Docs / What we have found

A registry number that is not this substance's

Part of What we have found, the class where the number on the flow is real, and belongs to something else.

Natural gas's number appears once in EF 3.1, on a flow that is not natural gas

EF 3.1 ships Gas, mine, off-gas, process, coal mining carrying 8006-14-2, which is natural gas's registry number. Gas that seeps out of a coal seam while the coal is being mined is not natural gas — different impurities, different heating value — and EF says so itself, on the neighbouring Pit Methane flow: "Pit methane is a different substance than natural gas (impurities, heating value etc.)".

EF keeps the two substances apart three ways — by name, by category (one is a material resource from ground, the other an energy resource) and by unit (m³ against MJ). The registry number is the single thing they share, and it is the only one of the four that decides identity downstream.

Meanwhile EF gives natural gas itself no identifier at all. 8006-14-2 appears exactly once in the whole of EF 3.1, and it is on the wrong flow.

What that cost was measured, and it was not nothing: enrichment looked the number up and wrote natural gas's names onto coal-mine gas, so Natural gas, Gas, natural and Sweet natural gas named two substances at once — and BAFU's two natural gas rows could then be placed on neither. ecoinvent shipped the same error and has already fixed it: 3.8 and 3.9.1 carry 8006-14-2 on this flow, and 3.10.1 onwards leave it off, under the same identifier (#412).

What the number can belong to instead

This is the plainest of the identity classes: the flow carries a registry number, and the number belongs to something else — a superseded entry for the same substance, or a neighbouring substance entirely — or it carries none at all and the substance cannot be reached.

The same class, four more shapes

Superseded. EF files borax under 12447-40-4. That number is genuinely borax's, but it has been superseded, and the number in current use is 1303-96-4 — which is how ChEBI files the mineral, twice, and what ecoinvent carries. Nobody disagrees about the substance; the two halves of the list simply never meet. Mecoprop is the same shape, EF holding 7085-19-0 where the rest of the world uses 93-65-2 (#564).

Borrowed from a neighbour. EF puts borax's number on six Borate rows as well as on its one borax row. Borate is the ion; borax is one particular salt of it — sodium tetraborate decahydrate, about a fifth boron by mass — so the two are not interchangeable in an inventory, and sharing a number fused them into one substance carrying borax's structure and trade names. ecoinvent tells them apart: its Borate rows carry 11129-12-7, the ion's own number.

Borrowed again, in the other list. ecoinvent gives Alpha-lindane, Beta-lindane and Delta-lindane all the gamma isomer's number, 58-89-9 — 21 flows across seven compartments each. The isomers have their own numbers (319-84-6, 319-85-7, 319-86-8) and different toxicity.

A number for something that is not a compound. All three lists put dinitrogen's number, 7727-37-9, on Nitrogen, organic bound — EF on five rows, ecoinvent on five, BAFU on four. Organic-bound nitrogen is not a compound at all: it is the nitrogen locked inside the organic matter of a discharge, reported as a mass of nitrogen, and no registry number can name it because a number names one compound and this is a measurement over many. The number is removed rather than corrected, because there is nothing to correct it to (#355).

Absent. Fifty of EF's part-numbered flows ship with no identifier of any kind, as do EF's own natural gas rows and BAFU's Anhydrite. A flow with no number is not wrong, but it is unreachable: nothing can look it up, and it will mint a substance of its own beside the one it should have joined. ecoinvent 3.8's Fosetyl-aluminium is what that costs. The row carries no number, 3.9.1 onwards call the same uuid Fosetyl-Al and register it 39148-24-8 — EF 3.1's fosetyl-aluminum, the fungicide sold as Aliette, which the JRC characterises in that very compartment — and the retired British spelling reaches EF's flow only as one of its synonyms, which is not enough to place a row against a substance that states a registry number the row does not. So the emission minted a substance of its own, and every 3.8 inventory's application of the fungicide scored zero while 3.12's and BAFU's identical rows scored. One curated match override answers it, guarded by the retired spelling so the releases that state their own number keep matching on it (#700).