Sie sind auf Seite 1von 8

Arrhaeometry 41,2 (1999), 413-420.

Printed in Great Britain

AN ARCHAEOMAGNETIC STUDY OF POTTERY KILNS FROM


BRUHL-PINGSDORF (GERMANY)*

J. REINDERS,’ U. HAMBACH,’ K. KRUMSIEK,’ M. SANKE2 and N. STRACK3


‘GeologischesInstitut der Universitat zu Koln, Ziilpicher Strasse 49 a, 50674 Koln, FRG
2Lehrstuhlf u r Archaologie des Mitielalters und der Neuzeit, Universitat Bamberg, Am Kranen 1-3,
96045 Bamberg. FRG
’Rheinisches Amt fur Bodendenkmalpfege, Endenicher Str. 133, 531I5 Bonn, FRG

Four pottery kilns from Bruhl-Pingsdorf (Germany) were dated by comparing their charac-
teristic remanent magnetization with a curve of the palaeo-secular variation of the geomag-
netic field. Archaeomagnetic results suggest last heating events between AD 775 and 1020.
The archaeological evidence generally supports the relative temporal succession and the
duration of use, but indicates that the kilns are about 150 years younger on average. Hence, a
systematic offset between these independently developed chronologies must be considered.
Rock magnetic parameters display radial variations through the kiln walls. These variations
are controlled by the maximum temperature the rock was exposed to. These parameters were
used to identify specimen which acquired a thermoremanent magnetization.

KEYWORDS GERMANY, PINGSDORF, MEDIEVAL, ARCHAEOMAGNETISM, KILN, POTTERY,


CERAMIC, DATING, SECULAR VARIATION

INTRODUCTION

Archaeomagnetic investigationsof burnt materials are interesting for various reasons. They may
provide a spot reading of the earth’s magnetic field at a certain place which can in turn be used to
date a heating event by comparing palaeo-directions (expressed as declination and inclination)
and palaeo-intensities of the geomagnetic field with time dependent calibration curves. The
application of the archaeomagnetic dating method is of special interest where pottery kilns can
be investigated because ceramics of various provenance are used for dating purposes as well.
Archaeomagnetic investigations, thus, can contribute to our understanding of both the earth’s
magnetic field and medieval chronology and trading relations.
The village of Briihl-Pingsdorf (latitude 50”50” N, longitude 6’54” E) near Cologne is famous
for its medieval ceramics production. The ceramics were traded over medieval Europe (from
Tromso (N), Basle (CH) and Winchester (UK) to Gdansk (PL)) and are nowadays used in
archaeology for chronological purposes.
The archaeological chronology of the Pingsdorf kiln-field was recently developed by Heege
(1995) and Sanke (in press). It distinguishes ten periods defined by arguments based on the
ceramic products and the evolution of their styles of ornamentation. The classical ‘Pingsdorfer
Ware’ (Pingsdorf products, periods 3-7) occurred first at the end of the ninth century and
vanished at the end of the twelfth century (Heege 1995; Sanke in press.). In 1993 four pottery
kilns, whose ceramics inventory differed, were sampled for archaeomagnetic dating. Following

* Received 11 December 1997. accepted 24 February 1999.

413
414 J. Reinders, (1. Hambach, K. Krumsiek, M.Sanke and N. Strack

the archaeological chronology,the kilns comprise ceramics that belong to periods 4 and 5 which
were produced from the mid-tenth to the beginning of the twelfth century.

SAMPLING AND ARCHAEOMAGNETIC TECHNIQUES

The kilns were constructed as pits in loam. Their radii range from 50 cm to 140 cm. They are
aligned on both slopes of a hollow running almost south-north. The greatest distance between
two investigated kilns amounts to approximately 12 m. No evidence for differential movements
of archaeological structures has been observed.
The sediments close to the firing chamber were baked in a zone which is up to 10cm thick. This
baked sediment is dark red but the colour changes gently to the yellowish-brown of the unheated
and unaltered loam. This baked layer is referred to as the 'kiln wall' in what follows. Four hand-
samples, c. 3Ox20x20cm in size, were taken per kiln. They were oriented with a magnetic
compass. Each sample includes kiln wall material and unaltered loam. Up to 40 specimen (8 ccm)
per sample were recovered from the kiln walls as described by Reinders and Hambach (1995).The
number of specimen per kiln varies as half of the samples broke upon cutting (Table 1).
The kilns were investigated with standard palaeomagnetic methods (e.g., Collinson 1983,
308-22,335-44). Only directional information was used in the archaeomagneticdating attempt.
The reference curve of the palaeo-secular variation of the earth's magnetic field (PSV-curve)
was taken from Daly and LeGoff (1996, table 2).
A spinner-magnetometer (Molyneux 1971) was used to measure the remanent magnetization.
The magnetic low-field susceptibility ( K ) was determined with a susceptibility-bridge (KLY-2,
Geofyzika, Bmo, CR). Stepwise thermal demagnetization was carried out with a Schonstedt
oven (Schonstedt GS-5, Virginia, USA). Alternating field (AF) treatment was conducted with a
2G degaussing system (Model 600, 2G-Enterprises, California, USA).

ROCK MAGNETIC INVESTIGATIONS

All concentration dependent quantities were normalized to each specimen's mass, since the
density of the material is almost constant but not the volume of the specimen.

Table 1 Kiln-number (MV: mean value), number of samples (Ns). number of specimens (Nsp),precision parameter K
and error estimate (09s). Characteristic remanences (Decl.: declination; Inel.: inclination) and their archaeomagneiic
ages which are inferredfrom the PSV-curve: ages I1 and 12 are inferred from the widest range of intersections of 95%
probability circles and 95% confidence band of the PSV-curve. The last column lists the age intervals derived from the
archaeological chronology (Sanke in press)

Kiln no. NS NsP K 095 Decl. Incl. Age I1 Age 12 Archaeological


(") (OE) ("N) (years AD) (years AD) chronology
(years AD)

22 3 31 70 3.3 11.9 71.9 175 878 950- 1050


23 2 18 192 1.7 14.7 69.5 867 890 1050-1 130
26 3 28 51 3.2 18.6 66.0 893 1025 1050- 1 130
(older than kiln 23)
45 1 8 148 1.3 19.7 63.6 962 1020 1130-1200

MV 4 424 4.5 16.7 67.8 770 1012 900- 1200


An archaeomagnetic study of pottery kilns from Briihl-Pingsdorf 415

Heating of rocks often causes alteration of existing mineral assemblages. The formation of
magnetite from paramagnetic minerals upon heating is frequently observed and results in a
drastic increase of both the remanent magnetization intensity and K (e.g., Le Borgne 1955;
Collinson 1983,435-8). The quotient of natural remanent magnetization (NRM) intensity and K
which is the modified Konigsberger value (Qmd) is a measure of the alignment efficiency of
magnetic grains parallel to the ambient field. The variation of this factor was, thus, used to
distinguish specimen from the unaltered loam and specimen from the kiln-walls which carry a
thermoremanent magnetization (TRM).
Figure 1 (a) shows the variation of NRM-intensity and K in a bilogarithmic plot. Two
populations can clearly be distinguished. The first population consists of specimen from the kiln
wall. A correlation following a power law is obvious (correlation coefficient r = 0.95). NRM-
intensities vary between =10 and =1000Am-lkg-' and K ranges from 0.06 to 1.00 SI unitskg.
The second population is more scattered (r = 0.53) and consists of specimen from the unaltered
loam with NRM-intensities about 2 Am-'kg-' and K-values from 0.01 to 0.05 SI unitskg
(shaded area). Specimen which plot to the upper right comer of the shaded area are not clearly
affected by heat, but absolute values of NRh4-intensities and K imply either a minor degree of
alteration or ineffective separation of kiln walls and unaltered loam upon cutting. Nonetheless,

Figure 1 (a) NRM-intensity versus magnetic low-field suscepiibility ( K ) .Power law fits are plotted forfitied data limits
(continuous lines). Dashed lines denote ihe extensions of ihe power law jits to the limits of ihe abscissa and are only
shown for convenience. (b) Bar chari of the modified Konigsberger value Qmod = NRM-iniensity/K.
416 J. Reinders, U.Hambach, K. Krumsiek, M.Sanke and N. Strack

there are obviously no transitional values from one population to the other but the two
populations are divided by an offset along the abscissa.
A bar chart of Qmd separates three almost log-normally distributed populations (A, B and C
in Fig. 1 (b)). Group A and B consist of specimen from the unaltered loam. The low median
value of group A, Qmd= 40 Am-' lies well within the typical range for sediments (e.g.. Hunt et
al. 1995). Group B specimen show a median value of G o d = 100 Am-' close to the upper
border of typical sedimentary Qmd-values (Hunt et al. 1995). Group C specimen are only found
in the dark red kiln walls and have a median value of approximately 600 Am-'. This high value
of Qmd indicates a better alignment parallel to the ambient field and can be explained as
resulting from the acquisition of a TRM which is carried by newly formed magnetic minerals.
Only group C specimen with K higher 0.05 SI unitskg were, therefore, used for archaeomagnetic
dating. In what follows, these specimen are referred to as 'specimen from the kiln walls' or
'specimen that carry a TRM'.
Acquisition curves of the isothermal remanent magnetization (IRM)were measured to
study the magnetic properties of the heated and of the unheated material. A steep increase of
the IRM in fields up to 0.1 T is visible. Specimen from the kiln walls have acquired about
90%, specimen from the loam about 70% of their saturation IRM.The curves are typical for
rocks where magnetite (Fe304) or maghemite (y-Fe203) dominate the magnetic behaviour.
High values of the coercivity of remanence from the unaltered loam (>40 mT) compared to
those from the walls (< 30 mT) suggest a significantly higher relative contribution of high
coercivity minerals (hematite or goethite) to the magnetic mineral assemblage in the unaltered
loam.
The change of induced magnetization with temperature of kiln wall specimen yields marked
Curie-temperatures which are characteristic of magnetite and, thus, confirm magnetite as the
dominant magnetic phase in the walls. A minor kink which is often observed above 600°C
indicates some amount of hematite or maghemite, which may cause the red colour of the kiln
walls.

ARCHAEOMAGNETIC ANALYSIS

Remanence directions are corrected for the reference site, Paris (France) (Nod and Batt 1990),
and for the local magnetic declination.
The characteristic remanent magnetization (ChRM) was obtained upon AF-demagnetization
and stepwise thermal demagnetization. Both methods yield the same results and none of the
group C experiments had to be rejected. Both demagnetization methods reveal only a weak
viscous overprint which is subsequently removed until the primary TRM is isolated at about
200 "C or 8 kA/m respectively. The natural remanent magnetization (NRM) is almost completely
destroyed at 590°C. Thus, newly formed hematite does not contribute significantly to the
primary TRM although AF-fields as high as 8 0 W m are not sufficient to destroy the NRM
completely. This behaviour is probably due to very fine grained, non-interacting magnetite
grains.
The ChRMs from the kiln walls are presented together with the PSV-curve in Figure 2. The
ChRMs from the kiln walls exhibit an obvious correlation with the PSV-curve. They are aligned
parallel to the course of the palaeo-field between about AD 800 and 1100. Kilns 22 and 23 are
younger than kilns 26 and 45 on a 95% confidence level. A further differentiation of these two
groups is not possible with the same statistical significance.
An archaeomagnetic study of pottery kilns from Briihl-Pingsdorf 417

Figure 2 Mean characteristic remanence of each kiln (stars). 95% confidence limits (circles) and the PSV-curve of the
last 2000 yeurs in Paris afrer Daly and LeGoff (1996) (open symbols. broken line) and 95% confidence range (shaded
area). ChRMs are converted to the reference site (Paris, France). Numbers denote years AD.

DISCUSSION

Archaeomagnetic data are often affected by magnetic refraction. The magnetic refraction affects
declination and inclination depending on the position of a specimen relative to the ambient field
and the shape of the heated object. A correlation as indicated in Figure 3 (a) is expected if
magnetic refraction affects the recording of the earth's magnetic field in a kiln wall. The
magnetic vector is deflected from the direction of the geomagnetic field, which was known to be:
declination =O" and inclination ~ 6 3 (Soffel
" and Schurr 1990).The magnitude of the deviation
depends on the susceptibility and, thus, on the temperature interval where the partial TRM was
acquired (Dunlop and Zinn 1980).
The striking alignment of the ChRMs parallel to the PSV-curve with strongly differing
inclinations and more or less steady declinations could well be caused by the magnetic
refraction. The data were, therefore, tested with respect to the positions of the hand-samples
relative to the field (Fig. 3 (a)) and with respect to the temperature dependent magnetization
(Fig. 3 (b)).
Figure 3 (a) presents declination and inclination of the ChRMs of the samples plotted as a
418 J. Reinders, U.Hambach, K. Krumsiek, M. Sanke and N . Strack

8
(D

0
0
v)

0
0
w

8
0

0
0
(v

-
8
0

I
I i - 0 c


An archaeomagnetic study of p o t t e y kilns from Bruhl-Pingsdorf 419

function of the azimuth. Directional scatter is obvious but no correlation with the position of the
samples relative to the ambient field is observed. Moreover, the deviations of declination and
inclination from their mean values are both of the same order. Since our specimen are taken from
the kiln walls magnetic refraction should show a stronger effect on the declination than on the
inclination (Duclop and Zinn 1980; Soffel and Schurr 1990).
Figure 3 (b) shows declination and inclination of the stacked thermal demagnetization curves
from specimen of kiln 22 and kiln 26 versus demagnetizationtemperatures. There is no dependence
of inclination or declination which can be correlated to the interval where the partial TRM was
acquired, although the greater scatter at high temperatures possibly obscures such a relationship.
The investigationspresented in Figure 3 (a) and (b) point to an acquisition of the TRM parallel
to the palaeofield. The ChRMs are likely to be a true record of the past geomagnetic field at
Briihl-Pingsdorf and the kilns can be dated with the archaeomagnetic method. Age intervals are
listed in Table 1 (columns 8 and 9). The age intervals are inferred from the widest interval in
which 95% probability circles and the 95% confidence band of the PSV-curve overlap. Error
estimates are non-symmetric with respect to the ages due to the non-linear time dependent
variation of the past geomagnetic field.
The ChRMs are presented in Figure 2 and Table 1 (columns 6 and 7). They are of different
statistical significance because varying numbers of samples were available per kiln (Table 1,
column 2). Palaeofield directions and, thus, ages of kilns 22 and 26 are statistically well defined.
The directions derived for kilns 23 and 45 can only be interpreted as palaeofield directions if one
takes into consideration that remanence directions of kilns 22 and 26 are probably not affected
by magnetic refraction. Kilns 23 and 45 consist of the same material with a similar history and
their rock magnetic behaviour cannot be distinguished from that of kilns 22 and 26. Therefore,
ages were calculated for kilns 23 and 45, too.
A piece of wood that was found in the subsequent filling of kiln 23 yielded a calibrated I4C age
of AD 1009 to 1027 (unpublished report of the ‘Institut fur Umweltphysik der Heidelberger
Akademie der Wissenschaften’ to the ‘Rheinisches Amt fur Bodendenkmalpflege’; lab. no.
Hd14617-14421).This age corroborates the archaeomagnetically derived date for kiln 23 since
this piece of wood came into the pit when this kiln was no longer used, but-as the
archaeomagnetic data indicate-the pottery kiln field still was in operation.
The archaeomagnetically derived temporal succession and time-span generally are in
agreement with the archaeological evidence. Both the archaeomagnetic and the archaeological
chronology suggest that kiln 22 is the oldest and kiln 45 is the youngest, but stratigraphic
observations demonstrate that kiln 26 is older than kiln 23 because it is cut by kiln 23. The main,
and significant, difference between the two chronologies is concerned with the absolute ages
which differ by about 150 years (see Table 1, column 10).
The generally good agreement of temporal succession and duration of use inferred from the
two dating methods suggests a systematic offset between the two independently developed
chronologies, the cause of which remains largely uncertain. Parts of the deviations may be
explained by a still imperfect reference curve, but the systematic alignment of the ChRMs with
the PSV-curve (Fig. 2) suggests a general reliability of the archaeomagnetic approach.

CONCLUSIONS
The pottery kilns from Briihl-Pingsdorf can be dated by comparing the ChRMs obtained from
the kiln walls with a curve of the palaeo secular variation of the earth’s magnetic field. The
J. Reinders, U.Hambach, K. Krumsiek, M. Sanke and N. Strack

pottery kilns are of different ages. The kilns were probably used over a time interval of more than
100 years.
In order to account for a systematic offset of about 150 years between the archaeomagnetic
and an archaeological dating approach, we suggest that the term ‘archaeomagneticage’ should
be used. The oldest kiln dates back to about AD 850 in the archaeomagnetic chronology. Future
work will hopefully yield an answer to the observed 150 year offset.
The modified Konigsberger factor can serve as a qualitative criterion to separate specimen
which acquired a TRM from specimen whose original magnetization was not affected by heat
and, thus, was preserved.

ACKNOWLEDGEMENTS
The samples were taken by Stefan Thomas and Peter Briicher; we thank them for their thorough work. Many thanks to
Michael Urbat for comments and discussions.

REFERENCES
Collinson. D. W., 1983, Methods in rock magnetism andpalaeomagnetism, Chapman and Hall, London.
Daly, L., and LeGoff, M., 1996, An updated and homogenous world secular variation data base. 1. Smoothing of the
archaeomagnetic results, Physics Earth Planetary Interiors, 93, 159-90.
Dunlop. D. J., and Zinn, M. B., 1980, Archaeomagnetism of a nineteenth-century pottery kiln near Jordan, Ontario, Can.
J. Earth Sci., 17, 1275-85.
Heege, A., 1995, Die Keramik desfriihen und hohen Mittelalters aus dem Rheinland, Archaologische Berichte, 5, Bonn.
Hunt, C. P., Moskowitz, B. M., and Banerjee, S. K., 1995, Magnetic properties of rocks and minerals, in Rockphysics and
phase relations, a handbook of physical constants, AGU reference shelf3 (ed. T. J. Ahrens), 189-204, AGU,
Washington.
Le Borgne, E., 1955, Susceptibilite magnetique anormale du sol supeficial, Annu. Geophys., 16, 159-95.
Molyneux, L., 1971, A complete result rock magnetometer for measuring the remanent magnetisation of rocks, Geophys.
J. Roy. Astronom. Soc., 24,429-33.
N&l, M., and Batt, C. M., 1990, A method for correcting geographically separated remanence directions for the purpose
of archaeomagnetic dating, Geophys. J. Int., 102, 753-6.
Reinders, J., and Hambach, U., 1995, A geomagnetic event recorded in loess deposits of the Tonchesberg (Germany):
identification of the Blake magnetic polarity episode, Geophys. J. Int., 122,407-18.
Sanke, M., in press, Die gelben Irdenwaren, in Handbuch zur mittelalterlichen Keramik Nordeuropas (eds. H. Liidtke and
K. Schietzel), Neumunster.
Soffel, H.C., and Schurr, K., 1990, Magnetic refraction studied on two experimental kilns, Geophys. J. Int., 102,551-62.

Das könnte Ihnen auch gefallen