# FDA "Sensitivity of the Method" final rule — 52 FR 49572 (31 Dec 1987)

*"Sponsored Compounds in Food-Producing Animals; Criteria and Procedures for Evaluating the Safety of
Carcinogenic Residues" — final rule, **52 FR 49572–49590** (31 Dec 1987); codified 21 CFR Part 500, Subpart E
(§§ 500.80–500.92) + conforming amendments to Parts 70, 514, 571. Docket 77N-0026; effective 29 Feb 1988.
Local: full FR issue PDF `papers/lateral/FDA_SOM-rule_52FR49572_1987-12-31.pdf` (+`.txt`; rule begins at txt
l. 20693). Operator drop 2026-06-12; READ IN FULL 2026-06-13.*
**Grade: [CONFIRMED-primary].** Register **H25**. Wishlist 🔴 row 2. The carcinogen-side 10⁻⁶, *inside* Delaney.

## What it is
The rule that put a hard **"1 in 1 million" lifetime-cancer number** into food law. It implements the **DES
Proviso** (the carcinogen exception written into all three Delaney clauses — §§ 409(c)(3)(A), 512(d)(1)(H),
706(b)(5)(B)) by giving "**no residue**" an *operational* definition: the residue concentration corresponding
to a 10⁻⁶ test-animal risk **is** "no residue."

## Verbatim (with `.txt` line locators)
- The number: the procedures determine "the concentration of residue of a carcinogenic compound that presents
  an insignificant risk… That concentration corresponds to **a maximum lifetime risk of cancer to the test
  animal on the order of 1 in 1 million.**" (ll. 20700–20710)
- **The legal theory is *not* de minimis** — a crucial distinction: "This legal theory… **differs from the
  application to the Delaney Clause of the *de minimis* doctrine**, i.e., that the law does not concern itself
  with trifling matters." FDA reaches the same number through "an **operational definition of 'no residue'**"
  instead. (ll. 20780–20786)  *(So the carcinogen-side 10⁻⁶ rides into Delaney on two distinct legal vehicles:
  this "operational no residue," and — for color additives — the de minimis doctrine struck down in*
  Public Citizen v. Young, 1987 *(file 10).)*
- The extrapolation engine: a "statistical extrapolation procedure" = **linearized-multistage / Gaylor-Kodell**,
  linear from the **upper 95% confidence limit at ED01 to zero** (ll. 20846–20852, 21428–21494).
- **The round number is openly a chosen consensus, not a derivation:** "there was a **clear consensus that
  risks on the order of 1 in 1 million were insignificant**, though the same degree of consensus did not obtain
  for potentially higher levels" — and commenters noted "**higher levels (e.g., 1 in 100,000 or 1 in 10,000)
  might also present insignificant risks.**" (ll. 21206–21228)
- One comment: the actual risk "is probably **some place between 1 in 1 million and a much lower indeterminable
  level**." (ll. 21244–21251)
- **The number is engineered to thread a needle** — FDA's two criteria for the permitted level: "(1) It must
  not significantly increase the human cancer risk and (2) **it must be high enough to permit the use of
  carcinogenic animal drugs, food** [additives, and color additives]." (ll. 21257–21265) — low enough to call
  "safe," high enough to keep the product legal.
- The "constituents/impurities" companion device: impurities are "treated **not as food or color additives**
  but as constituents… **not subject to the Delaney Clause**" (47 FR 14464). (ll. 21022–21030)
- Risk-assessment / risk-management split (the procedural, CCERP-derived frame): "the seemingly precise numbers
  derived from a risk assessment are **in part the product of reasonable scientific and policy judgments.**"
  (ll. 21020, 20953) Notes "multiple conservatisms contained in the procedures." (l. 20955)

## Why it matters
- This is **Frawley's "toxicologically insignificant level," metastasized into the carcinogen domain** and given
  a federal-rule number. Same logical structure (a round threshold of insignificance), now 10⁻⁶ instead of
  0.1 ppm, defended with the same over-conservatism language.
- Closes the chain Rodricks (H1) describes: Mantel-Bryan 10⁻⁸ (H11) → **Hutt's 10⁻⁶** → **this rule** → Rulis
  1992 (H13) → the 1995 Threshold of Regulation (D6/H10). The SOM rule is where 10⁻⁶ became *operative law*.
- The "differs from de minimis" passage is a fine analytic point for `10_DE_MINIMIS_LEGAL_LINEAGE.md`: FDA used
  *two* legal theories to install the same numerological device, hedging against exactly the Delaney-wall
  collision that *Public Citizen v. Young* (1987) and *Les v. Reilly* (1992) delivered to the de minimis route.