Fig 1.
Map of site locations and vegetation regimes in South Africa.
Fig 2.
A) Tool perimeter is divided by left and right sides based on maximum distal extent. Each edge is then divided into 1% intervals based on edge perimeter between the platform and distal maximum. In this way, each side contains 100 possible locations where damage could occur, regardless of size differences. Dorsal contains damage locations 1–200 and ventral contains damage locations 201–400. B) Photographs are taken from dorsal and ventral views onto a grid, then georeferenced and the outline digitized and edge damage scars traced. Presence or absence of edge damage in 1% intervals is calculated from the polygon shapefiles. C) Edge damage occurrences are aggregated (dashed line) based on the distribution of edge damage around individual tool edges (red histogram bars). The red bars indicate damage on point shown in B that would then be aggregated with other points from the same assemblage or experiment group, and thick black bar indicate platform-adjacent locations.
Fig 3.
A) Quartzite point hafted to wooden dowel. B) Points drying near heat source. C) Point lodged in carcass after being fired. D) Calibrated crossbow setup.
Fig 4.
Left, quartzite with mastic; right, quartzite in slot haft with mastic. White bar is 1cm.
Fig 5.
Butchery experiment completed with quartzite points.
A-B) Handheld tools during field dressing. C) Handheld tool during initial skinning. D) Hafted tool during defleshing.
Fig 6.
A) Cattle preparing for trampling. B) 3 x 3 m grid layout used at each site, with tools laid out alternating dorsal and ventral side-up. C) String used to lay out grid on the ground with four cells highlighted. D) Close-up view of highlighted cells showing tools prior to the trample experiment.
Fig 7.
Rock-tumbler experimental setup.
A) Two drums and digital timer to control tumble duration. B) Drum with water, gravel, quartz stone matrix.
Table 1.
Sample of experimental and archaeological points examined for edge damage.
Fig 8.
Experimental damage distributions (grey) and loess-spline (red) from: A) spear tipped armature use (black line is ironstone loess-spline scaled to right y-axis), B) field dressing butchery activity, C) defleshing butchery activity, D) long-term trampling by animals, E) rock-tumbler for five minutes.
Table 2.
Frequency of motion capture images captured by trampling location.
Fig 9.
Motion camera photos from trail.
A) endangered Humboldt Marten; B) cattle passing through trampling area during day; C) deer passing through trampling area at night; D) cattle lingering in trampling area; E) donkey passing through trampling area; F) Authors BJS and KSB recovering tools and piece plotting in the trampling area at the end of the experiment.
Table 3.
Recovery frequency of all artifacts by trampling location.
Table 4.
Edge damage frequency by trampling location and recovery ‘face up’.
Table 5.
Testing model fitting procedure with known distributions of edge damage.
Fig 10.
Temporally ordered (oldest to youngest) archaeological edge damage distributions (grey) and loess-spline (red) on points from A) Kathu Pan 1, B) the MIS6 layers at PP13B, C) the MIS5 layers at PP13B, D) layers 10–14 at DK1, and E) layers 6–9 at DK1.
Table 6.
Single best-fit general experimental parameter for each archaeological assemblage of points.
Table 7.
Results of model-fitting experimental edge damage distributions to archaeological points.
Parameter percentage of residual sum-of-squares contribution in parentheses.