highresolution (pixe) analyses of carbonate deposits …€¦ · 13094 aix-en-provence cedex 2,...

HIGH-RESOLUTION (PIXE) ANALYSES OF CARBONATEDEPOSITS IN A ROMAN AQUEDUCT (FRÉJUS, SE FRANCE):

IMPLICATIONS FOR THE STUDY OFPALAEOHYDROLOGICAL VARIABILITY AND WATER

RESOURCES MANAGEMENT IN SOUTHERN GAUL DURINGTHE ROMAN PERIOD*

C. BOBÉE,1† S. HUON,2 J.-L. GUENDON,3 J. SALOMON,4 C. GÉBARA,5 J.-M. MICHEL6

and M. REGERT4

1Université Pierre et Marie Curie (UPMC)—UMR 7619 Sisyphe, Case 105, 4, place Jussieu, 75252 Paris cedex 05, France2Université Pierre et Marie Curie (UPMC)—UMR 7618 Bioemco, Case 120, 4, place Jussieu, 75252 Paris cedex 05, France

3Maison Méditerranéenne des Sciences de l’Homme (MMSH)—UMR 6638, BP 647, 5, rue du château de l’Horloge,13094 Aix-en-Provence cedex 2, France

4Centre de Recherche et de Restauration des Musées de France (C2RMF)—UMR 171, Palais du Louvre, Porte des Lions, 14,Quai François Mitterrand, 75001 Paris cedex 01, France

5Service Départemental d’Archéologie, Conseil Général du Var, Le Clos de la Tour, 17, rue Gustave Bret,83300 Fréjus, France

6Institut National de Recherches Archéologiques Préventives (INRAP), 561, rue Etienne Lenoir, km delta,30900 Nîmes, France

Urban water was supplied to the Roman city of Forum Julii (Fréjus, southeastern France) forat least 200 years by a 39.4 km long aqueduct, operating in AD 50. Two perennial springs weresuccessively collected, the Foux and the Siagnole, located at the outlet of Triassic and Jurassickarstic reservoirs, respectively. In this study, we performed high-resolution PIXE (Particle-Induced X-ray Emission) measurements of Ca, Si, Fe and Sr concentrations on selectedcross-sections of laminated carbonate sampled along the sidewalls, before and after theconnection of the two collection channels. Seasonal variations of water composition, sus-pended sediment load and discharge are recorded by alternating clear and dark layers, andcan be traced by their Sr and Fe contents. On the basis of an annual bimodal high-dischargeregime for the two karstic reservoirs, the concentration measurements allow the derivation ofa chronological record (117.5 years) of water supply and maintenance activities for a part ofthe operational period of the aqueduct. The water level in the channel was mainly controlledby the extent of carbonate deposition and by dredging, repair and maintenance operationsrather than by the past hydrological regime of the two springs.

KEYWORDS: AQUEDUCT, FORUM JULII, CARBONATE DEPOSITS, PIXE ANALYSES, WATERRESOURCE MANAGEMENT, ROMAN PERIOD

INTRODUCTION

Urban water supply to the Gallo-Roman city of Forum Julii (Fréjus, southeastern France) wasaccomplished by building a 39.4 km long aqueduct collecting the perennial Foux and Siagnolesprings, located in the hinterland, north-east of the city. At the decline of Forum Julii, the

*Received 20 October 2009; accepted 8 April 2010†Corresponding author: present address LSCE, UMR CEA-CNRS-UVSQ, Boite courrier 132. 91191 Gif-sur-Yvette, France; email

[email protected]

Archaeometry 53, 2 (2011) 241–260 doi: 10.1111/j.1475-4754.2010.00544.x

© University of Oxford, 2010

aqueduct, devoted to public monuments, baths and fountains, was abandoned and part of itsstructure was dismantled (i.e., water intakes for irrigation and mills in medieval times, stonerecovery). The channel fed by the Siagnole spring in the remotest part of the aqueduct wasrepaired and partially reconstructed during the 19th century and is still used for water distribution(c. 11 ¥ 106 m3 yr-1; Gébara 2002a). The discovery in 1988 of a first channel collecting the Fouxspring, 24 km north of Forum Julii (Michel et al. 1994) renewed interest in the study of the waterresource management ability of the builders. As comparisons with present-day conditions arepossible using palaeo-environmental data covering this period (e.g., Jorda and Provansal 1996;Holzhauser 1997; Mann and Jones 2003; Holzhauser et al. 2005; Ollivier et al. 2006), derivationof information on the past hydrological regime of the springs during the period of Romansettlement in southern Gaul was also expected. The reasons for building a two-step watercollection system are complex and are attributed to an increasing need for additional waterresources, linked to the expansion of Forum Julii and the building of new public baths andfountains during the second half of the first century ad.

A means of answering this question is provided by the study of Ca-carbonate deposits (alsocalled travertine—Fouke et al. 2000; Pentecost 2005) accumulated along the internal sidewallsof the aqueduct channels (Adolphe 1973; Fabre and Fiches 1986; Bonté 1999; Blanc 2000;Guendon and Vaudour 2000; Joseph et al. 2000; Bobée 2002; Dubar 2006; Carlut et al. 2009).These deposits are organized as speleothems (cave deposits) with the succession of alternatingclear and dark millimetric to sub-millimetric laminae (Baker et al. 1993; Genty and Quinif 1996;Genty et al. 1997; Frisia et al. 2000). Although the calcification mechanism remains uncertain,some characteristics of laminated-type carbonates, such as organic matter content, colour,porosity or chemical and stable-isotope compositions, are currently used as palaeohydrologicalproxies, providing estimates of climatic parameters such as temperature and precipitation inten-sity (Gascoyne 1983; Shopov et al. 1994; Rousseau et al. 1995; Brook et al. 1999; Fairchild et al.2001; Genty et al. 2001a, 2002, 2003; Polyak and Asmeron 2001; Baker et al. 2002; Burns et al.2002; Frisia et al. 2003; Niggemann et al. 2003; Tooth and Fairchild 2003; Wang et al. 2005;McDonald et al. 2007; Zhang et al. 2008). In contrast to cave deposits, the clear layers ofcarbonate deposits in aqueducts are usually related to periods of high water discharge (‘rainyseasons’), whereas the dark layers are rather attributed to low water stages (‘dry seasons’)(Guendon et al. 1994, 2002; Bobée 2002). However, they may not only record seasonal changesin water composition but also fingerprint single hydrological events (flash floods). If cavedeposits result from water infiltration through the overlying soils and rocks, roman aqueducts insoutheastern France collect water at the outlet of perennial springs and convey regulated flowswithin underground channels. The Roman builders developed ‘skilful’ technologies to controldischarge, including upstream retention tanks, steep channel sections to dissipate flow kineticenergy, culverts or spillways for the release of excess storm water, distribution tanks andhydraulic cement application on the sidewalls (Hodge 1992; Chanson 2008). A common attributebetween cave and aqueduct deposits is the occurrence of dark layers enriched in organic matterand clay, oxide–hydroxide minerals or even charcoals (Baker et al. 1997; Ramseyer et al. 1997;Genty et al. 2001b; Perrette et al. 2008; Carlut et al. 2009). These inputs may also recordtemporary damage due to root penetration or local collapses of the tunnel structure. The watersupply must be interrupted to repair, clean and dredge the channels. Therefore, the overallcarbonate deposits provide an intricate record that accounts both for natural spring dischargevariations and periods of water shortage for maintenance activities.

The aims of this study were: (1) to perform high-resolution (100–200 mm) concentrationmeasurements of selected elements of laminated carbonate deposits of the Roman aqueduct of

242 C. Bobée et al.

© University of Oxford, 2010, Archaeometry 53, 2 (2011) 241–260

Forum Julii; and (2) to discuss the meaning and the reliability of these measurements forhydrological and climatic reconstructions. Two sites were selected. The first one is located in theupstream part of the aqueduct and corresponds to carbonate deposits solely in connection withthe remotest (Siagnole) spring. The second one accounts for the first collected proximal spring(Foux) and involves the additional contribution of the Siagnole. Better constraints on the chro-nology of water collection and on the modalities of water management, using geochemicalfingerprinting of water supply by carbonate deposits, are also expected from comparisonsbetween these two locations.

THE PHYSIOGRAPHIC SETTING OF THE FORUM JULII AQUEDUCT

The archaeological setting

From the foundation of the city (in around 50 bc) until the middle of the first century ad, all ofthe domestic quarters, public baths, fountains and water reservoirs of Forum Julii were fed bylocal water resources; that is, from collection and storage of runoff and rainwater, and from wells.Excavations and observations carried out during the 19th century provide evidence for anorganized sewerage system, completed by numerous cisterns and wells (Gébara 2002b). Towardsthe middle of the first century ad, major changes took place with the development of publicfacilities such as monumental fountains (nymphaeum) and baths. Street fountains were built andan additional sewerage system was implemented. This additional demand coincided with theconstruction of a 24 km long aqueduct, conveying water by gravity from the Foux spring to acollection tank (castellum divisorum), located at the highest position of the city (Fig. 1).However, accurate dating to the precise year is not possible. The construction of the aqueductmost probably took place after ad 20–25, when an ultimate local groundwater collection systemwas achieved (the ‘Porte d’Orée’ drain; Fig. 1). According to the chronology of building opera-tions in Forum Julii (i.e., the ‘Clos de la Tour’ sewer system and the ‘Moulin à Vent’ nymphaeum;Fig. 1) and to the occurrence of carbonate deposits in pipes and channels, water distribution bythe aqueduct was fully operational in ad 50 (Gébara 2002b).

During the following years, a 15.4 km long extension of the former aqueduct to the north wasbuilt to collect the Siagnole spring (Guendon et al. 1994, 2002). Several monuments (i.e., the‘Porte d’Orée’ nymphaeum, the cardo maximus fountain and the ‘Clos Saint Antoine’ baths;Fig. 1) bear witness to a continuous and enhanced water supply operating in ad 70. The datingof a newly excavated fountain (the ‘Mangin’ fountain; Pasqualini et al. 2006) confirms this fact.During the second century ad, a major thermal establishment was constructed (the ‘Porte d’Orée’baths; Fig. 1), benefiting from water distribution in Forum Julii (Gébara 2002b). Pipes wereinstalled in streets (the ‘Clos de la Tour’; Fig. 1) and repairs were carried out for two aqueductbridges and one of the public baths. The final interruption of the water supply by the aqueductmay only be estimated, as being between ad 250, when the repairs were carried out, and c. ad350, when most of the buildings and monuments located near the ramparts (including thecastellum divisorium) were no longer used.

The excavations show that sections of the Forum Julii aqueduct were generally built in coveredtrenches, with a concrete U-shaped internal channel about 0.70 m wide and 1.30 m deep (cae-menticium, a hydraulic cement made of a mixture of quicklime, sand and crushed bricks asdescribed by the Roman architect Vitruvius in De architectura, Book 8), surmounted by verticalsidewalls and covered with a vaulted roof (Fig. 1; Michel 2002). A fine finishing layer ofwaterproof mortar was added to the coarse caementicium. Direct access for maintenance or

Carbonate deposits in a Roman aqueduct (Fréjus, SE France) 243


Figu

re1

(a)A

sim

plifi

edge

olog

ical

map

ofth

ehi

nter

land

ofF

réju

s,w

ithth

elo

catio

nsof

the

mai

nsi

tes

cite

din

the

text

(mod

ified

from

Géb

ara

etal

.200

2b).

(b1–

b4)S

ketc

hes

ofth

em

ain

chan

nelt

ypes

(mod

ified

from

Mic

hel2

002)

.(c)

Asc

hem

atic

map

ofth

em

ain

hydr

aulic

stru

ctur

es,b

uild

ings

and

mon

umen

tsde

vote

dto

wat

ercu

ltsof

the

antiq

ueFo

rum

Julii

(Fré

jus)

,with

the

loca

tion

ofth

eaq

uedu

ct(m

odifi

edfr

omG

ébar

a20

02a)

.



inspection purposes was realized through vertical shafts. The slope of the aqueduct channelgradually increases from 0.13% (1.3 m km-1) near the Siagnole spring to 8.20% 10 km down-stream (Fig. 1). Between this latter location and the connection with the Foux channel, the slopedecreases to 1.52%. The crossing of valleys required the construction of numerous bridges onarches along the course of the aqueduct and tunnels were dug for three different sections (Gébaraand Michel 2002).

The geological setting

The Siagnole and Foux springs collect groundwater from contrasting geological formations, theJurassic karst plateau located between the Audibergue massif and the Artuby valley (Julianand Nicod 1984; Etienne 1987) and the Triassic formations of the Fayence–Callian depression(Michel et al. 1994; Gébara et al. 2002), respectively. The Roman aqueduct crosses three maingeological formations (Fig. 1), which are, from north to south: (1) the Provençal terrace, betweenthe Siagnole spring and the Saint-Cassien lake; (2) the Carboniferous formations of the Reyranvalley; and (3) the Permian formation of the Bas-Argens. The Provençal terrace is composed ofJurassic limestones and dolomites in its northern part and of Triassic limestones, dolomites,marls, clays and gypsum (Keuper formation) in its central part. The Permian metamorphic seriesare characterized by volcanic (rhyolite) and meta-sedimentary (pelites, sandstones, tuffs andconglomerates) formations. The major lithological contrast is the boundary between the Jurassic–Triassic limestones and the Carboniferous–Permian rocks (Bordet and Mennessier 1966; Toutin-Morin and Crévola 1994). The antique city of Forum Julii and its harbour were set up on Plioceneand Quaternary alluvial deposits overlying Permian bedrocks. Therefore, the carbonate depositsin the pipes and channels of the city are not linked to local groundwater collection and transport.Their occurrence indicates water collection from the karstic springs of the hinterland.

The hydrological setting

The present-day rainfall and Siagnole river discharge data at Mons have been downloadedfrom the MétéoFrance (http://www.meteofrance.fr) and DIREN (Direction Régionale del’Environnement, http://www.hydro.eaufrance.fr) websites. Average monthly precipitation anddischarge records have been calculated using the two databases (1998–2007, 220 measurementseach). Although the precipitation record is not directly linked to the discharge of the Siagnolespring, which drains a large watershed, its temporal variation should provide a first-order estimateof seasonal outflow for the springs located in the region. The transport of water to Forum Juliitakes place within covered channels or underground galleries (Fig. 1), and water temperaturechange is not the major process controlling carbonate deposition. Only a limited increase, from12°C at the spring outlet to 12.5°C 500 m below (the Roche-Taillée site; Fig. 1), has beenmeasured in July 2003 in the covered section of the modern channel.

MATERIALS AND METHODS

Sample collection

Chemical analyses were carried out on carbonate deposits sampled, before and after the connec-tion of the Siagnole and the Foux channels, at the Vallon de la Route site (site 6, Fig. 1) and theLes Adrets site (site 7, Fig. 1), respectively. The former site is located 500 m before the conflu-



ence. Only carbonate deposits are preserved. They now crop out at ground level (Fig. 2), with amaximum thickness of 12.2 cm and a major unconformity at 4.6 cm from the sidewall.

Les Adrets (also called Pra Bousquet) is located in the Reyran valley, c. 8 km below theSiagnole–Foux connection (Fig. 1). The sample, collected on the upper part of the sidewall, hasa total thickness of 19.3 cm (Fig. 2) and displays remarkable regressive layers (at 4.1–6.0 cm and9.6–12.5 cm from the sidewall) and transgressive layers (at 6.0–9.6 cm and 12.5–19.3 cm fromthe sidewall), with no major unconformity. The first deposits only linked to water supply by theFoux spring are missing for the studied section, but have been described for the Boson andJaumin sites (Fig. 1; Guendon et al. 2002), with which cross-correlations could be made (seediscussion). There is also evidence that part of the deposits has been removed by dredging thesidewall accumulations, but this operation was carried out above the studied section (Fig. 2).

Spectral analyses of hydrological data

Spectral analyses of precipitation and discharge data were carried out with the Arand softwarepackage (Howell 2001), following the method of Blackman and Tukey (1958). The rainfalldatabase was sampled at constant 1 month intervals and linearly de-trended before analysis. Theprocedure consists of computing the autocorrelation matrix of the data and in applying a Fouriertransform (80% confidence interval, number of lags calculated by dividing the number of valuesby 3, 0% pre-whitening).

High-resolution PIXE analyses of carbonate deposits

All measurements were carried out along linear profiles oriented in the growth axis of thecarbonate laminae, using the AGLAE accelerator facility (Accélérateur Grand Louvre AnalyseElémentaire—Calligaro et al. 2004; Salomon et al. 2008) of the C2RMF (Centre de Rechercheet de Restauration des Musées de France) in Paris. Particle-induced X-ray emission (PIXE)analyses were performed on polished sections to reduce beam deviations due to surface rough-ness. X-ray emissions were measured using two Si (Li) detectors (Calligaro et al. 2000). The

Figure 2 (a) An outcrop at ground level of the carbonate deposits for Vallon de la Route (photograph by J.-L. Guendon).(b) A view of the channel excavated in the Callian region (photograph by M. Cecchinato). (c) A carbonate sample fromLes Adrets (photograph by P. Foliot and M. Olive, modified from Guendon et al. 2002).



distance between two measuring spots was set at 100 mm for the Vallon de la Route sample and200 mm for the Les Adrets sample. Raw spectral data were recorded in ‘counts per second’converted to concentration units—that is, ppm (parts per million equivalent to mg kg-1)—usingthe GUPIX software package (Maxwell et al. 1995).

RESULTS

Present-day precipitation and discharge data

The distribution of monthly precipitation at Mons and monthly discharge of the Siagnole between1989 and 2007 (19-year record) is displayed in Figure 3. The data are characterized by highinter-annual variability, with a mean annual precipitation (11s) of 925 1269 mm yr-1 and a meanannual discharge (1 1s) of 478 1 209 mm yr-1.

Most years display two distinguishable ‘rainy seasons’, although the values bear highstandard variations (high inter-annual variability). The two rainfall maxima averaged (1 1s)92.7 1 64.6 mm and 132.0 1 90.7 mm, for April and October, respectively (Fig. 3). The‘dry seasons’ are characterized by rainfall minima in February (44.0 1 42.5 mm) and July(36.8 1 30.9 mm). The average discharge values of the Siagnole river are (1 1s) 30.8 1 26.2 mmand 3.4 1 6.7 mm for March and August (the ‘dry seasons’), respectively, and rise to47.1 1 37.2 mm and 84.4 1 77.4 mm for May and November (the ‘rainy seasons’), respectively.Spectral analyses of rainfall and discharge data provided prominent 6 and 12 month periods,consistent with a bimodal seasonal frequency (Fig. 3). Due to recent urbanization and enhancedgroundwater collection in the Fayence–Callian region, the Foux spring only delivers intermittentand very low outflows. We assumed that its past and present-day discharge distributions wereequivalent to that of the present-day Siagnole spring. This assumption is supported by studies thatreveal comparable climate conditions between the present-day and Gallo-Roman periods (forreferences, see the Introduction), although alternating low and high flood regimes connected toland degradation processes have been reported along the Rhône valley (i.e., Salvador et al. 2002;Van der Leeuw and ARCHAEOMEDES research team, 2005).

PIXE analyses of carbonate deposits

The composition of the carbonate deposits is inherited both from the chemical composition ofeach spring and from the occurrence of detrital components trapped within the carbonate depos-its. The Foux spring provides gypsum-saturated groundwater, whereas the Siagnole springinvolves diluted Ca-bicarbonate waters (Guendon et al. 1994). Carbonate layers with higheraverage Sr concentrations highlight major contributions of the Foux spring. Their incorporationand/or exchange with flowing water during carbonate crystallization have been extensivelydescribed for other travertine deposits (e.g., Gascoyne 1983; Tesoriero and Pankow 1996;Verheyden et al. 2000; Huang and Fairchild 2001; Huang et al. 2001). The distribution of the Ca,Si, Fe and Sr concentrations is displayed in Figure 4 for Vallon de la Route and in Figure 5 forLes Adrets (average concentrations in Table 1).

The average PIXE concentrations are consistent with previous measurements for both sites(Guendon et al. 1994). The carbonate deposits of the Vallon de la Route are composed ofalternating clear and dark laminated deposits, with no significant change in thickness (Fig. 6).Within a representative section, the thickness averaged (11s) 310 1 200 mm and 220 1 130 mmfor clear and dark laminae, respectively (262 1 167 mm for all laminae, 175 measurements).



Figu

re3

(a)

A19

-yea

rre

cord

(198

9–20

07)

ofth

epr

ecip

itatio

nat

Mon

san

dth

ew

ater

disc

harg

eof

the

Siag

nole

and

(b)

aver

age

mon

thly

disc

harg

esan

dpr

ecip

itatio

ns(s

tand

ard

devi

atio

nis1

1s;

dark

arro

ws

outli

neye

ars

char

acte

rize

dby

two

mar

ked

rain

yse

ason

s).(

c)A

plot

ofsp

ectr

alde

nsity

vers

usfr

eque

ncy

(or

peri

od)

for

the

mon

thly

prec

ipita

tion

atM

ons;

and

(d)

apl

otof

spec

tral

dens

ityve

rsus

freq

uenc

y(o

rpe

riod

)fo

rth

em

onth

lydi

scha

rge

ofth

eSi

agno

le(C

Ire

fers

toco

nfide

nce

inte

rval

).



No clear relationship links Ca, the major constituent of carbonate deposits, and other elements.However, several Ca minima are phased with high Fe, Sr and Si concentration peaks (Fig. 4).These relative drops correspond to more porous carbonate layers, either accumulating solidsuspended matter or incorporating higher dissolved loads during calcification. Fe and Si behavein a similar way but not systematically, and opposite trends are also displayed for Sr and Si(Fig. 7). A given clear and porous lamina is enriched in Sr with respect to the thin and darklaminae located on both sides, but the extent of the chemical discrimination is variable along thesection. The difference between dark and clear laminae is best highlighted by the ratio of the Srto Fe concentrations (Fig. 8).

In contrast to the Vallon de la Route, the thickness of the laminae at Les Adrets increasesthoroughly from the sidewall to the centre of the channel (Fig. 6). The first laminae have a lowthickness (550 1 340 mm, 28 measurements, 2.0–3.5 cm from the sidewall) with respect to theultimate laminae (1760 1 760 mm, 37 measurements 13.0–19.3 cm from the sidewall). Thesedifferences in size also correspond to changes in Sr content, which increases from 228 1 41 ppm

0100000200000300000400000500000

Ca (ppm)

0

1000

2000

3000Fe (ppm)

0

4000

8000

12000Si (ppm)

0

100

200

300

400Sr (ppm)

0 1 2 3 4

0.0

0.4

0.8

1.2Sr/Fe

5 6 7 8 9 10 11 12

side

wal

l

Distance from sidewall (cm)

Figure 4 Plots of Ca, Fe, Si and Sr concentrations and Sr/Fe versus distance from the sidewall measured by PIXE forVallon de la Route carbonate deposits. The spot resolution is 30 mm, with 100 mm intervals between each measurement.A sample photograph is displayed above the plots. Arrows refers to prominent Ca-depleted deposits linked to high Fecontents.



(2.0–3.5 cm, 75 measurements; Fig. 6) to 561 1 86 ppm for the last 6.3 cm (13.0–19.3 cm, 323measurements; Fig. 6). The former average concentration, although higher, approaches the valueobtained for the Vallon de la Route (104 1 32 ppm; Table 1), reflecting a major water supply bythe Siagnole spring. In contrast to Sr, the Si and Fe concentrations do not display any increasingtrend along the studied section (Fig. 5). The difference between dark and clear laminae is outlinedby changes in Si and Fe concentrations, as for Vallon de la Route, but the average concentrationsare significantly higher; that is, 8767 1 1642 ppm versus 4179 1 1357 ppm for Si (Table 1).Although carbonate deposition at Les Adrets involves collection of waters supplied by the twosprings, the thick and clear laminae still provide higher Sr/Fe than the thin and dark laminae, asshown for Vallon de la Route (Fig. 8).

Figure 5 Plots of Ca, Fe, Si and Sr concentrations and Sr/Fe versus distance from the sidewall measured by PIXE forthe Les Adrets carbonate deposits. Sr concentrations are also reported for the Boson (empty squares) and Jaumin (blacksquares) carbonate deposits. The spot resolution is 30 mm, with 200 mm intervals between each measurement. A samplephotograph is displayed above the plots (Rg, regressive layers; Tr, transgressive layers).



Table 1 Average concentrations for selected carbonate deposits

Elementconcentration(ppm)

Vallon de laRoute, PIXE(this study)

Vallon dela Route

(Guendonet al. 1994)

Les Adrets, PIXE(this study)

Boson (Guendonet al. 1994)

Mean* s* Mean s* Mean* s* Mean s

Ca 389 405 41 917 391 678 4 150 381 607 29 402 382 542 2 261Si 4 179 1 357 4 343 1 041 8 767 1 642 8 739 1 887Fe 751 428 752 153 982 607 984 250Sr 104 32 104 8 390 162 396 99

*Means and standard deviations of 1223 measurements for the Vallon de la Route site and 973 measurements for the Les Adrets site.

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Distance from sidewall (cm)

0.0

0.1

0.2

0.3

0.4

Thickness (cm)0

200

400

600

800

[Sr] (ppm)(b)

0 1 2 3 4Distance from sidewall (cm)

0

0.04

0.08

0.12

0.16

Thickness (cm) 0

100

200

300

400[Sr] (ppm)(a)

Figure 6 Plots of lamina thickness and Sr concentration versus distance from the sidewall for (a) a selected section ofthe Vallon de la Route carbonate deposits and (b) the carbonate deposits from Les Adrets.



DISCUSSION AND INTERPRETATION

The hydrological significance of laminated carbonate deposits of aqueducts

Deposition of Ca carbonates in the channel of the Forum Julii aqueduct is controlled bydegassing of surfacing CO2-rich groundwater, originating from two large karstic systems withcontrasting geological bedrocks. Excess water with respect to the maximum capacity of theaqueduct was routed to the rivers by the collection systems (Gébara and Michel 2002). Anyphysical, chemical and biological process that enhances CO2 degassing in the aqueduct—that is,slope breakdown or increases in sidewall rugosity, temperature and/or biological activity—willfavour carbonate deposition (Stumm and Morgan 1996). Over time, this process will also reducewater velocity and ultimately discharge in the collection system by increasing friction with thesidewalls. There is evidence that the Roman engineers were aware of the problem, as shown bythe application along the sidewalls of a fine-grained hydraulic cement to reduce turbulence andallow easier removal of the carbonate deposits. However, changes in the aqueduct slope—inparticular, before the connection with the Foux channel (0.13–8.20%)—indicate that the role ofturbulence and degassing processes was not fully understood. Discharge calculations were basedon reference sections of the aqueduct channel (Bonnin 1984) rather than on the mathematicalproduct of velocity and wet section area. The direct link between carbonate deposition and waterdischarge is, however, puzzling. Deposition models based on a thin stagnant water layer at thesurface of travertine predict an increasing rate, with a decreasing dissolved CO2 concentrationand an increasing layer thickness (Dreybolt 1988; Dreybolt and Buchmann 1991). The process isthoroughly enhanced by turbulence in the overlying water column. Therefore, an increase in the

Figure 7 Plots of Fe, Si and Sr concentrations versus distance from the sidewall for the first laminae of Vallonde la Route.



deposition rate is expected with an increase in the discharge, and vice versa. However, evidencefor positive as well as for negative links between carbonate accumulation and water discharge hasbeen reported for locations where water evaporates or substrates lead to erosion of the previousdeposits (Kano et al. 1999; Drysdale et al. 2003; Ihlenfield et al. 2003). The carbonatesaccumulated in the aqueduct channel only display small ‘erosion marks’ in their bottomparts (Guendon et al. 2002), and the balance between carbonate deposition and post-depositionalerosion may only partly explain the accumulation pattern. As spring water collection wasregulated, the progressive increase in thickness of the laminae after the connection of the Fouxand Siagnole channels (Fig. 6) reflects additional water collection but does not mean majorchanges in the past hydrological regimes of the karstic reservoirs. The best explanation is that theaccumulation of carbonates progressively modified the sidewall roughness and enhanced turbu-lence. Transgressive layers tend to reflect the rise of the water level due to a progressive reductionof the channel section by carbonate deposits rather than an increase of spring outflow (Figs 2 and5). Using height measurements of carbonate deposits along sidewalls, characteristic frictioncoefficients (Bailhache 1979) and slope measurements, Guendon et al. (2002) could determineinternal wet sections, from which water discharge estimates could be derived (180–250 l s-1 and

15 16 17 18 19

0.0

0.5

1.0

1.5

Sr/Fe

(b)

0.0 0.5 1.0

0.0

0.1

0.2

0.3

0.4Sr/Fe

(a)

Length (cm)

Length (cm)

Figure 8 Plots of Sr/Fe versus carbonate length for selected sections of (a) the Vallon de la Route carbonate depositsand (b) the Les Adrets carbonate deposits. The grey and white bands correspond to the dark and clear laminae,respectively.



60–280 l s-1, for the Foux and the Siagnole, respectively). When the water collection system wasfully operational, the total discharge exceeded 500 l s-1 beyond the confluence, for a maximumcapacity of 477–822 l s-1. Water supply by the remote Siagnole spring was thoroughly reduced orinterrupted several times during the following years, either for maintenance or repair activities.Several regressive layers account for these operations (Figs 2 and 5) and some of the high Fe andSi concentration peaks (Figs 4 and 5) may fingerprint local damage to the tunnel structure. Theseinterpretations are confirmed by the rupturing and abandonment of a primitive conduct at LaRoche Taillée (site 3, Fig. 1). Towards the end of the collection period, the total supply to ForumJulii decreased to 240–170 l s-1, with a major contribution from the Foux, shown by the persis-tently high Sr concentrations.

The PIXE analyses revealed that carbonate laminae might be described by their chemicalcomposition; in particular, by Sr/Fe ratios. Trace elements such as Sr enable discriminationbetween the types of laminae, as reported for other modern and archaeological carbonates (e.g.,Carlut et al. 2009). For a unique water supply (Vallon de la Route), the dark and compact laminaeare enriched in Fe and Si with respect to the clear and porous laminae due to transport of detritalparticles and their incorporation within carbonate deposits. Such conditions occur either duringlow water stages by settling of suspended matter that is usually exported further downstream in theaqueduct channel or following periods of enhanced infiltration of soil water linked to damage in thetunnel structure. The dark laminae were formed during these periods. When low precipitationconditions prevail in the watershed—that is, during the ‘dry seasons’ (winter and summer)—thewater levels within the karst reservoirs are low. They are only refilled by enhanced water infiltrationduring the ‘rainy seasons’ (spring and autumn) or after seasonal snow-melting in the watershed.Pre-existing water may be flushed out during these periods (e.g., Lastennet 1994; Baker et al. 1997;Plagnes and Bakalowicz 2002). The geochemical composition of spring waters is, however,controlled by complex types of hydrological behaviour (Ford and Williams 2007) and varyingfractions of the rainfall input will contribute to either slow or quick outflow generation (e.g., Pinaultet al. 2001; Aquilina et al. 2003). High dissolved load fluxes are expected when outflow involvesthe saturated zones of karstic reservoirs and displaces large amounts of pre-existing water. Incontrast, dilution by rainwater after major storms or flash floods should provide diluted waters(Dreiss 1989; Petelet-Giraud and Negrel 2007; Maréchal et al. 2008). Clear and porous laminaewith a high Sr content (and a low Fe content) were deposited during periods of flush flow, mostprobably at the beginning of the rainy seasons. In contrast, a rapid transfer of rainwater will supplyspring water with a lower mineralization status (a lower Sr content), due to more limited contactwith the bedrocks of the karst reservoirs. This situation should correspond to the deposition ofintermediate clear–dark layers. On the whole, the Sr/Fe ratios (Fig. 8) provide a first-orderfingerprint of clear and dark laminae that covers a large range of hydrological situations. Themixing of saturated groundwater collected by the Foux spring with diluted Siagnole water at LesAdrets (Guendon et al. 1994) should control the composition and the thickness of the laminae(Fig. 6). Higher Sr concentrations (with respect to Vallon de la Route) fingerprint a higher watercontribution of the Foux spring end-member in the mix. In contrast, a high contribution of theSiagnole end-member lowers the Sr concentration. This process is superimposed with the one dueto seasonal discharge variations on the basis of synchronous behaviour for the two karstic systems.

The derivation of a water supply record from the carbonate deposits of Forum Julii

Assuming a synchronous bimodal high discharge regime for the two springs, one hydrologicalyear should correspond to two ‘rainy’ and two ‘dry’ periods. On the basis of high concentrations



for high discharges (the clear laminae of the rainy periods) and low concentrations for lowdischarges (the dark laminae of the dry periods), two maxima and two minima should occur everyyear. A total of 470 maxima and minima could thus be extracted from the overall data, providing117.5 years (470/4) of carbonate deposition, equivalent to 1.64 mm yr-1 with respect to the totallength of the sample (193/117.5 mm yr-1; Fig. 5). Nearly equivalent results could be derived fromSr/Fe peak counts (114 years).

Because the first laminae are missing for Les Adrets, the carbonate deposits should not providea continuous record for the entire period of water supply to Forum Julii. A plot of Sr concen-tration versus distance from the sidewall allows comparison with other sites (Boson and Jaumin;Figs 1 and 5). Except for the first laminae, the overall compositions follow the same trend towardsthe centre of the aqueduct channel and are linked to the relative contribution of the two springs,fingerprinted by variable Sr concentration levels. The use of the palaeo-discharge model ofGuendon et al. (2002) strengthens this approach. Several periods, labelled P1–P6 in Figure 9,have been distinguished. During the first period (P1), water was solely supplied by the Fouxspring, as shown by high Sr concentrations in the carbonate deposits (559 1 8 and 533 1 6 ppmfor Boson and Jaumin, respectively; Fig. 5 and Guendon et al. 2002). Additional collection ofSiagnole water thoroughly increased the discharge from about 250 l s-1 to 530 l s-1 during thesecond period (P2) and matched, c. 25 years later, the extensive use of water for monuments inForum Julii. Deposition of the first laminae was initiated during this period of mixed watersupply. The increasing Sr content of the carbonate deposits (from 118 to 506 ppm; Fig. 5) of thethird period (P3) reflects an erratic and progressively decreasing contribution of the Siagnole. Theend of this increasing trend may be placed at the limit with the following period (P4), when watercollection from the Siagnole spring was interrupted. The P4 boundaries are highlighted by two Sr

Figure 9 The temporal evolution of the maximum water discharge estimated from the composite carbonate depositrecords for sites located after the connection between the Foux and the Siagnole channels and PIXE Sr concentrationmeasurements for Les Adrets (modified from Guendon et al. 2002). A, Collection from the Foux spring; B, additionalcollection from the Siagnole spring (+24 years); C, hydraulic improvements on bridges (+54–69 years); D, the firstdredging period (+77 years); E, the first collapse of the channel structure at Roche-Taillée (+119 years); F, the seconddredging period (+145 years); G, the second collapse of the channel structure (+164 years).



minima, at 118 ppm and 494 ppm, respectively (Fig. 9), between which a relatively constant highSr concentration (563 1 72 ppm) reveals a unique contribution of Foux spring water. Intermittentwater collection from the Siagnole proceeded during the following period (P5), as shown byhighly variable Sr concentrations for the last carbonate layers. According to our interpretation,the last period (P6) is not represented, most probably because carbonate deposits were dredgedor collapsed within the channel. Because the P4 period can be positioned within the Srconcentration curve, 117.5 years of carbonate deposition may be placed on the palaeo-dischargediagram, on the basis of two Sr maxima and two Sr minima per year (Fig. 9). The scaledcarbonate record provides a period of water supply spanning from +41 years (c. ad 91) to +155years (c. ad 209), with respect to the onset of water distribution in Forum Julii (c. ad 50; Gébara2002b).1 Relating these periods to the structural characteristics of carbonate deposits supports thischronological record. The first set of regressive–transgressive layers, linked to water-level varia-tions, provides evidence for repeated maintenance operations during the third period (P3), relatedeither to damage caused by excess water discharge or to warping by carbonate deposits (Guendonet al. 2002). Traces of repair and channel dredging have been spotted in the carbonate deposits atJaumin (+54 years), Esquine (+69 years) and in the Reyran valley (+77 years) using the samecounting technique. The second set of regressive–transgressive layers observed during the fourthperiod (P4, +119 to +139 years) is linked to the collapse of the channel structure during thisperiod (Roche-Taillée site; Fig. 1; also Guendon et al. 2002). Water supply to Forum Julii bythe aqueduct lasted for several more years (+181 to +218 years), as fingerprinted by carbonatedeposits of other sites (Guendon et al. 2002). Although the corresponding carbonate depositswere not recovered at Les Adrets, excavations performed in Forum Julii support a reduced butpersistent water supply until c. ad 350 (Gébara 2002b).

CONCLUDING REMARKS

In order to determine whether or not carbonate deposits of antique aqueducts are suitable forpalaeohydrological studies, high-resolution (100–200 mm) PIXE measurements of Ca, Si, Sr andFe concentrations were carried out on carbonate deposits of the Roman aqueduct of Forum Julii(Fréjus, southeastern France), which collected water from two perennial springs. The thick andclear laminae of the carbonate layers are enriched in Sr and depleted in Fe and Si with respect tothe thin and dark laminae. The former tend to characterize high discharges and high dissolvedloads that correspond to long periods of contact between groundwater and carbonate rocks of thegeological basements and flush flows from the karstic reservoirs during the ‘rainy seasons’. Thelatter tend to reflect low water stages (with low dissolved loads) typical of ‘dry seasons’ and/orenhanced detrital inputs due to damage in the channel structure. The overall geochemical recordof alternating dark and clear laminae covers a large range of hydrological situations that involvesperiods of water breakdown due to maintenance or repair activities. Using a simple assumptionon the past hydrological regime of the karstic reservoirs, the Sr concentration measurementsindicate that the carbonate sequences may provide a chronological record of collection andmixing of water originating from the two springs. Derivation of a time scale was also possible,but only for a part of the water supply period to Forum Julii (117.5 years), due to the absence,removal or collapse in the aqueduct channel of the first and last deposits. Therefore, the infor-mation that can be derived from aqueduct carbonate deposits must be considered with great careand may not be easily converted to efficient palaeohydrological proxies as in the case of cave

1According to Gébara (2002b), the building of the aqueduct took place in ad 20–25 and it was operational from ad 50.



deposits. The extension of the former Forum Julii aqueduct connecting the proximal Foux springto the remote Siagnole spring most likely corresponded to a concomitant expansion of the antiquecity, as revealed by the succession of water monuments, fountains, thermal baths, water pipes andsewerage networks operating between the middle of the first century ad and the end of the thirdcentury ad.

ACKNOWLEDGEMENTS

The authors are grateful to Thomas Calligaro (C2RMF) and Jean-Pierre Mohen (C2RMF) forproviding extensive help and support with AGLAE measurements.

REFERENCES

Adolphe, J.-P., 1973, Contribution à l’étude des encroûtements carbonatés de l’aqueduc du Pont du Gard, ComptesRendus de l’Académie des Sciences Paris, 277, 2329–32.

Aquilina, L., Ladouche, B., Doerfliger, N., and Bakalowicz, M., 2003, Deep-water circulation, residence time andchemistry in a karst complex, Groundwater, 41(3), 790–805.

Baker, A., Barnes, W. L., and Smart, P. L., 1997, Variations in the discharge and organic matter content of stalagmite dripwaters in Lower Cave, Bristol, Hydrological Processes, 11, 1541–55.

Baker, A., Proctor, C. J., and Barnes, W. L., 2002, Stalagmite lamina doublets: a 1000-year record of extreme winters inNW Scotland, International Journal of Climatology, 22, 1339–45.

Baker, A., Smart, P. L., Edwards, R. L., and Richards, D. A., 1993, Annual growth banding in a cave stalagmite, Nature,364, 518–20.

Bailhache, M., 1979, L’évolution des débits des aqueducs gallo-romains, Dossiers de l’Archéologie, 38, 82–7.Blackman, R. B., and Tukey, J. W., 1958, The measurement of power spectra from the point of view of communications

engineering, Dover, New York.Blanc, J.-J., 2000, La séquence des lamines carbonatées du chemin de fer au Grès (Sernhac), in L’aqueduc de Nîmes et

le Pont du Gard, archéologie, géosystème et histoire (eds G. Fabre, J.-L. Fiches and J.-L. Paillet), 249–61, CNRSéditions, Paris.

Bobée, C., 2002, Essai d’interprétation du fonctionnement des aqueducs romains de Nîmes et de Fréjus (à partir d’étudesgéochimiques du concrétionnement interne des conduits), Mémoire DEA Environnement et Archéologie, UPMC,Paris.

Bonnin, J., 1984, L’eau dans l’Antiquité. L’hydraulique avant notre ère, Eyrolles, Paris.Bonté, S., 1999, Essai d’interprétation du fonctionnement de l’aqueduc romain de Fréjus à partir d’une étude

pétrographique et isotopique du concrétionnement interne des conduits, Mémoire DEA Environnement et Archéolo-gie, UPMC, Paris.

Bordet, P., and Mennessier, G., 1966, Carte géologique de la France à 1/50000, 998, Fayence, Service de la cartegéologique de la France, Paris.

Brook, G. A., Rafter, M. A., Railsback, L. B., Sheen, S. W., and Lundberg, J., 1999, A high-resolution proxy record ofrainfall and ENSO since ad 1550 from layering in stalagmites from Anjohibe Cave, Madagascar, The Holocene, 9,695–705.

Burns, S. J., Fleitmann, D., Mudelsee, M., Neff, U., Matter, A., and Mangnini, A., 2002, A 780-year annually resolvedrecord of Indian Ocean monsoon precipitation from a speleothem from south Oman, Journal of GeophysicalResearch—Atmospheres, 107, D20 art., 4434, doi:10.1029/2001JD001281.

Calligaro, T., Dran, J.-C., Salomon, J., and Walter, P., 2004, Review of accelerator gadgets for art and archaeology,Nuclear Instruments and Methods in Physics Research, B226, 29–37.

Calligaro, T., Dran, J.-C., Poirot, J.-P., Querré, G., Salomon, S., and Zwaan, J. C., 2000, PIXE/PIGE characterization ofemeralds using an external micro-beam, Nuclear Instruments and Methods in Physics Research, B161–3, 769–74.

Carlut, J., Chazot, G., Dessales, H., and Letellier, E., 2009, Trace element variations in an archeological carbonate depositfrom the antique city of Ostia: environmental and archeological implications, Comptes Rendus Géosciences Paris,341, 10–20.

Chanson, H., 2008, The hydraulics of Roman aqueducts: What do we know? Why should we learn? World Environmentaland Water Resources Congress, 12–16 May 2008, Honolulu, Hawaii, ASCE.



Dreiss, S. J., 1989, Regional transport in a karst aquifer system, 1: component separation of spring flow hydrographs,Water Resources Research, 25(1), 117–25.

Dreybolt, W., 1988, Processes in karst systems, Springer-Verlag, Berlin.Dreybolt, W., and Buchmann, D., 1991, A mass transfer model of dissolution and precipitation of calcite from solutions

in turbulent motion, Chemical Geology, 90, 107–22.Drysdale, R. N., Lucas, S., and Carthew, K., 2003, The influence of diurnal temperatures on the hydrochemistry of a

tufa-depositing stream, Hydrological Processes, 17, 3421–41.Dubar, M., 2006, Approche climatique de la période romaine dans l’est du Var: recherche et analyse des composantes

périodiques sur un concrétionnement centennal (Ier–IIème siècle apr. J.-C.) de l’aqueduc de Fréjus, Archéosciences, 30,163–71.

Etienne, M., 1987, Les sources de la Siagnole de Mons (Var, canton de Fayence). Contribution à l’étude des sources etdu bassin d’alimentation. Etude et carte de la vulnérabilité des eaux souterraines à la pollution, Thèse Doctorat,Université de Montpellier, France.

Fabre, G., and Fiches, J.-L., 1986, Les concrétionnements de l’aqueduc romain de Nîmes, Méditerranée, 12,66–70.

Fairchild, I. J., Baker, A., Borsato, A., Frisia, S., Hinton, R. W., McDermott, F., and Tooth, A. F., 2001, Annual tosub-annual resolution of multiple trace-element trends in speleothems, Journal of the Geological Society of London,158, 831–41.

Ford, D., and Williams, P., 2007, Karst hydrogeology and geomorphology, Wiley, New York.Fouke, B. W., Farmer, J. D., Des Marais, D. J., Pratt, L., Sturchio, N. C., Burns, P. C., and Discipulo, M. K., 2000,

Depositional facies and aqueous-solid geochemistry of travertine-depositing hot springs (Angel Terrace, MammothHot Springs, Yellowstone National Park, USA), Journal of Sedimentary Research, 70, 565–85.

Frisia, S., Borsato, A., Fairchild, I. J., and McDermott, F., 2000, Calcite fabrics, growth mechanisms, and environmentsof formation in speleothems from the Italian Alps and Southwestern Ireland, Journal of Sedimentary Research, 70,1183–6.

Frisia, S., Borsato, A., Preto, N., and McDermott, F., 2003, Late Holocene annual growth in three Alpine stalagmitesrecords the influence of solar activity and the North Atlantic Oscillation on winter climate, Earth and PlanetaryScience Letters, 216, 411–24.

Gascoyne, M., 1983, Trace element partition coefficients in the calcite–water system and their paleoclimatic significancein cave studies, Journal of Hydrology, 61, 213–22.

Gébara, C., 2002a, Introduction, in L’aqueduc romain de Fréjus. Sa description, son histoire et son environnement (edsC. Gébara, J.-M. Michel and J.-L. Guendon), 23–39, Revue Archéologique de la Narbonnaise supplément 33,Montpellier.

Gébara, C., 2002b, L’alimentation en eau de Forum Julii, in L’aqueduc romain de Fréjus. Sa description, son histoire etson environnement (eds C. Gébara, J.-M. Michel and J.-L. Guendon), 261–303, Revue Archéologique de la Narbon-naise supplément 33, Montpellier.

Gébara, C., and Michel, J.-M., 2002, Description des vestiges, in L’aqueduc romain de Fréjus. Sa description, sonhistoire et son environnement (eds C. Gébara, J.-M. Michel and J.-L. Guendon), 43–86, Revue Archéologique de laNarbonnaise supplément 33, Montpellier.

Gébara, C., Michel, J.-M., and Dubar, M., 2002, L’aqueduc dans son contexte géologique et géomorphologique, inL’aqueduc romain de Fréjus. Sa description, son histoire et son environnement (eds C. Gébara, J.-M. Michel and J.-L.Guendon), 87–105, Revue Archéologique de la Narbonnaise supplément 33, Montpellier.

Genty, D., and Quinif, Y., 1996, Annually laminated sequences in the internal structure of some Belgian stalagmites—importance for paleoclimatology, Journal of Sedimentary Research, 66, 275–88.

Genty, D., Baker, A., and Barnes, W. L., 1997, Comparison of annual luminescent and visible laminae in stalagmites,Comptes Rendus de l’Académie des Sciences Paris, Série II, 325, 193–200.

Genty, D., Baker, A., and Vokal, B., 2001a, Inter and intra annual growth rates of European stalagmites, ChemicalGeology, 176, 193–214.

Genty, D., Baker, A., Massault, M., Proctor, C., Gilmour, M., Pons-Branchu, E., and Hamelin, B., 2001b, Dead carbonin stalagmites: carbonate bedrock paleodissolution vs. ageing of soil organic matter. Implications for 13C variations inspeleothems, Geochimica et Cosmochimica Acta, 65, 3443–57.

Genty, D., Blamart, D., Ouahdi, R., Gilmour, M., Baker, A., Jouzel, J., and Van-Exer, S., 2003, Precise dating ofDansgaard–Oeschger climate oscillations in Western Europe from stalagmite data, Nature, 421, 833–8.

Genty, D., Plagnes, V., Causse, C., Cattani, O., Stievenard, M., Falourd, S., Blamart, D., Ouahdi, R., and Van-Exter, S.,2002, Fossil water in large stalagmite voids as a tool for paleoprecipitation stable isotope composition reconstitutionand paleotemperature calculation, Chemical Geology, 184, 83–95.



Guendon, J.-L., and Vaudour, J., 2000, Concrétions et fonctionnement de l’aqueduc, in L’aqueduc de Nîmes et le Pont duGard, archéologie, géosystème et histoire (eds G. Fabre, J.-L. Fiches and J.-L. Paillet), 233–48, éditions CNRS, Paris.

Guendon, J.-L., Huon, S., Parron, C., and Bonté, S., 2002, Les concrétions calcaires, témoins du fonctionnement del’aqueduc, in L’aqueduc romain de Fréjus. Sa description, son histoire et son environnement (eds C. Gébara, J.-M.Michel and J.-L. Guendon), 163–215, Revue Archéologique de la Narbonnaise supplément 33, Montpellier.

Guendon, J.-L., Parron, C., Huon, S., Gébara, C., and Michel, J.-M., 1994, Premiers résultats de l’étude géochimique desconcrétions calcaires de l’aqueduc romain de Fréjus, Bulletin Archéologique de Provence, 23, 66–80.

Hodge, A. T., 1992, Roman aqueducts and water supply, Duckworth, London.Holzhauser, H., 1997, Fluctuations of the Grosser Aletsch Glacier and the Gorner Glacier during the last 3200 years: new

results, in Glacier fluctuations during the Holocene (ed. B. Frenzel), 35–38, Gustav Fisher Verlag, Jena.Holzhauser, H., Magny, M., and Zümbühl, H. J., 2005, Glacier and lake-level variations in west-central Europe over the

last 3500 years, The Holocene, 15, 789–801.Howell, P., 2001, ARAND time series and spectral analysis package for the Macintosh, Brown University, IGBP

PAGES/World Data Center for Paleoclimatology Data Contribution Series 2001-044, NOAA/NGDC Paleoclimatol-ogy Program, Boulder, CO.

Huang, Y. M., and Fairchild, I. J., 2001, Partitioning of Sr2+ and Mg2+ into calcite under karst-analogue experimentalconditions, Geochimica et Cosmochimica Acta, 65, 47–62.

Huang, Y. M., Fairchild, I. J., Borsato, A., Frisia, S., Cassidy, N. J., McDermott, F., and Hawkesworth, C. J., 2001,Seasonal variations in Sr, Mg and P in modern speleothems (Grotta di Ernesto, Italy), Chemical Geology, 175,429–48.

Ihlenfield, C., Norman, M. D., Gagan, M. K., Drysdale, R. N., Maas, R., and Webb, J., 2003, Climatic significance ofseasonal trace element and stable isotope variations in a modern freshwater tufa, Geochimica et Cosmochimica Acta,67, 2341–57.

Jorda, M., and Provansal, M., 1996, Impact de l’anthropisation et du climat sur le détritisme en France du sud-est (Alpesdu Sud et Provence), Bulletin de la Société géologique de France, 167(1), 159–68.

Joseph, C., Gilly, J.-C., and Rodier, C., 2000, Système hydrique de l’aqueduc et genèse des concrétions, in L’aqueduc deNîmes et le Pont du Gard, archéologie, géosystème et histoire (eds G. Fabre, J.-L. Fiches and J.-L. Paillet), 225–31,éditions CNRS, Paris.

Julian, M., and Nicod, J., 1984, Un karst subalpin méditerranéen: la région Audibergue-Mons, Alpes Maritimes et Var,Karstologia, 3, 52–8.

Kano, A., Kambayashi, T., Fujii, H., Matsuoka, J., Sakuma, K., and Ihara, T., 1999, Seasonal variation in water chemistryand hydrological conditions of tufa deposition of Shirokawa, Eihilime Prefecture, southwestern Japan, Journal of theGeological Society of Japan, 105, 289–304.

Lastennet, R., 1994, Rôle de la zone non-saturée dans le fonctionnement des aquifères karstiques. Approche par l’étudephysico-chimique et isotopique du signal d’entrée et des exutoires du massif du Ventoux (Vaucluse), Thèse Doctorat,Avignon, France.

Mann, M. E., and Jones, P. D., 2003, Global surface temperatures over the past two millennia, Geophysical ResearchLetters, 30(15), doi: 10.1029/2003GL017814.

Maréchal, J. C., Ladouche, B., and Dörfliger, N., 2008, Karst flash flooding in a Mediterranean karst, the example ofFontaine de Nîmes, Engineering Geology, 99, 138–46.

Maxwell, J. A., Teesdale, W. J., and Campbell, J. L., 1995, The Guelph PIXE software package II, Nuclear Instrumentsand Methods in Physics Research, B95, 407–21.

McDonald, J., Drysdale, R., Hill, D., Chisari, R., and Wong, H., 2007, The hydrochemical response of cave drip watersto sub-annual and inter annual climate variability, Wombeyan Caves, SE Australia, Chemical Geology, 244, 605–23.

Michel, J.-M., 2002, Le canal: caractéristiques et particularités, in L’aqueduc romain de Fréjus. Sa description, sonhistoire et son environnement (eds C. Gébara, J.-M. Michel and J.-L. Guendon), 122–36, Revue Archéologique de laNarbonnaise supplément 33, Montpellier.

Michel, J.-M., Gébara, C., and Béraud, I., 1994, De Mons à Fréjus, l’aqueduc romain revisité: bilan de cinq années derecherches, Bulletin Archéologique de Provence, 23, 49–65.

Niggemann, S., Mangini, A., Richter, D. K., and Wurth, G., 2003, A paleoclimate record of the last 17,600 years instalagmites from the B7 cave, Sauerland, Germany, Quaternary Science Reviews, 22, 555–67.

Ollivier, V., Guendon, J.-L., Ali, A., Roiron, P., and Ambert, P., 2006, Evolution postglaciaire des environnementstravertineux, provençaux et alpins. Nouveau cadre chronologique, faciès et dynamique morpho-sédimentaire, Qua-ternaire, 17, 51–67.

Pasqualini, M., Excoffon, P., Michel, J.-M., and Botte, E., 2006, Fréjus, Forum Iulii, fouilles de l’espace Mangin, RevueArchéologique de Narbonnaise, 38–9, 283–341.



Pentecost, A., 2005, Travertine, Springer-Verlag, Berlin.Perrette, Y., Poulenard, J., Saber, A.-I., Fanget, B., Guittonneau, S., Ghaleb, B., and Garaudee, S., 2008, Polycyclic

aromatic hydrocarbons in stalagmites: occurrence and use for analyzing past environments, Chemical Geology, 251,67–76.

Petelet-Giraud, E., and Negrel, P., 2007, Geochemical flood deconvolution in a Mediterranean catchment (Hérault,France) by Sr isotopes, major and trace elements, Journal of Hydrology, 337, 224–41.

Pinault, J.-L., Plagnes, V., and Aquilina, L., 2001, Inverse modeling of the hydrological and hydrochemical behavior ofhydrosystems: characterization of karst system functioning, Water Resources Research, 37(8), 2191–204.

Plagnes, V., and Bakalowicz, M., 2002, The protection of a karst water resource from the example of the Larzac karstplateau (south of France): a matter of regulations or a matter of process knowledge? Engineering Geology, 65,107–16.

Polyak, V. J., and Asmeron, Y., 2001, Late Holocene climate and cultural changes in the Southwestern United States,Science, 294, 148–51.

Ramseyer, K., Miano, T. M., D’orazio, V., Wildberger, A., Wagner, T., and Geister, J., 1997, Nature and origin of organicmatter in carbonates from speleothems, marine cements and coral skeletons, Organic Geochemistry, 26, 361–78.

Rousseau, L., Laafar, S., Pèpe, C., and de Lumley, H., 1995, Sterols as biogeochemical markers: results from ensembleE of the stalagmitic floor, Grotte du Lazaret, Nice, France, Quaternary Science Reviews, 14, 51–9.

Salomon, J., Dran, J.-C., Guillou, T., Moignard, B., Pichon, L., Walter, P., and Mathis, F., 2008, Present and future roleof ion beam analysis in the study of cultural heritage materials: the example of the AGLAE facility, NuclearInstruments and Methods in Physics Research, B266, 2273–8.

Salvador, P.-G., Vérot-Bourrely, A., Bravard, J.-P., Franc, O., and Macé, S., 2002, Les crues du Rhône à l’époquegallo-romaine dans la région lyonnaise, in Les paléoenvironnements fluviatiles et lacustres en France depuis 15000ans (eds. J.-P. Bravard and M. Magny), 215–21, Errance, Paris.

Shopov, Y. Y., Ford, D. C., and Schwarz, H. P., 1994, Luminescent microbanding in speleothems—high-resolutionchronology and paleoclimate, Geology, 22, 407–10.

Stumm, W., and Morgan, J. J., 1996, Aquatic chemistry, chemical equilibria and rates in natural waters, 3rd edn, JohnWiley & Sons, Inc., New York.

Tesoriero, A. J., and Pankow, J. F., 1996, Solid solution partitioning of Sr2+, Ba2+ and Cd2+ to calcite, Geochimica etCosmochimica Acta, 60, 1053–63.

Tooth, A. F., and Fairchild, I. J., 2003, Soil and karst aquifer hydrological controls on the geochemical evolution ofspeleothem-forming drip waters, Crag Cave, southwest Ireland, Journal of Hydrology, 273, 51–68.

Toutin-Morin, N., and Crévola, G., 1994, Carte géologique de la France au 1/50000, 1024, Fréjus-Cannes, éditions duBureau de Recherches Géologiques et Minières, Orléans.

Van der Leeuw, S. E., and the ARCHAEOMEDES research team, 2005, Climate, hydrology, land use, and environmentaldegradation in the lower Rhone Valley during the Roman period, Comptes Rendus Géosciences, Paris, 337, 9–27.

Verheyden, S., Keppens, E., Fairchild, I. J., McDermott, D., and Weis, D., 2000, Mg, Sr and Sr isotope geochemistry ofa Belgian Holocene speleothem: implications for paleoclimate reconstructions, Chemical Geology, 169, 131–44.

Wang, Y. J., Cheng, H., Edwards, R. L., He, Y., Kong, X., An, Z., Wu, J., Kelly, M., Dykovski, C. A., and Li, X., 2005,The Holocene Asian Monsoon: links to solar changes and north Atlantic climate, Science, 308, 854–7.

Zhang, P., Cheng, H., Edwards, L., Chen, F., Wang, Y., Yang, X., Liu, J., Tan, M., Wang, X., Lui, J., An, C., Dai, Z., Zhou,J., Zhang, D., Jia, J., Jin, L., and Johnson, K. R., 2008, A test of climate, sun and culture relationships from an1810-year Chinese cave record, Science, 322, 940–2.



highresolution (pixe) analyses of carbonate deposits …€¦ · 13094 aix-en-provence cedex 2,...

Documents