A hypothesis of sudden body fluid vaporization in the 79 AD victims of Vesuvius

In AD 79 the town of Herculaneum was suddenly hit and overwhelmed by volcanic ash-avalanches that killed all its remaining residents, as also occurred in Pompeii and other settlements as far as 20 kilometers from Vesuvius. New investigations on the victims' skeletons unearthed from the ash deposit filling 12 waterfront chambers have now revealed widespread preservation of atypical red and black mineral residues encrusting the bones, which also impregnate the ash filling the intracranial cavity and the ash-bed encasing the skeletons. Here we show the unique detection of large amounts of iron and iron oxides from such residues, as revealed by inductively coupled plasma mass spectrometry and Raman microspectroscopy, thought to be the final products of heme iron upon thermal decomposition. The extraordinarily rare preservation of significant putative evidence of hemoprotein thermal degradation from the eruption victims strongly suggests the rapid vaporization of body fluids and soft tissues of people at death due to exposure to extreme heat.


Introduction
Vesuvius is an active volcano situated about 12 kilometers from Naples, one of the metropolitan cities at highest risk in the world, with its population of more than three million [1]. Archaeological and volcanological site evidence show that Vesuvius tends to have a major (Plinian) eruption at least every 2,000 years [2][3][4]. In AD 79 a sudden Plinian event with subsequent volcanic pumice fallout and ash-avalanches affected an extensive area, causing total devastation and thousands of victims [5]. The initial fallout phase (pumice air-fall phase), driven by the dominant southerly and south easterly winds [6], was dispersed up to a distance of about 70 kilometers [7]. The later pyroclastic surges and flows (rapid gravity-driven currents of volcanic ash and hot gases generated by the collapse of the Plinian eruptive column) reached up to 30 kilometers northwest and west of Vesuvius [1,6,8]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 In the early phase of the eruption the first fatalities occurred in Pompeii as a result of roofs and floors collapsing due to pumice accumulation [9,10]. In the next hours, the remaining inhabitants of Herculaneum (ca. 4-5,000) [11,12] and Pompeii (ca. 20,000) [13,14], as well as those from nearby settlements and villas (e.g. Villa B at Oplontis) [15][16][17], who were not able to evacuate in time, were overwhelmed by the hot surge clouds [18,19].
At Herculaneum, 300 people who had taken refuge in 12 waterfront chambers along the beach (S1 Fig) were suddenly engulfed by the abrupt collapse of the rapidly advancing first pyroclastic surge (S1) [8] (S2 Fig). In just a few hours the towns of Herculaneum, Pompeii and Stabiae, situated respectively about 7, 10 and 16 kilometers from the vent, were definitively buried by subsequent pyroclastic currents [6], whose eruptive deposits reached a maximum thickness of 20 meters [8,[20][21][22]. In this area, the archaeological investigations of the last three centuries brought to light several Roman settlements and hundreds of human victims, even at a distance of 20 kilometers as far as Stabiae and close suburban villas in Gragnano [19,21,22].

Archaeological and osteological context
New excavations started in the 80s after the casual discovery of human remains in the suburban area of Herculaneum. On the beach and in 6 of the 12 waterfront chambers were uncovered approximately 140 victims of the eruption [23]. These skeletons, initially recovered and studied by Sara Bisel [24][25][26], were later the subject of several bioanthropological studies [27][28][29][30][31][32][33][34][35]. Further archaeological investigations conducted in the early 90s in 6 of the chambers not yet excavated brought to light an additional large group of human victims, left untouched within the ash surge deposit [36]. In the second half of the 90s, these skeletal remains were the subject of a valorization project through the making of fiberglass casts [37,38]. The latter were later placed in the original context of discovery, after archaeological investigation and final recovery of the victims' skeletons leaded by one of the authors (P. Petrone) [38,39]. The joint bioarchaeological and taphonomic study of the skeletons prior their removal allowed the investigators to verify the interactions between victims' bodies and the volcanic ash deposit. The resulting new site evidence combined with those from laboratory bones analysis allowed to obtain new information on the causes of death and the heat-induced effects on people [40][41][42][43]. Further bioanthropological and paleopathological studies were also carried out on the same skeletal sample [44][45][46][47].

Volcanological context
Recent global eruptions show that pyroclastic density currents are the greatest threat to life [48][49][50][51][52] and thus the dominant hazard in densely populated areas [53]. A dilute pyroclastic density current or surge is typically an intensely hot (200-500˚C) fast-moving cloud (100 to 300 km/hr) of fine ash, in an environment being low in free oxygen content and rich of superheated steam and other volcanic gases [6,48]. In such conditions survival is likely to be impossible, particularly due to the intense heat reached in areas closer to the vent [54]. In pyroclastic density currents, thermal injury may be at least as important as asphyxia in causing immediate death [48].
In the main body of a proximal surge temperatures may be as high as 400-500˚C, whereas in distal regions temperatures of 200-300˚C are more common [48,54]. These temperatures are analogous to those of the 79 AD pyroclastic surges that hit first Herculaneum and later Pompeii, as determined with various methods, including TRM on lithic clasts [55][56][57][58], bone analysis vs heating experiments [42], and charcoal reflectance [59].
Pyroclastic surge clouds are responsible for emplacement of the largest and widespread ash deposit in the suburban area of Herculaneum [6,8]. Rapid deposition of extremely fine-grained ash into thermally stratified volcanic deposit [60] on the beach and within the boat-chambers, and abrupt entrapment and burial of the victims by the hot ash surge [6,8] could have protected them from being bioturbated [61], thus resulting in the exceptional preservation of fully articulated skeletons in the last vital posture [5,39,40,48].

Heat effects and causes of death
The effects of the eruption on the inhabitants of Herculaneum and Pompeii as well as on the people living in the other urban and suburban settlements around the volcano have been the subject of several studies [5,9,14,19,21,27,28,[39][40][41][42]48]. Apart from the casualties occurred in the initial pumices fallout phase by buildings collapse in Pompeii [10,19,21], studies on the causes of death are mostly referred to the effects of heat associated with the pyroclastic surge clouds emplacement in both Herculaneum [41][42][43]49] and Pompeii [14,48].
Although the heat of the ash surges has been mostly accepted as a major cause of mass mortality in the 79 AD eruption, there are some differences in interpretation depending on the distance from the volcano and, within the same site, on the place where victims were found. As regards Herculaneum, more recent studies agree on the rapid death of people discovered on the sea shore area [34,[39][40][41][42][43], but some authors hypothesized a gradient of heat-induced effects. So, even if nearly every skeleton had some evidence of bone thermal exposure (changes in color, charring, fracturing) [28,34,40,42], the few victims found on the beach were assumed to show greater thermal effects compared to those sheltered inside the chambers [27,34]. Based on the previous assumption, it was also hypothesized that death was instantaneous only for people found on the beach, while those refugees in the chambers would have died from asphyxiation [27,28]. In fact, a comparative analysis of the full skeletal sample has not yet been achieved, being the victims' samples from different excavation surveys (conducted in the 80s by S. Bisel, and in 1997-1999 by P. Petrone) [26,39] studied separately. With regard to previous interpretations on the causes of death, also at a greater distance as in Pompeii, death by asphyxiation in both pumices fallout and pyroclastic surge phases has long remained the most accredited hypothesis [9,13,14,21,62].
The latter accounts are at variance with a first forensic interpretation concerning the victims found in the surge deposit [48], as well as with more recent multidisciplinary studies. Taphonomic, bioanthropological and volcanological site investigations and laboratory evidence [40,42], coupled with results from heating experiments on recent human bone samples [41], have shown that the Herculaneum residents were instantly killed by the extreme high temperature of the emplacing S1 surge, although it was previously believed that death had occurred by slow suffocation from ash inhalation. Skull and bone charring and cracking, as well as instant hand and foot contraction (flexor reflex by the nociceptive C fibers) [63] and spine hyperextension, have been described as thermally induced major effects on the victims' skeletons unearthed from both the beach and the sea-front chambers [27,28,34,40,42,43]. Histological and ultra-structural investigation have revealed linear and polygonal cracking of the intra-and inter-osteonic structure associated with incipient recrystallization (S3 Fig), bone changes typically induced by heat [64,65]. Evidence of sudden death is provided by the victims' corpses, appearing to be "frozen" in the last vital action (life-like stance) (Fig 1). The lack of voluntary self-protective reaction or agony indicates that any vital activity had to stop within a time shorter than the conscious reaction time, a state known as fulminant shock [66]. The widespread occurrence of life-like stance has been found consistent with cadaveric spasm, a rare but diagnostic form of instantaneous muscular stiffening (instant rigor), induced by instant thermal coagulation in victims from pyroclastic currents [48], which crystallizes the last vital activity prior to death [67,68]. As also observed in the other sites buried by the 79 AD eruption, the overall evidence including the predominance of life-like stance suggests the occurrence of thermally-induced instant death of the inhabitants in the Vesuvius area up to at least 20 kilometers from the vent [5,[40][41][42][43]48].
A thorough investigation on the Herculaneum victims' skeletons and their ash (hereafter, the term "ash" is always referred to volcanic ash) burial context revealed the preservation of atypical mineral residues encrusting the bones, which also impregnate the ash filling the skulls and the ash-bed. The iron and iron oxides amounts detected by inductively coupled plasma mass spectrometry and Raman microspectroscopy suggest such residues to be the final products of heme iron upon thermal decomposition. The significant putative evidence of hemoprotein thermal degradation and additional evidence of heat-induced effects seem to suggest the rapid vaporization of body fluids and soft tissues of victims resulting from exposure to the extreme high temperature of the ash-avalanches.

Materials and methods
The archaeological site of Herculaneum is placed at the foot of Vesuvius, at about six kilometers from metropolitan Naples. In 1997-1999, a large group of about 80 skeletons of victims of the 79 AD Vesuvius eruption were recovered from the ash surge deposit inside chambers 5, 10, 11, and 12 on the seafront area of the town [37][38][39][40][41][42]. Two subsequent archaeological excavation campaigns were conducted by one of the authors (P. Petrone), in collaboration with the director of the Herculaneum archaeological site (M. Pagano, Superintendence of Pompeii). The skeletal remains were carefully examined in the laboratory for the presence of certain mineral  Table). The child's corpse displays flexure only of the upper limbs, indicative of an incipient "pugilistic attitude". Full exhibit of this heat-induced stance is never found in the victims' corpses discovered at Herculaneum. residues, observed for the first time during the archaeological survey, which impregnated the ash burial deposit or encrusted the bones. Bones and ash samples were observed with a 10x-30x magnifying glass. Particular attention was also given to the detection of mineral residues from the ash filling the skulls. In order to avoid potential taphonomic complications, such residues were sampled only from those encrustations that affected the bones and the ash layers not at direct contact with metal artifacts like coins, rings and other types of personal objects found close to the victims [69]. As regards demographics and place of discovery of each skeleton, sex, age at death and chamber numbering are reported in S1 Table (see also [41]). Type and source of each of the 103 samples detected for iron content (Table 1) are specified in S2 Table. All necessary permits were obtained for the study of the Herculaneum specimen, which was approved

Inductively coupled plasma mass spectrometry (ICP-MS)
A selection was made to investigate the presence of iron [70] from 103 different archaeological samples. The samples collected were differentiated by the presence of red or black residuals. Each sample (10 mg) was digested in acid in a Teflon vessel in a microwave oven (Milestone Ethos 900-Mega II). Digestion was obtained by adding a mixture of 2 mL of 67% HNO3 and 4 mL of 37% HCl. HNO3 and HCl were Super Purity Solvent grade from Romil, Cambridge, UK. Acidic mineralization was achieved with the following microwave oven program: 20 min to reach 220˚C at 1400 W; 15 min at 220˚C and 1400 W; ventilation for 30 min. The solution was then quantitatively transferred into polystyrene liners and stored at 4˚C until ICP-MS analysis was performed. The analyses were carried out in triplicate on an Agilent 7700 ICP-MS, equipped with a frequency-matching RF generator and 3rd generation Octopole Reaction System (ORS), operating with helium as cell gas on diluted samples (1:10 v/v Milli-Q water). The parameters were set as follows: radiofrequency power 1550 W, plasma gas flow 14 L min-1; carrier gas flow 0.99 L min-1; He gas flow 4.3 mL min-1. The Octopole Reaction System was activated to improve metal quantification because of the interferences by polyatomic species produced by a combination of isotopes from plasma, reagents and matrix. Multi-element calibration standards were prepared in 5% HNO3 at four different concentrations (1, 10, 50, and 100 μg L−1). The standard addition approach for calibration on four concentration levels was used in order to keep the matrix-induced variations to a minimum. At least three replicates of each calibration standard were run. Moreover, in order to correct possible instrumental drifts, 103Rh was used as an internal standard (final concentration: 50 μg L−1). The error in the determination of the iron amount within the samples is within 10%.

Raman microspectroscopy
Of the 103 samples, 22 were investigated by Raman microspectroscopy in order to detect, quantify and discriminate the possible preservation of heme and heme degradation products [71]. Raman analysis of these samples selected on the basis of iron content data analyzed by ICP-MS was performed to identify or exclude various species containing iron and identify other non-ferrous species. A confocal Raman microscope (Jasco, NRS-3100) was used to obtain Raman spectra. The 514 nm line of an air-cooled Ar + laser (Melles Griot, 35 LAP431 220) or a 647 nm line of a water-cooled Kr+ laser (Coherent) was injected into an integrated Olympus microscope and focused to a spot diameter of approximately 3 μm by a 20x objective with a final 4 mW power at the sample. A holographic notch filter was used to reject the excitation laser line. Raman backscattering was collected using a diffraction lattice of 1200 grooves/ mm and 0.01-0.20 mm slits, corresponding to an average spectral resolution up to 1 cm -1 . Typically, it took 60 s to collect a complete dataset from a Peltier-cooled 1024x128 pixel CCD photon detector (Andor DU401BVI). Raman measurements were finally triplicated for the purpose of reproducibility for each spot sampled. Wavelength calibration was performed by using cyclohexane as a standard.

Proteomic analyses
Samples were treated in heterogeneous phase with different pre-treatments (either incubation with 6M urea, or extraction with TFA 0.1%, acetonitrile 10%, or extraction with RIPA buffer, or extraction with CH 3 CL 3 /CH 3 OH [6: 3; v/v]). This phase was followed by enzymatic digestion with trypsin at 37˚C for 16 hours, purification using a reverse-phase C18 Zip Tip pipette tip (Millipore), and nano LC-MS/MS analysis on a CHIP MS 6520 QTOF equipped with a capillary 1200 HPLC system and a chip cube (Agilent Technologies, Palo Alto, CA) [72]. Raw data were used for protein identification with a licensed version of MASCOT software (www. matrixscience.com) version 2.4. with 10 ppm MS tolerance and 0.6 Da MS/MS tolerance; peptide charge from +2 to +3. No fixed chemical modification was inserted, but possible oxidation of methionines, deamidation at asparagines and glutamines, and the addition of hydroxylation on prolines and lysines were considered as variable modifications to query the SwissProt database, with Homo sapiens as a taxonomy restriction.

Results
In the present work, in the wake of previous taphonomic, bio-anthropological and stratigraphic surveys within the seafront chambers on the beach of Herculaneum, the victims' skeletons and their ash burial context (S1 ash surge deposit) are examined in depth. The initial site investigation allowed to detect several features related to the sudden impact of the hot S1 ash surge on the people such as the life-like stance of bodies, the hyperflexion of the extremities, the spine hyperextension and the skeletons floating within the ash deposit. Moreover, the preliminary analysis of the bones highlighted some features typical of exposure to high temperature like color changes, carbonization and cracking.
Here we describe key evidence of body exposure to extreme heat provided by recurrent skull explosion (Fig 2), as detected by clear-cut fractures (Fig 2A), whose margins are sharp like those seen in cremated bones. In several cases, fracture lines radiate from a common center, thus showing a "stellate" appearance ( Fig 2B). A single skull may also be affected by multiple fracture centers. Dark staining is typically associated with these cracked areas, whose exposed surface is charred as well (Fig 2C). Spotted areas of dark stains affecting the outer bone can also be found in intact skulls (Fig 3B and 3C). Fractures are always limited to the charred bone areas.
A recurrent feature is the dark staining of the inner table beside a zone of unchanged color, with a distinct blackened/non-blackened pattern (Fig 2C), often associated with a typical effect of dark staining "exuding" from sutures. This feature is apparent close to the bone openings such as non-fused sutures (Fig 3C), orbits and the acoustic meatus (Fig 3B), the squamous part of the parietal bone ( Fig 3B, panel 1 and 3C), the foramen magnum, and the margins of the skullcap fractures as well (Fig 2A, 2B and 2C). In some cases dark staining of the intracranial cavity appears to be gradual, the bone progressively appearing natural/pale yellow (α), bright brown (β), dark brown and black (γ) (Fig 4A and 4B) [73]. Pale yellow is indicative of minor thermal exposure ( 200˚C), whereas darker bone coloration matches with higher temperatures (300 to 400-500˚C) [41,64,[74][75][76].
The careful examination of the skulls first on the site and then in the laboratory showed that the presence of blackened bone near bone of unchanged color depends on the victim's posture at death: whatever the orientation of the head, the color of the downward-facing bone stays unchanged, whereas the upward-facing bone is mostly blackened (Fig 3C, panel 1). This evidence is concomitant with the pattern of bone cracking, since dark stained areas often exhibit fissure fractures, or may even be exploded, while those of unchanged color are never affected by such major heat effects. A substantial feature associated with the charring of the inner cranial table is the brown coloration of the vascular grooves (sulci arteriosi and sulci venarum) (Fig 4C). In general, as detected from the death assemblage, victims' skeletons from less crowded chambers show greater heat effects (i.e., bone charring and cracking, skull explosion) than those where people were crammed into the available space (S1C Fig). An extraordinary find concerns skulls filled with ash, which indicates that after evaporation of the organic liquids the brain was replaced by ash (Fig 5). The presence of such an ash cast in all victims, even those showing minor heat effects, provides evidence that the S1 surge was sufficiently hot and fluid to penetrate the intracranial cavity soon after soft tissues and organic fluids disappeared, as supported by palaeomagnetic thermal measurements [56] and intracranial micro stratigraphy, the latter showing lamination by successive apposition of thin ash layers (Fig 6D).
New examination of the victims' skeletons and the taphonomic context now reveals the diffuse preservation of atypical red mineral residues encrusting the bones (Fig 7), which also impregnate the ash surge-bed below and around the corpses (Fig 8A and 8B), as well as the sand covering the chamber floor below the surge deposit ( Fig 8C). Red mineral residues were also found to encrust both the skullcap (Fig 3A) and the inner table (Fig 3A, panel 1), and impregnate the ash filling the intracranial space as well (Fig 6A), mostly in the form of spotted  11, chamber 12). B. Skullcap affected by a "stellate" fracture consisting of several cracks which radiate from a common center, characterized by charred outer margins (adult male, ind. 31, chamber 10). C. Exploded skull showing a partly dark stained inner table (right side, adult male, ind. 6, chamber 12); charring of the fractured margins is evident (white arrows) (bar scale 10 cm). The skull of this victim was lying in the ash bed on its left side.
https://doi.org/10.1371/journal.pone.0203210.g002 inflorescence-shaped aggregates (Fig 6C). The above red residues were found in adults, children and some surprisingly well-preserved fetal skeletal remains (Fig 9), in the last case encrusting as well the adult pelvis bones from which they were unearthed. Similarly, black mineral residues may encrust both the outer and inner table (Fig 3B, panel 1 and 3C, panel 1), or laminate the ash filling the intracranial space (Fig 6B).
On the basis of color and location, the mineral residues were classified as: i. red incrustations; ii. black incrustations; iii. ash or sand impregnated by red residues; iv. ash or sand impregnated by black residues. In order to identify their chemical nature, we carried out a multi-element analysis [77,78] by means of inductively coupled plasma mass spectrometry (ICP-MS) [70,79]. We selected 103 samples (S2 Table) from the mineral residues either encrusting the bones and the intracranial space ranging in color from light red to dark red and black (Fig 7), or impregnating both the ash bed layers in which the skeletons were embedded and the sand bottom. For comparison, additional samples of ash from the ash-bed deposit and the sand bottom not in direct contact with the skeletons were also analyzed.
We further investigated this question by searching for the possible presence of heme-containing protein residues by means of protein approaches. The red incrustation areas from several iron-containing samples were submitted to trypsin digestion after different extraction procedures, and the putative protein digests were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) [72]. No peptides originating from tissue proteins could be detected in the samples. However, although sampling did not notch into the bone but only the incrustation was taken, in several cases we were able to detect a few reliable peptides from human collagen proteins alpha-1(I) and alpha-2(I) (data from sample C5 are reported in S3 Table as an example). These results indicate that only proteins protected by the inorganic scaffold of the bones survived the unbearable temperature experienced, while proteins unprotected by the hard tissues were totally destroyed.
We then addressed the matter by Raman investigation on selected and representative samples in order to identify or exclude various iron-containing compounds, and identify other possible non-ferrous species. In particular, we sought to detect and discriminate the possible preservation of heme and heme-related degradation products within the red and black mineral residues. In all, 22 samples were selected on the basis of iron content data provided by ICP-MS and analyzed using Raman microspectroscopy with 514 and 647 nm excitation (Table 2). Microscopically heterogeneous residues showing spots of different colors (13,22,46,49,51,75 and 92) were sampled several times in different areas, typically red, dark reflecting and grey  (adult male, ind. 9, chamber 12). B. The inner bone surface progressively changing from pale yellow to black (α-γ). C. Brown residues encrusting the vascular grooves (scale bar in cm). Bone colors are based on the Munsell soil color chart [73].
The possible implications of the above Raman results to the fluid-body degradation is below discussed. Regarding the origin of the observed iron oxides (from mineral or from body fluids), indeed we have no direct indication. We can only find indirect evidence of compatibility with an organic origin.
Heme degradation has been extensively studied in the temperature range of meat cooking (70-100˚C). Already at such temperatures heme partially degrades, releasing iron without any porphyrin cleavage [86][87][88]. At higher temperature (400˚C), iron-free porphyrins undergo a two-stage degradation of the methylene bridges [89] and metallo-porphyrins also undergo cleavage reaction [90], generating anhydrous metal oxides as final products [91].
So, the expected heme-degradation products from victims of Vesuvius are only iron oxides. In our study hematite was both observed alone in some samples (C22c, C46b, C92c) and in combination with other iron oxides (e.g. magnetite in C51b). It is worth reminding that multiple mechanisms of interconversion of iron oxides occur as a function of temperature and oxidizing environment [92], so heme degradation is compatible with observations of multiple iron oxides. Furthermore, the coexistence of amorphous carbon (most likely product of organic combustion) and iron oxides with broad bands (thus not very crystalline) on the same  12). B. Exploded skull of a child filled by ash, from one of the rare skeletons preserved from the early 1900s' excavations of the town (ind. A, weaver's house). C. Perfect replica of the brain shape is apparent from the imprint of the coronal and sagittal open sutures (C). As also detected in the victims discovered in the sea-front area, the ash is impregnated by red mineral residues (B, C, white arrows), which also encrust the exposed bone margins (B, C, black arrows). Some reddish-brown stained areas affect the outer bone surface (B, right supraorbital region) (scale bar in cm).
Recent multidisciplinary research on the lethal effects of the pyroclastic surges induced by the 79 AD eruption in the Vesuvius area showed that in the vicinity of Pompeii heat was the main cause of death of those who had previously been thought to have died of ash suffocation [41]. This was also posited for the victims of Herculaneum, specifically for those who had taken refuge in waterfront chambers along the beach and were then sheltered from direct mechanical impact, but not from heat of the emplacing S1 surge [34,[39][40][41][42][43].
In the present work, careful inspection of the victims' skeletons revealed cracking and explosion of the skullcap and blackening of the outer and inner table, associated with black exudations from the skull openings and the fractured bone. Such effects appear to be the combined result of direct exposure to heat and an increase in intracranial steam pressure induced by brain ebullition, with skull explosion as the possible outcome [93]. As the final result of the victims' bodies being engulfed by the hot pyroclastic surge, the intracranial cavity is found to be filled by ash in the form of a brain-like ash cast. Experimental research shows black bone to be indicative of high thermal exposure (ca 500˚C) [29,41] even if coexistence of blackened bone and bone of unchanged color suggests the persistence of a cooler intracranial area possibly due to temporary preservation (prior to definitive vanishing) of a residual brain mass, since intense heat forces the dural layers to shrink, which in turn constricts the brain into a dense mass [93].
As to the effects of exposure to thermal destruction, a huge literature is available [54,[65][66][67][68]79,[93][94][95][96]. As to the finding of dark stained bone at Herculaneum, bone black in color represents carbonized skeletal material in direct contact with heat or flames [97,98]. Dark colors, particularly black, are related to the carbonization of collagen [64,99]. Heating can be related to combustion (with oxygen) or charring (without oxygen), both of which require the formation of char [100]. The heating process depends also on temperature, heating rate (C˚/min) and exposure time [101].
At Herculaneum, the direct contact of the soft tissues with the pyroclastic surge indicates that the charring was caused by hot-emplaced volcanic ash [102], a characteristic uncommon for victims of pyroclastic density currents, whose bodies are mostly preserved [48,103]. In the 79 AD eruption, assuming environmental reducing conditions (lack or low content of oxygen) at the surge emplacement [6,48], the dark staining of bones is likely to be due to a charring process affecting the victims engulfed within the hot ash cloud [104]. This particular condition seems confirmed by the results of experimentally heated bone vs victims' bones from the 79 AD volcanic context [41,100]. The soft tissues of a corpse act as a physical barrier against the heat and keep bones in anaerobic conditions [105]. The latter process and the unevenness of soft tissue thickness in the body and an unequal distribution of heat during exposure itself, possibly due to the different corpses distribution in the chambers (S1C Fig), may explain the difference in color alterations and the varying degrees of charred bones in the same individual or even on a single bone [106][107][108].
With regard to heat-induced effects on the skull, bone can display a sequence of charred, border and heat line zones which define the area of bone exposure to heat. As also detected in forensic cases [98,109], the changes in the visual appearance of thermally altered bone result in a scale that gradually evolves from a translucent yellow (unaltered bone) to an opaque white Victims' skeletons in their original context during excavation (chambers 10, 12): A. Red mineral residues impregnating the ash bed deposit at different depths (white arrows); B. Thick layer of red mineral residues at the contact surface of the skeleton with the ash deposit (white arrows); C. Red residues may locally impregnate the sand, which appears mostly blackened. https://doi.org/10.1371/journal.pone.0203210.g008 (heat line and border), to a blackened appearance (char). The particular evidence of skullcap fractures and sutures characterized by strikingly black staining has been interpreted as Thermal effects on a fetal skeleton. Intrauterine seven-month skeletal remains (specimen 15bis), unearthed from chamber 11. Spotted red residues, apparent on the skullcap surface, were found to be extremely rich in iron by ICP-MS analysis (55.587 mg/Kg), as well as those red residues (58.519 mg/Kg) encrusting the pelvis (sacrum) of the skeleton (specimen 15, adult female) within which they were found embedded (scale bar in cm).
https://doi.org/10.1371/journal.pone.0203210.g009 openings for fluids to vent from the brain case. This "venting" is said to trap fluids and tissue on the surface of the bone, causing it to be imbued black [110]. As regards the thermal origin of cracking detected on the skull of the 79 AD eruption victims, heat-induced fractures are always limited to the charred areas, since developing heat fractures do not have the energy to radiate out of charred areas into the uncharred bone [98]. This evidence is particularly significant since demonstrates the perimortem origin of the skull fractures induced by the hot ash surge, excluding postmortem causes like the weight of the ash deposit or the direct impact of the surge itself, as previously hypothesized [28,34]. In such cases, the skeletons would have been at least partly dismembered or crushed, which is not, as demonstrated by complete preservation of the victims' skeletons and their anatomical joint connection (Fig 1).
As to the mechanism of death at Herculaneum, evidence like the red residues rich in iron oxides detected from the ash filling the intracranial cavity and encrusting the inner and the outer table, as well as the brown coloration of the venous sinuses, strongly suggests massive heat-induced hemorrhage [111] and a rise in intracranial pressure, as appears clearly from recurrent skull explosive fracture [94]. In forensic cases of skull bursting, particularly in children, the expelled brain matter may form a circular pattern around the head [93], a feature also occurring in a few Herculaneum children (Fig 3A). Examination of fire victims has also shown the presence of heat hematoma [112], with brown bone color being associated with hemoglobin [94]. This is a heat-induced coagulation lying between the bone and the dura, caused by exudation from the venous sinuses of boiling blood, which becomes spongy and brown. The bone table overlying the hematoma is usually charred [113], as repeatedly seen in the victims' skulls at Herculaneum. An increase in pressure caused by bleeding in the various compartments of the brain is considered the most common mechanism of sudden death [95].
Evidence of a heat-induced process of rapid body flesh disappearing is given by the incipient "pugilistic attitude" testified by rare flexure of the upper limbs, but not yet evident in the lower ones (Fig 1A). This heat-induced posture results from denaturation of proteins and muscle fiber dehydration which cause rapid muscle contraction, with consequent abduction of the limbs to the body [94]. Since a body shows a pugilistic attitude soon after exposure to pyroclastic surge temperatures of around 200˚C to 250˚C [48] or burning for about 10 minutes in a crematorium at temperatures between 670˚C and 810˚C [96], the lack of a complete pugilistic pose in the victims' corpses at Herculaneum may indicate that the muscles disappeared more quickly than they contracted. This also seems attested by the "life-like" stance observed in the victims' corpses resulting from the extraordinarily well-preserved skeletal joints fixing the body shape in three-dimensional space (Fig 1B), that could only be explained by very rapid replacement of flesh by ash. In contrast, the widespread occurrence of a pugilistic attitude in the Pompeii victims is attributable to the long-lasting persistence of body flesh, apparent from the shape of the plaster casts, as a consequence of exposure to a lower temperature estimated to be around 250-300˚C [41], enough to cause muscle contraction but insufficient for soft tissues to vanish rapidly.

Conclusions
Here we show for the first time convincing experimental evidence suggesting the rapid vaporization of body fluids and soft tissues of the 79 AD Herculaneum victims at death by exposure to extreme heat, as testified by the unique preservation of iron and heme-iron degradation products as a result of thermally induced hemoprotein oxidation and denaturation. The occurrence of unexpectedly very high concentrations of iron was detected by ICP-MS analysis in samples showing red incrustations from bones, volcanic ash and sand compared to those unaffected, which might have originated from body fluids. Raman micro spectroscopic investigation of these samples reveals the presence of many iron-containing inorganic compounds, including several iron oxides, expected to be the final products of heme-iron upon thermal decomposition. The detection of such iron-containing compounds from the skull and the ash filling the endocranial cavity, coupled with brown coloration of venous sinuses, bone blackening and cracking, strongly suggests a widespread pattern of heat-induced hemorrhage, intracranial pressure increase and bursting, most likely to be the cause of instant death of the inhabitants in Herculaneum.
These findings highlight the need for thorough evaluation of key bioanthropological and taphonomic evidence during archaeological investigations. This is particularly true for the sites affected by the 79 AD Vesuvius eruption, given the high-risk scenario for three million people living today close to the volcano, even if sheltered within buildings.