Archaeometrical Identification of the Iberian Amphorae Productions at the Area of the Ancient Laietania (Now Maresme): Some Considerations on Production and Distribution Patterns

In the present contribution an integrated archaeological and geochemical study of Iberian amphorae found at the indigenous settlement of various Laietanian sites are presented. Two of the analysed sites are production sites and three are consumption sites. Two chemical groups were identified. The first is the reference group of the kiln site of Riera de Sant Simó and the second is the Paste Compositional Reference Group of the Iberian village (consumption site) Puig Castellar. The final results proved, on one hand, a diversification of the productions on the various analysed sites and, on the other, some wider commercial contacts. The chemical complexity reveals the existence of a multiplicity of productions in each geographical area and clearly indicates a commercial interaction between the Laietania and the Cossetania, at least. Adding to those results the typological information too, a certain standardization of forms in each of those regions was observed.

Provenance studies applying mainly chemical (XRF) and mineralogical (XRD) characterization published to date comprise Iberian amphorae dated to between the 6th and 2nd centuries BCE selected from 22 archaeological sites ( Figure 3) ( Table 1)  Arch & Anthropol Open Acc that has been also archeologically associated to the closely located kiln dump of Vista Alegre (Iberian Laietania, now Martaró) ( Figure  3). VA corresponds to the production of the kiln of Hortes de Cal Pons (northern Cossetania, now Villafranca Del Penedés) ( Figure  3) and finally Camp A and B are two variants of Campello's kilns (Contestania, now Alicante; Illeta de Banyets) ( Figure 3). As for the PCRU's, some of them can be related to specific regions according not only to the typological and archaeological information but also to the geochemical compatibility of the groups with specific geological areas [18] ( Table 1). PC is related to the site of Puig Castellar (Iberian Laietania, now Mataró). PV is geochemically compatible with certain non-calcareous Ordovitic and Siluric slatic and quartzitic areas (IGME 1982-Llíria) of Edetania (now a days it would be the broader area of Llíria) [20]. VB seems to be a variant of the RGVA, and it can be therefore associated to northern Cossetania (now Villafranca de Penedés). PCRU'S A and B ( Figure  3) are two variants of the same production that geochemically fit the continental and/or transitional quaternary areas of Southern coastal area of the Cossetania (Figure 3) (now geographical area of Garraf). Finally, TB is composed by individuals sampled at different consumption sites (Castellet de Banyoles, Mas Catellarde Pontós and also Alorda Park) (Figure 3). It is a broader chemical cluster with some subgroups that exhibit minor chemical differences inbetween them and they cannot be related, at least at the present, to none of the studied areas. Even though the groups VA and VB are quite similar, certain differences were observed that indicate that each one corresponds to a different chemical group. Finally, previous works [8,[18][19][20][21] in consumption sites from the Balearic Islands, has help to define the so-called PCRUIB, a chemical group that so far contains only individuals from the Balearic Islands. Nevertheless, IB's chemical composition suggests that the raw materials are associated to specific geological formations from the area of Alicante.
To summarize, the results from these previous works [6, [15][16][17][18] integrating chemical and typological suggest that, each general type ( Figure 1) was produced by different workshops at regional scale. For instance, type 2B is represented by the Laietaian chemical groups LAI and PC, and type 2C by the Cossetanian chemical groups A, B, VA and VB (Table 1). This, clearly, indicates that even if there is certain typological standardization in wach one of the regions, there is no technological homogenity nor the same materials were used for production.

Methodology and Sampling and Short Description of the Sampled Archaeological Contexts
Methodology Table 1: Summery of the description of the up to now identified Iberian pottery productions in pervious works.

Rural Iberian
Town-Consumption site.
6-2 nd century BCE 2C 16 Colour: from red to reddish brown to greyish-black depends on the sample's firing conditions Hardness: very hardc Surface feel: smooth Fracturein fresh break: smooth Macroscopic description: very fine and vitrified texture covered with very frequent white translucent crystals (mainly seem quartz and feldspars), black crystals (micas: biotite) and frequent iron rich particles.  Iberian amphora 2B ( Figure 1) 22

Arch & Anthropol Open Acc
The studied samples from the area of the Laitenaia are listed in Table 2. The chemical composition of the samples was determined by XRF using a Philips PW2400 spectrometer with a Rhexcitation source. A 25g specimen (to ensure it was representative) from each shard was powered and homogenized in a tungsten carbide mill and dried at 105 ⁰C for 24h. The major and minor elements were determined by preparing duplicates of glassy pills in a Philips Perl'X-3 high induction furnace using 0.3g of the powdered specimen in alkaline fusion with 5.7g of lithium tetra borate at 1/20 dilution. Trace elements and Na 2 O were determined in powdered pills made from 5g of dried specimen mixed with 2ml of Elvacite 2044 agglutinant, placed over boric acid in an aluminum capsule and pressed for 60 sat 200kN. The concentrations were quantified using a calibration curve based on 60 International Geological Standards. A total of 29 compounds and elements were determined:

Description Of The Studied Laietanian Production Sites
The kiln of Santa Cecilia, Riera de Sant Simó (Mataró) La Riera de Sant Simó is located in the northeast of Mataró ( Figure 4 & 5: A), in the area of Santa Cecilia. In 1968, the archaeological prospecting was carried out and the subsequent excavations highlighted the existence of three Iberian kilns located three meters from each other (Figure 4). The Iberian chronology of the kilns was established according to the typology of pottery found at the site. The pottery was Iberian amphorae and common wares. The only fragment of imported amphora was Greco-italic that dates the kilns between the 3rd and 2nd centuries BCE [11]. In 1973 one of these kilns was excavated to study the structure ( Figure 5: A). Most of the material that is preserved at the Mataró Museum, today, corresponds to fragments of Iberian amphora on the Catalan coast, belonging to variant 2B ( Figure 4).

The kiln dump of Vista Alegre (Mataró)
The kiln dump excavated at the camino de Vista Alegre [14] that located southwest, next to the Iberian kiln of Riera de Sant Simó, archeologically seems to be associated to this kiln ( Figure 5: B). The dump is situated next road that goes from the road of Valldeix (Riera de Sant Simó) to the neighborhood of Vista Alegre. Between 1981 and 1991, several archaeological surveys were carried out. Finally, in 1991 the excavation was completed [14]. The sector 10000 two archaeological layers were located that provided building materials (bricks and tiles), fragments of mural painting, large amount of Iberian ceramics (handmade common wares, amphora and typical Catalan grey wares). One part was Roman pottery: Italic and African Sigillata. The other part (in UE10007: Figure 5: B) was Iberian amphorae located at sector 10000 that dates the Iberian layer to the 3 rd . century BCE. The amphorae's typology corresponds mainly to subtype 2B (Figure 4).

The site of Can Bartomeu (Vall de Cabrera de Mar)
The importance of the Cabrera de Mar valley in terms of occupation models during the Iberian era is well known [24]. Cabrera de Mar is located in the north of Barcelona ( Figure 5: C), next to Can Segarra and the old field called Can Bartomeuthatisan Iberian village is about 300 meters from the oppidum of Burriac. Two chronological phases were defined for this Iberian village. The moment of its construction that dates to the beginning of 2nd century BCE, and a moment of re modelling/reconstruction and/or restructuration of the site, which dates to the second half of the 2 nd century BCE. 30 silos (grain storage) surround the village ( Figure  5: C) that were chronologically grouped into two groups. The first group, which contains 15silos, seems to be quite homogeneous and dates between the second half of the 3 rd century BCE and the beginning of the 2 nd century BCE. The second group is formed by 13 silos that are quite heterogeneous. Two of them were used up to the middle of the 2 nd century BCE and eight of them were used Arch & Anthropol Open Acc up to the last quarter of the 2 nd century BCE and/or the beginning of the 1 st century BCE [22]. Among the Iberian ceramics recovered in this site there are a large number of Iberian amphorae, however they haven't been defined typologically, because of the fragmented state they were found.

The necropolis of turó dels dos pinos (Vall de Cabrera de Mar)
Another probable consumption site is the Iberian necropolis of the Turó dels Dos Pins [21] that is located in the municipal district of Cabrera de Mar, such as the Can Bartomeu. The necropolis ( Figure

Chemical resulta and identification of the Refence groups and Paste Compositional Refence Units for the Laietania
The summary of the information on the analysed material is given in Table 2. Whereas, in Table 3 the normalized chemical composition of the analysed individuals is presented (see also Tsantini 2007). Most of the studied sites are situated in a geological area which is characterized by the main presence of granitic and/or granodioritic volcanic sequences. The chemical composition of the raw materials of that kind, normally, is characterized by relatively high Ba and MnO concentrations. Table 3 reflects exactly that fact for most of the analysed individuals. This is a compositional indication that the raw materials would be local/regional for most of the analysed pottery. For the statistical treatment, some of the elements determined (P 2 O 5 , Pb, Mo, Co and W) were left out to avoid misinterpretation due to very low values and or possible contamination and/or alterations.
The purpose of chemical analysis in the framework of provenience studies is to compare the chemical composition of the analysed individuals trying to identify clusters. This is normally done by using statistical treatments. If the statistical treatment is not done cautiously, erroneous results can be obtained, due to various problems that the compositional data present [24,25].
To avoid these problems, the raw data set is transformed into logarithms (additive log ratios) following the considerations of Aichison and Buxeda.
One of the most important steps in a chemical study, since the objective is to compare the composition between individuals, is to calculate somehow the existing variation within the dataset. The simplest way to control this variability is the calculation of the compositional variation matrix (MVC) [26]. This matrix includes all the necessary information to measure the variability in compositional data, such as the total variation and the degree of variability introduced by each chemical element in the data set. The total variation (vt) quantifies the total variability and if it exceeds 0.3 indicates that the analysed material is chemically heterogenic or polygenic that means heterogeneous chemical composition and diversified origin and/or provenience. In other words, the existence of more than one production/group. The variation matrix is used to define the most variable chemical elements and also the less variable chemical element that used as divider in the logarithmic transformation, as well.
To examine chemical variability in our dataset, first, we calculated the Compositional Variation Matrix (CVM) (Table 4), on the following sub-composition: Fe 2 O 3 , TiO 2 , CaO, K 2 O, SiO 2 , Al 2 O 3 , MgO, Ba, Ce, Cr, Cu, Nb, Ni, Sr, V, Zn and Zr. The total variation (vt) is 0.9778 that suggests a chemical heterogeneity [27]. After a logarithmic transformation using SiO 2 as a divisor, a Cluster Analysis was performed on that same sub-composition. The results from this analysis, applying the squared Euclidian distance and the centroid agglomerative algorithm, is the cluster tree presented in Figure 6.
Several observations can be done, after these first results. First, AMI211 (indicated with a red circle in Figure 6) is a chemical outlier with very high Ba wt%. AMI200, AMI023 and AMI024 (indicated with a circle in Figure 6) present higher CaO percentage sand lower Na 2 O. At the same time, the chemical data indicates some similarities between AMI023 and AMI024, but not between these two and AMI200. AMI032, AMI046, AMI047, AMI173, AMI175, and AMI177 are separated from the rest of the individuals, in a lower ultra metric distance. The similarities that these individuals present (border to low CaO; relatively high Al 2 O 3 , Fe 2 O 3 and Na 2 O) compared to the rest, grouped in the blue rectangle, in Figure 6, indicate similar geochemical origin (raw materials might come from the similar deposits). However, they don´t belong to the same production. After this first approximation, therefore: AMI023 AMI024, AMI032, AMI46, AMI47, AMI173, AMI175, AMI177, AMI200 and AMI211 cannot be well classified into a specific group since they present important differences from all the analysed material, coming from the Laietania. From these individuals: AMI023 AMI024 and AMI032 were sampled at the site Can Bartomeu. AMI046, AMI047, AMI173, AMI175 and AMI177 in the Iberian village of Puig Castellar and AMI211 and AMI200 came from the kiln dump of Vista Alegre, which is considered the kiln dump of the Laietànian workshop of Riera de Sant Simó (Mataró). Another individual VAC005, which is a common ware, is separated from the rest, and it was sampled in Vista Alegre.
We recalculated the CVM excluding of the above-mentioned chemical outliers (AMI023, AMI024, AMI032, AMI046, AMI047, AMI173, AMI175, AMI177, AMI200, AMI211 and VAC005), and the total variation is significantly lower: 0.4293. A vt of that range, certainly, indicates a common geological origin for all the reaming pottery, even if, they not necessarily form part of the same chemical group (or production). A new cluster analysis was performed in those remaining 103 individuals. The cluster tree of that analysis is presented in Figure 7. According to this graph, there are five individuals: AMI174, AMI180, AMI184, AMI166 and AMI170 that form a small chemical group (PC in Figure 7). All five were sampled at the Iberian site of Puig Castellar. From geological point of view, Puig Castellar is located to the north of Barcelona, in the area of the hill Coixarola, on a metamorphic and sedimentary ground, where the calcareous rocks are relatively abundant, opposite to the volcanic ground of the Maresme, where the rest of the sites are situated and, where the acidic igneous rocks like granite and granodiorite are dominant.
In Table 3, important similarities in the chemical composition of these five individuals, and the ommondifferencesfromrestoftheanalysedmaterialcanbeobserved. PCisacalcareousgroup (CaO between 8.98% and 10.76%) that also present lower concentrations in Fe 2 O 3 , MnO, TiO, Na 2 O, Th, Nb, Zr, Y, Ce, V, Zn and Ni. The calcareous character of this group is compatible with the location of the site where these five individuals came from. Therefore, the probability they were produced at this site and/or this are/region is high. Nevertheless, since there is no archaeological information on the existence of kiln/s on the site or

11/20
Arch & Anthropol Open Acc its surroundings, it can only be identified as Paste Compositional Reference Unite (PCRU) of Puig Castellar. The mean weight percentage of each element and their standard deviations can be seen in Table 5.
AMI029 is separated from PC since the only similarity with it is its calcareous character (14.32%). In this same cluster tree (Figure7) there is another larger group indicated with LAI that is formed by 59 individuals. The samples specified by the grey rectangles in Figure 7, at the right of LAI, according to the chemical data (Table 3) present some relatively important chemical differences (showed in this Figure 7) according to which they don't fit 100% inside this group. Despite that, in production terms, they must be local/regional variants of the group LAI. It is important to mention that one of the amphora included in this group (AMI053) have been sampled at the Cossetània, three come from Can Bartomeu (AMI036, AMI036 and AMI057), six from Turó dels Dos Pins (AMI019, AMI020, AMI021, AMI022, AMI028 and AMI054), eight from Puig Castellar (AMI182, AMI165, AMI169, AMI168, AMI172,AMI179, AMI185 and AMI176) and finally eight from the kiln site of Riera de Sant Simó (AMI188, AMI189, AMI190, AMI191, AMI192, AMI193, AMI194 and AMI195), which represents the total number of the sampled amphorae at this site, and the rest of the individuals belonging to the group coming from the kiln dump of Vista Alegre. The main typology of this production can be seen in Figure 9 and its mean chemical composition and the standard deviation in each analysed elements is given in Table 6. 12/20

Discussion and Conclusion
The chemical results lead to various conclusions. Regarding the Laietania (Figure 1), half of the individuals (50.43%) are classified in the RG of Rierade Sant Simó (LAI), while only the 5.22% belongs to the PCRU of Puig Castellar (Figure 10: A). The 24.35% of the nonclearly classified individuals are chemically related to Riera de Sant Simó's production. Their chemical composition indicates a common geochemical origin, and therefore, they must be regional variants of the group LAI. 17.39% of the individuals cannot be classified within any Laietanian group. However, they preserve geochemical relation with the Laietanian area. Finally, the 2, 61% individuals that are basically constituted by three individuals sampled at Can Bartomeu (AMI023, AMI024 and AMI029) and typologically identified as 2C, is not related to the Laietanian area at all. All the individuals (100%) sampled at kilns of Riera de Sant Simó belong to the LAI. That is why LAI was identified as the reference group of that kilns. 14/20

Arch & Anthropol Open Acc
More than a half of the individuals (57.63%) sampled at the dump of Vista Alegre is classified into LAI (Figure 10: B) and, generally, the 93.22% is somehow chemically related to this same production. There is only 6.78% that cannot be classified, nor can be related to the Laietanian productions (Figure 10: B). Regarding, now, the individuals sampled at Can Bartomeu (Cabrera de Mar) a high percentage is of unknown origin. 16.67% cannot even be related to the Laietanian area, while exactly the same percentage (16.67%) is clearly classified as LAI. The 11.11% seems to have Laietanian origin, even if they not belong to LAI (Figure 10: C). On the other hand, the 75% of the Iberian amphora sampled at the Necropolis of the Turró de dos Pins belongs to LAI and the rest of the individuals are distributed equally among unclassified, Laietanian origin and unclassified non-Laietanian origin ( Figure  10 Considering LAI, the RG of the Iberian kilns of La Riera de Sant Simó, it includes the totality (100%) of the samples from that specific kiln site that corresponds to the 13,56% of the production, while 57,63% of the individuals that belong to LAI were sampled at the kiln dump of Vista Alegre. A fact that clearly evidences that Vista Alegre was the kiln dump of Riera de Sant Simó. Additionally, the 11.86% of this production is represented by individuals coming from Puig Castellar, 10.17% from Turó de Dos Pins and 5.08% from Can Bartomeu (3 individuals).
It is important to mention that within LAI there is one single individual (AMI053) that was sample data Cossetanian site of Alorda Park (Calafell, Baix Penedès). This individual represents a very small part of the Laietana production (1.69%), nevertheless, it is clear indication that Iberian amphorae produced in the Laietanian site of Riera de Sant Simó travelled to southern Cossetania. This shows an exchange between the Cossetania and Laietània [28].
To conclude, there are other important facts to emphasize. LAI integrates both amphora and common wares that indicates the employment of the same recipe for the different type of ceramics. Additionally, from the amphorae sampled at Turró of the Dos Pins two that match with the smallest variants of the sub type 2C, which is archaeologically assigned to the Cossetania, belong to the RG of a Laietanian kiln site (Riera de Sant Simó), where the individuals mainly classified as local/regional subtypes: 2B, witnesses that the workshop, at least, produces imitations of other non-local Iberian amphora types. These specific results also stress out the local/ territorial character of the productions, which means the use of local/regional raw materials and the maintenance of a technological sequence in the production.
Another important aspect is that 24% of the analysed material of Laietan origin is chemically related to the kilns of Riera de Sant Simó, meaning similar geochemical origin, without necessarily having been produced at this same workshop (Figure10). Other 7.39% of the material is not classified within the LAI production, however geochemically related to the Laietanian area (Figure10) and only the 2.61% is of non-known origin (Figure10) and all of them sampled from Can Bartomeu. Regarding the composition of the non-classified individuals (Figure 10: G) 60% of the individuals were sampled at Can Bartomeu, 20% at Vista Alegre, 15% at Puig Castellar and only 5% at Turró dels Dos Pins. Therefore, the sight, where highest the diversity is Can Bartomeu.
To finish, the diversification of the participation of the different analysed sites in LAI (the reference group of the kilns of Riera de Sant Simó) evidences that this kiln was providing with Iberian pottery different areas of the Laietanian region. Also indicates some wider commercial inter-relations.
The high percentage of the unclassified material points to the existence of high divergence of productions that still have to be identified in the future. The results clearly indicate a commercial interaction between the Laietania and the Cossetania, for the moment. The chemical complexity reveals the existence of a multiplicity of productions in each geographical area, which still complicates the interpretation of all the available information. Adding to those results the typological information, even if, there is a diversification of production in the geographical areas (mean use of local/regional raw materials and specific technological processes in each kiln and production site) there is a certain standardization of forms in each of those regions. These results, of course, can be, and should be modified and expanded with future analysis.