You are on page 1of 12

Are ceramics and bricks reliable absolute geomagnetic intensity carriers?

Juan Morales
a,
, Avto Goguitchaichvili
a
, Bertha Aguilar-Reyes
a
, Modesto Pineda-Duran
a
, Pierre Camps
b
,
Claire Carvallo
c
, Manuel Calvo-Rathert
d
a
Laboratorio Interinstitucional de Magnetismo Natural (LIMNA), Instituto de Geofsica, Unidad Michoacn, Campus Morelia, Universidad Nacional Autnoma de Mxico, Mexico
b
Gosciences Montpellier, CNRS and Universit Montpellier 2, Case 060. 34095 Montpellier, France
c
Institut de Minralogie et de Physique des Milieux Condenss, Universit Pierre et Marie Curie, Paris, France
d
Laboratorio de Paleomagnetismo, Departamento de Fsica, Escuela Politcnica Superior, Universidad de Burgos, C/Francisco de Vitoria, s/n, 09006 Burgos, Spain
a r t i c l e i n f o
Article history:
Available online 1 July 2011
Edited by K. Zhang
Keywords:
Pottery
Bricks
Archeointensity
Open kiln
Western Mexico
a b s t r a c t
A detailed rock-magnetic and archeointensity study was carried out on materials baked by a western
Mexican artisan following traditional techniques to produce faithful reproductions of archeological
pieces of the Michoacn region (Western Mesoamerica). The eld strength at the site (41.0 0.5 lT)
was measured with a uxgate magnetometer and the temperature of the furnace during the baking pro-
cess was monitored continually by means of a thermocouple placed in the middle of the baking cavity.
Rock-magnetic experiments performed on the raw material (clay and paste) and on in situ prepared
baked ceramics and bricks included measurement of thermomagnetic curves (susceptibility and
strong-eld magnetization versus temperature), rst-order reversal curves (FORC), anisotropy of mag-
netic susceptibility (AMS) and anisotropy of thermoremanent magnetization (A-TRM). Magnetite and
probably hematite are present in the samples as carriers of the remanence. Hysteresis ratios suggest that
the samples fall in the pseudo-single-domain grain size region, which may indicate a mixture of multi-
domain and a signicant amount of single-domain grains.
Ceramic pieces and brick fragments were subjected to the Thellier-Coe archeointensity method and to
an alternative paleointensity experiment, with a TRIAXE magnetometer, in order to check whether they
are faithful recorders of the local geomagnetic eld strength.
Mean raw-intensity of sample M1 (pottery) overestimates a 7% the expected site intensity, while those
corresponding to the brick samples (LQ1 and LQ2) underestimate it 15%. Brick sample LNQ shows a
slightly lower intensity (7%), but agrees with the expected site intensity within the experimental uncer-
tainty. The intensity retrieved from the volcanic fragment also included closely reproduces the expected
intensity. After A-TRM and cooling-rate corrections, all mean raw values move closer to the expected
intensity.
Measurement of temperatures at different parts inside the kiln (bottom and upper parts of both central
and peripheral parts) revealed the existence of signicant thermal gradients, similar to those observed in
ovens from other localities. Different cooling rates are then expected in a single oven.
The scatter in the intensity determinations observed in this study, retrieved from pieces elaborated
together in the same oven, could arise from this differentiated cooling rate within the oven and thus,
to an inappropriate cooling rate correction in the archeointensity protocol. As this situation was probably
reproduced in the baking of ancient ceramic artifacts, a better knowledge of the temperature distribution
inside these types of kiln would be desirable in order to choose the appropriate cooling rate correction.
2011 Elsevier B.V. All rights reserved.
1. Introduction
More than fty years have passed since mile and Odette
Thellier proposed the method of paleointensity determination
(Thellier and Thellier, 1959). Although widely used for volcanic
rocks carrying thermoremanent magnetization, the Thellier laws
of reciprocity, independence and additivity were originally dened
for bricks and other baked clays. The ThellierThellier method is
still considered as the most trusted technique to retrieve absolute
paleointensities (e.g., Yu et al., 2004 and references therein).
Indeed, apart from a few recently proposed approaches using
partial thermoremanent magnetization (pTRM) production at a
single temperature in a variety of elds (Dekkers and Bhnel,
2006), most of the commonly employed methods are basically
variants of the original ThellierThellier protocol.
In many archeomagnetic studies, only the directional informa-
tion supplied by declination and inclination of the remanence
0031-9201/$ - see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.pepi.2011.06.007

Corresponding author.
E-mail address: jmorales@geosica.unam.mx (J. Morales).
Physics of the Earth and Planetary Interiors 187 (2011) 310321
Contents lists available at ScienceDirect
Physics of the Earth and Planetary Interiors
j our nal homepage: www. el sevi er . com/ l ocat e/ pepi
vector is available. However, the need for a better knowledge of the
variations of the Earths magnetic eld in historical time in order to
use it as a reliable dating tool also requires the determination of
the strength of the paleoeld vector. Unfortunately, the number
of reliable archeointensity data available is smaller than the num-
ber of directional data, since archeointensity determinations are
experimentally more difcult to obtain than estimations of the
direction of the paleoeld vector.
The basic assumption for any successful archeointensity deter-
minations is the ability of the analyzed material to faithfully record
the strength of the magnetizing eld. Therefore, besides standard
archeointensity investigations devoted to retrieve the unknown
eld intensity from archeological artifacts, it is interesting to carry
out simulated archeointensity studies dedicated to the analysis of
the ability of archeological materials to acquire a magnetization
with strength directly proportional to the intensity of the magne-
tizing eld.
However, detailed studies concerning the paleoeld strength
and the manufacturing processes of archeological artifacts are
scarce. The relationship between archeointensity determinations
and several parameters such as the magnetic properties of the ori-
ginal clay used for modeling of pieces, the maximum temperatures
reached during the baking process or the thermal homogeneity in
the ovens, still remains poorly understood. Nevertheless, some
experimental studies regarding these problems have been carried
out by different groups. Genevey and Gallet (2002) checked their
experimental procedure to retrieve the intensity of the geomag-
netic eld from ancient French pottery by implementing a preli-
minary test on new ceramic material, obtaining results 3% lower
than the expected local eld value. Gmez-Paccard et al. (2006)
carried out an archeomagnetic study of seven contemporaneous
Spanish kilns, which allowed discussing the different factors caus-
ing the observed dispersion. More recently, Catanzariti et al. (2008)
conducted a quality control test of the archeomagnetic method in a
modern partially heated structure, obtaining results consistent
with the known eld value in both direction and intensity. Aidona
et al. (2008) investigated the spatial distribution of magnetic
parameters in the oor-bricks of a test furnace, constructed using
similar materials and techniques than during the Roman period,
concluding that archeological kilns may need to be sampled very
carefully and at close spacing in order to nd the best areas for
archeomagnetic investigations because of the spatial limitation of
the re effect.
Many archaeomagnetic investigations were carried out in the
Americas during the last 40 years [Mesoamerica: Bucha et al.
(1970), Lee (1975), Aitken et al. (1991), Urrutia-Fucugachi (1999),
Bhnel et al. (2003), Lpez-Tllez et al. (2008), Morales et al.
(2009), Rodriguez-Ceja et al. (2009), and Alva-Valdivia et al.
(2010); Central America: Brandt and Costanzo-lvarez (1999);
Venezuela: Costanzo-lvarez et al. (2006) and Rada et al. (2008);
Peru: Shaw et al. (1996); Ecuador: Bowles et al. (2002), and in Bra-
zil: Hartmann et al. (2010)]. All these studies are mainly devoted to
recover the ancient eld intensity as well as to investigate the
magnetic properties of ceramics.
In order to obtain a better knowledge of the manufacturing pro-
cesses of Pre-Columbian western-Mesoamerican archeological
artifacts, as well as of the thermal behavior of kilns used for baking
those artifacts, we designed a synthetic archeointensity experi-
ment on an original open kiln from an artisan workshop founded
in 1815 in Zinapcuaro, Michoacn (Mexico) a small town lo-
cated 40 km to the NE of Morelia (19 51
0
35
00
N, 100 49
0
37
00
,
Fig. 1). This artisan workshop follows most of the ancestral
Fig. 1. Location map of the municipally of Zinapcurao, Michoacn.
J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321 311
manufacturing processes of the region and it is authorized by the
Mexican National Institute of Anthropology and History (INAH)
in Morelia to produce faithful reproductions of local archeological
pieces. However, it should be noted that since the main income of
artisans who work in this and other workshops of the region is al-
most exclusively due to the elaboration of these reproductions,
they use some modern tools like potters wheels and electrome-
chanical grinding machines in order to increase their production.
The eld strength at the site was measured with a MEDA
lMAG-01N Fluxgate Magnetometer, obtaining an averaged value
of 41.0 0.5 lT. This value is in good agreement with the observa-
tory data (41.1 lT) retrieved from Teoloyucan (290 km east of site)
and Coeneo (90 km west from the site) magnetic observatories
(http://132.248.6.186/COEonline.html). The IGRF11 value at the
site also agrees with the values from both observatories. Direct
eld measurement inside the kiln yielded a slightly lower value
(40.3 0.5 lT), which is still inside the experimental uncertainty
range. Thus it might be concluded that, no signicant magnetic
anomaly is observed at the production site.
2. Physical characteristics of the kiln
The studied kiln is a circular open structure, with a diameter of
100 cm and 80 cm high, made of clay blocks 20 cm wide and 40 cm
long. It consists of two chambers: (1) a burning cavity 20 cm high
and above it (2), the open baking compartment with an approxi-
mate height of 60 cm. The kiln oor, also made of clay blocks, is
normally covered with potsherds coming from broken or defective
pieces. Ocote a second class pinewood is used to feed the
burning chamber.
Once modeled and dried in air by the sun for several hours, raw
pieces are placed into the baking chamber. In our experiment, be-
sides some artisanal pieces (owerpot M1, spheres 1 & 2), a couple
of baked bricks were put into the oven; one was aligned with its
long axis parallel to magnetic north (LQ1) while the second one
was placed perpendicular to it (LQ2). The aim of this arrangement
was to investigate possible anisotropy effects of the thermorema-
nent magnetization (TRM), which is common in archeointensity
studies. A volcanic fragment of pumice (PZ) was also included as
part of the load with the aim of erasing its natural remanent mag-
netization (NRM) and imparting a new full TRM. At this point a
thermocouple was placed in the middle of the baking cavity, just
in contact with one of the bricks. This set-up allowed a continuous
monitoring of the temperature during the whole process by an
electronic unit to which the thermocouple was attached. Finally
the oven was covered with tiles and broken clay pieces to continue
with the baking of the pieces (Fig. 2).
Temperature was increased slowly up to approximately 100 C
and maintained at this value for 1 h in order to eliminate the
remaining water. After this dehydration process, temperature
was increased to much higher values. During regular heatings in
the kiln no temperature measurements are carried out by the arti-
sans, but their experience allows them to control the whole heat-
ing process. During our experiment the temperature increased
exponentially to reach a maximum of 658 C (Fig. 3). Then, the
oven was let to cool down naturally after the fuel was almost all
consumed, and once the oven was cool enough it was completely
opened, and the pieces were oriented in situ and recovered (Fig. 4).
3. Pre and post heating magnetic mineralogy of clays
Artisans usually use between two and three varieties of clays for
preparing the ceramic paste. These clays are obtained from
different deposits located within 3 to 8 km from the town
(Rojas-Navarrete, 1995). Each type of clay has different color and
hardness, and these characteristics are found in the paste as well.
Once at the workshop, the different clays are let to dry in the
sun and then pulverized and sieved. These materials are then
mixed at different proportions and water is added until obtaining
a homogeneous paste with the desired characteristics.
We collected some powdered clay and paste already prepared
in order to characterize the original magnetic mineralogy and to
check for chemical changes experienced by the magnetic minerals
during the baking process. For this purpose, continuous low- and
high-temperature vs. low-eld magnetic susceptibility (kT)
curves were obtained on raw and baked ceramic fragments with
both a Bartington Susceptibility Bridge equipped with a furnace
at the Laboratorio Interinstitucional de Magnetismo Natural (LIMNA)
and with a KLY-3S Kappabridge at the paleomagnetic laboratory of
the Institut du Physique du Globe de Paris (IPGP) in France.
The kT curve recording the thermomagnetic behavior of clay
(and also that of paste) is characterized by a very low initial mag-
netic susceptibility (Fig. 5a). Although the signal for these materi-
als is at the noise level of the instrument, a Curie temperature
compatible with that of magnetite could be determined in the
cooling branch. After cooling, the original grey color of these mate-
rials changes to red (as shown in Fig. 4), which probably reects
the formation of secondary hematite.
As the weakness of the susceptibility signal of the kT
curve of clay did not allow an unequivocal Curie temperature
Fig. 2. Picture of the kiln used in this study during baking of ceramic artifacts.
Fig. 3. Variation of the temperature with time within the kiln during baking of
artifacts.
312 J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321
determination, a strong-eld magnetization versus temperature
(J
S
T) curve of the same material was measured with a Variable
Field Translation Balance (VFTB) in the University of Burgos (Spain)
The J
S
T obtained (Fig. 5b) is characterized by the presence of a
strong paramagnetic component. In the heating-curve, a Curie-
temperature of 603 C could be determined, which points to
(slightly maghemitized) magnetite as the carrier of remanence.
The cooling curve displays a somewhat weaker magnetization
and a Curie-temperature of 583 C. These observations point to
an inversion of the magnetite-maghemite solid solution to a hema-
tite-magnetite intergrowth.
In contrast, the kT curve of the baked owerpot made with this
paste (M1) showed a stronger signal (one order of magnitude high-
er at room temperature), which monotonically increases with
increasing temperature without any noticeable transition along
the entire range, from 196 to 5 C (left part of Fig. 5c). Heating
and cooling curves of the high temperature experiment are quite
similar, which shows a reversible behavior with an extremely
low degree of alteration and the absence of phase transitions (right
part of Fig. 5c). We note, however, that the cooling branch runs
over the heating one from approximately 250 C to room temper-
ature; this may indicate the formation of new magnetite. A Curie
point of around 550 C was determined from the heating curve.
Similar behaviors were observed for the spherical pieces (esfera 1
and esfera 2, Fig. 5d), which show no magnetic contribution due
to the decoration of these pieces. No low-temperature curves are
available in these cases.
As shown by Costanzo-lvarez et al. (2006) and Rada et al.
(2008), the degree of irreversibility of susceptibility vs. tempera-
ture curves may be used as an indicator of the original ring con-
ditions of potsherds. Reasonably reversible curves would
correspond to ceramic pieces baked in open res and oxygen rich
atmosphere in which carbonaceous matter is almost completely
wiped out.
First order reversal curves (FORC) diagrams (e.g., Roberts et al.,
2000) are very useful to determine the domain-structure of mag-
netic carriers as well as the presence of magnetostatic interactions
among them. These measurements were carried out with an Alter-
nating Gradient Magnetometer at the Laboratoire des Sciences du
Climat et de lEnvironnement, Gif-sur-Yvette, France. An averaging
time of 100 or 200 ms (depending on the amount of noise on the
hysteresis loop) was used and 100 minor loops for each FORC dia-
gram were measured. The smoothing factor was set at 5 for all the
FORC diagrams. The three measured samples (M1, LQ2 and an
independent heated brick LNQ) have a similar behavior on the en-
tire eld range (Fig. 6): The peak of the FORC distribution occurs at
a very low coercivity but the distribution and the spread along the
H
U
axis is quite limited, even though it has to be kept in mind that
the relatively high smoothing factor causes the distribution to be
more spread out. The lowermost contours extend to high
Fig. 4. A view inside the kiln showing the pieces and their distribution within it.
The thermocouple (TC) is also shown.
Fig. 5. Thermomagnetic curves: (5a) susceptibility-versus-temperature of clay. (5b) strong-eld magnetization-versus temperature of clay. (5c) low and high temperature
vs. magnetic susceptibility curve of sample M1 (owerpot). (5d) high temperature vs. magnetic susceptibility curve of spheres.
J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321 313
coercivities on the H
C
axis. These features are characteristic of the
presence of pseudo-single-domain (PSD) and single-domain (SD)
grains with little magnetostatic interactions. In one case (L2,
Fig. 6b) two separate peaks are visible, which shows more clearly
the presence of the two distributions of PSD and SD grains. The size
distribution of these grains is quite large and probably spans the
SD and the whole PSD size range, as suggested by the FORC distri-
bution tail extending up to 80 mT.
Hysteresis ratios Hcr/Hc vs. Mrs/Ms (Fig. 7) of all the samples
fall in the pseudo-single-domain grain size region (Day et al.,
1977), which may indicate a mixture of multi-domain and a signif-
icant amount of single-domain grains (Dunlop, 2002).
Anisotropy of magnetic susceptibility (AMS) determinations
were performed at the paleomagnetic laboratory of LIMNA with
a Kappabridge MFK1-B. AMS is very low for all the studied sam-
ples. In addition, Anisotropy of Thermoremanent Magnetization
(A-TRM) measurements have been carried out at the palaeomag-
netic laboratory of Gosciences Montpellier (France), by inducing
in the sample a pTRM (560 C to room temperature) in six direc-
tions (i.e. +x, +y, +z, x, y, z). Zero-eld thermal demagnetiza-
tions at 570 C before each pTRM were used as a baseline. These
measurements were performed on the same specimen used for
archeointensity determination after the last heating step. Note that
we were not able to measure all samples because some of them ex-
ploded during the last heating step of the archeointensity mea-
surement (see Table 1). This kind of measurements is required to
correct the archeointensity determinations for TRM anisotropy ef-
fects (e.g. McCabe et al., 1985; Selkin et al., 2000; Chauvin et al.,
2000, among others). Whatever the material studied, the A-TRM
correction factor is very low less than 1%.
4. Field intensity determinations
4.1. Thellier-Coe determinations
Standard paleomagnetic samples were taken from both re-
heated bricks, as well as from the not reheated one. In the case
of the baked artisanal pieces (two owerpots), one fragment of
each piece was further divided into at least seven pieces; each
sub-fragment was pressed into a different salt pellet in order to
treat them too as standard paleomagnetic samples. Once marked
all the samples were subjected to the Thellier-Coe method for
archeointensity determination (Thellier and Thellier, 1959; Coe,
1967).
The experiments were carried out using an ASC Scientic TD48-
SC furnace; all heating/cooling runs were performed in air. Ten
temperature steps were distributed between room temperature
and 580 C, with reproducibility between two heating runs to the
same nominal temperature better than 2 C. The laboratory eld
was set to (30.00 0.05) lT. Partial TRM (pTRM) checks (i.e. rein-
vestigations of results from previous heating steps) were per-
formed after every third heating step.
Fig. 6. FORC diagrams of samples M1 (a), L2 (b) and LNQ (c). SF = 5 for all FORC
diagrams. Note that the scale on the H
u
axis is different than that on the H
c
axis.
Fig. 7. Room-temperature hysteresis-parameters ratios for samples analyzed in
this study plotted on Day diagram (see text for more details).
314 J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321
Thirty-two samples coming from four ceramic fragments (one
owerpot M1; two reheated bricks LQ1 and LQ2; one non re-
heated brick LNQ) and from one volcanic rock (PZ) were studied.
Representative Arai plots are shown in Fig. 8ae, while detailed re-
sults of archeointensity determinations, which have been consid-
ered successful after application of the following criteria, are
shown in Table 1:
(1) Directions of NRM end-points at each step obtained from
archeointensity experiments have to draw a reasonably
straight line, pointing to the origin in the interval chosen for
archeointensity determination. Greatest accepted value of
maximum angular deviation (MAD; Kirschvink 1980) of 15.
(2) No signicant deviation of NRM directions towards the
applied eld direction should be observed.
(3) A number of aligned points (n) on the NRM-pTRM diagram
P5; data suspected to correspond to the VRM acquired
in situ are rejected.
(4) NRM fraction factor (f, Coe et al., 1978) P0.4. This means
that at least 40 per cent of the initial NRM was used for
archeointensity determination.
(5) A quality factor q
f g
b
(Coe et al., 1978) P5 (see Table 1); g
being the gap factor (Coe et al., 1978), and b the relative
standard deviation of the slope.
(6) Archeointensity results obtained from NRM-pTRM diagrams
must not show an evidently concave up shape, since in such
cases remanence is probably associated to the presence of
multidomain grains (Levi, 1977).
(7) Positive pTRM checks, i.e., the deviation of pTRM checks
less than 15%.
Table 1
Individual intensity results: n is the number of heatings steps used for the intensity determination; Tmin-Tmax, the temperature interval of the intensity determination; f, the
fraction of extrapolated NRM used for intensity determination; g, the gap factor; q, the quality factor as dened by Coe et al. (1978); H
raw
, uncorrected intensity value, i.e.,
intensity before anisotropy and cooling rate corrections; 1r, standard error of H
raw
; H
An corr
, intensity value corrected for anisotropy effects; H
(An + CR) corr
, intensity value
corrected for anisotropy plus cooling rate affects; N/R: no result; N/A: not available; : q < 5.
J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321 315
Fig. 8. Representative Arai plots and associated thermal demagnetization curve obtained during archeointensity experiments. Arrows indicate the pTRM checks executed
every third heating steps. The slope was estimated from the linear segment determined from the full circles. Plots for (a): fragment M1, (b) brick fragment LQ1, (c) brick
fragment LQ2, (d) brick fragment LNQ, and (e) PZ: volcanic fragment.
316 J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321
4.1.1. Cooling rate correction
Within the artisanal kiln, the cooling time from highest temper-
ature to 100 C lasts around 3 h, whereas average cooling time in
laboratory (in a TD48 thermal demagnetizer with fan on) ranges
from30 to 45 min; this means a difference of almost 1 order of mag-
nitude. According to experimental results on synthetic SD-type par-
ticles, this difference in order of magnitude will produce an
overestimation up to 7% (Fox and Aitken, 1980; McClelland-Brown,
1984) in strength of the intensity. MD-type particles, on the con-
trary, wouldunderestimate the intensityina paleoor archeointen-
sity experiment (McClelland-Brown, 1984). There are, however, not
enough studies on the effect of cooling rate on the strength of mag-
netization for PSD-like materials, like the ones we are studying.
Cooling rate (CR) dependence of TRM was investigated follow-
ing a modied procedure to that described by Chauvin et al.
(2000). A new TRM (TRM
1
) (created in the same conditions as that
gained during last step of the Thellier experiment) was given to all
samples after last TRM check of the Thellier experiment. At the
same temperature, a second TRM (TRM
2
) was given to all samples
but using this time a long cooling time (2 h). Finally, a third
TRM (TRM
3
) was created using the same cooling time (of about
45 min) as that used to create TRM
1
. The effect of CR upon TRM
intensity was estimated by calculating percent variation between
the intensity acquired during a short and a long cooling time
(TRM
1
and TRM
2
). Changes in TRM acquisition capacity were esti-
mated by means of the percent variation between the intensity ac-
quired during same cooling time (TRM
1
and TRM
3
). It should be
noted that the cooling rate correction was applied only when the
corresponding change in TRM acquisition capacity was below 15%.
4.2. Alternative archeointensity determinations
In order to test unconventional approaches for the ancient geo-
magnetic eld strength recovery, we applied a new fast method for
Fig. 8 (continued)
J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321 317
archeointensity determinations. The TRIAXE magnetometer (Le
Goff and Gallet, 2004) allows continuous high-temperature (up
to w650 C) magnetization measurements and the acquisition of
TRM in any direction and eld intensity up to 200 lT. These three
features are perhaps the main distinct characteristics with respect
to conventional Thellier-type methodologies. Additionally, it also
permits the estimation of the ancient magnetic eld while taking
into account cooling rate differences in a rapid way (2.5 h approx-
imately). We applied this method on sister samples of both, cera-
mic fragments and the volcanic rock.
These experiments were carried out at the Institut de Physique
du Globe de Paris (IPGP). Three determinations were done in air,
applying a laboratory eld of 40 lT. Results obtained lie within ve
percent of the expected value (41.0 0.5 lT) for the pumice sam-
ple (Fig. 9). Intensity determinations on artisanal pottery were
not possible because of the extremely low magnetic moment car-
ried by the ceramic (10
8
10
9
) Am
2
, which was of the order of
magnitude of the TRIAXEs sensitivity (10
8
Am
2
).
5. Main results and discussion
As stated in point 1 of the above outlined acceptance criteria,
directions of NRM end-points at each step obtained from archeoin-
tensity experiments draw reasonably straight lines (MAD 6 15),
pointing to the origin in the interval chosen for archeointensity
determination, independently of the type of sample analyzed
(ceramic, brick or volcanic rock) as observed in the vectorial
demagnetization plots (right hand-side of Fig. 8 ae). It is worth
noting that while NRM of the ceramic fragment M1 demagnetizes
up to 85% of its original value at a temperature of 480 C, the ones
Fig. 9. Typical run plot obtained with the TRIAXE magnetometer, which shows the magnetization data acquired during a complete cycle of measurements for archeointensity
determination. Curve 1 shows the thermal demagnetization between 140 and 440 C of the natural remanent magnetization (NRM) together with the thermal variation of
the remaining (blocked) NRM. Curves 2 and 3 show the temperature dependence of the NRM fraction with unblocking temperatures > 440 C. Curve 4 shows the acquisition
of a new TRM in a known laboratory eld (here 40 lT). Curve 5 is a function of both the thermal demagnetization between 140 and 440 C of the new laboratory TRM and
the thermal variation of the remaining laboratory TRM and remaining NRM magnetization fractions.
Fig. 10. Mean intensity determinations for all artifacts studied together with estimation of TRIAXE for the pumice sample. M1: owerpot; LQ1: reheated brick 1; LQ2:
reheated brick 2; LNQ: not reheated brick; PZ
TC
: pumice intensity determined by the Thellier-Coe method; PZ
TR
: pumice intensity determined by the TRIAXE magnetometer.
318 J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321
corresponding to the reheated bricks (LQ1 and LQ2) only demagne-
tize up to 75% and 70% respectively at a temperature of 530 C.
Both corresponding Arai plots are quite similar. In case of the
non-reheated brick LNQ, the NRM lost up to 90% of its original va-
lue at a relatively low temperature of 450 C. Another point is that
while alteration in the LNQ begins at a moderately high tempera-
ture of 475 C (according to the pTRM checks) in the reheated
bricks it begins at temperatures higher than 530 C.
Individual site intensity determinations are reported in Table 1
while Fig. 10 shows mean intensity determinations from all arti-
facts studied, together with estimations from the TRIAXE magne-
tometer for the pumice sample.
Mean raw-intensity of sample M1 (pottery) overestimates 7%
the expected site intensity, while those corresponding to the brick
samples (LQ1 and LQ2) underestimate it 15%. Although substan-
tially below the expected value, their raw intensities are self-con-
sistent. The A-TRM correction factor is almost the same for both
bricks (less than 1.4%), independently of their orientation with re-
spect to the ambient geomagnetic eld, which indicates a very low
anisotropy of TRM for these burned artefacts. Brick sample LNQ
also shows a slightly lower intensity (7%), but agrees with the ex-
pected site intensity within the experimental uncertainty. The
intensity retrieved from the volcanic fragment closely reproduces
the expected intensity, and agrees well with the value obtained
Fig. 11. Variation of the temperature with time in Metepec kiln, evidencing the existence of high vertical thermal gradients between the lower and upper parts of it (up to
300 C), and in Capula kiln, conrming the existence of moderate radial thermal gradients within the central and peripheral parts of the kilns (up to 100 C). Modied from
Rojas-Navarrete (1995).
Fig. 12. Averaged thermal evolution of the different kilns showing the differences between the highest averaged temperatures reached in each oven (more than 230 C).
J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321 319
by the TRIAXE magnetometer. After A-TRM and CR corrections, all
mean values move closer to the expected intensity.
After analyzing the rock magnetic properties and the archeoin-
tensity estimates no correlation among them was found. On the
other hand, Rojas-Navarrete (1995) reported the thermal behavior
of artisanal kilns from different regions (Fig. 11). After comparing
their thermal behavior we can draw the following observations:
(1) The existence of high vertical thermal gradients between the
lower and upper parts within the same kiln (up to 300 C,
Metepec kiln, left part of Fig. 11) and
(2) The existence of moderate radial thermal gradients within
the central and peripheral parts of the kilns (up to 100 C,
case of Capula kiln, right part of Fig. 11).
Also, as stated in Section 2, no temperature measurements in
the kiln are carried out by the artisans during regular heatings;
control of the entire process depends on his experience. This is evi-
denced at Fig. 12, where the thermal behavior of kilns from differ-
ent regions shows also the differences in the process followed by
different artisans. This motivated us to evaluate the thermal gradi-
ents within the oven described in this study. During a second visit
to the artisanal workshop, we re-evaluated the thermal evolution
of the kiln, this time using four thermocouples distributed at the
bottom and upper parts of the central and peripheral areas of the
kiln. These measurements conrmed the existence of signicant
gradients in the oven, similar to those observed in ovens from
other localities.
These observations could account for signicant differences in
the archeointensity determinations even from pieces elaborated
together in the same oven, depending on the relative position
within it (upper or lower part; placed at the central part or close
to the ovens walls) since they would experience differences in
cooling rate. Under certain circumstances some pieces could even
carry a thermochemical remanence (TCRM) instead of a TRM (Carr-
ancho and Villalan, unpublished results), leading to erroneous
archeointensity determinations.
Taking into account the specic position of the analyzed pieces
in the oven (Fig. 4), we note that both bricks were placed at the
lower part of the kiln, just in contact with it, while the sampled
fragment of the M1 piece (the uppermost edge) was located almost
at the middle of the baking cavity. Moreover, while the bricks and
the volcanic fragment are solid compact structures, the owerpots
are thin wall open pieces which should cool faster than the former
structures due to its greater area. The specic combination of the
position within the oven and size/shape of the burnt pieces could
explain the values reported in Fig. 10.
6. Conclusions
A detailed thermomagnetic evaluation of materials baked by a
western Mexican artisan following traditional techniques was
achieved. This allowed knowing the original magnetic mineralogy
as well as the potential magneto-chemical changes experienced
by the minerals during the baking process.
kT (J
S
T) curves of clay and paste are characterized by a very
low initial magnetic susceptibility (magnetization) yielding Curie
temperatures compatible to that of magnetite. The same is true
for the burned artifacts that yielded however higher magnetic sig-
nal and almost reversible curves. Hematite is probably present in
the studied samples. However, its contribution to the remanence
seems to be negligible.
A full set of magnetic experiments (that included FORC dia-
grams, anisotropy of magnetic susceptibility and anisotropy of
thermoremanent magnetization) carried out on the burnt material
(ceramics and bricks) let to determine the domain-structure of
magnetic carriers, to estimate the anisotropy degree of magnetic
susceptibility of the materials, and to correct the archeointensity
determinations for TRM anisotropy effects. The TRM anisotropy ef-
fect seems to be almost negligible for these samples.
Ceramic pieces made in situ and fragments of bricks subjected
to the Thellier-Coe archeointensity method yielded corrected val-
ues within 5% of that of the nearby magnetic observatory value
(41.1 lT) and direct measurement in the kiln (40.3 lT) for a cera-
mic piece as well as for a brick baked independently in the region.
No correlation among rock magnetic properties and archeoin-
tensity estimates was found.
Measurement of temperatures at different parts inside the kiln
(bottom and upper parts of both central and peripheral areas) re-
vealed the existence of signicant thermal gradients, similar to
those observed in ovens from other localities.
While these observations could account for the signicant scat-
ter in the archeointensity determinations even from pieces elabo-
rated together in the same oven, depending on the relative
position within it, they could also put into question the suitability
of archeological artifacts for recording the intensity variations of
the ancient geomagnetic eld.
A reference thermal distribution obtained from similar kilns
of the region, that could take into consideration the different
cooling rates to which a set of pieces were subjected to, would
be useful at explaining the possible differences in archeointensi-
ty values from pieces retrieved from a burial when these differ-
ences could not be attributed to other causes (e.g., provenance,
temporality).
Acknowledgments
We really appreciate the hospitality of the Hernndez Cano
family at visiting and using their artisanal workshop. JM thanks
the IPGP staff (Saint Maur) for making the facilities available and
for the help with the measurements during his research stay. In
particular, we acknowledge Maxime LeGoff for the invaluable help
during the TRIAXE measurements. We are grateful to Carlo Laj and
Catherine Kissel for help with the FORC diagram measurements at
the LSCE. We acknowledge Mimi Hill, Elisabeth Schnepp and one
anonymous reviewer for the constructive comments and sugges-
tions made on an early version of the MS which substantially im-
proved it. Financial support was provided by CONACYT project #
54957, UNAM PAPIIT 103311 and the CNRS-CONACYT agreement
on bilateral cooperation. PC support was provided by PICS project
# 5319. MC acknowledges the nancial support from project
BU004A09 (Junta de Castilla y Len).
References
Aidona, E., Scholger, R., Mauritsch, H.J., Perraki, M., 2008. Remanence acquisition in
a Roman-style gold furnace. Phys. Chem. Earth 33, 438448.
Aitken, M.J., Pesonen, L.J., Leino, M., 1991. The Thellier paleointensity technique:
Minisamples versus standard size. J. Geomagn. Geoelectr. 43, 325331.
Alva-Valdivia, L.M., Morales, J., Goguitchaichvili, A., Hatch, M.P., Hernandez-Bernal,
M.S., Mariano-Matas, F., 2010. Absolute Geomagnetic Intensity Data from Pre-
Classic Guatemalan Pottery. Phys. Earth Planet. Inter. 180, 4151.
Bhnel, H., Biggin, A.J., Walton, D., Shaw, J., Share, J.A., 2003. Microwave
palaeointensities from a recent Mexican lava Flow, baked sediments and
reheated pottery. Earth Planet. Sci. Lett. 214, 221236.
Brandt, M.C., Costanzo-lvarez, V., 1999. A preliminary archaeomagnetic study of
prehistoric Amerindian pottery from Venezuela. Interciencia 24, 293299.
Bowles, J., Gee, J., Hildebrand, H.J., Tauxe, L., 2002. Archeomagnetic intensity results
from California and Ecuador: evaluation of regional data. Earth Planet. Sci. Lett.
203, 967981.
Bucha, V., Tylor, R.E., Berger, R., Haury, E.W., 1970. Geomagnetic intensity: changes
during the past 3000 years in the western hemisphere. Science 168, 111114.
Carrancho, A., Villalan, J., Unpublished results. Different mechanisms of
magnetization recorded in experimental res: archaeomagnetic implications,
Earth and Planetary Science Letters. submitted for publication.
320 J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321
Costanzo-lvarez, V., Surez, N., Aldana, M., Hernndez, M.C., Campos, C., 2006.
Preliminary Dielectric and Rock magnetic results for a set of Prehistoric
Amerindian Pottery Samples from different Venezuelan Islands. Earth Planets
Space 58, 14231431.
Catanzariti, G., McIntosh, G., Gmez-Paccard, M., Ruiz-Martnez, V.C., Osete, M.L.,
Chauvin, A., 2008. The AARCH Scientic Team Quality control of
archaeomagnetic determination using a modern kiln with a complex NRM.
Phys. Chem. Earth 33, 427437.
Chauvin, A., Garcia, A., Lanos, Ph., Laubenheimer, F., 2000. Paleointensity of the
geomagnetic eld recovered on archaeomagnetic sites from France. Phys. Earth
Planet Int. 120, 111136.
Coe, R.S., 1967. Paleo-Intensities of the Earths Magntic Field Determined from
Tertiary and Quaternary Rocks. J. Geophys. Res. 72 (12), 32473262.
Coe, R.S., Gromm, S., Mankinen, E.A., 1978. Geomagnetic paleointensities from
radiocarbon-dated lava ows on Hawaii and the question of the pacic
nondipole low. J. Geophys. Res. 83 (B4), 17401756.
Day, R., Fuller, M., Schmidt, V.A., 1977. Hysteresis properties of titanomagnetites:
grain-size and compositional dependence. Phys. Earth Planet Inter. 13, 206
267.
Dekkers, M.J., Bhnel, H.N., 2006. Reliable absolute palaeointensities independent of
magnetic domain state. Earth and Planet. Sci. Lett. 248, 508517.
Dunlop, D.J., 2002. Theory and application of the Day plot (Mrs/Ms versus Hcr/Hc) 1.
Theoretical curves and tests using titanomagnetite data. J. Geophys Res. 107, B3.
doi:10.1029/2001JB000486.
Fox, J.M.W., Aitken, M.J., 1980. Cooling rate dependence of the thermoremanent
magnetization. Nature 283, 462463.
Genevey, A., and Gallet, Y., 2002. Intensity of the geomagnetic eld in western
Europe over the past 2000 years: New data from ancient French pottery. J.
Geophys. Res. 107 (B11) 2285. doi:10.1029/2001JB000701.
Gmez-Paccard, M., Chauvin, A., Lanos, Ph., Thiriot, J., Jimnez-Castillo, P., 2006.
Archeomagnetic study of seven contemporaneous kilns from Murcia, Spain.
Phys. Earth Planet. Int., 57, (12).
Hartmann, G., Genevey, A., Gallet, Y., Trindade, R., Etchevarne, C., Le Goff, M., Afonso,
M.C., 2010. Archeointensity in Northeast Brazil over the past ve centuries.
Earth and Planet. Sci. Lett. 296 (34), 340352.
Kirschvink, J.L., 1980. The least-squares line and plane and the analysis of
palaeomagnetic data. Geophys. J. R. Astr. Soc. 62, 699718. doi:10.1111/
j.1365-246X.1980.tb02601.x.
Lee, S. S., 1975. Secular variation of the intensity of the geomagnetic eld during the
past 3000 years in North, Central, and South America. Ph.D. thesis, Univ. of
Okla., Norman.
Le Goff, M., Gallet, Y., 2004. A new three-axis vibrating magnetometer for
continuous high-temperature magnetization measurements. Earth Planet Sci.
Lett. 229, 3143.
Levi, 1977. The effect of magnetite particle size on paleointensity determinations of
the Geomagnetic Field. Phys. Earth Planet Int. 13, 245259.
Lpez-Tllez, J.M., Aguilar-Reyes, B., Morales, J., Goguitchaichvili, A., Calvo-Rathert,
M., Urrutia-Fucugauchi, J., 2008. Magnetic Characteristics and Archeointensity
Determination on Some Mesoamerican pre-Columbian Potteries: Case Study of
Quiahuiztlan Archeological Site (Veracruz, Gulf of Mexico, 900 - 1521 A.D.).
Geofsica Internacional 47 (4), 329340.
McCabe, C., Jackson, M., Ellwood, B., 1985. Magnetic anisotropy in the Trenton
limestone: results of a new technique, anisotropy of anhysteric susceptibility.
Geophys. Res. Lett. 12, 333336.
McClelland-Brown, E., 1984. Experiments on TRM intensity dependence on cooling
rate. Geophys. Res. Lett. 11 (3), 205208.
Morales, J. Goguitchaichvili, A., Acosta, G., Gonzlez-Morn, T., Alva-Valdivia, L.,
Robles-Camacho J., and Hernndez-Bernal, M.S., 2009. Magnetic properties and
archeointensity determination on Pre-Columbian pottery from Chiapas,
Mesoamerica, Earth Planets Space, Special issue on Magnetism of Volcanic
Materials, Tribute to Works of Michel Prevot, doi:10.19EPS2364.10.29.
Rada, M.A., Costanzo-lvarez, V., Aldana, M., Campos, C., 2008. Rock magnetic and
petrographic characterization of prehistoric Amerindian ceramic from the Dos
Mosquises Island (Los Roques, Venezuela). Interciencia 33 (2), 1291334.
Roberts, A. P., Pike, C. R., Verosub, K. L., 2000. FORC diagrams: A new tool for
characterizing the magnetic properties of natural samples. J. Geophys. Res., 105,
28,461 28,475.
Rodriguez-Ceja, M., Goguitchaichvili, A., Chauvin, A., Morales, J., Ostroumov, M.,
Manzanilla, L. R., Aguilar Reyes B., Urrutia-Fucugauchi, J., 2009. An Integrated
Magnetic and Raman Spectroscopy Study on Some Pre-Columbian Potteries
From Cuanalan (A Formative Village in the Valley of Teotihuacan) in
Mesoamerica. J. Geophys. Res., 114, B04103, doi:10.1029/2008JB006106.
Rojas-Navarrete, L.L., 1995. Desarrollo de un vidriado sin plomo de baja
temperatura para la alfarera tradicional Mexicana. M. Sc. thesis, Universidad
Autnoma Metropolitana, Unidad Iztapala. 127pp.
Selkin, P.A., Gee, J.S., Tauxe, L., Meurer, W.P., Newell, A.J., 2000. The effect of
remanence anisotropy on paleointensity estimates: a case study from the
Archean Stillwater Complex. Earth Planet. Sci. Letters 183, 403416.
Shaw, J., Walton, D., Yang, S., Rolph, T.C., Share, J.A., 1996. Microwave
archaeintensities from Peruvian ceramics. Geophys. J. Int. 124, 241244.
Thellier, E., Thellier, O., 1959. Sur lintensit du champ magntique terrestre dans le
pass historique et gologique. Ann. Gophysique. 15, 285376.
Urrutia-Fucugachi, J., 1999. Preliminary results of a rock magnetic study of
obsidians from central Mexico. Geofsica Internacional 38, 8394.
Yu, Y., Tauxe, L., Genevey A., 2004. Toward an optimal geomagnetic eld intensity
determination technique, Geochem. Geophys. Geosyst., 5, Q02H07,
doi:10.1029/2003GC000630.
J. Morales et al. / Physics of the Earth and Planetary Interiors 187 (2011) 310321 321

You might also like