,7+,& &+(57 48$55

1 downloads 0 Views 3MB Size Report
81 wadi terraces or scree slope deposits. In these local deposits, an important number of chert cobbles are present. Most of the cobbles are rolled and are well.

Vermeersch   P.M.,  P.  Van  Peer

  &E. !!"Paulissen, #$&%' $2002. $ #(Middle )!" Palaeolithic Chert Quarrying at Nazlet Khater 2, in Pierre M. Vermeersch (Ed.), . Leuven, Leuven University Press: 79-98.

±0,''/(3$/$(2/,7+,&&+(5748$55 1160

28N

6831159

27N

402682

26N

237410

25N

140236

24N

83139

23N

4982

22N

1-48

24

21N

23 0

Q U A R R Y

22

20N

21 20

31

30

29

28

27

26

25

24

23

22

21

20

19

18

W

21

20

19

W

W

W

Figure 4.6 - Artefact density in number of artefacts per square meter

Figure 4.5 - Excavated squares in the west sector

([FDYDWLRQV

([FDYDWLRQ5HVXOWV

On the southern part of the hill a base line has been outlined west-east. It served for the construction of a reference grid system as indicated on fig. 4.2. Using this system, each square is made individual by the north distance and by the west- or east distance in metres. The altitude has been estimated based on the elevation of the floodplain as indicated on the topographic map 1:25,000 of the British Survey. This estimation has been used in order to give an absolute, not necessarily exact, height for the top point of the highest boulder situated on top of the Boulder Hill. This point has been given an elevation of 67.12 m. Excavations proceeded by spits of 0.2 m in each separate stratigraphic unit. Artefacts have been collected separately by square metre. All excavation dump has been sieved through a mesh of 0.02 m. The artefacts from the pits dug for stratigraphic purposes, have not been collected.

From the initial excavations it was clear that most Middle Palaeolithic artefacts were no longer in their original archaeological position. Slope processes have been active for a long period of time, moving the surficial deposits, rich in local materials, alongside the hill slopes. Consequently, the excavations have been restricted to small surfaces, trying to collect a large sample of the assemblages. In addition to the archaeological excavation trenches, numerous other trenches have been dug in order to study the geological setting of the hill and the stratigraphic position of the lithic assemblages. The excavated areas comprise a west and an east sector.

:HVWVHFWRU The west sector is a restricted area (fig. 4.2), consisting of two different sub-sectors: a quarry face on the



82

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

north-western Boulder Hill slope and an area in the local deposits slightly higher on the hill.

the question remains as to why all this discarded material has been deposited in presumed extraction pits, dug in a Nilotic gravel, where raw material that could be extracted was absent. For that reason we would not completely exclude the interpretation that the local artefact rich deposits are sediments related to wadi and slope activity, reworking anterior large human deposits. If the latter hypothesis is accepted, then the disconformity between the local gravels and the slope deposits point to a situation where an important erosional activity has shaped the slope posterior to the deposit of the local deposits.

4XDUU\IDFHRQWKH%RXOGHU+LOOVORSH On the western hill slope, 22N29-31W, a small local quarry had been exploited. This quarry served as a starting point for a survey trench 22N24-28W. Middle Palaeolithic artefacts protruded from the quarry face (fig. 4.3). They have been collected and tridimentionally registered. However, no extensive excavation could be performed. At the base of this quarry face Nile gravels (fig. 4.4: 1) are outcropping up to an elevation of 64.70 m. They are covered with local deposits; first with gravely deposits, rich in Nilotic gravels (fig. 4.4: 2), later with sandy (fig. 4.4: 3) and silty (fig. 4.4: 4) deposits and finally with large local cobbles and blocks. At the interface between the Nile and the local deposits, and in the local gravels, a few white patinated chert artefacts were found. Incorporated in the local gravels, an artefact concentration (fig. 4.4: 5), produced from a single core, was present. The artefact concentration had an inclination different from the present hill slope, even cropping out of the hill where it is eroding. Apparently most artefacts had been caught behind large cobbles. In square 21N30W and 22N30W respectively 11 and 431 artefacts have been collected. In the Nile deposits, no artefacts have been found. Some artefacts from this concentration could be refitted and all of them are in a fresh state of preservation suggesting that displacement took place only over a very short distance or that they had been dumped at once. Because the artefact concentration is cut by the hill surface, the present hill morphology came into existence after the deposition of these deposits. The whole sequence is covered with a slope wash also containing a few artefacts. When we excavated the site, we had not yet been faced with the existence of large human-made deposits containing Middle Palaeolithic artefacts. At the time, we interpreted the local deposits as being the result of wadi activity, local run off and mass movement processes. Now, having understood the features of several extraction sites with huge extraction dump deposits, we are less sure about our original interpretation of the local deposits. We therefore have to accept the possibility that the local deposits are of human origin and could be considered as extraction dump material. If that interpretation is correct,

/RFDOGHSRVLWVRQWKH%RXOGHU+LOOWRS Stratigraphic position A profile (fig. 4.7) studied along 24-28N20W has been very informative concerning the position of the artefacts on the western sector of the Boulder Hill. The sterile Nile gravels, up to an elevation of 64.4 m, form the base of this profile. They have been lowered by at least three subsequent wadi channels or human digging activities, resulting in the deposition of sandy deposits with numerous artefacts (fig. 4.6). If we interpret the local deposits as related to wadi activity, the sequence of events can be summarised as follows: Above the green clays and sands, forming the Boulder Hill substratum, an exotic gravel (fig. 4.7: 1), presumably of Nilotic origin, has been aggraded up to an elevation of at least 64.4 m. Later, a local wadi provided the aggradation of a sandy deposit (fig. 4.7: 2) in which some artefacts, presumably derived from a nearby site, can be found. This aggradation phase was continued by fine local gravel (fig. 4.7: 3) in which numerous elements of the underlying Nilotic gravels remained included. This sedimentation unit comprises an important number of artefacts, which are probably derived from the same nearby site, and numerous very tiny charcoal fragments (from a depth of 60 to 80 cm), of which only 0.3g was preserved after sample preparation. 14C dating of this sample gave the following result: >35700 BP (GrN-10578). In a final stage of this wadi (or human) activity, the sediments are composed of gravels of mainly local origin, still containing artefacts. Later, wadi activity was continued by a new wadi bed (fig. 4.7: 5) aggrading very sandy deposits but also incorporating a large block of silicified limestone (fig. 4.7:11), similar to those that rest on the hill

83

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________

Figure 4.7 - Profile (24-28N20W) on the Nazlet Khater 2 site, with positions of artefacts (x) embedded in wadi (or dump) deposits. 1: Nile gravels; 2: sandy deposit; 3: deposit with local and Nile gravels; 4: deposit with mainly local gravels; 5: very sandy deposit; 6: gravely deposit; 7: aeolian sands; 8: loose slope deposits; 10: desiccation wedge with aeolian sands; 11: large silicified limestone blocks. Artefact positions have only been indicated in deposits 2, 3 and 4.

28N 28N > 64

27N

3364

26N

1732

25N

9-16

5-8

3-4

1-2

0

> 64

27N

3364

26N

1732

25N

9-16

24N

5-8

23N

3-4

22N

1-2

21N

0

20N

24N 23N 22N 21N 20N

19N

21

20

19

W

W

W

19N

Figure 4.8 - Levallois flake density in number/m²

21

20

19

W

W

W

Figure 4.9 - Levallois core density in number/m²

84

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

2WKHUDUHDVIURPWKHZHVWVHFWRU

surface. Finally, the wadi bed shifted, still in southern direction, and aggraded a gravely local deposit (fig. 4.7: 6) up to an elevation of at least 65.20 m. All over numerous artefacts were present in those deposits (fig. 4.6). In some squares densities of over 1000 artefacts per square metre were found. The successive “wadi” deposits can also be interpreted as large dumps from extraction activities nearby (?). Nowhere on the hill, however, have we located an exploitable chert cobble source. The present Boulder Hill topography, with its thin veneer of slope deposits (fig. 4.7: 8), intersects all previously mentioned deposits. This suggests an important denudation period responsible for shaping the hill, posterior to the human activity. In 23-24N20W, the deposits reach a thickness of 1 m. Artefacts are situated mainly at the interface between local and Nile deposits.

In trench 35-36N20W, the general sequence is similar to that of the previous sectors. Here the local deposits, containing artefacts, are of rather limited thickness, not exceeding 0.25 m. They rest on fine greenish gravels, granulous sands and consolidated silts with a calcrete in the upper part. The area around 10N1W had been disturbed by an 11N

10N

9N

8N

7N

Artefact distribution In the artefact count from fig. 4.6, chips are included. The artefact distribution is very dense but diversified, varying from square metre to square metre. A very important concentration occurred in square 21N20W. A high artefact number was also recorded in squares 27-28N19-20W. Artefacts were collected essentially from the sedimentation units 3 & 4 (fig. 4.7). At the time of the excavations we believed the artefacts were no longer in their original archaeological position, as we considered them to be transported by wadi activity. Therefore, no larger excavations have been planned. We presume that the human occupation debris did not undergo an important displacement, as artefact refitting (77 sets) was possible and the artefacts remained in a very fresh state of conservation. Most refitting occurred between material from sedimentation units 2, 3 and 4. However, refitting over a larger distance between different fills, was also possible: 1 set between layer 3 and 5 and 2 sets between 3 and 6. This situation suggests that the artefacts originated from a single site. In the present state of such restricted excavations, we prefer not to go into a detailed analysis of the artefact distribution pattern. It can, however be observed on fig. 4.8 and 9, that the distribution of Levallois cores and their products is similar to the general distribution.

29E

30E

31E

32E

Figure 4.10 - Artefact distribution (legend cf. fig. 4.5) Upper Palaeolithic burial (Vermeersch HWDO 1984a). In this area some Middle Palaeolithic artefacts have been collected from the slope scree, which surrounded and covered the burial fill.

(DVWVHFWRU The east sector is a restricted area of 30 m² including some trenches on the 29E-line and some more scattered trenches.

7UHQFK1( Artefacts, nearly LQ VLWX, have been recovered only in squares 9N29-30E (fig. 4.10). In the other squares within this sector most artefacts were found in a loose, superficial scree deposit. Moreover, the area has been disturbed by a human burial in 8N29E (Vermeersch HW DO. 1984a) from which the skeleton of the Nazlet Khater man has been recovered (Thoma 1984). Under a sandy scree slope deposit, containing artefacts, there is a bed of granulous sands and fine gravels,

85

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________

which, at its top, is consolidated by a calcrete. In the northern trench section a desiccation wedge, filled with aeolian sands, starts from just below the surface. In the southern trench section, artefacts can be found in loose scree deposits, resting upon 0.15-0.20 m reworked sandy Nilotic fine gravel, which is sterile, resting upon the sandy green clay substratum.

including only two artefacts. However, the largest set had 49 artefacts. Horizontal distances that are covered by refits are generally restricted. Indeed most do not exceed a distance of 1 m. Two refits are between squares 5 m apart from each other. A single one connects 9N29E with 27N19W and another one 9N29E with 20N20W. Artefacts from the upper part of the local deposits often connect with those from lower parts. Such observations suggest that most of the recovered material from the site is related to a single artefact deposit from which the artefacts have been scattered over the hill surface.

2WKHUVTXDUHV Squares 0N-0S25W, 0N5E, 3-4N29E, 5N7E, 1314N29E, 17-18N74E, 25-26N29E, 28N23-24E, 3536N29E and 45-46N29E (fig. 4.2) do not contain fresh archaeological materials. In 9N25E, an upper sandy layer contains some artefacts, that are rather fresh, but also some that are rolled, which according to their patina have been at the surface for some time. Below, artefacts become scarce. The square has been excavated down to the top of the green silts. The presence of many chips and bladelets is rather peculiar. The silts have been broken up by a 0.3 m large crack running in a north-south direction. The crack is filled with sands. In 30N26E, slope and local deposits have a thickness of 0.7 m, resting upon sterile sands, which have a thickness of about 0.15 m and are covering the silts. They have a sandy matrix. No Nile gravels are present. The local gravels are slightly consolidated at the base. Throughout the local deposits, artefacts can be found evenly distributed. Some of the artefacts are very sharp edged while others are not fresh.

$UWHIDFWV The bulk of the lithic artefacts recovered at Nazlet Khater 2 were incorporated in local deposits of the west sector. Much less artefacts were excavated in the east sector, in a stratigraphical position similar to that at the west sector. In terms of physical appearance and technological characteristics, both lithic assemblages are very similar as well. There are two refits that link west sector artefacts to east sector artefacts found at the surface. In view of the large degree of similarity between both assemblages and despite the rather weak refitting evidence, it was decided to combine the two into one sample to be studied as a homogeneous assemblage. Artefacts from other stratigraphical positions or other locations e.g. the Late Palaeolithic artefacts from the eastern side of the hill are not treated in this account. The artefacts are generally in fresh to very fresh physical condition. Occasionally, a rolled artefact may be present, but given the reworked position of the assemblage, this is not surprising. As a matter of fact, it is fairly easy to distinguish between the fresh and the rolled part of the assemblage. The latter was evidently excluded from the analysis to come. Numerous artefacts have suffered thermal damage and salt weathering. Sometimes cores are even completely shattered. The large majority of artefacts are out of local chert, unpatinated. A few small quartz cobbles were reduced as well.

Table 4.1 - Refits Artefacts involved less than 5 6 to 9 10 to 15 more than 15 Total

number of refits 88 4 3 3 98

5HILWWLQJ Refitting (1) has been applied to the material from the site. 98 sets have been conjoined, most of the time

(1) Refitting has been performed by the excavation team and later by G. Mertens, H. Van de Heyning and several students.

86

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

7HFKQRORJ\

The majority of the selected volumes are rather flat, round or ellipsoidal cobbles. Volumes of considerable elongation are quite rare. Sometimes, almost spherical shapes were selected for reduction. In general, the archaeological sample represents a proportional shape distribution that is different from the natural sample as present in the Nazlet Khater area. Clearly, shape selection took place, in view of the specific types of reductions to be performed.

*HQHUDO&RPPHQWV From a technological point of view the flake component is heavily dominating the Nazlet Khater 2 assemblage. Proportions of cores show that the Levallois reduction strategy was mostly used to produce the former. Clearly, this is a Middle Palaeolithic assemblage. Typological arguments in this respect are less strong because of the paucity of retouched tools as we will see later on.

Size of selected volumes The distribution of weight classes in the sample of precores (Fig. 4.16) is slightly skewed, the smaller classes being better represented. The median of the sample is 0.180 kg. We should of course take into account the minimal reduction of most of these cores but it is reasonable to consider these figures as fairly closely representing the original selected sizes. There are no weight data on the local gravels, but it is nevertheless beyond doubt that the archaeological sample is situated in the smaller part of the former’s size range.

Table 4.2: General technological inventory (1)

Flakes fragments Levallois Preparatory Elements type 1 type 2 type 3 Safaha blades fragments Levallois Endproducts fragments Blades fragments Cores Levallois fragments Kombewa discoidal single/multiple platform other and fragments precores and preparatory Leval. cores Debris Chips Burin spall TOTAL

N 2053 1699

% 21.96 18.17

205 76 566 22 343 193 75 17 18

2.19 0.81 6.05 0.24 3.67 1.84 0.80 0.18 0.19

105 33 2 15 18 31 133 522 3243 1 9370

1.12 0.35 0.02 0.16 0.19 0.33 1.42 5.58 34.69 0.01 100.00

7KH/HYDOORLV5HGXFWLRQ6WUDWHJ\ This is the best-represented reduction strategy in the assemblage. After selection of various types of reduction products (cores, endproducts and preparatory elements) and its validation by means of a number of reconstructed sequences, it is clear that the Levallois concept in its classic form has been applied. The specific position of the Nazlet Khater 2 Levallois reduction strategy within the classic concept will be further examined. Two Levallois cores are out of quartz, both very small.

Levallois methods The examination of core upper surfaces and dorsal faces of endproducts reveals the presence of different Levallois methods for flakes. Levallois point methods, for instance the Nubian 1 method, quite characteristic of a number of Nubian Complex assemblages, such as Nazlet Khater 1 and 3, are not represented. The majority of the cores have been reduced according to the classical Levallois method for flakes. The Safaha and Halfa methods for flakes are represented as well. In fact, both methods are strongly related to the classical flake method.

5DZ0DWHULDO6HOHFWLRQ All raw material volumes used at Nazlet Khater 2 are exclusively of local origin. Both chert and quartz cobbles were used, the latter to a very minor extent, however. Both types of raw materials are a major component of the local and Nilotic deposits.

Shape of selected volumes

87

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________

Figure 4.11 - Levallois core upper surfaces from Nazlet Khater 2 (5 and 9 with refitted endproducts); 4 and 14 are discoidal cores with refitted flakes

88

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

Figure 4.12 - Levallois flakes (1-2, 4-6, 9-10), simple ss (3), transversal scraper (7) and refitted nodule from Nazlet Khater 2

89

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________

Fig. 4.13 - Levallois flakes (1-10), Kombewa core (11), truncated flake (12) and Safaha blades (13-15) from Nazlet Khater 2

90

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

Figure 4.14 – Refittings from Nazlet Khater 2

fact, both methods are strongly related to the classical method. The Safaha method (Van Peer 1992) uses the same centripetal pattern of preparation but shows a very characteristic treatment of the distal core sector. After installation of a central distal ridge, it is removed by an elongated preparatory product with a very characteristic shape, a Safaha blade. The Halfa method for flakes as described in Marks (1968b) makes use of a bidirectional pattern of preparation as opposed to the centripetal one of the classical method for flakes. It appears, however, that this technical difference is not very explicit and that the two methods, in terms of the pattern of preparation, grade into one another. They should rather be considered as different "windows" in the range of variability that can

be present in a general classical Levallois method for flakes. For cores with the negative of a distally overpassed flake, the method can not be determined. Table 4.3 – Levallois method use in the assemblage Levallois method use as evidenced on Cores Classical and Halfa 48 Safaha 19 Nubian 1 4

Endproducts 178 15 0

Though the Nubian 1 method was stated not to occur, four cores of this type are shown in table 4.3. On closer examination, however, it is clear that all of them should be considered as preparatory stages of the Safaha method (at some point in the implementation of

91

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________

35

which a distal central ridge is created on the upper core surface). Morphologically, these cores are quite different from typical Nubian 1 cores in other Nile Valley assemblages.

percentage

Dynamic reduction aspects Twenty-four reduction sequences were selected for describing the dynamic aspects of Levallois reductions at Nazlet Khater 2.

25 20 15 10 5

N 4 11 8 1

600

550

35 30

As far as sequences involving cores and endproducts are involved, two of them show traces of a new Levallois surface preparation after the production of an endproduct. No more endproducts, however, were detached. One of those sequences was elaborated according to the Safaha method. Two sequences show the preparation of two consecutive Levallois surfaces, of

median = 91 N =43

25 20 15 10 5

300

275

225

175

25

0 125

refit 4:

2 1 1 2 1 1 1 1

75

refit 14:

450

which each produced an endproduct. In between the detachment of each endproduct, proximal striking platform preparation occurred. Again, one of these sequences shows the application of the Safaha method. One sequence finally shows the exploitation of two consecutive endproducts from the same surface and with no intermediate striking platform preparation. Of the sequences involving endproducts and preparatory elements, three show a renewed Levallois surface preparation after production of an endproduct. Three other sequences consist of two Levallois endproducts refitted onto one another. In between their detachment, the Levallois surface was renewed and a new striking platform was prepared. Finally two sequences show the exploitation of two endproducts from the same Levallois surface. In one case, the two endproducts are juxtaposed, in the other they are consecutive. In neither of the sequences was a new striking platform prepared for the second Levallois endproduct.

percentage

refit 11:

350

Figure 4.16 - Weight classes of selected nodules

Table 4.5 - Duration of reduction sequences for complete sequences Levallois surfaces endproducts/surface Levallois surfaces endproducts/surface Levallois surfaces endproducts/surface Levallois surfaces endproducts/surface

250

weight classes

The almost complete reconstructions (fig. 4.14) show that the same method is always maintained throughout one reduction sequence. In table 4.5 the number of prepared Levallois surfaces and Levallois flakes for each surface are listed. None of the complete refits evidences a very intensive reduction: three out of four had only one Levallois surface prepared. For incomplete sequences, the exact number of Levallois surfaces cannot be determined. Some of them, however, show that at least more than one Levallois surface was prepared.

refit 19:

150

50

0

Table 4.4 - Various types of reconstructed reduction sequences Type of refit Complete sequence Core with one or more endproducts Endproduct(s) with preparatory elements Preparatory element sequence

median = 180 N = 58

30

weight classes

Figure 4.17 – Weight classes in g of Levallois cores in early reduction state

92

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

Table 4.6 - Quantitative data on Levallois transport according to the present core sample (actual number comprise fragments)

60 median = 41,5 N = 60

percentage

50

Classical % 57 29 14 100

primary flakes unmodified secondary modified secondary Total

40 30 20

actual number of cores efficiency ratio Estimated Number of Endproducts Actual Number of Endproducts

10

300

175

225

175

125

75

25

0

weight classes Figure 4.18 - Weight distribution of the classical Levallois cores

138 1.8 248 268

Sample characteristics of cores and endproducts Various attributes of cores and endproducts were analysed. In this analysis, no distinction was made between the various Levallois methods represented, for reasons explained above.

According to the present refitting evidence, it appears that Levallois reductions generally did not last very long. Never was an example encountered where more than two consecutive Levallois surfaces were elaborated. The sample of 43 Levallois cores in an early reduction state has a median of 0.091 kg (fig. 4.17). It is interesting to observe that the skewed weight distribution has a long tail to the right. The percentage of the cores in preparatory state is quite large. The reduction ratio for such cores is 51 percent. To implement the Levallois concept on selected nodules (fig. 4.16), therefore, implies the removal of about half of the initial volume. The reduction ratio for abandoned Levallois cores (fig. 4.18) (calculated on the base of the median for Levallois cores in an early reduction state) is 46 percent. This means that the limited productivity of such cores required the removal of yet again about half of the available core volume. In table 4.6 the efficiency ratio- and ENE-data are brought together. An efficiency ratio of 1.8 is low. This confirms the refitting data. The expected number of Levallois endproducts is very similar to the actual number. This suggests that either no transportation of Levallois products from the site occurred or that both cores and endproducts were taken away in similar proportions.

Table 4.7- Attribute analysis summary for 59 cores Under surface Disposition of scars distal proximal left right

Ns 153 284 60 73

% 27 50 11 13

Total number of scars

Nc 59

M 9.7

Angles distal proximal left right

Nc 56 57 54 54

M 49.5 59.5 52.4 55.9

Upper surface

σ 3.9 σ 13.5 8.8 17.8 21.5

Scars distal left right

Nc 57 54 55

M 4.9 1.4 1.5

σ 1.6 1.6 1.5

Dimensions length width thickness elongation flattening

Nc 59 59 59 59 59

M 51.5 46.7 18.5 1.1 2.6

σ 13.8 10.1 5.6 0.2 0.6

Ns=number of scars; Nc=number of artefacts analysed.

93

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________ of the butts are prepared. Among those, FKDSHDX GH JHQGDUPH butts are rare. The flakes are rather small. In striking contrast to the cores, however, their elongation is considerable. An elongation index of 1.6 is high indeed for a classical flake method. This shows that endproduct elongation is to a large extent independent from core elongation. In the endproducts sample, there are a number of blanks which in a metric sense might be called blades as their elongation index is greater than 2. In other respects, however, they are not at all different from classical flakes. As a matter of fact, these "blades" often display plain butts. They have acquired a large elongation through a considerable reduction of width and thickness. It has been argued (Van Peer 1992) that such production is entirely possible within a classical method for flakes, especially if the pattern of preparation shows a tendency towards bidirectionality, which is the case here. This mechanism of increasing endproduct elongation, however, has nothing to do with real blade production. More significant adaptations of the core volume organisation are required for that. Close to 60 percent of the flakes are primary endproducts. This high amount is well in accordance with the refitting data on the basis of which a production of one endproduct per Levallois surface was put forward as the most common procedure at Nazlet Khater 2.

Table 4.8 - Attribute analysis summary for endproducts. 145 endproducts analysed. Disposition of scars distal proximal left right Total number of scars

N 307 386 235 223 Ne 145

Butts plain dihedral

*+,-&.,/102.43#.5202,26)71.

facetted convex facetted straight other

Reduction states primary secondary-unmodified secondary-modified Dimensions length width thickness elongation flattening

Ne 129 109 128 109 109

% 27 34 20 19

M 7.1

σ 2.5

Nb 9 12 20 86 6 12

% 6 8 14 59 4 8

Ne 82 42 21

% 57 29 14

M 45.7 30.5 4.8 1.6 7.1

σ 10.4 9.4 1.9 0.5 2.5

Conclusion Ns=number of scars; Ne=number of artefacts analysed; Nb=number of butts.

The Levallois reduction strategy, as applied in the Nazlet Khater 2 assemblage, is entirely within the limits of the definitions of the Levallois concept. In terms of efficiency, its potential appears rather limited. Two Levallois surfaces per reduction seem to be the maximal figure. This limited productivity is apparently not caused by raw material constraints. In the Nazlet Khater area, larger raw material volumes were present, if needed. Clearly, however, a selection in favour of smaller pebbles took place. Notwithstanding this relative lack of efficiency, the Levallois artefacts themselves, cores and endproducts, are quite "delicate". Both core and endproducts samples are quite homogeneous in various aspects up to a point where one might use the term standardisation.

The Nazlet Khater 2 Levallois cores have relatively few scars on their under surfaces, most of which are situated in the distal and, especially, the proximal sector. The low lateral scar percentages point to the fact that such sectors remained frequently unprepared. They display the characteristic opposition between distal and proximal angle values. Lateral angles are quite variable as compared to the former. The emphasis of upper surface preparation is on the distal sector. Mean numbers of scars for lateral sectors are low. The cores are small; their elongation is mostly very limited. According to the criteria of Crew (1975a) the pattern of preparation for endproducts is radial (centripetal). The distal scar percentage is considerably higher than that of lateral sectors. In this sense, the information of core upper surfaces is confirmed. The majority

94

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

2WKHUUHGXFWLRQVWUDWHJLHV

The last category is the most difficult to distinguish from Levallois cores. Their asymmetric opposition of surfaces and the functional opposition between them are almost similar to that of Levallois cores. However, the under surfaces of such cores lack the characteristic differences between the angles of the different sectors. Special striking platform treatment is not observed. Their upper surfaces display a number of radially disposed negatives of about the same amplitude and with quite pronounced ridges separating them. All of these discoidal cores produced small but sometimes rather wide flakes. According to the core striking platform areas, few of them have prepared butts.

Kombewa cores This reduction strategy was originally described by Owen (1938) and later by Tixier, Inizan & Roche (1980: 55). It is a very simple strategy where the ventral face of a thick flake is used as a striking surface for one flake. The striking platform is prepared on the former dorsal face. A reconstructed sequence shows the presence of a small flake struck at the ventral face of a thick flake, but from the distal end of the latter. A striking platform was carefully prepared for that flake. The ventral face has undergone some minor preparation prior to the detachment of the Kombewa flake and in that sense this does not represent a true Kombewa strategy. As a matter of fact, the relationship with the Levallois strategy is very close indeed. Moreover, in a typical Kombewa strategy, the flake is struck in the area where the bulb of percussion of the flake-core is situated.

Single/multiple platform cores The single platform type forms the majority of this group. Only three cores have two striking platforms. One of these is an opposed platform core, the others have adjacent platforms. Cores of these types are elaborated on rather thick pebbles, although they are not especially large when compared to Levallois pebbles. Striking platforms are created by detaching one large, or at most a few flakes, from one of the pebble edges. Careful platform preparations are not observed. Most of the cores produced flakes, sometimes quite short. Only two cores evidence the production of more or less elongated blanks. These cores, as most of the others, are in a very early reduction stage. The laminar products removed still preserve a considerable amount of cortex on their dorsal surfaces and are not blades in a technical sense. A true blade production strategy is not attested at Nazlet Khater 2. Blades (tab. 4.1) are very rare. Even these rare examples can hardly be given the qualification "blades". Many of them have large patches of cortex and irregular dispositions of dorsal ridges. Prepared butts are extremely rare, and seem to be accidents that occurred during unsystematic flake production.

Discoidal cores Few cores of this type are present, and among them, various types of volume organisation can be distinguished. However, they all show important differences with the volumetric organisation that is characteristic of Levallois cores. A first category (three cores) comprises cores of which the striking platform surface is much thinner than the debitage surface. Flakes are struck from the latter under a very large angle to the plane of intersection between the two core faces. These cores are easy to distinguish from Levallois cores since their upper surface morphology is completely different. To the second group belong two cores of which the striking platform surface is very thick and the debitage surface almost flat. The latter lacks the characteristic convexities of a Levallois core. The angle difference between distal and proximal surface sectors usually observed in Levallois cores is lacking here. Moreover, none of the striking platform areas evidence a more intensified preparation as compared to others. In the third category, comprising only one core, the functional opposition between the two core surfaces has disappeared. They both serve as striking platforms as well as debitage surfaces at the same time.

7\SRORJ\ Retouched tools are very rare, in essential count only twenty-three. Two of the lateralised flakes (Van Peer 1991a) here have the retouch on the left flake edge at the proximal end near the interface with the

95

P.M. Vermeersch, P. Van Peer & E. Paulissen _____________________________________________________________________________________________

butt, the other is retouched at the right edge, which is the normal situation observed in other Nile Valley assemblages.

Next to denticulates and notches, the most common tool types for the Nile Valley Middle Palaeolithic, quite a few Upper Palaeolithic types are present. The end notched piece is elaborated on a thick cortical flake. Apart from the lateralised Levallois flakes, six other tools are on a Levallois blank.

discussions about site function will have to take into account this assumption. The specific structure of the Nazlet Khater 2 assemblage indicates that lithic raw material processing was one of the site’s primary functions. Artefact types related to various reduction stages are present in proportions that suggest that the complete reduction chain took place at this spot. Next to processing, raw material was also extracted or at least collected at or nearby the site area. The observed weight distribution of selected nodules points to the choice of flat and relatively light chert nodules from the local gravel deposits. Middle Palaeolithic extraction features have now been recognised at various sites in the Nile Valley. Extraction features at Nazlet Khater 2, if they were present at all, were undoubtedly destroyed by the wadi activity that redeposited the archaeological material. The very low proportion of tools might be seen as a confirmation of the site’s raw material processing function.

6LWHIXQFWLRQ

&RQFOXVLRQ

From the field data it can be deduced that only very small surfaces have been excavated. The archaeological material that has been collected were no longer in its original position but has been transported by wadi activity and slope evolution processes or has been dumped by humans. However, the artefacts have not been transported over long distances. Indeed, the material is in fresh state of preservation and refitting between such artefacts has been performed. The wide distribution of the material over the hill surface may be an indication that several diachronous occupations are present. That this is the case is substantiated by the presence of a few rolled artefacts wherever the archaeological material has been recovered in slope deposits, and by the presence of numerous blanks and bladelets on the eastern hill slope. This material, however, has been excluded from this analysis. Even so, the question remains if the present artefact collection belongs to a single site. Field arguments can not be advanced but from the technological point the assemblage belongs to a single unit. We can only guess about the duration of site activity. All studied material has thus been considered as belonging to a single site. All

In the western sector of the hill, local deposits, superimposed upon a Nile gravel terrace, are well developed and preserved. They take the shape of shallow fluviatile beds. The Middle Palaeolithic artefacts are always imbedded in these local deposits or have an even more derived position in the loose slope deposits on top of the sector. Everywhere else artefacts are included in local deposits, which came down along the hill flank, probably by slope evolution. The well-developed local deposits with their important artefact content can be interpreted as the remains of wadi beds. The artefacts clearly attest an extraction activity on the site. Here again, we face the problem that no raw material is available on the spot itself, even when the area is rich in chert cobbles. This situation forces us to presume that close to the hill or on Boulder Hill raw material was present in local terrace remains, which we did not discover by due to our restricted excavations. The huge chert artefact content of the ‘wadi beds’ could, in the light of the existence of large and extensive Middle Palaeolithic extraction pits at other sites (this volume), also be interpreted as the remains of

Table 4.9 - Typological list of Nazlet Khater 2 Lateralised Levallois flakes single straight side scraper concave transversal scraper burin truncated flake/blade notch denticulate retouched flakes end notched piece Total

N 3 1 1 2 4 9 5 8 1 34

96

4 --- MIDDLE PALAEOLITHIC CHERT QUARRYING AT NAZLET KHATER 2 _____________________________________________________________________________________________

extraction pit fills. The problem with this interpretation is that there are no visible raw materials to be extracted from those areas. We therefore prefer to interpret the ‘wadi beds’ with their artefact content as real water laid deposits, reworking a nearby extraction dump. Such field data interpretation implies that the Middle Palaeolithic occupation is probably contemporary with the deposit of the local gravels. As these gravels occur on top of the hill and as their stratification is fully independent from the present hill topography, they must have been deposited before the Boulder Hill became an isolated hill through relief inversion. Upper Palaeolithic graves, dated around 33,000 BP, have been dug on the hilltop and flank. Both are well preserved, although that on the hill flank is less so, and we interpret this as an indication that only minor erosion took place since their burial. The Middle Palaeolithic occupation of Nazlet Khater 2 was thus contemporaneous or was followed by a very important fluviatile erosion activity, responsible for the present hill topography with its isolation from the limestone cliffs. The present hill topography was already in place around 33 ka BP, when Upper Palaeolithic people prepared the

two burials. The erosion activity was much more intense than that of the early Holocene. When viewed against the general climatic evolution of Northeast Africa (Wendorf & Schild 1992) we may presume that the erosion most likely took place in Marine Isotope Stage 5. The 14C-date of > 35700 BP is not in contradiction to such a dating. The reason why people settled on this locality is most probably related to the presence, in the local deposits, of large amounts of chert cobbles of good knapping quality. The original local deposit disappeared during the intense denudation following human occupation. The assemblage can be viewed as related to a chert extraction activity by Middle Palaeolithic humans (Vermeersch, Paulissen & Van Peer 1990a). Unfortunately, no extraction structures, nor settlement features have been discovered by our excavations. On the basis of the technology and the typology of the archaeological material the site can be attributed to the Lower Nile Valley Complex as defined by Van Peer & Vermeersch (1990a) and Van Peer (1998), who have used the Nazlet Khater-2 assemblage as the typeassemblage of that group.

97