Proposed methodology and hypothesis testing

Any attempt to derive salt (NaCl, halite) from seawater by evaporation (briquetage or salterns) goes through the same steps (Weller, 2015): A) CONCENTRATION, an initial enrichment to produce brine, B) CRYSTALLIZATION, artificial (fire) or natural (sun) evaporation process and, if needed, C) CONDITIONING (moulding) and D) PACKAGING. During the evaporation process, salts precipitate in this order: calcium carbonate (calcite), calcium sulphate (gypsum), sodium chloride (halite) and finally magnesium sulphate and other salts, referred to in the literature as ‘bitterns’. They give a strong bitter taste to the mixture and to obtain a good-tasting salt one needs to get rid of the bitterns. Archaeological features linked to each of these steps can be identified through a combination of data captured by Magnetic Gradiometry (MG) and Ground Penetrating Radar (GPR), which allow the identification of spatial structures, of which the nature and function can be checked by Coring Surveys (C) and Archaeological Excavations (E). We plan to apply in sequential order:

1. Ground Penetrating Radar. Using GPR (Conyers, 2018) we will identify specific features characteristic of the briquetage technique: the pits used for CONCENTRATION, the kiln structures (usually of stones) used in the CRYSTALLIZATION phase and other involved structures (settlement features, huts, fences).

2. Magnetic Gradiometry. MG surveys are instrumental to detect pyrotechnological features. Kilns used in briquetage and resulting dumps of pottery sherds produce high amplitudes and dipole characteristics (fig. 6). For salterns, intracellular ferromagnetic magnetite and/or greigite magnetosomes have been described, produced by magnetotactic bacteria that live in hypersaline aquatic environments (Bazylinski and Lefèvre, 2013). We will attempt to find a relation between the occurrence of magnetotactic bacteria and degree of salinity to interpret the MG surveys (but see Lin et al., 2012).

3. Coring. Following the GPR and MG surveys we will do coring to obtain soil samples to chemically characterize the soil. We will check for anomalous concentrations of Magnesium (Mg), Potassium (K), Sulfur (S) and Boron (B), which characterise the residual liquor discharged after the harvesting of the salt (NaCl). We will check for the presence in features of calcite and gypsum, which both characterise the CONCENTRATION step. We expect a gradual increase in the concentrations of Boron (B) and Bromine (Br) during the different stages of evaporation (fig. 3).

4. Excavation. Following the results of the GPR, MG and C, we will collect additional evidence about the detected features and structures through excavation.

To assess the use of the reddish jars we will use a combination of data from use-alteration traces, vessel content (salt?) and functional and physical characteristics. Diagnostic potsherds from excavations and existing collections will be analysed (fig. 5).

A first screening will be done using a pXRF (portable X-Ray Fluorescence, Frahm and Doonan, 2013; Donais and George, 2018) chosen for its rapidity, non-destructive analysis and in situ application. The range of elements that a pXRF can detect being limited, more powerful analytic techniques will be used on a selection of potsherds. We will then establish a classification based on stylistic attributes and chemical fingerprints and select sub-samples based on types. The chosen potsherds will be further processed to study:

1. Use alteration traces. This concerns surface use alteration traces like internal and exterior carbonization and sooting patterns, which can provide direct evidence for contact with open fire. Ceramic attritions (Skibo, 2015) will be recorded with optical microscopy. Among the non-abrasive attritional processes, the presence of spalls might be of particular interest since they can form during fermentation processes (Skibo and Blinman, 2008);

2. Salt traces? A SEM-EDS (Scanning Electron Microscopy/Energy Dispersive Spectroscopy) will be used to characterise qualitative and semi-quantitative chemical properties of the potsherds. In recent research, the focus has been on the penetration of Na and Cl into the walls of potsherds as a proxy for the presence of ancient salt traces. But since Na and Cl are highly mobile elements easily leached by infiltrating water, this method is not reliable. Moreover, in coastal areas, salt spray constitutes a potential cause of relatively high Cl and Na. We suggest to use Boron (B) and eventually Bromine (Br) as proxies for seawater used in the production processes. Boron is far more common in seawater than in a terrestrial environment and is less sensitive to leaching and salt spray (Sevink et al., 2020); Bromine is also mostly present in seawater and is characteristic of saline environments at times reaching very high concentrations (Dolphin et al., 2013; Moreno et al., 2017);

3. Chemical characterization. We propose to complement the SEM-EDS with ICP-MS analysis (Inductively Coupled Plasma – Mass Spectrometry) to improve the chemical characterization of the samples (Pollard and Heron, 2017). SEM-EDS is capable of performing analyses of selected point locations, but does not give absolute quantities and has poor sensitivity to trace elements lighter than Na. ICP-MS is a bulk analysis method, unable to trace trendlines across the potsherd walls, but has a much greater sensitivity to trace elements and gives absolute quantities. We will pay particular attention to the presence of sulphate which is less mobile than NaCl and is a proxy for the evaporation process due to the gypsum (sulphate mineral) precipitation from the seawater;

4. Mineralogical characterization. The petrographic and mineralogic characteristics of the selected samples will be described using a PLM (Polarizing Light Microscope) (Stoltman, 2015; Hunt and Bishop, 2016) according to the recording system already proposed by Orton and Hughes (Orton and Hughes, 2013, pp. 275–285).

5. Physicochemical properties. The combination of the results from point 3 and 4 will reveal the physicochemical properties of the reddish jars.

6. Functional attributes. To study the steps of C) CONDITIONING (moulding) and D) PACKAGING the capacity, stability, accessibility and transportability will be evaluated through technical considerations and parallels with already published ethnographic data (Orton and Hughes, 2013).

Interpretations of the inferred use (Rice, 1996) of the jars will be made on the basis of these results and the interpretation of production structures both published and found in our planned new excavations. Results will be used to enhance our understanding of the chaîne opératoire.

Goal 2A: the evolution of the early state territories

To reconstruct the political network in which the briquetage sites functioned, we review the existing literature about the reconstruction of the territories of the Early States. Several scholars studied this issue with different spatial but synchronic models, borrowed from geography, such as Thiessen polygons (Barbaro, 2010; Fulminante, 2014) and X-Tent (Redhouse and Stoddart, 2011). We opt for a diachronic model (Alessandri, 2016), linking chronological phases in complex time-space models (Bubble Model) to trace the evolution of the settled landscape before and after the birth of the Early States embedding the salt production sites in the latter’s territories.

Goal 2B: check the evolution of the geographic organization of the production units

To study the labour organization of salt production, we focus on the distribution and density of briquetage sites over the landscape. An even distribution of production units would point to a production geared at local demand while aggregated specialists, with few production units randomly distributed in the landscape, can be interpreted as evidence for regional or interregional exchange (Costin, 2005). Based on previous research, we envisage this transition to have occurred at the end of the Bronze age in the period of the formation of the early states (Alessandri et al. 2019). We use a model that describes the transition from seasonal household production to workshops in times of increased economic demand (Nijboer, 1998).

 

Goal 3A: assess the scale and evolution of the output
We will use collected potsherds from the excavations to infer change(s) in the output scale (Orton and Hughes, 2013; Banning, 2020) and link the evidence for change in the output scale to concentration of specialised sites established during early state formation (Late Bronze Age /start of the Early Iron Age). We compare the composition of assemblages using MNI (Minimum Number of Individuals) to reconstruct the amount of salt produced, calculating the volume of the jars (Rodriguez and Hastorf, 2013) and the original pottery population based on the brokenness and completeness of the samples (Felgate et al., 2013).

Goal 4A: dating the end of the briquetage technique
The only radiocarbon dates available from briquetage sites come from Pelliccione in the Pontine Plain and fall into the Late Bronze Age (Attema, 2004). However, reddish jars typologically dated to the VII century BCE have been found at Nettuno, south of Rome (Alessandri, 2013). To assess the relative chronology of the last evidence of the briquetage technique both north and south of the Tiber we will use the data from literature and the newly collected ceramics from the excavations. Absolute dates will be obtained by radiocarbon (organic) and thermoluminescence (pottery) (Aitken, 1985; Feathers, 2009).

Goal 4B: detecting ancient salterns
As mentioned (Goal 1A), salt production through evaporation of brine always follows the same steps and the methods used to reconstruct the chaîne opératoire of the briquetage sites can also be used to detect the presence of ancient salterns. We employ both GPR and MG to detect terrain adaptations, dug-out ponds and water channels transporting seawater or brine. Coring campaigns (C) will be conducted to chemically characterise detected ponds, analyse the concentration of Boron (B) and Bromine (Br) expecting their gradual increase in successive ponds and detect the presence of the bitterns in the final stage of the process. Test pits will be dug to double-check our interpretations.

Goal 4C: dating the beginning of the saltern mode of production
Importantly, we want to date the earliest evidence of salterns. In the literature, the beginnings of the Ostia and Veio salterns are dated around the 7th and 6th centuries BCE. While corroborated by the artificial increase in salinity of the lagoons, the salterns themselves have not yet been physically attested and dated. Study of the Puntone site (fig. 5) revealed that brine pits and associated sediments contain organic matter of largely microbial origin (Sevink et al., 2020), which is suited for absolute dating of phases in the salt production process. The same holds for salterns in which microbial mats form, the remains of which can be used to date the ponds. Combining C14 with thermoluminescence dating of collected ceramics will give a reliable chronological framework for the salterns.