Kilwa dynastic historiography: a critical study

33
Kilwa dynastic historiography: a critical study http://www.aluka.org/action/showMetadata?doi=10.5555/AL.CH.DOCUMENT.sip200052 Use of the Aluka digital library is subject to Aluka’s Terms and Conditions, available at http://www.aluka.org/page/about/termsConditions.jsp. By using Aluka, you agree that you have read and will abide by the Terms and Conditions. Among other things, the Terms and Conditions provide that the content in the Aluka digital library is only for personal, non-commercial use by authorized users of Aluka in connection with research, scholarship, and education. The content in the Aluka digital library is subject to copyright, with the exception of certain governmental works and very old materials that may be in the public domain under applicable law. Permission must be sought from Aluka and/or the applicable copyright holder in connection with any duplication or distribution of these materials where required by applicable law. Aluka is a not-for-profit initiative dedicated to creating and preserving a digital archive of materials about and from the developing world. For more information about Aluka, please see http://www.aluka.org

Transcript of Kilwa dynastic historiography: a critical study

Kilwa dynastic historiography: a critical study

http://www.aluka.org/action/showMetadata?doi=10.5555/AL.CH.DOCUMENT.sip200052

Use of the Aluka digital library is subject to Aluka’s Terms and Conditions, available athttp://www.aluka.org/page/about/termsConditions.jsp. By using Aluka, you agree that you have read andwill abide by the Terms and Conditions. Among other things, the Terms and Conditions provide that thecontent in the Aluka digital library is only for personal, non-commercial use by authorized users of Aluka inconnection with research, scholarship, and education.

The content in the Aluka digital library is subject to copyright, with the exception of certain governmentalworks and very old materials that may be in the public domain under applicable law. Permission must besought from Aluka and/or the applicable copyright holder in connection with any duplication or distributionof these materials where required by applicable law.

Aluka is a not-for-profit initiative dedicated to creating and preserving a digital archive of materials aboutand from the developing world. For more information about Aluka, please see http://www.aluka.org

Kilwa dynastic historiography: a critical study

Author/Creator Saad, Elias

Date 1979

Resource type Articles

Language English

Subject

Coverage (spatial) Northern Swahili Coast, Tanzania, United Republic of, KilwaKisiwani

Source Smithsonian Institution Libraries, DT19 .H673

Relation History in Africa, Vol. 6 (1979): 177-207.

Rights By kind permission of Elias N. Saad and History in Africa(published by the African Studies Association, Rutgers).

Format extent(length/size)

33 pages

http://www.aluka.org/action/showMetadata?doi=10.5555/AL.CH.DOCUMENT.sip200052

http://www.aluka.org

KILWA DYNASTIC HISTORIOGRAPHY:

KILWA DYNASTIC HISTORIOGRAPHY:A CRITICAL STUDYElias SaadNorthwestern UniversityOne of the unresolved problems in African historiography concernsthe Arabicand Portuguese versions of the so-called Kilwa Chronicle. Scholarswho haveused these sixteenth-century sources have tended to assume that the Portugueseversion, which is essentially a list of the kings of Kilwa up to around 1500, is atranscription of the Arabic version known under the title of Kitab al-Suiwa.' In therecent debate between Freeman-Grenville and Chittick, this assumption hascreated serious difficulties because the Portuguese account mentions kings whoare omitted in the Kitab. Freeman-Grenville attempted to resolve the difficulty byhypothesizing that the work was defectively abridged in the extant nineteenth-century copy. Relying on the regnal durations in the Portuguese account, hecomputed the dynastic chronology of Kilwa backwards to the tenth century.Subsequently, Chittick's excavations did not show Kilwa important enough tohave been the site of a kingdom prior to the thirteenth century. This became thebasis for an alternative explanation which denied the existence of gaps oromissions in the Kitab. Chittick argued instead that the longer list of kings in thePortuguese account may have resulted from dovetailing two sources together andduplicating their information. 3The present paper calls on genealogical evidence overlooked by both scholarswhich demonstrates that the divergence between the two sources results fromtheir varying perspectives on the dynastic politics and succession disputes. First,the Portuguese account, though occurring in Joao de Barros' Da Asiawrittenabout 1552, may represent an impromptu composition given to the Portugueseduring their occupation of Kilwa in 1505-12. The narrative reveals a consistentbias against the branch of the royal family which was in power at the time. Kitabal-SuZwa, on the other hand, bears evidence that its unnamed author, born in1499 and writing probably around 1550, did not go beyond a rough draft of sevenout of the ten chapters promised in hisHISTORY lIT AFRICA, 6(1979).

ELIAS SAADintroduction. Moreover, some of his omissions seem to relat to a controversialearly chronicle dating to the mid-1400s w apparently reached the author in adamaged (possibly censore state.Chittick did not undertake a close study of the Kitab, though he preferred itsevidence. It is not clear to what extent he relied on Strong's publication of theArabic text distinct from Freeman-Grenville's translation. Indeed, Freeman-Grenville does not seem to have had full recourse to! the original manuscript,

since there exist certain indicativ1 comments recorded on the marginswhich werementioned only b passing by Strong and were subsequently ignored altogether byFreeman-Grenville, whose translation, published separatel lacks adequatereference to his main study, and does not ever attempt to justify the theory ofdefective abridgement. On the contrary, it lacks annotation and any references tothe problems of interpreting the often-defective sentence struct As aresult, itgives the impression (followed by Chittick) that the narrative is a relativelycoherent and straightforw one. The gaps, as we shall see, are tobe distinguishedfrc other more "deliberate" omissions. Both, however, seem to i bear oncontroversial events in the history of Kilwa and cand be attributed to the errors ofa later abbreviator.Our approach will be to reconstruct the dynastic history Kilwa as muchaspossible before dealing with the peculiari' of each source. We willbegin with theShirazi dynasty, wh gave Kilwa its first period of importance during the thirtecentury. Secondly, we will focus on the Mahdali dynasty du the fourteenth andfifteenth centuries up to the Portuguese conquest. It will be seen that this extendedperiod, with i complex genealogical details, reveals succession disputes arecrucial for understanding the background of both sources Throughout we will tryto adhere to a chronological outline but because of the close linksbetweendynastic olitics an' historiography we often have to anticipate our discussion thesources.Our contribution to this subject rests mainly on a close study of Kitab al-SuZwa --primarily its genealogical evide This evidence is quite firm because itmay havebeen drawn directly from the pedigrees by which the kings are identifV in Arabic.The nomenclature involved, common in Islam, rev' the ancestry of each king andenables us to distinguish bet the numerous Muhammads, Husains,and Sulaimanswho succeeded each other. Moreover, it compensates the Kitab for its gap: thatcertain Mahdali kings whose reigns are affected by the are nonetheless mentionedas sultans in the pedigrees of the descendants. For example, [2]Sulaiman b.Hasan, the second A Mahdali king (known from coins bearing his name), is statedthe Portuguese account to have enjoyed a reign of fourteen

KILWA DYNASTIC HISTORIOGRAPHY 179years which ended when he was murdered on leaving the main mosque. The Kitabomits this reign, perhaps to suppress memory of this regicide, but still mentionsthe king as Sulaiman al,atcun (Sulaiman the Stabbed One) in the pedigree of hisson [4]Abu'l-Mawahib Hasan.5 In fact, all Mahdali reigns recognizable in thePortuguese account are identifiable either directly or indirectlyin the Kitab. Thismakes it possible to reconstruct the genealogy of the Mahdali dynasty with atolerable degree of confidence, despite the extraordinary proliferation of royallineages. Unfortunately, the same is not true of the shirazi dynastyduring andpreceding the thirteenth century.6 The Shirazi dynasty can be treated onlytentatively,especially since both our sources date to over two centuries after its demise. Theoutlines of this period are obscured by two factors which, in the mindsof Kilwatraditionists, may have been interrelated. The first is that the durations of Shirazi

reigns appear exaggerated when compared with the genealogical evidence. This initself casts doubt on the tenth-century date computed by Freeman-Grenville.Secondly, the Kitab omits the last nine Shirazi kings altogether and includes nodirect genealogical evidence on them. Chittick rejected the historicityof thesekings and this, indeed, is what gave rise to his theory of dovetailing. Our ownfindings call for a revision of this viewpoint on two main grounds. First, theearliest Shirazi kings (mentioned in both sources) would seem to havebeen basedon the nearby island of Mafia rather than at Kilwa itself. Secondly, thenine kingswho succeeded them may more adequately be assigned to the first period ofarchitectural importance (roughly spanning the thirteenth century)revealed by theexcavations.The Shirazi DynastyThe origins of the Shirazi dynasty are shrouded by legends which are open toseveral interpretations. For example, the detail in the Kitab that the Shirazismigrated in the third century of the Hijra might indicate their first appearance onthe east African coast but not necessarily at Kilwa. Likewise, the statement in thePortuguese account that Mogadishu was founded by settlers fromeastern Arabiaseventy years before the arrival of the Shirazis might attest'tothe antiquity ofMogadishu without relating the foundation date of Kilwa. Chittick placed far toomuch weight on a detail in Yaqut (d. 1222) that Mogadishu at the time was moreimportant than Kilwa. Yaqut's geographical dictionary, by far the mostcomprehensive Arab geography, mentions Brava, Malindi, Mombasa, Languya orLanguja (at Zanzibar), Tumbatu, Mafia, and, southernmost of all, Sofala.7 Thesecould not all have been founded shortly before Yaqut and it seems unnecessary tosingle Out Kilwa or Mogadishu for that conclusion. The Portuguese

ELIAS SAADaccount mentions that Shici settlers preceded both the foui of Mogadishu and theShirazis. 8 While these are identifiec the coast as Baduys, a Kilwa tradition ofrecent record associates the Shirazis with a certain "Musa Bin Amrani Elbedui."This may recall the Abu cAmran Musa whose name occurs on an inscription atthe Kizimkazi mosque in Zanziba dated 500 A.H. (c. 1107 A.D.).9 Although wecannot place excessive weight on this evidence, a close relationship bet the earlyShirazis and Zanzibar is certainly indicated for, one of two occasions when. Kilwa(and Mafia, it seems) was conquered by the Mutamandalin of Shagh, the reigningShira fled for twelve years to Zanzibar before regaining his thro Whatappearscertain is that direct contact between both sides of the Persian Gulf and thekingdom of Qanbalu (on th island of Pemba, north of Zanzibar) dates at least tothe t century. The evidence directly implicates Oman, in Easter Arabia, as well asSiraf, which served as a port for Shira Persia. This evidence comes from al-Mascudi (who apparentl made the direct journey by sea) and is strengthened byref much earlier to Qanbalu and Languja, in a work by al-Jahiz Altogether, if wetake account of the recorded pre-Islamic, contacts as well, it seems reasonable toassume that indivi and families bearing Shirazi, Mahdali, and other Middle Eanisbas began making their appearance on the east African co as merchants fromthe first centuries of Islam. It is thi process that gradually gave rise to the proto-

Swahili Musli culture which is evidenced by records and material remains 4Kilwa and elsewhere. Excavations conducted on a small sca-' Mafia, publishedby Chittick, have revealed much the same as those recovered at Kilwa. On theisland of Manda to th north of Zanzibar and Pemba, Chittick records ninth- and tecentury imported ceramics as well as massive walls built the encroachments of thethen-rising sea-table.12 This phenomenon, damaging to the remains of port-towns, has be, observed at several sites and is reflected in a number of submersionlegends. In any case, our inability to correlat documentary evidence with oraltraditions and archeological surveys suggests considerable fluctuation in earlysettleme sites. 13At Kilwa itself the excavations have yielded some clues f correlatingthe earliestarchitectural remains with the documentary evidence but, regrettably, these havenot been properly used. The very earliest coins, judging from their stratigraphy,bear the names of Daud b. Hasan and cAli b. Ha Chittick chose to identify thelatter coins as the earliest associated them with the Shirazi "founder" though theKital (whose evidence he prefers) describes the man as cAli b. Hu1 He relied inthis case on the Portuguese account, notwith-, standing its errors, because it lendsitself to the supposij that the founder was named cAli b. Hasan.14 For our part w

KILWA DYNASTIC HISTORIOGRAPHY 181prefer to look for these two kings, presumably brothers, among the nine omittedin the Kitab. The reason for this is quite simple; the largest numbersof recoveredcoins belong to the earliest Mahdali kings, and it is only reasonable toassume thatthe striking of coins in Kilwa began under their immediate predecessors ratherthan under the semi-legendary founder. 1The Portuguese account is especially confused on the Shirazi period because ittries to identify the kings by their pedigrees and produced a mixed result.[B]CAli b. Bashati becomes Ale Busoloquete while [E] Khalid b. Bakr becomesAle Bonebaquer. The phonetics suggest rather strongly that the account, thoughpurporting to be based on "a chronicle of the kings of Kilwa," represents a ratherhasty translation made from an oral recitation. The above two cases are notproblematical because the Kitab comes to our rescue. However, among theomitted kings we find one Ale Boni, presumably a certain cAli, whose father'sname was inadvertantly dropped. This could conceivably be the [P]cAli b. Hasanof the coins, especially since one of his immediate predecessors, a brother, beingnamed Daud, could answer the problem of the coins of [M]Daud b. Hasan. All weneed to confirm this is to find a reference to their father as Hasan. And, indeed,Daud's immediate predecessor, possibly a brother, is named Soleimao Hacen(presumably [L]Sulaiman b. Hasan). Furthermore, Ale Boni was succeeded by a"nephew" identified as Boni Soleimao. The parentage of the nephew, though hehimself is unnamed, suggests that Sulaiman, Daud, and cAli were brothers, sonsof a certain Hasan.16Before justifying our reconstruction any further, let uslook at the first sequence of Shirazi kings mentioned in common by both sources.These kings are assigned regnal durations totaling 172 years in Freeman-Grenville's chronology. But, judging from their nomenclature in the Kitab , they

belonged to three generations only. On basis of thirty to thirty-fiveyears betweengenerations, they need not have flourished for longer than one hundred years. Thetext leaves a firm impression that they were based, for the most part, at Mafiarather than Kilwa.- It requires some scrutiny to recognize this, butwith referenceto genealogical detail it can readily be ascertained. The kings in question includethe founder along with his sons and grandsons. The first is alleged to havemigrated directly to Kilwa from Shiraz but it transpires that he was buried atMafia. Two of his five known sons did not reign at all, while two reigned only atMafia. The fifth son, [C]Daud b. cAli, allegedly reigned two years inKilwa thenwent to visit his father's tomb at Mafia and decided to reside there because thatisland "pleased him." We are given to understand that his sons and nephews,representing the third generation, remained behind. However, we learn that they"lost" Kilwa for fifteen years to the Mutamandalin during the span of a quarter

±oU/ ELIAS SAADof a century, a period which might account for their entire floruit as kings.17It appears that Kilwa traditionists tended to "dovetail" histories of Kilwa andMafia. This might reflect a desire their part to rescue their kingdomfrom a ratherinsignific beginning when it was contested by Shirazis and Mutamandali Thelegend that the Shirazis were sons of kings even before left Persia comes intobetter focus at this point. The tra tionists would not have dwelt on the legendunless the Shir had been based at Kilwa itself for an extended period and h made astrong impact on its history. This impact is not apparent from the Kitab; hence, wehave to look for it in tcareers of the nine kings mentioned only in the Portugueseaccount.A clue to the relationship between the two sets of kings comes from the identityof the very last Shirazi. He is th( successor of Boni Soleimao, described as AleDaute. This a' certainly stands for [R]cAli b. Daud and resembles the nomeiclature of an earlier king named [D]CAli b. Daud b. CAli, of the grandsons of thefounder. Although it omits the Ilasi nine Shirazi kings, the Kitab leaves us anallusion to thelast in a peculiar comment on his earlier namesake, describi him as"the last from the seed of the [Shirazi] founder" ev( though it goes on to discuss acousin and a brother who suc ceeded him. The cousin is [F/H]Hasan b.Sulaimanb. CAli,, fled the Mutamandalin to Zanzibar, while the brother is [I]Hasan b.Daud b. cAli. Either one of the two could be t father of Soleimao Hacen([L]Sulaiman b. Hasan) and his brothers [M] Daud and [p]cAli whom we haveencountered abov But, if our analysis bears any resemblance to the facts, itappears probable that [I] Hasan b. Daud was their father. is given aratherlegendary image which helps to overshado his omitted nine successors, beingalleged to have succeeded the age of seventy and reigned for seventy yearsmore!18Remarkably enough, the Portuguese account, though it rev the namesof the ninekings, appears influenced by the same "interpretation," albeit alongdifferent lines.It, too, s to allude to the last Shirazi, [RI]Ale Daute or cAli b. Daud where itmentions his earlier namesake. As Ale Ben Daute, earlier king is transformed intotwo, one of the grandson other, who allegedly succeeded each other. The first reig

sixty years while the second reigned six years after whichh was killed,purportedly by popular agreement. This, so farl we can tell, mayreflect a self-conscious attempt at removiN the last Shirazi king from his real time context -- aperio of violence in which the Mahdalis seized the throne after struggle. sThe omitted nine kings appear to qualify as the architece of Kilwa's importancebeginning somewhere around 1200. The

KILWA DYNASTIC HISTORIOGRAPHY 183first two, including a certain [J]Sulaiman and his son [K]Daud, cannot be placedfully in their genealogical context because the former is described only as "ofroyal blood" with no indication of his parentage.2 Judging by the association ofDaud with the gold-trading port of Sofala to the south, Kilwa was at the thresholdof its first period of prosperity. The next king, [L]Soleimao Hacen,whom weidentify as son of []Hasan b. Daud of questionable longevity, is specificallyassociated with large-scale architectural works, allegedly including the firstbuildings in stone rather than wood. This places him at the beginningof Chittick'speriod III which is characterized by the introduction of coins and substantial stonebuildings. He was succeeded by a series of four brothers. The first we identifywith the few coins bearing the name [M]Daud b. Hasan, which appear to be theearliest minted in Kilwa. The third and fourth brothers, like Daud himself,probably reigned very briefly, since the fifth brother Ale Boniis described as "themost fortunate of his race because everything that he commenced he was able tocomplete," even though he himself ruled only ten years. This almost certainlyrefers to the architectural works begun by the eldest. Ale Boni, whom we presumeto be [P]CAli b. Hasan, is represented by a great number of coins. He wassucceeded by a son of [L]Soleimao Hacen (the unnamed [Q]Boni Soleimao) andthe latter by a son of Daud ([R]Ale Daute). It appears reasonable that the fivebrothers should be succeeded by sons of the two eldest.2'We have to admit our inability to reconstruct the Shirazi period beyond the abovesuggestions. If the omitted kings actually belonged to two generations of thedynasty previously established at Mafia, then we may ignore the 148 yearscomputed for them on basis of the regnal durations by FreemanGrenville. Instead,we may assign them a total floruit of perhaps 70 years terminating around 1275, adate suggested further by the durations assigned to the first few Mahdali reigns.These, in turn, are confirmed by the passage of three generations from theMahdali founder to his eldest grandson, [4]Abu'l-Mawahib Hasan, who wasvisited shortly before his death in 1332 by Ibn Batuta.22The Mahdaii DynastyWith the Hahdali dynasty we enter a period in which theexcavations have more or less substantiated the documents. The first four Mahdalikings, including [l]Hasan b. Talut, [2] Sulaiman b. Hasan, [4]Abu'l-MawahibHasan b. Sulaiman, and [3/5]Daud b. Hasan, are all known from coins bearingtheir names, The numismatic industry subsequently declined, but [l]Sulaiman b.Husain (a nephew of Abul'l-Mawahib) is represented by four coins, while his son[13]Muhammad al-cAdil is known from a larger number.23 Finally, there existtwo problematical coins which are probably to be dated to the mid-fifteenth

ELIAS SAADdentury. These include a large nunber bearing the name Na al-Din wa'l-DunyaMuhammad and a rare crude type bearing t name Muhammad cAbd al-Wadud.Conceivably, they belong respectively to a usurper noted for his wealth named[16] Muhammad b. Sulaiman, and to a certain [17]Muhammad al-Maz who.succeeded him and is stigmatized in one of our sources his "poverty." Both kingsreigned only briefly in the mid 1400s but they occupied important positions in' thesuccess disputes. So far as we can tell, the sixteenth century, a period of decline,produced no coins, unless those-of Muha cAbd al-Wadud could be assigned tothat period.24The problematical aspects of the Mahdali period all bea directly upon a complexpolitical and succession system wh gave rise to an extraordinary broadening of theroyal line The central principle was that no man could aspire to the honor unlesshis father had held it before him. "Sons of kings" (awlad al-muluk) appears as thekey phrase in the K which explains the repetitious recourse to the title "alsultan"in the nomenclatures. In the first two generation; this meant that succession wentto the eldest son. Subseqi however, the rights 6f surviving brothersbecameparamount they in turn gave way to cousins who vied for the throne ai them. Bythe sixth generation a' sdries of distant nephews cousins, largely on basis ofseniority by age, rapidly fol' each other. All had in common the fact of being sons6f k but some were believed more royal than others. Descent on mother's side(royal or otherwise) entered into the picturi while even the durations ofpreviousreigns (contested dowi the month or day) apparently affected therights of variou!contenders.25 The disputes naturally weakened the Mahdalii probably happenedto the Shirazis before them) and offered opportunity for two "usurpers" in thefifteenth century an( two more during the Portuguese occupation.Closely related to the above circumstances was a polariin the powerstructure ofKilwa between the throne (al-muli on the one hand, and another powerful postidentified as ti emirate (al-conr or al-imara). Indeed the whole system wasicharacterized by diffusion of influence. A patrician tend( in the kingdoms of thecoast is perhaps reflected in a stal by De Barros that the polity of Brava remainedunder the rt of its elders.26 In Kilwa the notables as a whole, both ro and non-royals, emerge as "the people of the major decisiot (ahi al-hall warl-Caqd) inconnection with one of the succ( disputes. The author of Kitab al-Suiwa indicatedthat his, maternal uncles formed part of a small delegation which ne tiated withDa Gama on the open waters, though as faqihs (T, jurists) they had no officialstatus. The viziers (holderE fiscal post, it seems) had a say in the dynastic politics, in at least one case the qadi (judge) had influence in a d)184

KILWA DYNASTIC HISTORIOGRAPHY27 The prerogatives of the amirs are nowhere indicated; however,the militaryconnotation of al-amr suggests that they commanded the kingdom's armed forces,while reference to them in the Portuguese account as "governors"suggests thatthey also had some administrative functions.

The emirate may have originated at the earliest periods when Kilwa was contestedfrom a distance by Shirazis and Mutamandalin, for one of the rulers appointed bythe latter is described as n amir who lacked the full symbols of sovereignty, thekhutba (the Friday sermon) being recited in the name of the monarch thenreigning at Shagh. More importantly, the emirate appears to have been the basefor the rise of the Mahdali dynasty. The relevant passage in the Kitab shouldperhaps be quoted here because the distinction between throne andemirate eludedFreeman-Grenville in his translation. After is summary treatment ofthe lastShirazis, the Kitab says:Then the throne (aZ-mu1k) passed from them to the Houseof Abu'l-Mawahib. The one among them who was assigned theemirate (wulliya al-amr) was al-Hasan b. TAlut, the one famous forinsight andcourage. He took the throne (almu1k) by gaining the upper hand despite thepresence ofthose to whom it [rightfully] belonged (maca wujud ahlihi).But they had become weak vis-a-vis the command of the army (al-amr), so hemonopolized power by himself andtook the kingdom (al-mamlaka) forcibly and by compulsion.28This passage introduces us to a most problematical aspect in the historiographicaltradition, for in denoting the new reigning family by the name of Abu'l-Mawahib,the founder's grandson, the text suggests that the dynasty did notacquire alegitimacy of its own until his reign. This curiosity may have resulted from thefact that the Shirazis actually continued to reign in Mafia so that the Mahdalifounder did not fully supersede them, while his son [2]Sulaiman, whose omissionis compensated for by the allusion to him as al-Matcun, apparently fell victim totheir machinations. In a digression the Kitab indicates that he was assassinated atthe instance of the people of Mafia, probably a reference to the Shirazis. It wasnot until Abu'lMawahib's reign that Mafia was conquered for the new family and,in a curious comment to be mentioned again, he is described as "the first to reignindependently over Mafia."29 This apparently marked the beginning of thedynasty as a fully sovereign and acknowledged one. The emirate atKilwa itself(and presumably a similar post at-Mafia) apparently survived in the hands ofmembers of the Mahdali family. For, although it disappears altogether from theaccounts of the first hundred years or so of the Mahdali dynasty,it re-emergesagain in the fifteenth century in the hands of a non-royal notable. Itsreversionsubsequently to descendants of the Mahdali line is featured in

ELIAS SAADone of the chapter headings as the "Restoration of al-Amr to House of Abu'l-Mawahib."The reappearance of the emirate occurs in a difficult pas which was alsoinadequately translated by Freeman-Grenville. Referring to the king reigning atthe time, the Kitab reads:He is [13]al-Sultan al-Malik al-cAdil Muhammad b.Sulaiman b. al-Husain. Then [25]Sabhati. When he tookcharge of the emirate (tawalla al-imara), the world

(al-dunya) came to him and the great men of the statefollowed him (inqadat lahu). Al-Sultan Muhammad al-cAdil remainedon thethrone (fi'l-mulk) twenty-two years thenpassed to God's mercy. Al-Wazir Sulaiman and al-AmirMuhammad b. Sulaiman remained together [in the vizierateand the emirate respectively] after him.30At the risk of anticipating too much our analysis of the Kitab, we have to indicatethe probable existence of a gap a affecting a few to several lines which, amongother things, should have provided more details on the king. For the man whoseimportance is emphasized in the second sentence (after the abrupt words "thenSabhati") must be the non-royal al-A Muhammad who we know occupied theemirate and outlived both, Muhammad al-cAdil and his son Sulaiman. Indeed,there exist possibility that the thitd sentence should have read: "[141 Al-SultanSulaiman b. al-Sultan Muhammad al-cAdil remained i the throne twenty-twoyears . . ." since the Portuguese acce assigns Muhammad a reign ofonly nineyears and reserves th twenty-two for Sulaiman. As it stands, the Kitab omits theduration of one of the two reigns.31Judging from other indications to be discussed in due co the gap in the Kitabmight reflect a deliberate attempt on part of a censor to exonerate Muhammad andSulaiman from th emergence during their reigns of a non-royal amirwho for timedominated the throne. The man's wealth is emphasized i where in the Kitab and isindicated here by the word al-dui which might qualify him further as Nasir al-Dinwa'l-Dunya Muhammad of the coins. 32 When [14]Sulaiman b. Muhammad al,died the amir-would not allow any of his brothers and sons succeed him, but thenotables insisted on a member of the family and brought forward to the throne[15]Ismacil b. Hust b. Sulaiman. However, this king too was dominated by theand, on his death, [16]Muhammad stepped forth and replaced formally. Hereigned only one year (apparently at a very advanced age) but his usurpation wasan important episode because it was the first since the establishment of the Mahdynasty.Our brief outline of Muhammad's career is crucial for u standing one of the mostsignificant genealogical clues in Kitab for, after recording the deathof the amir,the text

KILWA DYNASTIC HISTORIOGRAPHYtreats his successors under the title "The Restoration of the Emirate tothe Houseof Abu'l- Mawahib.'33 This emphasizes a fact not apparent elsewhere (and whichis in fact implicitly denied in the Portuguese account), namely, that[4]Abu'lMawahib left a whole sequence of direct royal descendants in yilwa.Other Mahdali kings were descended from his brother [7]Husain and thesuccession disputes, where gave undue influence to non-royal notables, werebetween the respective descendants of the two brothers. The descendants ofAbu'lMawahib gained the upper hand in the decades preceding the arrival of thePortuguese by virtue of their monopoly on the emirate.For the most part our reconstruction of the genealogical table at the end involves asimple process. The last king mentioned in Kitab al-Sulwa is identified as [36]al-

cAdil Muhammad b. Husain b. Sulaiman b. [13]Muhammad al-cAdil. The latter,in turn is already known to 'us from the quotation above as [13]al-cAdilMuhammad b. Sulaiman b. Husain. The grandfather in this nomenclature, namely[7]Husain, is undoubtedly [4]Abu'l-Mawahib's brother. Muhammad al-cAdil'sknown descendants, including seven kings, are consistently identified by theirpedigrees, leaving little room for error.A Meccan source recently mentioned by Martin helps us reconstructanother sub-branch descended from Husain. The source records the pilgrimage in 1410/11 of aKilwa identified as [12]Hasan b. Sulaiman b. Husain.35 This nomenclaturereveals the king as a brother of [13]Muhammad al-cAdil, to whom he probablygave way shortly before or after his pilgrimage. The Kitab mentionsthepilgrimage and the pilgrim in a digression concerning his son [20]Sacid b. Hasanb. Sulaiman. Sacid reigned at a much later time, in old age it seems, and after abrief interruption, was succeeded by two equally aged brothers. Named[22]cAbdullah and [23]cAli, the two are described as sons of al-Khatib Hasan.The most problematical lineage is that descended from [4]Abu'l-Mawahib. In reconstructing this lineage, we have reasonedthat therestoration of the emirate to this house can designate only [21]al-Amir Sulaiman,his brother [281al-Amir Muhammad Kiwab and Sulaiman's son [30]al-AmirIbrahim. These three assumed the emirate in succession after thedeath of thenonroyal [16]al-Amir M uhammed and, even after the Portuguese conquest, itseems the emirate never left their lineage, for there exists an inscription recordingthe death in 1554 of a certain cAisha bint al-Amir cAli b. al-Sultan Sulaiman. Thenomenclature shows the lady to be a niece of Ibrahim and her father,al-AmircAli, appears to be his brother. 36The royal descent of the amirs, though obscured in thePortuguese account, is not in doubt. [30]Ibrahim, for example, is identified as al-Amir Ibrahim b. al-Sultan Sulaiman b.

ELIAS SAADal-Sultan Muhammad al-Mazlum. The grandfather in this pedig [17]Muhammadal-Mazlum, is problematical mainly because Free Grenville construed onereference to him as applying to [36]Muhammad al-cAdil II. The reference inquestion identif him as Muhammad b. Husain b. Sulaiman and associates him witthe repair of the mosque in the mid-fifteenth century. This makes him abrother of[15]Ismacil b. Husain b. Sulaiman who reign, even in Freeman-Grenville'schronology, is assigned t this period. They had a non-reigning brother namedSayyid Hajj Rush b. Husain Ashazifiki who is also associated with t repair of themosque. Their father, known as [10]Husain b. Sulaiman from the pedigrees ofIsmacil and Muhammad, is mentioned in late-thirteenth century context as HusainAshazifi b. Sulaiman. His father, known as [6/9]Sulaiman from his pedigree,eluded Freeman-Grenville in his translation and chronology. The relevant passageabruptly introduces a cert al-Maczul (the Deposed One) and indicates that, after areig of slightly more than two years, he was succeeded by "his sa and regent al-Husain b. Sulaiman. '37 On comparison with the1 Portuguese account, the nameor adjective al-Maczul seems to compensate for the omission of an earlier reign

by the same Sulaiman. This [6/9] Sulaiman al-!aczul, as we will hencefo identifyhim, is the only 24ahdali king who, on genealogical and chronological grounds,qualifies to be the son of [4]Ab Mawahib from whom the amirs descended. Inother words, the nomenclature of [17]Muhammad (al-Mazlum) b. Husain(Ashazif b. Sulaiman (al-Maczul) supplies the crucial link between [4]Abu'l-Mawahib on the one hand and the powerful amirs descended from his house.Our genealogy may be tested against the chronology impl in the pedigrees asfollows: It shows that nine generation passed away in Abu'l-Mawahib's lineagefrom the founder down cAisha bint al-Amir cAli. A thirty-year average betweengenerations (acceptable because descent in this lineage was often through thefirst-born) would give us a total of 270; years. These fit remarkably betweentheYfounding of the d about 1277 and the death of cAisha in 1554. Naturally, wehave to adopt a higher average for the eight known generati in [36]Muhammad al-'Adil II's lineage. Besides being desce from the youngest of threebrothers, thislineage shows on likely case of descent through the first-born. Athirty-fi yearaverage in that case would give us a total of 280 year would place Muhammad al-cAdil II (whose dates are not preci known) somewhere at roughlythe same timeas cAisha.The chronology supplied to our table generally agrees wi that computed for theMahdali period by Freeman-Grenville. him we have essentially relied upon theregnal durations in Portuguese account. These have more or lessbeen confirmedthe excavations and they agree with the few absolute dates188

KILWA DYNASTIC HISTORIOGRAPHYour disposal. The durations assigned to Abu'l-Mawahib's successors down to theaccession of [29]al-Fudail amount to a total of 162 Muslim years, equivalent to158 years of the christian calendar. These fit reasonably betweenthe death of[4]Abu'l-Mawahib in 1332 and the probable accession of al-Fudail some timebefore the recorded birth of the author of KitabaJ-SulWa in 1499 during his reign.Naturally, we have had to make minor adjustments because over twenty reigns areinvolved and the durations of only a few are specified down to the month or day.The Kitab tends to complement the Portuguese account on the durations of thereigns and in no case does it fully contradict it. We have seen, forexample, thatwhile it "disagrees" with the Portuguese account on [13]Muhammadal-cAdil'sreign, this almost certainly is an error resulting from the gap in the text which,among other things, suppresses the regnal duration of Muhammad al-cAdil's son[14]Sulaiman and assigns his twentytwo years to the father. These exists a similargap affecting the transition from [3/5]Daud's second reign to [6/9]Sulaiman al-Maczul's first reign. The Kitab assigns Daud's second reign a duration of twenty-four days though it goes on to describe the king as "the man of proofsin his time."At the same time it omits [6/9]Sulaiman al- raczul's first reign even though it laterintroduces him abruptly as al-Maczul (the Deposed One). The Portuguese accountprovides a solution to this problem. It assigns twenty-four years for Daudfollowed by twenty days for Sulaiman. The obliteration of a line or two from thetext could have-combined the twenty-four years of Daud with the twenty days of

Sulaiman. Otherwise the two sources are remarkably in agreement; for example,the Portuguese account assigns Sulaiman's second reign a duration of two yearsand four months as against two years, four months and fourteen days in theKitab.38The only serious case of divergence concerns a period oftwenty-four years, four months and twenty days. These follow [6/9]Sulaiman al-Maczul's second reign and may be assigned roughly between 1360 and 1390. Theduration in question is assigned with precision in the Portuguese account to acertain [ll]Sulaiman whose reign was omitted in the Arabic work. This king isknown from his sons' pedigrees as Sulaiman b. Husain. The Kitab does not seemto suffer a gap in this case; it reads coherently and provides a different perspectiveon the quarter of a century involved, recounting that al-Maczul'sson, [10] HusainAshazifiki, occupied the throne for a few days after his father,then made thepilgrimage to Mecca, then "returned safely to his country" and enjoyed a longreign of twenty-three years. The author leaves some evidence of deliberation onthis point, saying that "no one has been mentioned to us as his regent during histravel to Mecca." Here, as in other cases189

190where the text is somehow inadequate, we encounter the expression "God knowsbest."'39The precision of the regnal durations at this juncture, coupled witha number ofother peculiarities, lead us to be that the period became one of controversy anddebate among Kilwa traditionists. [ll]Sulaiman b. Husain is knownfrom i coinsbearing his name at a period when few coins were stru in Kilwa. The historicity of[10]Husain Ashazifiki cannot I rejected either. In omitting him, the Portugueseaccount a faulty parentage to his son [15]Ismacil while describing thE latter'sbrother [17]Muhammad al-Mazlum, merely as "a poor D of the line of kings." Byand large, we seem to be dealing a period in which the kingdom was dividedbetween the two lineages, possibly with Abu'l-Mawahib's descendants reigning inMafia. First, we are expressly told that [6/9]Sulaiman s Maczul was buried nearMafia in the cemetery of a certain al Faqih Daud and we find a commenton themargin of the manusc which reads: "Find out about the cemetery of al-Faqih Daudthe manara." Secondly, Abu'l-Mawahib is in 6ne case describ as "thefirst to reignat Mafia to this our day."" Both pi of information, coupled with the generalarchitectural decli in Kilwa evidenced for this period by the excavations, suggerather strongly that the kingdom was divided. Its re-unific tion, possibly alongwith some revival, seems to date to the reign of [13]Muhammad al-cAdil I, a kingknown by the peculi nickname of al-Matar al-Jadid (the New Rain).4'Before turning to the Portuguese account, the main develo ments preceding thearrival of the Portuguese should be summarized. As already indicated,the non-royal [16]al-Amir Muhammad dominated Kilwa, possibly beginning the reign ofMuhammad al-cAdil, and throughout the reigns of [14]Sulaiman and 115]Ismacil.At his death, [17]Muhammad al-Mazlum apparently stepped forth with vigor andreplaced him, assign the emirate to his own son [21]Sulaiman. This, however, was

soon challenged by the other royal lineages, and Muhammad wa deposed ongrounds which, judging by his identification as al-Mazlum (the Ill-Judged One),appeared unjustified to some traditionists. After this the three main posts werepartiti1 among the major branches of the royal family. Sulaiman retained theemirate, by far the more stable and more powerful post. [18]Ahmad b. Sulaimanb. Muhammad al-cAdil became kin while Sacid b. Hasan became vizier. Thearrangement did not outlive the king who apparently negotiated it. [20]Sacideventually became king and he in turn-was dominated by [211 Sulaimanevenbefore the latter could assume the throne. Th( one aspect of the arrangementwhich survived was the retenti( of the emirate by al-Mazlum's sons andgrandsons.The career of [28]Kiwab b. Muhammad al-Mazlum as amirstrengthened the power base of his family but at the cost of propaganda which isfully reflected in both records. HeELIAS SAAD

KILWA DYNASTIC HISTORIOGRAPHYattempted, it seems, to abolish the sultanate altogether. As a first step, he assignedit to a non-royal notable, a certain [24/26]al-Hasan b. al-Wazir Sulaiman b. Yarik.The Kitab says that he had a "motive" of his own, while the Portugueseaccountexplains that he took this arbitrary step so that he "might have more absolutepower than the one he had raised to the throne. Whatever this means, we knowthat eventually Kiwab assumed full sovereignty (symbolized inpublic by theroyal umbrella) apparently without adopting the title of sultan. In this bid hefailed, however, because the Friday prayers at which his sovereignty wasproclaimed were attended by only a third of the normal congregation. Hesubsequently "deposed himself" but retained the emirate and passed it on to hisnephew Ibrahim. In the meantime, [29]al-Fudail b. Sulaiman b.Muhammad al-cAdil enjoyed the throne more as a figurehead than a ruler. [30]Ibrahim outlivedhim and was in sole control of Kilwa at the arrival of the Portuguese.43The Portuguese AccountWe have suggested that the Portuguese account is an impromptu compositiongiven to the Portuguese during their occupation of Kilwa between 1505 and 1512.Our basis for this suggestion is a consistent bias in the narrative which obscuresthe royal ancestry of [30]Ibrahim b. Sulaiman and his immediate kinsmen.Furthermore, there exists some evidence that the Portuguese at first acted on basisof the narrative and that later they recognized its error.Briefly, the Portuguese approached Kilwa from the mistaken view that Ibrahim (a"tyrant" in the text of De Barros) had no legitimacy in the power structure of thekingdom. When he retired to the mainland at the occupation of Kilwa,Almeidaraised to the throne a wealthy non-royal notable who had shown willingness tocooperate with an agent left behind in 1500 by Cabral. This notable,[31]Muhammad Rukn al-Din (Arkony or Ankony), was killed a year after hisaccession by an ally of Ibrahim. The Portuguese already had doubts about theirprevious policy but they nonetheless chose Rukn al-Din's son [32]Hasan tosucceed him. Troubles ensued which led to a partial exodus of the inhabitants and

paralyzed the lucrative commerce which the Portuguese had hoped to control.This caused them to reverse their policy and to enthrone Ibrahim's cousin,[33]Muhammad Mikati b. Muhammad Kiwab. They apparently hoped that Mikatiwould reconcile with Ibrahim and his forces but this did not eventuate and by1512 the Portuguese had to evacuate Kilwa and depart.The bias in the narrative reflects the hand of a sophisticated partisan whose familymay have had a long-term interest in the politics of the island. In describing[17]Muhammad al11azlum as "a poor man of the line of kings," he implied thathis

ELIAS SAADlineage was a peripheral fraction of the royal family. As already indicated, heconfused the parentage of [15]IsmaCil while omitting [10]Ashazifiki altogether.His most serious error lay in describing [6/9]Sulaiman al-Maczulas son of[3/5]Daud, occasioned by his statement that Daud had succeec [4]Abu'1-Mawahibbecause the latter had no sons.We know in fact that Daud succeeded Abu'l-M{awahib partly because of aprecedence often given to the eldest living brother. In this case, however, hisprecedence was also prE dicated on the fact that he had briefly enjoyed the royalhonor at his father's violent death during Abu'l-Mawahib's absence. The Kitabgoes into some length to explain how Dau graciously stepped aside onthe returnof his older brother. The story may imply that Daud's accession wassome sort ofexception made necessary by circumstance. Otherwise, we kno for certain thatAbu'l-Mawahib had at least one son, whom Ibn Batuta mentioned. Indeed,without naming him, he tells that the son had been designated heir apparentduring his father's lifetime, and expressed his regret at the accession Daud, anevent he learned about on his return journey from Kilwa. 44The bias in the Portuguese account is expressed most stro7 in its treatment of thethree amirs, [21]Sulaiman, [28]Kiwab and [30]Ibrahim. The narrative omits theirparentage and, ii so doing, effectively portrays them as a series ofusurpersunrelated to each other or to the royal line. Moreover, sin4 the three had beenpreceded in stepping from the emirate toI throne by the non-royal [16]al-AmirMuhammad (a real usurpat" duly recorded), the narrative may have given thePortugueseI impression that royal descent had ceased to be crucial as aqualification for assuming the throne. This might explain wl they enthroned[31]Rukn al-Din even though they were clearly concerned in maintaining thestructure of the kingdom unchar as much as possible.This naturally leads us to suspect the hand of Rukn al-Dit or his son [32]Hasanbehind the narrative. It is possible t Rukn al-Din (whose ancestry is not indicatedin his-nomencla ture) was son or grandson of the royal [16]al-Amir Muhan-OBoth were noted for their great wealth which in the case of Rukn al-Din may havebeen inherited. The author of the Kit praises Rukn al-Din and says that he"virtually owned the wh land" through his charity and generosity. However, heblam him for his contacts with the agent left behind by Cabral pr ceding theoccupation. During these contacts he allegedly him everything, big and small,because he was bent on old resentments and feuds." This statement may suggest

that the historical narrative in De Barros was given to the Portuguesby Rukn al-Din. It also gives added credibility to the possibility that Rukn al-Din wasdescended from al-Amir Muhamma since the "old feuds" could referto hisfamily's attempt to capture the emirate from [30]Ibrahim's lineage.192

KILWA DYNASTIC HISTORIOGRAPHYon the eve of the conquest, Rukn al-Din advised Almeida that the people of Kilwawere dissatisfied with Ibrahim and that "they all desired a king nearer to theirroyal line." The wording here suggests that Rukn al-Din had a preferredcandidate from [13]Muhammad al-cAdil's lineage and that, if he desired anythingfor himself, it was the emirate. But Almeida proceeded immediatelyto enthronehim, much to Rukn al-Din's amazement and misgiving. He tried to confer somelegitimacy on this imposed arrangement by namin a son of [24]al-Fudail as hisamir and prospective successor. However, when Rukn alDin was killed, his son[32]Hasan, determined to avenge him, abrogated this arrangement. It is possiblethat Hasan, rather than Rukn al-Din, gave the biased historical narrative to theportuguese at his father's death. De Barros records a hearing held at that time inwhich [33]Muhammad Mikati fully explained, with possible exaggeration, hisaugust descent, but his enthronement was postponed for a year, afactor whichsuggests that Hasan gave evidence to the contrary. 47 This evidence may be thedocument described by De Barros as "a chronicle of the kings of Kilwa" in hispreface to the historical narrative. As far as we can tell, only a translation of thedocument reached De Barros. The latter incorporated it into his DaAsia withoutrecognizing its inconsistency with other information which he duly recordedelsewhere in his compendium.In any case it is not necessary to determine whether Rukn al-Din or Hasan, or oneof their partisans, was the sponsor of the Portuguese account. All we know is thatit is biased against their opponents and almost certainly must be dated prior to theenthronement of Mikati in 1506/07. Likewise, although we identify it as animpromptu composition, this does not mean that no chronicle then existed fromwhich it could have been drawn. Chronicling in Kilwa, mainly in the form ofkinglists, may date to early in the Mahdali dynasty. In any case, coins andinscriptions helped keep alive the names and pedigrees of previouskings. In thediscussion which follows, we will posit the existence of an early chronicle datingto the mid-1400s. This chronicle is associated with Mikati's and Ibrahim'sgrandfather, [17]Muhammad al-Mazlum.Kitab al-SuiwaThe composition of the Kitab al-Suiwa belongs to a period much later than thePortuguese account. Freeman-Grenville assigned its compositionto around 1520(when its author was still quite young) as a result of confusing [36]MuhammadalCAdil II, who apparently sponsored it, with [17]Muhammad alMazlum, who isassociated with a chronicle of his own. Identifying the two kings as one requiredhim to argue that Muhammad al-Mazlum could not have lived long into thesixteenth century.

-, -ELIAS SAADOur own genealogical reconstruction distinguishes between th two and allows usto assign Muhammad al-cAdil II to somewher in the mid-1500s. The author of theKitab has left us littl information on the details of this period becausehis narratcame to an abrupt ending in the middle of the seventh chapte where he records anepisode involving the humiliation of his own uncles by da Gama in 1502.48 Fromthe titles of the'las three unfinished chapters (recorded in the introduction) welearn briefly that Ibrahim regained Kilwa after the departur of the Portuguese andassigned the throne to his brother [34]Sacid. The latter must have been succeeded,though not necessarily directly, by [35]Husain b. Sulaiman b. Muhammad al-cAdil, who is mentioned with the title of sultan in the pedigree of his son andsuccessor, Muhammad al-cAdil II. Si, chapter headings are not always correlatedto reigns, the soneed not have succeeded the fat1er directly. Judging from t'funerary inscription recording the death in 1554 of Ibrahim'; niece cAisha, itseems that the two royal branches were stil alive and well in the middle of thesixteenth century.In a digression on [24/26]al-Hasan b. al-Wazir Sulaimanb. Yarik, the author has left us a piece of indirect evidenc concerning the date ofhis own composition. After serving a, Kiwab's puppet, this scion of the vizierialfamily asserted I self and was eventually driven out of Kilwa with hisfollowisometime after 1490. Subsequently, he founded a new sultanal on the mainland ata place called Maghamghub. There, he was succeeded by his son Muhammad,husband to one of [30]Ibrahim' sisters. In turn, Muhammad was succeeded bytheir son Hasan This suggests that a long time had passed since Hasan'sdepa ture.The author confirms this by saying that "ever since tt day, they established thatcountry [and inhabited it] to thi. our own time." Moreover, he wrote that Hasan b.Muhammad wa reigning in Maghamghlub at the time of writing and described asa contemporary (muCasir) of [36]Muhammad al-cAdil I.49I all this is taken intoconsideration, it appears reasonable assign the composition ofKitab al-Suiwa atthe instance of Muhammad al-cAdil II to somewhere around 1550, when the authwas in his late middle age.That historiography was still a sensitive matter at that time is evident from thevery first paragraph of the Kitab. There, after recording thathe embarked on hiswork reluctant to oblige his king, the author says:I pray God to make it a book which would gladden Kthe eyes by its reading and put minds at ease by itsadoption (intiyajihi). I have collected it [from?]the collection (jamic) the unique of benefits andthe wanderings of intents [which is] like the chargedconstellation whenever it is read by its reader it

KILWA DYNASTIC HISTORIOGRAPHY 195drives him to the talk of sorrows . . . I have calledit the Book of ConsoZation on the History of Kilwa(Kitab al-Suiwa fi Akhbar Kuiwa). 50

T he structure of this passage is clearly defective and it would be useless to insistthat our translation of it is fully accurate. We have interpreted the word jamic asdesignating a ,,collection" or "work" because the phrase "the unique of benefitsand the wanderings of intents," rhyming as it does in the archaic fashion of booktitles in Arabic, may refer to the earlier chronicle whose existence wehavealready hypothesized. In any case, the author wanted to "gladden the eyes' of hisreader and so he could in no way have been referring to his own workasevocative of sorrow. Freeman-Grenville in his translation has the author say:"When I gave it to a critic to read, it drove him to the talk of sorrows." Thisinterpretation, which has nothing to justify it, obscures the main impressioncreated by the opening paragraph, namely, that a chronicle had been in existencein Kilwa which, possibly because it favored one lineage over another, needed tobe superseded by a more conciliatory approach.51 On two occasions the authorreferred to the authors orsponsors of the earlier chronicle as "the people of the history" (ahlal-taCrikh). Inthe first instance, concerning the migration of the Shirazis directly from Persia, hespoke of what the people of the history had "claimed" (zaCamu). Inthe secondinstance he attributed to them information which has been proven wrong by theexcavations. He quoted them saying that "at the time of [14]al-Sultan Sulaiman b.Muhammad alcAdil there was built the mosque which had collapsed atthe time of[4]Abu'l-Mawahib." In fact, the excavations are far more open to the idea that themosque was built at the time of Abu'l-Mawahib and was only partiallydamagedsomewhere around Sulaiman's reign in the mid-1400s.52The repair of the mosque is inconsistently treated in the Kitab. The textcreditsSayyid Hajj Rush, brother of [17]Muhammad al-Mazlum, with the initiative in themost roundabout way. He allegedly offered one thousand mithqalsto[14]Sulaiman b. Muhammad al-cAdil but the latter would not authorize him toundertake the repairs except at the king's own expense. Hajj Rush kept the king'sfunds aside, using his own, and subsequently returned them to Sulaiman's heirs.The controversial character of this story is highlighted by the ascription of povertyto Muhammad al-Mazlum in the Portuguese account. Indeed, the Kitabshows allMahdali branches to have been overshadowed in wealth at the time bythe non-royal [16] al-Amir Muhammad. Thus when [20]Sacid attempted to contest theenthronement of [15]Ismacil with the help of Zanzibar, the amir foiled the projectby sending a bribe of one hundred mithqals to the commander of the Zanzibariforces. Subsequently he

ELIAS SAADstepped to the throne himself, taking precedence over a hos contenders, precisely"because of the weakness of the kings the shortage of funds.",53It is possible thathe financed repairs to the mosque himself, either during or beforehis short reignand later a scramble apparently took place over the credit for the restoration.What we know for certain is that the author of the Kitab al-Suiwa, writing acentury after these events, received coi tradictory information for, besidesassigning the repairs t, [14]Sulaiman's reign, he also associates them with[17]Muhal al-Mazlum. Freeman-Grenville tried to resolve thisinconsistency at

the cost of a drastic departure from the text. relevant passage, which will bequoted below, has several implications. First, it is there that Muhammad al-Mazlum's pedigree (the same as that of [15]Ismacil) appears. Second' Muhammadal-Mazlum is there associated with the earlier chronicle and shown to have livedinto old age long after h reign. Thirdly, he is identified as an eye-witness to a gigantic tree which allegedly provided the timber for replaci the fallen columns.Construction in wood has a legendary quality in Kilwa; Ibn Batuta mentioned it inpassing but th excavations have not borne out his remark.54 After relati his ownstory, our author observed:I have seen someone who saw the tree in its saidshape, and he is al-Sultan Muhammad b. al-SultanHusain b. al-Sultan Sulaiman. He it was who ordered the building of the mosqueaforementioned. Al-SultanMuhammad was [Also] the one who ordered the composition of this history. Godknows best. 55The revelation that the author "saw" Muhammad al-Mazlum, pr sumably in hischildhood, suggests that the ex-king was stil living during the Portugueseoccupation. The coincidence tb he is associated with a chronicle, notwithstandingthe rese vation implied by "God knows best," opens wide scope for cw jecture. Asa first step, we may propose that al-Mazlum pre sented the Portuguese a copy ofthe chronicle in order to firm the rights of his grandson [33]MuhammadMikati at[311' Rukn al-Din's death. As a second step, we may go further an' suggest that[32]Hasan b. Rukn al-Din subjected the document to some form of censorshipwhereby he struck out unfavorabl details and altered other parts. It would not beillogical, given this line of reasoning,-to assume further that the document in itsdamaged form remained in the hands of the kings and was eventually passed on toour author so that he' might "correct" it and produce a less controversial work.Attractive as this thesis may be, the mere statement tha author saw theex-king,possibly an error in itself, is an t sufficient clue. All we know is that the Kitabsupplies moi196

KILWA DYNASTIC HISTORIOGRAPHYinformation on Muhammad al-Mazlum and his ancestors than on the ancestors ofthe very king who sponsored it. Additionally, ,early all the gaps andambiguitiesin the text pertain to the period up to the reign of al-Mazlum. The rest of thenarrative is detailed and straightforward with the single exceptionof the passagewhere the author records his own birth in 1499. There he says that hewas giventhe same name as the earlier author but without recording the latter's name. Thismight be an elusive reference to the fact that Muhammad al-Mazlum(b. gusain b.Sulaiman) bore similar nomenclature to Muhammad aI-CAdil (b. Husain b.Sulaiman b. Muhammad al-cAdil).sGFor the sake of avoiding repetition, we will henceforthexamine two closely related factors at the same time. The first is that the authorleft us only his rough'draft. The second is that he drew his rough draft, at least upto the reign of guhammad al-Mazlum, from the earlier chronicle. Our central

thesis is that the earlier chronicle was written in-the mid1400s and wassubsequently censored or some other way damagedbefore reaching the author.The tentative element in the Kitab is most strongly illustrated by a seriesofcomments occurring on the margin of the manuscript which we have alreadymentioned above. These comments are of a uniform nature and, once again,pertain to the period up to the reign of al-Mazlum. They seem to indicate theauthor's uncertainty about his information much like the expression"God knowsbest." The first of these comments suggested to us to read critically the portion ofthe Kitab covering the first sequence of Shirazi kings. For, next tothe account ofthese kings, we read on the margin: "Find out who was the first to reignin Kilwaand Mafia." Another comment concerns the non-sovereign amir appointed by themutamandalin. It reads: "Find out about the reign of alMandhiri during twelveyears." By far the most interesting of these comments accompanies the passagequoted above concerning the mosque and al-Mazlum. There, we read: "Find outwho ordered the composition of this history." 57 The author apparently inferredthat the earlier chronicle was sponsored by al-Mazlum, probablybasing hisinference on the grounds that it covered the history of Kilwa up to his reign andhighlighted the careers of his kinsmen over those of other royalbranches. The latter possibility may well have invited the hand ofa censor longbefore the time of Rukn al-Din and hisson Hasan.The strongest indication that the author relied on adamaged text occurs in-a passage which also illustrates the tentativeness of theKitab. The passage seemed to be concerned with the circumstances which delayedthe return of [4]Abu'lMwahib from Arabia, so that he was precededto the thronefor two full years by his brother [3/5]Daud. These circumstances197

ELIAS SAADappear to have had a strong bearing on later dynastic dispu The firstMahdalissucceeded each other from father to son. [l]Hasan b. Talut was succeeded by hisson [2]Sulaiman and, when he was killed, there was no attempt to enthrone one ofbrothers or uncles. With the second accession of Daud, how a strong precedentwas set in favor of the rights of surviv brothers, giving rise, as we have seen, to aproliferation royal lineages constantly at odds with one another.The relevant passage begins by mentioning Abu'l-Mawahib? travels to Arabia.After that the author says: "Ido not know his circumstances, but I will enquireabout his circun stances." 58 From there the text proceeds as follows:Then she hid her son Abu'l-Mawahib behind ascreen, so he could hear what his brother said whenhe came. She sent for her son Daud, and, when hearrived she said: What will be your position whenyour brother returns from the Hijaz? This is thetime when he is due to arrive, so what is to be done

when he arrives from Hijaz? Daud replied: There will be no quarrel between uswhen he comes. Theland is his land and I shall obey him. I am hisdeputy till he comes, if that is God's will.9This curious outline of events, beginning with an abrupt re ence to themother ofthe two kings, is inexplicable, even rough draft unless, as we have hypothesized,the author was copying verbatim from a damaged text. The earlier chronica ifindeed sponsored by al-Mazlum, would have had an interes in describing thedetails of Abu'l-Mawahib's uncontested as tion of the throne. 'However, in thecopy which apparently reached the author, these detail* had probably beensomehou altered or struck out.In light of the above, we should point out that the pers tive of the Kitabon thedynastic politics is essentially opposed to that of the Portuguese account. Thus,[IO]Ashaz is featured in it at the expense of [12]Husain b. Sulaiman. This maystem from its reliance upon the chronicle of al-Ma The gaps, on the other hand(which are to be distinguished the deliberate omission of Shirazi reigns), all seemto bea on controversial events. They affect the reigns of [2]Sulaal-MatcUn (theStabbed One), [6/9]Sulaiman al-!Maczul (the Deposed One), [12]Hasan b.Sulaiman (who inexplicably gave to Muhammad al-cAdil) and, finally,[17]Muhammad al-Mazlum Ill-Judged One). Some of the gaps may betray thebiases of partisan of [13]Muhammad al-cAdil's lineage. That partisan have turnedaround the crucial sentence mentioned above cot cerning the damage andrestoration of the mosque. As it st the sentence stigmatizes Abu'l-Mawahib for thecollapse of mosque in his time when, in fact, he was apparently the onE built it.60

KILWA DYNASTIC HISTORIOGRAPHY199Could it be that the earlier chronicle or else the Kitab itself was censored by theauthor's own king? We have considered this possibility on basis of certainpeculiarities which it ,ould be too tedious to relate. 61 All that can be said on thispoint is that the author has left us very little information, beyond nomenclature,on the ancestors of [36]Muhammad alYCAdil II. We may add that the authorembarked on his work reluctantly and never completed it, perhapsbecause hesimply did not wish to be implicated in the dynastic disputes.Before concluding this paper, a word must be said for thesake of clarity concerning the extant copy. Throughout we have worked on theassumption that it is a faithful reproduction of the author's original composition.So far there exists no evidence pointing to the contrary. We know thatit wasmade from a document which came into the hands of John Kirk whenhe wasBritish Consul in Zanzibar in 1877 and that Sultan Barghash also took an interestin the copying. The task was assigned to a copyist who appears to have been bothcapable and conscientious. He was careful to cite his name in full as CAbdullah b.Misbah al-Suwafi and to record the precise date when he completed the copy. Asfar as we can tell, he preserved all the information legible in the original, thoughminor scribal errors cannot be precluded. Certainly, neither he nor any of hiscontemporaries had any interest in altering the text substantially.When he

reached the abrupt ending in the seventh chapter, he indicated as much by noting,"here ends what I found.",62NOTES1. For the Portuguese account, see Joao de Barros, Da Asia(4 vols.: Lisbon, 1945-46), l:pp. 323-28. References will also be made to theEnglish translation in George M. Theal,Records of South-Eastern Africa (7 vols.: London, 18981903), 6:pp. 233-44. Theextant copy of the Kitab is atthe British Museum, MS Or. 2666. It was published inStrong, "History of Kilwa," Journal of the Royal AsiaticSociety, (1895), pp. 383-430.2. G.S.P. Freeman-Grenville, The Medieval History of the Coastof Tanganyika (London, 1962), p. 44 et passim. Idem,"Coin Finds and Their Significance for Eastern AfricanChronology," Numismatic Chronicle, 7th Ser., 11, (1971),pp. 283-301.3. H.N. Chittick, "The Shirazi 'Colonization' of East Africa,"JAH, 6(1965), pp. 275-94. Idem, "On the Chronology and Coinage of the Sultansof Kilwa," Numismatic Chronicle,7th Ser., 13, (1973), pp. 192-200.4. The translation is in Freeman-Grenville, ed., The EastAfrican Coast: Select Documents (Oxford, 1962), pp. 34-49.

ELIAS SAADFreeman-Grenville does not indicate his omission of t] marginal comments nordoes he note his departure from text at points where alternative readings seemedto h necessary. My disagreements with his translation concern the most difficultparagraphs and will be notedwe proceed.5. Compare Kitab, fols. 8-9 (Strong, pp. 415-17) with de Barros.The manuscriptof the Kitab is not'foliated to avoid confusion I follow the enumeration adopted b,Strong.6. A genealogical reconstruction was attempted by John Walker, "The History andCoinage of the Sultans of Ki Numismatic Chronicle, 5th Ser., 16 (1936), pp. 41-81., Walker had recourse to the pedigrees but often subord: ated them to theconfused (and, as we shall see, bias(information in the Portuguese account. Among otherproblems he showed the mid-fourteenth century Talut b, Husain asbrother of themid-fifteenth century Ismacit Husain b. Sulaiman. Additionally, he attached thela, Shirazis to the genealogy of the Mahdali family whiledissociating the last amirs from it.7. See Yaqut, Kitab MuCjam al-Buldan (Jacut's Geographis (6 vols.: Tehran,1965). Malindi is mentioned next tc (Brawa, misspelled Bwara) under the letter"B", but n( separately. Tumbatu is mentioned as a new city to wh.the inhabitants of Languja had moved. See, for examp 1: p. 485; 4: p. 366. Cf.Chittick, Kilwa: An IsZaz Trading City on the East African Coast (2 vols.: Nairc

1974), p. 237.8. For a discussion of possible ShiCi influence at Kilw see S. Hamdun and N.King, 2Tn Battuta in Black Afria(London, 1975), Appendix II.9. See S. Flury, "The Kufic Inscriptions of Kizimkazi Mo Zanzibar,"Journal ofthe Royal Asiatic Society (1922 pp. 257-64. The unknown monarchis identifiedas AbucAmran Musa b. al-Hasan b. Muhammad. Cf. C. Velten,Prosa und Poesie der Suaheli (Berlin, 1907), pp. 243' For possibleconnectionsbetween Shirazi and Shica of Zaidi sect, as well as "Amranis" see R.Pouwels,"The Medieval Foundations of East African Islam," JAHS, 11(1978), esp. p. 225.10. The identity of the Mutamandalin is problematical; tname may be a corruption of al-Mutamandhirin, i.e. "'i claiming Mandhiri descentor nisba." One of the rule appointed by them over Kilwa is identified as Muhamal-Husain al-Mandhiri. As is well known, the adopti a prestigious Middle Easternnisba, much as in the ca Shirazis and Mahdalis, does not preclude a local origiThe local home of the Mutamandalin is said to be Shag200

KILWA DYNASTIC HISTORIOGRAPHY 201name with many associations in the oral traditions butnonetheless obscure.II. Al-MasCudi, Muruj ai-Dhahab wa-MaCadin ai-Jawhar (4 vols.: 1948), 1: pp.59-65, 244-246, suggests that Omani andSirafi captains sailed directly on the open sea toQanbalu (Pemba) without putting in further north along thecoast. Al-Jahiz in the ninth century mentions Qanbalu asthe port par excellence to which Muslim navigators went.His brief and second-hand information suggests thatLanguja (at Zanzibar) was a major power ruled by blacks at the time.Musliminfluence at both places, by Shicifugitives who themselves may have been partly or fully black, cannot be ruledout. See al-Jahiz, Kitab Fakhrai-Sudan Cala al-Bidan, in Tria Opuscula Auctore, ed.G. Vloten (Leiden, 1903), pp. 57-66.12. See Chittick, Kisimani Mafia: Excavations at an IslamicSettlement on the East African Coast (Dar es Salaam, 1961).One of the early mosques now stands at the shoreline, partof it having already been destroyed by water, whilemasonry blocks attest to previous buildings in parts ofthe site now submerged. For Manda see Chittick, "Discoveries in the LamuArchipelago," Azania, 11(1976), pp.37-67. Elsewhere Chittick has not laid much emphasis onthe implications of the rising sea-table. See his "A New

Look at the History of Pate," JAH, 10(1969), pp. 375-91. 13. See James Kirkman,Men and Monuments on the East AfricanCoast (London, 1964). At Mkumbu on Pemba, believed to beQanbalu, Kirkman observed that the remains extended overland now submerged. One submersion legend, concerningthe island town of Ghama, is recorded in Kitab ai-Zunuj. 14. See Chittick, "On theChronology," pp. 193-94 and Kilwa,2 :pp. 269-70. The Portuguese account involves confusing the nomenclature ofCAli b. Husain b. CAli (the founder).At one point it is reduced to CAli b. cAli, (Ale Bumale).At another point he appears as Hocem, presumably a reference tohis father, whilein a third case he is identifiedsimply as "Ale."15. See Chittick, Kilwa, 2: pp. 274-301, for the provenance andstratigraphy of the coins. For reasons discussed byChittick the stratigraphy supplies only a rough chronological index, especially incases (such as the coins of Daud b. Hasan) where few samples havebeenrecovered.16. De Barros, Da Asia, 1: pp. 324-25 (Theal; Records, 6: pp.241-42). The Portuguese 'translator' at first treated "Ibn" (Bin, 'Bnu, depending oncontext) as an integralpart of the name. Subsequently, his informant must havecorrected him for he identified the Mahdali kings only bytheir personal names while indicating their parentage(often defectively) separately.

ELIAS SAAD17. Freeman-Grenville, The Medieval Coast, pp. 66-73, datethe Shirazis from 957 to 1277 and assigns the lastgrandson of the founder, the problematical Hasan b. Dato 1106-1129.18. See Kitab, fol. 8 (Strong, pp. 414-15). The statementthat CAli b. Daud b. cAli was the last descendant of tShirazi founder is questioned by the invocation "Godknows best."19. De Barros, Da Asia, pp. 324-25 (Theal, Records, 6:241The Portugueseaccount described Ale Daute, the lastShirazi, as grandfather of Hasan b. Talut, the Mahdali founder, but this may be anextension of the confusion between him and his earlier namesake. The Kitab and igenealogical evidence leave little doubt that the Mahd were a new dynasty,though kinship on the maternal sidcannot be excluded.20. Walker, "History and Coinage," understood the descriptof Sulaiman as "of royal blood" to mean that he may ha been a usurper fromwhom the Mahdalis descended. Chit "Shirazi colonization," followed thisproposition whil stipulating further that Sulaiman and his eight succes

duplicate the names of other kings.21. The Portuguese text reads as though succession among tfirst four of the omitted kings was from father to son The clue in the name ofSoleimao Hacen, however, alter the picture. As Sulaiman b. Hasan hecannot havebeen son of Daud. Likewise, Boni Soleimao cannot have been nephew of Daud(the second of that name among the omit kings) unless the latter was brother(rather than son) Soleimao Hacen. Compare De Barros, Da Asia, l:pp. 324 (Theal,Records, 6:pp. 241-42) with our genealogy of tShirazis.22. Ibn Batuta, Voyagesd' Ibn Batoutah, ed. and trans. C.Defremery and B.R. Sanguinetti (4 vols.: Paris, 1969[reprint of 1877-93 ed.]), 2: pp. 192-95.23. For reasons not altogether persuasive, Chittick believthat Muhammad al-cAdil's coins were minted at MogadishSee his "On the Chronology," p. 195.24. Concerning the coins of Nasir al-Din wa'l-Dunya MuhammChittick, ibid, p. 194, says "On archaeological groundcannot believe that these coins date from before the fifteenth century . . . If thename is really that ofsultan of Kilwa, I prefer a sixteenth century date, si presumably if he ruled beforec. 1520 he would be mentin the Chronicles." In fact, the Kitab mentions twoMuhammads from the fifteenth century, besides MuhammadcAdil.25. See, for example, Kitab, fol. 10 (Strong, p. 418). Thdeath of Sulaiman b. Muhammad al-cAdil confronted the

KILWA DYNASTIC HISTORIOGRAPHYkingmakers with a choice between "two grandsons who wereof equal descent on their fathers' and their mothers'sides." The two may be [20]Sacid b. Hasafi b. Sulaimanand [25]Sabhati b. Muhammad al-cAdil b. Sulaiman. Al-AmirMuhammad apparently objected to both, and therefore theirenthronement was long delayed. The notables finally chose [15]Ismacil b. Husainb. Sulaiman, whose claimswere apparently weak owing to questions surrounding thereign of his father [lO]Husain Ashazifiki.26. Yaqut, Kitab MuCjam, 4: p. 602, indicated that Mogadishuwas ruled by several elders, each representing a sectionof the population. The Portuguese text suggests that thissystem survived only at nearby Brava.27. For example, when Sacid b. Hasan unsuccessfully disputedthe enthronement of Ismacil, he sought refuge at thehouse of the qadi but found the latter unobliging. SeeKitab, fol. 11 (Strong, p. 419).28. Kitab, fol. 8 (Strong, p. 415). The reference to al-Amr

in this context confirms its identification with theemirate. Hasan b. Talut was assigned the emirate but he seized the throne byforce. The reference to ahlihi (itspeople or his people) could easily be mistaken (as inFreeman-Grenville, East African Coast, p. 38) as indicatingHasan's "people" or his family. Rather, it indicates "thepeople of the throne," or the Shirazi kings.29. Kitab, fol. 9 (Strong, p. 416). This certainly suggeststhat the Shirazis had continued to reign at Mafia; Abu'lMawahib was the firstMahdali to enjoy sovereignty overthat island.30. Kitab, fol. 9-10 (Strong, p. 417). Freeman-Grenville'stranslation, East African Coast, p. 40, recognizes the gaps or "lacunae" in thisparagraph but abbreviates and misreads the second sentence asfollows: "ThenSabhatiwas agreed as amir and the other notables of thegovernment?" As far as we can tell, Sabhati may havestood at the tail-end of one of the pedigrees, probablythat of al-Amir Muhammad. If by any chance he was adescendent of Bashati b. cAli, a grandson of the Shirazifounder (whose name in that case would appear to havebeen misspelled), then the Shirazis may have preserved someinfluence under the Mahdalis.31. Compare Kitab, fol. 10 (Strong, pp. 417-18) with DeBarros, Da Asia. In fact the reign of Sulaiman b. Muhammad al-cAdil (and henceits duration) would seem to be affectedby the gap. It is mentioned only in a digression concerning the restoration of themain mosque, which, as we shall see, provides inconsistent and controversialinformation.32. The word al-dunya (the world) implies wealth and power indistinction to at-din (religion) which is more concerned

ELIAS SAADwith the hereafter. Conceivably, Muhammad acquired tlhonorifics before he became sultan.33. See Kitab, fol. 11 (Strong, p. 420). In fact the texthas Abu'l-Madhhab in this case, apparently a scribalerror, although we considered the possibility that Abt Muwahib, having studied inArabia, may have been the fking to patronize the Shafici school (madhhab). IbnBatuta suggested that Orthodox Shaficism was prevalentalong the coast by the time of his visit.34. Muhammad al-cAdil is the only king whose ancestry istraced back to a great-grandfather. This may have beE deliberatelyintended todistinguish him from Muhammad(al-Mazlum) b. Husain b. Sulaiman. In one instance,

however, Muhammad al-CAdil is referred to more briefl3 as Muhammad b. al-Husain. Compare fols. 3, 15 with fc10 (Strong, p. 406, 427, 418).35. See B.G. Martin, "Arab Migrations to East Africa inMedieval Times," IJAHS, 7(1975), p. 374.36. The inscriptions are in Chittick, Kilwa, 2:pp. 259-64One of them bears the name of Hasan b. Sulaiman, presumably Abu'l-Mawahib.Another one, whose date is al no longer extant, bears the name of al-SultanSulaima al-Sultan Ismacil. This may be a son of [15]Ismacil Husainand hence abrother of [14]Hasan b. Ismacil. C genealogical grounds, he could not havereigned at a much later than 1512. Possibly, he succeeded [34]Sac b. Sulaiman b.Muhammad al-Mazlum sometime before the accession of [35]Husain b. Sulaimanb. Muhammad al-c The number and sequence of reigns after 1512in unkno37.Freeman-Grenville's translation, East African Coast,pp. 39-40, reads: "There succeeded him Talut Ibnal-Husain, whose reign was two years, four months and fourteen days. Hejourneyed to Mecca and left as his regent his son al-Husain b. Sulaiman." How,one won could Talut be father of Husain son of Sulaiman? The, translationconfuses between [8]Talut and [6/9] al-M (who clearly was named Sulaiman). Formy own readinsee note 38 below.38. If we were to reconstruct the gaps in the Kitab on.of the Portuguese account, the relevant passage wou as follows: "Then [Abu'l-11awahib] died and was sue by his brother Daud, the ascetic and pious king, thof proofs in his time. The duration of his reign wa twenty-four years. He wassucceeded by his nephew $ b. Hasan b. Sulaiman a7-MatCun !or twentydays.Then latter was deposed by al-Husain b. Sulaiman al-Matci reigned after him sixand a half years then went out.fight the pagans of the Mulli in a holy war and died martyr. Then Talut b. Husainsucceeded him Por oa and at his death the deposed Sulaiman once again a

KILWA DYNASTIC HISTORIOGRAPHYthe throne. The length of the reign of the deposed one was two years, four monthsand fourteen days. He thentravelled to [or towards] Mecca and left his sonal-Husain b. Sulaiman, the one nicknamed Ashazifiki, ashis regent . . .." The words in italics are added to compensate for gaps in the text.Compare Kitab, fol. 9(Strong, p. 417) with De Barros, Da Asia.39. For some reason Freeman-Grenville's translation reads"Hasha Hazifiki" in place of "Ashazifiki." We are unable to interpretthe meaningof his archaic Swahili nickname, but it seems to bear somehow on theconditionsof Ashazifiki's return, if at all, to Kilwa. Freeman-Grenville,East African Coast, p. 40.40. The statement may mean that Abu'l-Mawahib's descendants

uninterruptedly enjoyed the emirate of Mafia to the exclusion of other Mahdalilineages. If the kingdom was dividedduring the late fourteenth and early fifteenth centuries,then Abu'l-Mawahib's descendants may have claimedsovereignty at Mafia while their opponents reigned atKilwa. See Kitab, fol. 9 et passim (Strong, esp. pp. 41617).41. The reunification of the kingdom under Muhammad al-cAdilmay have given his descendants a claim on the thronestronger than that of other lineages. This may explainwhy the amirs recurrently enthroned them, though theyapparently enjoyed no real power.42. The Kitab (fol. 13, Strong, p. 423) attributes Kiwab'saction to "a motive he had in opposing the people of thesultanate," but it seems that his main opponents wereSacid's immediate kinsmen. Compare with De Barros, DaAsia, I, 326 (Theal, Records, 6, p. 243).43. In fact, Kiwab's action weakened Kilwa and left it in astate of disarray before the arrival of the Portuguese.Hasan b. al-Wazir Sulaiman enjoyed connections withZanzibar and managed to regain the throne one more time and outlived Kiwab.Ultimately, he fled Kilwa with hisfollowers, but only after a series of bloody struggles ata time when the Portuguese were making their first appearance along the EastAfrican coast. see Kitabfol. 13 ff. (Strong, pp. 423ff).44. Ibn Batuta, Voyages, 2, pp. 193-95. 45. Kitab, fols. 16-17 (Strong, p. 429).The statement thatRukn al-Din and his brother Ayyub "owned the land byvirtue of their generosity and their rescue of Muslims"may partly refer to the fact that they provided a ransomto Da Gama when Ibrahim refused to give him a tributedemanded by the admiral on behalf of the Portuguesemonarch in 1502. See Gaspar Correa, The Three Voyagesof Vasco Da Gama (London, 1869), pp. 293-99.205

ELIAS SAAD46. See Damiao deDom Emanuel,3:pp. 114-15. 47. De Barros, Dapp. 286-87).Gois, Chronicle of the Most Fortunate Ki Extracts and translation in Theal, RecorSee also ibid, 6:pp. 244-45. Asia, 7:p. 420 (Theal, Records, Records, Mikateclaimed that he was descended froRthe kings who "founded and populated the town." Connote 19 above.

48. The Kitab (fol. 17, Strong, p. 429) stops short ofrecording the full episode, but the details are knob from the Portuguese records.See Correa, Three Voycp. 298.49. Kitab, fols. 15-16 (Strong, p. 427). Reference to Fof Maghamghub as muCasir of Muhammad al-cAdil may su that the two came tothe throne at roughly the same However, it might also imply that Hasan(still livirthe time of writing) had predeceased al-'Adil, for tword usually designates the past.50. The ambiguity of this paragraph may result from a gaaffecting a sentence or two which might have been mc explicit concerning theearlier history. Strong mis "intiyajihi" which seems to representa coloquialisff"intihajihi." Kitab, fol. 2 (Strong, p. 405).51. Compare ibid. with Freeman-Grenville, East African Cp. 34.52. Where "the people of the history" are first mentionethe copyist added the plural form (reading "the peoF of the histories, ahi al-tawarikh") over the line iE manuscript. In the second instance, the textreadsdirectly ahl al-tawarikh, thereby leaving evidence c scribal error. Kitab, fols. 6, 10(Strong, pp. 411,18).53. The text here makes a point of indicating that al-AMuhammad sent the bribe to the Zanzibari amir in thname of al-Sultan Ismacil. See Kitab, fols. 10-11(Strong, p. 419).54. The details concerning the use of wood in the repairelate to the method of raising the domes rather t the actual materialused for thepillars. See IbnBVoyages, 2:p. 191. Cf. Chittick, Kilwa, l:passim 55. Freeman-Grenville'stranslation reads 'The latter[Sulaiman] gave the order for the rebuilding of th mosque . . .."There is nothingexplicit or impli which could signify "the latter." The phrase "MuhI b. Husain b.Sulaiman" was simply the man's full ni the system of nomenclature involved, sothat anytfollows refers to Muhammad rather than Sulaiman.Kitab, ff. 10-11, and Freeman-Grenville, East AfCoast, pp. 40-41.56. We have considered the possibility that the birtha1499 applies to the king rather than the author.iparin

KILWA DYNASTIC HISTORIOGRAPHYquite plausible, though, that the two authors shared acommon name, possibly even the name Muhammad, also sharedby the two kings. Kitab, fol. 16 (Strong, p. 428).

57. All the marginal comments begin with "gif CaZa" (Find outabout). Kitab, fols. 8-10 (Strong's publication omitsthem). These may have been added by the king to show hisdissatisfaction with the information in the text, butthe author himself showed his uncertainty by repeatingthe expression "God knows best."58. Freeman-Grenville' s translation reads: "I do not knowhis circumstances but will deduce them." Sa'astakhbiruCan halihi admits of no other reading than "I will enquireabout his circumstances." Cf. Kitab, fol. 9 (Strong,p. 416) with Freeman-Grenville, East African Coast, p. 39. 59. The BritishMuseum manuscript may have been copied fromthe author's rough draft. In that case, the original mayhave had a blank before the reference to the mother or aline or two in it may have been effaced. In any case, thecopyist would have reproduced only what was-.legible.60. The structure of the sentence is quite awkward. It mighthave read: "at the time of al-Sultan Sulaiman b.Muhammad al- CAdil there collapsed the mosque which had been built at thetime of Abu'l-Mawahib." Kitab, fols.9-10 (Strong, pp. 416-17).61. Following the defective opening paragraph andof the projected ten chapters, the Kitab includes a longprolegomenon which is quite coherent and detailed, though,as it deals with "the intellect and its arts," it haslittle bearing on the historical narrative. Among other possibilities, the author mayhave intended it as an invitation for his readers to apply inference inunderstanding his difficult text.62. Kitab, fol. 17 (Strong, pp. 429-430). One probable scribalerror has already been mentioned in note 52 above. In onecase, next to Manfasiyya (which evidently means Mafia),the copyist adds "that is, Mombasa." The Portuguesesources have Monfia while Yaqut (Kitab Mucjcon, 4: p. 969)has Manfiyya. A stretch preceding the Arabic 'y" ofManfiyya, for purposes of calligraphic style, apparently misledthe copyist to readManfasiyya, and in some casessimply Manfasa.207

3,4 4II" .~ 1~ 44it4 4,4 44 4 44

4¾4 4¾ r44~-444 4 44 ¾4 4444 4 4444~ 44 4444<4444444 444 44 4444 444~44~, 44 i4~44f4r4¾~ 444~ - 4 4444*4 4~,4¾ 444 4~~4 *~4~~t- 44: 44 4~'l4t ~-444 414 4~4~4 4~ 7444~44J4~ ~ 44:4444~44444444 414 4'4~&'44 444444~ 4444 44 444 44~ 44 444 44 ~4 444 444 444 ~4<44"44 44444444444 44 ~44 4 ~4~444 44 4 4 4444 444444444444 ~444? j44, 44 44 44~ 44 44444444 444~~~7~(1J ~ ~4. 4444 ~44 4:;~444 ~ 4 444444 44444q4~4444~l ~:4?4F 4444 44 444 44 4444444 4 444444444 44 44 44}44444 44441'~~ö 4 -~444 4444444 4144 44444444 44 4444 1 r 44444444 444 4 444444~44 ~444~44 444444 441444 144444 4 ~ ~4~t ~~444:~4' ~444444444 444~ 4 444 444 44 4 4142 ~44 444~44 4 44444444 4 4 4 44 44 444444444~ 44 444444444 ~4 4444144444 4444~444 4 4~14 444444444444444444444 44w 4 444444-~ 4 4444444~ 444444 ;44~4 44 ~44~ 444 ~ 4 444 4441r, 4 44 44 444444 44444 44444 444 ~44 4444444 ~44444 4- 44444 444444~444 4~44~4$4444444444 44------ 444 444 444444444444-4444 4 444 4444444444444444 ~~*4~44444444-4-~ 414444444 444~.44 ~

4,j444 44 ~4~44;4444, 44444444444444444444~4~4-4 444444~44 444 ~ 4444 4 44 4 44 4 24~$4~444:4fr ~ 44449 / 44 44444444 4414444144444 441'4 t~ 444444~44444~44444 4444 44:4444 4444 44~ 44444~ ~ .4444~44444444444 4 1 4444444 4444~44l444444~4444444 44~44~44444+4} 444444144444444444 4I~ 44 44 ~44444444,41444- 444-4444444 4- 444 444 ~44444 44~44~444~4-:4*44~44 4444 44 4444 4444444 4 444444444444441444424~4444 4444444444 4444~444~

NAF. iýkdt il d.'Afi "b Wnivemý y o gýa d 4_- N'wý 'Pbe rtin b sJD, Fage (Iýniversity of f*man (Ut&e ity of Wi consin, å Hair iversity Of Liv<(tin'D.R., Jmes (Unfversity of Lønd,ýn,ý, a* (university øf Stiffingyp ýÅýC Miu« (University øf Vir&iý ýy Of Løndrts (Universit of onM..-,Tw4ddle (University of Lon ýun ý)an tzig (University Of (Urivenity of WLou"1ýWo ry in Affica -is, pubbshed ann"lly by the Alo reptodu-ce matejrials from History in AM African Studies A~cigiort,Subscýiptions and ~tAd be ýsent tothc,,Affican Studies A~Ciatio»," vørýity,W.altham. Massaýhusetts 021,54, U.S.A. ma, respondence should bý sent toDavid ""enige, 8Wtý Wiýcon"', Madison, Wiscansiu,5370ý,ýý>l:ýfflc in His t6ýry In Africa represent ncither iý[tion nor- ýhow øfits officers, Thceditor Iýi, of the conteM of HiStory, 'in Africa and reser"g, deemedin4plýrópriate fi)i P4bliýation, R risý11 for the 'accuucy øf facts published re týmý7ýcditing of thiý:.a.e,