# Erratum — the reckoning on the electricity conversion factor **Press pack concerned**: *The reckoning on the electricity conversion factor* (OneDpe), released on **25 August 2026** at 8 a.m. **Erratum issued 19 September 2026.** It does not replace the pack: it states what, in the pack as it was sent, is no longer accurate, what was changed in it after sending, and since when. --- ## Why this erratum exists The **order of 19 August 2026** — the one that moves the electricity conversion factor from 1.9 to 1.7 on 1 January 2027 — was **signed on 19 August** and **published in the *Journal officiel*, France's official gazette, on 26 August 2026** (JO no. 0198, text 28 of 90). The pack was released on **25 August**, that is **the day before that publication**. It was therefore written and sent while the text was signed but not yet published, and it said so. ⚠️ **The pack was changed after sending, three times on 26 August 2026** (10:04, 10:10 and 12:09). The **press release** carries a visible update note at the top: that one was not done silently. The **method note**, the **figures** and the **data file**, however, were changed without notice. That should not have been done that way: a text already sent is corrected by a dated erratum. A newsroom comparing its 25 August copy with the online version will find differences, and this erratum exists so that it knows which. **The 26 August pass was moreover partial**: the entry corrected at point 1 below was not updated. --- ## 1. "The most recent text amending the conversion factor" — inaccurate since 19 August 2026 **Where**: method note, "Texts" section, entry for the order of 13 August 2025. **What the pack says**: > **Order of 13 August 2025**, NOR **ATDL2519132A**, *JORF* no. 0197 of 26 August 2025, text no. 9 — > the most recent text amending the conversion factor (2.3 → 1.9). Consulted on 17 August 2026. **What should be read**: the order of 13 August 2025 is **the text of the 2.3 → 1.9 step**. It is no longer the most recent to amend the factor: the **order of 19 August 2026** amends it again (1.9 → 1.7 on 1 January 2027) and **its article 5 repeals** the order of 13 August 2025. ⚠️ The same section of the pack already names the order of 19 August 2026 and its article 5 correctly, four lines below. **Within that entry**, the superlative "most recent" is the only point left at 25 August. **Effect on the figures: none.** The pack measures three named scales — 2.3, 1.9 and 1.7 — and none of its measurements depends on which one is the latest. --- ## 2. The regulatory watch is dated 15 August 2026, and its finding changed on 26 August **Where**: method note, end of the regulatory watch section. **What the pack says**: > ⏳ **This section is perishable.** The watch covers Légifrance alone; the last *Journal officiel* > swept is no. 0190 of 15 August 2026. **The "not published" status of the 1.7 order replays in one > query** — the consolidated version as at 1 January 2027 of the order of 31 March 2021 — and must be > replayed on the morning of any pick-up. **What should be read**: that "not published" status **ceased to be true on 26 August 2026**. The order appeared that day in JO no. 0198, text 28 of 90. The section announced its own expiry and said how to check it: it is that very check which, replayed, now returns a different result. **Effect on the figures: none.** A text's publication status does not change a count of ratings. --- ## 3. A clarification, not a correction — what Légifrance serves today **Where**: method note, "Texts" section, entry for the order of 31 March 2021. The pack states that the consolidated version as at 1 January 2027 served by Légifrance (`JORFTEXT000043353335`), **consulted on 17 August 2026**, carried **"+ 1.9"**. **That is accurate, and the consultation date is kept**: it is the fact as it was observed. ⚠️ But anyone replaying that consultation **today** reads **1.7** at that same date of 1 January 2027, since the order of 19 August 2026 substituted that value. The consultation of 17 August 2026 predates the signature of that order by two days. This point is added so that a checker does not conclude we made an error where there is only a gap between dates. --- ## 4. ⛔ THE DATA FILE CHANGED SCHEMA ON 26 AUGUST — a script written on the 25 August copy breaks **Where**: `donnees/coefficient-electricite-departments.en.csv` and its French twin `…-departements.fr.csv`. This is the most important point of this erratum for anyone who has **automated** a read of the file, because it breaks **with no error message**. Two changes, only one of which had been announced: | | 25 August 2026 (18 columns) | since 26 August 2026 (19 columns) | |---|---|---| | column 8 | `sorties_pas_2_conditionnel` | **`sorties_pas_2_2027`** — *renamed* | | column 12 | `rang_taux_cumule_sur_95` | **`taux_pas_1_sur_passoires_2_3_pct`** — *inserted* | | columns 12 to 18 | — | **shifted by one** (12→13, …, 18→19) | - The **renaming** of column 8 follows the order's publication: the 2027 step is no longer conditional. It was **announced nowhere**. - The **insertion** of column 12 is announced in the pack (method note: "served since 26/08/2026"; release: "added on 26 August"), but the announcement does not say that it **shifts every following column**. **What to do**: read the file **by column name**, never by position. A script reading column 12 by its rank has, since 26 August, been getting a different quantity, of the same numeric type, with nothing to signal it. **Effect on the figures: none.** No value was revised; one quantity was added and one column renamed. --- ## 5. What the release's update note does not say **Where**: press release, update note at the top. **What the pack says**: > **Update of 26 August 2026.** […] **No figure has changed — only the modality.** **What should be read**: **no published figure was revised** — that is accurate, and it is what the sentence means. But **figures were added**, and the online release is no longer identical to the one of 25 August: - the **headline** went from "Nearly 400,000 EPC-assessed homes have left the energy-sieve bracket with no works — and 600,000 that nothing will release" to the same sentence extended by "**, 160,000 more on 1 January 2027**"; - the **standfirst** received figures that already appeared elsewhere in the pack (the order of 275,000 homes, range 248,000 – 294,000; "third value in two years"); - a **whole paragraph** was added ("What your readers can check for themselves"); - **figure 5** (twin departmental map) was added, and the departmental paragraph rewritten with its percentages. A newsroom that quoted the **headline** of 25 August therefore finds a different headline online. That is what "only the modality" does not let one foresee. --- ## What does not change No measurement in the pack is revised by this erratum. The following remain accurate, and verifiable from the same sources: - the headline figure and the perimeter it rests on; - the three states of the stock at the 2.3, 1.9 and 1.7 scales, and the rates that link them; - the carbon ceiling and the homes it holds, whatever the coefficient; - the transposition to the national stock, its criterion written before the measurement, and the declared limits. **The three dates of the order of 19 August 2026 do not designate the same thing**, and both the pack and this erratum distinguish them: **signature** on 19 August 2026, **publication** in the *Journal officiel* on 26 August 2026, **entry into force** on 1 January 2027. Writing "the order of 26 August" or "the order of 1 January 2027" designates a text that does not exist. --- *OneDpe erratum of 19 September 2026, concerning the press pack of 25 August 2026. Issued by OneDpe (published by RIANN SAS, Paris). Free to reuse, including commercially, with the credit "Source: OneDpe.fr".*