See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236157626

The Fezzan Project 1999: preliminary report on the third season of work

Article in Libyan Studies · January 1999

DOI: 10.1017/S0263718900002831

CITATIONS READS
10 177

11 authors, including:

David Mattingly Nick Drake

University of Leicester King's College London
229 PUBLICATIONS 4,217 CITATIONS 189 PUBLICATIONS 9,720 CITATIONS

SEE PROFILE SEE PROFILE

Sue McLaren Ruth Pelling

University of Leicester Historic England
101 PUBLICATIONS 2,904 CITATIONS 43 PUBLICATIONS 610 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Nick Drake on 24 March 2016.

The user has requested enhancement of the downloaded file.

Libyan Studies
http://journals.cambridge.org/LIS

Additional services for Libyan Studies:

Email alerts: Click here

Subscriptions: Click here
Commercial reprints: Click here
Terms of use : Click here

The Fezzan Project 1999: preliminary report on the third season of work

David J. Mattingly, Mohammed al-Mashai, Phil Balcombe, Nick Drake, Stephanie Knight, Sue McLaren, Ruth
Pelling, Tim Reynolds, David Thomas, Andrew I. Wilson and Kevin White

Libyan Studies / Volume 30 / January 1999, pp 129 - 145

DOI: 10.1017/S0263718900002831, Published online: 03 March 2015

Link to this article: http://journals.cambridge.org/abstract_S0263718900002831

How to cite this article:

David J. Mattingly, Mohammed al-Mashai, Phil Balcombe, Nick Drake, Stephanie Knight, Sue McLaren, Ruth
Pelling, Tim Reynolds, David Thomas, Andrew I. Wilson and Kevin White (1999). The Fezzan Project 1999:
preliminary report on the third season of work. Libyan Studies, 30, pp 129-145 doi:10.1017/S0263718900002831

Request Permissions : Click here

Downloaded from http://journals.cambridge.org/LIS, IP address: 23.95.129.252 on 28 Feb 2016

 Libyan Studies 30 (1999)

The Fezzan Project 1999: preliminary report on the third season

of work
By David J. Mattingly, 1 Mohammed al-Mashai, 2 Phil Balcombe, 1 Nick
Drake, 3 Stephanie Knight, 1 Sue McLaren, 4 Ruth Felling, 5 Tim Reynolds, 6
David Thomas, 7 Andrew I. Wilson 8 and Kevin White 9

Abstract
This report summarises the work of the third season of the Fezzan project which took place in January 1999.
The main environmental findings of the project team of specialist geographers are providing confirmation of
dramatic climatic and environmental change over the last 100,000 years and give more precise dates for some
of these changes. The excavations in Old Germa (ancient Garama) have continued through Islamic levels, with
elements of five main phases of buildings now having been recorded. Additional standing structures, including
one of Germa's main mosques, have been surveyed. Field survey around Germa has revealed further new
settlement sites of prehistoric, Garamantian and Islamic date. Of particular importance is a series of lithic and
pottery scatters relating to neolithic occupation along the edge of the Ubari Sand Sea, to the north of Germa.
Further investigation of the irrigation channels (foggaras) has revealed significant new information about their
size, construction and probable date. The report concludes with a brief preliminary analysis of changing
settlement patterns over time.

Introduction
As in previous years, the work of the project in the Wadi al-Hayat (el-Agial) has focused
on a range of approaches designed to shed light on human adaptation to the desert
environment over the last 12,000 years, with some additional work setting this in the
context of longer term environmental, climatic and hydrological change (for previous
reports see Mattingly et al. 1997, 1998a/b, 1999; note also the links with previous British
work in the region: Edwards et al. 1999). This report will briefly describe the principal
results of the 1999 season under a series of headings, starting with the geographical
studies, proceeding to the major ongoing excavation at Old Germa, and then describing
the survey of buildings and of sites and irrigation works in the landscape around Germa.
The report concludes with a few tentative conclusions about the changing nature of human
settlement and lifestyle in the Germa region over the last 12,000 years.

Geographical and Geomorphological Studies

Work this season has concentrated on obtaining suitable materials for deriving absolute
ages of deposits of palaeoenvironmental significance that can be linked to the emerging
archaeological record. Most of the work was carried out in the immediate vicinity of
Germa, but some more wide-ranging reconnaissance was necessary to tie the results in
with other published work on Fezzan. This involved brief trips to the Murzuk and Ubari
Sand Seas and to the Wadi esc-Sciati. Further important geomorphological evidence, useful
for palaeoenvironmental reconstruction has been identified this season. This includes:
• Palaeolake sediments of Pleistocene age, perhaps coeval with the Wadi esc-Sciati
deposits studied by Petit-Maire (Petit-Maire et al. 1980). A significant exposure of
these sediments was identified south of Germa, at the base of the escarpment. Further

/ School of Archaeological Studies, University of Leicester.

2 Department of Antiquities, Tripoli.
3 Kings College, London.
4 Department of Geography, University of Leicester.
5 Oxford University Museum.
6 Cambridgeshire County1 Council.
7 MacDonald Institute, University of Cambridge.
8 Magdalen College, Oxford.
9 Department of Geography, University of Reading.

129
 D. J. MATTINGLY ET AL.

samples have been collected at Wadi esc-Sciati as part of this season's fieldwork in
order to clarify the relationship between the two sites.
• Palaeoshoreline features (a notch cut into the base of the escarpment immediately
above the palaeolake sediment mentioned above, and rounded gravel clasts interpreted
as palaeobeach deposits).
• Clayey-silt sediments of a probable Holocene lake exposed fortuitously during the
mechanical excavation of a drainage trench across the mudflat adjacent to the site at
Old Germa. Samples have been collected which will be analysed to see if they
contain any pollen or ostracods.
• Further exposures of large gypsum crystals at the interface between the Messak
sandstone unit and the underlying marls, interpreted as palaeospringline deposits.

Samples for absolute dating include:

• Spring line deposits along the escarpment edge. Large gypsum crystals related to
surface spring activity have been collected for Uranium-Thorium (U/Th) dating of
periods of high water level. These were located at El Greifa and Twesh. Stable
isotope analysis of oxygen and hydrogen of the water of crystallisation in these
materials should also indicate whether the source of water was dominated by
meteoric (rainwater percolating through the overlying Messak sandstone) or
groundwater (upwelling of the Continental Intercalaire) sources.
• Organic-rich layers interpreted as palaeolacustrine deposits. These will be dated
using U/Th analysis of the gastropod assemblage, dominated by Melanoides
Tubercolata. One date obtained from a Melanoides shell collected during the 1998
field season has yielded a U/Th age of 84,500 + 4,600 BP, which correlates with
Oxygen Isotope sub-stage 5a. Stable isotope analysis of the total organic carbon
fraction of these sediments, carried out on samples collected last year, has enabled
a typology of these organic-rich horizons to be developed, which has been of
considerable assistance in targeting sampling effort this year.
• Optically-stimulated luminescence (OSL dating) of the palaeolake sediments.
These were exposed in the drainage trench across the mudflat adjacent to Old Germa.
Samples of lacustrine clayey-silts and underlying aeolian sands were collected for
this purpose. These dates will be tied into radiocarbon dates associated with neolithic
artefacts that appear to be related to the northern edge of the palaeolake.

A suite of samples was also collected for chemical and stable isotope analysis to trace
the sources of evaporitic minerals in this area. These analyses will focus on sulphur
isotopes, to identify key aspects of the sulphur cycle in this area (essentially this is a way
of establishing through analysis of the sulphates in the sediments the balance between
groundwater and rainwater). Samples of bedrock, contemporary groundwater, and
evaporitic minerals will be used.

Recommendations for the next season of fieldwork are:

• Use of differential GPS techniques to determine the height of the palaeoshoreline
mentioned above. This will enable the exact extent of the palaeolakes (dated using
the approach outlined above) to be determined from digital elevation data.
• Use of Ground Penetrating Radar to analyse subsurface sedimentary features
associated with palaeoshorelines, surveying potential settlement sites identified by
130
 THE FEZZAN PROJECT 1999

Figure 1. Germa, site Gl, the Phase 4 building, looking south (Photo: T. Savage).

surficial artefact concentrations. Application of GPR to survey foggara channels is

unlikely to be successful due to the amount of channel collapse and infilling which
has been observed during this season. However, an experimental attempt may be
worthwhile.

The work carried out so far indicates that there has been significant environmental change
over the last 100,000 years. There is evidence for a lake level stand at the end of the last
interglacial. Palaeospringlines at the base of the hamada and other zones of upwelling in
the Wadi al-Hayat indicate evidence of high ground water tables in the past, which have
fallen significantly in recent times.

The Excavations in Old Germa

The 1999 Fezzan Project excavations at Old Germa made significant progress. Work
centred on the clarification and removal of the large Phase III building which was exposed
last season (Mattingly et al. 1998a, 126-29). The three main areas of investigation in 1999
were:
1. Rm. 3.1, a large room in the east of the excavation area;
2. The central, stone-lined well;
3. The southern area which was only partially excavated in 1998.
Rm. 3.1 had two wide, mud-brick walls in the north and east and a blocked doorway in
the west, leading into a series of small rooms with apparent storage installations which
were excavated in 1998. The size of the walls suggests that the room was important,
though virtually all of its Phase III surface was missing and the deposits came straight
down onto the tops of Phase IV walls. The Phase IV architecture proved interesting,
131
 D. J. MATTINGLY ET AL.

however, since the walls encompassed a rectangular room (Rm. 4.1) with a large expanse
of burnt fill on its surface which dipped significantly to the west (Fig. 1). Other Phase
IV walls were traced to the north, suggesting the presence of several more small rooms.
A large central 'plug' of compact deposits was found in the west of Rm. 4.1 and this,
along with the presence of the Phase III well immediately to the west, suggests that an
earlier, unlined well may have existed in this area, causing later deposits to slump into
it. An alternative explanation is that the pit dug for the insertion of the Phase III well
was far larger than the central core which was lined with stone around the well shaft. To
the east of the excavation site, the large stone wall 076 was found to be broadly
contemporary with the Phase III mud-brick walls of Rm. 3.1. A thick deposit of mud-
brick packing/paving was found to the east of wall 076, but the limited area of deposits
here means that the function of this presumably significant structure remains uncertain.
Ayoub's excavations above the Garamantian temple, which lies immediately east of our
excavations are believed to have demolished the central mosque of Old Germa (Ayoub
1967). If that is the case, the stone wall 076 and the mud-brick paving may have been
associated with that complex. A connection with one of the major public buildings of
medieval Germa would certainly be appropriate given the scale of these features.
A detailed stratigraphic study was made of the stone-lined well and the surrounding
paved area. The well was half-sectioned and found to cut through the Early Phase III
architecture, the surface and walls of Phase IV Rm. 4.1 and into a Phase V mud-brick
wall. The lining of the well appears to have been carefully constructed and a clayey
mortar/fill used to bond the stones. A solid retaining wall was constructed in the west
(again cutting through the earlier Phase III storage installations) and a jumble of stones
was then added to complete the structure. It is possible that the digging and construction
of the well, which must have required investment of time and labour, coincided with other
later Phase III alterations to the layout of the building, particularly those relating to routes
of access — for example, the blocking of the doorway in the west wall of Rm. 3.1 and
of the doorway in north-south wall 250 / 313 which controlled access from the external
courtyard area Ex. 3.2. There are interesting social implications involved in the
construction of a well (which appears not to have been open to communal access) on one
of the highest points of the site and in the modifications to the building's plan. The scale
of the building and these features suggest a relatively high status occupant.
The excavations in the southern part of the trench were very informative and added
significantly to our understanding of the Phase III building. This area seems to have been
the centre of domestic activity, with a tanur (bread oven) and presumed small storage
room in Rm. 3.9 and several hearths and ovens to the south, in a walled, but probably
open area. Before these features were reached, however, an intermediate phase of
architecture was found. Two rooms in the south-west closely mirrored Phase II rooms (Rm.
2.1 and Rm. 2.2) and their walls probably acted as foundations for these later structures.
An enigmatic square paved area, which dipped to the east, was located to the east of wall
293. Further investigations uncovered a stone-lined channel leading from this paved area
to a deep, stone-lined hole (Fig. 2). The hole seems to be too narrow to have functioned
as a well, so it is possible that it acted as a soakaway drain for a washing area. Major
east-west and north-south Phase III walls run under these intermediate phase structures in
the southern part of the trench.
The material culture excavated in 1999 remains meagre, although it appears to contain
more 'luxury' items such as painted pottery, ostrich egg-shell, glass and stone beads than
that of previous seasons' excavations. In the west of the excavated area, two clusters of
clay loomweights were found — in Rm. 3.4 and in the north of courtyard Ex. 3.1 —
suggesting that weaving took place in this area. More Roman fine ware (African Red Slip)
132
 THE FEZZAN PROJECT 1999

Figure 2. Stone-lined drain and soakaway in the southwest part of site Gl (Photo: I), Ihonuni.

rims have also turned up, as might be expected since the trench is now just over 1 metre
above Garamantian levels in the Daniels/Ayoub section (recorded by us as G2, see
Mattingly et al. 1997, 18; 1999). The 1999 excavations have provided a more detailed
picture of the layout, use and modification of the Phase III building, while the initial
indications are that Phase IV and Phase V buildings are just as complex.

Analysis of Animal Bones and Plant Remains

The faunal remains recovered this season have proven far more prolific than those from
the previous seasons, there being a total of c. 3,000 fragments (for previous work, see
Mattingly et al. 1998a, 126-27; 1999). These have not been in any better condition
however, and consist of almost 50 per cent unidentified fragments and splinters, similar
to last season's percentage and probably due to the same reasons, namely a saline post-
depositional environment and high degree of fragmentation in antiquity. Nevertheless, the
numbers provide considerable potential for investigation. It is hoped that any perceived
changes in animal husbandry over time will, with other specialist evidence, shed light on
the changing diet of the population in Germa.
A remarkable consistency exists between the bone remains from all three seasons
regarding the exploitation of sheep/goat (ovicaprids). These are by far the most common
species at almost two thirds of the identified bone fragment count. From toothwear
evidence, the majority seem to have been kept to a relatively mature age, probably in
order to utilise fully secondary products of milk and wool. Some neonatal sheep/goat
bones indicate these were being kept by the community rather than brought in from further
afield.
Frequent low quantities of cattle and chicken bones suggest these animals were
133
 D. J. MATTINGLY ET AL.

commonplace, whereas bones of camel and pig (the latter something of a surprise in a
medieval context) were found in higher densities, but in fewer contexts. There is no
evidence of butchery on the porcine bones, but camel, cattle and sheep were certainly
being processed post mortem, and frequent splitting of the bone suggests the removal of
marrow. This would also seem to confirm the status of the deposits as domestic refuse.
All elements of the carcass were recovered, and no specialised use of the bones seems to
have been occurring, suggestive of a self-sufficient community. The bones have not yet
been fully investigated by phase or context, but the quantity of material will repay further
work, hopefully illuminating the more subtle differences between use of animals during
phases and parts of the site.
Ageing of cattle was restricted by the paucity of mandibles recovered, but epiphyseal
fusion data may confirm the perceived trend for cattle to die immature, suggesting they
were raised as 'meat' stock. Cattle require more water for survival than do sheep or goats,
and the apparent rise in numbers in Phase III and IV contexts may be interpreted as
evidence of a wetter environment, perhaps suggesting that irrigation was then maintained
at a higher level than later.
Other species identified include gazelle, hare, and possibly ostrich. These wild animals
may be turned to in times of need as a supplement to the diet, but more detailed analysis
of frequencies is required to determine whether, and if so, when, this was the case.
Gnawing on some of the main meat-bearing bones of ovicaprids may have been due to
rats, dogs or jackals, all of which have been recorded in phase III and IV levels, and the
skull of one young canid had been butchered across the jaw and deliberately (?) deposited.
Patterns of butchery are distinct, and sudden changes in these from one period to the next
could indicate changes in population.
The sampling of deposits for the analysis of botanical remains continued during the
1999 excavation season. In total 35 samples were taken from 28 Phase III deposits. The
volume of material sampled was generally 10 litres with occasional smaller samples.
Deposits sampled included floor deposits and room fills, pit fills, possible courtyard
deposits and the ashy fills of hearths and ovens. Twenty-five samples have now been
processed. Samples were first dry sieved through a 0.5 mm mesh to remove the bulk of
the silt and fine sand. The remaining material was then placed in a bucket and processed
by hand using water flotation. The resulting flot was collected onto a 0.5 mm mesh. The
residues were washed through a 1 mm mesh. Both flots and residues were allowed to air
dry before being bagged up separately. The heavy residues will be sorted by eye during
the next excavation season.
The flots generated during the 1999 season are being analysed in Oxford, along with
the 43 samples taken during the 1998 season. Provisional analysis indicates that both
charred and desiccated remains are present in the samples. Cereal remains of bread wheat
and barley, cereals of Mediterranean origin, were present in the majority of samples,
generally forming the greatest component of the assemblages. A third cereal which was
frequent is sorghum, a cereal of sub-Saharan African origin. The Phase I and II samples
also contain a New World cereal, maize. Fruit remains were common throughout the
deposits, notably of date, but also including water melon, melon or cucumber, fig, grape
and peach. Other cultivated species included flax, coriander and another New World
species, chilli.

Continuing Survey of Islamic Architecture in Old Germa and Elsewhere in the

Germa Region
A further selection of mud-brick buildings of the Islamic town of Old Germa has been
surveyed. This work included the surveying of four domestic structures in the immediate
134
 THE FEZZAN PROJECT 1999

vicinity of the present Gl excavations (Germa buildings 53, 54, 83a/b — Fig. 3). In
addition the surviving remains of the southern mosque (Germa building 15, al-Kebir
mosque) were recorded. This mosque was also largely demolished by Ayoub in the 1960s
during his campaigns of excavation at the site, though he failed to locate any clear
Garamantian remains below the building (his work on this part of the site is unpublished).
Careful survey with a Total Station has allowed us to salvage the broad outline of the
plan of the mosque and to suggest a probable reconstruction of its interior arrangements.
The two best preserved of the town's gates, the East (Germa structure 19) and North-West
(Germa structure 34), were also surveyed, both in plan and elevation, again with the aim
of reconstructing the now destroyed archways.
In addition to the recording of buildings at Germa, work was carried out at the cemetery
of el-Charaig to establish the topographical and spatial relationship between the various
tomb types to be found there. Surveys were also made of the lines of three foggara
channels, at el-Charaig, Twesh and Fjej with the aim of recording their alignments and
likely outlet point in the oasis (see further below).
Beyond the site of Old Germa itself two gsur were surveyed in detail as part of a
project to build up a corpus of data on this style of building in the Wadi-el-Agial. This
work forms part of a general recording of Islamic gsur in the Germa region being
undertaken by Mohammed al-Mashai, with support from the British team. This involves
the mapping, planning and photographing of a number of sites between Ubari and Gasr
Budrinna (Fig. 4). So far a total of c. 20 sites has been noted and partly recorded (for
earlier work on these sites, see Ruprechtsberger 1997, pis 124-25 and Ziegert 1985). This
important work will continue next season.
At the suggestion of Mohammed al-Mashai a brief reconnaissance mission was made
to visit the sites of Gasr Mara and Gasr el-Sceraba to the south-east of Germa and west
of Murzuk (for previous references, see Daniels 1971, 268-71 and 1989, 58). Gasr Mara
is probably a wholly Islamic site as it stands, though there may have been an earlier
Garamantian habitation at this location. Gasr esc-Sceraba is an extraordinary stone-built
site, the size of a small city, with elements of regular planning (Fig. 5). It appears to be
a Garamantian centre of considerable importance, with some continuity into the Islamic
period. Further work at Gasr esc-Sceraba is recommended for next season, in particular,
with the aim of producing a plan of the site before it is further damaged by the visits of
4WD vehicles.

Fieldwalking in the Oasis

Surface collection and recording of artefactual materials continued the work of previous
seasons (see Mattingly et at. 1997; 1998a/b, for explanation of methods and preliminary
results). Fieldwalking was initially concentrated in the area of the oasis and valley floor,
with particular attention being paid to a series of probable Garamantian (and earlier and
later) settlement and cemetery sites between Germa and Twesh and around el-Charaig.
During the 1998 season, surface collections of artefacts had been made to the north and
to the west of Old Germa (in the immediate vicinity of Old Germa and c. 15 km west
in the area known as el-Hatiya); briefly, towards the end of the 1998 season, fieldwork
had been initiated to the east of Old Germa close to Twesh (at Germa 16) and this is
where it resumed in 1999. In addition to Germa 16, a further 8 'sites' were sampled during
the 1999 season: Germa 11, 15, 18, 26 and 27; Twesh 6 and 8; and el-Charaig 6/7. Several
of these appear to be extensive Garamantian settlements (tentatively interpreted as
'villages' — notably Germa 16 and Germa 27 (Saniat Suleiman Craida, just to the west
of the cemetery of Saniat ben Howedi) and el-Charaig 6/7); others relate to possible
Islamic settlements close to standing gsur, while two sites (Germa 11 and 16) also yielded
135
 D. J. MATTINGLY ET AL.

Walls
Roof beam position

Building 54

Building 53

10 Metres
3

Figure 3. Plan of traditional mud-brick houses (Buildings 53 and 54) at Germa.

136
 H
X
tn
aN
N
>

?o
O
tn
n
H

V > - •• ^.

'# •

^ Figure 4. Large mud-brick fort of Islamic date east of Fjej (Photo: T. Savage).
 D. J. MATTINGLY ET Ah.

- • • • - - — • - • • - - 1

figure 5. Gasr esc-Sceraba: general view showing surface preservation of the site (Photo: T. Savage).

substantial evidence of neolithic activity. A number of other probable sites were identified
by reconnaissance survey, but not systematically sampled by grid-walking.
During the latter part of the season, survey was targeted to investigate the southern edge
of the Ubari sand sea, specifically searching for prehistoric lithics and associated artefacts.
A senes of sites was located by preliminary reconnaissance (Germa 30-38) and several
subjected to gridded collections (Germa 31, 32, 33, 34 and 35). The methods used and
the results of the lithics sampling are described below.
The surface collection of artefactual materials has complemented the emerging
delineation of sites in the Wadi al-Hayat, especially so in relation to the identification of
settlement sites close to the well-known cemeteries of Saniat Ben Howedi and el-Charaig
The sheer bulk of collected or counted artefacts (potsherds and lithic materials) within the
50 x 50 m gridsquares is impressive. For example, over 10,000 potsherds were collected
or counted in a total of 126 grid squares (an average of 80 potsherds per grid square)
Germa 16 compnsed the largest contiguous survey area: 48 grid squares containing 5 433
potsherds (an average of 113 potsherds per grid square), 72 per cent of the finds being
concentrated in 23 (or 48 per cent) of the grid squares, with the highest concentrations
being in those gnd squares associated with probable structural evidence in the form of
eroded mud-brick walls and similar features. At el-Charaig 6/7 nine grid squares accounted
for 600 potsherds (an average of 66 potsherds per grid square). It remains now for these
data to be integrated emphatically and explicitly with the emerging geomorphic model as
well as with other interdisciplinary contributions.

Lithic Scatters and Analysis of Materials

The 1999 survey for lithics has concentrated upon the floor of the Wadi al-Hayat after
138
 THE FEZZAN PROJECT 1999

the successful identification of the exploitation of the hamada for raw materials (quartzite,
silcrete and silicified wood) last year. The hamada sites were all datable to the Pleistocene
but no clear Holocene activity was documented. However, at the end of the 1998 season,
the gridded survey in the wadi centre had recovered a series of flake scrapers from site
Germa 16. These scrapers have now been augmented by additional finds through
continuing the gridded survey. A clear series of blade-flake based endscrapers and double
ended scrapers has been recovered which can be ascribed to the 'Late Pastoral' through
to the early first millennium BC. Unfortunately, none of these tools can be attributed to
'sites' in the strict sense because of the depositional impact of successive phases of
agricultural and settlement activity. The gridded survey continues to monitor the
exploitation of the Wadi in a general manner and does now show Holocene activity. New
work at Germa 16 yielded a significant number of these scrapers, the remaining grids
repeated last year's generalised background of mixed material in mostly poor condition.
The series of scrapers were in a relatively fresh condition.
Despite these finds, there remained a gap in the sequence between the end of the
Mousterian at c. 40,000 BP and the first millennium BC. New work, involving the gridded
sampling of a number of lithic scatters on the north side of the Germa basin, adjacent to
the Sand Sea, has now partly filled this gap. This survey is based on collection within
five by five metre boxes, the surface spread being recorded at 1:10 before pieces are lifted.
This enables identification of activity areas and allows better site definition. Accuracy is
limited to c. 10 cm but this is adequate when the scatter is likely to have been affected
by post-depositional processes (in this case deflation of the sands and some down slope
slippage). A total of five such sites were recorded, Germa 31-35. The sites at 31, 32 and
35 were all collected over a 15 m X 15 m area but sites 33 and 34 were much larger
spreads of material and two 15 m X 15 m boxes and three 15 m X 15 m boxes,
respectively, were collected. The material from all sites was similar: a background of
Levallois Mousterian (no tanged artefacts were seen ruling out Aterian presence), overlain
by a mid-Holocene industry with bladelets, flake borers, projectile points (mostly bifacially
worked and tanged but also including a foliate bifacial point) and microliths. The
microliths comprise two bladelet based crescents and an obliquely blunted point. There
were few other formal tools, the occasional flake-based scraper and backed blade making
up these. At present no burins have been identified. There are also a significant number
of fragments (often quite small) of grinding stones. Associated with these lithics is a low
number of pot sherds and fragments of ostrich eggshell. A number of the eggshell
fragments had been broken during perforation and a number had been perforated to make
beads. Bead manufacture appears to be a component of all the sites so far analysed. The
raw materials exploited are dominated by those from the hamada — quartzite, silcrete,
and silicified wood but there is a fine grained brown flint with black speckles which must
derive from some distance away. The age of this industry would be from the sixth
millennium BC and could fill the gap to the first millennium BC material noted above.

Further Work on the Irrigation Systems (wells and foggaras)

Previous seasons' work on the foggaras (subterranean water channels for irrigation, see
Goblot 1979) had established their pre-Islamic origin and demonstrated that clusters of
foggaras were associated with Garamantian oasis floor settlements, and that they originated
chiefly in the zones of Garamantian cairn cemeteries at the foot of the hamada escarpment
(Mattingly et al. 1997; 1998a/b). Their introduction to the Fezzan can be placed securely
in the Garamantian period, and it seems that they must have been critical to the viability
of agriculture supporting the significant level of Garamantian settlement attested by the
field survey. Work in the 1999 season focused principally on determining the points of
emergence of the underground channels, and on elucidating the conditions under which
139
 D. J. MATTINGLY ET AL.

the foggaras were abandoned or fell into disuse. This involved the excavation and cleaning
of three foggara shafts and channels, near Twesh, el-Charaig and Fjej.
Reconnaissance around Twesh, where low-level air photos from the 1950s show square
features at the end of the visible foggara lines, revealed that these are in fact large
depressions formed by the collapse of successively re-dug wells. While it is possible that
they might once have been reservoirs through the floors of which modern wells have been
dug, this seems very unlikely as the foggara channels in this area are still several metres
below the surface. Cleaning of the nearby channel of foggara F48, exposed in a modern
pit near the Twesh mausoleum, revealed the channel floor to be 2.85 m below the modern
ground surface, and its emergence at the surface must lie some distance to the north.
Unfortunately, all along the wadi the area where the foggaras appear to have terminated
has been disturbed by modern cultivation, and no traces of terminal structures or
distribution arrangements are visible. Excavation and cleaning of the shaft and channel of
foggara F31, by the el-Charaig cemetery, combined with topographic survey of the surface,
revealed that the foggara should have emerged about 650 m from the escarpment; it is
precisely in this area that a scatter of Roman amphorae sherds and possible Garamantian
pottery, and lithics was observed. The location of oasis floor settlements may serve as a
rough guide to the terminal points of the foggaras, which (on the analogy of foggaras
elsewhere) doubtless provided water for the inhabitants as well as for irrigation of the
surrounding fields.
Excavation of F31 (el-Charaig) showed that the channel was of roughly rectangular
cross-section, with dimensions varying around 0.50 m wide and 1.60-2.00 m high, often
widened somewhat by erosion or collapse some 0.20-0.50 m above the bottom (Fig. 6).
The channel was almost entirely full of water-deposited sand, showing that it had not been
abandoned owing to exhaustion of the aquifer, but that some water had continued to flow
until the channel was choked. Three distinct layers of clayey silt near the bottom of the
channel seem to represent ponding of water in the foggara during the later stages of its
useful life, perhaps forming because of an obstruction or collapse downstream. The latest
of these showed a pitted surface formed by drops of water percolating through the rocks
above over a period of time, and may indicate the abandonment of the foggara as a
regularly maintained structure. A charcoal sample taken from approximately halfway up
the sediments in the channel may provide a terminus ante quern for the foggara's
abandonment.
The excavation of a shaft down to the channel (whose floor lay 3.73 m below the
modern surface), demonstrated that the aquifer tapped by the foggara was a fissured marl,
overlying an unfissured layer of similar marl that would have acted as an aquaclude. Some
water may also have been provided by the sandstone directly overlying this, although
water transport rates through the sandstone would have been slower than through the
cracks in the marl. The recharge catchment area is still unclear, but may have included
the hamada or the sand sea. Direct recharge from the immediate vicinity is less likely,
given the presence of a layer of impermeable clay marl overlying the aquifer a metre
below the modern ground surface. During the 1950s (according to local informants) the
shaft had been re-excavated and deepened to form a well, encountering water a metre
below the marl forming the foggara floor, but this was now dry.
Excavation of another foggara shaft and channel, F29 near Fjej, showed that much of
the channel roof has collapsed for some distance along its course. The shaft was entirely
choked with windblown and rainwashed sand and collapse. It is possible, though not
certain, that the settlement overlying this foggara may postdate not only the foggara's
construction but also its abandonment, as some of the surface pottery overlies filled-in
shafts.
140
 F31

3 silty crusts

F31

3
ffl
m
N
N
>

O
<—i
w
o
H
F31
Windblown sand

Red sandstone

Marl 7 /^cv*
ill 3 s Ity crusts
Fissured marl
W\
m§ Shale
F31

Clay
f§
^•** Charcoal

Figure 6. Foggara F31 by el-Charaig cemetery: transverse sections across channel (left) and plan and longitudinal section of access shaft (right).
,_! Transverse sections: 1 Channel at foot of access shaft, facing upstream 2 Channel at foot of shaft, facing downstream. 3 Channel exposed in modern
£ pit, facing upstream. 4 Channel exposed in modern pit, facing downstream. (AIW / SP)
 D. J. MATTINGLY ET Ah.

Four abandoned wells in the centre of the oasis were also studied; these all had
originated as the typical Arab well, originally operated by animals walking down an
inclined path to lift a self-dumping bucket (cerd) over a pulley; and discharging into a
hollowed palm trunk, remains of which, together with a working platform, survived.
Subsequently, these wells had been deepened, with installation of a motor to operate a
winch, and then again by sinking a bore-hole and pumping the water up through metal
pipes, witnesses to the pursuit of the falling water table by the successive introduction of
different lifting technologies. Remains of concrete tanks and metal pipes still in situ show
that the abandonment of some of these wells occurred within the last few years.
Photographs in the Daniels archive from the 1950s and 1960s show wells around Germa
with water standing to a depth of as little as 10 m below modern ground level.

Summary of the Evidence for Human Settlement in the Wadi al Hayat (el-Agial)
The Fezzan Project has continued to illuminate many aspects of human history in the Wadi
al-Hayat (el-Agial), from earliest prehistory to recent times. There are still gaps in the
story and still many questions require further analytical work, but the broad framework
of human adaptation to the Saharan environment is becoming clearer. Through the use of
a wide range of scientific techniques of analysis and dating we are trying to integrate the
archaeological picture with a detailed knowledge of changing environmental and climatic
conditions over time. Within our core survey area around Germa we now have evidence
of human activity in most phases of prehistory and historic times.
There was a Middle Palaeolithic exploitation of the area by handaxe — using
populations. This Acheulean phase would probably be late as there is a high number of
flake tools and use of Levallois technique. There is also a significant number of blades.
This pattern is elsewhere dated to 230-105,000 BP (OIS6/5). It is followed by a Levallois
Mousterian phase (broadly down to 40,000 BP) which is found in heavily rolled conditions
in the Wadi as well as on the Hamada. There was probably a high lake stand at this time.
There then follows a substantial period when no human presence can be seen, spanning
the late Pleistocene and early Holocene. In general, there is a poor representation of the
Aterian phase in Fezzan and this seems to mark a period of adverse environmental
conditions where population retreated to a few favourable locations. It was in the mid-
Holocene period, c. 6^1,000 BC, that populations using bladelet tools and pottery were
again present in sufficient numbers to be visible. These people seem to have congregated
around small relict lakes or marshes in the edge of the sand sea and a background 'noise'
of this earlier material may also be identified in the oasis band in the southern part of
the Wadi but it is too broken up by post-depositional processes to be characterised. These
neolithic people may well have had connections with the neolithic rock artists of the
Messak (see further, Cremaschi and di Lernia 1998). They were presumably responsible
for the sporadic rock engravings recorded along the escarpment to the south of Germa.
The large number of grinders from these sites attest to the increasing processing of basic
foodstuffs, including simple grains, in addition to the stock-raising that is well-attested in
the surviving rock art. Such populations continued with some slight changes to their
artefactual repertoire until the first millennium BC when large scraper-using groups are
attested both in the area of the later oasis (Germa 16) and on the southern side of the
Wadi (for example, on Zinchecra). No first millennium BC type scrapers have yet been
found on the sites on the northern side of the depression close to the Sand Sea, suggesting
that the settlements or encampments by the lakes/marshes there were now being
abandoned.
The faunal and botanical evidence from Zinchecra shows that by the first millennium
BC the people of the region were farmers, cultivating a range of crops which require
142
 THE FEZZAN PROJECT 1999

irrigation (wheat, dates, vines, etc.) and raising a variety of animals, notably sheep or
goats. The transition to agriculture in Fezzan was probably related to the onset of an
increasingly hostile desert environment, with communities adapting to the challenge by
finding new and more intensive ways to exploit the available groundwater through
sedentary settlements. The Garamantes of the historical record emerged, in part at least,
out of the neolithic communities occupying sites such as the hillfort of Zinchecra (Daniels
1989; van der Veen 1992). By the middle of the first millennium BC, if not earlier, they
were also constructing permanent settlements in the centre of the valley, as at Old Germa
(Garama) and in that case at least this was to reach the size and sophistication of a city.
The importance of Garamantian civilisation has been established by earlier work, though
attention has tended to focus on the city of Garama and the many Garamantian cemeteries
along the Wadi al-Hayat. Other elements in the settlement hierarchy have been unclear.
However, as a result of our present survey, many smaller settlements of
Garamantian/Roman date have now been recorded within the oasis zone. These confirm
the long-suspected model of Garamantian settlement all along the Wadi al-Hayat, matching
the extensive escarpment base cemeteries. It is clear that there must also have been many
oasis centre cemeteries similar to that known at Saniat Ben Howedi or those at el-Hatia.
The settlement picture of the Wadi al-Hayat is now more complicated and much fuller, in
a way that makes it easier to understand the dominant role played by the Garamantes in
the northern Sahara. All of this points to a massive demographic increase in the Wadi al-
Hayat during the Garamantian phase, particularly at the height of the Garamantian
civilisation (100 BC - AD 500).
The new insights into the foggara technology, which appears to have been introduced
to Fezzan in the Garamantian period, suggest that a far larger area of the valley was under
cultivation in this phase than was being sustained in the nineteenth and twentieth centuries.
Only in recent years, with improved wells and pumps, has modern cultivation been able
to rival the extent of Garamantian and early Islamic farming in the Wadi. It seems clear
that it was the introduction of foggara technology which enabled the demographic boom
in the Garamantian period.
The Islamic settlement pattern in the Wadi al-Hayat is still poorly understood,
principally because although many sites (notably gsur) are known, few have ever been
properly dated or excavated. However, it seems clear from our work thus far that many
of the Garamantian villages had some degree of continuity into the early Islamic period,
though with an increasing emphasis on defence re-emerging in the architecture and
morphology of settlements. The Arab conquest of Fezzan did not involve the displacement
of people, but the acceptance of Arab suzerainty by the Fezzanese (we do not know when
they stopped calling themselves Garamantes). Both through the excavations at Germa and
a wider survey of Islamic sites in the Germa region, we hope to be able to produce a
better framework for dating sites of the post-Garamantian era and improved knowledge of
the distribution of people in the landscape in the past 1300 years.
It is apparent that population levels at some point fell dramatically below the peak of
the Garamantian era and this decline is probably to be linked in part to changes in the
irrigation regimes in the Wadi al-Hayat. One of the key questions still to answer is when
and why the foggara irrigation systems broke down, and with them the ability to sustain
the very large population which characterised the Garamantian period. At some point in
the medieval period irrigation turned from the use of running water (the foggara
technology, which was expensive to construct and maintain) to simpler wells, each of
which could only serve a restricted area of ground in its close vicinity. The later phases
of occupation at Germa, covering the period from say 1600-1937, are notable for the
poverty of the material culture and diet of the population. There are already indications
143
 D. J. MATTINGLY ET AL.

that living standards were better at earlier dates in the Islamic period and we hope to be
able to fill out this picture as our excavation and analysis continues.
The continuing challenge of life in the Sahara and the remarkable creativity of Libyan
people in finding new possibilities to exploit is being revealed in our story of the Germa
region over time. Close integration of our new results with the publication of the earlier
British work by Daniels (Edwards et al. 1999 (this volume)) is bringing an added
dimension to the study. The next season of fieldwork will, we hope, produce further
contributions to the knowledge of Libya's Saharan heritage.

Acknowledgements
We wish to record our gratitude once again to the British Academy for providing the funding for the project,
through its grant to the Society for Libyan Studies, and to the University of Leicester for providing additional
financial support. The Arts and Humanities Research Board made a separate grant towards the post-excavation
analysis of the botanical remains. This report is a much condensed version of a longer article submitted for
publication to Libya Antiqua, the official journal of record of the Libyan Department of Antiquities. We wish
to thank the Department for permission to publish this summary account here.
The third season of the Fezzan Project took place from 12th January to 6th February 1999. A team of 13
archaeologists was in the field for four weeks, with a specialist group of three geographers, a photographer
and a lithics expert joining them for the final two weeks. The team was as follows: Professor David Mattingly
(Director), Philip Balcombe (field survey specialist, prehistorian, lithics), Franca Cole (conservator, finds
processor), Dr Nick Drake (physical geographer, arid zone specialist), John Duncan (surveyor, standing
building survey, soil micromorphology), Sophie Hay (excavation supervisor and standing building specialist),
Elanor Johnson (ceramic specialist), Stephanie Knight (animal bones), Dr Sue McLaren (physical geographer,
desert landforms specialist), Ruth Pelling (botanical remains), Simon Pressey (illustrator, excavation
supervisor), James Preston (surveyor, standing building survey), Dr Tim Reynolds (lithics specialist,
prehistorian), Toby Savage (photographer), Irene Schriifer-Kolb (slags and metallurgical specialist), David
Thomas (excavation assistant director), Dr Kevin White (physical geographer, remote sensing), Dr Andrew
Wilson (hydraulic features, especially foggaras). Participants from the Department of Antiquities were:
Mohammed al-Mashai (Islamic specialist, expert in archaeology of Fezzan) and Saad Saleh Abdelaziz (Curator
of Germa Museum and Representative of the Department in Germa).
The team employed a team of 18 workmen for the excavation and a cook and assistant cook. It is a
particular pleasure to be able to report that the diesel locomotive at Germa has now been repaired and is
running well.
The work of the team would not have been possible without the support and assistance of the President of
the Department of Antiquities, Ali Emhemmed al-Khadouri, and his staff in Tripoli — notably Giuma Anag,
Giuma Garsa, Mohammed Shakshuki, Mustapha Turjiman — and the Sebha Controller, al-Mahdi Mohammed
al-Azrak, and his staff, notably Mohammed Arreda. We also wish to acknowledge the help and material
support of Mike Buck and Mike Keane of LASMO Grand Maghreb and of the Head of the British Interests
Section in Tripoli, Dr Noel Guckian and his then Deputy, Neil Hammond.
The writing of the report was a team effort, with the first named author responsible for the overall editing
of this report. Several members of the team not named in the list of authors here contributed text which will
appear in the longer report submitted to Libya Antiqua. Thanks are also due to John Duncan, Jamie Preston
and Simon Pressey for the figures used in this report.

Bibliography
Ayoub, M. S. 1967. Excavations in Germa between 1962 and 1966. Tripoli.
Cremaschi, M. and di Lernia, S. (eds). 1998. Wadi Teshuinat. Palaeoenvironment and Prehistory in south-
western Fezzan (Libyan Sahara). Survey and Excavations in the Tadrart Acacus, Erg Uan Kasa, Messak
Sattafet and Edeyen of Murzuq, 1990-1995. Milan.
Daniels, C. M. 1971. The Garamantes of Fezzan. In Gadallah, F.F. (ed.), Libya in History. Proceedings of a
conference held at the Faculty of Arts, University of Libya 1968. Benghazi: 261-85.
Daniels, C. M. 1989. Excavation and fieldwork amongst the Garamantes. Libyan Studies 20: 45-61.
Edwards, D., Hawthorne, J., Dore, J. and Mattingly, D. J. 1999. The Garamantes of Fezzan revisited:
publishing the C. M. Daniels archive. Libyan Studies 30: 109-127.
Goblot, H. 1979. Les qanats: une technique d'acquisition de Veau. 'Industrie et artisanat 9'. Paris and New
York.
Mattingly, D. J., al-Mashai, M., Balcombe, P., Chapman, S., Coddington, H., Davison, J., Kenyon, D., Wilson,
144
 THE FEZZAN PROJECT 1999

A. I. and Witcher, R. 1997. The Fezzan Project 1997: methodologies and results of the first season. Libyan
Studies 28: 11-25.
Mattingly, D. J., al-Mashai, M., Balcombe, P., Chapman, S., Coddington, H., Davison, J., Kenyon, D., Wilson,
A. I. and Witcher, R. 1998a. The Fezzan Project I: research goals, methodologies and results of the 1997
season. Libya Antique ns 3: 175-99.
Mattingly, D. J., al-Mashai, M., Aburgheba, H., Balcombe, P., Eastaugh, E., Gillings, M., Leone, A., McLaren,
S., Owen, P., Pelling, R., Reynolds, T., Stirling, L., Thomas, D., Watson, D., Wilson, A. I. and White, K.
1998b. The Fezzan Project 1998: preliminary report on the second season of work. Libyan Studies 29:
115^14.
Mattingly, D. J., al-Mashai, M., Aburgheba, H., Balcombe, P., Eastaugh, E., Gillings, M., Leone, A., McLaren,
S., Owen, P., Pelling, R., Reynolds, T., Stirling, L., Thomas, D., Watson, D., Wilson, A. I. and White, K.
1999. The Fezzan Project II: preliminary report on the 1998 season. Libya Antiqua ns 4: 63-93.
Petit-Maire, N., Delibrias, G. and Gaven, C. 1980. Pleistocene lakes in the Shati area, Fezzan (27°-30'N), in
M. Sarnthein, E. Seibold and P. Rognon (eds), Sahara and the Surrounding Seas (Palaeoecology of Africa
12): 289-293.
Ruprechtsberger, E. M. 1997. Die Garamanten, Geschichte und Kultur eines Libyschen Volkes in der Sahara.
Mainz.
Veen, M. van der. 1992. Garamantian agriculture: the plant remains from Zinchecra, Fezzan. Libyan Studies
23: 7-39.
Ziegert, H. 1985. Budrinna. Die vorislamiche Besiedlung in Wadi el Adjal (Fezzan, Libyen). Tripoli.

145