Impact of prolonged heating on the color and crystallinity of bone

Abstract

Duration of an anthropogenic fire event is one aspect of fire use and maintenance that is linked to combustion feature function but has low archaeological visibility. In this study, we describe the transformations to fresh, modern cortical bone with prolonged exposure to heat in order to evaluate the utility of archaeological bone for the recognition of long duration thermal alteration. Cores of bovid cortical bone were heated exposed to air at 300, 550, and 750 °C in a sequence of experimental trials in a Nabertherm muffle furnace for periods of 10 minutes, 9 hours, and 48 hours, plus an extensive cooling period on heat retaining sediments (gravel or gravel compacted with fine quartz sand) to mirror the smoldering and extinguishing of actualistic fires. After heating, bone cores were analyzed with a color tool, Fourier-transform infrared spectroscopy, and X-ray diffraction to evaluate changes in structure, composition, and crystallinity of bioapatite as a function of different temperature thresholds and time. Results indicate that prolonged heating in air induces specific structural and chemical changes in bone compared to shorter duration burned counterparts. Coloration changes also demonstrate that white coloration, a primary characteristic utilized by zooarchaeologists to record information about burning intensity, is not an exclusive indicator of calcination at moderate to high temperatures but may also result from long duration exposures at low temperatures.

Introduction

Ethnoarchaeological documentation of fire use in modern hunting and gathering societies has highlighted the variable of duration as a distinguishing factor between different functional roles of fire (Henry et al. 2018; Mallol et al. 2007; Mallol and Henry 2017). In observations of use of fire by living hunter gatherer populations, prolonged duration of a fire event was noted to be necessary for activities such as meat curing and hide smoking, as well as for long-lasting communally maintained cooking fires (Henry et al. 2018; Mallol et al. 2007; Mallol and Henry 2017). In contrast to that, natural wildfires in environments with low woody biomass have been reported to be only of short duration (Table 1). Duration may therefore present a valuable proxy to differentiate the occurrence of a natural fire event from people constructing spatially constrained combustion features or hearths (Gowlett et al. 2017; Cutts et al. 2019) and for identifying and reconstructing ancient fire use practices. The length of archaeological fire events has proven challenging to document, however, due to the ephemeral nature of fire.

Table 1 Reported details of environment and duration, when reported, for observed natural fires or experimental fires in natural settings

Indirect fire proxies, such as thermally altered bone, provide opportunities to verify fire presence and document the conditions under which bone mineral is transformed with heat. To evaluate the impact of prolonged heating on bone we performed for this study a set of laboratory experiments heating modern mammalian (bovid) cortical bone at different durations (10 minutes, 9 hours, 48 hours) at controlled temperatures (300, 550, 750 °C). We recorded visible color with an RM200 CAPSURE Color Matching tool and utilized Fourier transform infrared spectroscopy (FTIR) with attenuated total reflectance (ATR) and X-ray diffraction (XRD) to provide information on the dual impact of duration and temperature on bone cortical tissue. Both observational and semi-quantitative spectroscopic data are considered here to reflect the practical and minimally invasive methodologies utilized by zooarchaeologists and to create a standardized dataset. Specific research objectives of this project include:

• 1. To investigate the impact to bone color, chemistry, and structure with prolonged heating.

• 2. To assess if prolonged heating at different temperature thresholds can be distinguished from each other.

• 3. To evaluate the application of these observations for interpreting fire origins and functions with archaeological burned bone assemblages.

Duration times in the context of natural versus anthropogenic fires

It is difficult to discern evidence for anthropogenic control of fire (hominin management of a local fire via fuel control), use of fire (hominin use of fire or fire altered materials naturally occurring in the landscape), or non-anthropogenic wildfire (Buenger 2003; Goldberg et al. 2017; Gowlett et al. 2017; Sandgathe and Berna 2017; Stahlschmidt et al. 2020). Natural wildfires are common and integral to many ecosystems and vary in intensity in relation to biome, climate, and vegetation (Scott 2009). Variations in intensity and length are also influenced by the available fuel, fuel moisture, weather conditions, and topography. Grass and brush fires are primarily documented to be ephemeral and fast moving, passing in minutes, opposed to the longer and more intense heat associated with peat, dung, and tree stump fires (Bailey and Anderson 1980; Bellomo 1993; David 1990; Gowlett et al. 2017; Pyne et al. 1996; Santana et al. 2013; Scott 2000; Stinson and Wright 1969; Whelan 1995; For review see Stahlschmidt et al. 2023, this issue; Table 1). Data regarding the average lengths and temperatures of forest fires in differing conditions, such as boreal and tropical forests, are difficult to obtain and are not regularly reported. For dry and heavily wooded environments, fires are commonly acknowledged to burn for hours.

Prior research on mammalian animal carcasses burned in fast-moving wildfires has shown that the skin, flesh, and organs can shield bones from the rapid initial heating (Gowlett et al. 2017). For animals like tortoises, which have exposed bony components, passing brushfires do leave visible burning indications on remains, but the impact is generally low to moderate (Stahlschmidt et al. 2023). The less commonly observed high temperature intensities seen on the tortoise remains are likely related to the presence of naturally occurring fuel sources on the landscape. These fuel sources increased the combustion potential and available fuel, leading to a higher intensity of burning in those cases, even though the fire event was not long lasting (Stahlschmidt et al. 2023). These findings align well with the known impact of flesh on burning dynamics, as initially flesh acts as a protective layer on bone elements when heated or exposed to flame (Ellingham et al. 2016). However, when the flesh combusts given the opportunity of high temperature intensities and time, it becomes a fuel source itself, contributing to the overall intensity of alteration to the bone components (Ellingham et al. 2016).

Neither the duration of a fire event nor high temperatures are inherently or intrinsically linked to fires either of anthropogenic or natural origin. Despite this, duration times are considered potentially more informative than high temperatures for assessing the likelihood of anthropogenic fire, as in many instances and environments natural fire cannot stay self-sustained for extended periods (Gowlett et al. 2017; Hlubik et al. 2019; Table 1). Many ethnographically documented human controlled and curated fires do require several hours for activities such as meat or hide smoking, or communal cooking fires, however (Henry et al. 2018; Mallol et al. 2007; Mallol and Henry 2017). An assessment of prolonged heating in bone can therefore contribute valuable information within an integrated evaluation of the likelihood of an archaeological fire curated by hominins following the microcontextual approach (Goldberg and Berna 2010; Goldberg et al. 2017). This methodological and theoretical framework recommends the integration of multiple multi-scale analytical techniques and emphasizes the context of archaeological material, including the surrounding matrix. Several independent lines of evidence of fire proxies and heat induced transformations to materials and sediment are necessary to comprehensively verify anthropogenic fire signatures and address the presence and properties of ancient fire.

When employing thermally altered bone as an indirect proxy for fire, it is essential to recognize that observed changes in burned or heated bone offer insights solely into the localized conditions of the exposure. These conditions may only represent the properties of a fire at a specific moment in time. It is also possible for bones to be reheated, and overall alteration may represent several heating events. The duration of heat experienced by bone can therefore provide a metric of the time of heating from an unknown number of occurrences. Indications of prolonged heating still can offer valuable insights into the relative properties of a fire, however, both for bone intentionally included in fires or incidentally burned around or under a feature. This information can provide additional details to contribute within the microcontextual approach that otherwise would remain undetectable.

Transformations to bone mineral and chemistry with thermal alteration

Bone is a useful indirect proxy of fire and heating conditions, as thermal alteration results in permanent changes to bone structure and chemistry which may be measured both in experimental and archaeological settings. By analyzing the properties of burnt bone, we can accumulate valuable information about the conditions under which the faunal material was thermally altered, verifying the presence of ancient fire (e.g., Berna et al. 2012) and addressing aspects of hominin pyrotechnology (e.g., Théry-Parisot 2002).

Living bone itself is comprised of the inorganic mineral bioapatite, organic proteins primarily of collagen, and water (Martin et al. 2015; Stout et al. 2019). Bone is porous, to a large extent for facilitation of Haversian systems, with trabecular bone exhibiting a higher porosity due to a more open structure (Figueiredo et al. 2010; Lyman and Lyman 1994; Stout et al. 2019). Bioapatite crystallites in vivo are characterized by a hydrated, large surface area (above 200 m²/g) necessary for the regulation of homeostasis (Berna et al. 2004; Buikstra and Swegle 1989; Drouet et al. 2018; Koeppenkastrop and Eric 1992; Martin et al. 2015; Reynard et al. 1999; Rollin-Martinet et al. 2013; Trueman et al. 2008; Ubelaker 2009), and small, plate-like morphologies (1–7 nm thick, 15–200 nm in length, 10–80 nm in width) which are cross-linked to the organic collagen fibrils (Bala et al. 2013; Berna et al. 2004; Rey et al. 2009). Water is integrated into and around the organic and mineral components but also found in void spaces and in the extracellular matrix (Drouet et al. 2018; Nyman et al. 2008).

Macroscopic properties influenced by heat include changes to fragment size, friability, texture, and mass at different temperature thresholds (Buikstra and Swegle 1989; Ellingham et al. 2015; Stiner et al. 1995; Shipman et al. 1984; Thompson 2004; Zazzo et al. 2009). These transformations are intricately linked to the chemical and structural organization of the mineral, organic, and water content of bone. While these changes are similar to modifications which occur during normal bone diagenesis, heat induced transformations above median temperatures (~ 500 °C) are more exaggerated in the extent and amount of alterations. The degree of transformation, the material properties, and the chemistry of bone exposed to heat are also related to the specific conditions under which the bone was burnt, including maximum temperatures and burning atmosphere (Ellingham et al. 2015; Gallo et al. 2021; Reidsma et al. 2016; Snoeck et al. 2014, 2016; Van Hoesel et al. 2019).

Studies have documented these specific primary structural and chemical changes to bone heated at different temperatures, including the rate of organic loss, alterations in porosity, the presence of different compounds reflective of burning atmosphere, and mineral bioapatite recrystallization and crystallite size growth (Ellingham et al. 2015; Gallo et al. 2021; Reidsma et al. 2016; Van Hoesel et al. 2019). High-resolution spectroscopic analyses such as FTIR and XRD can directly monitor these transformations to verify true burning, rather than staining, and elucidate maximum temperature thresholds and different burning atmospheric conditions (Ellingham et al. 2015; Piga et al. 2008; Shahack-Gross et al. 1997; Snoeck et al. 2016; Thompson 2004).

Four major overlapping and interconnected stages of heating which broadly describe the transformations to the mineral, organic, and water components are well documented with macroscopic, microscopic, spectroscopic, and radiometric studies for bone burnt and heated in oxygen atmospheres (Ellingham et al. 2015; Mayne Correia 1997; Thompson 2004). Combustion (heating in the presence of oxygen) is here distinguished from charring (heating without the presence of oxygen), as bones undergoing thermal alteration in full reduction atmospheres experience different changes which do not overlap neatly with their oxygenated counterparts (Reidsma et al. 2016; Van Hoesel et al. 2019; for review see Marques et al. 2021).

For bones burnt and heated with access to oxygen, stage 1, dehydration (100–600 °C), describes the evaporation of the water content of bone from within and between the organic and mineral components environments (Ellingham et al. 2015; Mayne Correia 1997; Thompson 2004). Stage 2, decomposition (300–800 °C), details the combusting and depletion of the organic portion (Ellingham et al. 2015; Mayne Correia 1997; Thompson 2004). Bone at this stage of burning loses a great deal of mass and has the highest degree of porosity, as the organic and water components have been removed and the bioapatite minerals remain small and flat (Figueiredo et al. 2010; Gallo et al. 2021; Pramanik et al. 2012). Stage 3, inversion (500–1100 °C), describes the loss of structural carbonates. Stage 4, fusion (> 700 °C, estimated have full transformation closer to 900–1000 °C), accounts for the coalescing of the remaining inorganic component (Ellingham et al. 2015; Mamede et al. 2018; Mayne Correia 1997; Thompson 2005). Bone which has undergone the fusion stage changes substantially in shape with a drastic increase in crystallite size of bioapatite, which additionally decreases porosity and produces a more crystalline structure (for review, see Ellingham et al. 2015; Gallo et al. 2021; Mamede et al. 2018).

Calcination in the context of thermally altered bone is a categorization broadly identified by white or grey coloration, due to the removal of the combusted and charred organics, but further describes micro- and nanoscopic transformations to the bone material which occur alongside the stages of burning previously outlined. Many structural and chemical changes occur during this process, including replacement of carbonates with hydroxyl groups and, even before the large-scale coalescence of the fusion stage, bioapatite growth (Etok et al. 2007; Mamede et al. 2018; Piga et al. 2016). This crystallite growth is very similar to growth seen as a part of diagenetic processes (Ostwald ripening) as crystallites evolve to more stable phases with a more ordered structure (i.e., become more crystalline) (Dal Sasso et al. 2018; Hedges 2002). The growth induced by high temperatures can be more intense than during typical diagenetic alteration, however, as after the removal of structural carbonates there are less distortions on the crystal lattice (Dal Sasso et al. 2018; Gallo et al. 2021; Mamede et al. 2018). Calcined bone for the purpose of this study is therefore defined by both the light coloration from the elimination of the combusted and charred, organic components (Ellingham et al. 2015; Ramirez-Gutierrez et al. 2017) but also the rapid transformation of the mineral phase with high temperature exposure: loss of structural carbonates, shifting vibrational frequencies of the bone mineral reflecting differing chemical compositions, large average crystallite sizes, and the corresponding high crystallinity indicating more ordered crystals and low lattice strain (Ellingham et al. 2015; Gallo et al. 2021; Mamede et al. 2018; Marques et al. 2018).

The reported temperatures under which bones are calcined varies, likely due to different experimental variables including the initial status of bone sampled, temperature steps utilized, and method of measurement (Table 2; Ravaglioli et al. 1996; Snoeck et al. 2014).

Table 2 Sample of reported temperatures of bone calcination, modified from Snoeck et al. (2014)

Color changes in thermally altered bone

The connections between burning stages and visible color are most clearly seen in reference to the state of organic decomposition, as the dark coloration (dark brown or black) of burnt bone is reflective of the charred or carbonized organic content (Snoeck et al. 2014). Dark brown and black surface colorations can also be indicative of discoloration or staining (from mineral manganese, soil, wood tannins, and/or the depositional environment) and not indicative of burning (Bradfield 2018; Dupras and Schultz 2014; Nicholson, 1993; Pollock et al. 2018; Shahack-Gross et al. 1997; Turner et al. 2018). For example, fauna can gain a saturated brown hue when in soils with high humic content (Dupras and Schultz 2014). White coloration occurs at higher temperatures associated primarily with calcination due to the elimination of the organic char, but bone can also become stark white in color with prolonged exposure to UV radiation (Dupras and Schultz 2014). This process, known as sun bleaching, also achieves depigmentation through the removal of organics with UV radiation (Dupras and Schultz 2014).

The color changes of heated bone are widely recognized, with many scholars noting that a sequence of color gradients can be connected broadly to different intensities of burning and heating, i.e., carbonized/charred and calcined (Costamagno et al. 2005, 2009; Ellingham et al. 2015; Shipman et al. 1984; Stiner et al. 1995). Observations of these color changes have established widely used zooarchaeological scales of heating, such as the Stiner et al. (1995) scale which utilizes macroscopic cues of color to indicate > / < 50% charred and > / < 50% calcined (Table 3). Studies which report and investigate heating through zooarchaeological methods often utilize these scales and similar schema to categorize the burning variability within an assemblage.

Table 3 Modified Stiner et al. (1995) burning scale with associated generalized coloration for recognition

Colorimetric studies that consider the hue (brightness) and chromatic qualities (color) alone to discern attributes such as temperature of thermal alteration in burnt bone have been demonstrated to be unreliable, however (Bradfield 2018; Fredericks et al. 2015). These issues are reasonably understood in the context of archaeological staining and depigmentation, as well as through an understanding of the origin of bone coloration. The changing color gradients within burnt bone at increasing temperatures reflect the charring and removal of the organic component and are not specifically tied to the temperature itself. Exemplifying these issues with coloration are studies which observe in laboratory settings bones that are black and brown after being heated at temperatures above 700 °C without access to oxygen, when white coloration is expected at temperatures ≥ 700 °C when bone is burned in the presence of oxygen (Marques et al. 2021; Reidsma et al. 2016; Walker et al. 2008). Oxygen is not needed to create the changes to the organic composition seen ~ 300 °C known as char, but oxygen is necessary for combustion, in which the char and present volatiles oxidize (Braadbaart et al. 2007; 2012; Reidsma et al. 2016). Further, in these anaerobic conditions, amorphous inorganic carbon is produced in these bone samples, contributing to the dark black coloration (Marques et al. 2021).

Solely utilizing color, therefore, can be problematic in studies utilizing bone as an indirect proxy for fire. It is only with consideration of other heat related variables and characteristics, such as surface warping and texture, weight and porosity transitions, and spectroscopic or radiometric measures of bioapatite transformation (including crystal size growth and shifts in the chemical arrangement) that heat alteration can be verified, and relative thresholds of burning conditions can be distinguished.

Evaluation of duration effects in prior experimental studies

Previous studies addressing the thermal alteration of bone have not systematically evaluated the impact of prolonged duration times. Prior experiments have incorporated a range of lengths when assessing questions of temperature intensity (Ellingham et al. 2016; Gallo et al. 2021; Reidsma et al. 2016; Shipman et al. 1984; Snoeck et al. 2014; Van Hoesel et al. 2019), burial environments (Bennett 1999; Herman and Bennett 1999; Stiner et al. 1995), carbon exchange (Snoeck and Schulting, 2013; Snoeck et al. 2014) and burning atmospheres (Marques et al. 2021; Reidsma et al. 2016; Van Hoesel et al. 2019; Table 4). Many studies utilizing durations limited to 1–3 hours have reported that time length of heat exposure has little impact on the transformations to bone mineral at different temperatures (Ellingham et al. 2016; Gallo et al. 2021; Krap et al. 2019). It is challenging to compare these results to experiments which do have duration times that exceed 3 hours, as many of these studies do not include spectroscopic data to provide insight into the degree of chemical and structural alterations (Table 4). This pattern is likely due to the relatively recent advancements in the ease of spectroscopic analyses for archaeological and experimental application and because of the difficulty of sustaining prolonged heating experiments at high temperatures.

Table 4 Reported details, including duration times, for a selection of faunal burning experiments

Discrepancies found between the different studies working with thermally altered bone, such as the reports of calcination from underground burning experiments in Bennett (1999) that recorded a maximum temperature of 506 °C, but not in Stiner et al. (1995), may be explained by considering the impact of length of the heat exposure and the initial status of the bone components (i.e., fresh or weathered) at time of the experiment (Bennett 1999; Ramirez-Gutierrez et al. 2017; Stiner et al. 1995). Additional factors which must be considered and vary between studies include fragment size and thickness, an initial status of powdered or solid bone, the degree of flesh covering, differences within and between species, and, potentially, duration.

An advantage in utilizing ATR-FTIR is that spectra characterizing the vibrations of bonds within the sample reflecting heat and diagenesis induced changes can be produced with minimal sample preparation (Beasley et al. 2014; Dal Sasso et al. 2018; Ellingham et al. 2015, 2016; Gallo et al. 2021; Gianfrate et al. 2007; Lebon et al. 2014; Lebon et al. 2016). These spectra provide semi-quantitative measures of bone composition that reflect structural and chemical changes induced by heat, such as specific peaks representing new chemical compounds and peak changes that correspond to higher degrees of crystallinity due to increasing bioapatite mineral order, size, and strain (Ellingham et al. 2015; Gallo et al. 2021; Mamede et al. 2018; Snoeck et al. 2014; Weiner and Bar Yosef 1990). The changes in width of diffraction peaks in XRD patterns provides complementary information on the sizes of bioapatite crystallites which experience substantial growth at temperatures > 650 °C (Ellingham et al. 2015; Etok et al. 2007; Figueiredo et al. 2010; Gallo et al. 2021; Pramanik et al. 2013; Ramirez-Gutierrez et al. 2017).

Materials and methods

Sample preparation and experimental methods

Cortical bone of bovid (Bos taurus) femurs, humeri, and radii were obtained from a local butcher (Fleischerei Engeln, Leipzig, Germany) to be utilized in this study. At time of purchase the bones segments were never frozen, and criteria for bone selection included the present bone availability and freshness. Bones were sourced in several waves so that samples for this study were only stored for a maximum of three days after purchase to preserve fresh cortical bone mechanical properties (Zhang et al. 2018), and trial order was determined by this time limitation. Each bone prepared was given a unique identifying label (A-L, SI Table 1).

The domesticated cow long bones were mechanically scraped of remaining flesh and periosteum and the diaphysis portions were cored utilizing a PROXXON Rotary Tool MICROMOT 60/EF drill with a Dremel 663 ¼ inch Diamond grit drill bit (SI Table 1). Resulting solid cortical bone core diameters were therefore standardized at ~ 4 mm, and length measurements varied between 2 and 7 mm, with an average of 3.92 mm (n = 240; Fig. 1). Bone cores of this size were chosen as heat would uniformly impact the entire sample. All samples were stored wrapped in saline (NaCl 0.9%) soaked gauze to maintain a hydrated state following removal of the periosteum and refrigerated at temperatures around 4 °C (Zhang et al. 2018). Bone segments for each trial were selected as to not be from the same element (e.g., both segments from the same left femur), but potentially could represent different elements of the same living animal (e.g., the left and right femur of the same animal).

Fig. 1

Representative sample of bone cores exposed to the variable heating duration experiments. All duration trials (T1–T8) are organized by time length (10 minutes, 9 hours, and 48 hours). Stark color differences exist between the 10 min duration period and bones heated at same temperatures but for 9 and 48 hours. Bones heated below the temperature of calcination (300 and 550 °C) but for the elongated period show a color resemblance to bones calcined at 750 °C at any duration. All samples organized by trial, crucible, sediment type, and depth (surface or − 2 cm depth) are provided in supplementary information (SI Figs. 1–47)

Bone tissues sourced from different species are observed to have small differences in bone hierarchical organization and in bone organic composition. The chemical formula for bioapatite itself is acknowledged to have small variances related to element and ion substitutions. Such variance is common even within bones and elements of a single species, however. As this study is primarily concerned with the changes to bone mineral crystals and the rate of organic loss, the use of bovid bone as a medium is not considered to be a problematic factor here and results can be broadly compared to prior studies using bone tissue from different species.

For each trial cores from three different bone segments (labeled A–L) were positioned in a crucible with the intention of randomizing the variables related to the natural state of the animal raw material that are outside of direct experimental control (SI Table 1). Each of the three bone cores were arranged approximately 1 cm distance from each other in consistent placements relative to labeling information on the crucible surface for later identification and retrieval (Fig. 2). For six of the trials, a layer of bone samples were also buried − 2 cm under the surface of the substrate (Table 5). Ceramic crucibles were 8 cm in maximum diameter and were filled either with substrates of loosely packed gravel (2–4 mm) or compacted gravel with fine quartz sand (1/8–1/4 mm) to emulate sediments with different heat retention properties (Fig. 2; Table 5).

Fig. 2

Muffle furnace and crucible setup: A Nabertherm muffle furnace; B crucible of gravel and fine quartz sand, crucible label visible which was used to orient samples; C schematic of numbered crucible placements in muffle furnace. Crucibles 6, 7, and 8 were only utilized for T1 (550 °C, 48 h). Bone cores (e.g., bones A–C) arranged in sequential order in a triangle formation relative to visible crucible label. D Picture of post-heating T1 (550 °C, 48 h) crucibles prior to excavation. Note rubified sediment composition and white bore cores in both gravel (crucibles 1–4) and compacted mixed gravel and quartz fine sand (crucibles 5–8)

Table 5 Trial parameters

Heating trials were designed to observe three temperature thresholds: 300 °C, a moderate temperature well beneath published temperatures of calcination; 550 °C, to test the brink of the calcination boundary; and 750 °C, firmly within the range of calcination. Length of trials were selected to represent a quick heating event for example during a grass wildfire (10 minutes) with expected minimal alteration, a fire event of a moderate length (9 hours) and a long length (48 hours) exceeding alterations currently reported with spectroscopic analyses in the literature. Inferences on moderate and long anthropogenic fire events were drawn from reported data on hunter-gatherer fires, of which include quick hunting or meal roasting fires, daily fires centered around an activity, and communal cooking fires which are kept continuously burning for days (Henry et al. 2018; Mallol et al. 2007). Trials were completed in a muffle furnace, programmed to reach desired temperature at the fastest rate achievable by the instrument (~ 25 °C/60 s) and to hold temperature for the length of the experiment (SI Fig. 48). Temperatures were monitored through the instrument display, and there is not an internal ventilation system. The muffle furnace was allowed to cool at a natural rate before samples were removed, mirroring the slow heat dissipation of an extinguished fire event (SI Figs. 48 and 49).

Sample analyses

Bone color was documented with a color matching tool on all bone cores post heating and prior to powdering with an agate mortar and pestle (Munsell Color (Firm), 2010).

ATR-FTIR and XRD instruments are used here to monitor the transformations that occur to bone mineral with heat and to evaluate changes in crystallinity. XRD specifically can provide information on the average size of the growing crystals even after the ATR-FTIR peak ratios related to crystallinity decrease above 800–900 °C when the lattice organization significantly alters. ATR-FTIR spectra (4000–400 cm⁻¹) were produced with a Nicolet 6700 Fourier transform infrared spectrometer equipped with a deuterated triglycine (DTGS) detector and a single bounce diamond crystal. ATR was specifically utilized to minimize sample preparation, and contamination (Bruno 1999; Hollund et al. 2013; Thompson et al. 2009; Nakamoto 2009). 256 scans were collected with a sample gain of 8 and resolution of 4 cm⁻¹.

To assess organic content and calcination, eight peaks and peak transformations were collected for all samples in this study (n = 234), and all measurements and equations were standardized and processed using OMNIC Macros Basic software (Tables 6 and 7). The CO/P ratio was monitored for the relative concentrations of organic components in bone, as this measurement assesses the functional group recognized as Amide I, a component of collagen proteins (Thompson et al. 2013; Mamede et al. 2018). Another measurement, the carbonate to phosphate (C/P) ratio, is utilized to understand the relative amount and loss of structural lattice carbonates within the sample (Thompson et al. 2013; Mamede et al. 2018). For reconstructing the degree of crystallinity (crystal order and lattice strain), a measure known as the infrared splitting factor (IRSF or SF) captures the splitting of the PO₄³⁻ v4 peak in bone, a trough which lengthens with increasing crystallinity until heat transformation above 800–900 °C, when the SF ratio experiences a decrease (Mamede et al. 2018; Weiner and Bar-Yosef 1990). This decrease is related to the chemical and structural changes with the fusion stage of bone mineral transformation, as bioapatite crystals melt and sinter (Mamede et al. 2018; Piga et al. 2018).

Table 6 FTIR-ATR wavenumbers associated with the likely functional groups relevant to this study and the thermal alteration of burnt bone (Ellingham et al. 2015; Gallo et al. 2021; Mamede et al. 2018; Reidsma et al. 2016; Thompson et al. 2013; Weiner and Bar-Yosef 1990)

Table 7 Equations utilized for this study

Finally, the appearance of two shoulders on different phosphate peaks indicate compositional changes within the chemical structure of the bone mineral which occur solely with high temperatures (~ 700–1100 °C) including the growth of a shoulder at ~ 1090 cm⁻¹ off the PO₄³⁻ v3 peak suggested to be the presence of francolite and a ratio measured from a shoulder at ~ 625 cm⁻¹ on the PO₄³⁻ v4 peak known as the phosphate high temperature (PHT) shoulder indicating an OH- librational mode associated with hydrogen bonding within the crystalline framework (Gallo et al. 2021; Mamede et al. 2013; Marques et al. 2018; Thompson et al. 2013).

To determine average crystallite size, diffraction peaks were obtained with a Bruker D2 Phaser diffractometers using CuKα radiation and run from 7 to 70° 2θ with 0.04° step and 0.02 θ increment. Divergent slits were set at 0.6 mm with a 1 mm anti-scatter screen, and a maximum opening of position sensitive detector (4.8°). The powdered bone samples were spread with ethanol on a zero background silicon holder. Average crystallite size was refined with whole pattern fitting (Rietveld refinement Rietveld 1969) using hydroxyapatite Ca₅(PO₄)₃OH structure (space group P63/m) with the GSAS-2 package (Toby and Von Dreele 2013).

Results

Color

Color observations are represented through the Munsell values given by color matching tool (Munsell Color (Firm), 2010). For the 10 minute duration trials, at 300 °C (T2A, T2B) bones from all crucibles were shades of browns, ranging from pale to darker hues including grey-brown (Fig. 3). At 550 °C for 10 minutes (T4), bones were categorized as darker greys, and at 750 °C (T5) heat altered bones were shades of light greys and whites (Fig. 3; SI Figs. 1–47).

Fig. 3

Munsell (Munsell color (Firm), 2010) color code results using a Color Gun RM200 CAPSURE Color Matching tool taken on bone cores post-heating for each trial (3–6 bone cores per crucible (C); Table 5). Trial 1 split by sediment type and is the only trial with 8 crucibles. Full variation of bone colors and hues can also be viewed in SI Figs. 1–47

For the 9 hour duration trials, bones heated at 300 °C still were recognized as brown in color, although considerably lighter browns and less saturated hues than the 10 minute trials (Fig. 3). At 550 °C for 9 hours (T3A, T3B), a range of greys including blueish-grey and brownish-greys were recognized (Fig. 3). These samples are very light in coloration and are likely, given standard scales of zooarchaeological classification, to be considered heated at a higher temperature based on visual observation. At 750 °C for 9 hours all heated samples were stark white and very desaturated greys as expected (Fig. 3).

Bones heated for 48 hours at 300 °C (T7) were very pale, resembling light tans and pale greys (Fig. 3). This degree of depigmentation would additionally be difficult to classify given the established color classification schemes for burnt bone, and may be mistaken to have experienced high temperatures and calcination based solely on coloration. A similar scenario is also seen in the bones heated at 550 °C for 48 hours (T1). All samples from this trial range from whites to light greys, with multiple samples determined as bluish-grey (Fig. 3).

All bones excluding 300 °C for 9 hours (T6) and 300° C for 10 min (T2A, T2B) would be likely classified as Stiner et al. (1995) Stage 6 based on coloration (white, grey, and heavily muted pale tans).

FTIR results

Ten-minute trials

Four trials were conducted with duration times of 10 min: 300 °C with dry, loose gravel (T2A), 300 °C with dry compacted gravel and fine sand (T2B), 550 °C in dry loose gravel (T4), and 750 °C in dry loose gravel (T5). At 300 °C, bones exhibited strong presence of Amide I and II (1630–1660 cm⁻¹), relevant to their strong organic tissue and water content, and have CO₃²⁻ v3 peak heights similar to fresh, unburnt bone (1400–1550 cm⁻¹), typical of bones burnt to this temperature for brief durations (Gallo et al. 2021; Fig. 4). For both 300 °C trials, there is no indication of high temperature alteration, primarily seen through the low SF ratio (565 and 605 cm⁻¹ peaks and the 595 cm⁻¹ trough) and the absence of a PO₄³⁻ v3 shoulder or a PHT shoulder at 1090 cm⁻¹ and 625 cm⁻¹, respectively (Fig. 4).

Fig. 4

FTIR-ATR spectra for the shortest duration trial (10 minutes; T2A, T2B, T4, T5) at 300, 550, 750 °C. Relevant peaks of interest representing different likely functional groups are highlighted, including the Amide I and II and the CO₃²⁻ v3 peak areas which are visibly depleted with increases in temperature. Also highlighted are the 3rd and 4th vibrations of PO₄³⁻ functional groups which exhibit high temperature induced shoulder growths at ~ 1090 cm⁻¹ and 625 cm⁻¹ (the PHT) with structural and chemical calcination. Also visible is the deepening of the trough at 595 cm⁻¹ on the PO₄³⁻ v4 peak with greater temperatures, a correlate to increasing SF

At 550 °C for 10 minutes, much of the organic components and the lattice carbonates have been eliminated from the samples as seen in the 1630–1660 cm⁻¹ and 1400–1550 cm⁻¹ wavelength regions (Fig. 4). This is consistent with the described first release of carbonates (CO₂) from organic combustion, but not full structural carbonate loss (Etok et al. 2007; Mamede et al. 2018). Despite the dehydration and carbonate loss, the physical and chemical structure of the bioapatite here has not radically changed and there are no indications of high temperature alteration or calcination when referencing the SF, the 1090 cm⁻¹ shoulder off the 1030 cm⁻¹ peak, or the PHT shoulder (Fig. 4). This spectrum is typical of bone burned to this temperature threshold for brief durations (Gallo et al. 2021).

For the highest temperature threshold, 750 °C, major changes are seen when compared to the lower temperatures. At 750 °C, all organics and water are eliminated and for the CO₃²⁻ v3 functional group (the structural lattice carbonate within the bone), there is either a complete absence or a barely present small doublet absorption (Fig. 4). Off the PO₄³⁻ v3 peak at 1030 cm⁻¹ there is a strong presence of a shoulder at 1090 cm⁻¹, linked to high temperature alteration also seen in bones burnt for short durations between 700 and 1100 °C (Gallo et al. 2021; Thompson et al. 2013; Fig. 4). For the PO₄³⁻ v4 peak area there is both a strong presence of a PHT shoulder as well as a much deeper trough at 595 cm⁻¹ between the 565 and 605 cm⁻¹ peaks which constitute a measurement of crystallinity through the SF ratio (Table 6; Fig. 4). These spectra are clearly calcined and exhibit a higher crystallinity than the samples burnt at 550 °C and 300 °C.

Nine-hour trials

Four additional trials were completed at 9 hours of sustained temperature duration: 300 °C in dry loose gravel (T6), 550 °C in both dry loose gravel (T3A) and dry compacted gravel and fine sand (T3B), and 750 °C in dry loose gravel (T8). For the lowest temperature experiment, the primary difference seen between the 9 hour and 10 minute trials is the depletion of the organic content as monitored by the Amide I and II functional group areas (1630–1660 cm⁻¹; Fig. 5). This is similar to the degree of reduced organics seen in archaeological bone, where organics are depleted through normal diagenetic processes (Gallo et al. 2021). There remains a strong presence of the lattice carbonate, and there is no high temperature shoulder at 1090 cm⁻¹ nor a PHT shoulder (Fig. 5). The infrared SF remains low and is not indicative of a highly crystalline sample nor of calcination (Fig. 5).

Fig. 5

FTIR-ATR spectra for 9 hour duration trials (T3A, T3B, T6, T8) at 300, 550, and 750° C. Amide I and II peak areas are present in the 300 °C trial but are weakly expressed compared to the 10-minute trials (T2A and T2B). The CO₃²⁻ v3 peak absorbance height remains similar to both shorter and longer durations at each respective temperature. The PO₄³⁻ v3 shoulder at 1090 cm⁻¹ and PO₄³⁻ v4 625 cm⁻¹ (PHT) shoulder are distinguished at 750 °C, as expected, but both can be seen weakly starting at 550 °C. The SF also visibly increases both at 550 and at 750 °C

Samples heat altered at 550 °C show no influence from the different substrates and have a lower organic presence than both the 300 °C, 9 hour trial (T6) and the 550 °C, 10 minute trial (T4). Weakly present shoulders emerge for both the 1090 cm⁻¹ area off of the 1030 cm⁻¹ PO₄³⁻ v3 peak and the PHT shoulder off of the PO₄³⁻ v4 peak, which are not found in the 550 °C 10 minute trial (T4). There are also deepened troughs at 595 cm⁻¹ which results in a higher SF ratio, but these spectra do not indicate samples which have been fully calcined (Fig. 5).

For the 750 °C trial, again we see clearly calcined bone with spectroscopic signatures of high temperature alteration through the very large splitting factor ratio on the PO₄³⁻ v4 peak, the PHT shoulder, and the PO₄³⁻ v3 1090 cm⁻¹ shoulder (Fig. 5). There are also no remaining organics as seen through the predicted Amide I and II functional group area (1630–1660 cm⁻¹) and there is only a small remaining doublet regarding the CO₃²⁻ v3 area (Fig. 5).

Forty-eight-hour trials

Two trials, 300 °C (T7) and 550 °C (T1), were held at steady temperature for 48 hours. Elevated temperatures also continued for a substantial cooling period after the 48 hour set experiment, as temperatures slowly disseminated for over 7 hours (SI Figs. 48 and 49). For the 300 °C trial (T7), crucibles were filled with dry, loose gravel, and for the 550 °C trial (T1) four crucibles had a dry, loose gravel sediment base, and four had dry, compacted gravel with fine sand.

Bones heated at 300 °C for 48 hours show reduced organics in the 1630–1660 cm⁻¹ wavelength region when compared to their 10 minute counterparts (Fig. 6). There are no indications of calcination on any part of the spectra, however, neither in the PO₄³⁻ v3 peak area nor in the PO₄³⁻ v4 region through the splitting factor ratio or in the presence of a PHT shoulder (Fig. 6).

Fig. 6

FTIR-ATR spectra for 48 hour duration trials (T1, T7) at 300 and 550 °C. Both temperature trials retain the presence of the CO₃²⁻ v3 lattice carbonate functional group, similar to their corresponding temperatures at other duration periods. At 48 hours, bones burnt at 300 °C clearly have only a weak absorption of Amide I and II functional group presence (~ 1630–1660 cm⁻¹), a striking difference from bones burnt to the same temperature but sustained at 10 minutes. There are no indications of structural or chemical calcination for the 300 °C samples on either the PO₄³⁻ v3 and v4 peaks and the SF is not elevated, represented here by a shallow dip between the 605 and 565 cm⁻¹ peaks. At 550 °C, two patterns are noted which map clearly with the crucible distribution within the muffle furnace. Half of the bone samples, located at the front of the muffle furnace close to the door, remain uncalcined with moderate changes to the SF values and no presence of a 1090 cm⁻¹ or 625 cm⁻¹ (PHT) shoulder off of the PO₄³⁻ v3 or v4 peaks (SI Figs. 50 and 51). The remaining samples from crucibles 1,2, 5 and 6 were located in the back of the furnace and we speculate that these samples experienced slightly elevated temperatures due to this placement. These samples do have heat induced changes to shoulders off the PO₄³⁻ v3 and v4 peaks (SI Figs. 50 and 51)

At 550 °C for 48 hours (T1), two different signatures are seen within the same trial. This pattern of differential alteration is potentially correlated with crucible placement within the furnace, as crucibles farthest from the furnace door (crucibles 1, 2, 5, and 6) exhibited the greatest amount of heat induced transformations with both the presence of the PHT shoulder, the PO₄³⁻ v3 shoulder at 1090 cm⁻¹, and elevated SF ratios (Fig. 6). These changes are not seen in the bones within crucibles near the door of the furnace, indicating that potentially temperatures varied within the same experimental trial (SI Figs. 50 and 51). This observation validates the selection of 550 °C as a temperature of interest, as it rests on a margin very close to temperatures capable of reaching full structural and chemical calcination in samples with prolonged heating. For all bones in this trial a small doublet peak representing the likely functional group of structural lattice carbonates, the CO₃²⁻ v3 peak area at 1400–1550 cm⁻¹, remains (Fig. 6). In an experimental reference collection utilizing the same FTIR-ATR instrument the full elimination of these carbonates was indeed not seen until temperatures above 900 °C are reached (Gallo et al. 2021). This indicates that despite the presence of heat induced changes to the mineral portion of the bone seen in the 550 °C, 48 hour trial (T1), a small amount of lattice carbonates remained (Gallo et al. 2021).

Summary of FTIR results

In sum, the results of the ATR-FTIR spectra for the short-term (10 minute duration) thermal alteration trials follow general expectations for each temperature threshold on the timing of heat induced stages to bone, dehydration, decomposition, inversion, and fusion. In the 10 minute trial, bones heated at 300 °C retain their organic components, at 550 °C have depleted organics and lattice carbonate, and at 750 °C have fully lost organics and the lattice carbonate (Figs. 7 and 8). With the 9 hour trial, the 300 °C trial results clearly indicate the further loss of organics, finally reaching depleted levels in range of 550 and 750 °C level of thermal alteration with 48 hours of prolonged heating (Fig. 7). Lattice carbonates are also further altered at 300 °C with long durations (9 and 48 hours), but do not reach the levels of 550 or 750 °C at any duration time (Fig. 8). This shows that low temperature prolonged heating of bone reaches the end of the decomposition stage and effectively the start of inversion stages without reaching higher temperatures.

Fig. 7

CO/P ratio, organized by duration and trial. Temperature and depth represented by color and shape, respectively. The 300 °C, 10 minute trials (T2A, T2B) have bones with the highest values, representing the strong presence of Amide I (collagen) within the sample, indicating the existence of organics. This signal is also visually observable through the strong brown color values from these samples (Figs. 1 and 3; SI Figs. 1–47). Bones heated at 300 °C for 9 hours have moderate values, while bones heated at 300 °C for 48 hours fall within similar ranges to the bones heated to 550 and 750 °C, transformations also visually recognized by the increasingly pale colors (Figs. 1 and 3; SI Figs. 1–47)

Fig. 8

Carbonate/phosphate (CP) ratio organized by duration and trial. Temperature and depth represented by color and shape, respectively. For the 10 minute trials, 300 °C (T2A, T2B) have the highest value, representing the presence of structural carbonates. Heated bone at 550 °C (T4) for 10 minutes still has detectable amounts of carbonate, whereas the 750 °C (T5) samples are completely depleted. When heating is extended for 9 hours, bones heated at 300 °C were measured at much lower C/P values than the 10 minute counterparts, representing the decreasing lattice carbonate content solely through duration. For the 550 °C, 9 hour (T3A/T3B) trials a moderate decrease is seen from 10 minute trials, and the 750 °C 9 hour (T8) samples exhibit the lowest values, as expected. Duration times of 48 hours demonstrate that the 300 °C trial (T7) values are closely aligned with the decreased presence of carbonates in the 300 °C 9 hour (T6) experiment, while the 550 °C heated bones are either at similar values to the 550 °C 9 hour (T3A, T3B) trial or lower values, representing the crucibles which experienced elevated temperatures

The crystallinity of bone heated at 300 °C for long durations (here 9 and 48 hours) does not increase, as seen by the stable SF ratios (Fig. 9). These values demonstrate that these 300 °C samples do not experience the final stages of inversion and fusion representing bioapatite crystal growth. They are therefore not calcined, despite pale coloration. The SF ratio does increase in the 550 °C and 750 °C trials at 9 hours when compared to their 10 minute counterparts (Fig. 9). Bone thermally altered for 48 hours at 550 °C show even larger SF values, as some begin the fusion stages of thermal alteration (Fig. 9). In the 750 °C for 48 hour trial, bones experience an expected decrease in SF ratio (Fig. 9). This is expected with thermal alteration, although usually seen at temperatures greater than 900 °C (Gallo et al. 2021), as the bone mineral lattice transforms to the extent that the SF ratio no longer represents the relative size of the bioapatite crystals. The decreasing ratio seen for prolonged heating at 750 °C therefore represents tremendous growth of the bioapatite crystals in the fusion phases of bone thermal alteration.

Fig. 9

Splitting factor (SF) organized by trial. Temperature and bone core depth represented by color and shape, respectively. For the 10 minute duration trials (T2A, T2B), bones burned to 300 °C have low infrared SF, as expected. At 10 minutes, 550 °C (T4) has a moderate value and 750 °C (T5) has a range of elevated values, all above the highest value of 550 °C. For the 9 hour trials, the SF of bones burnt at 300 °C (T6) remains consistent with the 10 minute experiment, and the 550 °C (T3A, T3B) values now fall within range of the 750 °C (T8) heated bones. This is likely due to the tremendous crystallite growth seen in this trial, and with coalescence and sintering of the crystallites the relationship the SF is measuring changes, a pattern noted by many studies which report falling SF values above 700–800 °C (Ellingham et al. 2016; Snoeck et al. 2014; Thompson et al. 2013; Lebon et al. 2010; Mamede et al. 2018). At 48 hours, the 300 °C trial (T7) heated bones again retains low crystallinity, whereas the 550 °C (T1) samples have bone cores with small crystallites with low crystallinity, and samples which experienced more heat and have larger crystallite growth and therefore are more crystalline with a higher SF value

SF values were further tested between differences of sediment (gravel or compacted gravel with fine quartz sand) and placement within the crucible (surface or buried at − 2 cm) for Trial 1. We first tested for normality using Shapiro–Wilk tests and the data was plotted to visually assess the distributions. T-tests were further used to evaluate both categories, and both showed that there is insufficient evidence to reject a null hypothesis of no difference for both variables (p = 0.98 for sediment, p = 0.27 for placement).

The presence of the heat-induced shoulder, measured with the PHT ratio, on the PO₄³⁻ v3 peak at 1090 cm⁻¹ is seen in both 750 °C trials as well as with a weak expression in all 550 °C/9 hour trials (T3A and T3B). For the 48 hour at 550 °C trial (T1), this peak at 1090 cm⁻¹ is seen fully present in the crucibles located at the back of the furnace (crucibles 1, 2, 5, and 6), with weak expression in the front crucibles near the furnace door (crucibles 3, 4, 7, and 8; Fig. 6; SI Figs. 50 and 51). The PHT ratio unsurprisingly highest in both 750 °C trials (Fig. 10). The combination of PHT and SF ratios supports the inference that the 750 °C/48 hour bone samples have very large crystal growth, as despite the lower SF ratio the PHT is unaffected and remains high (Fig. 11). Utilizing SF ratios in combination with other peak transformations related to high temperature alteration can assist with interpretations which would be difficult to assess using the SF ratio alone.

Fig. 10

Phosphate high temperature (PHT) ratio, organized by duration and trial. Temperature and depth represented by color and shape, respectively. Only trials with PHT shoulder presence represented: 10 minutes at 750 °C (T5), 9 hours at 550 °C in gravel sediment (T3A), 9 hours at 550 °C in compacted mixed sediment (T3B), 9 hours at 750 °C (T8), 48 hours at 550 °C in both gravel and compacted mixed sediment (T1)

Fig. 11

PHT and SF ratios for all trials with 625 cm⁻¹ shoulder present in FTIR-ATR spectra (T1, T3A, T3B, T5, T8). The largest SF values align with the largest PHT peak ratio, as expected with high temperatures. Note the large PHT ratio seen in the 750 °C 9 hour trial (T8), where the infrared SF value is decreased due to fusion stage thermal alteration but the PHT remains clearly expressed

XRD results

The average crystallite sizes as measured by XRD provide an independent calculation of average bioapatite crystallite growth, and therefore, heat-induced transformation. Results align closely with inferences taken from the ATR-FTIR spectroscopic data, with samples heated for 9 and 48 hours at 300° not experiencing growth in average crystallite sizes beyond reference samples burned either for 30 minutes or 1 hour and 30 minutes (Fig. 12, SI Figs. 52 and 53).

Fig. 12

XRD results of averaged crystallite size (nm) for sampled bone samples both from duration trials and from an experimental reference collection 200–1200 °C (Gallo et al. 2021). Strong overlap is seen at 300 °C between experimental trials at both temperatures and reference data. Reference collection duration times were primarily 30 minutes, with two sets of 300 and 700 °C samples cooled and reheated for an additional 30 minutes for no greater than 1 hour and 30 minutes of heating time in total. Average crystallite sizes of bone which has been chemically and structurally calcined can be inferred from this dataset to be of an average size >  ~ 30 nm, with sizes >  ~ 80 nm representing heated bone which has undergone the fusion stages of bioapatite sintering fully

Samples from the 550 °C trials at 9 hour also have average crystallite sizes similar to 30 minutes and 1 hour 30 minutes reference counterparts (Gallo et al. 2021; Fig. 12). With 48 hours, 550 °C samples show a range of crystallite sizes bridging the divide between bones which are and are not calcined at 30 minutes or 1 hour and 30 minutes (Gallo et al. 2021; Fig. 12). This also mirrors the ATR-FTIR spectral inferences, and samples with large crystal sizes do match the same samples which have high SF and PHT values from these trials.

Drastic crystallite growth is seen for 9 hours at 750 °C (T8) samples (Fig. 12). These large bioapatite minerals clearly show separation from bones burnt at lower temperatures both at the same duration and longer durations (Fig. 12). Bone cores burnt under these conditions demonstrate average crystallite sizes larger than twice the size of referential samples burned at 700 °C and even 1000 °C (Gallo et al. 2021). Such increase in crystallite size but without full elimination of the lattice carbonate functional group as seen in the ATR-FTIR spectra may indicate long duration heating at temperatures above calcination but where the carbonates have been completely removed as seen from reference spectra heated at 30 minutes and 1 hour and 30 minutes (Gallo et al. 2021). This observation indicates that long duration fires at high temperatures can likely be recognized archaeologically, despite all calcined samples not having any macroscopic coloration differences to distinguish them.

Discussion

The duration of an anthropogenic fire is deeply relevant and interconnected to fire function, construction, and maintenance. Additionally, as grass and brush wildfires are predominately rapid events, prolonged thermal alteration in bone can be a variable useful for assessing between likely natural or anthropogenic fire traces in certain environments (Buenger 2003; Gowlett et al. 2017; Table 1). Relative duration lengths in this way can provide a crucial lens for interpreting and understanding the properties of ancient fire which impacted heat altered bone in archaeological assemblages. This study explicitly includes heating length as an experimental variable to evaluate the degree of transformation in bone material at different temperatures. We find that by using the threshold of structural and chemical calcination, detected with spectroscopic and radiometric methods (ATR-FTIR, XRD), and macroscopic observations of coloration, duration of heating does impact properties of cortical bone in ways that can be measured and applied to the archaeological record, and are not seen in experimental studies using heating times around 3 hours or less (Table 4; Fig. 13).

Fig. 13

Compressed schematic summarizing the duration results of this study organized by temperature. Observations under 3 hours in the left blue window are well documented in studies characterizing burned bone under 3 hours (Table 4) and align well with the timing of burning stages detected by this study in the 10 minute trials. These prior works also have not noted any considerable impact of duration up to approximately 3 hours on the final properties of heated bone. Heat-induced transformations that do distinguish the impact of prolonged heating at 9 and 48 hours from the 10 minute counterparts are outlined for each temperature category in the pink shaded area. ATR-FTIR spectra monitored chemical and structural ratios which represented general heat induced transformations, whereas XRD provided the average crystallite sizes

The results of this study indicate that the timing of alterations with prolonged heat can be distinguished at different temperature thresholds (Fig. 13). Long-duration heating in air depletes the combusted organic content of bone at lower temperatures (here seen in samples heated at 300 °C for 9 and 48 hours) and therefore changes the visible color (Fig. 1 and 3, SI Figs. 1–47). These long duration but low temperature bones potentially would be evaluated by archaeologists to be calcined based on visual cues alone. The depigmentation and organic depletion from prolonged heat duration, and not calcination, can be observed here through the ATR-FTIR CO/P ratio (Fig. 7). The lowering CO/P ratio for bones heated at 300 °C 9 and 48 hours is seen to correlate with the loss of pigmentation as the combusted organic char is eliminated. Therefore, the recognition of thermally altered bones which are muted white, grey, and tan in color as represented from the Munsell color schema therefore are not always directly representative of high temperatures and calcination, but may also result low temperatures at long durations.

The average crystallite sizes monitored by XRD support the ATR-FTIR data, and can distinguish samples burned for 10 minutes and 9 hours at 750 °C (Figs. 12 and 13). The samples burned for 9 hours at 750 °C even experience crystal growth not seen at temperatures 700–1200 °C burned for either 30 minutes to 1 hour and 30 minutes in a referential sample (Fig. 12).

To summarize, with prolonged heating of 9 hours, bone heated at 300 °C shows less organic content than the 10 minute counterparts, correlating with the CO/P ratio, and also experiences a loss in structural carbonate, represented by the CP ratio (Figs. 7, 8, and 13). At 550 °C for 9 hours, bone also loses structural carbonate that is not seen in briefer heating events at the same temperature (Figs. 8 and 13). The organic component of these bones has already been depleted when brought to temperatures above 300 °C (Fig. 7). At a duration of 9 hours, bones thermally altered at 750 °C experience a tremendous growth in average crystallite sizes, more than double the sizes of bones kept at 750 °C for 10 minutes (Figs. 12 and 13).

At 48 hours of thermal exposure at 300 °C, bones are pale in color as any remaining organic content has been eliminated, but they are not calcined (Figs. 7 and 13). For 48 hours at 550 °C, some bones do become structurally and chemically calcined, although this is not seen uniformly throughout all samples, indicating that this trial temperature and duration is very close to the energetic conditions necessary to for bones to calcine (Figs. 7, 8, 9, 10, 12, and 13).

Our results also indicate that these observations can be used to recognize archaeological bone which has experienced prolonged heating. This is as all metrics considered here can be discerned from aspects of diagenesis, including crystallinity growth. Diagenetic crystal growth cannot reach the sizes measured by XRD for the prolonged heating at 750 °C, and diagenesis cannot cause the heat induced transformations what are seen to accompany calcination like the PHT peak visible in ATR-FTIR spectra.

Limitations predicted in detecting prolonged heating in archaeological bone do include issues of equifinality with different combinations of temperatures and heating lengths, as well as the impact of exogenous carbonate inclusions. This may make it challenging to assign specific temperatures or exact durations in archaeological settings. Despite this uncertainty, research investigating thermally altered archaeological bone can certainly investigate relative thresholds of duration times when discrepancies in pigmentation and bone structure and chemistry are seen, i.e., pale white or tan bones which are not calcined. Prolonged heating within bone samples which are calcined can also use the exponential crystallite growth documented in this study, as the average crystallite sizes are above what is possible to be reached in natural settings at short durations (Fig. 12). These observations can contribute valuable data even if it is not possible to discern with great confidence the exact duration length.

For integration into zooarchaeological methodology, we therefore recommend conducting spectroscopic testing primarily with ATR-FTIR on white, pale tan, and grey archaeological burned bone classified as calcined based on the macroscopic color evaluations. The purpose of this screening is to test for true signals of structural and chemical calcination: the presence of high temperature dependent peaks, high calculated SF ratio, little to no organic content. If they are not calcined, this would be a strong indication of depigmentation from long duration heating at temperature below the calcination threshold. Alternative scenarios which may result in light coloration of archaeological bone without burning is limited to processes such as sun bleaching, but these specimens can be distinguished from bones exposed to heat for long durations by differences in surface texture (Behrensmeyer 1978; Dupras and Schultz 2014; Shipman 1989).

Subsequent to the spectroscopic testing, we suggest XRD analyses on the verified calcined archaeological bones to calculate and compare to average crystallite sizes of referential assemblages. This is to infer the extent of the bioapatite transformations and to detect heating at high temperatures for extended time periods. Sampling strategies will likely fluctuate considering the different parameters and conditions of the archaeological assemblage, but large sampling procedures are made reasonable by the relatively low cost, minimal sample material, and minimal preparation needed for these analyses. The average crystallite sizes measured with XRD in this study for samples burned at 750 °C for 9 hours indicates how duration length of archaeological samples which would all be considered to fall within the same category of burning based on visual identification, calcination, can be distinguished. This would include all bones identified as Stage 6 following the Stiner et al. (1995) burning scale.

Future research, including work done in cremation studies, can provide further context on the threshold of these structural and chemical changes. Studies can also address the internal variation of temperature distribution in heating instruments to investigate internal variances in samples tested within the same experimental conditions. Work done to more precisely illustrate the timing of long-duration heat exposure at lower temperatures considering organic depletion and color changes, and at high temperatures measuring extreme crystallite growth, can contribute to the development prediction models to address the issues of equifinality and uncertainty for archaeological detection and interpretation.

The results of this study of short-term heating (10 minutes) on cortical bone tissue are found to be consistent with prior work characterizing heat alteration to bone and the timing of the stages of transformations (Table 4). The results of the trials investigating prolonged heat exposure do present new evidence for colorimetric studies and provide context for discrepancies within the burned bone literature. Calcination based on visible bone coloration alone in prior experiments (which did not reach temperatures above 700 °C and did not conduct spectroscopic analyses) potentially indicate bone which was depigmented due to depletion of organics achieved through prolonged duration of heating and not true structural and chemical calcination. Much like recent perspectives on colorimetric approaches to identifying burnt bone, this study therefore also recommends coloration in a supporting role to the spectroscopic and chemical investigations (Bradfield 2018; Krap et al. 2019).

For application of results to anthropological questions of hominin fire using behavior we recommend the inclusion of the zooarchaeological approach outlined here as a component of the microcontextual approach for investigating anthropogenic fire. This framework evaluates multiple independent lines of evidence for thermal alteration and crucially considers information within micromorphological thin sections. The detection of prolonged heating in bones cannot confidently identify anthropogenic fire alone, but when considered alongside their depositional context can provide valuable data on minimal duration times and maximum temperatures for integrated studies considering hominin pyrotechnology.

Conclusions

Duration times of anthropogenic fire are integrally related to the social, economic, and technological behaviors most important to understanding hominin fire use and control. Results from these experiments suggest that prolonged heating does impact and transform bone in distinguishable ways, as monitored with ATR-FTIR and XRD, from shorter duration counterparts at the same temperature. Thermally altered bone therefore provides a minimal measure of the length of a fire event. The identification of long duration fire impacted bones in archaeological settings can also help researchers more confidently distinguish between wildfire events and combustion features of variable duration within a microcontextual framework.

This present study intersects with study of bone as an indirect fire proxy and the greater zooarchaeological recognition of calcination. Calcined bone is a reliable source of inorganic C-14 for radiocarbon dating as the large bioapatite crystals of true calcined bone provide protections from diagenetic substitutions of carbon (Lanting et al. 2001; Zazzo and Saliege 2011). Our results call for caution in solely macroscopic identification of calcination, primarily white and light coloration. This is exemplified by the light tan and muted grey color of bones produced by prolonged heating of 48 hours at 300 °C. Bone which is muted in color and can be recognized as heated or burnt through other visible cues (e.g., surface texture, weight) but without the chemical or structural signatures of high heat alteration and calcination can therefore be explored further as examples of prolonged heating at low temperatures. Bone which is also recognized as white in coloration but has average crystallite sizes greater than 150 nm also can likely indicate prolonged heating at higher temperatures. Methodological recommendations resulting from this study are therefore to systematically use both color classifications and investigations into structural and chemical compositions to sample thermally altered faunal assemblages. This approach can combine the ease of analysis of traditional zooarchaeological methodologies, which first classify burnt bone based on macroscopic observations, alongside a sampling strategy of spectroscopic and radiometric methods.

Alongside the data presented here, future research on prolonged heating must also consider the starting condition of bone (i.e., fresh or weathered), different diagenetic conditions, and different types of bone tissue and species variability to fully understand the material properties of thermally altered bone and the full potential of this indirect fire proxy to extrapolate behaviors of fire using hominins.