The Villa della Pisanella at Boscoreale was buried under the eruption of Vesuvius in AD 79 and excavated in the late 1890s. Among the finds were ninety-seven amphorae from the villa's cellar, most of them already open or broken; fourteen of them, however, survived the eruption and the excavation in essentially sealed condition, with both the body of the vessel intact and the pitch-and-mortar plug in the neck unbroken.

Those fourteen amphorae have sat on a shelf in the Antiquarium di Boscoreale since 1908. Until last month nobody had asked them what was inside.

I should be clear: we did not open them. The whole point of a sealed Pompeian amphora is that it is sealed; cracking the plug would destroy the most informative thing about the object. What we did, with the Soprintendenza Speciale di Pompei and the Boscoreale curatorial team, was run a portable X-ray fluorescence (pXRF) campaign through the body of each vessel, mapping the elemental signature of the residues clinging to the inside walls. XRF is non-contact, non-destructive, and unaffected by the pitch plug.

It also told us something we did not expect.

The instrument and the protocol

We used a Bruker Tracer 5g handheld XRF unit, calibrated against a series of in-house garum and wine residue standards prepared by the Naples team in 2024. The standards matter: a generic pXRF calibration will give you elemental ratios that look credible and are wrong, because the absorption characteristics of a 4 mm-thick Pompeian terracotta wall are different from those of any commercial calibration set.

Each amphora was scanned at twelve points: four around the equator at the top of the body, four at the equator just above the base, and four on the base itself. We collected two-minute spectra at each point at 40 kV with the Rh secondary target installed for the lighter elements, and again at 50 kV for the heavier ones. Total scan time per amphora was around fifty minutes; total time on each campaign day, including the careful repositioning that did not let the instrument touch the vessel, was around ninety minutes for two amphorae.

The full elemental data is on the Open Context repository under embargo until October, at which point the journal paper will publish.

What we expected

Wine. Boscoreale was a wine-producing villa, the cellar is a wine cellar, the Dressel 2-4 is a wine amphora, and the contemporary literary record describes the Vesuvian slopes as one of the great wine regions of the early Empire. The expected elemental signature was the standard wine-residue pattern: elevated potassium and calcium, modest iron, traces of strontium and rubidium consistent with the trace-element fingerprint of the local volcanic-soil grapes.

Thirteen of the fourteen amphorae produced exactly that signature. Some are richer in calcium than others, some show evidence of resin sealants on the interior walls, all of them read as wine.

What we did not expect

Amphora B-RB-009 — the second-to-last on a shelf in the back of the storeroom, with no special label and a slightly more careful pitch plug than its neighbours — produced an elemental signature with significantly elevated phosphorus, sulphur, and chlorine, alongside a calcium-to-potassium ratio that is wildly off the wine-residue benchmark. The phosphorus reading is well above what fermented grape juice would produce, and the chlorine reading is roughly four times what we see in the cleanest of the surrounding wine amphorae.

That triplet of P, S, and Cl, with that Ca/K ratio, has one good organic-chemistry explanation in a Pompeian context: a fish-based sauce. Most likely garum — the fermented fish sauce that was the dominant condiment of the Roman Mediterranean and that we know from production sites at Pompeii itself and at Baelo Claudia in southern Spain to leave precisely this elemental fingerprint.

One sealed amphora in a wine cellar holding fish sauce instead of wine is not a stunning revelation about Roman cuisine. It is a small revision to what we think the Villa della Pisanella's cellar was actually for.

Wine cellars in well-documented Roman villas are usually wine cellars in the boring, monocrop sense. Boscoreale being a slightly mixed-purpose storage room — wine plus a small reserve of high-value condiments — fits with what we know from the textual record about how Vesuvian villas operated, but it has not previously been directly evidenced at this site. One amphora is a single data point. It is, however, a single data point with consequences for the inventory we have been working with for over a century.

Why this matters and what it doesn't

It matters for two reasons, both narrow.

1. Inventory accuracy. When researchers cite the Boscoreale cellar in arguments about Vesuvian wine production volumes, they should be citing thirteen confirmed wine amphorae and one confirmed condiment amphora, not fourteen wine amphorae. That number will appear in capacity-estimation papers for the next decade and being right about it is the kind of routine archaeological hygiene that adds up.
2. Garum trade economics. Garum was usually transported in smaller, distinctively shaped vessels — urcei or salsamenta amphorae — not in a Dressel 2-4 wine shape. Finding garum residue in a wine-amphora shape, in a stored cellar context, raises the modest question of whether the villa was decanting purchased garum into reused wine amphorae for cellar storage. There is some Pompeian literary evidence for exactly this practice, but the archaeological evidence has been thin.

It does not matter for any of the grander claims that are easy to reach for. It is not evidence that the cellar was a "garum store" — thirteen out of fourteen amphorae are wine. It is not evidence about Roman dietary preferences, which we already know in great detail from other sources. It is one well-measured data point that adjusts the inventory of one room in one villa.

What we are doing next

We are running the same protocol against the remaining sealed-amphora cellars at three other Vesuvian sites: the Villa Regina at Boscoreale, the cellar at the Villa of Mysteries at Pompeii, and a smaller suburban deposit at Oplontis. Total target: 41 additional sealed vessels by the end of the year.

If the Boscoreale result is part of a pattern — even one or two of those 41 producing garum signatures — we will have evidence that the practice of decanting condiments into reused wine amphorae was non-trivial in the early-imperial Vesuvian economy. If it is unique to Boscoreale B-RB-009, it is a curiosity. Both outcomes are useful; that's the comfortable position to be in when you start a campaign.

The October paper will include the full elemental dataset, the calibration standards, and the pXRF parameter file. Reproducibility on the standards is the part I'm most pleased with, because for a long time pXRF in archaeology was a "trust me" instrument. It does not have to be.

— Elara