The first scientific plan of the Mosque-Cathedral of Córdoba: Graphic and dimensional analysis of an oil painting from 1741

RESEARCH ARTICLE The first scientific plan of the MosqueCathedral of Co ´rdoba: Graphic and dimensional analysis of an oil painting from 1741 Juan Cantizani-Oliva a , Juan-Francisco Reinoso-Gordo b , Antonio Ga ´miz-Gordo c, * a Electrical and Automatic Engineering Department, University of Cordoba, Cordoba 14071, Spain b Architectural and Engineering Graphic Expression Department, University of Granada, Granada 18071, Spain c Architectural Graphic Expression Department, University of Seville, Seville 41012, Spain Received 8 December 2022; received in revised form 7 February 2023; accepted 24 February 2023 KEYWORDS Mosque-Cathedral; Co ´rdoba; Historical plans; Photogrammetry; 3D laser scanner Abstract This research analyzes the earliest located floor plan of the Mosque-Cathedral of Co ´rdoba, an anonymous oil painting in 1741, which has not been studied so far. The objective is to know the dimensional accuracy of the most relevant architectural forms drawn, considering the elements referenced in their legend and the graphic symbols used, to assess their documentary interest. It has also been compared with two important plans of the MosqueCathedral drawn in 1767 and 1868. The first task was a photogrammetric survey of the oil painting has been carried out. Subsequently, some data has been measured in the monument using a 3D scanner. The orthophoto of the oil painting has been overlapped to the digital model to verify its metric accuracy in a selection of points. For the first time, the legend and labels included in the oil painting have been transcribed. This precise graphic document contains reliable abundant data for future research about the transformations and restorations of a monument that is part of the UNESCO World Heritage List. The oil painting dating from 1741 can be considered as the first scientific plan of the Mosque-Cathedral of Co ´rdoba, and as an outstanding architectural survey of eighteenth-century Europe. ª2023 Higher Education Press Limited Company. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). * Corresponding author. E-mail address: [email protected] (A. Ga ´miz-Gordo). Peer review under responsibility of Southeast University. https://doi.org/10.1016/j.foar.2023.02.004 2095-2635/ª2023 Higher Education Press Limited Company. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Available online at www.sciencedirect.com ScienceDirect journal homepage: www.keaipublishing.com/foar Frontiers of Architectural Research 12 (2023) 587e600 1. Introduction 1.1. Brief historical data The Mosque-Cathedral of Co ´rdoba in southern Spain, a monument located in the historic downtown, next to the Guadalquivir river and the Roman bridge (Fig. 1a) has preserved both its religious use and architectural identity for more than a thousand years (Moneo-Valle ´s, 1985). Its construction process began in the middle of the 7th century and was completed in the late 10th century, after several extensions to the south and east, among which the Al-Hakam II one, built between 961 and 966, is of special architectural interest (Chueca-Goitia, 1965;NietoCumplido, 2007;Cantizani-Oliva et al., 2022). While other mosques in the Iberian Peninsula were demolished and replaced by churches or cathedrals, this one was consecrated as a Christian temple after the city occupation in 1236, and since then it has undergone numerous transformations. In the 16th century a large Gothic central nave was built inside and in the 18th century some vaults and new skylights were built in its naves, while its interior perimeter was populated by Christian chapels. The current bell tower that surrounds the Islamic minaret was completed in 1663, with five superimposed bodies (Herna ´ndez-Gime´nez, 1975): minaret, bell tower, clock, lantern, and crown with a sculpture of St. Raphael (Fig. 1b). Important architectural restorations began at the beginning of the 19th century and in 1882 the MosqueCathedral was declared National Monument. Since then, numerous architects have been in charge of its conservation and numerous interventions that have helped to recover significant fragments from different periods (HerreroRomero, 2016). The monument was inscribed on the World Heritage List in 1984, and this site was extended to include part of the Historic Center in 1994. 1.2. Graphic documentary sources The Mosque-Cathedral of Co ´rdoba has a valuable legacy of historical images prior to the proliferation of photography around 1850. These images constitute an important source for heritage research that is complemented by other historical or archaeological documentary sources, both Muslim and Christian (Nieto-Cumplido, 2007). The 1741 oil painting, object of this research, was published with poor graphic quality and without any study in the Bulletin of the Royal Academy of Sciences, Fine Arts and Noble Arts of Co ´rdoba in 1944 (issue 51, p. 456). It was also reproduced, without any type of analysis, in a book including plans and drawings of the Mosque-Cathedral (Nieto-Cumplido and Luca-de-Tena-Alvear, 1992); and in a book which includes a partial transcription of the legend (Cosano-Moyano, 1999). It has been mentioned in a book plenty of historic documentation (Nieto-Cumplido, 2007), in another with current plans of the monument (RuizCabrero, 2009), and in a paper about graphic sources of the Mosque-Cathedral until 1850 (Ga ´miz-Gordo, 2019). In addition, there are plans of the bell towerdalso drawn in the 1741 paintingddating from the 20th century in the aforementioned books by Nieto-Cumplido and Luca-deTena y Alvear, and in the one by Ruiz-Cabrero. The first known interior perspective was included in a book by Henry Swinburne in 1779, and the first collection of views was published by Alexandre Laborde in 1812 (Ga ´miz-Gordo and Garcı ´a-Ortega, 2012). The accuracy of Laborde’s interior perspectives has recently been analyzed (Ga ´miz-Gordo et al., 2022) plus that portrayed in Girault de Prangey’s important views in 1839 (Ga ´miz-Gordo et al., 2021). There is agreement that the first plans of the monument were promoted by the Royal Academy of Fine Arts of San Fernando and published in Las Antigu ¨edades A ´rabes de Espan ˜ain 1787 and 1804 (Almagro-Gorbea, 2015). This work includes a Mosque-Cathedral plan drawn by Juan Pedro Arnal, which has been considered the first scientific plan of the monument: “.Scientific criteria were followed, using geometry as the foundation of the graphic discipline to understand the composition and distribution of architecture, drawn with rigor and sensitivity .”(Ga ´miz-Gordo, 2019, pp. 149e152). In addition, the plans published by the Royal Academy of Fine Arts of San Fernando in an important collection of plates entitled Los Monumentos Arquitecto ´nicos de Espan ˜a (1852e1881) are of great interest (Almagro-Gorbea, 2015). Among them is included another plant of the MosqueCathedral of Co ´rdoba drawn by Mariano Lo ´pez Sa ´nchez in 1868. To understand how precise drawings were obtained, at an epoch when current computer technologies did not exist, you can consult various bibliographies on architectural drawing in the Spanish XVIII century treatises (Irisarri-Martı ´nez and Castan ˜o-Perea, 2014;Garcı ´aMorales, 1989). In the craft guilds related to architecture, the use of drawing would be common, there were manuals about measurement and also publications on geometry in the military engineers books (Gentil-Baldrich, 2021). It must be taken into account that the Royal Academy of Fine Arts of San Fernando de Madrid, established in 1752, would try to “promote and disseminate scientific knowledge and promote the arts and aesthetic”(AlmagroGorbea, 2015, p.13). In addition, the Royal Academy was involved in a new emerging scientific mentality at that time: “.concern for incorporating technical graphic information is, without doubt, a feature that characterizes the nascent scientific spirit that impregnates the works of this century and distinguishes them from previous ones .”(Almagro-Gorbea, 2015, p.14). The mentioned historic graphic documentation provides an important source for researching the architectural heritage, “.of great importance for studying later restoration works .”(Ga ´miz-Gordo, 2019, p.174). Its analysis, in relation to other documentary sources, offers new directions for future heritage research. For that reason, the surveys promoted by the Royal Academy have been subject of an exhibition with catalog (Almagro-Gorbea, 2015). However, there is little research on the documentary reliability of these drawings. Almagro-Gorbea compared the 1767 plan with current planimetry and concluded it to be “.a fairly accurate drawing in its general features, both metrically and in the interpretation of the structure J. Cantizani-Oliva, J.-F. Reinoso-Gordo and A. Ga ´miz-Gordo 588 and spatial organization .”(Almagro-Gorbea, 2015,p. 24), stating that “.This plan constitutes an exceptional document on the state of the cathedral, former mosque, of Co ´rdoba in 1767 .with a very correct and accurate planimetric representation of the monument .”(AlmagroGorbea, 2015, pp. 272e273). Until the present research, however, the level of accuracy had not been quantified. The plan of 1868 has also been analyzed, though not its metric accuracy. Another matter of great interest to understand the abstraction degree of the drawn architectural reality are the elements appearing in the legend or labeled on the plan; e.g. in the 1767 plan, the shapes of altarpieces and altars were replaced by “.small crosses identified by a number that refers to the legend .”(Almagro-Gorbea, 2015,pp. 272e273). The representation scale must be taken into account in order to assess the precision and documentary interest. 1.3. Research objective: graphic and dimensional analysis The objective of this research is to analyze and quantify the graphic and dimensional accuracy of the main architectural elements drawn in the 1741 oil painting in comparison to their current state, considering as a reference two other plans commissioned in 1767 and 1868 by the Royal Academy of Fine Arts of San Fernando in Madrid. An additional objective is identifying and transcribing the labels and references included in the legend or in the drawing itself, taking into account the graphic symbols used to represent its abundant details, thus assessing its reliability and documentary interest for further research on this unique edifice and heritage. 2. Materials and method. Graphic representation and current state 2.1. Brief description of the 1741 oil painting and the plans of 1767 and 1868 This anonymous oil painting is the first known plan representation of the monument. It is kept in the Archive of the Mosque-Cathedral of Co ´rdoba and it is dated 1741, according to the legend itself. The legend also states that it was commissioned by the Bishop of Co ´rdoba Pedro Salazar Go ´ngora, who held this position between 1738 and 1742. The painting measures 1030 mm 1650 mm and includes a plan of the monument on the right, an elevation of the bell tower on the left, a long legend, and some flower garlands as decoration. It displays two graphic scales, both in varas: one for the floor plan and one for the tower, using a careful symbology discussed later. Its initial layout was probably made with barely perceptible auxiliary lines and arcs, on which the oil was superimposed. To assess its accuracy, this research also analyzes two other important original plans published in 1767 and 1868. Both originals are kept in the archives of the Royal Academy of Fine Arts of San Fernando [sig. MA-0536; MA-0163], whose website contains digital reproductions of excellent quality. The first one was drawn by Juan Pedro Arnal in black ink and grey wash drawing on light-yellow laid paper measuring 523 mm 726 mm and includes graphic scale in pies castellanos measurement unit. The second plan was drawn by Mariano Lo ´pezSa ´nchez with black ink and grey, ochre, blue and green wash on paper, measuring 469 mm 616 mm, and its graphic scale is indicated in meters. Below, all three have been reproduced with their Fig. 1 a) The Mosque-Cathedral of Co ´rdoba and its urban environment (created by Google Earth). b) Bell tower (photography by Wikipedia). Frontiers of Architectural Research 12 (2023) 587e600 589 actual size on the same scale, to clearly appreciate the larger size of the 1741 oil painting (Fig. 2). 2.2. Graphic analysis methodology In recent years, many research works have focused on dimensional accuracy of historical maps, especially in the field of territory and urban cartography (Aguilar-Camacho, 2017;Algarı ´n-Ve ´lez, 1998;De-Cea-Garcı ´a, 2017;Go ´mezBlanco-Pontes et al., 2019;Ortega-Vidal, 2000). A frequently used methodology consists in the graphic overlay of a precise digitization of the historical plan and a representation of its current state. Sometimes, a mesh stressing the main deformations has been derived from that superimposition. In other cases, the positional accuracy between homologous points has been assessed using numerical indexes which indicate the average or global error. Finally, some recent research works have applied MapAnalyst software to assess the accuracy of historical maps. The reliability assessment of historical drawings of buildings is often based on overlapping accurate current drawings. In the case of the Mosque-Cathedral of Co ´rdoba, other previous works already mentioned have overlayed digital surveys with the plan of 1767 (Almagro-Gorbea, 2015), and even with interior perspectives (Ga ´miz-Gordo et al. 2021,2022). To assess the 1741 oil painting accuracy, this research has used very precise current planimetry and compared the two substantial plans dating from 1767 and 1868, as additional references. First of all, a photogrammetric survey of the 1741 oil painting has been made. Afterwards, a geometric model of the current state of the monument was built from the point cloud obtained by a 3D laser scanner in four areas: the outer perimeter, the patio, the surroundings of mihrab and the interior elevation of the tower. Next, the validity of the graphic scales and measurement units used in the oil painting has been tested by means of dimensional checks of the represented elements. Then, some control points have been defined and their vector coordinates computed. The graphical analysis undertaken is based on the overlay and representation of the deformation based on the same origin of coordinates and Fig. 2 Three plans of the Mosque-Cathedral of Co ´rdoba reproduced at the same scale: a) oil painting from 1741 (Archive of the Cathedral of Co ´rdoba); b) plans from 1767 and 1868 (Royal Academy of Fine Arts of San Fernando in Madrid). J. Cantizani-Oliva, J.-F. Reinoso-Gordo and A. Ga ´miz-Gordo 590 rotation angle. The following variables have been quantified: outer perimeter area, inner courtyard area, displacements according to x and y, displacement vector (module and angle), as well as percentage error as a ratio between the module of the displacement vector and the length from origin, and the Root Mean Squared Error (RMSE). The expression ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn iZ1ðPiOiÞ2 n rallows the RMSE computation, a measure frequently used to assess cartographic accuracy (Aguilar-Camacho, 2017). It allows to obtain a single analytical index that quantifies the error between two data sets, using the same unit. In this way, the mapping from points location in the historical planes (Oi) to the current state ones (Pi) is analyzed, where nis the number of points considered in the study area. Finally, to offer a broader assessment of the information represented in the different plans, the labels included in the legend and plan area have been analyzed, also identifying the main symbols or graphic codes used in the architectural representation. 2.3. Photogrammetry process applied to the 1741 oil painting A photogrammetry survey of the 1741 oil painting has been performed to obtain an orthoimage on which to take precise measurements of the drawn elements. Considering the state of conservation of the painting and its current location in the Archive of the Mosque-Cathedral of Co ´rdoba, hanging on the wall at a certain height, it was advisable not to move it to avoid any possible deterioration. For this reason, photographs were taken at a short distance using a camera pole that reached a height of up to 4 m (Fig. 3a), with a combination of natural and artificial light (mini spotlights with integrated battery). The camera used was a Sony Alpha ILCE-7K Full Frame and 24Mpixels. The software used to get the painting orthoimage was Metashape and the needed workflow to produce the point cloud, mesh and texture was as follow. - Alignment. At this stage the parameters values were accuracy (high), key point limit (40,000) and tie point limit (4,000). - Dense cloud building. The setting for this step was as follow: quality (high) and depth filtering (mild). - Mesh building. The parameters configurations got as follow: source data (dense cloud), surface type (arbitrary (3D)), face count (high), interpolation (enabled). - Texture building parameters: type (diffuse map), source data (images), mapping mode (keep uv), blending mode (mosaic), texture size/count (8192). Fig. 3 a) Camera mounted on a pole 4 m high to take the 130 photographs. The painting is hanged on the wall. b) Photographs relative position reconstruction taking the painting as the reference. c) Scaling and setting coordinate reference system for the orthoimage projection plane. Authors’ own elaboration. Frontiers of Architectural Research 12 (2023) 587e600 591 The locations of the photographic shots were arranged in horizontal alignments, you can see their relative positions in (Fig. 3b) (blue rectangles). Before producing the true orthoimage, two task should be carried out. - Scaling the model: A metric reference was used to scale the model between the point 4 and point 5 markers (red rectangles in Fig. 3c). Points 4 and 5 have been marked on the 3D model separated each other 1.5 m as it is shown on the scale bar in Fig. 3c. The exact dimension 1.5 m has been achieved pasting a metric tape under the painting (yellow tape in Fig. 3c) that is divided in millimetric marks. - Setting the coordinate reference system (CRS) for the projection plane. The CRS was established in 3 corners over the painting frame 3D model (points 1, 2 and 3 inside the green rectangle in Fig. 3c). The axis 1-2 (xaxis) and 1-3 (yaxis) are perpendicular and make up the plane for projecting the 3D model that produce the orthoimage shown in Fig. 3c. Fig. 4 a) Scan stations location around the Mosque perimeter. b) Detail of the traverse closing carried out between the 4th and 65th scan station. c) Mihrab plant point cloud view. d) Mihrab perspective point cloud view. e) Point cloud perspective outside the monument. Authors’ own elaboration. J. Cantizani-Oliva, J.-F. Reinoso-Gordo and A. Ga ´miz-Gordo 592 130 photographs were used to generate the digital mosaic of the final orthoimage (Fig. 2a) for a visible canvas of dimensions 1030 mm 1650 mm. The orthophoto pixel size was 50 mm. This value was selected after several tests in which a smaller pixel size produced a low-quality image, and a larger value did not improve the visual resolution of the image obtained. 2.4. Current state survey obtained from 3D laser scanner and point cloud A 3D laser scanner has been used to get an accurate survey of the exterior walls perimeter of the Mosque-Cathedral, the perimeter of the patio, the bell tower, and the Mihrab nave. The first scan was registered to a previous one that had been leveled with a Leica C10 scanner. The monument perimeter was surveyed using 65 new scans performed with a Leica BLK360 scanner and processed with the Cyclone 360 software. The BLK360 has three different density options to capture data (high, medium and low). The selected option in this study was the medium one which take points separated by 5 cm if the object was a plane placed 10 m away. The scan station (numbered 1 in Fig. 4a and b) was performed by C10 scanner and is the starting point for the traverse that fit the 65 scans used in the perimeter survey of the Mosque (Fig. 4a). In order to get a suitable accuracy, the traverse was closed by the 65th scan on the 4th scan getting a 4 mm fitting error (Fig. 4b). The traverse scan stations fitting guarantees that the accumulated errors are minimized (the error is 5 mm, both the whole and the cloud-to-cloud measurement). The registration for every scan station was carried out one by one inside Cyclone Register 360 software and it was used a cloud to cloud methodology. The registration is performed translating and rotating manually the one to be registered until it approximately be coincident to the reference and then order the software to optimize the registration. After the optimization task, an error distribution is shown by the software and the user has the option to accept or reject the result. The Mihrab survey is in the same coordinate system that the exterior walls, but it was processed by the Recap software. Fig. 4c and d represent the point cloud plant and a perspective view respectively for the Mihrab central nave. The spheres shown in Fig. 4d are the scan stations locations. The geometric model of the current state of the monumentdon the outer perimeter of the courtyard, Mihrab surroundings, and interior elevation of the bell towerdhas been elaborated by importing as reference the point cloud to CAD. Next, some sections have been performed on the point cloud, which have been then orthogonally projected to obtain the corresponding orthoimages. To define the geometry, certain commands have been used to obtain lines and edges, as well as planes and intersections. Each of the elements has been defined according to the degree of precision and definition of the model analyzed in each scale used. Fig. 4e shows a perspective for the whole point cloud of the capture data. 3. Results and discussion: dimensional analysis, legend and symbols 3.1. Quantification of scales and variables In order to know the equivalence of the measurement units used in the 1741 oil painting, it must be considered that the perimeter of the monument’s floor plan measures 175.34 m 128.52 m, according to our own survey carried out with a laser scanner. The legend included in the oil painting indicates that the floor plan measures 207.5 varas 152 varas, therefore it is deduced that 1 vara is equivalent to 0.84 m. On the other hand, the tower measures 11.71 m wide 59.65 m high and according to legend it measures 14 varas 70.5 varas, that is, 0.83 m in width and 0.84 m in height. These values are quite similar to 0.835905 m, which is the equivalence of 1 vara according to the Geographic and Statistical Institute of Spain (Direccio ´n, 1886;EscalonaMolina, 2009). Thus, considering that 1 vara is equal to 0.836 m and 1 pie to 0.279 m, the three planes scale is revealed. The dimensions of each plane along with their measurement units allow to obtain the numerical scale, computing it as a ratio between the distance on paper (d) and the real distance (D); the length of the graphic scale itself has been considered (Fig. 5). Fig. 5 Scales analysis. Authors’ own elaboration. Frontiers of Architectural Research 12 (2023) 587e600 593 Table 1 Control points and variables quantification. Authors’ own production. SURFACES 3D GEOMETRIC MODEL OIL PAINTING FROM 1741 PLAN OF 1767 PLAN OF 1868 Perimeter surface 21983.508 m 2 21462.67 (2.37%) 22401.48 (1.90%) 22180.44 (0.90%) Courtyard surface 5781.685 m 2 5439.53 (5.92%) 5911.87 (2.25%) 5597.98 (3.18%) VARIABLES XYVector modulus Error X Error Y Error vector modulus Error vector angle Error Error X Error Y Error vector modulus Error vector angle Error Error X Error Y Error vector modulus Error vector angle Error PLAN P1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00% 0.00 0.00 0.00 0.00 0.00% 0.00 0.00 0.00 0.00 0.00% P2 175.35 0.84 175.35 2.07 0.01 2.07 0.28 1.18% 0.28 0.00 0.28 179.59 0.16% 1.06 0.01 1.06 179.73 0.61% P3 174.43 127.68 216.17 2.67 1.48 3.05 28.95 1.41% 0.93 0.75 1.19 140.96 0.55% 1.03 0.31 1.08 163.08 0.50% P4 0.96 124.41 124.41 0.56 0.43 0.71 142.37 0.57% 0.39 1.75 1.80 77.56 1.44% 0.85 1.09 1.38 127.82 1.11% P5 8.25 7.86 11.40 0.59 0.62 0.85 133.48 7.50% 0.40 0.18 0.44 24.35 3.85% 0.14 0.72 0.73 100.83 6.40% P6 59.82 7.53 60.29 0.55 0.26 0.61 25.72 1.01% 0.03 0.17 0.17 100.28 0.29% 1.13 0.17 1.15 8.48 1.90% P7 58.91 117.47 131.41 0.11 2.43 2.43 92.53 1.85% 0.07 1.93 1.93 92.11 1.47% 0.39 0.23 0.45 30.86 0.35% P8 6.67 116.62 116.81 1.53 2.25 2.71 124.18 2.32% 0.05 1.25 1.25 92.15 1.07% 0.99 1.53 1.82 122.79 1.56% RMSE (perimeter) 1.71 0.77 1.88 0.52 0.95 1.09 0.85 0.57 1.02 RMSE (courtyard) 0.86 1.69 1.89 0.21 1.16 1.18 0.78 0.86 1.16 NAVE OF MIHRAB M1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00% 0.00 0.00 0.00 0.00 0.00% 0.00 0.00 0.00 0.00 0.00% M2 3.16 0.00 3.16 0.36 0.03 0.36 4.76 11.42% 0.09 0.02 0.09 13.28 2.89% 0.18 0.02 0.18 172.87 5.61% M3 6.32 0.01 6.32 0.45 0.07 0.45 9.05 7.14% 0.01 0.02 0.02 67.17 0.33% 0.00 0.07 0.07 89.12 1.03% M4 9.45 0.01 9.45 0.48 0.10 0.49 11.49 5.15% 0.15 0.07 0.16 25.14 1.72% 0.04 0.08 0.09 116.28 0.96% M5 12.63 0.00 12.63 0.53 0.05 0.54 5.56 4.25% 0.16 0.04 0.16 14.12 1.30% 0.04 0.06 0.07 54.94 0.55% M6 15.76 0.00 15.76 0.63 0.06 0.64 5.78 4.03% 0.18 0.01 0.18 4.11 1.15% 0.01 0.04 0.04 100.54 0.28% M7 18.91 0.03 18.91 0.54 0.02 0.54 1.90 2.87% 0.08 0.11 0.14 125.63 0.74% 0.09 0.09 0.13 43.75 0.69% M8 19.32 0.02 19.32 0.34 0.03 0.34 4.42 1.74% 0.04 0.10 0.11 69.00 0.55% 0.21 0.10 0.23 24.59 1.21% M9 22.21 0.10 22.21 0.38 0.02 0.38 3.01 1.71% 0.00 0.15 0.15 89.21 0.65% 0.11 0.20 0.22 60.92 1.01% M10 25.19 0.00 25.19 0.57 0.00 0.57 0.00 2.27% 0.21 0.00 0.21 0.00 0.84% 0.05 0.00 0.05 0.00 0.20% M11 18.86 2.58 19.04 0.02 0.25 0.25 95.59 1.29% M12 18.88 5.20 19.58 0.10 0.20 0.23 117.24 1.16% M13 0.01 7.81 7.81 0.03 0.55 0.55 93.21 7.08% 0.09 0.22 0.24 67.75 3.04% 0.02 0.08 0.08 104.36 1.08% M14 3.16 7.79 8.41 0.06 0.57 0.57 84.24 6.75% 0.08 0.23 0.24 71.49 2.88% 0.05 0.10 0.11 116.11 1.35% M15 6.33 7.78 10.03 0.21 0.55 0.59 69.57 5.88% 0.03 0.22 0.22 81.54 2.24% 0.03 0.05 0.06 57.68 0.58% M16 9.48 7.79 12.27 0.32 0.64 0.71 63.15 5.81% 0.19 0.18 0.26 43.92 2.15% 0.05 0.04 0.07 41.99 0.55% M17 12.65 7.80 14.86 0.45 0.62 0.77 53.98 5.18% 0.12 0.14 0.19 50.19 1.26% 0.12 0.02 0.12 11.40 0.82% M18 15.78 7.81 17.61 0.52 0.66 0.83 51.87 4.74% 0.23 0.12 0.26 28.43 1.48% 0.11 0.06 0.12 31.36 0.70% M19 18.92 7.82 20.47 0.47 0.73 0.87 56.93 4.23% 0.13 0.09 0.16 146.02 0.80% 0.09 0.01 0.09 5.77 0.44% M20 19.34 7.82 20.87 0.04 0.09 0.10 113.85 0.50% 0.22 0.05 0.22 12.86 1.08% M21 22.26 7.82 23.59 0.59 0.89 1.07 56.27 4.54% 0.07 0.02 0.07 161.05 0.30% 0.29 0.08 0.30 14.40 1.28% M22 25.17 7.81 26.36 0.56 0.96 1.11 59.84 4.22% 0.20 0.10 0.22 26.79 0.85% 0.00 0.08 0.08 87.95 0.32% RMSE 0.44 0.48 0.65 0.12 0.13 0.18 0.12 0.08 0.14 J. Cantizani-Oliva, J.-F. Reinoso-Gordo and A. Ga ´miz-Gordo 594 Table 1 includes the 44 control points and some variables and statistics: 8 at the corners of the building perimeter and at the corners of the patio, 22 at the axes of the columns in the Mihrab nave, and 14 at the tower. It should be noted that two columns were not drawn on the plans of 1741 and 1868; and that the tower is not represented in the plans of 1767 and 1868. 3.2. The outer perimeter of the floor plan and the patio After defining 4 control points at the corners of the outer perimeter of the monument and another 4 at the corners of the patio, the survey itself has been overlayed to the 1741 oil painting and both plans of 1767 and 1868, matching the vertex lower left and the direction of the lower facade (Fig. 6). All this has been scaled in meters, considering that 1 vara is equivalent to 0.836 m, 1 vara to 3 pies and 1 pie to 0.279 m (Direccio ´n, 1886;Escalona-Molina, 2009). The RMSE of the module of the displacement vector in the outer perimeter shows a deviation respect to the scanned survey of 1.88 m in the 1741 oil painting, 1.09 m in the 1767 plan, and 1.02 m in the 1868 plan. The same occurs in the perimeter of the patio, although the differences are smaller (1.89 m, 1.18 m, and 1.16 m, respectively). From the analysis of the variable surface enclosed by the outer perimeter, the greatest difference respect to the survey corresponds to the oil painting of 1741 (2.37%), and the minimum to the plan of 1868 (0.90%). The same occurs with the surface of the patio, corresponding the largest deviation to the oil painting of 1741 (5.92%) and the smallest to the plan of 1767 (2.25%). All this allows to conclude that the 1741 oil painting is somewhat less exact, considering the surface, than the 1767 and 1868 plans. 3.3. The Mihrab central nave The survey of the Mihrab nave has been overlayed to the vectorial transcription of the oil painting of 1741 and the plans from 1767 and 1868 (Fig. 7). 22 control points have been selected, corresponding to the columns axes of the main nave, matching the center of column M1 and the direction that joins the center of columns M1eM10. It must be noticed that columns M11 and M12 are not represented neither in the 1741 oil painting nor in the 1868 plan. The RMSE variable of the displacement vector module has been quantified. The maximum displacements in the oil painting from 1741 correspond to the M22 column (1.11 m), in the 1767 plan to the M11 (0.25 m), and in the 1868 plan to the M21 (0.30 m). The accuracy of the 1767 and 1868 plans (RMSE equal to 0.18 m and 0.14 m, respectively) is greater than the oil from 1741 (0.65 m). Therefore, the accuracy of the 1741 oil painting in the central nave of the Mihrab is slightly inferior than that of the outer perimeter of the monument and its patio. 3.4. The bell tower elevation The survey from the scanner has also been overlayed to the interior elevation of the tower represented in the 1741 oil painting, making the lower left end coincide with the BELL TOWER T1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00% T2 11.72 0.00 11.72 0.13 0.00 0.13 0.00 1.07% T3 17.09 0.00 17.09 0.09 0.00 0.09 0.00 0.55% T4 17.09 15.95 23.38 0.09 0.05 0.11 152.95 0.45% T5 0.45 21.68 31.68 0.35 0.56 0.66 57.62 2.07% T6 11.42 21.70 24.52 0.25 0.53 0.59 115.13 2.40% T7 1.67 31.52 31.56 0.26 0.74 0.78 70.89 2.47% T8 10.08 31.58 33.15 0.05 0.71 0.72 94.33 2.16% T9 1.86 38.60 38.64 0.25 0.68 0.73 110.41 1.88% T10 9.77 38.64 39.85 0.31 0.73 0.79 66.76 1.99% T11 3.56 47.23 47.37 0.03 0.22 0.22 98.02 0.47% T12 7.97 47.23 47.90 0.38 0.12 0.40 16.95 0.84% T13 5.68 54.62 54.91 0.16 0.17 0.23 47.12 0.42% T14 5.42 59.65 59.90 0.45 0.49 0.67 47.67 1.12% RMSE 0.24 0.46 0.52 Frontiers of Architectural Research 12 (2023) 587e600 595