LAMU

CASE STUDY OF THE SWAHILI TOWN

Usam Isa Ghaidan

A thesis submitted in fulfillment for the degree of

M.A. Architecture in the University of Nairobi.

1974
This thesis is my original work and has not been presented
for a degree in any other University.

This thesis has been submitted for examination with our

approval as University supervisors.

Professor ,F
Head, '
Department
University of Nairobi

Director,
British Institute of Archaeology in Eastern Africa,
Nairobi
i
CONTENTS

List of illustrations ................. 5

Summary ................................. 7

Introduction - the historical background 10

Part One - The area

1-1 The Lamu archipelago and its hinterland 22

1-2 The t o w n ................................. 40

1- 3 The house ............................ 55*^

Part Two - General conclusions

2- 1 Swahili concepts of space ............ 70

2-2 Structure of the Swahili town .. .. 86

Notes and references ................. 98

Bibliography ............................ 106

Illustrations
ILLUSTRATIONS

Frontispiece Lamu town - main street at Mkomani

Figure 1 The East African coast
2 Creek north of Lamu
3 Spread of Islam before the fifteenth
century (source - Roolvink)
4 The Ishikani tomb
(courtesy - J.de V. Allen)
5 The Lamu archipelago
6 Lamu island - land use
7 Shelia Friday mosque - view from east
8 Sheila Friday mosque - plan
9 Shelia Friday mosque - section
10 Sheila Friday mosque - view from
ablution
11 Sheila Friday mosque - north facade
12 Sheila Friday mosque - east facade
13 The coastal hinterland of the
archipelago
14 Matondoni Friday mosque
15 Northern Swahili mosques - comparative
plans (source - Garlake)
16 Mosque in Sheila - detail of mlhrab
17 Lamu - the stone town
(photo - Survey of Kenya)
18 Lamu - view of promenade
19 Lamu - the mud and wattle town showing
rope-walk (photo - Survey of Kenya)
20 Lamu town - piazza
(photo - H.R. Hughes)
21 Lamu town - rope-walk
22 Lamu town - streetscape
23 Lamu town - mltaa
24 Lamu town - house in Mkomani, ground
floor plan
1 25 Lamu town - house in Mkomani, first
floor plan
26 Lamu town - house in Mkomani, plans
27 Lamu town - carved door
28 Sheila Friday mosque - carved door
29 Lamu town - door motifs
30 Carved door from Dubai
31 House in Mkomani - kiwanda
32 House in Mkomani - msana wa tinj
33 House in Mkomani - msana wa yuu
34 House in Mkomani - ngao showing mwandi
35 House in Mkomani - view towards ndani
(photo - H. Snoek)
 6

36 House in Mkomani - ndani

(photo - H. Snoek)
37 House in Shelia - ruined zldaka
38 House in Shelia - ruined zidaka
39 Mzab - interior
(source - Roche)
40 Mesopotamian miniature
(courtesy - Bibliotheque Nationals,
Paris)
41 Church in Alahan (Anatolia) - detail
of window (source - Pevsner)
42 Tundwa - ruined house
43 Swahili houses - comparative plans
(source - Garlake)
44 Lamu town - baraza
45 Lamu town - main street in Langoni
46 Street in Lamu
47 Lamu town - boat launching
48 Lamu town - main street
49 Lamu lamia - the Friday prayers
50 Lamu lamia - after the Friday prayers
51 Shelia Friday mosque - interior
52 Lamu town - land use
53 Kilwa - Husuni Kubwa (model)
(source - Garlake)
 7

SUMMARY

The Swahili coast between Kismayu in the north and

the Zambezi river in the south, and many of the

islands facing it have been the locations of

important settlements. Maritime trade brought

these settlements into contact with various regions

of the Indian Ocean, and exposed them to their

c u l t u r e G r a d u a l l y new styles of expression began

to emerge. These are evident in all aspects of

Swahili material culture: the art, the architecture

and, not least, the towns which are the subject of

this thesis.

The thesis analyses the structure and discusses

the types and function of these towns. It does

this by focusing on the in-shore island town of

Lamu on the northern coast of Kenya. This town

is a convenient milieu for such a 3tudy because it

preserves more of its original character than any

of! the other Swahili settlements. It is one of the

few existing Swahili towns that managed to survive

physical destruction by war-like tribes, commercial­

ization by tourism and the urban surgery of progress.

The Introduction is a brief historical background

Against this background the three chapters of Part

One focus on Larau town and its immediate environment.

 8

Chapter 1-1 describes the Larau archipelago and

the coastal stretch of Kenya north of the Tana

river. It discusses the function, economy and

patterns of architecture of the towns of the area.

Chapter 1-2 focuses on the town of Lamu. It

discusses its form and traces its historical

development until the beginning of the present

century.

In Chapter 1-3 the plan of the Lamu house is

analysed in relation to patterns of behaviour and

compared to Swahili house plana of earlier dates.

Part Two draws conclusions about the Swahili town

generally. Chapter 2-1 di3cusses the factors that

appear to have dictated Swahili planning iaenls;

Swahili concepts of space are discussed under three

headings: Pedigree, Involvement and Privacy.

The final chapter, 2-2, deals with the various types

of the Swahili town. It discusses the influences

thAt appear to have affected the form of these

towns, and includes an assessment of the impact of

Islam on the acceleration of Swahili urban growth.

 9

The thesis is the result of research work carried

out in the north Kenya coast, on and off, since

July 1969. Moat of the field work was completed

by August 1971. On two occasions the author was

accompanied by students from the department of

architecture in tha University of Nairobi when he

wa 3 lecturer there, the measurements for the

buildings shown on Figures 8, 24, 25 and 26 were

taken by t h e m .

The time since August 1971 was spent in recording

and assessing the information collected, supplemen­

ting it with library research, and writing un.

This part of the research was supervised by

Professor Flemming Jorgensen and Mr. Seville Chittick.

M r • Chittick is a orimary source on the history of

the Fast African coast; grateful acknowledgements

arc due to him for his invaluable help and ftuid nee.

The author also wishes to record his gratitude to

his father, hr. Isa Ramzi Uhnidan of Baghdad, Iraq,

I

for the time he spent in copying out and sending

long extracts from some of the sources quoted in

the Introduction.
 10

INTRODUCTION

The Historical background

The East African littoral (Fig.l) may be divided

into two geographic regions: the Somali coast to the

north and the Swahili coast to the south. The two

regions meet at Kiemayu, the former extends north

along the Somali border and the latter stretches

south to the Zambezi river. The Somali coast ir

arid anu lined with a nd dunes, and it is inhabited

by a majority of nomads. ^ T h e Swahili coast, on the

other hand, enjoys good rainfall, it is fringed by

coral reefs and lined for Biost of its length with

thick mangrove forests (rig.2) which provide timber

for building and export. The oorsl reefs, together

with the large off-shore islands of Pemba, Mafia and

Zanzibar provide good shelter from the open oceon.

The Swahili coa t has a number of deep inlets which

in ,so!ne c ses enclose small islands. Such islands,

because of their protected positions and deep

nchorages attracted many settlements. A number of

the towns, e.g. *vilwa, Mombasa and the towns of the

lamu archipelago are situated on such sites. The

population of this part of the const and the islands

facing it are Swahili speaking Muslims.

 11

The littoral haa occupied a fairly prominent place

in the trade of the Indian Ocean since the early

centuries of our era, when it was known to the

Greeks by the name of Azania. It is shown on the

twelfth century map of Al-Idrisi where it is

divided into five different regions (1).

The earliest account of the trade of the East

African coast is in the Perlplus of the Erithraean

o e a . which is a traders* handbook to the commerce

of the Indian Ocean. By the time it was written

(probably second century A.D.), coastal trade was

connected to the trade of the Gulf of Aden and was

therefore a part of the commercial system of the

Indian Ocean.

The coast had a number of ports to which ships

sailed from south-west Arabia and western India with

the north-east monsoon, bringing grain, oil, sugar,

ghee, cotton cloth and a number of manufactured

I
commodities in exchange for cinnamon, frankincense,

palm-oil, fragrant gums, tortoise shells, ivory and

other natural products, dome of these items may have

been re-exports.

The Periplus mentions a number of places by their

Greek names which are difficult to identify now.

 12

One such name, the Pyralaon Islands appears to

refer to the Lamu archipelago; but the most

important plaoe mentioned in the Perinlus. Rhapta,

where iron weapons were exchanged for ivory, remains

unknown. Here Arab traders lived among the inhabi­

tants and in some cases intermarried with them.

Local chiefs ruled under the overall authority of

the ruler of the south Arabian kingdom of Himyer.

The next extensive piece of information after the

Periplus dates to the tenth century, at which time

the region between the upper Nile and Sofala was

known by the name of Bilad-al-Zen.1. the country of

the Zenj. The trade of the area seems to have

moved from the Gulf of Aden to Oman, whose ship­

owners employed navigators from the port of Siraf in

south-west Iran. The upper part of the coast was

known as Bilad Jifuni and the most important town in

the region was Uanbalu or Qanbala on an island by

the same name. Bilad Jlfunl overlooked the Gulf of

Barbara which was a part of the Sea of Zen;), a very

rough and treacherous waterway.

Al-Mas'udi, (died A.D.956), to whom we owe this

information, travelled on this sea a number of times;

once from Sin Jar, the O'pital of Oman, to t^anbalu

and "the last time I sailed on ihis sea was in the

 13

year 304 A.H. (A.D. 916) from the island of Uanbalu

to Oman” (2).

In Qanbalu, says Al-Mas’u d i , Muslims lived amon^

the non believing Zeal, suggesting that the latter

were the majority; but in another work he says the

population of the town were Muslims (3). "The town

is famous for ivory which is exported by the

merchants to the Muslim countries ... the diet of

the people consists of maize, bananas, meat and

honey" (4). At the outer stretches of the country

of the Zen.1 was Bilad Waa Waa "where there is much

gold" (5). The 3ite of Qanbalu, like that of

Hhapta, has not been found.

Al-Mas'udi does not refer to the export of slaves

from the coast although Arab historians of his time

and before it confirm the presence of large numbers

of Zenj in Arabia. These were employed in various

agricultural pursuits, mainly land improvement,

such as reclamation from swamps by drainage,

scraping the salty subsoil layers and so on. These

became major industries during the middle and late

Abbasid period (ninth century and after), with the

transformation of Abbasid economy from smallholdings

to large scale agriculture. This transformation was

encouraged by the state through the institution of

 14

ikta' (6). Under this system large tracts of barren

land were granted and the grantees were entitled to

claim freehold rights over them after Improving them

within a specified period.

Contemporary Arab historians refer to large numbers

of black slaves employed in soraping "mountains of

salt" from the land adjoining Basrah and other towns.

Al-Tabari (died A.D. 922) speaks of gangs consisting

of over 500 Zen.1 each; one working in Ahwaz had

1,500 men (7). These organized a revolt which

lasted fourteen years (A.D. 869-883), and their army

was estimated at 300,000 men, some of whom were

recent arrivals (8).

The earliest existing indication of the origins of

the Zen.1 of Arabia is given by the ninth century

writer Al-Jahiz (died A.D. 869)» who lived in Basrah.

He lists Qanbalu, Mkier, Mithkir, Barbara and

Linjwiya as Zen.1 homelands (9). The Omani boats

described by Al-Mas'udi no doubt took large numbers

of slaves as cargo on their return voyages to Oman

where they were resold. The existence of a slave

market in Oman is mentioned in a mid-tenth century

Persian account which states that slaves sold there

fetched up to thirty dirhems per he*d (10).

 15

The earliest known Swahili site is that of the ninth

century town of Manda which was excavated by Neville

Chittick in 1966. Chittick's view is that the town

was the creation of colonizers from overseas. This,

if correct, confirms a reference by Al-Mas'udi to

Muslim immigrants to the coast in the eighth

century A.D.

During the twelfth century the focus of this coast­

wise immigration appears to have been Mogadishu,

which became a place of consequence during the

twelfth, thirteenth and fourteenth centuries,

largely due to the opening up of the gold trade with

Sofala. These immigrants are remembered by coastal

traditions as of Shirazi origin.

Towards the end of the thirteenth century, Kilwa,

under a new Dhirazl dynasty, which issued its own

coins, took over the control of the gold trade.

v
Its prosperity during the opening decades of the

fourteenth century is attested by an abundance of

stone buildings including domed and vaulted struc­

tures, and by its import of large quantities of

Chinese porcelain and glazed beads.

The status of Mogadishu and Kilwa as major urban

centres on the East African coast is oonfirmed by

the eye-witness account of Ibn-Battuta who visited

 16

both towns around A.D. 1332. lie describes Mogadishu

as "endless" In size, where two hundred camels were

slaughtered dally to provide for the population.

It was a busy entrepot of trade and arriving

merchants were accommodated in the homes of local

agents. He found Kilwa wealthy and its rulers

pious. It used to conduct periodic raids against

the tribal hinterland.

Ibn Battuta's description is supplemented by some

archaeological evidence which indicates thnt all

along the coast were townships: Mogadishu, Barawa,

Lamu, Pate, Malindi, Mombasa, Kilwa, Zanzibar and

others, mostly independent of each other and living

in more or less general harmony with the surrounding

tribes.

The fifteenth century has left us a large number of

mosques and tombs, an indication that the process

*
of Islamization which probably began two centuries

earlier was completed during this century (Pig.3).

There is archaeological evidence for many small

»
settlements between the towns indicating an increase

in population. The more important towns were in the

region of Mombasa and Malindi, and what is now Lamu

District. At the time of the arrival of the

Portuguese, at the end of the century, Mombasa, a

 17

place of little importance during Ibn-Battuta'a

visit, appears to have become the main port of call

on the Swahili coast.

The prime aim of Portugal in the circumnavigation of

Africa was commercial: i.e. to gain control of the

oriental sources of wealth in India and the Far East.

Their purpose in East Africa was the establishment

of a transit base for which they needed the cooper­

ation of the existing Swahili city states. They

proposed to ensure this by the imposition of

treaties. With the exception of Malindi and to a

lesser degree Zanzibar these treaties were not

honoured and a number of t owns were sacked by the

Portuguese as a result: Kilwa (1505), Mombasa (1505

and 1528), Zanzibar (1503 and 1509), Oja (1505) and

Barawa (1505).

In the southern Swahili region, the Portuguese tried

to penetrate beyond the coast in an attempt to

control and monopolize the interior sources of gold;

but these attempts only succeeded in upsetting

existing patterns of trade and decreased the amount

of gold traded.

During the sixteenth century the Swahili towns

remained generally independant of Portugal's control

and under the government of their traditional

 18

Shirazi ruling families. However, due to Portugal's

disruptive commercial policy, the sixteenth century

was a period of decline. During the second half of

the century a new oceanic power, Turkey, made two

brief appearances on the Swahili coast and succeeded

in inciting local revolts against the Portuguese.

After defeating the Turks, the Portuguese decided

to consolidate their power in Hast Africa by keeping

a permanent garrison in order to ensure continued

control on their dominions.

The construction of Port Jesus of Mombasa began in

1593 to the designs of JSao Batista Cairato, an

Italian architect in the employ of the Portuguese

in India. It is a heavily fortified building with

elaborate outworks, moats and salients to counter

the effectiveness and accuracy of the new projectiles.

The salients were so arranged that any one bastion

could come to the aid of the other by means of cross

fire. Its plan consists of a central court, with

1
bastions at the corners. Gunports and turrets were

placed to control entering ships and the main streets

of the town. It is an example of High Renaissance

fortification, which was to enhance the power of

Mombasa in later years. In the same year of the

construction of the fort, the Portuguese acknowledged

their ally, the king of Malindi as ruler of Mombasa.

Thereafter Mombasa was headquarters for Portuguese

garrisons on the coast.

During the seventeenth century Portugal succeeded in

asserting its ascendency over the larger stretch of

the coast. Portuguese garrisons occupied several

points -in the area and the Portuguese kept a customs

house in Pate. During the second half of this cen­

tury Portugal's position in the Indian Ocean was

deteriorating in the face of intense competition

from Dutch and English commercial interests. This

deterioration, coupled with the emergence of a new

Arab maritime power, the Ya'rubi dynasty of Oman,

encouraged the Swahili city states to revolt against

the Portuguese.

Swahili dissent was led by Pate town which, aided by

Oman, rose against the Portuguese five times during

the seventeenth century. Portugal's end came with

the capture of Fort Jesus by the Omani Arabs in

1669 after a siege of thirty months.

However, the expulsion of the Portuguese from the

East African coast did not mean an automatic entry

by Oman. Attempts by Oman to impose garrisons on

the coast even led to plotting with the Portuguese

 20

and a reinstatement of their power for a short

period (1728-29). Oman's political influence

remained prevalent on a limited soale through the

Mazrui, a clan of Omani Arabs who established them­

selves as hereditary rulers of Mombasa. Under their

rule Mombasa dominated most of the towns of the

northern coast until the end of the eighteenth

century. By 1746 with the overthrow of the Ya'rubi

dynasty in Oman, the Mazruis declared their indepen­

dence from Oman. During the following decades

Mombasa grew in strength and signed a treaty of

alliance with its old time rival town, Pate. The

alliance came to a breaking point in 1812 after the

joint forces of Mombasa and Pate lost a decisive

battle against Lamu town. Lamu, fearful of similar

acts of agression appealed to Oman for protection.

This gave Oman's new leader, Sayyid Ja'id, an

opportunity for direct intervention and a base

which he later consolidated by terminating Mazrui

rule in Pate and then in Mombasa. In 1832 Sayyid

8a'id transferred his capital from Muscat to

Zanzibar.

The establishment of the Zanzibari sultanate in 1840

marks the beginning of a new era in coastal history.

Before the nineteenth century the Bast African coast

 21

vraa a part of the continent of Africa in a geo­

graphical sense only. During the decades following

Omani presence Swahili caravan trade was bringing

the coast and interior into continuous contact.

The influence of the trade routes which penetrated

over a thousand miles inland in places is attested

by the fact that Ki-Swahili is lingua franca in the

eastern regions of the Zair, and by the appointment

of Tipu Tip, one of Sayyid Barghash's men, viceroy

over Belgium's central African possessions in the

eighties of the last century(11).

The slave trade, first under French control, later

\

under the Arabs reached its highest peak during the

middle of the nineteenth century to provide cheap

labour for the new plantations societies in Zanzibar

Pemba, and elsewhere, and for export.

Zanzibar's economy was growing through the encourage

ment of foreign, mainly Indian investors and the

town was on its way to becoming an international

port.
 22

PART ONE: THIS AREA

1-1. The Lamu Archipelago and Its Hinterland

Lamu is the name of one of the districts of Kenya's

Coast Province; the name is also given to an archi­

pelago, one of its islands and to the largest town

on that island. The inhabitants of the town call it

Amu and from this the name of one of the three major

Ki-Swahili dialects, Ki-Amu, is derived. The more

popular version of the name, Lamu, may be a corrup­

tion of Al-Amu; the prefix being the definite article

in Arabic.

THE ARCHIPELAGO

The entrance to Kenya in the immediate vicinity of

Port Lurnford is marked by two tombs with high

pillars whose site may-have once been a pre-Islamic

market, possibly one of the emporia mentioned in the

Periplus. It is also believed to be the legendary

Shvpigwaya of African tradition, claimed by Kitab-ul-

Zonuj to have been the dispersal point for a large

number of African tribes during the 12th or 13th

centuries. Six miles to the south of Port Lurnford

are the remains of the walled town of Ishikani, with

a similar tomb. South of Ishikani is a large

striking rectangular panelled tomb (Pig.4) over

 23

1.2 m high covering an area of about eighty square

metres. Three of its walls are decorated with

asymmetrical, apparently abstract motifs in low

relief. The designs do not appear to bo Islamic.

No doubt the motifs were meant to symbolize some­

thing; magnificence perhaps, or even immortality;

but to the spectator of today they are only objects

d 'e r t . Ten miles further south are the ruin3 of a

mosque belonging to the site of a large settlement

on the island of Kiunga opposite, where there is

another tomb with a pillar in a bad 3tate of repair.

Kiunga stands at the head of a long chain of islands

running parallel to the coast and making a sheltered

navigable channel for about seventy miles. The

majority of the islands are uninhabited; but there

are a number of settlements on the mainland along

the coast. These islands, and the coastal strip

facing them, represent the farthest northern limit

of the Swahili cultural unit.

j f

At akokoni the coastline turns slightly to the west,

in a manner suggesting that a fault in the coral

ridge may have caused a partial collapse; the com­

bination of sea water and the shallow bed of soil

create excellent conditions for the growth of thick

mangrove forests. Here the more important islands

 24

of the arohipelago: Faza, Manda and Lamu, are

situated (Fig.5).

The largest of these is Faza Island, whioh has three

townships and a number of smaller settlements. The

most important of them, Pate town, is situated to

the south of the island, protected from the open sea

by the small uninhabited island of <isingati. Pate

was a city state of importance during the seventeenth

and eighteenth centuries. The town's own history,

the Pate Chronicle, claims that it was a place of

consequence as early as the fourteenth century when

the Nabahanis are supposed to have established their

sultanate there (12). During this time the town is

/
supposed to have commanded the large stretch of the

coast between the iienadir in the north and Songo

Mnara in the south. An examination by H.N.Chittick

of the chronicle in the light of archaeological and

external historical evidence however, shows that

Pate was of little importance before the sixteenth

f
century (13). In this study the establishment of

the Nabahani dynasty at Pate is dated to the

seventeenth century. A poem written in Pate in

1652 suggests that by this time the town was a

centre of literary activity (14).

Pate's relation with the Portuguese was one of

 25

perpetual defiance and submission. Among the

Swahili city states it was the only one to dare face

them in bloody street fightings finally forcing them

to fle8 in 1679 (15). However, because of later

disagreements with the Omani Arabs, Pate played a

part in helping to reinstate the Portuguese in 1728.

The town entered a phase of rapid decline after losing

a bloody battle in Lamu in 1812. Al-Inkishafi, a

poem written in Pate during the same decade, mourns

its days of gre' tness like this:-

"The lighted mansions are uninhabited,

The young bats cling up above,
You hear no whisperings nor shoutings,
Spiders orawl over the beds.

"The wall niohes for porcelain in the houses,

Are now the resting-place for nestlings,
Owls hoot within the house,
Mannikin birds and ducks dwell within." (16)

The Nabahanis were finally deposed and their last

leader, sultan Ahmed bin Pamoluti, fled to the

mainland about 1840 A.D. He established himself in

W i t u a n d later received German protection against

Zanzibar. Prom Witu he organized regular raids on

the mainland plantations of the Archipelago. He

died in 1888.

Pate is now occupied by a population of about 700

 26

who earn the main share of their living from tobacco

plantations. It has five mosques, eight shops and

one Koran School.

The next town, Faza, also called Rasini, is to the

north of the island, and has a slightly bettor

harbour. It is now an administrative centre - with

a population of 1500. During the Portuguese inter­

lude it kept friendly relatione with them and in the

middle of the seventeenth century actually helped -

them against Pate. During the Omani Arabs* seizure

of Port Jesus, the defence of the fort was for a

(
period led by a member of Paza's royal house.

The last town, Siyu, lies just south of Paza town.

It has the remains of a fort containing a ruined

mid-nineteenth century mosque. The Priday mosque

has an elegant mlnbar or lectern dated the equivalent

of 1521 A,D. Siyu had a substantial Somali minority

represented in thetown's government. It is famous

foij' skills in furniture-making and leather-work.

Its prosperity continued longer than that of Pate.

A channel about five kilometres wide separates Faza

island from that of itenda. This island, now deserted,

wa3 once the site of three towns including that of

the oldest known Swahili settlement, Manda town, on

the north eastern tip of the island overlooking a

 27

shallow oreek covered by thick mangrove. Recent

excavations have revealed a comparatively prosperous

ninth century level which lasted until the thirteenth

century. Wanda's excavator, Neville Chittick,

uncovered tenth century houses built of square

coral blocks in rough courses with mud and lime

mortar. There is evidence of ninth and tenth

century trade with Iran. Portions of the seaward

wall of the town, built of large coral blocks

weighing up to one ton, have survived.

On a low hill to the south of the island are the

ruin3 of Takwa, which, according to J.3. Kirkman,

belong to the sixteenth and seventeenth centuries

(17). Takwa has the remains of a large mosque;

above its mihrab is a pillar, as at the small domed

mosque of Kilwa. There are remains of another town

east of Takwa. The Pate Chronicle claims that one

7

of these towns succumbed to the power of Pate and

some of the inhabitants escaped across the creek to

1
Lamu island. The story is continued in the Lamu

chronicle which says that although the people of

lamu were prepared to give them shelter they did

Q Q t , however, permit them to build their houses of

stone.

The island of Lamu (Pig 6) is the most important in

 28

the archipelago; it has an excellent natural harbour

and is fringed along the west, north and north-east

by mangrove forests. No crops can be cultivated on

its 3andy soil, but there are shamba or cultivated

grounds west of the town where mangoes and coconuts

are grown. To the south of the island is a sandy

bench rising to a height of about twenty metres at

the estuary of Lamu Bay. The sand dunes are formed

by the north-east monsoon which blows between November

and February. Here is the village of 3hella, now

occupied by a population of only two to three hun­

dred. The site is probably over five hundred years

old (18) and reached its zenith in the middle of the

last century. It has an interesting Friday mosque

2
which covers an area of 290 m and has a conical

minaret approached by a spiral staircase of fifty

eight steps, (figs.7-12). Mosques like this, located

close to the sea on headlands, are a characteristic

feature of the Fast African coast (19). The present

lmf^m of the mosque remembers its musalla (prayer

hall) being filled by the congregation during the

second decade of this century, suggesting that the

town was much larger then Stigand's reference

to a qadhi in Bhella (20) at about the same period

confirms this. The township now has the largest

number of fishing vessels in the archipelago (fig.5)..

 29

To the west of Sheila the sand ia interrupted by

thicket8 and swamps. On the other side is the small

village of Kipungani, from which a roadstead runs

parallel to the shore to the fishing village of

Matondoni and oontinues in a southerly direction

towards Gamu town about two miles north of the open

sea.

The population of the islands and the mainland

strip opposite ia a mixture of many groups! Bajunis,

coastal Bantus, Arabs, Somalis and Indians. The

Bajunis, also called Watikuu, are the biggest group

and in the northern part of the region they are

virtually the only inhabitants. There is a

tradition which traces their dispersal centre to

Shungwaya along with a number of other Kenya African

tribes. In the hinterland are a few hunting peoples

who 3peak a Cushitic language.

The economy of the archipelago depends on dhow

tr-jde, the sale of mangrove poles, fishing and

cattle. A recent count showed that over 400 ships

visit its harbours annually, some from as far as

Kuwait, Iran and Iraq (21).

TUB CGABTAL HINTERLAND

Because of the sandy soil of the majority of the

 30

islands only limited cultivation is possiblei

tobacco plantations in Patef and coconuts, mangoes

and bananas in Lamu. For essential ground crops

suoh as millet, simsim, Indian corn and rice the

inhabitants of the island towns used to hold large

crop plantations (makonde. singular konde) on the

fertile strip of the mainland opposite the archi­

pelago, stretching from Bur Gao in the north to Ras

Tenewi in the south (Fig.13). Mainland cultivation

followed a system of land rotation. Agricultural

workers consisting of individual free men and

groups of slaves belonging to rich plantation owners

would clear an area by burning, each man would then

take a piece about a hectare and a half in area,

measuring 100 paces in width by 200 paces in length;

the measurement was determined by "the practice of

having a slave cultivate one ungwe per day, a narrow

strip of 5 paces by 200 paces" (22); the whole piece

would thus be cultivated by one slave in twenty days.

The plot8 would be planted for one or two seasons

before being abandoned in favour of others in the

same area, the deserted plots would be allowed to

regenerate over a period of ten to twelve years

before being cleared again by the same process of

burning. Many plantation villages, consisting of

mud houses, market banda. wells and mosques were

 31

founded. Due to the Impermanent nature of their

materials, most of these villages have disappeared;

but traoes of two of them can still be seen: Hindi

within the plantations of Lamu town and Mgini within

those of Pate. The sites are now deserted and over­

grown, but in Hindi three wells and the remains of

a mosque have survived. According to local infor­

mants it was the site of about two hundred and fifty

houses.

Plantation slaves (w-tumwa wa shamba) were either

granted their own patches of land where they were

allowed to work for two days in the week in return

for their labour during the remainder of the week

on the masters' plantations, or they worked on their

masters' land on their own account and paid him rent

(i.lara) (23). They were supervised by a manager

(nokoa) chosen by the owner from among the slaves.

Existing evidence shows that the fifties, sixties

and /seventies of the nineteenth century were years

of intense agricultural activity. In 1859 over

eight million pounds of simsim worth £20,000 was

exported from the area to Zanzibar for re-export (24).

In 1874 a European eye-witness found the area south

of Mokowe covered with huge plantations. "The rice

paddies produced enough rice for about seven thousand

 32

people." (25). This prosperity extended south as

far as Malindl which, during that period, held

plantations extending up to about twelve miles

inland and "only the fear of the Galla and Wasanye

prevented cultivation from extending to Tak-ungu" (26),

thir'y five miles to the south.

The next decade w a s one of deoay. Two eye-witnesses,

one in 1885 (27), the other in 1892 (28) found most

of the area derelict, overgrown and neglected. Its

deterioration is a reflection of the general state

of decline of the northern Swahili coast discussed

In the next chapter.

With the development of makonde plantations, country

and town were brought into a natural symbiotic

relationship; the former providing the food surplus

required to replenish the urban larder and at the

same time relying on the centralized power of the

latter to keep its system running. No doubt the

output of these plantations played a part in regu­

lating population densities in the towns.

Trimlngham's description of dwahili towns as

"parasitic in that they had no organic relation

with the region in which they were precariously

situated" (29) cannot therefore be accepted.

The area surveyed briefly above is known by the

 33

particular name of Jwahlllal to distinguish it from

the rest of the Swahili world. The term Swahillni

is generally understood to refer to the part of the

ooast between the Tana river mouth in the south and

Kismayu in the north. "There is a consensus of

opinion among most important authorities that this

part of the coast, i.e. the Lamu archipelago and

the mainland ju3t to the north and south of it, is

the original homeland of the Swahili cultural entity,

and that here also the language Kiswahili came into

being" ( 30 ) .

THE ARCHITECTURE

The most important building material on the East

African coast is coral of shich two varieties were

used: soft reef coral for jambs, lintels, mihrabs

and similar oarved elements; and hard terrestrial

coral for foundations, walls and other parts of the

structure, ^oral was also slaked to provide lime

for'mortax and plaster, ^ n g r o v e poles were used

as 'structural timber; either in round sections

(boriti) or in square, dressed sections (banaa ) , for

floor joists and roof rafters. The roofs were

covered with layers of coral and lime plaster to

thicknesses of 30 cm to 50 cm. Few of the existing

roofs so constructed are older than one hundred and

 34

fifty years. Roofs constructed in this manner did

not last very long because of wet rot in the

supporting wood. Water leaking through roofs

accumulates and penetrates by capillary action into

the end grain; excessive checking occurs when the

swollen wood dries, thus facilitating further

wetting. Nevertheless, this form of roofing is

widely used on the coast and almost universally in

domestic architecture. Some mosques depart from

this mode of construction, and are roofed with stone

vaults (Rig.14)■ A notable example is the small

mosque of rtvana north of the Tana delta which

possibly dates to the fourteenth century. It is

roofed by ten conical domes (supported by octagonal

cornices on square or rectangular bays) and two

semi-circular barrel vaults. There are examples

where the two common building materials, coral and

mangrove poles, were used in conjunction. In the

9th century site of Manda there is evidence of

mangrove poles used as horizontal wall reinforcement.

The late fifteenth oentury new Friday mosque of

Ungwnna on the Tana delta shows a similar technique.

The feature also occurs in a number of buildings in

Kilva. Wall thicknesses vary from 44 to 56 cm; which

appears to be the standard cubit (dhlraa) measure as

it has been found that dimensions of measured buildings

 35

are multiples of it (31). Frames to doors and

windows are commonly of dressed mangrove; but two

local varieties of hardwood are also used for

carved frames, centre posts and lintels.

iiuilding types in the area may be grouped under three

main headings: mosques, tombs and houses (32).

The musalla or prayer hall in most mosques is a

single rectangular space, divided into two, three

or four spans by square or rect-mgular stone piers

(fig.15). The size of the muoalla was determined

by the function of the mosque (Friday mosques being

larger than others) and the wealth of the town at

the time of the construction of the mosque. The

largest existing pre-nineteenth century mosque in

the archipelago is the seventeenth century Friday

2
mosque of toanda, which measures over 140 m . At

the short end of the rectangular musalla is the

prayer niche (mihrab). orientated towards the qibla

in Mecca. Moat mihrabs are round in plan, normally

arched and capped by a semi-dome and framed by a

decorated frieze which is sometimes rebated (Fig.16).

The view of the mlhrab in two or four span mosques

is obscured by the central row of columns. Almost

all surviving mosques in Pate are ao planned; so is

the small mosque in Siyu fort, built in the mid-

 36

nineteenth century, the large seventeenth century

^ancia Friday mosque and the small sixteenth century

mosque of Ungwana. The sixteenth century musalla

of Ta^wa mosque is flanked by two side aisles that

open on to the prayer hall through arched openings.

The two Ungwana mosques have one side room each

along the eastern wall of the musalla.

The ablution cistern is normally built along one

of the long sides of the mosque. Supply and

disposal of water is through stone conduits. Muslim

prayer is preceded by ablutionj the position of the

ablution cistern therefore determines that of the

entrance doors. Most mosques are approached from

the long side, perpendicular to the direction of

the prayer. This is the case in Taicwa, the domed

mosque of Mwana and the small mosque of Ungwana.

The new sixteenth century Friday mosque of Ungwana

is similar, but has two exit doorways on either side

of the mihrab (33). The musalla in the Friday

mosque of Manda and the mosque of Shala Fatani in

Faza town are approached from the south where the

ablution cistern is situated; but each mosque has

an exit door in one of the long walls.

None of the mosques has a sahn. or colonnaded

courtyard which is a common feature of mosque

 37

plana In Arab and Persian Muslim towns. There is

also no evidence to show that a maaaura screen (34)

was ever used. Minarets are rare too. In the area

under discussion only one exists, that of the mid -

nineteenth century Friday mosque of iihalla. Absence

of minarets has been attributed to possible influence

from Ibadhis (35), a purist Muslim sect who fled

from Iraq during the tenth century and settled,

among other places, in the Mzab area south of the

Sahara in Algeria where they founded a number of

towns. Until the arrival of the Shafii sharifs to

the Eaot African coast in large numbers, the major

Muslim sect was Ibadhi.

The erection of a pillar on top of mosque roofs is

a local invention. The Takwa mosque on Manda island

has a stone column about 2.5 m high at its northern

end. The feature appears once more in the fifteenth

century domed mosque of Kilwa where the stone pillar

is (fluted. Because of the association between pillars

and tombs on the East African coast, Kirkman, who

excavated Takwa mosque, has suggested that the feature

indicates a funeral mosque.

The tombs of the area provide a very rich variety

in form. Borne are rectangular, surrounded by low

boundary walls which are sometimes buttressed. In

 >8

some oases the walla step up at the c o m e r s then

rise sharply in a sweeping curve in tho form of

horns. 3uch tofcbs are ofton capped by pillars.

Thoro Is a tomb at 3iyu which probably aignifies a

later typo. It is a square structure carrying a

conical done with diagonal groins. The walls are

divided into rectangular panels framing arched

niches. The panels must have had porcelain bowls

set in them, for a number have still survived.

One feature of the Swahili tomb, the pillar, has

been the focus of some debate on its possible

phallic representation; it has been suggested that

it may derive from similar structures in southern

Ethiopia. In fact there is no reason to 3uppose

that its origin is any less Islamic than other

patterns of Swahili architecture. The custom of

erecting pillar like structures on tombs became

popular in the late Abbasid period. Two such

examples, namely the tomb of Zamarrad Khatun and

the tomb of Al-Sahrawardi (bdth in Baghdad) have

high conical towers in the form of pillars over

polygonal brick enclosures.

The largest number and probably best examples of the

third building type, houses, are preserved in Lamu

town, and are the subject of Chapters 1-3.

In viewing Swahili architecture one notes a striking
>
lack of precision in technique and an absence of

standardisation and geometric discipline. Even the

Husuni Kubwa in Kilwa, called by Gtarlake "the

fountainhead of all pre-eighteenth oentury (coastal)

architecture" (36) suffers from what would now be

regarded as unacceptable discrepancies (37). It

appears that precision as a criterion per se was not

considered by the Swahili master-builder as an

important quality in finished objects. It is however

not a quality that comes naturally to man. Lewis

Mumford has noted that precision and standardization

appeared at an early date in the formations, exer­

cises, and tactics of the army. The mechanization

of men is a first step toward the mechanization of

things" (38). Garlake highlights discrepancies in

Swahili bulldinge by comparing Last African coastal

architecture with that of Abbasid Islam. This is an

incorrect comparison as the Abbasid dynasty had an

established army and one of its caliphs, Harun

al-Rashid, once made a present of a clock, that great

symbol of precision, to Charlemagne (39). Prins's

suggestion that imprecision is a typical feature of

maritime cultures (40) cannot be accepted either as

it is precisely on the qualities of control and pre­

cision that maritime endeavour relies to distinguish

itself in the competitive world of trade.

 40

1-2. The Town

SPATIAL ORGANIZATION

Lamu town ia an administrative centre and the seat

of the coast's largest religious academy. It is

also a place of pilgrimage; its annual celebrations

of the Prophet's birthday (maulldi) attract large

numbers of visitors from all parts of the Muslim

coast. The town stretches between the sea to the

east and a low range of hills to the west for a

distance of about a kilometre and a third. Its

maximum width occurs roughly at the middle of its

ong axis where it measures a bout three hundred

metres. To the north and south of this point the

landward edge sweeps gently towards the sea giving

the town the shape of a segment of a circle.

It haB a population of about 6.000, of mixed origins,

all Muslims mostly dunnis of the dhafi'i sect. Their

livelihood depends largely on maritime endeavour

such as shipping, mangrove poles and fishing,

supplemented by fruits from the island's ahamba•

These lie to the west of the town; in some of them

summer houses and other amenities are provided. The

town produces the country's best varieties of mangoes

and coconuts. The latter provide raw material for a

few home industries such as rope making, mat weaving

 41

and oil extraction, Lamu district is also Kenya's

trading centre for Somali cattle.

Lamu has a good natural harbour, protected by the

island of Manda from the open ocean, which is here

due south-east. The promenade and harbour wall were

rebuilt after the first world war. At low tide the

water retreats to a distance of about twenty metres

and at high tide it covers the jetty and sometimes

spills on to the promenade. There are two jetties;

the main southern one faces the customs house, the

other is for the use of the local administration.

Lamu consists of three parts (Fig.52): the old

town west of the main street, where the Swahili

stone houses are to be found (Fig.17), the

nineteenth and early twentieth century Indian

additions along the promenade (Fig.18), and the

impermanent mud and wattle section mainly to the

south where the poorer families live (Fig.19). The

first and second parts meet at the main street which

runs north-south and accommodates over a hundred and

fifty shops. Opposite the main jetty, south of the

stone town, is a large piazza bound by the early

nineteenth century fort on the west and the lawn's

market on the south (Fig.20). Immediately north of

the fort is Pwani mosque which claims the old foun-

 42

datlon date of the equivalent of A.D. 1370. The

name Pwani (Swahili * coast) ia evidence that the

town'8 edge used to run fifty metres west of its

present position. When the fort was built it faced

the sea and its bastions oovered the harbour. The

mihrab of the .lamia or Friday mosque, in the northern

• part of the town, incorporates an inscription reading

the equivalent of A.D. 1511 which may belong to the

mihrab of the older Friday mosque on the site of

which the present one stands. In the middle of the

town is a medieval tomb; and on the south-west, at

the edge of the town, is a tomb with a fluted pillar.

None of the other existing buildings is likely to be

of very great age.

A number of the town's maritime industries, such as

boat building, sail mending and rope making, take

place in the open; the first two along the sea, the

last within the mud and wattle section to the south

(Fige.19 & 21). The mosque college of Lamu stands

in the middle of a large open space south of the

fort. The surrounding space is filled with dancers

during maulidi celebrations. Maulidl is a popular

religious festival held annually to oelebrate the

birthday of the Prophet; the custom is over a

hundred years old. The college was established in

1900; its founder Al-Habib .Jalih was a Comoran Arab

 43

of Hadhrami stock who arrived in Lamu at the end of

the 19th century and lived there until hie death in

1933. The academy runs courses lasting two to five

years and has students from countries as far apart

as the Congo and Madagascar. Al-Habib Ualih is

also responsible for the introduction of a new and

controversial cult, that of music and song in

mosques. The mosque he founded, Riyadha, which is

also the seat of the academy, holds song sessions

three times a week during which Arabic verse in

praise of the Prophet is sung to the music of

tambourines.

The town slopes down towards the sea; all lanes and ^

drainage channels follow this natural gradient,

taking waste and surface water down to the Ocean.

Night soil discharges into subterranean soakage pits

through large stone ducts. The main axis of the

town stretches at right angles to the direction of(

the kaskazl or north-east monsoon which blows in

as a pleasant breeze and provides natural thermal

comfort during the humid months of March to October.

The townscape is informal and intimate in scale

(Pig.22). As there is no wheeled traffic, streets

are entirely used for pedestrian communi ations.

They are lined with hard, richly textured surfaces \

producing Tarious densities of shadow. Facades are

often windowless, interrupted only by house entranoe

porches which are tunnelled through the coral walls

and flanked by stone seats. Some of the houses have

annexes spanning the street, oarried over mangrove

pole beams. It is not clear whether the town was

walled or not. One of the neighbourhoods south of

the fort is known by the name of Langoni (Sw = at

the door). The local inhabitants divine from the

name a reference to a gate in the town wall. It ia

difficult however to see the need for a wall in a

town which enjoyed such excellent protection from

the ocean. Pate town, sharing an island site with

two rival towns, was surrounded by a wall to which

there is a Portuguese reference in 1637 (41).

lamu is divided into a large number of small wards

(mltaa: singular mt a a ). each being a group of

buildings where a number of closely related lineages

*
live (Pig.23). Mitaa vary in size and character;

but the inhabitants of each mtaa enjoy the same

social status, and are often related by blood or v^.

united through clan or common ancestry. Most mltaa

have their own mosques and jointly shoulder the [

V

responsibility for their maintenance. The mitaa

are not always defined by roads, open spaoes or

similar urban edges and cannot therefore be easily

 45

Identified on a map. It is not uncommon for houses

in one block to belong to different mitaa: the

relation of houses to mitaa being more or less deter

mined by the position of the entrance door. Mitaa

names are in some c see derived from the urban

functions performed in them, e.g. madutcani or "at

the shops"; in others from distinctive features,

O.g. chfr,foyt o l or "roadside veil"; other

names refer to places of origin like landar Abbas

and Maskati referring to place names in Iran and

Oman respectively.

•
Until recently the affairs of each mtaa used to be

regulated by a local council of elders, wazee wa

mtaa. Councils of related clans were represented

on the town's government. Of such affiliated clans

Mombasa and Faza had three each, Siyu had two and so

on. Lamu town also had two: Suudl and Zelna who

had permanent representative councillors. The post

of i!amu President rotated between the two groups

every five years (42). The symbol of government was

a ceremonial horn or siwa of cast bronze blown on

important occasions and kept by descendants of an

old clan. The free men of the town could borrow the

siwa for blowing in social family functions for a

token consideration.
 46

HISTORY

The earliest known historical reference to the town

is preserved in an Arabic manuscript (43) which

describes a meeting between the fifteenth century

Arab historian Al-Maqrizi and the qadhi or judge of

Lamu in Mecca in the year 1441 A.D. The town is

described as a city state accruing its natural

wealth from the sea and from fruit orchards, and its

site is described as being engulfed by sand. The

qadhi impressed the Arab historian by his scholar­

ship. The office of a oadhi i3 normally a require­

ment of a large town, and his scholarship may be

regarded an indication of an evolved culture.

The town's own history, the Lamu Chronicle,

describes two independent townships, Hidabu and

Weyuni, to the north and south of the present town

respectively, as the forerunners of the present

town. After a period of wars, the two sections are

supposed to have jointly chosen the present town

site which has the advantage of a deep sea channel.

Hidabu was the subject of archaeological investi­

gations by Neville Chittick in 1966. His conclusions

are that its occupation "extends back at least to

the thirteenth century and ceased in the fifteenth

(44). The town described by Al-Maqrizi must have

been the original settlement on Hidabu Hill. The

 47

present site is therefore not likely to be older

than the fifteenth century and the date incorporated

in the mihrab of Pwani mosque is almost certainly a

later addition.

The town is mentioned by the Portuguese in 1506 when

TristSo da Cunha blockaded it and imposed a tribute

which was paid without resistance. In 1585 the

Turkish captain Mir Ali Beque visited the town and

took an ex-Portuguese captain prisoner. The town was

later punished twice for this, onoe in 1589 and again

in 1678; in each case the town's ruler was executed

by the Portuguese in Pate.

The seventeenth century was the period of Pate's

supremacy; during this time Lamu was a subsidiary of

Pate. The excellent siting of the towns of the a r c h i ­

pelago protected them against attack from the main­

land warlike tribes which during this century almost

destroyed many Swahili mainland towns together with

1
the island towns of Kilwa and Mombasa. The eighteenth

century witnessed renewed activity in stone buildings,

the Palace of Pate was built and many ruined mosques

were rebuilt. In 1812 the joint forces of Pate and

Mombasa were repulsed by Lamu in a fierce battle on

the beaches of Sheila. In the same year Lamu sought

and received protection from the sultan of Oman; the

 48

Lamu fort wa3 completed and garrisoned by Omani

soldiers. Lamu, thanks to its special relationship

with the Omani rulers who later established the

sultanate of Zanzibar, grew into a busy trade

entrepot. By the middle of the century its dhows

were trading in ivory, mangroves, oil seeds, hides,

grains, cowries, tortoise shells and hippopotamus

teeth in large quantities. Ivory was bought from

the Wasania hunters through the intermediary of

Kipini, Kau and the other settlements of the Tana

river. Kirk, writing at the end of 1873, talks of

oanoes carrying groups of fifty men leaving in

December and January, when the Tana was at its

height. "This riverine caravan trade was in time

extended further inland to reach Mount Kenya" (45).

The commodities listed above probably constituted

the traditional exports of the town. Its rise

during the nineteenth century was however due to

the town's participation in the Arab controlled

East African slave trade whioh grew in volume towards

the beginning of the nineteenth century with the

creation of a plantation economy on Zanzibar and

Pemba islands and the establishment of an Omani Arab

commercial empire in the north-western Indian Ocean.

The establishment of an organized state in Zanzibar

 49

was responsible for a great increase in the number

•\
of slaves being exported to Arabia. When the /

3ultan of Zanzibar, Barghash bin S a ’id, threatened

by a British naval blockade, was forced to sign the

Anglo-Zanzibari Treaty of 1873 banning the slave

trade, this trade was actually at its height. The

slave traders naturally began to explore new methods

of operation. Lamu was the northernmost point on

the coast along which the coastwise traffic in slaves

was allowed by the 1845 treaty. It was excluded

from the limitations Imposed by the agreement of

1873, which forbade export of slaves within the

domain of the Buitan of Zanzibar and abolished

existing slave markets. As a result large numbers

of slaves found their way north and north-east from

i^amu's harbour every season. Slaves smuggled from

Kilwa and Zanzibar normally stopped in Lamu before

sailing to the Benadir and Arabian ports in Lamu

dhows. In 1861 General Rigby saw 600 slaves awaiting

shipment from Lamu, having arrived from southern

ports, and in 1871 Kirk reported that 1901 slaves

reached Lamu during that year, compared with only

53 to Mombasa and 39 to Pemba (46). Many of these

were taken from Lamu district overland to Somalia.

In 1869 the Sultan of Zanzibar garrisoned the town

of Kismayu where a new settlement had been founded.

 50

"This new settlement and the subsequent expulsion

of the Galla from the neighbourhood opened up a

fairly safe land route for the slave traffic between

Lamu and the Benadir ports, thus rendering the sea-

watch by British cruisers less effective" (47).

Most slaves destined for export from Lamu were

brought from Kllwa and Lake Nyasa areas on foot.

Only a few of the slaves traded In Lamu and the

other Island towns were from the tribes of the near

mainland, either because they were considered

unsuitable or because they were not available.

During the third quarter of the century new waves

of migrants from various regions of the Indian Ocean

were arriving to Lamu, attracted by the town's

economic prosperity. Lamu's population, Judging by

the number of ruined houses, was probably twice

that of today. The Hindu community alone could

support an independent primary school. The arrival

of Asian traders to Lamu was part of a large

I

migratory wave prompted by Sayyid Sa'id's policy of

importing commercial skills and capital required for

the expansion of his own enterprises.

The prosperity did not last long. Masai raids and

the famine of 1884 were causing plantation villages

to be abandoned. The situation was worsened by the

cattle plague of 1889 (48) and the increased res­

trictions on slave labour. By this time the East

African ooast was beginning to attract European

interests. In the north the Germans were inciting

the ruler of Vitu to lay claim to the large coastal

stretch between Kipini and Kiwayu. He also insti­

tuted taxes on the produce of the islands'

plantations. Thus the traditional laissez-faire

understanding no longer operated and agricultural

activity dwindled. In the south, Sultan Barghash of

Zanzibar, unable to stem the tide of German imperial­

ism, conceded to a British Company, the East African

Association, the right to administer the ooast between

Vanga and Kipini in his name.

The combined result of these factors was the begin­

ning of the disintegration of the Swahili cultural

unit. During the following years the coast became a

stage for European rivalries, in 1895 it was formally

declared a protectorate of Britain. The twentieth

century brought new and superior technological media.

Colonial interests were directed more and more

towards the interior of East Africa. The slave

plantations and monsoon-based trade were no longer

sufficient to ensure a continued lease of life for

the towns of the archipelago and they, as a result,

entered a phase of rapid decline.

 52

After 1813, when Lamu became a protectorate of Oman,

the town was administered by local llwalie (viceroys),

answerable to the Omani sultans ruling first from

Muscat and, after 1840, from Zanzibar. After 1895

the llwalls of Lamu were linked to the British

oolonial administration through the llwall of the

coast sitting in Mombasa. Between 1813 and 1963

Lamu had twenty-four liwalls; the la9t, Aziz bin

Rashid, took office in 1948 and continued until

Kenya's independence in 1963.

Until the end of the nineteenth century the popu­

lation of the own and its hinterland consisted of

large numbers of slaves, watumwa and a smaller

number of free men, waum^wana. The majority of the

former, i.e. those working on the plantations, did

not live in the towns; the domestio slaves, who were

smaller in number, normally lived in the houses of

their owners. The latter sometimes intermarried

with their slaves.

i
I

The free men were divided into three groups: the

highest socially being the land-owning merchants,

descendants of influential lineages, who lived in

stone houses of the kind described in the next

chapter. They wore luxurious silk and cotton clothes,

and ate off imported porcelain bowls. Their women

 53

used gold and silver jewellery, including earrings

and bangles.

The second group of free men was that of the sharifs,

immigrants from the Hadhramout who, on account of

their supposed descent from the Prophet, constituted

the religious oligarchy. Their function included

teaching at mosque schools, arranging wedding and

divorce formalities and acting as "local doctors".

The last group was that of fishermen, artisans and

so on, who, possessing neither the pedigree of the

first group, nor the esteem of the second, were

limited in wealth and influence. The various social

groups were kept together by a unifying language and

a unifying religion.

The society was, as it still is, patriarchal;

polygamy was the rule rather than the exception.

Women walked in the streets inside shira1s . portable

tents supported on four wooden poles carried by

slaves'. Men meeting a shira* were required to stop

and turn towards the wall until it passed out of

sight.

Children received their education in mosque schools,

madarasa. where they learnt the Koran, religious

practice, ethics, Arabic language etc. Marriage was

usually arranged by parents whose duty it was to give

 54

their house to their newly wed daughter. Henoe the

Lamu proverb: "The decent girl drives her parents

out of their house, the bad one drives them out of

town" (49).

1
 55

1-3. The House

The stone-built house of Lamu is a self-contained

building housing all living, sleeping and service

accommodation that a labge family and its domestic

staff require. It stands on a small plot averaging

r

leas than 250 square metres in area, and except for

the internal courtyard, it covers the plot entirely,

giving a coverage ratio of over seventy per cent

per floor. Host existing houses in Lamu are double

storeyed, often with an additional pent-house. The

ground floor is, by tradition, the slaves* quarters;

the first floor contains the rooms of the free

owners (Fig.24-26).

The entrance to the house is through a porch daka

(plural madaka) about three metres wide, raised

one to three steps above the level of the street and

lined with stone seats. The daka has two carved

doors, usually double leafed, one leading to the

ground1floor, the other to the first floor. Door

carving is restricted to the frame, lintel and centre

post (Figs.27 & 28). The latter is not a structural

member and is nailed to one of the door panels.

Carved motifs are in some cases organic and some­

times include Kuranic inscriptions; in other cases

they consist of shallow geometric patterns (Fig.29).

 56

Carved doors of this type are common in tho domestio

architecture of the Indian Ocean region (Fig.30).

The door leading to the first floor opens on to the

staircase which half-way up gives access to a sabule

or guest room. This is not a constant feature in

all houses; it is sometimes plaoed at ground floor

level, accessible from the daka through a separate

external door. The staircase ends at a covered

landing, tekani. overlooking an open rectangular

courtyard, kiwanda (Fig.31). This is the nucleus

of the house where most of the daytime activities,

i.e. play, laundry, etc. take place and around

which the rest of the house spaces are organized.

Opposite the tekani is another verandah which serves

as the family's work space, attached to which is a

9

b throom and toilet. The third side of the kiwanda

very often has a staircase leading to the pent-house

where the kitchen (kidari cha meko) ie placed. The

fourth side opens on to a series of parallel rect-

I
angular galleries, which are the main spaces of the

house. Thus the kiwanda has traffic generated from

all sides and does not therefore have the customary

dead-end of some modern patios.

The first of the house spaces, msana >.a tlni

(Fig.32) is a verandah overlooking the courtyard.

 57

It is deep enough to sent a small group; but not

too deep to cut off the view of the sky. It is

separated from the court by wide piers, zioiva.

against which are placed benches of the same name.

The second rodra, msana wa vuu (Fig.33) is a more

private space; its two ends, nigao (singular, nga o )

are used as bed-spaces and are partitioned off by

curtains draped from round rails (taiwandl) built

into the walls (Fig.34). Ia each of these spaces

is a high bed (pavilao) reached by a low bench,

ntaganvao. The third and most private space, ndani

(Fig.35 & 36) is the master bedroom suite with a

private bathroom and toilet. 'The wall separating

the ndani from the room before it is called U3a wa

mato. In larger houses a fourth space, nyumba katl.

is sometimes also provided. This may lead to an

extra room, mtatato. which spans the street in the

form of a bridge, wikio (plural, mawiklo ) . supported

on mangrove pole beams running across the street at

a height of four to five metres above street level

(Fig.22).

Internal verandah openings are framed by decorative

plasterwork in the form of friezes along the top

and pilasters at the sides. Jambs and side walls are

pierced .by ornamental niches (gidaka za kue keleni)

capped by multi-foil arches. These features occur

in excavated houses of earlier dates and it is

 58

customary to describe them as niches for lamps.

In Lamu they are used for keeping copies of the

Kuran and informants agree that this in fact has

always been their function. One cannot help agree­

ing with this explanation as the shape of the

niches would out off a considerable portion of the

light if used for lamps. On the evidence of house

plans from earlier dates it may be assumed that

internal doorways were left as permanent openings,

i.e. with no wooden door leaves.

As one traverses the house away from tho oourt,

niches increase in frequency; the last wall of the

inner room is almost entirely covered by them.

Here, the zldaka. as they are called, take the form

of arched and rectangular niches of varying

proportions and, in most cases, a fixed module

(Figs.37 & 38). It is doubtful if these niches were

meant as storage alcoves in the normal sense; on the

other hand, due to the absence of any decorative

treatment at their back, it i3 difficult to believe

that they were meant to be viewed as patterns for

their own sake. The horizontal and vertical

repetition and the variety in shapes and sections of

individual niches do break the scale of the space

and hence the sense of monotony of the dark room

where the wife is expected to spend the rest of her

 59

life. This technique is used in the Hadhramout,

where niches are carved in the form of windows on

the external walls of buildings, giving them the

effect of multi-storey structures. The reference

to porcelain in zidaka in the early nineteenth

century poem quoted in Chapter 1-1, and information

received from elders in Lamu town, seem to suggest

that the zidaka were used for display rather than

storage. In the niches the wife arranged her show­

pieces: imported pottery, bronze artifacts, ornamen­

ted manuscripts, etc. Some existing zidaka in Lamu

have wooden shelves built across the niches seemingly

for this purpose. By varying the arrangement of

objects in the niches the look of the room could be

changed at will.

The plaster of which the zidaka and friezes are

made is prepared from lime, slaked from coral and

probably mixed with a retarding agent. When used in

vertical sections to divide niches from each other

the plaster is usually reinforced with broken shells.

It is not known whether carving was done by hand

after the plaster was applied, or whether it was

stamped by a mould. Probably a combination of both

techniques was used, i.e. wooden moulds being stamped

on the walls before the plaster was set and the

pattern being worked later by hand.

 60

Because of the discouragement of imagery in art by

Islam, figurative representation in plasterwork is

very rare. Motifs vary from stylised leaves in a

spiral surround to chain, zigzag or fluted patterns.

The turtle is a popular motif in plasterwork and a

number of stylised examples of it still exist (Pig.

42). The design, patterns and details of the

plasterwork are sophisticated, and pose an interes­

ting problem of scholarship. The zidaka are similar

to the niches in the houses of the tenth century

Ibadhi towns of the Mzab (Pig.39) and there is a

Mesopotamian miniature by the thirteenth century

Baghdadi painter and illustrator Al-Wasiti, which

shows what looks like wall zidaka with books in the

niches (Pig.40). The plasterwork around internal

doorways of Swahili houses are very similar to

carving in ashlar masonary on the fifth century

Anatolian church of Alahan (Kodja Kalesi)(50) shown

in Pig.41. However there is no evidence to indicate

I
direct links with either the Algerian Mzab or Turkey.

House walls are built of uncoursed coral in lime

mortar. Roofs and floors are of thick coral

supported on wood Joists at close centres, rarely

more than 30 cms (Fig.34). Rooms conform to a

constant module which limits their width to about

2.7 m to 3 m. This is a convenient planning grid

 61

for domestic buildings and has probably been

dictated by the spanning limitations of mangrove

poles. In mosques, where larger spans are required,

primary beams of twin soft wood members, rectangular

in section, were often used. Large wood sections

requiring advanced felling and seasoning techniques

are rare. Roofs are made of layers of coral lumps,

sealed with lime plaster. When this sets the roof

acts as a shallow arch buttressed by the thick side

walls.

Ruined sites are used as tambuu or betel tree

gardens. The nut of this tree is crushed and ohewed

and is said to have an intoxicating effect; it is

not prohibited by Islam and is therefore in widespread

use. Betel gardens were a common feature of Swahili

towns. In the sixteenth century town of Jumba ya

Mtwana, surface water from streets is drained into

the large sunken courtyards of houses which were

apparently used for growing betel trees (51).

Houses usually face north or south; apart from the

rationale of sun protection, it is possible that

facing the aibla has popularized this practice; the

Ki-Swahili term for north orientation is upande ya

Kib l a . This preference has sometimes produced

complex staircase layouts to make them land in the

 62

right direction. Such concern with orientation ie

prevalent throughout the coaat. Garlake states that

"only two complete houses in the entire coast face

westwards and both are subsidiary units within large

palace complexes" (52).

Wells provide sweet water fbr domestic use. In

some of the other towns of the archipelago, Pate

for instance, rainwater is collected from roofs and

stored. Many of the mosques of Lamu have their own

wells and there are a number of independent ones as

well. Water is emptied into a funnel, mllzamu.

placed outside the bathroom wall and connected to

the cistern, blrlka. through a half-round stone

conduit laid to fall along the inside surface of the

wall. The cistern is rectangular in plan, about

1.5 m long, 60-80 cms deep. A low partition

separates the bathroom from the toilet which is of

the pit-latrine type, capped by a coral stone plat­

form,. The back wall of the toilet is sometimes

semi-circular in plan in the shape of an apse.

Upper floor toilets discharge into the pit through

large stone ducts. A bidet consisting of two

elevated foot rests is often included in the bathroom

 63

DEVELOPMENT OP HOUSE FORM

The house described above does not date back further

than the mid-eighteenth century and we have no

evidence of earlier planning patterns from Lamu town.

Older structures do exist below ground level, but

until archaeological search uncovers earlier proto­

types it will not be possible to establish with any

certainty the lines the Lamu house has followed

during the course of its development.

There is a ruined two-storeyed house in the deserted

18th century site of Mtwapa, about ten miles north

of Mombasa. It was part of a compact terrace, and

/
consists of three room spaces behind a courtyard.

Few pre-eighteenth century multi-storeyed houses

have survived though we know from Portuguese accounts

that they did exist (53). In Kilwa, where the

Portuguese saw many two and three storey structures,

there are traces of two double-storeyed houses

1
behind the Great Mosque. There are also the remains

of two houses that appear to have been double­

storeyed in Omwe and She-Jafari on the mainland

opposite the island of Simambaya, about fifty miles

north of Lamu. They are similar in plan to the

houses of Lamu and are probably older than the

eighteenth century (54).

 64

The eighteenth century palaces of Kil a and Kua,

are double-storeyed; so is one of the houses within

the Kilwa palace enclosure.

In Bongo Mnara many houses are preserved, all single

storeyed. They include the smallest and the largest

known Swahili houses of this type, named by Garlake

"minor” and "double" respectively. In Gedi, four­

teen houses have been excavated. Like the Lamu

house, all these consist of series of long, narrow

spaces stacked behind an open court. The same

arrangement also exists in the oldest preserved

house plans, within the thirteenth century complex

of the palace of Husuni Kubwa in Kilwa. Here, north

and south of the Palace Court are a "normal" and a

"double" house respectively of similar plans to

those described above.

The houses of Bongo Mnara have entrance lobbies,

which lead to sunken courtyards (Pig.43). Adjoining

the entrance lobby is an isolated room for servants.

The house proper is entered from the court and con­

sists of a long anteroom, leading to a main room of

the same size and proportions. Behind are two bed­

rooms, eaoh half the main room in length. At the

sides of the built area of the house are two other

rooms, one on each side. In most of these, Jambs to

 65

openings are adorned with decorative niches.

Some of the houses of 3ongo Mnara show remains of

timber shelves built across the width of some of

the rooms, about 1 m wide and 1 m above the level of

the floor. This feature Is also found in the

thirteenth century Husuni Kubwa of Kilwa. It

appears twice in Gedi, where it is built of

masonry, and twice on the eighteenth century palace

of Kilwa. One of the rooms in the residential core

of the latter shows masonry supports for a

mezzanine that covered an area of 2.8 x 2 . 5 m at a

height of about 1.3 m above ground level. To all

these Garlake gives the designation "bed" (55).

i
However, these structures are unsuitable for this

function, being too high, and, in one case, too

short for a bed (56). It would have been easier and

cooler to sleep on the floor; it is difficult to

imagine these built-in structures being used for

anything other than storage.

The rooms marked (A) on the 3ongo Mnara plan

illustrated in figure 43, which are called by

Garlake Main Private Rooms, were probably guest

rooms. They hnve secluded entrances and in many

oases private latrines attached to them.

 66

Gedi houses are similar to those of Songo Mnara;

they have fewer niches but some Internal walls have

decorative pilasters. Some of the houses of Gedi

show the addition of a store, accessible from a high

trap door, reached by a ladder. Gedi's excavatar,

James Kirkman, believes that they may have been safes

for the storage of cowtie currency (5 7 ).

In the Lamu houses the servants rooms are said to

have occupied entire floors. According to the

inhabitants of the town, living, sleeping, work and

hygiene accommodation for the domestic slaves were

arranged on the ground floors below those of the

owners. This however seems hardly credible as in

some cases ground floors have more intricate

decorative plasterwork than upper floors (5 8 ).

The courtyards of Gedi and Kilwa are smaller than

those of Gongo Mnara and, due to the compactness of

the built-up areas of the towns, they are more

I
varied in arrangement, and often sunk to ensure a

deeper shadow. The two-storey Lamu house has two

courtyards, one for each floor. The first-floor

courtyard is smaller and covers part of the grounnd-

floor one, while the uncovered portion continues

through the first storey in the form of a well (see

Figs.24 & 25). This is probably the most striking

 67

difference between these houses and courtyard

houses in the Arab world and elsewhere where the

open well continues vertically in the form of a

single shaft.

The finely carved niches at jambs and end rooms,

and the decorative plasterwork in houses generally

are almost certainly attempts to counteract the

monotony generated by the practice of stacking the

lopg narrow spaces of the house behind each other.

Garlake's observation that decoration is less

common in Swahili domestic architecture than in

mosques (59) is true of the southern Swahili

houses, probably because ornamentation was there

substituted by hangings draped from wall pegs (60).

In the houses of the Lamu archipelago decorative

plasterwork appears to be the product of an

artistically mature period. The skill portrayed in

its execution represents the highest attainable

within ,the limitations of the medium of coral lime

worked with basic tools. Its chief interest lies

. in its intrinsic relationship to the architecture

that embodies it; a relationship in which decoration,

structure and plan patterns are complementary to

each other.

The houses discussed above share two important

 68

qualities: firstly a single-minded axiality in plan,

and secondly a constant module limiting room widths

to about three metres. The latter, as we saw, is

the result of the development of a domestic planning

grid dictated by the limiting technology of the

materials used. The first is more difficult to

explain; but, on account of its universality, it

must also be accepted as a permanent pattern. There

are some non-axial houses, but these are so few that

they can be considered as independent exceptions.

The large number of houses in the eighteenth century

island town site of Kua, about thirty miles east of

the Rufiji mouth, are a variant of the pattern

described above. They are twin houses, eaoh with

its own large court, but both sharing a common

entrance. Garlake believas they were the household

quarters of the two wives (61). If this is so they

are the only known Swahili houses where a privacy

of this kind was considered necessary. There are

remains of a house in Dondo (62) on the mainland

above Faza island, which is also not axial in plan.

A daka is flanked by two long rooms, one on each side

entered from an outer passage accessible from the

daka. The rest of the plan is not clear, but it

could have been similar in organization to those of

 69

Kua mentioned above.

The houses of the poor would have been built of mud and
wattle walls roofed with palm leaves or grass thatch.
These are impermanent materials which deteriorate within
a very short span of time. Although such dwellings
undoubtedly housed the great majority of the population
very little is known about them.

I
 70

PART TWO: GENERAL CONCLUSIONS

2-1. Swahili Concepts Of Space

The study of the modelling and use of urban spaces

is meant to find out how a building, a neighbourhood

or a town have come to be the result of the action

of available skills on existing materials to produce

a specific environment for a particular way of life.

Urban spaces and the manner in which people use them

are interdependent in the sense that systems of

behaviour can only fundtion within an appropriate

shell, and the shell, once formed, gives permanence

to these The patterning of space is a

culturally determined activity, "we can choose the

form of our houses no more than the cut of our

clothes. The one is imposed on us to the same degree

as the other by social usage." (6 3 )

Below is an attempt to analyse the form of Lamu town 4.

V—
as an expression of the specific "mood" of Swahili

culture, and to draw general conclusions on the

morphology of the Swahili town generally.

PEDIGREE

Figure 23 shows the town of Lamu divided into

thirty-six mitaa The group to the north, known as

Mkomani, is made up of large stone storeyed mansions

 71

whioh are the quarters of the town's Influential

lineages. The southern ones, called Langoni, are

the quarters of the poorer sections whose houses are

built of mud walls and thatch roofs of coconut palm.

Each group has its own mosques; but the important

Friday mosque is in Mkomani and its mkhatlbu. reader

of the Friday sermon, is always chosen from one of

the lineages of Mkomani.

Frins has noted that Lamu inhabitants distinguish

two types of citizens depending on the part of the

town they occupy: wa-Amu. the original inhabitants,

are the occupants of the stone sector, and watu wa

A m u , occupants of the mud and wattle sector. He

describes a tendency for patrician family heads to

move to the northern half of the town with advancing

years. He has also noted a resistance among stone

house owners to letting out houses or rooms to pros­

pective tenants who do not belong or aspire to

belong to "people of pedigree" (64). Even the

f
founder of Lamu's important mosque college, Al-Habib

3alih, as a new arriver to the town, had to build

his house outside the stone zone (65). The college

he founded is likewise situated there. It is of

course fitting that this should have been so, as the

academy houses a large number of foreign boarders

for whom it would have been very difficult to find

 72

accommodation in ilkomani anyway. During the last

day of the maulldi celebrations, when participants

collectively pay the customary homage to Al-Habib

Salih's grave, the two sections send separate

representatives as though they were two different

entities (6 6 ).

The stone sector is the traditional power oentre of

the city state. All decision-making rights such as

choice of ruler, declaration of war, peace settle­

ments etc., have traditionally remained the exclusive

rights of the occupants of the stone section. The

baraza. the traditional discussion forum, which con­

sists of benches placed along the harbour (Fig.4 4 ),

is still only used by the occupants of the stone

town (67). When a "stranger" greets a baraza in

session the customary karibu (3w. a welcome to join

us) is not said in reply.

The general Swahili term for descendants of influ-

1
ential lineages is Wa-Arabu (sing. Mwnrabu) in

contradistinction to recent migrants from Oman and

the Hadhramoht known as Wa-Manga and Wa-3hlhirl

respectively.

The term Wa-Arabu is not a distinction of race, as

it may imply, but of pedigree. The majority of

 73

people known by this name do not in fact speak

Arabic, and those who do use it as a second lan­

guage. irfhen Ibn-Battuta visited Mogadishu

(c.1332 A.D.), the sultan, who knew Arabic, spoke

the local language, and from the poor Arabic of the

Kilwa Chronicle (c.1530) we may divine that

Ki-Swahili, or a prototype of it, and not Arabic,

was the first language of its writer. The three

taifa of Lamu, the original tribal groupings from

which the town's present clans descend, all have

Swahili names (6 8 ).

In Lamu, the local name for Wa-Arabu is Ziloho

(sing. Kiioho). The word is derived from .1oho. the

name of their exquisitely embroidered ceremonious

gowns. Dress is traditionally a distance preserver

and has here given its name to the ruling class in

this pedigree conscious society. The pedigree of

Wa-Arabu is a function of the antiquity of their

ancestry on the East African coast. This was

f
noticed by Burton in Mombasa in 1857 when he found

the Mvita "the older and consequently the nobler of

the Swahili groups" (69). There were few women among

early waves of Arab migrants, which meant that most

men would have married African wives. As a result,

nobility in Swahili society is patriarchal, i.e.

a Mwarabu father keeps the pedigree for his offspring

 74

irrespective of the colour or ethnicity of their

mother. The situation is not the same in the case

of a Mwarabu wife with a non-pedigree husband. An

informant remembers his grandmother reserving in her

sitting-room a wooden stool for one of her visitors

whose husband had African blood from his father's

side, in distinction from other guests who sat on

padded chairs.

Reference has already been made to a tradition that

the present inhabitants of Sheila, the original

escapees from Manda, being strangers to Lamu, were

only given shelter on the condition that they did

not build in stone. Stone houses, no doubt because

of their permanence and grandeur, are regarded the

privilege of people, of pedigree. The general Swahili

word for house is nvumba; but a multi-storey stone

house is called belt. There is a current tradition

among rural Bajunis which says that at the turn of

this century an epidemic broke out in their islands,

killing about seventy percent of them. As a result

the demoralized survivors vowed never to build in

stone again. The story is interesting in that it

implies that rural Bajunis are original stone house

dwellers and hence people of pedigree. Recently, a

European who was finding difficulty in recruiting

local workmen for a factory he was intending to set

 75

up south of Kl 3 mayu found the people much more eager

when they realized that the factory was to be built

of stone. " 'Aiyee, dtone', came the surprised

murmur from all round, and I realized that in this

word lay the proof, in all their minds, of the

enduringness of the work. 'Yes’, I added ... 'and

the walls are as thick a 3 this ...' 'Aiyee - 30

thick? and the walls all of stone? Then it is a

building of ten generations!' Prom this point the

whole attitude of the villages changed" (7 0 ).

The difference in life-style between the two halves

of the town is reflected in the character of the

main street. Shops are concentrated in Langoni

(Pig.4 5 )i they become less in number and larger as

one approaches Mkomani (Frontispiece). At its

entrance they almost disappear, and the busy, crowded

main street changes into the quiet thoroughfare that

a rich aristocracy expects to find in its own

quarters. The absence of shops is axiomatic of the

Zllohoa* contempt for manual work. It is said that

until recently the gentlemen merchants of Lamu used

to send their garments for starching to the Hadhra-

mont. When the Jahadhmis, the town's first shop­

keepers, arrived from Oman during the seventeenth

century, the shop they set up in their house gave

its name to the whole mtaa : Madukani (Sw. at the

shops)

All the town's caf^s are concentrated in Langoni.

A Kljoho would not normally be seen in one. When

the present writer, out of ignorance, insisted on

one Joining him for coffee, the caf 6 was soon

surrounded by an astonished crowd. It was then

explained that that was the first time a Kijoho had

ever sat in a oaf 6. Until recently the madaka of

houses were used for this purpose. Visitors nor­

mally arrived after supper to spend the evenings

talking, fingering beads and eating halwa (71) with

coffee. Beads, coffee and halwa are traditional

adjuncts of talk in Swahili society; one of the

captains in Vasco da Gama's fleet was treated to

halwa by the Sheikh of Mozambique in 1498. He was

also made a present of a string of black pearl

beads (72).

INVOLVEMENT

A study of Swahili house groupings suggests that the

mtaa may have developed along the following lines:

a house was built on a large site; with the demand

for additional accommodation other units were added,

arranged around a central private access. Through

intermarriage, clans and mltaa sometimes merged.

After a while, la^d available for expansion became

 77

soarce, houses expanded vertically and, where

possible, bridged over streets. Densities rose,

plots shrunk, ground coverage reached maximum

limits, and party walls and sometimes other shared

facilities increased (7 3 ).

Within housing blocks units bulge into each other

in a manner suggesting that a high degree of co­

operation existed between neighbours and that

adjoining houses were built at the same time,

perhaps by related families. We have archaeological

proof of this from Gedi and Songo Mnara. In the

first, part of one house was taken over by an

adjoining one apparently while both were under

occupation (74). In the second, fifteen adjoining

houses were combined to form what is now called the

Songo Mnara palace (75). The room, spanning over

the street mentioned in Chapter 1-3, the so-called

w l k i o . belongs to one house but depends for

structural stability on external walls of others.

I
In Lamu there are twenty-three mawaikio (Fig.52),

the longest one measuring eighteen metres.

>•

The streets of the Swahili town provide usable

public space (Fig.46). Their function may be

contrasted to that of streets in modern cities

which are strictly meant for "going through", where

 78

"staying in" is synonymous with loitering, and is

sometimes punishable by law. In the Swahili street

one is constantly brought face to face with others;

as a result the common phenomenon of street

behaviour intensifies. To avoid an abrupt change

from the formal environment of the street to the

intimate atmosphere of the house (often a short­

coming of modern house designs), a form of tran­

sition becomes necessary to achieve a feeling of

arrival and to ease the momentum of tension

appropriate to street behaviour. In the Swahili

house this is achieved by the daka porch, where

the changes of level, materials and light intensity

help to break the effect of outside behaviour.

This form of involved living is common in

expatriate communities and has here been accelerated

by the high involvement ratio of the inhabitants'

cultures of origin in South Arabia and the Gulf

region. In interpersonal encounters this involve-

ment runs high. In funerals the coffin is followed

to the cemetery by a large procession of mourners

who take turns in carrying it. By the time it is

lowered each mourner will have carried the coffin

more than once. The launching of a new boat is also

a communal activity. It normally takes place on a

.Friday afternoon; the Friday noon congregation

 79

leaves the mosque directly for the launching ground

where they all join in pulling it to the water (fig.

47). The operation is accompanied by unison chants.

Pushing and shoving in public places is character-

istio of the 3wahili town (fig.48). When an informant

was told that in some oountries touching during an

argument could be legally claimed as assault he was

shocked. To the Swahili the sense of touch is a

natural means of communication; the habit of hand

kissing is but an extension of this.

Although many of the inhabitants know how to read and

write it cannot be said that the written word has

played a major role in communications in Swahili

society; the most important medium remained the spoken

word. News were announced by special town criers Just

as the oall to prayers is today shouted out from

mosque roofs. Important announcements were accomp­

anied by music (76). An early nineteenth century poem

begins by oalling upon the different categories of

f
criers to announce the news of war like this;

"Zez® and Asha Haraadi,

Say to our brave men,
'Those who may come,
Will find our door3 latched.'
Kyumbe, strike with the palm leaves the copper plates!
Blower, blow hard the alwa
Let the sound of the horn follow!
The ornaments of the siwa
Improve its resonance." (77)
 80

In Lamu important items of news such as new film

shows, bus departure times, etc., are still cried

out in this manner. Stigand, writing in 1909, com­

plains about the coastal man frilling to distinguish

between hearsay and direct evidence. 'He will relate

any story he has heard as if it has occurred to

himself, and thus one is abld to get first hand the

relation of so many marvels, all of which the

narrator alleges to have seen for himself." (7 8 ).

Marshall Mcluhan, writing about fifty years after

*»
Stigand, examines similar situations and concludes

that this is a natural state of affairs brought about

by the oral mode of communications. "The literate

society develops the tremendous power of acting in

any matter with considerable detachment from the

feelings of emotional involvement that a non-literate

society would experience" (79).

The high involvement ratio has set the scale of

streets and open spaoes at an intimate level. It

has also had the effeot of reducing the "personal

space" (80) of the Wa-Swahili to a minimum. The

prayer hall of the Lamu Friday mosque, which measures

le38 than 250 square metres, accommodates the entire

male population of the town during the Friday noon

prayers (Figs.49 & 50). This gives an area of less

than three-quarters of a square metre per person.

 81

Involvement is also responsible for the absence of

any expression of grandeur from Swahili archi­

tectural patterns. The Swahili mosque for instance

is not a large building, and cannot always be easily

distinguished from houses externally. As we saw,

it has no minaret nor is it built to the scale of

mosques of the great Islamic capitals. When a

number of Lamu elders were shown pictures of the

great mosques of bin Tolon, Qairawan and Cordova,

they were impressed by their grandeur, but said they

would not like to pray in them because they found

them too vast and impersonal (81). The Yumbe or

government house of Lamu is no more than a

fifteenth of a hectare in area and no different

from any other town house in plan, that of Gedi

being only slightly larger. This is generally the

sort of scale for rulers' residences throughout the

Swahili coast. The thirteenth century Husuni Kubwa,

which appears to be an exception, is referred to in

the next chapter.

PRIVACY

The plan of the Swahili stone built house gives

maximum length to communication lines between

rooms; a similar quality is achieved in the streets

by staggering the front doors on plan; both devices

 82

are brought about by considerations of privacy.

Another interesting dictate of privacy is found in

the 15th - 16th century houses in Gedi where house

plans "almost always ensure that the doors of outer

rooms are never placed directly opposite the doors

leading on to the inner rooms. Thus if both ante­

room and main room occur, the ante-room will have

two doors, and the main room a single central door

at the front. At the rear of the main room, the

bedrooms will, of course, again each have a door.

If there is no ante-room, the main house will be

entered from the court by a single central door

unless there is only a single bedroom with a cen­

tral door in which case the main room has two

doors" (82). In the north wall of the main block

of Gedi palace Kirkman found a blocked doorway which

he suggests was a small door through which pedlars

could sell their wares. "It is possible that

pedlars were not always pedlars and they sometimes

offered more interesting wares than cloth and beads"

(83).

The house is a closed box; despite the high

relative humidity no external window openings are

tolerated. The environmental requirements of day­

light and ventilation are performed through the

kiwanda or courtyard. As a result the street

 83

acquires a privacy of its own; in a sense it acts

as a public lounge, an extension of the madaka or

the semi-open porch/reception rooms of houses.

Street facades are uniform and, except for entrance

doors' openings, opaque. Front doors are endowed

with so much embellishment as to make them unique

components of their kind in domestic architecture

anywhere. This highly personalized treatment is

meant to restore the identity of the house in the

setting of the standardized facades; it is not

uncommon to find a carved door in front of a mud

hut built on the ruins of a stone mansion. The

carved door appears to be an early feature of the

architecture of the region; when the Portuguese

plundered the town of Paza in 1587, among the loot

carried away were a large number of carved doors

(84). The Portuguese historian Barbosa found the

doors in Kilwa houses "well carved with excellent

joinery" (8 5 ).
1

Like the Egyptian temple, the Swahili house is

axial in plan; as one follows the axis in the

direction of the interior of the house, spaces be­

come gradually darker, more decorated and more

intimate. The axis acts as an intimacy gradiant;

the further up a room is placed, the more private

 84

It is. The gradient is marked at its ends by the

daka and ndani. the most formal and informal places

respectively; the former is the antithesis to

privacy, the latter the antithesis to involvement.

The Swahili town has a large number of mosques.

There are twenty-three in Lamu town alone giving

the very high ratio of about one mosque per one

hundred adult males. It has been customary to

contribute this to exceptional preoccupation with

religion. Whilst this observation may not be in­

correct, it does not describe the entire function

of mosques on the East African coast. They serve

as the town's public lounges; the equivalents of

the common, the Georgian Square or the public park;

or the "social clubs" of the Swahili town. Like

clubs which generally draw their members from

specific social classes or ethnic groups, the

Swahili mosques draw their visitors from descendants

of specific lineages. In Lamu, for instance, the

mosques of Anisa, dated 1830 on Figure 23» and

Raskopu, dated 1797, are used by the non-pedigree;

Utukani, dated 1823, is used by descendants of the

Mhdali lineage; Pwani, dated 1370, and Nna Lalo,

dated 1753, are for the Ma'awis; Mpiya, dated 1845,

is for the Jahadhmis, and so on.

 85

The thick stone vails and cool matted floors of the

Swahili mosques provide the protection men need from

the external environment and the involved existance

of everyday living (Fig.51). Until recently,

contrary to Muslim practice, Lamu had a womens’

mosque where they too could spend some time in

solitude when needed.

 86

2-2. Structure of the Swahili Town

The shape and layout of the linear Swahili harbour

town, of which Lamu is an example, is determined by

a heirarchic road network. Lamu's main street is

the principal communication spine, as the suq or

covered market street is to the Muslim Arab town,

and the rienamanuB was to the Imperial Roman town.

Lamu's market square, facing the fort, continues to

the harbour forming an east-west hinge which bisects

the main street in the same manner as the Roman

corda bisects the decamanus at the forum. The

market is placed at the intersection just as it

would be in the market towns of medieval Europe

(Fig.52).

About two hundred and fifty metres north of this

axis is a straight wide road which connects the

sea-front to the town's traditional council chambers

(yumbe) and continues west to a large mtaa named

Utuku Mkuu, which means "great market". There is

no market here now; but it is probable that this

street was the original corda of the town during the

eighteenth century or maybe earlier. At its east

end is the mosque of Nna Lalo which stands in its

own precinct and has the equivalent date of A.D.

1755 inscribed on its mihrab. Dates in Lamu mosques

 87

are not always reliable, but that of Nna Lalo Is

probably the least suspect (8 6 ). The Ndia Kuu (main

street) of Mombasa Is placed In the town in the same

relative position as the main street of Lamu. This

is crossed by a square overlooked by the market and

now, the customs house. The market square opens on

to the town's old harbour.

The two axes lead to the mitaa through secondary

lanes which become smaller and less ordered the deeper

they penetrate the residential quarters; in some

cases terminating in cul-de-sacs at house daka

entrances. This is very similar to the road-net of

the linear merchant city of the Muslim orient (87).

The chief difference of the Swahili harbour town

from the Muslim town is one of scale. Because of

the seasonal nature of its trade, dictated by the

rhythm of the monsoon, and the relatively limited

volume of its merchandise, the Swahili market did

not grow to anything like the Muslim covered suq.

The large building complex of Husuni Kubwa (Fig.53),

which covers an area of about a hectare is thought

by its excavator to have been intended as an

emporium (8 8 ). If this is so its foundation repre­

sents a landmark in Swahili town development in

that it is the first example where the function of

trade is given architectural embodiment in a scale

 88

comparable to that of the larger merchant oltiea of

thdt time.

The spatial division of Lamu into a stone-built town

and a mud and wattle sector is a general pattern in

the spatial organization of the Swahili town. In

the mid-nineteenth century Burton found Qavana, the

old town of Mombasa, built of "narrow huts clustering

around a few one-storeyed flat roofed boxes of

glaring lime and coral rag" (89). Fifteenth century

Portuguese eye witnesses describe coastal towns in

similar terms. Some like Moz mbique and Oja had few

stone buildings, others like Malindi had many (90).

All had their own sections of mud and thatoh buildings.

The division of the town into mitaa being residen­

tial sectors for families of related clans is also a

general East African coastal pattern. The Arab

geographer laqut, writing between A.i). 1212 and 1229,

describes Moqadlshu as being occupied by "tribal

sections, having no sultan but each clan having a

sheikh whose orders they carry out" (91). This

draws upon precedents from Muslim Arab towns.

Ta'qubi describes the seventh oentury town of Kufa in

western Iraq as consisting of several tribal quarters;

each having its own mosque, public bath, and, in

some cases, a market. In Samarra, the Abbasid Caliph

 89

Al-Mu'tasim allocated independent wards to the

different ethnic units.

Unlike the mltaa of the mercantile towns of the East

African coast the neighbourhoods of the Persian and

Arab towns later developed into merchants’ guild

wards as the towns grew into trade termini. When

Ibn Jubair visited Baghdad in c.1184 A.D. he found

the eastern part of the town made up of seventeen

quarters. One of them, Itabiyyah, was famous for

silk and cotton weaving of various colours. The

word tapestry owes its origin to this name. This

diversification was reflected in the Muslim towns

where "producers or retailers of the same kind of

goods will always occupy adjacent stalls, in fact

each trade is likely to have one of the market lanes

completely to itself" (92). The Swahili harbour

town did not have the chance to follow this line of

development (93). It may therefore be described as

a variant of a Muslim market town; its development

towards a fully-fledged merchant city was arrested

for a variety of reasons, chief among them being a

scarcity of raw materials (9 4 ) and prolonged foreign

intervention.

Although there can be no doubt that the prime raison

d 13tre for coastal urbanisation was trade, it must

 90

not be assumed that all civilizing and town

building activities on the East African coast were

dictated directly by the demands and limitations of

trade. There was, for example, the town of Gredi,

four miles from the sea and two miles from Mida

creek which was during the fourteenth, fifteenth and

sixteenth centuries a large town covering an area of

about forty five acres within the outer walls. It

was not a commercial town or it would not have been

ignored by the Portuguese chroniclers. The reason

for its existence is not known, possibly it was a

resort for absentee landlords who belonged origin­

ally to Malindi (9 5 ).

There is also the town of Pate, which, despite a

very poor harbour, became a wealthy city-state of

considerable importance. Pate and perhaps Faza,

which has a slightly better but also poor harbour,

exemplify a type of economy which did not rely on

shipping movement in their harbours for their wealth.

I
They were^ not "depot" towns as was the case with

Lamu which relied on its excellent harbour to play

the role of the middle-man of trade, i.e. storing

commodities for reshipment. Pate was famous for its

fine coloured woven fabrics; apparently the quality

of the produce was so high that the te r m "Pate cloth"

became a general term to describe the richer varieties

 91

of fabrics. The town is also, according to its

own history, supposed to have manufactured the first

locally made ocean-going ships (96). It is men­

tioned in an account of a voyage by a Portuguese

captain undertaken in 1547 as having been a place

in which large vessels for the orossing to India

could be hired (97). Pate was a "workshop" in

distinction from a "depot" town. There is evidence

of industry elsewhere too: iron working in Manda,

Kilwa and possibly Malindi (9 8 ), cotton weaving in

Mogadishu the Kerimba islands and the adjacent

coast. The Portuguese historian Duarte Barbosa

records an interesting observation about weaving in

Sofala where locally produced white cloth had woven

into it blue and other coloured threads drawn from

material of Indian origin.

It is tempting to argue that had these towns had the

chanoe to develop naturally they could have grown

into industrial centres. However, such an argument

is offset by the puzzling fact that a town like

Manda which enjoyed four centuries of uninterrupted

existance, rather than developing along these lines,

died out by the end of the thirteenth century (9 9 ).

The makonde cultivations described in Chapter 1-1

produced the agricultural villages which housed the

 92

plantation workers (mostly slaves) and their

families. These were seasonal villages in places,

but at others they grew into permanent townships.

Judging by the remains of some of these villages,

Hindi for example, on the mainland opposite Lamu

island, it appears that they included a number of

mosques and had communal wells. The largest of the

mosques was used as a Friday mosque and was placed

at a central position. The sermon was delivered by

a travelling , m khatibu who belonged to the town

owning the plantations. "Muslim Canon law does not

insist on the faithful to recite his prayers in a

prayer house; but the Friday community prayer may

only be held in a fixed settlement with a permanent

population of whom at least forty adults must be

present to make the ceremony valid" (100). This

stipulation tended to accelerate the growth of

towns generally, without which the inhabitants would

have been precluded from full religious life.

One must pause a little to consider the reasons and

then the results of the success of Islam in the East

African coastal region. It demanded little in the

way of religious duties or ritual; but offered in

return an enhanced social position, signified by the

cap and gown, a membership in a large community and,

not least, a paradise with green lawns, rich orchards,

 93

attending hourls and so on. Islam did not abolish

slavery, although the Koran, In several places,

demands a fair treatment for them (101). The

religion appears to have provided an attractive

world for the multitudes of plantation slaves

emerging as they were from a secure existence where

man's destiny was seen through the collective destiny

of the tribe to the organized relations generated by

the new class society. The sherif soon distinguished

himself as the propounder of the divine message. He

took the place of the ritual leader, priest and

doctor of the tribal world. To a large extent he

still acts in these capacities nowadays.

A reading of the history of Muslim expansion and the

resulting acceleration of urban growth shows that

the religion embodies a strong urban message.

Muhammed was himself a townsman who, since the age

of twelve had been joining his merchant uncle on

caravan journeys to Syria and elsewhere. His

f
message came as the ethics of the sedentary over

those of the nomad. Trimingham quotes "an intelli­

gent and well read faaih (learned man)" in Ujiji as

saying that, from his experience, Islam needed an

urban centre to root itself (102); and Meek, dis­

cussing Islam in Nigeria, records a similar

observation (103).
 94

On the East African coast no evidence of pre-

Islamic urbanisation has yet come to light through

archaeology. This does not mean that no permanent

settlements existed. An examination of the exports

of the coast during the early centuries of our era

has led Neville Chittlck to suggest that permanent

settlements did exist (104). The same writer also

notes that the lower strata at Kilwa, showing

remains of rectangular houses may also be pre-Islamic.

The scaroity of visible traces of these pre-Islamic

settlements probably indicates that stone building

techniques were not known. The towns would have

been constructed of less permanent materials which

do not normally survive in this climate where every

man-made thing seems destined to be overwhelmed by

nature and forgotten.

These settlements, according to one theory, began

their life as seasonal towns and grew gradually

through trade into market towns (105). This

important change was probably facilitated by the

development of the makonde plantations which ensured

not only a surplus of food for the town dwellers,

but also goods for their trade. We do know that

some of the agricultural produce of these plan­

tations were exported to Arabia during the fifteenth

 95

century (106).

We do not know when the tnakonde work pattern took

form or how It developed. If the Kenya Land Com­

mission on evidence is anything to go b y f we may

assume that much of the hinterland had been in

effectual and effective occupation by the Lamu towns­

men or their slaves since the thirteenth oentury

(107). The word kundl appears in a short Jwahili

vocabulary recorded by Al-Idrisi in the twelfth

oentury from which some scholars divine a reference

to konde (108). However, these dates do not

indicate the origin of the system which is likely to

be older. Its growth is almost certainly linked to

the growth of towns as these plantations needed to

rely on an organized power structure of a type that

only an urban centre could provide. Equally, without

a food-producing hinterland of this kind, the

townsmen could not have turned their energy to the

requirements of trade.
I

ttany features of the 3wahili town show similarities

to prototypes in Arabia, Iran, India and other

regions of the Indian Ocean. James Kirkman has

attributed certain techniques of arch building found

on the Swahili coast to India. Neville Chittick

has compared one of the tenth century houses of rtanda

 96

to contemporary plans in the Persian port of Siraf;

and the thirteenth century Husuni Kubwa to the

eighth century Ahbasid Palace of Ukhaidir in West

Iraq. The n^alawa or Swahili double outrigger canoe,

the coconut and other items have been traced to

ancestral places in Indonesia and Malaya. The

marked spatial segregation in the Swahili town

between pedigree and non-pedigree quarters may be

compared to that found in Indian villages where the

quarters of the caste sections are divided from

those of the outcasts harlIans. The Swahili

cultural unit owes many of its elements to regions

in the Indian Ocean with which, until the nineteenth

century, it had more in common than with places on

the continent of Africa (109). It was not until the

opening up of the trade routes with the interior,

especially after the establishment of the Zanzibari

sultanate that the picture began to change. Prior

to this the coast's main lines of contact were

maritime. During this time many skills and tech­

niques were imported to the Swahili coast mainly

through migrations, some of which took place on a

large scale.

It is altogether fitting that this should be so.

"In so far as culture is an adaptation to a special

enviomment, it must be modified by transfer to a

 97

different environment and the degree of modification

is likely to be inversely proportional to the

culture's technological level" (110).

The imported impulses gradually underwent local

modifications resulting in a culture with its own

personality reflecting the social structure, state

of knowledge and skills of this distinct historical

period. In architecture for instance we saw that a

number of design patterns and building types are

sui generis (111). In the field of religious

institutions Trimingham cites local religious

practices where orthodox Muslim dogma is modified

by local mythology. Examples of other cultural

manifestations showing sililar modifications can

also be cited (112).

1
 98

NOTES AND REFERENCES

(1) . They are, from the Horn of Africa down: Bilad

Barbara (the Country of Barbara), a l - 'Ard
al-wafrah-al-Judan (the Barren Land of the Black
Peoples). 'Ard al-Zenj (the Land of Zen.1). *Ard
Sufalat al-Tibr (the Land of Sufala, where there
is gold) and Ard Waa Waa (the Land of Waq Waq).

(2) . Al-Mas'udi, tturu.l. Vol.2, pp.107-109.

(3) . Al-Mas'udi, Al-Tanbih. p.51.

(4) . Al-Mas'udi, Muruj. Vol.2, p.17.

(5) . Ibid.. Vol.2, p.6.

(6) . A form of administrative grant of land under

which portions were made over in semi-ownership
subject to tithe. Many types of 'ikta' were
practiced but the one that concerns us here is
that dealing with barren land: 'ikta' al-'ard
al-mawat. which became widespread during the
period under consideration. Al-Balathiri
(Putuh) pp.501, 503, 505-506, states that
tracts, some as large as the equivalent of
1,200 hectares were sometimes given over.
(See Encyclopaedia of Islam, Vol.3, pp.1088-
1091, and Al-Duri, A.A. (1970).

(7) . Al-Tabari, Tarlkh ... Vol.3, p.1750.

(8) . This is attested by the fact that the leader of

the Zen.1 revolt in Basrah, Ali Ibn Muhammed,
after delivering his famous 1Id-ul-Fitr speech
(A.D. 869), asked "those who understood to
translate to those who did not." See Al-Tabari,
op.cit.. Vol.7, p.547.

(9) . Al-Jahiz (ed. Van Vloten), pp.57-85.

(10) . Freeman-Grenville, G.S.P. (1966), p.9.

(11) . Tippu Tip (1970), p.121.

(12) . The Nabahani (Arabic Nabhani) is a clan of an

Arab tribe that ruled Oman for two and a half
centuries until the beginning of the fifteenth
century.

(13) . Chittick, H.N. (1969), passim.

(14) . The poem, entitled al-Hamziya, is written in

 99

the Ki-Amu dialect by a Bwana Mwenga. It is a

translation of a thirteenth century Arabic poem
written in Egypt. According to Dr. Knappert,
to whom I am indebted for this information,
this is the earliest known Swahili poem.

(15) . There is an account in 1648 of an adventurous

Portuguese plan made by a captain Salvador
Correa de Sa who wanted to start from West
Africa and traverse the whole continent in
order to force Pate to acknowledge allegience
to Portugal. See Strandes, J. (1968), p . (99 .

(16) .^uyumbe zao mbake ziwele tame,

Makinda ya popo iyu wengeme.
Husikii hisi wale ukeme,
Zitanda matandu walita ndiye.
Madake ya nyumba ya zishani,
Sasa walaliye wana wa nyuni.
Bumu hukoraoma kati nyumbani,
Zisiji na kotne waikaliye.
Harries, L. (1962), pp.96-98.

(17) .Kirkman, J.S. (1957), passim.

( 18) . Kirkman, J.S. (1964), p.73.

(19) .Garlake, P.S. (1966), p.98.

( 20) . Stigand, C.H. (1913), p. 122.

( 21) . Prins, A.H.J. (1965), p. 228.

(22). Ylvisaker, M. (1971), p. 6.

(23) .Trimingham, J .3. (1964), p. 147

(24) .Official correspondence quoted in Salim, A.I.

(1973), p. 24.

(25) .Greffulhe, H. (1878), pp. 3228-3233.

(26) . Official report quoted in Salim, A.I.

(1973), p. 38.

(27) . Jackson, P. (1930), p. 21, p. 37, p. 42 and p.91.

(28) . Chanler, W.A. (1896), pp. 17-18.

 100

(29) . Trimingham, J.3. (1964), p. 12.

(30) . Prins, A.H.J. (1967), p. 12.

(31) . GctrUk^pp. 78-79 and Kirkman, J.3. (1954), p.10.

(32) . In addition there are two nineteenth century

forts in the area, one in Siyu, the other in
Lamu. There are also the remains of a palace
in Pate which ia not excavated. 3uch structures
are classified by Garlake as an independent
building type; he also includes the stone lined
wells as a fifth type.

(33) . The reference to some mosque external doors as

exit doors is made on the knowledge that, during
prayer times, they could not have acted as
entrance doors as they by-pass the ablution
facilities.

(34) . This is placed in front of the mihrab to pro­

tect the ruler during prayers. It was intro­
duced by the first Umayyad Caliph (seventh
century A.D.).

(35) . Kirkman, J.S. (1964), p. 102.

(36) . Garlake, P.S. (1966), p. 98.

(37) . Ibid.. pp. 13-14. Garlake notes dimensional

irregularities and imprecision in the setting
out of the main axis.

(38) . Mumford, L. (1967), facing p. 84.

(39) . Hitti, P.K. (1967), p. 298.

(40) . Prins, A.H.J. (1965), pp. 263-275.

(41) . In a treaty signed between the Portuguese and

Pate in 1637 one of the conditions imposed on
Pate was the destruction of the town walls.
See Strandes, J. (1968), p. 186.

(42) . Orally from informants. Prins (1971), p. 47,

presumably also relying on oral evidence,
records similar conclusions.

(43) . Abu-l-Mahasim. Al-Manhal al-3afi wal-Jiustaufj

ba'dal wafi. Quoted in Guillain, G. (1856),
Vol.l, pp. 299-300.
 101

(44) . Chittick, H.N. (1967)

(45) . Sir John Kirk, The British Consul General,

quoted in Salim, A.I. (1973), p. 39.

(46) . Salim, A.I., (1973)

(47) . Ibid., p. 20.

(48) . For the cattle disease see Fitzgerald (1898),

p. 344-348.

(49) . "Mwana wakike akioka hukutoa ayumba, akioia

hukutoa mui."

(50) . Pevsner (1967), p. 195.

(51) . Rainwater from streets is drained down to the

courtyard through short, earthenware drains.
I am indebted for the interpretation mentioned
in the text to Dr.Kirkman who excavated the
site in 1972.

(52) . Garlake, P.S. (1966), p. 89. One of the

examples is at Gedi and the other is at Kilwa
within the Husuni Kubwa complex. Other west-
facing examples do exist in fact. See
Wilding, R. (1972), p. 43.

(53) . Freeman-Grenville, G.S.P. (1966), p.66.

(54) . Wilding, R. (1972), p. 43.

(55) . Garlake, P.S. (1966), pp. 92-93.

(56) . House No.12 Songo Mnara; see ibid, p. 93 and

Fig. 74.

(57) . Klirkman, J.3. (4963), p. 20.

(58) . Existing evidence indicates that slaves, having

lived with their masters for so long, had
adopted their style of life. "They cannot
always be distinguished from their masters as
they are allowed to imitate them in dress and
in other particulars." See Richards and Place
(1967), p. xxii, quoting Captain Owen on
domestic slaves in Mombasa (c.1833).

(59) . Garlake, P.S. (1966), p. 87.

(60) . Ibid.
 102

(61) . P- 1°8 *

(62) . Dondo la recorded by Chittiok (1967), p.66, a a

being the site of a Portuguese settlement
previously thought to be in Tundwa on Paaa
island.

(63) . Childe, V.O. (195«), p.7.

(64) . Prin8, A.H.J. (1972), p.^\

(65) . Leinhardt, P. (1959), p. 231.

(66) . I M i . , p. 239.

(67) . The harass consists of benches placed at a

prominent position that act as a meeting forum
for elders.

(68) . They are: Kina Mti, Kfamao and Ungwana wa Yumbe.

I am indebted for this information to Sheikh
Ahmed al-Jahadhmi. The Lamu Chronicle (p.13)
mentions a fourth group named Nayublli or
Wayunbili.

(69) . Quoted in Salim, A.I. (1973), p. 27.

(70) . Travis, W. (1967), p. 55.

(71) . A sweet jelly made of brown sugar, flour and

ginger.

(72) . Strandes, J. (1968), p. 18, quoting a Portuguese

source. The Portuguese historian refers to the
string of be^ds as a rosary. In fact these have
no religious significance.
• t

(73) . An example of shared facilities is the kino or

stone for sharpening swords. It is said to
have been brought from Oman to Lamu three
hundred years ago. It was built into the plinth
of the house of the original owner and made
available for public use. Sword sharpening is
the prelude to a special kind of dance.

(74) . Garlake, P.S. (1966), p. 90.

(75) . M m P. 118.

(76) . Leinhardt, P. (1959), p. 232, describes the

traditional annual bull sacrifice ceremony in
Lamu in which a bull is led round the northern
sector of the town before it is slaughtered and eaten
communally. The ceremony is accompanied by the blowing
of horns.
 103

(77) . zeza na Asha Hamadi,

Wahuburinl malenga,
Ayao napige hodi,
mllango tumeifunga,
Uivumize Muyumbe,
siwa yetu ya Myeo,
Na yaandamane na pembe,
Yamshabaka furungu,
Uvumizapo Muyumbe.

(78) . Stigand, C.H. (1913), p. 124.

(79) . Mcluhan, M. (1964), p. 79.

(80) . Hall, E.T. (1966), p. 112, defines personal

space as "the distance consistently separating
the members of non-contact species. It might
be thought of as a small protective sphere or
bubble that an organism maintains between it­
self and others”.

(81) . I am indebted to Mr. J. de V. Allen for this

information.

(82) . Oarlake, P.S. (1966), p. 95.

(83) . Kirkman, J.S. (1964), p. 107.

(84) . Strandes, J. (1968), p. 131.

(85) . Quoted in Richards and Place (1967), pp.xvii

and xviii.

(86) . On t)ie day before the traditional bull sacrifice

ceremony the animal is led round the town
starting from Nna Lalo mosque. See Leinhardt, P.
(1959), p. 233.

(87) . L. Torres Baibas (quoted in Von Grunebaum, G.E.

(1955), p. iss ) describes the road nets of
Muslim towns as follows:
"Lea villes musulmanes possedaient aussi un
certain nombre de voies transversales ou
radiales qui mettaient en communication les
porteo o p p o s e s de 1'enceinte fortifee de la
Medina et squi se prolongeaient a travers les
faubourges immediats. Mais sur elles se
greffaient des rues 6troites et tortueuses d'ou
partaient a leur tour un grand nombre de
ruelles sans issue, qui se remifiaient a la
fa?on d'un labyrinthe, commes les vienees du
crops humain”.
 104

(88) . See Chittick’s article in Zamanl. ed. Ogot and

Kieran (1968), p. 112.

(89) . Burton, R. (1872), p. 40.

(90) . See Strandes, J. (1968), pp. 17, 25 and 66.

(91) . Quoted in Trimingham, J.3. (1964), p. 5n.

(92) . Van Grunebaura, G.E. (1955), p. 146.

(93) . There is no evidence to substantiate a remark

by Trimingham (1964), p. 146, that Swahili
towns "have streets of artisans plying special­
ized occupations".

(94) . Coupland, R. (1968), pp. 56-57 notes that "the

economic exploitation of the East African
coast proved difficult enough with the aid of
modern science. In the seventeenth and
eighteenth and early nineteenth centuries it
was impossible".

(95) . Prins, A.H.J. (1965), p. 32, describes the

village of Kipungant on Lamu island as "the
former country town of the big city (of Lamu)
... many Lamu merchants had their mansions
here". He does not the source.

(96) . Pate Chronicle (Stigand version), p. 45.

(97) . See Strandes, J. (1968), p. 117.

(98) . Iron working in Malindi i3 mentioned by the

twelfth century Al-Idrisi. See Freeman-
Grenville, G.S.P. (1966), pp. 19 and 20.

(99) . There is a group of economic historians who

believe that because East Africa’s role in
international trade was a passive one, under­
development is concomitant with it. See
Alpers, E.A. (1973)* passim.

(100) . Von Grunebaura, G.E. (1955), p. 142.

(101) . The Koran, chapters 2:221, 5:89, 24:33 and

58:3.
 105

( 102 ) . Trimingham, J.S. (1964), p. 74.

(103) Meek,
. C.K. (1925), pp. 1, 4 and 5.

(104) Chit-kick's
. article in Lamanl. ed. Ogot and
Kieran (1968), pp. 105-106. Chittick here
examines the exports of the coast mentioned
in the Periolus. The list includes coconut
oil which shows "that some people at least
were living in permanent settlements" to
attend to the coconut shamba.

(105) Oliver
. & Mathew (1963), Vol.I, p. 115.

(106) Chittick's
. discussion of Kirkman's paper
(1966), p. 253.

(107) Kenya
. Land Commission on evidence (1934),
Vol.III, pp. 2610-2611.

(108) For .Al-Idrisi's passage and Sacleux’s

interpretation see Whiteley, W. (1969),
pp. 28-29.

(109). There is evidence of influence from the

interior too but this is comparatively
slight. There is reference to what sounds
like the Swahili ivory slwa horn in Ibn-
Battuta's description of the palace of
Sultan Mana3l of Mali. Ibn-Battuta visited
Mali in 1353• A game popular on the East
African coast known as bau played on a board
with rows of holes where nuts are distributed
is similar to one played in Sierra Leone where
it is called mankala; see Parrinder, G. (1969),
p. 110.

(110). Child, V.G., p. 137. ( 1956).

f

(111). Mathew ' 3 remark (Oliver and Mathew (1963),

Vol.I, p. 118) that "the domes in Swahili
buildings are reminiscent of Indian Muslim
architecture" cannot be aocepted.

( 112). This is evident in preserved dance patterns

with spirit connections. The practice is normally
anathema to Islam.
 106

BIBLIOGRAPHY

Allen, J.de V., 'A Proposal for Indian Ocean Studies',

in Mwazo. Vol.2, Nfi 1. June 1969.

Alpers, E.A., The East African Slave Trade. Historical

Association of Tanzania Paper N2 3.

'Rethinking African Economic History',

Kenya Historical Review, Vol.l. N®2,
1973.

Al-Baladhiri, Futuh Al-Buldan. Cairo, 1901 (Arabic)

Boxer, C.R., and Azevedo, C. de, Fort Jesus and the

Portuguese in Mombasa. London, I960.

Burton, R.F., Zanzibar. City Island and Coast. 2 Vols.

London, 1872.

Chanler, W.A., Through jungle and desert. London, 1896.

Childe, V.G., Piecing together the past. London } 1956.

Chittick, H.N., 'Notes on Kilwa', Tanganyika Notes

and Rgcords, Nfi 53, October 1959.

'The "Shirazi" colonization of East

Africa', Journal of African History.
Vol.VI, N2 3, 1965.

'Discoveries in the Lamu Archipelago',

Azania. Vol.II, 1967.

'A new look at the history of Pate',

Journal of African History. Vol.X,
N2 3, 1969.
1
'The East coast, Madagascar and the
Indian Ocean c.800-1600', The Oxford
History of Africa (forthcoming).

Coupland, R., East Africa and its invaders. London,

1968 .

Creswell, K.A.C., A short history of early Muslim

architecture. London, 1958.

Al-Duri, A.A., 'The birth of lkta' in Islamic

societies', Bulletin of the Iraq
Academy. Vol.20, 1970. (Arabic).
 107

Encyclopaedia of Islam.

Fitzgerald, W.W.A., Travels In the coaatlanda of

British East Africa and the islands
of Zanzibar and Pemba. London, 1898.

Freeman-Grenville, G.3.P., The East African coast:

select documents from the first to the
earlier nineteenth century. London,
1966.

Frisk, H. (ed.), The Periplus of the Erlthrean Se a .

Goteborg, 1927.

Garlake, P.S., The early Islamic architecture of the

East African coast. Nairobi & London,
1966.

Ghaidan, U.I., ’The stone houses of Lamu', The Journal

of the Architectural Association of
Kenya. May-June 1971.

'Swahili plasterwork', African A r t .

Los Angeles, Winter 1973.

Greffulhe, H., 'Voyage de Lamoo a Zanzibar', Bulletin

de la Society de Glographle et d'Etudes
Coloniales de Marseilles. 11, 1878.

Grotanelli, V.L., Pescatori dell'Oceano Indlano. Rome,

1955.

Guillain, M., Documents »ur 1'hlstoire. la geographic

et le commerce de l'Afrique Orientale.
3 vols., Paris, 1856,

Hall, E.T., The hidden dimension. New York, 1966.

Harris,;!., Swahili poetry. London, 1962.

Hitti, P.K., History of the Arabs. London, 1971.

Ibn-Battuta, Tuhfat ... , Cairo, 1928. (Arabic).

Al-Idrisi, Surat ul-'ard. edited and annotated map by

M.B. Al-Athari and M. Jawad, Baghdad,
1970 (Arabic).

Jackson, F., Early days in East Africa. London, 1930.

Al-Jahiz, Three treatise (ed. Van Vloten), Leiden,

1903 (Arabic).
 108

Kenya Land Commission on evidence. Nairobi, 1934.

Khasbak, S., 'Ibn Jubair's journey ...', Iraqi

Geographical Journal. Vol.VII,
November 1971 (Arabic).

Kirkman, J.S., The Arab city of Ged i . London, 1954.

'Takwa', Are Orlentalls. Vol.II, 1957.

'Excavations at Ras Mkumbuu on the

island of Pemba1, Tanganyika Notes and
Records. N2 53, October 1959.

Gedi: the palace. The Hague, 1963.

Men and monuments on the East African

coast, London, 1964.

'The coast of Kenya as a factor in the

trade and culture of the Indian Ocean',
paper presented at The International
Maritime history seminar. Beyrout,
September 1966.

Kitab-ul-Zunuj, Cerulli, E. (ed.), Somalia, scrittl

vral edj,U ed, lnediti, Rome, 1957.

The Koran.

Lamu Chronicle, Hitchens, W., (translator and editor),

Bantu Studies, 12, 1938.

Lienhardt, P., 'The Mosque College of Lamu and its

social background', Tanganyika Notes
and Records. Bp 53, October 1959.

Al-Mas'udi, Muru.1 -ul-Lhabab wa M a 'adin-ul-Jawhar.

, Cairo, 1964 (Arabic).

Al-T^nblj^wal-lshraf, Leiden, 1891 (Arabic).

Mcluhan, M., Understanding media. London, 1964.

Meek, C.K., The Northern tribes of Nigeria. London,

1925.

Moholy-Nagy, S., Matrix of man: ,an illustrated,,histpry

qf urban environment, London, 1968.
 109

Mumford, L., Technics and civilization. London, 1967.

Ogot, B.A., and Kieran, J.A. (eds.), Zamani. Nairobi,

1968.
Olabi, S .A ., Thawrat-al-Zen.l. Beyrout, 1961 (Arabic).

Oliver, R., and Mathew, 0., History of Bast Africa.

Vol.l, London, 1968.

Pevsner, N., 'The exploring eye', The Architectural

Review, London, March 1967, pp.95-97.

Pate Chronicle (two versions): Werner, A., 'A Swahili

history of Pate', Journal of African
Society. XIV (1913-14), p. 149 ffj and
Stigand, (see below) Chapters 2, 3, & 4.

Parrinder, G., African mythology. London, 1969.

Patterson, G.D., 'Some notes on the ecology and

agricultural economy of the Lamu
hinterland', unpublished manuscript,
1955.

Posnansky, M. (ed.), Prelude to East African history.

London, 1966.

Prins, A.H. J., The Swahili-speaking peoples of Zan­

zibar and the East African coast.
London, 1967.

Sailing from Lam u . Assen, 1965.

Dldemlc Lamu. Gruningen, 1971-

Richards, C. and Place, J. (eds.), Hast African

explorers. London, 1967-

Roche, M . , Le Mzab. Prance, 1970.

Roolvink, R., (compiler), Historical Atlas of the

Muslim Peoples. D.lambatan (undated).

Salim, A.I., Swahili-speaking peoples of Kenya's

coast. Nairobi, 1973.

Al-Samir, P., Tharat al-Zen.1. Baghdad, 1971 (Arabic).

Strandes, J., The Portuguese period in Bast Africa.

Nairobi, 1968.

Stigand, C.H., The land of Zln,1. London, 1913.

 no

Sutton, J.E.G., The East African Coast. Historical

Association of Tanzania Paper Nfi 1,
1966.

'Early Iron Age pottery', Azanla.

Vol.I, 1966.

Al-Tabari, Tarlkh al-Rusul wal-muluk. Leiden, 1964.

(Arabic.)

Tipu Tip, Maisha ya .... trans. W.H.Whiteley, •

Nairobi, 1970.

Travis, W., The voice of the turtle. London, 1967.

Trimingham, J.3., Islam in East Africa. London, 1964.

Von Grunebaum, Islam. London, 1955.

Whiteley, V., Swahili: the rise of a national

language. London, 1969.

Wilding, R., 'The ancient buildings of the north

Kenya coast', Journal of the Archi­
tectural Association of Ken y a .

Ylvisaker, M . , Shamba na konde. Seminar paper presented

at the University of Nairobi, October
1971.
1 The East African coast
 L i J
2 Creek north of Lamu

3 Spread of Islam before the fifteenth century

 i H fl H 1 1

k The Ishikani tomb

top to bottom: west, east, south and
north facades

Lamu archipelago
7 Sheila Friday mosque- View from east
 a 0

s> ® r

9 © ® a

©
[
© 0 f-

© o © E
0 © . i
cm
™ —
*!fU

°__ l i__ J 4U

8 Shelia Friday mosque- Plan

9 Sheila Friday mosque- Section

10 Sheila Friday mosquo- View from ablution
11 Sheila Friday mosque- North facade

12 Sheila Friday mosque- East facade

 I AM U TO VK P O S S £ S S lO S \

T£ ro *N A O S S f* ± / O A S
P A \ tN t T O *N P O L S iS S r O * *

SfU ro w * r'CSS/’S i/ O v l

n r u U JLT A N A T € W S » I SS /< W S

rOS±£5 i/OHS Of nAr**K*Mi **

3 n rw

P The coastal hinterland of the archipelago

lA Matondoni Friday mosque
 1

D
D

J
MWANA

0 Ia j 4 1 6 M

15 Northern Swahili moaques- Comparative plans

-l£ Mosque in Shelia- Detail of raihrab
13 Laura tovax- Promenade
19 Lamu- The mud and wattle town
showing rope-walk

20 Lamu town- Piazza

22 7-nrm town- Strootacape
23 Lomu town- Ilitaa* Mosque foundation dates
(copied from dhrab inscriptions) are not
necessarily reliable
25 Lomu- Houoo in Ilkonanl, first floor
 1

27 Laxnu- Carved door

28 Sheila Friday mosque, entrance door
29 Lamu- Door motifs
30 Carved door from Dubai

31 House in Mkomani- Kiv/anda

32 House in Mkomani- Msana wa tini

33 House in Mkomani- Ms ana v/a yuu

 3^ House in Mkomani- Ngao showing
mwandi

35 House in Mkomani- View towards ndani

37 House In Shelia* Ruined zidaka
 *

I 39 Mzab- Interior
4-0 Al-Wasiti- Thirteenth century Meoapotamian miniature

1
*fl Church in Alahan, Anatolia- Detail of
window
k3 Swahili houses- Comparative plans
a7
44 Lamu town- Baraza

>

4-5 Lamu- Main street at Langoni

1
47 Lamu town- Boat launching

48 Lamu tovm- Main street

49 Lamu jamia - The Friday prayers

51 Sheila Friday mosque- Interior
 V '

52 Lanu tovm- Land use

53 Kilwa- Husuni Kubwa