Organic geochemical evidence for the origin of ancient ...chidb/personal/content/paper8.pdf · Organic geochemical evidence for the origin of ancient anthropogenic soil deposits at

Organic geochemical evidence for the origin of ancientanthropogenic soil deposits at Tofts Ness, Sanday, Orkney

Ian D. Bulla, Ian A. Simpsonb, Stephen J. Dockrill c, Richard P. Eversheda,*aOrganic Geochemistry Unit, School of Chemistry, University of Bristol, Bristol, BS8 1TS, UK

bDepartment of Environmental Science, University of Stirling, Stirling, FK9 4LA, UKcDepartment of Archaeological Science, University of Bradford, Bradford, BD7 1DP, UK

Received 12 October 1998; accepted 11 February 1999

(Returned to author for revision 13 November 1998)

Abstract

Lipid biomarker components of soils constituting three Orcadian archaeological fossil soil pro®les were analysed.

The combined assessment of lipid distributional and compound speci®c stable carbon isotope data enabled theidenti®cation of grass turves as the most probable material used in the formation of the anthropogenic soil deposits.Appraisal of 5b-stanol components indicated a faecal input to one of the soils which, on considering distributional

evidence, was ascribed a human/porcine origin. Additional study of polar bile acids from this pro®le revealed adistribution exhibiting a predominance of deoxycholic acid indicating the primary faecal input to be mainly derivedfrom humans although the minor occurrence of hyodeoxycholic acid, a characteristic component of pig faeces,

attested to a limited porcine input. # 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Archaeology; Bile acids; Biomarkers; Lipids; Stable carbon isotopes; Soils; Stanols

1. Introduction

Analyses of archaeological soils have largely paral-leled those of more general soil formations in the pre-ferential use of bulk variables such as TOC, totalphosphate and other elements (Farswan and Nautiyal,

1997; Entwistle and Abrahams, 1997).Understandably, these types of analysis are attractive,the potentially high throughput of samples favouring

their joint use with physical prospective methods inexamining former settlement sites. In contrast, lipid

analysis is more time-consuming but provides a highlydiagnostic means of answering speci®c questions relat-

ing to subtle di�erences in soil organic matter engen-dered by various overlying vegetation changes oranthropogenic activity such as manuring (van Bergenet al., 1997; Bull et al., 1999 and references therein).

One prominent application of organic geochemicalmethods has been in the detection of ancient inputs offaecal matter to soils through the use of highly speci®c

biological markers, i.e. 5b-stanols and bile acids. Thepresence of characteristic 5b-stanol signatures in soilshas already enabled positive identi®cations of ancient

latrines and cess-pits to be made (Knights et al., 1983;Bethell et al., 1994; Evershed and Bethell, 1996), whilstsimilar analyses of ancient Cretian agricultural terracesoils have established the existence of an o�-site man-

uring regime in the Minoan period (Bull et al., in

Organic Geochemistry 30 (1999) 535±556

0146-6380/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.

PII: S0146-6380(99 )00020-0

* Corresponding author. Tel.: +44-117-9287671; fax: +44-

117-9251295.

E-mail address: [email protected] (Richard P.

Evershed)

press). A further study of modern-day agricultural

soils, where a 5b-stanol signal persisted for more than120 years after the cessation of manuring, attests to

the relative longevity of these compounds and supportstheir use in less exposed archaeological soils (Bull et

al., 1998). Recent advances by this laboratory (Bristol)

have shown bile acids to be diagnostic faecal markerswith an even greater resistance to degradation than 5b-stanols (Elhmmali et al., 1997). Together both com-

pound classes provide a reliable method by whichancient faecal deposition, and in many cases the actual

source of the faecal matter, may be determined

(Evershed and Bethell, 1996; Simpson et al., 1998;Simpson et al., in press).

Similar investigations to trace other components ex-

ogenous to the pedosphere are scarce. However, theprospect of determining further inputs is not unfeasi-

ble. It has already been demonstrated that a change inoverlying vegetation has a discernible e�ect on the

composition of low molecular weight soil components

(van Bergen et al., 1997; van Bergen et al., 1998).Hence, the potential exists for such changes to be

exploited and to provide information regarding di�er-

ences in ancient vegetation cover or physical transferof soils and vegetation from one environment to

another. Furthermore, techniques such as stable car-

bon isotope analysis can provide additional infor-mation regarding the source of organic soil

Fig. 1. A map depicting the location of settlement sites on the Tofts Ness peninsula of Sanday, Orkney. Insets: maps of anthropo-

genic soil depths associated with Mounds 4 and 11 as determined by auger. Dashed areas represent the extent of the settlement

mound and middens, including post abondonment spreading out of settlement site material, they do not represent the extent of

anthropogenic soils.

I.D. Bull et al. / Organic Geochemistry 30 (1999) 535±556536

Fig. 2. Soil pro®le descriptions of the pits dug around Mounds 4, 8 and 11.

I.D. Bull et al. / Organic Geochemistry 30 (1999) 535±556 537

constituents. As with soil organic matter such isotopeanalyses have been largely restricted to bulk measure-

ment of total organic carbon in studies of variousancient environments, such as Holocene habitat change(Ambrose and Sikes, 1991) and marine inputs to

Orcadian soil formations (Simpson, 1985; Davidson etal., 1986). Advances enabling the measurement of com-pound speci®c stable carbon isotope ratios have

further increased the level of information which maybe derived from the analysis of low molecular weightcomponents of soil (Hayes et al., 1990; Freeman et al.,

1990; Simpson et al., 1998; Simpson et al., in press).This study combines elemental, lipid, lipid bio-

marker and compound speci®c stable carbon isotopeanalyses to investigate the origin of three buried

anthropogenic soil formations at Tofts Ness, on theIsle of Sanday, Orkney, UK.

2. Description of study site

2.1. Anthropogenic soils

A buried, dark coloured loam soil horizon of

between 35 and 75 cm in thickness, embedded at var-ious depths between calcareous wind blown sanddeposits, can be identi®ed in three areas of the ToftsNess landscape. The close association with early settle-

ment sites and enhanced total phosphate levels indicatethat these fossil soils are anthropogenic in origin andformed by a process analogous to the plaggen manur-

ing process found extensively in north-west Europe(Dockrill and Simpson, 1994; Simpson et al., 1998).Such soils are thought to have arisen within a mixed

pastoral-arable economy as a result of e�orts to main-tain and enhance the fertility of arable areas. Heatheror grass turves were stripped from podsolic soils andused as animal bedding; the composted turf and ani-

mal manure were then applied to the arable landwhere the mineral component attached to the turfgradually contributed to the creation of a thick, arti®-

cial, soil horizon (Pape, 1970; Simpson, 1997).Radiocarbon dating and stratigraphic relationsbetween the fossil anthropogenic soils at Tofts Ness

and settlement sites indicate that the horizon is associ-ated with Bronze Age cultural landscape activity andmay have commenced formation during the late

Neolithic period (Simpson et al., 1998). Such a chron-ology places these anthropogenic soils amongst theearliest of their type.Soils around three settlement sites, Mound 4 (M4),

Mound 11 (M11) and Mound 8 (M8), in the ToftsNess peninsula (Royal Commission on the Ancientand Historic Monuments of Scotland notation; Fig. 1)

were surveyed to determine the extent and thicknessesof buried, fossil, anthropogenic soil horizons. Soils

around Mound 4 and Mound 11 were surveyed byhand auger where Munsell colours of 10YR 3/2 or

10YR 3/3 (dark brown/black) and hand textures ofsandy loam and sandy silt loam distinguished the hor-izons. At Mound 8, where the horizons of interest are

generally buried deeper and thicknesses could not befully ascertained by conventional hand auger, a ¯ux-gate gradiometer was used to determine the spatial

extent of the buried soil (Simpson et al., 1998). Theseobservations indicated that the buried soils are slightlyless than 1 ha in extent. Pits were dug in soil stratigra-

phies around each settlement mound (two from aroundMound 8) and samples (ca 400 g) excised from theexposed anthropogenic layers in the soil pro®le(Mound 4: 0±5, 30±35, 39±44 and 49±54 cm; Mound

8: 35±40, 45±50, 55±60, 65±70, 75±80, 85±90, 95±100and 105±110 cm; Mound 8/2: 60±65, 70±75 and 78±83 cm; Mound 11: 0±5, 33±36, 39±42, 45±48 and 51±

54 cm; Fig. 2). Sur®cial soil from Mound 8 was notanalysed since, unlike that of Mounds 4 and 11, it hadbeen subjected to more recent ploughing. In addition,

a number of contemporary materials of the type whichmay have been used in anthropogenic soil formationwere collected (1±5 g of each). These included single

samples of: cattle dung, midden (mixed turf and cattledung), roo®ng turf, surface vegetation litter and sea-weed (Fucus vesiculosus ). A number of soil analyseshave been made on archaeological sediments obtained

from Mounds 4, 8 and 11 (these studies include thinsection micromorphology, bulk d 13C analysis, totalphosphate analysis, radiocarbon dating and a gradio-

metric analysis of Mound 8; Simpson et al., 1998).

3. Experimental

3.1. Sample preparation and solvent extraction

All soil and putative inputs were supplied air dried.Samples were partially crushed with a pestle and mor-tar and, subsequently, sieved using 2 mm and 75 mmsieves yielding a single <75 mm particle size fraction.Dried vegetation samples were processed by the samemethod but liquid nitrogen was added to facilitate the

crushing process. All samples were Soxhlet extractedfor 24 h using a dichloromethane/acetone (9:1 v/v) sol-vent system to obtain a total lipid extract (TLE).

3.2. Acid/neutral separation of TLEs

TLEs were initially separated into two fractions,

àcid' and `neutral', using an extraction cartridge witha bonded aminopropyl solid-phase (JonesChromatography). Extracts dissolved in DCM/isopro-

panol (2:1) were ¯ushed through a cartridge pre-elutedwith the same solvent system. After further elution


with DCM/isopropanol (2:1, 6 ml) the collected `neu-tral' fraction was removed and the cartridge ¯ushed

with 2% acetic acid in diethylether (6 ml) thereby elut-ing the àcid' fraction. Both fractions were evaporatedunder reduced pressure, transferred to vials and re-

sidual solvent removed by a stream of nitrogen.

3.3. Column chromatography of neutral lipids

Columns were packed with dried activated silica gel60 (1608C, >24 h; Fluka) and pre-eluted with hexane.Samples were applied to the column as a mixture of

dissolved and ®nely suspended particulates in hexane.Gradient elution was performed under positive press-ure supplied by a stream of nitrogen providing an

elution rate of 015 ml minÿ1. The eluents used com-prised ®ve separate solvent systems: hexane, hexane/DCM (9:1), DCM, DCM/methanol (1:1) and metha-nol, applied in elutropic order to give ®ve fractions:

`hydrocarbon', àromatic', `ketone/wax-ester', àlcohol'and `polar', respectively. The relative volumes of sol-vents applied were determined by the ratio 2:1:3:2:2,

following the above elutropic series, and the size of thecolumn being used for a particular separation. Columnfractions were collected and solvent removed in an

identical manner to fractions from the acid/neutral sep-aration.

3.4. Urea adduction

Samples (010 mg) dissolved in hexane/acetone (2:1,1.5 ml), in a Pyrex test-tube, were agitated with a vor-

tex mixer whilst a warm, saturated solution of urea inmethanol (1 ml) was added dropwise forming a densewhite precipitate of urea. The solvent was removedunder a stream of nitrogen (ensuring the complete

evaporation of methanol) and 3 ml of DCM added toresuspend the urea precipitate. After centrifugation(2500 rpm, 15 min) the non-adduct in DCM was

removed by pipette and passed through a small plug ofpre-extracted cotton wool (ensuring complete removalof urea crystals). The adduction procedure was then

repeated and the second non-adduct solution combinedwith the ®rst. The adduct was recovered by washingthe urea crystals with DCM (10 ml) then dissolvingthem in double distilled water (10 ml) and extracting

with DCM (2 ml). The urea saturated aqueous layerwas removed and a further aliquot of double distilledwater (10 ml) was mixed with the DCM and left to

partition (ensuring complete removal of urea from theorganic layer). The organic layer was then removedand passed through a small column of anhydrous

sodium sulphate to remove any residual water. Solventwas removed from the eluates by evaporation under agentle stream of nitrogen.

3.5. Derivatization

All fractions except `hydrocarbons' and `n-alkanoicacids' were derivatized by heating sample aliquots with30 ml of N,O-bis(trimethylsilyl)tri¯uoroacetamide

(BSTFA), containing 1% trimethylchlorosilane(TMCS), at 708C for 1 h. Residual BSTFA was evap-orated under nitrogen and the sample redissolved in

hexane.n-Alkanoic acids were converted to their correspond-

ing methyl esters by reaction with a BF3-methanol

complex (14% w/v) at 708C for 10 min. Methyl esterswere isolated by adding 1 ml of double distilled waterand extracting into hexane.

3.6. Elemental analysis

Elemental analyses were performed using a PerkinElmer 240C elemental analyser to determine total car-bon, hydrogen and nitrogen compositions of the soils.Inorganic carbon content was determined using a

Strohlein Instruments Coulomat 702 carbon analyseradapted to analyse CO2 liberated from H3PO4 diges-tion; the TOC value was then calculated as the di�er-

ence between total carbon and total inorganic carbon.Each sample was analysed a total of four times and amean TOC value calculated; values typically exhibited

standard errors <20.1%dwt.

3.7. Gas chromatography (GC) and gas

chromatography/mass spectrometry (GC/MS)

GC analyses were performed using a Hewlett-

Packard 5890 series II gas chromatograph ®tted with afused silica capillary column (Chrompack CPSil-5 CB,50 m� 0.32 mm) coated with a 100% dimethylpolysi-loxane stationary phase (®lm thickness 0.12 mm).

Derivatized samples were injected (1.0 ml) via an on-column injector as solutions in hexane. Hydrogen wasused as carrier gas for all samples except the hydro-

carbon fraction when helium was used as carrier gas inorder to facilitate resolution of the internal standard.GC analyses of wax ester fractions were made using a

column capable of performing at elevated temperature(J & W Scienti®c DB1, 15 m� 0.32 mm� 0.1 mm ®lmthickness). The temperature was programmed from408C (1 min) to 2008C at a rate of 108C minÿ1 then to

3008C at a rate of 38C minÿ1 with a ®nal time of20 min when employing the former column and from508C (2 min) to 3508C at a rate of 108C minÿ1 with a

®nal time of 10 min when using the latter. The ma-jority of GC/MS chromatographic separations wereperformed using a Carlo Erba 5160 GC equipped with

a fused silica capillary column (Chrompack CPSil-5CB, 50 m� 0.32 mm) coated with a 100% dimethylpo-lysiloxane stationary phase (®lm thickness 0.12 mm, He


carrier gas) whilst `wax-ester' fraction analysesemployed a high temperature column (J & W Scienti®cDB1, 15 m� 0.32 mm� 0.1 mm ®lm thickness; He car-

rier gas). The GC oven temperature was programmedfrom 408C (1 min) to 2008C at a rate of 108C minÿ1

then to 3008C at a rate of 38C minÿ1 with a ®nal time

of 20 minutes when employing the former column andfrom 508C (2 min) to 3508C at a rate of 108C minÿ1

with a ®nal time of 10 minutes when using the latter

column. Samples were introduced using on-columninjection coupled, via a heated transfer line, to aFinnigan MAT 4500 quadrupole mass spectrometerscanning in the range of m/z 50±850 with a cycle time of

1.5 s. Electron energy was maintained at 300 mA with anion source temperature of 1908C. The mass spectrometerwas operated with an electron voltage of 70 eV.

3.8. Gas chromatography combustion/isotope ratio massspectrometry (GCC/IRMS)

Analyses were made on 1 ml sample aliquots using aVarian 3400 gas chromatograph ®tted with a fusedsilica capillary column (SGE BP-1, 50 m � 0.32 mm)

coated with a 100% dimethylpolysiloxane stationaryphase (®lm thickness 0.12 mm, He carrier gas), ®ttedwith a septum equipped temperature programmable

injector (SPI) coupled to a Finnigan MAT Delta Sstable isotope mass spectrometer (electron ionisation,

Fig. 3. Levels of total organic carbon determined for each soil pro®le.

Fig. 4. Histograms depicting the distribution of n-alkane com-

ponents exhibited by: (a) a typical mound excavated anthro-

pogenic soil, and (b) the soil associated with a roo®ng turf.


100 eV electron voltage, 1mA electron current, 3Faraday cup collectors masses 44, 45 and 46, CuO/Pt

combustion reactor set to a temperature of 8508C).

4. Results

4.1. Total organic carbon (TOC) analysis

Fig. 3 depicts the changes in TOC levels, with depth,

for each of the four soil pro®les analysed. Values areobserved to vary between 1.0 and 6.2% on a soil dryweight basis. Whilst the sur®cial samples of soils around

Mounds 4 and 11 (0±5 cm) contain the highest quan-tities of TOC (4.9 and 6.2% respectively), the majority

of anthropogenic soil samples [all except Mound 4 (0±5 cm), Mound 11 (0±5 cm) and Mound 11 (33±36 cm)]exhibit lower amounts (<2.7%). In the sub-40 cm

region of each pro®le, TOC levels can be seen to bothincrease and decrease within a much narrower range(1.0±2.7%).

4.2. Analysis of n-alkyl lipids

Extracts of each soil sample show practically identicalhomologies of dominant, odd n-alkanes typi®ed by a

Fig. 5. Histograms depicting the distribution of wax ester components exhibited by soils taken from the pro®les adjacent Mounds

4, 8 and 11; M8 (45±50, 55±60, 65±70, 75±80, 85±90, 95±100 and 105±110 cm) not analysed.


Fig. 6. Histograms depicting the distribution of n-alkanol components exhibited by soils taken from the pro®les adjacent Mounds

4, 8 and 11.


Fig. 7. Histograms depicting the distribution of n-alkanoic acid components exhibited by soils taken from the pro®les adjacent

Mounds 4, 8 and 11; M8 (45±50, 65±70, 85±90 and 95±100 cm) not analysed.


monomodal distribution with a maximum at C31 (Fig.4); the immediate peripheral homologues (C27, C29 and

C33) are also abundant. n-Alkane distributionsobtained from the turf soil, turf vegetation and cattledung reveal almost identical homologous series of n-

alkanes.Fig. 5 summarises the wax ester distributions

obtained from the majority of soil samples collectedfrom the Tofts Ness peninsula. Homologous series are

observed in the C36±C60 range with the C44±C52 hom-ologues generally exhibiting a markedly higher abun-dance. Further inspection of this latter range reveals

the C50 wax ester, in each case, as the characteristicallymost minor component barring a few notable excep-tions (M4Ð30±35 cm, M8/2Ð78±83 cm and M11Ð0±

5 cm). Interestingly, soils obtained from the M11 pro-®le exhibit C52 components which are signi®cantlymore abundant than peripheral homologues, especially

in the uppermost soil layer. The sur®cial sampleobtained from soils around the Mound 4 pro®le also

exhibits a dominant C52 homologue. Analysis of themoieties forming the observed wax esters reveals arange of components for each individual peak (n-alka-

nols, C24±C30; n-alkanoic acids, C16±C26). Of particularnote is the observation that the most abundant acidfragment ions, for each peak, correspond with the lossof a C26 n-alkanol moiety, this is particularly pro-

nounced for the C44 and C48 wax ester componentswhich exhibit dominant ions at m/z 285 (C18 n-alka-noic acid) and m/z 341 (C22 n-alkanoic acid) respect-

ively. Additionally, Fig. 5 gives details of thedistribution obtained from the roof turf vegetationwhich also exhibits a homologous series (C36±C60)

dominated by components from C42 to C56, inclusive,with an unpronounced maximum at C52.The relative abundances of n-alkanol homologues

Fig. 8. d 13C values determined for the n-alkanoic components from each soil sample.


observed for the soil samples are summarised in Fig. 6.Each homologous series lies within a C20±C34 range

with a central maximum at C26 and prominent butprogressively less abundant C28 and C30 components.The overall abundance of n-alkanol distributions ¯uc-

tuates between di�erent pro®les and depths. Inspectionof the n-alkanol distribution obtained from the roofturf soil reveals a homologous series almost identical

to those observed for the anthropogenic soils, albeitwith a slightly more dominant C26 component.n-Alkanoic acid distributions obtained for the ma-

jority of soil pro®le samples are shown in Fig. 7. Eachsample exhibits a bimodal distribution of componentswith a lower maximum at C16 and a higher maximumat C24, C26 or C28. Only two anthropogenic soils actu-

ally exhibit a latter maximum at C24, these being thetwo sur®cial samples from adjacent Mound 4 (0±5 cm)and Mound 11 (0±5 cm). Additionally, the distri-

butions of both sur®cial samples have even/odd CPIvalues which are signi®cantly higher (14.0 and 14.1 re-spectively) than those calculated for the remaining soil

derived distributions (2.6±4.9). Inspection of the distri-bution derived from the roof turf vegetation reveals asimilar relative abundance of latter homologues (rC22)

with a maximum at C24. In comparison, the homolo-gous series obtained for the associated roof turf soilexhibits C26 and peripheral homologues at a signi®-cantly higher relative abundance compared with the

C16 component. Whilst, the distribution observed forthe mixed vegetation litter has a latter maximum atC20 albeit a lower maximum still occurs at C16.

4.3. Stable carbon isotopic analysis of individual n-alkyllipids

d13C values obtained for the three dominant n-alkanes (C29, C31 and C33) have been reported pre-viously by Simpson et al. (1998). Values for homol-

ogues generally fell within a range of ÿ32 to ÿ35 -and exhibited a remarkable similarity between all four

soil pro®les.Further stable carbon isotope analyses were per-

formed on the n-alkanoic acid components of the

anthropogenic soil pro®les, the results of which aredepicted graphically in Fig. 8. Initial inspection revealsall values to lie within a range of ÿ27 to ÿ37 - with

homologues becoming progressively depleted in 13C indirect relation to chain length. The majority of d 13Cvalues obtained for any single homologue all lie within

a very narrow range. Two notable exceptions to thispattern are the 0±5 cm sur®cial soil samples takenfrom soils around Mounds 4 and 11. For bothsamples, the majority of n-alkanoic acid homologues

are depleted by ca 1.5 - relative to the isotopic com-position of identical homologues obtained fromsamples at greater depth. Interestingly, the C16 and C18

components are consistently 3±6 - less depleted in13C than the C20 homologue; a signi®cantly largerdi�erence than is observed between consecutive higher

homologues.

4.4. Analysis of steroidal componentsÐputative inputs

Table 1 summarises the relative abundance distri-butions of sterols and sterones in each of the potentialinputs. Distributional data are presented as a percen-tage of the identi®ed components in each sample.

Inspection of the results obtained for the cattle dungand dung-based midden reveals very similar distri-butions. Each is dominated by stanols, with the single

most abundant component in each sample being 24-ethyl-5b-cholestan-3b-ol(5b-stigmastanol; 1c) which occurs with its 3a-epimer

24-ethyl-5b-cholestan-3a-ol (2c). Other 5b(H)-stanolspresent are 5b-cholestan-3b-ol (coprostanol; 1a), 24-methyl-5b-cholestan-3b-ol (5b-campestanol; 1b), 24-

Table 1

A distributional summary of steroidal components in the putative inputs, % of identi®ed componentsa

1a 2a 4a 3a 1b 2b 1e 2e 4b 3b 1c 2c 4e 4c 4g 3c 5b 6b 5c 7b 6c 7c other components

Cattle Dung 6 2 ± 2 6 2 1 Trace ± ± 33 10 ± ± ± 17 ± ± 4 ± 2 2 12

Midden 3 2 1 2 4 1 3 Trace ± ± 21 10 ± ± ± 11 ± ± 4 ± 3 5 23

Roof Turf (Soil) ± ± 3 2 ± ± ± ± 8 7 ± ± 5 12 ± 11 ± 3 ± 3 10 12 9

Roof Turf (Vegetation) ± ± ± ± ± ± ± ± ± 5 ± ± ± 5 ± 6 ± 2 ± 18 14 50 0

Plant Litter ± ± ± ± ± ± ± ± 8 ± ± ± 7 36 ± 8 ± ± ± ± ± 19 22

Seaweed (Fucus vesiculosus ) ± ± ± ± ± ± ± ± ± ± ± ± ± ± 100 ± ± ± ± ± ± ± 0

a 5b-cholestan-3b-ol (1a); 5b-cholestan-3a-ol (2a); cholest-5-en-3b-ol (4a); 5a-cholestan-3b-ol (3a); 24-methyl-5b-cholestan-3b-ol(1b); 24-methyl-5b-cholestan-3a-ol (2b); 24-ethyl-5b-cholest-22-en-3b-ol (1e); 24-ethyl-5b-cholest-22-en-3a-ol (2e); 24-methylcholest-

5-en-3b-ol (4b); 24-methyl-5a-cholestan-3b-ol (3b); 24-ethyl-5b-cholestan-3b-ol (1c); 24-ethyl-5b-cholestan-3a-ol (2c); 24-ethylchole-stan-5,22-dien-3b-ol (4e); 24-ethylcholest-5-en-3b-ol (4c); 24-ethylcholest-5,24(24 ')E-dien-3b-ol (4g); 24-ethyl-5a-cholestan-3b-ol (3c);24-methyl-5b-cholestan-3-one (5b); 24-methyl-5a-cholestan-3-one (6b); 24-ethyl-5b-cholestan-3-one (5c); 24-methylcholest-4-en-3-one

(7b); 24-ethyl-5a-cholestan-3-one (6c); 24-ethylcholest-4-en-3-one (7c).


ethyl-5b-cholest-22-en-3b-ol (1e) and their correspond-ing 3a-epimer analogues. Only two stanols with a

5a(H) con®guration are observed to occur, i.e. 5a-cho-

lestan-3b-ol (5a-cholestanol; 3a) and 24-ethyl-5a-cho-lestanol-3b-ol (5a-stigmastanol; 3c), although this latter

compound is still the second most abundant com-

Fig. 9. Partial gas chromatograms showing the sterol distributions observed for the sur®cial (0±5 cm) and deeper (33±36 cm) soils

taken from the soil pro®le adjacent Mound 11. For structures see appendix.


Fig. 10. Chromatograms m/z 129 and m/z 215 mass and reconstructed ion chromatograms obtained for the sterol fractions from

the uppermost (35±40 cm) and deepest (105±110 cm) soils taken from the soil pro®le adjacent Mound 8. For structures see appen-

dix A.


ponent of each distribution. Additionally, 24-ethyl-5b-cholestan-3-one (5c), 24-ethyl-5a-cholestan-3-one (6c)and 24-ethylcholest-4-en-3-one (7c) are observed, albeit

in low abundance.The soil associated with the roof turf exhibits a less

complicated sterol distribution lacking a single domi-

nant component. Inspection of the distribution revealsthe presence of the unsaturated phytosterols 24-methyl-cholest-5-en-3b-ol (campesterol; 4b), 24-ethylcholest-5,22-dien-3b-ol (stigmasterol; 4e) and 24-ethylcholest-5-

en-3b-ol (sitosterol, 4c) as well as the C27 congenercholest-5-en-3b-ol (cholesterol, 4a). Also present arethe 5a(H)-stanol analogues 5a-cholestan-3b-ol (3a), 24-methyl-5a-cholestan-3b-ol (3b) and 24-ethyl-5a-chole-stan-3b-ol (3c). Of note is the particularly high pro-portion of 24-ethyl-5a-cholestan-3-one (6c) and 24-

ethylcholest-4-en-3-one (7c).Vegetation associated with the roof turf yields a

similarly uncomplicated distribution of components in

which the only sterols 24-methyl-5a-cholestan-3b-ol(3b), 24-ethylcholest-5-en-3b-ol (4c) and 24-ethyl-5a-cholestan-3b-ol (3c) exist at low abundance. The domi-nant component at 50% is 24-ethylcholest-4-en-3-one

(7c) whilst 24-methylcholest-4-en-3b-ol (7b) and 24-ethyl-5a-cholestan-3-one (6c) are present at a moder-

ately high abundance and a 2% proportion of 24-methyl-5a-cholestan-3-one (6b) is also observable.The distribution observed for the miscellaneous

plant litter is very simple. The dominant compound is24-ethylcholest-5-en-3b-ol (4c; 36%) with a corre-spondingly high proportion of 24-ethylcholest-4-en-3-

one (7c). 24-methylcholest-5-en-3b-ol (4b), 24-ethyl-cholest-5,22-dien-3b-ol (4e) and the 5a(H)-stanol ana-logoue 24-ethyl-5a-cholestan-3b-ol (3c) all reside atlower levels of abundance.

Seaweed (Fucus vesiculosus ) represents the onlyputative input of marine origin. As expected, the`sterol' fraction obtained is signi®cantly di�erent to

those of previous samples. The distribution consists ofa single component identi®ed as 24-ethylcholest-5,24(24 ')E-dien-3b-ol (fucosterol; 4g) recognised as the

major sterol consitiuent of nearly all macroscopicbrown algae (Phaeophyceae; Volkman, 1986).

4.5. Analysis of steroidal components in theanthropogenic soils

The majority of anthropogenic soils exhibit steroidal

components at abundances amenable to GC analysis.However, the sterol fractions of the Mound 4 fossil

Fig. 11. Reconstructed ion chromatogram showing the bile acid distribution obtained from the deepest (105±110 cm) soil taken

from the soil pro®le adjacent Mound 8. Structures given in appendix.


soils are extremely low in abundance, hampering any

reliable analysis, although the sur®cial sample (0±5 cm)does exhibit a sterol pro®le dominated by 24-ethylchol-est-5-en-3b-ol (4c) with lower, but still signi®cant,

quantities of 24-methylcholest-5-en-3b-ol (4b), 24-ethyl-cholest-5,22-dien-3b-ol (4e) and 24-ethylcholest-4-en-3-one (7c).

Anthropogenic soils around Mound 11 contain thehighest quantity of sterols relative to aliphatic com-

ponents. Fig. 9 depicts two representative sterol frac-tions from soils around Mound 11 as determined byGC. The sur®cial sample excised from 0±5 cm depth is

overwhelmingly dominated by 24-ethylcholest-5-en-3b-ol (4c) and the relatively less abundant components 24-

methylcholest-5-en-3b-ol (4b), 24-ethylcholest-5,22-dien-3b-ol (4e) and 24-ethylcholest-4-en-3-one (7c), lar-gely paralleling the sur®cial soil from around Mound

4. Cholest-5-en-3b-ol (4a) is also present as are the5a(H) reduced analogues of this component and thephytosterol congeners mentioned above. Soils taken

from the pro®le at greater depth generally exhibitmostly the same steroidal components although the

5a(H)-stanols are all observed at higher relative abun-dances than their stenol analogues. Additionally,minor quantities of 24-methylcholest-5,22-dien-3b-ol(4d) and 5b-cholestan-3b-ol (1a) are also observed tooccur at low concentrations. Mound 4. Cholest-5-en-

3b-ol (4a) is also present as are the 5a(H) reduced ana-logues of this component and the phytosterol conge-ners mentioned above. Soils taken from the pro®le at

greater depth generally exhibit mostly the same ster-oidal components although the 5a(H)-stanols are allobserved at higher relative abundances than their ste-

nol analogues. Additionally, minor quantities of 24-methylcholest-5,22-dien-3b-ol (4d) and 5b-cholestan-3b-ol (1a) are also observed to occur at low concen-trations.The sterol distributions for the anthropogenic soils

from around Mound 8 (8 and 8/2) show many simi-larities to those of fossil anthropogenic soils from

around Mound 11. However, a number of minor com-ponents observed in soils around Mound 11, e.g. 24-methylcholest-5,22-dien-3b-ol (4d), are not detected.

Fig. 10 depicts total ion traces and mass chromato-grams (m/z 129 and m/z 215) for the sterol fractions ofthe shallowest and deepest fossil soils taken from

around Mound 8. Of particular signi®cance is theincreasing abundance of 5b-cholestan-3b-ol (1a), rela-tive to 5a-cholestan-3b-ol (3a), observed at greaterdepth. The m/z 215 mass chromatogram is particularlyuseful since under EI conditions (70 eV) the 3b,5b-mol-

ecule is observed to fragment to produce a lower yieldof ions of m/z 215 compared with the 3b,5a-epimer.Hence, the m/z 215 mass chromatogram unequivocally

con®rms the greater abundance of 5b-cholestan-3b-ol(1a) compared with 5a-cholestan-3b-ol (3a). Again par-

alleling soils around Mound 11, soils from aroundMound 8 do not exhibit any detectable levels of 24-

ethylcholest-5,24(24 ')E-dien-3b-ol (4g).

4.6. Bile acids analysis of soils associated with Mound 8

Fig. 11 depicts the bile acid distribution of the soilexcised at 105±110 cm depth from the Mound 8 pro-

®le. It may be readily observed that, disregarding theinternal standard, the distribution is dominated by alarge peak corresponding to the secondary bile acid

deoxycholic acid (9). A number of related primary andsecondary bile acids are present at a much lower abun-dance amongst which a small peak corresponding tohyodeoxycholic acid (12) is observed.

5. Discussion

5.1. General organic composition

Information regarding changes in the overall compo-sition of material used to construct the mounds may

be obtained by analysing soil TOC levels. The rela-tively large amount of organic carbon present in thesur®cial layers around Mound 4 (0±5 cm) and Mound

11 (0±5 cm) is most likely the result of contributionsfrom modern-day ¯ora and fauna. Whilst the narrowerrange of TOC levels exhibited by all samples deeper

than 40 cm indicates a relatively uniform compositionof material deposited over time, the low levels of or-ganic matter being consistent with an input of mineral

topsoils (Davidson et al., 1986; Simpson, 1993).

5.2. Identi®cation of fossil anthropogenic soilconstituents using n-alkyl components

The identi®cation of material(s) used in the for-mation of the Tofts Ness anthropogenic soils is facili-

tated somewhat by environmental history dataconcerning the Orkneys. For example, due to increas-ingly adverse weather conditions and increased anthro-

pogenic activity by 3000 BC the only remainingvegetation in the Orkneys consisted of scrub and opengrassland. Additionally, the Isle of Sanday had fewdeposits of peat so seaweed, turf and animal dung

were used as primary fuel sources, the burning of turfgiving rise to iron rich orange/red soil deposits whichmay be observed at Tofts Ness (Ritchie, 1995). Such

knowledge severely limits the number of putativeinputs to fossil soils, the most probable of these being:cattle dung, dung-turf midden, grass turf, miscella-

neous low lying vegetation and seaweed. Lipid analysisprovides a highly diagnostic method which, as well asenabling di�erentiation between these various inputs,


can provide further information regarding the pedolo-

gical fate of lipid components derived from them.Further evidence for a majority input of unchanging

composition is provided by the practically identical

homologues of n-alkanes exhibited by all anthropo-genic soils (Fig. 3). The large odd-over-even predomi-

nance observed for the components being indicative ofan input from terrestrial vegetation (Peters andMoldowan, 1993). Additionally, the maximum at C31

supports the notion that grasses may have been amajor organic input to each anthropogenic soil sincethe distribution observed is consistent with those of

most temperate grasses (Ma�ei, 1996). The similaritybetween n-alkane distributions from fossil anthropo-

genic soils and those from turf soil, turf vegetation andcattle dung lend further support to a grass turf input,the cattle dung distribution having derived from the

original food source, i.e. grass.Wax esters commonly occur as components of epicu-

ticular leaf waxes, usually exhibiting a suite of homol-ogues (C30±C56; Walton, 1990). Vegetation-derivedesters commonly contain C26 and C28 n-alkanol moi-

eties (depending upon the dominant free n-alkanolcomponent) and a range (C18-C22) of n-alkanoic acids(Tulloch and Ho�man, 1973a, b; Tulloch 1984, 1987).

The larger ranges of constituent moieties exhibited byesters extracted from the fossil soils are most likely the

result of transesteri®cation processes operating withinthe soil (AmbleÁ s et al., 1994; Jambu et al., 1995).Despite this, the preferential occurrence of the C26 n-

alkanol moiety parallels the occurrence of a dominantC26 n-alkanol component in the free n-alkanol distri-butions derived from turf soil and turf vegetation (see

below). The exceptionally high abundance of the C52

wax ester homologue, in the two sur®cial samples

(Mound 4: 0±5 cm and Mound 11: 0±5 cm), may beattributed to a more recent input of higher plant detri-tus, the predominant C26,C26 n-alkanol:n-alkanoic acid

composition of the ester lending support to this con-clusion. The relative prominance of the C52 component

in all samples from the Mound 11 pro®le is note-worthy and is possibly the consequence of enhancedpreservation compared with the soils of other mound

pro®les. Whilst it may be compared favourably withdistributions from the pro®les associated with Mounds4 and 8 the wax ester distribution obtained from the

roof turf vegetation lacks the more dominant C52 com-ponent characteristic of samples from the Mound 11

pro®le. However, given the age and former environ-mentally exposed position of the roof turf, one wouldexpect the sample to have su�ered considerable degra-

dation with no new input of lipid from vegetation.Under such a scenario the dominant C52 componentwould exhibit considerable degradation (Jambu et al.,

1993) whilst the gradual increase in abundance, withchain length, observed may well arise from an inverse

relationship between rate of loss and chain length as

previously reported for aliphatic compounds (AmbleÁ set al., 1994).It is interesting to note the overall abundance of n-

alkanol distributions ¯uctuating with both pro®le anddepth. Jambu et al. (1993) reported an increase in free

primary n-alkanols from L to F to A1 horizon in asoil pro®le beneath a cover of pine trees (Pinus mari-tima ssp.). The soil samples in this study di�er in that

they have, within any pro®le, been largely excised fromthe same horizon which originates as a consequence ofanthropogenic and not natural processes. Therefore,

the ¯uctuations observed probably arise from increasedinput of vegetation and/or increased post-depositional

environmental exposure, the latter leading to enhanceddegradation. The similarities in distributions observedbetween fossil soils and the turf soil infer a probable

majority input of turf soil and related materials. Asmentioned above, enhanced degradation of the moreabundant C26 component could yield an identical dis-

tribution; a phenomenon previously observed by boththis laboratory (Bristol; Bull, 1997) and other authors

(Disnar and HeÂ roux, 1995). However, it is possiblethat the apparent increase in abundance of the periph-eral homologues is also the result of other additional

processes, e.g. enzymatic reduction of n-alkanoic acids(Kolattukudy, 1971, 1975). The same process may be

used to explain the ìncrease' of peripheral homologuesbetween the roof turf vegetation and the roof turf soil.Appraisal of n-alkanoic acid distributional data pro-

vides both further evidence concerning fossil anthropo-genic soil formation materials and insights into soillipid diagenesis. The high even/odd CPI values exhib-

ited by distributions from the sur®cial soils (Mounds4: 0±5 cm and Mound 11: 0±5 cm) re¯ect recent incor-

poration of plant detritus. The lower values exhibitedby distributions from deeper soils indicate the actionof pedogenic microbial processes (Peters and

Moldowan, 1993). The similarity between the sur®cialsoil and turf vegetation distributions implies, not sur-

prisingly, that recently incorporated plant matter is ofthe same type associated with the ancient turf.Lehtonen and Ketola (1990) have suggested that the

release of n-alkanoic acids from decaying plant detritusis the process reponsible for observable increases ineven long chain n-alkanoic acids. Whilst the hydrolytic

release of long chain n-alkanoic acids from ceridesundoubtedly contributes to the increase of higher hom-

ologues in the roof turf soil distribution, the most im-portant contributory mechanism is the terminaloxidation of long-chain n-alkanols. AmbleÁ s et al.

(1994) report signi®cant correlation between the car-bon number of even n-alkanoic acid components andthe even n-alkanoic acid:n-alkanol ratio in a hydro-

morphic forest podzol, it was furthermore noted thatthe extent of oxidation did not decrease with the


increase in chain length of even n-alkanols. This pro-vides an explanation for the relative increase in C26,

C28 and C30 n-alkanoic acids, observed in distributionsderived from anthropogenic soils, since the respectiven-alkanol analogues are the dominant components in

each corresponding soil n-alkanol distribution. The n-alkanoic acid distribution obtained from the miscella-neous vegetation litter does not correlate well with any

of the derived distributions including, most impor-tantly, the relatively young sur®cial soils therebya�rming grassy turf as the most probable major com-

ponent used in anthropogenic soil formation.Stable carbon isotope analysis of the n-alkanoic acid

components increased the evidence for a major terres-trial input to the Tofts Ness anthropogenic soils. The

narrow range of d 13C values between soils, obtainedfor each homologue, is consistent with the idea of afairly uniform composition of formation material thus

agreeing with n-alkane stable carbon isotope data pre-viously reported by Simpson et al. (1998). The ca 1.5- depletion exhibited by homologues from the two 0±

5 cm sur®cial soils may be attributed to the assimila-tion of isotopically lighter, contemporary CO2 (Keelinget al., 1984; Friedli et al., 1986). This result holds par-

ticular importance regarding the question of mobilityof n-alkanoic components in the soil environment. Anumber of studies (e.g. Ja�eÂ et al., 1996) have high-lighted the physical movement of lipids within soil/

sediments. However, the relative immobility of n-alka-noic acids in the mounds studied is supported by thesedata since the n-alkanoic acids in the sur®cial soils are

isotopically distinct from deeper homologues therebyinferring little, or no, detectable vertical migration.Furthermore, this result also supports the notion that

lipids in the deeper fossil soil are indeed ancient in ori-gin and not derived from modern-day sources althoughideally 14C dating of lipids would provide better sup-port for this proposition (Huang et al., 1996).

5.3. Identi®cation of anthropogenic soil constituents

based on steroidal components

Steroidal components remain one of the most useful

suites of biomarkers for inter- and intra-environmentalinvestigations having been the subject of much study(Mackenzie et al., 1982; Volkman, 1986). The speci-®city of certain sterols for particular sources of organ-

ics makes them ideal indicators for determining thenature of material used in anthropogenic soil for-mation. Additionally, the availability of modern-day

inputs for comparative purposes not only enhances thereliability of source determinations but also facilitatesfurther study of the chemical behaviour of sterol com-

ponents in the natural environment. It should be notedthat the fractions discussed in the following section,whilst designated as, and predominantly composed of

sterols, do contain a limited number of sterone com-

ponents.All of the putative inputs, except seaweed, exhibit

distributions containing phytosterols, commonly

observed in higher plants, and/or their diagenetic pro-ducts both reductive (5a-stanols) and oxidative (ste-

nones/stanones) (Mackenzie et al., 1982; Goad, 1991).If the original materials used to form the anthropo-genic soils were signi®cantly terrestrial in origin then

the fossil soils would be expected to yield a similarsuite of components and this is indeed the observedcase. Whilst lower in abundance, soils around Mound

8, and to a greater extent around Mound 11, exhibitcharacteristically terrestrial steroidal distributions. The

sur®cial (0±5 cm) and deeper (33±36 cm) soil sterol dis-tributions depicted in Fig. 9 are of particular interest.When compared with the sterol distributions of the po-

tential inputs, that of the 0±5 cm sample most re-sembles the plant litter whilst the deeper (33±36 cm)sample bears a strong resemblance to the distribution

from the mineral soil of the roo®ng turf. The lowerrelative abundance of 24-ethylcholest-4-en-4-one (7c), a

product of oxidative degradation (Ren et al., 1996), inthe fossil soil indicates a higher state of preservationcompared with the roof turve. Signi®cantly, 24-ethyl-

cholest-5,24(24')E-dien-3b-ol (4g) is not detected in anyanthropogenic soils indicating little, or no, input of

Fucus vesiculosus and/or any other related brownmacroscopic algae. This agrees favourably with bulkd 13C values previously obtained for the anthropogenic

soils which were strongly terrestrial (ÿ26.0 to ÿ26.1 -; Simpson et al., 1998). Whilst, based on these data,seaweed cannot be considered a major component of

anthropogenic soil formation there still remains thepossibility that small amounts have remained unde-

tected due to degradation of 24-ethylcholest-5,24(24 ')E-dien-3b-ol (4g). The D24(24 ') unsaturation isa relatively labile functionality and burial in the soil

environment could well result in the complete conver-sion of this sterol to other steroidal and/or non-ster-oidal compounds. A search for these compounds

would be a potentially useful future avenue ofresearch.

The single most abundant component in both thecattle dung and the midden is 24-ethyl-5b-cholestan-3b-ol (1c). 5b-Stanols are widely known as intestinal

reduction products of D5 unsaturated sterols in mam-mals (e.g. Knights et al., 1983; Laureillard and Saliot,

1993; Bethell et al., 1994). The presence of a dominantC29 5b-stanol homologue, resulting from reduction ofthe phytosterol 24-ethylcholest-5-en-3b-ol (4c), attests

to the ruminant faecal source (Evershed and Bethell,1996; Evershed et al., 1997). However, inspection ofthe fossil soil sterol distributions reveals the total

absence of this component indicating that cattle man-ures were not a signi®cant constituent of fossil soil for-


mation. Moreover, the only detectable 5b-stanols in

the fossil soils are 5b-cholestan-3b-ol (1a, Mounds 8and 11) and 5b-cholestan-3a-ol (2a, Mound 8). 5b-cho-lestan-3b-ol (1a) is present as a minor component in

the fossil soils from around Mound 11 and althoughthere is a slight increase in abundance, with depth, itspresence is most likely attributable to the multi-sourced background of 5b-stanols that is observed to

occur in many soils (Bethell et al., 1994). Conversely,5b-cholestan-3b-ol (1a) and 5b-cholestan-3a-ol (2a) areobserved to occur at much higher abundances, relative

to other sterol components, in soil pro®les fromaround Mound 8. Whilst it can also derive from por-cine manures, the potential use of 5b-cholestan-3b-ol(1a) as a biomarker of human faecal deposition hasalready been extensively investigated and reported(Hatcher and McGillivary, 1979; Knights et al., 1983;

Bethell et al., 1994; Evershed and Bethell, 1996;Evershed et al., 1997). Previous work has made use ofa stanol ratio, ®rst proposed by Grimalt et al. (1990)and later modi®ed, for archaeological soil studies, to

include the epimer 5b-cholestan-3a-ol (2a) as a morereliable proxy for determining faecal deposition in anti-quity, i.e.:

�1a� 2a�=�1a� 2a� 3a� �i�

When applied to the C27 stanol components, deter-

mined for soils around Mound 8 and Mound 11, an

increase in the ratio with depth is observed for all pro-

®les (Fig. 12). Whilst the values generated for soils

around Mound 11 remain R0.42 in line with the

above conclusion of only a minor faecal background

contribution to these soils, those from around Mound

8 increase to a much greater extent approaching the

theoretical 0.7 threshold for faecal deposition (as deter-

mined by Grimalt et al., 1990) in the deepest sample,i.e. Mound 8 (105±110 cm; 0.63). Given the possibility

of partial degradation of soil components with time it

is not unreasonable to expect a decrease in this lower

limit required to positively satisfy this parameter for

faecal deposition. Given their inherently greater ther-

modynamic stability, 5a-stanols will be less susceptible

to abiotic degradation than 5b-stanols (Mackenzie et

al., 1982). Hence, we would interpret these results to

indicate faecal material as a signi®cant input during in-

itial formation of anthropogenic soils around Mound 8

(>78 cm) and a more minor component during later

episodes of soil formation (45±78 cm). As mentionedabove, since the C27 5b-stanols are the only detectable

5b-stanol components, the most likely faecal source is

either human or porcine. Di�erentiation between these

two sources requires additional analysis of bile acid

components (Bethell et al., 1994).

Fig. 12. A plot of values obtained by applying equation (i) to the C27 stanol components detected in anthropogenic soil adjacent

Mounds 8 and 11.


5.4. Appraisal of bile acid components to distinguishbetween pig and human inputs

Bile acids represent another group of metabolic pro-ducts, structurally related to stanols, which have been

used in the detection of faecal deposition (Evershedand Bethell, 1996; Elhmmali et al., 1997). Whilst thepredominant secondary bile acid in human faeces is

deoxycholic acid (9) the main secondary bile acid com-ponent of porcine faeces is the C-6 hydroxylated ana-logue hyodeoxycholic acid (12), thereby o�ering a

potential method of faecal source discrimination insoils around Mound 8 (Bethell et al., 1994).Considering the prominance of deoxycholic acid in thedeepest (105±110 cm) soil taken from around Mound 8

the faecal signal can be attributed to a majority humanfaecal input. However, it is important to note theoccurrence of hyodeoxycholic acid (12) albeit at a low

level. Since the compound is not produced by micro-¯ora in the human gut it is most likely derived from aminor input of porcine faecal material. Interestingly,

other bile acid analyses concerning anthropogenicOrcadian top-soils, surrounding a medieval farm settle-ment, have recorded porcine faeces as the major com-

ponent of a faecal input (Simpson et al., in press).

5.5. Archaeological implications

Evidence obtained from the lipid analysis of fossilsoils from around Mounds 4, 8 and 11 indicates that

these soils arose from the systematic deposition ofgrass turves. This agrees with both bulk d 13C analysespreviously made on the same samples and observed

morphological features such as the occurrence of phy-toliths (Simpson et al., 1998). The additional humanfaecal input determined for soils around Mound 8 isconsistent with an ancient input of household waste

from the adjacent settlement and is further supportedby the occurrence of small bone shards, another indi-cator of domestic waste application, in the soils.

Radiocarbon measurements of soils from the pro®lesindicate a Bronze Age date for each mound with for-mation continuing into the Iron Age (Simpson et al.,

1998). The soils can be considered to have been anintegral component in the intensive manuring regimerequired in order to sustain any viable agricultural ac-tivity in the Bronze Age landscape of Tofts Ness,

where the abundance of calcareous sand would haveinhibited water retention during summer periods thuspresenting a high risk of erosion when cultivated. One

distinctive feature is the human origin of the faecalmaterial input to soils around Mound 8. The apparentlack of high amounts of animal derived manures is sig-

ni®cant and infers either the practice of direct appli-cation of this material to agricultural plots or itspreferential use as a fuel source. Soils around Mound

8 appear to have been intimately associated with ahuman settlement although the relative proportion of

faecal material applied to the mound decreased afterinitial formation.

6. Conclusions

This investigation has involved a synthesis of lipiddistributions, speci®c lipid biomarkers and compound

speci®c d 13C measurements in an attempt to identifythe organic components that contribute to the for-mation of three Bronze Age anthropogenic fossil soildeposits located at Tofts Ness, Sanday, Orkney. The

study has yielded a number of interesting points,namely:

. Analysis of aliphatic lipid components has led to theidenti®cation of grassy turves as a major componentof anthropogenic soil formation,

. Analysis of sterol and n-alkanoic d 13C data failed toreveal any evidence supporting the use of seaweed asa major component of anthropogenic soil formation.

. Use of a stanol index ratio enabled the detection ofa signi®cant faecal input to soil around Mound 8 atinitial stages of formation, dropping to a moderateinput at later stages,

. Further analysis of bile acid components from soilsaround Mound 8 identi®ed the faecal material aspredominantly human in origin with possibly a very

minor porcine contribution.

This study has shown that organic geochemical analysis

is both an applicable and highly useful tool for investi-gations concerning archaeological soils and can revealinformation unobtainable by more traditional forms ofsoil and archaeological analysis.

Acknowledgements

Mr J. F. Carter and Mr A. R. Gledhill are thanked

for their help with GC/MS analyses and the use of theNERC Mass Spectrometry Facilities (Grant: GR3/2951, GR3/3758, FG6/36101) is gratefully acknowl-

edged. A grant from the School of Chemistry,University of Bristol, to IDB is gratefully acknowl-edged. IAS acknowledges ®eld support from theUniversity of Stirling Research Fund. The excavation

programme was supported by Historic Scotland.

Associate EditorÐS.J. Rowland


Appendix AI.D

.Bullet

al./

Organic

Geochem

istry30(1999)535±556

554

References

AmbleÁ s, A., Jambu, P., Parlanti, E., Jo�re, J., Ri�e, C., 1994.

Incorporation of natural monoacids from plant residues

into an hydromorphic forest podzol. European Journal of

Soil Science 45, 175±182.

Ambrose, S.H., Sikes, N.E., 1991. Soil carbon isotope evi-

dence for Holocene habitual change in the Kenya Rift

Valley. Science 253, 1402±1405.

van Bergen, P.F., Bull, I.D., Poulton, P.R., Evershed, R.P.,

1997. Organic geochemical studies of soils from the

Rothamsted Classical ExperimentsÐI. Total lipid extracts,

solvent insoluble residues and humic acids from Broadbalk

Wilderness. Organic Geochemistry 26, 117±135.

van Bergen, P.F., Nott, C.J., Bull, I.D., Poulton, P.R.,

Evershed, R.P., 1998. Organic geochemical studies of soils

from the Rothamsted Classical ExperimentsÐIV.

Preliminary results from a study of the e�ect of soil pH on

organic matter decay. Organic Geochemistry 29, 1779±

1795.

Bethell, P.H., Goad, L.J., Evershed, R.P., Ottaway, J., 1994.

The study of molecular markers of human activity: The

use of coprostanol in the soil as an indicator of human

faecal material. Journal of Archaeological Science 21, 619±

632.

Bull, I.D., 1997. New molecular methods for detecting natural

and anthropogenic inputs to soils and sediments. Ph.D.

dissertation, University of Bristol, 222 pp.

Bull, I.D., van Bergen, P.F., Poulton, P.R., Evershed, R.P.,

1998. Organic geochemical studies of soils from the

Rothamsted Classical ExperimentsÐII, Soils from the

Hoos®eld Spring Barley experiment treated with di�erent

quantities of manure. Organic Geochemistry 28, 11±26.

Bull, I.D., Simpson, I.A., van Bergen, P.F., Evershed, R.P.,

1999. Muck `n' molecules: Organic geochemical methods

for detecting ancient manuring. Antiquity 73, 86±96.

Bull, I.D., Betancourt, P.P., Evershed, R.P., Chemical evi-

dence supporting the existence of a structured agricultural

manuring regime on Pseira Island, Crete during the

Minoan Age, Malcolm Weiner Festschrift (in press).

Davidson, D.A., Harkness, D.D., Simpson, I.A., 1986. The

formation of farm mounds on the island of Sanday.

Geoarchaeology 1, 45±60.

Disnar, J.R., HeÂ roux, Y., 1995. DeÂ gradation et lassivage des

hydrocarbures de la formation ordovicienne de Thumb

Mountain encaissant le g|Ã te Zn-Pb de Polaris (Territoires

du Nord-Ouest, Canada). Canadian Journal of Earth

Sciences 32, 1017±1034.

Dockrill, S.J., Simpson, I.A., 1994. The identi®cation of pre-

historic anthropogenic soils in the Northern Isles using an

integrated sampling methodology. Archaeological

Prospection 1, 75±92.

Elhmmali, M.M., Roberts, D.J., Evershed, R.P., 1997. Bile

acids as a new class of sewage pollution indicator.

Environmental Science and Technology 31, 3663±3668.

Entwistle, J.A., Abrahams, P.W., 1997. Multi-element analysis

of soils and sediments from Scottish historical sites. The

potential of inductively coupled plasma-mass spectrometry

for rapid site investigation. Journal of Archaeological

Science 24, 407±416.

Evershed, R.P., Bethell, P.H., 1996. Application of multimole-

cular biomarker techniques to the identi®cation of faecal

material in archaeological soils and sediments. ACS

Symposium Series 625, 157±172.

Evershed, R.P., Bethell, P.H., Reynolds, P., Goad, L.J., 1997.

Molecular markers of manuring: analysis of modern exper-

imental material and the archaeological potential. Journal

of Archaeolgical Science 24, 485±495.

Farswan, Y.S., Nautiyal, V., 1997. Investigation of phos-

phorus enrichment in the burial soil of Kumaun, mid-cen-

tral Himalaya, India. Journal of Archaeological Science

24, 251±258.

Freeman, K.H., Hayes, J.M., Trendel, J.M., Albrecht, P.,

1990. Evidence from carbon isotope measurements for

diverse origins of sedimentary hydrocarbons. Nature 343,

254±256.

Friedli, H., Lotscher, H., Oescher, H., Siegenthaler, U.,

Stau�er, B., 1986. Ice core record of the 13C/12C ratio of

atmospheric CO2 in the past two centuries. Nature 324,

237±238.

Goad, L.J., 1991. Phytosterols. In: Charlwood, B.V.,

Barnthorpe, V.V. (Eds.), Methods in Plant Biochemistry,

vol. 7. Academic Press, New York, pp. 369±434.

Grimalt, J.O., Fernandez, P., Bayona, J.M., Albaiges, J.,

1990. Assessment of faecal sterols and ketones as indi-

cators of urban sewage inputs to coastal waters.

Environmental Science Technology 24, 357±363.

Hatcher, P.G., McGillivary, P.A., 1979. Sewage contami-

nation in the New York Bight. Coprostanol as an indi-

cator. Environmental Science Technology 13, 1225±1229.

Hayes, J.M., Freeman, K.H., Popp, B.N., 1990. Compound-

speci®c isotopic analyses: A novel tool for reconstruction

of ancient biogeochemical processes. Organic

Geochemistry 16, 1115±1128.

Huang, Y.S., Bol, R., Harkness, D.D., Ineson, P., Eglinton,

G., 1996. Post-glacial variations in distributions, 13C and14C contents of aliphatic hydrocarbons and bulk organic

matter in three types of British acid upland soils. Organic


Ja�eÂ , R., Elisme, T., Cabrera, A.C., 1996. Organic geochemis-

try of seasonally ¯ooded rain forest soils: molecular com-

position and early diagenesis of lipid components. Organic


Jambu, P., AmbleÁ s, A., Jaquesy, J.C., Secouet, B., Parlanti,

E., 1993. Incorporation of natural alcohols from plant

residues into a hydromorphic forest-podzol. Journal of

Soil Science 44, 135±146.

Jambu, P., AmbleÁ s, A., Magnoux, P., Parlanti, E., 1995.

E�ects of addition of clay minerals on the fatty acid frac-

tion of a podzol soil. European Journal of Soil Science 46,

187±192.

Keeling, C.D., Carter, A.F., Mook, W.G., 1984. Seasonal,

latitudinal and secular variations in the abundance and

isotopic ratios of atmospheric CO2 2. Results from

oceanographic cruises in the tropical Paci®c Ocean.

Journal of Geophysical Research 89, 4615±4628.

Knights, B.A., Dickson, C.A., Dickson, J.H., Breeze, D.J.,

1983. Evidence concerning the Roman military diet at

Bearsden, Scotland, in the 2nd century A.D. Journal of

Archaeological Science 10, 139±152.

Kolattukudy, P.E., 1971. Enzymatic synthesis of fatty alco-


hols in Brassica oleracea. Archives of Biochemistry and

Biophysics 142, 701±709.

Kolattukudy, P.E., 1975. Biochemistry of cutin, suberin and

waxes, the lipid barriers on plants. In: Galliard, T.,

Mercer, E.I. (Eds.), Recent Advances in the Chemistry

and Biochemistry of Plant Lipids. Academic Press, New

York, pp. 203±246.

Laureillard, J., Saliot, A., 1993. Biomarkers in organic matter

produced in estuaries: a case study of the Krka estuary

(Adriatic Sea) using the sterol marker series. Marine

Chemistry 43, 247±261.

Lehtonen, K., Ketola, M., 1990. Occurence of long-chain

acyclic methyl ketones in Sphagnum and Carex peats of

various degrees of humi®cation. Organic Geochemistry 15,

275±280.

Mackenzie, A.S., Brassell, S.C., Eglinton, G., Maxwell, J.R.,

1982. Chemical fossils: The geological fate of steroids.

Science 217, 491±504.

Ma�ei, M., 1996. Chemotaxonomic signi®cance of leaf wax

alkanes in the Gramineae. Biological Systematics and

Ecology 24 (1), 53±64.

Pape, J.C., 1970. Plaggen soils in the Netherlands. Geoderma

4, 229±256.

Peters, K.E., Moldowan, J.M., 1993. The Biomarker Guide:

Interpreting Molecular Fossils in Petroleum and Ancient

Sediments. Prentice-Hall, New York, p. 363.

Ren, D., Li, L., Schwabacher, A.W., Young, J.W., Beitz,

D.C., 1996. Mechanism of cholesterol reduction to copros-

tanol by Eubacterium coprostanoligenes ATCC 51222.

Steroids 61, 33±40.

Ritchie, A., 1995. Prehistoric Orkney. BT Batsford Ltd,

London.

Simpson, I.A., 1985. Stable carbon isotope analysis of anthro-

pogenic soils and sediments in Orkney. In: Fieller, N.R.J.,

Gilbertson, D.D., Ralph, N.G.A. (Eds.),

Palaeoenvironmental Investigations, vol. 258. BAR

International Series, Oxford, pp. 55±65.

Simpson, I.A., 1993. The chronology of anthropogenic soil

formation in Orkney. Scottish Geographical Magazine 109

(1), 4±11.

Simpson, I.A., 1997. Relict properties of anthropogenic deep

top soils as indicators of in®eld management in Marwick,

West Mainland, Orkney. Journal of Archaeological

Science 24, 365±380.

Simpson, I.A., Dockrill, S.J., Bull, I.D., Evershed, R.P., 1998.

Early anthropogenic soil formation at Tofts Ness, Sanday,

Orkney. Journal of Archaeological Science 25, 729±746.

Simpson, I.A., van Bergen, P.F., Perret, V., Elhmmali, M.M.,

Roberts, D.J., Evershed, R.P., Lipid biomarkers of manur-

ing practice in relict anthropogenic soils, The Holocene (in

press).

Tulloch, A.P., 1984. Epicuticular waxes of 4 Eragrostoid

Grasses. Phytochemistry 23, 1619±1623.

Tulloch, A.P., 1987. Epicuticular waxes of Abies-Balsamea

and Picea-GlaucaÐOccurrence of long-chain methyl-esters.

Phytochemistry 26, 1041±1043.

Tulloch, A.P., Ho�man, L.L., 1973a. Leaf wax of oats. Lipids

8, 617±622.

Tulloch, A.P., Ho�man, L.L., 1973b. Leaf wax of Triticum

aestivum. Phytochemistry 12, 2217±2223.

Volkman, J.K., 1986. A review of sterol markers for marine

and terrigenous organic matter. Organic Geochemistry 9,

83±99.

Walton, T.J., 1990. Waxes, Cutin and Suberin. In: Harwood,

J.L., Bowyer, J.R. (Eds.), Methods in Plant Biochemistry,

vol. 4. Academic Press, London, pp. 105±158.


Documents

Organic geochemical evidence for the origin of ancient ...chidb/personal/content/paper8.pdf · Organic geochemical evidence for the origin of ancient anthropogenic soil deposits at