Identity of the −2 σ generator — the bit is a persistence threshold; the biological candidate is origin jitter, not ρ nor buffer failure; ReSTAGE persistence is not estimable in any public (L, FP, LP) of 36–44 y

1. Executive scientific abstract

M1-2 does not reopen the M1-1 map. The STRAW+10 −3→−2 crossing is already a consecutive-variance operator: first persistent |L_t − L_{t−1}| ≥ 7 (recurrence ≤10 cycles). The glued ternary in the question —(A) selection jitter with material ρ=Var_w(LP)/Var_w(L), (B) a luteal/uterine residual with stable selection, (C) a tautological SD(L) threshold that adds no ovarian node— soldered a generator, a metrological gate and an already-dead ovarian prohibition. This report splits them and freezes the factual base that REPORT_ES locked.

ρ is not the clock. It is gate G0-IDENTITY. Henry, Shirin, Goshtasebi and Prior 2024 (PMID 39320898; n=53, 21–41 y, mean 33.9, ≥8 cycles, mean 13, 676 ovulatory) and Cole, Ladner and Byrn 2009 (PMID 18433748; Var_w n=31, k=5, 18–36 y) show a material luteal residual (ρ̂≈0.97 and 0.93) in mid-reproductive life. That kills the global identity SD(L)≡SD(FP) and does not crown H2L as the generator of the bit. Henry’s 2.4% outside 21–36 d (body; abstract «98%») is band occupancy, not the consecutive operator: a woman can live in 22/30/23/31 and fire |ΔL|=8 with every cycle «normal». Fehring, Schneider and Raviele 2006 (PMID 16700687; n=141, mean 29.0 y, mean 5.2 cycles) tabulates a >7 d within-woman range in 42.5% of women, not consecutive jumps nor ReSTAGE persistence. The abstract’s 25% is the fertile window on days 10–17; the body’s is mean |Δ| LP/cycle. Neither is ρ.

Adversarial review split the conjunction that hypothesis-generation had inscribed. M1-2-HG-H1 v1 required the bit to fire when Cov(FP,LP)<0 failed. That predicate is an unmeasured non sequitur in −3. Fehring r(L,FP)=0.829≫r(L,LP)=0.255 is algebra of L=FP+LP when Var(FP)>Var(LP), not proof of a buffer whose collapse fires −2. The buffer is only r(FP,LP)=−0.323. Henry 6 vs 8 SOD (Var FP 2.4 vs 15.3 and Var LP 1.4 vs 13.9 without a difference in L) is polarity against the trigger: maximum phase jitter with L in band. M1-2-HG-H1 v2 keeps the biological candidate —jitter of the origin of the dominant’s growth segment, anchored to the LH surge, in −3 by bleeding regularity alone— without a buffer-failure predicate. Robertson, Lee and Baerwald 2021 (PMID 33855196) is not a −3 anchor: LRG is 48.7±1.37 y, one IOI, and the authors read an extra wave.

The operational contrary is M1-2-HG-H3 v2: the bit is a persistence threshold on SD(L). Huang, Elliott and Harlow 2014 (PMID 24729638; TREMIN n=617, 95 246 segments) delivers population Λ at age 35 (α0=3.313, β0=−4.795 ⇒ SD(L)≈2.5 d). iid mean-field gives P(|ΔL|≥7)≈0.05 for one consecutive pair. That is not «of the order» of Fehring’s 42.5% (1−0.95^{4.2}≈0.19). Ecochard, Stanford, Fehring et al. 2024 (PMID 38598621, PMC11006216, body reconfirmed this turn): ACF lag-1 not rejected (Europe P=0.9480; NA P=0.3482); Europe lags 2–4 yes (P<0.0001 / 0.0004 / 0.0024). Mean-field specifies the isolated jump; not persistence. diag(Ω) absent: D0-OPERATOR-MIX not executable. Clause (vi) of M1-1-HG-H1 v1 —«therefore there is no node of the process»— remains withdrawn.

Experimental design converts that metrology into a desk protocol with four non-interchangeable operators. Band ≠ range ≠ isolated consecutive ≠ ReSTAGE persistence (k≥11). Zero public (L, FP, LP) archive in regular 36–44 y estimates persistence: Cole k=5 and Fehring mean 5.2 fail the floor; Henry mean 13 would pass it in a subset of the wrong window, QBT, prescreening, operator not tabulated, data on request. Fehring >35 y (cycle 27.9 SD 2.6; n=24) sits below the n≥30 floor for the IC90% of ρ vs 0.20: the allowed informative failure is inconclusive, not a null of the bit. D1 stays closed (six cycles estimate isolated jumps and fail persistence). Freeze (Liu 2026: n=15, cessation 32 vs 36 weeks, growing follicles ↓) and do(TGFBR3L) (Brûlé: larger litter) remain illicit. G8 closed. No prescription. No Lua data.

Scientific delta of this report (M1-2-REPORT-EN-C1): the English twin freezes the bilingual factual base that REPORT_ES locked, after adversarial-review and experiment-design split the hypothesis-generation and synthesis conjunctions, and after the recorded bilingual-QA failure (headings lacked the substrings mecanismo/mechanism; numeric bags diverged on CI wording, a repeated percent, and YAML extras). Both drafts now carry those substrings in headings; the Spanish twin is heading-corrected only. The operational hypotheses are M1-2-HG-H3 v2 (wins the bit), M1-2-HG-H1 v2 (biological candidate without a buffer trigger), M1-2-HG-H2L v2 (seed, not title) and M1-2-HG-HT1 v2 (parked). Ecochard 2024 Table 2 is reconfirmed this turn. Four operators do not vote for one another. ReSTAGE persistence in 36–44 y with phases is not estimable in public. D0-IDENTITY is the smallest discriminant and is now a four-operator protocol, leadership vs Cov, and QBT vs LH. D1 closed. (vi) is not restored.

2. Scientific question and relevance

The 1.2 question is not whether a first licit node exists for delaying peri —M1-1 mapped that and left the generator open—. It is who generates the σ that the bit reads in regular −3 women. Answering it requires separating objects that the title, the glued ternary and several intermediate stages fuse:

1. Is the −2 bit the classifier or the generator? Fusing them rehabilitates (vi).
2. Is ρ a clock or an index gate? Materiality frees SD(FP)-menses as a candidate; identity forbids it. Neither votes ovary.
3. Are the 21–36 d band, max−min range >7 d, isolated consecutive |ΔL| and persistence ≤10 cycles the same estimand? No. Henry 97.6% and Fehring 42.5% do not vote the bit.
4. Is leadership r(L,FP)≫r(L,LP) a Cov(FP,LP)<0 buffer? No. The first is partition algebra; the second is compensation. HG soldered them; AR splits them.
5. Is the trigger «jitter when the buffer fails»? That conjunction is withdrawn. Cov collapse is a separable link (H2L vs dating artefact vs H1), unmeasured in −3.
6. Is a location shift of selection (Klein 10.7 vs 11.0 d) within-woman jitter? No. One cycle estimates a mean, not SD(day | LH).
7. Is Robertson LRG 48.7 y architecture of −3? No. It is ARA, one IOI, extra wave.
8. Does a mean-field quantile P≈0.05 explain Fehring 42.5% or ReSTAGE persistence? No. Range ≠ consecutive; lags 2–4 require mixture/Ω.
9. Is delaying the bit or FMP success? No. The coprimary is competent ovulatory months. G8 closed.
10. Are freeze, TGFBR3L, LPDF-ARA, AMH-census or Lua data the 1.2 node? No.

The relevance is high because «identity of the generator» is the only object that later authorizes asking do(node) without manipulating the classifier. A do validated against delaying the −2 binary without an exogenous-to-bleeding index is tautology. A freeze that lengthens the calendar by cutting growing follicles fails the fertile clause. An AMH+FSH score that predicts FMP answers another title. The correct unit remains the woman in −3 by bleeding regularity alone, approaching a non-circular crossing.

3. Scope, population and life stage

The primary biological estimand is the adult woman in natural reproductive aging, STRAW+10 written by bleeding regularity alone. The principal stratum is late reproductive (−3b/−3a), 36–44 y (LRA window): it is the only human calendar where the operator is not yet the state, H1-mean may already have moved, ovulation is conserved (Hale LRA 0/16 anovulatory) and ρ could be estimated before the crossing. The classification outcome is the −3 → −2 crossing itself. Early transition (−2) is an arrival state and a negative control. −1 and +1 are non-targets: L7-4 already works −2/−1.

Mid-reproductive 21–41 y (Henry) and 18–36 y (Cole) are transport controls, not the n of −3: they kill global identity and do not close ρ in the window. ARA 45–55 / EMT / LMT are non-targets of identity: Vanden Brink ARA arrives late; Robertson LRG 48.7 y lives there; L7-4 describes luteal loss inside −2/−1.

Inferential unit: the woman. Cycles, follicles, Huang segments and embeddings are not n. One cycle per woman = uninformative for Var_w, not a null. Between-woman SEM ≠ within-woman SD. A Lua-user arm does not exist.

Out of scope: pregnancy, lactation, mixed PCOS, endometriosis, POI, ovarian cancer, oophorectomy, iatrogenic gonadal suppression; current Lua data, PHI, PII, private genomes; prescription of AMH, calcitriol, senolytics, NMN/NR or a non-indicated ovarian biopsy; FMP or a symptom as onset; p>0.05 as equivalence; Henry or Cole as ρ of −3; Robertson LRG as a −3 anchor; Fehring 42.5% as P(|ΔL|≥7) or ReSTAGE; Henry 97.6% as a null of the bit; r(L,FP)≫r(L,LP) as proof of Cov; mean-field as closure of persistence; (vi); LPDF-ARA as first; inventing diag(Ω).

This project does not gavage, does not dose, does not recruit.

4. Background and mechanism map

4.1 Operational definitions

• Onset of perimenopause: the −3 → −2 crossing under STRAW+10: persistent difference ≥7 d in consecutive cycle lengths, persistence = recurrence ≤10 cycles (Harlow 2012, PMID 22344196). It is not −3a. It is not −1. It is not FMP. It is not «around 45». It is not the 21–36 d band. It is not max−min range.
• Four operators of «L variance» (ED, not interchangeable): (i) 21–36 d band occupancy; (ii) within-woman max−min range >7 d; (iii) isolated consecutive |ΔL|; (iv) ReSTAGE persistence, floor k≥11.
• −3a machine / H1-mean: fewer preantral–small-antral cohort → dimeric inhibin B ↓ → earlier and higher FSH → advanced selection, FSH-peak→LH slope conserved → shorter FP. Uniform location leaves |ΔL|≈0.
• σ generator / H1 v2: variance of the origin of the dominant’s growth segment (the day an already-growing antral reaches the dominance threshold), anchored to the LH surge, not to menses. Does not require LPDF-ARA. Does not require Cov(FP,LP) failure.
• ρ: within-woman Var_w(LP)/Var_w(L). Gate, not generator. Materiality: IC90% ≥0.20. Identity: IC90% <0.20. CI covering 0.20 = inconclusive.
• Leadership vs buffer: leadership = within-woman r_w(L,FP) vs r_w(L,LP) (Fisher-z). Buffer = r_w(FP,LP) and Var_w(L)?<Var_w(FP). Not soldered.
• Inhibin B: INHA–INHBB heterodimer of preantral/small-antral granulosa. Reader of mass. Not a product of FSH (Welt 2001). Not called a biomarker.
• AMH: product of growing granulosa. Absent in human primordials (Weenen n=12). Not a census. Not called a biomarker.
• Fertile function: months of competent ovulatory cycles. Not high AMH. Not late FMP. Not delaying the bit.
• Observation ≠ do(node). Mapping an order does not authorize dosing.

4.2 Chain the literature would have to demonstrate

Preantral / small-antral cohort (not the primordial not encoded by AMH)
  → dimeric inhibin B ↓                         [−3a machine]
  → gonadotrope reads via TGFBR3L
  → earlier/higher FSH
  → ADVANCED selection, slope conserved         [Klein 2002: 10.7 vs 11.0 d]
  → shorter FP = STRAW −3a                      [location ⇒ |ΔL|≈0]

Origin noise (σ candidate; H0 of the anchor)
  antral N jitters cycle-to-cycle               [Hansen/Scheffer; I to −3]
  origin of the segment | LH jitters            [ABSENT in −3]
  → FP jitter
  → L jitter  (partition algebra; buffer is confounder, NOT trigger)

Classifier
  persistent |ΔL|≥7 = −2 bit                    [H3 v2 wins the bit]
  mean-field: P(isolated jump)≈0.05             [Huang Λ; Ecochard lag-1 not rejected]
  persistence ≤10 cycles: second threshold      [requires Ω; not executable]

do(node) that lowers that σ
  WITHOUT cutting competent months              [G4]
  WITHOUT displacing organ harm                 [G5, G8]
  WITHOUT manipulating the classifier           [tautology]

Today no human link anchors origin jitter to the first non-circular −2. Within-woman SD(day | LH) vs the crossing: absent. ρ in 36–44 y: absent. ReSTAGE persistence with phases in that window: not estimable. Primordial freeze attacks the inlet valve and cuts throughput. Human TGFBR3L occupancy in −3: absent; the female KO raises litter size.

5. Evidence method

Unit: the woman. Classes: H human, A animal, V in vitro/ex vivo, C computational, I inferred, R review (does not vote causality). Polarity is to the σ generator of the persistent operator in −3, not to FMP nor to −3a. p>0.05 is not equivalence, a null of H2L nor identity. FMP is not −2. AMH is not a census. PdG is not ovulation. Pre/post is not STRAW. Observation is not do. Supplement hits, SEO, Farr/KEEPS as peri, senolytics, NAD, S1 indices and any user cohort do not vote.

Prior stages audited the map (2026-10-20), reread bodies of Henry, Huang, Fehring, Cole, Paramsothy, Robertson and Ecochard, corrected Huang Λ, Fehring 25%, Cole n, Klein 1996≠Welt, and split the conjunctions in AR (2026-10-21) and ED (2026-10-21). Compute: decision: skip (2026-10-20). This turn reconfirms Ecochard 2024 Table 2 (PMC11006216) and does not reopen GEO nor request microdata. Intercepts of Ω are not invented. Henry or Cole are not transported as ρ of −3.

6. Evidence map

7. Contradictory evidence and null findings

8. Multiscale mechanism synthesis

8.1 Molecular — two machines, one buffer that does not fire

H1-mean (direct between women at approach, H0 for the crossing). Groome 1996: dimeric assay, early-FP peak. Welt 2001: preantral/small-antral source; FSH 30 ng/mL does not raise inhibin B. Klein 2004: inhibin A and activin A do not mediate the intercycle FSH of 40–45 ovulatory women. Lewis/Li: betaglycan reads inhibin A; Brûlé: TGFBR3L reads inhibin B. Klein 2002: short FP = advanced selection, not acceleration (slope 10.7 vs 11.0 d, P=0.74).

Closure of −3a: small-cohort granulosa ↓ → INHBB/inhibin B ↓ → less TGFBR3L–ACVR2–SMAD antagonism on FSHB → earlier/higher FSH → advanced selection, growth rate conserved → shorter FP. If every cycle shortens 28→24, |ΔL|≈0. The operator requires a variance generator.

Mid-reproductive buffer (architecture, not trigger). If L ≈ FP + LP, Var(L) = Var(FP) + Var(LP) + 2 Cov(FP,LP). Henry and Cole show Var_w(L)<Var_w(FP); Fehring shows r(FP,LP)=−0.323. That is compensation at 18–41 y. AR withdraws reading it as the switch of the bit: (i) follicular leadership is algebra when Var(FP)>Var(LP), even with Cov=0; (ii) Pearson of 1060 cycles mixes between-woman and within-woman; (iii) QBT injects Var(ε)≈2.25 d² into Henry’s Cov (delay 2.4±1.5 d); Cole/Fehring with LH conserve the sign, not the magnitude nor −3; (iv) Henry 6 vs 8 shows the pattern «both phases explode, L does not» with buffer intact, in SOD. The buffer remains a D0 confounder and a separable link, not a firing clause.

8.2 Cellular / follicular — origin, not slope, not LPDF, not Robertson-as-−3

The random variable of (A) is the origin of the FSH-peak→LH segment, not its velocity. Klein 2002 fixes the slope. Hansen 2003 (n=50 IVF): pooled AFC SD 3.0±0.3 — N noise exists far from −3. Order inference (MS-C2, retained as I): Var(min of n hitting times) rises when n falls and the slope is fixed. That is not the anchor to the crossing.

Robertson 2021 measures mass (inhibin B linear in AFC r=0.70) and a P=0.054 of between-woman timing, one IOI, LRG 48.7 y. The authors prefer extra wave = the LPDF-ARA family already not-first (M1-1-K5). Using it as first-step architecture in −3 recycles the degraded object (AR-C4). Vanden Brink 2013: LRA n=10, one IOI; prevalence NS; ARA kinetics arrive late. van Zonneveld 2003: recruitment from late-LP couples the prior CL as a seed of (A), one cycle, possible −3/−2 mix.

8.3 System — classifier, four operators, uterus

STRAW −2 writes the bit on the endometrium. Huang quantifies σ-before-mean in TREMIN 1934–1939. ReSTAGE isolates persistent >6 d and names ovarian activity. None observes inhibin B, ovulation or selection day relative to LH. Paramsothy links long menses to fibroids; the consecutive-difference operator is not the endpoint. Burger admits endometrial FMP and, at the same time, a follicular generator of erratic cycles (clause 3 feeds H1 v2, not (vi)).

The ED correction is metrological and changes the synthesis: treating band occupancy or range as the bit manufactures a classifier, not a generator. ReSTAGE persistence requires k≥11. Zero public (L, FP, LP) body in 36–44 y delivers it.

8.4 Direct versus inferred

Favoured chain provisional, not assumed, post-AR:

Already-small antral N, cycle-to-cycle noisy (−3a mass machine) → origin of the FSH-peak→LH segment jitters, slope conserved → SD(selection | LH) [ABSENT] → FP jitter → L jitter by partition algebra → the classifier thresholds persistence on SD(L). ρ is measured before calling SD(FP)-menses exogenous. The Cov<0 buffer is mid-reproductive architecture and a dating confounder, not the switch. The bit can be tautological given SD_10(L)+age without the process lacking an ovary.

9. Computational layer

Decision: skip. Retained (2026-10-20T16:08Z). NCBI GEO: peri/STRAW × (inhibin|INHBB|AMH) × omics = 0; peri × pituitary × omics = 0; TGFBR3L = GSE198907 mouse n=4; LPDF phrase = bovine IVF. Henry, Cole and Robertson: data on request or laboratory, not anonymous deposit. Najmabadi 2020 measures max−min range with mucus peak, mean 27.1 y, not Var_w in 36–44 y. Huang Λ permits mean-field; absent diag(Ω) blocks MIX. TREMIN is not FASTQ. BioNeMo does not observe |ΔL|, Cov(FP,LP) or selection|LH. BIMORA/ICPSR 4452: no inhibin B, AMH or US. BioCycle (1–2 cycles) does not estimate Var_w. SWAN DHS: DUA and age 42–52 y, not LRA 36–44 as first.

D0-OPERATOR-MEANFIELD is arithmetic on already-published Λ (P≈0.048 if iid), not a FASTQ nor a DUA. Discriminant: calibration of the isolated jump. Ecochard does not reject lag-1: the specification survives. Not run as a gate: it does not close persistence, does not observe ovary, does not split H1 from H3. Comparing it with Fehring 42.5% is the withdrawn mapping. Inventing diag(Ω) is a protocol failure. Apple WHS / app calendars that estimate SD(L) by age without FP/LP or STRAW do not discriminate. Lua data invalidate.

10. Primary hypothesis

M1-2-HG-H3 v2 — the −2 bit is a persistence threshold on SD(L); mean-field specifies isolated jumps, not Fehring-range nor persistence under mixture; manipulating the classifier without an exogenous is tautology; the generating process remains open

Role: primary of object after adversarial review. Promotes SCOPE-H3. Inherits M1-1-HG-H1 v2 clauses (i)–(v). Does not restore (vi). Does not treat CP_σ as selection. Does not treat FMP as success. Does not map P≈0.05 to Fehring 42.5%.

Falsifiable statement: in natural reproductive aging, (i) the operational object «when perimenopause begins» under STRAW+10 is the first persistent |L_t−L_{t−1}|≥7 (recurrence ≤10 cycles), a consecutive-variance operator on the same L_t process Huang modelled; (ii) H1-mean produces −3a and does not fire −2 if shortening is uniform; (iii) Huang 2014 Table 1 delivers population Λ at age 35 (α0=3.313 ⇒ mean exp(α0)≈27.5 d; β0=−4.795 ⇒ σ_log≈0.091 ⇒ SD(L)≈2.5 d); if consecutive cycles were independent, P(|ΔL|≥7)≈0.05 for one pair — Ecochard 2024 does not reject ACF lag-1, so that isolated-jump specification survives; (iv) Fehring 42.5% is within-woman range over mean 5.2 cycles, not that quantile, and persistence ≤10 cycles is a second threshold that requires mixture/Ω (European lags 2–4), not mean-field; (v) conditioning the binary on SD_10(L)+age collapses its classifier information (decision margin ΔAUROC 0.02, not a measured result); (vi) withdrawn both the prohibition of any node of the process and the mapping Fehring 42.5% ≡ 0.05 quantile. An exogenous index (selection | LH or dimeric inhibin B) whose leadership survives conditioning on SD(L) restricts this primary to the bit and leaves HG-H1 v2 alive as the biological object. Reproducing CP_σ before CP_mean kills nothing (M1-1-K2). Inventing diag(Ω) is a protocol failure.

Lineage: M1-2-SCOPE-H3 v1; M1-2-HG-H3 v1 split; M1-2-MS-C5; M1-2-EV-C1; M1-2-CD-C4; M1-2-AR-C5; M1-2-ED-C4; M1-1-HG-H1 v2; M1-1-K1; M1-1-K2. Harlow PMID 22344196. Huang PMID 24729638. Ecochard PMID 38598621. Fehring PMID 16700687. Burger PMID 18574431.

Mechanism: the bit is a function of consecutive differences. SD(L) is the second moment of the same process. Conditioning the bit on recent SD(L) is nearly identity — and that does not vote an upstream ovary. Mean-field of Λ writes the isolated jump; ReSTAGE persistence is 2–10-cycle memory that Ω does not let us simulate.

Predictions:

1. In any cohort with daily bleeding and without ultrasound/inhibin B, the first persistent |ΔL|≥7 will not add material AUROC (IC90% of the increment <0.02) to SD_10(L)+age under the mixture. If it adds, (v) is weakened — and it still is not an ovary until an exogenous enters. Mean-field does not close this predicate.
2. Re-estimating Huang (if a DUA existed) will reproduce CP_σ before CP_mean. That strengthens (iii); it does not kill HG-H1 v2.
3. Isolated consecutive compatible with 1−(1−p)^{k−1} (p≈0.05) and persistence not estimable → this primary wins the isolated jump; the second threshold stays open.
4. Band or range, without consecutive, do not vote this primary nor kill it.
5. Burger is retained: a one-visit FSH/E2/inhibin B panel does not classify early vs late better than bleeding itself.
6. G8 remains closed. Delaying CP_σ or FMP is not a success prediction.
7. Lua data invalidate.

Evidence for: Harlow 2012 (definition); ReSTAGE 2008 (only separable early); Huang 2014 (Λ verified, CP_σ); Ecochard 2024 lag-1 not rejected (reconfirmed this turn); Fehring 42.5% as proof that the isolated threshold is not the crossing (range, young); Burger clauses 1–2; S1-K3 (residualising the classifier on the process that defines it is circular).

Evidence against: ReSTAGE names ovarian activity (M1-1-K3); Burger clause 3 feeds H1 v2 (M1-1-K4); Henry shows that SD(L) does not exhaust phase architecture; Huang anchors age/FMP, missingness 512/617, population CP ≠ index cycle; diag(Ω) absent: hierarchical persistence not demonstrated as a quantile; a selection | LH node can add information to the bit after SD(L) — then this primary restricts to «the binary was a threshold», which is exactly the written kill.

Kill criteria: an exogenous-to-bleeding node whose change point precedes the first non-circular −2 after conditioning on SD(L) and common cause → restricted to the bit; HG-H1 v2 wins the biological object. Restoring (vi) or inventing diag(Ω) invalidates the call. Treating MEANFIELD as closure of persistence invalidates the call.

Status: revised_competing_contrary in HG/AR; in this report, primary of object. Maturity: H1 of literature for the operator and for CP_σ-before-mean; H0 for ΔAUROC under mixture; H0 for the Fehring mapping (withdrawn). Confidence: 0.80 that the bit is a σ threshold; 0.34 that it does not add material AUROC under the mixture; 0.08 that mean-field explains Fehring’s 42.5% (residual).

11. Competing hypothesis

M1-2-HG-H1 v2 — the first candidate biological generator of that σ is jitter of the origin of the dominant’s growth segment in regular −3, with ovulation and P4 AUC stable, without requiring LPDF-ARA or Cov(FP,LP) failure

Role: competitor. Primary of the candidate generator, no longer of the joint trigger. Contrary to reading Huang/ReSTAGE/Ecochard as closure of the generator. Promotes SCOPE-H1 without clause (iv) of v1. Refines M1-1-HG-H2 v2. Absorbs MS-C2 (origin, not slope). Cedes the buffer as architecture/confounder.

Falsifiable statement: in −3b/−3a women defined only by bleeding regularity (36–44 y), (i) growing mass has already fallen (dimeric inhibin B linear in AFC) and that is the −3a machine — a uniform location shift of selection does not fire |ΔL|≥7; (ii) the σ generator is the variance of the origin of the dominant’s growth segment, anchored to the LH surge, not the slope of that segment and not LPDF prevalence; (iii) ρ is measured before calling SD(FP)-menses exogenous: if IC90% <0.20, that index dies and (ii) survives only with selection|LH or dimeric inhibin B; (iv) withdrawn the requirement that the bit fire by collapse of Cov(FP,LP); (v) LPDF-ARA is not (ii); (vi) Robertson LRG (48.7 y, extra wave, one IOI) is not the anchor of (ii) in −3. Calling Fehring leadership «therefore there is a buffer that fires» violates the statement and kills this hypothesis by category. Restoring «jitter with material ρ» violates MS-C1.

Lineage: M1-2-SCOPE-H1 v1; M1-2-HG-H1 v1 split; M1-2-MS-C2; M1-2-AR-C1; M1-2-AR-C4; M1-1-HG-H2 v2; M1-1-K5; Klein PMID 12466381; Fehring PMID 16700687; Welt PMID 11232020.

Kill: SD(day | LH) or dimeric inhibin B does not lead SD(|ΔL|) at equalised ovulation/P4, or the leadership disappears on conditioning SD(L) → H3 v2 wins the bit; H2L inherits if LP/P4 do lead. Using illicit SD(FP)-menses invalidates the call, does not kill (ii) until the non-collinear fails. P4/ovulation fall before selection → ceases to be first. Restoring the buffer trigger or «jitter with material ρ» = violation.

Status: revised_primary of the generator. Maturity: H1 of literature for mass/−3a and for follicular leadership of L in mid-reproductive; H0 for the anchor to first −2; H0 as do(node). Confidence: 0.36 in first-step as candidate; 0.14 that that process is anchored to first −2.

M1-2-HG-H2L v2 — named luteal rival: seed, not title

Role: not the titular contrary (that is H3 v2). Luteal residual / SOD seeds phase noise or the origin of (A); does not write the bit while follicular leadership is conserved.

Falsifiable statement: at comparable SD(day | LH), (i) Var_w(LP) or a fall of P4 AUC / ovulation leads SD(|ΔL|); (ii) the prior CL / SOD can seed partition jitter without writing L (Henry 6 vs 8 pattern); (iii) van Zonneveld couples late-LP as a seed of (A); (iv) Visby Madsen is cited neither as a null nor as proof. H2U (Paramsothy duration; Jacobson T4→PdG) is not promoted: D0-UTERUS confounder, absent on the operator.

Kill: selection | LH leads at equalised LP/P4 → H1 v2 wins. ρ IC90% <0.20 and stable P4/ovulation → the title falls (does not prove C-ban).

Status: revised_competing_luteal. Maturity: H1 of existence of the residual; H0 of leadership of the operator in −3. Confidence: 0.28 that LP/P4 lead the bit with stable selection; 0.62 that the prior CL/SOD seeds phase noise or the origin of (A).

12. Translational hypothesis

M1-2-HG-HT1 v2 — freeze and do(TGFBR3L) are not the 1.2 node; if HG-H1 v2 delivers a non-collinear index, the research stick is regularity of the already-growing cohort, scored in competent ovulatory months + SD(selection | LH), not in the −2 bit nor in FMP, not in buffer failure

Role: translational, negative with respect to the binary and the obvious levers, parked as a biomarker. Inherits SCOPE-HT1 and M1-1-HG-HT1 v2. Does not prescribe. Does not recruit. HUMAN_QA_REQUIRED. G8 closed. Not H5.

Falsifiable statement: (i) a primordial-activation perturbation (AMH, PI3K–PTEN–FOXO3, pharmacological VD3) that delays cessation cuts growing follicles and does not raise output (Liu 2026: n=15, cessation 32 vs 36 weeks, pups slightly smaller) — fails G4; (ii) do(TGFBR3L) raises folliculogenesis and litter (Brûlé 2021) — sign contrary to delaying human σ; occupancy −3 absent; dKO infertile; (iii) FSH alone does not restore inhibin B (Welt 2001) — pushing gonadotropin advances selection (−3a), does not stabilise σ; (iv) an AMH-census + FSH + −2 binary or FMP parameter will not win out-of-sample competent months against age + recent SD(L); (v) if and only if H3 v2 does not absorb the object and H1 v2 delivers a selection index exogenous to bleeding that survives conditioning on SD(L), the licit stick is reducing atresia/irregularity of the already-growing cohort while conserving monthly recruitment, scored in competent months + SD(day | LH), not in delaying −2 or FMP, not in illicit SD(FP)-menses, not in buffer failure; (vi) Ruth/G8: late ANM redistributes harm — FMP is not the validation event. No user n enters. No compound is named in this project.

Kill: freeze that conserves output + competence, without cytostasis, with named competing risks → (i) dies; not prescribed. TGFBR3L occupancy that lowers SD(day | LH) without cutting ovulations → (ii) is particularised. Marking H5, biomarker, prescription, recruiting, or a Lua dataset invalidates the call. Wrong event (FMP, symptom, bit as success, extraovarian pleiotropy) = object kill.

Status: revised_translational, parked. Maturity: H1 of literature for the freeze veto and the contrary sign of TGFBR3L; H0 for a positive lever of growing-cohort regularity. Confidence: 0.82 that freeze and do(TGFBR3L) are not the 1.2 node; 0.18 that a growing-cohort perturbation that lowers SD(day | LH) while conserving throughput exists today.

13. Falsifiable predictions and kill criteria

14. Discriminating experiment

No outcome of this project recruits. The de-risk sequence, frozen in EXPERIMENT_DESIGN and not reordered by this report:

D0-OBJECT-SPLIT   Does the archive tabulate band, range,     NO → does not vote the bit
       │           consecutive or persistence?
       ↓ yes (at least one named operator)
D0-IDENTITY       ρ, leadership and Cov in regular 36–44 y,  Absent tables → inconclusive
       │           k and n above the floor?                  Henry/Cole = transport, not −3
       ↓
D0.5-SPLIT        Does negative Cov survive an LH anchor?    Collapses with LH → Henry magnitude was dating
       │
D0-OPERATOR-MEAN  Calibration: P(|ΔL|≥7) under iid Λ         NOT a kill. Ecochard lag-1 not rejected
       │
D0-OPERATOR-MIX   Does the bit add AUROC to SD_10(L)+age     NOT EXECUTABLE (diag Ω absent)
       │           under Huang’s mixture?
       ↓ not run
D0-UTERUS         Do fibroids/TSH associate with the         Absent ≠ null
       │           consecutive operator, not menses ≥10 d?
       ↓
D0-INVENTORY      Public archive with day|LH or inhibin B    NO → D1 stays CLOSED
       │           + bleeding in regular −3?
       ↓ YES (today: no)
D1                six cycles, TVUS anchored to LH,           not opened in this project
                  dimeric inhibin B                          (k=6 does not estimate persistence)

D2-G4             in PARALLEL, preclinical of throughput     Liu 2026 already fails freeze

Margins frozen before seeing extracted n. ρ 0.20; ΔAUROC 0.02. p>0.05 is not equivalence. Lua n invalidates.

Floors of k and n (ED, not a licence to enrol). ρ and Var_w: k≥8, n≥30. Within-woman r_w (Fisher-z, not Pearson of cycles): k≥8, n≥30. Isolated consecutive: k≥5, n≥40. ReSTAGE persistence: k≥11, today not estimable. Fehring mean 5.2 and Cole k=5 fail persistence. Henry mean 13 would pass k in a 21–41 y subset, QBT, prescreening, operator not tabulated, data on request. Fehring >35 y n=24 sits below n≥30: expected result = IC90% covers 0.20 = inconclusive.

D1, if a human opened it (not this stage). Within-woman observational, STRAW −3 by bleeding regularity alone, 36–44 y. Estimand: within-woman partial R² of SD(selection day | LH) predicting SD(|ΔL|), after SD(L) and SD(LP), with ovulation and P4 AUC stable. Partial r 0.30, α 0.05, power 0.80 → n≈85 completers / 122 enrolled. Six cycles estimate isolated jumps and fail persistence; a D1-β k≥11 is not recruited here. Robertson LRG 48.7 y is not a −3 arm. Wrong event (FMP, symptom, AMH-census, bit as success) relabels the study. Inventory 2026-10-20 = 0.

This design does not demonstrate extended human fertility nor longevity. It does not prescribe.

15. Biomarkers and stratification

Inhibin B, AMH, AFC, LPDF, SD(FP), SD(L), TGFBR3L, ρ and the −2 bit are not called biomarkers.

Prespecified stratification of a future D1: age 36–39 vs 40–44; BMI; fibroids yes/no; TSH tertiles; LH vs QBT anchor. No fishing. No ethnicity as mechanism.

16. Individual variability

Huang Fig. 2: women with early CPs have ~4–5 y between σ and mean; late ones ~1–2 y. The population «3.39 y» is not each woman’s crossing. ReSTAGE: 12.4% of TREMIN have persistent >6 d and late ≥60 d within <1 y; 87.6% do not. Fires <40 y do not approximate FMP.

Henry 2024: within-woman LP variance already ranges 0.70–7.9 d at 21–41 y prescreened; 55% had >1 short luteal phase and 17% ≥1 anovulatory cycle despite two normal entry cycles. That underestimates the population (authors’ limitation) and is not transported to −3. The 6 vs 8 SOD (fragile n) shows extreme heterogeneity of both phases with L conserved: H2L seed / arrival territory, not regular −3.

Hale: anovulation 0/16 LRA vs 9/23 LMT — competence heterogeneity explodes after the crossing. Ferrell: Var_w(L) minimum 33–40 y then rises — transport contradiction versus Fehring >35 y (less SD, selected cross-section: 3–13 cycles contributed to a fertility study). Recorded; not smoothed. Vollman (via Henry): FP shortens in early peri and lengthens/becomes variable in late peri — location then arrival variance, not identity of the generator in −3.

Huang cohort: University of Minnesota students 1934–1939. Transporting 42.84 y as a universal clock is overclaim. Ecochard: lags 2–4 memory significant in Europe and weak/inconsistent in NA; lunar of opposite phase across continents. A circamonthly clock is neither crowned nor ignored as a confounder of persistence.

17. Pharma relevance and maturity

There is no regulatory use context, clinical utility, net benefit, external replication or partnering readiness. Not H5. HUMAN_QA_REQUIRED. Healthspan and longevity out (G8). No compound, dose, efficacy trial, companion diagnostic or negotiation is proposed. The stick does not wait for Cov collapse (HT1 v2).

The nearest opportunity is not a drug to «delay peri». It is, at most, a research-use-only (RUO) assay of regularity of the already-growing cohort — dimeric inhibin B + selection day relative to LH + coprimary of competent ovulatory months — conditioned on D0-IDENTITY freeing a non-collinear index and on a human D1 (not this project) showing partial R². Competing risks of late ANM (Ruth: bone/T2D vs hormone-dependent cancer) are named, not «balanced» in a compound.

18. Limitations

• Zero ρ and zero within-woman SD(selection | LH) vs first non-circular −2 in 36–44 y is not proof they do not exist; it is published absence (map, verification, compute 2026-10-20, ED 2026-10-21).
• ReSTAGE persistence with phases in that window is not estimable, not null. Declaring it null because Henry lives in band or Fehring reports range is a protocol failure.
• Huang: hierarchical population CP, missingness 512/617, MAR, FMP anchor, 1934–1939 cohort. Λ yes; diag(Ω) no. Mean-field does not close persistence.
• Hale, Klein, Welt, Vanden Brink, Robertson: cross-sectional or one IOI. Between women ≠ trajectory. Robertson LRG 48.7 y is not −3.
• Vanden Brink LRA n=10 could not declare LRA≡MRA equivalence; AR should not have read «indistinguishable» as TOST. ED converts it to TOST n≈50/group, secondary and closed.
• Henry 2024: QBT with delay 2.4±1.5 d vs LH peak; «variances» in days (SD vs days² ambiguity); prescreened ovulatory; mean 33.9 y; underestimates population SOD; data on request. Does not close ρ in −3. 97.6% is not the operator.
• Cole 2009: k=5, extraordinaries excluded, 18–36 y. Var_w unit ambiguous (variance formula, label in days). Henry fused n with Cole hCG 2009: mash-up, not a finding.
• Fehring: Pearson of cycles, not r_w; range, not consecutive; k mean 5.2; n>35 = 24. Two distinct 25%s. EV reported the >35 y cycle as 27.2 d; the body re-read in AR/ED freezes 27.9 SD 2.6. The later reread is retained.
• Ecochard: mid-reproductive, calendars, no FP/LP, no STRAW. Confounder of persistence, not a rival of the generator. Lunar does not enter the graph.
• Visby Madsen 2025: n=29, commercial device, COI, «presumed perimenopausal» by age >42, no STRAW, no operator. Underpowered null.
• Liu: exposure from 3 weeks of life, not human −3; n=15; «slightly smaller» without CI. Human crossing polarity: I.
• Brûlé: the mouse has no STRAW. Developmental KO ≠ adult occupancy.
• D0-OPERATOR-MIX was not executed: intercepts of Ω are not invented. MEANFIELD is not run as a gate.
• Lua data forbidden. This report does not recruit and does not prescribe. It is not written into memoria agentes/Lua Labs/reportes/.

19. Conclusions

The 1.2 title asks for the identity of the generator of the σ that the −2 bit reads in regular −3 women. What exists today, after splitting the conjunctions, is:

1. A persistence classifier (STRAW −2 / ReSTAGE persistent >6 d / Huang CP_σ 42.84 y) on a continuum that staging itself already declared in motion at −3a. M1-2-HG-H3 v2 wins that object. Huang mean-field specifies isolated jumps (P≈0.05; Ecochard does not reject lag-1). It does not specify persistence (Ω absent; lags 2–4). The mapping to Fehring 42.5% is withdrawn. The clause that forbade any node of the process remains withdrawn.
2. A biological candidate —jitter of the selection origin of the already-growing cohort, ovulation and P4 still stable— that nobody anchored to the first non-circular crossing. M1-2-HG-H1 v2 keeps that candidate without the trigger «when the buffer fails» and without Robertson LRG as an anchor. Klein is location. SD(day | LH) vs first −2: absent.
3. A material luteal residual that does not lead L (r(L,LP)=0.255; Henry Var(FP)>Var(LP)) and that can seed phase noise (Henry 6 vs 8; van Zonneveld). M1-2-HG-H2L v2 is seed, not title. H2U absent on the consecutive operator.
4. An inconclusive ρ gate in −3. Henry∩Cole kill global identity in mid-reproductive life and do not donate 36–44 y. SD(FP)-menses remains forbidden.
5. Four operators that the primaries do not interchange. ReSTAGE persistence with (L, FP, LP) in 36–44 y is not estimable. The allowed informative failure is inconclusive, not a null of the bit.

The lever that does move the end (primordial freeze) cuts throughput. The pituitary reader (TGFBR3L) has contrary animal sign. FSH does not restore the granulosa that makes inhibin B. Manipulating the bit without an exogenous to bleeding is tautology.

The smallest next step is not recruiting perimenopausal women nor dosing calcitriol. It is the D0-IDENTITY desk with split operators, leadership split from Cov, and LH vs QBT sensitivity, without recruiting. While IDENTITY is inconclusive, SD(FP)-menses is not a node. D1 stays closed. G8 stays closed. Fertile function is not played by delaying a bleeding bit.

20. References

1. Harlow SD, Gass M, Hall JE, et al. Executive summary of the Stages of Reproductive Aging Workshop + 10. J Clin Endocrinol Metab. 2012. PMID 22344196. PMC3319184.
2. Harlow SD, Mitchell ES, Crawford S, et al. The ReSTAGE Collaboration: defining optimal bleeding criteria for onset of early menopausal transition. Fertil Steril. 2008. PMID 17681300. PMC2225986.
3. Huang X, Elliott MR, Harlow SD. Modeling menstrual cycle length and variability at the approach of menopause using hierarchical change point models. J R Stat Soc Ser C Appl Stat. 2014. PMID 24729638. PMC3979630. DOI 10.1111/rssc.12044. α0=3.313, β0=−4.795.
4. Ecochard R, Stanford JB, Fehring RJ, Schneider M, Najmabadi S, Gronfier C. Evidence that the woman's ovarian cycle is driven by an internal circamonthly timing system. Sci Adv. 2024;10(15):eadg9646. PMID 38598621. PMC11006216. DOI 10.1126/sciadv.adg9646. Table 2 reconfirmed this turn: ACF lag-1 Europe P=0.9480, NA P=0.3482; Europe lags 2–4 P<0.0001 / 0.0004 / 0.0024.
5. Henry S, Shirin S, Goshtasebi A, Prior JC. Prospective 1-year assessment of within-woman variability of follicular and luteal phase lengths in healthy women prescreened to have normal menstrual cycle and luteal phase lengths. Hum Reprod. 2024;39(11):2565–2574. PMID 39320898. PMC11532606. DOI 10.1093/humrep/deae215. n=53, k mean 13, 2.4% outside 21–36 d, QBT 2.4±1.5 d vs LH.
6. Cole LA, Ladner DG, Byrn FW. The normal variabilities of the menstrual cycle. Fertil Steril. 2009;91:522–527. PMID 18433748. DOI 10.1016/j.fertnstert.2007.11.073. Var_w n=31, k=5, 2.8/3.9/2.6 d. Not Cole hCG PMID 19438167.
7. Fehring RJ, Schneider M, Raviele K. Variability in the phases of the menstrual cycle. J Obstet Gynecol Neonatal Nurs. 2006. PMID 16700687. DOI 10.1111/j.1552-6909.2006.00051.x. n=141, 1060 cycles; range 42.5%; r(FP,LP)=−0.323; r(L,FP)=0.829; r(L,LP)=0.255.
8. Klein NA, Harper AJ, Houmard BS, et al. Is the short follicular phase in older women merely a consequence of advanced recruitment? J Clin Endocrinol Metab. 2002. PMID 12466381. FSH-peak→LH 10.7 vs 11.0 d, P=0.74.
9. Klein NA, Illingworth PJ, Groome NP, McNeilly AS, Battaglia DE, Soules MR. Decreased inhibin B secretion is associated with the monotropic FSH rise in older, ovulatory women. J Clin Endocrinol Metab. 1996. PMID 8675606. n=10 vs 13; dominant day 10.6±0.4 vs 14.5±1.0. First author Klein, not Welt.
10. Robertson DM, Lee CH, Baerwald A. Anti-Müllerian hormone, inhibin B, FSH and antral follicle count in late reproductive age. Endocrinol Diabetes Metab. 2021;4:e00172. PMID 33855196. PMC8029535. LRG 48.7±1.37 y; r(inhibin B, AFC)=0.70; B rise day 18 vs 14 P=0.054. Out of D0 as a −3 anchor.
11. Hale GE, Zhao X, Hughes CL, et al. Endocrine features of menstrual cycles in middle and late reproductive age and the menopausal transition classified according to STRAW. J Clin Endocrinol Metab. 2007. PMID 17550960. LRA 0/16 anovulatory.
12. van Zonneveld P, Scheffer GJ, Broekmans FJ, et al. Do cycle disturbances explain the age-related decline of female fertility? Hum Reprod. 2003. PMID 12615813.
13. Vanden Brink H, Chizen D, Hale G, Baerwald A. Age-related changes in major ovarian follicular wave dynamics during the human menstrual cycle. Menopause. 2013. PMID 23571519. LRA n=10.
14. Vanden Brink H, Robertson DM, Lim H, et al. Antral follicle dynamics and hormone production with age. J Clin Endocrinol Metab. 2015;100:4553–4562. PMID 26465392.
15. Hansen KR, Morris JL, Thyer AC, Soules MR. Reproductive aging and variability in the ovarian antral follicle count. Fertil Steril. 2003. PMID 12969701.
16. Scheffer GJ, Broekmans FJ, Dorland M, et al. Antral follicle counts by transvaginal ultrasonography are related to age in women with proven natural fertility. Fertil Steril. 1999. PMID 10560988.
17. Welt CK, Smith ZA, Pauler DK, Hall JE. Differential regulation of inhibin A and inhibin B by luteinizing hormone, follicle-stimulating hormone, and stage of follicle development. J Clin Endocrinol Metab. 2001. PMID 11232020.
18. Groome NP, Illingworth PJ, O'Brien M, et al. Measurement of dimeric inhibin B throughout the human menstrual cycle. J Clin Endocrinol Metab. 1996. PMID 8636341.
19. Weenen C, Laven JS, Von Bergh AR, et al. Anti-Müllerian hormone expression pattern in the human ovary: lack of expression in primordial follicles. Mol Hum Reprod. 2004. PMID 14742691.
20. Paramsothy P, Harlow SD, Greendale GA, et al. Bleeding patterns during the menopausal transition in the Study of Women's Health Across the Nation. BJOG. 2014. PMID 24735184. PMC4199918.
21. Jacobson MH, Howards PP, Kesner JS, et al. Thyroid hormones and menstrual cycle function in a longitudinal cohort of premenopausal women. Paediatr Perinat Epidemiol. 2018. PMID 29517803. DOI 10.1111/ppe.12462. Jacobson, not Plowden.
22. Burger HG. The menopausal transition—endocrinology. Menopause. 2008. PMID 18574431.
23. Ferrell RJ, O'Connor KA, Rodríguez G, et al. Monitoring reproductive aging in a 5-year prospective study: aggregate and individual changes in luteinizing hormone and follicle-stimulating hormone with age. Menopause. 2005;12:567–577. PMID 16145311. DOI 10.1097/01.gme.0000172265.40196.86. Var_w(L) minimum 33–40 y.
24. Najmabadi S, Schliep KC, Simonsen SE, et al. Menstrual bleeding, cycle length, and follicular and luteal phase lengths variation by mucus peak day. Paediatr Perinat Epidemiol / related. PMID 32104920. PMC8495765. max−min range, mean 27.1 y.
25. Visby Madsen C, Adornetto G, Kirkegaard K. The role of short luteal phases on cycle regularity during the perimenopausal transition. Menopause. 2025;32(3):258–265. PMID 39718569.
26. Brûlé E, Wang Y, Li Y, et al. TGFBR3L is an inhibin B co-receptor that regulates female fertility. Sci Adv. 2021. PMID 34910520.
27. Liu S, et al. Calcitriol is involved in maintaining primordial follicle reserve through the inhibition of the PI3K/Akt signaling pathway. J Ovarian Res. 2026. PMID 42169137. DOI 10.1186/s13048-026-02135-5.
28. Ruth KS, Day FR, Hussain J, et al. Genetic insights into biological mechanisms governing human ovarian ageing. Nature. 2021. PMID 34349265.
29. Sugino N, Suzuki T, Sakata A, et al. Angiogenesis in the human corpus luteum. Reprod Med Biol. 2008. PMID 29699289.
30. Reame NE, Kelche RP, Beitins IZ, et al. Age effects of FSH and pulsatile LH secretion across the menstrual cycle of premenopausal women. J Clin Endocrinol Metab. 1996. PMID 8636360. One cycle; not Var_w.
31. Lenton EA, Landgren BM, Sexton L, Harper R. Normal variation in the length of the follicular phase of the menstrual cycle: effect of chronological age. Br J Obstet Gynaecol. 1984. DOI 10.1111/j.1471-0528.1984.tb04830.x. FP location, not Var_w.
32. Lenton EA, Landgren BM, Sexton L. Normal variation in the length of the luteal phase of the menstrual cycle: identification of the short luteal phase. Br J Obstet Gynaecol. 1984. DOI 10.1111/j.1471-0528.1984.tb04831.x.
33. Harlow SD, Lin X, Ho MJ. Analysis of menstrual diary data across the reproductive life span: applicability of the bipartite model approach and the importance of within-woman variance. J Clin Epidemiol. 2000. PMID 10941950.
34. Baerwald AR, Adams GP, Pierson RA. Characterization of ovarian follicular wave dynamics in women. Biol Reprod. 2003. PMID 12748128.
35. Klein NA, Houmard BS, Hansen KR, et al. Age-related analysis of inhibin A, inhibin B, and activin A relative to the intercycle monotropic FSH rise. J Clin Endocrinol Metab. 2004. PMID 15181087.
36. Lewis KA, Gray PC, Blount AL, et al. Betaglycan binds inhibin and can mediate functional antagonism of activin signalling. Nature. 2000. PMID 10746731.
37. Li Y, Fortin J, Ongaro L, et al. Betaglycan (TGFBR3) functions as an inhibin A, but not inhibin B, coreceptor in pituitary gonadotrope cells in mice. Endocrinology. 2018. PMID 30364975.
38. Sowers MR, Eyvazzadeh AD, McConnell D, et al. Anti-Müllerian hormone and inhibin B in the definition of ovarian aging and the menopause transition. J Clin Endocrinol Metab. 2008. PMID 18593767.
39. WHO Task Force on Methods for the Determination of the Fertile Period. Temporal relationships between ovulation and defined changes in the concentration of plasma estradiol-17β, luteinizing hormone, follicle-stimulating hormone, and progesterone. Am J Obstet Gynecol. 1983. Via Fehring Table 1.
40. Prior JC, Vigna YM, Schulzer M, Hall JE, Bonen A. Determination of luteal phase length by quantitative basal temperature methods: validation against the midcycle LH peak. (Prior 1990b, Henry’s QBT–LH validation; not reopened as a 13-cycle series.)

Inherited and not reanalysed this turn: Welt 1999 PMID 9920069; Hale 2009 LOOP PMID 18978637; Castrillon 2003; Reddy 2008; Durlinger 2001; Visser 2007; Gaytán 2000; Abel 2000; Freeman 2012; Broer 2011; Vollman 1977 (via Henry); Prior 1990a (Henry’s QBT cohort).

Claims of this report: M1-2-REPORT-EN-C1 (delta: bilingual factual base post-AR/ED twinned; four operators; H3 v2 / H1 v2 / H2L v2 / HT1 v2); M1-2-REPORT-EN-C2 (Ecochard 2024 Table 2 reconfirmed: lag-1 not rejected, EU lags 2–4 yes; mean-field = calibration, not persistence); M1-2-REPORT-EN-C3 (ReSTAGE persistence with phases in 36–44 y is not estimable: Cole k=5 and Fehring k=5.2 fail G1-K; Henry k mean 13 in the wrong window); M1-2-REPORT-EN-C4 (Henry 2.4% is band, not operator; Fehring 42.5% is range, not P(|ΔL|≥7) nor ReSTAGE; Fehring leadership is algebra, not Cov); M1-2-REPORT-EN-C5 (H1 v1 buffer trigger withdrawn; Robertson LRG 48.7 y is not a −3 anchor; (vi) not restored); M1-2-REPORT-EN-C6 (four-operator D0-IDENTITY first; D1 closed; freeze/TGFBR3L illicit; G8 closed; HUMAN_QA_REQUIRED; not written into reportes/).

Hypotheses touched: M1-2-HG-H3 v2 (primary of object, revised_competing_contrary); M1-2-HG-H1 v2 (competitor / primary of the generator, revised_primary); M1-2-HG-H2L v2 (luteal rival, seed); M1-2-HG-HT1 v2 (translational, parked, HUMAN_QA_REQUIRED); M1-2-HG-H1 v1 / H3 v1 / H2L v1 / HT1 v1 split, not in force; SCOPE-H0 v1 gate not promoted; SCOPE-H2U v1 not promoted; lineage M1-1-HG-H1 v2, H2 v2, HT1 v2.

Not written into memoria agentes/Lua Labs/reportes/. Laboratory memory is not consolidated. No prescription. No Lua data. Next stage: BILINGUAL_QA.