Ethnographic toolstone heat treatment reveals distinctive motives and patterns of material transformation

Abstract

The heat treatment of rocks to improve their fracture qualities for stone tool production is a key technological innovation of Pleistocene humans. Because the intentionality and technicality of this transformative process are often associated with cognitive prowess, the topic has attracted considerable archaeological research interest worldwide. Yet, with few ethnographic examples, assumptions about archaeological heat treatment are almost always based upon laboratory experiments. Here we report contemporary Konso craft specialists from southern Ethiopia, who represent the last known stone tool makers to regularly heat treat their toolstone. We found that, while material transformations commonly used as indicators of enhanced knappability (e.g., fracture toughness, elastic modulus, improved homogeneity, and newly formed lustre) are discernible in the studied heat-treated toolstones, other qualities of enhanced knappability (e.g., density and hardness) are also informative. Our results highlight the diversity and peculiarity of intentions and mechanisms through which toolstone workability is improved. The motives, techniques, and material transformations involved in Konso toolstone heat treatment afford important insights into various aspects of archaeological heat treatment that are unavailable with experiments alone.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-97207-9.

There is burgeoning research interest in ancient technologies and behaviors that imply complex technical and cognitive processes^1–3. Notwithstanding the lack of consensus on the nature, timing, mode of appearance, and even validity of some of these traits, the accumulating evidence supports the emergence during the African Middle Stone Age (MSA) of distinct technologies and behaviors considered complex. The deliberate heat treatment of stone raw materials to improve their knappability represents one such significant Pleistocene technological innovation^2,3.

The role of thermally altering the microstructure of stone raw materials in improving fracture response during flake production was recognized over half a century ago^4–9. Renewed research on the thermal engineering of toolstones and its implications for early human behavior largely followed its positive identification in the South African MSA². Evidence from multiple prehistoric sites in South Africa, Europe, and Australia is reviving debates on heating techniques, intentionality of the process, time and energy investments, and the bearing of these on the behavioral and cognitive prowess of early toolmakers^2,3,10. By contrast, ethnographic data on toolstone heat treatment mostly derive from secondary descriptions that did not offer archaeologists the opportunity to observe the technology directly^10,11. The few exceptions were limited to cursory descriptions in broader studies^7,11,12. Consequently, assumptions about human intentionality, techniques, processes, outcomes, and implications in archaeological contexts regarding thermally transforming toolstone properties are routinely tested experimentally^2,3,13–17.

Carefully designed experiments represent one of the two major sources of actualistic information about past technologies, the second being ethnoarchaeology^18–21. Experimental tests must often be conducted as part of long-term research programs since aspects of past human behavior involve multiple sources of variability that restrict experimenter control¹⁸. As such, “[e]xcept where constrained by physics, chemistry, or biology, everything humans do we do in more than one recognizably different way”^{22, p.2}. Accordingly, all things being equal, experiments assessing artifact form and fracture attributes, for instance, provide more replicable results²³ than those investigating various aspects of toolstone heat treatment^13–17. Given such limitations, and the impracticality of recreating the complex contexts in which technologies were produced prehistorically, our functional and behavioral inferences can benefit substantially from ethnoarchaeological data.

We present here results of the first detailed ethnoarchaeological investigation of lithic raw material heat treatment by Konso craft specialists in southern Ethiopia who produce and use stone tools as part of their material culture (Supplementary Notes). Our in-depth investigation of this transformative technology, including its contexts and material traces, allows inferences about archaeological heat treatment to be enhanced through careful relational analogy, and the accuracy of results from experimental studies to be validated.

The use of stone scrapers for hide working was widespread in Ethiopia until the end of the 20th century^24–28. Today, it is reported only among the Hadiya, Gamo, and Konso of southern Ethiopia (Fig. 1a). Various aspects of this quickly vanishing tradition have been studied over the past two decades^29–38. While stone scraper and compound adhesive production, hafting onto wooden handles, retouching, use, dehafting, and discard remain largely similar across these groups, the heat treatment of toolstones for enhanced flakeability is unique to Konso hideworkers (Supplementary Notes). Konso stone scrapers are made on heat-treated chalcedony and chert nodules while rock crystal was historically used without heat treating³⁴. Heat treatment is conducted in preparation for subsequent scraper production (Fig. 1b, Supplementary Fig. 1). Scrapers are inserted into the single socket of a straight wooden handle and secured with compound adhesive made by mixing Balanites aegyptiaca resin with black soot powder (Fig. 1b, Supplementary Fig. 2a, b). Hafted scrapers are used to remove excess fat from the inner layers of cowhides and are resharpened, while in the haft, multiple times during their use lives (Supplementary Fig. 2d).

Fig. 1.

Konso villages, quarries, and material culture. (a) Map showing the studied Konso villages and raw material sources. (b) Chalcedony nodules, and wooden handles with hafted hidescrapers.

Most Konso hideworkers are women, another aspect that makes the group unique. The craft tradition is quickly vanishing, with only three Konso hideworkers currently practicing (Supplementary Notes). The present work represents the first detailed study of the heat treatment process and resultant material transformations.

Results

Technique, steps, and motives

Preparations for toolstone heat treatment start with the reduction of nodules, including by splitting relatively large ones via the bipolar technique (Fig. 1, Supplementary Fig. 1b). Between five and eight nodules are prepared for a single heat-treatment session. Heat treatment takes place indoors, in a dedicated shallow pit dug into the ashy edge of the hearth, ~ 20 cm from its center, and measures ~ 23 cm in diameter and ~ 13 cm in depth (Fig. 2a). Once the pit is preheated by the ambient temperature from the hearth fire, the hideworker places the chalcedony nodules in the pit, with attention to their even arrangement. A layer of wool is used for insulation before the pit is covered with a large piece of broken pottery whose concave side faces the pit (Fig. 2b). Hot coals are then placed directly on top of the pottery cover and a mixture of hot ash and fine embers is added as a final cover to ensure a slow and persistent heating process (Fig. 2c).

Fig. 2.

Heat treatment. (a) Chalcedony toolstones are placed in the preheated shallow pit. (b) The pit is then covered with a piece of pottery. (c) Hot ash and embers are placed on top of the pottery cover before the nodules are left to heat up overnight. (d) Fully heat-treated nodules are placed near the hearthstones for a slow cooling.

Hideworkers leave their nodules in the sealed pit for ~ 24–30 h, and up to 48 h in some cases (Fig. 2c). The duration of heating depends on the size and cortex cover of the toolstone as well as the amount and consistency of heat received during the first day, which each hideworker carefully judges. The average ambient and pit temperatures recorded at the start of the heat-treatment process were ~ 108 °C and 148 °C, respectively. Within an hour of placing the larger hot coals on the pottery cover, the pit temperature rose to an average of 223 °C. The temperature continued to slowly rise until it reached a peak of 387 °C two hours into the heating process. The temperature then stayed relatively stable, dropping only to 352 °C at its lowest; replenishing the heat source after 8–12 h brought the pit temperature back up to the higher readings (Supplementary Fig. 3). The final slow drop followed the eventual removal of the heat source. After a successful heat treatment, the pit is left to cool for at least an hour before the toolstones are removed and placed on the non-fire side of the hearthstones to continue to slowly cool, usually overnight (Fig. 2d).

The studied hideworkers emphasized that the heat treatment improved toolstone flakeability through smooth and uninterrupted fractures that allowed predictability and success in removing untruncated products. In addition to fracturing better when knapping, heat-treated toolstones reportedly resulted in a stronger working edge and were shinier and more ‘attractive’³⁴. This “is the only way the rock becomes workable”. By contrast, the hideworkers maintained that non-heated toolstones require greater blow force, resulting in a higher fragmentation incidence and step-terminating/truncated knapping products. Given raw material transport cost, limited availability, and small original toolstone size, heat treatment thus provided the required enhanced workability through better fracture qualities and success rates via less attenuated and straighter fracture propagation.

Material transformations

We analyzed a sample of heat-treated (fullyHT), partially heated (semiHT), and non-heated (unHT) chalcedonies using various methods. The fullyHT toolstones were more homogeneous and lustrous (Fig. 3a). FT-IR measurements (Table 1; Fig. 4a) show statistically significant changes within the observed bands (p < 0.05; two-tailed t-test for equal variance) after heat treatment. The 4545/4469 cm^{− 1} ratio rose from 0.96 to 0.99 in fullyHT samples while standard deviation (SD) dropped from 0.02 to 0.01. There was similarly a change in the 4535/4450 cm^− 1 ratio, rising from 0.97 in unHT samples to 0.99 in fullyHT ones, while SD again dropped from 0.02 to 0.01, possibly indicating material homogenization after heat treatment. Unlike unHT and semiHT samples, freshly flaked fullyHT samples appear lustrous (Fig. 3a), reflecting the mobilization of water indicated by IR-spectra (Table 1, Supplementary Table 1).

Fig. 3.

Raw materials. (a) Macroscopic aspect of the Konso rhyolite and chalcedonies. The chalcedony samples represent different heat treatment stages of a single nodule split into parts for our controlled study. (b) Petrographic thin-section photomicrographs showing (top left) rhyolite microstructure with phenocrysts of quartz in fined-grained matrix and chalcedony crystallized into a void, XPL; (top right) magmatically corroded quartz in the same rhyolite, XPL; (bottom left) chalcedony in XPL; (bottom right) LF chalcedony indicated by the orange hue in XPL with 550 nm retardation plate.

Table 1.

Results of the performed infrared- (FT-IR), mechanical-, and density measurements.

FT-IR ratios						Density			Mechanical tests
Samples	4545/4469	SD	4535/4450	SD	N.meas.	[g.cm-3]	SD	N.meas.	HM[MPa]	SD	N.meas.	EIT[MPa]	SD	N.meas.		KIC[MPa.m− 1/2]	SD		N.meas.
unHT	0.96	0.02	0.97	0.02	16	2.574	0.004	23	3960.00	47.91	19	66.37	1.18	19		2.23	0.21		28
semiHT	-	-	-	-	-	2.570	0.005	16	3919.71	48.22	30	65.36	1.33	16		1.45	0.12		12
fullyHT	0.99	0.01	0.99	0.01	11	2.559	0.005	28	4284.04	96.92	28	74.00	2.04	23		1.04	0.06		33

Martens hardness (HM), indentation elastic modulus (E_IT), and fracture toughness (K_IC) for each type of measurement. The number of samples measured by each test is provided in the respective ‘N.meas.’ columns.

Fig. 4.

Change in properties. (a) Absorbance changes, as determined by FT-IR, in unHT and fullyHT chalcedony samples. (b) Change in density between unHT, semiHT and fullyHT chalcedony samples. (c–e) surface texture comparisons of unHT and fullyHT samples: fractal complexity (Asfc), mean depth of furrows (MeDF), and core roughness depth (Sk).

Mean density values and surface roughness decreased considerably with heat treatment. Density for unHT vs. semiHT, and semiHT vs. fullyHT samples dropped significantly (p < 0.05) from a mean of 2.574 to 2.570 and 2.559 g.cm^− 3, respectively (Table 1; Fig. 4b). Similarly, a considerable drop in surface roughness in the fullyHT samples compared to unHT ones is shown by the results of our 3D surface texture analysis (Supplementary Tables 2–5; Supplementary Fig. 4). The most noticeable changes included in the commonly used rugosimetric parameters (ISO 25178, Fig. 4c-e, Supplementary Table 6) of peak height and curvature, pit depth, core roughness, inverse areal material ratio, texture aspect-, and developed interfacial area ratios, material and void volumes, and area roughness measures using the watershed algorithm. In addition, the SSFA parameter for area-scale fractal complexity (Asfc), commonly used in dental microwear texture analyses, also shows a significant drop in fullyHT samples compared to unHT ones, indicating a dramatic reduction in surface roughness and improved heterogeneity after heat treatment (Fig. 4c, d). Combining the ISO and SSFA parameters, and for the first time the three new furrow parameters, improves our understanding and objective quantification of differences in surface textures between the two toolstone groups. Most rugosimetric parameters are usually strongly correlated and require further dimensionality reduction using multivariate analyses (e.g., PCA) to identify which variables are the most strongly correlated. Given the extremely high differences between the fullyHT and unHT specimens, with no range overlap for most parameters, such analyses are not deemed necessary for this study (Supplementary Table 7 and Supplementary Fig. 5).

FullyHT toolstones varied from semiHT and unHT ones in their hardness and elasticity, while their fracture toughness was reduced, and their surface became less resistant to marring. Martens hardness (HM) changed little, dropping from 3960 to 3920 MPa in semiHT samples compared to unHT ones, but there was a significant increase to 4284 MPa in fullyHT samples (Table 1; Fig. 5a). Similarly, indentation Young’s (elastic) modulus (E_IT) remained practically unchanged when unHT samples were semiHT (66.37 vs. 65.36 MPa) whereas there was a significant increase in fullyHT samples (74 MPa; Table 1; Fig. 5a). SD was largest in fullyHT samples in both HM and E_IT values.

Fig. 5.

Change in properties. (a) Change in Martens hardness (HM) and elastic modulus (E_IT) and (b) in fracture toughness (K_IC) between unheated (unHT), semi-heated (semiHT) and fully heated (fullyHT) samples of Konso chalcedony. (c) Normal probability plot of fracture toughness of unHT, semiHT and fullyHT samples indicates no change in fracture behavior apart from the drop of K_IC. (d) Post-scratch depth in unHT, semiHT and fullyHT samples shows lower resistance of HT materials to scratching.

Fracture toughness (K_IC) showed a steady and significant drop from 2.23 to 1.45 (unHT vs. semiHT samples) to 1.04 MPa.m^1/2 in fullyHT samples (Table 1; Fig. 5b). This time, there was a decrease of SD upon heat treatment, further implying homogenization of the material. When plotted in a cumulative graph (Fig. 5c), K_IC shows similar sloping trends in unHT, semiHT and fullyHT samples. This indicates that the fracture behavior of all chalcedonies is similar, except for lower K_IC (i.e., improved knappability) in fullyHT samples.

Post-scratching depth test was conducted to assess the effects of heat treatment on the mechanical properties of the Konso stone tools. Figure 5d summarizes the test results, with the initial smooth part of the curves illustrating a clean scratch where the depth is proportional to the applied load. The corrugated part of the curves represents the locations where the indenter tip caused severe damage to the flake surface with spikes indicating decreased−and valleys increased−depth. Both semiHT and fullyHT samples are relatively susceptible to damage by scratching compared to unHT samples; fullyHT samples show the least resistance (Supplementary Fig. 4).

Discussion

Increased hardness was observed in Konso heat-treated flints, which is considered to cause brittleness and lead to sharper tool edges^39,40. The FT-IR measurements of the Konso chalcedony show changes also observed on experimentally heated flints from Germany and Jurassic cherts from Czechia^41,42. The changes in 4545/4469 and 4535/4450 cm^− 1 ratios (Fig. 4a) indicate that temperature-induced water evaporation in Konso chalcedony causes transgranular fracture, the formation of fluid inclusions because of water mobilization, or both^6,39,43. This is reflected in the lustre on freshly flaked surfaces of fullyHT samples (Fig. 3a). The loss of water through evaporation probably leads to a progressive drop in density from unHT to fullyHT chalcedony samples. Although a reverse (increasing) trend was observed in experimentally heat-treated Jurassic cherts⁴⁴, decreasing density was also observed in other case studies^6,45, indicating that density can change in multiple ways upon the heat treatment of cherts. The increased HM of Konso chalcedonies upon HT (Fig. 5a) is unusual, as hardness did not change in other studies^39,43,46 suggesting that there is no uniform hardness change in cherts when heat treated.

Konso heat-treated chalcedony presented changes that improved knappability, such as increased elasticity and decreased fracture toughness. Increase in elasticity (E) is a common phenomenon in heat-treated cherts^6,39,47, although it is shown to drop in coarse-grained silcretes upon heat treatment⁴⁸. Together with decreasing fracture toughness (K_IC) (Fig. 5b, c), it is considered the predominant factor in improving knappability^6,49. In practice, the knapper can obtain longer and finer flakes as the fracture is more regular and predictable, and requires lower knapping force. The high K_IC of unHT Konso chalcedony samples (2.23 MPa·cm^1/2) explains why the raw material is difficult to knap without heat treatment. In addition, the lower SD values of their K_IC suggest the overall homogeneity of HT samples, as independently shown by the significant changes in the rugosimetric parameters. The latter notably indicate considerable changes in surface textures related to heating, with reduced surface roughness characterized by overall much shallower features in fullyHT samples. Such increased homogeneity in surface texture means faster and more predictable fracture propagation during knapping. Overall similar results were also previously observed, albeit using different methods, in silcrete heat treatment^48,50. Lower resistance to scratch in fullyHT chalcedonies means higher wear^51–53, which is an obvious trade-off with the improved knappability and smooth working edge.

Details about the techniques and processes of heat treatment of raw materials have proved difficult to readily decipher from archaeological remains. Inferences about such aspects have sometimes led to debates. For example, there are divergent views among researchers on whether the heat treatment of silcrete during the South African MSA was successfully accomplished aboveground, by placing nodules directly in an open fire or embers, rather than underground in a pit. Based on the identification of wood tar residue and distinct fractures on heat-treated archaeological and experimental pieces, a case was made for a less demanding process³ while other experiments involving an underground heat treatment of silcrete have led to the inference that this was likely a complex process involving more planning and temperature control, hence greater cognitive investment¹³. Others have suggested based on experimental work that the differential tolerance for heating rates and temperatures of silcrete raw materials recovered from discrete geological sources may have necessitated the application of more than one heat treatment method, depending on technological needs and/or raw material qualities¹⁶. The Konso heat treatment example presented here involves a different type of siliceous rocks – chalcedony. Yet, it provides a case where the careful control of heating rate and temperatures is indispensable to attain the desired outcome.

It is unclear whether the Konso heat treatment evolved from a practice in the deep past where materials other than ceramic were used as thermal insulators. In the archaeological record, the presence of baked clay in a pit containing heat-treated silcrete nodules underneath has been interpreted as a similar underground heat treatment where clayey sediments were possibly used as thermal insulators⁵⁴. If the contemporary Konso toolstone heat treatment tradition has a pre-ceramic origin, such sediments could likely be used for thermal insulation. A shorter chronology for the Konso heat treatment tradition would, on the other hand, beg alternative explanations. In that regard, it is interesting to note the absence of obsidian sources in the Konso territory, as opposed to all other stone tool-using groups in southern Ethiopia. Such unique stone raw material contexts of the Konso may have thus necessitated innovative ways of improving the flaking qualities (e.g., predictability) of the available raw materials through a process that requires careful control and high time, resource, and cognitive investment.

Archaeological inference inherently involves some form of analogical reasoning. Unlike direct analogies based on similarities between ethnographic and archaeological objects of interest, relational analogies emphasizing shared natural or cultural underlying principles enhance the interpretation of archaeological data. The potentially pertinent aspects of Konso toolstone heat treatment within its unique context provide insights into the drivers, techniques, and patterns of material transformation with which archaeological heat treatment can be better understood and interpreted.

Konso material scientists

Konso technologies, such as toolstone heat treatment, demonstrate a long history of transferring generational knowledge garnered through observation and experimentation, i.e., science. The knowledge, skills, and procedures involved in the Konso heat treatment attest to the fact that this technology requires long-term planning, multi-tasking, and attention-switching^34–36,55. In addition to the tremendous length and complexity of the process, the Konso knappers demonstrate a complex knowledge of thermal shock through their use of a preheated pit, an ash-ember mix and insulators to maintain uniform and stable temperature, pottery as a good thermal retainer-conductor, and the gradual cooling of fullyHT toolstones. Given these complex processes, the high-fidelity knowledge transfer, and the time, energy, and raw material investments involved, understanding the material transformations in the Konso toolstones require a careful consideration of the context in which the technology is produced. As recognized early on⁵⁶, the optimized attributes of a heat treatment process can be difficult to interpret without understanding the knowledge and skills behind the development of specific technical responses as solutions to specific problems^16,17.

The Konso knappers heat treat their chalcedony toolstones to attain the desired flakeability, predictability, edge strength and shine, which are reflected in improved homogeneity and smoothness, increased hardness, and newly formed lustre. From a mechanical point of view, the Konso identify edge strength in terms of improved edge sharpness. However, our results show that heat treated scrapers blunt more easily during use due to the increased susceptibility to abrasion^52,53,57, although this is offset by continuous resharpening of the scraper’s working edge. While the antiquity of the Konso toolstone heat treatment tradition is difficult to establish, the invention was likely driven by the need for attaining the desired functional qualities and predictability by attenuating the raw material constraints³⁸. The intentional enhancement of toolstone workability and flake predictability through thermal engineering, therefore, signifies a technological invention conditioned by the interplay between unique ecological contexts and technical choices, including associated opportunity costs.

Materials and methods

Study group and documentation

We surveyed all Konso districts to identify stone tool-using hideworkers. We closely studied the raw material occurrence, and procurement, heat treatment, knapping, and hafting activities of the three practicing hideworkers in Teshmelle and the border with Gelabo localities between 2022 and 2024. Our ethnographic study mainly involved observations, photographic and video documentation, interviews, and the collection of samples for laboratory analyses. Hideworkers were closely observed as they procured, heat treated and knapped their siliceous stone raw materials. The hideworkers from Teshmelle village procure chalcedony nodules from the Ubaba source situated ~ 5.5 km away while the third hideworker uses the Shoshone source, ~6 km away (Fig. 1). The studied hideworkers provided informed consent. This study was granted approval by the Ethical Review Board of Arba Minch University (Approval No. HRE-SE24-AMU035), with an exemption from a full review due to its nature and objectives. The research was carried out in accordance with relevant guidelines and regulations.

Raw material

Chalcedony occurs in the form of fist-sized geodes with thin whitish cortex and blocky veins in weathered rhyolitic bedrocks that widely crop out in the region (Fig. 1, Supplementary Fig. 1). Based on macroscopic examination and optical microscopy of petrographic thin sections, we identify the parent rock as a reddish-grey, flow-banded rhyolite of porphyric and fluidal texture (Fig. 3a). It is possibly identical to the Sharenga Rhyolite of Miocene age which widely crops out across the east-central parts of Konso in the form of piles and necks. Patchy rhyolite exposures are also encountered within larger bodies of rhyolitic tuffs that crop out in the area from which chalcedonies are collected. In a polarized-light microscope, the rhyolite is formed by euhedral phenocrysts of quartz in a fine-grained matrix of devitrified glass, sanidine and opaque minerals (Fig. 3b). The quartz grains are sometimes magmatically corroded as they were partially melted by the still unsolidified lava. Larger crystals (phenocrysts) of sanidine are rather exceptional. Frequent, on the other hand, are larger grains and aggregates of Fe-rich opaque minerals. Secondary Fe-oxides sometimes form rims around quartz grains. Chalcedony can be observed crystallizing into voids, again, with rims of Fe-oxides (Fig. 3b).

Whitish cortical nodules of the Konso chalcedony (Fig. 1, Supplementary Fig. 1) are sometimes embedded in the rhyolite as layers and lenses. The microstructure is formed by bundles of fibrous chalcedony that go to extinction simultaneously. Much like in the sample reported elsewhere⁵⁸, there are alternating areas of coarse-grained chalcedony and fine-grained matrix, the latter showing fine banding (Fig. 3b). There seem to be sheaves of chalcedony at margins from where the nucleation started, filling voids during a hydrothermal process simultaneous with cooling of the parent rhyolitic lava. This is in concordance with micropetrographic observations made in the rhyolite sample. Introducing a 550 nm retardation plate, it is obvious that fibers of chalcedony are of the length-fast type (LF; Fig. 3). The patches of nonfibrous microquartz are relatively scarce.

Samples

To understand the compositional and physical transformations in the Konso heat treatment of chalcedony toolstones, data were collected from the studied groups across three field seasons. All samples analyzed in the present study come exclusively from the Ubaba source. Collected samples represented different stages of the heat treatment process. Unheated samples (unHT, n = 23) were collected from the same nodules whose remaining parts were eventually heated. Fully heated samples (fullyHT, n = 51) were collected once the heating process was completed. In addition to these, partially heated samples (semiHT, n = 15) deemed not yet ready, based on a visual assessment of the roughness and lustre by the hideworkers, were collected when the hideworkers opened the pit to check on the heating progress. Heating temperature was measured by a handheld RS Pro K Probe Digital Thermometer (model RS52 206–3738), which has a dual thermocouple input and an accuracy of ± 0.1% of the reading + 0.7 °C at 1 °C resolution for measurements ranging between − 99.9 °C and + 999.9 °C. A handheld digital infrared thermometer (model IR1150) capable of measurements of -30 °C to 1150 °C was used to double check the ambient and rock temperature readings. Samples were exported for specialized compositional and physical analyses in laboratories at Palacký University in Olomouc and the Institute of Physics of Materials of the Czech Academy of Sciences in Brno, Czechia. The various types of specialized analyses required various sample dimensions and sizes. In addition, some of the samples were used for multiple analyses while other samples were used for a specific type of analysis as deemed suitable. Supplementary Table 8 provides a list of the samples, and the respective analyses performed on each.

Infrared spectroscopy

In the first step, FT-IR spectra known to change upon the heat treatment of SiO₂-rich lithologies were measured on the Nicolet iS50 FT-IR spectrometer (Thermo, USA) in transmission mode. Relatively large preparation flakes knapped by the Konso were selected as suitable for this analysis. Prior to the measurement, all samples were soaked in water for 48 hrs^cf..ref.49. As single peaks do not always discriminate between heated and unheated samples, we plotted their ratios. Some of these ratios were similarly measured in previous studies⁴² and include absorbance ratios of 4545/4469 cm^− 1 and 4535/4450 cm^− 1. These ratios generally reflect the amount of water in chalcedony in the form of silanol (SiOH in crystal defects and on crystal surfaces, i.e., grain boundaries⁵⁹) and H₂O (in crystal structure⁵¹). The measured pieces were 16 unHT and 11 fullyHT flakes (Table 1). The limited SemiHT samples had to be saved for other tests and were thus not include in the IR spectra measurement. The 4545/4469 cm^− 1 ratio reflects the amount of water retained in the pore space of the samples⁴⁶. Fine-grained silica lithologies usually show an increase in the 4545/4469 cm^− 1 ratio, which represents the currently most frequently measured ratio in heat treatment studies^41,42. The 4535/4450 cm^− 1 ratio shows the implication of silanol in H-bonding. When heat treating a material, the H-bond should be weakened, and this will be reflected by the increase of the 4535/4450 cm^− 1 ratio in temperatures exceeding 300 °C⁴².

Confocal microscopy

We performed rugosimetric analysis at a microscopic scale to assess changes in surface textures. A sample of HT (n = 7) and unHT (n = 7) chalcedony flakes were collected from a single Konso hideworker. In order to take into account intra-flake variability, we measured 3–4 randomly selected loci on each freshly exposed ventral surface, following a widely used approach in rugosimetric analyses of stone tools and other materials^[e.g.,60. Surface measurements were collected using a Sensofar S-Neox confocal microscope. Each measure locus represents a surface of 663 × 500 μm, corresponding to the stitching of four fields of view measured with a long-working-distance 50x objective (numerical aperture = 0.45; spatial sampling of 0.26 μm; white light). The resulting high-resolution height maps (saved as .plux files) were analyzed using the SensoMAP Premium 8.2 software.

Post-acquisition treatment, using the built-in SensoMAP operator, adapted procedures detailed elsewhere^61,62, which included: (i) levelling the surface (least-square method); (ii) removing isolated outliers and noise⁶³; (iii) removing, manually with the ‘retouch’ operator, additional artefacts; (iv) filling non-measured points using interpolation from neighbor points; (v) extracting a 400 × 400 μm-surface; (vi) removing the form with a second-order polynomial; and (vii) applying a Gaussian filter of 80 μm as a cut-off to remove waviness from roughness.

For each 400 × 400 μm-surface (saved as a .sur file), after a last step of levelling, various rugosimetric parameters were compiled. These combined ISO 25,178 parameters, Scale-Sensitive Fractal Analysis (SSFA) parameters, and new parameters provided by the latest versions of SensoMAP to quantify depth and density of furrows (hereafter “furrow parameters”). ISO 25,178 and SSFA parameters efficiently quantify and detect differences in surface textures on a wide range of biological and archeological objects^61,64–67. For each flake, we finally compiled the average value of the 3 or 4 loci and visualized the differences between the two groups of flakes by using box plots (Supplementary Table 1, Fig. 5).

Mechanical testing

A separate set of cores form the Konso chalcedonies was selected for mechanical testing. These comprised eight unHT, four semiHT, and six fullyHT cores collected from the studied Konso knappers. As in a previous study by some of the present authors⁴¹, fracture toughness was determined on notched beams 4 × 3 × 25 mm in size in a three-point bending configuration using an Instron 8862 electromechanical machine. The chevron notch beam method (CNB; ČSN EN 14425-3 for ceramics) was used to determine fracture toughness values K_{I, cnb} [MPa.m^1/2] calculated from the specimen’s geometry, maximum applied force, and compliance function calculated via the slice method. Chalcedony beams were cut using a precise diamond saw Brillant 220 (ATM GmBH, Germany) and ground down to 10-µm diamond abrasive to eliminate cutting defects. The chevron notch was machined into the beams using a 0.15-mm-thick diamond cutting wheel. In total 73 K_{I, cnb} measurements were realized (Table 1 and Supplementary Material). We did not further distinguish which beam came from which core as long as it belonged to its respective category (unHT, semiHT or fullyHT). The details about the method and a comparison with the indentation techniques can be found elsewhere^68,69. Prior to K_{I, cnb} measurements, density was measured on selected beams (Table 1 and Supplementary Material) by Archimedes method according to EN 623-2 standard (hydrostatic weighing) and accuracy of estimation of 0.001 g/cm³.

One beam from each sample category (i.e., unHT; semiHT; fullyHT) was mounted onto epoxy resin, ground, and polished on progressively finer abrasive polishing material down to 1 μm. Subsequently, indentation techniques were used to acquire indentation Young’s (elastic) modulus E_IT [GPa] and Martens hardness HM [GPa] (Table 1 and Supplementary Material).

Instrumented hardness was measured using a mechanical testing machine (Z2.5/ZHU0.2 Zwick/Roell, Germany) at 5 kg/49.03 N loading. As multiple indentations can be performed on a single sample, we acquired 19 (unHT), 30 (semiHT), and 28 (fullyHT) valid hardness measurement data from the respective sample categories. The same number of measurements was planned for E_IT. A total of 19 measurements was successfully conducted for the unHT sample set but, due to the inhomogeneity of the samples, only 16 semiHT and 23 fullyHT samples were successfully measured for E_IT. The nanomechanical machine (ZHN Zwick/Roell, Germany) was used for scratch tests. These were conducted on one sample from each category (unHT; semiHT; fullyHT). The experimental set-up used a Vickers type diamond tip which was laterally moved across the surface with increasing normal force. The overall lateral distance was set to 300 μm and the normal force increased from 0 to 300 mN. The scratch test was performed over a period of 30 s. The scratch depth was measured after the test as the normal displacement with respect to the lateral distance. In total 20 scratch test measurements were made for each sample, and the average scratch depth was calculated for each.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Y.S. conceived the project. Y.S., M.M., S.A., H.H., Z.C., F.Š., L.K., A.S., S.A.B. and K.W.A. conducted field and/or laboratory research. Y.S., M.M., H.H., Z.C. F.Š., L.K. and A.S. analysed and interpreted the data. Y.S. and M.M. drafted the manuscript. All authors commented on and contributed to the manuscript.

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.