关键词: 石制品;拼合研究;埋藏学;技术行为;花石浪龙牙洞

Abstract: A systematic refitting analysis was carried out in 1999 on the stone artefacts excavated from the Longyadong cave, Luonan Basin, China. 18608 lithic artefacts excavated from the inner cave in 1995 were selected as the refitting unit. It included two strata: (i) Layer 10, the disturbed layer, and ( ii) the upper part of Layer 4, a red clay soil layer, the principal sediment at the cave.
Out of a total of 18 499 stone artefacts ( except the manuports, hammer stones, anvil stones, and some of the burnt stones) , 94 refitted groups were identified. These sets included 212 elements, made up of 27 cores, 67 flakes, 37 broken flakes, 71 chunks, one small piece of flaking debris, and nine retouched flakes. The raw materials range from various coloured quartzite ( cream, red, and dark grey colour etc. ), quartz, sandstone, flint, and igneous rock. The refitting rate is about 1115%.
There are two patterns of refitting in current work. The first is where flake(s) or chunk(s) can be refitted to a core, or flakes can be refitted together and represent a succession of removals but where the core cannot be identified. The second pattern identified is those artefacts that have broken either during the manufacturing process or through post2depositional disturbance. The former pattern will be referred to as conjoins, the latter as joins. Joins include all broken artefacts whether broken from the percussion point or medial snaps.
Seventy2eight groups, or 82198% in total of 94 refitted groups, consist of two elements, and include conjoined core with flake (s), conjoined flakes, or joined flakes and chunks. These refits usually possessed a cortical platform. Eleven groups ( 1117% ) consist of three refitted elements. Four groups ( 4126% ) consist of four refitted pieces. The largest refitted group conjoined together was made up of five elements. This comprised of 1106% out of the 94 refitted groups.
Spatial Distribution of Refitted Lithic Artefacts
For the core and flake ( s) or chunk (s) conjoined sequences, the maximum and minimum horizontal separation distance is relatively narrow compared to the other refitted types. The horizontal rectilinear distance between conjoining flakes ( chunks) is relatively longer than the other types. The joined broken artefacts have a lower mean distance suggesting that these are relatively more undisturbed than the other groups. The retouched type reveals the opposite pattern. The rectilinear distance between retouched flakes is the largest with the nearest rectilinear distance 9913cm, which is significantly greater than other refitted types.
The vertical separations of retouched flakes are significantly different and are separated over larger distances. Their mean vertical separation is nearly twice that of the other refitted types. Most of refitted artefacts are simple conjoins or joins of two artefacts. They have a vertical separation from one to 20 cm. However, some refitted groups are separated by over 20 cm, and are mostly associated with retouched flakes.
Taphonomic Issues
The results from experiments have shown that lithic scatter patterns are influenced by a number of factors, including the method of manufacture, the movements of the knapper, animals inside the cave, the amount of artefacts, the habits of the knapper ( e. g. , the height of knapper. s hand above the ground standing or sitting, etc. ), trampling, and the strategies of tool use such as re2use of former blanks or tool re2sharpening. Because of the possible range of both natural and behavioural influences on the distribution of artefacts within the cave, alternative explanations must be considered. Technological explanations cannot be the only influences operating on horizontal distributions of refitted artefacts. As we have refitted that some refitted artefacts lay up to 6133 m apart, while some other sets lay close together.
There are 87 groups, in total of 94 refitted groups that have been measured in situ while the other artefacts came from the disturbed Layer 10. Nine refitted groups ( 912%) of 87 groups lay more than 4 m apart. The percentage of different refitted types is composed of two conjoining groups ( 10% ) from a total of 20 core2flake groups, two joining groups ( 8133%) in a total of 24 groups of broken pieces, four groups ( 10181% ) in total of 37 groups of conjoining flakes, and one group ( 16167% ) in total of six groups of conjoining retouched flakes. The results suggest that serious artefact disturbance has taken place in the cave. However, the rectilinear distance of some refitted groups lay less than 4 mapart, their horizontal distribution pattern can be interpreted as representing part of a hominid knapping event, since it is clear that the inner cave wall configuration separates some of the refitted artefacts, and argues against simple natural process.
The horizontal distribution of related broken flakes or chunks, particularly, the elements that are the closest together, suggests that they lie near their original positions. Fifteen groups of refitted broken flakes are broken or separated by distances of less than one metre. In these cases their distribution in almost certainly the result of primary knapping. Here it is argued that they more likely reflect behavioural patterns rather than natural disturbance.
Analysis of the distribution of the broken pieces also provides important information about post2 depositional events. Several pieces lay close together without any gap between them indicative of trampling or some other external pressure such as pressure caused by the weight of the cave deposits.
The stratigraphy and the TL dating suggest that the deposit has been built up over a period of tens of thousands of years. Given that the thickness of the analytical unit is about 70 ) 90cm and built up nearly 100 kyr between 35616 ? 1718kyr and 27319 ? 1317kyr, it is likely that the lithic artefacts were exposed on the surface for a substantial period of time. No stratigraphic unconformity or sediment gap within the upper Layer 4 has been identified which would indicate that the artefacts were exposed for a considerable time before deposits covered them. The artefacts therefore have undergone a process of hominid or non2human animal trampling before they were buried.
Careful analysis of the sediments of the Longyadong cave shows that there is no evidence for significant fluvial action. The presence of intact, very thin 1 ) 2mm thick deposits interleaved with a number of 2 ) 3mm trampling strata in the inner cave, suggests no major disturbances. Fluvial action does not appear to have been a major influence on the artefacts distribution patterning.
Overall the evidence suggeststhat artefacts separated by large distance can be interpreted in terms of hominid behaviour. These appear to have been selected for use or transport to another location. Conversely, the joined broken pieces with distances less than 10cm suggested that they were formed during in situ lithic manufacture or broken through trampling.
The study and interpretation of the vertical movement of lithic artefacts in the Longyadong cave is relatively complex when compared to the horizontal distributions. The results show that most of the refitted artefacts at the cave underwent vertical movements from zero to 20 cm. However, the largest differential separation are those found in the conjoined core and flake(s) groups of up to 48 cm, while conjoined retouched pieces were found to be 40 cm apart. There are four refitted groups ( 416% ) of the 87 refitted groups with vertical displacement of over 30 cm. Ten refitted groups ( 11149%) have vertical displacements of 20 to 30 cm. Twenty2three refitted groups ( 26144%) have vertical displacement of 10 to 20cm, while the other 50 groups ( 57147%) with have a vertical separation of less than 10 cm. The results suggest that conjoined groups, particularly the conjoined retouched flakes are significantly different from the other refitted groups. Their mean vertical separation is nearly twice that of the other refitted types.
It is impossible to indicatewhich agent was responsible for the vertical displacement of artefacts in the Longyadong cave. Some experiments and excavations of archaeological sites with sandy sediment matrix show that when artefacts were deposited, the heavier pieces tend to penetrate further. However, at the Longyadong cave, the relationship is reversed to some extent. This may be related to the surface area of lithic artefacts and the differing sediments in which these were deposited.
Only the heavier conjoined core and flake( s) group is displaced toward the lower levels. The other three refitted groups challenge the conclusion that heavier artefacts tended to travel further down into the deposits over time. Generally, the results show that the degree of vertical movement of refitted groups does not support the previous findings at the other sites with a fine sand matrix.
In an assessment of the principal factors, biological activities may be ruled out as a major agent. The cave site deposits have clear stratigraphic ordering with 2 ) 3mm thick strata preserved in the unit. Although some plant roots were found, the matrix was nearly horizontal, and the macro level biological activities do not seem to have played an important role in displacing artefacts.
A major factor in the formation processes is the alternate wetting and drying of the deposits that probably played a significant role in vertical movement of refitted artefacts. The clay soils are liable to wetting and drying, with a formation of cracks allowing smaller flakes to penetrate into lower layers more easily than larger, heavier artefacts. Rainfall slowly percolates from the cave roof in droplets wetting the sediments and allowing it to expand. When dry, it contracts and cracks. The artefacts located in the softer soil inside the cave in northwest corner have a deeper vertical distribution than other areas. The influence of differential clay moisture content appears to have a considerable effect on vertical displacement of artefacts. The cave deposits are loess and there is no evidence that shows regular flooding through the cave particularly in the Layer 4.
Trampling may have played a major role in vertical displacement, since the cave was occupied by early hominids and non2human animals for a very long period. It was quite probable that lithic artefacts had been exposed on the surface for a long time before they were buried. The density of artefacts in the central parts of cave was not as high as the surrounding areas and the fact that the soil here was more compact appears to have led to a greater displacement of artefacts due to trampling.
It appears that the effects of artefact weight, sedimentary environment, hominid trampling, other bio2disturbances, and soil properties have had a significant influence on artefact distribution inside the cave. Hominid behaviour is one factor but other site formation factors have been identified. The wetting and drying of sediment has played an essential role in the dispersal of lithic artefacts in the cave.
Technological Issues
Experimental replication of flint2knapping shows that the d bitage are scattered within 50 cm around the knapper. It has also provided information on how much dispersion of flaking debris can be expected in situ. Although the exact method of lithic reduction cannot be identified, the characteristics of special flakes show that the former occupants of the cave used anvil2chipping, bi2polar, and hard hammer percussion techniques for flake production. Based on experiments, it was found that anvil2chipping technique was used to detach flakes. Quartzite river cobbles were used and when struck scattered d bitage up to a maximum of 150 cm from the knapping site. Ninety percent of shatter was distributed within 70 cm of the knapper. It contrasts with the distribution of refitted artefacts in the cave, where the mean distance of all refitted groups except the broken flake type is greater than 150 cm. If 50 cm is used as the common criterion for hard hammer percussion, then there are only 12 refitted groups ( 13179%) that fall into this range. When the common criterion of 70 cm of rectilinear distance is used for anvil2chipping technique, only 16 groups ( 18139%) fall within this range.
It has been demonstrated that there is a close relationship between the position of knapper and the resultant d bitage scatter. That is, the further away from the ground surface the knapper is positioned, the greater dispersal of the d bitage, with some individual flakes travelling up to 4 m away. A more concentrated pattern is likely to result when the knapper is seated. This means that the rectilinear distance of dispersion of refitted stone artefacts might not be useful for identifying the specific knapping technique in the Longyadong cave.
Unfortunately the refitted groups cannot be used for studying specific lithic reduction technology, although other patterns can be identified through an analysis of refitting artefact groups. For example the distance between core and flake(s) or chunk(s) show a narrower range of dispersal when compared to other types. This implies that under some experimental circumstances, when the first flake is removed from a core in a sitting position, d bitage scatters in a fan where the distance between core and d bitage equals that of the radius of the fan. However, the conjoined results of flake(s) or chunk (s) show a different pattern. The distance between each set is variable. At face value the interpretation could be that the conjoining flakes were picked up and selected for re2sharpening or directly used as a tool after they were detached. It suggests that the flakes underwent post2knapping removal and discard. Moreover, since the conjoined retouched flakes were chosen for re2sharpening, the rectilinear distance between them suggests that even after retouching, the tools were taken away from the cave. It seems that the knapper did not work at same location during the knapping sequence. They appear to have moved from one area to another. Such activity would transport the artefacts from their original location to another work area.
However, the groups of joined broken flakes differ from retouched groups. They appear to have formed in two ways. One is the result of knapping processes, where the pieces were reduced by percussion processes or the detached flakes hit another object in percussion processes to made the flake break. The second is by post2depositional processes, such as trampling or striking another rock. It is not hard to imagine that the rectilinear distance for broken pieces remained short compared to the other refitted types unless they were chosen for further retouching or removed during heavy disturbance. The experiments and the analysis of the joined artefacts groups support this hypothesis.

Key words: Palaeolithic; Refitting analysis; Taphonomy; Technological behaviour; Longyadong cave