Religion, politics and an apocryphal admonition: the German East African “Mecca letter” of 1908 in historical-critical analysis

Abstract

This article analyses a Muslim missive, which was circulated in German East Africa in 1908. Erroneously dubbed the “Mecca letter”, it called believers to repentance and sparked a religious revival, which alarmed the German administration. Their primarily political interpretation of the letter was retained in subsequent scholarship, which has overlooked two important textual resources for a better understanding of the missive: the presence of similar letters elsewhere and the fourteen copies still available in the Tanzanian National Archive. Presenting the first text-critical edition of the letter, together with a historical introduction of the extant specimens and a textual comparison to similar missives elsewhere, the article argues that the East African “Mecca letter” of 1908 was nothing more than a local circulation of a global chain letter. As such, its rapid transmission was not connected to a single political agency, but was likely prompted by a large variety of motivations.

In the summer of 1908 a religious missive alarmed the colonial administration in German East Africa. It was found in all major coastal towns and some places in the interior, and according to German reports it coincided in many places with signs of a religious revival: mosques were full, ḏikr was intensified, repentance was preached, people anticipated the end of the world, women joined in prayers, and concubines withdrew from their arrangements with soldiers or German officers. This sudden and fairly disruptive appearance of this letter in German East Africa took German colonial officials by surprise, and they suspected that this was a co-ordinated attempt to incite an uprising in the name of Islam.

The content of the letter itself gave little rise to such suspicions. Mislabelled as the “Mecca letter” in the German debate, the missive claimed to originate from the guardian of the Prophet's tomb in Medina and urged its recipients to a more faithful adherence to the tenets and practices of Islam. The ostensible author, a rather generically named Šayḫ Aḥmad, reports that he saw Muḥammad in a dream and delivered his warning to the ummah, accusing the community of the (not so) faithful of a litany of aberrations, from neglecting prayer and alms to drinking alcohol and engaging in gossip. This is followed by a warning about the final judgement being at hand, giving Muslims one last chance to repent of their sins and separate themselves from sinners. The letter closes with various assertions of veracity and instructions to copy and spread its message.

Despite the lack of overt political instructions, the Millennial undertones of the letter, its call for separation, and its effect on the Muslim population lent themselves to a political reading. Accordingly, members of the colonial apparatus suspected political intentions and began to look for a mastermind behind the spread of the letter. A suspect was found in the former slave and ivory trader Muḥammad bin Ḫalfān, better known as Rumaliza. He and his family had a considerable stake in the pre-colonial caravan economy around Lake Tanganyika, and Rumaliza had resisted the installation of German and Belgian rule for as long as he could (Swann 1910; Martin 1969). While some of his associates got caught up in the German conquest, Rumaliza escaped fairly unscathed and resettled in Zanzibar, from where he continued to trade and even won a settlement in the German courts against a former business partner. When some of the first “Mecca letters” obtained were traced to the Rumaliza clan, the colonial administration took this as evidence of Rumaliza's continued “resentments” against German rule and concluded that he had attempted to mobilize Muslim piety and apocalyptic hopes against them via this only ostensibly harmless letter.

Fears of a Muslim uprising proved to be unfounded and reports of religious fervency quickly subsided. Yet the letter nonetheless marked something of a turning point in German colonial debates about Islam. Ever since the anti-Arab rhetoric of the conquest war of 1888/9 (Glassman 1995; Haustein 2018), the German policy towards Islam in East Africa had been one of accommodation and integration as the Empire sought to inherit and supplant the socio-economic fabric of the Omani-Swahili elites of the coast. Local jurisdiction and governance was left intact, and slavery – though a primary reason for the German conquest in the first place – was never abolished (Haustein 2017). Christian missions were tolerated but not promoted, and the government built up its own “religiously neutral” school system which was small in comparison but fed directly into the colonial apparatus. As a result the non-German layers of administration, police, and military were almost exclusively Muslim, and with the continued importance and spread of Swahili culture and language to the interior, conversions to Islam increased noticeably. Missionaries bemoaned these developments and attacked the government's alleged “Islam-friendly” policies, but their voices were largely ridiculed in the colonial press and ignored in policy debates (Haustein 2018).

From the “Mecca letter” onwards, however, the missionary warnings against the political potency of Islam found their way back into mainstream colonial thought. This was due to several factors. First, the sudden appearance of the letter provoked comparisons with the Maji Maji war of 1905–06, which had been driven, at least in the German interpretation, by chiefs mobilizing “traditional religions” and “sorcery” to incite a concerted rebellion (Monson 2010). Although there was no discernible Muslim involvement in the Maji Maji insurgency, the quick spread of the “Mecca letter” and its accompanying rumours lent it to be read as another “religious” uprising attempt.¹ A second reason was entirely due to German colonial politics. Governor Albrecht Rechenberg, appointed in 1906 to rebuild the colony in the wake of the Maji Maji war and various economic failures, pursued a new strategy of promoting indigenous production and trade for the stimulation of tax revenues. This brought him into regular conflict with the interests of settlers, who essentially strove for a state-subsidised plantation economy on the back of African wage labour (Iliffe 1969). Inasmuch as the “Mecca letter” could be used to undermine Rechenberg, settler papers now joined the formerly dismissed missionaries warning against the “political danger” of Islam.

A third factor in the catalytic effect of the “Mecca letter” on the German assessment of Islam in East Africa was the contribution of scholarship. For the nascent Islamwissenschaft the “Mecca letter affair”, as it came to be known, was an opportune incident to demonstrate the usefulness of their expertise and exert some influence on policy making. The two experts consulted about the letter were Carl Heinrich Becker of the Colonial Institute in Hamburg, one of the first German scholars to break with the Orientalist tradition towards a more contemporary analysis of Islam, and Max von Oppenheim, whose later strategy papers would shape German efforts to instrumentalize Islam in the First World War (Schwanitz 2004). Both scholars agreed with the government's perception that the letter was an effort of agitation against German colonial rule, but offered different ideas about its origin. Becker saw the most likely point of origin as the “fanatic” population of the Somali coast from where the letter reached East Africa through one of the ṭuruq.² Von Oppenheim, by contrast, mused that French propaganda had utilized East African Muslim networks as part of their anti-German efforts since the First Moroccan Crisis.³ What is common to both is the understanding of a primary political intent behind the letter's spread (Haustein 2018). In Becker's later scholarship (1909, 1911), the “Mecca letter” became one of the exhibits for the political potency of Islam in the German colonies. While he did not follow the general warnings against an “Islamic danger” and emphasized the plurality of Islam, he contended that the political ideal of Islamic unity could easily be employed by Arabs or other interested parties in general agitation efforts, and that therefore a strict surveillance of Islam was necessary, as well as a quick and firm reaction to propaganda efforts just like the colonial authorities had demonstrated in the case of the “Mecca letter”.

Post-colonial scholarship retained the impetus of identifying the “Mecca letter” with a main political agent or motivation, even as their assessments of the actual reasons differed. B.G. Martin's (1969; 1976: 153–76) interpretation was largely based on the German colonial claims about Rumaliza's involvement, while offering an additional political layer by considering the role of Sufi ṭuruq and Omani political factions. The network behind the spread of the “Mecca letter” was therefore presented as a “Muslim alliance that included Rumaliza, his sons and relatives, a number of Comorians and Bravanese, and some Swahilis – most of whom shared a membership in the Qadiriya brotherhood and many of whom had ties to the Hinawi group” (Martin 1976: 173).⁴

Although Martin failed to substantiate his claims of Qādirī membership for a number of people involved, and overestimated the role of others in the ṭarīqah (Nimtz 1980: 202, n. 8), the involvement of Qādirī networks seemed plausible to other scholars as well (e.g. Iliffe 1979: 211–12). Michael Pesek (2000; 2002; 2003) plausibly centred his analysis on the political dynamics behind the letter on the spread of Sufism in the colony. He contended that German ignorance of the spreading Qādirīya had left them blind-sided about the actual dynamics of the letter's spread, while the “Orthodox establishment” used the government's nervousness about the “Mecca letter” to rid themselves of unwanted Sufi preachers while asserting their political loyalty.

Felicitas Becker (2010) saw a close political analogy between the Maji Maji uprising and the “Mecca letter” on account of a shared “millenarian hope”. She acknowledged that the letter reached far beyond the Maji Maji area, but argued that it appealed to the same stratum of society that had risen up there, namely fairly uneducated, poor Africans. Asserting that the “letters promised deliverance from the present political predicament and a new kind of citizenship unfettered by European rule in a universal Muslim piety”, she concluded that now “Islam could authorise claims and expectations similar to those of the maji” (F. Becker 2010: 311–2).

Jennifer Kopf (2007) has offered another interpretation of the political dynamics behind the spread of the “Mecca letter”. Focusing on female agency as evident in the reports about women's prayer and the breaking of concubinages, Kopf interpreted the Mecca letter as a female challenge to the male-dominated sexual order in the colony. While she did not allege that the Mecca letter was conceived or spread solely to this end, rebellion against sexual oppression nonetheless became the central dimension of her analysis of the letter's political potency and the German reaction to it.

These attributions of the “Mecca letter” with a single political intent or dynamic tend to overlook two bodies of sources that may yield a slightly different perspective. The first consists of very similar letters from different geographical regions and different times. Snouck Hurgronje (1923, orig. 1888) had already recorded such a letter in Indonesia in the early 1880s, something which only Becker (1909: 168; 1911: 45) briefly acknowledged, albeit without much consequence for his analysis. More recently, Jonathan Katz (1994) has presented three different versions of the letter held in the Bibliothèque nationale in Paris, dating from 1844 onwards and obtained in North Africa, Iraq, or India. Juan Cole's (1999: 205) study of the ʿUrābi movement pointed to a version of the letter appearing in Cairo in 1877, an English translation of which is still extant in the UK Foreign Office files.⁵ Gajendra Singh (2014) has discussed a version found among Indian soldiers fighting in the British Expeditionary Force in the First World War. Finally, the letter has even survived into the information age, as evident in a forum exchange about a very similar chain email (Islamic Board 2005). All of these versions are close to the East African “Mecca letter” in overall content and thrust, indicating the need to place the specimens of 1908 more firmly into a global and longstanding circulation of the letter.

The global presence of the letter does not disprove political intent in relation to the German East African circulation, however. In fact, Becker (1911: 45) reconciled the existence of the Indonesian version with the question of political intent, by asking in whose interests it might have been to spread the letter on the East African coast. This is where the second, neglected, body of sources comes in: the extant copies in the German East African archives, completely overlooked by scholars so far. Becker only managed to obtain one specimen of the letter as the basis of his text and translation (1911), and later scholars have relied on Becker's version and the German political correspondence only. Yet the multiple Arabic copies held by the Tanzanian National Archives (TNA) allow for a much more detailed and historical-critical analysis of the letter's content and transmission, alongside the specific information about each copy in the accompanying correspondence.

This article presents a text-critical edition of the East African “Mecca letter”, alongside a comparative analysis of its content with previously known versions of the same missive. It aims to provide a better understanding of the letter's spread in German East Africa, including the question of political intent. The analysis begins with an individual presentation of extant Arabic copies, including their historical context. This is followed by a presentation of the text-critical edition of the letter, sectioned by the internal structure of the letter and discussed in comparison with the versions outside of East Africa. The textual evidence will establish very clearly that the letter was part of a global circulation, spread in different varieties, along multiple paths, and via various agents. As the conclusion will argue, this establishes the character of the missive as that of a chain letter, which in turn necessitates more plural analyses of the intentions and agencies behind its proliferation than scholarship has previously offered.

1. Extant manuscripts

The German records in the TNA contain 14 Arabic versions of the letter. They were collected between mid-1908 and early 1909. One is of unknown origin and the others were obtained in Lindi (1 copy), Kilwa (2), Bagamoyo (3), Zanzibar (2), Mahenge (1), Kilossa (2), and Tanga (2). Most had been collected and forwarded by German officials in response to a request by Governor Rechenberg. All copies and the accompanying reports were kept in a new file titled “Religious Movements”, which had been created in response to the “Mecca letter affair” and became the first archival file with a systematic collection of information on Islam in the German East African administration.⁶ The following presentation discusses the specimens in the order of their appearance in the archive, which does not necessarily mirror their creation or circulation date.

1a. Version of unknown origin (Ukn)

Ukn is inserted in the beginning of the file, directly after the first correspondence about the “Mecca letters” in the German archive.⁷ Unlike the other versions, the document has no administrative markings and is not referenced in an accompanying letter. It is directly followed by the Lindi version (Lnd, see below), which is marked as a handwritten copy of the original and was submitted together with a draft translation.⁸ The most straightforward hypothesis therefore would be that Ukn was the original to the Lnd administrative copy, but a number of important textual deviations make this an unconvincing hypothesis.⁹

Another possible place of origin for Ukn would be Dar es Salaam. Governor Rechenberg indicated that he had been able to obtain a copy of the letter in Dar es Salaam, copies of which he sent to Major Schleinitz of the Protectorate Army and to various district offices as reference.¹⁰ Rechenberg forwarded a copy of this letter to Berlin, where it was kept in an envelope marked “Original of so-called Mecca letter”.¹¹ Upon request, this specimen was sent to C.H. Becker and returned later that year, but it is no longer found in its original place.¹² Becker's (1911: 43–4) transcript of the letter, however, contains multiple deviations from Ukn. Many of them could be seen as corrections of obvious language mistakes, but there are significant deviations at the end where Becker presents phrases and sentences not contained in Ukn.¹³

Ukn is written in thick but uneven black ink, as would have been produced by dip pen. It was folded immediately after completion as evidenced by the faint mirror imprint of the bottom characters on the top part of the letter. The writing is neat if a bit hurried, and in addition to a few smudges, text is crossed out in two places. The text is fully vocalized.

1b. Lindi version (Lnd)

This specimen follows immediately after Ukn and is marked as folio 4a. It was signed by the district officer Karl Wendt as “Copy of the Mecca letter (from the original) 31/7 Wendt” and submitted together with a draft translation (folios 4b and 5). The manuscripts may have been enclosed in Wendt's full report of 1 August 1908, though there is no reference to appendices here.¹⁴ Another possibility is that they were handed to Major Schleinitz, sent to inspect the situation in Lindi in early August.

Lnd is written in even, thin black ink, most likely in fountain pen as would have been typical for the German administration. There are green pencil marks between sentences at the end of the letter. The letter also contains about two dozen minor corrections in red pen from a different hand, all of which clarify spellings or aim to correct grammatical errors. An accompanying note by the experienced colonial officer Hans Zache states: “The red additions show in which places the Text does not show correct Arabic. 12/8” Becker's (1911) rendition of the text appears to be based on a copy of this manuscript and offers similar corrections to Zache.

According to Wendt's reporting, the town's qāḍī ʿUmarī (“Omari”) had received the letter on 8 July from Naṣr bin Ḫalfān, brother of Muḥammad bin Ḫalfān alias Rumaliza. Allegedly, Rumaliza had sent the letter to Naṣr via his son Ḥilālī. This family connection led Wendt to believe that he was witnessing a plot, and consequently he arrested Naṣr bin Ḫalfān, believing him to be in the epicentre of the letter's spread on the coast.

1c. Kilwa versions (Kwα, Kwβ, Kwγ)

The next three manuscripts follow a letter by the Kilwa district officer, Ludwig Schön.¹⁵ On 2 August 1908, Schön was alerted by Schleinitz to the “Mecca letters” in Lindi. He was told to investigate whether the letters had arrived in Kilwa via ʿAbdallah bin ʿUmarī, who according to Wendt had travelled there from Lindi. Within a day, Schön had obtained two copies of the letter and forwarded them to Dar es Salaam on 5 August 1908.

According to Schön's report, both letters had originated in Zanzibar, but reached Kilwa at different times. One had been brought two months earlier by Rumaliza's son Hemedi and (via a third party) had reached the mwalim of the Kilwa mosque, who copied and distributed it to “coloureds” in the Kilwa district. The other version had arrived more recently via a mwalim Ḫamīs (“Hamiss”) from Zanzibar. According to Schön, Ḫamīs made a number of contradictory claims when interrogated, but admitted after a “long cross-examination” that he had received the letter from a member of the “Rumaliza clan”. For Schön, this was enough to incarcerate Ḫamīs and ʿAbdallah bin ʿUmarī “for danger of collusion”.

The appendices to Schön's report contain three Arabic and two Swahili versions.¹⁶ The first Arabic version of the letter (Kwα, f 40) is labelled as “Anlage I” (appendix I), but seems to be the second version of the letter mentioned by Schön, because it is signed by the “šayḫ servant of the order sworn into the tradition of the Qadiriyya, šayḫ Ḫamīs bin Aḥmad bin Mafum bin Yūsuf Širazī”.¹⁷ It is dated to 2 jumādā al-ūlā 1326, which translates to 2 June 1908. The manuscript is written in black ink (most likely dip pen), in neat and clean handwriting, as one would expect from a learned Zanzibar sheikh. It is not vocalized, but is consistent in its use of šaddāt. There are a few secondary markings of German origin, one of which denotes the file number, identical to the Kilwa report.

Kwβ has no original signature or date. It is marked in pencil as “Anlage II” (appendix 2), but lacks a folio number or reference number. The text is written in thin black ink, most likely fountain pen. With minor exceptions, it is not vocalized, and the writing is a bit uneven with a couple of crossed out corrections.

Kwγ is not signed or dated. It is inserted in the file as folio 41 and labelled as “Anlage III” (appendix 3). The file number (in pencil) is identical to Kwα. The letter is written in purple pencil, most likely copying pencil. The handwriting is hurried, with some crossed-out mistakes and superscribed corrections. The text is almost identical with Kwβ, apart from minor spelling deviations and a more frequent (but equally inconsistent) vocalization. Some vocalizations are added in slightly darker pencil and are probably secondary. The writing implement, the textual proximity to Kwβ, and the added vocalizations make it most likely that this was an administrative copy of Kwβ for translation purposes.

1d. Bagamoyo versions (Bgα, Bgβ, and Bgγ)

Based on the submissions by Lindi and Kilwa, governor Rechenberg drew up a report for the Colonial Office, dated 12 August 1908, which was also distributed to the district offices and the German consulates in Mombasa and Dar es Salaam with a request to report on similar movements.¹⁸

The first response to this request came from the provisional district officer of Bagamoyo, the court clerk Dinkelacker.¹⁹ He reported that three letters had been found, which he identified with Roman numerals in his report. The first one (I) he understood to be the “original letter” and the second (II) as a “not wholly precise copy” of I. The third manuscript (III) was introduced as a free adaption of I “if not the copy of an entirely different letter”. According to Dinkelacker the “original” had been received towards the end of July by a certain “Abubakar”, characterized as an “Arab mwalim” who taught “ilmu” (ʿilm al-kalam) to about 50 students, administered oaths, and performed weddings and other religious ceremonies. Martin (1969: 484) identified him – most likely correctly – as the Qādirīya šayḫ Abū Bakr b. Ṭaha al-Ǧabrī al-Barāwī, who was well connected to Qādirīya šuyūḫ elsewhere in the colony and to the Rumaliza family.²⁰

Three Arabic versions follow the report (f 50–52), all of which carry Roman numerals and the file number of Dinkelacker's report, but there is some confusion as to the correct identification of the appendices. Bgα is inserted directly after Dinkelacker's letter, marked in blue as “I” and as folio no. 50. In pencil it also carries a “II” and the same file number as the cover letter. It is likely that these pencil marks are primary and the blue marks were added secondarily when the archival file was organized, re-sorting the two appendices. This would make Bgα the second letter Dinkelacker mentioned in his report. It contains no date, sender or addressee, nor did Dinkelacker mention how and from whom this letter was obtained. However, his assertion that this letter was a “not wholly precise” copy of the “original” (Bgβ) is unlikely due to several significant deviations and additions. As the critical edition and discussion below will show, this version belongs to a different branch of copies which already developed in Zanzibar (if not before). The text is unvocalized and written in a fairly neat hand in thick black ink (most likely dip pen). A notable feature is a closing colophon in sentence 17 (“ه ه ه”) which marks the end of the purported original message. This is followed by a closing prayer from the copyist, followed by the date of 12 jumādā al-āḫirah 1326, i.e. 12 July 1908.²¹

Bgβ is labelled in blue as “II” and folio no. 51. Written in pencil at the top, it has the same file number as Dinkelacker's letter, a crossed out “I”, a crossed out “III”, pointing to the later reorganization of the letters. At the bottom there is a repetition of the file number (in black ink) as well as the name “Abubakar” (in pencil). This is how Dinkelacker spelled Abū Bakr, whom he identified as the recipient of the “original letter” in Bagamoyo. A further indication that Bgβ is indeed the letter Dinkelacker identified as Abū Bakr's “original” copy is in its close textual proximity to Lnd, Kwβ/Kwγ, and Znα, as the text-critical analysis will establish. These letters had been identified with members of the Rumaliza family, something Abū Bakr also admitted when later questioned in Dar es Salaam.²² Bgβ is written in thin, but fairly uneven black ink (dip pen). The text is unvocalized in even writing, though less neat than Bgα.

Bgγ (only labelled in blue as “III”, folio no. 52) is different from the other two. It is written in black ink (most likely fountain pen) on thin lined paper turned sideways. The handwriting is small and hurried, the text is unvocalized. Due to the brittle nature of the paper, fragments are missing from the text and a fold in the bottom third further obfuscated parts of text, but the text could be reconstructed from an earlier photograph of the archive file and the German transliteration. The text of Bgγ deviates significantly from all of the other copies and was not therefore considered for the text-critical edition below.²³ These deviations simplify and condense the letter's message, which makes it unlikely that the text was arranged by Abū Bakr, as Dinkelacker claimed. A more likely interpretation is that it was a written recitation from memory by someone who had read or heard the letter.

1e. Zanzibar versions (Znα and Znβ)

The next submission of letters came from the German consul in Zanzibar, dated 6 September 1908.²⁴ The first specimen was brought by members of the Rumaliza clan seeking clemency, as by now Naṣr bin Ḫalfān was not the only one in German prison; Rumaliza's son Hemedi was there too.²⁵ Ḥilālī bin Moḥammad, Rumaliza's other son, who had himself been accused of bringing the letter to Lindi, now approached the German consulate. Leaving a “depressed and fearsome impression” on the consul,²⁶ he denied having transmitted the letter to the coast himself. Upon the consul's request, he brought a copy of the letter which had been on display in the mosque in Ng'ambo, the new city adjacent to Stone Town.

The second letter was brought by a relative of the likewise arrested ʿAbdallah bin ʾUmarī, who admitted that ʿAbdallah had brought the document to the coast, but insisted that he and his family did not know the Rumaliza clan. Furthermore, he stated that the letter was spread with purely religious motives, and that there were hundreds of others who had spread the letter in East Africa. He had received his letter from a šarīf Muḥammad bin Aḥmad, who, in turn, was summoned for testimony. Šarīf Muḥammad told the consulate that the letter had been sent to him from the Ḥiǧāz, where he had connections owing to a prior visit, but he failed to provide (or actively withheld) further leads. He claimed that the letter had been sent anonymously in a closed envelope, with a postal seal from the Ḥiǧāz but no stamp. Having disposed of the envelope, he could not remember what town the stamp was from.

In the archive, the specimen from the Ng'ambo mosque (Znα) immediately follows the report from Zanzibar (folio 63). It is marked as “Anlage 1” (appendix 1) and carries a file number matching the report.²⁷ As per the consul's report, the submitted document is a copy and not the original, which had visible marks from being fixed to a wall and was probably returned. The text is penned in black ink in fairly neat and practised handwriting. It is unvocalized and contains no dates or signatures.

The second specimen (folio 64, here Znβ) is a copy of the letter brought by the relative of ʿAbdallah bin ʿUmarī. It is marked as “Anlage 2” and carries the same file number. The copy is also penned in black ink in a very similar handwriting to Znα, suggesting that both letters were copied by the same clerk. The text is unvocalized and contains no dates or signatures. The original, according to the Consulate's report, was seen by multiple “local Arabs”, who estimated that it did originate from the Ḥijāz due to its flawless style and lettering.

1f. Mahenge version (Mhn)

On 31 August 1908 Captain Gideon von Grawert of the military post in Mahenge submitted two letters to Dar es Salaam, which he had obtained from the local imām, Tum bin ʿAlī.²⁸ The first one von Grawert understood to have been of a different genre from the other “Mecca letters”, more akin to what the Germans believed to be a widespread “exegesis” of the letter with a more overtly political bent. The letter was written by a person from Kilwa to a recipient near Liwale, and was brought to mahenge (approx 100 km from Liwale) by an unknown person and given to the son of the imām. According to von Grawert, this letter stated explicitly that the mahdī would arrive in four years, something which he had also found to be spreading as an oral rumour. Regrettably, the letter is no longer extant. In 1910 it was forwarded to the district office in Kilwa to aid with the investigation against the imprisoned persons there, but the relevant district office files were lost in the First World War.²⁹ It is unclear whether von Grawert's reading was verified in Kilwa, but it seems to have been of little use and is not mentioned in subsequent reports about the proceedings there. It is possible, therefore, that it was just another rough copy like Bgγ or a completely unrelated document.

The second letter von Grawert submitted had been sent to Tum bin ʿAlī anonymously from Morogoro. The specimen is still contained in the archive in its original form, and its content agrees with the other “Mecca letters”. It was written in ink pen (dip pen) in fairly neat handwriting, with one minor ink spill in the middle. It is not dated or signed by the copyist, but carries the usual secondary administrative markings which clearly identify it as the relevant appendix to von Grawert's report. The text is not vocalized and contains no šaddāt.

1g. Kilossa versions (Ksα and Ksβ)

These versions were submitted in September 1908 by the district officer of Morogoro, Arnold Lambrecht.³⁰ In his rather general accompanying report about a rising religious intensity among Muslims in his district, he simply noted that he was submitting “a number of Arabic writings that were confiscated in the district or handed in by the recipients themselves”, and were alleged to be copies of the “Mecca letter”.

The letter notes that five appendices were submitted, but only three are extant in the archives, with very brief accompanying notes by Lambrecht. The first is a Swahili letter in Ajami script which, according to the note, was written by Mwalim Punja in Kisara to Wali Hamid in Kilossa and received in early June 1908. The two others, in turn, are Arabic manuscripts, both of which, according to Lambrecht's accompanying note, were sent from šarīf ʿAbdallah bin Aḥmad Ṣadīq in Bagamoyo to the “Arab” Ṣaliḥ bin ʿAbdallah in Kilossa.

Ksα directly follows Lambrecht's note as folio 82. It has no other administrative markings. The manuscript was written in fairly thick black ink on thin chequered paper. The handwriting is dense but fairly even, and there is only one scribbled out mistake. There is no vocalization and no šaddāt.

Ksβ follows as folio 83, with no further administrative markings, but some secondary pencil annotations. One of these names the author of the letter and the others seek to clarify the spelling of individual Arabic words. The manuscript is written in thin black ink, probably fountain pen. The handwriting differs from Ksα: it is less condensed and practised, but even nonetheless. The note has one scribbled out correction. Like Ksα, the text is unvocalized and contains no šaddāt. Both letters are very close in content, with only minor deviations.

1h. Tanga versions (Tnα and Tnβ)

The last two version of the “Mecca letter” in the archive were submitted by the district officer of Tanga, Max Nötzl, on 30 October 1908.³¹ In his brief report, he mentioned that the imām of a mosque in the nearby town of Korogwe, Muḥammad bin Sultāni, had requested a copy while visiting Zanzibar in August, which he received by way of a šayḫ ʿAmir.³² Passing through Tanga, he made a copy and gave it to the local qāḍi Ḥamis, who in turn distributed it further to the imām of Tongoni, an old fishing village south of Tanga. The letter had not caused any stirs, and the local authorities had only heard about it in response to repeated requests for information.

The second copy submitted, Nötzl mentioned, had been obtained from a certain Majidi, imām in the Tanga quarter of Chumbageni. The district officer seemed to think that this was a copy of the same letter and did not therefore submit any additional information. The two letters differ significantly, however, as will be shown below.

The order of the letters in the archive is opposite to that in Nötzl's report. The first letter (Tnα, folio 109) has the same file number as Nötzl's letter (2855) and is marked (in pencil) as “copy of the Mwalim Majidi, [?] in Tanga”.³³ It is written in black ink and a fairly uneven hand. The text is unvocalized and uses šaddāt only intermittently. The letter is dated to Thursday, 25 Jumādā al-ūlā 1326, which translates to 25 June 1908. It is not signed by the copyist. On the bottom, the letter has two lines written in Ajami Swahili, which appear to come from the same hand.³⁴

The second letter in the archive (Tnβ, folio 110) also carries the file number 2855 and is marked in pencil as “Copy brought by Mohamed bin Sultani from Korogwe”. It is written in black ink and a very uneven hand, with the first four lines being considerably larger than the remainder of the letter and the handwriting differing between both sections. Nötzl noted that the letter was only partially copied by Muḥammad bin Sultāni, but offers no hypothesis as to who copied the second part. There are numerous corrections throughout the letter. It is not signed or dated.

2. Critical edition and interpretation

With the exception of Bgγ, all letters are similar enough in content to warrant the compilation of a single edition with a critical apparatus.³⁵ In order to keep the text manageable, only semantically relevant differences have been recorded, not divergent spellings of the same word or phrase. Spelling deviations have only been corrected if at least one specimen contained the standard spelling.

Despite the overall textual agreement, two major areas of difference emerge. First, the text was traditioned in two major variants in sentences 8–10, the shorter of which most likely emerged from a copying error. Second, there are major variations in the closure of the letter, with copyists often adding their own assertions of veracity, admonitions, or dates.

The critical edition will be presented, translated and discussed in nine sections which comprise the internal structure of the letter.

2a. Letter opening

1. 1. بسم الرحمن الرحيم

2. 2. وصلى ⸂على سّيدنا محمد⸃ و^⸆آله وصحبه وسلم³⁶

3. 1. In the name of God, the Most Gracious, the Most Merciful.

4. 2. God bless our lord Muḥammad, his family, and his companions and [give them] peace.

The opening follows standard conventions with little variation between manuscripts. Kwα replaces the phrase “our lord Muḥammad” with “His highest One”, but this does not seem to be of special significance as the manuscript does use the name of Muḥammad further below. None of the letters offer personal greetings or a formal transition to the content of the letter, which highlights its character as an impersonal missive.

2b. Šayḫ Aḥmad's vision

1. 3. قال ^⸀الشيخ احمد °^⸁خادم ^1⸀الحجرة النبوية على صاحبها افضل الصلاة ^⸆والسلام^⸇37

2. 4. ^⸆رايت النبيّ ⸂صلى عليه وسلم⸃ في المنام ⸋ليلة الجمعة⸌ وهو يقرا ^⸇ القران °العظيم³⁸

3. 5. فقال ⸂لي يا شيخ⸃ احمد المؤمنين ⸄حالهم تعبان⸅ ^1⸀من شدة معصيتهم³⁹

4. 3. Šayḫ Aḥmad, guardian of the tomb of the Prophet, upon whose inhabitant be the most splendid blessing and peace, said:

5. 4. I saw the Prophet, God's blessing and peace be upon him, in [my] sleep on Friday night and he was reading the exalted Quran.

6. 5. Then he said to me: Oh Šayḫ Aḥmad, the faithful are weary [with regard to] their state, because of the intensity of their disobedience.

As contemporary observers noted (C.H. Becker 1911: 45), the purported role of the author makes the term “Mecca letter” misleading, since the Prophet's tomb is located in Medina.⁴⁰ All versions of the East African circulation identify him as Šayḫ Aḥmad. This is not unique to this particular tradition of the missive, but was found in the earlier North African and later Indian version of the letter, and is still the case in the much more recent email version of the letter (Katz 1994; Cairo specimen; Singh 2014; Islamic Board 2005). Only the Indonesian version uses a different name (Hurgronje 1923: 133), while one of the versions reported by Katz (1994: 159) has a scribal corruption of the name to Ḥamza. The generic nature of this name suggests that the recipients of the letter were not expected to be able to verify the identity of the missive's author. Instead, the authority of the letter was supported in the purported function of its author, its supposed origin from the Ḥiǧāz, and the trustworthiness of whoever passed it along. The Cairo version adds an additional element of authentication by claiming that the letter was received directly from Šayḫ Aḥmad via telegram (FO 141/111).

The defective spelling for “tomb” (هجرة) in a group of letters also signals a considerable distance from the site itself and may have been due to Swahili influence. It is worth noting that all specimens with this spelling belong to the group of letters which will emerge as secondary further below. Further variations in line 3 pertain to added venerations of the Prophet, the second of which (“pure salutations”) reveals a commonality between Bgα and Kwα.

The appearance of the Prophet in a dream is not unusual in Islamic tradition (Schimmel 1998) and according to Becker (1911: 46) may have lent the dream more credibility, because it was considered impossible for the devil to appear in the figure of Muḥammad. The dating of the dream to a Friday, the day of congregational prayer, provides additional weight and was considered an essential detail by all but one of the East African versions. It is reported in other versions of the letter as well (Hurgronje 1923; Katz 1994), though the Cairo translation notes Thursday night (FO 141/111).

The Prophet's first line of speech delivers the main thrust of his message, which is to bring the faithful to repentance. Schimmel (1998: 292) suggested that this dream was essentially a penitential sermon, while others (and certainly the contemporary German observers) focused on the later millennial undertones of the letter in favour of a more political interpretation.⁴¹

There are two interesting variations in line 5. The first is a clear edition: Bgα makes Šayḫ Aḥmad the speaker of this sentence, who complains to the Prophet about the faithful. The second major variation is a linguistic corruption: the somewhat cumbersome phrase حالهم تعبان (their state is weary/tired) seems to have presented difficulties for copyists. Lnd, Kwβ, and Bgβ appear to derive from a scribal distortion of تعبان to لعبان, which none of the copies manage to restore. Bgα and Kwα (or their common source), in turn, feel the need to intensify the “weariness” of the faithful by adding an element of aggravation or nuisance (via idiosyncratic derivations of ضاق). In Bgα, Šayḫ Aḥmad asserts that the faithful are “his [God's or Muḥammad's] nuisance and weary because of the intensity of their disobedience”, whereas in Kwα the Prophet notes that “with regard to their weary states, the faithful are aggravating in the intensity of their disobedience”. Outside of the East African circulation, only two letters mention the condition of the faithful without further qualifying it as either good or bad,⁴² while the Cairo letter states that the “faithful are suffering for their misdeeds” (FO 141/111). The term “suffering” here may be a loose translation of تعبان, but if not, the allegation of weariness or tiredness in religious matters would be a unique accusation of the East African circulation of the letter.

2c. Celestial intercession

1. 6. °^⸀فانّي سمعت الملائكة ¹°وهم يقولون ^{⸆ ⸁}تركوا ^1⸀ذكر ^⸇ ⸂سبحانه وتعالى⸃ فاراد ربك ان ^2⸀يقبض ²°عليهم⁴³

2. 7. ⸂فقال النبى صلى عليه وسلم⸃ يا رب ارحم أمّتي ⸋فانك انت الغفور الرحيم⸌⁴⁴

3. 8. ⸋وانا ⸂عليهم بذلك⸃ يتوبون ^⸀فان لم يتوبوا الامر اليك⸌⁴⁵

4. 6. And truly I heard the angels and they said: they have neglected the invocation of God, may he be praised and exalted, so that your master already wanted to seize them.

5. 7. And the Prophet, God's blessing and peace be upon him, said: oh master, have mercy on my people, for behold you are the forgiving and merciful one.

6. 8. I will be upon them about this and they will repent, but if they do not repent, the matter is yours.

This section of the letter raises the stakes of Muḥammad's accusation, as the angels now specify the allegation and the dream develops into something of a throne room vision, with the Prophet interceding for his followers before God. Outside of the East African circulation, only the Cairo letter mentions the angelic intervention, whereas two others omit this part but mention the Prophet's entreaty.⁴⁶

There is an ambiguity as to whether the angels appear to Šayḫ Aḥmad or the Prophet, because the فانّي in line 6 can be read as a continuation of the Prophet's speech in line 5 or as the Šayḫ taking over again as speaker. Only Bgα offers clarification as it undoes the earlier reversal of speaker roles in line 5: the Prophet is the one to report the angelic vision. Together, both lines put Muḥammad in an intermediary position: Šayḫ Aḥmad complains to the Prophet, who in turn has the throne room vision and intercedes for the faithful before God. Bgα also omits line 8, making Muḥammad's intercession perpetual and his role more distant: he is not the one to admonish the faithful, but neither is his effort before God his final one. Both of these may be a theological correction or clarification by the copyists of this specimen, the Qādirīya-adherent Abū Bakr bin Ṭaha in Bagamoyo, intended to elevate the status of the Prophet.

Another important insertion supports the assumption of a Sufi correction elevating Muḥammad: in Bgα and Kwα the faithful are not only guilty of neglecting the “invocation of God”, but of “God and his prophet”. This addition aligns the term used, ḏikr (ذكر), more closely with the homonymous Sufi practice. Only Bgα and Kwα have this modification, pointing again to their common source. This observation limits the plausibility of B.G. Martin's and Michael Pesek's assertions that the letter spread mainly through Qādirī networks (Martin 1969; 1976: 173–4; Pesek 2002). If that had been the case, it would be unclear why only the two manuscripts clearly identified with the Qādirīya offer a deviant reading that is not taken up by any of the other copies.

2d. Sins and separation

1. 9. ^⸀وهم ⸂قد ^⸁ارتكبوا المعاص⸃ ^1⸀والكبائر وتركوا ^2⸀الدعاء ^3⸀واتّبعو الزنا ⸄ونقصو الكيل⸅^{⸆ 4⸀}ويشربو ^5⸀الخمر^6⸀ومشوا بالغيبة والنميمة ¹⸂^7⸀واحتقرو الفقير والمسكين ولا يعطو الفقير حقّه¹⸃ ^⸇ وتركو الصلاة ¹⸋ ^8⸀ومنعو الزكاة⸌⁴⁷

2. 10. ⸋ ^⸀فاخبرهم يا شيخ احمد °بذلك وقول لهم لا ^⸁تركو الصلاة ^1⸀واتو الزكات⸌⁴⁸

3. 11. ⸋^⸀واذا ^⸁مر عليكم تارك الصلاة⸌ ^⸆ لا ^1⸀تسلّموا ^2⸀عليه واذا ^⸇مات ¹⸆ لا ⸂تمشوا في ^3⸀جنازته⸃ ^4⸀وانتبهوا⁴⁹

4. 9. They have committed sins and great transgressions, they neglect the invocation, they practise fornication, they reduce the dry measure, they drink wine, they engage in slander and defamation, they despise the poor and the beggar and they do not give the poor justice, they neglect prayer and withhold alms.

5. 10. Inform them about this, oh Šayḫ Aḥmad, and tell them not to neglect the prayer and to offer the alms.

6. 11. And if you encounter one who neglects his prayer, do not greet him, and when he dies, do not follow his funeral procession and be careful.

This section delivers the full “catalogue of sins” that the believers are accused of and exhorts them above all to return to their core religious duties (prayer and alms), while exerting social pressure on those who do not follow suit. And this message was not lost. The most continuous feature in the German reports from towns where the letter had left its mark was that the mosques were full and people displayed a renewed religious fervour, whereas political unrest – the alleged purpose of the letter – was only feared but never observed in direct connection with the letter.

While the content of this admonition is fairly straightforward, the text-critical analysis of this portion is complex. It is the most divergent part of the letter, apart from the various copyists’ additions at the end. This is also the case in the letters outside the East African circulation, making a comparison difficult, as each letter attacked its target communities for a different list of shortcomings. Only the Cairo letter provides a list very close to that offered in the East African versions some 31 years later.

More importantly, the variations within the East African circulation form two distinct clusters, which allow important conclusions about the transmission of the letter. Ukn, Lnd, Znα, Kwβ, Kwγ, Bgβ, Mhn, Tnα are missing a large portion from the end of the first sentence to the beginning of the third, which is present in all other versions. The omission itself is clearly haplographic: a copyist skipped from الصلاة (al-ṣalāh) near the end of line 9 to the same word in line 11. This omission created a grammatical distortion, which two versions in this haplographic branch (Lnd, Mhn) correct in different ways. This basic bifurcation is also related to two further textual variations: Ukn, Lnd, Znα, Kwβ, Kwγ, Bgβ, Mhn, Tnα all accuse the faithful of neglecting religion (الدين, al-dīn) rather than the prayer or invocation (الدعاء, al-duʿā), and they also recommend to “keep away” (انتهوا, intahū) from sinners rather than to “be careful” (انتبهوا, intabihū).

This textual bifurcation makes it highly unlikely that the letter was spread in a centralized fashion as the German administration asserted. Even if one were to assume that the letter's origin was in Zanzibar, it would have been transmitted in two disjunct streams, because the haplographic divergence was already present in the two Zanzibar copies. Thus Ukn, Lnd, Kwβ, Kwγ, Bgβ, Mhn, and Tnα would be seen to depend on Znα, whereas Kwα, Bgα, Ksα, Ksβ, Tnβ would depend on Znβ. This, in turn, makes it extremely unlikely that the spread of the letter was co-ordinated by the Rumaliza family, while it is of course possible that they participated in one of the streams of distribution, namely that related to the letter displayed in the Ng'ambo mosque (Znα). This would also de-couple the Rumaliza family from the Qādirīya – a claim by Martin (1976: 174), which Nimtz (1980: 202, n. 10) has already rightly cast in doubt – because the clearly Sufi versions of Bgα and Kwα belong to the other branch. Moreover, in three major coastal towns, Kilwa, Bagamoyo, and Tanga, the letter was present in both of its traditional versions. This too points to multiple local agents behind its proliferation, because if only one group or person had spread the letter, only one version should have survived as the differences would have been merged.

A second cluster of variations relates to the accusation of tampering with the dry measure. Almost all specimens with the haplographic omission offer a very different reading here: the faithful yearn for the night. This may stem from a scribal corruption of الكيل (al-kayl) to ليلا (laylā) with the verb following suit. Mhn, Ksα, and Ksβ offer the semantically slightly odd reading of “they scrutinize the dry measure” (وتمحصوا الكيل, watamaḥaṣū al-kayl)¸ but the more interesting anomaly here is Mhn. While it normally follows the branch with the haplographic omission, in this case it sits with the other group of letters. This may indicate some cross-fertilization between copies (Morogoro is proximate to both Mahenge and Kilossa), or it may point to another group of texts that sits between both branches but is no longer extant.

Finally, Kwα and Bgα once again reveal their mutual proximity by radicalizing the separation from the religiously negligent. For them it is not enough for believers to abstain from the funeral of sinners; for them, sinners are to be denied a burial in Muslim graveyards altogether.

2e. Final warning

1. 12. ^⸀واستيقظو واجتنبوا ^⸁الفواحش ما ^1⸀ظهر منها ⸂وما بطن⸃⁵⁰

2. 13. وقل لهم ^⸀الساعة قد قربت ولم ^⸁ يبقى ^1⸀لهم الا ^2⸀قليلا وتظهر الشمس من ^3⸀مغربها⁵¹

3. 14. ^⸀فارسلت ^⸁اليهم وصيّة وبعد ^1⸀وصيّة ⸋فلم يزدادوا الا طغيانا وكفرا ونفاقا⸌⁵²

4. 15. ⸋وهذا ^⸀اخر ^⸁وصيّة⸌⁵³

5. 12. Wake up and refrain from the depravities in what is visible of them and what is hidden.

6. 13. Tell them, the hour has already come near, and there is but little [time] left for them, and then the sun will rise from the west.

7. 14. I have sent them warning after warning, but they have only increased in tyranny, blasphemy, and hypocrisy.

8. 15. This is the final warning.

These are the final words of the Prophet in Šayḫ Aḥmad's dream and they highlight the urgency of his reproof. A drastic awakening is called for, and Muslims are running out of time. The eschatological horizon is signalled clearly with reference to a ḥadiṯ of the Prophet announcing the reversal of the sun's course as a sign of the end of times. Yet contrary to the German interpretation or the local “exegeses” they reported, the letter does not link this eschatological imminence to a political revolution or the arrival of the mahdī, but to a call to repentance before the final judgement day arrives. This repentance was to be total, addressed not only to visible religious observances but also to hidden sins. The related German reports suggest that this bore fruit, as the most frequently reported behavioural changes were increased mosque attendance and the withdrawal of female concubines from their arrangements with German officials or African soldiers.

There is a small tension between a more individualistic and “objective” eschatological vision that is visible in the text variants as well. The branch with the haplographic omission asserts that not much time is left for the world (min al-dunyā), whereas the other branch notes that time is limited “for them” (lahum). All of the East African letters insist, however, that time is short. Only Tnα seems to limit the eschatological urgency somewhat by omitting the Prophet's statement that this was his last warning.

A similar tension can be seen outside the East African circulation as well. Versions A and B reported by Katz (1994) fail to provide a clear warning about the end times, but mention that this is the Prophet's last admonition. This is justified by the limits of the human lifespan: in light of certain death and God's judgement after death, repentance is of the essence. The others offer a more comprehensive eschatological outlook: mountains will burn (Singh 2014: 1025), there will only be one star at night, the sun “will lower itself with the earth”, and the writing of the Quran will disappear (Islamic Board 2005), God will withdraw a series of positive characteristics from humans (Hurgronje 1923: 137), and some copies even determined the specific year of the world's demise (Katz 1994: 168). There are some indications that the oral transmissions accompanying the letter in East Africa also contained specifications of time,⁵⁴ but the letter's stated eschatology makes no such pronouncements.

2f. End of dream and further instructions

1. 16. ⸂وقال الشيخ احمد⸃ ^⸀قد ^⸁استيقظت من ^1⸀المنام ⸄فوجدت وصيّة⸅ ^2⸀ °مكتوبا ^3⸀بجانب ^4⸀الحجرة النبوية ^{⸆ 1}⸂بخطّ اخضر¹⸃⁵⁵

2. 17. ⸉فقال النبىّ صلى عليه وسلم من قراها ولم ^⸀ينقلها ^⸆كنت خصيمه يوم القيامة⸊ ⸋ومن ⸂قراها ونقلها ¹⸋من بلد الى ^⸁بلد¹⸌ °كنت شفيعه يوم القيامة⸃⸌⁵⁶

3. 16. Šayḫ Aḥmad says: then I woke up from my sleep and found a warning written on the side of the Prophet's tomb in green writing.

4. 17. So the Prophet, God's blessing and peace be upon him, said: whoever reads it and does not forward it, I will be his adversary on the day of reckoning, and whoever reads it and forwards it from country to country, I will be his advocate on the day of reckoning.

This passage provides material confirmation for Šayḫ Aḥmad's dream and links it with the letter itself. The miraculous appearance of a written warning by the Prophet himself (green ink), erases any doubts about the dream having been an illusion, and increases its authority and urgency. Moreover, the tacit link between sentences 16 and 17 implies that the instruction to spread the letter was the very content of the Prophet's miraculous writing. Four other versions outside the East African circulation of this letter also contain a miraculous writing by the Prophet (Hurgronje 1923; Katz 1994, version A and B; Cairo specimen), and in all but one case, this writing is equally linked with the content of the letter itself and the instruction to spread it. In the East African circulation, Ukn, Tnα, Tnβ, Bgα, Ksα, and Ksβ even seem to suggest that the missive itself was found on the tomb: Šayḫ Aḥmad finds the warning on the Prophet's tomb, so presumably the warning letter itself.

Sentence 17 instructs the reader to spread the letter, reinforced by the Prophet's symmetry of threat and promise for the day of reckoning. The continuation of the transmission chain from “country to country” thus becomes a religious duty. Znα, Bgα, Kwα (and possibly Tnβ) only contain the Prophet's threat and not his promise of blessing. There is no precedent for this outside of the East African circulation. Those which have similar passages either contain both the promise and the threat (Katz 1994, all versions; Cairo specimen), or the promise only (Hurgronje 1923). One hypothesis would be that the branch without the haplographic omission did not contain the blessing and that Ksα and Ksβ added this for the sake of symmetry or through cross-fertilization from another source. It is also possible that the omission in Znβ, Kwα, Bgα, and Tnβ was caused by theological reservations against making the copying of the letter a condition for the Prophet's intercession on judgement day. But there could also be individual textual reasons. Bgα moved the blessing by the Prophet further down, and Kwα instead paired this sentence with the warning about doubts, linking both to the day of reckoning.⁵⁷ In Tnβ a partially illegible superscript may pertain to the promise blessing, while in Znα the omission may have been a mistake by the administration's copyist.

2g. Safeguarding authenticity

1. 18. ⸂^⸀فقال الشيخ احمد⸃ و °العظيم و °العظيم و العظيم ⸄قسما ثلاث⸅ ^⸀ان ¹°كنت كاذبا ^{⸆ ⸁} فاخرج من الدنيا ²°على ^⸁غير ^⸇الاسلام⁵⁸

2. 19. ⸋فمن بدّله بعد ما سمعه فانّما ^⸀اثمه على الذين يبدلونه ان ⸂سميع عليم⸃⸌⁵⁹

3. 20. ⸋فمن شكّ ¹⸋في ذلك¹⸌ ^⸀ فقد كفر⸌⁶⁰

4. 18. Then Šayḫ Aḥmad said: On God the Mighty, on God the Mighty, on God the Mighty in triple oath: if I am lying, I will pass from this earth outside of Islam.

5. 19. And whoever amends it after he has heard it, his sin is like those of them who amend it [the Quran], for God is hearing and knowing.

6. 20. And whoever doubts this, has already become an infidel.

As Šayḫ Aḥmad's missive enters its concluding phase, a thick layer of protection is added to the text with a triple oath on its veracity, a severe judgement against all who would dare to alter it, and the threat of excommunication for those who doubt. Apart from the Cairo manuscript, this triple safeguard of textual veracity is a unique feature of the East African circulation, though constituent parts are contained in all other versions. Nevertheless, there are considerable variations among the East African letters. Znβ, Kwα, Ksα, and Ksβ omit the last threat of excommunication (sentence 20), whereas Bgα replaces sentence 20 with a closing colophon and continues with sentence 21. Following a text-critical rule of thumb, one would have to suggest that elaborations are secondary, and therefore view sentences 19–20 as later additions. This would be supported by the fact that all the versions without these sentences belong to the branch without the haplographic omission, i.e. the branch with textual precedence. However, Tnβ, Kwα, and Bgα are problematic in this regard as they too belong to this branch but contain both (Tnβ) or one (Kwα, Bgα) of the sentences in the same wording as letters from the other branch. A cross-fertilization between both branches in Tanga, Kilwa, and Bagamoyo seems unlikely as this should have corrected the haplographic omission in sentences 9–11. A more likely hypothesis is that sentences 20 and 21 were added following formulas used in other, similar, missives that purportedly originated in Arabia.⁶¹

An example of such formulaic adjustments can be seen in Šayḫ Aḥmad's oath in sentence 18, which is quite divergent in the different copies in both main strands. It is likely that this formula was adjusted to whatever was the most common phraseology, as long as the main content of the oath remained intact.

2h. Letter closure and signatures

1. 21. ⸋ ⸂^⸀عليكم ^⸁بتقوى تنجو ^⸆المهالك⸃⸌⁶²

2. 22. ⸋وصلََّى ⸉ على سيّدنا محمّد ^{⸆ ⸀}وآله وصحبه ^⸁وسلّم⸌⁶³

3. 23. ⸋⸂^⸀ فانى نقلتها كما رايتها في نسخة الاولى و اعلم⸃⸌⁶⁴

4. 24. ⸋ ⸂تمّت بالخير وعونه⸃⸌⁶⁵

5. 21. In the fear of God, you will save yourselves from the place of your perdition.

6. 22. God bless our lord Muḥammad, his family, and his companions and [give them] peace.

7. 23. And I have passed it on as I have seen it on the primary copy, and God knows.

8. 24. Closing in goodness and His help.

In these four closing sentences, variance between the letters increases significantly. The version chosen for the main text here is no longer representative of any majority of letters, but comes from the Lnd letter which formed the basis for Becker's (1911) text and translation. The critical apparatus thus shows the extent to which Becker's version is not representative here.

Sentences 21 and 22 consist of a final admonition and blessing of the Prophet which is somewhat symmetrical to the opening sentence. Two letters introduce slight variations by including the Prophet's mother (Ukn, Tnβ) and father (Ukn). Four letters omit these closing lines altogether, which makes for a more urgent ending of the letter with Šayḫ Aḥmad's blessing and curse. Bgα instead echoes Šayḫ Aḥmad's curse with a pious request:

May God protect us from this and beatify us, may he approach you and the sins of the Muslims from His love and His mercy, and may God grant us his advocacy.

Sentences 23 and 24 are clearly a copyist's addition, verifying the authenticity of the copy and sending an additional greeting. Most letters leave this out altogether, while three offer very different sentences with the same purpose:

Kwα: God [is my witness] to the congruence between the letters. I have relayed it as I have seen it, free of deceit. This is the note of his apparition. Hasten quickly to pay heed to it, according to the warning of the messenger of God (SAW). The Šayḫ, servant of the path in the tradition of the Qadirīyā, Šayḥ Ḫamīs ibn Aḥmad Mafūm ibn Yusuf, Širazi [?] [?] dated 2 jumādā al-ūlā 1326.⁶⁶

Tnα: [?] By God, if one leaves this world with the saying: there is no God but God and Muḥammad is the messenger of God, the blessing of God and peace will be upon him. [? ?] Thursday, 25 Jumādā al-ūlā in the year 1326.

Bgα: Amen, Amen, Amen. 12 jumādā al-āḫirah 1326.

The commonality in the copyist's signature between Znα, Lnd, Bgβ, and Kwβ (and its scribal copy Kwγ) is interesting and points to a common source for these letters.⁶⁷ Textual proximity between these four copies is high throughout the letter.⁶⁸ They thus appear to form a sub-branch of the main branch with the haplographic omission in sentences 9–11. All four specimens had been associated with relatives of Rumaliza in the German investigations, making this copyist signature an idiosyncratic feature of the Rumaliza's version of the letter. Since such a signature would be difficult to drop in subsequent copies, it provides a solid textual basis for delimiting the Rumaliza family's involvement in the spread of the letter to four of the extant fourteen copies.

3. Conclusion

The German “Mecca letter affair” has produced a unique archive for analysing the spread of this particular religious missive. Unlike other studies of the same or very similar letters, which were based on isolated specimens, the extant fourteen copies of the East African circulation of 1908 form a chronologically and geographically compact cache of manuscripts. The text-critical comparison has revealed a basic congruence between almost all of the specimens, as well as a number of significant differences, which allows us to draw three main conclusions about the letter's character and spread.

First, the so-called “Mecca letter” of 1908 was a global chain letter in local circulation, predating the German East African “affair” and continuing until the present. This is supported by the similarities of the East African specimens with versions of the letter found elsewhere, ranging from very proximate texts (especially the Cairo specimen of 1877) to more distant varieties. The form and message of the missive bears the marks of a chain letter as well: it is sufficiently general to be copied in various contexts, it transports urgency, blesses those who spread the letter and curses those who fail to do so, and it safeguards its authenticity through various vows and signatures. Furthermore, the textual deviations between the specimens are also compatible with the nature of a chain letter's spread: the text bifurcates with the introduction of smaller and larger deviations, which are retained and further transformed in later copies. Renditions from memory, such as in Bgγ, or the existence of Swahili translations, also point to the fraying and snowballing of the missive in typical chain letter fashion, as do theologically motivated alterations such as in Bgα.

Second, in light of what can be established about the textual variations and the mode of the letter's spread, the German suspicion of a co-ordinated campaign must be rejected, along with later explanations attributing the dispersion of the “Mecca letter” to a primary agent. The involvement of Rumaliza, the Germans’ foremost suspect, was confirmed, but also clearly delimited by the textual analysis to only four of the 14 extant copies. There were two versions with clear Qādirīya affiliations, but these differ noticeably from all the others, for which no Qādirīya involvement can be established from the available sources. The letter spread in (at least) two main versions, and these were never reconciled in the German East African circulation, nor is there clear evidence of substantial cross-fertilization. Moreover, the fact that both versions were found to co-exist in Bagamoyo, Kilwa, and Tanga, also points to fairly disjunct networks of proliferation, even at the local level. This is also supported by the relatively sparse notes in the archive about how the letters were obtained, pointing to very different actors: merchants, walimu, šuyūḫ, a qādi, an imām, or less specifically, a “notable Arab” (šarīf). Specimens of unidentified origin, Swahili translations and highly deviant copies on low-grade paper such as Bgγ further substantiate the multiplicity of social actors involved in the letter's dispersion.

Finally, the distribution of the letter cannot be modelled as a unidirectional process of diffusion from centre to periphery. There are good reasons for the German assumption that Zanzibar was an important point of origin, since in each of the main textual variants two letters had arrived directly from there (Kwα, Tnβ vz. Lnd, Kwβ) – one of each as early as June 1908 (Kwα, Lnd). Yet at the same time, an interior German East African circulation of the letter was attested by Mhn, Ksα and Ksβ (and even a Swahili specimen from early June), whereas the origin of five letters was not determined. So even if one wanted to maintain a general direction of travel from Zanzibar to the coast, one would still have to reckon with a highly versatile and impactful mainland circulation. The text-critical analysis showed several deviations, version clusters, possible cross-fertilizations and formulaic adjustments that do not fit into a neat stemmatic diagram but point to the likely presence of further specimens and versions. This is supported by the historical archive as well. Not all district officers who noted the presence of the letters submitted specimens, and those who did send specimens concluded their investigations once two or three copies were obtained. The arrests of those who had been found to have spread the letters would also have served as a deterrent from bringing more copies to the administration's attention.

These observations all point to the same conclusion: the “Mecca letter affair” of 1908 was an overdetermined historical event, that is, an incident that needs to be interpreted in light of multiple causalities, forces, and factors. Due to its content, form, and mode of dispersion, the letter enabled a broad variety of actors and intentions to attach themselves to its proliferation. Like its global predecessors and successors, the East African circulation travelled along various interpersonal and institutional networks, fragmenting and snowballing in the process. Interpretations and reactions evidently varied, from chiliastic fervour to sceptical indifference, from expectations of anti-colonial unrest to the demarcation of doctrinal difference. This conclusion does not exclude the possibility of the political and religious dynamics that contemporaneous actors and subsequent scholarship attributed to the event, but it does contest their historiographical hegemony over its interpretation. Rather than determining a primary agency or interest in spreading the letter, the salient question for postcolonial scholarship of Islam should be: under which circumstances and scholarly paradigms has the interpretation of the “Mecca letter” as a single political event become its primary frame of reference?