Archaeological fieldwork at Songo Mnara, 2011

Archaeological Investigations at Songo Mnara, Tanzania: 2011 Field Season Preliminary Report Submitted to the Department of Antiquities, Republic of Tanzania 2013

Jeffrey Fleisher, Ph.D. Department of Anthropology Rice University Houston, Texas, USA Stephanie Wynne-Jones, Ph.D. Department of Archaeology University of York York, United Kingdom

Cover image: Carnelian beads from SM017 (context 17012) found with a hoard of over 300 Kilwa-type coins.

2

Table of Contents Introduction……………………………………………………………………………………………………………...……..9 Research Problem………………………………………………………………………………………….….………..….10 Methods and Research Procedures Conducted…………………………………………………..………..…11 Excavation Results……………………………………………………………………………………………..……….…..14 Trenches in domestic architecture…………………………………………………………..……...….14 SM016…………………………………………………………………………………………..………...15 SM017……………………………………………………………………………………………....…….17 SM018……………………………………………………………………………………………..……...19 SM021……………………………………………………………………………………………..….…..19 SM023…………………………………………………………………………………………….....…...22 Trenches in Open areas………………………………………………………………………………..……...25 SM019………………………………………………………………………………………………........25 SM020………………………………………………………………………………………………........27 SM022………………………………………………………………………………………………..…...29 Central Mosque and Adjacent Area………………………………………………………………..…...29 SM030/SM029…………………………………………………………………………………..…....29 Burial Excavations…………………………………………………………………………………………........31 SM024………………………………………………………………………………………………..…...32 SM025……………………………………………………………………………………………….…….35 SM026………………………………………………………………………………………………….....36 SM027………………………………………………………………………………………………….….36 Geochemical/Phytolith Testing……………………………………………………………………….…..37 Geophysical survey at Songo Mnara and Kilwa Kisiwani……………………………...…..….38 Archaeological Survey on Songo Mnara Island (Jack Stoetzel)…………………………………….…..41 Archaeobotany: Preliminary Field Report (Sarah Walshaw)………………………………………...….45 Maritime Survey of Songo Mnara in South-East Tanzania, June 2011 (Edward Pollard)…..51 References Cited……………………………………………………………………………………………………………..69 Appendix 1: Archiving and Storage………………………………………………………………………………..73 Appendix 2: Artifact Tables………………………………………………………………………………………...….75 3

Figures and Tables Introduction and Excavation Results Figure 1: East African coast Figure 2: Kilwa region Figure 3: Site plan of Songo Mnara Figure 4: Location of trenches SM016, SM017, SM018 Figure 5: Plan of SM016 Figure 6: Deposit of Kilwa-type coins, SM017 Figure 7: Plan of SM017 Figure 8: Grand staircase to House 34 Figure 9: Location of trenches SM019, SM020, and SM021, on northern wall of settlement Figure 10: Plan of SM021 Figure 11: Location of Trench SM023, in blue (with trenches from 2009 in green) Figure 12: Plan of SM023, with aragonite bead deposit in northwest corner Figure 13: Aragonite beads (above) from SM023, context 23014 (right) Figure 14: Profile of east baulk of SM023 Figure 15: Context 19006 and 19012 in SM019, lime manufacturing area Figure 16: SM019, profile of north wall Figure 17: Plan of SM020 Figure 18: Plan of excavated central mosque, SM030 Figure 19: Excavated central mosque (left); celadon bowl inset in mosque birika (right) Figure 20: Porites blocks (from mihrab) heaped in northeast corner of central mosque Figure 21: Location of burial excavation trenches (SM024, SM025, SM026, SM027) Figure 22: Quartzite offering stones from SM024 Figure 23: Test pit grid in western and central open spaces Figure 24: Possible open space surrounded by earthen houses; red dots indicate the presence of daub in test pits Table 1:

Skeletons Excavated

4

Archaeological Survey (Stoetzel) Figure 1: Environments of Songo Mnara Figure 2: Shovel test pit survey of Songo Mnara Figure 3: Positive shovel test pits Figure 4: Sites recorded at Songo Mnara

Preliminary Maritime Survey Report (Pollard) Figure 1: Kilwa estuary Figure 2: Survey routes taken by the author around Sangarungu Harbor Figure 3: Surveying the reef crest on Songo Mnara Figure 4: Sites around Sangarungu Harbour Figure 5: Copse where Mkuje graves are located Figure 6: Shermani where a German lived during the colonial period Figure 7: North side of Watiro islet Figure 8: An almost complete decorated pot on Sanje ya Kati (Kat009/11) Figure 9: Kwale Ware from Mikumbe, Pande Figure 10: Excavation of Sangarungu TP1 Figure 11: Sangarungu TP1 profile

5

6

Acknowledgements We gratefully acknowledge the financial support provided by many generous organizations. Archaeological research at Songo Mnara was funded by the National Science Foundation (BCS 1123091) and the Arts and Humanities Research Council (AH/J502716/1). Additional funding was supplied by Rice University Archaeological Field School and the Leverhulme Trust.

Geophysical survey at Kilwa Kisiwani was funded by the Society of Antiquaries of London. Edward Pollard’s maritime survey was partially funded by the British Institute in Eastern Africa.

Fieldwork was conducted in collaboration with the Antiquities Department of Tanzania, and we are very grateful to the Director, Mr. Donatius Kamamba, for his assistance. We also acknowledge the help of our Antiquities representative, Mr. Revocatus Bugumba. Mr. Mohammed Chidoli, also of the Kilwa Antiquities office, provided additional support.

Dr Kate Welham and Ms Charlene Steele of Bournemouth University were responsible for the geophysical survey; GIS data collection and site mapping was done by Mr. Mark Dover.

Geoarchaeological work was completed by Dr Federica Sulas of the University of Pretoria. Dr Kate Robson Brown and Francesca Migliaccio of the University of Bristol analyzed the human remains. Sarah Walshaw of Simon Fraser University oversaw paleobotanical sampling and flotation with the assistance of Mr. Dominic Pistor. Jack Stoetzel conducted pedestrian survey and phytolith sampling on the island. Dr Edward Pollard carried out a maritime survey of the foreshore and tidal zones. Prof. Heinz Ruther, Roshan Bhurtha and Stephen Wessels conducted laser survey of many of the standing ruins at the site, continuing work from 2009, on behalf of the Zamani Project; we thank them especially for the excellent GIS that they have created for Songo Mnara and the Kilwa region.

A hearty thanks to all the graduate and undergraduate students, for their hard work and good cheer: Nathan Anderson, Grace Apfeld, Juliette Chaussen, Daisy Chen, Tom Delany, Kylie Klein, Hannah Leighton, Khadija McBain, Erick Miller, Courtney Ng, John Perkins, and ErĂŠndira Quintana Morales, Sheena Shah-Simpson, and Amanda Wicker.

Finally, we want to thank the residents of Songo Mnara Island who so graciously host us each field season. We would not be able to do our research without their hard work; we are thankful for their hospitality each time we arrive on their amazing island. Our camp was run ably by Ahmed Kassim who made sure delicious food and ample water always awaited our return. A hearty thanks also to Bw. Polo for his remarkable problem-solving abilities. For assistance in our camp we would like to thank: Zurafa Mpate, Time Bakari, Amina Saidi, Hailaka Mohamedi, Sofiya Mopo, Rukiya Juma, Fauziya Mohamed, Abdallah Twalibu, Nurubini Omari, Muhamedi Mwinyi, Omari Hakimu, Yahaya Mohamedi, Pipo Sulieman, Oga Mohammedi, Zubere Mohammedi, Hatibu Mburula. For work on the excavations, we thank: Mohamed Yahaya, Abdu Changi, Mahibu Saidi, Rahema Abdurahman, Bakari Mfalame, Toreha Hamad, Rafa Tunda, Selemani Mohamed, Ahmad Issa, Yusuf Changi, Rithwani Mohamed, Sulieman Saffi, Sulieman Masoudi, Hasani Kinyaga, Sulieman Abdallah, Ali Mohemedi, Aisha Ahmadi, Hamisa Mshehe, Mzee Nabu Hamisi, Zenabu Hamisi, Makame Mbwana, Yahaya Kesse, Shaibu Mohamed, Saidi Mushaha, Issa Hasani, Suliemani Kachi, Juma Mwinyi, Hawa Mohamedi, Salima Saidi, Mzee Majida, Zubeda Amuri.

7

8

Introduction This report details research carried out at the site of Songo Mnara in southern Tanzania from June 6 to July 18, 2011. This project focuses on the use of space as the setting for daily practice at this ancient urban center, investigating public (centrally-located religious architecture, cemeteries, and open areas) and private (stone house) spaces. In comparing public and private spaces, this project aims to delineate particular material practices related to the constitution of social identities through domestic activity and through ritual/ commemorative practices associated with centrally-located cemeteries and open areas. Archaeological research at Songo Mnara began to address these issues in 2009; this report summarizes findings from the second season of research.

Songo Mnara, nestled in the Kilwa archipelago on the southern coast of Tanzania (Figs. 1 and 2), has proven to be a fantastic location to address these theoretical concerns. Despite its prominence and excellent preservation, most of the work completed there by previous investigators focused on recording the extensive architecture (Dorman 1938; Garlake 1966), or conducted limited, exploratory test excavations (Chittick 1961; Mathew 1959; Pradines and Blanchard 2005a, 2005b).

Most research in the region has been focused on the more famous and well-known site of Kilwa Kisiwani (Chittick 1974; Sutton 1998; Fleisher 2004; Pollard 2008; WynneJones 2007).

The site of Songo Mnara is dominated by the well-preserved remains of more than 40 large domestic room-blocks, six mosques, and numerous tombs (Fig. 3).

Room blocks wrap around and enclose an open, central area of the site where tombs, a walled cemetery and a small mosque are located. The site was occupied for a maximum of 200 years, between the late 14th Fig. 1: East African Coast and early 16th centuries. This relatively short lifespan (in comparison with the 800-year occupation of Kilwa) means that the stratigraphy is simple, with little vertical complexity, making it an ideal site at which to explore horizontal complexity in the form of household and public spaces.

9

Research Problem Research at Songo Mnara adopts a spatially-integrated approach to Swahili economic and ritual/religious practice that is allowing us to isolate both deliberately-planned elements of the town (central cemeteries and open spaces) and elements that might have been created through exigency, particularly in the realm of domestic architecture.

Our ongoing research explores the urban space of Songo Mnara at four interlinked scales: (1) household activities, pursued through systematic excavations within and around buildings at the site; (2) public and communal practices investigated through geoarchaeology, geophysical survey, and excavations across the open areas and monuments of the site; (3) the town plan, described and accurately plotted through laser scanning and mapping; and (4) the broader landscape within which the site sits, explored through off-site terrestrial and maritime surveys and geophysics. This multi-scalar research at Songo Mnara will contribute to discussions of urban planning in Swahili and other urban contexts. The exceptional preservation at Songo Mnara is allowing for the construction of a robust dataset to explore the way that Swahili towns were both planned and unplanned, with their components are offering an insight into the priorities and social negotiations of their inhabitants.

Fig. 2: Kilwa Region This work will contribute substantially to discussions of organizational principles and levels of meaning in ancient town plans more generally, and especially to research on aspects of town layouts that were recursively linked to movement and activity. By thinking through the function of city spaces, in terms of how places were used and experienced at all levels, we supplement the regional understandings of urban 10

character with an appreciation of the spaces of the towns themselves; we simultaneously bring the insights from study of the regions into the town. Swahili towns are, we suggest, ideal contexts in which to explore the emergent qualities of town plans as a nexus of various levels of meaning, incorporating ‘high-level’ concerns of religion and cosmology as well as planning principles that stem from the needs and negotiations of groups and individuals within society. In this way, Swahili town plans offer a dramatic case study in which to break down the ‘false dichotomy’ between planned and organic, and explore the importance of spatial practice in the negotiation of global and local economic and religious practice.

Methods and Research Procedures Conducted The field work at Songo Mnara in 2011 aimed to build on the successes of the 2009 season, exploring the open spaces more intensively (based on geophysical surveys completed) and begin systematically testing deposits within houses. Our goals included:  Comparative excavations in stone houses, sampling entrance rooms, courtyards and back rooms  A systematic test pit program in the central and western open area  Finish geophysical survey in the town itself  A pilot study of burials in the town Exploration of the anthropogenic features in the island landscape, through terrestrial and maritime surveys on the island.

This report will describe what we accomplished, and our preliminary interpretations.

Excavations and Artifact Processing Trenches were dug in natural layers, with trowels, shovels, and hoes. All deposits were sieved, using a 2mm mesh. This proved the most effective way of recovering small artifacts, like beads, fish bones, and coins. Artifacts designated ‘special finds’, such as imports, coins, beads, metals and any other rare or unusual objects, were recorded in situ and their locations mapped.

As in the 2009 season, the project used a modified version of the Museum of London Archaeology Service single-context recording system (MoLAS 1994), recording all contexts and cuts on context recording sheets. Each new layer or feature was given a separate context number; context numbers were recorded consecutively within each trench (i.e., context 10003 is the third context recorded in Trench 10). Additional recording sheets documented coral architecture, and environmental samples. Scale plans and profiles were drawn of all layers; contexts were also photographed. In general, trenches were dug to sterile soil, unless otherwise noted below.

All artifacts were bagged in the field, labeled with contextual information, and then returned to the field lab, where they were washed, sorted, and preliminarily recorded. Some artifact classes, like imported pottery, were analyzed and catalogued in the field lab. Other materials, like beads, were exported for further study and more detailed recording. In the field laboratory, local pottery was sorted into diagnostic and undiagnostic sherds, all artifacts were counted and weighed, and then undiagnostic sherds—undecorated bodies only—were 11

discarded, placed into excavated trenches prior to backfilling. All stored artifacts were placed in archival quality plastic bags, as well as bug-proof cotton bags; these were organized by trench in wooden boxes and stored in the storeroom at Songo Mnara, alongside finds from 2009.

Laser Scan Survey A team of surveyors from the University of Cape Town and the Zamani Project, directed by Prof. Heinz Ruther, conducted laser scanning at the site during the first two weeks.

This was a continuation of work carried out in 2009. Hundreds of scans were made of the standing architecture. These data will be used to construct plans of the site as well as 3D reconstructions of particular buildings.

Geophysical Surveys Geophysical surveys were conducted by a team from Bournemouth University, directed by Dr Kate Welham. The work in 2011 included a smalls area at Songo Mnara using magnetometry and electromagnetic survey. The geophysical team also worked at Kilwa Kisiwani, experimenting with different geophysical survey techniques, including ground penetrating radar. The latter survey has been reported in the Journal of African Archaeology (Fleisher, Wynne-Jones, Steele and Welham 2012)

Fig. 3: Site plan of Songo Mnara, with mosques and open areas noted 12

Geochemistry and Geoarchaeology Geochemistry and geoarchaeology were coordinated by Dr Federica Sulas, following up on research completed in 2009 (Sulas and Madella 2012), and included the test pit sampling program in the central and western open areas. In each of these test pits (see below for complete description and maps), 20g of soil was taken from the main cultural layer; part of each sample was sent to the ASL Chemex Laboratory in Seville Spain for ICP-AES multi-element analysis (four acid digestion MEICP61). Phytoliths will be extracted from the remaining soils from each sample.

Flotation/Botanical Remains In order to recover palaeobotanical remains, the project systematically took 2-5 liter soil samples from excavations and subjected them to flotation. This procedure, overseen by Dr. Sarah Walshaw of Simon Fraser University (who was assisted by Mr. Dominic Pistor), allows for the recovery of charred plant remains as well as other macroscopic artifacts. The samples from this work are currently being processed and analyzed by Dr. Walshaw.

Faunal remains All bones were collected, and sieving ensured good recovery of fish as well as mammal bones. The mammal bones are being analysed by Dr. Thomas Biginagwa of the University of Dar es Salaam. The fish remains will be studied by Dr. Erendira Quintana Morales of the University of Bristol.

Finds analysis Ceramic analysis was conducted in the field by Dr. Freda Nkirote of the National Museums of Kenya. The work continues at Rice University, under Jeffrey Fleisher’s direction. Beads are being analysed by Dr. Marilee Wood, University of Uppsala. The coins are being identified

by John Perkins of the British Museum/University of Bristol (see Fleisher and Wynne-Jones 2009; Wynne-Jones and Fleisher 2012).

Maritime and Terrestrial Surveys of Songo Mnara Island Two surveys were carried out in 2011: a sample transect survey by Mr Jack Stoetzel of the University of Virginia, and a maritime survey by Dr Edward Pollard. The terrestrial survey was part of Stoetzel’s PhD research, and aimed at recovering phytolith samples from across the island to investigate ancient landscapes and changes to it. In the process, he also recorded a number of sites, visible on both the surface in the shovel tests that he dug. This has been reported fully in Nyame Akuma (Stoetzel 2011), a copy of which is included in this report. Dr. Pollard carried out an intensive maritime survey of the island, walking the coastline and foreshore, as well as some limited diving to investigate features in the surrounding waters. This survey has been documented and discussed in the Journal of Maritime Archaeology (Pollard, Fleisher and Wynne-Jones 2012).

13

Fig. 4: Trenches SM016, SM017, SM018

Excavation Results What follows is a summary of each excavation trench dug in the 2011 season. In all, twelve trenches were excavated across the site. Six trenches were located within domestic architecture (SM016, SM017, SM018, SM020, SM021, SM023). One trench was located just outside the walls of these houses, spanning an enclosed open space (SM019). One trench was excavated in the southern open area (SM022) as a guide to sediments for the test-pit survey.

Four trenches were excavated into burials (SM024, SM025, SM026, SM027) at different parts of the sites. The central mosque was excavated as a single excavation unit (SM030).

Trenches in domestic architecture A variety of different rooms were sampled across the site (Fig. 4), building on work in 2009 which had identified differing patterns between two houses on the northern and southern sides of the site. Although this was a sampling program, the focus on spatial practice meant that rooms were excavated in their entirety: test pitting would be inappropriate here.

This aspect of sampling is therefore slower than the test pits elsewhere on site, and is likely to continue into future seasons. In total, six units (SM016, SM017, SM018, SM020, SM021, 14

SM023) were excavated in the houses/domestic spaces (house numbers are those assigned by Garlake [1966]):  Two entrance rooms were excavated in Houses 31 (SM016) and 34 (SM018)  Two back rooms were excavated in Houses 31 (SM017) and 40 (SM023)  One unit (SM021) explored a back room associated with a courtyard. The courtyard was also excavated (SM020) One of the entrance rooms, off the courtyard of the largest and grandest house to the north of the site (SM018) was disappointing as it became obvious that it had been excavated previously, a fact that had been obscured by the falling in of its ceiling after clearance. This likely occurred during the work of the 1950s/60s and the renovations of Mr Charles Little. In fact, a thicker plaster floor overlaid the original, cracked floor, and was probably laid by the conservators. The other front room, however, was extremely densely packed with materials, and the remains of multiple domestic activities. As in 2009, a buried ceramic pot was found to one side of the entrance, containing a rich sediment which was sampled and floated. The entire room was also full of ashy deposits and ceramic concentrations, suggesting a wealth of activity in this entrance hall.

The back rooms, in contrast, were both largely clear above floor level. The numbers of ceramics were low in both units, and suggest that the plaster floors were kept clean. Both had a drainage sump in the centre of the room down which would have helped in washing them down. Beneath the floors, though, both rooms yielded buried deposits which had been plastered into the foundations. A hoard of 361 Kilwa-type coins was found beneath the floor of SM017, accompanied by a carnelian necklace. SM023 contained instead a small pit into which eight large aragonite (clam shell) beads or spindle whorls had been placed. These comparable but different deposits suggest deliberate acts of structured deposition that might be linked to the meanings and value of the houses, and their founding as domestic spaces.

The courtyard and associated back room of House 47 were much more reminiscent of the open area excavations, and directly correspond to those undertaken this season in SM019.

As outside, the courtyard contained a dispersed midden associated with domestic activity and cooking, with some clear areas of burning but no clear concentrations or midden pits. In fact, the deposits were less rich than those immediately outside the wall of this courtyard, suggesting that debris was simply thrown over into the external space.

Excavations of the houses are therefore slowly building a picture of both diversity (in the uses of different rooms) and shared practices (such as the structured deposits) across the site. We are also beginning to be able to speak to the different uses of the range of space across the site, and the ways that domestic activities were defined by, and overspilled the bounds of, the domestic structures themselves. Future seasons will continue this process, and explore as many as possible of the range of houses at Songo Mnara.

SM016 (supervised by John Perkins) SM016 was an excavation unit of approx. 8 x 1.80 m, excavated in the entrance room of House 31 and based on the size of the room. A baulk of approximately 40cm was left on the southern side to offer support to a tilting wall (this was done on the advice of Mr Pierre Blanchard, who also gave instructions for the wall to be shored up using mangrove poles, work 15

that was carried out by Mr Idiphonse Mloka), and 30cm at the western and eastern ends to afford a trench profile after excavation. The room had doors on all four sides, with the main entrance to the house on the northern edge, and a large door directly opposite leading into the house. At the western end, the door was significantly higher than the former ground surface; traces of a line in the plaster of the wall suggest there might previously have been a wooden stairway at this end.

The ground surface before excavation was quite uneven, and significantly higher in the western end, relating to different quantities of fallen rubble and debris in the different areas.

This fill had also unevenly impacted on the buried deposits, making the stratigraphy of the trench difficult to recognize in excavation. The plaster floor and the fills that would have underlain it were encountered at different levels in different places. The sandy fills that underlay the floor were replete with pits and buried deposits, which resulted in a large number of contexts recorded. The overall stratigraphy became clear only in retrospect and in reference to the trench profile.

In general, the sequence was of a floor layer overlain by a loose rubbly soil containing large pieces of coral rag and porites fragments (#16001), which was broken down into a series of essentially-similar context numbers to allow spatial control between the eastern (#16002) and western (#16003) ends of the unit and the area in front of the door (#16005). This layer contained most, if not all, parts of a carved coral arch from the doorway at the western end of the room. Beneath the rubble, a semi-compact sandy silt deposit (#16004) was encountered, which contained many fragments of pottery, bones and shells. Many of these finds showed

Fig. 5: Plan of SM016 16

signs of burning. Towards the bottom of #16004 a pottery lens was discovered along the southern section, which also contained some bone and shell (#16008).

The plaster floor (#16011 and #16014) was only visible in small, discrete areas of the room, having been substantially demolished by the rubble and fill above. Across the rest of the unit, a series of different sediments were recognized, apparently representing lenses and deliberate pits in the sandy fills that had underlain the floor. The sand here was rich in ash and micro-charcoal, as well as ceramic and bone debris. This is likely due to a combination of debris left by the builders of the house, and re-deposition of debris contained within the sands prior to construction. The western end of the trench was particularly rich in debris, perhaps representing a locus of activity during construction of the house; two lime burning pits were found at this end in the lowest levels (#16040 and #16041).

Some deliberate deposition was visible in amongst the more general debris. In particular, #16006 beneath the doorway was a clearlydelineated pit of reddish sandy silt, containing pottery, bones and shell. Along the north wall on the western side (Fig. 5), a pit (#16013) had been excavated into the foundations of the wall which contained a complete pot (#16018).To the west of #16014 along the north wall a sandy patch was assigned context #16013.

The sandy fill that underlay the floor had a series of different layers to it, recorded as separate contexts and with varying amounts of faunal and ceramic debris and evidence of burning. Eventually, clean yellow sand was encountered, deemed the natural subsoil (#16044).

SM017 (supervised by ErĂŠndira Quintana Morales) The eastern back room of House 31 was also excavated in its entirety as unit SM017. The trench was approx. 4 x 2.5m, covering the extent of the room except for a 30 cm baulk left for a section profile on the east wall. A doorway on the west end of the north wall leads into the middle room of House 31.

As elsewhere, the top layers of the unit consisted of a rubble fill, containing coral and plaster fallen from the ceiling and walls (#17001). A thin layer of sediment (#17002) was encountered beneath this fill and above the plaster floor, with the area around the doorway (#17003) separated off to give spatial control. These layers were quite poor in finds, suggesting that the plaster floor of the room 17

Fig. 6: Deposit of Kilwa-type coins, SM017

Fig. 7: Plan of SM017

(#17004) was kept largely clean. Although now quite damaged, the floor was encountered in patches across the unit, particularly in the east where the plaster was thicker due to an underlying drainage sump on the eastern end. A triangular sandstone with bored hole covered this sump, and was set into the plaster floor (#17005). The rubble within the doorway lintel was cleared (#17006) and the steps sent into the doorway were identified as context #17008. The steps were made up of large coral blocks covered in plaster. There was a concentration of artifacts, including 3 coins (SF17012-14) just inside the doorway (#17007).

The floor was supported by a layer of compact rubble (#17009) atop a sandy fill (#17013). These surrounded the plaster sump structure (#17014), which extended into the eastern wall of the room, draining in that direction. Two buried deposits were excavated as part of those fills. The first was a buried deposit containing 361 Kilwa-type coins and a cache of carnelian beads (Figs. 6 and 7; #17011 and #17012) placed into the floor fills in the northwestern corner of the room, to the west of the door. Another pit on the southern wall (#17010 and 18

#17016) contained quantities of pottery. Otherwise, the floor fills contained low quantities of artifacts. The basal layers consisted of a layer of porites chips (#17017), above the sterile sand that constitutes the natural subsoil here.

SM018 (supervised by Courtney Ng) A unit of 3.5 x 2.0 meters was excavated in the front room of House 34, a particularly grand house that faces the Great Mosque (Fig. 8). This unit, SM018, covered the western half of the front room, looking onto the stepped court of the structure. A 30 centimeter baulk was left on the southern wall for later profiling. The room was extremely rubbly, with substantial quantities of coral and plaster fallen from the ceiling and walls (#18001). This contained some interesting finds, but concealed the fact that the room seems to have been cleared previously, likely during the 1950s/1960s campaign of work here. This was evidenced by extremely low quantities of artifacts found Fig. 8: Grand staircase to House 34 during excavations, and confirmed when a comparatively modern plaster floor (#18013) was uncovered, somewhat above the original floor of the room. The fills above this floor had been recorded as contexts 18001 – 18012, different types of fallen debris that contained human teeth, glass and daub, perhaps from the roof of the house.

The original floor of the room (#18016) was a much thinner layer, and had been covered with white sand (#18015) to provide a base for the restored floor. The original floor had instead sat upon sandy fills (#18017) and a thick layer of compacted red earth (#18019). Although these fills were excavated to a total depth of >2m, the fills were largely sterile and only various types of sand were encountered (#18020 – 18024). It must be assumed that any previous archaeology was cleared at the time of restoration, but also that the floor was always supported by fairly clean sand.

SM021 (supervised by Sheena Shah-Simpson) Trench SM021 was a 3 x 4m trench located in the back room of House 47, on the northwestern side of the site. The room is interesting in that the only entrance is from an open courtyard (also excavated as SM020) rather than from the other rooms of the structure. It is not completely clear how the rest of the structure was accessed, but it seems to have been a rare example of a house with an entrance to the south (Fig. 9). House 47 is part of a complex of rooms and courtyard spaces that sit next to the town wall on the northern side. The room excavated as SM021 is believed to have had a makuti roof due to a sloping top to the walls here. The trench covered half of the room (Fig. 10), including the space directly in front of the entrance.

19

Fig. 9: Location of Trenches SM019, SM020, and SM021, on northern wall of setttlement

The room had been used to store lime plaster during the restoration campaign of 2008, and a layer of lime had to be removed from the surface before excavation. Beneath this, the unit had a simple stratigraphy composed of sandy topsoils above a red packed earth floor (#21008). Although these topsoils were excavated in a series of spatially discrete stages, there was little variation evident in the concentrations of finds through the sediments; contexts were distinguished mainly by color. Context #21001 was a relatively rubble free layer of loose brown topsoil consistent across the trench, containing low numbers of beads, a coin, quartz, fish bone, and high concentrations of shell and pottery. Beneath this, a rubbly layer was divided into four contexts (#21002, #21003, #21004, #21005) to give some spatial control (Fig. 10), although there proved to be little difference between these contexts. All had large concentrations of pottery, some beads, coins, quartz, charcoal, and shell. It seems likely that the differences in soil were superficial ones resulting from the rubble.

Underneath contexts #21002-5 was a layer of sandy soil. This was excavated as two contexts: #21006a and #21006b, but there was no discernible difference in the number and type of artifacts coming from #21006a, the western half of the trench, compared to #21006b, the eastern half. The northwestern corner of the trench contained different sediment, recorded as #21007 and containing a few beads, a spindle whorl, a coin, and some quartz, as well as pottery and shell.

Beneath these ultimately quite similar deposits, a red soil covered the entire trench.

20

Fig. 10: Plan of SM021

This was interpreted as a packed earth floor (#21008) although the loose and sandy nature of the sediments here meant that it was difficult to define a floor surface. Beneath the floor, the natural yellow sand was encountered (#21009). This contained some lenses of differentcolored fill (#21010 and #21011) but became more sterile as it went down. One interesting find was a set of young goat bones in the sandy fill (#21012).

In general, the sediments in SM021 were poorly-delineated. Finds were quite consistent between contexts, packed into the layer interpreted as a floor as well as the fills that overlay it. This was due to their very loose nature. No features were encountered.

21

Fig. 11: Location of SM023, in blue (with Trenches from 2009 in green)

SM023 (supervised by Grace Apfeld) Trench SM023 was a unit of c. 2.5 x 3.5 m, located within the back room of House 40.

The unit covered most of the room, with a small baulk left on the eastern wall for a soil profile (Fig. 11). The room was cleared of a deep rubbly topsoil (#23001) which contained occasional finds including several coins. The doorway was also cleared (#23002) and a thin layer of packed red soil was found to overlie the coral steps down into the room (#23003).

Beneath the rubbly layers, excavation quickly encountered a packed earth floor, easily distinguished as it was bright red in color (#23007). Along the wall to the east of the doorway a loose sandy deposit overlay this floor (#23006) and produced a large concentration of ceramics and a further 10 coins. In addition, the floor held a central drainage stone (#23008); the limey plaster that provided the conduit for this drain had broken up and was visible in the eastern side as a compact grey sediment (#23005) which continued down through the belowfloor fills.

The floor was supported by a series of fills that surrounded the plaster drainage channel. These were excavated as contexts separated by color and consistency into #23009, #23010, #23011, #23012 and #23015 (Fig. 12). All contained low levels of artefacts, perhaps redeposited. One particular concentration of bones (mainly rodent) was recorded as context #23013. These overlay a lower compact earth layer (#23016): either an earlier floor or part of 22

Fig. 12: Plan of SM023, with aragonite bead deposit in northwest corner

the packing for the upper floor. Excavation of this compact layer revealed several special finds including quartz, spindle whorls, several lamps, beads (ivory), pottery (Husuni Modelled Ware), and bone as well as under-floor deposits. Notably, a large overturned pot (#23014) covered a pit containing 8 aragonite spindle whorls (#23016; Fig. 12—13). These lay in packed red earth in a round pit excavated into the white sand beneath (#23017). Elsewhere a similar and perfectly round pit excavated into the fill (#23019) and with a hardened clay edge (#23013) contained only packed red earth and a few pieces of pottery and shell.

As in other units, the sandy fill beneath the floor contained a series of lenses and different fills (#23018, #23020, #23021) and a further deposit of red clay (#23024). This all overlay the white sand subsoil (#23025; Fig. 14).

23

Fig. 13: Aragonite beads (above) from SM023 context 23014 (right)

Fig. 14: Profile of east baulk of SM023

24

Fig. 15: Context 19006 and 19012 in SM019, lime manufacturing area

Trenches in Open areas SM019 (supervised by Kylie Klein) Trench SM019 was a 13.2 x 2 meter trench that stretched across a courtyard between houses 47 and 48 (Fig. 9). It ran from the western external wall of House 48, to a wall stump that may have served to divide the larger courtyard from one behind House 47. Geophysical survey in 2009 revealed a large anomaly through which the trench cut. In general, the trench revealed a scatter of materials related to both productive activities (spindle whorls) and trash deposition. At the center of the trench was located a large area of burned coral that likely

Fig. 16: SM019, profile of north wall 25

represents the remains of a lime manufacturing area.

The first, topsoil layer (#19001) was composed of fine sandy silt covering a majority of the trench extending from the western coral wall. Extending out from the eastern wall was fine silty sand fill with much coral rubble (#19002). Below context 19001 was a medium silty sand fill that was arbitrarily divided into three parts, which later became four, in order to gain spatial control over the trench (from west to east: #19003a, #19003b, and #19003c). Context #19003b seemed to produce fewer artifacts than #19003a and #19003c. Context #19002 sealed 19003d , a layer which contained a higher frequency of artifacts than others. A lens of reddish sand with few artifacts was excavated along the southern trench wall (#19004). In removing contexts #19003a, #19003b, and #19003c two darker patches were exposed which contained high concentrations of coral rubble and lime (#19005, #19006).

Context #19005 was cut by the northern wall of the trench, at the eastern end. It included lenses of soft coral rubble, charcoal/ash, and lime. Context #19006 had a similar composition as #19005 but higher concentrations of charcoal/ash. Initially we interpreted these two contexts as areas for manufacturing lime. After defining and excavating #19005 and #19006 we switched to new context numbers in the rest of the trench, despite little change in fill; this was a way of controlling the dominant fills in the trench, and noting the point at which contexts #19005/06 were noted and excavated. The fill along the western coral was a sandy context much coral rubble, likely from wall fall (#19007). In this context were found several spindle whorls, including one very large aragonite one (SF#19057), perhaps indicating the deposits resulted from people throwing waste over the wall or thread production in the area.

Along the eastern coral wall was a similar context defined by large amounts of rubble (#19008). The contexts in this area contained a relatively higher density of artifacts, including #19008 and the layers above it (#19002, #19003d). In between #19007 and #19008, two contexts were defined, arbitrarily divided into 19009a and 19009b to gain horizontal control over the trench. Since we interpreted #19005 and #19006 as two separate lime making pits, we excavated them individually based on what we thought was their extent (Figs. 15 and 16).

Context 19005 was removed and its cut recorded (#19011). Between the two possible lime making pits was an intact local pot, perhaps a bowl, suspended in the trench’s northern wall. While excavating #19009a and #19009b we revealed another layer of coral rubble and lime plaster (#19012) that seemed to connect #19005 and #19006. This new context was interpreted as being part of a larger lime making pit comprised of the three contexts together (#19005, #19006, #19012) or as a rubble discard area between the two smaller lime making pits (#19005 and #19006). Context #19012 contained a large amount of charcoal. At the base of #19009b, in the transition to subsoil, we revealed a lens of darker sediment; similarly, at the base of #19009a, we came down on a lens of reddish sediment with pieces of burnt coral. The soil discoloration may have resulted from coral associated with the lime burning area that was not heated enough to become lime. Cut into subsoil near the western wall was a shallow lens of grey sediment containing a high density of charcoal (#19015). The final context recorded was #19016, which was subsoil.

A couple of general interpretations of the contexts in this trench might be offered.

First, the trench revealed an area of coral burning, likely related to the production of lime.

Whether this lime was used in the adjacent houses is not clear. It is likely that the contexts 26

Fig. 17: Plan of SM020

related to this (#19005, 06, 12) were, in fact, all part of a single coral burning mound: the more intense burning seen at the center of these area (#19012) would result from a mound of wood and coral. The distribution of artifacts across the space, with evidence of denser deposits of artifacts at the eastern and western extent may indicate a number of practices: the disposal of material directly outside the walls of houses, a greater intensity of activities along the walls of the houses, or the sweeping of the central part of the courtyard, resulting in the buildup of artifacts at the margins.

SM020 (supervised by ErĂŠndira Quintana Morales) SM020 was a 5 x 2 meter trench in a semi-internal courtyard of House 47, on the northern wall of the town (Fig. 9). The trench was located west of SM019, which was staggered south, just outside the east wall of the courtyard. These trenches were meant to pro27

vide a long cut between an exterior and interior courtyard space, and to compare deposits from interior and exterior spaces. In this regard, SM021 is also related to these trenches.

A topsoil layer (#20001) was removed from the entire trench, after clearing grasses from the surface. This exposed an area of darker soil with charcoal inclusions (#20002) on the northern edge of the trench, close to the eastern end. Higher concentrations of bone were noted in this context, and relatively complete iron knife was collected (SF20002) from the edge of the trench. At this point, the trench was extended north, now abutting a wall stump, to include an additional 1.5 meters, making the trench now 5 x 3.5 meters in size. This extension allowed for better definition of context #20002, and to include the north doorway, now at the western end of the new trench boundaries. In the topsoil from this extension (#20003), there was a concentration of bone in the NE corner. It is possible that this context post-dates the main occupation of the town; the presence of coral rubble beneath this context, likely from wall fall, supports this conclusion. The rubble layer (#20004) was restricted to the eastern end of the north wall, and probably represents the fallen upper part of the north wall.

There was a small continuation of the dark charcoal soil (#20002) found over and within the southern edge of the wall fall (#20004).

After removing the rubble, a concentration of large shells was found in the corner of the north and east walls (#20005). In the trench extension, a homogenous layer of brown silty sand, coterminous with #20001 was removed. With this context removed, a series of welldefined contexts were mapped and excavated (Fig. 17), including: an orange-tinged sandy loam with frequent worn shell inclusions covering the east end of the trench (#20006); a reddish sandy loam (#20007) extending over most of the trench, interrupted at the western end by a lighter silty, fine sand (#20008), which ultimately was determined to be the top of subsoil.

The original extent of #20008 covered most of the western end of the trench but was redefined and restricted to a smaller area along the middle of the western edge of the trench. A small context of dark fill along the western edge of the trench was also mapped (#20009).

After beginning to excavate #20007, a small- to medium-sized circular feature of porites rubble (#20010) within this context. After clearing around the rubble, it was determined that #20010 was possibly a pit, as it reduced in size as it was excavated. Underlying the rubble was an area of dark gray silty sand with dark patches of charcoal. This context was smaller in size as it was excavated, and sealed subsoil. Other areas of #20007 sealed fragments of plaster floor (#20011), in the SE corner, and small sections extending out of the north and east walls. At the southern edge of the trench, an area of brown soil with turtle shell fragments was also defined (#20012), which also sealed subsoil. Finally, an area of brown soil with higher concentrations of bone was located (#20014) next to a small circular area of dark ashy fill (#20013).

At this point, subsoil had been exposed in most of the trench except along the north and east walls. The northern area contained reddish sandy loam (#20007) around and beneath #20013 and #20014. Context #20007 was deepest along this wall. The soils of this context in front of the doorway were recorded separately (#20007B), as higher concentrations of artifacts were noted. An ashy deposit (#20018) was located while clearing #20007 along the north wall, beneath #20013 and #20014. The area along the east wall was characterized by a mottled brown-orange gritty sand, similar to #20006 but with few artifacts (#20015). When 28

local pottery sherds were found in this context, in the NE corner, the context was removed as #20016. On the southern end of the eastern wall an area with shell concentrations and local pottery (sometimes stacked), was distinguished from the fill above (#20007), and recorded as #20007A. This concentration extended to the north under and around #20006 and probably meeting the edge of #20016, although the boundary was not very clear.

Dark gray soil was found under both #20016 and #20007A (#20019A, #20019B), separated by an area of subsoil. Context #20019A seemed to be a pit dug into subsoil, while #20019B (in the NE corner) was a shallow patch of dark charcoal soil a few cm from the wall. A layer of rubble including sand, lime, shell and charcoal inclusions (#20021) was located under #20019A/B extended underneath the north wall; this rubble sealed subsoil. It is possible that these two features represent the remains of a cooking area because of the high concentrations of shell and pottery overlying ashy sediment cut into the subsoil. Finally, two small pits were found in the middle of the trench, cutting subsoil; both were about 10 to 12 cm in diameter and approximately 10 cm deep (#20017, #20020). Context #20017 included a large cowrie shell, and #20020 included only a few local potsherds.

SM022 (supervised by Federica Sulas) SM022 was a small, 50 x 50 cm trench dug in the area of the geochemical test pit survey (see below) in the central/western open areas. The trench was dug next to GT061. The trench was placed in an area where two cultural layers were visible, which was unique in that most of test pits located only a single cultural layer. The top eight centimeters included the topsoil (#22001) of dark brown very fine sand silty loam. Context (#22002), including the next 30 cm, comprised the first cultural layer, a dark brown silt sandy loam. A second cultural layer (#22003) was found in the next 10 cm, a reddish brown find sandy loam. This sealed subsoil, a white beach sand deposit.

The cultural layers included local pottery, fish bones, a folded copper coin, glass beads and vessel fragments and shell fragments and seemed to represent midden deposits. A micromorphology sample was taken from the wall of the trench at 20-39 cm below surface.

Central Mosque and Adjacent Area SM030/SM029 (supervised by Mark Horton) The central mosque was fully excavated in 2011. These excavations (SM030) included removing the rubble and fill within the mosque itself, as well as limited excavation of areas to the north, east and south of the structure (see Figs. 18 and 19). To the west, a larger area was opened, to explore the space surrounding an associated well. The excavations revealed a small structure, with entrances on the east, west and south, and verandas on the east and west sides. The roof was likely supported by wooden posts, set into the plaster floor. The mihrab was apparently robbed of some of its porites embellishments after the structure went out of use. A concentration of porites blocks (#30015) was recorded in the northeast corner of the building, resting on the plaster floor (Fig. 20) These are what remained of the mihrab decorative elements. The entire mosque was raised up above the level of the surrounding area, built on fill evidently taken from another location. As such, the mosque was set on a series of plinths, and sandstone steps were constructed at each entrance. Water tanks (birika) were 29

Fig. 18: Plan of excavated central mosque (SM030)

located outside the southern wall of the building, one on either side of the doorway. A sluice was built below the southern steps to allow water to flow from one tank to the other. This southern area, and the space surrounding the well, was likely the ablution areas for the mosque. These excavations also revealed that the northern wall of the walled cemetery to the southwest was likely rebuilt to make space for the mosque. Perhaps most surprising in these excavations was the presence of numerous head/footstones just south of the building, and in 30

Fig. 19: Excavated central mosque (left); celadon bowls inset in mosque birika (right)

the area to the north and west. These grave markers were not evident at the ground surface.

SM029 was a 2 x 2 m trench dug into the floor of the central mosque, in the southwest corner of the prayer hall (Fig. 18). This trench was excavated to explore the deposits that predated the mosque, to determine if there were previous mosques, and to attempt to resolve the question of how the roof was supported.

The first two contexts, approximately 40cm in depth, included the plaster floor itself as well as coral rag and sand laid to support the floor. Once through these rubble layers, a thin layer of redeposited cultural material was excavated (#29003); this layer sealed a another shallow layer of cleaner beach sand (#29004). However, unlike other mosques where a deep layer of beach sand is found under the floor, this this 10 cm layer was removed to expose a rich and deep layer of midden deposits (#29005). This dark brown sandy loam contained imported and local pottery, beads, coins, spindle whorls, faunal material, all indicative of common midden materials at the site. This layer was likely moved to this site as a way to build up the platform on which the mosque stands.

In the northeast corner of the trench a rubble-filled pit was cleared to expose a flat coral slab, surrounded by mortared coral rag. This feature probably supported a wooden roof pillar.

Burial Excavations A sample of Songo Mnara’s many burials were excavated in 2011, as a pilot program to gauge feasibility for a larger study looking at demography and disease within the town. We were very grateful to receive permission from the local community for these excava-

Fig. 20: Porites blocks heaped in northeast corner of central mosque

31

Fig. 21: Location of burial excavation trenches (SM024, SM025, SM026, SM027)

tions. Fourteen skeletons were excavated, across four excavation units (SM024, SM025, SM026, SM027; Fig. 21, Table 1). These sampled simple graves (SM024), a stone-built tomb in the central area (SM025), the central graveyard (SM026) and a cemetery outside the walls (SM027). Analysis was completed by Dr Kate Robson Brown and Francesca Migliaccio of the University of Bristol. All bones were subsequently re-buried in the same positions, and a blessing was completed at the end of the season.

Initial results were excellent, with wonderful bone preservation in the sandy soils. Full results can be found in the preliminary report by Robson Brown and Migliaccio. Overall, it was notable that the excavations were instructive on burial practice. It was clear from the offset tombs and stones that graves were often marked or memorialised sometime after the initial burial. In addition, offerings are visible at the former ground level, including ceramics and imported wares, and one grave with nearly 6,000 rounded quartz pebbles (Burton 1872; Fleisher and Wynne-Jones 2012). What follows is a description of the contextual data from each trench that was aimed at excavating burials.

SM024 (supervised by Sarah Walshaw) SM024 was a 1.5 x 2 meter trench in the graveyard located immediately west of a tomb and trench excavated in 2009 (SM012). This trench is notable for the presence of several 32

SK# SK1

Trench# SM024

M/F ?

Age ?

grave shafts, well-preserved human remains (specifically SK4 and SK8), an enorSK2 SM025 ? neonate, 34-38 weeks mous amount of offering stones, and a SK3 SM025 male 26-44 years cowrie shell concentration associated with SK4 SM024 male 25-30 years SK8. The trench was placed over two large SK5 SM025 ? neonate headstones visible at the surface, adjacent SK6 SM026 ? neonate, 6-9 months to the stepped coral-rag tomb, found on SK7 SM024 ? ? the eastern edge of the trench. The first SK8 SM024 male 30-50 years context, #24001, including topsoil, was reSK9 SM026 female 18-23 years moved from the initial 1.5 x 2 meter trench; SK10 SM027 ?

5-6 years before this fill was fully removed, the SK11 SM026 ? neonate, 6-12 mon trench was expanded to 2.5 x 2 meters to SK12 SM025 male 25-44 years accommodate the extent of potential skeletons. Context #24002 was an arbitrary conSK13 SM027 male 25-45 years text that comprised the topsoil of the unit SK14 SM026 ? child expansion plus approximately 20 cm of fill Table 1: Excavated skeletons beneath #24001. This context, #24002, included slag, local and imported ceramics, beads and offering stones. Within this context were also observed three additional gravestones on the north and south sides of the trench, which had been buried by subsequent fill. Near a gravestone in the northeast corner of the trench (Gravestone 4), two large pieces of local ceramic were found, and geochemistry sample was taken (ES24003) from inside one of the sherds. Contexts #24001, #24002, #24003, and #24004 were excavated by hoe, trowel, and hand-shovel, they were arbitrary levels of cultural deposits with little stratigraphic definition. Human remains were located in contexts #24005 and #24006, and excavation methods were shifted to use finer tools including wooden modeling spatulas, trowels, and fine brushes.

One of the most distinctive aspects of this unit was a large number of offering stones, with approximately 5,800 stones located, weighing 68.4 kg (Fig. 22). We first came across the stones at the bottom of #24001 (n=118), where they were located in the center of the trench, one meter west of Gravestones 4 and 5. The stones continued through #24002 (n=1623) and increased dramatically in #24003 (n=4031); this context was clearly the depositional layer for the stones.

Just below, in context Fig. 22: Quartzite offering stones from SM024 33

#24004, only 39 stones were located.

Gravestones were removed when their bases were exposed and became loose. They were individually plotted, labeled, and stored nearby for reburial at the close of the unit.

Gravestones 1 and 2 were removed at the closing of #24003. Gravestone 3 lay on its side at the northern edge of the trench; it was pedestalled and kept in situ during the excavation.

Gravestone 6 lay on its side at the center of the trench within #24003 but it was not evident whether this corresponded to a grave shaft or human remains. It was removed at the opening of #24004. Gravestones 1b (south half), 4, and 5 were removed during the excavation of #24004. A fragment of coral believed to be a gravestone support (Gravestone 7) was found and removed in #24004.

Context #24004 marked a sediment transition that would continue into context #24005/06: fine beach sand interrupted by darker fill in the grave shaft. The sand was quite light, characterized by abundant shell but a deficit of artifacts. Darker sediments indicating burial shafts contained denser, coarser fill with abundant artifacts (predominantly local ceramics, with some non-human bone). Shell was not kept. Special finds included a small number of votive stones, beads, imported ceramics, slag and other metal, three coins, and glass.

Two articulated human skeletons were found in the darker soils, in contexts #24005 and #24006 (the south and north halves of this layer, respectively). Context #24005 revealed an adult skeleton (SK4) lying on its right side with the head directed towards the N/NW. The human remains were in very good condition and we encountered little difficulty in exposing and lifting the skeleton for further identification in the on-site osteology lab. SK4 may be commemorated by Gravestones 4 and 5 which lay a meter above the head. The only grave goods associated with SK4 were a local pottery base sherd immediately to the west of the skull and two large local pottery sherds located north of the innominate bones. Other finds from #24005 included non-human bone, beads, a few fragments of glass, imported pottery, and very few offering stones. Context #24005 also contained three concentrations of disarticulated human bones. SK1, identified only by toe bones, appears to have come from a grave shaft located in the SE corner of the unit. From the southern wall in this same grave shaft, we found disarticulated finger bones at approximately 110 cm below surface. It seems unlikely that these toe and finger assemblages belong to the same individual, given the east-west orientation expected for primary burials in a Muslim context. Disarticulated foot bones were also found on the north side of #24005 and left in situ until the excavation of SK8 (context #24006; see below), whereupon it was discovered that SK8 was complete and the foot bones from #24005 were from a separate individual. In total, seven instances of disarticulated human bone or bone concentrations were bagged as special finds.

Context #24006 was the northern equivalent of #24005. The beach sand fill was cleared until an articulated skeleton was located (SK8). The skeletal remains were associated with a concentration of cowrie shells, which were kept as special finds (SF#24050, from above the burial; SF#24051 from below); these may have been natural, related to the beach sand fill surrounding the burial. Other finds included sherds of local pottery, non-human bone, glass, metal, and one offering stone. Like #24005, this context contained other human remains; these were mostly left in situ because they were adjacent to the wall and thus precluded full excavation. This included finger bones located in the north wall (1m west; 110cm below sur34

face), disarticulated bones (SF#24047) behind and below the skull of SK8, and splinters of friable long bone to the immediate south of the innominate bones of SK8. Following the lifting of SK8, the fill immediately below the skeleton was excavated and sieved for any further bones.

Only tiny bone fragments were found; it was unclear whether these derived from SK8 and were given to the osteology lab for further consideration.

Before the skeletons were returned to the trench for reburial, a thin layer of red sandy back fill from SM026 was spread over the trench; this deposit should indicate the extent of 2011 excavations should the trench be re-opened in subsequent seasons.

SM025 (supervised by Kylie Klein and Courtney Ng) SM025 was located within a rectangular tomb enclosure located in the central open area of the site. The tomb was constructed of coral rag and measured 3.4 x 2 meters. The tomb seemed to be divided into two sections, a smaller northern section (1.4 x 2 meters) and a larger southern portion (2 x 2 meters). The eastern and western walls of the southern enclosure buttressed those of the northern portion, leading us to believe that the northern section was constructed first and that the larger southern portion was added at a later date. Excavations were carried out using a trowel and hand shovel unless otherwise noted. A 30 cm wide baulk was left along the southern wall but was later removed as context #25008.

Context #25001 spanned the entirety of the trench and comprised a thin topsoil layer of silty loam with large coral rubble inclusions. Beneath the topsoil in the southern portion was context #25002, a sandy silt with less rubble. In the northern partition, context #25003 was a sandy silt deposit 50 cm deep at its greatest depth and, containing a significant amount of coral rubble. Both of these contexts contained very few artifacts, including small amounts of local pottery, glass, a few beads, quartz, a round stone, and a small round clay disk. Both contexts were excavated by hoe and shovels.

Below context #25003 (in the northern portion) was a distinct soil change, a darker sand that marked the appearance of a burial shaft. This darker sand (of the burial shaft) was recorded as context #25004 and the lighter subsoil sand to the north of it context #25005. Although we photographed these contexts, we eventually decided that it would not be feasible to excavate the contexts separately because the subsoil was so soft, whenever #25004 was removed, #25005 would slump. Therefore, they were removed together, as context #25009.

Context #25009 yielded similar artifacts as context #25003. At a depth of 70 cm below surface, in #25009, an adult skull was revealed; a full skeleton, SK 3, was subsequently exposed. SK 3 was positioned laying on its right side with its head facing north. We carefully uncovered, photographed, and planned it, and removed it in anatomical units for analysis. Upon the removal of SK 3, we stopped excavating in the northern portion of the tomb.

In the southern section, context #25006 was located beneath #25002, but was virtually identical in composition to context 25002 except with slightly more artifacts. In #25006, three burial shafts were revealed – a thin one that extended into the southern wall, a deeper one that was completely enclosed in the southern section and stretched across its entire width, and a smaller patch of dark sand that appeared in the western section of the southern partition. We decided not to dig the burial shaft that extended into the southern wall because it was likely that the body within it would be partially under the tomb wall and therefore very 35

difficult to excavate. Because of this, at about 1 meter below the surface, we left a second, 30cm-wide baulk along the southern wall. Context #25007 was a very fine layer of sandy soil removed beneath context #25006, but which was only dug for a short time before locating context #25010.

At a depth of approximately 93 cm below surface, the skeleton of a small very wellpreserved baby (SK 2) was located in #25010. Since it appeared in the largest grave cut, it was believed that there was an adult body buried in this area of the tomb as well and thus continued to dig in this area. Shortly afterwards, a second baby skeleton (SK 5) was located in the small western grave cut; this skeleton was that was slightly larger than SK 2, but not as well preserved. Both babies were buried on their right sides with their faces turned northward.

Following the removal of the two baby skeletons, the frequency of shell in context #25010 increased dramatically. At a depth of about 1.72 meters below surface, an adult skeleton (SK 12) was located in the southern portion. It was a poorly-preserved individual oriented in much the same way as the other skeletons. Based on the position of the bones, both adult skeletons appeared to have been tightly shrouded. Once SK 12 was removed, excavation ceased, and the trench was prepared for the reburial of the bodies.

SM026 (supervised by Juliette Chaussen and Daisy Chen) SM026 was a 2.5 x 3.5 trench located within the small walled graveyard to the south of the central mosque, in the central open area. Excavations revealed three main layers including contexts #26001/02/03 and three contexts associated with human burials (#26004/05/06).

Context #26001 was the topsoil layer, stretching across the entire trench, consisting of dark reddish-brown sandy loam. The layer was rich in coral rubble and roots. Context #26002 was also a reddish sandy fill, with less brown loam included than in the topsoil; as the layer was excavated, the soil’s color became more intensely red. This context was slightly more compact than that of #26001, and included small amounts of local pottery and frequent finds of clumped reddish sand. Context #26003 was clean white sand, the subsoil deposit.

Four burials were located in this trench, including two infants, one juvenile and one adult. The first skeleton encountered was a poorly-preserved infant (SK 6), located in the northwest sector of the trench. Stratigraphically, the skeleton was positioned at the base of #26002 and in the transition from #26002 and #26003.

The remaining three burials were indicated by burial shafts dark red sandy soil cut into the subsoil (#26003). Context #26004 was one of these, a grave that contained a wellpreserved adult skeleton (SK 9), located in the north central part of the trench. Three pieces of pottery were found by this skeleton—a sherd found on both sides of the left shoulder and one sherd by the knees of the skeleton. Moderate numbers of cowrie shells were found in this context; these were likely from the natural sand subsoil deposit. Context #26005, toward the center of the trench, contained a poorly-preserved infant skeleton (SK 11). Context #26006, in the southeast part of the trench, contained the skeleton of a juvenile in good condition (SK 14).

While SK6 and SK9 were located stratigraphically higher than the other two burials, it should be noted that there was not more than 20 cm of fill between them, suggesting that these burials may have been interred over a short time period.

36

SM027 (supervised by John Perkins and Grace Apfeld) This trench was located outside the eastern wall of the town, to the east of Houses 28 and 29. A bit farther to the east of the trench is a cluster of coral rag tombs, with one just to the south of this trench. The trench was sited in a location with many head and footstones, to explore burial deposits outside of the town wall. Seven contexts were recorded in this trench, four of which were either natural or related to more general deposition deposits of trash outside the town wall; two were related to burials and one may represent offerings made at one of these burials.

Topsoil was removed as context #27001, a dark brown sandy silt with small coral rag rubble and 20-25 cm in depth throughout the trench. This layer was rich in artifacts, suggestive of a midden deposit that was broadcast outside the town walls. This is somewhat surprising as the area contains a number of head and footstones, and tombs. Context #27002, a dark brown sandy silt with many artifacts was the soil in which the head and footstones were set, mainly differentiated by a decrease in rubble and roots. This layer might represent a transition between the topsoil and #27004, the main cultural layer. Context #27003, scatter of pottery, bone and shell was recorded separately, and may represent a discrete deposit placed by a headstone (#3), in the eastern end of the trench, perhaps as an offering.

Context #27004 was a dense, rich midden layer (and thus #27003 may have simply been the top of this). This layer was 20-30 cm in depth and comprised of reddish brown sandy silt. This context sealed a number of contexts, including #27005, the yellow/white sand subsoil layer, and area of darker soils in the center of this trench. This area contained two different fills, #27006 and #27007, each related to a burial, SK 10 and SK 13 respectively. The first burial excavated, SK 10, was of a child with relatively good preservation. SK 13 was an adult male, very well preserved but situated strangely within the grave cut. It looked as if this body was close to face down in the grave cut, with its hands behind its back, evident in both the arrangement of the arms and the position of the shoulder blades.

Other possible grave cuts were noted in the northern wall of the trench, but were left unexcavated. The number of head and footstones and possible grave cuts in this trench suggest that this area is likely dense with graves.

Geophysical survey at Songo Mnara and Kilwa Kisiwani The geophysical survey team completed the remaining unsurveyed areas at Songo Mnara, focusing on the southwest and southeast parts of the site. Both magnetometry and EM survey was conducted. This completes the work begun in 2009; as then, definite areas of activity were visible on the magnetic survey. No further spikes associated with iron-working were located. Instead, magnetometry has illustrated a pattern of anomalous areas that seem to be linked to wattle and daub architecture and associated areas of domestic refuse. This populates the spaces of the site and gives a context for much of the sampling work undertaken here. A Digital Elevation Model was also completed, to combine with the other spatial data in a site GIS, constructed by Mark Dover.

In addition, the geophysical survey team conducted some trials at the neighbouring site of Kilwa Kisiwani (Fleisher, Wynne-Jones, Steele and Welham 2012). This aspect was fund37

ed separately by a small grant from the Society of Antiquaries of London, and the aim was to explore the wider urban landscape into which the famous Kilwa monuments fit. In addition to magnetometry, the team used Ground-Penetrating Radar (GPR) survey at Kilwa, to help take account of the much deeper stratigraphy. The results are still being processed, but it seems that Kilwa is more complex methodologically, with fewer clear signatures across the open spaces. This may be due to wholesale clearance of ground during the 1960s excavations. Still, results in certain delimited areas have been clearer, particularly in the enigmatic enclosure of Husuni Ndogo, and the full results look set to be very interesting.

Geochemical/Phytolith Testing Research in open areas also included a testing program over a large section of the central/western open area and excavations in a courtyard between two houses in the north (SM019). The testing program was an extension of work begun in 2009, during which soil samples were taken along three transects in the central open space to examine soil chemistry and phytoliths (Sulas and Madella 2012). The 2011 testing program was more intensive, and covered an area of roughly 130m by 70m, with shovel tests excavated every five meters on a stag-

Fig. 23: Test pit grid in western and central open spaces 38

Fig. 24: Possible open space surrounded by earthen houses; red dots indicate the presence of daub in test pits

gered grid (Fig. 23). In total, 266 shovels tests were excavated; soil samples were taken from the primary cultural deposit and all fill was sieved and artifacts collected. Soil samples have been submitted to the ALS Laboratory in Seville, and will be subject to ICP-AES analysis to determine chemical concentrations. Phytoliths will also be extracted from these samples. The testing program was supervised by Dr Federica Sulas, who will be overseeing the geochemical and phytolith processing.

The test zone corresponds with an area in which geophysical surveys in 2009 indicated a possible open plaza, bordered by numerous circular anomalies. It also extended into the central open space, covering the central cemetery with tombs and gravestones, as well as spaces adjacent to and between houses (Fig. 23-4). While we await the chemical and phytolith data to examine possible activities across this space, an initial plotting of artifact distributions from the test pits reveals that earth houses once bordered the plaza space on the north and south; concentrations of daub correlated directly with geophysical anomalies located in 2009 (Fig. 24). Artifact distributions have also allowed for a more precise mapping of an iron smithing area, as well as variable and likely patterned distributions of materials across the space. A similar testing program will cover the northern open area in 2013.

39

40

Archaeological Survey on Songo Mnara Island Jack Stoetzel University of Virginia Introduction This paper reports results of an archaeological survey of Songo Mnara Island, in the Kilwa archipelago off the southern coast of Tanzania. The purpose of this research was gain an appreciation of settlement patterns and chronologies beyond the walls of the stonetown, also Songo Mnara, located in the north- west tip of the island. The survey includes a paleoethnobotanical component, results forthcoming, which aims to reconstruct a diachronic link between Swahili land-use strategies and maintenance or change in plant communities. With the overarching environmental nature of this research in mind, I rely heavily on contemporary physical conditions (namely substrate type and character) in order to frame the results and analysis of this work.

Collaboration with J. Fleisher and S. Wynne-Jones (2010) during their 2009 field season at Songo Mnara motivated this research. While investigating daily use of public and private space across the urban center, the researchers uncovered a simple stratigraphy that indexes a

Fig. 1: Environments of Songo Mnara Mangrove (blue), Land Thicket (purple), Coral (yellow), Sand and Mud (brown), and Land Scrub (black)

200 year window of occupation between the 14th and 16th centuries AD (Wynne- Jones & Fleisher 2010). This survey was conducted in conjunction with the Wynne-Jones and Fleisher 2011 field season and looks to augment their ongoing research through the addition of contextual evidence of settlement patterns across the site as a whole.

41

Archaeological Survey I conducted an archaeological survey of the island in order to locate previously unknown archaeological sites. The island features five environmental zones assayed by satellite survey (Fig. 1). Of the five zones, the survey was relegated exclusively to “land-scrub” as this was the only accessible area. In order to achieve full coverage of the “land scrub” regions I divided the island into a series of transects spaced no closer than 250m from one another. Previ-

Fig. 2: Shovel test pit survey of Songo Mnara. Note, that Land Scrub is the only one included in image

ous archaeological surveys conducted along the Swahili coast (LaViolette et al. 1989; Fawcett et al. 1989; Fleisher 2003; Pawlowicz 2009) demonstrates that sub-surface testing is more successful in identifying archaeological sites than exclusive reliance on walking survey. With this in mind, I began each transect with the intention to excavate a shovel test pit every 250m along each transect (Fig. 2).

Sub-surface excavation was not possible in several locations which, judging by satellite images, appear survey-able. The primary impediment to excavation was an abundance of fossilized coral rag either protruding from the ground- surface or located within 10cm of the soil surface. Songo Mnara Island features a heterogeneous composition of porous limestone, forminiferal limestone, and fossilized coral substrates surrounded or covered by soils that range from reddish-brown clay loam to coarse yellow- grey sand. This range of substrate types is characteristic of the general Kilwa-Lindi region of southern Tanzania (Haughton 1938; Nicholas et al. 2006). In order to mitigate limitations imposed on the sub- surface survey by coral- and limestone dominated areas, I sought crags or heavily eroded areas within a 15m radius of intended location. This survey also excluded (a) rice fields, (b) dense bushland, and (c) mangrove forests which otherwise appear survey-able in satellite images. Walking survey represents the only source of data for areas where substrate or local environments prohibited excavation.

42

Survey Results The archaeological survey recovered artifacts from 32 units, 39.02% of the 83 excavated test pits. Positive units were distributed through 70% of transects (Fig. 3). The tidal island, Sanga Rungu, directly north of Songo Mnara was the only area that did not produce any sub-

Fig. 3: Positive shovel test pits

surface artifacts. Not surprisingly, positive test units correlate with the past urban center and more recent village and agricultural areas, which are concentrated in the southern part of the island. The spread and range of positive units between these known occupation zones offers a new appreciation of the geographic extent to which Swahili people have occupied and interacted with physical conditions across Songo Mnara Island.

While demonstrative of Swahili familiarity with the island as a whole, I do not accept each positive test unit as a distinct archaeological site. A majority of sites contained a single

Fig. 4: Sites recorded at Songo Mnara 43

undiagnostic pot sherd. In order to filter such units from further analysis, I define five artifacts as the minimum threshold necessary to index an archaeological site. Twelve test units, 37.5% of the 32 positive units, therefore constitute the archaeological sites discovered through the course of this archaeological survey (Fig. 4).

Distribution of archaeological sites is more constricted than that of positive units. Four transects boast the presence of an archaeological site and, of these, 11 of the 12 sites are restricted to the three northernmost transects. These sites all cluster around the 14 th to 16th century AD stonetown; the furthest site is located roughly 3km south east of the former town wall. Considering artifact assemblages specific to each, a wave pattern of material richness becomes evident. A wave pattern of material richness becomes evident when considering artifact assemblages specific to each site. I say a wave pattern because sites within 1km of or 2.5 to 3km from the former town wall sport a rich and diverse material assemblage relative to that recovered at sites 1-2km from the urban center.

Initial ceramic analysis conducted by S. Wynne- Jones and J. Fleisher (2011, personal communication) identifies five sherds with chronologically diagnostic characters. Four of these sherds date to the 14th to 15th centuries whereas two demonstrate features consistent with 19th century ceramics in the Kilwa region. As Figure 6 demonstrates, the sites bearing 14th to 15th century ceramics are adjacent to the urban site whereas those with 19th century characters are associated with an area that, in the modern context, is devoted to rice agriculture. The chronology of sites identified by this archaeological survey indicates two pulses of Swahili occupation at Songo Mnara: one during the 14th to 16th centuries associated with the urban center at the northern end of the island, the other beginning in the 19 th century and located in the southern portion of the island.

Discussion Initial survey of Songo Manra Island indicates that the urban center did not terminate at the town wall. Instead, the stonetown was surrounded by a non-stone community extended over 3km south and east toward the Indian Ocean. Imported material culture was not restricted to the urban center as non-stone sites beyond the wall feature Islamic monochrome pottery and glass beads. A more exhaustive sub-surface survey of this area just beyond the urban center will yield further evidence of a rich and bustling earth and thatch community engaged in economic pursuits similar to those of their stonetown neighbors.

Geography offers an explanation for the wave pattern of artifact density and richness surrounding the stonetown. The fact that sites nearest the urban center demonstrate the richest material assemblage is not surprising. These sites could represent habitation areas of lower echelon merchants or individuals otherwise marginally connected to the Indian Ocean commercial system. Site richness 3km away from the stonetown may be the result of these sites representing the area of initial interaction between merchants from Songo Mnara and those arriving from the Indian Ocean. The sites are adjacent to expansive sand beaches accessible at high tide and exposed at low tide, fashionable harbor conditions according to historic texts.

44

Archaeobotany: Preliminary Field Report Sarah Walshaw Simon Fraser University Introduction Archaeobotanical research conducted in conjunction with the Songo Mnara 2011 field season included three sub-projects: (1) flotation of sediment for archaeobotanical analysis, (2) collection of plant comparative specimens, and (3) ethnobotanical interviews. Here I report the progress made for all three sub-projects, and I indicate future arenas for botanical research in the laboratory as well as anticipated fieldwork in 2013 and 2015.

Flotation Sediment samples from 76 contexts across the site were floated to retrieve macrobotanical specimens, including carbonized seeds and fragments of wood, nutshell, and parenchyma. Excavators normally took 3-4 liter samples from the freshly exposed sediment of occupation layers; feature contexts often yielded smaller flotation samples. Following protocols previously established by Walshaw (2005, Walshaw and Pistor 2010), manual flotation was carried out using predominantly locally available materials (e.g., buckets) and using as much recycled water as possible. The light fraction (also known as the ‘flot’ – buoyant material) was caught in a .500mm standard geologic sieve. This included uncarbonized plant matter, insects and insect eggs, and carbonized organic matter – mostly plant remains, but some bone also. The heavy fraction comprised materials in the sediment that sank in water, caught in a 2.00mm sieve.

These included fragments of ceramics, daub, heavier bones, and occasionally items such as beads. All diagnostic cultural materials were returned to the artifact lab for identification and recording. After visual inspection in the field, 19 heavy fractions were turned over for fish/ shell analysis, and 47 were discarded. The remaining heavy fractions, and all light fractions, were transported to the Archaebotany Laboratory at Simon Fraser University. They will be analyzed by Walshaw in later 2011 and 2012.

The flotation methods used warrant a brief description here, in particular the water recycling system used. Freshwater is optimal for flotation but is scarce on the island of Songo Mnara. The Songo Mnara Archaeological Project brought freshwater from the mainland to the island at great expense, and water conservation/recycling was paramount. The flotation system used at Songo Mnara relied on a mix of fresh and recycled water. During flotation, water containing buoyant material was poured through a flot sieve nestled in the lid of a receiving bucket. Water was recycled by letting this water in the receiving bucket settle for several hours or overnight so that clay and fine silt particulates descended, then pouring the relatively clear water through the finest mesh sieve used in the procedure (.335 mm). Thus, intersample contamination was negligible, because potential contaminants in the recycled water were too small to be collected in the light- or heavy-fraction sieves. The final float for each sample was conducted using fresh water – this ensured the maximum recovery of buoyant materials in the light fraction, and also served to rinse those light fraction materials floated with recycled water. The heavy fraction was largely rinsed with recycled water, with less need 45

for de-silting. Light and heavy fractions were dried in the shade in fine-mesh mosquito netting for several days, then transferred to plastic bags containing silica gel packets for transportation and storage.

Preliminary Results Analysis of the archaeobotanical remains from the 2011 field season is only just beginning, however some preliminary findings are available. SM 020 yielded a flotation sample with evidence suggesting household cooking activities. Wood charcoal was dominant but the sample also yielded a few pieces of parenchyma (undifferentiated plant tissue from storage organs), one seed fragment of baobab or cotton, two probable mung beans (Vigna radiata), one unidentified bean (cf Vigna) and one unidentified seed. Mung bean, a south Asian legume, was also reported from the site of Chwaka on Pemba Island (Walshaw 2005). It is not known whether mung bean was grown on the island of Songo Mnara, or shipped to its residents.

By contrast, a sample located directly above human remains in a cemetery context (SM 024) was dominated by uncarbonized elements (roots, seeds, shell, worm castings.) Interestingly, wood charcoal fragments were present but rare, and one fragment of baobab seed was found. These findings may be attributable to tracking from other areas of the site, however if further wood charcoal and burnt foot remains are found in other cemetery contexts then ritual food offerings may be responsible for these deposits.

The dominance of wood charcoal in the archaeobotanical assemblages of both a household and a burial context warrants further discussion, as it suggests that the primary fuel source of Songo Mnara was wood (rather than coconut). The site was likely flanked by mangrove trees in the past, as it is today, and these are a potential source of fuel. However, other trees grow in this region, and it would be valuable to determine which trees were used for fuel. Taxonomic identification of the wood charcoal from Songo Mnara awaits further analysis, aided by the compilation of a comparative collection (see J. Stoetzel’s research, this volume.)

Baobab (Adansonia digitata) seeds were observed in both light fractions reported above, and also in the heavy fractions of two contexts: a mosque mihrab (SM 030) and in SM 019 (context 3). I fully expect further baobab specimens from the 2011 flotation samples. Baobab trees were important in the past, as they are today, for their uses as food and medicine, but also as culturally significant markers on the landscape. Interestingly, baobab is often not cut down for fuel, due to its cultural value and poor combustion properties of its timber. This tree visually dominates the flora supporting the site today, and modern baobabs have grown up alongside, and in many cases infiltrating, standing architecture.

Comparative Specimen Collection Botanical comparative specimens aid in the identification of archaeobotanical materials. As eastern African floral comparative collections are few and far between (and typically incomplete or poorly suited to archaeological plant remains) I have been collecting specimens for herbaria since 2002. Collections are housed at the National Herbarium in Arusha, the University of Dar es Salaam, and the Missouri Botanical Garden. I have also developed my own comparative collection targeting the taxa and plant parts most likely to be found in archaeo46

logical contexts. This season I completed a plant collection pilot study at Songo Mnara to begin to document the local flora, focusing on collecting the useful parts of plants, as well as those that could be contaminants in archaeological contexts. I was greatly assisted by several local people who, once they learned that I was interested in plants – particularly foods – brought me specimens from their fields or yards, or from uncultivated spaces. Local people also were especially helpful in providing local names and uses of plants, and in demonstrating how to process plants for food (see below for further details).

The pilot study confirmed that there are enough different wild and cultivable plants at the site to warrant a more complete botanical collection program in 2013. I plan to have a Tanzanian botany student assist me with the collecting, labeling, drying, and transportation of plant comparative materials. To this end, I intend to apply for a small grant to fund a student to accompany me on the project in 2013 and possible 2015. Our goals will be to document the local flora that occupy the site today as well as the useful plants of Songo Mnara. These collections would reside in herbaria in Tanzania and the United States for future consultation and research.

Ethnobotany Understanding how and why local people use plants today aids greatly in the interpretation of the archaeobotanical record. It also helps document local practices of nonmechanized agriculture, which are increasingly rare in the world today. The offshore islands of Tanzania did not experience the agricultural reform and community upheaval associated with Ujamaa during Tanzania’s early independence period. As a result, farming on Songo Mnara has been influenced moreso by local practices of family farming than industrial or commercial farming. During the 2011 field season we conducted interviews with three farmers and their families concerning rice and sorghum cultivation. We additionally obtained information about the processing of the root/corm of the mlanga plant, which today grows wild near the archaeological site. Mlanga’s starchy underground storage organ can be used to prepare a flour by grinding it against a rock to shred it, then squeezing the water from the pulp and letting the water evaporate. The flour can be added to other starchy flours to stretch one’s food source, and it can be used to make the sweet treat halwa.

Sorghum farming near Songo Mnara The family of Mama Zuberi has sorghum fields and a homestead within a five-minute walk of the archaeological site. This great-grandmother and farmer kindly agreed to speak with us on two separate occasions, the second of which resulted in a detailed discussion about past and recent sorghum cultivation and processing practices (translations and discussions aided by Julie McGruder). We also took video footage of how to grind sorghum using a stone wheel.

Mama Zuberi described several attributes of sorghum farming that are notable for archaeologists and archaeobotanists. Fields are cleared for planting by felling vegetation with a panga (machete) and then burning the field. This practice of “slash and burn” may be archaeologically visible in micromorphological analyses. Sorghum grains were stored for food as well as for planting the next season. Sorghum seeds destined for planting were not selected for any 47

specific trait; rather, “sorghum is just sorghum”, or “mtama ni mtama tu.” The growing season is approximately five months long, with two weeding episodes necessary. Toward the end of the growing season, the seeds become visible and are susceptible to pests such as birds, wild pigs and vervets. Older children patrol the fields at this time to scare away the birds. I noted field houses (raised platforms) nearby, however Mama Zuberi did not describe these. Instead, she discussed storage containers at length, a practice which I have not observed elsewhere. In the past, she said, every cultivator had his/her own threshing tank built of sticks, rope, and leaves woven together to make thatch. This would rest on a platform raised above the ground on poles. A fire could be built underneath to dry and fumigate the threshed seeds She described this as a specialty craft that only a few had the knowledge to build. She reported that today people are more likely to use such platforms to store their sorghum in the kitchen area.

Mama Zuberi noted that Songans today store their sorghum with the seed coating (pericarp) intact, and removed it just prior to cooking or grinding by pounding and winnowing with a mortar (kinu) and pestle (twanga). She stores her sorghum with a white powder (dawa; commonly used to refer to medicine as well as insecticides, fertilizers, etc) that is purchased in the mainland market town of Kilwa Masoko.

Rice farming in Madaweni Approximately 1.5 hours north of the site, by foot, lies a series of low-lying square to rectangular rice plots separated from one another by 1-2m tall earthen berms. This area was desalinated and a dam was erected in the early 1970s to prevent nearby ocean water from encroaching on the fields. Recently, however, the dam was breached and at least three fields were rendered useless through the presence of saltwater. At least three families were forced to move elsewhere, and the entire farming community remains at risk for further souring of the soil. Local peoples consider this a very high priority and expressed disappointment that the government had not organized a reclamation effort. One local mzee, an engineer who worked in Zanzibar for much of his life, argued that the repair of the dam was a conservation issue as important as the efforts to preserve the standing architecture at Songo Mnara.

We interviewed Mzee Suleiman Abdallah and his wife Bi Mkanje concerning rice farming in this area, and their son Hamisi kindly gave us a tour of the dam and provided us with a local history. The mzee described a fairly conservative and small-scale rice-farming practice.

Each rice plot is owned by one family, whose sons inherit and divide the plot once they come of age. Seed rice may be obtained from one’s store, from neighbors or family, or sometimes from the government. Rice is sown in the low-lying areas directly (no transplantation mentioned) with rainwater being the sole water source. No dawa is used to reduce weeds or to fertilize the soil, however dung deposited by cattle grazing on stalks after the harvest provides nutrients to the soil. Harvesting is done by hand, and rice is stored in the chaff after threshing.

Family members do the rice farming work. If workers are needed, they are compensated by getting a portion of the harvested rice or earning a favour in return; farming help is not paid in money. We also visited Hamisi’s house and harvested some rice by hand, during which we were introduced to novel techniques of threshing by his female relatives. I have previously observed that farmers in Pemba thresh rice in several ways, including: a) putting the harvested rice in a bag and beating it with a stick to release the florets; and b) twisting feet on the har48

vested plant to remove the florets from the stalk. The women in Hamisi’s household showed me yet another method: by pulling the stalk quickly but carefully between the thumb and the blunt edge of a harvesting knife, the florets were easily removed.

Conclusions The archaobotanical contributions to the 2011 Songo Mnara field season included the recovery of macrobotanicals through flotation, the collection of plant comparative specimens, and learning about wild plant uses and agricultural practices from local experts. Results from the archaeobotanical analyses should be available early in 2012. These are expected to complement the microbotanical data from Songo Mnara, and permit a more focused archaeobotanical sampling during the 2013 field season. Collection of plant specimens and local knowledge concerning the flora and agriculture of Songo Mnara will be an important component of the 2013 field season, for which I hope to obtain a Tanzanian field assistant trained in botany.

Acknowledgements I am grateful to my very capable field assistant, Dominic Pistor. Thanks also to the many field school students who helped out in the flotation banda. Site directors Stephanie Wynne-Jones and Jeff Fleisher were endlessly resourceful and supportive. Finally, I am humbled by willingness of so many local farmers to share their expertise, crops, and stories with me.

49

50

Preliminary maritime survey report of Songo Mnara , June 2011 Edward Pollard Orkney College UHI Introduction This maritime archaeological survey of Songo Mnara took place between 13-30 June 2011. Songo Mnara is an island located in the southern part of the Kilwa estuary or bay (Fig. 1). The aim was to record the exploitation of the inter-tidal zone around Songo Mnara and the harbour of Sangarungu. The survey joined the expedition to Songo Mnara organised by Drs. Stephanie Wynne-Jones, University of York, and Jeff Fleisher, Rice University.

The fieldwork included inter-tidal survey around low tide and coastal survey around

Fig. 1: Kilwa estuary showing location of Fig. 2.

51

high tide. The routes covered can be seen in Figure 2. It included Sangarungu Harbour and its approaches. This involved north Songo Mnara, south Kilwa Kisiwani, Sangarungu, Sanje ya Kati and the mainland area of Msekera. The final day was spent visiting the mainland at Pande to the south in order to ascertain potential for a future more intensive survey. Two locals were used as guides for safety, carrying survey equipment and local knowledge (Fig. 3). On three occasions students from the Songo Mnara expedition joined the inter-tidal survey.

Fig. 2: Survey routes taken by the author around Sangarungu Harbour.

The survey was done by either walking from the camp site on Songo Mnara, or by dhow from Songo Mnara to a landing place on another part of the harbour and walking from there. Areas that were difficult to survey were the dense mangrove forests on the sheltered parts of the bay and estuary, and the mainland around Msekera and Sanga, which is relatively uninhabited and large wild animals are present (evidence for elephant, hippo and buffalo was recorded). The coast exposed to the Indian Ocean is fringing reef and lagoon, which is easier to explore though care has to be taken from the presence of sea urchins, eels and sharp eroded fossil coral. As low tide is later, by approximately half an hour every day, the inter-tidal survey had to take place later each day making the surveyors more exposed to midday heat.

Snorkelling and diving around the reef west of the camp site, and to the north of Kilwa Kisiwani, indicated similar loose coral material to that on the fringing reef.

A small test pit was also excavated into Sangarungu Island to determine formation of the sand spit and to find dating evidence. Sangarungu Island is the most important modern 52

landing place on Songo Mnara and has a fishing village. Investigation was necessary to determine if it was used during the occupation of the ruins on Songo Mnara.

Sangarungu Harbour Sites Sites are shown in Figure 4 with preliminary descriptions following.

Figure 3: Surveying the reef crest on Songo Mnara

Songo Mnara Previous visits to Songo Mnara by Pollard (2008) described the area immediately surrounding the ruins.

Mna007/11 Site Number Msanapamba Place name S09˚02.262 E039˚33.682 (GPS8) GPS location Ruin of a house Site Type Stone walls Features 18th – 19th century (Pradines et Blanchard, 2005: 22) Period Facing North Orientation/aspect Stone building with N-facing side eroding into the sea. It has two Description compartments [4.5m including the walls, E-W the total is 9.2m including the walls]. The walls are 0.44 m thick. The building is very overgrown. From the south wall there is 90cm of collapsed stone from the back wall to the HWM. There are two wattle and daub huts within a coconut plantation nearby.

53

Fig. 4: Sites around Sangarungu Harbour

Site Number Place name GPS location Site Type Period Description

Mna008/11 Baraka -9.038975 39.579625 (GPS16) Pot scatters Late 13th century to 17th century Loose reef coral on top of an islet on the reef crest. Sand and loose reef coral sand spits are at the north and south ends of the islet. Very overgrown with terrestrial vegetation on the islet and the sand spits. Dense mangroves are on the sheltered sides landward of the island. On the inter-tidal zone on the sheltered landward side of the island are two pottery concentrations: GPS above and -9.038565 39.579233. The southern scatter is water worn with rounded edges but the trace of rims can still be discerned and the shape of Type 22 Husuni Modelled Ware on one shard. Period IIIa. One piece with a flatter rim could be Type 19, food bowl Period IIIbIV. The northern pottery scatter had a piece of imported blue and white.

54

Site Number Place name GPS location Site Type Period Orientation/aspect

Mna009/11 Mvinja S09.036322 E39.564459 (GPS45) Stone house, pot find spot c.1400 to 1500AD for pot, 18th – 19th century for house (Pradines et Blanchard, 2005: 22) Overgrown stone house overlooking the coast. Kifinge and Pagogo have loose and cemented reef coral on reef crest seaward of the Mvinja house. Kifinge was a findspot of pottery. The pottery is pockmarked as it is worn by being exposed on the reef to the waves. This indicates a period of time on the reef. Although worn, a lip can be made out on the pot. Possibly similar to Type 33c water-pot or Type 34a Later Water-jar.

Site Number Place name GPS location Site Type Period Description

Mna010/11 Magomba S09˚01.473 E039˚33.417 Navigation reef marker, pot find spot Unknown Natural jagged coral limestone standing proud of the reef crest. It is surrounded by loose reef coral on the landward side. The reef coral is up to cobble size. A piece of thick pot was recorded on the east side with orange and white coarse specks in a black matrix. Five fish trap baskets were in the mangrove trees on the sheltered side.

Site Number Place name GPS location Site Type Period Description

Mna012/11 Ukaya S09.042734 E39.583807 (GPS64) Pot find spot Unknown Ukaya ‘island’ consists of loose reef coral and cemented coral above the limestone bedrock that forms islets. Pot is found in the small harbour formed between the islets and loose reef coral on the landward side.

55

Site Number Place name GPS location Site Type Period Description

Mna013/11 Mswala S09.046256 E39.586355 (GPS142) Find spot 15th century?

Pottery on steep sandy beach with lagoon seaward. Grooved pot fragment probably from a lamp.

Site Number Place name GPS location Site Type Period Description

Mna014/11 Mkuje S09.059816 E39.599692 (GPS143) Panel tombs 14th – 15th century (Pradines et Blanchard, 2005: 22) Two stone Sherif graves are surrounded by smaller upright single stones graves. Porcelain removed from the more elaborate graves as the cup holes remain. This site is in an overgrown area landward from the beach (Fig. 5).

Fig. 5: Copse where Mkuje graves are located

56

Site Number

Mna015/11

Place name

Shermani

GPS location

S09.061956

Site Type

Ethnographic evidence

Period

Late 19th to 20th century

Description

A German lived on this raised coral islet from the Berlin Conference after partition of Africa until the end of war (Fig. 6). Modern fishermen’s huts were to the south of the coral islet and cliffs.

E39.610628 (GPS146)

Fig. 6: Shermani where a German lived during the colonial period

Site Number Place name GPS location Site Type Period Description

Mna016/11 Kivurugu S09.058828 E39.617454 (GPS156) Shipwreck ballast Unknown Reef coral piled on cemented coral platform. Modern concrete datum point found on the reef coral causeway. Basalt on the reef crest could have come from a ship hitting the reef and loosing ballast.

57

Site Number Place name

Mna017/11 Msete

GPS location

S09.037169

Site Type

Pot find spot

Period

Unknown

Description

Pot found in grey clay amongst rock pools. Large bowl, possibly thinning to the rim

Site Number Place name

Mna018/11 Kifigo

GPS location

-9.046958

Site Type

Pot scatter

Period

1400 to 1700AD

Description

A scatter of local pot, and blue and white pot in red soil above limestone cliff. Local pot is small fragments, some with comb incision and incised incision, flat rim, rounded rim and rim with small lip. Possibly food bowls, Period IIIb/IV

Site Number

Mna019/11

Place name

Kisima Chamkwaju

GPS location

-9.046142

Site Type

Well

Period

Unknown

Description

Deep well, rock carved c.12m deep c.5m wide. Smaller circular basin on south side of well with fresh water at base. Lots of ants on a red soil with pot.

Site Number Place name

Mna020/11 Kisiwa Nduo

GPS location

-9.038453

Site Type

Ethnographic evidence

Period

20th century

Description

A 15.3m wide causeway. Appears to be bedrock below 7.5m wide loose reef coral that is 0.5 to 1.0m high. During the 20th century there was an airborne disease on Songo Mnara called Nduo.

Ndui means smallpox. The sick were put on the island until they died so they wouldn’t infect the rest of village.

E39.557682 (GPS161)

39.555413 (GPS172)

39.557792 (GPS176)

39.570652 (GPS11)

58

Chani The survey at Chani on the south coast of Kilwa Kisiwani continued on from that recorded in Pollard (2011).

Site Number Place name

Cha001/11 Watiro Islet

GPS location

-9.005159

Site Type

Pot scatter

Period

c. late 13th century to 1400AD

Description

Pot was discovered, mostly on its west side, around the edge of an uninhabited islet. The islet was densely vegetated but it was possible to climb up on the top on NE side where the wave-cut notch cliff has collapsed. An arch and a baobab is on the east side of islet (Fig. 7). The undecorated pot is similar to local pottery type 22b (Period 3a)

39.542105 (GPS124)

Fig. 7: North side of Watiro Islet.

Site Number Place name

Cha002/11 Chani

GPS location Site Type Period Description

-9.015996 39.539012 (GPS132) Pot find spot Unknown Eroded piece of pot found in lagoon.

59

Sanje ja Kate A previous visit, with sites, to Sanje ya Kate in 2004 is described in Pollard (2008).

Site Number

Kat002/11

Place name

Uvula Village

GPS location

-9.059035

Site Type

Pot scatter

Period

Unknown

Description

Some pot lies about the village and coconut plantation.

Site Number

Kat003/11

Place name

Kiungani landing place

GPS location

-9.060478

Site Type

Pot scatter

Period

Unknown

Description

Kiungani LP has a few pieces of eroded pot. There is a wide passage through the mangroves (10-20m wide) for the landing place.

Mangrove whelk shell middens along with some pot are also here.

It is possible to see Pande from here and a small island for fishermen called Swelani.

Site Number

Kat004/11

Place name

Kwaomarimmadi landing place

GPS location

-9.059351

Site Type

Pot scatter

Period

Under investigation

Description

Sparse mangroves on a sand spit, canoe, pot along beach. Mangroves behind. Steep sand then flat sand, Sanje ya Majoma on opposite side

Site Number Place name GPS location Site Type Period Description

Kat005/11 Matikiti -9.055705 39.535146 (GPS89) Pot findspot Unknown A small cut through the mangroves with erosion occurring along this stretch. Mangrove cutting is occurring. 1 piece of eroded pot.

A sand spit is landward.

39.526626 (GPS81)

39.529682 (GPS85)

39.531655 (GPS86)

60

Site Number

Kat006/11

Place name

Jangwani

GPS location

-9.051714

Site Type

Pot scatter

Period

Unknown

Description

On the beach terrace close to a modern shell midden beside the mangrove path.

Site Number

Kat007/11

Place name

Jangwani

GPS location

-9.050572

Site Type

Pot scatter

Period

Late 12th to late 13th century

Description

Jangwani after shell midden till sand spit on mangroves, brought up by coconut planting.

Undecorated pottery probably Early Kitchen Ware Type II Period II (Chittick, 1974: Fig97c).

Site Number

Kat008/11

39.531766 (GPS94)

39.529084 (GPS96)

Place name GPS location

-9.054398

39.527332 (GPS99)

Site Type

Pot scatter

Period

Unknown

Description

Cassava field with a baobab. Some unidentified pot in disturbed agricultural soil. Also nearby shells (fundule, gowegowe, tondo) lime making, charcoal.

Site Number Place name GPS location Site Type Period Description

Kat009/11 -9.056486 39.526207 (GPS106) Pot scatter Late 12th to 14th century (GPS106) Pottery amongst cassava burnt possibly Types 12 and 22, and M19 and M22 consisting of jars and bowls (Period IIIa).

An almost complete pot with stick incisions around neck was found besides blocks of sandstone piled up against orange tree.

The pot could be early kitchen ware Period II Type 2 late 12th to 13th century Fig 98b (Fig. 8). Other pieces Fig 98d, 99g, 98f (GPS101) Decorated pot on upper beach, mangroves seaward, broken shelly coarse sandstone cliffs. Local pottery Type 3 (Late TIW) Period II 61

Fig. 8: An almost complete decorated pot on Sanje ya Kati (Kat009/11).

Msekera Site Number Place name GPS location Site Type Period Description

Mse001/10 Nyati, Msekera -9.038459 39.481888 Lithic scatter LSA A lithic scatter on the jangwani inter-tidal zone. Wild animals including elephant and hippopotamus are common in this area. A quick exit was made after sighting a buffalo in the bush. A hut was nearby.

Site Number Place name GPS location Period Description

Mse002/10 Kinazi -9.043042 39.487832 (GPS201) th th 11 to 13 century Mingi LP, fishermen shelters, fish trap basket, fish drying rack, rope. Pot found was late TIW, quartzite pounding stone for mangrove whelk. The mangrove whelk is broken at the mouth of shell.

62

Site Number

Mse003/10

Place name

Mgongowangisi, Msekera

GPS location

-9.042772

Site Type

Pottery

Period

Unknown

Description

Rim of pot, also some undecorated pot at Kavogo -9.039588 39.492151 (GPS210)

39.491112 (GPS207)

Pande Site Number

Pan001/10

Place name

Mitonga

GPS location

S09.101702

Site Type Period

Lithic scatter LSA

Description

On jangwani near overgrown bush, quartz lithics, must be eroding out of dense bush landward

Site Number

Pan002/11

Place name

Mtomkuru

GPS location

-9.109248

Site Type

Eroding midden

Period

Late 12th to late 13th century

Description

(GPS224) Single pot on salt pan (GPS225) Mtomkuru sand spit: Lots of pottery along a modern salt works sand bank that is near a sand spit with lots of pottery, mangrove whelk, reef coral, burnt conglomerate. A jangwani is landward and seaward of spit. The profile here is steep sandy beach leading to HWM where beach ridge is present. The pottery has the thinning to rim and flat rim typical of Type II Early Kitchen ware. (GPS219) small pot on jangwani found near rice field

E39.567051 (GPS215)

39.577406 (GPS225)

63

Site Number Place name GPS location Site Type Period Description

Pan003/11 Mikumbi -9.110194 39.579734 (GPS227) Pot scatter EIA, 12th-14th century and 19th century (GPS227) Pottery on beach amongst dark mangrove soil, containing Kwale Ware (Fig. 9), Local pot Period II and M51. This is located at the place where the spit (GPS225) reaches the water channel. The pottery appears to be in the mangrove soil and might not be in situ. (GPS228) Mikumbi landing place has a dhow, steep sandy beach, beach ridges landward, canoe, coconut plantation, this coast has little mangrove width, pumice and some pot.

Opposite is Mwanikiwambi Mosque on Songo Mnara.

Fig. 9: Kwale Ware from Mikumbe, Pande.

Sangarungu Island TP1 Sangarungu TP1 (a 1.0x1.0m test pit) was excavated, in a coconut plantation, 16.9m west (060/240Ëš) of the SW corner of a ruin of a 19th-century house on Sangarungu Island at S09.034146 E039.549187 (Figs. 10 and 11). It was excavated to determine if the foundation sediments of the sand spit consisted of reef coral similar to the causeways and platforms on the east coast, and to date the formation and use of the sand spit (to determine if it is contem64

Fig. 10: Excavation of Sangarungu TP1.

Fig. 11: Sangarungu TP1.

65

66

Large pebble-sized piece of reef coral

Mangrove whelk shell

Loose light yellow 50% fine sand, 50% medium sand

Loose wet beige shelly 10% coarse sand, 45% medium sand, 35% fine sand, 10% shells

5

4

0

Contexts from Samgarungu TP1

0

Subangular worn 3-4.5mm thick as before in fine grained matrix

4

7

0

No evidence of exploitation

No evidence of exploitation

Carinated pot, flat rim. Possibly Type 38 Late Kitchen Ware 17th to 19th century

Loose light beige-grey 50% medium sand, 45% fine sand. Shell fragments.

2

Angular, orange and dark grey/ black color, 1 sharp angled piece, size 0.5 to 46mm, 3 fish vertebra, 2 other fish bones

3

34

Small little lip, similar Late Kitchen Ware to Context 2 but coin suggests late 19th century to early 20th century

Some black burnishing, angular 56mm to 8mm, orange and black color, generally less than 1cm thick, 1 bone, 1 German East African coin (1 Heller copper coin), 2 small white beads

98

0

Loose beige-grey 70% medium sand, 30% fine sand. Shells and charcoal.

3

2

73

Date

Description of artefacts

247

No. of decorated pieces

Loose well sorted light browngrey, 59% medium sand, 40% fine sand, reef coral 1%. Lots of roots, some charcoal

No of rims

1

No. of pieces

Weight of pot

Description of contexts

Context

porary with the ruins on Songo Mnara). The results could help understand where the landing place used during occupation of the ruins is located. Sangarungu Island is used today as the port. The pit was left open after excavation as the headman wanted to put a well there.

Preliminary conclusions The inter-tidal and coastal survey has revealed further evidence of Late Stone Age (LSA) occupation around the bay including at Pande and Msekera. It has shown contemporary links between Sanje ya Kati and the mainland at Msekera during the 11th to 13th centuries with the presence of late TIW discovered at Msekera. The Early Iron Age (EIA) Kwale Ware found at Pande is significant in that it is early evidence of iron-age farmers on coast exploiting the maritime enviornment. Wynne-Jones (2005: 102) has recorded Kwale ware in Kilwa region, but not on the coastal plain. However, the find was disturbed and the EIA and TIW sites would need to be at looked more closely on a future visit.

The investigation of the reef coral causeways and platforms, previously observed from satellite imagery, did not reveal the man-made features expected. The causeways snaked across the lagoon and consisted of a mixture of sand and reef coral. This was different to the linear features of reef coral observed on the east coast of Kilwa. Ethnographic and placename evidence suggests use and ownership of this resource such as Funguyabintimwenye and Jomalambwana. The reef coral along the coast is a vast resource for the production of lime.

Further archaeological evidence of its exploitation could be from pottery scatters at Baraka and Watiro Islet. Further survey is planned in 2012 to the south of Songo Mnara, which will reveal further details on these features.

Preliminary analysis of Sangarungu TP1 has not revealed evidence of being founded on reef coral. Further analysis of the pottery is necessary. There was no evidence of its use as a port though another excavation closer to the modern landing place would reveal more information.

Acknowledgements The author is grateful to Dr Ambreena Manji, Director of the British Institute in Eastern Africa, Dr Colin Breen of the University of Ulster, Dr Stephanie Wynne-Jones of University of York and Dr Jeffrey Fleisher from Rice University for supporting the research.

Jack Stoetzel, Erendira Quintara Morales and Juliette Chaussen accompanied on part of the field survey. The locals of Songo Mnara who helped in the survey were Abdu Chenji and Mahibu Saidi. A grant from the BIEA allowed the author to participate in this project.

67

68

References Burton, R. F. 1872 Zanzibar: City, Island and Coast. 2 vols. vols. Tinsley Bros., London.

Chittick, H. N. 1961 Excavations at Songo Mnara. Annual Report of the Department of Antiquities. Govern ment Printer, Dar es Salaam, pp. 4-6.

1974 Kilwa: an Islamic trading city on the East African coast. Memoir 5. British Institute in Eastern Africa, Nairobi and London.

Dorman, M. H. 1938 The Kilwa Civilization and the Kilwa ruins. Tanganyika Notes and Records 6:61-71.

Fawcett, W., A. LaViolette and N. J. Karoma 1989 Final Report: Archaeological Investigations at Coastal Sites Between Dar es Salaam and Bagamoyo. Archaeology Unit, History Department, University of Dar es Salaam.

Fleisher, J. B. 2003 Viewing Stonetowns from the Countryside: An Archaeological Approach to Swahili Regional Systems, AD800-1500. Unpublished PhD thesis, Department of Anthropology, University of Virginia.

2004 Behind the Sultan of Kilwa's 'Rebellious Conduct': Local Perspectives on an Interna tional East African Town. In African Historical Archaeologies, edited by A. Reid and P. Lane, pp. 91-124. Kluwer Academic/Plenum Publishers, New York.

Fleisher, J. B. and S. Wynne-Jones 2010 Archaeological Investigations at Songo Mnara, Tanzania: Urban Space, Social Memory and Materiality on the 15th- and 16th-century Southern Swahili Coast. Submitted to Antiquities Division, Dar es Salaam.

2010 Kilwa-type coins from Songo Mnara, Tanzania: New Finds and Chronological Implica tions. Numismatic Chronicle 170:494-506.

2012 Finding Meaning in Ancient Swahili Spatial Practices. African Archaeological Review 29(2/3):171-207.

Fleisher, J. B., S. Wynne-Jones, C. Steele and K. Welham 2012 Geophysical Survey at Kilwa Kisiwani, Tanzania. Journal of African Archaeology 10(2).

Garlake, P. S. 1966 The early Islamic Architecture of the East African Coast. BIEA Monograph. Oxford University Press, London.

69

Haughton, S.H. 1938 Lexicon de Stratigraphie, vol. 1: Africa. Thomas Murby, London, p. 203, 229.

LaViolette, A., W. B. Fawcett and P. R. Schmidt 1989 The Coast and the Hinterland: University of Dar es Salaam Archaeological Field Schools, 1987-88. Nyame Akuma 32:38-46.

Mathew, G. 1959 Songo Mnara. Tanganyika Notes and Records 53:154-160.

Museum of London Archaeology Service 1994 Archaeological Site Manual (Third Edition). London: Museum of London.

Nicholas, C. et al.

2006 Stratigraphy and sedimentology of the Upper Cretaceous to Paleogene Kilwa Group, southern coastal Tanzania. Journal of African Earth Sciences 45: 431-466.

Pawlowicz, M. 2009 Archaeological Exploration of the Mikindani Region of the Southern Tanzanian Coast.

Nyame Akuma 72:41-51.

Pollard, E. 2008 The Archaeology of Tanzanian Coastal Landscapes in the 6th to 15th Centuries AD (The Middle Iron Age of the Region). Cambridge Monographs in African Archaeology 1873.

BAR International Series, Oxford.

2011 Safeguarding Swahili Trade in the 14th and 15th Centuries: a Unique Navigational Com plex in South-East Tanzania. World Archaeology, 43 (3): 458-477.

Pollard, E., J. B. Fleisher and S. Wynne-Jones 2012 Beyond the Stone Town: Maritime Architecture at Fourteenth - Fifteenth Century Songo Mnara, Tanzania. Journal of Maritime Archaeology 7(1).

Pradines, S. and P. Blanchard 2005a Archéologie et préservation du patrimoine: le projet franco-tanzanien de Kilwa, 20022005 Nyame Akuma (63):20-26.

2005b Kilwa al-Mulûk. Premier bilan des travaux de conservation-restauration et des fouilles archéologiques dans la baie de Kilwa, Tanzanie. Annales Islamologiques 39:25-80.

Stoetzel, J. 2011 Field Report: Archaeological Survey of Songo Mnara Island. Nyame Akuma 76:9-14.

70

Sulas, F. and M. Madella 2012 Archaeology at the micro-scale: micromorphology and phytoliths at a Swahili stonetown. Archaeological and Anthropological Sciences 4(2):145-159.

Sutton, J. E. G. 1998 Kilwa: A History of the Ancient Swahili Town with a Guide to the Monuments of Kilwa Kisiwani and Adjacent Islands. Azania 33:113-169.

Wynne-Jones, S. 2005 Urbanisation at Kilwa, Tanzania, AD800 – 1400. Doctoral dissertation, University of Cambridge.

2007 Creating urban communities at Kilwa Kisiwani, Tanzania, AD 800-1300. Antiquity 81:368-380.

Wynne-Jones, S. and J. B. Fleisher 2010 Archaeological Investigations at Songo Mnara, Tanzania, 2009. Nyame Akuma 73:2-8.

2011 Archaeological Investigations at Songo Mnara, Tanzania, 2011. Nyame Akuma 76(3-8).

2012 Coins in Context: Local Economy, Value and Practice on the East African Swahili Coast.

Cambridge Archaeological Journal 22(1):19-36.

71

72

Appendix 1 Archiving and storage: Songo Mnara 2011 Local ceramics Most locally-produced ceramics from the site were taken to Rice University, Houston, for analysis. The cataloguing is ongoing; a full list will be made available to the Antiquities Unit, Dar es Salaam, as soon as it is available. All ceramics will be returned to Tanzania when analysis is complete, and storage arrangements will be agreed with the Antiquities Officers.

Imported ceramics The imported ceramics from Songo Mnara were bagged and labelled by context, and stored in wooden boxes in the site house on the island. Analysis was completed in the field, and the numbers are presented fully in the trench descriptions above. The imported ceramics will be moved to more appropriate storage along with the locally-produced ceramics, when returned.

Beads As reported, the beads from Songo Mnara are currently being analysed in the USA by Ms Marilee Wood.

Coins The coins found during 2011 excavations are being analysed by Mr. John Perkins. The coins themselves are being stored in acid-free coin slides, as per current museum practice.

They will be returned to Dar es Salaam as soon as the project is complete.

Plant macrofossils The results of flotation on soil samples from Songo Mnara are under analysis by Dr Sarah Walshaw at Simon Fraser University, Vancouver. A copy of the report will be made available once it is complete. The samples will remain at the laboratory facilities in Vancouver.

Phytoliths and micromorphology slides from the 2011 excavations are being analysed

by Dr Federica Sulas. A copy of the report will be made available once it is complete. The samples will remain at laboratory facilities at the University of York once analysis is complete.

Bones The bones from Songo Mnara are being studied by Dr. Erendira Quintana Morales. The bones will be returned for storage along with other finds from the site.

Reports A copy of this report will be deposited with the Tanzanian Commision for Science and Technology (COSTECH), the British Institute in Eastern Africa, Nairobi, and made available via the Songo Mnara project website (http://www.songomnara.rice.edu/). All subsequent reports and artefact information will be made available in the same places.

73

74

Appendix 2: Artifact Tables

75

76

context

16029

16028

16027

16025

16024

16021

16020

16019

16018

16017

16016

16015

16013

16012

16011

16010

16009

16008

16007

16006

16005

16004

16003

16002

16001

Local pottery wt.

Local pottery count

11 47 93 35 59

141 495 1755 488 1090

117 1007 67 1522 37 541 890 11803 123 952 153 1369 10 115 26 603 2 25 193 2398 16 117 172 1676 8 61 46 600 270 4769 3 23 46 1726 3 38 14 182

Glass bead

3

1 4 10

4 23

6

7

12 9 6 25 3 2

Shell beads

1

1

1

1 1 3 8

aragonite bead/disc

1

1

1

other beads

1 1

4

6 1 6

copper coins count

1

1

1

1

2

Imported pottery count

1

1

3 5

5

5

1

2

5

1 1

Import pottery wt.

0.1

10.4

13.5 32

39.4

69.3

10.6

7.5

5.7

65 21.3

Glass count

11

14

3

2 1 4

1 1 2 5 1 1

Glass wt.

2

1.5

0.3

0.1 0.1 2.2

0.1 0.1 4.4 5.1 0.1 0.1

1

1

2

2

1 1 5

spindle whorls

spit

SM016 Artifacts

copper wt.

copper count

2 1.3

1 0.1

2 0.4

quartz wt.

1.9

19.5 19

2

fish bone wt.

ground stone

92

15 34 116

10 12 41 5 6

11 5 24 151 19 41 267 201 37 21 19 47 8 4 9 8 262 62 17 88 39 6 6 77 151 80 9 5 18 6 2 27 7

other bone wt.

iron slag wt.

iron wt.

iron count

lamp frag

jiko frag

77

16043

16042

16039

16038

16037

16036

16035

16034

16033

16032

16031

16030

70 13 15 7 13 6 12 85 236 3 177 3

864 91 114 67 107 114 156 1115 3530 43 2037 143 9

1 5 34

1

4 12 1

4

6

1

2

1

1

1

1

2.5

1

4.6

0.1

1.5

1

0.2

2 28.2 1 2.9

1

1

1

1

1

0.8 2 0.2

2 1.7

1.2

31 23 8 8 5 6 10 72 224 1 56 22

3 20 15 8 22 60

8

78 17018

17017

17016

17015

17013

17012

17011

17010

17009

17007

17006

17004

17003

17002

17001

context spit

46 505 149 1188 5 46 8 110 2 16 2 69 26 269 5 62 1 10 1 13 32 334 2 22 107 249 5 62 9 85

Local pottery count Local pottery wt.

1

4

2

4

5 1

Glass bead

1

7 1

Shell beads aragonite bead/disc

3

3

1 1

1 67 10 294 3 5 1 1 2 3 1

other beads copper coins count

1

Imported pottery count

1

Import pottery wt.

1 0.1

SM017 Artifacts

Glass count Glass wt. spindle whorls jiko frag lamp frag iron count iron wt. iron slag wt. copper count copper wt.

6.4

0.8

quartz wt. ground stone

10

8 28 17 3

6

6

fish bone wt.

6

10 4

5

other bone wt.

79 18028

18024

18023

18022

18021

18020

18019

18018

18017

18015

18014

18013

18011

18010

18009

18007

18006

18005

18004

18003

18002

18001

context spit

3 8 2 1 109 2 11 7 18 11 2 4 2 3

2

127 1 13 13 218 51 119 65 270 100 19 41 9 83

13

0

0 18

Local pottery count

0 2

0

Local pottery wt.

3 1 5

42 18

Glass bead

1

1

1

2 1

Shell beads

1

aragonite bead/disc

1

2

other beads copper coins count

3 1

1

1

1

Imported pottery count

13 5

23

10

18.5

Import pottery wt.

SM018 Artifacts

Glass count Glass wt. spindle whorls jiko frag lamp frag iron count iron wt. iron slag wt. copper count copper wt. quartz wt. ground stone

70 28 21 10 97 41 7 12 3 47 9

1

47 48 13 12 4 6 5 4

fish bone wt.

14

13 3

9 11

3

36 18 5 4 6 18

other bone wt.

80

context

b

c

d

19003

19003

b

19009

19016

19014

19013

19012

a

19009

19009

19008

19007

19006

19005

19004

a

19003

spit

19003

19003

19002

19001

Local pottery wt.

Local pottery count

41

116 37 146 0 25 81 419 159 559 286 73 14

367

393 654 2997 0 218 1057 5567 1860 7027 3192 1350 213

8 140 157 1106 479 9193

Glass bead

1 1

7 7 11

2 13

Shell beads

1

1 1

aragonite bead/disc

1

1 1

1 3 2

other beads

1

7 3

1 5 2

2 1 2

1

copper coins count

2

2 1

Imported pottery count

22.3

Import pottery wt.

2

10 1 1 12 4.5

55.6 4 4.9 31.4

1 2.5 1 1 14 158.3 4 58.1

5

Glass count

2

1

3 1

1

2 4 13

Glass wt.

1

0.3

0.7 0.2

0.1

0.8 0.8 3.8

2 6

4 1

2

2

spindle whorls

SM019 Artifacts

jiko frag

2

iron count

1

4 1

iron wt.

0.9

4 13.6

iron slag wt.

8.6

copper wt.

copper count

2 1.1

1 0.1

quartz wt.

9 1

14

0.2 0.8

fish bone wt.

ground stone

104 34 5 2 25 38

10 20 4.6 59 19 3 11 13 121 85 32

4 12

3

29 11

8 7 20 3

3 5 22 20

other bone wt.

lamp frag

81

context

20020

20019

20017

20016

20015

20014

20013

20012

20010

20009

a

b

20007

20008

a

spit

20007

20007

20006

20005

20004

20003

20002

20001

Local pottery wt.

Local pottery count

39 2

8 48

456 13

77 690

1952 24193 3 92 140 519 2 14 3 354 80 1012 633 6855 115 1649 74 1040 213 2364 3 78 5 61 13 91 0 0

Glass bead

1

1

1

3 16

37

Shell beads

1

1 1

2

aragonite bead/disc

1 1

1

1

1

other beads

1

1

2

1

7

copper coins count

1

1

1 3

4

Import pottery wt.

Imported pottery count

1

1 1

1.3 0.7

1 0.1 9 138.3 4 49.6 3 85.2 10 45.8

1

28 293.7

Glass count

6

4.1

56

Glass wt.

1

0.1

21 14.6

13

1

14

1

4 2 1 2

10

spindle whorls

SM020 Artifacts

lamp frag

3

iron count

1

3

iron wt.

0.9

28.7

iron slag wt.

7.5

copper wt.

copper count

4 3.4

12 4.1

1 7.5

fish bone wt.

ground stone

quartz wt.

150

14.9 62

5 6

9

47 8 9 108 125 10 6 28 25 28 8 1 3 5 205

15.7 349.6 648 223 13 54 217 21 18

other bone wt.

jiko frag

82 22002

22001

21012

21011

21010

21009

21008

b

21006

21007

a

21006

21005

21004

21003

21002

21001

context spit

Local pottery wt.

1557 1882 1783 945 630 1491 1506 266 161 351 144 60

Local pottery count

109 82 84 58 50 106 114 19 12 88 8 5 5 10 3 3

2

2

9

Glass bead

2

Shell beads

1

1

aragonite bead/disc

2 2 2

1

1

2

other beads

1

1

1 1

1

1 1

copper coins count

2

2 1

5.5

20 6.5

10.1

28.2

8

3 131.2

3

1

2

Imported pottery count Import pottery wt.

1

5

0.2

0.8

1

1 1 1 4 1

1

2

SM021 Artifacts

Glass count Glass wt. spindle whorls jiko frag lamp frag iron count iron wt. iron slag wt. copper count copper wt.

7.7

21.5

334

98

quartz wt. ground stone

4 8 27 9 17 4 9 15 5 10 47 8 15 13 274 3 19 12 37 8 7 2 8 184

fish bone wt. other bone wt.

83 23025

23021

23019

23017

23016

23015

23014

23013

23012

23011

23010

23009

23007

23006

23005

23001

context spit

Local pottery wt.

1052 359 302 1281 2859 358 198 41 2 485 875 9446 453 1402 600 34

Local pottery count

79 19 17 68 161 23 15 2 1 2 39 419 24 47 34 4 4

3

1

Glass bead

1

5

Shell beads

5

8

aragonite bead/disc

4

other beads

19 2 9 11 9

copper coins count

2 1

1

Imported pottery count

10 7

13.4

Import pottery wt.

1 0.2

1

5

SM023 Artifacts

Glass count Glass wt. spindle whorls jiko frag

5

lamp frag

1

iron count

4.5

iron wt.

4.5

iron slag wt. copper count copper wt.

25.5

quartz wt. ground stone

23

24 5

113 7 9 33 15 153 40 5 22

7 5 3 27 46 21 7

fish bone wt. other bone wt.

84

context

25010

25009

25008

25007

25006

25004

25003

25002

25001

24006

24005

24004

24003

24002

24001

Local pottery wt.

81 1540 1204 3674 2206 1174 62 1556 1224 59 2142 181 544 376 1046

Local pottery count

8 90 137 479 179 88 6 102 86 4 148 13 44 42 86

Glass bead

1 10

Shell beads

1

aragonite bead/disc

1

other beads

23

1

2 1

copper coins count

5

Imported pottery count

2

2

1

2 2 5 3 2

Import pottery wt.

5.7

24.3

2

3.5 2 24.3 8.5 6

Glass count

1

1

1 3 1

Glass wt.

0.4

0.3

0.1 0.3 0.3

spindle whorls

2 1

1

iron count

lamp frag

jiko frag

spit

SM024, SM025: Artifacts from Burial Excavations

iron wt.

18

iron slag wt.

12.7 80.1 2.6 227 62.8

copper wt.

copper count

1 0.1

quartz wt.

5.3 109.9 6.5

fish bone wt.

10 4 4 7 16

35 107 40 14 6 6 14

3

6 9

15 6

13

19 49 26 19

5

other bone wt. ground stone

85 27007

27006

27005

27004

27003

27002

27001

26005

26004

26003

26002

26001

context spit

275 3717 207 3243 115 1625 4 100 0 0 131 1308 281 3286 60 2330 722 10460 222 3253 20 582 9 334

Local pottery count Local pottery wt.

Glass bead Shell beads

1

1

aragonite bead/disc

1

1

other beads

1 2 11.6 23.2

30.3

73.8

19 136.3 4 12.3 1 11.5

5 9

1

4

3 1 2 0.5 0.1 0.6 2

1

1

1 2

SM026, SM027: Artifacts from Burial Excavations

copper coins count Imported pottery count Import pottery wt.

Glass count Glass wt. spindle whorls jiko frag lamp frag iron count

43.9

iron wt. iron slag wt. copper count copper wt.

4.8

389 25.8

quartz wt. ground stone

26 19 8 115 40 21 12

30

fish bone wt.

5

15 26 11 31 17

18 2

other bone wt.

86

SOA

30018

30013

30012

30011

30008

30006

30005

30004

30003

30002

30001

29005

29004

29003

29002

context spit

49 7 5 136 48

20 366 4 35 80 1969 1108 23559 4 241 560 10098 5 138 60 2104 18 445

Local pottery count

669 117 116 1940 2078

Local pottery wt.

Glass bead Shell beads

2

1

1

4

aragonite bead/disc

70

2

1

1

other beads

9

1

1 4 1 3

2 3

1

69 86.3

118

13.3

9 163.1 1 133 5 59.3

1

7

14

1 8

3.2

3.3

0.4 2.8

1

8

4 3 402.1

SM029, SM030: Artifacts from the Central Mosque

copper coins count Imported pottery count Import pottery wt.

Glass count Glass wt. spindle whorls jiko frag lamp frag iron count iron wt. iron slag wt.

19 7.3

copper count copper wt. quartz wt. ground stone

10

14

18

123

fish bone wt.

51

97

other bone wt.