J ournal of Archaeological Science(1998) 25, 149163

Archaeomagnetic DatingandPalaeochannelsSediments: Data

FromtheMediaeval Channel Fillsat Hemington, Leicestershire
C. Ellis
AOC (Scotland) Ltd, Edinburgh, U.K.
A. G. Brown
Department of Geography, University of Exeter, U.K.
(Received 28 May 1996, revised manuscript accepted 28 February 1997)
Thispaper illustrateshowarchaeomagneticdatingcan beusedfor palaeochannel llswithin oodplain sequences. The
sitesinvestigatedarepalaeochannelsassociatedwiththemediaeval braided/anastomosingchannelsof anunstablereach
of theRiver Trent at Hemington, north-west Leicestershire. This reach has revealed abundant archaeological remains
including sh weirs, anchor stones, a Norman mill and three mediaeval bridges. The studies used here show good
agreement between archaeomagnetic datingand other evidencefromarchaeology, dendrochronology and radiocarbon
dating. Fabricandmineral analysesof thesedimentsareusedto identifythesourceof theremanent magnetizationand
it isbelieved that thisispredominantly theresult of thebiogenic synthesisof magnetic mineralsin high pH but anoxic
conditions. This study illustrates thetypeof alluvial sediments that aremost suitablefor archaeomagnetic dating.
1998 Academic Press Limited
Keywords: ARCHAEOMAGNETI SM, ALLUVI UM, PALAEOCHANNELS, ANI SOTROPY OF
SUSCEPTI BI LI TY, PYRI TE.
Introduction
B
uried landscapes within oodplain sedimentary
sequences can giveuniqueinsights into human
activity, whether this is long-term or seasonal
occupation, industrial exploitation, pastoral or arable
cultivation (Brown & Keough, 1992; Brown, 1997).
Rivers and their valleys have always attracted human
attention and exploitation primarily because of the
proximity of water but also because of fertile alluvial
soils, potential food resources and therelativeeaseof
transport. The study of uvial and alluvial sediment
in association with archaeological material can pro-
vide information concerning the specic site setting,
the local habitat, the regional environment and
theprocesses of sitepreservation and destruction. The
absolute dating of alluvial units is essential for the
development of an accurate alluvial chronology
which will facilitate the exploration of the relation-
ships between human activity, climatic conditions,
vegetation and geomorphology.
Thevarying direction of thegeomagnetic eld over
the last 2000 years has been broadly established for
Britain(Tarling&Dobson, 1995). TheBritisharchaeo-
magnetic reference curve has been constructed from
466 archaeological sites in Britain and in north-east
France (Clark, Tarling & Noel, 1988). However, in
addition somene-grainedsediment hasbeen foundto
record the earths magnetic eld (Clark, Tarling &
Noel, 1988; Batt &Noel, 1991; Clark, 1990, 1992; Batt,
1992). Most of the data used in the construction of
theBritish geomagnetic secular variation curveof the
earthseldwerederivedfrominsituredarchaeologi-
cal material andto compensatefor magneticdistortion
all inclinations fromoors wereincreased by 24' and
12' from mixed walls and oors, with only a small
proportion obtained fromsediment (Clark, Tarling &
Noel, 1988). Archaeomagnetic directional dating is
achieved through themeasurement of themean angles
of dip (inclination) and declination of magnetic min-
erals in samples. Thenatural remanent magnetization
(NRM) is then matched against thereferencecurveto
obtain a dateeither of deposition (which is referred to
as detrital remanencemagnetization (DRM)), or post-
deposition (referred to as post-depositional remanence
magnetization(PdRM)). Unfortunately thecalibration
curvecontains a number of tight crossovers which can
result in ambiguity and potential alternative dates
(Clark, Tarling & Noel, 1988). The calibration curve
also changes at varying rates and so thereareperiods
of good and poor dating resolution. The density of
datapointsisextremelygoodfor thesecondhalf of the
13th century , whereas the 1000100 curve is
constructedlargelyfromathree-point runningmeanof
149
03054403/98/020149+15 $25.00/0/as970203 1998 Academic Press Limited
combined lake data of Turner & Thompson (1981).
The dark ages are again a period of little cultural
dierentiation and development in Britain and hence
thereis alack of independent datapoints. Theloop of
600900 was observed in lake sediments (Clark,
Tarling & Noel, 1988) and there is a lack of primary
dated material for this period. However, despitethese
problems it has been suggested that a dating of 20
years (68%condencelevel) is possiblewhen usingthe
British archaeomagnetic reference curve from 850
topresent becausethecurveisrelativelyopen, withone
crossover, and is of relatively good resolution (Clark,
Tarling & Noel, 1988). I t is also suggested that a date
of 10 years is possibleafter 1600.
Anisotropy of magnetic susceptibility (AMS) is a
method of three-dimensional petrofabric analysis
(Hamilton, 1967; Hamilton & Rees, 1970; Ellwood,
1984). The purpose of studying AMS is to obtain
information on the depositional processes of all the
grains, although the magnetic content is a minute
proportion of thewholesample(Hrouda, 1982; J ohns
& J ackson, 1991). Thetechniqueis used in theestima-
tion of preferred orientation of grains in sediments
(Rees, 1961). There are two types of microfabric ele-
ments; therst is a planar structuredueto theeects
of gravity and the second is linear, due to the action
of uid forces, for example hydrodynamic forces
(Hamilton, 1967). I t isalso important to determinethe
extent of AMS in specimens because of the potential
distorting eect on NRM.
This paper presents data from the archaeological
site of Hemington Fields, north Leicestershire (SK
461307). Thesiteisaworkinggravel quarry, locatedat
the conuence of the River Derwent and River Trent
(Figure 1). The oodplain at this point is some 4km
wide; on theeast sideof theoodplain lies theconu-
enceof theRiver Sour. Thesiteprovided an excellent
opportunity to extract orientedsamplesof ne-grained
sediment and to establish whether thismaterial carried
a natural remanent magnetization (NRM). A number
of analytical techniques wereused to investigatedepo-
sitional and post-depositional processes and aid in the
reconstruction of thesedimentary history of thesite.
FieldandLaboratoryProcedures
Eight vertical sections through palaeochannels were
created during gravel extraction (Figure 2). Sampled
sediment wasnegrained, insituand physically undis-
turbed. Thesections werecleaned, then described and
recordedintheformof sedimentarylogs. Undisturbed,
orientated samplesweretaken usingaluminiummono-
lith tins (54119cm) where the tops of the tins
werehorizontal and thelength vertical. Themonolith
tins were oriented to true north using a magnetic
compass. Thesampleswerewrapped in foil and plastic
sheet to prevent loss of soil moisture, inhibit bioturba-
tion and giveprotection when transported. At sections
1 and 2 orientated perspex cubes (of approximate
dimensionsof 222222cm) wereinsertedintothe
face by hand. The advantage of using monolith tins
was that a relatively large sample could be taken and
therefore the bulk of the sample was protected from
physical distortion (Gravenor, Symons & Coyle, 1984)
and loss of moisture (Noel, 1980). Bulk samples
were also taken from the main clay, silt and sand
stratigraphical horizons of each section.
Using a stainless steel blade and knife, orientated
cubic specimens werecut fromtheundisturbed mono-
lith samples. Each cut specimen was transferred to a
cubic specimen holder, theend cleaned and thespeci-
men capped and sealed. Volumesusceptibility () was
measured using a balanced a.c. bridge circuit. Mag-
netic fabric/anisotropy of susceptibility (AMS) was
measured using a computer linked Molspin Minisep
anisotropy delineator instrument. Various anisotropy
parameters were also calculated. Natural remanent
magnetization (NRM) measurements were made in a
Minispinslowspeedcomputerizeduxgatemagnetom-
eter. Mean directions were calculated using an inter-
activepalaeomagnetic data analysis softwarepackage,
I APD (Torsvik, 1992). These were corrected to
Meriden (5243'N:162'W), following the methods of
I rving (1964) and Noel & Batt (1990). The corrected
mean directions and their associated cone of con-
dence were plotted onto a digitized version of the
British archaeomagnetic reference curve, taken from
Clark, Tarling& Noel (1988). To establishthestability
of magnetization stepwise, alternating eld demagnet-
ization was carried out on two specimens fromsection
1, eleven from section 2 and eight from section 7.
Specimens were subjected to progressively increased
elds up to 60mT.
Particlesizeanalysis was carried out usingstandard
dry sieving methods (B.S.I ., 1967). Thesilt- and clay-
sizefractionwasdispersedin01%calgon, placedinan
ultrasonic bath for 10 min and subsequently analysed
using a Microtechnics model 5000ET SediGraph.
X-ray diraction and X-ray uorescenceanalysis were
carried out by A. Smith, University of Leicester. The
water content and loss on ignition were determined
following the methods of Bascombe (1974). Three
thin sections, one from unit 2 and two from unit 4,
were prepared by Mr K. Schrapel of Queens College,
University of London. Freeze-dried specimens from
unit 4 were analysed in a scanning electron micro-
scope (SEM) and by energy-dispersive X-ray analysis
(EDXA).
SiteDescriptions
Archaeology
During gravel extraction an abundance of mediaeval
archaeological artefacts and structures have been ex-
posed at Hemington Fields. The plethora of archaeo-
logical material is a consequenceof thesites location
150 C. EllisandA. G. Brown
and its physical nature. The location was used as a
crossing point of the Trent between Leicester and
Derby/Nottingham and the site was probably chosen
because of the existence of a number of shallow
channels. However, thereach was unstableand struc-
tures were repeatedly destroyed and buried by gravel
(Brown & Salisbury, in press). Wooden posts, timbers
and anchor stones have been recovered from the
surface of the Mediaeval gravel and in some palaeo-
channels. Theseartefacts havebeen interpreted as the
remainsof shweirsandpossiblyanchor stones, net or
basket weights(Salisbury, 1987, 1988, 1995). I n1985, a
12th century mill and mill damwererapidly excavated
by the Leicestershire Archaeological Unit (Clay &
Derby
Nottingham
Hemington
Fields
R
. D
e
r
w
e
n
t
R
.
T
r
e
n
t
R
.

S
o
a
r 0 5 km
N
2
4
1
7
9
3
3
Extrapolated course
of palaeochannels
Section
11th century bridge
Early 13th century bridge
Mid 13th century bridge
Old county boundary
Waterlogged
0 100 m
11
10
P
o
r
t
w
a
y
1
1
t
h

1
3
t
h

c
e
n
t
u
r
y

p
a
l
a
e
o
c
h
a
n
n
e
l
s
N
R. Derwent
R
.

T
r
e
n
t
Figure1. Location of Hemington Fields and a plan of thesiteof Hemington Fields. Shown arethelocations of thesampled palaeochannels
and their extrapolated course, theold county/parish boundary and thelocation of thethreeexcavated bridges.
Archaeomagnetic DatingandPalaeochannelsSediments 151
c m
7
0 0
1
3
0
1
0
2
0
3
0
4
0
5
0
6
0
8
0
9
0
1
0
0
1
1
0
1
2
0
7
0 0
2
2
0
1
0
2
0
3
0
4
0
5
0
6
0
8
0
9
0
1
0
0
1
1
0
1
2
0
2
1
0
2
0
0
1
9
0
1
8
0
1
7
0
1
6
0
1
5
0
1
4
0
1
3
0
7
0 0
1
0
2
0
3
0
4
0
5
0
6
0
8
0
9
0
1
0
0
1
1
0
1
2
0
2
0
0
1
9
0
1
8
0
1
7
0
1
6
0
1
5
0
1
4
0
1
3
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
1
0
0
0
1
0
0
1
5
0
2
0
0
2
5
0
3
0
0
5
0
3
5
0
0
8
0
1
2
0
1
6
0
2
0
0
2
4
0
4
0
2
8
0
2
0
1
0
0
1
4
0
1
8
0
2
2
0
2
6
0
6
0
3
0
0
U
n
i
t

6
0
4
0
2
0
1
0
6
0
3
0
5
0
U
n
i
t

1
U
n
i
t

2
U
n
i
t

3
U
n
i
t

4
U
n
i
t

5
U
n
i
t

6
U
n
i
t

1
U
n
i
t

2
U
n
i
t

3
U
n
i
t

4
U
n
i
t

6
U
n
i
t

1
U
n
i
t

3 U
n
i
t

2
U
n
i
t

4
U
n
i
t

6
U
n
i
t

5
U
n
i
t

1
U
n
i
t

3
U
n
i
t

2
U
n
i
t

4
U
n
i
t

6
U
n
i
t

1
U
n
i
t

3
U
n
i
t

2
U
n
i
t

4
U
n
i
t

6
U
n
i
t

1
U
n
i
t

3
U
n
i
t

2
U
n
i
t

4
U
n
i
t

6
U
n
i
t

5
U
n
i
t

1
U
n
i
t

2
U
n
i
t

3
U
n
i
t

4
U
n
i
t

5
U
n
i
t

2
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C l a y
S i l t
F i n e s a n d
M e d i u m s a n d
C o a r s e s a n d
G r i t
S t o n e s
P e b b l e s
C o b b l e s
B o u l d e r s
C
R
o
o
t
l
e
t
s
M
a
c
r
o
s
c
o
p
i
c

c
h
a
r
c
o
a
l
C
u
t

s
t
o
n
e

(
p
i
e
r
s

o
f

m
i
d

1
3
t
h

c
e
n
t
u
r
y

b
r
i
d
g
e
)
F
e
w

s
t
o
n
e
s
W
a
v
y

c
o
n
t
a
c
t
S
e
c
t
i
o
n

1
S
e
c
t
i
o
n

2
S
e
c
t
i
o
n

3
S
e
c
t
i
o
n

4
S
e
c
t
i
o
n

7
S
e
c
t
i
o
n

9
S
e
c
t
i
o
n

1
0
0
3
0
0
4
2
0
4
6
0
1
0
0
2
0
0
4
0
0
4
4
0
4
8
0
S
e
c
t
i
o
n

1
1
F
i
g
u
r
e
2
.
S
e
d
i
m
e
n
t
a
r
y
l
o
g
s
f
o
r
t
h
e
e
i
g
h
t
s
a
m
p
l
e
d
s
e
c
t
i
o
n
s
.
F
o
l
l
o
w
i
n
g
t
h
e
m
e
t
h
o
d
o
f
T
u
c
k
e
r
(
1
9
8
9
)
.
C
i
s
c
h
a
r
c
o
a
l
.
152 C. EllisandA. G. Brown
Salisbury, 1990). Excavation also revealed a series of
10 palaeochannels dated between the 7th and 20th
centuries .
I n19931994threemediaeval bridgeswereidentied
(Cooper, Ripper & Clay, 1994). Theearliest bridgewas
preserved as two box-like caissons constructed from
massivetimbers; thetimbers werefelled between c.
1066 and 1096 (Cooper, Ripper & Clay, 1995). The
second bridge lay some 50msouth (upstream) of the
11th century one; this bridge was constructed of pile-
driven posts and dendrochronology revealed it to date
to 1216(Salisbury, 1995). Thelast bridgelay some
20m further upstream and dendrochronology pro-
duced a felling dateof between 1230 and 1240 for
two oak timbers from its stone and timber starlings
(Cooper, Ripper & Clay, 1995). A track known as the
Portway was aligned with this bridge, conrming the
hypothesis that this area was an important crossing
point of the River Trent (Cooper, Ripper & Clay,
1994). Quarryingbegan at theend of 1993on theeast
side of an ox-bow lake; this channel once formed the
old county and parish boundary (Figure1). Thechan-
nel was abandoned before 1830, when avulsion
changed thecourseof theRiver Trent to that followed
at present (Clay & Salisbury, 1990). Three large
palaeochannelshavebeenuncoveredinthisarea; asyet
no archaeology has been found in relation to these.
Palaeohydrological analysis of thebridges and associ-
atedchannelshasshownhowtheowof theTrent was
progressively switched fromthewestern to theeastern
side of the oodplain during the 12th13th centuries
(Brown, 1996; Brown & Salisbury, in press).
Geomorphology
The simplied stratigraphy at Hemington Fields con-
sists of Mercia mudstone bedrock. Lying unconform-
ably above the irregular bedrock surface were
approximately 225m of tabular and shallow cross-
bedded travels. Organic sediment from an eroded
palaeochannel cut into thebaseof thegravelsand into
bedrock hasprovided two radiocarbon datesof 12,530
(uncal. beta-85499) and 11,730 (uncal. beta-
85448). This, with the common occurrence of ice-
wedge casts in the gravels, indicates that the gravels
weredepositsduringtheLateGlacial Stadial (Younger
Dryas) between c. 11,500 and 10,500 (Brown &
Salisbury, in press). These gravels were capped by
approximately 225mof Mediaeval gravel. Cut into
the Mediaeval gravel are a complex series of palaeo-
channels. The Holocene oodplain is blanketed by
c. 12m of silt/clay overbank sediment. The contact
between the Devensian and Mediaeval gravels can be
identied, in somelocations by a subtlecolour dier-
ence from a pink/reddish to a yellow/brown hue, an
erosional contact made obvious by truncated ice
wedges and the occurrence of subangular to sub-
rounded boulder sized rocks, interpreted as anchor
weights (Salisbury, pers. comm.).
During the mediaeval period the river gradually
evolved through a braided/anastomosing phaseinto a
meandering, single channel form. The palaeochannels
werea consequenceof this changein thesystem, they
are characterized by a basal inll of black silty sand,
10 17/1. Thesediment contained alimited quantity
of small detrital organic material. When freshly
exposed the sediment was black, homogeneous and
very moist. Contact with air caused a rapid colour
changetobrownishblack, 10 2/3, andtheformation
of coarse prismatic peds. Ancient rootlets occurred
in both the overbank and palaeochannel sediment; in
the latter these were made visible by the presence of
brownish black aureoles, the result of oxidation. A
summary description of thesedimentary units is given
in Table1 and Figure2.
Results
Fabric data
A total of 675 specimens were measured for AMS. A
summary of theresults and illustrativeexamples from
themajor unitsof specicsectionsaregivenbelow(see
Ellis, 1995 for a detailed discussion of theresults).
Section7Themagneticfabricmeandirectionof unit 2,
section7(Table2) layonanorth-eastsouth-west axis,
Table 1. Description of the main sedimentary units at Hemington
Fields. The units described are shown in the sedimentary logs of
Figure2
Unit 1. Clayey silt. 75 2/3, very dark brown. Mottled, 5 3/6,
dark reddish brown. Coarseblocky peds. Circa. 5%rootlets.
Unit 2. Clayeysilt. 10 4/1, dark greytobrown10 4/4. Mottled,
5 3/4, dark reddish brown, especially along root channels and
laminae; iron concretions. Occasional manganese nodules. Some
amorphous organic fragments.
Unit 7. Bright brown clayey silt.
Unit 2/3. Clayey silt. 10 4/3, greyish yellowbrown.
Unit 3. Clayey silt. 10 4/2, greyish yellow brown. Occasional
vivianite. Occasional visible laminae, manganese ecks and iron
nodules.
Unit 4. Silty sand. 10 17/1, black and 10 3/2, brownish
black zones. Vivianite largely associated with this unit. Occasional
laminae. Rare sandstone clasts. Some amorphous organics,
occasional dead rootlets and brous organic materials.
Unit 9. Coarse to ne, well sorted sand. 10 4/4, dark yellowish
brown, 10 3/1, very dark grey. Top of bands often stained black.
Occasional grit-sized material.
Unit 10. 1cmthick silt clay unit (section 11). 10 3/1. Vivianite.
Unit 5. Clayey silt. 10 4/2, greyish yellow brown to 10 3/2,
brownish black.
Unit 11. Coarsesand with pebbles (section 11).
Unit 6. Gravel. At boundary with upper, younger unit generally
matrix supported, coarsening downwards to clast supported.
Archaeomagnetic DatingandPalaeochannelsSediments 153
perpendicular to themajor axis of thechannel at this
location (Figure 1). The upper two samples were
dominated by foliar fabric and the lower two by a
lineated one (Table 3). The ne-grained nature of the
sediment, coupled with theanisotropy of susceptibility
data, shows that deposition was under low velocity
conditions. However, hydrodynamic forces dominated
grain alignment in the lower portion of the unit and
gravity the upper portion. The samples exhibited
scattered K
min
axes of susceptibility and dipping
K
max
(Figure 3); this pattern is indicative of post-
depositional root disturbance of the original planar
fabric. All four samples had low mean intensities of
susceptibility, and post-depositional dissolution and
reprecipitation of Fe-rich minerals was also indicated
by the occurrence of mottles and iron concretions.
Similar depositional and post-depositional processes
aresuggested for unit 2.
Therewas no indication of post-depositional distur-
bance in unit 2/3, section 7. The domination of foli-
ation showed that theunit wasdeposited in still water.
Theeight samples fromunit 3, section 7had a similar
sequence of deposition to unit 2; a reduction in the
inuence of water currents with time. There is no
indication of post-depositional disturbance.
The magnetic fabric of unit 4 was again predomi-
nantly foliar, with no indication of physical post-
depositional disturbance(thiswasalso thecasefor unit
4insections1, 2, 4, 9and11). Occasional sampleswere
dominated by a linear fabric. No discerniblepreferred
orientation of grains in unit 3 (sections 7 and 10) or
unit 4(sections9and11) wasobservedunder theSEM;
thiswasaconsequenceof thene-grainednatureof the
sediment and the domination of a platy, face-to-face,
foliar structure. Horizontal microlaminae, highlighted
by organic matter, were observed in thin section
derived fromsection 4, unit 4.
Table2. Mean directions of theprincipal axis of maximumsusceptibility (K
max
) for unit 2, section 7
Sample Depth (cm) N Dec I nc R
95
Ro
102110 2512535 10 47 21 86 207 643 503 3068
88101 2532575 14 53 18 115 192 527 598 3453
5987 258267 29 31 20 232 136 486 865 3672
3443 26852785 10 47 16 83 233 526 503 3401
N isthenumber of specimens; Dec isthemean declination; I nc isthemean inclination, R istheresultant vector (Butler, 1992);
95
isthecone
of condence(Fisher, 1953), is theprecision parameter (Fisher, 1953), Ro is thesignicant point (I rving, 1964) and theangular standard
deviation (Butler, 1992).
Table3. Parameters of anisotropy of susceptibility for unit 2, section 7
Sample
Mean
K
Mean
A
Mean
L
Mean
F %F
Mean
q
102110 462 040 0013 0060 0059 0045 0076 0031 700 052 029
88101 419 027 0015 0003 0068 0035 0080 0035 714 065 036
5987 488 132 0015 0040 0075 0040 0070 0033 517 074 040
3443 731 247 0020 0006 0105 050 0090 0059 600 082 045
Mean K is themean of thethreeprincipal axes (K
max
, K
int
and K
min
) (Tarling, 1983); Mean A is themean total anisotropy (100) (Owens,
1974); Mean L isthemean degreeof lineation (100), theintensity of linear-parallel orientation (Khan, 1962); Mean F isthemean degreeof
foliation (100), theintensity of theplanarparallel orientation (Khan, 1962); %F is thepercentageof foliation; Mean q is themean shape
of theellipsoids (Granar, 1958) and is thestandard deviation.
270 90
180
0
258267 cm
Figure3. Magnetic fabric of unit 2, sample5987, whereK
max
()
is the principal axis of maximum susceptibility, K
int
() is the
intermediateaxisof susceptibilityandK
min
() theminimumaxisof
susceptibility. Declination is shown as an angle measured in a
clockwisedirectionfromnorth(90') andinclinationasananglefrom
thecentral vertical point (90'). +=Mean.
154 C. EllisandA. G. Brown
Section 9 Similar conditions of deposition and post-
deposition areproposed for unit 2and unit 4as those
described abovefor thesameunits of section 7. How-
ever, rounded clasts of unit 4 wereobserved in unit 9;
thelatter was a well-sorted sand.
Section 10 A similar scenario to that given for unit 2,
section 7 is suggested for unit 2, section 10. However,
theunit 2 of this section was dominated by lineation,
indicatingaslightly higher owregime. Unit 3, section
10was dominated by a foliar fabric. Thebasal sample
of unit 3 exhibited dipping K
max
and K
min
axes of
susceptibility, indicative of a disturbed fabric. The
samplewasderivedfrom2cmaboveasandstoneblock
which formed part of the13th century bridgepier. I t is
probablethat thefabricmirroredtheundulatingbedof
sandstoneblocks.
Archaeomagnetic data
Alternating-eld stepwise demagnetization demon-
strated that all the samples of unit 3 and unit 4 (all
sections) carriedastableNRM (Figure4). Themedian
destructive elds fell in the range of 3040mT. The
nal archaeomagneticresultsfor eachsectionaregiven
in Table 4. The remanence directions of all the
samples, except that of section4, weretightlyclustered.
Theintensity of NRM varied signicantly with lithol-
ogy and the colour of the sediment (Table 5). I t was
found that on oxidation in thelaboratory therewas a
signicant loss of intensity of NRM and magnetic
susceptibilityandachangeintheremanencedirections
in units 3 and 4 (Tables 6 & 7).
Mineralogical data
Thebasic mineral components of unit 4(sections 1, 2,
4, 7, 9 and 11) were largely derived from reworked
Devensianalluvium. X-raydiractionrevealedthat the
dominant minerals were quartz and potassium feld-
spar, with illite/muscovite, chlorite and kaolinite clay
minerals. I t is signicant that no other major iron-
bearing minerals, apart from paramagnetic chlorite,
were detected. X-ray uorescence analysis established
722% Fe and 096% TiO
2
content; vital components
of the titanomagnetites group (Butler, 1992). The
sediment was dominated by 6775% silica, 1628%
aluminium.
Framboids (spherical mineral aggregates) were ob-
served fromunit 4(section 4) in thin section and units
3 and 4 (sections 4, 7, 9 and 10) under the scanning
electron microscope. The framboids were most com-
mon in organic-lined voids and root channels. How-
ever, somewereobserved in thematrix and thesewere
regularly associated with vivianite (Fe
2
(PO
4
)
2
.8H
2
O),
a secondary phosphate occurring in oxidized zones
containing pyrite and organic fragments (Hamilton,
Woolley & Bishop, 1990). The framboids were con-
structed fromindividual octahedron crystals arranged
inspheresof between10and50mindiameter (Figure
5). The framboids were elementally analysed using
EDXA, which identied the major elements as Fe
and S. This is believed to betherst timesuch mono-
sulphuric sediments havebeen recorded in a Holocene
uvial environment of thiskind (seelater discussion of
sulphur mineralogy).
Diatoms of Campylodiscus noricus, Nitzschia species
and Naviculaspeciesand other small unidentied frag-
ments were observed in unit 4, section 4. The mean
organic content of unit 4 (all sampled channels) lay
between 6 and 14%; most of the organic content was
amorphous under themicroscopeand probably origi-
nated fromalgae. Themean soil moisturecontent for
unit 1 was 85%, unit 2 between 9 and 25%, unit 3
between 11and 40%and unit 4between 75and 40%.
H9
270 ,Up
180 , 180
Exp: 3
Sample: H9 North
Mo/M
0 800C, 0e 400 200 600
(b)
(a)
Figure 4(a) and (b). Alternating eld demagnetization of specimen
9, section 7. (a) The points represent orthogonal projection of the
successive positions of the resultant magnetization vector with
increasing peak demagnetizing eld. (b) Variation of the total
intensity and successive directions of magnetization with peak
demagnetizing eld.
Archaeomagnetic DatingandPalaeochannelsSediments 155
The characteristic grain size of the units is given in
Table 7. During grain size analysis magnetic particles
becameattached to amagnetic bean (used to maintain
slit and clay suspension). However, no magnetic par-
ticles were noted when the sediment had been treated
with hydrogen peroxideto removeorganic matter.
Discussion
Depositional environments
The generally planar, foliated nature of unit 4 (in all
the sampled channels) demonstrated that most of the
sediment settled out fromsuspension and that gravita-
tion was the major force acting upon the grains and
occules. The palaeochannels can be envisaged as
sedimentarysinks, wheretrappedoodwater andalgal
growth resulted in stagnant, anoxic bottom-water con-
ditions. Allochthonous and autochthonous organic
matter was readily preserved. Theoccasional lineation
of unit 4 was a result of rare and short-lived inuxes
of sediment-laden ood waters. A sudden increase in
velocity was apparent from the sedimentary log of
section 9; where unit 9 contained rounded lenses of
unit 4, thought to be scoured out as the oodwater
entered the palaeochannel. The deviation of some
meanmagneticfabricdirectionsfromthemajor axisof
agiven palaeochannel isexplained, rstly by thedomi-
nationof foliationandsecondlybythemeasurement of
two or moreowdirections. Theseowdirections are
probably relicsof occasional multidirectional currents,
caused by wind action or secondary events within the
palaeochannels.
Grain size, organic content and water content data
indicate that the depositional environment of unit 3,
section 7 was not signicantly dierent from that of
unit 4, section 7. However, theunits arecharacterized
Table 4. Archaeomagnetic data for palaeochannel sediment, Hemington Fields, North Leicestershire. Hemington Fields, latitude 5287 and
longitude 1317. Dec and I nc arenormalized to Meriden, latitude5234, longitude 162
Hemington
Fields Sample N
Mean
mAm
1

Depth
(cm) Dec I nc Dec I nc
95

Calendar date
(68%condence)
Section 1, unit 4 1 pushed 11 51 30 79110 14 77 136 768 31 217 172550
Unit 4 2 13 124 44 118124 13 76 127 757 29 209 172550
Unit 5 3 8 92 92 1245127 5 75 49 727 35 526 167550
Section 2, unit 4 1 7 143 65 180187 7 62 70 618 40 225 c. 13001350
Unit 4 2 14 148 88 200211 25 63 247 626 20 378 c. 1100
Section 3, unit 3 1 14 22 19 185193 6 58 59 575 22 333 140050
Section 4, unit 4 1 7 35 5 5 69 49 686 47 169 c. 16501450/400600
Section 7, unit 2 (1) 5987 21 5 4 258267 14 65 148 646 20 254 115050
Unit 2 (1) 3358 26 167 15 2565273 19 66 188 656 19 213 100050
Unit 2/3 (1) 113 13 190 95 27652785 13 74 129 737 14 829 c. 700
Unit 2/3 (1) 1525 12 178 110 279281 20 72 198 726 12 1272 c. 900
Unit 3 (1) 2632 7 443 158 28152835 16 70 158 696 14 1974 c. 900
Unit 3 (2) 117 16 51 147 2785283 1 73 09 727 24 237 165050
Unit 3 (2) 1826 9 467 198 28552875 16 77 156 767 14 1397 c. 175050
Unit 3 (2) 2735 9 145 169 288290 7 76 68 757 21 603 170050
Unit 3 (2) 3653 18 200 143 2912955 17 69 168 686 16 476 c. 900/1500
Unit 3 (2) 5471 18 103 70 2965301 31 68 306 676 14 642 100050
Unit 4 (2) 7288 16 90 46 30253065 26 72 257 716 10 1238 c. 90050
Unit 4 (2) 89120 30 206 101 3073165 31 68 306 676 06 1707 100050
Unit 4 (2) 121137 18 70 52 318322 28 69 277 686 14 635 100050
Unit 4 (2) 138143 6 31 14 32283248 19 72 188 716 21 1004 90050
Section 9, Unit 4 163 65 415 168 231248 10 75 99 747 06 840 75050
Unit 4 64117 53 1323 332 248260 20 74 198 737 05 1289 ?
Section 10, unit 3 10, 11, 12 27 76 44 2633 0 58 01 575 15 359 140050
Unit 3 13, 14 17 75 22 3339 5 57 47 566 21 286 140050
Unit 3 16, 17 18 78 25 4246 2 56 19 555 22 257 137550
Section 11, unit 4 135 35 142 103 4534635 14 60 138 595 10 589 122550
(1) and (2) are the monolith numbers froma section 7; N is the number of specimens in a sample; Mean mAm
1
is the mean intensity of
remanence; is thestandard deviation of theintensity of remanence;
95
is theconeof condenceat the95%condencelevel and is the
precision parameter. Section 1, sample 1 is the only sample where cubic specimen holders were pushed directly into the eld section, other
specimens wereobtained by subsampling frommonolith tins.
Table 5. The change in intensity of remanence within and between
dierent coloured lithologies
Unit Colour mAm
1
Unit 2 Dark grey and brown 06316737
Unit 2/3 Dark grey and brown 257190
Unit 3 Greyish yellowbrown 1436467
Unit 4 Black 313162433
mAm
1
is theintensity of remanence.
156 C. EllisandA. G. Brown
by dierent magnitudes of magnetic mineral content.
Unit 3, section 10 was deposited in a deep pool
environment (Pitts, 1995), where the water at the
sediment/water interface was still; however, Berner
(1980) notes that a minimal quantity of oxygen is
needed for the formation of pyrite framboids, which
also only formunder pH 78 (Marnette et al., 1993).
The avulsion and abandonment of the channel was
probablyaconsequenceof aoodwhichdestroyedthe
13th century bridge and resulted in the deposition of
unit 3. Field observationsand magnetic fabric analysis
indicate that unit 2/3 of section 10 had experienced
post-depositional turbation and the dissolution and
oxidation of iron compounds. However, the archaeo-
magnetic data indicatethat thepost-depositional pro-
cesseswerenot of asignicant magnitudeto distort or
destroy theNRM.
The presence of Campylodiscus noricus diatoms is
indicativeof alkalineconditionsof pH 7or greater and
they are common in standing or nutrient-rich water.
The Nitzschia species are a large group and are most
commonineutrophicwater. TheNaviculaspeciesarea
varied group of diatoms and are generally found in
runningwater with a high nutrient content and a wide
range of pH; however, some do prefer alkaline or
neutral pH. Thediatomsconrmthealkalinenatureof
thesediment and water.
Archaeomagnetism
Thecyclic natureof theBritish archaeomagnetic refer-
encecurvehasresultedinanumber of tight curvesand
loops, which can result in alternative dates for one
independent sample. A second drawback of therefer-
ence curve is that the distance between each half
century marker is not uniform and so when a data
point fallsonto theinterpolatedcurveandthe
95
cone
of condenceis drawn, the in years aredicult to
estimate. The amount of change in secular variation
is dierent in specic periods, for example the line
between 1500 and 1000 is rather large, while that
between 600 and 700 is tiny and therefore the
assignment of a moreprecisedateis morelikely in the
former than in thelatter. Coupled with thisproblemis
the resolution of the curve; some of the curve is
constructed from many data points (independently
dated) while others have very few anchor points. For
example Batt (1992) noted that there were 50 points
between 0and 250but only four pointsbetween
500 and 750; again this will aect the reliability of
dating independent samples. The archaeomagnetic re-
sults from Hemington Fields are given below, and
because of the lack of sequence dated levels coupled
with the problems of curve, data errors are broadly
estimated to 50 years at the68%condencelevel.
Table 6. Mean loss of intensity of remanence and volume susceptibility and mean remanence direction change on
oxidation for units 3 and 4, Hemington Fields
Section and unit
Mean %loss
mAm
1
(dry)
Mean loss vol
SI (dry)
Mean Dec
dierence
Mean I nc
dierence
Section 9, unit 4 9990 005 9671 061 5214 2736 3717 2589
Section 10, unit 3 9884 072 8746 423 4691 7648 1816 1641
mAm
1
istheintensity of remanence; vol isthevolumesusceptibility; Dec isthedeclination; I nc theinclination
and thestandard deviation.
Table 7. Characteristic grain size for the major sedimentary units,
Hemington Fields
Unit Grain size
Unit 1 Silty clay
Unit 2 Silty sand
Unit 3 Silty sand
Unit 4 Sandy silt
Figure 5. Pyrite framboids in a degraded root channel, unit 3,
section 10. Scalein m.
Archaeomagnetic DatingandPalaeochannelsSediments 157
Section 1 Theupper two samples derived fromunit 4,
section 1, have been dated to 172550. Unit 5,
stratigraphically lower than unit 4, has been dated to
167550years. I t can thus beconcluded fromthe
archaeomagnetic evidence that the channel was aban-
doned shortly before 167550. I t is interesting to
note at this stage that the samples comprising eld-
pushedcubicspecimenshavealower meanintensityof
remanence than those specimens cut from an undis-
turbed column. The lower mean moment was caused
by a loss of moistureover a period of 2 weeks before
measurement andthehigher meanintensitywasdueto
amoreeectiveretention of moisturewithin thelarger
undisturbed sampleand themeasurement of thespeci-
mensdirectly after subsampling. However, themethod
of sampling, pushing cubic specimen holders directly
into thesediment versus subsamplingfroma monolith
tin, did not aect thearchaeomagnetic data.
Section 2 The two sample mean directions are signi-
cantly dierent at the 95%condence level. Sample 1
has a high
95
level and lies at thelimit of acceptable
data. Neither samplewith thecondencelimitsplotted
intersected the curve, preventing the assignment
ofyears. Thetwo archaeomagnetic dates for unit 4,
section 2 may lie some 175325 years apart, over a
maximumof 13cm. Thus it is possiblethat remanence
within this unit has been acquired gradually and
sequentially.
Section 3 Due to a crossover in the reference curve
there are two possible dates for unit 3, section 3:
140050 and 127550. The stratigraphy of the
section is characteristic of the top of a point bar
(Figure 2), where unit 3 and unit 7 are covered by
overbank alluvium(unit 1). I t is likely that the point
bar is a relic of the westwardly meandering 13th
centurypalaeochannel (Figure2). I f so, theearlier date
of 127550 is more probable. This date ts with
tentative dating on stratigraphical and archaeological
grounds (Salisbury, pers. comm.).
Section 4 The archaeomagnetic results fromsection 4
wereinconclusive. Thesampled unit was covered by c.
13th century gravels and silts and so a date of c.
1450can berejected. Mappingindicated that section 4
and section 7 were the same channel (Salisbury &
Brown, pers. comm.). Assumingthistobecorrect, then
a date of c. 400600, although rather broad, is in
agreement withthesedimentaryandgeomorphological
data. This dateis also in reasonableagreement with a
radiocarbon date on brushwood with bark from the
base of the ll (cal 605, 1 sigma cal 550645
Beta-87291).
Section 7 Thebaseof unit 4, section 7 has been dated
to 90050, rising to 100050 at a depth of
301cm, some21cmabovethelowermost sample. The
sediment acquired a remanence over a maximum
period of 150 years; unfortunately the evidence of
sequential acquisition is minimal. Sample 5471, unit
3, yielded a date of 100050 and has the same
valuesof declinationandinclinationassample89120,
unit 4. This suggests that theremanenceacquisition in
unit 4and at thebaseof unit 3was simultaneous. The
data point of sample7288does not fall onto or cross
theinterpolated referencecurve(Figure6). I t could be
argued that the sediment shows a systematic inclina-
tion error (systematic becausethe
95
of thesampleis
very small). However, when viewed within thecontext
of theunit, thissampleisnot perceivedasproblematic.
The base of unit 2 is dated to 100050. An
archaeomagnetic anomaly within the sequence is
present in unit 3 (with the exception of the lower
sample) and unit 2/3. The dates for the central three
samples of unit 3 aresimilar and thereforeremanence
acquisition was concurrent. However, the samples
below, as well as thoseabove, areapparently younger
and so there is no stratigraphical progression of the
dates (Table 4). Stepwise AF demagnetization was
performed on eight specimens from units 2/3 and 3.
Thesediment carriedaverystableNRM (040/50mT),
with no indication of a viscous or weaker secondary
chemical remanence. Of interest is the intensity of
remanence, generally higher in units 2/3 and 3 than in
units2or 4. I t isthought that prior to c. 1500most
of unit 3and unit 2/3experienced a post-depositional,
geochemical transformation, resulting in the produc-
tionof astrong, secondarychemical remanence. Unit 2
carries a stable and credible remanent magnetization.
The action of post-depositional processes, such as
seasonal wetting and drying and root penetration, are
attested by the presence of mottles, iron concretions
and a disturbed magnetic fabric; but these processes
40
50
80
30
Declination
I
n
c
l
i
n
a
t
i
o
n
10
55
60
65
70
75
20 0 10 20 30
1950 AD
600 AD
1200 AD
Figure6. Thearchaeomagnetic datafromsection 7plotted onto the
British archaeomagnetic reference curve (digitized from Clark,
Tarling & Noel, 1988), where each 100 years is marked by a solid
square. Theerror barsarecalculatedfromthe
95
andshowthecone
of condenceat the95%level. &, , Clark, Tarling& Noel, 1988;
+, section 7.
158 C. EllisandA. G. Brown
have not destroyed the magnetic remanence. The low
intensity of remanence of the top sample may be the
result of oxidation and destruction of the magnetic
content, a reection of the environment (which may
not have favoured magnetotactic bacteria), or even a
consequenceof a detrital sourceof magnetic grains.
Section 9 I nitial data analysis revealed that the 12
samplescould begrouped into two populations(Table
4). However, two samples, sample 4654 (sampled
from a depth of 243245cm) and sample 108117
(sampled from a depth of 258260cm), had similar
values of declination (016' dierence) and inclination
(069' dierence). Thisindicatesthat theundatedlower
sample (undated because the remanence data did not
intersect the reference curve), despite a declination
error, is of a similar date to the top sample,
75050. Thereasonfor theapparent declinationerror
is unclear. One possible explanation is the incorpor-
ationof sandlensesfromunit 9; thealignment of these
by hydrodynamic forces could have physically dis-
torted theremanence. However, AMS iscarried by the
larger multi-domain grains and the NRM by the
smaller multi-domain and single-domain grains and
the sample is dominated by foliar ellipsoids. Also,
given thelowdegreeof anisotropy (0007), it does not
seempossiblethat coarser sand grains aremasking or
distortingtheremanence. Themean magnetic moment
of the lower original samples was much greater than
the upper samples, ranging between a minimum of
315mAm
1
andmaximumof 1309mAm
1
higher. I t
issuggested thedeclination error may berelated to the
higher intensity of remanence phenomenon. This
would indicatethat therewas a magnetic and perhaps
geochemical dierence within the unit of a post-
depositional origin.
Section 10 The archaeomagnetic curve is rather com-
plexaround 12501450. Datapointsof varyingages
are clustered (see Clark, Tarling & Noel, 1988). The
useof thereferencecurve, without considerationof the
archaeological data, would yield a tentative calendar
date for unit 3, section 10, of c. 12501300. How-
ever, a dendrochronological felling date for timber
used in theconstruction of thebridgeisthelate1230s.
Therefore, it is reasonable to assume that the date
above is too close to that of the construction of the
bridge. This date would allow only a minimumof 60
yearsof use; without takinginto considerationthetime
for the construction of the bridge, its destruction, the
deposition of unit 3and theacquisition of remanence.
A large scale ood was documented to have aected
this area in 1401 and is proposed as the one
that destroyed the bridge (Brown, 1996; Brown &
Salisbury, in press). A dateof 140050 correlates
well withthearchaeological andhistorical data. Unit 3
must have been deposited fairly rapidly and the
remanence, if not detrital, was acquired soon after.
Sulphidemineralogy
Determinationof themagneticmineralogyisnecessary
beforedrawingconclusions regardingtheorigin of the
natural remanent magnetization (NRM). Studies of
recent sediment have shown that generally detrital
magnetite (titanomagnetites and titanomaghaemites)
is theprincipal magnetic phasepresent and thecarrier
of depositional (DRM) or post-depositional (PdRM)
remanent magnetization (Batt, 1992; Butler, 1992).
Thepotential ferromagnetic and ferrimagnetic content
of the source rocks for the Hemington palaeochannel
sediment is lowand thene-grained detritus fromthe
Merciamudstonehas alowmean susceptibility, domi-
nated by haematite. However, the volume suscepti-
bilities and intensities of NRM of units 3 and 4 were
exceptionally high and so another, non-detrital, mag-
netic source has to be sought. Other main remanence
carriers in recent sediments are iron oxides (maghae-
mite and haematite), the ironmanganese oxyhydrox-
ides and certain iron sulphides, for example
ferrimagnetic greigite (Fe
3
S
4
), which are associated
with in situ authigenic or diagenetic chemical changes
andso will carry achemical remanence. Becauseof the
similar grain size of magnetite and greigite and very
similar AF demagnetization behaviour, medium de-
structive eld >38 and 35mT, respectively (Snowball
& Thompson, 1990; Fassinbinder & Stanjek, 1994;
Reynolds et al., 1994), it is very dicult to distinguish
between thetwo minerals.
I t ispostulatedthat thebrowncolour of unit 2isdue
to hydroxides and theblack colour of unit 4 is dueto
the presence of iron monosulphide minerals. The
former can beclassied as thehydroxidezoneand the
latter as the monosulphuric zone, as dened by Love
(1967). Unit 3, the grey sediment, is analogous to
the bisulphuric or pyrite zone (Love, 1967) which is
notablefor thepresenceof pyriteframboids (Al-agha
et al., 1995).
Sulphideformation isprimarily abiological process,
requiring a supply of dissolved sulphate, organic
matter and iron (iron oxideand hydroxide) (Vaughan
& Lennie, 1991; Sawlowicz, 1993); these conditions
were fullled in units 3 and 4. The breakdown of
organic matter is achieved by the bacterial reduction
of sulphate, via bacterial anaerobic decomposition
of organic matter and the production of hydrogen
sulphide, H
2
S. The H
2
S reacts with iron minerals to
produceblack monosulphides (Berner, 1970).
Anoxic sedimentary conditions most often occur in
theupper few millimetres of a sediment (Leslie, Lund
& Hammond, 1990) and especially in oxygen-depleted
waters (Karlin, 1990). Geochemical processes of or-
ganic matter degradation and sulphur reduction have
beenshownto aect themagneticmineral composition
of sedimentsintwo ways. Therst isthedissolutionof
magnetic grains and the second is the production of
authigenic magnetic minerals which can actually in-
crease the ferromagnetic/ferrimagnetic concentration,
Archaeomagnetic DatingandPalaeochannelsSediments 159
increasing theintensity of magnetization and thepro-
duction of a CRM (Snowball, 1991). Conditions for
monosulphide production can be envisaged for the
Hemington palaeochannels, where trapped ood
waters in an organic-rich environment resulted in the
depositionof anorganic-rich, siltysandunder reducing
conditions.
None of the monosulphides are thermodynamically
stable (Berner, 1970). A phenomenon often noted in
lake sediments is a loss in the intensity of NRM and
magneticsusceptibility. Thisisthought to bedueto the
oxidation of labilegreigiteand biological oxidation of
pyrite (Hilton et al., 1986; Hilton, 1990; Snowball &
Thompson, 1990; Crockford & Willet, 1995). On air
drying there was dramatic loss of intensity of NRM
and volumesusceptibility in both units 3 and 4 and a
dramatic change in remanence direction (Table 6).
However, a loss of susceptibility and a change in
remanence direction during oxidation should not be
signicant if single-domain magnetite oxidizes to
maghaemite(Henshaw& Merrill, 1979). Hilton (1990)
shows that a loss in susceptibility and intensity in an
anoxic lakesediment is caused partially by themicro-
bial degradation of magnetite under oxidizing con-
ditions and the oxidation of authigenically formed
greigiteto alessstrongly magnetic form. Hilton (1990)
observed a large intensity loss when the sediment
was treated with hydrogen peroxideand subsequently
dried (this was also observed in unit 4). This phenom-
enon was deemed consistent with the presence of
greigite. XRD analysis of the anoxic Esthwaite lake
sediment revealed no evidence of a magnetic material
(aswasthecaseinunit 4), but it wassuggestedthat this
was due to the total Fe content being lower than the
5% needed to register a reasonable signal (Hilton,
1990). However, extraction of the magnetic fraction
from freeze-dried samples by Hilton (1990) and
Snowball & Thompson (1990) revealed the presence
of greigite.
The formation of greigite demands strong reducing
conditions, but not totally without oxygen (Al-agha
et al., 1995). The production or formation of greigite
can rstly beof an inorganic origin, biologically medi-
ated by sulphate-reducing bacteria and, secondly, the
result of the diagenesis of bacterial magnetite (Mann
et al., 1990; Homan, 1992). Single domain magnet-
ites, <01m, pyrrhotite (FeS
1+x
), greigite and pyrite
areproduced by magnetotactic bacteria (seeChang &
Kirschvink, 1989; Heywood et al., 1990; Stolz, Lovley
& Haggerty, 1990; Bazylinski et al., 1991; Moskowitz,
Frankel & Bazylinski, 1993; Bazylinski, Frankel &
Garratt-Reed, 1994). Most of thesemagnetotacticbac-
teria require molecular oxygen and produce magnetic
minerals intracellularly in theformof a chain within a
magnetosome(Stolz, Lovley & Haggerty, 1990). Other
forms of magnetotactic bacteria containingsomeform
of iron sulphide have been identied in hydrogen
sulphide-rich, anoxic sediments (Bazylinski et al.,
1991). Magnetotactic bacteria have also been found
to produce ferrous phosphate, vivianite, through
biologically induced mineralization. The vivianite is
thought to be a by-product of the export of ferrous
ions (Bazylinski et al., 1991). The magnetic minerals
producedbymagnetotacticbacteriaareusedfor orien-
tation. Many of the bacteria are located in the upper
few millimetres of sediment. On death they fall out
of suspension and may align with the ambient geo-
magnetic eld to produce a DRM. Both anaerobic
magnetotacticbacteriaandsulphate-reducingbacteria,
such as GS-15, are proposed as primary producers of
authigenic iron sulphides and magnetite, respectively,
in anoxic sediments (Stolz, Lovley & Haggerty, 1990;
Bazylinski et al., 1991).
Post-depositional greigite has been observed in
Cretaceousstrata(Reynoldset al., 1994), withinrecent
anoxic marine sediments (Leslie, Lund & Hammond,
1990) and recent lacustrine deposits (Hilton, 1990;
Snowball & Thompson, 1990; Snowball, 1991).
Homan (1992) argues that greigite, rather than being
a relatively raremineral, is in fact common in anoxic,
sulphide-rich sedimentsassociated with organic matter
and is often observed with other non-magnetic sul-
phides such as pyrite. Pyrite is cubic in form and
paramagnetic at room temperature. The presence of
pyrite framboids in old root channels and ssures
showed that they formed after root penetration and
werethereforeauthigenicor diageneticinorigin. Pyrite
framboids form indirectly from or via iron monosul-
phides (Sawlowicz, 1993). Sweeney & Kaplan (1973)
and Vaughan & Lennie(1991) suggest that framboids
develop when the initial iron sulphide precipitate is
transformed to greigite. Berner (1970) suggests pyrite
only occurs when thereis an excess of sulphur and the
sediment is characteristically grey (e.g. unit 3), but it
will not formwhen there is no excess of sulphur and
thenthesediment ischaracteristicallyblack andanoxic
(e.g. unit 4). Theconsensus is that pyriteforms at the
interface between oxic/anoxic zones and that vital
ferrous and sulphideions aresupplied by thereducing
sediment and the oxidizing component from the
oxidized zone. I t isperhapstheseprocessesthat can be
envisaged in unit 3 at Hemington.
Theremanencecarrier in units 3and 4has not been
formally identied, but preliminary results indicated
that an iron monosulphide, probably greigite, is re-
sponsible for the exceptional quality of the archaeo-
magnetic data. Theauthigenic or diagenetic formation
of greigitenecessitates that theremanenceis chemical;
however, the timing of this remanence is dicult to
determine. Hallam & Maher (1994) suggest that
greigitewasformedduringearlydiagenesisandso they
regardtheremanenceasprimary, rather thanasecond-
ary overprint. The archaeomagnetic date of unit 3,
section 10, is some100 years after theconstruction of
the bridge, so it seems reasonable to presume that if
greigite is the remanence carrier, its authigenic or
diagenetic production took place soon after sediment
deposition.
160 C. EllisandA. G. Brown
Conclusions
(1) Relic features of depositional and post-
depositional disturbance (mostly in units 2 and
3, sections 1, 2, 7 and 10 in the form of rootlet
penetration and chemical alteration) were ident-
ied using AMS and thin section analysis. Unit 4
(sections1, 2, 4, 7, 9and 11) wasgenerally charac-
terizedbyafoliated, still water, depositional fabric.
However, in some instances wind or secondary
currents are thought to have aected the orienta-
tion of thelarger mineral grains and may account
for the general lack of correlation between the
orientation of channel banks and the mean AMS
directions. Low velocity conditions dominated
sediment deposition in the palaeochannels. The
presenceof water in thepalaeochannels was main-
tained by seepage from the main channel and
occasional overbank ooding.
(2) The use of a variety of techniques has led to the
conclusion that unit 4(sections1, 2, 4, 7, 9and 11)
was monosulphuric. Anaerobic depositional and
post-depositional conditions prevailed with the
authigenic/diagenetic production of monosulphide
minerals, includinglimited pyriteformation. I t has
been postulated that the origin of the magnetic
mineral(s) was linked to magnetotactic bacteria or
iron-reducing non-magnetotactic bacteria, rather
than detrital in origin. I t is also thought that the
main remanence carrier was greigite. The grey
sediment of unit 3 has been classied as bisul-
phuric, where sediment depositional and post-
depositional processes took place in oxygen-poor
conditions. These conditions were suitable for the
formation of monosulphide minerals and pyrite
framboids.
(3) The remanent magnetization of all the units was
very stable and, despite the rather complex post-
depositional processes envisaged for units 3 and 4
(sections 3 and 10 and 1, 2, 4, 7, 9 and 11
respectively), all the units produced statistically
acceptableremanencedataandmost wereassigned
calendar dates which were in general agreement
with other evidence of palaeochannel age. How-
ever, theprocessesof magnetic mineral production
and the verication of the remanence carrying
mineral(s) require clarication. I t has been found
that weak magnetic fabrics, whether depositional
or post-depositional in origin, did not mask or
distort theNRM.
(4) The data from Hemington indicate that palaeo-
channelsandox-bowlakes, which areoften closely
associated with settlement, areideal environments
for successful archaeomagnetic dating, especially
when thesediment hasremained waterlogged since
deposition. Sediment fabric and geochemical
analyses suggest why this environment is so suit-
able for archaeomagnetic studies. Such sediments
aretypicallyfoundonandnear lowlandoodplain
sites (Brown & Keough, 1992) where they have
frequently been utilized for a variety of purposes
from ax retting to refuse disposal. Such sites in
conjunctionwithadjacent buriedlandsurfacesand
structures can providehigh quality environmental
data if adequately dated.
Acknowledgements
Theauthorswouldliketothank ChrisSalisburyfor his
helpandenthusiasmandP. Clayandother membersof
theLeicestershireArchaeological Field Unit as well as
the land owners Ennemix Ltd. The work was done
whileoneof us (CE) was in receipt of a ScienceBased
Archaeology/NERC research award.
References
Al-agha, M. R., Burley, S. D., Curtis, C. D. & Esson, J . (1995).
Complex cementation textures and authigenic mineral assemblage
in recent concretions from the Lincolnshire Wash (east coast,
U.K.), driven by Fe(0) and Fe(I I ) oxidation. J ournal of the
Geological Society, London 152, 157271.
Bascomb, C. L. (1974). I n (B. M. Avery & C. L. Bascombe, Eds)
Soil Survey. Technical Monograph No. 6. Soil Survey Laboratory
Methods. Harpenden, London.
Batt, C. M. (1992). Archaeomagnetic dating: investigations of new
materials and techniques. Ph.D. Thesis, University of Durham.
Batt, C. M. & Noel, M. (1991). Magnetic studies of archaeological
sediment. I n(P. Budd, B. Chapman, L. J ackson, R. C. J anaway&
B. S. Ottaway, Eds) Archaeological Sciences. 1989 Proceedings
of a Conference on the Application of Scientic Techniques and
Archaeology, pp. 234241.
Bazylinski, D. A., Frankel, R. B., Garratt-Reed, A. J . & Mann, S.
(1991). Biomineralization of iron sulphides in magnetotactic
bacteria from suldic environments. I n (R. B. Frankel & R. P.
Blakemore, Eds) I ron Biominerals. New York: Plenum Press,
pp. 239255.
Bazylinski, D. A., Frankel, R. B. & Garratt-Reed, A. J . (1994).
Electron microscope studies of magnetosomes in magnetotactic
bacteria. Microscopy Research and Technique27, 389401.
Berner, R. A. (1970). Sedimentary pyrite formation. American
J ournal of Science268, 123.
British Standards I nstitution (1967). Methods of Testing of Soils for
Civil Engineering Purposes. Gr 10. British Standards I nstitution,
BS1377.
Brown, A. G. (1996). Human dimensions of palaeohydrological
change. I n (K. J . Gregory, J . Branson & A. G. Brown, Eds)
Palaeohydrology and Global Change. London: Geological Society
Special Publications 115, 257265.
Brown, A. G. (1997). Alluvial Geoarchaeology: FloodplainArchaeol-
ogy and Environmental Change. Cambridge University Press,
Cambridge.
Brown, A. G. & Keough, M. K. (1992). Palaeochannelsand palaeo-
landsurfaces: the geoarchaeological potential of some Midland
(U.K.) oodplains. I n (S. Needham & M. Macklin, Eds)
Archaeology Under Alluvium. Oxford: Oxbow, pp. 185196.
Brown, A. G. & Salisbury, C. R. (in press). Geomorphology. I n
(P. Clay, Ed.) Hemington Bridges Project. Leicester: Leicester
University Archaeology Unit.
Butler, R. F. (1992). Palaeomagnetism: Magnetic Domains and
Geological Terranes. Boston: Blackwell Scientic Publications.
Chang, S. R. & Kirschvink, J . L. (1989). Magnetofossils, the
magnetisation of sediments and the evolution of magnetite bio-
mineralisation. Annual Review of Earth Planetary Science 17,
169195.
Archaeomagnetic DatingandPalaeochannelsSediments 161
Clark, A. J . (1990). Archaeomagnetic measurements. I n (M. Bell,
Ed.) Brean Down Excavations, 19831987. English Heritage:
Archaeological Report no. 15, 113116.
Clark, A. J . (1992). Magnetic dating of alluvial deposits. I n (S.
Needham& M. G. Macklin, Eds) Alluvial Archaeology inBritain.
Oxford: OxbowBooks. OxbowMonograph 27, 3742.
Clark, A. J ., Tarling, D. H. & Noel, M. (1988). Developments
in archaeomagnetic dating in Britain. J ournal of Archaeological
Science15, 645667.
Clay, P. & Salisbury, C. R. (1990). A Norman mill damand other
sites at Hemington Fields, Castle Donnington, Leicestershire.
Archaeology J ournal 147, 276307.
Cooper, L., Ripper, S. & Clay, P. (1994). Bridging the gap to
Domesday at Hemington. Leicestershire Archaeological News 8,
47.
Cooper, L., Ripper, S. & Clay, P. (1995). The Hemington bridges.
Current Archaeology 140, 316321.
Crockford, R. H. & Willet, I . R. (1995). Dryingandoxidationeects
of the magnetic properties of suldic material during oxidation.
Australian J ournal of Soil Research 33, 1929.
Ellis, C. (1995). The archaeomagnetism of ne-grained alluvial
sediment. Ph.D. Thesis, University of Leicester.
Ellwood, B. B. (1984). Anisotropyof magneticsusceptibility: empiri-
cal evaluation of instrument precision. Geophysical Research
Letters 11, 645648.
Fassinbinder, J . W. E. & Stanjek, H. (1994). Magnetic properties of
biogenic soil greigite (Fe
3
S
4
). Geophysical Research Letters 21,
23492352.
Fisher, R. (1953). Dispersion on a sphere. Proceedings of theRoyal
Society A217, 295305.
Granar, L. (1958). Magnetic measurements on Swedish varved
sediments. Arkiv foer Geofysik 3, 140.
Gravenor, C. P., Symons, D. T. A. & Coyle, D. A. (1984). Errors in
theanisotropy of magnetic susceptibility and magnetic remanence
of unconsolidated sediments produced by sampling methods.
Geophysical Research Letters 11, 836839.
Hallam, D. F. & Maher, B. A. (1994). A record of reversed polarity
carried by the iron sulphide greigite in British early Pleistocene
sediments. Earth and Planetary ScienceLetters 121, 7180.
Hamilton, N. (1967). The eect of magnetic and hydrodynamic
control on the susceptibility anisotropy of redeposited silts.
J ournal of Geology 75, 738743.
Hamilton, N. & Rees, A. I . (1970). The use of magnetic fabric in
palaeocurrent estimation. I n (S. K. Runcorn, Ed.) Palaeogeo-
physics. London, NewYork: Academic Press, pp. 454464.
Hamilton, W. R., Woolley, A. R. & Bishop, A. C. (Eds) (1990).
Minerals. Rocks and Fossils. London: Hamlyn Publishing Group
Ltd.
Henshaw, P. C. & Merrill, R. T. (1979). Characteristics of drying
remanent magnetizationinsediments. EarthandPlanetaryScience
Letters 43, 315320.
Heywood, B. R., Bazylinski, D. A., Garrat-Reed, A., Mann, S. &
Frankel, R. B. (1990). Controlled biosynthesis of greigite(Fe
3
S
4
)
in magnetotactic bacteria. Naturwissenschaften 77, 536538.
Hilton, J . (1990). Greigite & the magnetic properties of sediments.
Limnology Oceanographic 35(i), 502520.
Hilton, J ., Long, G. J ., Chapman, J . S. & Lishman, J . P. (1986). I ron
mineralogy in sediments. A Mossbauer study. Geochemica et
Cosmochimica Acta 50, 21472151.
Homan, V. (1992). Greigite (Fe
3
S
4
): magnetic properties and rst
domain observations. Physics of theEarthandPlanetary I nteriors
70, 288301.
Hrouda, F. (1982). Magnetic anisotropy of rocksand itsapplication
in geology and geophysics. Geophysical Survey 5, 3782.
I rving, E. (1964). Palaeomagnetismand its Application in Geological
and Geophysical Problems. New York, London, Sydney: J ohn
Wiley & Sons.
J ohns, M. & J ackson, M. (1991). Compositional control of aniso-
tropy of remanent and induced magnetisation in synthetic
samples. Geophysics Research Letters 18, 12931296.
Karlin, R. (1990). Magnetic mineral diagenesis in suboxic sediments
at Bettis Site, NE Pacic sea. J ournal of Geophysical Research 95
(B4), 44214436.
Khan, M. A. (1962). The anisotropy of magnetic susceptibility of
some igneous and metamorphic rocks. J ournal of Geophysical
Research 67, 28732885.
Leslie, R. W., Lund, S. P. &Hammond, D. E. (1990). Rock magnetic
evidence for the dissolution and authigenic growth of magnetic
minerals within anoxic sediments of the California continental
borderland. J ournal of Geophysical ResearchSolid Earth and
Planets 95(NB4), 43834389.
Love, L. G. (1967). Early diagenetic iron sulphide in recent
sediments of theWash (England). Sedimentology 9, 327352.
Mann, S., Sparks, N. H. C., Frankel, R. B., Bazylinzki, D. A. &
J amasck, H. W. (1990). Biomineralisationof ferrimagneticgreigite
(Fe
3
S
4
) and iron pyrite(FeS
2
) in magnetotactic bacterium. Nature
343, 258261.
Marnette, E. C. L., Van-Breemen, N., Hordijk, K. A. &Cappenberg,
T. E. (1993). Pyrite formation in two freshwater systems in the
Netherlands. Geochemica et Cosmochimica Acta 57, 41654177.
Moskowitz, B. M., Frankel, R. B. & Bazylinski, D. A. (1993). Rock
magnetic criteria for the detection of biogenic magnetite. Earth
and Planetary ScienceLetters 180, 283300.
Noel, M. (1980). Surfacetensionphenomenainthemagnetisationof
sediments. Geophysical J ournal of theRoyal Astronomical Society
62, 1525.
Noel, M. & Batt, C. M. (1990). A method for correctinggeographi-
cally separated remanencedirections for thepurposeof archaeo-
magnetic dating. Geophysical J ournal I nternational 102, 753756.
Owens, W. H. (1974). Mathematical model studies on factors
aecting the magnetic anisotropy of deformed rocks. Tectono-
physics 24, 115131.
Pitts, F. (1995). A palaeoecological reconstruction of theRiver Trent
using subfossil insect remains. B.Sc. Dissertation, University of
Loughborough.
Rees, A. E. (1961). The eects of water currents on the magnetic
remanence and anisotropy of susceptibility of some sediments.
Geophysical J ournal of theRoyal Astronomical Society 5, 235251.
Reynolds, R. L., Tuttle, M. L., Rice, C. A., Fishman, N. S.,
Karachewski, J . A. & Sherman, D. M. (1994). Magnetisation and
geochemistry of greigite bearing Cretaceous strata, North Slope
Basin, Alaska. Americal J ournal of Science294, 485528.
Salisbury, C. R. (1987). A watching brief at Hemington Fields,
Castle Donnington. Leicestershire Archaeological and Historical
Society. Transactions LXI, 9091.
Salisbury, C. R. (1988). A watching brief at Hemington Fields,
Castle Donnington. Leicestershire Archaeological and Historical
Society. Transactions LXII, 7476.
Salisbury, C. R. (1995). Floodplain archaeology: the excavation of
Hemington Fields. Current Archaeology 145, 3437.
Sawlowicz, Z. (1993). Pyrite formation and their developmenta
newconceptual mechanism. GeologischeRundschau 32, 148156.
Snowball, I . F. (1991). Magnetic hysteresis properties of greigite
(Fe
3
S
4
) and a new occurrence in Holocene sediments from
SwedishLappland. Physicsof theEarthandPlanetaryI nteriors68,
3240.
Snowball, I , & Thompson, R. (1990). A stable remanence in
Holocene sediments. J ournal of Geophysical Research 95 (B4),
44714479.
Stolz, J . F., Lovley, D. N. & Haggerty, S. E. (1990). Biogenic
magnetite and the magnetisation of sediments. J ournal of
Geophysical Research 95, 43554361.
Sweeney, R. E. & Kaplan, I . R. (1973). Pyriteformation; synthesis
and marinesediments. Economic Geology 68, 618634.
Tarling, D. H. (1983). Palaeomagnetism. Principles andApplications
inGeology, Geophysics andArchaeology. London and NewYork:
Chapman and Hall.
Tarling, D. H. & Dobson, M. J . (1995). Archaeomagnetism: anerror
assessment of red material observations in British directional
database. J ournal of Geomagnetismand Geoelectricity 47, 518.
162 C. EllisandA. G. Brown
Tucker, M. (1989). Techniques in Sedimentology. Oxford: Blackwell
Scientic Publications.
Turner, G. M. & Thompson, R. (1981). Lake sediment record of
the geomagnetic secular variation in Britain during Holocene
times. Geophysical J ournal of the Royal Astronomical Society 65,
703725.
Torsvik, T. H. (1992). I nteractiveAnalysis of Palaeomagnetic Data.
(Unpublished handbook and software.), N.G.V., P.B.3006 Lade,
N-7002 Trondheim, Norway.
Vaughan, D. J . & Lennie, A. R. (1991). Theiron sulphideminerals;
their chemistry and rolein nature. ScienceProgress Edinburgh75,
371388.
Archaeomagnetic DatingandPalaeochannelsSediments 163