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 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 b A R T I C L E I N F O A B S T R A C T Keywords: Smartphone REE pED-XRF Imaging CIELAB Mortars Islamic period The present work shows the results of the chemical, mineralogical and colorimetric characterisation of the ancient mortars from Silla Islamic Tower (Valencia, Spain) and Fuengirola Castle (Malaga, Spain). The samples were characterised from the mineralogical point of view by X-ray diffractometry and mid-infrared attenuated total reflection spectroscopy, while portable energy dispersive X-ray fluorescence spectroscopy and inductively coupled plasma mass spectrometry were employed to obtain the concentrations of major and trace elements, including rare earth elements. Data analysis through multivariate statistics was used to evaluate features to 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 characterise mortars and carry out the study of construction phases in historical complexes. 1. Introduction 1 The analysis of architecture and ancient building materials is of particular interest for different fields of study being a fundamental step for the archaeological interpretation of architectural heritages construction phases, development of a complex, the employed construction techniques, the provenance of raw materials and their manufacture, as well as the implication of all these aspects for social history, and to program restoration interventions (Ferris, 1989; Francovich and Bianchi, 2002; Brogiolo, 2007; Dessales, 2017; Columbu et al., 2018; Azkarate, 2020). Several methods have been employed to characterise ancient mortars, including the use of different analytical techniques such as optical microscopy (OM), thermogravimetry with differential scanning calorimetry (TG-DSC), X-ray diffractometry (XRD), energy dispersive X-ray fluorescence (ED-XRF) and inductively coupled plasma mass spectrometry (ICP-MS) (Crisci et al., 2004; Lezzerini et al., 2014; 2018). The use of these archaeometric methods, often carried out during archaeological works or in the study of a monument, can provide important information concerning the employed raw materials and the production technology of mortars (Miriello et al., 2015; Lezzerini et al., 2016; Sitzia et al., 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/bync-nd/4.0/). M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433 wall stratigraphy, recognising similarities and diversities among group of samples driven by different manufacturing recipes (Corti et al., 2013; Chiarelli et al., 2015; Lezzerini et al., 2018). Mortars trace elements have been employed as markers of different construction phases (Miriello et al., 2010a). For example, in the last few years, Gallello and colleagues successfully used the rare earth elements (REE) data, processed though multivariate statistics, as elemental markers of the different construction phases for the structures of Sagunto Castle and its surroundings (Gallello et al., 2017; Ramacciotti et al., 2018). The present study shows the results of the archaeometric characterisation of the ancient mortars from the Islamic Tower of Silla (Valencia, Spain) and the Castle of Fuengirola (Malaga, Spain; Fig. 1) in order to understand the construction phases of the wall structures, implementing non-invasive screening approaches such as image analysis on smartphone photos. The Islamic Tower of Silla (Fig. 1b) was founded in the 11th century CE as a defensive building. This was part of the defensive belt of the city of Valencia and is built on the basis of Roman ashlars. Its function was to 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 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 Germanía of 1521, being abandoned during the 19th century (Alapont et al., 2016). About Fungirola Castle (Fig. 1c), such as the Silla Tower, it was built by the Muslims in the 11th century CE to control an area of the coast. It was occupied by Christians in 1485 and suffered several reforms in the 16th and 18th centuries, when its towers were demolished and several canvases were enlarged for the placement of artillery. In the 19th century one of the walls was rebuilt and at the end of that century it became a police station (Aguilar Cuesta et al., 2019). Multielement analysis employing a portable ED-XRF (pED-XRF) and an ICP-MS, and mineralogical characterisation through XRD were carried out for all the samples from the two historical complexes. The mortars from Fuengirola Castle were also characterised by attenuated total reflection infrared spectroscopy (ATR-IR). Finally, colour features of mortars from both the sites were measured by a portable Visspectrophotometer and through image processing of smartphone photos as a test to evaluate the performance and the possible employment of this last device in ancient mortar archaeometric studies for construction phases identification. Colour features were characterised in previous studies by colorimetric devices in order to investigate technological aspect of the production of ancient mortars (e.g.: Miriello et al., 2010b; Centauro et al., 2017). However, although smartphone as a nondestructive, fast and cheap analytical tool is more and more employed in various fields of analytical chemistry (e.g.: Cruz-Fernández et al., 2017; Rezazadeh et al., 2019; Herreros-Chavez, 2021), its use is still underdeveloped in archaeological science (De Luca et al., 2021). In this study, mortars were analysed and for the first time imaging data from smartphone photos were cross-referenced with other analytical data to evaluate the image analysis effectiveness in distinguishing among samples from different periods. The test was especially developed for the study of historical complexes characterised by several building interventions through the centuries. Indeed, the proposed developed approach could become useful as a prior screening step before running further invasive sampling strategies for standard methods. 2. Materials and methods 2.1. The mortars The samples collected from both the Silla Islamic Tower and Fuengirola Castle are shown in Table 1. Twenty-one samples come from Silla Islamic Tower. Nine samples were collected from the rammed earth external northern (N3A-B, N4), western (O8-10) and southern (S6, SA-B) walls of the tower, pertaining to different construction phases from the Islamic period to the past century. Four samples of bedding mortar were collected from an Islamic structure (P0M1-2), found in the ground floor (P0) of the Tower, and from the remains of a Roman Villa (P0M3-4). Eight samples were collected in the interior walls of the tower: four samples come from the first floor (P1), three from the second (P2) and one from the third (P3). P1M1-2, P2M1-2 and P3M1 come from the rammed earth structure and dated back to the Islamic phase of the tower, except for P2M3 which is possibly a modern restoration. P1QU comes from the ceiling lathwork 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 mineralogical qualitative identification was performed by DIFFRAC. EVA v4.1.1 software (by Bruker). Table 1 Samples from the two sites, localisation and dating. Sample Location* Dating Location* Dating Islamic Christian (?) 20th c. Second floor (int.) P2M2 Second floor (int.) P2M3 Second floor (int.) P3M1 Third floor (int.) Fuengirola Castle 20th c. M01 Section 3rd 14th c. M02 Tower 3 13th or 13th15th c. (?) 14th c. (?) Islamic M03 Section 2nd 14th c. (?) (?) M04 Section 4th (?) M05$ Tower 4 Islamic M06 Section 8th Beginning of 19th c. 10th or 12th c. (?) End of 18th c. Islamic M07 Tower 1 12th c. Roman M08 Tower 1 16th c. Roman M09 (?) M10 Section 10th Section 1st Ending 18th c. (?) 14th c. (?) Islamic M11B (?) Islamic M11W Islamic M12 Section 9th b Section 9th b Section 7th P2M1 Silla Islamic Tower N3A N3B N4 O8 O9 O10 S6 SA SB P0M1 P0M2 $ P0M3 P0M4$ P1EN P1QU P1M1 P1M2 Northern wall (ext.) Northern wall (ext.) Northern wall (ext.) Western wall (ext.) Western wall (ext.) Western wall (ext.) Southern wall (ext.) Southern wall (ext.) Southern wall (ext.) Ground floor (int.) Ground floor (int.) Ground floor (int.) Ground floor (int.) First floor (int.) First floor (int.) First floor (int.) First floor (int.) Sample Islamic Islamic Notes: * ext. = external, int. = interior; uncertain dating; c.: century. $ 2.3. Attenuated total reflection infrared spectroscopy (ATR-IR) Islamic The powdered samples of Fuengirola Castle were analysed with a portable spectrometer 4300 Handheld FT-IR by Agilent Technologies. Each spectrum is the average of 50 scans. Background was measured between an analysis and the others. Resolution of 4 cm−1 and Boxcar apodization were employed. Modern restoration Islamic 2.4. Multielement analysis Major elements (Al, Si, K, Ca, Ti and Fe) concentrations were measured in all the samples with a portable energy dispersive X-ray fluorescence spectrometer (pED-XRF) S1 Titan by Bruker, equipped with a Rh X-ray tube (50 kV) and X-Flash® SDD (resolution: 147 eV; FWHM: 5.9 keV). Geochem-trace application was employed for the analysis. Although its high limits of detection compared with the standard laboratory spectrometers, pED-XRF is increasingly employed in archaeometry, also in order to obtain quantitative and semi-quantitative elemental data of ancient mortars (Donais et al., 2009; Tenconi et al., 2018). Internal calibration was adjusted using geologic certified reference materials, and the accuracy and precision of the analytical data were controlled employing NIM-GBW07408 (Soil) and NCS DC 73375 (Limestone) reference materials (Annex 1). Trace elements (Ba, Bi, Cd, Cr, Co, Cu, Pb, Li, Mn, Mo, Ni, Sr, Th, Tl, U, V, Zn), including rare earth elements (REE: La, Ce, Pr Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, and Sc and Y), concentrations were analysed by an inductively coupled plasma mass spectrometer (ICP-MS) Elan DRCII by Perkin Elmer. The powdered samples were previously brought in solution by acid attack (aqua regia). Details on sample preparation and equipment setting can be found in Gallello et al., 2017. Accuracy and precision of the analytical data were controlled employing NIMGBW07408 and NCS DC 73375 reference materials. 14th c. (?) 14th c. (?) indicates earth mortars; (?) indicates 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 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 others as lime mortars, characterised by sandy aggregate, whitish-light gray binder and presence of lumps, except for P1EN in which aggregate was not visible by naked-eye. Thirteen samples of mortars were collected from different sectors of the Fuengirola Castle. Samples M01-03, M05, M09-12 come from rammed earth structures. Most of the samples have been attributed to structures of uncertain dating. Anyway, structures of M07 and M08 are from phases of the 12th and 16th centuries respectively, while M04 and M06 from Modern times interventions. Except for M05, which looks earthy, the other samples are lime mortars. M11 was divided in two samples due to the different colour of the binder, which is whitish for M11W and brownish for M11B. Binder fraction is whitish to light grey for the lime mortars. All the lime samples present a sandy aggregate fraction with mainly greyish clasts except for M06 which shows mainly whitish clasts. 2.5. Colour analysis Colour features were measured in powdered samples from the two buildings and in intact samples of Fuengirola Castle. The analyses were carried out employing reflectance spectroscopy in the visible region (Vis-SP) as reference technique through a portable CM-26d spectrophotometer by Konica Minolta, calibrated with reference white and black to convert the signal to CIELAB colour space coordinates. In the CIELAB colour space, L* represents the perceptual lightness, ranging from 0 (black) to 100 (white), while a* and b* correlate with chroma perceptions on green–red and blue-yellow axes, respectively (Fairchild, 2013). Powdered samples were put in a small cylindric container and analysed with a 3 mm measurement spot, while intact samples were analysed with an 8 mm spot in five different random areas in which both aggregate and binder fractions were present, and median values were used. Concerning imaging (see Annex 2 for a flowchart resuming the procedure), the smartphone photos were made in a controlled environment 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; 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: R2019b) by MathWorks and Colorlab toolbox (Malo and Luque, 2002). Firstly, the photos were cropped in order to process data from mortar surface only, RGB parameters for each pixel were converted to XYZ tristimulus values and, finally, to the parameters of the CIELAB colour space employing the photo of a reference white taken in the same conditions of the other ones. Median L*, a* and b* were employed. 2.2. X-ray diffraction (XRD) All the analyses were carried out on the whole sample, without the separation of aggregate and binder fractions. As a first step, each sample was powdered and homogenised by agate mortar and pestle prior to the analysis. The analyses were carried out through using a Bruker D2 Phaser X-ray diffraction instrument, Cu Kα radiation λ = 1.5418 Å, range 5-65◦ 2θ with 0.02◦ 2θ per step and accumulation time of 10 s. The 3 M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433 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 lime mortars. These major elements profiles are consistent with mineralogical features evidenced by mineralogical profiles obtained by XRD which pointed out higher amounts of quartz and aluminosilicates in earth mortars than in lime ones, and lower amounts of carbonates. Earth mortar samples are also richer in REE and most trace elements (Bi, Cd, Co, Cr, Li, Mn, Mo, Ni, Th, Tl, and V) and lower in Sr than lime mortars. The chemometric exploration of the multielement analysis data conducted by PCA allowed to highlight similarities and differences among the lime mortars. The first model (Annex 4) shows the results obtained on all the considered elements. Earth mortars were excluded from this analysis since their peculiar features could have hidden the differences among lime ones in the model. Annex 4a shows the scores of the two first PC for the samples, while Annex 4b-c shows the variables loadings. The first two components explain 53.3% and 12.9% of the total variance respectively. The first PC is positively correlated with most of the elements and negatively with Ca and some trace elements (Pb, Mo, Sr and U), while PC2 shows the most intense positive correlations with 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 chronologies are scattered in the samples/scores plot and clusters related to the different construction phases cannot be clearly observed. Most Islamic samples have lower PC1 scores than most mortars from the following phases. However, P2M3 falls among them. The unclassified sample of SB fall isolated in the first quadrant. N3B is in the third quadrant, close to the two samples of the 20th century (O8-9), the unclassified sample of plaster and the 14th century one (O10). The second model was made employing REE as variables (Fig. 2), since previous archaeometric studies proved their effectiveness as chronological markers in mortars (Gallello et al., 2017; Ramacciotti et al. 2018). The first two PC explain more than 95% of the total variance. The first PC (Fig. 2b) is positively correlated with all the variables, while PC2 (Fig. 2c) shows a fractionation of REE since lighter REE (from La to Pr) and heavier REE (from Tm to Lu) have negative coefficients, while the others (from Nd to Er) have positive ones. Sc and Y have negative and positive coefficients respectively. Concerning the scores diagram (Fig. 2a), samples from the centuries 14th (O10) and 20th (O89) have lowest scores for PC1 or higher scores for PC2 than Islamic samples, which are scattered on the two PC-axes. The sample from the modern restoration (P2M3), which previously grouped with the Islamic mortars, falls isolated due to high PC2 scores. The unclassified samples group clearly with the mortars from the Islamic constructive phases, except for the plaster one, which shows PC1 scores closer to the 14th and 20th century mortars but negative PC2. 2.6. Exploratory data analysis Statistics and data visualisation were carried out in R (version 4.0.2; R Core Team, 2020) employing the following R packages: factoextra (version: 1.0.7; Kassambara and Mundt, 2020), ggplot2 (version: 3.3.3; Wickham, 2016), signal (version: 0.7–6; signal developers, 2013), ggrepel (version: 0.9.1; Slowikovski, 2021), ggpubr (version: 0.4.0; Kassambara, 2020). Principal component analysis (PCA) was used separately in Silla Tower samples and in Fuengirola Castle ones to reduce dimensionality and analyse the main variance tendencies of the datasets, evaluating the presence of groups. Concerning multielement analysis data, for both the sites, two models were built: the first with all the elemental concentrations, the second only with REE as variables. The concentrations were standardised prior to the analysis. Principal component analysis was carried out also for ATR-IR data in the region between 2000 and 700 cm−1. Each spectrum was pre-processed by Savitzky-Golay filter (polynomial order: 2nd, length: 13, derivative: 2nd) and signals were mean-centred prior to the PCA. Cluster analysis (CA method: average linkage) employing the first two PC as variables was also carried out for Fuengirola Castle samples. 3. Results and discussion 3.1. The mortars from Silla Islamic Tower 3.1.1. X-ray diffraction results The mineralogical analysis (Table 2) conducted on the samples allowed to highlight the qualitative mineralogy of the mortars under examination. In qualitative mineralogy, what immediately stands out is the substantial difference of the mortars of the Roman period (P0M3 and P0M4) compared to the other subsequent periods. These mortars have a relatively high quartz content compared to calcite. High amounts of quartz are not unusual in earth mortars aggregate fraction, as observed in previous studies (Gómez Morgade et al., 2021). The samples P2M1 and P2M2, from Islamic period presents traces of gypsum. Gypsum-lime mortars are known for the same period in other places in Spain and Portugal (Genestar and Pons, 2003; Freire et al, 2008; Vitti, 2021). However, minor amount of this mineral can probably be better explained by chemical weathering products (Lanas et al., 2005). The P1EN sample has also traces of gypsum, but a lower quartz content than the previous ones. Samples SA and SB are qualitatively similar to each other. Likewise, the O8, O9 and O10 samples are mineralogically similar, although O10 is archaeologically attributed to the 14th century. 3.1.2. Multielement analysis Major elements concentrations are reported in Table 3, while the trace elements and REE ones are reported in the Annex 3a and Annex 3b, respectively. The difference between earth mortars (P0M3-4) and lime mortars 3.1.3. Archaeological evidence Resuming the obtained results from an archaeological point of view, it is interesting to observe that the first discrimination is between the 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 N3A N3B N4 S6 SA SB O8 O9 O10 P0M1 P0M2 XX XXX XX XXX XXX XXX XXX XXX XXX XX XXX tr XX XX XX XX XX XX tr tr tr XX XX X tr tr X X X Pl Phyll. Others tr tr tr tr tr tr tr tr tr tr tr tr tr tr tr tr tr tr Sample Cal P0M3 P0M4 P1EN P1QU P1M1 P1M2 P2M1 P2M2 P2M3 P3M1 X X XXX XXX XXX XXX XXX XXX XXX XXX Dol Qtz Kfs Pl Phyll. XXX XXX X XX XX XX XX XX XX XX tr tr tr tr tr tr tr tr tr tr tr tr Others Gp (tr) Gp (tr) Gp (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 N3B N4 S6 SA SB O8 O9 O10 P0M1 P0M2 0.90 0.82 1.03 0.78 0.98 0.89 0.88 0.38 0.70 1.16 1.29 9.85 5.63 9.78 8.05 9.76 9.55 2.06 3.06 2.82 9.52 10.04 0.09 ND 0.27 0.16 0.41 1.06 ND ND ND 0.60 0.62 24.96 32.33 25.24 30.18 25.79 21.28 37.26 30.36 32.40 27.44 26.18 0.06 0.05 0.06 0.06 0.07 0.05 0.02 0.04 0.02 0.10 0.12 0.6 0.53 0.59 0.57 0.65 0.57 0.34 0.5 0.38 0.89 0.92 P0M3 P0M4 P1M1 P1M2 P1EN P1QU P2M1 P2M2 P2M3 P3M1 4.03 4.41 1.32 1.57 0.78 1.05 0.86 0.80 0.74 0.57 18.53 19.26 7.36 9.12 2.19 5.88 8.16 8.48 8.61 7.21 2.23 2.59 0.35 0.70 ND 0.14 0.20 0.49 0.07 0.07 8.82 8.00 27.58 25.11 34.76 29.39 29.05 25.54 23.87 28.67 0.32 0.33 0.08 0.12 0.02 0.06 0.05 0.07 0.05 0.05 2.22 2.27 0.61 0.93 0.23 0.5 0.48 0.63 0.47 0.47 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 observed also in the structures from the Roman Republican phase of Sagunto Castle (Gallello et al., 2017) and it is possibly a recurring technique in the area during this period, although more studies are needed in order to confirm this hypothesis. The mortars from the 14th (O10) and the 20th (O8-9) centuries structures cannot be discriminated through multivariate statistics, but their elemental concentrations and especially REE ones are quite different from those of samples from the previous periods. Concerning the unclassified samples, the plaster sample (P1EN) does not show clear relationship with the others, maybe because it is a completely different type of mortar, while N4, SA and SB have REE levels similar to the Islamic mortars. The results for N4 seem in contradiction with the archaeological hypothesis, which suggested a more recent dating (Christian period), however, it could also be explained by a continuity in raw material use and mortar manufacturing recipe also in a part of the period following the Islamic occupation. Table 4 Results of XRD analyses of Fuengirola Castle mortars. Sample Cal M01 M02 M03 M04 M05 M06 M07 M08 M09 M10 M11B M11W M12 X X X X X X XXX XX X XX XX XX XX Dol X X XXX X X Qtz Kfs XXX XXX XXX XX XXX tr XX XXX XXX XXX XXX XXX XX tr tr tr X tr tr tr tr tr Pl X tr X tr tr tr Phyll. tr tr tr X X tr tr tr XX tr X Others Arg (X) Arg (tr), Hl (tr) Note: Semiquantitave estimates are reported from traces (tr), and to small amounts (X) to large amounts (XXX). Cal: calcite, Dol: dolomite, Qtz: quartz, Ffs: K-feldspar, Pl: plagioclase, Phyll.: phyllosilicates, Arg: aragonite, Hl: Halite. 3.2. The mortars from Fuengirola Castle 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 samples M01 and M02 seem to be quite similar. The presence of aragonite in the samples M04 and M11W could be due to the presence of crushed shells in the composition of the mortar. Sample M09 is characterized by the marked presence of phyllosilicates. (Fuentes et al., 2006; Olivares et al., 2009; Shillito et al., 2009). The wide band at ~ 3400 cm−1 and the one of ~ 1635 cm−1 in some samples suggest the presence of water (Silva et al., 2005), maybe linked to phyllosilicates or other hydrated phases. Some samples seem to be earthy (M02, M11B) and M05 is an earth mortar. The PCA was carried out to determine the main characteristics concerning the variance of the dataset and to evaluate the presence of groups. Cluster analysis was also employed to validate the interpretation. The results of the data analysis can be observed in Fig. 4. The first two PC explain 75.2% and 13.7% of the overall variance respectively. The variables that have the highest influence for PC1 are signals close to 1400, 875 and 715 cm−1 (Fig. 4b). These wavenumbers correspond to bands of carbonates and their intensities in the original spectra are positively correlated to the first PC. It can be observed in the scores plot (Fig. 4a), corroborated by the dendrogram, that M07 (12th c.) and M12 can be distinguished due to the higher scores of PC1. It is worth noting that although the earth mortar M05 does not differ significantly from the other samples according to the dendrogram, it is characterised by the 3.2.2. Infrared spectroscopy The ATR-IR spectra are shown in Fig. 3. As can be observed, the main bands of the spectra correspond to the vibrational modes of carbonates between 1420 and 1405 cm−1, between 725 and 715 cm−1, and at ~ 875 cm−1 (Clark et al., 1990; Silva et al., 2005; Bruckman and Wriessnig, 2013). Less intense band probably linked to carbonates can be observed in all the samples except for the one of earth mortar (M05) at ~ 1800 cm−1 (Silva et al., 2005). Another relevant feature can be seen at ~ 990 cm−1 in all the mortar samples, except for M06, and it is probably caused by the presence of aluminosilicates (Shillito et al., 2009). Bands at ~ 1160 and ~ 1085 cm−1, as well as those of about 795 and 780 cm−1 could mark the presence of aluminosilicates as well or silicon dioxide 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. Carbonate bands have a relevant influence also on PC2 (Fig. 4c). 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 of the bands at about 1420, 885 and 730 cm−1 in the original spectra are linked to higher scores on the PC2, while the intensity at about1395 cm−1 is linked to lower values of the same PC. The feature at ~ 1340 cm−1, observable as a shoulder of ~ 1400 cm−1 band in M11W and M11B, and the one at ~ 850 cm−1, shoulder of the ~ 875 cm−1 band, are possibly caused by the presence of organic matter (Legan et al., 2020) and aragonite (Pronti et al., 2020), respectively. On this direction, a band related to the aluminosilicates has a certain influence, as indicated by loadings around 1000 cm−1. In this case, higher intensity corresponds to lower score. Sample M06 (18th c.) is characterised by the highest PC2 score. Its peculiarity, compared to the other samples is evidenced also by the dendrogram. Looking at the spectrum (Fig. 3), it can be observed 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 calcite one at ~ 715 cm−1. These two features suggest a relevant amount of dolomite (Bruckman and Wriessnig, 2013). This characteristic was pointed out also by XRD analysis which revealed large amount of dolomite in M06, while the other samples were characterised by a small to moderate presence of this carbonate mineral in the other sample. Sample score is also coherent with the absence of aluminosilicate band at ~ 990 cm−1. These facts indicate that for the production of this sample (M06) the employed raw materials were considerably different from those of the other lime mortars. Table 5 Major elements concentrations for Fuengirola Castle samples. Sample Al Si K Ca Ti Fe M01 M02 M03 M04 M05 M06 M07 M08 M09 M10 M11B M11W M12 0.98 1.19 ND 1.41 3.35 ND ND 0.42 2.81 0.41 1.86 0.60 0.28 18.63 11.92 13.70 12.95 16.46 0.45 4.72 14.09 15.01 7.85 10.37 5.44 8.02 0.26 0.60 0.16 0.60 2.07 ND ND 0.10 1.25 0.05 1.63 1.14 ND 15.08 15.47 15.06 14.58 5.90 24.27 22.11 12.83 9.85 19.26 9.38 14.65 20.49 0.11 0.10 0.06 0.18 0.35 0.01 0.03 0.08 0.27 0.06 0.23 0.12 0.04 1.93 1.64 1.31 3.19 4.23 0.20 0.67 1.48 3.35 1.18 2.97 2.07 0.75 Note: Concentrations are expressed as mass percentage. ND = not detected. 3.2.3. Multielement analysis Results for pED-XRF analysis can be observed in the Table 5, while results from ICP-MS in the Annex 5a (trace elements) and Annex 5b (REE). From the point of view of major elements, some peculiarities can be observed: Al and K could not be detected by the spectrometer in the sample M06, which show the lowest levels of Si, Ti and Fe, while it has the highest level of Ca. As previously stated, this sample is characterised by a different aggregate also from the macroscopic point of view and has high amounts of carbonate minerals and low ones of quartz and aluminosilicates 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. Among the other samples, M05, M09 and M11B have the lowest levels of Ca and the highest of Al. In the case of M05, the high amount of Al is combined with the highest level of Si, which can be explained by the presence of earth. In order to explore the dataset reducing the variable number, PCA was employed. A first PCA (Annex 6) was carried out with all the analysed elements as variables and CA with PC1 and PC2 was also carried out to guide scores plot interpretation. The first two PC explain 80.6% of the overall variance. The first PC loadings (Annex 6b) is influenced in the positive direction by elements probably linked to aluminosilicates (Al, Si, K, Ti, Fe, REE and most trace elements), while Ca and few trace elements (Mo, Sr and U) are correlated in the negative direction with this PC. As regards PC2 (Annex 6c), loadings show intense negative correlations with Si, Fe, Cd, Cr, Co, Ni and Sr, and positive ones with K, Pb, Mo, Tl, V y U. Scores diagram (Annex 6a) and CA dendrogram show that the most important separation is the one between earth (M05) and lime mortars. The lime mortars can be divided in other sub-groups. M09 and M11B, rich in aluminosilicates elements, cluster together and are half-way between earth and lime mortars. M04 (19th c.) is closer to them but shows completely different PC2 scores. M07 (12th c.) and M12, and further, M06 (18th c.) and M03 pertain to another sub-cluster; while 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 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). The first two PC explain more than 99% of the overall variance. The first PC is linked to the total amount of REE and all the variables show negative coefficients (Fig. 5b). The second PC (Fig. 5c) shows two kinds of REE fractionation. Indeed, loadings seem to be positively correlated to elemental mass, and Eu has opposite coefficient compared to its neighbouring elements (Sm and Gd). Scores diagram (Fig. 5a), corroborated by PC1-2 dendrogram, evidences the distance between earth mortar (M05), rich in REE, and lime mortars. M09, M11B and M04 are also in this case midway between the two groups on PC1, although closer to lime mortars. M06 (18th c.) has very low levels of REE and the highest PC1 score. The other samples group in two subclusters (1st: M01, M12, M03, M07 (12th c.); 2nd: M02, M08 (16th c.), M10, M11W). 3.2.4. Preliminary considerations Due to the reduced sample size, the obtained results permitted only preliminary considerations. The analytical results and data processing evidenced the presence of three types of mortars, according to their chemical features: sample M05, an earth mortar, sample M06 (18th c.), characterised by a different aggregate and a relevant amount of dolomite, and M04 (19th c.), which showed characteristic elemental levels evidenced by the PCA employing all the elements (Annex 6). Features of M09 and M11B are probably related to the presence of earth, although the cause is not clear (contamination or voluntary addition?). The difference among the other samples is less evident but data analysis suggests a possible similarity of M07 (12th c.) and M12. 3.3. Colour characterisation As previously stated, colour characterisation was carried out employing imaging of smartphone pictures and Vis-SP analysis as reference method (results are shown in the Annex 7). Powdered samples 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 results 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 (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 unclassified mortars, most of them are plotted within the Islamic samples suggesting this possible classification, which is consistent with multielement analysis data. P1EN1 shows instead colour properties different from the other samples which is consistent with elemental and mineralogical 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 VisSP and imaging data show strong linear relationship only for L* (R2 = 0.87; Annex 9a). CIELAB diagrams (Annex 9b) evidence separation between M05 and M11B and the other mortars, although, on the contrary Fig. 6. Biplots for L* vs a* (top) and L vs b* (bottom) employing imaging data of Silla Islamic Tower powdered samples. Fig. 7. Biplots for L* vs a* (top) and L* vs b* (bottom) employing imaging data of Fuengirola Castle powdered samples. of powdered samples, in this case M11W is plotted closer to these and not to the other lime mortars. Intact M07 has the highest L* and the lowest b*, as for powdered sample, while M06 is plotted within the other lime mortars. The issues in obtaining colour values consistent with those of the reference technique in intact mortars is probably related to the difficulty in characterising samples with irregular and heterogeneous surface. Further studies are needed with a large sample size to improve the smartphone method as a viable non-invasive method to characterise ancient mortars colour features. Table 6 Pearson correlations (r) between colour parameters (imaging) and major elements in lime mortars from the two architectural complexes. Mortars Silla (n = 19) Fuengirola (n = 12) All (n = 31) L* a* b* L* a* b* L* a* b* Al Si K Ca Ti Fe −0.54 0.78 0.68 −0.79 0.58 0.68 −0.68 0.61 0.64 −0.69 0.53 0.75 −0.32 0.08 0.10 −0.47 0.21 0.20 −0.74 0.71 0.68 −0.93 0.81 0.90 −0.87 0.77 0.69 0.57 −0.40 −0.58 0.80 −0.59 −0.65 0.65 −0.32 −0.10 −0.73 0.86 0.92 −0.87 0.67 0.76 −0.87 0.67 0.57 −0.81 0.89 0.91 −0.82 0.58 0.70 −0.81 0.47 0.27 4. Conclusion Different analytical techniques were employed to characterise the samples from Silla Islamic Tower and Fuengirola Castle. The results of multielement analysis suggested that most of the Note: ND for Al and K were changed with half of the lowest detected concentration. Barred coefficients are statistically not significant (p >.05). 9 M. Ramacciotti et al. Journal of Archaeological Science: Reports 43 (2022) 103433 unclassified samples from Silla Tower date back to the Islamic construction phase, while some characteristic mortars could be observed among Fuengirola Castle from the phases of 12th century and from the restorations of the past two centuries. However, to confirm the construction period of undated structures, a wider number of samples should be studied and additional standardised techniques should be used. As regards the use of imaging on smartphone photos as an innovative archaeometric screening method for ancient mortars, it is possible to observe that data obtained from powdered samples showed strong linear relationship with those of the employed reference technique (Vis-SP). In lime mortars, L* has positive correlation with Ca, and negative with the other major elements, while both a* and b* have correlations opposite to L*. This suggests that lightness increases together with the amount of carbonate minerals, and a* and b* together with other minerals. This fact is also corroborated by XRD and ATR-IR analyses which pointed out higher levels of carbonates in the samples characterised by the higher L*. The variance of these parameters in the samples could be linked to differences in mortar recipes or raw materials, as well as to different degradation levels. Furthermore, it is worth noticing that the classification of Silla undated mortars obtained by colour parameters is consistent with that of multielement analysis. On the other hand, about the Fuengirola Castle sample set, the most relevant difference is among earthy samples and lime mortars, although also dolomite-rich sample (M06) seems to have slightly different colour features which were not evident by naked-eye examination of powdered samples. This preliminary smartphone test shows the potential that colour characterisation by imaging have, being a cheap and fast method to carry out chronological classification of ancient mortars, at least for powdered samples, which could be a first screening step in order to carry out more invasive approaches. However, more studies should be carried out in order to optimise the methodological approach and corroborate the results of this first study. Concerning intact samples, the test on the mortars of Fuengirola Castle gave more challenging methodological issues. However, L* shows good agreement with Vis-SP values, suggesting that further studies are needed in order to improve non-invasive image analysis of mortars. To go deeper and understand if issues such as mortar recipes, raw materials (binder/aggregate ratios, higher amounts of aluminosilicates and oxides in the aggregate fraction, limestone impurities or pozzolanic materials addition), or contamination and weathering are highlighting smartphone colour differences, further studies, employing standard methods such as petrographic analysis, should be carried out investigating the relationship between the colour features and the properties of mortar binder and aggregate fractions. This archaeometric study further improves the archaeological knowledge on the studied monuments, highlighting the diachronic features of the construction phases related to different mortar manufacturing aspects, pointing out the possibilities of a multianalytical approach based on portable and innovative methodologies together with widely used analytical techniques. the project “Smartphone and Green Analytical Chemistry” (PROMETEO 2019-056) funding and the related predoctoral scholarship. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.jasrep.2022.103433. References Alapont, L.l., Pitarch, A., Orengo, H.A., 2016. Arqueologia de l’Horta Sud: L’origen d’una comarca histórica. Perifèric Edicions, Catarroja. Aguilar Cuesta, Á.I., Vallina Rodríguez, A., García Juan, L., 2019. Fuengirola a través del Catastro de Ensenada. In: Martín Ruiz, J.A. (Ed.), De Suel a Fuengirola: Actas I Jornadas de Arqueología y Patrimonio. La Serranía, Fuengirola, pp. 151–187. Azkarate, A., 2020. Archaeology of architecture: buildings archaeology. In: Orser, C.E., Zarankin, A., Funari, P.P.A., Lawrence, S., Symonds, J. (Eds.), The Routledge Handbook of Global Historical Archaeology. Routledge, New York, pp. 517–536. Brogiolo, G.P., 2007. Dall’Archeologia dell’architettura all’Archeologia della complessità. Pyrenae 38, 7–38. Bruckman, V.J., Wriessnig, K., 2013. Improved soil carbonate determination by FT-IR and X-ray analysis. Environ. Chem. Lett. 11, 65–70. https://doi.org/10.1007/ s10311-012-0380-4. Cantisani, E., Calandra, S., Barone, S., Caciagli, S., Fedi, M., Garzonio, C.A., Liccioli, L., Salvadori, B., Salvatici, T., Vettori, S., 2021. The mortars of Giotto’s Bell Tower (Florence, Italy): raw materials and technologies. Constr. Build. Mater. 267, 120801 https://doi.org/10.1016/j.conbuildmat.2020.120801. 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. J. Archit. Herit. 11, 670–684. https://doi.org/10.1080/15583058.2017.1287978. Chiarelli, N., Miriello, D., Bianchi, G., Fichera, G., Giamello, M., Turbanti Memmi, I., 2015. Characterisation of ancient mortars from the S. Niccolò archaeological complex in Montieri (Tuscany – Italy). Constr. Build. Mater. 96, 442–460. https:// doi.org/10.1016/j.conbuildmat.2015.08.023. Clark, R.N., King, T.V., Klejwa, M., Swayze, G.A., Vergo, N., 1990. High spectral resolution reflectance spectroscopy of minerals. J. Geophys. Res.: Solid. Earth 95, 12653–12680. https://doi.org/10.1029/JB095iB08p12653. Columbu, S., Antonelli, F., Sitzia, F., 2018. Origin of Roman worked stones from St. Saturno Christian Basilica (South Sardinia, Italy). Mediterr. Archaeol. Ar. 18, 17–36. https://doi.org/10.5281/zenodo.1256047. Crisci, G.M., Franzini, M., Lezzerini, M., Mannoni, T., Riccardi, M.P., 2004. Ancient mortars and their binder. Period. Miner. 73, 259–268. Cruz-Fernández, M., Luque-Cobija, M.J., Cervera, M.L., Morales-Rubio, A., de la 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. De Luca, R., Barca, D., Bloise, A., Cappa, M., De Angelis, D., Fedele, G., Mariottini, S., Micieli, D., Muto, F., Piluso, E., Pucino, F., Taranto, M., Tropea, M., Miriello, D., 2021. RecoStones: a New Tool to Identify Calabrian Stone Materials Through Image 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 architecture. Ann. Hist., Sci. Soc. – English Edition 72 (1), 69–86. https://doi.org/ 10.1017/ahsse.2019.6. 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 spectrometry. X-Ray Spectrom. 39, 146–153. https://doi.org/10.1002/xrs.123. Fairchild, M.D., 2005. Color appearance models, second ed. John Wiley & Sons Ltd., Chirchester. Ferris, I.M., 1989. The Archaeological Investigation of Standing Buildings. Vernac. Archit. 20, 12–17. https://doi.org/10.1179/vea.1989.20.1.12. Francovich, R., Bianchi, G., 2002. L’archeologia dell’elevato come archeologia. Arqueol. Arquit. 1, 101–111. Freire, T., Santos Silva, A., Veiga, M.R., Brito, J.D. (2008). Characterization of Portuguese historical gypsum mortars, in: Laboratório Nacional de Engenharia Civil (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., Amador, U., 2006. Room-temperature synthesis of apatite-type lanthanum silicates by mechanically milling constituent oxides. Solid State Ion. 177, 1869–1873. https://doi.org/10.1016/j.ssi.2006.02.032. Gallello, G., Ramacciotti, M., Lezzerini, M., Hernandez, E., Calvo, M., Morales, A., Pastor, A., de la Guardia, M., 2017. Indirect chronology method employing rare earth elements to identify Sagunto Castle mortar construction periods. Microchem. J. 132, 251–261. https://doi.org/10.1016/j.microc.2017.02.009. 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. culher.2003.02.001. 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. Rep. 37, 102931 https://doi.org/10.1016/j.jasrep.2021.102931. Herreros-Chavez, L., 2021. Estrategias verdes para el análisis de muestras alimentarias y medioambientales. University of Valencia. PhD thesis. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements Gianni Gallello acknowledges the financial support of the Beatriz Galindo Fellowship (2018) funded by the Spanish Ministry of Science and Innovation and Ministry of Universities (Project BEAGAL18/00110 “Development of analytical methods applied to archaeology”). M. Luisa Cervera, Agustín Pastor and Mirco Ramacciotti acknowledge the Ministry 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 Version 0.4.0. https://CRAN.R-project.org/package=ggpubr. Kassambara, A., Mundt, F., 2020. factoextra: Extract and Visualize the Results of Multivariate Data Analyses. R package version 1.0.7. https://CRAN.R-project.org /package=factoextra. Klowikowski, K., 2021. ggrepel: Automatically Position Non-Overlapping Text Labels with ’ggplot2’. R package version 0.9.1. https://CRAN.R-project.org/package=ggre pel. Lanas, J., Sirera, R., Alvarez, J.I., 2005. Compositional changes in lime-based mortars exposed to different environments. Thermochim. Acta 429(2), 219–226. DOI: 10.1016/j.tca.2005.03.015Legan, L., Leskovar, T., Črešnar, M., Cavalli, F., Innocenti, D., Ropret, P., 2020. Non-invasive reflection FTIR characterization of archaeological burnt bones: reference database and case studies. J. Cult. Herit. 41, 13–26. https:// doi.org/10.1016/j.culher.2019.07.006. Lezzerini, M., Legnaioli, S., Lorenzetti, G., Palleschi, V., Tamponi, M., 2014. Characterization of historical mortars from the bell tower of St. Nicholas church (Pisa, Italy). Constr. Build. Mater. 69, 203–212. https://doi.org/10.1016/j. conbuildmat.2014.07.051. Lezzerini, M., Ramacciotti, M., Cantini, F., Fatighenti, B., Antonelli, F., Pecchioni, E., Fratini, F., Cantisani, E., Giamello, M., 2016. Archaeometric study of natural hydraulic mortars: the case of the Late Roman Villa dell’Oratorio (Florence, Italy). Archaeol. Anthr. Sci. 9, 603–615. https://doi.org/10.1007/s12520-016-0404-2. Lezzerini, M., Raneri, S., Pagnotta, S., Columbu, S., Gallello, G., 2018. Archaeometric study of mortars from the Pisa’s Cathedral Square (Italy). Meas. 126, 322–331. https://doi.org/10.1016/j.measurement.2018.05.057. Malo, J., Luque, M.J. (2002). COLORLAB: a color processing toolbox for Matlab. https ://www.uv.es/vista/vistavalencia/software/colorlab.html. Miriello, D., Barca, D., Bloise, A., Ciarallo, A., Crisci, G.M., De Rose, T., Gattuso, C., Gazineo, F., La Russa, M.F., 2010a. Characterisation of archaeological mortars from Pompeii (Campania, Italy) and identification of construction phases by compositional data analysis. J. Archaeol. Sci. 37 (9), 2207–2223. https://doi.org/ 10.1016/j.jas.2010.03.019. Miriello, D., Bloise, A., Crisci, G.M., Barrese, E., Apollaro, C., 2010b. Effects of milling: a possible factor influencing the durability of historical mortars. Archaeom. 52, 668–679. https://doi.org/10.1111/j.1475-4754.2009.00494.x. Miriello, D., Barca, D., Pecci, A., De Luca, R., Crisci, G.M., López Luján, L., Barba, L., 2015. Plasters from Different Buildings of the Sacred Precinct of Tenochtitlan (Mexico City): Characterization And Provenance. Archaeom. 57, 100–127. https:// doi.org/10.1111/arcm.12074. Olivares, M., Tarriño, A., Murelaga, X., Baceta, J.I., Castro, K., Etxebarria, N., 2009. Nondestructive spectrometry methods to study the distribution of archaeological and geological chert samples. Spectrochim. Acta Part A: Mol. Biomol. Spectrosc. 73, 492–497. https://doi.org/10.1016/j.saa.2008.12.036. Pronti, L., Romani, M., Viviani, G., Stani, C., Gioia, P., Cestelli-Guidi, M., 2020. Advanced methods for the analysis of Roman wall paintings: elemental and molecular detection by means of synchrotron FT-IR and SEM micro-imaging spectroscopy. Rend. Lincei. Sci. Fis. Nat. 31, 485–493. https://doi.org/10.1007/s12210-02000888-9. R Core Team, 2020. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria https://www.R-project.org/. Ramacciotti, M., Rubio, S., Gallello, G., Lezzerini, M., Columbu, S., Hernandez, E., Morales-Rubio, A., Pastor, A., de la Guardia, M., 2018. Chronological Classification of Ancient Mortars Employing Spectroscopy and Spectrometry Techniques: Sagunto (Valencia, Spain). J. Spectrosc. 2018, 9736547. https://doi.org/10.1155/2018/ 9736547. Rezazadeh, M., Seidi, S., Lid, M., Pedersen-Bjegaard, S., Yamini, Y., 2019. The modern role of smartphones in analytical chemistry. TrAC-Trend. Anal. Chem. 118, 548–555. https://doi.org/10.1016/j.trac.2019.06.019. Shillito, L.M., Almond, M.J., Wicks, K., Marshall, L.J.R., Matthews, W., 2009. The use of FT-IR as a screening technique for organic residue analysis of archaeological samples. Spectrochim. Acta Part A: Mol. Biomol. Spectrosc. 72, 120–125. https:// doi.org/10.1016/j.saa.2008.08.016. Signal developers (2013). signal: Signal processing. http://r-forge.r-project.org/projects/ signal/. Silva, D.A., Wenk, H.R., Monteiro, P.J.M., 2005. Comparative investigation of mortars from Roman Colosseum and cistern. Thermochim. Acta 438, 35–40. https://doi.org/ 10.1016/j.tca.2005.03.003. Sitzia, F., Beltrame, M., Columbu, S., Lisci, C., Miguel, C., Mirão, J., 2020. Ancient restoration and production technologies of Roman mortars from monuments placed in hydrogeological risk areas: A case study. Archaeol. Anthrop. Sci. 12, 147. https:// doi.org/10.1007/s12520-020-01080-8. Tenconi, M., Karatasios, I., Balaawi, F., Kilikoglou, V., 2018. J. Cult. Herit. 33, 100–116. https://doi.org/10.1016/j.culher.2018.03.005. Vitti, P., 2021. Mortars and masonry—structural lime and gypsum mortars in antiquity and Middle Ages. Archaeol. Anthr. Sci. 13, 1–19. https://doi.org/10.1007/s12520021-01408-y. Wickham, H., 2016. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag, New York. https://doi.org/10.1007/978-3-319-24277-4. 11