, the form al-Sharafī chose for his atlases does not reproduce them, but rather alludes to patterns used in early modern printed books. Although the opposite half of al-Sharafīʼs title page has been cut out by an earlier owner or reader, it is not improbable that it contained an image, as is often the case in early modern printed title pages. The calligraphy of the extant half of al-Sharafī’s title page reflects, on the other hand, another step to integrate the atlas into a local Muslim experience of public religious art – the art of tombstone configuration. Extant tombstones from Sfax from the early modern period show clearly similar properties in their calligraphy and coverage of the entire surface.

Another example may be identified in al-Sharafī’s privileging of poetic forms in textual situations, where similar kinds of sources use standard prose. This can be regarded as a minor case of ‘oblique’ practice because these instances are not marked as poetry but appear as integrated elements of a prose text and thus violate the standard rules of prose. As is well known, poetry was used as a didactic medium throughout the Muslim world given its mnemotechnic qualities. Calendrical information was presented in such a format too.²⁹ Specific features of this mode of expression seeped into al-Sharafī’s text, where we can see, in a number of places, poetic licence in addition to the rhyme. One of them is the deletion of the final syllable of a word to fit the metric of the verse. An example is the circular table of the shadow lengths for the daily prayers. In it, al-Sharafī deletes the final syllable of words in the dual -ni, thus writing taṣilā (‘both of them arrive’), qadamay (‘two feet’), thulthay (‘two thirds’), and ithnay (‘two’) instead of taṣilāni, qadamayni, thulthayni and ithnayni.³⁰ A more traditional way of interpreting such deviations from prose norms is to see them as an expression of al-Sharafī’s limited linguistic skills and a lack of care in executing the transfer from poetry to prose.

A third challenging textual phenomenon is contained in the transliteration of the toponyms. Among them are names like Fīnizīā (Venice), al-Nāṭūlī/al-Nāḍūrī (Anatolia), al-Rūmīlī (Rumelia), al-Būliya (Apulia), Anklatīra (England), al-Qanabriya (Calabria), Lāwalūnā/Lāwalūna/Awalūna (La Valona), Utrantu (Otranto), Nābuli (Naples), Ṭurṭūsha (Tortosa), or Barshīnūna (Barcelona). Instead of using the standard Arabic renditions of these places, al-Sharafī presents an approximate phonetic transliteration in Arabic letters. This suggests that he worked either with an Italian model, from which he adapted the names according to phonetic approximation, or with people speaking a Romance language, or possibly both. An example of the first interpretation is Anklatīra, with the second exemplified by Fīnizīā. In the second case, the term is identified as Bunduqiya in an explanatory subclause, which indisputably shows that he was relying on an older Arabic source rather than adhering to contemporary Ottoman usage. This is a rare explicit verbal instance of cultural translation. The entire set of phonetic transliteration needs a substantial linguistic analysis. It is by no means particular to al-Sharafī. It also appears in other Arabic portolan charts and in later Ottoman translations of atlases brought as gifts or commodities from various map-making and print centres in Europe.³¹ This seems to be a phenomenon similar to what Gideon Toury has called ‘source-oriented translation’, whereby the properties of the source language are preserved even if they violate the properties of the target language.³² By privileging the phonetic properties of the Latin (or Romance) words over either the standard Arabic rendition of these names or adherence to the rules of Arabic orthography, our author chose to preserve the foreign nature of the terms instead of domesticating them.

However, the preservation of the Indo-European orthographic and phonetic properties of the toponyms, as well as of terms like ṭabla (tabula) or qunbāṣ (compasso), could also attest to the prevalence of the Mediterranean lingua franca (also known as Sabir), a Romance-based contact language used in different parts of the Mediterranean region.³³ Many questions regarding this language remain unresolved: it is hard to determine how geographically widespread its use was or how many people spoke it. More importantly, the sociological portrait of its speakers is not completely clear either: while earlier scholarship worked under the assumption that it was used as a trade language, the surviving documentation does not support this hypothesis.³⁴ Jocelyne Dakhlia has found evidence that it was often spoken in locations far removed from the coastal areas due to the phenomenon of captivity.³⁵ It is thus not completely clear whether its use went beyond the western Mediterranean realm. The many unresolved questions surrounding this issue and the gaps in the documentation make it difficult to determine the extent to which the toponyms in al-Sharafī’s atlas indeed coincide with the names of these places in the lingua franca. This is therefore a hypothesis that requires further investigation.

2.2 Iconographic Elements

The charts of the atlas are adorned with frames featuring a wide variety of knots. Their patterns are the most immediately noticeable decorative elements of the atlas as a whole. They strongly resemble decorative norms used primarily in manuscripts of Andalusi and North African Qur’ans and other religious texts, including legal works from the Mālikī school.³⁶ Among the extant copies of this kind are Qur’ans made in Tunis – that is, in the vicinity of Sfax (see Fig. 13.1).³⁷ The similarity between the 1551 atlas and religious texts is accentuated by the atlas’s 20 × 25 cm format, which corresponds to that of medieval Maghribi Qur’ans. It is clear that the atlas was designed to mirror the aesthetics of the Muslim Holy Book.³⁸ This is a remarkable decision, as maps made in Islamicate societies often abstain from using explicit religious symbolism beyond representations of the Ka‘ba and occasionally Medina. By contrast, charts and atlases from Christian milieus document a progressive inclusion of scriptural references in their imagery and highlight elements of sacred geography – especially Jerusalem, St. Catherine’s monastery on Sinai, the “True Cross”, or symbols of paradise. Applying decorative patterns used for religious and historical texts to other genres is a development in Islamic book art that can be observed in other forms in other Islamicate societies too. But while al-Sharafī’s atlases and the charts by al-Ṭanjī and al-Mursī reference Andalusi and Maghribi decorative norms of religious literature, this is not the case for other geographical works produced in that region.³⁹ This suggests that al-Sharafī adhered to a decorative norm specifically shared by North African portolan chart producers.

Fig. 13.1

Sura al-fātiḥa, known as “Charles V Qur’an”, BnF, ms arabe 438, fol. 1^v

The knot patterns expressing the integration of the atlas into religiously acknowledged artwork are bold claims of belonging to a concrete religious cultural space. Al-Sharafī was aware of the imagery of Christian charts and atlases. Trying to maintain their expressions of religious identity, he used the knot patterns as a recognizable expression of religious identity for Muslims in North Africa. He transferred them into his atlas in a way that legitimizes the translation of the Christian atlas into a Muslim one. The framing is the foundation on which his acts of cultural translation are implemented. This translation is specific and expresses the cultural norms and aesthetics of al-Andalus and the Maghrib only. It is not an act of norm-breaking as in the cases of textual imitations, transliterations, or translations discussed in the previous section, nor is it an innovation. It draws on well-established and cross-communally shared patterns of religiously condoned representation. Through this strategy of decorative embedding, it expresses a conservatism that ennobles the atlas as a truly domestic and familiar object of regionally shared culture and identity. Other steps of familiarization can be seen in not just the inclusion but the representational integration of the calendrical, timekeeping, and qibla items within the atlas. With these strategies for anchoring the charts of the Mediterranean and the Black Sea in various Muslim contexts, al-Sharafī is inviting the customer to accept the whole atlas as a Muslim object and the knowledge it offers as genuinely belonging to them.

The presence of the knots in the frames and corners of the atlas clearly translate the religious artworks illustrating the corners in many fourteenth- and fifteenth-century atlases from the northern Mediterranean. These illustrations cover biblical stories and the four apostles, as well as angels, saints, and members of the Holy Family (such as the Virgin and Child). Portolan charts from those regions placed similar illustrations at the neck of the parchment or on other appropriate locations. But while both cases can be considered as strategies of cultural contextualization, the paths that were chosen differ. In the case of al-Sharafī the source is book art, while in the case of the northern Mediterranean charts and atlases inspiration came more often from artwork that followed the tradition of Byzantine religious icons or imitated religious sculpture. Book art also seems to have occasionally contributed (see Fig. 13.2). These similarities are not limited to the composition of individual charts. They rather permeate the overall structure of the atlas. Where atlases like that of Battista Agnese (d. 1564) include Christian images, in this case the scene of the Crucifixion,⁴⁰ al-Sharafī included a qibla chart.⁴¹ In other words, al-Sharafī’s atlas results from a cultural adaptation of the visual practice of incorporating sacred elements into atlases found in late medieval and early modern Christian exemplars.

Fig. 13.2

Nativity scene in a compass rose, Chart of Juan de la Cosa, 1500, Naval Museum Madrid, MN 257

A second important visual component of al-Sharafīʼs atlas is its flags and banners. With a total of sixty-three flags, it significantly outnumbers those on most other portolan charts and atlases, including his own 1571 atlas.⁴² A widespread element in this type of map, flags were used to express the military and political importance of a place.⁴³ However, this does not seem to be the case in the 1551 atlas, because it gives flags to locations of little-known importance in the sixteenth century.⁴⁴ Another unique characteristic of the flags in the atlas is that, while some of them appear only once and thus correspond to one place alone, others are used in different locations: forty flags are recurrent, while only twenty-three occur just once.⁴⁵ We are not aware of any other examples of recurrent flags, whatever their linguistic or cultural identity. The question of what meaning al-Sharafī inscribed into the atlas by his choice of number, shape, colour, symbolism, and placement is difficult to answer. Earlier proposals, such as Kahlaouiʼs suggestion that the flags serve as demarcations between the lands of western Christianity and Islam, lack substance, since many of the recurrent flags appear in locations that correspond to different religious and political realms.⁴⁶

Applying concepts and perspectives from translation studies, art history, and Mediterranean studies might open up ways of making sense of al-Sharafī’s flags without, however, offering simple, straightforward identifications. We tried to identify al-Sharafīʼs main translation practices in this respect. The main method involves making comparisons across cultural, political, and religious boundaries. We applied this to flags surviving from the Islamicate world, representations of flags in Muslim and Christian artworks, coats of arms, and travel accounts. The few surviving military and nautical flags from the Merinid, Hafsid, and Ottoman dynasties share a number of common features but have little kinship – and this also only in specific cases – with those depicted by al-Sharafī. The former are all quadratic or rectangular, while al-Sharafī displays a greater variety of shapes (see Fig. 13.3a – e).⁴⁷ Thus, they do not provide direct models for al-Sharafī’s iconography (see Figs. 13.4 and 13.5).

Fig. 13.3

Paris, BnF, ms arabe 2278; a: fol. 6^r, Nakīrʼs flag; b: fol. 5^v, Alexandria’s flag; c: fol. 3^v, Oran’s flag; d: fol. 4^v, Sibenikʼs flag; e: fol. 6^v, Tripoli’s flag

Fig. 13.4

Merinid flag of Abū l-Ḥasan (r. 1331–48) dated 1339–40, Cathedral of Toledo, inv. 1516

Fig. 13.5

North African flags, Petrus Russus, 1508, Barcelona, Maritime Museum Barcelona, no. 841

Two sources resemble al-Sharafīʼs flags surprisingly closely despite the fact that there is no possibility they could have inspired him: (1) depictions of Ottoman flags in the notebooks of Luigi Marsigli (1658–1730); and (2) visual claims by the Turkish Naval Museum about historical naval flags of the Ottoman Empire.⁴⁸ This resemblance is currently unexplainable, as are the drawings by Marsigli and the Turkish Naval Museum. Hence, further research is clearly needed.

By contrast, artworks – in particular, miniatures, tiles, bowls, or cups – may indeed have served as a major source of inspiration. They contain representations of flags that are related to those chosen by al-Sharafī without being clear-cut ancestors in terms of their shape, colour, or symbolism. This applies, in particular, to triangular and swallowtail pendants. They appear, for instance, in miniatures in Abū Muḥammad al-Ḥarīrīʼs (d. 1122) al-Maqāmāt (see Fig. 13.6) and in Turkish Ottoman chronicles illuminated during the reigns of Sulaymān I (r. 1520–66), Selim (or Salīm) II (r. 1566–74), and Murād III (r. 1574–95).⁴⁹ These works – which contain depictions of naval or land-based battles – include flags whose forms or colours chime in some respect with al-Sharafīʼs choices without, however, being identical (see Fig. 13.7). Ottoman nautical charts and atlases, on the other hand, rarely contain flags. But when they do, they follow relatively closely the repertoire of Majorcan-type charts.⁵⁰ The theory that al-Sharafī might also have imitated examples from such charts or atlases of the northern Mediterranean is not fully borne out. Only a few portolan charts of the late fifteenth and sixteenth centuries agree in specific elements with al-Sharafīʼs representations. This applies, for instance, to the peculiar form of hexagonal flags across Jaume Bertránʼs chart of 1482,⁵¹ the triangular pendants in North Africa in Petrus Russus’s chart of 1508 (see Fig. 13.5 and 13.6),⁵² the swallowtail and triple-tail flags in North Africa, and the triangular pendants in North Africa and Anatolia of Vesconte Maggiolo’s 1548 chart.⁵³ But despite these similarities, al-Sharafī’s flags differ significantly from them.

Fig. 13.6

Al-Sharafīʼs atlas of 1551, Paris, ms arabe 2278. Fig. 10: fol. 5^r, Istanbul’s flag; Fig. 11: fol. 6^v, Gabesʼs flag; Fig. 12: fol. 4^r, Binzertʼs flag; fig. 13: fol. 4^r, Algiers’s flag

Fig. 13.7

Cantigas manuscript, El Escorial, Library, ms T-1–1. Fig. 14: fol. 82^r

The only working hypothesis that can be formulated at the moment is that al-Sharafī did not copy any precise ancestor but drew inspirations from a mixed body of Ottoman, Majorcan, Italian, and perhaps older Muslim charts, paintings, and possibly lost naval flags. The question, however, of whether such a creative mix could be the result of translation practices needs to remain open for the time being.

The analysis of the symbols in al-Sharafī’s flags yields clearer results about cultural contexts and potential sources of inspiration, despite the fact that none of them appear exactly as such in any of the available atlases, charts, world maps, illuminated historical chronicles, or religious books. The symbols used by al-Sharafī are simple geometrical shapes in single or compound forms: swastika, rhombus (sometimes with a small circle or square inside), a big yellow circle with another small red/blue one inside, a cross (sometimes with four small dots in its corners), a kind of flower or bird, a crescent (sometimes two), a rhombus divided into four squares with four red dots in each of them, a symbol looking like the modern infinity symbol, two small yellow dots at the border of a triangular flag, and other figures that seemingly try to represent Kufic script or a building (a mosque, a castle?).

Their repertoire of meaning can be subdivided into four groups: (1) the crescent and/or fake Kufic script; (2) a single or a double Persian form of the letter hā; (3) multiple stripes coloured differently; and (4) sets of simple geometrical figures used in Maghribi ceramic tiles and in registers of decorative elements in Castilian book art. Flags with one or more crescents had been used to express Muslim identity on Italian and Majorcan portolan charts or in atlases since the fourteenth century.⁵⁴ Al-Sharafī used them rather sparingly (altogether only seven times) and outside North Africa. Again, as in other cases, no ancestor for the specific distribution of flags with crescents could be identified.

A second identifiable symbol is inscribed in the flags used for Gabes and Istanbul (see Figs. 13.6a, b). It has the shape of the infinity symbol (in one case doubled). But since this is a new mathematical creation of the late nineteenth century, this could not have been al-Sharafīʼs source of inspiration. It seems instead to be the representation of the Arabic letter hā, written in Persian style, which was used in Ottoman battle banners to represent the Sufi dhikr expression huwa (He: i.e., God).⁵⁵

A third kind of symbolism al-Sharafī employed for demarcating flags involves coloured stripes. He placed striped flags at the North African coast in the cities of Badis, Algiers, Binzert (Bizerta), and Tunis, the island of Djerba, and the Ottoman town of Alanya. The use of striped flags of different colours (red-yellow-green or red-yellow-blue) is attested in the Ottoman provinces of North Africa (Algiers, Tunis, Tripoli) during the sixteenth century (see Figs. 13.6c, d).⁵⁶ However, at the beginning of 1551 Tunis and Tripoli were not yet Ottoman possessions. Tunis was partially in the hands of the Spaniards, and Tripoli was controlled by the Ottomans starting only in that year.⁵⁷ This makes it unlikely that the striped flags were meant to express current events or possessions. If they had indeed a meaning beyond mere decoration, they might signal desires, expectations, or rumours. Since these types of flags can also be found in Ottoman historical depictions of battlefields and on portolan charts or atlases produced in towns of the northern Mediterranean since the early sixteenth century, the search for ‘cultural equivalence’ in which al-Sharafī apparently was engaged, as argued above for the frames, might have motivated his choices.⁵⁸

This search for ‘cultural equivalence’ seems to have also guided his decisions with regard to the remaining symbols in the atlas. Many of them are identical to those used in public buildings in North Africa and in book art at the court of Alfonso X of Castile (r. 1252–84). An architectural example of this can be found in the ceramic tiles in the mosque of Qayrawan.⁵⁹ Alfonsine book art, surprisingly closely related to al-Sharafīʼs choices of symbols, is represented in the manuscripts of the Cantigas de Santa María, a work composed in the scriptorium of Alfonso. It contains a total of 2,400 vignettes.⁶⁰ These reflect the multicultural nature of the newly emerging trends in book art in Christian Castile and León. The scribes at Alfonso X’s court came from different backgrounds: French, Italian, Castilian, Jewish, Mozarab, and Arab.⁶¹ Together they created a new style of illumination, later called alfonsí.⁶² Among the possible stylistic affiliations in the Cantigas, Menéndez Pidal, Domínguez, and Treviño have stressed the importance of Muslim artistic practices and motifs.⁶³ Early twenty-first-century studies have pointed out similarities between the military scenes in the Cantigas and some of the vignettes in an illuminated manuscript of al-Ḥarīrī´s Maqāmāt from Baghdad.⁶⁴ So far it seems unlikely that the scribes of the Alfonsí scriptorium would have had access to a copy of al-Ḥarīrī´s Maqāmāt, but they were certainly familiar with Islamic and Christian traditions of illuminating Arabic books through the many manuscripts preserved in al-Andalus, for example in recently conquered Seville (1248).⁶⁵ In addition to Arabic manuscripts, Muslim architecture from the Iberian Peninsula will have provided the designers of the Cantigas with abundant access to the vast and culturally complex iconographic palate available in al-Andalus and the Maghrib. Through their choices, they strengthened the presence of Muslim art elements in the visual forms of Castilian and Aragonese religious, courtly, and everyday life, as reflected in architecture, textiles, reliquaries, and book art between the twelfth and the sixteenth centuries.⁶⁶ Several of the geometric motifs present in al-Sharafīʼs flags (the swastika, rhombus, cross, flowers, circle, square with a circle inside) appear, for instance, in Catholic altar cloths, some of which even have inscriptions in Kufic script (see Fig. 13.7).⁶⁷ It is widely accepted that elite art in the Christian kingdoms of the Iberian peninsula was a continued practice of imitation, adaptation, translation, and integration of different cultural elements from different areas of Europe, as well as North Africa and the Middle East.⁶⁸ The many Muslim and Jewish immigrants from the peninsula transferred this experience and taste to the southern Mediterranean and spread it along the North African coast at the very least as far as Ifriqiya.⁶⁹

Hence, the cultural complexity of these iconographic elements in al-Sharafī’s atlas and their connections with the Andalusi aesthetic heritage reflect one of the shared cultural spaces in the western Mediterranean that decreased the necessity for translation and increased the immediacy of comprehension. Although it is unlikely that al-Sharafī had access to a copy of the Cantigas, he could have had access to material objects from the Iberian peninsula and the Maghrib containing some of the symbols al-Sharafī used, like the swastika, which is scarcely documented in the local art forms of Ifriqiya.⁷⁰ Thus, in addition to the shapes and patterns that were omnipresent in different kinds of objects and buildings in Iberia and North Africa and which al-Sharafī and the illuminators of the Cantigas could easily ‘copy’, ‘imitate’, and ‘modify’, the manifold mutual interconnections also enabled and encouraged acts of ‘transfer’, ‘integration’, and ‘transformation’.⁷¹ Al-Sharafī’s obviously limited artistic skills seem to have added elements of simplification and sterilization. This can be detected, in particular, in comparison with the symbols on other portolan charts showing single or double eagles or a crossbow with arrow. Such effects can perhaps be considered as cases of failed ‘direct translation’.

In addition to direct and intermediary ‘appropriation’ and ‘adaptation’ from public buildings and book art, al-Sharafī may also have acquired some of his symbols from everyday objects. The use of some of the geometrical motifs depicted in his flags and in the Cantigas is well attested in various objects from North African Berber dynasties. Although further analysis is needed of the vast repertoire of geometrical symbols included in the Cantigas, the presence of some of them almost three centuries later in al-Sharafī’s flags not only proves their accessibility in sixteenth-century Sfax but also indicates the choices al-Sharafī made. First, by including them, he decided to associate several flags – and hence several locations – with each other, giving his works a specific note. Second, he extended the frames of reference of his atlas by uniting them with motifs shared with many other portolan charts and atlases. The overlap in elementary geometrical symbols between these different traditions of decorative art decreased the burdens of cultural translation and circumvented the obstacles that such acts often encounter. In this manner, al-Sharafī introduced the locally rooted and widely shared mix of Islamic and Berber traditional symbolic patterns into an object that had its own specific semiotic repertoire. In a subtle manner, his choices strengthened the appearance of his atlas as a part of the North African religious culture and book art, as well as the shared cultural space of nautical charts and atlases, making it recognizable to all those who were conversant with the main components of these three spaces. Its full comprehensibility, however, will have been limited to those who were multilingual within the linguistic and visual scopes of the atlas.

3 Conclusions

Drawing on several distinct approaches to translations, we have sought to identify the different processes of cultural translation al-Sharafī applied in his 1551 atlas. In our quest to understand such processes, we have relied on the concepts of ‘cultural equivalence’, ‘oblique translation’, ‘source-oriented translation’, and lingua franca.

In the fourteenth century, map-makers in such places as Genoa, Venice, and Majorca developed a multicultural repertoire for representing their knowledge of the geographical, cultural, and political characteristics and identities of the coastal as well as inland territories. They drew on medieval Latin world maps, ancient encyclopaedias, travel accounts in several Romance languages, Arabic maps and geographies (particularly on al-Idīsī), Islamic works of art, Byzantine copies of Ptolemy’s Geography, and the linguistic, natural, cultural, and political knowledge of merchants, diplomats, clerics, sailors, and crusaders. Similar, though not always as rich, cross-culturally informed charts and atlases were produced during the fifteenth century, spreading the models worked out in the previous century across the western Mediterranean, including a new generation of Arabic native speakers as producers of nautical charts and perhaps even atlases. The processes of establishing shared frames of textual and visual information, together with the spread of such frames across several Mediterranean milieus, were enabled by the numerous acts of copying, imitating, translating, and integrating, as well as by the intellectual and physical mobility of the people involved. In this paper, we have argued that the atlas at the centre of our study was the result of multifaceted acts of copying, imitating, translating, and integrating linguistic, religious, nautical, geographical, agricultural, and astronomical information and the patterns and symbols of the decorative arts in use among different ethnic, linguistic, and political groups in North Africa and on the Iberian peninsula. As in the fourteenth century, cross-cultural translation took place as a combination of multiple skills and information drawn from different cultural sources. Which of the different acts of translation characterizing ‘Alī al-Sharafī’s 1551 atlas were his own and which he owed to the works of his grandfather and father cannot be decided beyond his own claims in the 1571 atlas and the 1579 world map, in which he states that he relied on their works and profited from them.⁷² But the compound nature of all of his products is undeniable. Cultural translation was clearly a cherished mode of production.