A multidisciplinary approach for the archaeometric study of pozzolanic aggregate in roman mortars:...

A MULTIDISCIPLINARY APPROACH FOR THEARCHAEOMETRIC STUDY OF POZZOLANIC

AGGREGATE IN ROMAN MORTARS: THE CASE OFVILLA DEI QUINTILI (ROME, ITALY)*

C. M. BELFIORE,1,2 G. V. FICHERA,1† M. F. LA RUSSA,2 A. PEZZINO,1

S. A. RUFFOLO,2 G. GALLI3 and D. BARCA2

1University of Catania, Dipartimento di Scienze Biologiche, Geologiche e Ambientali—Sezione di Scienze della Terra,Corso Italia 57, 95129 Catania, Italy

2University of Calabria, Dipartimento di Biologia, Ecologia e Scienze della Terra (DiBEST), Via Pietro Bucci,87036 Arcavacata di Rende (Cs), Italy

3Soprintendenza Speciale per i Beni Archeologici di Roma, Villa dei Quintili, via Appia Nuova 1092, 00197 Roma, Italy

This contribution focuses on the study of historical mortars from a Roman archaeological siteknown as Villa dei Quintili, a monumental villa located in the south-eastern part of Rome(Italy). The study was carried out on 38 mortar samples, collected from several edifices withinthe complex. A multi-analytical approach, including polarized optical microscopy, scanningelectron microscopy coupled with energy-dispersive system and laser ablation inductivelycoupled plasma mass spectrometry, was used to analyse the pozzolanic material used for thepreparation of mortars. Data obtained on both major and trace elements were compared withthe compositions of Pozzolana Rossa, Pozzolana Nera and Pozzolanella samples from theAlban Hills volcanic district, collected from a historic quarry nearby, as well as with literaturedata. The results of such a multidisciplinary approach allowed us to recognize the use ofall three pozzolan types for the aggregate fraction of examined mortars from the Villa deiQuintili.

KEYWORDS: ROMAN MORTARS, LATIUM, POZZOLAN, CLINOPYROXENE, SEM–EDS,LA–ICP–MS

INTRODUCTION

From the third century bc onwards, the ancient Roman builders were able to recognize the maincharacteristics of building materials, but it was only from the first century bc that they refined theproduction techniques for mortars by accurately choosing a particular volcanic aggregate in orderto formulate mortars with specific properties (Collepardi 2003; Jackson et al. 2005; Lancaster2009). With this expertise the Romans, during the Imperial age (27 bc–ad 476), created gran-diose monuments that still exist today, thus testifying to the durability of the Roman mortars(Jackson and Marra 2006).

A vast literature concerns the archaeometric study of Roman mortars. Many papers havefocused attention on the mineralogical and petrographic features of mortar components, aimingat determining the old production techniques used by the Roman builders (Sánchez-Moral et al.2005; Silva et al. 2005; Giavarini et al. 2006; Jackson and Marra 2006; Jackson et al. 2006; Pavía

*Received 20 July 2013; accepted 17 December 2013†Corresponding author: email [email protected]

bs_bs_banner

Archaeometry ••, •• (2014) ••–•• doi: 10.1111/arcm.12085

© 2014 University of Oxford

mailto:[email protected]

and Caro 2008; Jackson et al. 2009, 2010). Other authors have investigated the hydraulicproperties of pozzolans and analysed the processes leading to the formation of hydrate phases,such as calcium silicate hydrates (CSH) and calcium aluminate hydrates (CAH) (Day 1990;Massazza 1998; Moropoulou et al. 2004; Pavía 2008; Fernández et al. 2010).

In this work, Roman mortars taken from a relevant archaeological site known as the Villa deiQuintili, in Rome (Italy), were studied (Fig. 1) (Fichera et al. 2012). The monumental complexwas built in the first half of the second century by the family of the Quintili brothers, on thepromontory formed by the ‘Capo di Bove’ lava flow, erupted during the volcanic activity of theAlban Hills (Rotondi 2012). The villa, which belonged to several emperors, consists of manyedifices (all in excellent condition despite having stood for more than 1900 years) that arearticulated into various centres and characterized by different building techniques ascribable todiverse construction phases (Frontoni 2000; Frontoni et al. 2010). The main goal of this study isto identify the type of pozzolanic material used in the preparation of the examined mortars. Itis known that the Roman builders, before and during the Imperial age, used to employ threedifferent types of pozzolans (Pozzolana Rossa, Pozzolana Nera and Pozzolanella), all of whicherupted from the Alban Hills volcanic district (Jackson et al. 2005; Jackson and Marra 2006;Conticelli et al. 2010; Jackson et al. 2010) (Fig. 1). For such a purpose, a geochemical approachbased on the investigation of major and trace elements of specific volcanic phases, which hasproved to be a powerful tool for provenance studies of archaeological ceramics (Barone et al.2010; Belfiore et al. 2014), was applied here, for the first time ever, to the study of historicalmortars. Specifically, analyses through scanning electron microscopy coupled with an energy-dispersive system (SEM–EDS) and laser ablation inductively coupled plasma mass spectrometry(LA–ICP–MS) were carried out on both pozzolanic rock fragments and clinopyroxene crystalsoccurring in the aggregate fraction of examined mortars. Clinopyroxenes were chosen among theother observed volcanic minerals since they represent the sole ubiquitous mineralogical phaseexhibiting a large chemical variability that is able to record the variations of the Alban Hillsmagmas during their petrological evolution (Dallai et al. 2004; Freda et al. 2006). The appliedapproach has also included a chemical comparison with samples of the three above-mentionedtypes of pozzolans, taken from a historic quarry sited near to the villa.

This work, which represents the first archaeometric study ever carried out at this renowned site,was performed in collaboration with the Archaeological Superintendence of Rome.

THE ARCHAEOLOGICAL SITE OF THE VILLA DEI QUINTILI

The main nucleus of the monumental villa was built in the first half of the second century,probably at the behest of the family of the brothers Sesto Quintilio Condiano and Sesto QuintilioValerio Massimo, as testified by lead pipe inscriptions with their names, found during theexcavations of 1828 conducted by the archaeologist Antonio Nibby. Later (ad 182), the EmperorCommodus killed the brothers and confiscated the villa, transforming it into an imperial resi-dence until the third century ad, with the Emperors Septimius Severus (ad 193–211), SeverusAlexander (ad 222–35) and Antonius Gordianus (ad 238–44) (Rotondi 2012).

The monumental complex consists of several edifices (characterized by different buildingtechniques referred to diverse construction phases, mainly those of Trajan, Hadrian, Commodusand Severus), which are articulated into various distinct cores, as shown in Figure 2. The mainnucleus of the villa is represented by the large residential area, which is divided into publicand private zones. In the south-eastern part is the basis villae, the service sector of the villa,consisting of several rooms used by the servants (such as kitchens, warehouses, ovens etc.). The

2 C. M. Belfiore et al.

© 2014 University of Oxford, Archaeometry ••, •• (2014) ••–••

Figu

re1

Age

nera

lize

dm

apof

Rom

ean

dit

ssu

rrou

ndin

gs,s

how

ing

the

loca

tion

ofth

eV

illa

deiQ

uint

ilial

ong

wit

hth

eM

onti

Saba

tini

and

Alb

anH

ills

volc

anic

dist

rict

s.O

nth

eri

ght,

anov

eral

lvi

ewof

the

arch

aeol

ogic

alsi

te.

Pozzolanic aggregate in Roman mortars: Villa dei Quintili (Rome, Italy) 3


thermal environments, located in the north-western zone, are characterized by three differentareas: the Calidarium, the Frigidarium and the Tepidarium. Between these three buildings is theLudus/Viridarium, a structure that was born as a small amphitheatre for gladiatorial games andlater changed to a panoramic garden. Finally, the Nymphaeum, a monumental fountain overlook-ing the Appia Antica, stands near the original entrance to the villa (Frontoni and Galli 2006, 2010;Frontoni et al. 2010).

THE SAMPLED MATERIALS

Archaeological mortars from the Villa dei Quintili

An extensive sampling (38 mortars) was performed on several edifices of the monumentalcomplex; that is, the thermal areas (the Calidarium and Frigidarium), the Ludus/Viridarium, theresidential area (both public and private) and the Nymphaeum (Fig. 2). The sampling was carriedout with the assistance of archaeologists from the Archaeological Superintendence of Rome, inorder to take samples that were representative of the different construction phases. Stainless steeltools and surgical lancets were used and the main criteria for sampling included the type ofmortar, the good state of preservation, the available quantities and the representativeness. Thesamples taken from the above-mentioned edifices consist of bedding, coating and paving mortars.

Figure 2 A satellite view of the Villa dei Quintili and details of the sampled areas: A, Nymphaeum; B,Ludus/Viridarium; C, public zone of the residential area; D, Frigidarium; E, Calidarium; F, private residential area.



Most of them appear to be macroscopically intact and with a high degree of cohesion, except forsome samples that are more friable (Fig. 3). A list of the examined mortars, along with samplingsites, typology and main macroscopic characteristics, is given in Table 1.

Pozzolan samples from the Alban Hills

With the aim of identifying the type of pozzolanic materials used as the aggregate fraction inthe examined mortars, some samples of Pozzolana Rossa (red), Pozzolana Nera (black) andPozzolanella (brownish) were taken from a historic quarry located in the ‘Tenuta di TorMarancia’, close to the archaeological site.

The Roman pozzolans are pyroclastic deposits, partially lithified, erupted between 561 and285 ka from the Alban Hills volcanic district (Karner et al. 2001). The entire successionis known as the ‘Pozzolane Tuffs succession’ (500–355 ka), within which Pozzolana Rossa(460 ka), Pozzolana Nera (410 ka) and Pozzolanella (355 ka) are the main ignimbrites(Conticelli et al. 2010). The latter represent the three final eruptive cycles of the Tuscolano–Artemisio explosive phase (De Rita et al. 1995). The main features of the three pozzolan typesare briefly described as follows. Pozzolana Rossa is a dark red, massive and chaotic ignimbriteand, generally, its thickness varies from 10 to 20 m, with a total volume of 35 km3 (Giordano

(a) (b)

(c) (d)

Figure 3 Representative photographs of the sampled mortars.



Table 1 A list and macroscopic description of the examined mortar samples from the Villa dei Quintili

Sample Location* Typology† Bindercolour(Munsell)

Aggregateabundance‡

Size Cohesion§

VQ1 E b.m. 7.5 YR 8/1 +++ Up to ∼2 cm ++++VQ2 E b.m. 7.5 YR 8/1 +++ Up to ∼2 cm ++++VQ3 E c.m. 10 YR 8/1 +++ From ∼1 mm to ∼2 cm ++++VQ4 E b.m. 10 YR 8/1 +++ From ∼2 mm to ∼1–1.5 cm ++++VQ5 E b.m. 10 YR 8/1 +++ From ∼2 mm to ∼1 cm ++++VQ6 E b.m. 10 YR 7/1 +++ From ∼1 mm to ∼5 mm +++VQ7 E p.m. 10 YR 8/1 + Up to ∼2 mm ++++VQ8 E b.m. 7.5 YR 8/1 +++ Up to ∼3–4 cm +++VQ9 E b.m. 10 YR 8/1 +++ Up to ∼3 cm +++VQ10 E b.m. 10 YR 8/1 +++ Up to ∼3 cm ++++VQ11 E c.m. 10 YR 8/1 + Up to ∼1 cm ++VQ12 E c.m. 7.5 YR 8/1 + Up to ∼1.5 cm ++VQ13 D b.m. 7.5 YR 7/1 + Up to ∼1 cm ++VQ14 D b.m. 10 YR 8/1 + Up to ∼1 cm ++++VQ15 D b.m. 7.5 YR 8/1 + Up to ∼2 cm +++VQ16 D b.m. 10 YR 7/1 + Up to ∼3 mm ++VQ20 C b.m. 2.5 YR 7/1 ++ From ∼1 mm to ∼3 cm +++VQ21 B b.m. 10 YR 7/1 + Up to ∼5 mm ++VQ22 B b.m. 10 YR 7/1 + Up to ∼5 mm ++VQ23 B b.m. 10 YR 8/1 ++ Up to ∼2 cm ++VQ24 B b.m. 10 YR 8/1 ++ Up to ∼5 mm ++VQ28 T b.m. 7.5 YR 8/1 ++ Maximum 1 mm +++VQ30 F b.m. 7.5 YR 7/1 ++ Up to ∼1 cm ++VQ31 F b.m. 10 YR 8/1 + Up to ∼5 mm +++VQ32 F c.m. 10 YR 8/1 ++ Up to ∼1.5 cm +++VQ33 F b.m. 10 YR 8/1 +++ Up to ∼2 cm +++VQ34 F b.m. 10 YR 8/1 +++ Up to ∼1 cm +++VQ35 C b.m. 10 YR 7/1 + Up to ∼5 mm ++++VQ36 C c.m. 7.5 YR 8/1 + Up to ∼0.5 mm +++VQ37 C c.m. 10 YR 8/1 + Up to ∼0.5 mm ++VQ38 C p.m. 7.5 YR 8/1 + Up to ∼2–5 mm ++VQ39 C p.m. 7.5 YR 7/1 + Up to ∼2–5 mm +++VQ40 A b.m. 7.5 YR 7/1 ++ From ∼1 mm to ∼2 cm ++VQ41 A b.m. 10 YR 5/2 ++ Up to ∼5 mm ++VQ42 A b.m. 10 YR 7/1 +++ Up to ∼1–2 mm +++VQ43 A b.m. 7.5 YR 7/1 + Up to ∼5 mm +++VQ44 A b.m. 7.5 YR 8/1 +++ Up to ∼1 cm +++VQ46 A b.m. 10 YR 7/1 +++ Up to ∼1 cm ++++

*Location: E, Calidarium; D, Frigidarium; B, Ludus/Viridarium; T, Tepidarium; C, public residential area; F, private residential area;

A, Nymphaeum.

†Typology: b.m., bedding mortar; p.m., paving mortar; c.m., coating mortar.

‡Aggregate abundance: +++, very abundant; ++, abundant; +, scarce.

§Cohesion: ++++, very high; +++, high; ++, medium.



et al. 2006). The matrix of this ignimbrite (60–90 vol%) primarily consists of coarse-ash shardsalong with crystal fragments of leucite, clinopyroxene and biotite (Trigila et al. 1995). Theframework is largely constituted by reddish to dark porphyritic and vesicular to poorly vesicularscoria (Giordano et al. 2006). Despite its name, the Pozzolana Rossa is not red in all locations orat all depths. The chemical composition is tephritic (Trigila et al. 1995) or tephritic/basanitic,according to Fornaseri et al. (1963). The pyroclastic unit of Pozzolana Nera shows a variablethickness (from 1 to 10 m) and a total volume of 15 km3 (De Rita et al. 1995). A dark grey toblack colour and a massive and chaotic structure characterize these ignimbrites. The matrix(80–90 vol%) is mostly composed of fine-ash shards and crystal fragments of leucite,clinopyroxene and biotite, whereas the framework is made of dark grey porphyritic and vesicularscoria (Giordano et al. 2006). The composition is tephri-phonolitic (Trigila et al. 1995). Thepyroclastic unit of Pozzolanella is characterized by massive and chaotic ignimbrite deposits, witha colour variable from dark grey to dark red. The deposits are generally 20–80 m thick in theproximal areas. The matrix of the ignimbrite (60–90 vol%) is constituted by coarse-ash shardsand abundant crystal fragments of leucite, clinopyroxene and biotite. The framework is charac-terized by scoria with a dark reddish to black colour, highly porphyritic and variously vesiculated(Giordano et al. 2006). The composition varies from phono-tephritic to tephri-phonolitic (Trigilaet al. 1995; Freda et al. 1997).

ANALYTICAL TECHNIQUES

All mortar and pozzolan samples underwent petrographic analyses through polarized opticalmicroscopy. Thin-section observations were performed using a Zeiss AxioLab microscope,equipped with a digital camera to capture images. The characteristics observed were as follows:the grain size and shape of the aggregate fraction, the mineralogical composition and texturalfeatures of both aggregate and binder, and the percentage ratio between the two components.

SEM–EDS analyses were performed to obtain information on the microstructure and majorelement composition of both mortars and pozzolan samples. Analyses were carried out at theDipartimento di Scienze Biologiche, Geologiche e Ambientali of Catania (Italy), using a TescanVega-LMU scanning electron microscope equipped with an EDAX Neptune XM4-60 micro-analyser operating via an energy-dispersive system characterized by an ultra-thin Be windowcoupled with an EDAX WDS LEXS (wavelength-dispersive low-energy X-ray spectrometer)calibrated for light elements. The operating conditions were set at 20 kV accelerating voltage and0.2 nA beam current for the analysis of major element abundances; the low values of the beamcurrent reduce the effect of alkali loss during the running of the analysis. Repeated analyses at thebeginning and end of analytical runs performed on clinopyroxene and glass inner standards,which have been verified through certified international reference materials, ensure precision forall the collected elements of the order of 3%. Chemical analyses were performed in raster modefor the investigation of the whole rock and in spot mode for the analysis of clinopyroxenecrystals.

Concentrations of trace elements in both mortar and pozzolan samples were determined bylaser ablation inductively coupled plasma mass spectrometry (LA–ICP–MS) at the Dipartimentodi Biologia, Ecologia e Scienze della Terra, Università della Calabria, using an Elan DRCeinstrument (PerkinElmer/SCIEX), connected to a New Wave UP213 solid-state Nd:YAG laserprobe (213 nm). Measurements were carried out on 100-μm-thick sections. The ablation wasperformed with spots varying between 50 and 80 μm, with a constant laser repetition rate of10 Hz and a fluence of ∼ 20 J cm−2. In all analyses, a transient signal of intensity versus time was



obtained for each element using a 60 s background level (acquisition of gas blanks), followed by40 s of ablation and then 60 s of post-ablation at background levels. Data were transmitted to aPC and processed by the GLITTER program. The calibration was carried out through a glassreference material (NIST 612-50 ppm) used as external standard (Pearce et al. 1997), whileinternal standardization to correct instrumental instability and drift was achieved using CaOconcentrations from SEM–EDS analyses (Fryer et al. 1995). Accuracy was evaluated on BCR 2Gglass reference material, and the resulting element concentrations were compared with referencevalues from the literature (Gao et al. 2002). Accuracy, as the relative difference from referencevalues, was always better than 10%, even if most elements plotted in the range of ±5%.

RESULTS

Petrographic analysis

Archaeological mortars Thin-section analyses of mortars allowed us to identify four differentgroups (A, B, C and D), on the basis of the type/abundance of the aggregate and the maincharacteristics of the binder (colour, texture and degree of compactness) (Table 2). Specifically,the mortars of types A and B (Figs 4 (a) and 4 (b)) show a highly compact binder, while thosebelonging to types C and D (Figs 4 (c) and 4 (d)) have a moderately to scarcely compact one. Thecolour of the binder is generally greyish, except for samples of type C, where it is beige. Thetexture varies from micritic to microsparitic in the same sample, with the only exception of typeD, where it is predominantly microsparitic. Abundant areas of dissolution and recrystallization ofthe binder were observed in both primary (small and spherical) and secondary (large andirregular-shaped) voids. Furthermore, most of examined samples exhibit the occurrence of limelumps, sometimes with a size up to 2 mm. With regard to the aggregate fraction, it is mainly

Table 2 A list of the mortar samples belonging to the four petrographic groups and a brief description of the maincharacteristics of the binder and aggregate

Typology Samples Binder Aggregate

A VQ1, VQ2, VQ10, VQ12,VQ16, VQ21, VQ22, VQ32,VQ40, VQ41, VQ42, VQ43,VQ44, VQ46

Very compact, homogeneousenough with colour frombrown to dark grey

Only volcanic fragments

B VQ3, VQ6, VQ7, VQ8, VQ11,VQ13, VQ14, VQ15, VQ20,VQ23, VQ24, VQ30, VQ31,VQ33, VQ34, VQ35, VQ38,VQ39

Very compact, homogeneousenough with colour from lightto dark grey

Volcanic fragments andsubordinate cocciopestofragments

C VQ4, VQ28, VQ36, VQ37 Compact and homogeneousenough, with colour frombrown to beige

Cocciopesto fragments andsubordinate volcanicfragments

D VQ5, VQ9 Rarely compact, highlyheterogeneous, with colourfrom light to dark grey

Cocciopesto and volcanicfragments in similarproportions



constituted by volcanic fragments (pozzolans), cocciopesto (grog) and single crystals ofclinopyroxene, biotite and leucite. Sporadic fragments of marble were also recognized in somesamples. On the whole, all pozzolan fragments are characterized by a dark brown colour, aporphyric to oligophyric texture, with a groundmass ranging from vitrophyric to intersertal, andphenocrysts of leucite, clinopyroxene and biotite, in variable proportions. Some fragments do notshow any phenocrysts, while others display abundant and large-sized (up to ∼5 mm) leucitecrystals. The shape of the volcanic aggregate varies from angular to sub-rounded and the particlesize distribution is poorly sorted. All examined mortar samples show clear evidence of reactionrims around the pozzolanic fragments. The type and the abundance of the aggregate fractionvaries in the above-mentioned four groups. Mortars of type A display the exclusive presence ofpozzolans, whereas predominant volcanic fragments and subordinate cocciopesto characterizethe type B. In type C, cocciopesto predominates on the pozzolanic fragments and in type D theyoccur in similar proportions. The aggregate/binder ratio is highly variable even within a singletype of mortar. The porosity, evaluated through a rough estimation with respect to the entire fieldof view of the thin section, ranges from about 20 to 40 vol% and consists of both primary(spherical in shape) and secondary (irregularly shaped) pores.

Pozzolan samples from the quarry The three pozzolan types (Pozzolana Rossa, Pozzolana Neraand Pozzolanella) were examined in thin section in order to investigate their main petrographic

(a) (b)

(c) (d)

Figure 4 Microphotographs of the four different petrographic groups identified among the examined mortars: (a) typeA; (b) type B; (c) type C; (d) type D. The images were taken in cross-polarized light, magnification 5×.



features. The Pozzolana Rossa samples (RED; Fig. 5 (a)) show an oligophyric structure,with very rare phenocrysts of leucite (0.25–2.5 mm) and clinopyroxene (0.2–1.5 mm). Thegroundmass exhibits a dark brown colour and a vitrophyric texture. The Pozzolana Nera samples(PNR; Fig. 5 (b)) have a porphyritic structure, with abundant phenocrysts of leucite (0.1–5 mm),subordinate clinopyroxene (from 0.25 mm to 1.5 mm) and rare biotite (up to 0.5 mm). Thegroundmass, brownish in colour, exhibits an intersertal texture, with a glassy fraction alternatingto microlites (mainly with tabular shape) too small to be identified. The Pozzolanella samples(VSN2; Fig. 5 (c)) show a porphyritic structure, with phenocrysts of leucite (0.2–1.5 mm),clinopyroxene (0.2–2 mm) and biotite (0.2–2 mm) in similar proportions. The mafic phases aremore abundant than in the other two pozzolan types. The groundmass ranges from vitrophyric tointersertal and has a dark brown colour.

SEM–EDS analysis

Archaeological mortars Micromorphological investigations of the pozzolanic rock fragmentsoccurring in the examined mortars revealed that their textural characteristics are highly variable,also within the same sample. Some BSE images, referring to volcanic fragments within themortar VQ24, are shown in Figure 6, as an example. The shape varies from polygonal (with sharpedges) to sub-rounded and the size from micrometric (50 μm) to centimetric (to 2 cm). Thegroundmass generally exhibits a vitrophyric to intersertal texture, with glass amounts varyingfrom sample to sample and also within a single one. Some fragments showing microlites ofleucite and clinopyroxene were also observed. The frequency, shape and size of the vacuoles arehighly variable, even between pozzolan fragments of the same mortar.

Thirteen mortar samples, representative of the four identified petrographic groups, wereselected for SEM–EDS analyses in order to obtain the major element composition of bothpozzolanic fragments and clinopyroxene crystals within the examined mortars. Investigationswere carried out in raster mode for the pozzolan fragments and in spot mode for clinopyroxenes.Furthermore, analyses of the volcanics were performed in the central area of the biggest frag-ments, in order to avoid possible contaminations due to reactions with the binder. The full dataset obtained is available as electronic supplementary material. In the TAS diagram (Fig. 7), thecomposition of pozzolanic aggregate ranges from basaltic andesite to tephri-phonolite, showinga highly variable alkali content (Na2O ranges from 0.07 to 9.96 wt% and K2O from 0.16 to20.80 wt%) and a limited variability of SiO2 (44.60–59.60 wt%).

The analysed clinopyroxenes were classified according to Morimoto et al. (1988). In theQUAD diagram (Fig. 8), they mainly fall into the diopside field. All analysed crystals (see theelectronic supplementary material) show wide compositional variations with regard to most ofmajor elements (considered as cationic values after the formula calculation on the basis of sixoxygens). In particular, larger intervals are presented by Ti (0.011–0.057 a.p.f.u.), Al (0.099–0.448 a.p.f.u.), Mg (0.506–0.950 a.p.f.u.), Fe (0.099–0.408 a.p.f.u.), Na (0.001–0.109 a.p.f.u.)and Ca (0.676–0.955 a.p.f.u.).

Pozzolan samples from the quarry Micromorphological analyses of the three pozzolan types(Pozzolana Rossa, Pozzolana Nera and Pozzolanella) were carried out in order to better inves-tigate their textural features. Backscattered electron (BSE) images show that the texture ofgroundmass, and also the abundance, size and shape of the vacuoles, are rather variable evenwithin the same sample. On the whole, the texture of the groundmass ranges from vitrophyric tointersertal and the vacuoles are characterized by an irregular to spherical shape and a size ranging



(a)

(b)

(c)

Figure 5 Microphotographs of the three pozzolan samples from the quarry: (a) Pozzolana Rossa (RED); (b) PozzolanaNera (PNR); (c) Pozzolanella (VSN2). The images were taken in plane-polarized light, magnification 5×.



Figu

re6

BSE

–SE

Mim

ages

show

ing

the

diff

eren

tm

icro

text

ures

that

char

acte

rize

the

pozz

olan

frag

men

tsoc

curr

ing

wit

hin

the

aggr

egat

efr

acti

onof

sam

ple

VQ

24.



from a few tens of microns to ∼1 mm (Fig. 9). The groundmass of Pozzolana Rossa, in someportions, shows microlites of clinopyroxene (∼10 μm sized) and Fe oxides (Figs 9 (a)–9 (c)),whereas the Pozzolana Nera exhibits abundant microlites of leucite and subordinateclinopyroxene (both with a size ranging from ∼20 to ∼50 μm) (Figs 9 (d)–9 (f)). Finally, inthe Pozzolanella type an intersertal groundmass, with microlites of leucite (∼20 μm), almost

Figure 7 Total alkali versus silica classification diagram (Le Bas et al. 1986) for the pozzolanic aggregate within theexamined mortars.

Figure 8 QUAD diagram (Morimoto 1988) for the classification of clinopyroxenes within the volcanic aggregate ofmortars.



completely altered into analcime, and more rare clinopyroxenes (∼10 μm), was observed(Figs 9 (g)–9 (i)).

Raster analyses were carried out by SEM–EDS on the three pozzolan types to determine themajor element composition. The results (in wt%) are reported in Table 3. The data obtaineddisplay a wide compositional range in the TAS diagram (Fig. 10), in terms of serial characters anddegree of evolution. In particular, the compositions of Pozzolana Rossa are the less evolvedamong the three types, falling into the fields of tephrite, basaltic andesite and basaltictrachyandesite. The SiO2 values vary between 47.82 and 55.54 wt%, while contents of Na2O andK2O are in the ranges 0.30–7.22 wt% and 0.55–2.45 wt%, respectively. The Pozzolana Nerasamples are characterized by intermediate compositions between those of Pozzolana Rossa andPozzolanella. On the whole, the data distribution shows a typical enrichment in alkali withrespect to the silica content. Concentrations of SiO2 vary from 52.11 to 58.07 wt%, whereas thoseof alkali are in the ranges 1.29–5.07 wt% for Na2O and 2.21–8.48 wt% for K2O. Products fall intothe fields of andesite, basaltic trachyandesite and tephri-phonolite. The Pozzolanella composi-tions are the most evolved among the three analysed ignimbrites, showing concentrations of SiO2

from 51.26 to 62.47 wt%. Products vary between basaltic andesite, andesite, trachyandesite and

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Figure 9 BSE–SEM images showing the different textural features observed for Pozzolana Rossa (a–c), Pozzolana Nera(d–f) and Pozzolanella (g–i). Microlites of clinopyroxenes characterize the groundmass of RED samples (c), whereasmicrolites of leucite can be observed in PNR (f) and VSN2 (i), in the latter almost totally replaced by analcime.



Tabl

e3

Maj

orel

emen

tco

ncen

trat

ions

(wt%

)de

term

ined

thro

ugh

SEM

–ED

San

alys

isof

pozz

olan

sam

ples

from

the

quar

ry(P

NR

,Poz

zola

naN

era;

RE

D,P

ozzo

lana

Ros

sa;

VSN

2,Po

zzol

anel

la)

Sam

ple

PN

RR

ED

Ana

lysi

snu

mbe

r1

23

45

67

8A

vera

geSt

.Dev

.1

23

45

67

89

SiO

255

.47

52.5

856

.84

58.0

752

.11

53.0

253

.17

53.0

054

.28

2.22

47.8

249

.23

53.0

753

.91

53.5

855

.54

51.2

251

.06

52.4

2T

iO2

1.05

1.06

1.09

0.79

0.74

0.95

1.06

0.98

0.97

0.13

0.94

1.16

1.01

0.75

1.06

0.26

1.12

1.27

0.99

Al 2

O3

16.6

616

.19

14.8

321

.120

.81

17.7

119

.31

19.3

218

.24

2.26

16.5

715

.35

18.1

15.5

913

.137

.24

16.6

717

.49

10.2

5Fe

O8.

739.

379.

946.

948.

449.

046.

896.

678.

251.

267.

4211

.26

10.7

36.

349.

003.

3210

.77

7.45

10.0

9M

nO0.

500.

460.

430.

300.

360.

342.

852.

951.

021.

160.

190.

290.

190.

180.

300.

200.

290.

280.

53M

gO2.

913.

563.

512.

522.

373.

05–

–2.

241.

445.

574.

461.

534.

597.

220.

343.

704.

967.

12C

aO6.

758.

799.

444.

456.

599.

016.

686.

197.

241.

7012

.10

9.85

5.55

10.3

611

.55

0.94

7.79

10.7

16.1

7N

a 2O

1.77

4.19

1.29

1.72

5.07

1.64

3.07

2.39

2.64

1.36

5.68

6.01

6.57

5.76

2.44

0.30

6.34

5.00

0.78

K2O

5.66

2.99

2.21

3.57

3.53

4.80

6.97

8.48

4.78

2.14

2.45

1.05

1.26

1.22

0.63

0.76

0.56

0.72

0.76

P 2O

50.

310.

530.

410.

25–

0.43

––

0.24

0.22

0.88

1.03

1.37

1.30

0.93

0.24

0.95

0.75

0.62

Cl 2

O0.

090.

08–

0.07

––

––

0.03

0.04

0.15

––

–0.

080.

770.

400.

080.

16SO

30.

100.

21–

0.23

––

––

0.07

0.10

0.23

0.31

0.64

–0.

110.

110.

180.

240.

11

Tota

l10

010

0.01

99.9

910

0.01

100.

0299

.99

100.

0010

0.00

100.

0010

0.00

100.

0210

0.00

100.

0010

0.02

99.9

910

0.00

100.

00

Na 2

O+

K2O

7.43

7.18

3.50

5.29

8.60

6.44

10.0

410

.87

7.42

8.13

7.06

7.83

6.98

3.07

1.06

6.90

5.72

1.54

Sam

ple

RE

DV

SN2

Ana

lysi

snu

mbe

r10

11A

vera

geSt

.Dev

.1

23

45

67

89

1011

1213

Ave

rage

St.D

ev.

SiO

251

.01

52.6

251

.95

2.19

54.9

854

.460

.48

62.3

151

.26

57.1

655

.19

61.1

762

.47

59.1

455

.69

55.6

555

.84

57.3

73.

45T

iO2

0.88

0.92

0.94

0.27

0.73

0.87

0.98

0.23

1.44

0.82

0.64

0.66

1.15

0.87

0.93

0.83

0.98

0.86

0.28

Al 2

O3

16.9

316

.53

17.6

26.

8836

.94

31.7

20.4

624

.18

14.3

425

.73

37.0

422

.88

18.0

921

.25

20.4

20.7

619

.90

24.1

37.

01Fe

O8.

368.

988.

522.

344.

255.

697.

131.

4412

.86

5.44

4.10

4.28

8.35

5.08

7.24

6.86

7.43

6.17

2.73

MnO

4.70

4.51

1.06

1.75

0.05

0.23

0.05

0.08

0.58

0.15

0.08

0.18

–0.

992.

222.

292.

330.

710.

93M

gO–

–3.

592.

720.

271.

160.

630.

254.

522.

080.

461.

340.

88–

––

–0.

891.

26C

aO8.

969.

249.

383.

870.

374.

232.

711.

4311

.98

5.55

0.46

4.46

3.86

3.29

5.12

5.27

5.41

4.16

2.96

Na 2

O7.

226.

384.

772.

430.

130.

505.

768.

662.

211.

770.

063.

892.

498.

187.

47.

327.

224.

283.

26K

2O0.

550.

760.

970.

550.

581.

031.

360.

960.

460.

750.

800.

931.

611.

191.

001.

020.

890.

970.

30P 2

O5

1.19

–0.

840.

420.

270.

130.

160.

080.

290.

450.

270.

181.

01–

––

–0.

220.

28C

l 2O

0.20

0.07

0.17

0.23

1.35

0.05

0.13

0.23

0.06

0.04

0.71

0.04

0.09

––

––

0.21

0.39

SO3

––

0.18

0.19

0.08

–0.

160.

15–

0.07

0.19

––

––

––

0.05

0.07

Tota

l10

0.00

100.

0010

0.00

99.9

910

0.01

100.

0010

0.00

100.

0110

0.00

100.

0110

0.00

100.

0010

0.00

100.

0010

0.00

Na 2

O+

K2O

7.77

7.14

5.75

0.71

1.53

7.12

9.62

2.67

2.52

0.86

4.82

4.10

9.37

8.40

8.33

8.11

5.24



trachyte. Concentrations of alkali are markedly variable, ranging from 0.06 to 8.66 wt% for Na2Oand from 0.30 to 1.61 wt% for K2O.

Spot analyses were carried out by SEM–EDS on clinopyroxene phenocrysts occurring in theexamined pozzolan samples. Concentrations of major elements are available as electronic supple-mentary material. The results show that the clinopyroxene compositions of Pozzolana Rossaand Pozzolanella are mainly diopsidic (with some terms more Ca-enriched), while compositionsof Pozzolana Nera have a higher variability, falling into the diopside and augite fields of theQUAD diagram (Fig. 11), even in this case with some Ca-enriched terms. Some major elements(e.g., Al, Mg, Fe and Ti), considered as cationic values after the formula calculation, cover a widecompositional range for all analysed clinopyroxenes of the three ignimbrite types (Fig. 12).Specifically, the concentrations (a.p.f.u.) of the above elements vary as follows: Ti between 0.012and 0.052 in RED, 0.011–0.063 in PNR and 0.013–0.043 in VSN2; Al from 0.084 to 0.416 inRED, 0.069–0.383 in PNR and 0.123–0.388 in VSN2; Mg between 0.586 and 0.887 in RED,0.537–1.036 in PNR and 0.555–0.903 in VSN2; Fe from 0.139 and 0.323 in RED, 0.076–0.378in PNR and 0.124–0.376 in VSN2. The diagrams of Figures 12 (a)–12 (c) show that all of theanalysed crystals define a main trend of evolution, although a slightly different behaviour isshown by the compositions of clinopyroxenes in Pozzolana Rossa. Indeed, some crystals ofthe RED type exhibit lower Al contents, at the same concentrations of Fe and Mg (Figs 12 (a) and12 (b)), and higher Ti values, at the same Al contents (Fig. 12 (c)), with respect to PNR and

Figure 10 Total alkali versus silica classification diagram (Le Bas et al. 1986) for the three types of pozzolan collectedfrom the quarry: Pozzolana Rossa (RED), Pozzolana Nera (PNR) and Pozzolanella (VSN2).



VSN2. With regard to the Ca versus Na diagram (Fig. 12 (d)), the Ca concentrations (a.p.f.u.)display a limited variability in the RED and VSN2 types (0.837–0.926 in RED and 0.837–0.935in VSN2): conversely, in PNR, the same element exhibits a wide compositional range (0.714–0.918). With regard to the Na contents, the values in RED type range from 0.012 to 0.038, whilein PNR and VSN2 the concentrations show a higher variability (from 0.012 to 0.082 in PNR and0.012 to 0.072 in VSN2).

LA–ICP–MS analysis

Clinopyroxenes within the aggregate of mortars A total of 84 spot analyses was carried out on52 selected clinopyroxene crystals from the aggregate fraction of examined mortars to obtainthe trace element concentrations. Two analyses (core and rim) were performed on the biggerclinopyroxenes, while only one was performed on the smaller crystals. The full data set obtainedis available as electronic supplementary material. On the whole, Sc, V andY along with rare earthelements (REE), both light and heavy, exhibit the most significant variations among the suiteof elements analysed by LA–ICP–MS. The minimum and maximum values of some noteworthyelements are reported as follows: Sc, 11.1–156 ppm; V, 74.7–484 ppm; Y, 10.4–59.7 ppm;La, 9.72–82.6 ppm; Nd, 31.3–62 ppm; Sm, 7.13–34.4 ppm; Eu, 1.49–7.02 ppm; Dy, 3.20–

Figure 11 QUAD diagram (Morimoto 1988) for clinopyroxenes of the three pozzolan types from the quarry: PNR,Pozzolana Nera; RED, Pozzolana Rossa; VSN2, Pozzolanella.



Figu

re12

Bin

ary

diag

ram

ssh

owin

gth

eco

ncen

trat

ions

ofth

em

osts

igni

fican

tmaj

orel

emen

ts(p

lott

edas

cati

onic

valu

es:

a.p.

f.u.)

incl

inop

yrox

ene

crys

tals

from

the

pozz

olan

sam

ples

from

the

quar

ry:

Cpx

_RE

D,P

ozzo

lana

Ros

sa;

Cpx

_PN

R,P

ozzo

lana

Ner

a;C

px_V

SN2,

Pozz

olan

ella

.



16.68 ppm; Er, 0.64–5.62 ppm. To better investigate such compositional variability observedamong clinopyroxenes in the mortars, a graphical representation of data was made through binarydiagrams. Two transition elements—that is, Sc and V—were chosen as index elements of theevolution degree, since they are strongly compatible in clinopyroxenes and can be directlycorrelated with the degree of differentiations of the liquid. In Figure 13, reporting some of themost significant diagrams, the trace element concentrations of clinopyroxenes within the volcanicaggregate of mortars show a rather high degree of dispersion, which prevents us from defining aunique trend. This could be due to the existence, among the suite of analysed mortars, of morethan one family of clinopyroxenes with variable REE concentrations at constant values of Sc andV (Fig. 13).

Clinopyroxenes of the pozzolan samples from the quarry Seventy spot analyses by means ofLA–ICP–MS were carried out on 45 clinopyroxene phenocrysts occurring in the three pozzolan

Figure 13 Binary diagrams of the most significant trace elements in clinopyroxenes within the volcanic aggregate ofmortars.



types sampled (Pozzolana Rossa, Pozzolana Nera and Pozzolanella). The obtained concentra-tions are reported as electronic supplementary material. Even in this case, significant variationswere observed for all the rare earth elements (REE), both light and heavy, as well as for Sc, V andY. In particular, the lower contents of REE refer to clinopyroxenes from Pozzolana Nera,compared to Pozzolana Rossa and Pozzolanella, which show relatively higher and comparableconcentrations, especially for V,Y, Nd, Sm, Eu, Dy, Ho and Er. Minimum, maximum and averagevalues (in ppm) of the most significant REE in the three types are reported in Table 4.

With regard to Sc, clinopyroxene crystals from the Pozzolana Rossa type are characterizedby lower average concentrations (minimum 19.7, maximum 107, average 60.5) compared withPozzolana Nera and Pozzolanella ones (minimum 44.3, maximum 167, average 99.1 in PNR;minimum 43.9, maximum 159, average 87.0 in VSN2).

The binary diagrams of Figure 14 better display the variability of Sc and V (used again as indexelements of the evolution degree) and REE in the three pozzolan types. In particular, in thediagrams of Sc versus REE (Figs 14 (a)–14 (d)), clinopyroxenes define two distinct groups,whose separation is at about 45 ppm of Sc. Crystals of PNR and VSN2 types entirely fall in thegroup characterized by higher concentrations of Sc, whereas those of RED type are distributedin both groups. With regard to the binary diagrams of V versus REE (Figs 14 (e)–14 (h)), onthe whole, a positive correlation was observed, although two main trends can be identified,as indicated in figure: the first trend, defined by clinopyroxenes of Pozzolana Rossa andPozzolanella, is characterized by higher concentrations of REE, whereas the second, referring tocrystals of Pozzolana Nera, displays lower REE contents, at the same concentrations of V.

DISCUSSION

The petrographic analysis carried out on mortars from the Villa dei Quintili allowed us to identifyfour different groups on the basis of textural and compositional features (the type/abundance ofthe aggregate, and the colour, texture and degree of compactness of the binder). However,the common presence of volcanic aggregate (in variable amounts) characterizes all of them. Itconsists of both pozzolan fragments and single crystals of leucite, clinopyroxene (both colourlessand green) and biotite, which are, on the whole, compatible with products of the ultrapotassicRoman Magmatic Province. In this respect, due to the location of the investigated archaeological

Table 4 Minimum, maximum and average values of the trace element concentrations (ppm), determined throughLA–ICP–MS analysis, of pozzolan samples from the quarry (PNR, Pozzolana Nera; RED, Pozzolana Rossa;

VSN2, Pozzolanella)

Samples V Y Nd Sm Eu Dy Ho Er

RED Minimum 82.2 10.9 30.9 4.56 1.26 2.11 0.27 0.21Maximum 229 26.7 96.6 20.1 4.19 9.59 2.02 5.15Average 152 19.7 64.8 13.5 2.83 5.29 0.87 2.16

PNR Minimum 45.5 6.52 19.3 3.99 0.97 1.44 0.15 0.31Maximum 331 23.3 81.2 17.3 3.44 7.14 1.32 2.76Average 140 13.3 45.2 9.74 1.80 3.66 0.51 1.21

VSN2 Minimum 98.1 14.0 37.7 8.16 1.68 2.79 0.53 1.31Maximum 273 39.1 101 23.1 4.66 9.16 1.56 3.63Average 174 22.9 63.2 14.6 2.99 6.05 0.97 2.17



Figure 14 Binary diagrams of the most significant trace elements in clinopyroxene crystals occurring within thepozzolan samples from the quarry: RED, Pozzolana Rossa; PNR, Pozzolana Nera; VSN2, Pozzolanella. The dashed linesrepresent the R2 values for each group of clinopyroxenes (RED + VSN2 and PNR), while the continuous line representsthe R2 value of the entire data set.



site and indications from the existing literature, a correspondence of such a volcanic aggregatewith products from the Colli Albani volcanic district can be drawn. In addition, the frequentpresence of abundant phenocrysts of leucite in all examined mortars allows us to exclude anaffinity with the other volcanic district rising near to the site; in other words, the Monti Sabatinivolcanic district, whose products show rare leucite phenocrysts (this mineral being restricted tothe groundmass) and the common presence, in the phonolitic terms, of plagioclase and sanidine,which are totally absent in the aggregate of mortars (Peccerillo 2005). Comparisons between thepetrographic features of pozzolan fragments within the mortars and those of the three pozzolantypes from the quarry (Pozzolana Rossa, Pozzolana Nera and Pozzolanella) suggest the use of allthree types in the preparation of the examined mortars. Specifically, the presence of large leucitephenocrysts and the abundance of femic phases (clinopyroxenes and biotite) are distinctivefeatures of Pozzolana Nera and Pozzolanella, respectively, whereas the occurrence of fragmentscharacterized by small leucite phenocrysts and the low abundance of femic phases are charac-teristic of the Pozzolana Rossa type.

Micromorphological investigations (the texture of the groundmass, and the size and abundanceof vacuoles) of both the volcanic aggregate and the pozzolan samples show a marked variability.However, these features are dependent upon the cooling rate of the pyroclastic deposit andtherefore can be variable even in outcrops characterized by limited thickness and extension.Therefore, micromorphology cannot be considered as an effective discriminating feature of thepozzolan type(s) used in the preparation of the mortars. For such a purpose, a detailed study ofthe chemical composition of the volcanic aggregate and pozzolan samples was carried out. Froma first observation, it was noted that the chemical compositions of the volcanic aggregate in thesingle petrographic groups do not differ considerably from each other, so that in the followinggraphical representations of the results, all of the data set was considered as a unique group. Inthe TAS diagram (Fig. 15), compositions obtained by the SEM–EDS analysis of the volcanicaggregate and quarry pozzolan samples were compared with those taken from the literaturereferring to the three pozzolan types from the Alban Hills (Fornaseri et al. 1963; Bakos et al.1994; Trigila et al. 1995; Boari et al. 2009; Jackson et al. 2009, 2010; Dickie 2010). The diagramshows a large overlapping area between the literature data for the three pozzolan types, although,on the whole, the compositions of Pozzolana Rossa are the less evolved, those of Pozzolanellainclude the most evolved terms, and the Pozzolana Nera samples display intermediate compo-sitions. Although there is a certain degree of dispersion towards the low-alkali terms, thecompositions of each pozzolan type from the quarry (RED, PNR and VSN2) match those fromthe literature well. The decrease in the alkali content (at the same SiO2 values) of some data canbe due to weathering processes of the cineritic fraction, which may have produced a selectivedepletion in alkali (Conticelli et al. 2010). In the same diagram (Fig. 15), the compositions of thevolcanic rock fragments within mortars fall in the area where the three types overlap, thussuggesting the possible use of all of the pozzolan types. In this respect, the compositional datarelative to major elements are in agreement with the above-mentioned hypothesis based onpetrographic observations.

In order to obtain more detailed information about the three pozzolan types, a compositionalstudy of clinopyroxene crystals occurring within the volcanic aggregate of mortars and pozzolansfrom the quarry was performed. Clinopyroxenes were chosen among the other volcanic mineralssince they represent the most abundant mineralogical phase, along with leucite, in the Alban Hillsproducts and, in addition, exhibit a large chemical variability that is indicative of the magmaticaffinity of the erupted products (Dallai et al. 2004; Freda et al. 2006). Even in this case, nonoticeable differences were observed among the clinopyroxene compositions of the four



petrographic groups. In the binary diagrams of Figure 16, the variability of major elementsin clinopyroxenes within the volcanic aggregate of mortars and those from the RED, PNR andVSN2 pozzolan types was compared. In these diagrams, the compositional range identified bycrystals from the mortars outlines a trend that perfectly superimposes that defined by crystals ofthe RED, PNR and VSN2 pozzolan types (Figs 16 (a)–16 (d)).

Attention was then paid to the study of trace element concentrations in the same crystalsthrough the use of laser ablation inductively coupled plasma mass spectrometry (LA–ICP–MS).In clinopyroxenes of both mortars and pozzolans from the quarry, the most significant variationsamong the suite of analysed elements were observed for all the rare earth elements (REE), bothlight and heavy, as well as for Sc, V and Y (Figs 13 and 14). Comparisons between the traceelement compositions of the entire suite of analysed samples are shown in Figure 17, where themost representative binary diagrams are reported. Observations of such diagrams reveal that thetrace element concentrations in clinopyroxenes within the pozzolanic aggregate of mortars coverboth the trend typical of PNR, characterized by high Sc concentrations (Figs 17 (a)–17 (d)) andlow REE contents at the same concentrations of V (Figs 17 (e)–17 (h)), as well as the trenddefined by the RED and VSN2 types, with higher REE values at the same V contents. Theseresults, along with petrographic evidence and data of major element contents in clinopyroxenesand pozzolan fragments of both mortars and samples from the quarry, seem to support the

Figure 15 Total alkali versus silica classification diagram (Le Bas et al. 1986) for the volcanic aggregate of mortars andcomparison with data of the three pozzolan types from both the quarry (RED, Pozzolana Rossa; PNR, Pozzolana Nera;VSN2, Pozzolanella) and the literature.



Figu

re16

Bin

ary

diag

ram

ssh

owin

gth

eco

ncen

trat

ions

ofth

em

ost

sign

ifica

ntm

ajor

elem

ents

(plo

tted

asca

tion

icva

lues

:a.

p.f.u

.)in

clin

opyr

oxen

ecr

ysta

lsfr

ompo

zzol

ans

ofbo

thm

orta

rs(C

px_V

Q)

and

rock

sfr

omth

equ

arry

(Cpx

_RE

D,P

ozzo

lana

Ros

sa;

Cpx

_PN

R,P

ozzo

lana

Ner

a;C

px_V

SN2,

Pozz

olan

ella

).



Figure 17 Binary diagrams of trace element concentrations (ppm) of clinopyroxenes within the volcanic aggregate ofmortars (Cpx_VQ) and the pozzolan samples from the quarry (RED, Pozzolana Rossa; PNR, Pozzolana Nera; VSN2,Pozzolanella).



hypothesis that the Roman builders used the Pozzolana Rossa, Pozzolana Nera and Pozzolanellatypes as the aggregate fraction for the preparation of the examined mortars.

CONCLUDING REMARKS

In this work, several Roman mortars taken from the monumental complex of the Villa dei Quintiliwere analysed. It is a relevant archaeological site near Rome (Italy), which was used as animperial residence for at least two centuries (from the second to the third century). A distinctivegeochemical approach based on the investigation of major and trace elements of clinopyroxenecrystals (up to now successfully applied only to provenance studies of ceramics), was employedfor the first time to the study of pozzolanic materials used as the aggregate fraction in theexamined mortars. It is known that the Roman builders, before and during the Imperial age,used to employ three different types of pozzolans (Pozzolana Rossa, Pozzolana Nera andPozzolanella), all of which erupted from the Alban Hills volcanic district. From the obtainedresults, the following considerations can be drawn:(a) no correlations were observed between the petrographic groups and chemical data of theexamined mortars, due to an overall homogeneity that characterizes the aggregate fraction of allsamples; and(b) both the petrographic and chemical evidence point to the use of all three pozzolan types inthe analysed mortars, thus emphasizing the importance of a multidisciplinary approach.

ACKNOWLEDGEMENTS

The authors are grateful to Dr Rita Paris, Archaeologist Director of Soprintendenza Speciale peri Beni Archeologici di Roma, for allowing archaeological samples to be taken. They also wouldlike to thank Dr Marco Viccaro of the University of Catania, for his help in interpretation ofvolcanic data, and Dr Guido Giordano, of the University of Rome ‘La Sapienza’, for kindlyproviding pozzolan samples from quarries.

REFERENCES

Bakos, F., Cimitan, L., Rossi, P., and Zanetti, A., 1994, Caratterizzazione petrologica e chimica di malte pozzolanicheantiche, Materiali e strutture: Problemi di conservazione, 1, 1–20.

Barone, G., Belfiore, C. M., Mazzoleni, P., Pezzino, A., and Viccaro, M., 2010, A volcanic inclusions based approach forprovenance studies of archaeological ceramics: application to pottery from southern Italy, Journal of ArchaeologicalScience, 37, 713–26.

Belfiore, C. M., La Russa, M. F., Barca, D., Galli, G., Pezzino, A., Ruffolo, S. A., Viccaro, M., and Fichera, G. V., 2014,A trace element study for the provenance attribution of ceramic artefacts: the case of Dressel 1 amphorae from alate-Republican ship, Journal of Archaeological Science, 43, 91–104.

Boari, E., Avanzinelli, R., Melluso, L., Giordano, G., Mattei, M., De Benedetti, A., Morra, V., and Conticelli, S., 2009,Isotope geochemistry (Sr–Nd–Pb) and petrogenesis of leucite-bearing volcanic rocks from ‘Colli Albani’ volcano,Roman Magmatic Province, central Italy: inferences on volcano evolution and magma genesis, Bulletin of Volcanol-ogy, 71, 977–1005.

Collepardi, M., 2003, La lezione dei romani: durabilità e sostenibilità delle opere architettoniche e strutturali, inProceedings of III Convegno AIMAT ‘Restauro e Conservazione dei Beni Culturali: Materiali e Tecniche’, 3–4October 2003, Cassino, Italy.

Conticelli, S., Boari, E., Avanzinelli, R., De Benedetti, A. A., Giordano, G., Mattei, M., Melluso, L., and Morra, V., 2010,Geochemistry, isotopes and mineral chemistry of the Colli Albani volcanic rocks: constraints on magma genesis andevolution, in The Colli Albani volcano (eds. R. Funiciello and G. Giordano), 107–39, Special Publications of IAVCEI,3, The Geological Society, London.



Dallai, L., Freda, C., and Gaeta, M., 2004, Oxygen isotopic geochemistry of pyroclastic clinopyroxene monitorscarbonate contributions to Roman-type ultrapotassic magmas, Contributions to Mineralogy and Petrology, 148,247–63.

Day, R. L., 1990, Pozzolans for use in low-cost housing, a state of the art report, 38–80, Department of Civil Engineering.University of Calgary. International Development Research Centre, Ottawa.

De Rita, D., Faccenna, C., Funiciello, R., and Rosa, C., 1995, Stratigraphy and volcano-tectonics, in The volcano of theAlban Hills (ed. R. Trigila), 33–71, Università degli Studi di Roma ‘La Sapienza’, Roma.

Dickie, J. M., 2010, Mineralogical and geochemical indicators of subaerial weathering in the Pozzolane Rosse ignimbrite(Alban Hills Volcanic District, Italy), Geosciences Theses, Paper 23, http://digitalarchive.gsu.edu/geosciences_theses/23 (accessed 16 May, 2013).

Fernández, R., Nebreda, B., Vigil de la Villa, R., García, R., and Frías, M., 2010, Mineralogical and chemical evolu-tion of hydrated phases in the pozzolanic reaction of calcined paper sludge, Cement and Concrete Composites, 32,775–82.

Fichera, G. V., Barca, D., Belfiore, C. M., La Russa, M. F., and Pezzino, A., 2012, Technological and geochemical studyof historical mortars from the Roman ‘Villa dei Quintili’ (Rome, Italy), Rendiconti Online Società Geologica Italiana,21(1), 659–60.

Fornaseri, M., Scherillo, A., and Ventriglia, U., 1963, La regione vulcanica dei Colli Albani: Vulcano Laziale, C.N.R.(Consiglio Nazionale delle ricerche), Roma.

Freda, C., Gaeta, M., Palladino, D. M., and Trigila, R., 1997, The Villa Senni Eruption (Alban Hills, central Italy): therole of H2O and CO2 on the magma chamber evolution and the eruptive scenario, Journal of Volcanology andGeothermal Research, 78, 103–20.

Freda, C., Gaeta, M., Karner, D. B., Marra, F., Renne, P. R., Taddeucci, J., Scarlato, P., Christensen, J. N., and Dalla, L.,2006, Eruptive history and petrologic evolution of the Albano multiple maar (Alban Hills, central Italy), Bulletin ofVolcanology, 68, 567–91.

Frontoni, R., 2000, Il cantiere antico a Villa dei Quintili, Forma Urbis—Itinerari nascosti di Roma Antica, Roma, ANNOV(7/8), 42–7.

Frontoni, R., and Galli, G., 2006, s.v. Quintiliorum praedium, in Lexicon topographicum urbis Romae, Suburbium IV,Roma.

Frontoni, R., and Galli, G., 2010, Calce e calcara nella Villa dei Quintili, Arkos, Scienza e Restauro, 25, Ottobre–Dicembre, 66–73.

Frontoni, R., Galli, G., and Paris, R., 2010, Villa dei Quintili: guida pieghevole, Ed. Electa, Milano.Fryer, B. J., Jackson, S. E., and Longerich, H. P., 1995, The design, operation and role of the laser-ablation microprobe

coupled with an inductively coupled plasma-mass spectrometer (LAM–ICP–MS) in the Earth sciences, The CanadianMineralogist, 33, 303–12.

Gao, S., Liu, X., Yuan, H., Hattendorf, B., Gunther, D., Chen, L., and Hu, S., 2002, Determination of forty-two major andtrace elements in USGS and NIST SRM glasses by laser ablation-inductively coupled plasma mass spectrometry,Geostandards Newsletter: The Journal of Geostandards and Geoanalysis, 26, 181–96.

Giavarini, C., Samuelli, M., Ferretti, A., and Santarelli, M. L., 2006, Mechanical characteristics of Romanopus caementicium, in Fracture and failure of natural building stone (ed. S. K. Kourkolis), 107–20, Springer,Dordrecht.

Giordano, G., De Benedetti, A. A., Diana, A., Diano, G., Gaudioso, F., Marasco, F., Miceli, M., Mollo, S., Cas, R. A. F.,and Funiciello, R., 2006, The Colli Albani mafic caldera (Roma, Italy): stratigraphy, structure and petrology, Journalof Volcanology and Geothermal Research, 155, 49–80.

Jackson, M. D., and Marra, F., 2006, Roman stone masonry: volcanic foundations of the ancient city, American Journalof Archaeology, 110(3), 403–36.

Jackson, M. D., Deocampo, D., Marra, F., and Scheetz, B., 2010, Mid-Pleistocene pozzolanic volcanic ash in ancientRoman concretes, Geoarchaeology: An International Journal, 25(1), 36–74.

Jackson, M. D., Kosso, C., Marra, F., and Hay, R., 2006, Geological basis of Vitruvius’pirical observations of materialscharacteristics of rock utilized in Roman masonry, in Proceedings of the Second International Congress of Construc-tion History, Queen’s College, University of Cambridge (eds. M. Dunkeld, J. Campbell, H. Louw, M. Tutton,B. Addis and R. Thorne), 1685–702, The Construction History Society, 2, London.

Jackson, M. D., Marra, F., Hay, R. L., Cawood, C., and Winkler, E. M., 2005, The judicious selection and preservationof tuff and travertine building stone in ancient Rome, Archaeometry, 47, 485–510.

Jackson, M. D., Logan, J. M., Scheetz, B. E., Deocampo, D. M., Cawood, C. G., Marra, F., Vitti, M., and Ungaro, L.,2009, Assessment of material characteristics of ancient concretes, Grande Aula, Markets of Trajan, Rome, Journal ofArchaeological Science, 36, 2481–92.



http://digitalarchive.gsu.edu/geosciences_theses/23

http://digitalarchive.gsu.edu/geosciences_theses/23

Karner, D. B., Marra, F., and Renne, P. R., 2001, The history of the Monti Sabatini and Alban Hills volcanoes:groundwork for assessing volcanic–tectonic hazard for Rome, Journal of Volcanology and Geothermal Research,107, 185–219.

Lancaster, L. C., 2009, Concrete vaulted construction in Imperial Rome: innovations in context, Cambridge UniversityPress, Cambridge, UK.

Le Bas, M. J., Le Maitre, R. W., Streckeisen, A., and Zanettin, B., 1986, A chemical classification of volcanic rocks basedon the total alkali diagram, Journal of Petrology, 27, 745–50.

Massazza, F., 1998, Pozzolana and pozzolanic cements, in Lea’s chemistry of cement and concrete (ed. P. C. Hewlett),471–602, Arnold, London.

Morimoto, N., Fabries, J., Ferguson, A. K., Ginzburg, I. V., Ross, M., Seifert, F. A., Zussman, J., Aoki, K., and Gottardi,G., 1988, Nomenclature of pyroxenes, Mineralogy and Petrology, 39, 55–76.

Moropoulou, A., Cakmak, A., Labropoulos, K. C., Van Grieken, R., and Torfs, K., 2004, Accelerated microstructuralevolution of a calcium–silicate–hydrate (C–S–H) phase in pozzolanic plaster using fine siliceous sources: comparisonwith historic pozzolanic mortars, Cement and Concrete Research, 34, 1–6.

Pavía, S., 2008, A petrographic study of mortar hydraulicity, in Historical mortars conference: characterization,diagnosis, conservation, repair and compatibility, LNEC General Civil Engineering Laboratory, Lisbon.

Pavía, S., and Caro, S., 2008, An investigation of Roman mortar technology through the petrographic analysis ofarchaeological material, Construction and Building Materials, 22, 1807–11.

Pearce, N. J. G., Perkins, W. T., Westgate, J. A., Gorton, M. P., Jackson, S. E., Neal, C. R., and Chenery, S. P., 1997, Acompilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glassreference materials, Geostandards Newsletter: The Journal of Geostandards and Geoanalysis, 21, 115–44.

Peccerillo, A., 2005, Plio-Quaternary volcanism in Italy—petrology, geochemistry, geodynamics, Springer-Verlag,Berlin.

Rotondi, A., 2012, La Villa dei Quintili e i suoi antichi proprietari, Forma Urbis—Itinerari nascosti di Roma Antica,Roma, ANNO XVII(2), 6–8.

Sánchez-Moral, S., Luque, L., Cañaveras, J. C., Soler, V., Garcia-Guinea, J., and Aparicio, A., 2005, Lime pozzolanamortars in Roman catacombs: composition, structures and restoration, Cement and Concrete Research, 35, 1555–65.

Silva, D. A., Wenk, H. R., and Monteiro, P. J. M., 2005, Comparative investigation of mortars from Roman Colosseumand cistern, Thermochimica Acta, 438, 35–40.

Trigila, R., Agosta, E., Currado, C., De Benedetti, A. A., Freda, C., Gaeta, M., Palladino, D. M., and Rosa, C., 1995,Petrology, in The volcano of the Alban Hills (ed. R. Trigila), 95–165, Università degli Studi di Roma ‘La Sapienza’,Roma.



A multidisciplinary approach for the archaeometric study of pozzolanic aggregate in roman mortars:...

Documents

Transcript of A multidisciplinary approach for the archaeometric study of pozzolanic aggregate in roman mortars:...