Transcript Aucun titre de diapositive

Potential of archeomagnetism as
a dating tool for archeology:
Examples from France
Maxime Le Goff 1, Yves Gallet 1, Nicolas Warmé 1,2, Agnès Genevey 3
IPGP1, INRAP2, C2RMF3
Archeomagnetic sampling following
Thellier’s technique
Magnetic measurements
« Big » sample inductometer at Saint Maur IPGP Laboratory
Methodology:
The directional results are determined and selected
on the basis of viscosity experiments (Thellier, 1981)
1) Direct position during 15 days
measurements
2) Inverse position during 15 days
measurements
By vectorial substraction
VRM and TRM
Example of archeomagnetic results (Rungis, France)
VRM
2000
TRM
fichier RUN06.AMZ
Mean TRM direction: D=-0.7°, I=70.6°, a95=0.4° (N=14)
Archeomagnetic dating
A reference directional curve is needed
Changes in direction of the Earth’s magnetic field
in France as deduced from archeomagnetic data
Thellier,
PEPI, 1981
Bucur,
PEPI, 1994
Sliding window of 80 years shifted every 25 years
Age distribution of the French
archeomagnetic results selected by Bucur (1994)
Constructing a reference curve taking into
account the non-homogeneous
age distribution of the data
Each individual datum is defined by a direction (D, I, k, N)
and an age bracket
For each time window (i), intercepted data (j) are weighted
following the proportion of time contained
in the window (0< wj <1)
The mean directions (i) are estimated using the bivariate
extension of Fisher’s statistics ( wj Tj )
A total weight is obtained for each time window ( wi )
The width of the window is step-by-step increased
until the minimum weight required is attained
Comparison between the
French curves constructed
using moving windows
of varying (double line) and
fixed (dashed line) duration
Weight and duration of each
time window considering
a threshold value of 2.5
(Le Goff, Gallet, Genevey, Warmé, PEPI, 2002)
Determining an archeomagnetic age…
Use the rejection test developed
by McFadden and McElhinny (1990)
Modified in order to compare a Fisherian mean (to be dated)
and non Fisherian means (ellipses defining the reference curve)
 The angular distances between the respective means
allow to determine an archeomagnetic age bracket at 95%
 We can also estimate the probability P (in %) of making
an error if an undated archeomagnetic direction is assumed
different from any sliding window direction
Circumvent a contradiction…
Perfect agreement
Archeomagnetic
dating
still possible
high
quality
but low
quality
Dating the end of use of a Roman water conduit
(Rungis, France) supplying the south of Paris
kiln
Water
conduit
Age of the oven found in Rungis
Archeomagnetic age: AD 625-725
(more probable: AD 685-725)
The case of a kiln found
with a filling containing several tens of
Roman and some undetermined potsherds
(Argenteuil,
France)
Has this kiln a Roman age ?
No, the kiln is Merovingian (AD 625-725)
The story of a domestic kiln…
1) Excavating a ditch
4) Cooking meals
2) Digging the kiln in silt
3) Heating the kiln
5) Abandonment of the kiln
And digging a new one
Is there a significant time interval between several
domestic kilns sharing the same working area ?
1m
Site of Marines (Val d’Oise, France)
Dating of
4 kilns
same archeomagnetic age: AD 705-895
(more probable: AD 705-845)
Individual to group
Archeomagnetic ages at 95%
Isolated kiln
Site of Marines
600
700
800
900
1000
600
700
800
900
1000
Nb. Structures
500
Most probable
archeomagnetic ages
Non isolated Kiln
Nb. Structures
Constraining the historical
change in the use of kilns:
500
Age (AD)
Conclusions:
The archeomagnetic dating technique
is already operational in France,
But
We still need additional well-dated archeomagnetic
directions to define better
the reference archeomagnetic secular variation curve.