Docs / What we have found

A substance is given its chemical relative's common name

Part of What we have found, the class where the plain common name sits on the salt, the hydrate or the racemate rather than on the chemical the name belongs to.

Paraquat is not paraquat dichloride, and a tonne of gypsum is not a tonne of anhydrite

EF 3.1 puts the plain common name on the relative for six substances, over 66 flows:

Take paraquat. The chemical called paraquat is a doubly charged ion, 4685-14-7. What is sold and sprayed is its chloride salt, paraquat dichloride, 1910-42-5 — 38% heavier for the same amount of active herbicide. EF 3.1 ships both, and puts the plain name on the salt.

Three witnesses, and the third is the source itself. Every one of the thirteen flows carries the dichloride's registry number and the dichloride's EC number, 217-615-7. Every one of the fifty-odd synonyms EF ships with them names the salt — Paraquat dichloride, 1,1'-Dimethyl-4,4'-bipyridinium dichloride, Methyl viologen dichloride, and the products Gramoxone, Esgram, Pillarxone; the bare word Paraquat appears in none of them. And EF states the rule itself, in the general comment shipped on these flows:

Correct mapping CAS and flow to be verified: yet unclear whether this is the sodium-, hydrochloride-, calcium-, potassium- or other salt or the pure substance that is emitted. Note that the CAS No is the relevant identifier, to which also the ILCD LCIA characterisation factors relate.

That is the source telling the reader to trust the number over the name.

The substance the plain name belongs to is already in the list, in all thirteen compartments, under its systematic name 1,1'-dimethyl-4,4'-bipyridinium — and that flow's synonyms include PARAQUAT, Paraquat ion and Paraquat dication. Nothing is lost by taking the bare name off the salt.

What it cost, and who caused it. This is where the attribution has to be careful, because the naming error and the mapping error have different authors.

GLAD got three of the five right. Its file maps ecoinvent's paraquat to EF's 1,1'-dimethyl-4,4'-bipyridinium — the ion, correctly — at rows 3659 and 3660; mcpa to 2-methyl-4-chlorophenoxyacetic acid, the acid, at rows 2928 to 2931; and mecoprop-p to (r)-2-(4-chloro-2-methylphenoxy)propionic acid, the pure enantiomer, at row 2937. Every one of those is a registry-number match onto the right substance, and none of them uses EF's mislabelled flow. The misroutings this project measured in its own builds — five paraquat rows onto the salt, four MCPA rows onto the sodium salt, six mecoprop-P rows onto the racemate — come from rows added to the table after GLAD, by name, in compartments GLAD never mapped. They are this project's own and are not a finding about published data.

Two of the five are GLAD's, and both are forced by EF. gypsum, in ground (row 2183) and anhydrite, in ground (row 331) are both mapped onto EF's single gypsum flow — the one carrying anhydrite's registry number — because EF 3.0 has exactly one flow for the two minerals. Both rows are marked FLOWNAME_MANUAL (PROXY); two different minerals, one target, and the target is named after the hydrate while it is numbered as the anhydrous salt. And flupyrsulfuron-methyl (row 2051) was a true number match in 3.7 and became wrong when ecoinvent renumbered its flow, which is the identifier-reuse class.

What the EF naming error costs regardless of any table is the thing this section is really about: any step that matches on a name — and every list without a correspondence table matches on names — sends a row naming paraquat to a flow that is paraquat dichloride, and a row naming gypsum to a flow that is anhydrite. The name is the trap, and it is EF that set it.

The decision. Correct the name on EF's flow, keep the registry number, and do not retain the replaced name as a synonym — because it is a name of a different substance, and retaining it would leave every name-matching step landing exactly where it did before. Where the correct target exists in EF, the ecoinvent rows are re-pointed onto it; where it does not — the flupyrsulfuron parent acid, gypsum in the compartments ecoinvent uses — the mapping is declined and the substance is published under its own name (#597).

Mecoprop needed one more thing: EF files it under 7085-19-0, and the number in current use is 93-65-2. Asked for the old number, CAS Common Chemistry returns a record whose own registry number is 93-65-2 — the service resolving the superseded entry onto the current one. ecoinvent and BAFU both already carry 93-65-2, so EF was the only list holding the old number, and correcting it moved eight ecoinvent rows and one BAFU row from matching on a name to matching on a number.

What it cost, by source list:

Common names, and the thing that is actually sold

What EF did with paraquat is not peculiar to EF, or to herbicides.

Pesticides, minerals and pharmaceuticals are known by a common name — paraquat, MCPA, gypsum — and the thing actually sold, sprayed or mined is very often not the chemical that name belongs to. It is a salt of it, a hydrate of it, an ester of it, or one of its two mirror images. The relatives differ in molecular weight, in the dose applied per hectare, and sometimes in what they do.

A flow list that puts the plain common name on the relative is not making a spelling mistake. It is making a claim about identity that its own registry number denies, and — because correspondence tables and name matching both work on names — every list downstream inherits it, including lists that had the pair right.