Herculaneum victims of Vesuvius in AD 79

Abstract

The town of Herculaneum, lying at the foot of Mount Vesuvius on a cliff overlooking the sea, was buried by a succession of pyroclastic surges and flows (currents of volcanic ash and hot gases generated by collapse of the eruptive column) during the plinian eruption of AD 79. The skeletons of 80 of 300 people who had taken refuge in 12 boat chambers along the beach have now been unearthed from the first surge deposit.
We have investigated how these people were killed by this surge, despite being sheltered
from direct impact, after its abrupt collapse (emplacement) at about 500 7C on the beach. The victims’ postures indicate that they died instantly, suggesting that the cause of death was thermally induced fulminant shock1 and not suffocation, which is believed to have killed many of the inhabitants of Pompeii and of Herculaneum itself.

Full text

T
he town of Herculaneum, lying at the
foot of Mount Vesuvius on a cliff over-
looking the sea, was buried by a succes-
sion of pyroclastic surges and flows (currents
of volcanic ash and hot gases generated by
collapse of the eruptive column) during the
plinian eruption of
AD 79. The skeletons of
80 of 300 people who had taken refuge in 12
boat chambers along the beach have now
been unearthed from the first surge deposit.
We have investigated how these people were
killed by this surge, despite being sheltered
from direct impact, after its abrupt collapse
(emplacement) at about 500 7C on the
beach. The victims’ postures indicate that
they died instantly, suggesting that the cause
of death was thermally induced fulminant
shock
1
and not suffocation, which is believed
to have killed many of the inhabitants of
Pompeii and of Herculaneum itself.
The first surge was generated 12 hours
after the eruption started
2
. Unlike the sub-
sequent surges, it billowed through the
evacuated town of Herculaneum without
damaging it and without leaving any
deposit or even disturbing small and fragile
heat-resistant objects. The surge advanced
as a deflating current with low momentum
until it reached the 20-metre cliff drop,
when its basal, denser component emplaced
abruptly on the beach, halted probably by
hydraulic-jump effects, bursting into the
waterfront chambers and enveloping the
people hiding inside.
We have studied 80 intact skeletons
unearthed from chambers 5 (3 skeletons
out of 14), 10 (40 skeletons), 11 (5 out of
30) and 12 (32 skeletons). Their life-like
stance reflects their posture at the time
when the first surge emplaced. These indi-
viduals, who did not suffer mechanical
impact, do not display any evidence of vol-
untary self-protective reaction or agony
contortions, indicating that the activity of
their vital organs must have stopped within
a shorter time than the conscious reaction
time, a state known as fulminant shock
1
.
The natural posture of the skeletons has
been preserved by virtue of the survival of
their anatomical bone connections.
The skeletons, entombed in the ash from
the first and covered by the subsequent
surges, are lying down or partially leaning
up to a few tens of centimetres above the
chamber floor, probably because of incipi-
ent ‘floating’. Their positioning indicates
that during emplacement the ash must have
expanded slightly and then suddenly deflat-
ed, becoming denser and engulfing the
bodies, cooling them and fixing them in
their positions at the same time.
Some of the skeletons have articulated
fractures, as seen in incinerated bodies
3,4
, and
the inner skull surfaces, cranial openings and
unclosed sutures are blackened from the
effects of high temperature on the skull cap
under increased intracranial pressure. The
victims show transversal clear-cut fractures
with blackened edges, and longitudinal frac-
tures in long-bone diaphyses or flat bones, as
well as cracked tooth enamel, which are also
evident after incineration
4,5
. The patterns of
dental enamel cracks and of bone coloration
indicate that the victims were exposed to a
temperature of about 500 7C, based on
observations of fire victims
6
, ancient burnt
bones
7
and human bone tissue, and on teeth
heat-treated in the laboratory
5,8
. This tem-
perature is compatible with the estimated
480 7C determined palaeomagnetically from
a tile collected from outside chamber 12.
The observed flexion of the hands and
feet (caused by the thermally induced noci-
ceptive or flexor reflex
9
) (Fig. 1) and occa-
sional spine extension of the skeletons are
evidence of instantaneous muscle contrac-
tion having occurred before the ash bed
compacted. The pugilistic attitude charac-
teristic of limb flexures that result from
tendons and muscles shortening post
mortem, typical of fire victims and deaths
in pyroclastic flows
10
, is only apparent on
some of the victims.
The signs of bone carbonization and the
preservation of joint connections indicate
that most soft body tissues were destroyed
by the intense heat and then replaced rapid-
ly by ash. A thermodynamic calculation for
a chamber filled by 30 people, on average,
indicates that a sudden cooling must have
occurred inside the chamber as heat from 8
cubic metres of scorching ash passed into
the bodies (corresponding to 4.5 m
3
of
organic matter) over a contact surface area
of about 20 m
2
. The heat of the ash was just
sufficient to vaporize most of the organic
matter, so the initial violent vaporization
caused a sudden drop in ash temperature.
This could explain why the most marked
thermal effects are limited to teeth and
those parts of the bones least protected by
fat and tissue, and would also account for
the slower disappearance of some residual
soft tissue. The strongest thermal effects are
exhibited by bones and teeth from people in
the less crowded chamber (5), consistent
with our thermodynamic results. The lack
of partial thermal remanent magnetization
in tiles from inside chamber 12, in contrast
with the specimen outside, supports the
proposed rapid drop in temperature (over a
few tens of minutes
11
).
Our findings indicate that the emplace-
ment of the first surge caused the instant
death of these 80 people as a result of fulmi-
nant shock
1
. They were killed before they
had time to display a defensive reaction (in
less than a fraction of a second), their hands
and feet underwent thermally induced con-
traction in about one second (estimated
time based on the mean conduction veloci-
ty of nociceptive C fibres
9
), the positions of
their bodies were fixed by the sudden defla-
tion of the ash bed occurring over the next
few seconds; their soft tissues were vapor-
ized and the temperature then fell over a
few tens of minutes, inhibiting the progress
of the pugilistic stance and the disappear-
ance of residual soft tissue.
Giuseppe Mastrolorenzo*, Pier P. Petrone†,
Mario Pagano‡, Alberto Incoronato§,
Peter J. Baxter||, Antonio Canzanella¶,
Luciano Fattore#
*Osservatorio Vesuviano, Via Manzoni 249,
80123 Napoli, Italy
†Centro Musei delle Scienze Naturali, Museo di
Antropologia, ¶Centro Interdipartimentale di
Servizio di Analisi Geomineralogiche, and
#Dipartimento di Biologia Evolutiva e Comparata,
Università degli Studi di Napoli Federico II,
brief communications
NATURE
|
VOL 410
|
12 APRIL 2001
|
www.nature.com 769
Herculaneum victims of Vesuvius in AD 79
The eruption’s first surge instantly killed some people sheltering from the impact.
Figure 1 The feet of a child’s skeleton recovered from a water-
front boat chamber after being entombed in the first surge of the
AD 79 Vesuvius eruption. The toes show hyperflexion (flexor reflex)
of the feet (chamber 12, juvenile): proximal phalanxes are dorsi-
flexed, whereas medial and distal phalanxes are plantar-flexed.
The feet also show a contracture on the longitudinal axis due to
both eversion and inversion, with opposition of the first and the
fifth metatarsi and toes (scale rule, 10 cm).
© 2001 Macmillan Magazines Ltd

2. Carey, S. & Sigurdsson, H. Geol. Soc. Am. Bull. 99, 303–314
(1987).
3. Bohnert, M. et al. Forensic Sci. Int. 87, 55–62 (1997).
4. Bohnert, M., Rost, T. & Pollak, S. Forensic Sci. Int. 95, 11–21
(1998).
5. Yamamoto, K. et al. Bull. Kanagawa Dent. Coll. 18, 55–61 (1990).
6. Holden, J. L., Phakey, P. P. & Clement, J. G. Forensic Sci. Int. 74,
17–28 (1995).
7. Holck, P. thesis, Anatomical Institute, Univ. Oslo (1986).
8. Shipman, P., Foster, G. & Schoeninger, M. J. Archaeol.Sci. 11,
307–325 (1984).
9. LaMotte, R. H. & Campbell, J. N. J. Neurophysiol. 41, 509–528
(1978).
10. Baxter, P. J. Bull. Volcanol. 52, 532–544 (1990).
11. Butler, R. F. in Paleomagnetism 319 (Blackwell, Boston, 1992).
Tuna swim by restricting lateral undula-
tions to the most caudal body segments, but
maintain their sizeable red-muscle mass in
the mid-body region
7
. We believe that this is
possible because of the novel physical
uncoupling of the action of deep red muscle
from local body bending shown here. Deep
red muscle at the mid-body in yellowfin
tuna shortens in phase with body bending
some 20% more to the posterior, support-
ing the idea that the tuna’s complex tendon
system and elongate myotomes provide a
force-transmission pathway to the tail
8,9
. We
find that deep red fibres undergo strains as
large as, or larger than, the strains in super-
ficial fibres, allowing much greater work
output during swimming than might be
expected from their deep location.
The architecture and physiology of tuna
muscle allow it to generate more work than
is possible in other fish lacking this special-
ized anatomy. Although this increased
power enables the tuna to achieve higher
aerobic speeds, it must also burn more
metabolic fuel. At aerobic swimming
speeds, tuna do maintain a higher total
metabolic rate than salmonids of similar
size
10
, making it hard to draw conclusions
about overall efficiency. But the tuna still
enjoys an advantage in being able to main-
tain higher aerobic speeds than its prey.
Stephen L. Katz*, Douglas A. Syme†,
Robert E. Shadwick‡
770 NATURE
|
VOL 410
|
12 APRIL 2001
|
www.nature.com
Via Mezzocannone 8, 80134 Napoli, Italy
‡Soprintendenza Archeologica di Pompei,
Scavi di Ercolano, 80056 Ercolano, Italy
§Università degli Studi di Napoli Federico II,
Dipartimento di Scienze della Terra,
Largo S. Marcellino 10,
80138 Napoli, Italy
e-mail: incorona@unina.it
||University of Cambridge Clinical School,
Addenbrooke’s Hospital, Hills Road,
Cambridge CB2 2QQ, UK
1. Brinkmann, B. et al. Rechtsmedizin 83, 1–16 (1979).
brief communications
High-speed swimming
Enhanced power
in yellowfin tuna
T
una are distinctive among bony fish for
their elite swimming ability and their
muscle anatomy, having loins of red,
aerobic fibres deep within the body where
other fish have only white, anaerobic
fibres
1,2
. Here we record the performance of
the red muscle of yellowfin tuna in vitro and
in vivo to show how this specialized muscle
architecture can double the cruising power
of these fish, revealing a functional link
between this biomechanical design and
high-speed swimming.
Active muscle must develop force and
shorten in order to do work. In most fish
the relative shortening, or strain, of swim-
ming muscle is the product of body curva-
ture and the distance of the muscle from the
backbone
3
(that is, strain can be accurately
calculated as if the body were a homo-
geneous, bending beam). If a tuna’s body
deformed like a bending beam, then the red
muscle’s internal position (Fig. 1a) would
limit strain and therefore the work and
power produced, a seeming paradox for fish
using high-performance locomotion
1
.
We measured muscle strain in yellowfin
tuna (Thunnus albacares) swimming in a
large water tunnel. Sonomicrometry trans-
ducers were implanted
3,4
in superficial and
in deep red muscle at the longitudinal mid-
point of four animals in order to measure
muscle shortening directly. We also used
videography and beam theory
3
to predict
muscle strain by assuming that the fish
bends as a homogeneous beam (Fig. 1b).
We found that for superficial red muscle
there was close agreement between the mea-
sured and predicted strain amplitude and
phase (Fig. 1c). However, shortening of deep
red muscle measured by sonomicrometry in
all fish (55.31%, s.e. 0.63) was almost dou-
ble that predicted by beam theory (52.78%,
s.e. 0.43), and it lagged behind body curva-
ture by almost 10% of a complete cycle.
Remarkably, this phase difference pro-
duces brief intervals in which superficial
muscle is lengthening while the adjacent
deep red muscle is shortening, and vice
versa. There is thus a large degree of shear
between superficial and deep fibres (the net
strain is double that in other fish), and con-
sequently an uncoupling of deep muscle
strain from local body bending, which is
not observed in other fish
3
.
Using work-loop techniques
5,6
to quantify
work output from muscle strips, we found
that the increased strain results in signifi-
cantly more work (Fig. 1d). Specifically, the
average strain in deep fibres measured by
sonomicrometry produces twice the work
compared with the average strain predicted
by local curvature (24.0 versus 12.7 joules
per kg per cycle). Using even larger strains
(58%) does not increase work output sig-
nificantly, indicating that tuna muscles are
designed to work near maximum capacity
at the strain they experience in the animal.
Figure 1 Superior performance of tuna red muscle is due to its
anatomical and biomechanical design, as well as to its physiology.
a, Tuna axial muscle in cross-section; sonomicrometer crystals
are shown as black dots with trailing wires. In tuna and non-tuna
fish, red muscle forms a wedge close to the skin (yellow), but tuna
have red muscle deep in the myotome as well (red). b, Video
image of a yellowfin tuna swimming at 2.3 body lengths per sec-
ond. Overlaid digitized points (magnified 25) were used to calcu-
late body margins and the curvature of the body midline
3
and to
predict muscle strain from beam theory. Cross, dorsal reflective
position marker. Arrow, mid-body position of crystals. c, Compari-
son of red-muscle strain over 4 tailbeats as predicted from
videography by beam theory
3
(dots) and measured from sonomi-
crometry (full lines) for tuna swimming at 2.9 body lengths per
second (statistics calculated for a minimum of eight tailbeats).
Top, peak strain in superficial muscle: predicted, 54.79%; mea-
sured, 54.47% (predicted strain lags behind measured by only
4.69 deg of phase). Bottom, peak strain in deep red muscle: pre-
dicted, 52.41%; measured, 55.46% (measured strain lags
behind predicted by 30.7 deg). Orientation of sonomicrometry
transducers was verified post mortem. d, Work loops from seg-
ments of deep red muscle. The muscle was stimulated phasically
during sinusoidal length oscillations; onset (arrows) and duration
of stimulation were adjusted to give maximum net work (loop
area), and were similar to activation timing
in vivo
4,9
. Loops run
anticlockwise, so that the force is relatively low during lengthening
and high during shortening. Frequency of the length oscillation
(tailbeat frequency) was 3 Hz. Peak amplitude of imposed length
change (muscle strain as a percentage of resting length) was
52.75% (solid line), 55.5% (broken line), or 58% (dashed
line). The net work done per cycle is shown with the strain for
each loop. Muscle resting length, 4.4 mm, corresponding to zero
length change on the graph.
a
b
c
0
1
2
3
4
5
6
7
8
2.75%
5.5%
8%
12.7 J kg
–1
d
24.0 J kg
–1
29.4 J kg
–1
–0.4 –0.3 –0.1
0.0
0.1
0.2
0.3–0.2 0.4
Force (mN)
Muscle-length change (mm)
0
0.5
1 1.5
6
3
0
–3
–6
6
3
0
–3
–6
Muscle strain (%)
Time (s)
1 cm
© 2001 Macmillan Magazines Ltd