THE NEOLITHIC SITE OF HAJ YUSIF ( CENTRAL SUDAN )

RESUMEN Informe de la excavación de 1989, que incluye la descripción y análisis preliminar de la cerámica, útiles líticos y fauna. A pesar de su condición alterada por ocupaciones posteriores, en e! yacimiento se han podido detectar dos áreas con distinta proporción de motivos decorativos en la cerámica, lo que implica diferencia funcional o, más probablemente, cronológica. El asentamiento parece haber sido de corta duración y corresponde a un grupo pastoril de alta movilidad, con rasgos similares a otros conocidos en e! Sudán central durante el «Neolítico de Jartúm», tipo Shaheinab, en la segunda mitad de! quinto milenio a. C.

in press).It is located about 10 Km east of Khartoum North, beside the earth track to Abu Deleiq, halfway between Haj Yusif and Gereif East, in the area administratively known as Haj Yusif New Extension.It is 5 Km east of the Nile river (the coordinates are 32 2 , 3S', IS" E, 15 2 36' 30" N) (Fig. 1.1.),To our knowledge, the area has not yet been archaeologically inspected, with the exception of the salvage digging by P. Lenoble at the Meroitic cemetery of Gereif East (Geus, 1984: 12-3;Geus, 1986;32-3).
In recent years the zone has started to become urbanized, and numbers of new houses have been erected, sparsely distributed, sharing the arid land wilh rough temporary shacks occupied by refugees fram the South (two of these, Dinkas, were employed as workers in the dig, together with eight Arabs fram Haj Yusif).Possibly as a result of the building activities, a large area was excavated for quarrying just in the middle of the site before our arrival in 1981, destroying a great deal and seriously menacing the rest (Fig. 1.2.).
The site was first inspected on surface, by collecting pot-sherds and recording the most important features.A planimetric survey was made on this essentially flat area (the greatest difference in altitude, between the low ridge along the telephone line and the bottom of the quarry pit, is about 1.50 mts.),This survey marked the outer limits of Neolithic pottery distribution (points in fig.1.2.), the two heaps of red bricks north of the line (possibly the remains of a Christian medieval church), and the quarrying hole.A quick inspection of the sherds picked up from the different areas did not reveal significant differences, but special concentrations were clearly detected (more dense points in the map) and were taken into account when deciding where to digo An overall site grid was established, using a line paraHel to the telephone line, 100 mt to the NE, as the longest axis, and the datum point was put to the N of the surface distribution (Fig. 1.2.).The trenches were consequently numbered in a two dimensional system of square meters; trench No. 1 (measuring 5 X 2 meters) is listed, for example, as BO-BS /116-117.
The distribution of Neolithic sherds covers 300 X 150 mts, the complete surface of the site thus amounting to about 4,5 hectares.The subsequent excavation, however, revealed this figure to be an exaggeration of the real size of the settlement.No doubt due to water and wind erosion, and to later activities in Meroitic and Christian times, the Neolithic sherds and lithic implements had spread beyond their original distribution.This seems to have been located in the middle of the area and, consequently, most of it has been destroyed by the modem quarrying activities.A total of 9 trial pits were dug in the more promising zones, but only those in the middle ares (Nos.3, 4, 5, 6, S and 9) unearthed enough artefacts, more or less in their original position for the place to be considered a part of the actual settlement.Trench n 2 . 1 yielded a few Neolithic sherds mixed, even in inverse stratigraphic position, with sherds of the Christian periodo Trench n!! 2 revealed the top of a partially destroyed (plundered) tumulus of Meroitic or (more probably) post-Meroitic date, together with a very few prehistoric artefacts.Trench n 2 7 yielded nothing at aH.
The geological interpretation, based exclusively on surface and subsurface observations at the site, and on general descriptions of the area North of Khartoum by the italian Mission, must be viewed as merely provisional and preliminary.Two kinds of deposits, below the loase sand on the surface, were recorded during the dig: a light brown mixture of sand and small watered pebbles, and a much darker grey /black, fine grained clay of harder consistency.The former was detected in the more e1evated areas, such as trench N2. 1, which has deposits 1.4.mt thick (completley sterile after the initial 60 cms).Isolated remains of this formation could be seen in other, lower parts as in trenches 4 and S, while the excavated areas in the middle zone, of still lower elevation, revealed a continuos stratum of dark mud, over which the Neolithic settlers left their debris.This flat band probably corresponds to the upper clay member of the Gezira formation, that lies almost exclusively in the east bank of the river and may be dated at the terminal Pleistocene.The overlying lenses of gravel and sand could be ascribed to the early and middle Holocene (before ca.5000 B. P.), when the Blue Nile flowed along a few channels cutting the Gezira formation.During this wetter phase, the Neolithic settlements usually occupied isolated hillocks on the edge of the alluvial plain along the water coursc, which was alreadv in the present westem channel before the end of the Neolithic period (Marcolongo, 1983: Marcolongo, Palmieri, 1988).
The Neolithic strata.apparently intact, in trenches 3, 4, 6 and 8, were of only 10-20 cm in thickness, containing a big number of potsherds and millstones, and a few lithic tools and waste.No remains of huts or hearths were detected, and the faunal remains, shclls included, were scarce.The Christian sherds, usually thick, black bumished potsherds with sorne incised decoration, were collected in smaU numbers only in the upper sand layer aboye the Neolithic stratum.These deposits were thinner than usual in the area (60-70 ems are attested at Shaheinab and Kadero;40-50 cms at El Geili;Caneva 1988: 24).During the Mesolithic period the sites show thicker strata, suggesting a more regular occupation or less in tense erosive processes (70-135 cms at Saggai, Caneva, 1983: 18).In spite of this fact, the site of Haj Yusif was not disturbed by subsequent Late Neolithic burials, as in Geili or Saggai, and the cultural assemblage belongs exclusively to a single prehistorie phase.
The trial excavation in the central area of the site, trenches 5 and 9, joined by numbers 9/1 to 9/3 (squares FF-FS/143-14 in the general grid), pro ved to be mueh more interesting.In this zone an almost superficial, thin layer of Christian remains, eonsisting of small hearths, bones and sherds, suggests temporary activity linked to the nearby buildings.In sorne places (n Q S, 9 and 9/1), the irregular, discontinuous lenses overlay patches of intact Neolithic deposit.By intact we mean here a cóndition somewhat different to the one present in the other trenches previously mentioned: the horizontal position of sherds, grinders and shells suggest a more stable occupation of the area.In zone n Q 9/2, however, the Neolithic layer had been destroyed by a Christian hearth (or never existed at all), and in 9/3 all the remains had been washed out by superficial water runoff.
A sterile level (of only 7 cm) in trench n Q S, under the Christian debris and aboye the remains of two hearths, moved us to think that these could be of Neolithic date.The hearths were of a more consistent character than those found in the superficial level, and they had a very few Neolithic sherds in the nearby.The radiocarbon analysis of two charcoal samples, however, revealed that they actually were made at the beginning of the Middle Ages (Groningen nos.GrN-16554, 1265 ± 25 BP, 1205 ± 25 BP).
In the nearby zone of trench 9, a larger area was recorded with potsherds, broken milling stones and querns (one of them associated to a piece of red ochre), and a fairly large number of limicolaria and fewer Pila sheUs.Although the remains appeared to constitute a «living flooD), the pottery sherds carne from a lot of different vessels.Only a few sherds, actually those that lay together on the ground, could be assembled together.

The cultural assemblage
The pottery (Fig. 2) was the most abundant artefact at the site, and a preliminary comment will be attempted here.The previous classifications of Khartoum Neolithic impressed ware, by Arkell (1953: 68-77), which was followed by the Mission in the first report on the site (Fernández in press), Haaland (1987: 144-80), Mohammed Ali (1982: 74-82), Chlodnicki (1984) and Caneva (1988: 67-114), differ with respect to which decorative atributes are considered more diagnostic, and consequently are very difficult to compare or follow as a whole.The typology elaborated by Isabella Caneva, hierarchically based on the sequence of decorative technique, implement used to perform the impressions, motif and structure, and not merely on the visual impression of the final result by the classifier, seemed to us the most culturally significant and simple to apply, and so has be en the system followed in the analysis of our ceramic material.
A total of 1833 pottery sherds were recovered from the excavated areas, the overall percentages of the types being as• follow: rocker stamp resulting in packed zig-zags, with evenly serrated edge (33.7 %) ur with unevenly serrated edge (26.7 %); rocker stamp, evenly serratcd, but resulting in spaced zig-zags (17.2 %); rockcr stamp uf plain and curved edge (2 %); alternately pivuting stamp wilh double pronged implement, resulting in paired lines of single dots (8.1 %), and of opposed triangles (4.7 %); simple impression of dotted lines, with serrated edge (0.3 %); incisión of simple and double lines (6.7 %) or scraping comb (0.2 %); red burnished black topped ware, with black triangles on the rim (0.4 %).Only a small fragment of dotted wavy line, loo eroded for the technique be conspicuous, was picked up on lhe surface of lrench nI! 9.The rim decoralion was mostly of oblique dotled lines.
The percentages look very similar to the Geili data (Caneva, 1988: lable 2), lhe differences being in lhe rocker comb lechnique (77.2 % in Haj Yusif and 44.8 % in Geili), the simple impression of comb (0.3 % as opposed lo 7,9 %) and undecorated sherds (14.1 % in Geili; negligible number in Haj Yusif).In Geili, however, the sherds from two occupations (sorne of them are Late Neolithic) were mixed up and so the data are hardly comparable.Another problem comes from the fact that the excavators in Geili díd not sort out the spaced from the packed zig-zags in the rocker stamp technique, a distinetion of probable chronological significance in our sÍte.
The table of frequencies and percentages of types in the trenches was analyzed by several statistical multivariate techniques (namely Cluster, Principal component and MDSCAL anaIysis) and a significant c1uslering of trenches and correIation of pottery types were c1early apparent in aH the results.The area in the middIe of the site (i.e. trenches 5 and 9) is characterized by a higher frequency of spaced zig-zags (30.7 %), alternately pivoting stamp (14.4 %, 10.8 for the dots), and by a lower quantity of packed zig-zags (52 %) and incision (2.6 %).In contrast, the outer parts of the site (trenches 3, 4, 6 and 8) has Iess spaced zig-zags (10.6 %) and aIternately pivoting stamp (5.9 %, 3.9 for the dots), and more packed zig-zags (72.5 % the increase being bigger in the unevenIy serrated edge, from 18.4 % to 35.5 %) and incision (10 %).The chronoIogical significance of these differences is indíeated not onIy by the increase of incised pottery (a continuous trend during the Neolithic) in the periphery of the site, but a1so by the faet that the onIy dotted wavy line sherd oceurred in the central part and the onIy black topped sherds in the outer area, thus suggesting the first as older than the second.
The faH in the pereentage of spaeed zig-zag pottery has not been observed up to now in other sites of the area, but this kind of decoration, as opposed to the packed zig-zag (where the rocker technique, is not so evident al firsl sight), seems more frequent in the Mesolithic sites sueh as Saggai (where it was not counted apart from the whoIe rocker eomb tecnique, Caneva, 1983: fig. 13, 1-3, 6) and Early Khartoum (ArkeH, 1949: PIs. 65, 70-2 and possibIy 90-2).The alternateIy pivoting double tooth pottery was aIread y present in the Mesolithic (Caneva, 1983: fig. 16;ArkeH, 1949: PIs. 83-84), and apparently inereased in frequeney until it eventually repIaee the rocker comb technique in the Late Neolithie (Caneva, 1988: 112), this being in eontradiction with the deerease observea in Haj Yusif.The rise in percentage for the unevenIy serrated edge pottery, unknown in the Mesolithic and typical of the early Neolithic ([bid), is in accordance with the supposed typoIogical trend in the Central Sudanese Neolithic sites.
Vessel shapes at Haj Yusif include open-mouth forms (6.4 %), vertical walls (56 %) and closemouth ones (37.1 %), with minor variation in the rim shape and thickness.No possible identification of bottom sherds was possible, sinee the decoration seemingly occupied aH the vesseI surfaee, but rounded bowls were probabIy the most eommon or unique shape.With respeet to the association between shape and deeoration, vertical waHs are the usual form in all the types (also the closemouth is connected to the evenly serrate edge), except in the rare undecorated rim sherds, where inverted rims appear to be slightly more abundant.
All the aforementioned percentages were computed on the basis of sherd counts, yet a rough estimate of surfaee area was made for every sherd.The mean size was greater for the paeked zigzag (about 12 ctn 2 ) than in the other categories (about 8-9 cm 2 ), suggesting that this decoration was applied to bigger vessels.This agrees with differences in rim diametre, greater in the packed zig-zag (mean about 36 cm) than in spaeed zig-zag (32 cm), pivoting stamp (30 cm), incisión (29 cm) and black topped (25 cm).Differences in mean size were not observed between the two supposed chronological areas of the site.
The lithic assemblage (Fig. 3) was extremely scarce in the areas excavated this season, the raw material being rhyolite (60.8 %), basalt (9.8 %), fossil wood (9.8 %), and others (including quartz, gneiss and a type of caramel coloured fIjnt).No differentiation was noticed between the trenches and areas, and the complete list of retouched tools is as follows: two scrapers (one core-scraper), one steep retouched perforator, one retouched notch, five si de scrapers (mostly simple convex and one possibly made on a fragmented celt), and a truncated blade.Sir aciall~ retouched or polished implements included six worn or fragmented gouges, one complete, bifacially retouched celt; and three broken, completely polished axes.Up to sixteen unretouched or slightl ~ retouched flakes, two blades (one retouched) and two globular cores were also recovered from surface and excavated areas.Not a single geometric microlith was found, although most of the earth from the trenches was carefully sifted.
The ground stone implements were abundant, and the list coincides with the known typologies of the sudan ese Neolithic, with the exception of rings and mace-heads, absent in our site.Querns were scarce and always in a very fragmentary condition, the contrary being the case with milling stones, which ineluded circular and oval shapes, pestles, rubbers and spheroids.The raw materials was usually sandstone with sorne examples made on gneiss.No bone or shell implements were found in the excavation.

The faunal remains
As with the lithic tools, the animal bones recovered in the dig were very scanty, so that calculating their percentages would appear unsound.The preliminary identification of the different species is coincident with many of the remains from other Neolithic sites (e. g.Gautier, 1983;1988), yet the diversity looks considerably lower.Freshwater mollusks in elude Pila wernei, Aspatharia (possibly A. rubens and A. hartmanm) and Etheria elUptica.The landsnails group consisted of very abundant remains of Limicolaria cailliaudi, ubiquitous in every part of the site including the surface.As far as fish are concerned, only a single bone of catfish (Siluro ind) was found, from a very big specimen.The mammal bones were very eroded and fragmented, the identity being dubious in most cases: the big antelope Tragelaphus strepsiceros was present, as well as several remains of Capra aegagrus (or possibly Ovis ammon in one case).From trench nº 4 come several bones of a large bovid, yet the identification of 80s has not been possible.

Conclusions
As has been elearly stated, the site is in a disturbed condition, and we have not been able to ascertain wether this is due to the subsequent occupations of the zone or to the fact that the settlement belonged to a very mobile group that left few traces.The suggested chronological division of the remains, an earlier phase in the central area and a slightly later occupation in the peripherical zones, seems probable after the analysis of the respective pottery atributes, but the possibility of a functional explanation for that difference cannot be preeluded.The depositional conditions are different in the two areas, but a control for functionality, based on the lithic or faunal reIÍlains, was not possible because of their scarcity.
As an overall interpretation of the site, the abundance of grinders and scarcity of flint, specially microlithic, tools, would, in the light of the model proposed by Randi Haaland (1987) suggest that Haj Yusif was a permanent base site, agriculturally based, similar to Kadero (Krzyzaniak, 1978;1984), where similar proportions of implements existed (Nowakowski, 1984).Nonetheless, the poorly preserved, scanty remains of domes tic fauna, .together with the thin condition of the deposits, FIG, 1.-1.The location 01 the Neolithic site 01 Ha; Yusi/, in the vicinity 01 Khartoum 2. Plan 01 the site, showing the arfUl 01 Neolithic pottery distribution, Christian and post•Meroitic remains, and the modern quarry excavation.