Çatalhöyük 1993 Archive Report

ÇATALHÖYÜK 1993 ARCHIVE REPORT

Çatalhöyük 1993

Roger Matthews and Ian Hodder

Introduction

A first season of renewed fieldwork at the site of Çatalhöyük took place between the 1st and 30th of September 1993. The principal aim for the 1993 and 1994 seasons is to obtain information from surface work which, on the one hand, will yield results, significant in their own right, concerning surface and subsurface distributions of artefacts and architecture and, on the other, will in future provide guidance for a long term programme of excavation at the site commencing in 1995. Previous work at Çatalhöyük, while relatively extensive by modem excavation standards, was almost entirely restricted to one area of the main mound so that other parts of the site remain obscure. In the 1993 season a range of techniques was employed in order to retrieve maximum information from as much of the site as possible without excavation, as detailed below. This work took place exclusively on the main or east mound. The west mound, and more of the main mound, will be the subject of surface work in 1994.

Topographic survey

Using a total station EDM a detailed contour survey of the entire site was begun. This season the EDM was frequently employed in the laying out of a site grid for the purposes of surface collection and surface scraping (see below), thus diminishing the amount of time spent on contouring. Nevertheless, a large proportion of the main mound received complete coverage. The site of Çatalhöyük consists of two principal mounds, a main mound and a west mound. This season no work took place on the west mound, whose surface remains indicate a predominantly Chalcolithic occupation, but this mound will be fully treated in the future. Our work in 1993 has now made it clear that the main mound consists of three distinct eminences: a large southern one, which rises 15 m above the surrounding modem plain and whose western flank was the site of the 1960s excavations, a smaller northern one, rising 9 m above the plain, and a wide, low eastern one, 5 m above the plain. These three eminences are not separate mounds, as considerable depths of cultural deposits lie in the lower areas between peaks thus joining the three eminences into one main mound.

The area of this main mound has now been established as 14 hectares, of which a total of 6 hectares received detailed contour survey this season (fig. 1). The mapping was carried out on a 2 m grid with contours drawn at 20 cm intervals, thus allowing fine resolution of surface features, some of which may relate to buildings or major wall alignments immediately underlying the mound's surface. The detailed contour survey carried out this season and in the future will provide the base map of the entire site, giving a sound and accurate basis for all future work across the site.

Surface artefact collection

A programme of collection of surface artefacts from 2 by 2 m squares at 20 m intervals was executed across most of the main mound. A uniformly dense coverage of vegetation across the mound rendered surface collection difficult and unreliable, so that the technique was devised of clearing vegetation from each 2 by 2 m square before taking from that square a set volume, 36 litres, of surface earth to be passed through a 5 mm mesh from which all artefacts were retrieved. Initial experiments with a 2 mm mesh soon established that its use was prohibitively time consuming and that no significant improvements in artefact recovery were thereby achieved. Artefacts recovered from each 2 by 2 m square were recorded on set forms under the headings of ceramic, bone, ground stone, shell, obsidian, flint, and other materials. Each category of object was divided into two size classes: below 2 cm (largest dimension) and above 2 cm. Within these classes each artefact category was then counted and weighed. For the obsidian and flint, the use of three size classes - below 1 cm, 1-2 cm, and above 2 cm - provided finer resolution. All ceramic sherds were further divided into wheel made or handmade categories.

The aims underlying these fairly labour intensive procedures are manifold, but have in common a quest for significant pattering in the nature of surface artefacts. Firstly, we are looking for absolute densities of artefacts, in total and divided into categories, across the mound and, secondly, for the relative frequency with which one or more types of artefact occur in association with other types. Patterns of obsidian distribution or of handmade sherd distribution, for example, may thus be detected, and correlations between bone and ceramics, for example, may be established. Thirdly, degrees of fragmentation of all categories of artefact, in absolute and relative terms, can be established and examined across the mound. Fourthly, we hope to establish the extent to which patterning of artefacts on the surface of the mound relates to the distribution and nature of subsurface artefacts and features. These patterns are looked for in the full knowledge that they are certain to be the result of a complex web of cultural, depositional and post-depositional factors which are unlikely fully to be apprehended in our interpretations. A significant proportion of the artefacts collected from the surface of a mound have their cultural and depositional origin in higher levels of the mound which have now eroded away, and therefore bear no contextual relationship to cultural deposits on or under the surface.

Surface and subsurface work at a number of Near Eastern sites and elsewhere, however, has demonstrated that surface artefact patterns may significantly relate to patterns of occupation in deposits immediately underlying the mound's surface and, in a more general sense, have demonstrated the validity of employing a systematic approach to the collection and analysis of surface material. The first explicitly systematic attempt to deal with surface artefacts was made in the 1960s by Binford at the native American site of Hatchery West, where clusters of classes of artefacts were interpreted as traces of specific occupations or activity loci. Subsequent excavation of the site, however, detected no significant correlations between surface and subsurface materials and features (Binford 1972).

The prehistoric site of Ayngerm in southeastern Anatolia was treated in a similar manner by Whallon, again in the 1960s, with artefacts being collected from 10% of the site's surface. Material was collected and analysed from a total of 99 squares, each measuring 5 by 5 m. The results were presented in a series of contour maps, showing densities of different types of artefact across the mound's surface, with the aims of indicating chronological variations in occupation, the differential distribution of artefact types and the correlation between various artefact categories, all with the intention of providing information upon which to design an excavation strategy (Whallon 1980). The sites of Çayönü and Girikihaciyan, both in southeastern Anatolia, were also subjected to systematic surface treatment, this time by Redman and Watson, in the 1960s. In both cases 10% of the mounds, in 5 by 5 m squares, was collected. Contour maps of artefact densities show ratios between artefact categories. Test trenches at both sites established a very strong correlation between the nature of surface and subsurface artefact assemblages, to a depth of up to 50 cm (Redman and Watson 1970; Watson and Leblanc 1990).

Our programme of surface collection at Çatalhöyük is designed as an integral element of a tripartite strategy involving surface work, subsurface scraping and excavation. Of these three elements, the first two are being employed in 1993 and 1994. To some extent, each element is regarded as informative and significant on its own terms while at the same time providing data which will shape the wider strategy. Thus, our 1993 surface studies have provided important information which can stand by itself, but which also has relevance for subsurface scraping and, in the future, for full scale excavation. A great deal of the data can most usefully be provided in the form of distribution maps which indicate the densities of artefact types across the contours of the main mound (Fig. 2; Fig. 3; Fig. 4; Fig. 5; Fig. 6; Fig. 7; Fig. 8; Fig. 9; Fig. 10).

In all, a total of 242 squares, each measuring 2 by 2 m, was collected at Çatalhöyük in 1993. This represents a total area of 968 m2 , or approximately 0.7% of the total surface area of the main mound if calculated at 14 hectares. Further surface collection in 1994 should increase this proportion to about 1%. Although not in absolute terms a large proportion, we are confident in the validity of our conclusions given the relatively close spacing, at 20 m intervals, of collection squares and the systematic use of 5 mm sieves throughout. The overall artefact category make-up is as follows:


Item	Number recovered	%age of total
Ceramics	11,419	54.06
Bone	6,540	30.96
Clay/daub/tile/plaster	1,474	6.98
Obsidian	1,416	6.70
Charred remains	118	0.56
Flint	53	0.25
Ground stone	45	0.21
Other (metal, slag, shell, glass)	58	0.27
Total	21,123	99.99

Thus, a total of 21,123 artefacts was recovered from 968 m2, giving an average density across the mound of 22 artefacts per m2. At the Neolithic site of Çayönü 15,000 artefacts were retrieved from 2,100 m2 (84 squares, each 5 by 5 m), an average density of just over 7 artefacts per m2, while at the Halaf period site of Girikihaciyan 15,000 artefacts were recovered from 2,725 m2 (109 squares, each 5 by 5 m), an average density of 5.5 artefacts per m2. These lower densities may be at least partly accounted for by the lack of use of sieves at these sites.

Figs. 2-10 show the distribution of a range of types of artefacts across the mound. These maps employ a 20 by 20 m grid: the actual collection squares, each measuring 2 by 2 m, were located in the southwestern corner of each 20 by 20 m square. All the surface distributions have been shown over a basic contour plan of the site, with contours at 1 m intervals. The surface distribution of all artefacts, en masse, is depicted in fig 2, showing a fairly even spread across much of the mound, with notably higher frequencies on and around the three eminences. The most consistently high densities occur to the southeast of the principal southern eminence, while the lowest frequencies are found to the west, north and east of the northern end of the main mound. These differential distribution patterns are largely explained by fig. 3, which shows the distribution of the total number of ceramic sherds. Here we see very high densities of sherds to the southeast of the main southern eminence, slight increases in density around the smaller two eminences, and a relatively even spread of sherds in all other areas. The distribution of handmade sherds, fig. 4, shows a very distinctive pattern of higher densities around the three eminences, but study of fig. 5, showing the distribution of handmade sherds as a percentage of all sherds, clearly demonstrates the extent to which the handmade material is completely swamped by wheel made sherds in all of the southern area of the site. This map shows at the same time a relatively unadulterated spread of handmade pottery across the northern and eastern reaches of the mound, a clear indication of the likelihood of finding undisturbed Neolithic occupation immediately under the surface. We shall see below how well these surface indications match with results obtained from subsurface scraping.

Fig. 6 and fig. 7 show the distribution of sherds measuring less than 2 cm as a percentage of all sherds, by number and by weight. These maps show an increase in smaller sherds towards the edges of the mound and in areas where handmade sherds are more frequent. Smaller sherds are perhaps likelier to have been moved to the edges of the mound by erosional factors, but the phenomenon may also be accounted for by the appearance at the mound's edges of Neolithic levels containing higher frequencies of small sherds, caused by the greater friability of handmade, as against wheel made, sherds. In more central areas of the mound these Neolithic levels are clearly covered by later occupation. As a methodological point, fig. 6 and fig. 7 show that both counting and weighing surface sherds in size groupings lead to approximately the same results, indicating that, in average terms, sherds less than 2 cm each weigh about the same amount, as do sherds larger than 2 cm, again in broad average terms. This means that in future we can probably dispense with the time-consuming practice of weighing all sherds, once having divided them into size groups.

Fig. 8 shows the distribution of bone. This distribution has a marked similarity to that of fig. 5, strongly suggesting a connection between the distributions of bone and handmade pottery, with high densities on the northern and eastern eminences and a low density in the area of high wheel made sherd density. This may reflect a Neolithic practice of dumping bone, as kitchen refuse, in areas immediately adjacent to dwellings, in contrast to, for example, a practice in Classical times of taking refuse off the site to be dumped in special areas.

The distribution of obsidian pieces is shown in fig. 9. Once again we see higher densities around the three eminences but also in the northwestern lower slopes of the mound. As with bone, but to a lesser extent, the distribution of obsidian is similar to that of handmade pottery. Fig. 10 depicts the distribution of obsidian pieces smaller than 1 cm, as a percentage by number. Here we see some slight increase in the frequency of smaller pieces around the peripheries of the mound, perhaps caused by erosional movement, but also perhaps by the occurrence of earlier levels on the mound's fringes. Detailed study of the typology of obsidian across all areas of the mound may also suggest the practice of carrying out obsidian knapping on the peripheries of the mound, in order to keep dangerous waste material at some remove from habitation areas.

Subsurface scraping - Introduction

During the 1993 season a programme of subsurface scraping was begun at Çatalhöyük, this season entirely on the main mound. As with surface collection, this programme is very much an integral part of the total strategy, while at the same time providing significant results on its own terms. In overall terms, subsurface scraping in the 1993 season has greatly enhanced our knowledge of occupation at Çatalhöyük, yielding important indicators for future work. The main aim of scraping is to investigate and establish patterns of architectural and occupational activity across the site, upon which chronological and functional distinctions can be based. This information, often extremely detailed, may then be used as a prelude to full excavation in selected areas.

The technique of subsurface scraping is one that has long been employed by Near Eastern archaeologists, but generally purely as a prelude to excavation. Its use as a full-blown technique in its own right, in order to provide detailed information without excavation, originates in work at the Sumerian city of Abu Salabikh in southern Iraq, where a substantial proportion of the surface architecture has been planned in this way (Postgate 1983; Matthews and Postgate 1987). Good results have also been obtained by subsurface scraping at Jemdet Nasr in central Iraq (Matthews 1990).

At Çatalhöyük, the density of surface vegetation and possible soil build-up initially raised some doubts about the applicability of the scraping technique. Once in use, however, it was soon established that the site is greatly suited to scraping, with architecture and features clearly visible at depths varying from 5 to 50 cm below the surface. In general, greater amounts of soil had to be removed as scraping progressed further down the mound's slopes. In several areas, however, extremely clear architecture could be traced only a few centimetres below the modern surface of the mound. In essence, there are three stages in the technique of scraping. Firstly, after removal of vegetation all surface artefacts are collected and kept for study. Secondly, a spit of surface earth is removed using pick-axes, shovels and baskets. All artefacts from this earth are again kept separately. Thirdly, the resulting surface, which may be 10 cm below the modern surface, is scraped clean with wide-bladed hoes. These hoes are not locally available in the region of Çatalhöyük, but the local blacksmith was able to manufacture some first class tools for our use. The blades should be wide and very sharp, in order to cut the mound's surface and scrape back all loose earth which is then removed by shovels and baskets. If after this scrape no architecture or soil changes are visible it may be necessary to remove a further spit of earth before rescraping. Again, all artefacts from the scraping itself are kept separately. Once the square has been scraped clean, all architecture and features are planned and recorded in detail using standard forms and soil description methods.

Our original intention had been to scrape sample areas measuring 20 by 20 m within a 100 m grid across the site. Several factors persuaded us to change our sample area to 10 by 10 m within a 50 m grid. Firstly, the smaller than expected total area of the main mound, 14 hectares instead of 20, meant that the separation of sample squares from each other by 100 m in each direction suggested we might end up completely missing vital areas of the mound. Secondly, substantial soil build-up on the lower slopes of the mound would have meant a large investment in time and labour in order to clear 400 m2 (20 by 20 m), whereas an area of 100 m2 (10 by 10 m) could be cleared and scraped in one day or, at most, two, an important consideration given the relatively short duration of our season. In one case, at the northern end of the northern eminence of the main mound, we did expand an initial 10 by 10 m square into a full 20 by 20 m expanse, due to the exceptional clarity and accessibility of the subsurface architecture. In all, 9 sample areas were scraped, 8 of them measuring 10 by 10 m, and one measuring 20 by 20 m, yielding a total scraped area of 1,200 m2, or 12% of one hectare. The distribution of these scraped squares is shown in fig. 11.

In general terms, subsurface scraping at Çatalhöyük in 1993 established that Neolithic dwellings occur at almost all points of the main mound so far investigated, but also that in many areas the Neolithic levels have been substantially disturbed by much later, Classical, occupation, thus rendering much of the Neolithic occupation not comprehensible without full scale removal of the overlying deposits. At other points, however, more or less completely undisturbed Neolithic buildings were exposed immediately underlying the mound's surface. In the following analysis, the material from all scraped squares will be discussed square by square, under the heading firstly of architecture and features and secondly of artefacts. In the process, the subsurface evidence will be compared to the surface evidence already considered above.

Architecture and features

Coherent rooms and parts of structures were identified in scraped squares all over the mound, but some architectural features were too badly affected by later disturbance or erosion to be comprehensible. The following account of architecture and features, as illustrated in figs. 12-22, is based on the site grid system, commencing with the southwesternmost scraped square. All grid co-ordinates refer to the southwestern corner of the scraped square.

In square 940, 1040 (fig. 12), located only 20 m to the north of the northern edge of the 1960s' trenches on the western slopes of the mound, considerable amounts of earth had to be removed before the appearance of any features. Even after the removal of some 45 cm of earth across the square only extremely fragmentary remains could be detected, including one short stretch of wall, constructed of large silty bricks, running east-west along the southern edge of the trench. The size of these bricks, at least 60 by 25 cm, suggests a Neolithic date. Apart from faint bricky patches elsewhere in the square, the bulk of the deposits here were clearly affected by severe Classical disturbance, as indicated by large quantities of Hellenistic and later pottery.

Square 990, 1040 (fig. 13), just to the north of the summit of the main mound, yielded some coherent features after the removal of 15 cm of surface earth. A further 5 cm were removed, as an experiment, in the northwestern quadrant of the square. Several pits, at least two of which are brick-lined, and most of which may be graves, dot the square. The series of walls in the southeastern corner form at least three sides of a heavily burnt room, with singed wall plaster and burnt room fill. Parts of in situ pots, all wheel made, and the size of the bricks, here and in all walls in this square, with a maximum dimension of less than 45 cm, indicate that all visible architecture in this square belongs to the Classical period, although their precise dating is uncertain.

In square 990, 1090 (fig. 14), on the lower northern slopes of the main eminence, a largely coherent plan of a building, on a north-south alignment, was revealed after the removal of 12-15 cm of earth. All walls were built of the same fine textured pale orange bricks, with a maximum measurement of over 60 cm, but with little detectable wall plaster. At least two stretches take the form of double walls. The alignment and construction of all walls in this square strongly suggest a Neolithic date.

Our most exciting results in the scraping campaign came from the 20 by 20 m square at 1030, 1180 (fig. 15). Initial scraping of square 1040, 1190 produced results sufficiently encouraging for us to clear three further 10 by 10 m squares, covering an area of 400 m2 in total. All walls and features in this area, to the north of the northern eminence, were revealed after the removal of only 10-12 cm of earth. It is immediately clear from the plan that there is very little in the way of later disturbance in this area, apart from a few erosion gullies or pits. In general terms, the square contains clear remains of a conglomeration of closely packed buildings, all of Neolithic date. There are two complete buildings in the centre of the scraped area surrounded by parts of nine or ten other buildings. These structures bear many similarities to those excavated in the 1960s, over 200 m to the southwest. The range of room sizes, technique of wall construction, using large bricks often over 1 m in length, use of parallel double, occasional triple, walls are all features found in Neolithic structures excavated at Çatalhöyük in the 1960s. Several rooms have multilayered white plaster, in some cases clearly forming moulded features on interior wall faces. No traces of painted plaster were found. There are no obvious entrances into or between rooms, supporting the idea of access through the roof or crawl holes at some height above floor level. In the northwestern and southwestern quadrants series of floor deposits occur, showing that, at least in these areas, we are close to the very base of these buildings. Future subsurface scraping, further up the mound immediately to the south, is likely to expand our picture of these well-preserved buildings.

In square, 1040, 1040 (fig. 16 and fig. 17), features were firstly planned after removal of some 25-30 cm of earth (fig. 16), and a second plan was made after the removal of a further 10 cm of earth (fig. 17). Apart from an extensive series of pits and scattered heavily burnt deposits, rich in animal bone, the main features are two east-west walls constructed of large, fine textured, pale orange bricks, some over 1 m in length, of Neolithic type. A semicircle of laid bricks was also found in the southwestern corner. After the removal of 20 cm of earth from square 1040, 1090 (fig. 18), no clear features were revealed, so that a further 10 cm were removed from the southern half of the square. The wall running to the northwest from the southeastern corner is built of large bricks, maximum dimension 80 cm, of probable Neolithic date. Other features, including pits and part of a circular structure are of uncertain date.

In square 1040, 1140 (fig. 19), close to the northern eminence of the main mound, some walls and features were visible. The clear wall in the northwestern corner, of bricks 40 by 25 cm, is of probable Classical date, but other walls, all unclear in detail, may be earlier. Features and walls in 1040, 1240 (fig. 20), located towards the northern limits of the main mound, did not show clearly even after the removal of 40 cm of earth. Deposits were partly covered in wash material from higher up the mound and partly disturbed by modem trenching. The dating of the few visible remains is uncertain.

Little coherent architecture was revealed in square 1090, 1040 (fig. 21), immediately to the west of the eastern eminence of the main mound. Most features took the form of pits, some of which are likely to have been burials. One definite burial, in the northwestern quadrant of the square was exposed and fully excavated (fig. 22). The burial was capped by two terracotta tiles, each with diagonal finger-marks (fig. 22a), which covered the burial of a small infant (fig. 22b) without grave goods. The body was aligned east-west, with the head to the west. The nature of the tiles suggests a Byzantine date for this burial.

Artefacts

Before examining the nature of artefacts from particular scraped areas of the mound, we can make some observations on the relationship between artefacts from surface collections and those from subsurface scraping. As a general point, it should be kept in mind that the amount of soil removed from different scraped areas varied enormously, depending on the depth needed to uncover clear architecture, so that absolute comparisons with the more tightly controlled surface material may be misleading. Nevertheless, some patterns in the distribution of ceramics seem to emerge. Fig. 23 shows the total number of ceramic body sherds across the scraped areas, which can be compared to fig. 3. The densities of ceramics generally agree on the two maps, particularly as regards the very low sherd density in area 1030, 1180. Indeed, it can be surmised that low surface artefact density correlates strongly with the presence immediately under the mound's surface of intact, undisturbed deposits and features, as is clearly demonstrated in square 1030, 1180. Again, the distribution of handmade sherds, shown in fig. 24, to some extent agrees with that shown in fig. 4, while high densities in the southeastern scraped squares may be a factor of the large quantities of earth removed. When converted into percentages, fig. 25 and fig. 5, there is much greater closeness of fit, with higher frequencies to the north and east. These comparisons strengthen the idea that fig. 5 provides the clearest indications of where on the mound to scrape and/or excavate in order to find relatively undisturbed Neolithic occupation.

Considerable quantities of diagnostic sherds were recovered and processed this season. Much work remains to be done on various aspects of ceramic analysis, but some preliminary comments are made here. In essence the pottery of the main mound at Çatalhöyük derives from two main periods, the Neolithic and the Classical (comprising Hellenistic, Roman and Byzantine). In gross terms, these two periods can be equated respectively with handmade and wheel made pottery. There is no evidence for Bronze Age or Iron Age occupation at the site, so that a gap of some 5,000 years separates the Neolithic from the Hellenistic settlements, with some evidence of scattered Chalcolithic occupation. Apart from the 1960s' material from Çatalhöyük, the best parallels for Neolithic pottery from the site are to be found at Can Hasan (French 1967), while few parallels can be traced with the Neolithic pottery of Mersin (Garstang 1953). Of the Classical material, some parallels occur with ceramics from Tarsus (Goldman 1950). Study of the few published illustrations of Çatalhöyük pottery from the 1960s' excavations and of material now in onya Museum, indicates that almost all the Neolithic pottery recovered by us from the surface and subsurface in 1993 compares to levels VII upwards of Mellaart's trenches, particularly levels V-Ill. No examples were found of the large, straw-tempered vessels of levels XII-IX (Mellaart 1964; fig. 33; 1966; fig. 4).

Figs. 26-32 illustrate selected sherds from all scraped squares, which can be briefly treated in turn. From square 940, 1040 large amounts of Classical sherds were recovered, fig. 26:2-4, along with some Neolithic examples, fig. 26: 1. Both Neolithic and Chalcolithic sherds occur in square 990, 1040, fig. 27:1-3, and Neolithic sherds in square 990, 1090, fig. 27:4-5. Almost exclusively Neolithic sherds were found in square 1030, 1180, fig. 28:1-2. From square 1040, 1040 came many examples both of Classical sherds, fig. 28:3-4, and a wide range of Neolithic types, fig. 28:5, 9:1-5, including vertically pierced lugs, open forms, small bowls and vessels with horizontal incised bands around the outer rim, these latter occurring in levels III-II of the 1960s trenches (Mellaart 1962; 54). From square 1040, 1090 came Classical sherds, fig. 29:6, fig. 30: 1, Chalcolithic sherds, fig. 30:2, and Neolithic examples, fig. 30:3-5. In square 1040, 1140 Neolithic sherds were found, fig. 31:1-2, while in square 1040, 1240, at the northern end of the mound, a greater range of Classical types was recovered, fig. 31:3-5. A few Classical types occur in square 1090, 1040, fig. 31:6, but from this square came a host of Neolithic types, fig. 31:7, fig. 32:1-5, comprising mainly open, everted forms, suggesting a date late in the Neolithic.

Artefacts other than pottery were found in small numbers. Several bone tools, principally awls, fig. 33:6, were retrieved, as were fragments of small polished stone axes, fig. 33:5. Clay objects include figurines, fig. 33:3-4, and an unusual miniature table or stand with hemispherical feature, fig. 33:2. One of the most striking finds, from square 1040, 1040 but of unknown date, is a ceramic face with faint traces of red paint, fig. 33: 1.

References cited:

Binford, L.R. (1972) 'Hatchery West: site definition - surface distribution of cultural items', in L.R. Binford An archaeological perspective (New York: Seminar Press), 163-181.

French, D. H. (1 967) 'Excavations at Can Hasan, 1966. Sixth preliminary report', Anatolian Studies 17, 165-178.

Garstang, J. (1953) Prehistoric Mersin (Oxford: Clarendon Press).

Goldman, H. (ed) (1950) Excavations at Gözlü Kule, Tarsus I. The Hellenistic and Roman periods (Princeton: Princeton University Press)

Matthews, R.J. (1990) 'Excavations at Jemdet Nasr, 1989', Iraq 52, 25-39.

Matthews, R.J. and Postgate, J.N. (1987) 'Excavations at Abu Salabikh, 1985-86', Iraq 49, 91119.

Mellaart, J. (1962) 'Excavations at Çatal Hüyük. First preliminary report, 1961', Anatolian Studies 12, 41-65.

Mellaart, J. (1964) 'Excavations at Çatal Hüyük, 1963. Third preliminary report', Anatolian Studies 14, 39-119.

Mellaart, J. (1 966) 'Excavations at Çatal Hüyük, 1965. Fourth preliminary report', Anatolian Studies 16, 165-191,

Postgate, J.N. (1 983) The West Mound surface clearance (London: British School of Archaeology in Iraq. Abu Salabikh Excavations 1).

Redman, C.L. and Watson, P.J. (1970) 'Systematic, intensive surface collection', American Antiquity 35, 279-291.

Watson, P.J. and Leblanc, S. (1990) Girikihaciyan: a Halafian site in southeastern Turkey (Los Angeles: UCLA. Institute of Archaeology Monograph 33).

Whallon, R.(1980) The systematic collection and analysis of surface materials from a prehistoric site in southeast Anatolia', in H. Çambel and R.J. Braidwood The joint Istanbul-Chicago Universities' prehistoric research in southeastern Anatolia I (Istanbul: Istanbul University Faculty of Letters), 207-219.