**4. Discussion**

Provenance of the numerous marble varieties that have been used in archaeological sites worldwide still remains one of the most debated topics [11]. This matter is of critical importance due to the need of determining a suitable source that can supply the optimum restoration material, in terms of physicochemical, mechanical and historical compatibility [2,3,12]. Furthermore, it can also reveal crucial information regarding the history of a monument and shed light on transportation routes and commercial practices. In this section, the results of the petrographic and isotopic characterization of the examined samples will be used to establish the provenance of the Holy Aedicule marbles and a discussion will be made in relation to possible intra-site discriminations and in relation to historical testimonies and evidence.

## *4.1. Establishing the Provenance for the Holy Aedicule Marbles*

Based on the petrographic study results, all six samples display critical similarities—textural and mineralogical—that indicate a common origin.

The studied marble samples are characterized by explicit textural similarities: all six samples display a strong heteroblastic fabric, which is typical for marbles that have suffered a certain degree of deformation during (or after) peak of metamorphism. This fabric comprises xenomorphic calcite crystals that commonly form two distinct generations: the first refers to calcite porphyroblasts, which are scattered in a matrix of fine-grained calcite neoblasts, forming the second generation. This deformation-inferred fabric is not observed in marbles that have been formed under equilibrium conditions with no subsequent deformation, thus exhibiting isodiametric grains comprising the so-called homeoblastic fabric (e.g., the marbles from Carrara).

The studied samples exhibit a strong heteroblastic fabric, commonly described as "mortar-type", which is characterizing marbles originating from Proconnesos and is a strong vector towards provenance determination (Antonelli and Lazzarini 2015 [12] and references therein). Beyond the "mortar-type" fabric, all the studied samples carry more evidence of deformation, like the polysynthetic twinning, often with bent twinning lines, in the calcite grains. In addition, two of the samples, namely OM10 and OM49, exhibit another feature: they are characterized by grey-colored bands that can be observed with naked-eye as well. This feature gives the rock a relatively foliated texture, which is common to many marbles (e.g., those from Pentelikon mountain, Proconnesos etc. [12]). In the samples under investigation, these gray-colored bands are composed by preferentially oriented calcite grains and minor presence of accessory minerals, mainly micas (phlogopite and/or muscovite). Despite the strong deformation features that characterize the studied marble samples, local equilibrium conditions were also identified, as evidenced by the rare presence of straight-sided grains that conjunct into triple points.

The maximum grain size is a parameter that could also be used to identify the provenance of a marble sample, because along with textural features, it is related to the tectonometamorphic conditions of the rock. In terms of archaeometry, marbles are distinguished in three categories: fine-grained (MGS up to 2 mm), average-medium grained (MGS 2–5 mm) and coarse-grained (MGS >5 mm). By using solely this parameter, and according to the coarse categorization proposed by Antonelli and Lazzarini 2015 [12], the studied samples, can be characterized as medium-grained, since they exhibit MGS that range mostly around 2 mm. This category consists of marbles from Proconnesos, Aphrodisias (Aydin) and some Parian and Thassian varieties (Pa2,3,4 and T3 respectively). This parameter allows the exclusion of marble quarries that produce both fine- and coarse-grained marbles. Based on this ascertainment, marble producing areas that should not be considered as raw material sources for the Holy Aedicule marbles are: Penteli and Hymmetos, Paros 1/Lychnites, Carrara and Afyon (Dokimeion) which produce fine-grained varieties and Thasos 1-2 and Naxos, which produce very coarse-grained marble varieties. MGS identified in the present study, range from 1.6 to 2.3 mm, gathered mostly around 2 mm, and fit well in the published MGS range for all the medium-grained marble varieties (Proconnesos, Aphrodisias, Paros 2,3,4 and Thasos 3), thus suggesting that by defining this parameter only, no safe conclusions regarding provenance can be made.

In parallel, accessory minerals have been also identified as possible key indicators useful in provenance determination [11]. The six studied samples contain the same accessory minerals paragenesis, which consist of (in decreasing volumetrically order) dolomite, mica (phlogopite, muscovite), apatite and pyrite. Mica group minerals, namely phlogopite and muscovite-paragonite-margarite (white micas, which are not easily distinguishable by optical microscopy only) are common accessory phases in many marbles (e.g., Proconnesos, Thassos, Paros, Naxos etc. [11]). The presence of apatite is also common in many marbles used in antiquity (e.g., Proconnesos, Carrara, Naxos, Thassos). Dolomite is present in marbles from a great number of different localities and so is pyrite. This fact precludes their use as provenance indicators.

By composing the above-mentioned data with the isotopic signature of the studied samples, their provenance should be searched between the Proconnesos, Thasos and Paros-3 marble quarries (see Figure 17), suggesting that a multi-parametric comparison of the studied samples to literature data for the above-mentioned localities is necessary.

Regarding the Thassian marble, despite its textural similarity to the studied samples, its heteroblastic texture is characterized as "mosaic-type" and not "mortar-type", as evident in the studied samples. Furthermore, the predominant grain boundary shape in Thassos marbles is the curved type, but in the samples from the Holy Aedicule, sutured and rarely embayed grain boundaries were the only identified types. The isotopic signature of the studied samples only partly (two out of six samples) overlaps the field of Thassos-3 (Vathy), indicating low isotopic relations. Finally, serpentine, which is common accessory mineral in the Thassian marbles (especially the dolomite-rich varieties), was not observed in any of the studied samples, and along with the previously-mentioned differences, Thassos should also be excluded as a possible source for the studied samples.

The Parian variety 3 (and 2), originating from the Chorodaki quarries, displays significant isotopic similarities to the studied samples, as four out of six, plot inside the relative subfield. Its MGS is also comparable to the observed MGS values. Textural and mineralogical variations though, are substantial to exclude this variety as well: the Parian marbles are characterized by mixed homeoblastic and heteroblastic, mosaic/lineated fabric, composed of curved and embayed crystals, and comprises accessory amphiboles, rutile, zircon ± serpentine [11,12]. Beyond scarce grey-colored bands in the studied samples, which could be considered as a common feature of the studied samples to the Parian marble, the absence of accessory phases like amphibole, zircon and serpentine also enhance another sourcing locality rather than Paros.

Although a partial overlapping of three out of the six studied samples is noticed with the Carrara subfield in the isotopic diagram (Figure 17), this provenance has to be excluded. Carrara marbles present significantly lower MGS values and they are characterized as a fine-grained variety in contrast to the Holy Aedicule samples, which are classified as medium to coarse-grained marbles. Furthermore, no evidence of plagioclase crystals was found in the Holy Aedicule samples, which is however, a characteristic accessory phase for the Carrara locality. In addition, their fabric is not homeoblastic-polygonal, as described for the Carrara marbles [12,18].

The final variety to be examined is the Proconnesian marble. In the isotopic diagram (Figure 17), it is clear that all the studied samples plot in a very narrow field in the center of the Proconnesos-1 subfield, indicating a critically similar isotopic signature between the Holy Aedicule samples and the published data on Proconnesos marbles ([15,16]). The observed "mortar-type" heteroblastic fabric is another significant similarity that should not be neglected and points toward a Proconnesian provenance for the studied samples. This is also enhanced by the remarked accessory mineralogical components, (phlogopite ± other micas, e.g., muscovite+apatite+minor dolomite+pyrite). Furthermore, in terms of MGS values, which range from 1.6–2.3 mm (mostly around 2 mm), the observed values fit well in the field of the published data.

Taking into consideration all aforementioned data, and having excluded other marble-producing localities with similar isotopic signature, it is strongly suggested that the examined marbles of the Holy Aedicule originate from the Proconnesos island quarries.

#### *4.2. Proconnesos Quarries: Intra-Site Discriminations*

The Proconnesian marble is one of the most popular white marbles used in antiquity [16], attracting the interest of the scientific community in relation to provenance studies in archaeometry. Its use flourished during the Roman era, with numerous buildings and art materials (sarcophagi, sculptures etc.) made out of Proconnesos marble. During the second and third centuries AD, the use of Proconnesian marble has been documented throughout the whole Imperial territory [16,71–73]. This continued to the Byzantine times, since the close proximity of the quarries to the capital city of Constantinople was an additional advantage.

More than twenty-three ancient and contemporary quarrying locations have been identified so far in the northern part of Proconnesos (Marmara Island) [16]. In many of them, remnants of sculpted and or half-worked items are present (e.g. columns). The abundance of quarries and the widespread use of their marbles lead to the publication of detailed works, defining mineralogical and isotopic signatures of these quarries, in an attempt to define any possible intra-site discriminating features [15,16,18].

Based on isotopic characterization, Asgari and Mathews (1995) [15] already distinguished two possible Proconnesos marble varieties: The Proconnesos-1 and the Proconnesos-2, with the later exhibiting highly negative δ18O values. This discrimination was later confirmed by Gornoni et al. 2002 [18] and Attanasio et al. 2008 [16], who extended knowledge on the Proconnesian marble properties. Attanasio et al. 2008 [16] proposed the use of an isotopic threshold (δ18O = <sup>−</sup>5.00), as a discriminating factor of the two varieties. As shown in Figure 18, the Holy Aedicule samples are characterized by relatively low-negative <sup>δ</sup>18O values, ranging from <sup>−</sup>2.49 to <sup>−</sup>1.13, thus revealing an isotopic affinity to the Proconnesos-1 variety.

**Figure 18.** Isotopic discrimination plot between the so-called varieties Proconnesos-1 and Proconnesos-2. Fields for Proconnesos-1 and Proconnesos-2 varieties are adopted from Attanasio et al., 2008 [16].

Even though the isotopic values for the majority of the quarries display significant overlaps, a fact that allows limited chances for safe intra-site topographical discriminations, an effort was made to couple selected values of Proconnesos quarries with the respective values displayed by the marble samples of the Holy Aedicule.

In particular, for each sample of the Holy Aedicule, comparison was made with the published MGS values and isotopic values (δ18O and δ13C), as given by Attanasio et al. 2008 [16], for the quarries that the later examined, excluding the modern-only quarries and the open-air museum artifacts.

In Table 5, the comparison of isotopic and MGS values between the Holy Aedicule samples and the literature data for quarrying locations on Proconnesos island [16] is presented. Highlighted boxes indicate pairing of the isotopic values of the Holy Aedicule samples (both δ18O and δ13C) with the published range of values for a certain quarry, while (√) stands for pairing of the MGS values as well. Obviously, when both aforementioned criteria are satisfied, the probability of a sample deriving from that specific quarry is increased.

**Table 5.** Comparison of isotopic and MGS values between the Holy Aedicule samples and literature data for quarrying locations on Proconnesos island (Attanasio et al. 2008 [16]). Highlighted boxes indicate pairing of the isotopic values of a Holy Aedicule sample (both δ18O and δ13C) to the published range of values for a certain quarry; (√) stands for respective pairing regarding the MGS values.


Based on this comparison, seven quarries, namely the Filiz, Saraylar, C5, C6b, C7, C8 and C16, demonstrate values which exclude them as sources of raw material for the examined samples of the Holy Aedicule.

According to the set criteria, sample OM13, which corresponds to the internal facing opposite the Holy Tomb, could have derived only for the C5b quarry. Sample OM49, which corresponds to the marble fragment found inside the Holy Tomb, could have originated only from quarries C1 and C5b. Although intra-site discrimination can be considered successful for the aforementioned samples, this is not the case for samples OM10 and OM11, which correspond to the lower and the upper plate of the Holy Tomb respectively. OM 10 could have originated from quarries Altintas, C2, C3, C12, C14 and C5b; the latter being the only common quarry with samples OM13 and OM49. OM11 is correlated with quarries OC13, Mandira, C1, C2, C4, C5t, C6, C7i, C11, C13, C14, C15, making any comparison with other samples difficult.

Samples OM50 and OM51 were taken from the internal facings of the Chapel of the Angel, right and left of the entrance to the Holy Tomb Chamber, respectively. OM50 could have been quarried from Altintas, Harmantas, Mandira and C6, while OM51 could have originated only from Altintas and Mandira quarries.
