GI2

The second excavation was carried out in the western part of the site. One square metre was excavated to 36 cm depth with four subsurface spits. Although the upper part was rocky, the base of the excavation was almost pure calcrete powder (Figure 8). Again, no anthropogenic stratigraphy was evident.

Figure 8: View of the GI1-2 excavation. The scale bar is 0.5 m long.

The characteristic and more interesting artefacts found in the 1 m2 excavation are as follows:

Surface

Initially doesn’t look very MSA
Retouch, almost Adze like
Notch
Denticulate
Radial core (short tabular flakes removed)
Convergent flake
Convergent scraper
Recurrent Levallois core

Spit 2

3 Convergent flakes – These are common in MSA assemblages, but are found throughout the MSA. They are products from preferential Levallois.
2 >70mm blades – These also form part of the MSA, but are typically later,
Extensive Retouch including 3 notched pieces and 6 scrapers
4 platform Levallois cores with: 2 blade removals, 2 blade removals, 2 flake removals and 2 blade removals respectively.
2 preferential Levallois cores with a single convergent removal from each
1 recurrent Levallois core with multiple short tabular removals.

Spit 3

3 Platform cores
Retouch, 2 nice notches
Recurrent Levallois core
Scraper
Radial core
Rotated core
Large amounts of retouch

Spit 4

Platform preparation (multiple instances)
Convergent scraper
2 very clear notches
Nice retouched piece.

1.1.1. GI1-3

The third excavation was also carried out in the western part of the site, but 20 m to the east of GI1-2. This hole followed a similar course to GI1-2 but it was stopped at 30 cm depth where the substrate was just powdery calcrete (Figure 9).

Figure 9: View of the GI1-3 excavation. The scale bar is 0.5 m long.

The characteristic and more interesting artefacts found in the 1 m2 excavation are as follows:

Surface

Large platform core, with regular flake removals.
3 Convergent scrapers
Notched piece
2 small scrapers
Platform preparation present on many of the flakes

Spit 2

Very few cores relative to the high number of flakes.
Large notch
Convergent scraper
Denticulate
Notched levallois flake
Possible small levallois preferential core
Small platform core, almost robberg
Very nice scraper retouch on distal and lateral margins.
Convergent scraper (almost a unifacial point)
Adze-like retouch/use

Spit 3

Non-descript flakes.
1 bipolar flake (outils-like)
Retouched piece
Convergent flake
Single large notch on flake
1 artefact with signs of heating (potlid fracture)

Spit 4

A number of non-descript flakes with platform preparation, otherwise nothing special.

1.1.2. GI1-4

No excavation as carried out at this point but an outcrop of banded iron formation with evidence of quarrying was located here (Figures 10 & 11). Flakes had been removed directly from the outcrop.

Figure 10: View of the banded iron formation outcrop    Figure 11: Close-up showing the flaked edge of the
at GI1-4.                                                                        outcrop at GI1-4.

1.1.3. Summary

The excavations at GI1 revealed that the artefact scatter reached variable depth across the site but was deepest in the east. It is clear that the rock here was strongly favoured for artefact production and that the site must have been extensively used over a long period of the Middle Stone Age as a source of stone. The common occurrence of faceted platform flakes supports an ascription to the MSA. Preliminary examination of the material from the excavations has allowed some characterisation of the assemblage. Faceted platform flakes are common, while the cores include Levallois and discoidal cores. Although blades are relatively rare, a number of broken blades were noted. Retouched artefacts are common (perhaps more so than expected in what appears to be a quarry site assemblage), but their identification is complicated by the presence of extensive amounts of edge-damage. Retouch in general seemed more common near the surface and the assemblage included scrapers, notches and adze-like pieces. The abundant edge-damage may well relate to trampling damage during the prehistoric use of the site.

The presence of convergent flakes and long blades (> 70 mm) could indicate a relatively late period of the MSA dating before about 35 000 years ago, but such artefacts can be found on occasion throughout the MSA. The denticulates could suggest a greater age, but there is nothing diagnostic like backed artefacts or Still Bay points. A single unifacial point could support an age around 45 000 years ago, but again this is not a given.

1.2. GI2

This site is a small rock shelter in a low, south-facing cliff on the inside slope of the inselberg (Figure 12). It was noted by Morris (2013) as having very little evidence of prehistoric human occupation – he recorded a single quartz flake. However, examination of the talus slope showed that a number of stone artefacts were present and that evidence of occupation of the shelter should certainly be expected.

Figure 12: View towards the north showing the               Figure 13: View of the entrance of the shelter at GI2.
location of the shelter at GI2.

We collected all the visible artefacts from the talus slope. These included fifteen of quartz and eleven of other materials (banded iron formation and quartzite; Table 1). A small test excavation just outside the mouth of the shelter and the generally rocky nature of the substrate showed that there was no deposit in this area. No attempt was made to map the artefacts as this would not provide any meaningful information. However, some MSA characteristics and the degree of patination/weathering of some artefacts suggest material of that age, while one flake was clearly an LSA flake. The remaining artefacts are too generic in character to assign an age to.

Table 1: Stone artefacts from the GI2 talus slope.

Quartz

Banded Iron Formation

Quartzite

Bipolar core

1

Irregular core

1

Edge-damaged flake

3

4

Blade

1

Flake

10

3

1

Chunk

1

Within the rock shelter a grid of 1 m2 squares was laid out as shown in Figure 14. Six squares were excavated, although two of these were not full square meters (Figures 14 & 15). Excavation followed natural stratigraphy, although this was found to be very difficult due to the heavily compacted and consolidated nature of the deposits. In one layer it was even difficult to chip the deposit out with a spade. The deposit was very shallow towards the rear (north) where bedrock rose to meet the walls of the shelter, but towards the front (south) the deposit reached a maximum depth of about 30 cm (Figure 16). Unfortunately oil had been dumped in the shelter at some point in the past with the result that much of the excavated deposit was contaminated. The oil has dried sufficiently to enable excavation but no doubt this has added to the compaction and consolidation of the deposits. The two westernmost squares (A2 and A3) were found to be free of oil and it is anticipated that the deposits beneath the rocks in the south-western corner will also be clean. This meant that radiocarbon dating samples had to be hand-picked from the deposits in the western side of the shelter. Only two good samples, a charcoal nodule and a wad of grass, could be obtained in this manner.

1.2.1. Stratigraphy

Altogether nine stratigraphic layers were identified, although, as noted above, it was not always easy to separate these during excavation. Table 2 presents a brief description of these layers.

Table 2: Description of identified stratigraphic layers in GI2. The grey shading indicates the way in which these layers have been grouped for the purposes of the analysis that follows. From top to bottom they are Layer 1, 2 and 3.

Layer

Description

1. Surface (Surf)

Loose, powdery surface deposit with modern rubbish and some gravel content.

1. Below Surface (BSu)

Harder soil with gravel and many insect cocoons and still some modern rubbish.

1. Brown Sticks (BSt)

Soft deposit with much vegetation in it. Limited to area of intersection of squares B2, B3, C2 and C3. Still some modern rubbish.

2. Hard Layer (HL)

Compact deposit with rock fragments and still some cocoons. Some areas incredibly hard and some modern rubbish still noted.

2. Soft Grassy (SG)

Small patch of softer grass containing a fair bit of grass. Limited to square C2.

2. Hard Orange Brown (HOB)

Variable hardness, but extremely hard towards the south. Towards the north the orange parts seem to be due to decomposing bedrock.

2. Hard But Softer (HBS)

Variable hardness, paper fragment in C2.

3. Dark Organic (DO)

Soft deposit with small sticks and droppings.

3. Softer Brown (SB)

Generally soft, chocolate brown deposit but some extremely hard patches. Does not reach the north edge of C2.

3. Hard Base (HB)

Hard, lighter brown deposit with much decomposing bedrock.

Figure 14: Schematic plan of the rock shelter showing squares excavated (solid lines) and squares proposed for future excavation (dotted lines). The large rock and smaller rocks behind it are assumed to have been from a collapse that resulted in the arch opening up. They lie on the surface. The roof is dome-shaped with a high point above the southeast corner of square C2 at 2.3 m above the deposit surface.

Figure 15: View towards the west showing the                Figure 16: View of the southern section of square A3,
excavation area on completion of work. Scale bar          the only section not contaminated by oil. Scale bar in
in 10 cm intervals.                                                         10 cm intervals.
1.2.2. Stone artefacts

Flaked stone artefacts were the most common cultural items recovered from the excavation, although there were only 144 in all. Table 3 shows the typological break down per layer. As is generally expected in western South Africa, quartz dominates, but banded iron formation (BIF), cryptocrystalline silica (CCS), silcrete (Silc) and quartzite (Qzite) all also occur in smaller frequencies (Figure 17). What is clear from the graph is that in Layer 1 there is far less diversity in materials with quartz used in more than 90 % of artefacts. The overall numbers of artefacts are also far greater in the lower levels of the deposit. Retouched artefacts are generally expected to be more frequent in deposits pre-dating 2000 years ago, but in the hinterland of the Northern Cape this is not always the case (Beaumont et al. 1995). The CCS backed bladelet in Layer 3 had evidence of mastic adhering to it suggesting that the artefact was once mounted in  handle of some sort (Figure 18).

Table 3: Stone artefacts from GI2.

 

Layer 1

Layer 2

Layer 3

Artefact type

Qtz

BIF

CCS

Silc

Qzite

Qtz

BIF

CCS

Silc

Qzite

Qtz

BIF

CCS

Silc

Qzite

Bipolar core

1

2c

Irregular core

1a

Blade

1

Bladelet

1

5

Flake

10

1

1

12

5

1

5

37

7

2

1

2

Chunk

3

1

1

2

1

1

Chip

5

8

1

2b

12

3

1

1

Endscraper

1

Scraper fragment

1

Backed bladelet

1d

Ochre, n (g)

2 (0.3 g)

-

-

Shiny black rock

1

3

-

Other items

1 water worn qtz pebble
1 qtz crystal
1 tiny water worn qtz pebble
1 small, dark, heavily worn/worked rock fragment, possibly pigment
1 ground stone pendant on a thin slab of mica-rich rock.

 
 

Notes:
a Layer 1 qtz irregular core may also be a hammerstone.
b Layer 2 silcrete is red and might be CCS.
c Layer 3 has one quartz flake with crystal facet cortex.
d Layer 3 CCS backed bladelet has mastic stains.

Figure 17: Stone material frequencies in the three layers of GI2. Quartz has been omitted and comprises the remainder to 100 % in each case.

Figure 18: The dorsal (left) and ventral (right) surface of the CCS backed bladelet with mastic adhering to it (arrowed) from Layer 3. Scale in 5 mm intervals.

A stone pendant was found in Layer 3. Such finds are very unusual but a few have been found in South Africa (Dart 1949; Fitz-Simmons 1926; Heese 1926; Orton & Halkett 2007; Rudner 1953). Unfortunately this one has had its upper and perforated end broken off and lost – it might perhaps even have been reused with the break smoothed off again.

Figure 19: View of the intact end of the stone pendent found in Layer 3. Scale in 5 mm intervals.

1.2.3. Ostrich eggshell fragments

Ostrich eggshell fragments were common in the excavation, although the 120 fragments found only weighed a total of 69.6 g, substantially less than the weight of one whole egg. Figure 20 shows their distribution through the three layers. In Layer 3 one small fragment of OES was found to have been engraved with a single straight line (Figure 21).

Figure 20: Graph showing the distribution of ostrich eggshell fragments in the three layers of GI2.

Figure 21: The engraved ostrich eggshell fragment from Layer 3 of GI2. Scale in 5 mm intervals.

1.2.4. Ostrich eggshell beads

Five OES beads were found, two in each of Layers 1 and 2 and one more in Layer 3. Figure 22 shows the size distribution of these beads as well as their aperture diameters and thicknesses. Following Orton (2012), there are one medium bead, three large beads and one very large bead. This kind of size signature is typical of relatively recent sites dating within, perhaps, the last 1000 years.

Figure 22: Graph showing the ostrich eggshell bead size distribution in the three layers of GI2.

1.2.5. Pottery

Three mineral tempered pot sherds were found in Layer 2. Their statistical data are shown in Table 4. All three are body sherds. The sherd from B2, HL shows evidence of coil manufacture.

Table 4: Statistical data for the three potsherds from Layer 2 of GI2.

Provenience

Weight

Min. thick

Max. thick

Ave. thick

Colour

B2, HL

12.9

5.26

5.96

5.61

dark brown to black

B3, HBS

6

6.91

8.52

7.715

dark brown to black

C3, HBS

0.6

4.48

4.86

4.67

dark brown to black

1.2.6. Other finds

Among the botanical material recovered from the site were two small fragments of cut reed (Figure 23). They are from the genus Phragmites and may have originated from a reed mat or some other item made from reeds.

Figure 23: One of the cut reed fragments from Layer 2 of GI2. Scale in 5 mm intervals.

There were also many fragments of bone and micro-mammal bones from all three layers. The micro-mammal bones may well have originated from a bird of prey roosting on a small ledge in the cave. The bones have yet to be analysed but this will be done for publication purposes once an appropriate specialist is available. It is likely that many are klipspringer (Oreotragus oreotragus) since many hairs matching those on a dead animal found nearby were also found in the deposit. Some droppings that are likely from an animal the size of klipspringer or sheep were found in Layer 1, while some larger droppings were found in all three layers. A piece of porcupine quill was also found in Layer 1

The site also contained fragments of charcoal but these were only retained at times because of the contamination arising from the oil spill.

A variety of modern materials were also found. The vast majority were in layer 1 and included many matches and cigarette butts, some newspaper bearing a closing date in the jobs section of 27th January 1978, a sawed bone and some glass. Layer 2 contained modern material in its uppermost excavation unit only, while in Layer 3 the only modern item found was a small metal cog. In addition, there were two metal poles that had been hammered into the deposit and bedrock. Not all of the modern materials were collected. It is assumed that these items relate to some sort of activity from earlier mining and that this activity likely took place in late 1977.