You are on page 1of 26

Received: 15 May 2023 | Revised: 11 August 2023 | Accepted: 19 September 2023

DOI: 10.1002/gea.21980

REVIEW ARTICLE

Writ in water—Unwritten histories obtained from carbonate


deposits in ancient water systems

Gül Sürmelihindi1 | Cees Passchier2

1
School of Archaeology, University of Oxford,
Oxford, UK Abstract
2
Institute for Geosciences, University of Calcium carbonate deposits from ancient water systems such as aqueducts are
Mainz, Mainz, Germany
a hidden archive for archaeology and environmental sciences. These deposits
Correspondence formed wherever carbonate‐rich water was in contact with a water‐containing
Gül Sürmelihindi, School of Archaeology, structure and recorded water composition, temperature, biological content, the
University of Oxford, 2 South Parks Road,
Oxford OX1 3TG, UK. operation or nonoperation of a water system segment, flow discharge and velocity,
Email: surmelih@uni-mainz.de the shape of disappeared segments of water structures, the number of years a water
Scientific editing by Kevin Walsh supply system was active, disruptions of the water supply and water management
such as repairs, adaptations and cleaning. Indirectly, urban development, resilience,
population‐ and socioeconomic dynamics can be studied through the stratigraphy of
carbonate in water systems. Carbonate archives can also give insight into long‐term
changes in paleoclimate and on environmental pollution, deforestation, extreme
floods, droughts, earthquakes and volcanic eruptions. Archaeological and environ-
mental investigations of carbonate deposits can provide data with up to daily
resolution over decades to centuries. Although absolute dating of carbonate from
water systems is still problematic, each study on the aqueduct of an ancient city,
together with its carbonate deposits, provides its own microstory in Roman life.

KEYWORDS
archaeohydrology, carbonate deposits, geoarchaeology, geochemistry, history of water
management, microstratigraphy, Roman aqueduct

1 | INTRODUCTION literature with a total length of at least 5000 km (Figure 1c) (www.
romaq.org). The engineering, construction and modifications of these
Ancient water systems, and especially Roman aqueducts, belong to the structures have been extensively studied from remaining ruins (Ashby,
greatest technological achievements of past civilisations and represent 1935; Hodge, 1992; Kessener, 2000b; Wikander, 2000; Wilson, 1997,
an important cultural heritage of humankind (Bonnin, 1984; 1999, 2000, 1996) but this does not provide information on how many
Hodge, 1992; Mays, 2010; Viollet, 2007). In this review paper, we years an aqueduct transported water; on water discharge, temperature
discuss carbonate deposits from long‐distance urban water supply and composition and its change with time and on maintenance which
systems in the Mediterranean area that served major population reflects on changing local socioeconomy and population dynamics.
centres (Figure 1a,b,d). More than 2400 ancient aqueducts, mostly Such information, and more, can be obtained from the laminated
from the Roman Imperial period, are presently known from the calcium carbonate deposits that formed in ancient water systems from

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
© 2023 The Authors. Geoarchaeology published by Wiley Periodicals LLC.

Geoarchaeology. 2024;39:63–88. wileyonlinelibrary.com/journal/gea | 63


15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
64 | SÜRMELIHINDI and PASSCHIER

F I G U R E 1 (a) Typical masonry underground Roman channel with an inspection shaft, Cologne Eifel aqueduct, Germany. (b) Cross‐section of
the Cologne aqueduct channel, showing carbonate deposits, recognisable by their tapering shape along the channel wall. (c) Map of presently
known larger Roman aqueducts published in www.romaq.org. (d) Masonry channel of the Zaghouan Roman aqueduct of Carthage, Tunisia on top
of an arcade. (e) Cartoon of a typical buried Roman aqueduct channel, with deposited carbonate (yellow). (f) Perforated block from the inverted
siphon of Laodicea, Turkey, which was running full and deposited concentric rings of carbonate. (g) Cartoon of deposition in a full pipe.
(h) Ceramic water pipe from Naxos, Greece that was initially running partly full and later full, covering the entire pipe wall with carbonate.
(i) Cartoon of a partially filled water pipe with carbonate deposition. Photo (a) Klaus Grewe; other photos by Cees Passchier.

carbonate‐loaded water (Figure 1b,f,h). These deposits can act as et al., 2002). (Lime)scale is a common term for these deposits in
archives for archaeology and environmental studies. modern pipes, heat exchangers and other machines.
This review paper focusses on carbonate deposits in Roman water The content of this review paper is based on the literature
systems, notably aqueducts, but conclusions drawn will equally apply to on aqueduct carbonate deposits and our own unpublished results,
water supply systems of other cultures and ages, which we cannot all reflecting our experience studying carbonate from over 100 aqueducts
treat here. Notable examples of other ancient water systems with around the Mediterranean (Figure 1c).
carbonate deposits are the Classical Greek and Hellenistic water supply
systems of Greece and Turkey (Angelakis et al., 2012, 2017; Lolos, 1997;
Öziş, 1994), Nabatean water systems in Jordan (Oleson, 1997, 2018) and 2 | PROPERTIES OF CARBONATE
pre‐Columbian water channels in the Americas (Caran & Neely, 2006; DEP O S I TS I N W A T E R SY S T E M S
Neely, 2001, 2017; Neely et al., 2022; Winsborough et al., 1996).
Several alternative terms have been used to describe carbon- 2.1 | Calcium carbonate deposition
ate deposits in water systems, notably travertine (e.g., Keenan‐
Jones et al., 2022; Neely, 2017; Neely et al., 2022), tufa (e.g., Calcium carbonate deposition in water supply systems depends on the
Keenan‐Jones et al., 2008), sinter and Kalksinter (e.g., Blyth et al., equilibrium of CO2 and Ca2+ dissolved in water1 (Figure 2a) (Andrews
2017; Grewe, 1991; Maier, 1998; Müller, 2000; Pentecost, et al., 2004; Langmuir, 1997, pp. 193−230; Müller, 2000; Pedley &
2005; Schmitz, 1978; Schulz, 1986; Sürmelihindi, Passchier, Rogerson, 2010; Pentecost, 2005, pp. 197–240). CO2 forms carbonic acid
Baykan, et al., 2013; Sürmelihindi, Passchier, Spötl, et al., 2013;
Wilson, 2000) and carbonate concretion or incrustation (e.g., 1
This is a simplified treatment of the process of carbonate deposition; a full and more
Benjelloun et al., 2019; Garczynski et al., 2005; Guendon technical treatment can be found in the quoted textbooks.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 65

F I G U R E 2 (a) The carbonate cycle: each sphere represents water with dissolved carbonic acid (white) and calcium (yellow). The higher the
carbonic acid content of the water, the more calcium it can take up from bedrock. Rainwater (I) penetrates the soil where it will be enriched in
CO2 from roots and decaying plants (II) leading to the dissolution of calcium carbonate or the take‐up of calcium from bedrock formations (III).
When this enriched groundwater is captured to supply an aqueduct at the caput aquae (IV), excess CO2 degasses, leading to supersaturation in
calcium carbonate that will be forced to precipitate in aqueduct channels or pipes (V–VIII). Maximum thickness is usually attained in the central
section of an aqueduct (VI) after which it decreases (VII) until deposition ceases (IX). Deposition may be enhanced in steep sections (VIII, XIII).
Plant growth (X) or evaporation (XI) can locally cause supersaturation and deposition. In pipes, degassing is reduced and the deposition rate
decreases (XIV). Porous carbonate, formed by high biological activity, causes locally thicker carbonate (XII) and is more commonly found in
low‐gradient sections of an aqueduct. (b–d) Variety in carbonate thickness along the 50 km long aqueduct of Nîmes, France. (b) The initial
section of the aqueduct is almost devoid of carbonate except in bends where turbulence triggered some deposition (arrow). (c) Central near the
Pont du Gard bridge with massive deposits. (d) Castellum aquae in Nîmes, where the aqueduct ended and water was distributed, with minor
carbonate deposits in the entrance tunnel (white arrow in inset—white rectangle marks the site in the tunnel). Photos by Cees Passchier.

(H2CO3) in water, which dissociates into H+, HCO3− and CO32− ions, the aquifer is hosted in limestone or marble and can take up calcium ions
lowering the pH. Carbonic acid concentration in rainwater reflects CO2 if it is hosted in other rocks such as volcanic deposits2 (Figure 2a‐III)
pressure in the atmosphere (Figure 2a‐I) but CO2 pressure is orders of (Berner, 1992; Brasier, 2011; Meheruna & Akagi, 2006; Anderkó, 2006).
magnitude higher in soil because of the decay of organic matter and When water thus enriched in calcium ions and carbonic acid exits at a
respiration of plant roots (Hinsinger et al., 2006; Pentecost, 2005). spring (Figure 2a‐IV), it will start to degas CO2 to equilibrate with CO2 in
Rainwater that penetrates the soil takes up this extra CO2 to form
additional carbonic acid (Figure 2a‐II) which decreases the pH. When this 2
Most ancient water supply systems that deposit carbonate are fed by karst springs from
water joins an aquifer in the underlying bedrock, it can dissolve CaCO3 if aquifers hosted in carbonates such as limestone, marble or dolostone.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
66 | SÜRMELIHINDI and PASSCHIER

F I G U R E 3 Aqueduct carbonate deposits in various settings. Aqueducts of (a) Béziers, France; (b) Divona, Cahors, France; (c) Aspendos,
Turkey: note the secondary plaster in the middle of the sequence. The width of samples in (a), (b) and (c) is 31, 35 and 44 cm, respectively. (d)
Aqueduct bridge of Pont de la Lône, Nîmes aqueduct, France: massive leakage deposits cover the bridge, formed by tapping of water for
irrigation in antiquity. (e) Schematic presentation of the gradual filling of an aqueduct channel with carbonate. In general, the height of carbonate
deposited on the walls increases as carbonate accumulates, giving the deposits an upward‐tapering shape. However, this gradual increase in
height is interrupted if there is a temporary decrease in discharge as shown here. Besides deposits formed from water flowing in the aqueduct
(e1), leakage can also form deposits (e3). Carbonate can also form on the channel vault, or on the walls and bottom of an abandoned channel by
percolation of water through the mortar of an aqueduct (e2). Such deposits are known as calthemite and are not discussed in this paper. Photos
by Cees Passchier.

the atmosphere, losing carbonic acid and increasing the pH (Figure 2a‐V). Once equilibrium is reached, carbonate will not precipitate further
CaCO3 will then precipitate on any surface in contact with water in the (Figure 2a‐IX) (Andrews, 2006; Bono et al., 2001) unless the water
form of the minerals calcite or aragonite (Bar‐Matthews et al., 1991; temperature increases (Baatz, 1978; Dreybrodt, 1982); the CO2 level is
Holland et al., 1964; White, 1976).3 brought below that of the atmosphere, for example, if cyanobacteria
Water that exits at springs is usually unsaturated in CaCO3, (Merz, 1992; Spiro & Pentecost, 1991), algae (McConnaughey, 1991) or
(Figure 2a‐IV) and some CO2 should degas for water to become higher plants grow in the water and extract CO2; or if evaporation
supersaturated so that carbonate deposition can start (Sürmelihindi & reduces the water volume (Figure 2a‐X,XI) (Curie & Petit, 2014; Harmon
Passchier, 2013). Deposition therefore usually starts at some distance et al., 1983). Such conditions exist on the outside of channels in leakage
downstream (Figure 2a‐V,b) (Guendon et al., 2002). Carbonate deposition water (Figures 2‐X,XI and 3d) (Sürmelihindi & Passchier, 2013) or in a
rate may then increase downstream to some maximum (Figure 2a‐VI,c), channel if it is an open system or if the vault is damaged (Figure 2a‐X,XI)
after which it decreases (Figure 2a‐VII,d) until carbonic acid in water (Sürmelihindi, Passchier, Baykan, et al., 2013). A porous deposit with
approaches an equilibrium concentration with atmospheric CO2 again. algal filaments and plant imprints is formed in such cases,
similar to ‘tufa’ (Figures 3d, 4e and 5d) (Fairchild & Baker, 2012;
3 Fouke, 2011; Neely, 2017; Pentecost, 2005; Sürmelihindi &
Aragonite is known to deposit in warmer water, and water rich in Mg. It has so far not been
found in aqueduct carbonate but is common in cave deposits with similar composition. Passchier, 2013). Some aqueducts locally show massive leakage (tufa)
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 67

F I G U R E 4 Aqueduct carbonate microfabric as seen in samples from various sites. (a–d) Sparite‐dominated dense carbonate from aqueducts of
(a) Cologne, Germany with fine irregular lamination. (b) Béziers, France with irregular laminae thickness. (c) Fréjus, France from a steep section with
some event horizons. (d) Aspendos aqueduct, Turkey, with dominantly sparitic deposits and bimodal laminae alternation. (e) Micrite‐dominated
porous carbonate from the aqueduct of Saintes, France. Thickness of carbonate: (a) 121 mm, (b) 220 mm, (c) 142 mm, (d) 22 mm,
(e) 33 mm, respectively. Photos by Cees Passchier.

deposits on the outside, even if there is no deposition inside the channel rates can be expected on a steep slope or at other sites with high
(Figure 3d) (Combes et al., 1997; Guendon & Vaudour, 1986, 2000; turbulence, for example, in sharp curves (Figure 2b), drop shafts, steps or
Hauck & Novak, 1987). at constrictions in the channel such as sluice gates (Figure 2a‐VIII, XIII)
(Brinker, 1986; Chen et al., 2004; Grohmann, 1978; Passchier, Sürmeli-
hindi, Spötl, Mertz‐Kraus, et al., 2016; Sürmelihindi et al., 2018, 2023;
2.2 | Carbonate thickness Ventura, 2002; Zhang et al., 2001).
In longer aqueducts with a relatively uniform slope as in Nîmes
The thickness of carbonate that is deposited per time unit on a certain (Guendon & Vaudour, 2000), the maximum carbonate thickness is in
section of a water channel depends on the concentration of calcium and the central part of the aqueduct (Figure 2a,c) and decreases from
carbonic acid dissolved in the transported water, the local precipitation there downstream. However, there are exceptions where the thickest
rate and the density of the precipitated carbonate (Joseph et al., 2000; deposits are found in the first section, for example, in Patara
Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; Sürmelihindi (Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier,
et al., 2018, 2023). Spötl, et al., 2013), Aspendos (Kessener, 2000b, 2016) and Béziers
Calcite will precipitate in a water system if water is locally (Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016) or in the last
supersaturated in CaCO3 (Figure 2a) but the precipitation rate in section, for example, in Aix‐en‐Provence (Claude et al., forthcoming),
g/cm2/day depends mostly on the location in the channel, on discharge, apparently since steep slopes are present there (Figure 2a‐XIII).
on flow speed and turbulence and on biological activity (Claude Long sections of closed pipes, and possibly some parts of
et al., forthcoming; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; closed aqueduct channels (Gilly, 1971; Schmitz, 1978) may have less
Sürmelihindi & Passchier, 2013). A higher flow speed and discharge or no deposition because degassing of CO2 is hampered there
implies a larger volume of water passing per time unit, which will increase (Figure 2a‐XIV). However, this does not apply to shorter pipe sections
the local carbonate deposition rate (Wróblewski et al., 2017). A high (Figure 1f,g) or partly filled pipes where degassing is still possible
degassing rate of CO2 increases supersaturation and contributes to the (Figure 1h,i) (Kessener, 2001, 2016; Sürmelihindi, Passchier, Baykan,
carbonate deposition rate (Ford & Williams, 2007). Such high degassing et al., 2013).
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
68 | SÜRMELIHINDI and PASSCHIER

F I G U R E 5 Microfabric in thin sections. (a) Dominantly sparitic deposit with prominent brown bands, alternating with thin micritic laminae. Barbegal,
France; (b) Alternating sparite, microsparite and micrite. Thickness of micritic laminae increases upwards, probably because discharge decreased.
Valpuentes aqueduct of Córdoba, Spain. (c) Dominantly micritic fabric, sparite crystals isolated. Some of the calcite crystals are very elongated and extend
through most laminae. Prominent brown bands show periodicity. (d) Pure micrite deposit with large pore spaces. Patara, Turkey. Images (c) and (d) are
from steeper and shallower parts of the same aqueduct channel, respectively. (e) Isolated sparite crystals in micrite with regular brown lamination, thought
to represent daily deposition intervals and (f) alternation of micrite (mi) and sparite (sp) with brown bands (bb). Each sparite/micrite couplet is thought to
represent 1 year of deposition Jerash, Jordan. Scale bars (a, b) 5 mm; (c–e) 1 mm; (f) 0.1 mm. Photos by Cees Passchier.

Aqueduct carbonate is commonly visibly laminated, especially in aqueducts. Examples of annual thickness are 1.1 mm/year for the
dense and coarse‐crystalline deposits (Figures 3–5). Lamination inside Traconnade aqueduct of Aix‐en‐Provence (Claude et al., forthcoming);
a channel or pipe is generally parallel to the wall and floor of the 0.2–1.5 mm in Pompeii (Filocamo et al., 2018); 11 mm for Béziers
conduit in which it forms (Figures 1e–i and 3). Laminae commonly get (Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016); 2 mm for Fréjus
thinner towards the top of the deposit in conduits that are partly (Guendon et al., 2002); 1–2 mm for Nîmes (Benjelloun et al., 2019) and
filled with water (Figure 3a,e) and are concentric in pipes that were 2.5 mm for Jerash (Passchier et al., 2021).
running full (Figure 1f,g) (Kessener, 2016). Younger laminae usually
ascend progressively higher on the walls, an effect of mounting water
level when carbonate accumulates, except if the water level 2.3 | Microstratigraphy and microfabric
fluctuates (Figure 3e) (Hodge, 1992; Nöggerath, 1858).
The density and porosity of carbonate play a role in lamina thickness All carbonate deposits are composed of aggregates of crystals.
so that porous aqueduct carbonate can have thicker laminae than dense Aqueduct carbonate in individual laminae can be micritic, with
deposits at the same stratigraphic level (Figures 2a‐XII and 5c,d). The crystals <4 μm in diameter, microsparitic (4–30 μm) and sparitic
thickness of annually deposited aqueduct carbonates can be up to 11 mm (>30 μm) (Figures 5 and 7d) (Frisia, 2015; Frisia et al., 2000;
but is usually around 1 mm in the maximum‐deposition section of larger Pentecost, 2005). If deposits are dominantly porous and micritic,
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 69

lamination is less visible (Figures 4e and 5d) (Durlet et al., 2023; laminae (Figure 5c). The shape of elongate sparite is also described as
Fronteau et al., 2023). ‘columnar’ (Blyth et al., 2017; Brasier, 2011; Frisia, 2015; Frisia
In aqueduct channel sections that are steep and where no daylight et al., 2018; Hostetter et al., 2011).
penetrates, deposits are usually sparitic and dense due to fast water flow Growth competition leads to the formation of bundles of elongate
and reduced biological activity, while gently dipping sections, with slow crystals that fan outwards and create a bulbous surface structure
water flow or sections where light is admitted to the channel, tend to be (Figures 4c and 5a) (González et al., 1992). Under certain flow conditions,
dominated by more micritic and porous deposits (Figures 2a‐X,XI, 4e these can become asymmetric and take a shape similar to ripples
and 5d) (Kessener, 2019; Passchier & Sürmelihindi, 2019; Sürmelihindi, (Keenan‐Jones, 2015; Keenan‐Jones et al., 2014, 2022; Passchier
Passchier, Baykan, et al., 2013; Sürmelihindi et al., 2021). However, et al., 2021).
exceptions exist such as the gently sloping arcade of the Anio Novus Visible lamination in aqueduct carbonate is formed partly by
aqueduct of Rome which was documented to have dense and coarse alternating laminae of different dominant crystal size and shape, and
crystalline carbonate (Carrara & Persia, 2001; Motta, et al., 2015, 2022). partly by bands of inclusions or coloration and organic matter content
Examples of aqueducts with dominantly sparitic carbonate are Patara and (Genty & Quinif, 1996). Commonly, laminae of elongate sparite and
Aspendos (Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, micrite alternate (Figure 5a,b,e), locally with sharp transitions
Passchier, Spötl et al., 2013), Aix‐en‐Provence (Claude et al., forthcomoing), between sparite with acute or rounded crystal terminations and
Béziers (Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016), Hier- overlying micrite (e.g., Passchier et al., 2021, Passchier, Sürmelihindi,
apolis (Scardozzi, 2007, fig. 12) and Jerash (Passchier et al., 2021). Spötl, Mertz‐Kraus et al., 2016; Sürmelihindi, Passchier, Spötl,
Porous micritic deposits found in many aqueducts such as Ephesus et al., 2013). The alternation is attributed to changing growth
(Kessener, 2019; Passchier & Sürmelihindi, 2019), Saintes (Figure 4e), conditions caused by seasonal or other variations in supersaturation,
Poitiers (Durlet et al., 2023), Butrint (Sürmelihindi & Passchier, 2020) and discharge, temperature, the presence of suspended sediment and of
Meaux (Fronteau et al., 2023) seem to derive mostly from the presence of bio‐influenced calcite crystal growth (Filocamo et al., 2018;
biofilms, which decompose after deposition, leaving pore spaces behind Frisia, 2015; Sürmelihindi, Passchier, Baykan, et al., 2013).
(Figures 4e and 5d) (Frisia, 2015; Gradziński et al., 2010; Sürmelihindi, Many aqueduct carbonate deposits contain bundles of fine laminae
Passchier, Baykan, et al., 2013). Such porous micrite may contain small of brown coloration, mostly in sparite crystals, that seem to represent
branching structures known as travertine shrubs (Sivaguru et al., 2022). time horizons of growth with a special water composition (Figure 5a,c,e,f)
Favourable conditions for the development and persistence of biofilms (Keenan‐Jones et al., 2022; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus,
are found in slow‐flowing water, in water with a high detrital load; and et al., 2016; Sivaguru et al., 2022). These brown laminae show high
where light is admitted to the channel (Figures 2‐X,XI, 4e and 5d) fluorescence (Figure 7c) (Frisia et al., 2018; Keenan‐Jones et al., 2014;
(Kessener, 2019; Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmeli- Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; Passchier
hindi et al., 2018, 2021). Cyanobacteria produce extracellular polymeric et al., 2021; Sivaguru et al., 2022) and are probably composed of
substances (EPS) that can promote micrite nucleation (Emeis et al., 1987; included organic matter such as humic and fulvic acids, proteins and
Kandianis et al., 2008; Pedley et al., 2009; Riding, 2000; Rogerson colloids (Nöggerath, 1858; Perrette et al., 2005; Shopov et al., 1994;
et al., 2008). Sivaguru et al., 2022). The finest, regular brown lamination (Figure 5e)
Deposition of carbonate in aqueduct channels can be a dominantly may represent a daily lamination (Passchier, Sürmelihindi, Spötl, Mertz‐
chemical process in fast‐flowing turbulent water where biofilms Kraus, et al., 2016; Passchier et al., 2021; Schulz, 1986).
cannot grow (Gradziński, 2010; Pedley, 1992), in which case, large Lamination can be irregular (Figure 4a–c), or regular and cyclic
elongate sparite crystals of calcite can form; or biofilm‐dominated (Figure 4d). Cyclic, regular lamination is most common in carbonate from
producing more porous and micritic fabrics (Frisia, 2015; González the eastern Mediterranean (Figures 4d and 5e), probably because of the
et al., 1992; Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, strongly bimodal climate there, with rainfall almost exclusively in
Passchier, Spötl, et al., 2013). Pipes usually contain solid deposits because winter and dry summers (Carrara & Persia, 2001; Passchier et al., 2021;
biofilm growth is hampered (Lieber, 1990a; Sürmelihindi, Passchier, Spötl, Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier,
et al., 2013). In some flowstones in caves, micrite can also form in the Spötl, et al., 2013). In the western Mediterranean and Northwestern
absence of biofilms, possibly nucleating on fine‐grained sediment particles Europe, aqueduct carbonate tends to show less regular lamination
(Wróblewski et al., 2017). The same mechanism may also apply to some (Figures 3a,b and 4a–c) (Passchier, Sürmelihindi, Spötl, Mertz‐Kraus,
dense micrite in aqueduct deposits. et al., 2016; Sürmelihindi et al., 2023).
Sparitic crystals tend to be elongate normal to the lamination As an alternative explanation, elongate, columnar sparite in
(Blanc, 2000; Botturi & Parecinni, 1991; Rodier et al., 2000) and are aqueduct carbonate, as described above, has been suggested to be
inferred to start as small crystals which grow upwards and increase in the result of ‘diagenesis’, postdepositional secondary crystal
width by growth competition with the neighbouring crystals, usually growth over an initially micritic fabric4 (Sivaguru et al., 2022).
because of a favourable crystallographic orientation (Figures 5e
and 7d) (Frisia, 2015; González et al., 1992; Sürmelihindi, Passchier,
4
This process of postdepositional crystal growth, where new crystals grow over older,
Baykan, et al., 2013; Urai et al., 1991; Wroblewski et al., 2017). Some
usually smaller crystals, replacing them, is known in materials science and geology as
of these large, elongated crystals extend through many overlaying ‘recrystallisation’.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
70 | SÜRMELIHINDI and PASSCHIER

However, some fabrics such as alternating sparite and micrite Williams, 2007; Pentecost, 2005). Stable isotope studies of O and C
horizons with sharp boundaries, sparite crystals that exclusively in CaCO3 are particularly useful for aqueduct carbonate.5 Stable
grew upwards in the stratigraphy and preserved delicate inclu- oxygen isotope composition of aqueduct carbonate, expressed as
sions, brown banding and chemical profiles seem hard to reconcile δ18O, depends on the isotopic composition and temperature of the
with diagenesis (Figure 5) (Passchier, Sürmelihindi, Spötl, Mertz‐ source water and is further mainly affected by evaporation and by
Kraus, et al., 2016; Passchier et al., 2021; see also Brasier et al. seasonal temperature changes of water in the aqueduct channel
[2011], Frisia [2015], Frisia et al. [2018], Melim and Spilde (Figure 6a) (Fairchild & Baker, 2012; Sürmelihindi, Passchier, Baykan,
[2011, 2021]). Also, similar fabrics with alternating elongate et al., 2013; Sürmelihindi, Passchier, Spötl, et al., 2013; Sürmelihindi
sparite and micrite were observed in modern waterpipes and in et al., 2021, 2023).
flowstone in caves, grown in monitoring experiments where no δ18O of rainwater varies with the seasons (Figure 6a) but this
diagenesis was reported (Wróblewski et al., 2017). Most authors variation is commonly depolarised in the soil and aquifer where water
on aqueduct carbonate have therefore interpreted elongate is retained and mixed, so that δ18O and temperature of spring water
columnar sparite as dominantly formed by primary growth (e.g., do not show a regular seasonal change6: δ18O merely reflects the
Benjelloun et al., 2019, 2023; Carlut et al., 2009; Claude mean rainwater composition above the aquifer (Figure 6c) (Ford &
et al., forthcoming; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, Williams, 2007; Williams & Fowler, 2002).
et al., 2016; Sürmelihindi, Passchier, Spötl, et al., 2013). However, As long‐distance water conveying systems, aqueducts are
the subject should be further investigated. Local diagenesis may exposed to annual temperature variations in the soil in which they
occur in some fine‐crystalline deposits (Sivaguru et al., 2022) and are buried. Water temperature in the channel will therefore vary
should not be ignored. So far, no change in the distribution of with the seasons, warmer in summer and cooler in winter,
stable isotopes and trace element patterns on a microscopic scale especially far downstream from the source (Figure 6a)
has been observed in relation to diagenesis (Sivaguru et al., 2022). (Baatz, 1978; Sürmelihindi et al., 2021, 2023; Sürmelihindi,
Some fabric elements seem to be asymmetric in cross sections Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier, Spötl,
parallel to the flow direction in the channel, for example, oblique et al., 2013). When carbonate is deposited from aqueduct water,
‘foliation’ structures in micrite (Sürmelihindi, Passchier, Spötl, its δ 18 O will be higher than that of the water, but the increase is
et al., 2013, fig. 3c), or asymmetric bundles of sparite crystals strongly temperature‐dependent. This effect, known as
(Passchier et al., 2021, fig. 4), some forming ripple‐shapes (Keenan‐ ‘temperature‐dependent kinetic isotope fractionation’ will gener-
Jones et al., 2022; Motta et al., 2017). ate a variation in δ 18 O in carbonate with the seasons, giving high
Microfabric analysis of carbonate can be done using an optical values in winter and low values in summer (Figure 6a) (Clark &
microscope for the identification of mineral content and microfabric, Fritz, 1997; Ford & Williams, 2007; Passchier et al., 2021;
for laminae counting and for identifying microfossils such as diatoms Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; Sürme-
(Figure 5) (Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; lihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier,
Sürmelihindi & Passchier, 2020). More detailed work at higher Spötl, et al., 2013). Evaporation, if present, will be strongest in
magnification is possible by various types of high‐resolution and summer and will have the opposite effect but is thought to play a
specialised optical microscopy (Sivaguru et al., 2022) and by scanning minor role in aqueducts since they are covered structures, closed
electron microscopy (SEM) for the characterisation of the smaller to avoid evaporation (Filocamo et al., 2018; Hodge, 1992;
microfabric elements and determination of minerals, notably the Sürmelihindi, Passchier, Baykan, et al., 2013). δ18 O in carbonate
presence of aragonite. Electron backscatter diffraction (EBSD) is used therefore depends mainly on the isotopic composition
to determine crystal orientation (Figure 7d). Three‐dimensional (3D) of the source water, on temperature‐dependent kinetic isotope
microcomputed tomography can help to see 3D details inside fractionation processes between water and carbonate, and
samples (Sivaguru et al., 2022). Localisation of organic compounds possibly on evaporation (Figure 6a) (Durlet et al., 2023; Sürme-
is done by examining fluorescence properties of samples through lihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier,
techniques such as spectrofluorescence or synchrotron‐radiation Spötl, et al., 2013).
based micro X‐ray fluorescence (SR‐μXRF) (Figure 7c) (Frisia
et al., 2018; Perrette et al., 2005).
5
Stable isotopes are measured from powders produced by micromilling parallel 0.2‐mm wide
stripes of carbonate and measurement of the produced powders in an isotope ratio mass
spectrometer. The ratio of rare heavy to common light isotopes (18O and 16O for oxygen, 13C
2.4 | O and C stable isotope geochemistry
and 12C for carbon, respectively) is expressed as δ18O for oxygen and δ13C for carbon by the
 Rsample − Rstandard 
equation δ (rare isotope) =   × 1000 where R = abundance of rare isotope .
Carbonate depositing in an aqueduct reflects the chemical and isotopic  Rstandard  abundance of common isotope

δ Values for carbonate are expressed in ‰ VPDB (Vienna PeeDee Belemnite). This refers to
composition of the water from which it forms, although the actual marine Belemnite fossils from the Cretaceous PeeDee formation in Southern Carolina, USA,
concentration of ions included in carbonate depends on fractionation which is used as a standard. For water, δ values are expressed in ‰ VSMOW (Vienna
Standard Mean Ocean Water).
processes, which in turn depend on temperature, degassing and in 6
Annual variation in source water composition and temperature can be checked by
some cases, biological activity (Clark & Fritz, 1997; Ford & monitoring water of the original aqueduct source.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 71

F I G U R E 6 Stable isotopes. (a) Schematic diagram of factors affecting δ18O and δ13C in aqueduct carbonate. The initial values and the
changes during mixing, equilibration and fractionation are shown. Values of δ18O in aqueduct carbonate depend mainly on values in the source
water and temperature‐dependent kinetic fractionation, which is stronger at lower temperatures, causing a seasonal variation as shown. For
carbon, values of δ13C in aqueduct carbonate depend mainly on the mixture of carbon from plant matter, with very low δ13C value and that of
carbon in bedrock, with high values. Besides this effect, δ13C values depend on the CO2 degassing rate, where rapid degassing causes an
increase in δ13C. The combined effect can again give a seasonal cyclic signal, which will be opposite that of δ18O in the eastern Mediterranean
with dry summers and wet winters. Further explanation in the text. (b) Selected stable isotope profiles for aqueduct carbonate throughout the
Mediterranean. Cyclic δ18O signals (red) are common in nearly all aqueducts, except when close to the spring (Aqua Marcia). δ13C signals (blue)
are more variable, from antithetic to δ18O to more complex patterns. Profiles of Aspendos, Constantinople and Cahors were published in
Sürmelihindi, Passchier, Spötl, et al. (2013) and Sürmelihindi et al. (2021, 2023), respectively. (c) Range of variation in stable isotopes observed
for 64 Roman aqueducts studied by the authors (unpublished data). VPDB, Vienna PeeDee Belemnite.

Carbon in rainwater is mostly present in dissolved carbonic acid degassing of CO2 and biological activity such as algal growth in an
(H2CO3), which originally derives from CO2 (Figure 6a). Carbon in aqueduct channel will increase δ13C since the lighter carbon isotope is
bedrock, especially in limestone, usually has higher δ C values than
13
preferentially included in escaping CO2 and taken up by living
carbon in atmospheric CO2, while plant matter has much lower values organisms (Dreybrot, 2019; Ford & Williams, 2007; Shiraishi
12
since plants preferentially use the lighter C carbon isotope et al., 2008). Since the carbon reservoir of water is small compared
(Figure 6a) (Fairchild & Baker, 2012; Ford & Williams, 2007). δ13C of to that of oxygen, this has a significant effect. Temperature‐dependent
carbon dissolved in aquifer water is therefore strongly affected by the kinetic fractionation of carbon isotopes is less important than that of
contact of water with either plant matter in the soil or bedrock. In oxygen isotopes, so δ13C in aqueduct carbonate depends mainly on
winter, when rainwater percolates through the soil, δ C entering the
13
the effects of soil and bedrock and on degassing and biological activity
aquifer will be low, and high precipitation may cause water to be (Figure 6a) (Ford & Williams, 2007; Genty et al., 2001). Over a
flushed rapidly through the conduit network of the aquifer to the multidecadal period, δ13C in speleothems can record changes in
springs, restricting contact with the bedrock (Ford & Williams, 2007; vegetation (Fairchild & Baker, 2012, section 8.4) and this has also been
Goldschneider, 2015; White, 2015). Such spring water may have low suggested for aqueduct carbonate (Benjelloun et al., 2019). Since the
δ13C (Frisia et al., 2012). In the summer or dry season, little soil‐ interaction with bedrock, degassing and biological activity are all
affected water with low δ13C will enter the aquifer, and spring water highest at low water levels in an aqueduct channel in summer and
may contain a large component of water from the fracture network in lowest at high water levels in winter, δ13C can also show seasonal
the epikarst or rock matrix. Such water has been in contact with cyclicity but opposite to that of δ18O (Figure 6b) (Passchier, Sürmeli-
bedrock for a longer time and has taken up the heavier carbon from hindi, Spötl, 2016; Sürmelihindi & Passchier, 2013; Sürmelihindi,
the bedrock and therefore has a high δ13C (Figure 6a) (Shopov Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier, Spötl,
et al., 1994). Hence, seasonal water availability driven by bimodal et al., 2013). This antithetic correlation of both isotopes is best
changes in the amount of rainfall impacts the movement of water from developed in the eastern Mediterranean (Figure 6b). Carbonate
the aquifer to the springs, and δ13C of spring water can vary strongly samples taken in downstream locations of aqueducts commonly show
with the seasons (Ford & Williams, 2007). Besides this effect, the best cyclical δ18O and δ13C pattern due to increasing seasonal
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
72 | SÜRMELIHINDI and PASSCHIER

variations in temperature and degassing rate along the channel 2.5 | Trace elements and organic compounds
(Benjelloun et al., 2019; Claude et al., forthcoming; Passchier, Sürme-
lihindi, Spötl, 2016; Sürmelihindi & Passchier, 2020, Sürmelihindi The concentration of trace elements in aqueduct carbonate depends on
et al., 2021; Sürmelihindi & Passchier, 2013; Sürmelihindi, Passchier, their concentration in the original source water and their partitioning
Baykan, et al., 2013; Sürmelihindi, Passchier, Spötl, et al., 2013). into the calcite crystal lattice (Figure 7a,b) (Ford & Williams, 2007;
In aqueduct carbonate, δ18O and δ13C values usually lie in a Passchier et al., 2021). Trace element analysis can be useful to obtain
relatively narrow range (Figure 6c). δ O is mostly in the range of
18
fine‐tuned information on changes in the water composition of the
−4‰ to −9‰ Vienna PeeDee Belemnite (VPDB) (Figure 6c) but source and its possible causes (Fairchild & Baker, 2012, section 8.3;
tends to be higher in the western Mediterranean (Benjelloun Frisia et al., 2018). For example, the Mg concentration of water reflects
et al., 2019; Claude et al., forthcoming; Durlet et al., 2023) and more water temperature (Section 3.6) and the residence time of water in
negative to the eastern Mediterranean and to higher altitude carbonate aquifers (Fairchild & Treble, 2009; Jones, 2010; Passchier
(Passchier et al., 2021; Sürmelihindi, Passchier, Spötl, et al., 2013). et al., 2021; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016).
Evaporation in the aqueduct can also lead to higher values. Hot Similarly, phosphorous (P) is freed in the soil by the decay of organic
springs and hot baths will produce low δ O values in carbonate
18
matter and accumulates in the dry season, so that high concentrations
(Curie & Petit, 2014; Curie et al., 2018). δ13C is more variable because of P in water can coincide with the onset of the first rains of autumn and
of the variable influence of soil and bedrock in the aquifer. In single rainstorms (Figure 7a,b). P‐enrichment can also be associated with
aqueducts with karst springs, it usually varies between −6‰ and enhanced biological activity (Frisia et al., 2012; Jones, 2010). Lantha-
−12‰ VPDB, but if the aquifer is hosted in volcanic rocks or lacks nides and yttrium (Y), which are bound to colloids in the soil, peak in
organic input, δ13C values can be higher and even positive (Figure 6c) aqueduct water when they are washed out of the soil by major
(Schroyen et al., 2000). rainstorms and are therefore also important proxies to use for

F I G U R E 7 Results of geochemical and other techniques. (a, b) Distribution of selected trace elements in a profile of carbonate from the
aqueducts of (a) Béziers, France and (b) Jerash, Jordan. Trace elements are cross‐correlated with the stable isotope patterns. Interpretation given
in the text. Trace elements measured by LA‐ICPMS. (c) Spectrofluorescence results for the aqueduct of Arles, France. High values represent
enrichment in organic matter. (d) EBSD map of micrite (left) and sparite (right) for the aqueduct of Aspendos, Turkey. Width of view 1.3 mm.
EBSD, electron backscatter diffraction; LA‐ICPMS, laser ablation inductively coupled plasma mass spectrometry.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 73

environmental studies (Figure 7a) (Fairchild & Baker, 2012, section 1.4.2; analysis as described above. Even in aqueducts with dominant
Frisia et al., 2018; Passchier et al., 2021; Passchier, Sürmelihindi, Spötl, porous, micritic deposits, such material can commonly be found and
Mertz‐Kraus, et al., 2016). The concentration of other trace elements sampled at sites of enhanced turbulence (e.g., Passchier et al., 2013;
such as U, Sr, Ba and Rb will also vary in source water with time and Sürmelihindi, Passchier, Baykan, et al., 2013). Detailed instructions on
therefore in deposited calcite (Claude et al., forthcoming; Passchier sampling are given in the Supporting Information: File S1.
et al., 2021; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016).
Besides dissolved trace elements, water can contain a fine, dispersed
clay fraction that can be included in carbonate and which is revealed by 3 | APPLICATION
enhanced values of K, Mg, Fe, Si, Al and Th (Figure 7a) (Passchier
et al., 2021; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; On 7 October 1520, German painter Albrecht Dürer wrote ‘In
Sürmelihindi et al., 2021; Wróblewski et al., 2017). Finally, dissolved Aachen I saw … columns with their nice capitals of porphyry … and of
organic matter such as humic and fulvic acids, proteins and colloids will be gossenstein (aqueduct carbonate)’.7 After this early fortuitus descrip-
caught in carbonate (Section 2.3) and their concentration in source water tion and early work in the 19th (Eick, 1867; Nöggerath, 1858) and
can change with the seasons (Huang et al., 2001; Passchier et al., 2021; beginning of the 20th century (Ashby, 1935; Esperandieu, 1926),
Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016). the study of aqueduct carbonate deposits has rapidly evolved over
The electron microprobe (EPMA) can be used to determine the last 50 years. The first modern analytical work was done on the
major element concentrations and to produce element aqueducts of Nîmes in France (Adolphe, 1973; Gilly, 1971) and
distribution maps, but this method can only analyse a limited number Cologne in Germany (Baatz, 1978; Brinker, 1986; Grewe, 1982, 1991;
of elements (Sürmelihindi et al., 2021; Sürmelihindi, Passchier, Spötl, Grohmann, 1978; Schmitz, 1978; Schulz, 1986) while subsequent
et al., 2013). Higher resolution analyses, with more elements, can be work focussed on aqueducts in France, Italy and Turkey (The full list
carried out using techniques such as laser ablation inductively of investigated aqueduct sites is given in Supporting Information:
coupled plasma mass spectrometry (LA‐ ICPMS; Figure 7a,b), X‐ray File S2). This section describes thematically what subjects have been
spectroscopy (scanning transmission electron microscopy [STEM]/ and can be studied with aqueduct carbonate.
energy dispersive X‐ray spectroscopy [EDS]), SR‐μXRF (Carlut
et al., 2009; Frisia et al., 2012, 2005, 2018; Passchier et al., 2021;
Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016; Sürmelihindi, 3.1 | Operation of water systems
Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier, Spötl,
et al., 2013) and laser‐induced breakdown spectroscopy (LIBS; Caceres The presence of carbonate deposits in a water system can be taken
et al., 2017; Harmon & Senesi, 2021; Marín‐Roldán et al., 2014). as evidence that it was in use where this is in doubt (Figures 1 and 2)
(Hodge, 1992; Sürmelihindi & Passchier, 2013, 2014) as in the case
of the Lincoln aqueduct, UK (Vince, 2006). However, absence of
2.6 | Lateral changes evidence is no evidence of absence: water systems that transport
carbonate‐unsaturated water, for example, rainwater, will not
At any site along an ancient aqueduct, carbonate microfabric and deposit carbonate. In fact, unsaturated water may etch the mortar
composition in terms of crystal size and shape, δ18O and δ13C, trace lining8 of a water structure in a characteristic way, with a smooth
elements, clay and organic matter will vary through the carbonate surface above the waterline, equally proving a functioning system.
stratigraphy and form a visible alternation of laminae, brown bands Aqueducts fed by rivers and lakes, or from sources in a geological
and a chemical stratigraphy. Most important are repetitive, cyclical setting without carbonate or volcanic rocks will in most cases not
patterns that are common in many carbonate deposits and differ form any deposits.
from one site to another. Most long‐distance Roman aqueducts, however, deliberately
Most of the stratigraphic aspects mentioned will change along used karst springs: these are large and usually perennial and will have
the length of a water supply system, even in a single layer of the same water even in dry summers (Fiorillo, 2014; Ford & Williams, 2007). In
age (Gilly, 1986). These lateral changes in carbonate stratigraphy are some cases where a choice existed, karst sources may have been
caused by depletion due to prior calcite precipitation (PCP) in the selected since carbonate deposition would prevent corrosion of the
channel (Fairchild & Baker, 2012; Frisia, 2015), by changes in water waterproof mortar lining in long water channels. The Roman
temperature and composition downstream and by changes in flow aqueduct of Cologne tapped two small limestone outcrops in a slate
speed due to changes in channel dimensions and slope. belt, where many other water sources with soft, corroding water

7
What Dürer saw according to his diary for this day in the cathedral of Aachen, Germany,
2.7 | Optimal samples for analysis were spolia of carbonate taken from the nearby Roman aqueduct of Cologne, fashioned into
columns because of their layered, marble‐like quality. Gossenstein was a local name for this
stone.
Dense, coarse‐crystalline, sparite‐dominated deposits with clearly 8
Also known as cocciopesto. In many publications, including our own, this red plaster is
visible layering are best suited for geochemical and microfabric referred to as Opus signinum.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
74 | SÜRMELIHINDI and PASSCHIER

F I G U R E 8 Carbonate casts. (a, b) Casts formed on the wood of the Barbegal watermills. (a) Cast of a millrun flume with remains of
carbonate‐encrusted debris at the bottom and a hook shape profile of the sidewall deposits, indicating overflow (inset). (b) Cast of the spoke of a
waterwheel, Barbegal. (c) Carbonate filling of the Değirmendere aqueduct of Ephesos, Turkey. The carbonate filling remains, showing the shape
and highest water level, hence the minimum height of the channel, while the masonry walls have fallen away. (d) Impression of the waterwheel
of Venafro, Italy. Photos by Cees Passchier.

would have been closer at hand (Grewe, 1986, 1991). On the other Augst (Fellmann, 1988, p. 123) and Windisch (Hagendorn, 2003,
hand, the long aqueducts of Lyon, which contain several inverted pp. 141, 156; Maier, 1998, p. 79). Imprints of the wood of gutters and
siphons made of lead pipes that could easily have been clogged by mill wheels were discovered in the Roman Barbegal watermills in
carbonate, used carbonate‐unsaturated water sources (Hodge, 1981). France (Figures 8a,b and 9a) (Benoit, 1940; Sürmelihindi
Minor carbonate deposition is not a disadvantage in aqueduct et al., 2018, 2019), while in Venafro, Italy, a complete wooden
channels and has no health consequences and it is conceivable that Roman waterwheel was encased in carbonate, preserving its shape
Roman engineers were aware of this option in the choice of available (Figure 8d) (Jacono, 1938). Carbonate can also preserve the imprint
water sources. of masonry or ashlar walls that have fallen away, for example, in
In the dry part of water systems built from concrete, penetrating Sikyon (Lolos et al., 2018), Perge, Hierapolis (Scardozzi, 2007) and
groundwater can deposit a white, translucent carbonate on the Ephesos (Figure 8c) (Passchier & Sürmelihindi, 2019). Finally, in the
interior of the system in the form of crusts, stalactites and stalagmites caliphal palace of Medina Azahara, Córdoba, loose carbonate blocks
known as calthemite (Figure 3e‐2) (Fabretti, 1690, p. 413; preserve the shape of large diameter lead pipes that were recycled9
Smith, 2016). Calthemite should not be confused with carbonate (Passchier & Sürmelihindi, 2023, fig. 11).
deposited from aqueduct water. Aqueduct carbonate is darker in
colour and less dense and translucent (Figure 4); it forms homoge-
neous lamination along the floor and walls with a horizontal upper 3.3 | Function of special structures
limit and is usually tapering upward in thickness (Figure 3e).
The geometry of carbonate incrustations in water structures can give
detailed information on their function. This has been applied to the
3.2 | Shape of disappeared segments of water function of inverted siphons and other water pipes
systems (Kessener, 2001, 2016, 2022); of sluice gates (Kessener, 1995, 2000a)
and of special water gutters of watermills (Passchier et al., 2020;
Some parts of water systems were built of wood, such as the Sürmelihindi et al., 2020).
channels of some smaller aqueducts, water pipes and the gutters and
wheels of watermills. Carbonate deposited on such parts may
preserve the shape of the original structure long after the wood 9
The outside of the blocks is smooth and trace element analysis of the first layer of
has decomposed. Casts of Roman wooden waterpipes were found in carbonate shows strong enrichment in lead.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 75

F I G U R E 9 Repairs, Reconstruction, modification and cleaning. (a‐c) Reconstruction of the replacement of wood segments in the Roman
watermills of Barbegal, France. (a) Fragments B and C grew in wooden millrun flumes that fed the waterwheels, while Fragment A grew most
likely on a waterwheel, based on its shape and stratigraphy with uneven layering. Fragment B has a thick carbonate stratigraphy, because it has
the oldest laminae of all, while Fragment A, from the wheel, starts later, at Level II. Possibly, the wheel, or part of the wheel was replaced at this
time, leaving the flume of Fragment B undisturbed. Carbonate from another millrun flume C started growing even later (Level III), at the same
time as cleaning was done to carbonate on the wheel Fragment A. (b) Possible placement of the fragments. (c) Reconstructed history. (d) Traces
of cleaning in the deposits of the Roman aqueduct of Cahors, France. Arrows indicate tool marks on an erosion surface where carbonate was
manually removed by a maintenance crew. (e) Traces of cleaning in a water gutter of Ephesos, Turkey. The gutter was cleaned of carbonate, but
an edge along the corner was left in place and was later covered with new deposits, creating an unconformity. (f) Final section of carbonate from
flume C in (a), showing a sudden change from brown sparite to white micrite, with an increase in porosity and with plant remains, inferred to
reflect a sudden admittance of light due to damage or modification of the roof structure of a watermill (Barbegal, France).

3.4 | Water level, flow velocity and discharge there is a margin of error since (1) most deposition takes place in deeper
parts of the channel and the upper boundary of deposition may be
Flow velocity (m/s) and discharge (m3/s) of water in an aqueduct can be difficult to determine (Figure 1b,e) (Guendon et al., 2002; Keenan‐
calculated from the shape and size of the channel, the slope, the water Jones, 2015); (2) as there is constant carbonate deposition and
depth and a roughness factor, using the empiric Manning narrowing of a channel, water rises to ever higher levels (Figure 3e);
or Chézy formulas (e.g., Akan, 2006; Chanson, 2004; Chow, 1959, (3) water level may have varied with time (Figure 3e) (Bobée et al., 2011;
pp. 9–19; Hodge, 1992). In aqueducts, the upper limit of carbonate Guendon et al., 2002, fig. 125) and (4) carbonate stratigraphy may have
deposits on the wall is usually taken to estimate water depth and been removed during aqueduct maintenance (Section 3.10) (Passchier
calculate flow velocity (Figure 3e) (Bailhache, 1979, 1983; Brinker, 1986; et al., 2015; Rigal et al., 2023; Sürmelihindi et al., 2023). It is advisable to
Christensen, 1988; Guendon & Vaudour, 1986; Hauck & Novak, 1987; carry out velocity and discharge calculations at various sites along an
Hodge, 1992; Keenan‐Jones, 2015; Kessener, 2019; Lockett, 1988; aqueduct line and to investigate the carbonate profile: without the
Sürmelihindi, Passchier, Spötl, et al., 2013; Verrez, 1985). However, addition of water sources or diversions, discharge estimates and
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
76 | SÜRMELIHINDI and PASSCHIER

carbonate profile should be similar along a channel. In the Nîmes cases, the crystalline nature of carbonate in the structures can give
aqueduct, the profile changes dramatically over a short distance. This some indication since carbonate fabric is strongly affected by the
was attributed to a fall in water level in the downstream part of the presence or absence of daylight (Sürmelihindi et al., 2018, 2019).
aqueduct, possibly due to leakage after an earthquake, or the use of a Observations in several aqueducts suggest that the porosity of
water‐lifting machine in the channel for agricultural activity (Benjelloun aqueduct carbonate depends not only on flow speed but also on
et al., 2019; Carbon et al., 2005; Volant et al., 2009). Other types of biological activity and indirectly on daylight for photosynthesis
legal/illegal diversion of water are also suggested and can have a similar (Section 2.3) (Mahapatra et al., 2015; Sürmelihindi &
effect (Guendon & Vaudour, 1986, 2000; Hauck & Novak, 1987). Passchier, 2020, 2013, 2014). A sudden change in the fabric can
Most aqueduct channels show a gradually upward tapering indicate a change in flow speed (Sürmelihindi, Passchier, Baykan,
profile of carbonate deposits (Figures 1b,e and 3e) but other shapes et al., 2013) or in light conditions, for example, by breaching of an
such as ridges, trapezoidal shapes and thin or absent deposits at the aqueduct vault (Figure 9f) (Sürmelihindi et al., 2018, 2019, 2023).
bottom are also observed (Figure 2d) (Filocamo et al., 2018; Guendon The Patara aqueduct shows an alternation of porous, micritic and
& Vaudour, 1986; Müller, 2000; Sürmelihindi et al., 2023; Supporting dense, sparitic carbonate along its length, linked to the nature of the
Information: File S1). Such shapes can be attributed to complex flow aqueduct channel and the slope: steep sections and closed pipes have
conditions near the aqueduct walls and poor water mixing (Grewe & dense, sparitic deposits while gentle slopes and masonry channels
Blackman, 2001; Müller, 2000); to preferential cleaning (Sürmelihindi with cover stones, that probably let in some light, have porous
et al., 2023); or to the presence of clastic sediment along the bottom micritic carbonate (Sürmelihindi, Passchier, Baykan, et al., 2013). In
of the channel on which carbonate could accumulate, to be the Barbegal mills, a sharp transition between dense, coarse‐
transported downstream (Ashby, 1935, p. 226; Sivaguru et al., 2022). crystalline sparitic deposit and overlying white, porous micritic
In carbonate casts from wooden water gutters, a tapering or hook‐ carbonate with imprints of plant remains (Figure 9f), is attributed to
shaped profile indicates whether the gutter was partly full or overflowing the removal or collapse of a roofed structure over the millrun flumes
(Figure 8a) (Sürmelihindi et al., 2018, 2019). If ceramic pipes were running and basins, admitting daylight and enhancing evaporation, degassing
full from the start, deposits will follow the shape of the pipe, though and biological activity (Sürmelihindi et al., 2018, pp. 1049, 1050). In
bottom deposits may be thicker (Figure 1f,g) (Kessener, 2001, 2022). In Patara, δ13C values increase significantly in open sections of the
partly filled pipes, deposits are initially crescent‐moon shaped (Figure 1i) aqueduct, probably due to increased degassing by comparison to pipe
but may fill the pipe completely as carbonate accumulates (Figure 1h) samples (Sürmelihindi, Passchier, Baykan, et al., 2013, fig. 8).
(Kessener, 2001; Müller, 2000; Lieber, 1990a, 1990b).
Seismic activity or landslides10 can change the slope of aqueduct
channels after they have stopped functioning (Section 3.11), in which 3.6 | Water temperature and composition
case flow speed and discharge calculations are no longer possible.
However, an independent tool to calculate discharge and flow speed The stable isotope and trace element composition of aqueduct
in aqueducts exists in the form of ripple shapes in the fabric (Keenan‐ carbonate provides information on the composition and temperature
Jones et al., 2022; Motta et al., 2017). of the original source water and on any changes in water chemistry
Carbonate microfabric is also dependent on flow speed and temperature that developed on the way from the water source to
(Section 2.3) (Keenan‐Jones et al., 2022; Passchier et al., 2021; the castellum aquae (Figures 6 and 7) (Benjelloun et al., 2019;
Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier, Bobée, 2002; Bobée et al., 2011; Carlut et al., 2009; Gilly, 1986, 1990;
et al., 2013). The Değirmendere aqueduct of Ephesos has a relatively Guendon et al., 2002; Guendon & Vaudour, 2000; Hauck &
gentle slope and dominant porous micritic carbonate throughout, but Novak, 1988; Sürmelihindi, 2013; Sürmelihindi, Passchier, Baykan,
dense carbonate is found in the narrow entrance of a basin where et al., 2013; Sürmelihindi, Passchier, Spötl, et al., 2013). This is a
flow velocity was enhanced (Kessener, 2019; Passchier et al., 2013; particularly important source of data if the original water sources
Sürmelihindi et al., 2019). have dried out (Passchier et al., 2021).
Based on the assumption that temperature‐dependent fraction-
ation in aqueduct water is the most important factor determining
3.5 | Exposure to light—Open or closed nature δ18O fluctuations in aqueduct carbonate (Section 2.4) (Benjelloun
of water structures et al., 2019; Claude et al., forthcoming; Passchier et al., 2021;
Passchier, Sürmelihindi, Spötl, 2016; Sürmelihindi, Passchier, Baykan,
Since the upper part of water structures is commonly ruined, it can be et al., 2013; Sürmelihindi, Passchier, Spötl, et al., 2013; Sürmelihindi
hard to determine if they were originally covered or open. In such et al., 2021, 2023), long‐term trends may indicate climatic changes
(Dubar, 2006a, 2006b; Garczynski et al., 2005). Absolute water
temperature in antiquity can be reconstructed from δ18O in aqueduct
10
Landslides can be self‐inflicted if an aqueduct on a steep slope is not well sealed and leaks carbonate if the original isotopic composition of the water is
water into the underground, which can trigger landslides. This may be an explanation why
assessed, for example, from fluid inclusions or spring monitoring.
aqueducts are commonly affected by landslides during their period of use if built along steep
slopes. This has been applied in a bath complex in Djebel Oust, Tunisia, but is
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 77

usually difficult because the water composition cannot be accurately shown by filaments (Adolphe, 1973; Carlut, 2011; Filocamo
known and because of the nonequilibrium deposition of carbonate et al., 2018; Guendon & Vaudour, 1986, 2000; Rodier et al., 2000),
(Curie et al., 2018). Clumped isotope thermometry can be used to Chironomidae tubes (Guendon et al., 2002, fig. 12612; Passchier &
determine the temperature of carbonate deposition independent of Sürmelihindi, 2012, 2019; Passchier et al., 2013; Sürmelihindi
water composition (Eiler, 2011; Huntington & Petersen, 2023) and is et al., 2023), diatoms (Sürmelihindi et al., 2020) and freshwater snails.
promising for application on aqueduct carbonates but has so far only
been applied to natural deposits (e.g., Kele et al., 2015; Kluge
et al., 2018). 3.7 | Identification of water sources and
Trace element composition of aqueduct carbonate can provide connection of channels
information on water composition and indirectly on processes in the
aqueduct, the aquifer and the soil of the catchment area (Section 2.5) Aqueduct carbonate from different branches of an aqueduct system
(Carlut et al., 2009; Keenan‐Jones et al., 2008). This includes can be linked to specific sources by means of chemical fingerprint-
recognition of dry and wet periods (Section 2.5). Mg/Ca composition ing if water or, better, modern carbonate samples are available for
is partly dependent on water temperature (Ford & Williams, 2007, those water sources and if they are significantly different in
p. 317; Roberts et al., 1998) but this effect seems to be commonly chemical composition. This application is useful where multiple
masked by the variable Mg/Ca composition of source water, caused water sources were used, where water systems have been badly
by variable residence time of water in the aquifer (Fairchild & ruined, where large parts are buried and inaccessible, or where
Treble, 2009; Jones, 2010; Passchier et al., 2021; Passchier, Sürme- multiple changes in the arrangement of water supply and distribu-
lihindi, Spötl, Mertz‐Kraus, et al., 2016). tion were realised. Examples where this has been applied or
suggested are Fréjus (Bobée, 2002; Bobée et al., 2011; Guendon
et al., 1998, 1994); Ostia (Carlut et al., 2009); Rome (Lombardi
3.6.1 | Water quality et al., 2005; Puliti et al., 1986, 1992); Sebestia (Sabri, 2016; Sabri
et al., 2015); Constantinople (Sürmelihindi et al., 2021); Arles
Discussion of the quality of Roman aqueduct water started with (Benjelloun et al., 2023) and in Pompeii (Matsui et al., 2009;
Vitruvius and Frontinus11 (Aicher, 1995), but determining the water Ohlig, 2001) but with conflicting results (Hostetter et al., 2011;
quality of an ancient source from carbonate deposits is challenging. Keenan‐Jones, Motta, et al., 2015). If the multiple sources are from
Water quality can be defined in terms of contamination with aquifers with identical geology, no distinction is possible, for
pathogenic microorganisms, or water chemistry such as high amounts example, in Cologne (Grewe, 1986, 1991), Ephesos
of salt or toxic chemicals. Research on carbonate from Cologne (Wiplinger, 2019), Cahors (Sürmelihindi et al., 2023) and Jerash
(Grewe, 1992; Lieber, 1990a, 1990b), Pompeii (Keenan‐Jones (Passchier et al., 2021). The fabric of aqueduct carbonate can also
et al., 2011) and the harbour sediments of Naples (Delile et al., 2017) be characteristic of certain channels: Ashby (1935) used the
have indicated that aqueduct water was locally contaminated stratigraphy of carbonate deposits to identify individual aqueducts
with lead. of Rome in the field.
A change or addition of water sources in an aqueduct system could
increase water quantity but also quality and this can be assessed from
carbonate composition. The aqueduct of Forum Julii (Fréjus) initially used 3.8 | Usage period of water structures
a source of brackish water from Triassic evaporites that was partly
replaced by a freshwater karst spring from Mesozoic limestones (Bobée Many deposits of aqueduct carbonate, especially in the eastern
et al., 2011). In the aqueduct system of Arelate (Arles), a southern branch Mediterranean, show a regular bimodal alternation of laminae
with small springs was abandoned and replaced by more abundant defined by crystal shape, crystal size and colour (Figures 3a, 4d
sources from a more distant, northern provenance. The southern branch and 5a,e). It has long been suggested that such regular lamination
was then used to supply the mills of Barbegal after which the water could be annual, since the most important seasonal changes in water
drained away. Water quality was suggested as a reason for this change composition are annual, while the number of observed cycles is on
in use (Guendon & Leveau, 2005). the order of magnitude of the suspected life cycle of the water
structure (Adolphe, 1973; Benjelloun et al., 2019, 2023; Carlut
et al., 2009; Carrara & Persia, 2001; Claude et al., forthcoming;
3.6.2 | Water biology Dubar, 2006a, 2006b; Gilly, 1971; Grewe & Blackman, 2001;
Guendon, 2014; Schulz, 1986). Other aqueducts show more irregular
Like modern water supply systems, Roman aqueducts contained an (Figures 3b and 4a–c) or weak or absent lamination, for example, in
ecosystem of living organisms. Carbonate preserves remains of algae micritic carbonates (Figures 4e and 5d).

11 12
Vitruvius, De Architectura VIII, 3; Frontinus De Aquis I, 12–15. These were misidentified as plant roots.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
78 | SÜRMELIHINDI and PASSCHIER

Many aqueducts show a regular cyclicity of δ18O, and to a lesser Rome, so much so that Ashby (1935) mapped part of the buried
extent δ13C, suspected to be annual (Section 2.4 and Figure 6), which, aqueducts using these deposits at the surface. Similar deposits were
especially in eastern Mediterranean aqueducts, coincides with the found outside manholes in Arles (O. Badan, personal communication,
regular visible lamination (Figure 5) (Sürmelihindi & Passchier, 2013; June 13, 2016) and around drop‐shafts in Cordoba.
Sürmelihindi, Passchier, Baykan, et al., 2013; Sürmelihindi, Passchier, Evidence for carbonate removal can also be found in deposits
Spötl, et al., 2013). This stable isotope cyclicity is thought to within aqueduct channels in the form of unconformities in the
reflect seasonal temperature variation in the water channel carbonate deposits, covered by younger carbonate or plaster
(Benjelloun et al., 2019; Sürmelihindi & Passchier, 2013, 2014; (Figures 3b,c and 9a,d,e); by tool marks (Figure 9d) and on a
Sürmelihindi, Passchier, Spötl, et al., 2013; Sürmelihindi microscopic scale, by the presence of deformation twins in sparitic
et al., 2018, 2021, 2020, 2023) and can therefore be used to calcite crystals (Burkhard, 1993; Passchier & Trouw, 2005, p. 37;
determine how many years a water structure functioned, even in Mirtović ‐Woodell et al., 2023) and included fragments of carbonate,
samples with poorly developed visible layering. Where trace element sometimes developed into spherulites (Sürmelihindi et al., 2023).
patterns show annual cyclicity, for example, Mg and P in Jerash Evidence that carbonate was removed at least once, in the
(Figure 7b), this can in principle also be used to determine the period form of tool marks or unconformities was found in aqueducts in
of use (Passchier et al., 2021). Rome (Aicher, 1995; Ceccherelli & Mancioli, 2001, pp. 174, 175;
Coates‐Stephens, 2004, pp. 54, 55); Cahors (Passchier et al., 2015;
Rigal et al., 2023; Sürmelihindi et al., 2023); Fréjus (Bobée
3.9 | Repairs, reconstructions and their relative et al., 2011; Guendon et al., 1998, 2002, p. 174, 1994); Ephesos
dating (Passchier et al., 2015); Béziers (Passchier, Sürmelihindi, Spötl,
Mertz‐Kraus, et al., 2016); Constantinople (Sürmelihindi
Repairs, reconstruction and other modifications of water channels or et al., 2021); Nîmes (Guendon & Vaudour, 1986, 2000; Hauck &
breaks in their activity all leave traces in carbonate deposits. Novak, 1988); the Serino aqueduct (Dessales et al., 2019; Keenan‐
Interruptions in the water supply for longer than 1 month can be Jones, 2015); Reims (Fronteau et al., 2023) and in the Roman water
recognised from truncations in the regular cyclicity of stable isotope mills of Saepinum (Guendon, 2014) and Barbegal (Sürmelihindi
patterns, as shown for the Barbegal water mills (Sürmelihindi et al., 2018, 2019). Evidence for periodic cleaning has been
et al., 2018). Repairs including replastering over older carbonate attested in the aqueduct of Divona (Cahors, France, where at least
was observed in Rome (Garbrecht & Manderscheid, 1995), Fréjus 14 cleaning events over a period of 88 years were recorded by
(Michel, 2002, fig. 91), Aspendos (Kessener, 2000b, fig. 11), Ephesus unconformities and the presence of tool marks and deformation
(Kessener, 2019, fig. 5) and Cahors (Sürmelihindi et al., 2023). twins. This is evidence that regular cleaning operations were
Comparison of small differences in stratigraphy between samples scheduled here, as proposed by Frontinus (Figure 3b) (Sürmelihindi
from different parts of a water structure can establish the relation and et al., 2023).
history of use of these parts. In the Barbegal mill complex, carbonate Not all aqueducts were cleaned of carbonate. Some of the larger
fragments from flumes and a wheel of the water mill show evidence for aqueducts with wide channels such as those of Nîmes and Cologne
the replacement of wooden elements (Figure 9a–c) (Sürmelihindi (Grewe, 1986, 1991, 1992; Guendon & Vaudour, 2000) were
et al., 2018, 2019). The history of use of adjacent and connected water probably not cleaned because they were purposefully over‐
channels was reconstructed from carbonate deposits for the Arles dimensioned (Fahlbusch, 1991; Hodge, 1992). Many aqueducts with
aqueduct (Guendon & Leveau, 2005), the Serino aqueduct (Filocamo smaller channels and pipes and stone inverted siphons (Figure 1f,h)
et al., 2018) and for bypasses to replace sections of aqueduct in Ephesos were apparently not cleaned, probably because they did not work for
and Fréjus (Fronteau et al., 2023; Michel, 2002; Wiplinger, 2019). a long time, or because they could not be accessed by maintenance
teams. Such structures were probably abandoned once they filled up
with carbonate. In some extreme cases of carbonate deposition, like
3.10 | Carbonate removal in water systems and Béziers, France (Esperou & Roques, 2009, pp. 23, 25, 30;
cleaning surfaces Passchier, Sürmelihindi, Spötl, Mertz‐Kraus, et al., 2016), Antibes
(Garczynski et al., 2005); Sikyon, Greece (Lolos et al., 2018) and
Carbonate deposits in ancient aqueducts reduced the flow capacity Hierapolis, Turkey (Scardozzi, 2007, pp. 247, 250), channels were
by gradual filling of channels and were regarded as an undesirable filled up completely. In Petra, a clogged‐up water pipe was opened at
formation where accumulation was significant (Hodge, 1992). In the top to allow further use (Oleson, 2018). In Béziers, Sikyon
De aquis, Frontinus13 advised regular maintenance and removal of and Hierapolis, a new channel was built on top or beside the old one
carbonate for the aqueducts of Rome. Masses of cleaned‐out to replace lost capacity (Esperou & Roques, 2009; Lolos et al., 2018;
carbonate were indeed found around manholes of the aqueducts of Scardozzi, 2007). These aqueducts had very high deposition
rates (>10 mm/year) and dense carbonate, which probably hampered
cleaning, leading to the replacement of the channels after they
13
Frontinus, De Aquis II, 122: ‘limo concrescente, qui interdum in crustam indurescit, iter
aquae coartatur’. filled up.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 79

In some cases, for example, in Nabatean aqueducts (2003) and Benjelloun et al. (2019) have done first paleoclimatic
(Oleson, 2018) and in the Roman channels of Argos (Lolos et al., 2018) reconstructions using δ18O profiles from aqueduct carbonate. Finally,
and Patara, loose tiles or pipe fragments were placed at the bottom large‐scale vegetation changes such as deforestation can be traced
of a channel, probably to make complete removal of carbonate using pollen included in carbonate (Schroyen et al., 2000) and long‐
deposits easier. Besides mechanical cleaning, Fahlbusch (1991), term changes in δ13C (Frisia, 2003).
suggested that vinegar could have been used to remove carbonate
in less easily accessible water system segments such as pipes, but no
evidence was found. 3.11.3 | Earthquakes, volcanic eruption, floods and
droughts

3.11 | Paleoenvironmental data Long‐distance aqueducts were vulnerable during earthquakes


because their unusual length implies that many were close to or
3.11.1 | Nature of feeding aquifers even crossed by active geological faults (Passchier et al., 2012).
Movement on such faults in earthquakes could cause structural
Much information about the nature of an aquifer can be obtained from damage to the aqueduct specus and even complete offset of the
the discharge, temperature and composition of spring water (Berglund channel and interruption of the water supply. Carbonate deposi-
et al., 2019; Dewandel et al., 2003; Fiorillo, 2014; Ford & tion in the channel can be interrupted by stirred‐up sediment or
Williams, 2007). These spring water signals are captured in aqueduct debris fallen into the channel that will be included in the deposits if
carbonate, creating an archive of aquifer behaviour. For example, the water flow continues. Structural damage to the channel could
aqueducts of Béziers and Jerash show a characteristic ‘sawtooth’ cause fractures in the carbonate and damage to calcite crystals in
distribution of stable isotopes and trace elements, notably δ13C, Mg the form of deformation twins (Passchier et al., 2021; Volant
and P (Figure 7a,b) (Passchier et al., 2021; Passchier, Sürmelihindi, et al., 2009), although these can also form due to cleaning
Spötl, Mertz‐Kraus, et al., 2016). This can be explained by a small (Section 3.10) (Sürmelihindi et al., 2023). Earthquake damage to
aquifer with a strong sudden increase in water flow after winter rains aqueducts and the use of carbonate deposits to analyse this
(peak of P and drop in Mg and δ13C), followed by a gradual depletion damage was reported for the aqueducts of Nîmes (Gilly &
and falling water levels in summer (decrease in P and increase in Mg Levret, 2001; Levret et al., 2008; Volant et al., 2009), Ephesos
and δ C). The changes in Mg and P are especially pronounced in
13
(Passchier et al., 2013), Venafro (Galli et al., 2010), Patara
Jerash due to the local arid climate (Figure 7b). Large aquifers do show (Passchier, Sürmelihindi, Spötl, 2016), Iznik (Benjelloun et al., 2018),
fluctuations, but less extreme. Since many springs that were used to Köln (Hoffmann et al., 2019) Antioch (Benjelloun et al., 2015) and a
feed ancient water systems are now dry, aqueduct carbonate can give villa aqueduct near Rome (Marra et al., 2004). The Al Harif
a unique insight into the properties of now‐depleted aquifers. aqueduct in Syria was displaced several times by earthquakes over
a total of 14 m and partly repaired, as could be reconstructed and
dated with carbonate in the channel (Meghraoui et al., 2003;
3.11.2 | Paleoclimate and the environment Sbeinati et al., 2010). The Değirmendere aqueduct of Ephesos was
displaced over 3 m by a single earthquake and rebuilt after the
Like speleothems, the microstratigraphy of Roman aqueduct carbon- event (Passchier et al., 2013). Earthquakes can also fracture
ate contains information on paleoclimate for periods up to several aqueducts and cause them to leak (Combes et al., 1997; Levret
centuries (Gilly et al., 1978; Benjelloun et al., 2019; Claude et al., 2008; Passchier et al., 2012) which can trigger landslides,
et al., forthcoming; Dubar, 2006a, 2006b; Garczynski et al., 2005; destroying part of an aqueduct. Tilting of the terrain associated
Schulz, 1986). Since aqueducts show seasonal temperature fluctua- with earthquakes or volcanic eruption may affect the aqueduct
tion, a water temperature signal with up to daily resolution can in slope, for example, the Serino aqueduct before the Vesuvius
some cases be retrieved (Figure 5e). Aquifer water composition and eruption of AD 79 (Filocamo et al., 2018; Keenan‐Jones, 2015).
discharge depend on rainfall and hence, information on rainfall is Theoretically, carbonate chemistry could also record ashfall from
stored in the δ13C, trace element and water level variation signals. volcanic eruptions (Frisia et al., 2005, 2008).
Potentially, ancient aqueduct carbonate can therefore provide very Like earthquakes, floods can damage aqueducts, cause
high‐resolution climate information for antiquity. Application of the erosional unconformities in the carbonate (Guendon, 2014;
tool, however, is still hampered by several problems, associated with Keenan‐Jones et al., 2014) or fill them with sediment, which
the separation of anthropogenic effects (Sections 3.7, 3.9 and 3.10), may be recorded in the deposits as in the aqueduct of Argos
calibration and in particular, precise dating. First attempts to retrieve (Lolos et al., 2018) and the Byzantine watermills of Ephesos
paleoclimatic data were made by Gilly et al. (1971), Garczynski et al. (Passchier & Sürmelihindi, 2015).
(2005) and Dubar (2006a, 2006b) who, using laminae thickness in the Evidence for droughts in the form of drying out of the channel
aqueducts of Nîmes, Antibes and Fréjus, respectively, recognised was observed in Béziers (Passchier, Sürmelihindi, Spötl, Mertz‐Kraus,
signals of the North Atlantic Oscillation using spectral analysis. Frisia et al., 2016) and Jerash (Passchier et al., 2021).
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
80 | SÜRMELIHINDI and PASSCHIER

3.12 | Damage and demise (Passchier, Sürmelihindi, Spötl, 2016) and Ephesos (Passchier
et al., 2013) with a resolution on 1 year, while coin finds in or
In many aqueducts, the upper part of the stratigraphy differs from below carbonate helped to date the Barbegal watermills
the main part, usually with more micritic and porous deposits (e.g., (Leveau, 1995) and the Nîmes aqueduct (Veyrac, 1992) with a
Carlut, 2011; Garbrecht & Manderscheid, 1995; Guendon, 2014; precision of a few years.
Guendon & Vaudour, 1986; Passchier et al., 2021; Sürmelihindi Two radiogenic isotope dating techniques are commonly used
14
et al., 2023). If the transition is gradual, this can be due to for terrestrial carbonate, C and U/Th dating (Pentecost, 2005,
14 14
decreasing flow speed due to the narrowing of the channel with pp. 243–250). The C method is based on the decay of C which
increasing carbonate accumulation, but in Jerash, Barbegal and forms in the atmosphere and is no longer replenished once it is fixed
14
Cahors, the final deposits contain excessive clay and imprints of in biological or crystalline materials. In most cases, the C method
plant remains (Figure 9f) (Passchier et al., 2021; Sürmelihindi cannot be applied to date the aqueduct carbonate itself, since it will
et al., 2019, 2023). Probably, these final stratigraphic levels indicate be affected by the massive presence of ‘dead carbon’ from the
decreasing maintenance of the aqueduct. Roman aqueducts were dissolved bedrock of the aquifer (Blyth et al., 2017; Meghraoui
surrounded by a protected zone where no trees were allowed to et al., 2003; Pentecost, 2005, pp. 243–250; Pons‐Branchu
14
grow and where agriculture was forbidden (Burdy, 2001). If trees et al., 2018; Sbeinati et al., 2010). An alternative is C dating of
are no longer cleaned above a buried aqueduct, tree roots will soon freshwater snails or of organic matter included in the carbonate
breach the vault and cause the entrance of stormwater, soil and (Blyth et al., 2017; Neely et al., 2022; Winsborough et al., 1996),
admittance of light, causing changes in deposited carbonate as using δ13C values to detect possible contamination with inorganic
described above. A gradual change in aqueduct carbonate quality carbon. Finally, charcoal in the aqueduct inner plaster lining can be
was attributed to a decline in maintenance in Nîmes and Arles dated. This method was applied to date the aqueduct of Jerash
(Wilson, 2000, p. 602). (Boyer, 2022).
In Cahors, evidence for deterioration was not only found at the The U/Th dating method (Frank et al., 2002; Pentecost, 2005,
top but also at the centre of the deposits, indicating a period of lack p. 249) is the most directly applicable to carbonate and has been
of maintenance, followed by recovery. δ18O and δ13C values showed successfully applied in some water systems (Claude et al., forthcoming;
a gradual rise towards the top of these central deposits (Sürmelihindi Frumkin et al., 2003; Passchier, Sürmelihindi, Spötl, Mertz‐Kraus,
et al., 2023). The final deterioration and abandonment of the same et al., 2016; Pons‐Branchu et al., 2018). Aqueducts, however,
aqueduct took place only a few years after the last cleaning event commonly contain clay and other detrital particles in sub‐
230
(Sürmelihindi et al., 2023). microscopic particles, which contain Th (initial Th) which signifi-
The presence of leakage deposits outside aqueducts can also be cantly increases the error of the method (Wenz et al., 2016). Dating is
an indication of abandonment and lack of maintenance during their therefore presently best attempted in white, sparitic translucent
234
final years. The aqueduct of Nîmes shows such deposits with aqueduct carbonate with low clay content and high U levels. In
imprints of beams in its upstream parts, suggesting the aqueduct such cases, an error of ±15 years is attainable (Claude
was tapped, illegally or after abandonment14 (Figure 3d) (Guendon et al., forthcoming).
& Vaudour, 1986, 2000; Hauck & Novak, 1987). In any case, a
sudden decrease in maintenance or enforcement of aqueduct
protection may indicate a change in local or general socioeconomic 5 | REU SE AS SP OLI A
conditions, that could not be revealed by any other means
(Wilson, 1999, 2000). Aqueduct carbonate can be extracted from ancient water systems
and used as building stone and, in the case of dense laminated
carbonate, as ornamental stone. Carbonate from the Cologne
4 | D A TI N G O F CA R B O N A T E D E P O S I TS aqueduct was used to produce decorative ‘marble’ columns, altar
plates and gravestones in churches in Germany, the United Kingdom
In most aqueduct carbonate deposits, it is possible to determine the and Denmark (Grewe, 1982, 1991, 1992, 1996; Grewe &
period of usage by counting of number of annual δ18O cycles. Blackman, 2001). In Hierapolis, Turkey, a house was partly built
Absolute dating of such ‘floating’ carbonate time series can be done from carbonate fragments formed in ceramic pipes (Figure 10d)
using conventional radiogenic isotope dating techniques and, since (Scardozzi, 2007) while carbonate of the Nîmes aqueduct was used to
the carbonate deposits are linked to anthropogenic structures, using build churches and castles in the Middle Ages (Figure 10a)
epigraphy or datable foreign bodies in the carbonate or the (Paillet, 1986; Paillet & Gilly, 2000).
aqueduct structure. Epigraphy related to earthquakes helped to Where carbonate from water systems was already used in
date carbonate time series of the aqueducts of Patara antiquity as a building material, it can be applied to date the time of
abandonment of the affected water system or maintenance work.
Examples have been found in Rome (Ashby, 1935), Barbegal, France
14
Similar deposits are present on the aqueducts of Caesarea Maritima and Salona. (Leveau, 1995) and Pompeii (Figure 10b,c).
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 81

F I G U R E 10 Examples of the use of carbonate deposits as a building material. (a) The church of Saint‐Bonnet, France, mostly built with blocks
of carbonate deposits from the Nîmes Roman aqueduct. (b) Carbonate segment from the bottom of a wooden millrun flume of the Barbegal
watermills, France, built into the walls of a late‐Roman building. (c) Fragment of carbonate (centre) with attached red mortar from a water
channel or basin, used as building stone in a Roman house in Pompeii. (d) Sinter house, Hierapolis, Turkey. Dense calcium carbonate formations
from water pipes with remains of the terracotta pipes (inset) were used for the construction of the walls. Photos by Cees Passchier.

6 | DISC US SION is unique in its intricate link to archaeology, since it forms inside
ancient structures that provided habitation centres, villas, mines and
Carbonate from ancient water systems is a type of freshwater deposit industrial sites with water. As such, these carbonate deposits can
comparable to flowstone in caves and tufa in rivers, but special, with provide a wealth of archaeological data and information on local
its own depositional environment and characteristics (Ford & environmental aspects, linked through water management. Carbonate
Williams, 2007; Wróblewski et al., 2017). The deposit can be exploited from ancient water systems can be used to test whether these systems
as a new environmental archive that can be studied with the same were operational (Section 3.1) to determine the shape of lost elements
tools as applied to other freshwater carbonates. However, the material of a water system, notably if these were made of wood or metal
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
82 | SÜRMELIHINDI and PASSCHIER

(Section 3.2) and to find details on the function of special structures depositional conditions and flow mechanisms. These carbonate deposits,
such as inverted siphons and water gutters (Section 3.3) and the water with their direct connection between water supply and centres of human
level and indirectly, the flow speed and discharge in a water system, activity, give us a unique chance to study human history and its
which is important to determine the volume of water available at a site connection with water, a relationship as old as the first human
per day (Section 3.4). Variations in discharge along a channel can reveal settlements.
leaks and diversion, or additions of water. Further, it is possible to Only 5% of the 2400 presently known ancient aqueducts in the
establish the closed or open nature of water‐bearing structures where Mediterranean have so far been studied in terms of their carbonate
the top is destroyed (Section 3.5), especially important in the case of deposits and few outside this area. The carbonate deposits in these
water mills or basins. It is even possible to determine when buildings aqueducts and other ancient water systems represent unique archives
were ruined and the light was admitted in later stages (Section 3.5). for archaeology and environmental sciences that should be studied
Water temperature and composition, pollution and water biology can before they are destroyed by human action or weathering. Ancient
be analysed (Section 3.6) and it is possible to determine which water water structures used energy‐efficient technology, some of which may
sources were used, and the type of water each one once provided still be applicable today. They provide a historical benchmark of local
(Section 3.7), especially useful where water supply systems are climate conditions and the functioning of aquifers in the past that will
complex and frequently altered, and where buried parts of a water be a boon for humanity to help solve the problems we face in a drying
supply system are inaccessible. Other applications are the determina- and warming world.
tion of the number of years a system functioned (Section 3.8), repairs,
modifications and reconstructions (Section 3.9); carbonate cleaning A UT H O R C O N T R I B U TI O NS
and maintenance schedules (Section 3.10); and the collection of local Gül Sürmelihindi: Methodology and writing. Cees Passchier: Writing
paleoenvironmental data (Section 3.11) about aquifer hydrology, and original draft.
weather, climate and extreme events such as floods, earthquakes
and droughts, all with high resolution and directly linked to habitation ORC I D
centres; finally, the termination of water system use is important Gül Sürmelihindi http://orcid.org/0000-0001-7874-8631
(Section 3.12) including the question why these systems
were abandoned and how sudden this cessation was. Indirectly, RE F ER EN CES
aqueduct carbonate can possibly give information on groundwater Adolphe, J.‐P. (1973). Contribution à l'étude des encroûtements carbon-
pollution, deforestation and volcanic eruptions. In archaeology, urban atés de l'aqueduc du Pont du Gard. Comptes Rendus de l'Académie
des Sciences Paris, 277, 2329–2332.
development, population dynamics, socioeconomic developments
Aicher, P. J. (1995). Guide to the aqueducts of ancient Rome. Bolchazy‐
and resilience can be studied through aqueduct carbonates. Clearly, Carducci Publishers.
understanding the processes, efforts and modifications made to water Akan, A. O. (2006). Open channel hydraulics. Elsevier.
systems helps us to understand changing paleoenvironmental condi- Andrews, J. E. (2006). Palaeoclimatic records from stable isotopes in
riverine tufas: Synthesis and review. Earth‐Science Reviews, 75,
tions and anthropogenic impact on water sources and on nature.
85–104.
Andrews, J. E., Brimblecombe, P., Jickells, T. D., Liss, P. S., & Reid, B.
(2004). An introduction to environmental chemistry. Blackwell.
7 | C ONC LUS I ON S AND FU TU RE WORK Anderkó, K. (2006). Savaria Vízvezetéke. Savaria ‐ A Vas Megyei Müzeumok
Értesítöje 30, 9–46.
Angelakis, A. N., Chiotis, E., Eslamian, S., & Weingartner, H. (2017).
Several trends have been set in motion over the last decade that
Underground aqueduct handbook. CRC Press.
promise to make aqueduct carbonate an important archive in Angelakis, A. N., Mays, L. W., Koutsoyiannis, D., & Mamassis, N. (2012).
archaeology and paleoenvironmental studies. The interdisciplinary Evolution of water supply througout the millennia. IWA Publishing.
and complex nature of the subject warrants the set‐up of a separate Ashby, T. (1935). The aqueducts of ancient Rome. Clarendon Press.
Baatz, D. (1978). Temperatur und sinterbildung. Berichte aus der Arbeit des
field of archaeohydrology, serving archaeology, history of science and
Museums, 6, 90.
environmental sciences, applied not only to Roman aqueducts but
Bailhache, M. (1979). Le vieillissement des aqueducs et la baisse de leur
also to water systems of other time periods and of other cultures. débit. Les aqueducs romains. Les Dossiers d'Archéologie, 38, 62–71.
Analytical techniques can be further developed and expanded. Tests Bailhache, M. (1983). L'évolution des débits des aqueducs gallo‐romains.
have shown that for trace elements in aqueduct carbonate deposits, a Dossiers de l'Archéologie, 38, 82–87.
Bar‐Matthews, M., Matthews, A., & Ayalon, A. (1991).
resolution of hours is theoretically possible, allowing reconstruction of
Environmental controls of speleothem mineralogy in a karstic
paleoclimate with unprecedented resolution. The dating of carbonate dolomitic terrain (Soreq Cave, Israel). The Journal of Geology, 99,
deposits should be improved, possibly by increasing the resolution of U/ 189–207.
Th dating, while the correlation with tree rings and cave speleothems Benjelloun, Y., Carlut, J., Hélie, J.‐F., Chazot, G., & Le Callonnec, L. (2019).
Geochemical study of carbonate concretions from the aqueduct of
should be explored. Monitoring studies in a working aqueduct over a
Nîmes. Scientific Reports, 9, 5209.
longer period could help boost the interpretation of aqueduct carbonate
Benjelloun, Y., Helie, J.‐F., & Carlut, J. (2023). Étude géochimique des
deposits. Further potential trends are studies of fluid inclusions, pollen, concrétions carbonatées des aqueducs romains de Nîmes et d'Arles
ancient DNA and studies on fabric elements that can give information on (France). Archéologie, Société et Environnement, 1.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 83

Benjelloun, Y., de Sigoyer, J., Carlut, J., Hubert‐Ferrari, A., Dessales, H., Caran, S. C., & Neely, J. A. (2006). Hydraulic engineering in prehistoric
Pamir, H., & Karabacak, V. (2015). Characterization of building Mexico. Scientific American, 295, 78–85.
materials from the aqueduct of Antioch‐on‐the‐Orontes (Turkey). Carbon, D., Fabre, G., Volant, P., Fiches, J.‐L., Levret, A., & Combes, P.
Comptes Rendus Geoscience, 347, 170–180. (2005). L'aqueduc de Nîmes dans la haute Vistrenque: Analyse
Benjelloun, Y., de Sigoyer, J., Dessales, H., Garambois, S., & Şahin, M. (2018). interdisciplinaire d'un tronçon souterrain. Gallia, 62, 69–86.
Construction history of the aqueduct of Nicaea (Iznik, NW Turkey) Carlut, J. (2011). Observations sur une concrétion de l'aqueduc antique de
and its on‐fault deformation viewed from archaeological and geo- Bellegarde (Languedoc). Bulletin de l'école Antique de Nîmes, 29,
physical investigations. Journal of Archaeological Science: Reports, 21, 311–314.
389–400. Carlut, J., Chazot, G., Dessales, H., & Letellier, É. (2009). Trace element
Benoit, F. (1940). L'usine de meunerie hydraulique de Barbegal (Arles). variations in an archeological carbonate deposit from the antique
Revue Archéologique, 15, 19–80. city of Ostia: Environmental and archeological implications. Comptes
Berglund, J. L., Toran, L., & Herman, E. K. (2019). Deducing flow path Rendus Geoscience, 341, 10–20.
mixing by storm‐induced bulk chemistry and REE‐variations in two Carrara, C., & Persia, F. (2001). Indagini mineralogico‐petrographiche e di
karst springs: With trends like these who needs anomalies? Journal diffrazione dei raggi X sulle incrostazioni calcaree e sulle malte. In D.
of Hydrology, 571, 349–364. Mancioli & G. Pisani Sartorio (Eds.), Gli acquedotti Claudio e Aniene
Berner, R. A. (1992). Weathering, plants, and the long‐term carbon cycle. Nuovo nell' area della Banca d'Italia in via Tuscolana (pp. 193–197).
Geochimica et Cosmochimica Acta, 56, 3225–3231. Istituto Poligrafico e Zecca dello Stato.
Blanc, J.‐J. (2000). La séquence des lamines carbonatées du chemin de fer Ceccherelli, A., & Mancioli, D. (2001). Le fasi costruttive delle acque
au Grès (Sernhac). In G. Fabre, J.‐L. Fiches, & J.‐L. Paillet (Eds.), Claudia‐Anio Novus dalle origini fino all'epoca tardo‐antica. In D.
L'aqueduc de Nîmes et le Pont du Gard, archéologie, géosystème et Mancioli, & G. Pisani Sartorio (Eds.), Gli acquedotti Claudio e Aniene
histoire (pp. 249–261). CNRS Editions. Nuovo nell' area della Banca d'Italia in via Tuscolana (pp. 71–187).
Blyth, A. J., Hua, Q., Smith, A., Frisia, S., Borsato, A., & Hellstrom, J. (2017). Istituto Poligrafico e Zecca dello Stato.
Exploring the dating of “dirty” speleothems and cave sinters using Chanson, H. (2004). Hydraulics of open channel flow. Elsevier.
radiocarbon dating of preserved organic matter. Quaternary Chen, J., Zhang, D. D., Wang, S., Xiao, T., & Huang, R. (2004). Factors
Geochronology, 39, 92–98. controlling tufa deposition in natural waters at waterfall sites.
Bobée, C. (2002). Essai d'interprétation du fonctionnement des Sedimentary Geology, 166, 353–366.
aqueducs romains de Nîmes et de Fréjus (à partir d'études Chow, V. T. (1959). Open‐channel hydraulics. McGraw‐Hill.
géochimiques du concrétionnement interne des conduits), Mém- Christensen, B. A. (1988). Discussion of interaction of flow and
oire DEA. Environnement et Archéologie, 1–41. incrustation in the Roman aqueduct of Nîmes by George F. W.
Bobée, C., Huon, S., Guendon, J.‐L., Salomon, J., Gébara, C., Michel, J.‐M., & Hauck and Richard A. Novak (February, 1987, Vol. 113, No. 2).
Regert, M. (2011). High‐resolution (PIXE) analyses of carbonate Journal of Hydraulic Engineering, 114, 462–464.
deposits in a Roman aqueduct (Fréjus, SE France): Implications for the Clark, I. D., & Fritz, P. (1997). Environmental isotopes in hydrogeology. Lewis
study of palaeohydrological variability and water resources manage- Publishers.
ment in Southern Gaul during the Roman period. Archaeometry, 53(2), Claude, C., Vidal, L., Passchier, C., Angeletti, B., Guillou, A., Deschamps, P.,
241–260. Ricolleau, A., Sürmelihindi, G., Sonzogni, C., Delanghe, D., Fino, B.,
Bonnin, J. (1984). L'eau dans l'Antiquité, L'hydraulique avant notre ère. Eyrolles. Fuhry, M., Leveau, P., Marié, L., Panneau, M., Gassier, G., & Nin, N.
Bono, P., Dreybrodt, W., Ercole, S., Percopo, C., & Vosbeck, K. (2001). (forthcoming). High resolution paleo‐hydrological record from
Inorganic calcite precipitation in Tartare karstic spring (Lazio, central carbonate deposits in the roman aqueduct of Traconnade (Aix‐en‐
Italy): Field measurements and theoretical prediction on depositional Provence, SE France). Chemical Geology.
rates. Environmental Geology, 41, 305–313. Coates‐Stephens, R. (2004). Porta Maggiore. Monument and landscape:
Botturi, G., & Parecinni, R. (1991). Antichi acquedotti del territorio Archaeology and topography of the southern Esquiline from the Late
Bresciano. Edizioni ET. Republican period to the present (Vol. 12). L'Erma di Bretschneider.
Boyer, D. D. (2022). Water management in Gerasa and hinterland: From Combes, P., Fabre, G., Fiches, J. L., Gazenbeek, M., & Paillet, J. L. (1997).
the Romans to AD 750 (Vol. 10). Brepols. L'aqueduc romain de Nîmes et l'enregistrement d'événements
Brasier, A. T. (2011). Searching for travertines, calcretes and speleothems sismiques. In J.‐L. Fiches, B. Helly, & A. Levret (Eds.), Archéologie et
in deep time: Processes, appearances, predictions and the impact of sismicité: Autour d'un grand monument, le Pont du Gard (pp. 75–123).
plants. Earth‐Science Reviews, 104, 213–239. Actes des journées d'étude Nîmes.
Brasier, A. T., Andrews, J. E., & Kendall, A. C. (2011). Diagenesis or dire Curie, J., & Petit, C. (2014). Geoarchaeology of “anthropogenic” travertine:
genesis? The origin of columnar spar in tufa stromatolites of central A story of water and life etched in stone. European Geologist, 38,
Greece and the role of chironomid larvae Sedimentology, 58, 21–24.
1283–1302. Curie, J., Petit, C., Ben Abed, A., Broise, H., & Scheid, J. (2018). Les dépôts
Brinker, W. (1986). Überlegungen zur Hydrologie und Hydraulik der carbonatés en contexte archéologique, mémoire d'une gestion de
Eifelleitung. In K. Grewe (Ed.), Atlas der römischen Wasserleitungen l'eau: l'exemple du site de Jebel Oust, Tunisie. In V. Brouquier‐
nach Köln (pp. 235–247). Rheinland Verlag. Reddé, & F. Hurlet (Eds.), L'eau dans les villes du Maghreb et leur
Burdy, J. (2001). Une nouvelle borne de protection de l'aqueduc romain territoire à l'époque romaine (Vol. 54, pp. 273–286). Ausonius
du Gier (Lyon). Bulletin de la Société Nationale des Antiquaires de Mémoires.
France, 1997, 152–161. Delile, H., Keenan‐Jones, D., Blichert‐Toft, J., Goiran, J.‐P., Arnaud‐Godet,
Burkhard, M. (1993). Calcite twins, their geometry, appearance and F., & Albarède, F. (2017). Rome's urban history inferred from Pb‐
significance as stress‐strain markers and indicators of tectonic contaminated waters trapped in its ancient harbor basins.
regime: A review. Journal of Structural Geology, 15, 351–368. Proceedings of the National Academy of Sciences of the United States
Cáceres, J. O., Pelascini, F., Motto‐Ros, V., Moncayo, S., Trichard, F., of America, 114, 10059–10064.
Panczer, G., Marín‐Roldán, A., Cruz, J. A., Coronado, I., & Martín‐ Dessales, H., Carlut, J., & Filocamo, F. (2019). L'entretien d'un aqueduc
Chivelet, J. (2017). Megapixel multi‐elemental imaging by laser‐ face au risque sismique: Le cas de l'aqueduc du Serino (Italie).
induced breakdown spectroscopy, a technology with considerable Instandhaltung und Renovierung von Straßen und Wasserleitungen von
potential for paleoclimate studies. Scientific Reports, 7, 5080. der Zeit der römischen Republik bis zur Spätantike, 135–160.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
84 | SÜRMELIHINDI and PASSCHIER

Dewandel, B., Lachassagne, P., Bakalowicz, M., Weng, P., & Al‐Malki, A. Frisia, S. (2015). Microstratigraphic logging of calcite fabrics in spe-
(2003). Evaluation of aquifer thickness by analysing recession leothems as tool for paleoclimate studies. International Journal of
hydrographs. Application to the Oman ophiolite hard‐rock aquifer. Speleology, 44, 1–16.
Journal of Hydrology, 274, 248–269. Frisia, S., Borsato, A., Drysdale, R. N., Paul, B., Greig, A., & Cotte, M.
Dreybrodt, W. (1982). A possible mechanism for growth of calcite (2012). A re‐evaluation of the palaeoclimatic significance of
speleothems without participation of biogenic carbon dioxide. Earth phosphorus variability in speleothems revealed by high‐resolution
and Planetary Science Letters, 58, 293–299. synchrotron micro XRF mapping. Climate of the Past, 8, 2039–2051.
Dreybrodt, W. (2019). Physics and chemistry of CO2 outgassing from a Frisia, S., Borsato, A., Fairchild, I. J., & McDermott, F. (2000). Calcite
solution precipitating calcite to a speleothem: implication to 13C, fabrics, growth mechanisms, and environments of formation in
18
O, and clumped 13C18O isotope composition in DIC and calcite. speleothems from the Italian Alps and Southwestern Ireland. Journal
Acta Carsologica, 48, 59–68. of Sedimentary Research, 70, 1183–1196.
Dubar, M. (2006a). Approche climatique de la période romaine dans l'est du Frisia, S., Borsato, A., Fairchild, I. J., & Susini, J. (2005). Variations in
Var: Recherche et analyse des composantes périodiques sur un atmospheric sulphate recorded in stalagmites by synchrotron micro‐
concrétionnement centennal (Ier–IIème siècle apr. J.‐C.) de l'aqueduc XRF and XANES analyses. Earth and Planetary Science Letters, 235,
de Fréjus. Archéosciences, 30, 163–171. 729–740.
Dubar, M. (2006b). Recherche et interprétation climatique des micro- Frisia, S., Borsato, A., & Hellstrom, J. (2018). High spatial resolution
cycles du concrétionnement travertineux de l'aqueduc romain de investigation of nucleation, growth and early diagenesis in spe-
Fréjus (Var, France). Quaternaire, 17, 79–85. leothems as exemplar for sedimentary carbonates. Earth‐Science
Durlet, C., Brenot, J., Philippe, É., Curie, J., Teboul, P.‐A., Cocquerez, T., & Reviews, 178, 68–91.
Bruneau, L. (2023). Les dépôts carbonatés de l'aqueduc de Poitiers‐ Frisia, S., Borsato, A., & Susini, J. (2008). Synchrotron radiation
Fleury (Vouneuil‐sous‐Biard, 86): l'enregistrement d'une dynamique applications to past volcanism archived in speleothems: An over-
hydro‐sédimentaire instable? Archéologie, Société et Environnement, view. Journal of Volcanology and Geothermal Research, 177, 96–100.
1, 1–18. Fronteau, G., Couturier, D., & Rabasté, Y. (2023). Les concrétions
Eick, C. A. (1867). Die römische Wasserleitung aus der Eifel nach Köln, mit carbonatées des aqueducs gallo‐romains de Villenoy (Seine‐et‐
Rücksicht auf die zunächst gelegenen römischen Niederlassungen, Marne, France) et de la Suippe à Reims (Marne, France): Etude
Befestigungswerke und Heerstrassen. Cohen & Sohn. pétrographique et questionnement sur l'origine des séquences
Eiler, J. M. (2011). Paleoclimate reconstruction using carbonate enregistrées. Archéologie, Société et Environnement, 1, 1–14.
clumped isotope thermometry. Quaternary Science Reviews, 30, Frumkin, A., Shimron, A., & Rosenbaum, J. (2003). Radiometric dating of
3575–3588. the Siloam tunnel, Jerusalem. Nature, 425, 169–171.
Emeis, K. C., Richnow, H. H., & Kempe, S. (1987). Travertine formation in Galli, P. A. C., Giocoli, A., Naso, J. A., Piscitelli, S., Rizzo, E., Capalini, S., &
Plitvice National Park, Yugoslavia: Chemical versus biological Scaroina, L. (2010). Faulting of the Roman aqueduct of Venafrum
control. Sedimentology, 34, 595–609. (southern Italy): Methods of investigation, results, and seismotec-
Esperandieu, É. (1926). Le Pont du Gard et l'Aqueduc de Nîmes. Petites tonic implications. In M. Sintubin, I. S. Stewart, T. M. Niemi, & E.
monographies des grands édifices de la France. Henri Laurens. Altunel (Eds.), Ancient earthquakes (Vol. 471, pp. 233–242). Geologi-
Esperou, J. L., & Roques, P. (2009). L'Aqueduc Romain de Béziers. Pro cal Society of America.
Baeterris. Garbrecht, G., & Manderscheid, H. (1995). Die Wasserversorgung der
Fabretti, R. G. (1690). De Columna Traiani syntagma. Accesserunt explicatio Caracallathermen durch die Aqua Antoniana. Antike Welt, 26, 193–202.
veteris tabellae anaglyphae Homeri Iliadem atque ex Stesichoro Arctino Garczynski, P., Foucras, J., & Dubar, M. (2005). L'aqueduc d'Antipolis dit
et Lesche ilii excidium continentis. Emissarii Lacus Fucini descriptio. de la Bouillide (Alpes‐Maritimes): Aqueducs de la Gaule méditerra-
Forgotten Books. néenne. Gallia, 62, 13–34.
Fahlbusch, H. (1991). Maintenance problems in ancient aqueducts. In A. T. Genty, D., Baker, A., Massault, M., Proctor, C., Gilmour, M., Pons‐Branchu,
Hodge (Ed.), Future currents in aqueduct studies (pp. 7–14). Francis E., & Hamelin, B. (2001). Dead carbon in stalagmites: carbonate
Cairns Publications. bedrock paleodissolution vs. ageing of soil organic matter. Implica-
Fairchild, I. J., & Baker, A. (2012). Speleothem science: From processes to tions for 13C variations in speleothems. Geochimica et Cosmochimica
past environments. Wiley‐Blackwell. Acta, 65, 3443–3457.
Fairchild, I. J., & Treble, P. C. (2009). Trace elements in speleothems as Genty, D., & Quinif, Y. (1996). Annually laminated sequences in the
recorders of environmental change. Quaternary Science Reviews, 28, internal structure of some Belgian stalagmites importance for
449–468. paleoclimatology. Journal of Sedimentary Research, 66, 275–288.
Fellmann, R. (1988). Die Römer in der Schweiz. Theiss. Gilly, E., & Levret, A. (2001). Recherche d'un archéoséisme dans les
Filocamo, F., Carlut, J., Dessales, H., Burckel, P., & Borensztajn, S. (2018). concrétions du pont de la Lône (Vers‐Pont‐du‐Gard, France). Datation
Roman builders facing the risk of disaster: Coupling archaeological d'un niveau repère. Colloque Riviera 2000—Géomorphologie et
and geochemical analyses on a section of the ‘Aqua Augusta’. tectonique active (pp. 1–7).
Archaeometry, 60, 915–932. Gilly, J.‐C. (1971). Les dépôts calcaires de l'aqueduc antique de Nîmes.
Fiorillo, F. (2014). The recession of Spring Hydrographs, focused on Karst Bulletin Annuel de l'école Antique de Nîmes, 6–7, 1971–1972.
Aquifers. Water Resources Management, 28, 1781–1805. Gilly, J.‐C. (1986). Étude géochimique des incrustations de l'aqueduc
Ford, D., & Williams, P. (2007). Karst hydrogeology and geomorphology. Wiley. romain conduisant les eaux d'Uzès à Nîmes: Détermination de
Fouke, B. W. (2011). Hot‐spring systems geobiology: Abiotic and biotic l'origine des eaux d'alimentation. Méditerranée, 57, 131–139.
influences on travertine formation at Mammoth Hot Springs, Gilly, J.‐C. (1990). L'aqueduc romain de Béziers—Pente, tracé, technique
Yellowstone National Park, USA. Sedimentology, 58, 170–219. de construction et incrustations sur les parois. Études Héraultaises,
Frank, N., Mangini, A., & Korfmann, M. (2002). 230Th/U dating of the 1990, 20–21.
Trojan “Water quarries”. Archaeometry, 44, 305–314. Gilly, J.‐C., Coudray, J., & Plegat, R. (1978). Zonation et géochimie des
Frisia, S. (2003). Towards an understanding of the relationships between incrustations calcitiques de l'aqueduc romain du Pont du Gard
environmental changes and human societies: Analytical techniques, comme témoin de la paléoclimatologie et de la paléohydrologie des
and examples from the Alps. Preistoria Alpina, Museo Tridentino di cinq premiers siècles de notre ère. Symposium de l'institut des
Scienze Naturali, 39, 39–47. hautes études scientifiques, Implication de l'hydrogéologie dans les
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 85

autres sciences de la terre. Mémoire du Centre d'études et de Hagendorn, A. (2003). Zur frühzeit von Vindonissa (Vol. 18). Veröffentli-
recherches géologiques et hydrogéologiques, 2, 659–673. chungen der Gesellschaft Pro Vindonissa.
Gilly, J.‐C., Plegat, R., & Coudray, J. (1971). Note préliminaire sur les Harmon, R. S., Atkinson, T. C., & Atkinson, J. L. (1983). The mineralogy of
incrustations calcitiques de l'aqueduc romain du Pont du Gard, Castleguard cave, Columbia Icefields, Alberta, Canada. Arctic and
indicateurs paléoclimatique et paléochronologique des cinq premiers Alpine Research, 15, 503–516.
siècles de notre ère. Comptes Rendus Hebdomadaires des Séances de Harmon, R. S., & Senesi, G. S. (2021). Laser‐induced breakdown
l'Académie des Sciences, 273, 1668–1670. spectroscopy—A geochemical tool for the 21st century. Applied
Goldschneider, N. (2015). Overview of methods applied in karst Geochemistry, 128, 104929.
hydrogeology. In Z. Stevanovic (Ed.), Karst Aquifers‐Characterization Hauck, G. F. W., & Novak, R. A. (1987). Interaction of flow and
and Engineering (pp. 127–145). Springer Verlag. incrustation in the roman aqueduct of Nîmes. Journal of Hydraulic
González, L. A., Carpenter, S. J., & Lohmann, K. C. (1992). Inorganic calcite Engineering, 113, 141–156.
morphology: Roles of fluid chemistry and fluid flow. Journal of Hauck, G. F. W., & Novak, R. A. (1988). Closure to “interaction of flow and
Sedimentary Petrology, 62, 382–399. incrustation in the Roman Aqueduct of Nîmes” by George F. W.
Gradziński, M. (2010). Factors controlling growth of modern tufa: Results Hauck and Richard A. Novak (February, 1987, Vol. 113, No. 2).
of a field experiment. In H. M. Pedley, & M. Rogerson (Eds.), Tufas Journal of Hydraulic Engineering, 114, 465–466.
and speleothems: Unravelling the microbial and physical controls (Vol. Hinsinger, P., Plassard, C., & Jaillard, B. (2006). Rhizosphere: A new
336, pp. 143–191). Geological Society Special Publications. frontier for soil biogeochemistry. Journal of Geochemical Exploration,
Gradziński, M., Chmiel, M. J., Lewandowska, A., & Michalska‐ 88, 210–213.
Kasperkiewicz, B. (2010). Siliciclastic microstromatolites in a Hodge, A. T. (1981). Vitruvius, lead pipes and lead poisoning. American
sandstone cave: Role of trapping and binding of detrital particles Journal of Archaeology, 85, 486–491.
in formation of cave deposits. Annales Societatis Geologorum Hodge, A. T. (1992). Roman aqueducts and water supply. Bristol Classical
Poloniae, 80, 303–314. Press.
Grewe, K. (1982). Wo ist der Kalksinter der römischen Wasserleitung Hoffmann, G., Kummer, S., Márquez, R., & Valdivia Manchego, M. (2019). The
geblieben? Frontinus‐Mitteilung, 13, 1–2. Roman Eifel aqueduct: Archaeoseismological evidence for neotectonic
Grewe, K. (1986). Atlas der römischen Wasserleitungen nach Köln. Rhein- movement at the transition of the Eifel to the lower Rhine embayment.
land Verlag. International Journal of Earth Sciences, 108, 2349–2360.
Grewe, K. (1991). Aquädukt‐Marmor. Kalksinter der römischen Eifelwasser- Holland, H. D., Holland, H. J., & Munoz, J. L. (1964). The coprecipitation of
leitung als Baustoff des Mittelalters. Bonner Jahrbuch, 191, 278–343. cations with CaCO3‐II. The coprecipitation of Sr+2 with calcite between
Grewe, K. (1992). Aquädukt‐Marmor. Konrad Verlag. 90° and 100°C. Geochimica et Cosmochimica Acta, 28, 1287–1301.
Grewe, K. (1996). Aquädukt‐Marmor der römischen Eifelwasserleitung als Hostetter, E., Fouke, B. W., & Lundstrom, C. C. (2011). The last flow of water
Schmuckstein des Mittelalters. Bericht über die 38. Tagung für to, and through, the baths of Caracalla: Age, temperature and chemistry.
Ausgrabungswissenschaft und Bauforschung der Koldewey‐ Journal of Ancient Topography. Rivista di Topografia Antica, 21, 53–90.
Gesellschaft vom 11. Bonn. Huang, Y., Fairchild, I. J., Borsato, A., Frisia, S., Cassidy, N. J.,
Grewe, K., & Blackman, D. (2001). Excursion on calx. In D. McDermott, F., & Hawkesworth, C. J. (2001). Seasonal variations
Blackman, & A. T. Hodge (Eds.), Frontinus' legacy (pp. in Sr, Mg and P in modern speleothems (Grotta di Ernesto, Italy).
109–115). The University of Michigan Press. Chemical Geology, 175, 429–448.
Grohmann, A. (1978). Chemie und Sinterbildung. Das Rheinische Huntington, K. W., & Petersen, S. V. (2023). Frontiers of carbonate
Landesmuseum Bonn, 6, 91. clumped isotope thermometry. Annual Review of Earth and Planetary
Guendon, J.‐L. (2014). Les concrétions du canal d'amenée d'eau au moulin: Sciences, 51, 611–641.
examen pétrographique. In J.‐P. Brun & M. Leguilloux, Les installa- Jacono, L. (1938). La ruota idraulica romana di Venafro. L'ingegnere, 12,
tions artisanales Romaines de Saepinum. Tannerie et moulin hydrauli- 850–853.
que. Publications du Centre Jean Bérard. Jones, B. (2010). Microbes in caves: Agents of calcite corrosion and
Guendon, J.‐L., Huon, S., Parron, C., & Bonté, S. (1998). Les concrétions precipitation (Vol. 336, pp. 7–30). Geological Society.
calcaires, témoins du fonctionnement de l'aqueduc. In C. Gébara, I. Joseph, C., Gilly, J.‐C., & Rodier, C. (2000). Système hydrique de l'aqueduc
Bérand, & L. Rivet (Eds.), Fréjus antique, guide archéologique de la et genèse des concrétions. In G. Fabre, J.‐L. Fiches, & J.‐L. Paillet
France (pp. 163–215). Imprimerie nationale. (Eds.), L'aqueduc de Nîmes et le Pont du Gard (2nd ed., pp. 225–231).
Guendon, J.‐L., Huon, S., Parron, C., & Bonté, S. (2002). Les concrétions CNRS Editions.
calcaires, témoins du fonctionnement de l'aqueduc. In C. Gébara, & Kandianis, M. T., Fouke, B. W., Johnson, R. W., Veysey, J., & Inskeep, W. P.
J.‐M. Guendon (Eds.), L'aqueduc romain de Fréjus. Sa description, son (2008). Microbial biomass: A catalyst for CaCO3 precipitation in
histoire et son environnement (Vol. 33, pp. 87–105). Éditions de advection‐dominated transport regimes. Geological Society of
l'Association de la Revue Archéologique de la Narbonnaise. America Bulletin, 120, 442–450.
Guendon, J.‐L., & Leveau, P. (2005). Dépôts carbonatés et fonctionnement Keenan‐Jones, D. (2015). Somma‐Vesuvian ground movements and the
des aqueducs romains. Gallia, 62, 87–96. water supply of Pompeii and the Bay of Naples. American Journal of
Guendon, J.‐L., Parron, C., Huon Gébara, C., & Michel, J.‐M. (1994). Archaeology, 119, 191–215.
Premiers résultats de l'étude géochimique des concrétions calcaires Keenan‐Jones, D., Hellstrom, J., & Drysdale, R. (2008). Trace element and
de l'aqueduc romain de Fréjus. Bulletin Archeologique de Provence, other analyses of tufa from ancient water systems in Campania and
23, 66–80. Petra. Cura aquarum in Jordanien. Schriften der DWhG, 12, 329–339.
Guendon, J.‐L., & Vaudour, J. (1986). Les concrétions de l'aqueduc de Keenan‐Jones, D., Hellstrom, J., & Drysdale, R. (2011). Lead contamination in
Nîmes. In G. Fabre, J.‐L. Fiches, & J.‐L. Paillet (Eds.), Travertins LS et the drinking water of Pompeii. In E. Poehler, M. Flohr, & K. Cole (Eds.),
évolution des paysages holocènes dans le domaine méditerranéen (Vol. Pompeii: Art, industry and infrastructure (pp. 131–148). Oxbow Books.
57, pp. 140–151). CNRS. Keenan‐Jones, D., Motta, D., Foubert, A., Fried, G., Sivaguru, M., Perillo, M.,
Guendon, J.‐L., & Vaudour, J. (2000). Concrétions et fonctionnement de Waldsmith, J., Wang, H., Garcia, M. H., & Fouke, B. W. (2014).
l'aqueduc: Étude morpho‐stratigraphique. In G. Fabre, J.‐L. Fiches, & Hierarchical stratigraphy of travertine deposition in ancient Roman
J.‐L. Paillet (Eds.), L'aqueduc de Nîmes et le Pont du Gard (2nd ed., pp. aqueducts. Archaeological Institute of America Annual Meeting, Seattle
233–248). CNRS Éditions. (pp. 293–295).
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
86 | SÜRMELIHINDI and PASSCHIER

Keenan‐Jones, D., Motta, D., Garcia, M. H., & Fouke, B. W. (2015). Mahapatra, A., Padhi, N., Mahapatra, D., Bhatt, M., Sahoo, D., Jena, S.,
Travertine‐based estimates of the amount of water supplied by Dash, D., & Chayani, N. (2015). Study of biofilm in bacteria from water
ancient Rome's Anio Novus aqueduct. Journal of Archaeological pipelines. Journal of Clinical and Diagnostic Research, 9, DC09–DC11.
Science: Reports, 3, 1–10. Maier, F. B. (1998). Vindonissa: Rückblick auf die Feldarbeiten im Jahr
Keenan‐Jones, D., Motta, D., Garcia, M. H., Sivaguru, M., Perillo, M., 1997. Gesellschaft Pro Vindonissa. Jahresbericht, 1998, 77–85.
Shosted, R. K., & Fouke, B. W. (2022). Travertine crystal growth Marín‐Roldán, A., Cruz, J. A., Martín‐Chivelet, J., Turrero, M. J.,
ripples record the hydraulic history of ancient Rome's Anio Novus Ortega, A. I., & Cáceres, J. O. (2014). Evaluation of laser induced
aqueduct. Scientific Reports, 12, 1239. breakdown spectroscopy (LIBS) for detection of trace element
Kele, S., Breitenbach, S. F. M., Capezzuoli, E., Meckler, A. N., Ziegler, M., variation through stalagmites: Potential for paleoclimate series
Millan, I. M., Kluge, T., Deák, J., Hanselmann, K., John, C. M., Yan, H., reconstruction. Journal of Applied and Laser Spectroscopy, 1, 7–12.
Liu, Z., & Bernasconi, S. M. (2015). Temperature dependence of Marra, F., Montone, P., Pirro, M., & Boschi, E. (2004). Evidence of active
oxygen‐ and clumped isotope fractionation in carbonates: A study of tectonics on a Roman aqueduct system (II–III century A.D.) near
travertines and tufas in the 6–95°C temperature range. Geochimica Rome, Italy. Journal of Structural Geology, 26, 679–690.
et Cosmochimica Acta, 168, 172–192. Matsui, S., Sorrentino, L., Sakai, S., Shimizu, Y., & Iorio, V. (2009). La
Kessener, P. (1995). The entrance channel of the castellum divisorium at provenienza dell'acqua potabile nell' antica Pompei: Un'ipotesi
Nîmes. Babesch, 70, 179–191. basata sull' analisi chimica dei residui calcarei degli impianti idrici.
Kessener, P. (2000a). Incrustations at the Castellum Divisorium at Nîmes. Journal of Fasti Online, 1–10.
In G. C. M. Jansen (Ed.), Cura aquarum in Sicilia. (pp. Mays, L. W. (2010). Ancient water technologies. Springer.
169–177). Peeters Publishers. McConnaughey, T. (1991). Calcification in Chara corallina: CO2 hydroxyl-
Kessener, P. (2000b). The aqueduct at Aspendos and its inverted siphon. ation generates protons for bicarbonate assimilation. Limnology and
Journal of Roman Archaeology, 13, 104–132. Oceanography, 36, 619–628.
Kessener, P. (2001). Vitruvius and the conveyance of water. Babesch, 76, Meghraoui, M., Gomez, F., Sbeinati, R., Van der Woerd, J., Mouty, M.,
139–158. Darkal, A. N., Radwan, Y., Layyous, I., Al Najjar, H., Darawcheh, R.,
Kessener, P. (2016). The Aspendos Siphon and Roman Hydraulics. Hijazi, F., Al‐ Ghazzi, R., & Barazangi, M. (2003). Evidence for 830
Babesch, 27, 261–274. years of seismic quiescence from palaeoseismology, archaeoseis-
Kessener, P. (2019). The discharge of the Değirmendere Aqueduct. mology and historical seismicity along the Dead Sea fault in Syria.
In G. Wiplinger (Ed.), Der Değirmendere Aquädukt von Ephesos Earth and Planetary Science Letters, 210, 35–52.
(pp. 525–534). Peeters Publishers. Meheruna, A., & Akagi, T. (2006). Role of fine roots in the plant‐induced
Kessener, P. M. (2022). Roman water transport: Pressure lines. Water, 14, 28. weathering of andesite for several plant species. Geochemical
Kluge, T., John, C. M., Boch, R., & Kele, S. (2018). Assessment of Journal, 40, 57–67.
factors controlling clumped isotopes and δ 18 O values of Melim, L. A., & Spilde, M. N. (2011). Rapid growth and recrystallization
hydrothermal vent calcites. Geochemistry, Geophysics, of cave pearls in an underground limestone mine. Journal of
Geosystems, 19, 1844–1858. Sedimentary Research, 81, 775–786.
Langmuir, D. (1997). Aqueous environmental geochemistry. Prentice Hall. Melim, L. A., & Spilde, M. N. (2021). The rise and fall of cave pearl pools:
Leveau, P. (1995). Les moulins romains de Barbegal, les ponts‐aqueducs Highly variable growth, recrystallization and demise of a mine pearl
du vallon des Arcs et l'histoire naturelle de la Vallée des Baux (Bilan site. Sedimentology, 68, 2165–2194.
de six ans de fouilles programmées). Comptes Rendus des Séances de Merz, M. U. E. (1992). The biology of carbonate precipitation by
l'Academie des Inscriptions et Belles‐Lettres, 139, 115–144. Cyanobacteria. Facies, 26, 81–101.
Levret, A., Volant, P., Carbon, D., Combescure, D., Verdel, T., Piant, A., Michel, J.‐M. (2002). Le canal: Caractéristiques et particularités, in
Scotti, O., & Laurent, P. (2008). L'aqueduc romain de Nîmes (France): L'aqueduc romain de Fréjus. Sa description, son histoire et
Nouveaux résultats en archéosimicité. In A. Levret (Ed.), Archéosismicité son environnement. In C. Gébara, J.‐M. Michel, & J.‐L. Guendon
et Vulnérabilité: Patrimoine bâti et société (pp. 165–188). Groupe APS. (Eds.), Revue Archéologique de la Narbonnaise supplément (Vol. 33,
Lieber, W. (1990a). Kalksinter‐Abscheidungen in Wasserleitungen, Der pp. 122–136). Montpellier.
Aufschluss, 41, 249–261. Mitrović‐Woodell, I., Tesei, T., Plan, L., Habler, G., Baroň, I., Grasemann, B.
Lieber, W. (1990b). Unliebsamer Wasser‐Härte. Kalksinter‐Bildungen in (2023). Deformation of columnar calcite within flowstone spe-
Wasserleitungen. Calcit, Baustein des Lebens, (pp. 98–110). leothem. Journal of Structural Geology, 174, 104924.
Münchner Mineralientage Fachmesse. Motta, D., Keenan‐Jones, D., Garcia, M. H., & Fouke, B. W. (2017).
Lockett, W. G. (1988). Discussion of interaction of flow and incrustation in Hydraulic evaluation of the design and operation of ancient Rome's
the Roman Aqueduct of Nîmes by George F. W. Hauck and Richard Anio Novus Aqueduct: Hydraulic evaluation of ancient Rome's Anio
A. Novak (February, 1987, Vol. 113, No. 2). Journal of Hydraulic Novus aqueduct. Archaeometry, 59, 1150–1174.
Engineering, 114, 464–465. Müller, W. (2000). Bildung von Sinterablagerungen in Wassersystemen. In
Lolos, Y., Passchier, C., & Sürmelihindi, G. (2018). Carbonate deposits in C. M. G. Jansen (Ed.). Cura aquarum in Sicilia Proceedings of the Tenth
Roman aqueducts as a data source: Examples from Corinthia and Argolis. International Congress on the history of water management and
In (Eds: E. Zimis, A. Vasiliki Karapanayiotou & M. Xanthopoulou) Tο hydraulic engineering in the Mediterranean region (pp. 185–189).
αρχαιολογικο εργο στην πελοποννησο (ΑΕΠΕΛ1). (The archeological Neely, J. A. (2001). A contextual study of the ‘fossilized' prehispanic canal
project in the Peloponneses—AEPEL1) Proceedings of the Interna- systems of the Tehuacan Valley, Puebla, Mexico. Antiquity, 75, 505–506.
tional Conference Tripoli (p. 915–921). Neely, J. A. (2017). The beginnings of water management and agricultural
Lolos, Y. A. (1997). The Hadrianic aqueduct of Corinth (with an appendix intensification in Mesoamerica: The case of the Prehistoric San
on the roman aqueducts in Greece). Hesperia, 66, 271–314. Marcos Well, the Purrón dam and the ‘Fossilized’ canal Systems of
Lombardi, L., Coates‐Stephens, R., & Barbieri, M. (2005). L'aquedotto the Tehuacán Valley, Puebla, Mexico. In N. Sanz (Ed.), The origins
antoniniano: l'alimentazione idrica delle Terme di Caracalla. In I. I. of food production/los orígenes de la producción de alimentos
Riera (Ed.), In binos actvs lvmina. Rivista di studi e ricerche sull'idraulica (pp. 146–159). UNESCO Publishing.
storica e al storia della tecnica (pp. 211–220). Atti del Convegno Neely, J. A., Aiuvalasit, M. J., & Winsborough, B. M. (2022). Relict canals of
Internazionale di Studi Archeologia e società. L' idraulica degli the Tehuacán Valley, Mexico: A Middle‐ to Late‐Holocene dryland
Anticha fra Passato e Futuro. socio‐hydrological system. The Holocene, 32, 1422–1436.
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SÜRMELIHINDI and PASSCHIER | 87

Nöggerath, J. (1858). Die Marmorgewinnung aus den römischen Pedley, H. M., & Rogerson, M. (2010). Tufas and speleothems: Unravelling the
Wasserleitungen in der preußischen Rheinprovinz. Westermann's microbial and physical controls (p. 336). Geological Society of London.
Jahrbuch der illustrierten deutschen Monatshefte, 4, 165–171. Pedley, M. (1992). Freshwater (phytoherm) reefs: The role of biofilms and their
Ohlig, C. P. J. (2001). De Aquis Pompeiorum. Das Castellum Aquae in Pompeii: bearing on marine reef cementation. Sedimentary Geology, 79, 255–274.
Herkunft, Zuleitung und Verteilung des Wassers (Circumvesuviana) Pedley, M., Rogerson, M., & Middleton, R. (2009). Freshwater calcite
(Vol. 4). Books On Demand. precipitates from in vitro mesocosm flume experiments: A case for
Oleson, J. P. (1997). The origins and design of Nabataean water‐supply biomediation of tufas. Sedimentology, 56, 511–527.
systems. In K. Amr, F. Zayadine, & M. Zaghloul (Eds.), Studies in the Pentecost, A. (2005). Travertine. Springer‐Verlag.
history and archaeology of Jordan (Vol. 5, pp. 707–719). Perrette, Y., Delannoy, J.‐J., Desmet, M., Lignier, V., & Destombes, J.‐L.
Oleson, J. P. (2018). Strategies for water supply in Arabia Petraea during (2005). Speleothem organic matter content imaging. The use
the Nabataean through Early Islamic Periods: Local adaptations of of a fluorescence index to characterise the maximum emission
the regional ‘technological shelf’. Berlin Studies of the Ancient World, wavelength. Chemical Geology, 214, 193–208.
53, 17–42. Pons‐Branchu, E., Bergonzini, L., Tisnérat‐Laborde, N., & Branchu, P.
Öziş, Ü. (1994). Historische Wasserbauten in Anatolien, ein Fundstellen (2018). 14C in urban secondary carbonate deposits: A new tool for
und Literatuurverzeichnis. Schriftenreihe der Frontinus‐Gesellschaft, environmental study. Radiocarbon 60, 1269–1281.
18, 89–108. Puliti, C., Borgioli, A., & Tezano, C. (1986). Studio chimico‐fisico su
Paillet, J.‐L. (1986). L'utilisation des concrétions de l'aqueduc de Nimes en formazioni calcaree prelevate da antichi acquedotti romani. Il Trionfo
tant que matériau de construction. Méditerranée, 57, 152–160. Dell'Acqua, 195–198.
Paillet, J.‐L., & Gilly, J.‐C. (2000). Une deuxième vie de l'aqueduc ou le Puliti, C., Borgioli, A., & Terzano, C. (1992). I depositi calcarei nella
remploi de ses concrétions dans des monuments médiévaux. In G. identificazione di opere idrauliche antiche. In A. M. Liberati Silverio, &
Fabre, J.‐L. Fiches, & J.‐L. Paillet (Eds.), L'aqueduc de Nîmes et le Pont G. Pisani Sartorio (Eds.), Il trionfo dell'acqua. Gli antichi acquedotti di Roma:
du Gard (2nd ed., pp. 423–441). CNRS Editions. problemi di conoscenza, conservazione e tutela (pp. 229–234). Comune
Passchier, C., Rigal, D., & Sürmelihindi, G. (2015). Preuves du nettoyage di Roma.
des concrétions calcaires de l'aqueduc antique de Divona‐Cahors. Riding, R. (2000). Microbial carbonates: The geological record of calcified
In L. L. Borau, & A. Borlenghi (Eds.), Aquae Ductus (Vol. 33, bacterial—Algal mats and biofilms. Sedimentology, 47, 179–214.
pp. 233–241). Aquitania Supplement. Rigal, D., Passchier, C. W., & Sürmelihindi, G. (2023). Origines,
Passchier, C., Sürmelihindi, G., Boyer, D., Yalçın, C., Spötl, C., & Mertz‐Kraus, développement et preuves du nettoyage des concrétions calcaires
R. (2021). The aqueduct of Gerasa—Intra‐annual palaeoenvironmental durant le fonctionnement de l'aqueduc gallo‐romain de Divona‐
data from Roman Jordan using carbonate deposits. Palaeogeography, Cahors (Lot). Archéologie, Société et Environnement, 1, 1–7.
Palaeoclimatology, Palaeoecology, 562, 110089. Roberts, M. S., Smart, P. L., & Baker, A. (1998). Annual trace element
Passchier, C., Sürmelihindi, G., Spötl, C., Mertz‐Kraus, R., & Scholz, D. variations in a Holocene speleothem. Earth and Planetary Science
(2016). Carbonate deposits from the ancient aqueduct of Béziers, Letters, 154, 237–246.
France—A high‐resolution palaeoenvironmental archive for the Rodier, C., Joseph, C., & Gilly, J.‐C. (2000). Étude à la microsonde de la
Roman Empire. Palaeogeography, Palaeoclimatology, Palaeoecology, géochemie de concrétions internes dans l'aqueduc à Bezouce. In G.
461, 328–340. Fabre, J.‐L. Fiches, & J.‐L. Paillet (Eds.), L'aqueduc de Nîmes et le Pont
Passchier, C. W., Bourgeois, M., Viollet, P.‐L., Sürmelihindi, G., Bernard, V., du Gard (2nd ed., pp. 262–271). CNRS Editions.
Leveau, P., & Spötl, C. (2020). Reconstructing the hydraulics of the Rogerson, M., Pedley, H. M., Wadhawan, J. D., & Middleton, R. (2008). New
world's first industrial complex, the second century CE Barbegal insights into biological influence on the geochemistry of freshwater
watermills, France. Scientific Reports, 10, 17917. carbonate deposits. Geochimica et Cosmochimica Acta, 72, 4976–4987.
Passchier, C. W., & Sürmelihindi, G. (2012). Sinter deposits in Roman Sabri, R., Merkel, B., & Tichomirowa, M. (2015). Has the water supply
aqueducts. Aouras, 6, 273–290. network of Sebestia been connected to that of Nablus? Freiberg
Passchier, C. W., & Sürmelihindi, G. (2015). Geochemical observations on Online Geoscience, 41, 46–64.
sinter of Ephesos. In S. Wefers (Ed.), Die Mühlenkaskade von Ephesos Sabri, R. N. R. (2016). Geochemical and isotope investigations of
(pp. 120–123). RGZM. carbonate sinter—2000 years of water supply management in
Passchier, C. W., & Sürmelihindi, G. (2019). Carbonate deposits of the Palestine. Freiberg Online Geoscience, 47, 1–127.
Değirmendere Aqueduct. In G. Wiplinger (Ed.), Der Değirmendere Sbeinati, M., Meghraoui, M., Suleyman, G., Gomez, F., Grootes, P.,
Aquädukt von Ephesos (Vol. 36, pp. 511–522). Babesch Supplements. Nadeau, M., Al Najjar, H., & Al Ghazzi, R. (2010). Timing of
Passchier, C. W., & Sürmelihindi, G. (2023). Arqueología de las earthquake ruptures at the Al Harif Roman aqueduct (Dead Sea
concreciones calcáreas en sistemas de agua antiguas/Archaeologi- fault, Syria) from archaeoseismology and paleoseismology. Geological
cal information obtained from carbonate deposits in ancient water Society of America, 471, 243–267.
systems. Libro de ponencias, VI Congreso Internacional de Scardozzi, G. (2007). Hierapolis di Frigia, Applicazioni informatiche alle
Ingeniería Romana. ricognizione archeologiche e telerilevamento da satellite; l'és-
Passchier, C. W., Sürmelihindi, G., & Spötl, C. (2016). A high‐resolution sempio degli acquedotti della città. Archeologia e Calcolatori, 18,
palaeoenvironmental record from carbonate deposits in the Roman 331–353.
aqueduct of Patara, SW Turkey, from the time of Nero. Scientific Schmitz, W. (1978). Kalksinter im Römerkanal Zur Sinterbildung der
Reports, 6, 28704. Eifelwasserleitung. Berichte aus der Arbeit des Museums, 4, 55–57.
Passchier, C. W., & Trouw, R. A. J. (2005). Microtectonics (p. 289). Springer. Schroyen, K., Vermoere, M., Librecht, I., Degryse, P., Muchez, P.,
Passchier, C. W., Wiplinger, G., Güngör, T., Kessener, P., & Sürmelihindi, G. Viaene, W., Smets, E., Paulissen, E., Keppens, E., & Waelkens, M.
(2013). Normal fault displacement dislocating a Roman aqueduct of (2000). Preliminary study of travertine deposits in the vicinity of
Ephesos, Western Turkey. Terra Nova, 25, 292–297. Sagalassos: Petrography, geochemistry, geomorphology and paly-
Passchier, C. W., Wiplinger, G., Sürmelihindi, G., Kessener, P., & Güngör, T. nology. Sagalassos V. Report on the survey and excavation
(2012). Aqueducts as indicators of historically active faults in campaigns of 1996 and 1997. Acta Archaeologica Lovaniensia
the Mediterranean Basin. 2nd INQUA‐IGCP‐567 International Monographiae, 11, 757–781.
Workshop on active tectonics. Earthquake Geology, Archaeology Schulz, D. (1986). Schichtungen im Kalksinter der römischen Wasserlei-
and Engineering, 186–189. tung nach Köln. Eine Hilfe zur relativen Datierung. In K. Grewe (Ed.),
15206548, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/gea.21980 by Arizona State University Acq & Analysis, Lib Continuations, Wiley Online Library on [13/12/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
88 | SÜRMELIHINDI and PASSCHIER

Atlas der römischen Wasserleitungen nach Köln (pp. 263–268). Ventura, A. (2002). Los acueductos romanos de Cordoba y su rehabilitación
Rheinland Vlg. Köln. Omeya. Empuries, 53, 113–128.
Shiraishi, F., Reimer, A., Bissett, A., de Beer, D., & Arp, G. (2008). Microbial Verrez, P. (1985). Contribution à l'étude des dépôts carbonatés de l'aqueduc
effects on biofilm calcification, ambient water chemistry and stable romain du Pont du Gard (Mémoire de D.E.A.). Université de
isotoperecords in a highly supersaturated setting (Westerhöfer Bach, Montpellier, USTL.
Germany). Palaeogeography, Palaeoclimatology, Palaeoecology, 262, Veyrac, A. (1992). Contribution à l'étude hydraulique du castellum aquae de
91–106. Nîmes. Archéologie de la France—Informations.
Shopov, Y. Y., Ford, D. C., & Schwarcz, H. P. (1994). Luminescent Vince, A. (2006). Assessment of the Roman Aqueduct pipe from
microbanding in speleothems—High‐resolution chronology and Nettleham Road, Lincoln (NERL‐06). AVAC Reports, 2006, 127.
paleoclimate. Geology, 22, 407–410. Viollet, P.‐L. (2007). Water engineering in ancient civilisations. 5000 Years of
Sivaguru, M., Fouke, K. W., Keenan‐Jones, D., Motta, D., Garcia, M. H., & history. IAHR.
Fouke, B. W. (2022). Depositional and diagenetic history of Volant, P., Levret, A., Carbon, D., Scotti, O., Combescure, D., Verdel, T.,
travertine deposited within the Anio Novus aqueduct of ancient Piant, A., & Laurent, P. (2009). An archaeo‐seismological study of the
Rome. Geological Society of America Special Paper, 557, 541–569. Nîmes Roman Aqueduct, France: Indirect evidence for an M > 6
Smith, G. K. (2016). Calcite straw stalactites growing from concrete seismic event? Natural Hazards, 49, 53–77.
structures. Cave and Karst Science, 43, 4–10. Wenz, S., Scholz, D., Sürmelihindi, G., Passchier, C. W., Jochum, K. P., &
Spiro, B., & Pentecost, A. (1991). One day in the life of a stream: A diurnal Andreae, M. O. (2016). 230Th/U dating of carbonate deposits from
inorganic carbon mass balance for a travertine‐depositing stream ancient aqueducts. Quaternary Geochronology, 32, 40–52.
(waterfall beck, Yorkshire). Geomicrobiology Journal, 9, 1–11. White, W. (2015). Chemistry and karst. Acta Carsologica, 44, 349–362.
Sürmelihindi, G. (2013). Roman aqueducts and calcareous sinter deposits as White, W. B. (1976). Cave minerals and speleothems. In T. D. Ford, & T. D.
a proxy for environmental changes (PhD thesis). University of Mainz. H. Cullingsham (Eds.), The science of speleology (pp. 267–327).
Sürmelihindi, G., Leveau, P., Spötl, C., Bernard, V., & Passchier, C. W. Academic Press.
(2018). The 2nd century AD Roman Watermills of Barbegal: Wikander, Ö. (2000). Handbook of ancient water technology: Technology
Unravelling the enigma of one of the oldest industrial complexes. and change in history (Vol. 2). Brill.
Science Advances, 4(9), eaar3620. Williams, P. W., & Fowler, A. (2002). Relationship between lime oxygen
Sürmelihindi, G., & Passchier, C. (2020). Testing the water: A survey of the isotopes in rainfall, cave percolation waters and speleothem calcite
water supply of Ancient Butrint (Buthrotum). Using carbonate at Waitomo, New Zealand. New Zealand Journal of Hydrology, 41,
deposits from the Aqueduct system. Schriftenreihe der Frontinus‐ 53–70.
Gesellschaft, 32, 63–91. Wilson, A. (1997). Water management and usage in Roman North Africa: Social
Sürmelihindi, G., Passchier, C., Crow, J., Spötl, C., & Mertz‐Kraus, R. (2021). and technological study (D. Phil. dissertation). University of Oxford.
Carbonates from the ancient world's longest aqueduct: A testament of Wilson, A. (1999). Deliveries extra urbem: Aqueducts and the countryside.
Byzantine water management. Geoarchaeology, 36, 643–659. Journal of Roman Archaeology, 12, 314–331.
Sürmelihindi, G., Passchier, C., & Leveau, P. (2020). Giving voice to the Wilson, A. (2000). The aqueducts of Italy and Gaul. Journal of Roman
incrustations from a Roman watermill complex at Barbegal: An Archaeology, 13, 597–604.
impression. L'eau dans tous ses états. Perceptions antiques. Wilson, R. J. A. (1996). Tot aquarum tam multis necessariis molibus.
Archéologies Méditerranéennes, 161–171. Recent studies on aqueducts and water supply. Journal of Roman
Sürmelihindi, G., & Passchier, C. W. (2013). Sinter analysis—A tool for the Archaeology, 9, 5–29.
study of ancient Aqueducts. In G. Wiplinger, Historische Wasserlei- Winsborough, B. M., Caran, S. C., Neely, J. A., & Valastro, Jr., S. (1996).
tungen Gestern—Heute—Morgen (Vol. 49, pp. 269–287). Babesch Calcified microbial mats date prehistoric canals—Radiocarbon assay
Supplement 24/ÖAI Sonderschriften. of organic extracts from travertine. Geoarchaeology, 11, 37–50.
Sürmelihindi, G., & Passchier, C. W. (2014). A high‐resolution archive for Wiplinger, G. (2019). Der Degirmendere Aquädukt von Ephesos. Peeters.
palaeo‐environmental studies: calcareous sinter in ancient aque- Wróblewski, W., Gradziński, M., Motyka, J., & Stankovič, J. (2017).
ducts. In I. K. Kalavrouziotis, & A. N. Angelakis (Eds.), IWA Regional Recently growing subaqueous flowstones: Occurrence, petrography,
Symposium on water, wastewater and environment: Traditions and and growth conditions. Quaternary International, 437, 84–97.
culture (pp. 1331–1341). Hellenic Open University. Zhang, D. D., Zhang, Y., Zhu, A., & Cheng, X. (2001). Physical mechanisms
Sürmelihindi, G., Passchier, C. W., Baykan, O. N., Spötl, C., & Kessener, P. of river waterfall tufa (travertine) formation. Journal of Sedimentary
(2013). Environmental and depositional controls on laminated freshwater Research, 71, 205–216.
carbonates: An example from the Roman aqueduct of Patara, Turkey.
Palaeogeography, Palaeoclimatology, Palaeoecology, 386, 321–335.
Sürmelihindi, G., Passchier, C. W., Leveau, P., Spötl, C., Bourgeois, M., &
SUPP ORTING INFO RM ATION
Bernard, V. (2019). Barbegal: Carbonate imprints give a voice to the
first industrial complex of Europe. Journal of Archaeological Research: Additional supporting information can be found online in the
Reports, 24, 1041–1058. Supporting Information section at the end of this article.
Sürmelihindi, G., Passchier, C. W., Rigal, D., Wilson, A., & Spötl, C. (2023).
Roman aqueduct maintenance in the water supply system of Divona,
France. Scientific Reports, 13, 12035.
Sürmelihindi, G., Passchier, C. W., Spötl, C., Kessener, P., Bestmann, M., How to cite this article: Sürmelihindi, G., & Passchier, C.
Jacob, D. E., & Baykan, O. N. (2013). Laminated carbonate deposits (2024). Writ in water—Unwritten histories obtained from
in Roman aqueducts. Sedimentology, 60, 961–982.
carbonate deposits in ancient water systems. Geoarchaeology,
Urai, J. L., Williams, P. F., & van Roermund, H. L. M. (1991). Kinematics of
crystal growth in syntectonic fibrous veins. Journal of Structural 39, 63–88. https://doi.org/10.1002/gea.21980
Geology, 13, 823–836.

You might also like