Contents

APPENDIX 2

PRELIMINARY REPORT ON THE MICROMORPHOLOGY OF
DEPOSITIONAL SEQUENCES AT THE FIRST MILLENNIUM BC
URBAN SETTLEMENT AT KERKENES DAG, CENTRAL ANATOLIA.

Dr Wendy Matthews
Research Associate, McDonald Institute for Archaeological Research, Cambridge

INTRODUCTION

A range of depositional sequences in key context types have been sampled from the urban settlement at Kerkenes Dag, c. 550 BC for microscopic analysis in large resin impregnated thin sections. The aim is to assess the potential of the technique for examining the nature of depositional sequences and identifying traces of use of space in conjunction with geophysical survey and test trench excavation strategies at Kerkenes Dag. Seven contexts have been sampled including in situ fuel in a large oven and adjacent floor (STT4); floors in a typical two-roomed building (TT 16); a large columned hall (TT15), and a stone paved area (STT5).

Analyses of depositional sequences at Kerkenes are providing important data on the nature of depositional sequences within settlements in more temperate climates in Anatolia, for comparison to sequences further south at catal Hoyuk, Asikh HOyuk and Kilise Tepe (Matthews et al. 1996). The settlement at Kerkenes lies within Zohary’s Xero-Euxinian steppe forest of Querco-Artemisietea anatolica geobotanical zone, in a region with sandy clay soils on top of granite, which contrast to the calcareous soils in the Konya basin and Goksu valley (Zohary 1973). The Iron Age levels at Kerkenes are being compared to Iron Age levels at Kilise Tepe, and to earlier sequences at all three of the other settlement currently being studied.

METHOD

Each depositional sequence was cleaned, photographed and drawn. Intact block samples were cut out of the section faces using a Swiss Army knife and wrapped tightly in tissue and tape. It should be noted that the block samples were difficult to cut out from the section faces at Kerkenes Dag due to the hardness of the sandy clay sediments when dry. The slots cut around each block had to be moistened with a gentle jet of water, and sediments pared away c. 5-10mm at a time. Each block took up to 2 hours to cut out, in contrast to the usual maximum time of half an hour. Furthermore, the conflagration which destroyed much of the settlement in the Sixth Century BC had baked a range of deposits, particularly in the columned hall, and meant that some samples were solid and had to be sawn out of the section face. After oven drying at 400C, the blocks were impregnated with a crystic polyester resin under vacuum, and cut, ground and polished into large thin sections, 13.5 x 6.5cm, in the Geoarchaeology Laboratory, University of Cambridge (Murphy 1986).

The thin sections have been analysed as hand held specimens and at both low and high magnifications from x5-400. They have been described using internationally standardised terminology and procedures (Bullock et a!. 1985; Courty et a!. 1989). The relevance and interpretation of each micromorphological attribute to study of site formation processes and uses of space is discussed in Courty et a!. 1989; Matthews 1995.

Previous applications of micromorphology to studies of occupation sequences and use of space within archaeological sites include analysis of deposits within Palaeolithic and Neolithic caves, and a range settlements in the Near East and elsewhere in the world in conjunction with experimental and ethnoarchaeological research (Courty et a!. 1989; Matthews and Postgate 1994; Macphail and Goldberg 1995; Matthews et al. 1997).

RESULTS

Depositional components
Mineral sediments
A wide range of particle sizes are present in the thin section samples from Kerkenes, including clay (<2µm), silt (<50µm), sand (50µm-2mm) and rock fragments up to 32mm in diameter. The mineralogy of these sediments has been analysed in collaboration with Dr Piril Oren, Department of Geology, Middle East Technical University. The mineral and rock fragments principally comprise quartz and plagioclase and orthoclase feldspars derived from the local granite bedrock. The individual crystals range from microcrystalline structures to large crystals which cooled and formed during increasingly long periods of time. The minerals and rocks present in the settlement deposits are more weathered than those in the parent bedrock, due to ongoing pedogenetic processes in the soils, in particular of oxidisation and kaolinisation. Other minerals include: mafic amphiboles, pyroxines, hornblende, ?epidote, sphene, apatite, and sparse volcanic rock fragments. The latter could either be derived from natural outcrops, or from grindstones used within settlement. Silt, sand and rock particles sizes constitute up to 60% of deposits, particularly in packing or levelling deposits below floors, and would have provided a firmer substrate.

The organisation of the groundmass (clays) in the settlement deposits at Kerkenes varies considerably according to context and deposit type:

Further library research is being conducted in order to investigate the significance of these variations with regard to the origin, deposition and post-depositional alterations of deposits.

Inorganic materials of organic origin
The fragments of bone in these samples are sparse and small, <2mm, and include both burnt and unburnt fragments. Bone occurs in deposits adjacent to the oven in STT4 and in the columned hall in IT 15.

Organic remains
The principal organic remains in the samples analysed are charred plant remains, which constitute <2-50% of deposits. Charred remains are particularly abundant in lenses on floors adjacent to the large oven in STT4. The charred plant remains include a range of coniferous and deciduous woods, including Quercus type; reeds, grasses, seeds and cereal grains. The preservation of these charred plant remains varies from ‘mashed’ (due perhaps either to shrink! swell action of clays, or trampling), to well preserved remains in uncompacted deposits. Siliceous plant remains, often referred to as phytoliths, are not abundant, and only occur as 2% of deposits in in-situ oven fuel, and as traces in roofing material. Melted plant silica from reeds, with identifiable phytoliths embedded in a vesicular melted silica matrix, occur in occupation deposits on top of the floor of the columned hail and in overlying collapse. Silica melts in temperatures greater than c. 800°C.

Pseudomorphic voids of vegetal remains which have since decayed characteristically occur in floor plasters, but in low concentrations c. 2%, and in greater concentrations in mudbrick and particularly roofing fragments, at 5-10%. These voids probably represent the remains of vegetal stabilisers which would have provided tensile strength and flexibility (Norton 1986).

Possible fragments of dung occur in some deposits, represented by 2-5% finely fragmented plant remains in in-situ oven fuel (96.03 [2]), and ?digested plant remains included in floor plaster materials in the two roomed building (96.05 [2]).

Artefactual remains and anthropogenic aggregates
Few fragments of artefactual remains have been identified in these samples. Only one fragment of pottery is present in the thin sections, in deposits above the stone paving in STT5. Fragments of basaltic rock may have been abraded from grindstone surfaces during use, but may also be of natural origin from volcanic outcrops in the environs. The proximity of such outcrops is being investigated by the Department of Geology, METU.

Sub-rounded aggregates of unburnt floor plaster may originate from floor sweepings, and occur in accumulated deposits next to the oven in STT4, and in the unroofed stone paved area in STT5. Burnt aggregates occur in in-situ oven fuel, collapsed structural debris, and in the ?unroofed area of the two-roomed building.

The sparsity of anthropogenic debris, including charred plant remains, in all building materials (plaster, mud bricks and roofing) corresponds with the nature of the new and rapid construction of the settlement at Kerkenes. Virgin soil has been exposed below some buildings, and current chronological evidence suggests the settlement may have only been occupied for c. 40 years. Building materials on longer lived tell or hOyuk mound sites tend to include higher percentages of anthropogenic debris in at least some instances, from the use of source materials in and close to the halo of refuse around a settlement, or materials excavated from abandoned areas of a site.

Post depositional alterations
No reprecipitated salts have been identified in any of the thin sections analysed from Kerkenes, in contrast to settlements in more semi-arid and calcareous environments. Deposits at Kerkenes have been subject to greater physical disturbance and reworking of the original microstructure particularly in the two-roomed building in TT 15, where aggregates of plaster floor have been vertically displaced by more than 4cm.

The agents of this reworking include extensive root action attested by presence of modern roots, and a network of channels and chambers. The alternate shrink! swell of the clays during wetting, drying, and winter freeze thaw action, probably also accounts for some of the physical reworking. The microstructure of baked deposits by contrast is better preserved, as in the burnt plaster floor in two roomed building (96.05 [2]) and the baked floor and collapsed mudbrick in the columned hall (96.04 [1 and 3]). Organic staining from the decay of organic remains occurs in deposits in the two roomed building.

Types of floors

Different types of floors have been identified on the basis of their micromorphological characteristics. These types of floors moreover, vary according to context.

Floor plasters at Kerkenes are characterised by:

• embedded related distribution of coarse and fine sediments
• reticulate and grano-striated groundmass
• pseudomorphic voids of vegetal remains which have since decayed, and probably represent remains of vegetal stabiiisers which would have provided some tensile strength and flexibility
• c. 40% mineral inclusions (predominantly quartz and plagioclase)

The possible unprepared earthen floor in the ?unroofed area in the two roomed building by contrast:

• does not have a grano-striated groundmass
• has no pseudomorphic voids of vegetal remains
• has a higher concentration of mineral inclusions (60%) and resembles packing/levelling.

Underlying packing/levelling deposits in the large room with the oven have:

• high concentrations, c. 60%, of sand and rock fragments
• absence of strong striations in groundmass

Impact on floors

Some floors, particularly in the columned hail have sub-horizontal cracks below the surface, which at other sites are interpreted as characteristic of trampling (Ge et al. 1993; Davidson 1992).

Burning during a conflagration has oxidised the plaster floor surface in the roofed room of the two-roomed building, which is yellowish orange brown in colour. The plaster fabric below the surface was burnt in more reducing conditions, and is dark brown in colour. Bioturbation has vertically displaced plaster floor fragments up to 44mm in the two roomed building.

Origin and deposition of occupation deposits

Some of the attributes which distinguish occupation deposits from floors include:

Interpretation of uses of space in different contexts

A list of the principal mieromorphological attributes of each depositional unit [n] and interpretation of each sequence is presented in Table 2.

96.01 STT4 Schmidt trench. Room with large oven
This room has a moderately well prepared plaster floor. Overlying lenses of accumulated oven rake-out are rich in diverse charred plant remains, including coniferous and deciduous woods, and seeds. Other food related debris includes a ?grindstone fragment, and sparse burnt bone. These lenses of oven rake-out alternate with lenses of sediment rich deposits. The overlying layer of building collapse includes a ?roof fragment with a charred reed still adhering to the surface.

96.03 STT4 Schmidt trench. Large oven.
The large oven in the room described above, had been subject to extensive bioturbation, evident also in the animal burrows visible in the field. Surprisingly well preserved charred plant remains survive in disturbed burrow fill. The surviving remnants of in-situ fuel include charred wood, 2% phytoliths, <2% melted silica, and 2-5% ?dung represented by finely fragmented plant remains. This fuel has been mixed with collapsed superstructure aggregates, which resemble the fabric of the adjacent floor plaster, 96.01 [2].

96.05 TTJ6 Two-roomed building. Room.
The plaster floor in this room was also moderately well prepared. The surface of this plaster has been oxidised during the conflagration which destroyed this building. Some plaster floor aggregates have been vertically displaced by bioturbation. No occupation deposits are distinguishable from the overlying infill. This may due either to the fact that the room was well maintained and swept, perhaps, leaving no microscopic residues, or to extensive bioturbation which has dislodged floor plaster fragments.

96.07 TT1 6 Two-roomed building. ? Unroofed area.
The floor in this area appears to have been an earthen unprepared floor [2]. It is distinguishable from the overlying infill deposits at a microscopic scale, by the nature of the groundmass which is mosaic speckled, lacking oriented striations. The spongy/vughy microstructure of the infill deposits further suggests extensive bioturbation. These deposits include 2% charred remains and burnt aggregates, perhaps from a fire-installation somewhere in the vicinity, or brought in from the street.

96.04 TTI6 Columned hall.
The plaster floor in this hall was baked hard by the intensity of the heat from the conflagration which destroyed the building. The surface has subhorizontal cracks which are often characteristic of trampling (Ge et al 1993, Davidson 1992), but may also have been affected by the burning. A thin lens of occupation deposits includes melted reed (perhaps also from the root) and sparse slivers of bone. These deposits are irregularly dispersed, and mixed with 50% aggregates of collapsed and baked mud brick.

96.02 STT5 Schmidt trench. Stone paved area.
This deposit has little internal lensing, in contrast to deposits in the room with the oven in STT4. Deposits are unoriented and include 2-5% charred remains, a pottery fragment <5mm, calcareous rock fragment, and subrounded aggregates with charred remains adhering to the surface, perhaps from floor sweepings. The high concentrations of cereal grain retrieved during excavation were less visible in the field section and the thin section, perhaps suggesting localised deposition/survival.

96.06 Roofing fragment
This roofing fragment has macroscopic impressions of reeds on one surface. The microstructure has the highest concentration of pseudomorphic voids from decayed vegetal remains in this sample set, at 10%. This high figure suggests plant remains were vital to construction of roof to add greater tensile strength, flexibility and resistance to shrink/swell during temperature differentiations, and would have enabled reduction of the weight of deposits on the roof

CONCLUSIONS

Although the deposits at Kerkenes were difficult to sample, and have been subject to a range of post-depositional disturbances, micromorphology is enabling us to analyse a wide range of mineral, organic and artefactual components, and their preserved depositional and contextual relationships. Clear variation in deposit types, and indications of different uses of space are emerging. These variations will be better understood in the light of co-ordinated controlled volume wet-sieving and flotation which will enable examination of the concentrations and degrees of fragmentation of bioarchaeological and artefactual remains in larger sample sizes for further statistical analyses. The nature of the deposits at Kerkenes is new to the researcher, and can be further elucidated by additional library research, and comparison and discussion of samples with other micromorphologists at the International Soil Micromorphology Workshops which take place annually. This set of samples is providing important comparative information on the impact of human activities on depositional sequence in more temperate climates and regions with sandy clay soils on granite.

BIBLIOGRAPHICAL REFERENCES

Bullock, P., Fedoroff, N., Jongerius, A., Stoops, G. and Tursina, T.
1985. Handbook for soil thin section description. Wolverhampton, Waine Research.

Courty, M.A., Goldberg, P. and Macphail, RI.
1989. Soils and micromorphology in archaeology. Cambridge, Cambridge University Press.

Davidson, D.A., Carter, S. and Quine, T.A.
1992. An evaluation of micromorphology as an aid to archaeological interpretation. Geoarchaeology: An International.Journal. 7 1, 5 5-65.

Gé, T., Courty, M.A., Matthews, W. and Wattez, J.
1993. Sedimentary formation processes of occupation surfaces. In Formation Processes in Archaeological Context. Monographs in World Archaeology No. 17 (eds P. Goldberg, D.T.Nash and M.D. Petraglia). Madison: Prehistory Press, 149-64.

Macphail, RI. and Goldberg, P.
1995. Recent advances in micromorphological interpretations of soils and sediments from archaeological sites, in Archaeological Sediments and Soils: Analysis, Interpretation and Management, eds A.J. Barham and R.I.Macphail. London: Archytype Books, 1-24.

Matthews, W. and Postgate, J.N. with Payne, S., Charles, M.P. and Dobney, K.
1994. The imprint of living in a Mesopotamian city: questions and answers. In Whither environmental archaeology? (eds R. Luff and P. Rowley Conwy). Oxford: Oxbow Books, 171-212.

Matthews, W.
1995. Micromorphological characterisation and interpretation of occupation deposits and microstratigraphicsequences at Abu Salabikh, Iraq. In Archaeological. Sediments and Soils, Analysis, Inteipretation and Management (eds A.J. Barham and Ri!. Macphail). London: Archetype Books, 4 1-76.

Matthews, W., French, CAl., Lawrence, T., Cutler, D.F and Jones, M.K.
1996. Multiple surfaces: the micromorphology. In On the swface: Catal Hoyuk 1993-95 (ed. I. Hodder). Cambridge:McDonald Institute for Archaeological Research and British Institute of Archaeology at Ankara, 301-42.

Matthews, W, French, C.A.I., Lawrence, T., Cutler, D.F and Jones, M.K.
1997. Microstratigraphic traces of site formation processes and human activities. High definition archaeology: threads through the past. World Archaeology 29.2, 281-308.

Murphy, C.P.
1986. Thin section preparation of soils and sediments. Berkhampsted, A.B. Academic Publishers.

Norton, J.
1986. Building with earth. A handbook. Leamington Spa, Salvo Print.

Zohary, M.
1973. Geobotanical foundations of the Middle East. Stuttgart, Gustav Fischer Verlag.

Sample	Location	Packing/levelling	Floors	Impact on floors	Occupation deposits	Collapse	Post-depositional alterations	Interpretation
96.01	STT4S Section Room w ith large oven	[1] Unoriented sandy clay loam. 60% deposit: granite rock fragments > 25 mm. Embedded in mosaic, local parallel + grano- striated brown groundmass.	[2] Unoriented sandy clay-sandy clay loam. 40% fabric: quartz and feldspar minerals. Embedded in grano- and reticulate striated yellovv- orange brown groundmass. Subangular blocky microstructure. <2% pseudomorphic voids from decayed vegetal stabilisers.	Diffuse, irregular undulating boundary, <0.5mm thick.	[3] Undulating lenses of sandy silt loam with a dark brown mosaic speckled groundmass, + up to 50% charred plants including seeds + wood (coniferous* + ?oak) <2.7 mm in size; <2% burnt bone, + a ?grindstone frag <4mm, 5% subrounded aggregates (?sweepings)	[4] Includes orange brovvn sandy clay loam aggregate (?roof fragment) with charred reed adhering to undulating surface	Bioturbation: channels and chambers from root and insect activity 20% (floor [1]) - 50% (occupation [2]), modern roots stili visible, <2%. Some charred plant remains in occupation deposits [2] are cracked and not well preserved.	Packing/levelling [1] includes 60% granite frags which would have provided a firm substrate for the moderately well prepared plaster floor [2]. Occupation deposits [3] comprise lenses of öven rake-out and floor sweepings. Collapse [4] includes ?roof frag.
96.03	STT4 Large oven			knife edge, irregular boundary (in areas where there has not been anv bioturbation)	[2] Undulating sandy silt loam. Complex packing void microstructure. Mosaic speckled + undifferentiated groundmass, bridged + intergrain agg.10% charred remains (wood), 2% phytoliths, <2% melted silica, 2-5% ?dung, 20-40% burnt + unburnt aggs, 5-10% rock	[3] Unoriented sandy clay loam with reticulate + slightly grano-striated groundmass + embedded related distribution. 2% voids from vegetal stabilisers.	Extensive bioturbation, including large burrovv 11.5x4 cm, between burnt aggregates. Large frag of bone and well preserved charred plant remains (inci. wood) in disturbed deposits in burrow.	In-situ fuel mixed with burnt and unburnt aggregates. Fuel included coniferous and deciduous wood, 'grasses', and ?dung. Fabric of collapsed öven superstructure [3] is similar to plaster floor fabric [96. 01 Unit [2],
96.05	TT16E Section Two- roomed building. Room		[2] Unoriented sandy clay loam (c. 40% quartz and feldspar). Embedded. grano- and reticulate striated groundmass. 2% voids and charred remains from vegetal stabilisers. ?2% charred dung.	Burning in conflagration has oxidised plaster floor surface y- orange bro\\n, + reduced lower plaster v dark brown.	none detectable separately from collapsed debris, perhaps due to extensive bioturbation.	[3] Unoriented sandy clay loam. 50% quartz and feldspar minerals. <2% basalt grain. Embedded mosaic speckled brown groundmass. ?Spongy/vughy microstructure.	5% chambers \\ithin hard plaster floor, but 30-40% floor surface has been dislodged by bioturbation, and moved vertically >44mm. Modern roots > 1 5 mm. 2% dark brown staining. 25 % channels and chambers in [3].
96.07	TT162 roomed building ?Unroofe d area		[1] Unoriented sandy clay loam \\ith 60% quartz and feldspar minerals and rock frags . Embedded, reticulate striated and ?crvstallitic y-orange groundmass. <2% charred plants (wood). 5% crack microstructure	very diffuse boundary	none detectable separately from collapsed debris, perhaps due to extensive bioturbation.	[2] Unoriented sandy silt loam. Embedded, mosaic speckled brov,Ti groundmass, with microcontrasted particles. Some cracks and vughs in microstructure. <2% charred plants, 2% burnt aggregates and 1 0% unburnt aggregates.	10-15% channels and chambers. At least 20% bioturbation in floor zone. >60% in overlying deposits.	?Unprepared/poorlv prepared earth floor (with no vegetal stabilisers nor a grano-striated groundmass, but a higher % of sand + rock frags). Occupation could not be distinguished from collapse /infill due to extent of bioturbation. Sparse ?FI rake-out.
96.04	TT15 Columne d Hail building		[1] Unoriented sandy clav. 40% quartz and feldspar minerals and rock frags. Embedded reticulate and grano- striated dark y-or brown groundmass. Subangular blocky microstructure.	Sub-horizontal cracks belovv plaster floor surface, characteristic of trampling (Ge et al + Davidson*)	[2] Unoriented sandy silt loam 2-7. 5mm thick. Faint reticulate striated dark ör brown groundmass. Embedded, bridged + intergrain aggregate Complex packing void + crack microstructure. 50% building aggs. (2% melted). 2-5% melted reeds, 2% bone frags	[3] Collapsed baked mud- brick. Unoriented loamy sand-sandy clay loam (quartz + feldspar). Undifferentiated v dk brown groundmass. 5- 10% voids from vegetal stabilisers. <2% charred plants. Large aggregate of vesicular melted silica 55 mm, w. reed cells	Less bioturbation in floor and collapsed deposits in this sequence, due to indurate nature of fabrics baked bv heat, Some of the cracks may be due to heat. 50% fabric bioturbated in occupation [2], due to irregular distribution of components	Baked plaster floor with sub- horizontal cracks from ?trampling/heat. Thin lens of occupation deposits includes melted reed and fine bone frags, later mixed vvith up to 50% aggregates from collapsed building materials.
96.02	STT5 Unroofed paved area				Heterogeneous deposits, including 2-5% charred plant remains < 1 5 mm, aggregates with charred material adhering to surfaces (from ?discarded sweepings). pottery frag, 2% amorphous crystallitic aggregates, <2% calcareous rock fragment, + <2% cracked bone
96.06				?whole reed stem impressions		Unoriented sandy clay loam- sandy clay. 50% quartz and feldspar: <lmm + <15mm) Embedded, dark red ?undifferentiated groundmass. Massive microstructure. 10% voids from vegetal stabilisers, <2% siliceous plant remains/modern roots.	Silt clay-silt loam coating 140um thick on edge of roofing in curving impression, '7translocation/dust/mud adhering to (now absent) ?reed	Roofing material with intact ?reed impressions. This material and that of other plasters and mudbricks does not include reworked occupation debris, perhaps attesting to new and rapid foundation of architecture in this settlement. Highest % stabilisers.

Contents