# Excerpt — the RfD and the uncertainty-factor tower (the factor of 100, metastasized)

*Source: `papers/lateral/epa2002_RfD-RfC-process-review.pdf` — U.S. EPA, Risk Assessment Forum, *A Review of
the Reference Dose and Reference Concentration Processes*, EPA/630/P-02/002F (Dec 2002). Grade
**[CONFIRMED-primary]**.*

The Reference Dose is the modern, official descendant of the first essay's factor of 100. A point of departure
(usually a **NOAEL**) is divided by a *product* of round tens — one 10-fold default factor for each
acknowledged gap in knowledge:

- **Interspecies UF** (×10) — animal → human extrapolation
- **Intraspecies UF** (×10) — variation among humans, "to account for variations in susceptibility within the"
  human population (the doc weighs whether the intraspecies 10 can be split or replaced by data for "the most
  susceptible subpopulation(s)")
- **Subchronic→chronic UF** (×10) — short-study → lifetime
- **LOAEL→NOAEL UF** (×10) — when no no-effect level was observed
- **Database-deficiency UF** (×10) — missing studies

The defaults multiply: 10 × 10 = 100; with a third gap, 1,000; with more, 10,000. The 2002 panel repeatedly
debates "the **10-fold default factor**" and whether each "should continue to be assessed" — i.e., the device's
own custodians treat the tens as conventions to be justified, not measurements.

**The through-line to essay one:** the practice traces explicitly to **Lehman & Fitzhugh (1954)**, "100-Fold
Margin of Safety" (**now held, register H16**), via **Dourson & Stara (1983)** (**now held, H18**) — the paper
that gave EPA its formal RfD machinery. **Correction (from the held primary):** Dourson & Stara do **not** concede
the factors are "rather arbitrary"; they argue they "**are not arbitrary as is commonly perceived**" — the paper
is a *defense* of the round tens (the giveaway being that a defense was needed because the factors are
"commonly perceived" as arbitrary). The "appears to have been rather arbitrary" wording earlier attributed here
is **not** in Dourson & Stara 1983 — **do not quote it to them**; **Barnes & Dourson (1988)** stays
unverified/[P-cite]. The other pole of the debate is now held too: **Weil (1972, H17)** "deprecates" statistical
extrapolation and recommends a safety factor (his "1/5000 for cancer") set by "scientific judgment" — the
judgment-over-statistics camp (cf. Frawley's 1981 manifesto). This EPA 2002 review is the held primary for the
**structure** of the tower (the "order of magnitude" RfD definition). Ties to `reckoningscience/sections.md` §1
(the ignorance factor) and §11 (propagation); analysis `06` §5.3.