# Referee-checklist results (2026-07-19)

## 0. RCW117 ADDED (from item-1 follow-up)
G345.407-0.952 = RCW117/IRAS17059-4132: CORNISH-S UCHII 384mJy + MMB 6.7GHz
maser at same position + ATLASGAL 138Jy clump; vlsr=-21 (NH3/CS/10 clumps);
d = 1.4 (Urquhart arm-prior) / 2.2 (near kin, ADOPTED) / 2.6 (WISE); far 13.9
excluded (z, optical, extinction); DECaPS rise at 2.6-3.0, none at 1.4; no
maser parallax (southern gap). L_SED = 8.9e4/2.2e5/3.1e5 @ 1.4/2.2/2.6.
Added to curated_nearby_sample (kinematic-unvetted flag). Samples 47/101.
NEW LADDER: numerator <3kpc = 10; joint-Poisson rho = 61 [44-87] 68%,
[33-129] 95%; tau_emb = 70 [49-98] kyr. Inside previous interval.
AUDIT HOLE FOUND: RCW/Sh entries with NO RMS counterpart were never
SED-checked (RCW117 fell through this). sweep_norms_rcwsh.py running.
ms.tex renumbering DEFERRED until sweep completes (one consistent pass).

## New reviewer rounds (Kuiper, Vazquez-Semadeni) — queued action items
Kuiper: (i) restate photon-budget via envelope-replenishment chain (disk
drains in ~1e3 yr unfed; FIR census excludes the envelopes) — airtight form;
(ii) ADD dust-Eddington paragraph (L>1e5 -> Gamma_dust >> 1, unshielded dusty
gas expelled dynamically: "none could persist if found"); (iii) post-
revelation disk-drain ~1e3 yr closes breakout-timing gap; (iv) bloating
direction = numerator inflation = conservative; (v) simulation-comparison
table (nice-to-have).
VS: (i) accelerating-SFR ensemble => tau_emb OVERestimate 1.3-2x — needs
small ensemble MC, strengthens result; (ii) REFRAME discussion: sever
reservoir-assembly clock (Myr, unconstrained, GHC/IIM both) from stellar
accretion clock (measured) — P&G attach wrong clock to luminous phase;
(iii) feedback-decoupling paragraph (clump still collapsing / star finished,
uses Wells+24 positively); (iv) cluster-Poisson variance x(1+Nbar_c) ~ +20%;
(v) active-region top-truncated IMF = conservative sign.

Running notes; feeds ms.tex revision + response-to-referees.

### DONE (2026-07-19) — Kuiper + VS folded into ms.tex
- Kuiper(ii) dust-Eddington: new para in sec:reservoir. Gamma_dust=kappa_d L/
  (4piGMc)~50 at L>1e5 -> unshielded dusty gas expelled; "none could persist."
- Kuiper(i)+(iii) envelope-drain: sec:bondi closure extended. Shielded disk
  drains on viscous time ~1e3 yr unfed -> needs replenishing envelope (=submm
  structure census excludes); same time closes breakout-timing loophole.
- Kuiper(iv) bloating: added to sec:budget conservative list (over-luminous at
  fixed mass -> enters numerator too readily = conservative); cite Hosokawa2009.
- VS(i) accelerating-SFR bias: referee_accel_sfr.py. Exponential B(t), ensemble
  MC over tau_acc 2-10 Myr, tMS 3.5-5 Myr -> bias R_obs/R_true = 1.40 [1.26-1.78]
  68%, matches Quintana OB-overdensity bound 1.3-2x. tau_emb overestimate ->
  true ~5e4 yr. Folded into sec:budget conservative para.
- VS(ii) two-clocks reframe: NEW subsection sec:twoclocks. Sever reservoir-
  assembly clock (~Myr, unconstrained, GHC=IIM, = P&G tau0) from stellar-
  accretion clock (~1e5 yr, measured). P&G attach wrong clock to luminous phase.
- VS(iii) feedback-decoupling: same subsection. Wells2024 clump-scale inflow
  ~7e-4 Msun/yr feeds reservoir clock + lower-mass members, NOT the revealed
  star -> clump inflow != evidence against short stellar clock. cite Wells2024.
- VS(iv) cluster-Poisson: sec:stats sentence. 10 members/8 complexes ->
  var x(1+<c(c-1)>/<c>)=1.4 (~20% wider); correlated-complex variant IS the
  cluster-aware bar.
- VS(v) top-truncated IMF: added to conservative list (young clusters may lack
  most-massive member -> numerator high-L suppressed = conservative).
- Skipped Kuiper(v) sim-comparison table (nice-to-have).
- Bib: added Hosokawa2009, Wells2024 (both %%CHECK). Compiles 8pp, 0 errors,
  all citations resolved.

### ALSO folded: Mdot refit with post-S140 bins (EMB=[5,5,5,3,2])
Salpeter p=2.08 [1.32-2.93]; steep p=2.33 [1.56-3.19]; A=(0.95-1.45)e-4;
constant disfavored 2.9-3.2 sigma. Updated abstract/sec:mdot/concl to
p=2.1+-0.8 (was 2.3+-0.9). Still ~3sigma, still brackets KH m^2.7.
Fixed stale tau_emb 6e4->7e4 in phase-sequence + conclusion item 2
(canonical = sec:clock 7e4 [4.9-9.8], <=1.3e5); ladder concl 1:11-13:53-71.

## 1. RMS nearby completeness (Mottram) — DONE, conservative-bounded
`referee_completeness.py`
- Compact (r<0.15pc) WISE-K entries at d<3: all 12 carry broken kinematic
  distances (0.1-0.5 kpc at inner longitudes); 8/12 are RMS sources within 20"
  (at RMS's correct distances). Zero verified missed nearby compact members.
- CORNISH-S UC HII (radio-selected, N=255) recovery by RMS: 61% overall, but
  **rises with flux: 45/48/72/81% by quartile; 88% (23/26) in brightest decile**
  = the regime any d<3kpc L>1e4 UC HII occupies. Extended sources recovered
  BETTER (76%) than compact (59%) — directly counters the "nearby extended
  missed" concern.
- The 3 missed brightest-decile: G333.1350-0.4322 = complex component of
  census member G333.1306 (26" away); G345.4070-0.9516 (IRAS 17009-4042
  region?) and G320.2340-0.2836 = candidates at UNVETTED distances.
  **TODO(Adam): vet G345.41-0.95 distance — possible census addition.**
- Worst case (both at <3kpc, L>1e5): numerator 9->11, rho->48, tau->9e4 yr —
  inside the existing 68% interval. Completeness correction <=20% on numerator.

## 2. Shell-ratio gradient decomposition (Padoan) — DONE, decisive
`referee_gradient.py`
- Raw emb/OB ratio (L>1e4) restricted to the SOLAR GALACTOCENTRIC ANNULUS
  (7.5<R_gal<9.5, R0=8.2): 0.011 / 0.008 / 0.013 / 0.105 / 1.04 for shells
  0-2/2-3/3-4/4-6/6-9 kpc. FLAT (~0.010+-0.004) to 4 kpc, then x10, x100 —
  at FIXED galactocentric radius, where no physical gradient can operate.
  Rise onset = the optical extinction wall. Selection, proven.
- Supporting: the ratio is intensive (lifetime ratio; SFR density cancels).
  Real gradient requires tau_emb varying with environment; P&G's own
  tau0 ~ 1/Sigma scaling gives factor ~2-3 ring-vs-local, not 300.

## 3. ZAMS L-m mapping sensitivity (McKee/Padoan) — DONE (no new computation)
From the existing threshold scan: tau_emb = <tMS>/rho at logL thresholds
4.65/4.8/5.0/5.2/5.4 = 100/79/64/32/15 kyr. A factor-2 (+-0.3 dex) systematic
mis-mapping of accreting-protostar luminosities (Hosokawa swelling, accretion
luminosity) therefore moves tau_emb within ~25-100 kyr — never toward Myr.
Ladder ratio never drops below 55 under any threshold placement. Add one
manuscript sentence + cite Hosokawa & Omukai (L within ~x2 of ZAMS at fixed m
for m>20).

## 4. P&G recipe applied to shells (Tan) — DONE, mechanism demonstrated
`referee_pg_recipe.py` — their Appendix-C recipe verbatim (V=2h*pi*R^2, h=39pc;
g(R) from reference ranges emb[3.5,4.5]/OB[4.0,4.6] normalized g(3)=1;
C(L)=max_R phi_g/phi_g(R0)), in L space:
- phi_g,emb(L>1e5) RISES monotonically with R (4.5 -> 11.6 kpc^-3, x2.6 from
  R=3 to 18): one g(R) per catalogue cannot describe L-dependent differential
  selection, so the residual inflation survives correction.
- The max_R completeness estimator then anchors the final density at R=18 —
  the MOST selection-inflated distance (beam-summing + ring) — by construction
  (sup of a biased sequence).
- Final corrected ratio = 0.043 = 2.5x the volume-complete 0.017. Their model
  needs 0.185; the remaining x4 lives in their G_abs-space OB side (A_G=0
  assumed for 2989/12640 stars scatters reddened bright stars faint,
  suppressing the bright denominator) + shape-weighted joint fit.
- Bottom line for ms: their own recipe, on the same data, inflates the high-L
  ratio 2.5x mechanically; corrected phi_g rising with R violates their own
  completeness logic (density should fall beyond the horizon, not rise).

## 5. Disjoint-bin Mdot(m) fit (Offner/McKee) — DONE, trend survives
`referee_mdot_fit.py` — disjoint L bins [4.65,4.8,5.0,5.2,5.4,6.2], joint
Poisson over Mdot=A(m/30)^p:
- Salpeter tail: p = 2.27 [1.47-3.17] 68%, [0.76-4.15] 95%; A=1.45e-4 @30Msun.
- Steep tail (3.35): p = 2.53 [1.71-3.44]; A=1.02e-4.
- Constant rate (p=0) disfavored: dchi2=9.1/11.0 -> ~3.0-3.3 sigma.
- Exponent nearly IMF-independent (2.3 vs 2.5) — the feared Mdot-IMF
  degeneracy is subdominant. Consistent with KH-argument m^2.7.
- ms.tex updated: abstract + sec 3.5 + conclusions now quote p=2.3+-0.9 and
  "constant disfavored at ~3 sigma" instead of the naive m^2.8 claim.

## 6. Rebuttal edits — DONE in ms.tex
- Abstract + conclusions: ">20 Msun" qualifiers; merger sentence tempered.
- New robustness paragraph in sec 3.3: CORNISH completeness (88% bright
  decile; worst case rho->48), correlated-complex variant (rho=72[49-113],
  tau=59[38-88] kyr), L-m mapping factor-2 (tau within 25-100 kyr),
  episodic-accretion duty cycle (photospheric L forbids hiding above 25 Msun).
- New P&G-section paragraphs: solar-annulus decomposition (flat 0.010 to 4
  kpc at fixed R_gal); intensive-ratio argument; their-recipe-applied result
  (0.043 = 2.5x mechanically, remainder = A_G=0 OB side); 1653-vs-1800
  resolution (live DB vs 2013 static table, low-L only).
- Keto 2002 trapped-ionization lineage + anisotropic-escape closure in 4.2.
- Merger k -> 1-4 range (2-8.6 kyr visibility), "if all events are mergers",
  order-of-magnitude framing.