Smartphone Application For Ancient Mortars Identifcation Developed by A

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of Archaeological Science: Reports 43 (2022) 103433

Contents lists available at ScienceDirect

Journal of Archaeological Science: Reports


journal homepage: www.elsevier.com/locate/jasrep

Smartphone application for ancient mortars identification developed by a


multi-analytical approach
Mirco Ramacciotti a, b, Gianni Gallello a, *, Marco Lezzerini c, Stefano Pagnotta c, Andrea Aquino c,
Llorenç Alapont a, Juan Antonio Martín Ruiz d, Alejandro Pérez-Malumbres Landa e,
Ramón Hiraldo Aguilera f, David Godoy Ruiz g, Angel Morales-Rubio b, M. Luisa Cervera b,
Agustín Pastor b
a
Department of Prehistory, Archaeology and Ancient History, University of Valencia, Avenida de Blasco Ibañez 28, 46010 Valencia, Spain
b
Department of Analytical Chemistry, University of Valencia, Edificio Jeroni Muñoz, Dr. Moliner 50, 46100 Burjassot, Spain
c
Department of Earth Sciences, University of Pisa, via S. Maria 53, 56126 Pisa, Italy
d
Universidad Internacional de Valencia, c/del Pintor Sorolla 21, 46002 Valencia, Spain
e
Department of Culture and Historical Heritage of Andalusian Government, Edificio Eurocom, c/Mauricio Moro Pareto 2, 29006 Málaga, Spain
f
Institute of Studies of Ronda and la Serranía, c/Virgen de la Paz 15, 29400 Ronda, Málaga, Spain
g
Múrex Arqueólogos, S. L. c/Santa Teresa 6, 29651 Mijas, Málaga, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: The present work shows the results of the chemical, mineralogical and colorimetric characterisation of the
Smartphone ancient mortars from Silla Islamic Tower (Valencia, Spain) and Fuengirola Castle (Malaga, Spain). The samples
REE were characterised from the mineralogical point of view by X-ray diffractometry and mid-infrared attenuated
pED-XRF
total reflection spectroscopy, while portable energy dispersive X-ray fluorescence spectroscopy and inductively
Imaging
CIELAB
coupled plasma mass spectrometry were employed to obtain the concentrations of major and trace elements,
Mortars including rare earth elements. Data analysis through multivariate statistics was used to evaluate features to
Islamic period discriminate among the mortars from the different construction phases and to classify undated samples. Finally,
colour features of powdered and intact samples were characterised by smartphone photo processing and with a
Vis-spectrophotometer as reference technique to evaluate the effectiveness of smartphones for archaeometric
studies of historic mortars. The analytical results permitted the classification of most undated samples from Silla
and evidenced the presence of peculiar chemical characteristics in some samples from Fuengirola. Imaging data
for powdered samples showed a good potential as a reliable, cheap and non-destructive fast method to char­
acterise mortars and carry out the study of construction phases in historical complexes.

1. Introduction Several methods have been employed to characterise ancient mor­


tars, including the use of different analytical techniques such as optical
1
The analysis of architecture and ancient building materials is of microscopy (OM), thermogravimetry with differential scanning calo­
particular interest for different fields of study being a fundamental step rimetry (TG-DSC), X-ray diffractometry (XRD), energy dispersive X-ray
for the archaeological interpretation of architectural heritages con­ fluorescence (ED-XRF) and inductively coupled plasma mass spectrom­
struction phases, development of a complex, the employed construction etry (ICP-MS) (Crisci et al., 2004; Lezzerini et al., 2014; 2018). The use
techniques, the provenance of raw materials and their manufacture, as of these archaeometric methods, often carried out during archaeological
well as the implication of all these aspects for social history, and to works or in the study of a monument, can provide important information
program restoration interventions (Ferris, 1989; Francovich and Bian­ concerning the employed raw materials and the production technology
chi, 2002; Brogiolo, 2007; Dessales, 2017; Columbu et al., 2018; Azka­ of mortars (Miriello et al., 2015; Lezzerini et al., 2016; Sitzia et al., 2020;
rate, 2020). Cantisani et al., 2021) and can be a support for the interpretation of the

* Corresponding author.
E-mail address: gianni.gallello@uv.es (G. Gallello).
1
This paper is part of the Special Issue “Advances in Archaeometry during COVID-19 pandemic” edited by Celestino Grifa, Donata Magrini e Francesco Izzo.

https://doi.org/10.1016/j.jasrep.2022.103433
Received 20 November 2021; Received in revised form 7 February 2022; Accepted 5 April 2022
Available online 11 April 2022
2352-409X/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

wall stratigraphy, recognising similarities and diversities among group photos as a test to evaluate the performance and the possible employ­
of samples driven by different manufacturing recipes (Corti et al., 2013; ment of this last device in ancient mortar archaeometric studies for
Chiarelli et al., 2015; Lezzerini et al., 2018). Mortars trace elements have construction phases identification. Colour features were characterised in
been employed as markers of different construction phases (Miriello previous studies by colorimetric devices in order to investigate techno­
et al., 2010a). For example, in the last few years, Gallello and colleagues logical aspect of the production of ancient mortars (e.g.: Miriello et al.,
successfully used the rare earth elements (REE) data, processed though 2010b; Centauro et al., 2017). However, although smartphone as a non-
multivariate statistics, as elemental markers of the different construction destructive, fast and cheap analytical tool is more and more employed in
phases for the structures of Sagunto Castle and its surroundings (Gallello various fields of analytical chemistry (e.g.: Cruz-Fernández et al., 2017;
et al., 2017; Ramacciotti et al., 2018). Rezazadeh et al., 2019; Herreros-Chavez, 2021), its use is still under­
The present study shows the results of the archaeometric charac­ developed in archaeological science (De Luca et al., 2021). In this study,
terisation of the ancient mortars from the Islamic Tower of Silla mortars were analysed and for the first time imaging data from smart­
(Valencia, Spain) and the Castle of Fuengirola (Malaga, Spain; Fig. 1) in phone photos were cross-referenced with other analytical data to eval­
order to understand the construction phases of the wall structures, uate the image analysis effectiveness in distinguishing among samples
implementing non-invasive screening approaches such as image analysis from different periods. The test was especially developed for the study of
on smartphone photos. historical complexes characterised by several building interventions
The Islamic Tower of Silla (Fig. 1b) was founded in the 11th century through the centuries. Indeed, the proposed developed approach could
CE as a defensive building. This was part of the defensive belt of the city become useful as a prior screening step before running further invasive
of Valencia and is built on the basis of Roman ashlars. Its function was to sampling strategies for standard methods.
defend the population from possible enemy attacks and served also as a
refuge for the population. Later, it was employed as prison and as a 2. Materials and methods
warehouse. It is mentioned by historical fonts in Wars of the Union
(1347–1348), in War of Castile (1356–1375) and in the episodes of 2.1. The mortars
Germanía of 1521, being abandoned during the 19th century (Alapont
et al., 2016). The samples collected from both the Silla Islamic Tower and Fuen­
About Fungirola Castle (Fig. 1c), such as the Silla Tower, it was built girola Castle are shown in Table 1.
by the Muslims in the 11th century CE to control an area of the coast. It Twenty-one samples come from Silla Islamic Tower. Nine samples
was occupied by Christians in 1485 and suffered several reforms in the were collected from the rammed earth external northern (N3A-B, N4),
16th and 18th centuries, when its towers were demolished and several western (O8-10) and southern (S6, SA-B) walls of the tower, pertaining
canvases were enlarged for the placement of artillery. In the 19th cen­ to different construction phases from the Islamic period to the past
tury one of the walls was rebuilt and at the end of that century it became century. Four samples of bedding mortar were collected from an Islamic
a police station (Aguilar Cuesta et al., 2019). structure (P0M1-2), found in the ground floor (P0) of the Tower, and
Multielement analysis employing a portable ED-XRF (pED-XRF) and from the remains of a Roman Villa (P0M3-4). Eight samples were
an ICP-MS, and mineralogical characterisation through XRD were car­ collected in the interior walls of the tower: four samples come from the
ried out for all the samples from the two historical complexes. The first floor (P1), three from the second (P2) and one from the third (P3).
mortars from Fuengirola Castle were also characterised by attenuated P1M1-2, P2M1-2 and P3M1 come from the rammed earth structure and
total reflection infrared spectroscopy (ATR-IR). Finally, colour features dated back to the Islamic phase of the tower, except for P2M3 which is
of mortars from both the sites were measured by a portable Vis- possibly a modern restoration. P1QU comes from the ceiling lathwork
spectrophotometer and through image processing of smartphone structure, which pertains to the Islamic phase as well, while P1EN is a

Fig. 1. Localisation of Fuengirola and Silla towns in the Iberian Peninsula (a), Silla Islamic Tower (b) and 3D reconstruction of Fuengirola Castle (c).

2
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

Table 1 mineralogical qualitative identification was performed by DIFFRAC.


Samples from the two sites, localisation and dating. EVA v4.1.1 software (by Bruker).
Sample Location* Dating Sample Location* Dating

Silla Islamic Tower P2M1 Second Islamic


floor (int.) 2.3. Attenuated total reflection infrared spectroscopy (ATR-IR)
N3A Northern Islamic P2M2 Second Islamic
wall (ext.) floor (int.) The powdered samples of Fuengirola Castle were analysed with a
N3B Northern Islamic P2M3 Second Modern portable spectrometer 4300 Handheld FT-IR by Agilent Technologies.
wall (ext.) floor (int.) restoration
N4 Northern Christian P3M1 Third floor Islamic
Each spectrum is the average of 50 scans. Background was measured
wall (ext.) (?) (int.) between an analysis and the others. Resolution of 4 cm− 1 and Boxcar
O8 Western 20th c. Fuengirola Castle apodization were employed.
wall (ext.)
O9 Western 20th c. M01 Section 3rd 13th or 13th-
wall (ext.) 15th c. (?)
O10 Western 14th c. M02 Tower 3 14th c. (?) 2.4. Multielement analysis
wall (ext.)
S6 Southern Islamic M03 Section 2nd 14th c. (?) Major elements (Al, Si, K, Ca, Ti and Fe) concentrations were
wall (ext.) measured in all the samples with a portable energy dispersive X-ray
SA Southern (?) M04 Section 4th Beginning of
wall (ext.) 19th c.
fluorescence spectrometer (pED-XRF) S1 Titan by Bruker, equipped with
SB Southern (?) M05$ Tower 4 10th or 12th a Rh X-ray tube (50 kV) and X-Flash® SDD (resolution: 147 eV; FWHM:
wall (ext.) c. (?) 5.9 keV). Geochem-trace application was employed for the analysis.
P0M1 Ground Islamic M06 Section 8th End of 18th c. Although its high limits of detection compared with the standard
floor (int.)
laboratory spectrometers, pED-XRF is increasingly employed in
P0M2 Ground Islamic M07 Tower 1 12th c.
floor (int.) archaeometry, also in order to obtain quantitative and semi-quantitative
P0M3 $
Ground Roman M08 Tower 1 16th c. elemental data of ancient mortars (Donais et al., 2009; Tenconi et al.,
floor (int.) 2018). Internal calibration was adjusted using geologic certified refer­
P0M4$ Ground Roman M09 Section Ending 18th ence materials, and the accuracy and precision of the analytical data
floor (int.) 10th c. (?)
P1EN First floor (?) M10 Section 1st 14th c. (?)
were controlled employing NIM-GBW07408 (Soil) and NCS DC 73375
(int.) (Limestone) reference materials (Annex 1).
P1QU First floor Islamic M11B Section 9th (?) Trace elements (Ba, Bi, Cd, Cr, Co, Cu, Pb, Li, Mn, Mo, Ni, Sr, Th, Tl,
(int.) b U, V, Zn), including rare earth elements (REE: La, Ce, Pr Nd, Sm, Eu, Gd,
P1M1 First floor Islamic M11W Section 9th 14th c. (?)
Tb, Dy, Ho, Er, Tm, Yb, Lu, and Sc and Y), concentrations were analysed
(int.) b
P1M2 First floor Islamic M12 Section 7th 14th c. (?) by an inductively coupled plasma mass spectrometer (ICP-MS) Elan
(int.) DRCII by Perkin Elmer. The powdered samples were previously brought
$ in solution by acid attack (aqua regia). Details on sample preparation and
Notes: * ext. = external, int. = interior; indicates earth mortars; (?) indicates
uncertain dating; c.: century.
equipment setting can be found in Gallello et al., 2017. Accuracy and
precision of the analytical data were controlled employing NIM-
GBW07408 and NCS DC 73375 reference materials.
very thin layer of plaster from an uncertain phase. The macroscopic
features of the Roman samples suggest that they can be classified as
earthen mortars, an ancient technique in which the mortar was the
2.5. Colour analysis
result of the mixing of soil and water and, in some cases a stabiliser, such
as lime, plant or organic matter (Gómez Morgade et al., 2021) while the
Colour features were measured in powdered samples from the two
others as lime mortars, characterised by sandy aggregate, whitish-light
buildings and in intact samples of Fuengirola Castle. The analyses were
gray binder and presence of lumps, except for P1EN in which aggre­
carried out employing reflectance spectroscopy in the visible region
gate was not visible by naked-eye.
(Vis-SP) as reference technique through a portable CM-26d spectro­
Thirteen samples of mortars were collected from different sectors of
photometer by Konica Minolta, calibrated with reference white and
the Fuengirola Castle. Samples M01-03, M05, M09-12 come from ram­
black to convert the signal to CIELAB colour space coordinates. In the
med earth structures. Most of the samples have been attributed to
CIELAB colour space, L* represents the perceptual lightness, ranging
structures of uncertain dating. Anyway, structures of M07 and M08 are
from 0 (black) to 100 (white), while a* and b* correlate with chroma
from phases of the 12th and 16th centuries respectively, while M04 and
perceptions on green–red and blue-yellow axes, respectively (Fairchild,
M06 from Modern times interventions. Except for M05, which looks
2013). Powdered samples were put in a small cylindric container and
earthy, the other samples are lime mortars. M11 was divided in two
analysed with a 3 mm measurement spot, while intact samples were
samples due to the different colour of the binder, which is whitish for
analysed with an 8 mm spot in five different random areas in which both
M11W and brownish for M11B. Binder fraction is whitish to light grey
aggregate and binder fractions were present, and median values were
for the lime mortars. All the lime samples present a sandy aggregate
used. Concerning imaging (see Annex 2 for a flowchart resuming the
fraction with mainly greyish clasts except for M06 which shows mainly
procedure), the smartphone photos were made in a controlled envi­
whitish clasts.
ronment with a Samsung Galaxy S7 Edge (powdered samples camera
setting: ISO: 100, f: 1/350 s, T: 5500 K, focus: manual, zoom: 3.0 x;
2.2. X-ray diffraction (XRD) intact samples camera setting: ISO: 80, f: 1/500 s, T: 3700 K, focus:
manual, zoom: 3.0 x). Image were processed with MATLAB (version:
All the analyses were carried out on the whole sample, without the R2019b) by MathWorks and Colorlab toolbox (Malo and Luque, 2002).
separation of aggregate and binder fractions. As a first step, each sample Firstly, the photos were cropped in order to process data from mortar
was powdered and homogenised by agate mortar and pestle prior to the surface only, RGB parameters for each pixel were converted to XYZ
analysis. The analyses were carried out through using a Bruker D2 tristimulus values and, finally, to the parameters of the CIELAB colour
Phaser X-ray diffraction instrument, Cu Kα radiation λ = 1.5418 Å, range space employing the photo of a reference white taken in the same con­
5-65◦ 2θ with 0.02◦ 2θ per step and accumulation time of 10 s. The ditions of the other ones. Median L*, a* and b* were employed.

3
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

2.6. Exploratory data analysis can be observed from major elements composition, since the first ones
show higher levels of Al, Si, K, Ti and Fe and lower ones of Ca than the
Statistics and data visualisation were carried out in R (version 4.0.2; lime mortars. These major elements profiles are consistent with miner­
R Core Team, 2020) employing the following R packages: factoextra alogical features evidenced by mineralogical profiles obtained by XRD
(version: 1.0.7; Kassambara and Mundt, 2020), ggplot2 (version: 3.3.3; which pointed out higher amounts of quartz and aluminosilicates in
Wickham, 2016), signal (version: 0.7–6; signal developers, 2013), earth mortars than in lime ones, and lower amounts of carbonates. Earth
ggrepel (version: 0.9.1; Slowikovski, 2021), ggpubr (version: 0.4.0; mortar samples are also richer in REE and most trace elements (Bi, Cd,
Kassambara, 2020). Principal component analysis (PCA) was used Co, Cr, Li, Mn, Mo, Ni, Th, Tl, and V) and lower in Sr than lime mortars.
separately in Silla Tower samples and in Fuengirola Castle ones to The chemometric exploration of the multielement analysis data
reduce dimensionality and analyse the main variance tendencies of the conducted by PCA allowed to highlight similarities and differences
datasets, evaluating the presence of groups. Concerning multielement among the lime mortars. The first model (Annex 4) shows the results
analysis data, for both the sites, two models were built: the first with all obtained on all the considered elements. Earth mortars were excluded
the elemental concentrations, the second only with REE as variables. from this analysis since their peculiar features could have hidden the
The concentrations were standardised prior to the analysis. Principal differences among lime ones in the model. Annex 4a shows the scores of
component analysis was carried out also for ATR-IR data in the region the two first PC for the samples, while Annex 4b-c shows the variables
between 2000 and 700 cm− 1. Each spectrum was pre-processed by loadings. The first two components explain 53.3% and 12.9% of the total
Savitzky-Golay filter (polynomial order: 2nd, length: 13, derivative: variance respectively. The first PC is positively correlated with most of
2nd) and signals were mean-centred prior to the PCA. Cluster analysis the elements and negatively with Ca and some trace elements (Pb, Mo,
(CA method: average linkage) employing the first two PC as variables Sr and U), while PC2 shows the most intense positive correlations with
was also carried out for Fuengirola Castle samples. Si, Bi, Cd, Cu, Pb, Mn, V and Zn, and negative ones with Ca, Ba, Cr, Co, Li
and Sr. As can be observed in the Annex 4a, mortars from different
3. Results and discussion chronologies are scattered in the samples/scores plot and clusters
related to the different construction phases cannot be clearly observed.
3.1. The mortars from Silla Islamic Tower Most Islamic samples have lower PC1 scores than most mortars from the
following phases. However, P2M3 falls among them. The unclassified
3.1.1. X-ray diffraction results sample of SB fall isolated in the first quadrant. N3B is in the third
The mineralogical analysis (Table 2) conducted on the samples quadrant, close to the two samples of the 20th century (O8-9), the un­
allowed to highlight the qualitative mineralogy of the mortars under classified sample of plaster and the 14th century one (O10).
examination. The second model was made employing REE as variables (Fig. 2),
In qualitative mineralogy, what immediately stands out is the sub­ since previous archaeometric studies proved their effectiveness as
stantial difference of the mortars of the Roman period (P0M3 and P0M4) chronological markers in mortars (Gallello et al., 2017; Ramacciotti
compared to the other subsequent periods. These mortars have a rela­ et al. 2018). The first two PC explain more than 95% of the total vari­
tively high quartz content compared to calcite. High amounts of quartz ance. The first PC (Fig. 2b) is positively correlated with all the variables,
are not unusual in earth mortars aggregate fraction, as observed in while PC2 (Fig. 2c) shows a fractionation of REE since lighter REE (from
previous studies (Gómez Morgade et al., 2021). The samples P2M1 and La to Pr) and heavier REE (from Tm to Lu) have negative coefficients,
P2M2, from Islamic period presents traces of gypsum. Gypsum-lime while the others (from Nd to Er) have positive ones. Sc and Y have
mortars are known for the same period in other places in Spain and negative and positive coefficients respectively. Concerning the scores
Portugal (Genestar and Pons, 2003; Freire et al, 2008; Vitti, 2021). diagram (Fig. 2a), samples from the centuries 14th (O10) and 20th (O8-
However, minor amount of this mineral can probably be better 9) have lowest scores for PC1 or higher scores for PC2 than Islamic
explained by chemical weathering products (Lanas et al., 2005). The samples, which are scattered on the two PC-axes. The sample from the
P1EN sample has also traces of gypsum, but a lower quartz content than modern restoration (P2M3), which previously grouped with the Islamic
the previous ones. Samples SA and SB are qualitatively similar to each mortars, falls isolated due to high PC2 scores. The unclassified samples
other. Likewise, the O8, O9 and O10 samples are mineralogically group clearly with the mortars from the Islamic constructive phases,
similar, although O10 is archaeologically attributed to the 14th century. except for the plaster one, which shows PC1 scores closer to the 14th and
20th century mortars but negative PC2.
3.1.2. Multielement analysis
Major elements concentrations are reported in Table 3, while the 3.1.3. Archaeological evidence
trace elements and REE ones are reported in the Annex 3a and Annex 3b, Resuming the obtained results from an archaeological point of view,
respectively. it is interesting to observe that the first discrimination is between the
The difference between earth mortars (P0M3-4) and lime mortars walls from the Roman phase and the one the following periods, being the

Table 2
Results of XRD analyses of Silla Islamic Tower mortars.
Sample Cal Dol Qtz Kfs Pl Phyll. Others Sample Cal Dol Qtz Kfs Pl Phyll. Others

N3A XX tr XX X P0M3 X XXX tr tr tr


N3B XXX XX P0M4 X XXX tr tr tr
N4 XX XX tr P1EN XXX X Gp (tr)
S6 XXX XX P1QU XXX XX tr
SA XXX tr XX tr tr tr P1M1 XXX XX tr
SB XXX tr XX tr tr tr P1M2 XXX XX tr
O8 XXX X tr tr tr P2M1 XXX XX tr Gp (tr)
O9 XXX X tr tr tr P2M2 XXX XX tr Gp (tr)
O10 XXX X tr tr tr P2M3 XXX XX tr
P0M1 XX XX tr tr P3M1 XXX XX tr
P0M2 XXX XX tr tr tr

Note: Semiquantitave estimates are reported from traces (tr), and to small amounts (X) to large amounts (XXX). Cal: calcite, Dol: dolomite, Qtz: quartz, Kfs: K-feldspar,
Pl: plagioclase, Phyll.: phyllosilicates, Gp: gypsum.

4
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

Table 3
Major elements concentrations for Silla Tower samples.
Sample Al Si K Ca Ti Fe Sample Al Si K Ca Ti Fe

N3A 0.90 9.85 0.09 24.96 0.06 0.6 P0M3 4.03 18.53 2.23 8.82 0.32 2.22
N3B 0.82 5.63 ND 32.33 0.05 0.53 P0M4 4.41 19.26 2.59 8.00 0.33 2.27
N4 1.03 9.78 0.27 25.24 0.06 0.59 P1M1 1.32 7.36 0.35 27.58 0.08 0.61
S6 0.78 8.05 0.16 30.18 0.06 0.57 P1M2 1.57 9.12 0.70 25.11 0.12 0.93
SA 0.98 9.76 0.41 25.79 0.07 0.65 P1EN 0.78 2.19 ND 34.76 0.02 0.23
SB 0.89 9.55 1.06 21.28 0.05 0.57 P1QU 1.05 5.88 0.14 29.39 0.06 0.5
O8 0.88 2.06 ND 37.26 0.02 0.34 P2M1 0.86 8.16 0.20 29.05 0.05 0.48
O9 0.38 3.06 ND 30.36 0.04 0.5 P2M2 0.80 8.48 0.49 25.54 0.07 0.63
O10 0.70 2.82 ND 32.40 0.02 0.38 P2M3 0.74 8.61 0.07 23.87 0.05 0.47
P0M1 1.16 9.52 0.60 27.44 0.10 0.89 P3M1 0.57 7.21 0.07 28.67 0.05 0.47
P0M2 1.29 10.04 0.62 26.18 0.12 0.92

Note: Concentrations are expressed as mass percentage. ND: not detected.

Fig. 2. Samples/scores plot (a) and variables/loadings plots of PC1 (b) and PC2 (c) for the lime mortars of Silla Islamic Tower employing REE as variables.

former made with earth mortars. The use of this technology was
Table 4
observed also in the structures from the Roman Republican phase of
Results of XRD analyses of Fuengirola Castle mortars.
Sagunto Castle (Gallello et al., 2017) and it is possibly a recurring
technique in the area during this period, although more studies are Sample Cal Dol Qtz Kfs Pl Phyll. Others

needed in order to confirm this hypothesis. The mortars from the 14th M01 X XXX tr tr
(O10) and the 20th (O8-9) centuries structures cannot be discriminated M02 X XXX tr tr
M03 X X XXX tr tr
through multivariate statistics, but their elemental concentrations and
M04 X X XX X X X Arg (X)
especially REE ones are quite different from those of samples from the M05 X XXX tr tr X
previous periods. Concerning the unclassified samples, the plaster M06 X XXX tr X tr
sample (P1EN) does not show clear relationship with the others, maybe M07 XXX X XX tr tr tr
because it is a completely different type of mortar, while N4, SA and SB M08 XX XXX tr tr
M09 X XXX XX
have REE levels similar to the Islamic mortars. The results for N4 seem in
M10 XX XXX tr tr
contradiction with the archaeological hypothesis, which suggested a M11B XX XXX tr tr X
more recent dating (Christian period), however, it could also be M11W XX XXX tr Arg (tr), Hl (tr)
explained by a continuity in raw material use and mortar manufacturing M12 XX X XX
recipe also in a part of the period following the Islamic occupation. Note: Semiquantitave estimates are reported from traces (tr), and to small
amounts (X) to large amounts (XXX). Cal: calcite, Dol: dolomite, Qtz: quartz, Ffs:
3.2. The mortars from Fuengirola Castle K-feldspar, Pl: plagioclase, Phyll.: phyllosilicates, Arg: aragonite, Hl: Halite.

3.2.1. X-ray diffraction results


From the point of view of the mineralogical composition (Table 4),
the presence of traces of halite in the M11W sample is highlighted,

5
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

probably due to the interaction with the marine environment. The (Fuentes et al., 2006; Olivares et al., 2009; Shillito et al., 2009). The
samples M01 and M02 seem to be quite similar. The presence of wide band at ~ 3400 cm− 1 and the one of ~ 1635 cm− 1 in some samples
aragonite in the samples M04 and M11W could be due to the presence of suggest the presence of water (Silva et al., 2005), maybe linked to
crushed shells in the composition of the mortar. Sample M09 is char­ phyllosilicates or other hydrated phases. Some samples seem to be
acterized by the marked presence of phyllosilicates. earthy (M02, M11B) and M05 is an earth mortar.
The PCA was carried out to determine the main characteristics
3.2.2. Infrared spectroscopy concerning the variance of the dataset and to evaluate the presence of
The ATR-IR spectra are shown in Fig. 3. As can be observed, the main groups. Cluster analysis was also employed to validate the interpreta­
bands of the spectra correspond to the vibrational modes of carbonates tion. The results of the data analysis can be observed in Fig. 4. The first
between 1420 and 1405 cm− 1, between 725 and 715 cm− 1, and at ~ two PC explain 75.2% and 13.7% of the overall variance respectively.
875 cm− 1 (Clark et al., 1990; Silva et al., 2005; Bruckman and Wriessnig, The variables that have the highest influence for PC1 are signals close to
2013). Less intense band probably linked to carbonates can be observed 1400, 875 and 715 cm− 1 (Fig. 4b). These wavenumbers correspond to
in all the samples except for the one of earth mortar (M05) at ~ 1800 bands of carbonates and their intensities in the original spectra are
cm− 1 (Silva et al., 2005). Another relevant feature can be seen at ~ 990 positively correlated to the first PC. It can be observed in the scores plot
cm− 1 in all the mortar samples, except for M06, and it is probably caused (Fig. 4a), corroborated by the dendrogram, that M07 (12th c.) and M12
by the presence of aluminosilicates (Shillito et al., 2009). Bands at ~ can be distinguished due to the higher scores of PC1. It is worth noting
1160 and ~ 1085 cm− 1, as well as those of about 795 and 780 cm− 1 that although the earth mortar M05 does not differ significantly from the
could mark the presence of aluminosilicates as well or silicon dioxide other samples according to the dendrogram, it is characterised by the

Fig. 3. Spectra of ATR-IR of the Fuengirola Castle mortars (spectra are offset to avoid excessive overlapping).

6
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

Fig. 4. Samples/scores plot for ATR-IR analysis with dendrogram (a) and variables/loadings plots for PC1 (b) and PC2 (c).

lowest PC1 scores.


Table 5
Carbonate bands have a relevant influence also on PC2 (Fig. 4c).
Major elements concentrations for Fuengirola Castle samples.
Indeed, the loading peaks close to 1420, 1395, 885 and 730 cm− 1 are
probably ascribable to this class of minerals. In particular, the intensity Sample Al Si K Ca Ti Fe

of the bands at about 1420, 885 and 730 cm− 1 in the original spectra are M01 0.98 18.63 0.26 15.08 0.11 1.93
linked to higher scores on the PC2, while the intensity at about1395 M02 1.19 11.92 0.60 15.47 0.10 1.64
M03 ND 13.70 0.16 15.06 0.06 1.31
cm− 1 is linked to lower values of the same PC. The feature at ~ 1340
M04 1.41 12.95 0.60 14.58 0.18 3.19
cm− 1, observable as a shoulder of ~ 1400 cm− 1 band in M11W and M05 3.35 16.46 2.07 5.90 0.35 4.23
M11B, and the one at ~ 850 cm− 1, shoulder of the ~ 875 cm− 1 band, are M06 ND 0.45 ND 24.27 0.01 0.20
possibly caused by the presence of organic matter (Legan et al., 2020) M07 ND 4.72 ND 22.11 0.03 0.67
and aragonite (Pronti et al., 2020), respectively. On this direction, a M08 0.42 14.09 0.10 12.83 0.08 1.48
M09 2.81 15.01 1.25 9.85 0.27 3.35
band related to the aluminosilicates has a certain influence, as indicated
M10 0.41 7.85 0.05 19.26 0.06 1.18
by loadings around 1000 cm− 1. In this case, higher intensity corresponds M11B 1.86 10.37 1.63 9.38 0.23 2.97
to lower score. Sample M06 (18th c.) is characterised by the highest PC2 M11W 0.60 5.44 1.14 14.65 0.12 2.07
score. Its peculiarity, compared to the other samples is evidenced also by M12 0.28 8.02 ND 20.49 0.04 0.75
the dendrogram. Looking at the spectrum (Fig. 3), it can be observed Note: Concentrations are expressed as mass percentage. ND = not detected.
that its carbonate band close to 1400 cm− 1 peaks at ~ 1420 cm− 1 and,
furthermore, it has an intense band at ~ 730 cm− 1, which hides the 3.2.3. Multielement analysis
calcite one at ~ 715 cm− 1. These two features suggest a relevant amount Results for pED-XRF analysis can be observed in the Table 5, while
of dolomite (Bruckman and Wriessnig, 2013). This characteristic was results from ICP-MS in the Annex 5a (trace elements) and Annex 5b
pointed out also by XRD analysis which revealed large amount of (REE).
dolomite in M06, while the other samples were characterised by a small From the point of view of major elements, some peculiarities can be
to moderate presence of this carbonate mineral in the other sample. observed: Al and K could not be detected by the spectrometer in the
Sample score is also coherent with the absence of aluminosilicate band sample M06, which show the lowest levels of Si, Ti and Fe, while it has
at ~ 990 cm− 1. These facts indicate that for the production of this the highest level of Ca. As previously stated, this sample is characterised
sample (M06) the employed raw materials were considerably different by a different aggregate also from the macroscopic point of view and has
from those of the other lime mortars. high amounts of carbonate minerals and low ones of quartz and alu­
minosilicates compared to the other mortars, as indicated by ATR-IR and

7
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

XRD analyses. Therefore, it could be the main cause of this difference. of REE fractionation. Indeed, loadings seem to be positively correlated to
Among the other samples, M05, M09 and M11B have the lowest levels of elemental mass, and Eu has opposite coefficient compared to its neigh­
Ca and the highest of Al. In the case of M05, the high amount of Al is bouring elements (Sm and Gd). Scores diagram (Fig. 5a), corroborated
combined with the highest level of Si, which can be explained by the by PC1-2 dendrogram, evidences the distance between earth mortar
presence of earth. (M05), rich in REE, and lime mortars. M09, M11B and M04 are also in
In order to explore the dataset reducing the variable number, PCA this case midway between the two groups on PC1, although closer to
was employed. A first PCA (Annex 6) was carried out with all the ana­ lime mortars. M06 (18th c.) has very low levels of REE and the highest
lysed elements as variables and CA with PC1 and PC2 was also carried PC1 score. The other samples group in two subclusters (1st: M01, M12,
out to guide scores plot interpretation. The first two PC explain 80.6% of M03, M07 (12th c.); 2nd: M02, M08 (16th c.), M10, M11W).
the overall variance. The first PC loadings (Annex 6b) is influenced in
the positive direction by elements probably linked to aluminosilicates 3.2.4. Preliminary considerations
(Al, Si, K, Ti, Fe, REE and most trace elements), while Ca and few trace Due to the reduced sample size, the obtained results permitted only
elements (Mo, Sr and U) are correlated in the negative direction with preliminary considerations. The analytical results and data processing
this PC. As regards PC2 (Annex 6c), loadings show intense negative evidenced the presence of three types of mortars, according to their
correlations with Si, Fe, Cd, Cr, Co, Ni and Sr, and positive ones with K, chemical features: sample M05, an earth mortar, sample M06 (18th c.),
Pb, Mo, Tl, V y U. Scores diagram (Annex 6a) and CA dendrogram show characterised by a different aggregate and a relevant amount of dolo­
that the most important separation is the one between earth (M05) and mite, and M04 (19th c.), which showed characteristic elemental levels
lime mortars. The lime mortars can be divided in other sub-groups. M09 evidenced by the PCA employing all the elements (Annex 6). Features of
and M11B, rich in aluminosilicates elements, cluster together and are M09 and M11B are probably related to the presence of earth, although
half-way between earth and lime mortars. M04 (19th c.) is closer to the cause is not clear (contamination or voluntary addition?). The dif­
them but shows completely different PC2 scores. M07 (12th c.) and M12, ference among the other samples is less evident but data analysis sug­
and further, M06 (18th c.) and M03 pertain to another sub-cluster; while gests a possible similarity of M07 (12th c.) and M12.
in the other there are M01, M08 (16th c.), M02, and M10 and M11W. It
is worth noticing that ATR-IR PCA evidenced a compatibility between
3.3. Colour characterisation
M07 and M12 as well, due to their high content of carbonates. A second
PCA with CA was made employing only REE as variables (Fig. 5).
As previously stated, colour characterisation was carried out
The first two PC explain more than 99% of the overall variance. The
employing imaging of smartphone pictures and Vis-SP analysis as
first PC is linked to the total amount of REE and all the variables show
reference method (results are shown in the Annex 7). Powdered samples
negative coefficients (Fig. 5b). The second PC (Fig. 5c) shows two kinds
were analysed from both the architectural complexes, while only in

Fig. 5. Samples/scores plot and CA dendrogram (a), and loadings for PC1 (b) and PC2 (c) for the PCA employing REE as variables.

8
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

Fuengirola Castle samples the two techniques were employed on intact


fragments.
Concerning powdered samples, it can be observed that imaging re­
sults are a good predictor of colour properties analysed by the reference
method (Annex 8), although L* values are higher for imaging than for
Vis-SP, while chromatic parameters (a* and b*) are instead lower.
Indeed, the coefficient of determination (R2) ranges between 0.88 (a*)
and 0.96 (L*). Pearson correlation coefficient (r) were calculated to
investigate the possible relationship between imaging detected colour of
powdered samples and elemental properties in lime mortars (Table 6).
Earth mortars (P0M3-4 from Silla and M05 from Fuengirola) were
excluded to avoid misleading results. In general, it is possible to observe
that L* is negatively correlated with Al, Si, K, Ti and Fe, and positively
with Ca. On the other hand, a* and b* are positively correlated with Al,
Si, K, Ti and Fe and negatively with Ca. It suggests that mortars L* is
mainly driven by the amount of carbonate minerals, while coordinates
on chromatic axes (a* and b*) by the other mineralogical phases such as
aluminosilicates.
Fig. 6 shows the scatter plots for L* vs a* (top) and L* vs b* (bottom)
imaging data for Silla Islamic Tower powdered samples. Roman earth
mortar samples (P0M3-4) were excluded to improve image clarity,
anyway, they are visibly different from lime mortars due to higher a*
and b* values, and for lower L*, due to the very different composition Fig. 6. Biplots for L* vs a* (top) and L vs b* (bottom) employing imaging data
of Silla Islamic Tower powdered samples.
(Annex 7a). From the point of view of L* axis, Islamic samples have
lower L* than the others from the following construction phases,
although the distributions overlap. However, they can be differentiated
from the one of the 18th century on the a* axis and from those of 14th
and 20th centuries on b* axis. It is also worth noticing that samples from
the ground and the first floors (P0M1-2, P1M1-2) show slightly higher a*
values than the others. At this point, the analytical results do not give
any hint on the possible causes of this difference. As regards the un­
classified mortars, most of them are plotted within the Islamic samples
suggesting this possible classification, which is consistent with multiel­
ement analysis data. P1EN1 shows instead colour properties different
from the other samples which is consistent with elemental and miner­
alogical data pointing out higher amounts of carbonates.
Concerning Fuengirola Castle powdered samples (Fig. 7), the earth
mortar (M05) and the brownish stratum of M11 (M11B) are plotted
isolated in the diagram due to the low L* and the high chromaticity
compared to the other samples. M06, M07 and M12 have the highest L*.
These samples are characterised by low levels of Al and Si, and high ones
of Ca, while XRD and ATR-IR pointed out high amounts of carbonate
minerals. It is worth noticing that mineralogical analyses pointed out a
very high amount of dolomite for M06, which have higher L*/a* ratio
than the other samples.
A small test was also carried out employing the photos from the
intact samples of Fuengirola Castle. However, we can observe that Vis-
SP and imaging data show strong linear relationship only for L* (R2 =
0.87; Annex 9a). CIELAB diagrams (Annex 9b) evidence separation be­ Fig. 7. Biplots for L* vs a* (top) and L* vs b* (bottom) employing imaging data
tween M05 and M11B and the other mortars, although, on the contrary of Fuengirola Castle powdered samples.

of powdered samples, in this case M11W is plotted closer to these and


Table 6 not to the other lime mortars. Intact M07 has the highest L* and the
Pearson correlations (r) between colour parameters (imaging) and major ele­ lowest b*, as for powdered sample, while M06 is plotted within the other
ments in lime mortars from the two architectural complexes. lime mortars. The issues in obtaining colour values consistent with those
Mortars Al Si K Ca Ti Fe of the reference technique in intact mortars is probably related to the
Silla (n = 19) L* − 0.54 − 0.69 − 0.74 0.57 − 0.73 − 0.81
difficulty in characterising samples with irregular and heterogeneous
a* 0.78 0.53 0.71 − 0.40 0.86 0.89 surface. Further studies are needed with a large sample size to improve
b* 0.68 0.75 0.68 − 0.58 0.92 0.91 the smartphone method as a viable non-invasive method to characterise
Fuengirola (n = L* − 0.79 − 0.32 − 0.93 0.80 − 0.87 − 0.82 ancient mortars colour features.
12) a* 0.58 0.08 0.81 − 0.59 0.67 0.58
b* 0.68 0.10 0.90 − 0.65 0.76 0.70
All (n = 31) L* − 0.68 − 0.47 − 0.87 0.65 − 0.87 − 0.81 4. Conclusion
a* 0.61 0.21 0.77 − 0.32 0.67 0.47
b* 0.64 0.20 0.69 − 0.10 0.57 0.27 Different analytical techniques were employed to characterise the
Note: ND for Al and K were changed with half of the lowest detected concen­ samples from Silla Islamic Tower and Fuengirola Castle.
tration. Barred coefficients are statistically not significant (p >.05). The results of multielement analysis suggested that most of the

9
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

unclassified samples from Silla Tower date back to the Islamic con­ the project “Smartphone and Green Analytical Chemistry” (PROMETEO
struction phase, while some characteristic mortars could be observed 2019-056) funding and the related predoctoral scholarship.
among Fuengirola Castle from the phases of 12th century and from the
restorations of the past two centuries. However, to confirm the con­ Appendix A. Supplementary data
struction period of undated structures, a wider number of samples
should be studied and additional standardised techniques should be Supplementary data to this article can be found online at https://doi.
used. org/10.1016/j.jasrep.2022.103433.
As regards the use of imaging on smartphone photos as an innovative
archaeometric screening method for ancient mortars, it is possible to References
observe that data obtained from powdered samples showed strong linear
relationship with those of the employed reference technique (Vis-SP). In Alapont, L.l., Pitarch, A., Orengo, H.A., 2016. Arqueologia de l’Horta Sud: L’origen d’una
comarca histórica. Perifèric Edicions, Catarroja.
lime mortars, L* has positive correlation with Ca, and negative with the Aguilar Cuesta, Á.I., Vallina Rodríguez, A., García Juan, L., 2019. Fuengirola a través del
other major elements, while both a* and b* have correlations opposite to Catastro de Ensenada. In: Martín Ruiz, J.A. (Ed.), De Suel a Fuengirola: Actas I
L*. This suggests that lightness increases together with the amount of Jornadas de Arqueología y Patrimonio. La Serranía, Fuengirola, pp. 151–187.
Azkarate, A., 2020. Archaeology of architecture: buildings archaeology. In: Orser, C.E.,
carbonate minerals, and a* and b* together with other minerals. This Zarankin, A., Funari, P.P.A., Lawrence, S., Symonds, J. (Eds.), The Routledge
fact is also corroborated by XRD and ATR-IR analyses which pointed out Handbook of Global Historical Archaeology. Routledge, New York, pp. 517–536.
higher levels of carbonates in the samples characterised by the higher Brogiolo, G.P., 2007. Dall’Archeologia dell’architettura all’Archeologia della
complessità. Pyrenae 38, 7–38.
L*. The variance of these parameters in the samples could be linked to Bruckman, V.J., Wriessnig, K., 2013. Improved soil carbonate determination by FT-IR
differences in mortar recipes or raw materials, as well as to different and X-ray analysis. Environ. Chem. Lett. 11, 65–70. https://doi.org/10.1007/
degradation levels. Furthermore, it is worth noticing that the classifi­ s10311-012-0380-4.
Cantisani, E., Calandra, S., Barone, S., Caciagli, S., Fedi, M., Garzonio, C.A., Liccioli, L.,
cation of Silla undated mortars obtained by colour parameters is
Salvadori, B., Salvatici, T., Vettori, S., 2021. The mortars of Giotto’s Bell Tower
consistent with that of multielement analysis. On the other hand, about (Florence, Italy): raw materials and technologies. Constr. Build. Mater. 267, 120801
the Fuengirola Castle sample set, the most relevant difference is among https://doi.org/10.1016/j.conbuildmat.2020.120801.
earthy samples and lime mortars, although also dolomite-rich sample Centauro, I., Cantisani, E., Grandin, C., Salvini, A., Vettori, S., 2017. The influence of
natural organic materials on the properties of traditional lime-based mortars. Intern.
(M06) seems to have slightly different colour features which were not J. Archit. Herit. 11, 670–684. https://doi.org/10.1080/15583058.2017.1287978.
evident by naked-eye examination of powdered samples. Chiarelli, N., Miriello, D., Bianchi, G., Fichera, G., Giamello, M., Turbanti Memmi, I.,
This preliminary smartphone test shows the potential that colour 2015. Characterisation of ancient mortars from the S. Niccolò archaeological
complex in Montieri (Tuscany – Italy). Constr. Build. Mater. 96, 442–460. https://
characterisation by imaging have, being a cheap and fast method to doi.org/10.1016/j.conbuildmat.2015.08.023.
carry out chronological classification of ancient mortars, at least for Clark, R.N., King, T.V., Klejwa, M., Swayze, G.A., Vergo, N., 1990. High spectral
powdered samples, which could be a first screening step in order to carry resolution reflectance spectroscopy of minerals. J. Geophys. Res.: Solid. Earth 95,
12653–12680. https://doi.org/10.1029/JB095iB08p12653.
out more invasive approaches. However, more studies should be carried Columbu, S., Antonelli, F., Sitzia, F., 2018. Origin of Roman worked stones from St.
out in order to optimise the methodological approach and corroborate Saturno Christian Basilica (South Sardinia, Italy). Mediterr. Archaeol. Ar. 18, 17–36.
the results of this first study. Concerning intact samples, the test on the https://doi.org/10.5281/zenodo.1256047.
Crisci, G.M., Franzini, M., Lezzerini, M., Mannoni, T., Riccardi, M.P., 2004. Ancient
mortars of Fuengirola Castle gave more challenging methodological is­ mortars and their binder. Period. Miner. 73, 259–268.
sues. However, L* shows good agreement with Vis-SP values, suggesting Cruz-Fernández, M., Luque-Cobija, M.J., Cervera, M.L., Morales-Rubio, A., de la
that further studies are needed in order to improve non-invasive image Guardia, M., 2017. Smartphone determination of fat in cured meat products.
Microchem. J. 132, 8–14. https://doi.org/10.1016/j.microc.2016.12.020.
analysis of mortars.
De Luca, R., Barca, D., Bloise, A., Cappa, M., De Angelis, D., Fedele, G., Mariottini, S.,
To go deeper and understand if issues such as mortar recipes, raw Micieli, D., Muto, F., Piluso, E., Pucino, F., Taranto, M., Tropea, M., Miriello, D.,
materials (binder/aggregate ratios, higher amounts of aluminosilicates 2021. RecoStones: a New Tool to Identify Calabrian Stone Materials Through Image
and oxides in the aggregate fraction, limestone impurities or pozzolanic Processing. Geoheritage 13, 41. https://doi.org/10.1007/s12371-021-00564-w.
Dessales, H., 2017. The archaeology of construction: a new approach to Roman
materials addition), or contamination and weathering are highlighting architecture. Ann. Hist., Sci. Soc. – English Edition 72 (1), 69–86. https://doi.org/
smartphone colour differences, further studies, employing standard 10.1017/ahsse.2019.6.
methods such as petrographic analysis, should be carried out investi­ Donais, M.K., Duncan, B., George, D., Bizzarri, C., 2009. Comparisons of ancient mortars
and hydraulic cements through in situ analyses by portable X-ray fluorescence
gating the relationship between the colour features and the properties of spectrometry. X-Ray Spectrom. 39, 146–153. https://doi.org/10.1002/xrs.123.
mortar binder and aggregate fractions. Fairchild, M.D., 2005. Color appearance models, second ed. John Wiley & Sons Ltd.,
This archaeometric study further improves the archaeological Chirchester.
Ferris, I.M., 1989. The Archaeological Investigation of Standing Buildings. Vernac.
knowledge on the studied monuments, highlighting the diachronic Archit. 20, 12–17. https://doi.org/10.1179/vea.1989.20.1.12.
features of the construction phases related to different mortar Francovich, R., Bianchi, G., 2002. L’archeologia dell’elevato come archeologia. Arqueol.
manufacturing aspects, pointing out the possibilities of a multi- Arquit. 1, 101–111.
Freire, T., Santos Silva, A., Veiga, M.R., Brito, J.D. (2008). Characterization of
analytical approach based on portable and innovative methodologies Portuguese historical gypsum mortars, in: Laboratório Nacional de Engenharia Civil
together with widely used analytical techniques. (Portugal) (Ed.), Characterization, diagnosis, conservation, repair, compatibility:
HMC08, 1st Historical Mortars Conference. LNEC, Lisbon, pp. 1–16.
Fuentes, A.F., Rodríguez-Reyna, E., Martínez-González, L.G., Maczka, M., Hanuza, J.,
Declaration of Competing Interest Amador, U., 2006. Room-temperature synthesis of apatite-type lanthanum silicates
by mechanically milling constituent oxides. Solid State Ion. 177, 1869–1873.
The authors declare that they have no known competing financial https://doi.org/10.1016/j.ssi.2006.02.032.
Gallello, G., Ramacciotti, M., Lezzerini, M., Hernandez, E., Calvo, M., Morales, A.,
interests or personal relationships that could have appeared to influence
Pastor, A., de la Guardia, M., 2017. Indirect chronology method employing rare
the work reported in this paper. earth elements to identify Sagunto Castle mortar construction periods. Microchem. J.
132, 251–261. https://doi.org/10.1016/j.microc.2017.02.009.
Acknowledgements Genestar, C., Pons, C., 2003. Ancient covering plaster mortars from several convents and
Islamic and Gothic palaces in Palma de Mallorca (Spain). Analytical
characterisation. J. Cult. Herit. 4 (4), 291–298. https://doi.org/10.1016/j.
Gianni Gallello acknowledges the financial support of the Beatriz culher.2003.02.001.
Galindo Fellowship (2018) funded by the Spanish Ministry of Science Gómez Morgade, T., Rivas Brea, T., Carrera-Ramírez, F., Barbi Alonso, V., 2021. Earth
mortars in the “Castro de Santa Trega” (A Guarda, Pontevedra, Spain). J. Arch. Sci.
and Innovation and Ministry of Universities (Project BEAGAL18/00110 Rep. 37, 102931 https://doi.org/10.1016/j.jasrep.2021.102931.
“Development of analytical methods applied to archaeology”). M. Luisa Herreros-Chavez, L., 2021. Estrategias verdes para el análisis de muestras alimentarias y
Cervera, Agustín Pastor and Mirco Ramacciotti acknowledge the Min­ medioambientales. University of Valencia. PhD thesis.
istry of Education, Culture and Sport of the Valencian Government for

10
M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433

Kassambara, A., 2020. ggpubr: ’ggplot2’ Based Publication Ready Plots. R Package Olivares, M., Tarriño, A., Murelaga, X., Baceta, J.I., Castro, K., Etxebarria, N., 2009. Non-
Version 0.4.0. https://CRAN.R-project.org/package=ggpubr. destructive spectrometry methods to study the distribution of archaeological and
Kassambara, A., Mundt, F., 2020. factoextra: Extract and Visualize the Results of geological chert samples. Spectrochim. Acta Part A: Mol. Biomol. Spectrosc. 73,
Multivariate Data Analyses. R package version 1.0.7. https://CRAN.R-project.org 492–497. https://doi.org/10.1016/j.saa.2008.12.036.
/package=factoextra. Pronti, L., Romani, M., Viviani, G., Stani, C., Gioia, P., Cestelli-Guidi, M., 2020. Advanced
Klowikowski, K., 2021. ggrepel: Automatically Position Non-Overlapping Text Labels methods for the analysis of Roman wall paintings: elemental and molecular
with ’ggplot2’. R package version 0.9.1. https://CRAN.R-project.org/package=ggre detection by means of synchrotron FT-IR and SEM micro-imaging spectroscopy.
pel. Rend. Lincei. Sci. Fis. Nat. 31, 485–493. https://doi.org/10.1007/s12210-020-
Lanas, J., Sirera, R., Alvarez, J.I., 2005. Compositional changes in lime-based mortars 00888-9.
exposed to different environments. Thermochim. Acta 429(2), 219–226. DOI: R Core Team, 2020. R: A language and environment for statistical computing. R
10.1016/j.tca.2005.03.015Legan, L., Leskovar, T., Črešnar, M., Cavalli, F., Innocenti, Foundation for Statistical Computing, Vienna, Austria https://www.R-project.org/.
D., Ropret, P., 2020. Non-invasive reflection FTIR characterization of archaeological Ramacciotti, M., Rubio, S., Gallello, G., Lezzerini, M., Columbu, S., Hernandez, E.,
burnt bones: reference database and case studies. J. Cult. Herit. 41, 13–26. https:// Morales-Rubio, A., Pastor, A., de la Guardia, M., 2018. Chronological Classification
doi.org/10.1016/j.culher.2019.07.006. of Ancient Mortars Employing Spectroscopy and Spectrometry Techniques: Sagunto
Lezzerini, M., Legnaioli, S., Lorenzetti, G., Palleschi, V., Tamponi, M., 2014. (Valencia, Spain). J. Spectrosc. 2018, 9736547. https://doi.org/10.1155/2018/
Characterization of historical mortars from the bell tower of St. Nicholas church 9736547.
(Pisa, Italy). Constr. Build. Mater. 69, 203–212. https://doi.org/10.1016/j. Rezazadeh, M., Seidi, S., Lid, M., Pedersen-Bjegaard, S., Yamini, Y., 2019. The modern
conbuildmat.2014.07.051. role of smartphones in analytical chemistry. TrAC-Trend. Anal. Chem. 118, 548–555.
Lezzerini, M., Ramacciotti, M., Cantini, F., Fatighenti, B., Antonelli, F., Pecchioni, E., https://doi.org/10.1016/j.trac.2019.06.019.
Fratini, F., Cantisani, E., Giamello, M., 2016. Archaeometric study of natural Shillito, L.M., Almond, M.J., Wicks, K., Marshall, L.J.R., Matthews, W., 2009. The use of
hydraulic mortars: the case of the Late Roman Villa dell’Oratorio (Florence, Italy). FT-IR as a screening technique for organic residue analysis of archaeological
Archaeol. Anthr. Sci. 9, 603–615. https://doi.org/10.1007/s12520-016-0404-2. samples. Spectrochim. Acta Part A: Mol. Biomol. Spectrosc. 72, 120–125. https://
Lezzerini, M., Raneri, S., Pagnotta, S., Columbu, S., Gallello, G., 2018. Archaeometric doi.org/10.1016/j.saa.2008.08.016.
study of mortars from the Pisa’s Cathedral Square (Italy). Meas. 126, 322–331. Signal developers (2013). signal: Signal processing. http://r-forge.r-project.org/projects/
https://doi.org/10.1016/j.measurement.2018.05.057. signal/.
Malo, J., Luque, M.J. (2002). COLORLAB: a color processing toolbox for Matlab. https Silva, D.A., Wenk, H.R., Monteiro, P.J.M., 2005. Comparative investigation of mortars
://www.uv.es/vista/vistavalencia/software/colorlab.html. from Roman Colosseum and cistern. Thermochim. Acta 438, 35–40. https://doi.org/
Miriello, D., Barca, D., Bloise, A., Ciarallo, A., Crisci, G.M., De Rose, T., Gattuso, C., 10.1016/j.tca.2005.03.003.
Gazineo, F., La Russa, M.F., 2010a. Characterisation of archaeological mortars from Sitzia, F., Beltrame, M., Columbu, S., Lisci, C., Miguel, C., Mirão, J., 2020. Ancient
Pompeii (Campania, Italy) and identification of construction phases by restoration and production technologies of Roman mortars from monuments placed
compositional data analysis. J. Archaeol. Sci. 37 (9), 2207–2223. https://doi.org/ in hydrogeological risk areas: A case study. Archaeol. Anthrop. Sci. 12, 147. https://
10.1016/j.jas.2010.03.019. doi.org/10.1007/s12520-020-01080-8.
Miriello, D., Bloise, A., Crisci, G.M., Barrese, E., Apollaro, C., 2010b. Effects of milling: a Tenconi, M., Karatasios, I., Balaawi, F., Kilikoglou, V., 2018. J. Cult. Herit. 33, 100–116.
possible factor influencing the durability of historical mortars. Archaeom. 52, https://doi.org/10.1016/j.culher.2018.03.005.
668–679. https://doi.org/10.1111/j.1475-4754.2009.00494.x. Vitti, P., 2021. Mortars and masonry—structural lime and gypsum mortars in antiquity
Miriello, D., Barca, D., Pecci, A., De Luca, R., Crisci, G.M., López Luján, L., Barba, L., and Middle Ages. Archaeol. Anthr. Sci. 13, 1–19. https://doi.org/10.1007/s12520-
2015. Plasters from Different Buildings of the Sacred Precinct of Tenochtitlan 021-01408-y.
(Mexico City): Characterization And Provenance. Archaeom. 57, 100–127. https:// Wickham, H., 2016. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag, New
doi.org/10.1111/arcm.12074. York. https://doi.org/10.1007/978-3-319-24277-4.

11

You might also like