This article was downloaded by: [Murdoch University Library]On: 18 June 2015, At: 18:28Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Click for updates

Isotopes in Environmental and HealthStudiesPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/gieh20

Tracing formation and durability ofcalcite in a Punic–Roman cistern mortar(Pantelleria Island, Italy)Martin Dietzela, Frerich Schönb, Jens Heinrichsc, Artur P.Deditiusad & Albrecht Leise

a Institute of Applied Geosciences, Graz University of Technology,Graz, Austriab Institute for Classical Archaeology, University of Tübingen,Germanyc Karlsruhe University of Applied Science, Karlsruhe, Germanyd School of Engineering and Information Technology, MurdochUniversity, Murdoch, Australiae Institute of Water, Energy and Sustainability, JoanneumResearch, Graz, AustriaPublished online: 11 Mar 2015.

To cite this article: Martin Dietzel, Frerich Schön, Jens Heinrichs, Artur P. Deditius & Albrecht Leis(2015): Tracing formation and durability of calcite in a Punic–Roman cistern mortar (PantelleriaIsland, Italy), Isotopes in Environmental and Health Studies, DOI: 10.1080/10256016.2015.1016430

To link to this article: http://dx.doi.org/10.1080/10256016.2015.1016430

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies, 2015http://dx.doi.org/10.1080/10256016.2015.1016430

Tracing formation and durability of calcite in a Punic–Romancistern mortar (Pantelleria Island, Italy)

Martin Dietzela∗, Frerich Schönb, Jens Heinrichsc, Artur P. Deditiusa,d and Albrecht Leise

aInstitute of Applied Geosciences, Graz University of Technology, Graz, Austria; bInstitute for ClassicalArchaeology, University of Tübingen, Germany; cKarlsruhe University of Applied Science, Karlsruhe,

Germany; dSchool of Engineering and Information Technology, Murdoch University, Murdoch, Australia;eInstitute of Water, Energy and Sustainability, Joanneum Research, Graz, Austria

(Received 20 July 2014; accepted 19 January 2015)

Dedicated to Professor Dr. Jochen Hoefs on the occasion of his 75th birthday

Ancient hydraulic lime mortar preserves chemical and isotopic signatures that provide important infor-mation about historical processing and its durability. The distribution and isotopic composition of calcitein a mortar of a well-preserved Punic–Roman cistern at Pantelleria Island (Italy) was used to trace theformation conditions, durability, and individual processing periods of the cistern mortar. The analyses ofstable carbon and oxygen isotopes of calcite revealed four individual horizons, D, E, B-1 and B-2, ofmortar from the top to the bottom of the cistern floor. Volcanic and ceramic aggregates were used for theproduction of the mortar of horizons E/D and B-1/B-2, respectively. All horizons comprise hydraulic limemortar characterized by a mean cementation index of 1.5 ± 1, and a constant binder to aggregate ratio of0.31 ± 0.01. This suggests standardized and highly effective processing of the cistern.

The high durability of calcite formed during carbonation of slaked lime within the matrix of the ancientmortar, and thus the excellent resistance of the hydraulic lime mortar against water, was documented by (i)a distinct positive correlation of δ18Ocalcite and δ13Ccalcite; typical for carbonation through a mortar hori-zon, (ii) a characteristic evolution of δ18Ocalcite and δ13Ccalcite through each of the four mortar horizons;lighter follow heavier isotopic values from upper to lower part of the cistern floor, and (iii) δ18Ocalcitevarying from − 10 to − 5 ‰ Vienna Pee Dee belemnite (VPDB). The range of δ18Ocalcite values rule outrecrystallization and/or neoformation of calcite through chemical attack of water stored in cistern.

The combined studies of the chemical composition of the binder and the isotopic composition of thecalcite in an ancient mortar provide powerful tools for elucidating the ancient techniques and processingperiods. This approach helps to evaluate the durability of primary calcite and demonstrates the importanceof calcite as a proxy for chemical attack and quality of the ancient inorganic binder.

Keywords: calcite; carbon-13; hydraulic lime mortar; hydrogen-2; isotope fractionation: isotope geo-chemistry; oxygen-18; Pantelleria; Punic–Roman cistern

1. Introduction

Lime is one of the most common and important inorganic binders used since the Neolithic era[1,2]. In spite of the exposure of lime mortar to physico-chemical weathering, it is preservedthroughout millennia. However, the reasons for the relatively high durability of lime mortar

*Corresponding author. Email: martin.dietzel@tugraz.at

© 2015 Taylor & Francis

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

2 M. Dietzel et al.

remain poorly understood [3–5]. Knowledge about the durability of the ancient lime mortar, inparticular historical processing and performance, is important for developing advanced, tailoredinorganic binders in the fields of material science, restoration, architecture, and archaeology [6].

Commonly, hydraulic lime mortar is used for the preparation of water resistant constructionslike cisterns to hold bricks and stones together as well as to line cistern walls against water ifapplied as render. Hydraulic lime mortar is a composite material composed of a binder-formingmaterial, which consists of slaked lime and latent hydraulic components, and aggregates such asfragments of rocks and/or synthetic materials. In water, the setting of the hydraulic lime mortaris caused by ‘activation’ of the latent hydraulic components under alkaline conditions. The latenthydraulic components comprise alkali-rich volcanic tuffs (cretoni, pozzolan), crushed ceram-ics or separately calcined kaolin [7,8]. As a result of the ‘activation’ process, calcium silicate(C-S-H), calcium aluminate (C-A-H), and calcium ferrite hydrates (C-F-H) form a water resis-tant binder. The excess of calcium oxide reacts with water and carbon dioxide to form CaCO3

[9]. The low solubility of these hydrates results in high performance of hydraulic binders andthus their utilization as constructing material for cisterns, aqueducts, and duct drains. The C-S-H, C-A-H, and C-F-H phases of hydraulic lime mortar are difficult to characterize due to theirlow degree of crystallinity, small size of individual grains, and complex chemical composition[9]. On the other hand, crystalline calcium carbonate precipitates in the binder can be easily iden-tified using diffraction methods, in the absence of other carbonate-containing aggregates, suchas limestone and/or dolostone.

The information about the chemical composition and distribution of stable carbon and oxygenisotopes of the calcium carbonate binder is a powerful tool to elucidate the environmental condi-tions of its formation and subsequent transformation [3,10–16]. However, the combined isotopicand micro-chemical investigations of an entire profile of hydraulic lime mortar are sparse. Inthe present contribution, the changes in mineralogical, chemical, and isotopic composition ofhydraulic lime mortar along a vertical profile of the cistern from Pantelleria Island (Italy) arestudied. The 13C/12C and 18O/16O ratios of the mortar were investigated to trace the formationconditions, individual processing periods, and alteration of the calcium carbonate bearing binder.

2. Site description

The volcanic island of Pantelleria is located within the Mediterranean Sea about 110 km south-west of Sicilia, and 70 km east of the Tunisian coast (Figure 1). Pantelleria was inhabited sincethe early Neolithic period (5000 B.C.). The strategic geographical position of the Island betweenthe Eastern and Western part of the Mediterranean Sea increased its importance for the ancientcivilizations. Since the 8th century B.C., the Island was settled by Carthaginian or Punic people,whereas from 217 B.C. to 439 A.D., Pantelleria was embedded into the Roman Empire [17–21]. The latter period is known to be of high prosperity for Pantelleria, which is documented bylocally manufactured and widely distributed ceramic products called ‘Pantellerian ware’, oftenused for archaeological dating [22,23].

Samples of mortar from the central cistern #1 of 65 m3 volume come from the main ancientsettling site Cossyra located about 1.5 km southeast of the ancient harbour (Figure 1). Archae-ological dating based on stratigraphic evidence, pottery chronology, and construction detailsindicates a utilization period of the cistern for water storage from about 300 B.C to 200 A.D.,which approximates the Punic and Roman eras [24–28]. Fifty ancient cisterns were found inthe acropolis area of the ancient town of Cossyra, among in total of 690 cisterns located on theIsland [29,30]. The cisterns were used for collection and storage of rainwater due to insuffi-cient resources of freshwater during the time of increasing population. The volcanic character

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies 3

Figure 1. (a) Pantelleria Island within the Mediterranean Sea. (b) Ancient settlement site Cossyra and the ancient har-bour located at Pantelleria Island. (c) Overview about the ‘acropolis’ excavation at Pantelleria Island. Ancient cisterns aremarked by blue dots. (d) 3D image of cistern #1 which has a length, width, and depth of 6.2, 2.0, and 5.3 m, respectively.The mortar sample was drilled from the cistern floor (black arrow).

of the island promotes mixing of meteoric water and seawater as well as contamination ofwater resources by intense water–rock interaction in the active geothermal systems of Pantel-leria. Therefore, this locally occurring water was of no use for domestic, manufacturing, andagriculture purposes [31]. The leached rocks comprise basalts and tuffs from distinct periods oferuption between 120 and 9 ka B.P. [32].

3. Methodology

Sampling of meteoric water from two nearby sound cisterns was carried out in May 2011 (#26and #30 in Figure 1; see details in Schön [27]). The mortar sample was drilled from the floor ofcistern #1 (core diameter: 10 cm). Cistern #1 was chosen for analyses because it was used forwater storage in ancient times and it showed complex processing history; that is, different mortarlayers, which are typically referred to as re-coating of cisterns to repair damage and/or to re-gaina waterproof performance [24,25].

Part of the drilling core was treated with a milling cutter (Metba, MB-1) to crush and separatebulk mortar at millimetre step; in total, 41 samples starting from the top surface of the cisternfloor. The fine-grained samples were dried at 40 °C and prepared for mineralogical, chemical,and isotopic analyses. Mineralogical composition of the powder samples was investigated bymeans of X-ray diffraction (XRD) using a Siemens D5000 with CuKα radiation, and collectionangles from 5 ° to 45 °2� (scanning step size: 0.02°2�; counting time: 4 s/step; conduction

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

4 M. Dietzel et al.

current: 40 mA; acceleration voltage: 40 kV). Thermogravimetric analyses (TGA) of the bulksamples were conducted from 30 to 1000 °C with a heating rate of 10 °C min−1 using a MettlerToledo TGA/SDTA851e under N2 atmosphere. The carbonate content (carbonated lime) of thebulk sample was calculated from the weight loss at reaction temperatures from 600 to 900 °Cwith an analytical inaccuracy < 5 %. Acidic treatment of the fine-grained bulk mortar was doneby reacting 100 mg of the sample with 5 mL of 10 % HCl for 5 min. This short-term acid treat-ment preferentially dissolves the binding matrix instead of the siliceous aggregates [5,33,34].The solutions were filtered through 0.45 µm cellulose acetate membranes. Subsequently, the Ca,Mg, Si, Al, and Fe concentrations of the filtered solutions were measured by inductively coupledplasma optical emission spectroscopy (ICP-OES) with a precision of better than ± 5% (PerkinElmer Optima ICP-OES 4300; Merck multi-element standard).

The other part of the drill core was embedded in blue coloured epoxy resin. Subsequently,polished thin and block sections were prepared using diamond paste < 0.1 µm for opticalmicroscopy and electron microprobe analyses (EMPA), respectively. Quantitative EMPA werecarried out utilizing Jeol JXA8200 with wavelength-dispersive spectroscopy. Due to the beamsensitive nature of the investigated material, the analytical conditions were as follow: the accel-erating voltage of 15 keV, beam current of 10 nA, and focused beam of ∼ 1 µm scanning overa raster of 4 µm × 6 µm. The single-spot analyses were calibrated against natural and syntheticstandards, and included following elements with characteristic spectral lines: Ca Kα and Mg Kα

(dolomite); Si Kα, Al Kα, and Fe Kα (almandine); the counting time for all elements was 20 s(s) on peak and 10 s on the background position on each side of the peak. Only the analyses thathave an error < 5% were considered. The lower analytical totals are due to the presence of CO2,H2O, and porosity (Table 1). The elemental mappings of 1000 × 1000 pixels resolution, 2 × 2mm in size, were collected using a focused beam ∼ 1 µm in size, and a dwell time of 20 ms perstep. The EMPA single-spot analyses refer to areas of the matrix of the cistern mortar (bindingmatrix), excluding aggregates.

The hydraulic property of the binder of the present mortar was evaluated by the cementationindex (CI) defined by Boynton [35] (CI = (2.8 SiO2 + 1.1 Al2O3 + 0.7 Fe2O3) / (CaO + 1.4MgO)). CI > 0.3 indicates a significant hydraulic property of a lime mortar, which correspondsto the formation of C-S-H, C-A-H, and C-F-H phases [36].

The stable carbon and oxygen isotope ratios of calcite in the fine-grained bulk samples weremeasured with a Thermo Fisher Scientific (Bremen, Germany) Gas Bench II carbonate prepa-ration device connected to a Finnigan DELTAplus XP isotope ratio mass spectrometer [37,38].Sample vessels were cleaned with diluted phosphoric acid, then rinsed three times with deionizedwater (18.2 M� cm, ELGA PURELAB Maxima), and dried overnight at 70 °C. An aliquot of1–6 mg of the powdered mortar sample, containing approximately 400 µg of calcite, consideringthe calcium carbonate concentration measured by TGA, was introduced in the Gasbench exe-tainers, sealed and flushed with helium gas to remove residual air. Prior to analyses, phosphoricacid was injected into the individual sample vials. Each sample was analysed twice. The samplesand international reference materials (NBS_19 and NBS_18) were simultaneously analysed. Theresults of carbon and oxygen isotopic composition are reported in the delta (δ) notation as theper mil (‰) deviation relative to the Vienna Pee Dee belemnite (VPDB) standard according tothe following equation:

δsample = Rsample − Rstandard

Rstandard× 1000, (1)

where for C and O isotopes R is 13C/12C and 18O/16O, respectively. The overall precision (2σ )for the measurements of δ13Ccalcite and δ18Ocalcite is ± 0.1 ‰ and ± 0.08 ‰, respectively.

Water samples from the cistern were analysed for δ2Hwater and δ18Owater values of H2O. Thestable isotopes of hydrogen were measured using a Finnigan DELTAplus XP mass spectrometer

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopesin

Environm

entalandH

ealthStudies

5

Table 1. Chemical composition of the binder (matrix of the mortar) from horizons E, D, B-1, and B-2 analysed along the profile of the cistern lining at the given depths (see Figure 1).

Horizonbulk Depths (mm) Calcite CaO(carb) CaO (n-carb) CaO(total) MgO SiO2 Al2O3 Fe2O3 B B/A CI

B-1 2–3 12.9 7.22 3.81 11.03 5.13 4.61 1.53 0.59 22.89 0.30 1.4B-2 11–12 13.5 7.55 4.33 11.88 7.34 2.47 1.57 0.51 23.77 0.31 0.6B-2 18–19 13.2 7.39 5.09 12.48 1.92 6.77 1.71 0.56 23.43 0.31 2.7E 29–30 15.2 8.51 4.82 13.33 4.22 4.03 1.02 1.41 24.00 0.32 1.2D 33–34 27.5 15.39 2.59 17.98 3.63 2.05 0.89 0.49 25.04 0.33 0.9spotB-1 0.1–5.1 5.47 8.06 12.65 5.91 2.73 2.6(n = 5) ± 3.11 ± 1.5 ± 8.55 ± 1.99 ± 2.75B-2 10.4–13.0 2.23 24.83 16.32 7.65 0.52 1.5(n = 7) ± 0.85 ± 1.62 ± 0.81 ± 2.99 ± 0.09B-2 16.1–21.5 2.88 22.85 15.69 8.99 0.79 1.6(n = 14) ± 1.91 ± 2.87 ± 2.50 ± 2.54 ± 0.88E 28.2–31.4 11.09 12.66 15.00 5.33 10.32 1.9(n = 9) ± 4.72 ± 3.45 ± 4.45 ± 1.59 ± 5.62D 31.8–34.0 7.80 16.16 16.89 6.18 1.48 1.8(n = 7) ± 2.71 ± 1.17 ± 2.34 ± 0.62 ± 0.73

Note: Values are given in wt%. The bulk values are based on acidic treatment and analyses of the leached fraction. The calcite content is measured by TGA (equivalent to CaO(carb); see Table 2). CaO(n-carb):non-carbonate based CaO calculated by total CaO concentration of the binder (CaO(total)) minus CaO(carb). B: sum of total concentration of oxides in the binder in wt%. B/A: binder to aggregate (A = 100 – B)ratio. The chemical composition of the matrix is also obtained by EMPA single-spot analyses (spot). CI: cementation index calculated according to Boynton [35] (see text).

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

6 M. Dietzel et al.

working in continuous flow mode by the chromium reduction technique [39]. The oxygen iso-topic composition was analysed with a Finnigan DELTAplus mass spectrometer using the classicCO2–H2O equilibrium method [40]. The δ2Hwater (analytical precision: ± 0.8 ‰) and δ18Owater

values ( ± 0.05 ‰) are given relative to Vienna Standard Mean Ocean Water (VSMOW).The oxygen isotope fractionation factor α between calcite and water is given by the following

equation:

αcalcite−water = (18O/16O)calcite/(18O/16O)water. (2)

At 10 °C, the value of 1000 ln αcalcite−water at isotopic equilibrium is 32.9 ‰ using theequation, 1000 ln αcalcite−water = 17.4 (1000/T) – 28.6 (T in Kelvin), revised by Coplen [41](considering former data sets, e.g. of Kim and O’Neil [42]). The 1000 ln αcalcite−water value( ≈�18Ocalcite−water = δ18Ocalcite – δ18Owater) is based on isotopic values relative to VSMOW,where the conversion of δ18Ocalcite values from VPDB to VSMOW standard is given by δ18O(VSMOW) = 1.03091 δ18O(VPDB) + 30.91 [43].

4. Results

4.1. Composition and microtexture of the cistern mortar

The textural analyses of the mortar sample revealed the presence of four horizons, D, E, B-1, andB-2, from the bottom to the top of the investigated profile of the cistern mortar (Figure 2). TheXRD analyses of the bulk mortar samples show that calcite, quartz, and feldspar are the dominantconstituents of all investigated samples from horizons D to B-2. In the mortar samples of thehorizons D and E, pyroxene and haematite are also frequently noted, whereas in the horizons

C

A D

B

Figure 2. Profile of the mortar sample drilled from the bottom of the cistern #1 (see Figure 1). (a) Reflected lightimage; (b–d) transmitted light images, 1 polarizer. White squares indicate the enlarged areas of transmitted light images ofhorizons E/B-2 (b and c) and B-1 (d). Fractures between distinct horizons are marked with white arrows. The arrowheadspoint to the exterior layer of horizon E.

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies 7

B-1 and B-2 mica are occasionally detected. The XRD analyses were inadequate to confirm thepresence of other inorganic binders such as less crystalline C-S-H or C-A-H phases, which areexpected as hydraulic reaction products.

The oldest horizon of the mortar, horizon D, occurs at the bottom of the cistern floor and ischaracterized by white to violet, and red colours (Figure 2(b)). Its composite material consists ofaggregates of basaltic rocks and slags of < 3 mm in diameter embedded in a fine-grained matrix.The matrix of this horizon is formed by a binder, which is suggested to be made of calcinedlimestone (quick lime) and fine-grained volcanic material [24,25]. The horizon D is separatedfrom the more reddish–brownish horizon E by a fracture, which width varies between 50 and200 µm. This kind of discontinuity typically occurs at the boundaries between two horizons (seewhite arrows in Figure 2(a) to (c)). In contrast to horizons D and E, where aggregates of volcanic

Table 2. Concentration (equivalent to CaO(carb) in Table 1) as well as stable carbon and oxygenisotope ratios of calcite from the top to the bottom of the fine-grained cistern mortar (see Figure 1).

Horizon bulk Depths (mm) Calcite (wt%) δ13Ccalcite (‰) δ18Ocalcite (‰)

B-1 0–1 18.0 − 14.62 − 7.83B-1 1–2 20.3 − 12.74 − 6.63B-1 2–3 12.9 − 13.19 − 6.54B-1 3–4 7.0 − 13.68 − 6.45B-1 4–5 7.2 − 13.95 − 5.98B-1 5–6 8.0 − 13.85 − 5.90B-1 6–7 9.5 − 13.54 − 5.96B-1 7–8 10.0 − 13.58 − 6.36B-1 8–9 12.4 − 13.63 − 5.46B-2 9–10 15.7 − 14.68 − 7.26B-2 10–11 15.5 − 14.14 − 6.82B-2 11–12 13.5 − 13.93 − 5.99B-2 12–13 12.7 − 13.46 − 5.59B-2 13–14 12.5 − 13.57 − 6.33B-2 14–15 13.3 − 13.99 − 6.39B-2 15–16 14.0 − 14.21 − 5.79B-2 16–17 14.2 − 14.46 − 6.14B-2 17–18 12.7 − 13.72 − 6.01B-2 18–19 13.2 − 13.44 − 6.04B-2 19–20 13.5 − 12.89 − 5.37B-2 20–21 13.8 − 12.83 − 5.83B-2 21–22 13.6 − 13.45 − 5.90B-2 22–23 14.5 − 13.33 − 5.63B-2 23–24 16.5 − 13.06 − 6.09B-2 24–25 20.0 − 12.93 − 5.75B-2 25–26 32.4 − 12.69 − 6.02B-2 26–27 41.7 − 13.24 − 6.61E 27–28 28.0 − 13.60 − 9.26E 28–29 14.8 − 12.27 − 8.04E 29–30 15.2 − 12.72 − 7.80E 30–31 22.7 − 13.47 − 8.51D 31–32 26.0 − 14.72 − 9.31D 32–33 25.9 − 13.90 − 9.78D 33–34 27.5 − 13.74 − 8.92D 34–35 26.1 − 13.32 − 8.80D 35–36 28.2 − 11.77 − 8.60D 36–37 27.5 − 11.86 − 8.98D 37–38 25.9 − 12.46 − 8.61D 38–39 26.2 − 12.58 − 8.71D 39–40 25.9 − 13.02 − 8.84D 40–41 26.0 − 13.11 − 8.86

Note: Isotope ratios are given in δ13Ccalcite and δ18Ocalcite values in respect to the VPDB standard.

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

8 M. Dietzel et al.

A B

Figure 3. (a) BSE image of the boundary area between horizons B-1 and B-2 associated with (b) the EMPA elementalmap (investigated area is marked in Figure 2). The warm colours of the elemental map correspond to the higher con-centrations of Ca. White arrows mark areas with elevated Ca concentrations, which correspond to domains of calcite(Cc).

origin are found, the aggregates of horizons B-1 and B-2 consist of ceramic fragments up to 10mm in diameter. The matrix of horizons B-1 and B-2 has a light pink colour, which suggests thatits preparation involved a binder made of quick lime and fine-grained ceramic material [24,25].

The concentrations of the major components of the matrix of the studied mortar are shown inTable 1. The portion of CaO related to hydraulic phases is calculated by subtracting the amountof CaO bound to calcite (obtained by TGA; Table 2) from the CaO content of the matrix as bothcarbonates and hydraulic phases of the binder are dissolved by acidic treatment. The carbon-ate (calcite) and non-carbonate (hydraulic portion) fractions of CaO are highly variable, but thebinder to aggregate ratio is remarkably constant (B/A = 0.31 ± 0.01; n = 5); where for simpli-fication, the concentrations of metal oxides in the matrix (acidic treatment) are used to estimatethe chemical composition of the binder (B) and the remaining non-dissolved solid is related tothe aggregate (A; see Table 1). The analyses of the dissolved fraction of the binder show sig-nificant amounts of SiO2 (2.0–6.8 wt%), Al2O3 (0.9–1.6 wt%), Fe2O3 (0.5–1.4 wt%), and MgO(1.9–5.1 wt%).

The calcite concentrations along the whole mortar profile of the cistern are highly variable, andthe bulk analyses range from 7 to 42 wt% of calcite (see Table 2). A typical heterogeneous dis-tribution of Ca through mortar horizons B-1 and B-2 is shown in Figure 3. The areas marked byyellow to red colours are related to calcite domains, which is confirmed by single-spot analysesof these areas varying between 51 and 54 wt% of CaO (average value: 52.2 ± 1.4 wt% of CaO;n = 5; data not shown in Table 1). This corresponds to 97 wt% of calcite at maximum. The car-bonation of slaked lime is a much slower process than the formation of the hydraulic phases [5].Therefore, it is suggested that fine-grained calcite forms aggregates of needle- and/or laminar-like shape that fills interstitial spaces in addition to Ca-phases of the matrix of the lime mortar(Figure 3). The areas enriched in calcite often occur at the boundaries between different mortarhorizons, which is shown for the boundary between the horizons B-1 and B-2 (Figure 3(b)).

4.2. Isotopic composition of calcite through the cistern mortar

The δ13Ccalcite and δ18Ocalcite values of the mortar samples from the cistern #1 range from − 15 to− 11 ‰ and from − 10 to − 5 ‰ (VPDB), respectively (Table 2). The plot of carbon and oxygenisotopic values displays two separate positive trends for horizons B-1/B-2 and E/D (Figure 4).These trends in isotopic data from our mortar samples follow the correlation between δ13Ccalcite

and δ18Ocalcite obtained by Kosednar-Legenstein et al. [15] for a Roman lime mortar horizon

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies 9

–15

–10

–5

–16 –15 –14 –13 –12 –11

d18O

calc

ite

(o/ o

o, V

PD

B)

d13Ccalcite (o/oo,VPDB)

Figure 4. Variation of stable carbon and oxygen isotopes of calcite from the mortar from the top to bottom of the cisternfloor coring (data from Table 2). The isotopic data of the horizons B-1 (�) / B-2 (�) and E (�) / D (�) are approxi-mated by the regression lines: δ18Ocalcite = 1.23 δ13Ccalcite + 10.6 (r2 = 0.95); and δ18Ocalcite = 0.98 δ13Ccalcite – 4.0(r2 = 0.96), respectively (VPDB; including limiting isotopic values for calcite formed in a strongly alkaline environ-ment by absorption of atmospheric CO2: δ18Ocalcite ≈ − 22 ‰ and δ13Ccalcite ≈ − 26 ‰, VPDB; see text for detailson calculation of latter isotopic values). The dashed line is a reference for profiles from Roman mortar in Styria, Austria[15]. Black arrow indicates a shift to higher oxygen isotopic values of calcite in the matrix of the mortar horizons E/Dversus B-1/B-2.

(Austria). However, the isotopic values reported by these authors [15] are shifted from thosemeasured in horizons B-1/B-2 and E/D by δ18Ocalcite values of about − 1 and − 4 ‰, respec-tively (arrow in Figure 4 indicates shift in δ18Ocalcite values considering horizons B-1/B-2 andE/D).

The isotope analyses of the water samples from the cisterns #26 / #30 result in δ2Hwater andδ18Owater values of –27.3/ − 22.4 ‰ and − 5.2/ − 4.4 ‰ (VSMOW), respectively. Both isotopicvalues fit within the range of locally occurring meteoric water at Pantellaria [44] and follow theglobal meteoric water line given by Craig [45] (δ2Hwater = 8 δ18Owater + 10). Similar δ2Hwater

and δ18Owater values ( − 25 ‰ and − 5 ‰, respectively) were measured for precipitation on thewest coast of Sicilia [46].

5. Discussion

5.1. Formation of hydraulic lime mortar

The studies of ancient hydraulic mortars from various localities showed the importance of theuse of volcanic material, ceramics or calcined kaolin, besides slaked lime, to obtain the waterresistant mortar binder that sets under water saturated conditions. The examples include mortarsof cisterns, reservoirs, and pipelines in Petra [47]; water channel mortars and floor screeds inCyprus [2]; floor screeds in Jericho [48]; mortars of ponds and water channels in Spain [49,50];mortars from the historical town of Mertola [51]; and mortars of cisterns of the Colosseum inRome [52]. The investigation of the mortar profile of the cistern from Pantalleria revealed achange in the technology of hydraulic mortar production. The older horizons D and E containvolcanic rock, while ceramic fragments were found in younger horizons B-1 and B-2. Accord-ing to archaeological dating, this change is a characteristic for the production of cistern mortarin Punic and Roman eras, respectively. Identical or similar render sequences are also found in11 other cisterns with this chronology in the excavation area [27,28]. The estimated B/A ratio

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

10 M. Dietzel et al.

of 0.31 ± 0.01 of the studied mortar is in the typical range of ancient and recent lime mortar(Roman lime mortar B/A = 0.33 ± 0.17 (n = 15) [15]; lime mortar, B/A = 0.3 ± 0.1 [9]). Themarginal variability of B/A parameter through the four mortar horizons proves the sophisticatedand well-established technology of hydraulic mortar production by ancient builders. Although,distinct C-S-H, C-A-H, and C-F-H phases of the studied mortar could not be evaluated by XRDpattern and petrographic analyses, the chemical composition of the matrix (Table 1) suggests theaddition of silicate-based latent hydraulic material to slaked lime to form the hydraulic binder[9], most probably fine-grained volcanic and/or ceramic material.

The chemical analyses of the bulk matrix obtained by acidic treatment revealed that the CIvaries from 0.62 to 2.7 with a mean CI of 1.4 ± 0.8; the CI was calculated according to theequation given by Boynton [35] (Table 1). The CaO(n-carb) values were used for the calculationof the CI to assess the hydraulic components of the binder. The calculated CI obtained from theEMPA single-spot analyses of the binding matrix shows the mean value of 1.9 ± 0.5; domains ofcalcite (CaO ≥ 51 wt%; see chapter 4.1) were excluded from CI calculation. All of the calculatedCIs plot within the range of EMPA analyses reported by Elsen et al. [5] for hydraulic lime binderof ancient mortar from Tournai (0.3 > CI > 3.6). Thus, the results of acidic treatment and single-spot EMPA of the binder show the same range of CI values and indicate its hydraulic property.

In spite of the variability of CI values of the studied hydraulic lime binder along the investi-gated profile, this parameter cannot be used for identification of individual horizons from D toB-1. The variability of the CI values may be related to (i) contamination from partly non-reactedlatent hydraulic material or small fragments of aggregates, and (ii) the ancient processing pro-cedure of mixing of binder components that resulted in chemically heterogeneous binder, at themicroscale [5].

5.2. Tracing the durability of ancient mortar

The durability of a hydraulic lime mortar in the perspective of building material science isreferred to its resistance to degradation. The resistance is based on the change of its miner-alogical, chemical, and microstructural properties due to chemical and/or mechanical corrosion.It is highly challenging to distinguish between the primary and the secondary mortar based onthe individual differences in the chemical composition, particularly for well-preserved hydrauliclime mortar.

Here, the stable carbon and oxygen isotopes of the well-characterized calcite in the binderwere used to trace the chemical alteration of the ancient mortar and thus its durability throughouttime. The chemical alteration of the primary calcite by newly precipitation and/or recrystalliza-tion should change the isotopic signature of calcite in the binder. The range of δ18Ocalcite andδ13Ccalcite values and the positive trend between both isotopic values (Figure 4) are typical forthe precipitation of calcite due to absorption of atmospheric CO2 into alkaline Ca2+-bearingenvironments [12,53–59]. Usdowski and Hirschfeld [14] and Kosednar-Legenstein et al. [15]found an analogous correlation of isotopic data measured along a profile of Roman lime mortarand from calcite cementation in a brickwork experiment, respectively. In both studies, a signif-icant discrimination of the heavy (13C and 18O) versus the light isotopes (12C and 16O) occursfrom the exterior versus the interior layer of a lime mortar profile. Analogous relationships arefound in the present case by plotting the measured isotopic data as a function of the depths ofthe mortar profile (Figure 5). At the boundary between the horizons D and E, the isotopic com-position of the calcite shifts from δ18Ocalcite ≈ − 10 to − 8 ‰ (VPDB) from the upper to thelower parts of horizon D (Figure 5). This shift corresponds to the variation in the isotope ratiosof the exterior and sequential interior layers, respectively [14,15]. The observed change in theδ18Ocalcite values of horizon D (ancient lining of the cistern) is interpreted to be a consequence of

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies 11

Figure 5. Distribution of stable carbon and oxygen isotopes (a) and concentration of calcite (b) in the mortar from thetop to the bottom of the cistern floor (data from Table 2). Dashed lines separate the distinct horizons B-1, B-2, E andD, which are visualized on the Figures 2 and 3. Exterior layers of mortar horizons are located at the right side of eachdashed line (upper part).

its formation conditions during processing. A similar shift is evident for δ18Ocalcite values at all ofthe boundaries of mortar horizons E, B-2, and B-1, where the values change from − 9 to − 8 ‰,− 7.5 to − 6 ‰, and − 8 to − 6 ‰, respectively, considering the exterior and sequential interiorlayers of the horizons (illustrated by the red lines in Figure 5). An analogous shift is shown forδ13Ccalcite values as a function of the depth of the mortar profile in Figure 5. This is also approx-imated by the positive correlation between the values of oxygen and carbon isotopes (Figure 4).The observed δ18Ocalcite and δ13Ccalcite evolution as a function of the depths of the studied mortarprofile reflects the setting of each mortar horizon, where atmospheric CO2 is incorporated intothe lime binder via hydroxylation [14,15].

The isotopic composition of calcite, which is formed in a strong alkaline environment byabsorption of atmospheric CO2, is calculated as lowest and limiting δ18Ocalcite and δ13Ccalcite val-ues for the exterior mortar layers during their formation as lining of the cistern (see arrowheadsin Figure 2(b) and (c)). The unidirectional reaction of CO2 with OH− to HCO−

3 (and subsequentdeprotonization of HCO−

3 to CO2−3 ) proceeds via a kinetically controlled isotope fractionation in

a mixture of the isotopic composition of CO2 and OH− at a 2:1 ratio in the precipitated calcite[15,60]. Briefly, the isotopic composition of the precipitated calcite of δ18Ocalcite = –22 ± 3 ‰( = 1/3 ( − 87) ‰ + 2/3 ( + 10) ‰, VPDB) is calculated by consideration of the isotopic com-position of the local meteoric water with δ18Owater = –35 ± 1 ‰ and δ18OOH– ≈ –87 ‰ (corre-sponding to the measured δ18Owater value in section 4.2, but converted to the VPDB scale), and

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

12 M. Dietzel et al.

the atmospheric CO2 value of δ18OCO2 = + 10 ± 2 ‰ (VPDB; accepted range throughout thepast 2000 years [43,61,62]). In the case of the fractionation of stable carbon isotopes via hydrox-ylation, 13CO2 diffuses and reacts with OH– slower than 12CO2 to form HCO3

–. The overallkinetic isotope fractionation between the precipitated calcite by absorption of atmospheric CO2

in a strong alkaline environment is given by �13Ccalcite−CO2 = δ13Ccalcite–δ13CCO2 ≈ − 18 ‰[15,59,60]. Considering a range of δ13CCO2 values from − 6 to − 9 ‰ (VPDB) of the Earth’satmosphere throughout the past 3000 years [61,62], an isotopic composition of calcite ofδ13Ccalcite = − 26 ± 2 ‰ (= − 18 ‰ + (− 8 ± 2) ‰, VPDB) is calculated. These limitingδ18Ocalcite and δ13Ccalcite values for calcite formation induced by absorption of atmospheric CO2

in a strong alkaline environment (δ18Ocalcite = − 22 ± 3 ‰; δ13Ccalcite = − 26 ± 2 ‰, VPDB)may not be found in an exterior mortar layer due to (i) the insufficient spatial resolution of thesamples collected from the exterior mortar layer directly exposed to the atmospheric CO2 duringsetting and/or (ii) the impact of decreasing pH during the setting of the lime mortar [14]. How-ever, the positive trend of stable carbon versus oxygen isotopic data given in Figures 4 and 5 isnot affected by aforementioned aspects [14,54]. Accordingly, two individual regression lines areobtained by fitting the measured isotopic values for horizons B-1/B-2 and E/D considering theaforementioned calculated limiting isotopic values for the exterior mortar layer (Figure 4). Thedistinct shifts in isotopic composition from the left to the right for each horizon, correspondingto the exterior to the interior mortar layers, can be explained by a Rayleigh isotope fractionationprocess (Figures 4 and 5). As calcite formation continued during setting of the lime mortar, theremaining gaseous CO2 in the pore space was subsequently enriched in 13C and 18O versus 12Cand 16O. This caused subsequently formed calcite to be isotopically heavier along the setting pathfrom the left to the right for each horizon (Figure 4). The Rayleigh isotope fractionation approachduring proceeding carbonation along a lime mortar horizon was verified by laboratory experi-ments, where calcite precipitation by the uptake of gaseous CO2 into series-connected alkalineCa(OH)2 solutions shows an analogous relationship for stable carbon and oxygen isotopic datacompared to that shown in Figure 4 [15].

The shift of the regression lines for δ18Ocalcite versus δ13Ccalcite may be a result of different iso-topic composition of the water used for slaking (black arrow in Figure 4). It is plausible that themeteoric water was used for slaking at Pantelleria Island. The water from Lake Venere and localsprings is of no use for slaking processes due to its high ion concentrations – a result of mixingmeteoric precipitation, seawater and hydrothermal solutions [44,63–65]. The isotopic composi-tion of meteoric precipitation in the area of Pantelleria Island is almost constant within the lastdecades (δ18Owater = − 5 ± 1 ‰; VSMOW). No continental and/or altitude effects could sig-nificantly modify the isotopic signature of the Pantelleria Island, and the isotopic composition ofprecipitation at the open Mediterranean Sea during the past 2000 years is suggested to be almostconstant. However, a significant change of isotopic values can be induced by evaporation duringthe storage and use of water for the processing of the mortar. For instance, evaporation of waterduring processing results in the enrichment of the remaining solution in H2

18O versus H216O

[66]. Such phenomenon may be responsible for the higher δ18Ocalcite values in the mortar hori-zons B-1/B-2 in comparison with that of horizons E/D (see black arrow in Figure 4 and distinctbaselines for the red lines in Figure 5).

Tracing the durability of a hydraulic lime mortar by analysing the stable carbon and in par-ticular oxygen isotopes of calcite in the binder allows for elucidation of primary carbonationof the mortars or secondary alteration, like subsequent recrystallization and/or neoformation ofcalcite. A significant change of the primary δ18Ocalcite value inherited by calcite during settingof the mortar may be caused by such chemical alteration reactions [15]. Newly precipitatedand/or recrystallized calcite carries the isotopic signature of the pore water in which it formed.In the analysed samples, the alteration of calcite was most likely induced by migration of porewater of meteoric origin stored in the cistern through the mortar. For temperatures between 5

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies 13

and 15 °C in the cistern, a 1000 ln αcalcite−water value of 33 ± 1 ‰ is calculated (see Equation(2)). Considering the measured δ18Owater of about − 5 ‰ (VSMOW) of the cistern water, thevalue of δ18Ocalcite should be + 28 ± 1 ‰ (VSMOW) (corresponding to − 3 ± 1 ‰ in theVPDB scale) if the calcite composition was affected by recrystallization and/or formation ofnew grains. The latter isotope value is out of the range of the measured δ18Ocalcite values of thecistern mortar, which varies between − 10 and − 5 ‰ (VPDB; see Table 2). On the other hand,using Equation (2) and a measured δ18Ocalcite value, for example, − 10 ‰ (VPDB) of our cis-tern mortar, it is possible to calculate the theoretical isotopic composition of water from whichcalcite is recrystallized. This calculation results in δ18Owater of − 12 ‰ (VSMOW). Such 18Oversus 16O depleted isotopic composition of the meteoric water is highly unlikely for precipita-tion at Pantelleria Island (see discussion above). In addition, some water could evaporate fromthe cistern. However, the δ18Owater value of − 12 ‰ for the remaining water in the cistern isnot reasonable to indicate evaporation of cistern water. This is because the evaporation of cis-tern water of δ18Owater ≈ − 5 ‰ (VSMOW) results in the opposite trend for the remaining water(δ18Owater > − 5 ‰), as 18O is enriched versus 16O in the remaining H2O [66]. Thus, the rangeand the positive trend of δ18Ocalcite and δ13Ccalcite values of the investigated cistern mortar suggestthat calcite was formed during the setting reactions of the mortar of the distinct horizons, but notbecause of the subsequent recrystallization and/or neoformation of calcite from cistern and porewater.

6. Summary and conclusions

The studied mortar profile from the Punic–Roman cistern #1 from Pantelleria Island (in usebetween 300 B.C and 200 A.D.) shows four distinct horizons, D, E, B-1, and B-2, from the bot-tom to the top of the cistern floor. Volcanic and ceramic aggregates were used for the productionof the mortar of horizons E/D and B-1/B-2, respectively. This change in the manufacturing ofthe cistern is archaeologically dated to Punic and Roman eras. The chemical composition of thematrix of the cistern mortar results in a CI = 1.5 ± 1, which is characteristic for hydraulic limemortar. Non-systematic variability of CI values and a remarkable constant binder to aggregateratio, B/A = 0.31 ± 0.01, along with the analysed mortar profile, indicate highly standardizedprocessing of the studied hydraulic lime mortar. The binder was most likely made of quick lime,fine-grained aggregate material, and water from local precipitation.

Calcite precipitated as a result of carbonation of slaked lime and forms aggregates of needle-and/or laminar-like shape embedded in the matrix of the hydraulic lime mortar. It fills interstitialspaces between the products of hydraulic reaction and/or the aggregates. Areas enriched withcalcite often occur at the boundaries between the distinct horizons of the mortar and indicate anelevated proportion of slaked lime in the binder used for finishing the outer mortar layer. Suchprocessing intensified the activation of latent hydraulic compounds and improved the adhesivestrength of the adjacent layers. The durability of calcite within the investigated ancient mortaris a result of the carbonation and hydraulic setting behaviour of the materials used for construc-tion. This feature can be assessed by understanding the factors controlling the stable oxygenand carbon isotope distribution in the carbonate of the binder matrix that appear to be asso-ciated with the materials durability. In the present case, the ancient manufacturing of a highlydurable cistern mortar was traced and verified by (i) a positive correlation between δ18Ocalcite

and δ13Ccalcite values, (ii) a characteristic evolution of δ18Ocalcite and δ13Ccalcite values througheach mortar horizon, and (iii) the range of δ18Ocalcite values. The latter range of isotopic valuesof the calcite helped to rule out the recrystallization and/or formation of new calcite from porewater.

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

14 M. Dietzel et al.

Acknowledgements

The EMPA analyses were carried out utilizing Jeol JXA8200 equipment at the Montan University in Leoben (Austria)within UZAG cooperation. Chemical analyses were conducted at NAWI Graz Central Lab for Water, Minerals and Rocks.The constructive comments of two anonymous reviewers are greatly appreciated.

Disclosure statement

No potential conflict of interest was reported by the author(s).

References

[1] Kingery WD, Vandiver PB, Prickett M. The beginnings of pyrotechnology, part II: production and use of lime andgypsum plaster in the pre-pottery Neolithic Near East. J Field Archaeol. 1988;15:219–244.

[2] Theodoridou M, Ioannou I, Philokyprou M. New evidence of early use of artificial pozzolanic material in mortars.J Archaeol Sci. 2013;40:3263–3269.

[3] Rehm-Berbenni C, Druta A, Aberg G, Neguer J, Külls CGP, Pacha T, Kienzle P, Bugini R, Fiore MG. Isotope tech-nologies applied to the analysis of ancient roman mortars. In: European Commission, editor. Results of the CRAFTproject (EVK4 CT-2001-30004). Luxembourg: Office for Official Publications of the European Communities; 2005.p. 65.

[4] Maravelaki-Kalaitzaki P, Galanos A, Doganis I, Kallithrakas-Kontos N. Physico-chemical characterization of mor-tars as a tool in studying specific hydraulic components: application to the study of ancient Naxos aqueduct. ApplPhys A. 2011;104:335–348.

[5] Elsen J, Van Balen K, Mertens G. Hydraulicity in historic lime mortars: a review. In: Válek J, Hughes JJ, GrootCJWP, editors. Historic mortars: characterisation, assessment and repair. vol. 7, RILEM Bookseries. Dordrecht:Springer; 2012. p. 125–139.

[6] Válek J, Hughes JJ, Groot CJWP, editors. Historic mortars – characterisation, assessment and repair. Dordrecht:Springer; 2012.

[7] Lea FM. Investigations on pozzolanas. Building Res. 1940;27:1–63.[8] Vitruvius M. The ten books on architecture. Morgan MH, translator. New York: Dover Publications Inc.; 1960.[9] Stark J, Wicht B. Zement und Kalk – Der Baustoff als Werkstoff. Basel: Birkhäuser Verlag; 2000.

[10] Pachiaudi C, Marechal J, van Strydonck M, Dupas M, Dauchot-Dehon M. Isotopic fractionation of carbon duringCO2 absorption by mortar. Radiocarbon. 1986;28:691–697.

[11] van Strydonck M, Dupas M, Keppens E. Isotopic fractionation of oxygen and carbon in lime mortar under naturalenvironmental conditions. Radiocarbon. 1989;31:610–618.

[12] Letolle R, Gegout P, Rafai N, Revertegat E. Stable isotopes of carbon and oxygen for the study of carbona-tion/decarbonation processes in concretes. Cem Concr Res. 1992;22:235–240.

[13] Rafai N, Letolle R, Blanc P, Gegout P, Revertegat E. Carbonation-decarbonation of concretes studied by the way ofcarbon and oxygen stable isotopes. Cem Concr Res. 1992;22:882–890.

[14] Usdowski E, Hirschfeld A. The 13C/12C and 18O/16O composition of recent and historical calcite cement andkinetics of CO2 absorption by calcium hydroxide. N Jb Miner Mh. 2000;11:507–521.

[15] Kosednar-Legenstein B, Dietzel M, Leis A, Stingl K. Stable carbon and oxygen isotope investigation in historicallime mortar and plaster – results from field and experimental study. Appl Geochem. 2008;23:2425–2437.

[16] Dotsika E, Psomiadis D, Poutoukis D, Raco B, Gamaletsos P. Isotopic analysis for degradation diagnosis of calcitematrix in mortar. Anal Bioanal Chem. 2009;395:2227–2234.

[17] Tusa S. Archeologia e storia di un’isola nel Mediterraneo. In: Tusa S, editor. Pantellerian ware. Archeologiasubacquea e ceramica da fuoco a Pantelleria. Palermo: Dario Flaccovio; 2003. p. 15–24. [Italian].

[18] Schäfer T. Archäologische Forschung auf Pantelleria. Zusammenfassung und Interpretation der Befunde und Fundeauf der Akropolis von S. Teresa. In: Schäfer T, Schmidt K, Osanna M, editors. Cossyra I. Die Ergebnisse derGrabungen der Kampagnen 2000–2011 auf der Akropolis von Pantelleria. Rahden: Marie Leidorf; in press.

[19] Abelli L. Archeologia subacquea a Pantelleria. “ . . . de Cossurensibus et Poenis navalem egit . . . ”. Bologna: AnteQuem; 2012.

[20] Schäfer T, Schmidt K, Osanna M, Almonte M, Cossyra II. Ricognizione topografica. Storia di un paesaggiomediterraneo. Rahden: Marie Leidorf; 2013. [Italian].

[21] Schäfer T, Schmidt K, Osanna M, Cossyra I. Die Ergebnisse der Grabungen der Kampagnen 2000–2012 auf derAkropolis von Pantelleria. in press.

[22] Montana G, Fabbri B, Santoro S, Gualtieri S, Iliopoulos I, Guiducci G, Mini S. Pantellerian ware: a comprehensivearcheometric review. Archaeometry. 2007;49:455–481.

[23] Montana G, Randazzo L, Castiglia A, La Russa MF, La Rocca R, Bellomo S. Different methods for soluble saltremoval tested on late-roman cooking ware from a submarine excavation at the island of Pantelleria (Sicily, Italy).J Cult Herit. 2014;15:403–413.

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

Isotopes in Environmental and Health Studies 15

[24] Schön F, Schäfer T, Heinrichs J, Gerdes A. Hydraulische Zisternenverputze aus Karthago und Pantelleria (Italien).In: Dolenz H, Flügel C, editors. Römische und byzantinische Großbauten am Decumanus Maximus. Karthago IV.Die deutschen Ausgrabungen in Karthago. Mainz: von Zabern; 2012. p. 235–248.

[25] Schön F, Schäfer T, Heinrichs J, Gerdes A. Damnum nocet, damnum docet. Werkstoffanalyse von antiken Zis-ternenverputzen. In: Kreiner R, Letzner W, editors. SPA Sanitas Per Aquam. Tagungsband des InternationalenFrontinus-Symposiums zur Technik- und Kulturgeschichte der antiken Thermen (Babesch Supplements 21).Aachen: Peeters; 2012. p. 235–240.

[26] Heinrichs J, Gerdes A, Schön F. Technologietransfer in die Moderne – Vergleich von antiken und modernenBeschichtungssystemen für Wasserreservoire. In: Schäfer T, Schön F, Gerdes A, Heinrichs J, editors. Antike undmoderne Wasserspeicherung. Rahden: Marie Leidorf; 2014. p. 123–134.

[27] Schön F. Insulare Wasserversorgung. Antike Regenwassersammel- und -speicheranlagen auf Pantelleria und Linosa(Italien). In: Schäfer T, Schön F, Gerdes A, Heinrichs J, editors. Antike und moderne Wasserspeicherung. Rahden:Marie Leidorf; 2014. p. 103–118.

[28] Schön F, Schmidt K, Laube I. Die punischen und römischen Befunde auf der Hügelkuppe von S. Teresa. Grabungs-bericht der Kampagnen 2000–2012, Schnitt I/VIII, XIX. In: Schäfer T, Schmidt K, Osanna M, editors. CossyraI. Die Ergebnisse der Grabungen der Kampagnen 2000–2011 auf der Akropolis von Pantelleria. Rahden: MarieLeidorf; in press.

[29] Mantellini S. Water cisterns survey in Pantelleria Island (Italy). In: Angelakis AN, Koutsoyannis D, editors. Pro-ceedings of the 1st IWA International Symposium on Water and Wastewater Technologies in Ancient Civilizations.Iraklio: National Foundation for Agricultural Research; 2006. p. 469–476.

[30] Mantellini S. Water harvesting on Pantelleria Island. Archaeological survey and analysis of the bottle-shaped cis-terns. In: Schäfer T, Schön F, Gerdes A, Heinrichs J, editors. Antike und moderne Wasserspeicherung. Rahden:Marie Leidorf; 2014. p. 73–101.

[31] Gianelli G, Grassi S. Water–rock interaction in the active geothermal system of Pantelleria, Italy. Chem Geol.2001;181:113–130.

[32] Avanzinelli R, Braschi E, Marchionni S, Bindi L. Mantle melting in within-plate continental settings: Sr–Nd–Pband U-series isotope constraints in alkali basalts from the Sicily Channel (Pantelleria and Linosa Islands, SouthernItaly). Lithos. 2014;188:113–129.

[33] Alvarez JI, Navarro I, Martín A, García Casado PJ. Study of the ancient mortars in the north tower of Pamplona’sSan Cernin church. Cem Concr Res. 2000;30:1413–1419.

[34] Callebaut K, Elsen J, Van Balen K, Viaene W. Nineteenth century hydraulic restoration mortars in the SaintMichael’s Church (Leuven, Belgium): natural hydraulic lime or cement? Cem Concr Res. 2001;31:397–403.

[35] Boynton RS. Chemistry and technology of lime and limestone. New York: Wiley; 1980.[36] Lindquist JE. Sub-hydraulic binders in historical mortars. Conference on “Repair mortars for historic masonry”.

Delft: Delft University of Technology; 2005. p. 224–230.[37] Révész KM, Landwehr JM. δ13C and δ18O isotopic composition of CaCO3 measured by continuous flow isotope

ratio mass spectrometry: statistical evaluation and verification by application to Devils Hole core DH-11 calcite.Rapid Commun Mass Spectrom. 2002;16:2102–2114.

[38] Spötl C, Vennemann TW. Continuous-flow isotope ratio mass spectrometric analysis of carbonate minerals. RapidCommun Mass Spectrom. 2003;17:1004–1006.

[39] Morrison J, Brockwell T, Merren T, Fourel F, Phillips AM. On-line high-precision stable hydrogen isotopic analyseson nanoliter water samples. Anal Chem. 2001;73:3570–3575.

[40] Horita J, Ueda A, Mizukami K, Takatori I. Automatic δD and δ18O analyses of multi-water samples using H2-and CO2-water equilibration methods with a common equilibration set-up. International Journal of RadiationApplications and Instrumentation. Part A. Appl Radiat Isot. 1989;40:801–805.

[41] Coplen TB. Calibration of the calcite-water oxygen-isotope geothermometer at Devils Hole, Nevada, a naturallaboratory. Geochim Cosmochim Acta. 2007;71:3948–3957.

[42] Kim S-T, O’Neil JR. Equilibrium and nonequilibrium oxygen isotope effects in synthetic carbonates. GeochimCosmochim Acta. 1997;61:3461–3475.

[43] Hoefs J. Stable isotope geochemistry. 6th ed. Berlin: Springer-Verlag; 2009.[44] Dongarrà G, Hauser S, Alaimo R, Carapezza M, Tonani F. Hot waters on Pantelleria island. Geochemical features

and preliminary geothermal investigations. Geothermics. 1983;12:49–63.[45] Craig H. Isotopic variations in meteoric waters. Science. 1961;133:1702–1703.[46] Longinelli A, Selmo E. Isotopic composition of precipitation in Italy: a first overall map. J Hydrol. 2003;270:75–88.[47] Bonazza A, Ciantelli C, Sardella A, Pecchioni E, Favoni O, Natali I, Sabbioni C. Characterization of hydraulic

mortars from archaeological complexes in Petra. Period Mineral. 2013;82:459–475.[48] Malinowski R, Garfinkel Y. Prehistory of concrete. Concr Int. 1991;13:62–68.[49] Franquelo ML, Robador MD, Ramírez-Valle V, Durán A, Jiménez De Haro MC, Pérez-Rodríguez JL. Roman

ceramics of hydraulic mortars used to build the Mithraeum house of Mérida (Spain). J Therm Anal Calorim.2008;92:331–335.

[50] Robador MD, Perez-Rodriguez JL, Duran A. Hydraulic structures of the Roman Mithraeum house in Augustaemerita, Spain. J Archaeol Sci. 2010;37:2426–2432.

[51] Silva AS, Ricardo JM, Salta M, Adriano P, Mirão J, Candeias AE, Macias S. Characterization of Roman mortarsfrom the historical town of Mertola. In: Fort R, Alvarez de Buergo M, Gomez-Heras M, Vazquez-Calvo C, editors.Heritage, weathering and conservation. London: CRC Press; 2006. p. 85–90.

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5

16 M. Dietzel et al.

[52] Silva DA, Wenk HR, Monteiro PJM. Comparative investigation of mortars from Roman Colosseum and cistern.Thermochim Acta. 2005;438:35–40.

[53] Macleod G, Fallick AE, Hall AJ. The mechanism of carbonate growth on concrete structures as elucidated by carbonand oxygen isotope analyses. Chem Geol. 1991;86:335–343.

[54] Dietzel M, Usdowski E, Hoefs J. Chemical and 13C/12C- and 18O/16O-isotope evolution of alkaline drainage watersand the precipitation of calcite. Appl Geochem. 1992;7:177–184.

[55] Clark ID, Fontes J-C, Fritz P. Stable isotope disequilibria in travertine from high pH waters: laboratoryinvestigations and field observations from Oman. Geochim Cosmochim Acta. 1992;56:2041–2050.

[56] Dietzel M. Messungen der stabilen Isotope des Kohlenstoffs an Kalk-Versinterungen von Beton [Measurements ofthe stable carbon isotopes in calcite sinters on concrete]. ZKG Int. 2000;53:544–548. [German].

[57] Krishnamurthy RV, Schmitt D, Atekwana EA, Baskaran M. Isotopic investigations of carbonate growth on concretestructures. Appl Geochem. 2003;18:435–444.

[58] Flehoc C, Girard J-P, Piantone P, Bodenan F. Stable isotope evidence for the atmospheric origin of CO2 involvedin carbonation of MSWI bottom ash. Appl Geochem. 2006;21:2037–2048.

[59] Rinder T, Dietzel M, Leis A. Calcium carbonate scaling under alkaline conditions – case studies and hydrochemicalmodelling. Appl Geochem. 2013;35:132–141.

[60] Usdowski E, Hoefs J. Oxygen isotope exchange between carbonic acid, bicarbonate, carbonate, and water: a re-examination of the data of McCREA(1950) and an expression for the overall partitioning of oxygen isotopesbetween the carbonate species and water. Geochim Cosmochim Acta. 1993;57:3815–3818.

[61] Ghosh P, Brand WA. Stable isotope ratio mass spectrometry in global climate change research. Int J Mass Spectrom.2003;228:1–33.

[62] Eiler JM, Schauble E. 18O13C16O in Earth’s atmosphere. Geochim Cosmochim Acta. 2004;68:4767–4777.[63] Aiuppa A, D’Alessandro W, Gurrieri S, Madonia P, Parello F. Hydrologic and geochemical survey of the lake

“Specchio di Venere” (Pantelleria island, Southern Italy). Environ Geol. 2007;53:903–913.[64] Cangemi M, Bellanca A, Borin S, Hopkinson L, Mapelli F, Neri R. The genesis of actively growing siliceous

stromatolites: Evidence from Lake Specchio di Venere, Pantelleria Island, Italy. Chem Geol. 2010;276:318–330.[65] Censi P, Cangemi M, Brusca L, Madonia P, Saiano F, Zuddas P. The behavior of rare-earth elements, Zr and Hf

during biologically-mediated deposition of silica-stromatolites and carbonate-rich microbial mats. Gondwana Res.2015;27:209–215.

[66] Gat JR, Gonfiantini R. Stable isotope hydrology: Deuterium and oxygen-18 in the water cycle. IAEA TechnicalReport Series 210. Vienna: International Atomic Energy Agency; 1981.

Dow

nloa

ded

by [

Mur

doch

Uni

vers

ity L

ibra

ry]

at 1

8:28

18

June

201

5