Every number KitchenMetrics prints is traceable to a published source: a national regulation, an international standard, or the peer-reviewed literature. This document maps each figure to its origin so that anyone — a chef, an IT reviewer, a food-safety inspector — can verify it independently, without taking the app’s word for anything.
This is not a certification. No public body certifies calculation apps on request. What this document offers is stronger than a badge: full traceability. The app cites the source; you check the source. Where a figure is an estimate rather than a regulation, it is labelled as such — see Honest limitations at the end.
Ogni valore stampato da KitchenMetrics è riconducibile a una fonte pubblicata: una norma nazionale, uno standard internazionale o la letteratura scientifica. Questo documento collega ogni dato alla sua origine, così che chiunque — cuoco, revisore IT o ispettore — possa verificarlo in autonomia.
Last reviewed: 21 July 2026. On 16 July four primary legal texts were read directly (Switzerland, Spain, Argentina, Chile) and eight clarification requests were sent to the bodies that own the remaining figures. The Status column below is the position on 21 July, five days on — and most of it is still silence, which is the honest thing for this table to show.
Where a reply exists but is not quoted. Two bodies have answered and their answers are not reproduced here: permission to cite private correspondence has been requested and not yet granted. What is recorded is only what we may state ourselves — that a reply arrived and on what date. No conclusion in this document rests on an unquoted reply; where an entry changed, it changed on grounds stated in the open, and the Italy note below is the worked example of that.
| Sent (CEST) | Body | Channel | Status on 21 July |
|---|---|---|---|
| 11:36 | 🇮🇹 Ministry of Health — DGISAN | email · vigency of Circular 1/1991 | Replied 17 July. A follow-up was sent 18 July and the definitive answer is outstanding. The reply is not quoted pending permission to cite. The Italy row was demoted to disputed on 19 July on grounds stated openly below — that nobody on this project has read the Circular — and not on the strength of that reply |
| 12:37 | 🇩🇪 DGF — Deutsche Gesellschaft für Fettwissenschaft | email · replied the same day: forwarded to their experts | Forwarded to their experts on 16 July. Nothing on the merits since. The 24 % figure remains attributed as a DGF recommendation, which is what it was before the query |
| 14:07 | 🇲🇽 COFEPRIS | email → auto-routed, not answered: that desk replies that it “no atiende trámites, ni es vinculatorio, únicamente es de carácter informativo” and lists the channels. Re-sent the same day to the Citas Técnicas channel (reunionesfomento@), whose stated purpose is “consultas precisas sobre la aplicación de las disposiciones técnico-regulatorias” — though that channel is meant for sectores regulados, which we are not, and we said so |
No answer from either channel. Neither the original desk nor Citas Técnicas has replied on the merits |
| 14:37 | 🇮🇪 FSAI | email · Advice Line (staffed by food scientists) | No answer in five days |
| 14:38 | 🇨🇦 CFIA | email → routed, not answered: on 2026-07-18 Public Enquiries replied that Ask CFIA handles regulatory questions and instructed us to resubmit through the online contact form, selecting Industry/Business → Food, with a stated reply time of 20 business days. Resubmitted 2026-07-19 through that form (Country: Other → Italy), stating in the body that we are not a food business operator and are using that route only because Ask CFIA directed us to it. No receipt or case number was issued. The 20-day clock, if it runs at all, runs from 19 July — not from the original email | No answer. If the 20 business days run, they end about 16 August |
| 14:48 | 🇬🇧 FSA | FOI request (Freedom of Information Act 2000) · acknowledged in 12 minutes by the Information Governance Manager · statutory reply due within 20 working days | Acknowledged only. The statutory clock runs from 16 July, so the reply is due about 13 August. This is the one deadline in the table with law behind it |
| — | 🇺🇸 FDA | web form 3907 (FCIC / FSMA Technical Assistance Network), category Retail Food | No answer, and no receipt exists to chase it with |
| — | 🇦🇺🇳🇿 FSANZ | web enquiry form | No answer, same caveat |
(The two web-form submissions carry no timestamped copy — unlike an email, they leave the sender no receipt. Both were submitted the same afternoon. The same is true of the CFIA resubmission of 19 July: the form returned no confirmation number, so its only trace is this note.)
Three letters concern the thermal side of the app rather than frying-oil limits, and had no record in this file. They are logged here for the same reason as the table above: a claim is worth what its trail is worth.
| Sent | Body | Subject | Status on 21 July |
|---|---|---|---|
| 19 July | 🇺🇸 ASHRAE — Handbook Editor | A suspected typographical error in the water-density row of the Choi–Okos thermal-property table: the quadratic coefficient is printed with an exponent three decades from the value the SI form converts to. The letter states the scope of the check — 41 coefficients converted, 40 matching — and the two things it did not check | Replied 21 July. The reply is not quoted and its author is not named: permission to cite has been requested in the response sent the same day and has not been granted. What may be stated is our own side — the error was reported, and the matter is not closed. No number in this app rests on that reply. The letter as sent is kept in the private repository |
| 19 July | 🇮🇹 CREA — Alimenti e Nutrizione | Provenance of the food-composition figures the thermal model uses | No answer in two days |
| 19 July | 🇮🇹 Ministry of Health | Good Hygiene Practice guidance | Draft in the private repository |
(A fourth request, a FOIA to the Italian Ministry of Tourism, belongs to a different project and is not tracked here.)
Every request asks the same thing and refuses to ask anything else: whether the norm exists and what it says. None asks for an assessment, endorsement or approval of the application — no public body certifies calculation software, and asking would only earn a refusal worth quoting against us.
Each entry below carries the date it was checked. A pending request is not an answer, and this table will not pretend otherwise until the replies arrive.
Total Polar Compounds (TPC) limits are national, and they disagree. There is no EU, EFSA, FDA or Codex Alimentarius limit — a claim the app deliberately corrects, because citing a non-existent threshold in a HACCP tool is something an inspector can falsify on the spot.
The app always applies the stricter of the fat’s own chemistry limit and the local legal ceiling; a jurisdiction can only ever tighten the threshold, never relax it.
Where the “read” column is ticked, the primary legal text was consulted directly on the date shown, not taken from a review paper. A citation without a consultation date is worth little: laws change.
| Country | Limit | Status | Legal basis (as cited in-app) | Primary text read |
|---|---|---|---|---|
| Italy | 25% | disputed — see note | Ministry of Health Circular no. 1 of 11/01/1991 | ❌ not read. Vigency query sent to the Ministry (DGISAN) 2026-07-16, 11:36 CEST, follow-up 2026-07-18 |
| Switzerland | 27% | statutory | ODOV/VLpH art. 6 para. 4 — DFI Ordinance of 16.12.2016 (RS 817.022.17) | ✅ 2026-07-16 (Fedlex; version in force 01.01.2026) |
| Spain | 25% | statutory | Orden de 26 de enero de 1989, art. 6.3 (BOE-A-1989-2265) | ✅ 2026-07-16 (boe.es) |
| Argentina | 25% | statutory | Código Alimentario Argentino art. 552 bis (Res. Conjunta 17/2021); method ISO 8420:2002 | ✅ 2026-07-16 (boletinoficial.gob.ar) |
| Chile | 25% | statutory | Reglamento Sanitario de los Alimentos (D.S. 977/96) art. 266 let. c | ✅ 2026-07-16 (consolidation to Aug 2017 — see caveat below) |
| Mexico | — | no maximum | NMX-F-068 defines how to measure, sets no maximum (verify with COFEPRIS) | ⏳ COFEPRIS asked 2026-07-16; the citizen desk auto-routed without answering, re-sent to Citas Técnicas the same day |
| United Kingdom | 24%* | no statutory limit | FSA — general food-safety law; *industry practice | ⏳ FOI request (Freedom of Information Act 2000) sent to the FSA 2026-07-16, 14:48 CEST — acknowledged the same day by the FSA’s Information Governance Manager; statutory reply due within 20 working days |
| Ireland | 25%* | no statutory limit | FSAI; *industry benchmark | ⏳ query sent to the FSAI Advice Line 2026-07-16, 14:37 CEST |
| United States | 25%* | no statutory limit | No federal TPC limit found (FDA); *voluntary benchmark, some states differ | ⏳ query sent to the FDA (FCIC/FSMA TAN, form 3907, category Retail Food) 2026-07-16 |
| Canada | 25%* | no statutory limit | CFIA; *industry benchmark | ⏳ asked twice, not yet answered on the merits: email 2026-07-16, routed on 2026-07-18 to the Ask CFIA online form, resubmitted there 2026-07-19 (no receipt issued) |
| Australia | 27% | disputed | Reviews cite 27%; industry says TPM is not a legal limit — confirm with state regulator | ⏳ query sent to FSANZ 2026-07-16 |
| New Zealand | 25%* | no statutory limit | FSANZ Food Standards Code; *industry benchmark | ⏳ covered by the FSANZ query 2026-07-16 |
This table listed Italy as statutory until 19 July 2026. That was an overclaim, and it is withdrawn here on the same grounds it was withdrawn in the app.
Two things are wrong with it. First, nobody working on this project has read Circular no. 1 of 11/01/1991. The 25% reaches us through review papers and industry guides that cite it to one another; not one line of the instrument itself has been consulted. Every other row marked statutory in this table carries a date in the “primary text read” column. Italy never did — it carried a pending query instead, and a pending query is not a reading.
Second, and independent of what the Circular turns out to say: a ministerial circolare is an administrative act addressed to the administration. Its binding force is not that of a statute. Even a faithful reading of it would not obviously support the word statutory.
The app still discards at 25% in Italy. The operative threshold is unchanged and was never in question — 25% is the strictest figure in this table, the engine can only ever tighten, and a demotion in status must never loosen a limit in a kitchen. What changed is the claim about why, which is a claim about Italian law and was not ours to make on an unread text.
To resolve it: read the Circular in the Gazzetta Ufficiale. If it sets a maximum, this row becomes statutory with a reading date like every other. If it recommends one, it joins Germany’s 24% as a criterion rather than a law.
No TPC limit found for: the European Union / EFSA, the US FDA, and Codex Alimentarius.
The wording matters. “No limit found” is what the evidence supports; “sets no limit” asserts a fact about a body’s entire corpus, which nobody here has exhaustively searched. The app used to mark the US claim as verified against the primary legal text — for a claim that no such text exists. An absence cannot be read; it can only be attested by the body itself, and none has been asked. Where a negative is only ever “widely repeated and never contradicted”, this document says so.
The 24% figure widely quoted for Germany — including the “strictest anywhere” figure the app displays — originates from the DGF (Deutsche Gesellschaft für Fettwissenschaft), established at the 3rd International Symposium on Frying Fats (2000). German food-control authorities use it as an assessment criterion, alongside sensory findings, polymeric triglycerides (max. 12%) and acid value — not as a single statutory maximum. Germany is deliberately absent from the registry above, because no primary legal text has been established for it.
A clarification has been requested directly from the DGF (16 July 2026); their Managing Director confirmed receipt and forwarded the question to their experts. This document will be updated with their answer. Until then the app calls 24% a reference figure, never a limit.
Four primary texts were read on 16 July 2026, and not one of them sets the polar figure on its own. Every text embeds it in a package of further requirements — sensory in Spain and Argentina, purely chemical in Chile and Switzerland:
| Jurisdiction | What the law actually requires |
|---|---|
| Spain (art. 6) | 6.1 free of substances foreign to frying · 6.2 organoleptic characters must not taint the food · 6.3 polar components below 25% |
| Argentina (art. 552 bis) | (a) alterations in smell or taste — listed first, and joined by “y/o”, so it condemns the oil on its own — and/or (b) TPC ≥ 25% |
| Chile (art. 266) | (a) free acidity > 2.5% · (b) smoke point < 170 °C · (c) polar compounds > 25% — three chemical/physical criteria, no sensory leg |
| Switzerland (ODOV art. 6) | para. 2 acidity (Säuregrad) · para. 3 trans fatty acids ≤ 2 g/100 g · para. 4 polar fraction ≤ 27% — again no sensory leg |
(Germany’s DGF criteria do put sensory findings “im Vordergrund”, alongside polymeric triglycerides (12%), polar compounds (24%) and acid value (2%) — but the DGF is a learned society and its criteria are not a legal text, so they do not belong in the count above. See the Germany section.)
Switzerland is the instructive case: its statute has no sensory requirement at all, yet its own enforcement laboratory approaches the oil from exactly that side — the Ticino report lists the signs that condemn an oil with no instrument at all: rancid or scratchy taste, incipient smoking, foaming, viscosity, a pungent smell over the hot fryer. What no text does is let the polar number stand alone.
Correction, 2026-07-18: this section previously claimed that every text read pairs the polar figure with a sensory test. That was false. It is true of Spain and Argentina; Chile’s and Switzerland’s further criteria are chemical, and the table listed Germany — not a legal text — in place of Switzerland, which was one of the four texts actually read. Art. 6 of the ODOV was re-read at source before writing the corrected row.
KitchenMetrics models the polar-compound figure only. That is stated here rather than glossed over — and it is not, on reflection, a weakness: the TPC number is precisely the leg of the stool a cook cannot judge by eye, which is why an instrument is worth having. But the app does not replace looking at, smelling and tasting the oil, and in none of the four jurisdictions whose text we read does the number alone decide compliance.
Threshold wording: Spain requires the content to be “inferior al 25 por 100” and Argentina condemns oil at “igual o superior al 25 %” — so 25.0% is already non-compliant, not the last acceptable value. The app’s arithmetic always treated the threshold as inclusive; its wording was corrected on 2026-07-16 from “discard when TPC exceeds X%” to “discard at or above X%”.
Art. 266 of the RSA sets three limits, any one of which condemns the oil: “a) acidez libre expresada como ácido oléico superior al 2,5%; b) punto de humo inferior a 170 ºC; c) 25% de compuestos polares como máximo.” Art. 265 caps linolenic acid at 2% for industrial/institutional frying; art. 267 bans reusing discarded frying oil in other human food. KitchenMetrics currently models only the 25% — an honest gap, not a claim. Note the wording “aceites y mantecas“: Chile covers animal fats explicitly. Version read: consolidation to August 2017; post-2017 amendments not checked, and article numbering shifts between consolidations — verify on BCN.
The Laboratorio cantonale del Ticino publishes its frying-oil checks. Its February 2026 report on the 2025 campaign (“Qualità dell’olio per frittura”, Bellinzona) cites the same provision this document does — “Ai sensi dell’Ordinanza del DFI sulle derrate di origine vegetale, i funghi e il sale commestibile (ODOV) del 16 dicembre 2016 tale valore, nei grassi e negli oli commestibili per friggere, non deve superare il 27 per cento (270 g/kg)” — which is an official enforcement body reading the ordinance exactly as we do.
Its findings are worth stating, because they show the limit is not theoretical:
The same report also corroborates two things stated elsewhere in this document: that a frying oil’s smoke point falls as the oil is used (“il quale si abbassa man mano che l’olio viene utilizzato” — so a database of fresh smoke points cannot stand in for measuring the used oil), and that degradation is recognisable without any instrument — rancid or scratchy taste, incipient smoking, increased foaming, viscous fat, pungent smell over the hot fryer.
Source: Laboratorio cantonale, Ticino — Olio di frittura, campagna 2025 (PDF, read 2026-07-16).
Art. 6 para. 4 of the ODOV/VLpH (RS 817.022.17) reads: “Der polare Anteil in Speisefetten und Speiseölen zum Frittieren darf 27 Prozent nicht übersteigen.” Version in force: 1 January 2026. Open question: the ordinance’s scope (art. 2, art. 1 let. b) covers foodstuffs of plant origin, and art. 6 sits in the chapter on plant oils and fats — so whether the 27% also binds animal frying fats (lard, tallow) is not settled by this text. The app does not assume it does.
Primary source for the registry: Song et al., “Feasibility of total polar compound … aspect
of regulations of various countries,” Food Chemistry (2022), cross-checked against primary
legal texts where marked official in the app (tpc_limits.js).
How to verify: each basis string above is a searchable legal reference. BOE-A-1989-2265
resolves on boe.es; the Italian Circular is citable to the Ministero della Salute; the Argentine
article is in the Código Alimentario Argentino online.
Smoke point, safe frying temperature, crystallisation, density, fatty-acid composition,
oxidation risk and allergens for 148 fats and oils. Each record carries its own src field —
the database is a single audited source of truth, and a second registry that consumes it is
test-pinned to agree with it.
Sources cited across the records include:
Allergen policy: allergen labels are never machine-translated, because the allergen hangs off the food’s name; a mistranslation would be a safety bug, not a typo. Translated fat names keep the English original alongside (e.g. “Strutto (Lard)”) for the same reason.
Cook and pasteurisation times are computed from transient heat-conduction physics, not lookup tables:
λ·tan(λ) = Bi) — not the naïve
Bi→∞ shortcut, which under-predicts cook time up to 3.4× in the unsafe direction. A sanity
gate refuses to output if any derived property goes non-physical, and flags results where the
one-term series itself stops being valid (Fo < 0.2).Regulatory / standards references cited in-app:
Aligned with the Uniform System of Accounts for Restaurants (USAR), published by the National Restaurant Association. (There is no ISO/IEC standard for food costing.)
The engine is a pure function with a public regression test suite (cost_engine.test.html).
This document holds the app’s figures to the primary legal text of twelve countries. The costing
engine, meanwhile, prints € for all of them — the symbol is welded into the formatter, and
there is no notion of currency or country anywhere in the costing path.
Nine of those twelve countries do not use the euro (CH, AR, CL, MX, GB, US, CA, AU, NZ). The Spanish interface serves Argentina, Chile and Mexico — none of them euro. The English interface serves the UK, US, Canada, Australia and New Zealand — none of them euro either. A cook in Zurich is given ODOV art. 6 para. 4 cited to the paragraph, and then a plate cost in euros.
Nothing is miscalculated: the arithmetic is currency-agnostic, so only the label is wrong. But it is stated here rather than left to be discovered.
Two further European assumptions, for the same reason:
Prime Cost % < 65% benchmark is likewise American, applied everywhere.The fix, and the design decision behind it: the country becomes the single selector and the
currency follows from it — not the other way round. Using the currency to infer the applicable
law was considered and rejected: the euro alone spans Italy (25%), Germany (24%, and a DGF
recommendation rather than an established statute) and Ireland (no statutory limit at all), and
the $ sign spans seven countries in this very registry whose rules range from 25% to none to
disputed. Currency does not identify jurisdiction, and inferring it would err in the unsafe
direction more often than not.
Amounts will never be converted, only relabelled: conversion needs exchange rates, which needs
the network, which would end the zero-permission guarantee.
Transparency cuts both ways. These figures are estimates, and the app says so:
* in the TPC table are practice, not law. The app
shows them as benchmarks and never presents them as statutory.If you work in food science, food-safety regulation, or thermal engineering and want to scrutinise any figure here, that scrutiny is welcome — it is exactly how the app’s worst bug (the thermal model above) was caught. Open an issue on this repository or get in touch.