Introduction: The Aqquyunlu takeover of Mardin in 835/1432 according to narrative and diplomatic sources

The Aqquyunlu takeover of the town and citadel of Mardin from its Qaraquyunlu governor in 835/1432 is exceptionally well documented by several complementary narrative and diplomatic sources. Until the beginning of the ninth/fifteenth century, the town had been ruled by the Artuqid dynasty for several centuries.Footnote ¹ Their rule initially outlasted the campaigns of Timur. However, subsequent Aqquyunlu raiding prompted the last Artuqid ruler, al-Malik al-Ṣāliḥ Aḥmad, to deliver the town and its citadel to the Qaraquyunlu ruler qara Yūsuf in 811/1409.Footnote ² Although the town itself was subsequently strengthened by a rebuilding of its defensive walls in the same year of 1720 as/1409,Footnote ³ its hinterland continued to be raided by forces affiliated to the Aqquyunlu ruler qara ʿUthmān,Footnote ⁴ whose realms were centred on the fortified town of Diyarbakır. See Figure 1 for a map of Northern Mesopotamia c. 1420 ce.

© d-maps.com https://www.d-maps.com/carte.php?num_car=699&lang=de

Figure 1. Northern Mesopotamia and adjacent areas, c. 1420

The transregional scale of these mutual raids centred on qara Yūsuf Qaraquyunlu and qara ʿUthmān Aqquyunlu is exemplified by the events of the year 821/1418. As narrated among the events of the preceding year 820/1417 by the towering Mamlūk scholar Ibn Ḥajar, qara ʿUthmān had notified the Mamlūk authorities in Aleppo of a truce between him and qara Yūsuf that had been concluded in late summer 820/1417.Footnote ⁵ According to the stipulations of this agreement, the Aqquyunlu ruler transferred the town of Savur (madīnat ṣawur) to qara Yūsuf Qaraquyunlu in exchange for one million dirham, as well as a hundred horses and a hundred camels.Footnote ⁶ The elimination of Savur as a defensible outpost used in Aqquyunlu raiding must have contributed greatly to the consolidation of Qaraquyunlu rule over Mardin, which constituted the anchor of the south-eastern salient of the realms ruled by qara Yūsuf.

Notwithstanding the truce of 820/1417, the following year saw a resumption of mutual raids:

As incidentally attested sub anno 1729 as/1418 by the complementary report of Nūḥ (Breviarium, 469), qara ʿUthmān subsequently followed the retreating Qaraquyunlu army and even briefly occupied the town of Siirt, some 200 kilometres north-east of Mardin, before surrendering it “to its master” (sallamahā li-ṣāḥibihā), presumably the local Ayyubid ruler of Hasankeyf.Footnote ¹¹

The geographical span of some 600 kilometres covered by the raiding armies between Siirt and Aleppo is impressive and illustrates the scale of “etatist fluidity” across early fifteenth-century northern Mesopotamia. Although secondary towns and castles could be occupied during raids, the citadel of Mardin and the strongly fortified town of Diyarbakır served to stabilize the hegemony of qara Yūsuf and qara ʿUthmān respectively over the surrounding areas. Accordingly, the intermittent raids by “Turkmen” and other forces resulted in a devastation of rural areas and secondary centres that was weathered only by the exceptionally strong fortifications of Mardin’s citadel and the town walls of Diyarbakır.

This equilibrium of pervasive raiding sparing only the strongest fortifications continued across the transition of leadership within the Qaraquyunlu from qara Yūsuf to his son Iskandar,Footnote ¹² paradigmatically changing with the Aqquyunlu occupation of Mardin’s citadel in 835/1432. Significantly, this extension of qara ʿUthmān’s realms did not result from a sustained siege or a successful assault against the citadel, but rather from an internal revolt of the non-military entourage of the Qaraquyunlu governor Nāṣir al-DīnFootnote ¹³ left in control of the citadel while Nāṣir al-Dīn and his army descended to defend the town. This revolt is described by the already-cited Arabic chronicle of Ibn Ḥajar (Inbāʾ: vol. 3, p. 473), as well as by the non-courtly Arabic chronicle of al-GhiyāthFootnote ¹⁴ and one of the anonymous continuations of the Syriac Chronography of Barebraeus.Footnote ¹⁵ In addition, a diplomatic missive sent by the Qaraquyunlu governor Nāṣir al-Dīn to the Ottoman sultan Murād II immediately before the internal revolt has been published together with the Ottoman response in a nineteenth-century anthology of rulerly missives (munshaʾāt) preserved in Istanbul.Footnote ¹⁶ The Syriac continuation of Barebraeus and the account of al-Ghiyāth may preserve oral narratives circulating in the vicinity of Mardin in the second half of the fifteenth centuryFootnote ¹⁷ and al-Ghiyāth has been suggested to cite the Inbāʾ al-Ghumr of Ibn Ḥajar elsewhere.Footnote ¹⁸ Nonetheless, the remarkably consistent picture emerging from these four sources for the Aqquyunlu appropriation of Mardin from its Qaraquyunlu governor cannot be explained by direct intertextual dependencies.

This absence of direct intertextual dependencies between the preserved sources is particularly significant due to the dense sequence of pervasively attested narrative motifs or topoi in the most detailed account contained in the Syriac continuation of Barebraeus, which is echoed across the other three sources. Due to the recurrence of the motifs in this detailed Syriac narrative across the other three texts, it accordingly appears that (some of) these narrative topoi already structured narratives of the Aqquyunlu occupation of Mardin in 835/1432 immediately after the event. In the following overview, I tentatively classify these motifs according to the rubrics of Thompson (Reference Thompson1975–76) to illustrate their narrative function as Wandermotive beyond their immediate context of the Aqquyunlu occupation of Mardin.

When qara ʿUthmān’s son Ḥamza came to raid Mardin in September 835/1431,Footnote ¹⁹ he was captured by the Qaraquyunlu governor Nāṣir al-Dīn and imprisoned in the citadel.Footnote ²⁰ At this point, Nāṣir al-Dīn on his own authority wrote to the Ottoman sultan offering to extradite Ḥamza, possibly contemplating a switch of his political allegiance to the Ottomans.Footnote ²¹ After the capture of his son, qara ʿUthmān moved against Mardin with another army.Footnote ²² When Nāṣir al-Dīn moved out from the citadel with his troops to defend the town and its surroundings, Ḥamza was unexpectedly declared ruler of Mardin by a civil revolt.Footnote ²³ As partially implied by al-Ghiyāth and explicitly suggested by the Syriac continuation of Barebraeus, this civil revolt was mounted by “the women and slaves of Nāṣir al-Dīn” (nšē d-nṣr w-ʿabdēhū) who had remained in the citadel.Footnote ²⁴ The support of the subaltern members of Nāṣir al-Dīn’s provincial court for the Aqquyunlu prince is motivated by their recognition of Ḥamza as a “handsome youth of noble stature” (ṭalyā šapīrā w-šaqīḥ qūmtā).Footnote ²⁵ Caught between the closed gates of the citadel and the besieging Aqquyunlu army led by qara ʿUthmān, Nāṣir al-Dīn was forced to flee.Footnote ²⁶

The recognition that the extant sources describing the Aqquyunlu occupation of Mardin and the installation of Ḥamza as its governor are densely structured by narrative motifs also attested in other contexts certainly does not rule out the possibility that the events could in fact have unfolded as claimed by the sources. Nonetheless, the density of these Wandermotive across all four sources cautions against any over-confident interpretation beyond the recognition that Ḥamza had indeed been a prisoner in the citadel before being declared ruler by the “civil” (or at least non-soldierly) revolt. Hitherto unrecognized Arabic epigraphic sources preserving the self-representation of Ḥamza and qara ʿUthmān immediately after the occupation of Mardin help us move beyond this impasse.

Two epigraphic sources commemorating the Aqquyunlu occupation of Mardin: description and critical edition

This article presents the edition of an Arabic inscription formerly displayed above the gate connecting the town of Mardin to the citadel (henceforth inscription 835 Mardin, see Figure 2), which was commissioned by Ḥamza b. qara ʿUthmān immediately after the Aqquyunlu takeover in 835/1432. As the original has been lost, reconstruction of this crucial source proceeds based on a photograph taken in 1911 by the British scholar Gertrude L. Bell.Footnote ²⁷ Inscription 835 Mardin in turn facilitates some emendations and crucial context to an undated Arabic inscription that was commissioned together with a monumental new lock of Mardin’s citadel on the same occasion. Although this lock was held as #378 in the Çinili Köşk of the Topkapı Sarayı Müzesi during the first half of the twentieth century, it cannot currently be located.Footnote ²⁸ Accordingly, my reconstruction of the text inscribed on this object proceeds based on the edition by Halil Etem [Eldem],Footnote ²⁹ as checked against additional photographs published elsewhere.Footnote ³⁰ In the present section, I suggest a critical edition followed by a detailed commentary for both inscriptions.

Bell (Gertrude) Archive, Newcastle University Library

Figure 2. The gate connecting the town to the citadel and displaying inscription 835 Mardin as photographed by Gertrude Bell in 1911 (GB/3/1/18/1/145).

The concluding outlook contextualizes both texts as sources for the Aqquyunlu takeover of Mardin in 835/1432 and within Islamicate epigraphic habits of the fifteenth century.

Originally composed of six lines of Arabic text displayed on three tiers of stones set within a frame above the main gate between the town and its citadel, inscription 835 Mardin commemorated the commission of a renovation (ʿimāra)Footnote ³¹ of the gate immediately after the Aqquyunlu takeover (see Figures 3 and 4). Both inscription 835 Mardin and this gate collapsed at some point after 1932.Footnote ³² As its former location is currently inside the perimeter of the active military base on Mardin’s citadel, I have not been able to investigate whether stones of inscription 835 Mardin may still be extant among the ruins and rubble covering the steep slope below its former location. One of the two lions formerly surmounting inscription 835 Mardin is now held in the Mardin Müzesi.Footnote ³³

Figure 3. The citadel of Mardin as seen from the town in 2024

Figure 4. The former location of the gate displaying inscription 835 Mardin in 2024

As already noted by Gabriel (Reference Gabriel1940: vol. 1, p. 13), the stones of inscription 835 Mardin were quite deteriorated and their sequence was garbled even before the eventual collapse of the gate after 1932. This garbled sequence, including two stones that had been turned upside down, indicates a restoration or rebuilding of the gate at some point between 835/1432 and 1911. For ease of reference, I have assigned letters to the stones of inscription 835 Mardin from right to left in the sequence in which they are documented on Bell’s photograph. Each of the three tiers of stones displays two lines of Arabic text (see Figure 5).

Figure 5. The sequence of the stones displaying inscription 835 Mardin as documented by Bell in 1911

Upper Tier: A, B, C, D, E, F

Middle Tier: G, H, I, J, K

Lower Tier: L, M, N, O, P

Based on the legible letters displayed on each stone, this garbled sequence can confidently be corrected as follows, marking the two stones that were reset upside-down with asterisks.

Upper tier displaying the original lines 1 and 2: A, B, C, I, O*

Middle tier displaying the original lines 3 and 4: G, H, P, N, D

Lower tier displaying the original lines 5 and 6: L, M, E, F, K, J*

As indicated by the preservation of the stones displaying the beginning of all six lines in situ, the rebuilding that caused the garbled sequence in which the stones were displayed during the first half of the twentieth century may only have affected the left part of the inscribed surface.

According to this reconstructed sequence of its stones, the legible portions of inscription 835 Mardin can be edited as follows, accepting reasonable conjectures where alternative, but largely synonymous, readings may be possibleFootnote ³⁴ and indicating passages that cannot be ascertained with confidence as lacunae:

1. 1) bi-smi llāhi l-raḥmāni l-raḥīmi wa-mā tawfīqī illā bi-llāhi ʿalā man tawakkaltu wa-huwa rabbu l-ʿarshi l-ʿaẓīmi [pastiche of Quran 11: 88 and Quran 9: 129]

2. 2) wa-lammā fataḥa llāhu taʿālā ʿalā l-amīri l-aʿẓami fakhri l-dunyā wa-l-dīni ʿuthmāna bahādura khallada llāhu mulkahū wa-ayyada wa-sallaṭa waladahū ḥamzata bi-niyābati mārdīna abbada llāhu taʿālā dawlatahū wa-amadda lahū bi-niyābatihā

3. 3) amara bi-ʿimārati hādhā l-bābi l-sharīfi l-ʿālī l-amīru l-aʿẓamu wa-l-khāqānu l-muʿaẓẓamu mawlā mulūki l-ʿarabi wa-l-turki wa-l-ʿajami … shāhrukhu bahāduru

4. 4) khallada llāhu taʿālā mulkahū wa-sulṭānahū wa-ḥakama ʿalā āmida l-maḥrūsati wa-arqanīna wa-l-ruhā wa-kharbirta wa-tarjīla wa-l-raqqati wa-raʾsi l-ʿayni wa-l-ṣawuri wa-arzinjāna wa-kamākha wa-jimishkazaka wa-qarāḥiṣāra wa-kūnja wa-bāburta wa-āqshahra wa-kīghī wa-bālū wa-akil wa-qalʿati mārdīna

5. 5) … yā dāwūdu jaʿalnāka ḥākima l-ruʾūsi [paraphrasing Quran 38: 26] wa- … ʿuthmānu taḥta qalʿati mārdīna wa-hajama nāṣiru l-dīni minhā … allāhi faʾlu … l-dunyā wa-l-dīnī yawma

6. 6) l-aḥadi awwala shahri l-muḥarrami min shuhūri llāhi sanata khamsatin wa-thalāthīna wa-thamānimiʾatin taḥāwala wa-qahara … wa-kāna wāliduhū mallakahū wa-ḥāṣara qalʿata mārdīna wa-fī hādhā … sanata [835].

1. 1) In the name of God the Merciful and Compassionate. My success is exclusively due to God in whom I trust, He is the Lord of the mighty throne [pastiche of Quran 11: 88 and Quran 9: 129].

2. 2) When God, exalted be He!, gave victory to the greatest amīr, the splendour of the world and the religion ʿUthmān bahādur, may God make his reign everlasting!, and [when God] supported and empowered his [ʿUthmān’s] son Ḥamza by granting him governorship over Mardin, may God, exalted be He!, prolong his incumbency and make his governorship lasting!,

3. 3) [at this time], the rebuilding of this noble and high gate was ordered by the greatest amīr and empowered khāqān, the master of the kings of the Arabs, the Turks, and the Persians, … Shāhrukh bahādur,

4. 4) may God, exalted be He!, make his reign and rule everlasting! He [qara ʿUthmān and/or Shāhrukh] reigns over well-protected Diyarbakır, Ergani, Urfa, Harput, Tercil, al-Raqqa, Raʾs al-ʿAyn, Savur, Erzincan, Kemah, Çemişgezek, Karahisar, Genç, Bayburt, Akşehir, Kiğı, Palu, Eğil and the citadel of Mardin.

5. 5) … oh Dāwūd, we have appointed you ruler of the people [paraphrasing Quran 38: 26] and … ʿUthmān was below the citadel of Mardin when Nāṣir al-Dīn made a sortie from it … God [granted] the omen of … the world and the religion, on the day of

6. 6) Sunday, the first of the month of Muḥarram among the months of God in the year eight hundred and thirty-five. … he overcame … after his father had appointed him commander to siege the citadel of Mardin when … of the year of [835].

In addition to the lacunae indicated in this reconstruction, the reading of the narrative passages in lines 5 and 6 suggested above is tentative due to the singularity of this portion of the text and the state of preservation in which it was documented by the sole remaining witness of Bell’s photograph. Nonetheless, reconstruction and interpretation of inscription 835 Mardin can confidently depart from the suggestion that it commemorated the Aqquyunlu takeover that installed qara ʿUthmān’s imprisoned son Ḥamza as governor in Mardin’s citadel. The text must also be interpreted as composed in an erudite register of Arabic rhymed prose (sajʿ) studded with Quranic citations and intertextual references to the current titles used by qara ʿUthmān, his son Ḥamza, and their Timurid overlord Shāhrukh. Within this historical and intellectual context, the structure of inscription 835 Mardin can confidently be described as follows:

1. 1) Invocation of God, followed by a pastiche of two Quranic verses attributing the Aqquyunlu success to divine intervention.

2. 2) [First part of building inscription:] At the time when God effected the Aqquyunlu takeover headed by qara ʿUthmān and his son Ḥamza,

3. 3) [Second part of building inscription:] The Timurid ruler Shāhrukh ordered the rebuilding of the gate between town and citadel that is ostentatiously commemorated by inscription 835 Mardin.

4. 4) Eulogy for Shāhrukh, followed by a catalogue of the Aqquyunlu realms in 835/1432 that acknowledged the leadership of qara ʿUthmān. This catalogue concludes with the citadel of Mardin and grammatically simultaneously attributes rule over these localities to Shāhrukh and to qara ʿUthmān.

5. 5) [Narrative of the Aqquyunlu takeover of Mardin:] …, paraphrasis of Quranic verse. [Narrative of Ḥamza’s capture:] …, the sortie of Nāṣir al-Dīn breaking a siege of Aqquyunlu troops acknowledging the authority of qara ʿUthmān …, due to God’s divine wisdom, on

6. 6) Sunday, Muḥarram 1, 835, … [capture of Ḥamza] … [eventual reversal of fortune resulting in the installation of the imprisoned Aqquyunlu prince as governor in the citadel of Mardin] … [date].

Accordingly, the text of inscription 835 Mardin fuses the occasion and common template of a renovation inscription (amara bi-ʿimārati hādhā l-bābi etc.) that ostentatiously commemorates the Timurid overlord of qara ʿUthmān as the patron of the rebuilding with an epigraphically stabilized account and rationale for the Aqquyunlu takeover. Within its highly visible location above the main gate connecting the town of Mardin to its citadel, this commemoration of the Aqquyunlu takeover must be understood as a programmatic declaration manifesting the commitment of the new rulers to a specific standard of rule in their administration of Mardin. In this regard, the narrative focus on the capture of the Aqquyunlu prince Ḥamza by the former governor Nāṣir al-Dīn on Muḥarram 1, 835/September 9, 1431 (see below) arguably underlines the Aqquyunlu programmatic commitment to just rule against the actors who in fact facilitated Ḥamza’s subsequent liberation and the ousting of the Qaraquyunlu governor.

Due to the deterioration of the end of line 6, we cannot know whether the final date, likely concluding with the indication of the year in lost numerals following the legible word sana, year, commemorated the date of the internal revolt against Nāṣir al-Dīn or the rebuilding of the gate. Nonetheless, the inscription can confidently be attributed to the year 835/1432 due to its coherence with the text displayed on the lock to this gate that is presented below. Among the events of 835, Ibn Ḥajar (Inbāʾ: vol. 3, p. 473) indicates that a major military campaign that had been prepared by the Mamlūk sultan Barsbāy in response to qara ʿUthmān’s occupation of Mardin was abandoned after the arrival of the keys to Mardin’s citadel (mafātīḥ qalʿat mārdīn). As it is highly likely that these keys pertained to the lock of the rebuilt gate formerly held in the Çinili Köşk of the Topkapı Sarayı, we can confidently attribute the inscriptions commemorating the Aqquyunlu takeover on the gate and its lock to the year 835.Footnote ³⁵

Line 1 begins with a basmala, followed by a pastiche of two Quranic verses (iqtibās)Footnote ³⁶ that complements the introductory invocation of God with the prose rhyme of al-raḥīm/al-ʿaẓīm. The same prose rhyme extends across lines 2–6 with the exception of line 3 (2: fakhr al-dunyā wa-l-dīn/bi-niyābat mārdīn, 4: wa-arqanīn/wa-qalʿat mārdīn, 5: taḥta qalʿat mārdīn/nāṣir al-dīn/al-dunyā wa-l-dīn, 6: khamsa wa-thalāthīn/qalʿat mārdīn), tying the various “Aqquyunlu components” of inscription 835 together and showcasing the erudition of its patron(s).

Within its epigraphic setting, the Quranic pastiche of line 1 must be understood as spoken jointly by the patron(s) and readers of the inscription, who are framed as pious Muslims capable of recognizing the Quranic words. The joint declaration of ruler(s) and ruled to entrust their matters to God specifically resonates with the subsequent attribution of the Aqquyunlu takeover to divine agency, promising similarly astonishing success to whoever emulates the trust in God that transported Ḥamza b. qara ʿUthmān from the prison to the throne. In addition, the emphasis on divine agency in effecting the Aqquyunlu takeover arguably serves to downplay Ḥamza’s indebtedness to the local civic and non-military courtly elites who are credited with the civil revolt in the narrative accounts discussed in the introduction to this article.

Beyond this resonance of the content of line 1 with the joint declaration of trust in God by rulers and ruled in Mardin, the choice of Quran 11: 88 and Quran 9: 129 for epigraphic display in this prominent location further resonates with the context attributed to both verses in Quranic exegesis. Within its original Quranic context, Quran 11: 88, is spoken by the pre-Islamic prophet Shuʿayb. In his story, as referenced by Quran 11: 84–95, repeated emphasis is placed on the suggestion that he was sent to the people of Midyan by God to chide them for adulterating established measures and weights (al-mikyāl wa-l-mīzān, Quran 11: 84 and 85). Accordingly, the prominent deployment of a paraphrasis of Quran 11: 88 in line 1 arguably declares the commitment of the new Aqquyunlu administration to uphold established measures and weights. This may possibly also imply a claim that these standards had not been observed during the preceding governorship of Nāṣir al-Dīn on behalf of qara Yūsuf Qaraquyunlu.

As already noted, the two Quranic verses 11: 88 and 9: 129 are not cited verbatim, but rather combined in a composite form that clearly references the Quranic wording while reformulating it into something of a pastiche. While certainly showcasing the erudition of the scholar(s) responsible for coming up with the elaborate sajʿ of the text of inscription 835 Mardin on behalf of the new Aqquyunlu rulers, this citation of Quranic verses in a paraphrased form is unusual. The same mode of referencing Quranic verses reappears in line 5 (see below). As noted by the concise fifteenth-century summary of nine centuries of Quranic exegesis written under Mamlūk auspices by al-Jalālayn (Tafsīr: 264), Quran 9: 128–9, was held by some scholars to constitute the last verses that had been revealed to Muḥammad. Accordingly, the epigraphic deployment of the concluding part of Quran 9: 129 in line 1 of inscription 835 Mardin evokes the command of the divine voice of Quran 9: 129 to respond to deviation by declaring one’s trust in God (fa-in tawallū fa-qul etc.). Within the Aqquyunlu takeover of Mardin, this implies that the former Qaraquyunlu governor Nāṣir al-Dīn “deviated”, likely both specifically in imprisoning the Aqquyunlu prince and generally in his administration of Mardin. As spoken by the new Aqquyunlu governor Ḥamza b. qara ʿUthmān, the evocation of Quran 9: 129 arguably also declares his commitment to just rule and an unwillingness to acknowledge any debt to those who liberated him from prison and installed him as governor.

This downplaying of specific obligations to those who facilitated the Aqquyunlu takeover continues in the attribution of the Aqquyunlu occupation of the citadel of Mardin to divine intervention in line 2. Against the possible expectations of those who mounted the civil revolt to be favoured by the new administration, line 2 credits God alone with giving victory to qara ʿUthmān and empowering his son Ḥamza. In this regard, the description of God’s agency towards Ḥamza with the word sallaṭa, to empower, is particularly significant as this pertains to the same root as the term sulṭān (divinely supported) rule, which constitutes a key component of the titles deployed by pre-industrial Islamic rulers.Footnote ³⁷ Although this may be coincidental, the same root also figures prominently in the proclamation of Ḥamza’s rule during the civil revolt against Nāṣir al-Dīn in the Syriac continuation Barebraeus (Chronography: 195v, col. i). Here, the “women and slaves” in the citadel are presented as proclaiming their civil revolt by calling out “long live sulṭān Ḥamza b. ʿUthmān” (sūlṭān [!] ḥmzh br ʿūtmn), before closing the gate. Accordingly, attribution of Ḥamza’s empowerment to God at this very same gate also counters potential claims of his human backers to have facilitated his appointment as ruler over Mardin and its citadel.

Line 3 features the restoration inscription proper, commemorating the rebuilding of the gate, ostensibly furnishing the occasion for the commission of inscription 835 Mardin. Its unequivocal attribution of agency to the Timurid ruler Shāhrukh amounts to a declaration of allegiance to Shāhrukh by qara ʿUthmān Aqquyunlu and his son Ḥamza. The reading of Shāhrukh’s name in line 3 also sustains my proposed identification of his name on published photographs of the inscribed lock of this gate pace Etem [Eldem]’s suggested reading of this name as Ḥamza. This commemoration of a Timurid overlord in the medium of monumental epigraphy in south-eastern Anatolia is highly unusual, the only parallel being the foundation inscription at the Yelmaniye Cami or Medrese at Çemişgezek, which named Shāhrukh’s father Timur in 806/1403.Footnote ³⁸ Within the Timurid–Mamlūk struggles for hegemony in the 1430s,Footnote ³⁹ the unequivocal declaration of allegiance to Shāhrukh on a gate and its lock, while dispatching the keys of this very gate and lock to Mamlūk Cairo as a token of allegiance to Barsbāy, attests to the chutzpa of qara ʿUthmān’s Schaukelpolitik between the two supraregional powers.

Within inscription 835 Mardin, the declaration of allegiance to Shāhrukh in line 3 is singled out due to its adherence to a different prose rhyme (sajʿ). Against the rhyme on -īm/-īn that structures lines 1–2 and 4–6, line 3 alone features a rhyme on –am (al-amīr al-aʿẓam/al-khāqān al-muʿaẓẓam/al-ʿajam). The overarching structure of the entire inscription also carefully navigates the question of the extent of qara ʿUthmān’s factual subordination under Shāhrukh. Accordingly, the Aqquyunlu occupation of Mardin constitutes a victory granted to qara ʿUthmān (and not to Shāhrukh) by God, whereupon Shāhrukh ordered the rebuilding of the gate. A small lacuna between stones N and D stands within the titles of Shāhrukh: the remains of letters visible on Bell’s photograph do not readily lend themselves to his regnal laqab of Muʿīn al-Dīn as featured on Shāhrukh’s early coins.Footnote ⁴⁰

After a benediction of Shāhrukh, the subsequent catalogue of qara ʿUthmān’s holdings in 1432 as presented in line 4 is formulated in deliberately ambiguous terms. The governing verb ḥakama, “to rule”, may pertain grammatically either to qara ʿUthmān or to Shāhrukh, however, the exact correspondence of the toponyms listed in this catalogue to the Aqquyunlu realms clearly signals that factual rule is exerted by qara ʿUthmān.Footnote ⁴¹ Accordingly, the framing of the catalogue deliberately subordinates the entirety of qara ʿUthmān’s realms under Shāhrukh while presenting the Aqquyunlu holdings, among which the citadel of Mardin constitutes the latest addition, as a coherent entity. Line 4 thereby signals to local readers that Mardin is now integrated among neighbouring centres including Diyarbakır, Savur, al-Raqqa, and Urfa, implicitly promising an end to the raiding of Mardin’s hinterland and political stability within the Aqquyunlu realms centred on qara ʿUthmān (see Figure 6).

© d-maps.com https://www.d-maps.com/carte.php?num_car=699&lang=de

Figure 6. The toponyms listed in the catalogue of the Aqquyunlu realms in 1432 in line 4.

Grammatically, the toponyms of Urfa (al-ruhā), al-Raqqa, Raʾs al-ʿAyn, and Savur are distinguished among the catalogue of line 4 by their presentation in a form that includes the Arabic definite article al-. According to the normativities of Arabic fuṣḥā, this differing treatment of the toponyms in line 4 has been indicated in the transliteration given above by vocalizing the toponyms in question as triptotic (ʿalā … l-raqqati) as opposed to diptotic (ʿalā āmida). While many toponyms continue to be current today, Tercil, contemporary Yerhisar/Hazro/Diyarbakır was the old centre of the district now centred on Hazro/Diyarbakır. I tentatively read the letters possibly spelling *KWNJ* on stone N as Genç, indicating qara ʿUthmān’s rule over the castle of this name formerly located above the Murat River at contemporary Kale or Kaleköy/Solhan/Bingöl.Footnote ⁴² Akşehir (āqshahr) was correctly localized south-west of (Şebin-) Karahisar/Giresun by Woods (Reference Woods1999: 55). The name of the former regional centre lives on in the name of the hamlet of Akşar/Suşehri/Sivas. Finally, the three toponyms tentatively read as Kiğı, Palu, and Eğil stand on a very dilapidated part of stones N and D. My interpretation of the remains of letters as spelling these three toponyms is decisively motivated by the difficulty of coming up with a convincing explanation for why they should have been left out of a catalogue of qara ʿUthmān’s holdings in 1432.

So far, the Quranic intertextuality of line 1, the stability of the regnal names and titles in lines 2 and 3, and the historical context of qara ʿUthmān’s holdings in 1432 have facilitated a confident reading of lines 1–4. By contrast, the narrative of the captivity of Ḥamza in the citadel of Mardin and his subsequent installation as governor in lines 5–6 is unique among known Qaraquyunlu and Aqquyunlu “Turkmen” epigraphy. While continuing the Persianate tradition of an epigraphic rendering of text pertaining to the genre of “conquest narratives” (fatḥnāma),Footnote ⁴³ the only comparable fifteenth-century example of an Arabic fatḥnāma inscribed in stone in the Near East is the narrative of a Mamlūk campaign against Ottoman troops at the foundation of Yaʿqūb shāh al-Mihmāndār in Cairo dated to 901/1495–1496.Footnote ⁴⁴ This lack of context renders any interpretation of lines 5–6 very tentative.

Beyond the suggestion that line 5 on stone M paraphrases Quran 38: 26Footnote ⁴⁵ and that lines 5–6 together present a narrative of Ḥamza’s capture and subsequent appointment as Mardin’s governor, I can only suggest a confident interpretation of the date written across the end of line 5 and the beginning of line 6 as Sunday, Muḥarram 1, 835/September 9, 1431. The date and the day of the week are internally consistent. However, interpretation as the date of the internal revolt against the Qaraquyunlu governor is impossible due to the explicit indication of the Syriac sources that date the Aqquyunlu occupation to 1743 as, beginning on 1 October, 1431. This motivates my suggestion that this must be interpreted as the date of Ḥamza’s capture. According to this interpretation, the narrative of lines 5–6 was centred not on the internal revolt in 1432, but rather on Ḥamza’s capture and imprisonment, underlining his steadfast trust in God already suggested by the Quranic pastiche of line 1. This interpretation additionally resonates with the citation of the story of Dāwūd in line 5 as alluded to in Quran 38: 26, as Dāwūd was similarly remembered as exemplifying divine deliverance from a difficult situation. As the final portion of line 6 on stone J may spell the Arabic word for “year” (sana), inscription 835 Mardin may originally have concluded with the date of the Aqquyunlu takeover, if this was not the date of the conclusion of the rebuilding of the gate.

In its location above the main gate connecting the town of Mardin to the citadel, inscription 835 Mardin was complemented by a second inscription that was displayed on a massive metal lock, likely hanging in the right wing of the opened gate as facing a visitor ascending to the citadel (see below). Ibn Ḥajar (Inbāʾ: vol. 3, p. 473) states that qara ʿUthmān pre-empted an already organized Mamlūk invasion by sending the keys of the citadel of Mardin (mafātīḥ qalʿat mārdīn) to the Mamlūk ruler in 835/1432. Although these keys do not appear to have been preserved, it is highly likely that this ornamental lock pertained to one of them, facilitating its confident dating to the same year of the Aqquyunlu occupation of Mardin and its citadel.Footnote ⁴⁶ As the lock, formerly held as #378 in the Çinili Köşk of the Topkapı Sarayı Müzesi, cannot currently be located (see above), the edition of its inscription follows Etem [Eldem] (Reference Etem [Eldem]1936: 141–7), as complemented by the photographs published by Kühnel and Ogan (Reference Kühnel and Ogan1938: 31), as well as Arseven (Reference Arseven1952: 122) and the historical context of inscription 835 Mardin discussed above.

In lines 1 and 6, clearly readable letters are inscribed in a nonsensical sequence, which must have been intended to be emended ad sensum by the reader. These instances are briefly discussed in the commentary below. I retain the nonsensical sequence of the inscribed letters in my edition of the Arabic text, while emending to the implied form in the transliteration.

The inscription on the ornamental lock of the gate surmounted by inscription 835 Mardin can be edited as follows:

1. 1) ṣuniʿa bi-mā shāʾa llāhu taʿālā li-qalʿati mārdīna l-maḥrūsati bi-smi l-amīri l-kabīri ʿuthmāna bahādura khallada llāhu mulkahū

2. 2) ʿumila bi-rasmi l-sulṭāni l-aʿẓami māliki riqābi l-umami mawlā mulūki l-ʿarabi wa-l-turki wa-l-ʿajami shāhrukha bahādura khāna

3. 3) ʿamala l-muʿallimu ḥasanuni bnu rasūlin ghafara llāhu la-hū

4. 4) al-ʿizzu l-dāʾimu wa-l-iqbālu l-zāʾidu

5. 5) wa-li-ṣāḥibihī l-saʿādatu wa-l-salāmatu wa-ṭūlu l-ʿumri

6. 6) mā nāḥat ḥamāmatun wa-lā *ghurābatun W ilā* yawmi l-qiyāmati

1. 1) In accordance with the will of God, exalted be He!, this has been made for the citadel of well-protected Mardin in the name of the great commander ʿUthmān bahādur, may God make his reign eternal!

2. 2) It was made due to the command of the greatest sulṭān, the owner of the necks of the people, the master of the kings of the Arabs, the Turks, and the Persians, Shāhrukh bahādur khān.

3. 3) Made by the master Ḥasan b. Rasūl, may God forgive him!

4. 4) Eternal glory and growing fortune!

5. 5) May its owner be fortunate and well, enjoying a long life!

6. 6) May no dove coo nor raven croak until the Day of Judgement!

As described by Etem [Eldem] (Reference Etem [Eldem]1936: 141), the lock measured 535 x 33 x 6 milimetres and weighed 6 kilograms.Footnote ⁴⁷ According to Etem [Eldem] (Reference Etem [Eldem]1936: 141 and 147), it was made of iron, while Arseven (Reference Arseven1952: 122) suggests that it was made from bronze. In either case, the letters were incised into the metal without any suggestion that they were ever inlaid with another material.

The lock’s enormity agrees well with its proposed purpose of serving as an ornamented lock to the main gate connecting the citadel to the town, which is implied by the suggestion that it was made “for the castle of Mardin” in line 1. Within this utilization, visitors entering the citadel from the town would probably have seen the lock hanging in one of the two wings of the opened gate, displaying the main narrative inscription of 1 and 2 running downward along the main corpus of the lock. According to this reconstruction, the keyhole would have been oriented horizontally, within the beginning of lines 1 and 2. Parts of the latter sections of lines 1 and 2 would have been obscured by a fragmentarily preserved ornamented metal ring on a stick of unclear purpose that was fastened with a massive pin between lines 3 and 4.

In contrast to lines 1 and 2 that were displayed along the length of the corpus of the lock, lines 3 and 4 were inscribed on a metal plate covering its lower part. In the reconstructed context suggested above, they would have been legible written horizontally from right to left. Lines 5 and 6 are displayed on another metal plate facing 90 degrees away from the side that features lines 1–4. As this side joins the side displaying lines 1–4 clockwise, within the reconstructed display of the lock as hanging from one of the wings of the opened gate, lines 5 and 6 would have been visible to visitors entering the gate and looking to their right, if the lock hung to the right of them. This strongly indicates that the lock would have been displayed hanging in the right-hand wing of the opened gate as faced by a visitor ascending to the citadel.

Although the lock has been well preserved, part of the metal plate surrounding the keyhole and originally displaying the beginning of line 1 has broken off. Accordingly, the reconstruction suggested here is very tentative and based on the clearly visible letters Alif – Shīn – Alif at the beginning of the extant part and the expectation that the opening verb would have referenced the production of the lock “for the castle of Mardin” in some form. As already indicated by Etem [Eldem] (Reference Etem [Eldem]1936: 143), the nonsensical letters clearly readable as Qāf – ʿAyn – Lām at the beginning of line 1 must be emended ad sensum to spell qalʿa, citadel, just as the reader must supply the missing mater lectionis of Wāw in the passive participle of stem I, al-maḥrūsa, the well-protected. Although these misspellings are curious, they pose no problem to the confident reconstruction of the text that was intended to be communicated by the inscription on the lock.

Due to his reading of the name concluding line 2 as that of qara ʿUthmān’s son Ḥamza, Etem [Eldem] (Reference Etem [Eldem]1936: 143) suggested that the lock must be dated to after the death of qara ʿUthmān. Based on the photographs published by Etem [Eldem] (Reference Etem [Eldem]1936: 142), Kühnel and Ogan (Reference Kühnel and Ogan1938: 31), and Arseven (Reference Arseven1952: 122), however, the name in line 2 must be read as that of the Timurid ruler Shāhrukh. This reading is corroborated by the confident identification of Shāhrukh’s nominal overlordship over qara ʿUthmān as declared in lines 3–4 of inscription 835 Mardin as reconstructed above. Accordingly, lines 1–2 of the lock reinforce the declaration of joint Aqquyunlu–Timurid rule over Mardin following the occupation of its citadel in inscription 835 Mardin. In this context, the suggestion of line 2 that the lock was made due to a command of Shāhrukh echoes the parallel claim that the gate was built at the behest of the Timurid ruler in line 3 of inscription 835 Mardin. Similarly, the comparatively lesser titles awarded to qara ʿUthmān in line 1 of the lock, as compared to the more grandiose framing of his nominal overlord Shāhrukh in line 2, parallels the hierarchized presentation of both in lines 1–4 of inscription 835 Mardin.

Line 3 contains the signature of a craftsman. Although his name has been read as Naṣr by Etem [Eldem] (Reference Etem [Eldem]1936: 141), the published photographs make a reading as Ḥasan preferable. The name Rasūl given for his father is indicated as doubtful by Etem [Eldem] (Reference Etem [Eldem]1936: 141). However, the use of Rasūl as a personal name is attested for three individuals in the monumental biographical dictionary dedicated to the ninth/fifteenth century by al-Sakhāwī (Ḍawʾ: vol. 3, p. 200)(2003). Two of these are introduced with the nisba al-Kurdī, while the third is given the nisba al-Qayṣarī, possibly from the modern town of Kayseri. As Etem [Eldem]’s reading of the name as Rasūl fits the visual evidence of the published photographs, we can build on al-Sakhāwī’s attestation of Rasūl as a personal name in fifteenth-century eastern Anatolia to accept the reading of the name of the craftsman in line 3 of the lock as Ḥasan b. Rasūl as unproblematic.

Lines 4–5 display slogans of stereotypical praise of the artefact’s owner that also appear elsewhere in fifteenth-century Middle Eastern inscribed metalworks.Footnote ⁴⁸ The occurrence of these slogans on the lock may indicate that its manufacture was entrusted to a metalworker experienced in the production of portable metal artefacts inscribed with standard templates of praise. Accordingly, we should not overinterpret the occurrence of these slogans in lines 4–5, as they may simply have been inserted to decorate some less visible parts of the lock with suitable calligraphically arranged text.

By contrast, line 6 displays a rhythmicized and rhymed passage of Arabic sajʿ that is not attested elsewhere. The two words nāḥat ḥamāmatun at the beginning of line 6 intertextually reference the famous qaṣīda opening with aqūlu wa-qad nāḥat bi-qurbī ḥamāmatun, I say while the dove coos near me, of the tenth-century poet Abū Firās al-Ḥamdānī (Dīwān: 282)(1414/1994).Footnote ⁴⁹ This intertextual reference to Abū Firās (Dīwān: 282) in line 6 is very well suited to the epigraphic commemoration of the Aqquyunlu takeover of Mardin and its citadel, as this poem is one of the most emblematic passages of Abū Firās’ Rūmiyyāt commonly held to have been composed in Byzantine captivity.Footnote ⁵⁰ Accordingly, the first words of line 6 constitute a highly appropriate literary allusion to Ḥamza’s captivity preceding the civil revolt of Mardin in 835/1432.

Against the despair permeating the qaṣīda of Abū Firās, however, line 6 deploys the poetic image of the dove cooing near the prisoner in negated form, while extending the motif of the bird seen outside the window of an inescapable fortress to include a raven or crow (ghurāb).Footnote ⁵¹ As no raven appears in the qaṣīda of Abū Firās (Dīwān: 282), the addition of this bird likely constitutes a conscious innovation by the author of the sajʿ of line 6. Accordingly, the raven should here be interpreted as a harbinger of bad luck as current in Arabic–Islamic literary traditions.Footnote ⁵² The joint deployment of the dove and the raven in the wish that “neither should coo or croak again until the end of time!” in line 6 thereby amounts to an auspicious prayer safeguarding the newly installed Aqquyunlu rulers of Mardin from imprisonment and bad luck. Against the generic slogans of lines 4–5, line 6 amounts to a highly erudite and skilled deployment of poetic intertextualities and imagery that must have been composed expressly for the conclusion of the inscription displayed on the newly commissioned lock of the gate in 835/1432.

In their original context surrounding the main gate connecting the town of Mardin to the citadel, inscription 835 Mardin and the inscription on the lock complemented each other. Beyond the display of impressive craftsmanship in stone and metal, the Arabic text of both inscriptions is highly nuanced and signals the presence of impressive scholarly creativity and erudition among the courtly elites centred on the newly established Aqquyunlu rulers of the town. This scholarly erudition at the disposal of the new rulers is reflected in the careful composition of both inscriptions in Arabic rhymed prose (sajʿ) and the skilful implication of qara ʿUthmān’s formal subordination under the Timurid ruler Shāhrukh through their respective titles and the spatial arrangement of the texts. In addition, the mastery of multi-levelled intertextualities is particularly visible in the non-verbatim citation of Quranic verses in inscription 835 Mardin, as well as in the navigation of poetic intertextualities and motifs in the concluding line (6) of the lock.

Against the narrative imagery of the non-epigraphic accounts discussed in the introduction, this display of Arabic–Islamic learning arguably signals a commitment of the new rulers to administer Mardin according to the established normativities of Islamic law. Beyond the immediate rule of qara ʿUthmān and his son Ḥamza, this commitment to rule justly may additionally be reinforced by the explicit reference to Shāhrukh’s overlordship. Accordingly, the two texts displayed at the gate linking the town to the citadel declare the commitment of the new administration to rule in accordance with Islamic normativities, while integrating Mardin in the regional administrative configurations of the Aqquyunlu realms and the supra-regional standard of Timurid hegemony.

Outlook and conclusion

Together, inscription 835 Mardin and the inscription on the lock that was commissioned to close the gate surmounted by inscription 835 Mardin crucially complement the narrative and diplomatic accounts of the Aqquyunlu occupation in 835/1432. While corroborating the chronological sequence of Ḥamza’s capture, the subsequent campaign of his father qara ʿUthmān, and the internal revolt ousting the Qaraquyunlu governor, the reconstruction of both inscriptions as undertaken in the present article is particularly important as evidence for the self-representation of the new Aqquyunlu rulers immediately after their occupation of Mardin.

As suggested in the introduction, the independent narrative accounts of Ibn Ḥajar, al-Ghiyāth, and the anonymous continuator of the Chronography of Barebraeus are focused on the agency of the Qaraquyunlu governor Nāṣir al-Dīn and the subaltern members of his court. By highlighting the narrative turning points of the civil revolt with topic Wandermotive, these narrative accounts resonate with the storytelling of fables and legends, arguably framing Mardin as a site of contemporary mirabilia ultimately attributed to divine intervention. Although the two epigraphic sources commemorating the Aqquyunlu takeover of Mardin in 835/1432 also attribute this outcome to divine agency, they instead declare a stable political hierarchy of Ḥamza, his father qara ʿUthmān, and their nominal overlord Shāhrukh. Intertextually, the two inscriptions reference Quranic and poetic text, as well as the established titles and benedictions used in the courtly representation of Timurid and “Turkmen” rulers. Against the fairytale resonances of the narrative sources, the two inscriptions manifest a consolidated claim and commitment by the new masters of Mardin to rule fairly. In its phrasing, this would have resonated immediately with the two standard vectors of Islamic rulerly self-representation in the Friday-prayer (khuṭba) and in coinage (sikka).

Within the corpus of Qaraquyunlu and Aqquyunlu “Turkmen” courtly epigraphy, inscription 835 Mardin is unique as a conquest narrative (fatḥnāma) that was epigraphically displayed at the nexus between the town of Mardin and its citadel, which served as the seat of the newly installed Aqquyunlu administration. Although it is not as transparently narrative, the navigation of Arabic poetic intertextualities and imagery in the concluding line 6 of the inscribed lock similarly references the extraordinary circumstances leading to the incorporation of Mardin in the Aqquyunlu realms. As a portable metal object inscribed with text selected expressly for its intended designation, the inscription on the lock of the citadel of Mardin is paralleled within “Turkmen” courtly epigraphy only by the inscribed ornamental stand for an oil lamp (chirāghdān), which was commissioned by qara ʿUthmān’s grandson uzun Ḥasan for the shrine of a certain Bayrām bābā walī, possibly at Ankara, as edited by Melikian-Chirvani (Reference Melikian-Chirvani1987: 126–31). Although both pieces are undated, the lock produced for the citadel of Mardin differs from the lampstand by constituting part of a coherent ensemble that can be reconstructed and dated.

As indicated above, inscription 835 Mardin and the inscription on the lock are also unique within extant Aqquyunlu and Qaraquyunlu “Turkmen” epigraphy in declaring (nominal) subordination to the Timurid ruler Shāhrukh. This epigraphic declaration of qara ʿUthmān’s subordination under Shāhrukh is almost certainly conditioned by the political context of his occupation of Mardin in 835/1432. Notwithstanding qara ʿUthmān’s sending of the keys of the citadel to the Mamlūk sulṭān, the following year of 836/1433 saw a massive Mamlūk invasion led by Barsbāy himself that culminated in an unsuccessful siege of Diyarbakır between 28 May and 2 July 1433 ce, which qara ʿUthmān waited out at the nearby fortress of Ergani.Footnote ⁵³ As recently argued by Leube (Reference Leube2025: 120–24), the successful weathering of the Mamlūk invasion was subsequently commemorated by qara ʿUthmān together with the Armenian bishop of Diyarbakır in the foundation of an Armenian church inside the citadel of Ergani. Although the Armenian foundation inscription of this church has been destroyed in the early twentieth century, the text as reconstructed by Leube (Reference Leube2025: 121) only refers to the political agency of qara ʿUthmān without any reference to Shāhrukh’s (nominal) overlordship. Accordingly, the performative attribution of the commission of the new gate and its lock to Shāhrukh reflects the precarious situation of qara ʿUthmān navigating Mamlūk and Timurid overlords after his occupation of Mardin. Subsequently, his attempted Schaukelpolitik was conclusively resolved by the failure of the Mamlūk invasion and siege of Diyarbakır, leading to the unambiguous assertion of qara ʿUthmān’s claim to independent rule after 836/1433.