The oxide layer argument

The claim: a glassy skin of aluminum oxide seals the lead inside scrap-melt cookware, so acid can't touch it. Run that claim next to the real metal, at the same timestamps, and watch where they split.

Scene 1 of 6

The claim

The argument says the lead is dissolved evenly through the aluminum and sealed under a continuous glassy oxide layer that acid cannot disturb. If that were the metal in your kitchen, leaching would be near zero unless you scraped through the glass.

the claim
vinegar lands, sits, never crosses continuous glassy Al₂O₃, as claimed lead “evenly dissolved” in the metal, as claimed
predicted leach: ~0 µg unless scraped
what's actually there
native film, a 2-4 nm hairline smear streaks from rolling discrete globules at grain boundaries
every surface globule: direct exposure
Scene 2 of 6

The real metal

Lead is essentially immiscible in solid aluminum. As a scrap melt cools it freezes out as discrete globules at grain boundaries, and rolling smears the soft surface globules (Mohs ~1.5 against aluminum's ~2.75) into streaks along the working direction. That is why an MABr spray under 365 nm UV shows green specks and streaks, never a uniform field.

The glow is the killer contradiction. The perovskite reaction needs available Pb²⁺, so lead sealed under glass could not glow at all. Anything the spray can light up, acid can reach.
365 nm
1 melt cools, lead freezes out 2 rolling smears surface globules into streaks 3 MABr spray + UV: specks and streaks glow, aluminum stays dark
Scene 3 of 6

The scale problem

The native oxide film is 2 to 4 nanometers thick and grows only from the aluminum itself, so it stops dead at the edge of every lead globule. A surface globule is roughly 1,000 times wider than the film is thick, and it caps itself with its own soft lead oxide instead.

Al₂O₃ film, 2-4 nm. Truly to scale it would be invisible here, so it's drawn as a hairline. a self-oxide grows from the aluminum, so it stops dead at the phase boundary lead globule: ~1,000x wider than the film is thick the globule's own cap: PbO / lead carbonate, soft and acid soluble
Scene 4 of 6

Vinegar, pH 2.5

Lead oxide plus acetic acid gives lead acetate, sugar of lead, the reaction the Romans ran by keeping wine in lead vessels. So the cap goes first. The film goes second: pH 2.5 sits below the passive window of Al₂O₃ (pH 4 to 8.5), so the armor itself slowly dissolves while galvanic corrosion undercuts the globules. Two doors, and the fast one was never locked.

the claim
acid bounces off glass sealed, untouched, as claimed
counter: ~0 µg unless scraped
what's actually there
4% acetic · pH 2.5 cap dissolves fast: lead acetate, sugar of lead, the Roman reaction Al₂O₃ film dissolves below pH 4, off as aluminum acetate (grey) lead gone, sockets stay, ions in the liquid galvanic undercut (Al is the anode, the matrix corrodes first)
pH 048.514
vinegar 2.5 sits outside the Al₂O₃ passive window (shaded, pH 4-8.5)
Pb²⁺ into solution: 0 (schematic)
Scene 5 of 6

The data

King County ran 250 mL of 4% acetic acid in each pot, simmered 15 minutes, sampled, then let it sit and sampled again at 24 hours. Each mark below runs from the 15 minute value to the 24 hour value. Hover or tap any item for exact numbers.

brass aluminum 15 min → 24 h
Brass is supply-limited: the exposed surface lead strips out fast in the hot phase and then runs out, because the copper matrix is noble (plain acid cannot dissolve copper), so no new lead gets exposed. The small overnight drift, 1.6 to 1.9x, is slow zinc loss and dissolved-oxygen attack. Aluminum is the opposite: the matrix itself keeps dissolving at pH 2.5 all night, undercutting and exposing fresh globules, so release climbs 13 to 30x overnight. The acid is not rinsing the surface, it is mining the metal. If a sealed oxide layer were in charge, every aluminum row here would sit near zero.
The Taiwan steamer reads a median of just 24 ppm on an XRF, would pass Washington's 90 ppm content limit, and still delivers 56.5 µg per serving, about 25x the child limit. Surface XRF content does not predict leaching.
view the data as a table
itemgroup15 min (µg)24 h (µg)
Harjee saucepan (India)brass19.7532.25
Unknown tope (India)brass104.5182
Taluka pital pot (India)brass1,4302,700
Chef Valley caldero (China)aluminum0.1852.46
MSR cookpot (US camping)aluminum0.4583.85
GSI cookpot (US camping)aluminum3.255.18
Unknown steamer (Taiwan)aluminum53.556.5
Wee's Beyond caldero (Colombia)aluminum2.3862.3
LBB skilletaluminum3.73106
Hindalium item (India)aluminum11.4340
Scene 6 of 6

The pressure cooker gap

The two Afghan pressure cookers were leached as open pots, so the vinegar never touched the brass vent parts in the lid. Run the pressure cooker like a pressure cooker (vinegar in, normal cycle, natural release, then sample) and the worst numbers in the dataset should only go up.

how it was tested
open pot: simmer 15 min, sit to 24 h lead globules in the body wall: the source of the 22-24x overnight climb body: 538-605 ppm 64,852 ppm 48,193 ppm lid off to the side brass vent + valve: dry, untouched

Open-pot results: 45.75 → 990 and 48.5 → 1,157.5 µg per serving. The 22-24x overnight climb comes from the lead globules in the aluminum body wall, no fast brass jump, because the liquid never met the lid.

how it's used
64,852 ppm 48,193 ppm body: 538-605 ppm steam runs the brass vent every cycle condenses, drips back into the food

Bodies scan 538-605 ppm, but the brass vent pipe and valve in the lid spot-read 48,193 and 64,852 ppm. So the published 990 and 1,157.5 µg figures are likely undercounts: the leach liquid never met the highest-lead parts.