# Crump, Hoel, Langley & Peto (1976) — "Fundamental Carcinogenic Processes and Their Implications for Low Dose Risk Assessment"

*K. S. Crump, D. G. Hoel, C. H. Langley (NIEHS) & R. Peto (Oxford), **Cancer Research** 36(9 Pt 1):2973–2979,
September 1976. PMID 975067 (no clean DOI). Local: `papers/lateral/crump-hoel-langley-peto1976_cancer-res_fundamental-carcinogenic-processes.pdf` (+`.txt`). Operator drop 2026-06-12; READ IN FULL 2026-06-13.*
**Grade: [CONFIRMED-primary].** Register **H26**. Wishlist §🔵 row 4 (the no-threshold "method bridge").

## What it is
The paper that supplied the **mechanistic justification for low-dose *linearity*** — the move that retired
Mantel-Bryan probit extrapolation and seeded EPA's linearized-multistage default. The argument: if a carcinogen
acts **additively** with already-ongoing ("spontaneous") carcinogenic processes — and humans already carry a
~20% lifetime cancer background — then **under almost any model the dose-response is linear at low dose.**

## Verbatim (with locators; `.txt` line numbers)
- (Summary) "if carcinogenesis by an external agent acts additively with any already ongoing process, then
  under almost any model the response will be linear at low dose." (ll. 19–23)
- (Intro) "An 'estimate' of risk is **as arbitrary as the interpolation scheme that produced it.**" (ll. 53–55)
- (Discussion) "essentially absolute human safety (e.g., a risk of 10⁻⁸) in general cannot be guaranteed by
  extrapolation from the results of animal carcinogenicity tests." (ll. 494–496)
- They turn **against** Mantel-Bryan probit: "our arguments suggest that in general it is not correct… thus led
  us, **reluctantly, away from Mantel's hopes** and back to the familiar ground." (ll. 444–451, 483)
- **The arbitrariness of the target number, in their own footnote 3:** "There is **no compelling reason why the
  guarantee should be of an extra risk of 10⁻⁸ or less**; if, for example, we merely wanted to guarantee an
  extra risk of less than **10⁻⁶, a few ppm might be permitted.**" (ll. 507–512)
- The radiation precedent, named explicitly: linear extrapolation should be "publicly agreed for such
  substances, **as it was for radiation 20 years ago.**" (ll. 460–462) → 1976 − 20 ≈ 1954–56 = the NCRP/NBS
  Handbook 59 "acceptable risk" move (register **H29**); the deeper single-hit root is Lea 1946 (**H31**).

## Why it matters
1. It is the hinge from Mantel-Bryan's *probit* "virtually safe dose" to the modern **linear no-threshold**
   default — the technical reason regulators stopped looking for a threshold and started picking an
   acceptable-risk *number*. Ties to Rodricks (H1), Crump 2018 (H2), Mantel-Bryan (H11).
2. **Footnote 3 is load-bearing for the numerology thesis:** the authors say outright that 10⁻⁶ vs 10⁻⁸ is a
   *policy choice with no compelling scientific basis* — corroborating Kelly (H12: "pulled it out of a hat")
   and the SOM rule's own admission (H25). The slope is biology; the cutoff is numerology.
3. Uses **vinyl chloride** as its worked example of a low-background (high-p) carcinogen (ll. 333–337) — ties to
   `08_VINYL_CHLORIDE_CAMPAIGN.md`.