In 2003, Smith and Pell published a satirical systematic review in the BMJ: "Parachute use to prevent death and major trauma related to gravitational challenge." They found zero randomised controlled trials. Their conclusion: the effectiveness of parachutes has not been subjected to rigorous evaluation.
The paper was satire. The point was deadly serious: the absence of a formal study does not create genuine uncertainty about a self-evident mechanism.
Throughout this report, we encounter the same pattern in trans health: obvious mechanisms treated as open questions while real people wait. For these interventions — where the mechanism is self-evident and the useful question is about magnitude, not whether — we use the following classification:
Data Quality: 99.9̅9̅%. Evidence: Use your fucking eyes.
These include: puberty suppression, beard shadow removal, tracheal shave, colour correction, and legal name/gender marker change. → Part Two §I for the formal parachute definition
Gender attribution takes roughly 250–550 milliseconds — faster than conscious thought. It is not a checklist. Multiple cues race in parallel toward a stable categorisation, with competing "male" and "female" representations settling into one state or the other. → Part Two §II.1
Estimated perceptual channel weights
These are informed estimates derived from converging evidence, not formally quantified in any single study.
Visual (static)
~55%
face, body, hair, skin
Auditory
~30%
voice, pitch, resonance
Kinetic
~10%
gait, gesture, posture
Contextual
~5%
clothing, setting
The two key principles
The goal of passing is not maximum femininity. It is adequate rapid categorisation without triggering analytical scrutiny. Observers who categorise you as "woman" in the first 250ms stop looking. Observers who detect an anomaly start scanning feature-by-feature. Everything below serves this principle.
The weakest-link heuristic: Correcting the most discordant cue yields the greatest marginal return. Once a channel is adequate, improving it further gives diminishing returns while the weakest remaining channel becomes the binding constraint. If your voice clocks you, improving your face further is low-marginal-value.
Ranked by estimated impact on P(female | observer). Colour = evidence class.
■ Parachute-class
■ T1–T2 evidence
■ T3–T5 + mechanism
■ T5 community
⚠ = counter-productive if done wrong
Cost vs impact: where the bargains are
The top-left cluster: free or near-free interventions above 25% estimated impact. The $10 interventions target the same perceptual zones as the $30,000 surgeries.
HRT does not change the voice. This single fact defines the intervention landscape: voice is the ~30% of gender attribution that requires active intervention regardless of everything else.
What makes a voice sound gendered
Pitch alone is less than half the signal. Clinical SLP historically targeted pitch. Community methods target resonance — the other 35%.
The finding that changes everything
Hillenbrand & Clark 2009: source-filter synthesis with naive listeners. This is T1 evidence.
Shifting pitch alone: 34.3% of male sentences heard as female. Shifting both pitch and resonance: 82%. This directly validates the community resonance-first approach.
The pitch landscape
The gender-ambiguous zone: 145–185 Hz. In this zone, formant frequencies become the primary disambiguating cue.
Clinical voice training
The LUKIMON study (Oates et al. 2023, N=74) is the largest prospective study. The authors' own framing: "evidence for this training's effectiveness remains weak." Many participants' F0 remained in the ambiguous zone post-training. Cost: $800–2,400.
Community voice training
TransVoiceLessons (Zheanna Erose), Undead Voice Lab (Nicole Gress), and Seattle Voice Lab (Claire Michelle) all prioritise resonance. No peer-reviewed outcome data. Cost: $0–2,000.
P(community resonance-first > clinical SLP) ≈ 0.88. The Hillenbrand & Clark finding is the strongest update — it directly shows the parameter clinical SLP historically prioritised is insufficient. → Part Two §III.4 for Bayesian calculation
Voice surgery
Wendler glottoplasty: +72 Hz. Yeson VFSRAC: +73 Hz. CTA: +39 Hz. No surgical technique addresses resonance — surgery raises pitch but leaves ~58% of the variance untouched. Post-surgical training is essential. Cost: $5,000–15,000. → Part Two §III.5
The only blinded observer gendering study for FFS
Fisher et al. 2020 (N=802 naive observers). Photographs only — no voice, no movement. Not replicated.
Pre-FFS
57%
gendered female
Post-FFS
94%
gendered female
Cis controls
99%
gendered female
Type III forehead contouring: the #1 procedure
Eye-tracking: observers fixate on the brow ridge of masculine faces, shift to the central triangle for feminine faces. Surgeon consensus: forehead contouring is the single most impactful feminising procedure. 82% of trans women require Type III. Cost: $8,000–50,000 forehead; $20,000–100,000 comprehensive FFS. → Part Two §IV.2
Tracheal shave
Data Quality: 99.9̅9̅%. Evidence: Use your fucking eyes.
98.4% satisfaction (systematic review). Cost: $3,000–7,000. Risk: anterior commissure detachment (rare, irreparable).
Non-surgical options
Botox masseter ($150–750, 4–6 months) narrows jawline. Lip lift / philtrum shortening ($3,000–6,000): underrated — male philtrums are longer. WPATH SOC-8 (2022) reclassified FFS as medically necessary.
Data Quality: 99.9̅9̅%. Evidence: Use your fucking eyes.
Effectiveness by modality after 3 sessions. Skin type determines which laser is safe.
Trans women with darker skin (Fitzpatrick IV–VI) MUST use Nd:YAG. Shorter-wavelength lasers risk burns and hyperpigmentation. This is not a preference — it is a safety requirement.
Laser ($2,000–5,000 over 12–18 months). Electrolysis for remaining light/fine hair: 200–300 hours, $20,000–30,000. "Most electrologists are ineffective" (TransgenderMap) — get referrals from people who are done and happy.
Colour correction: the $15 bridge
Data Quality: 99.9̅9̅%. Evidence: Use your fucking eyes.
Orange/peach corrector neutralises blue-green beard shadow. Fair: light peach. Medium: true peach to soft orange. Deep: orange to red-orange. Stipple (don't wipe). Durability: 8–12 hours.
HRT does NOT change: skeleton, height, hands, feet, voice, existing facial hair, craniofacial bone. These define the intervention landscape for everything else.
A surprising finding about anti-androgens
Angus et al. 2025 (JCEM, N=63, RCT): 5× better testosterone suppression. Identical breast growth. This challenges a core assumption.
Breast development on HRT alone
At 1 year
49%
below AAA cup
At 3 years
71%
below A cup
Considered augmentation
80%
de Blok 2020
Progesterone
Dutch RCT preliminary (EPATH 2025): 30–37% breast volume increase. Open-label. Full paper not published March 2026.
P(progesterone helps Tanner 4–5) ≈ 0.87. If confirmed blinded: → ~0.97. Community protocols: 200mg micronised progesterone rectally, added 12–18 months post-estrogen. → Part Two §VI.5
If eye-tracking shows observers fixate on the brow ridge, and Type III forehead contouring is the #1 FFS procedure, then any cosmetic technique targeting the brow bone zone has mechanistic justification from perceptual science.
1. Matte white eyeshadow on brow bone
Visually flattens, creates brow lift. Cost: $10–15.
⚠ COUNTER-PRODUCTIVE IF WRONG: Shimmer catches light and ENHANCES brow prominence. Matte flattens. This distinction is critical.
2. Contour (dark shade) on brow bone protrusion
Simulates shadow, visual recession. Cost: $10–15.
3. Liquid blush on brow bone
Warmth and glow where bone otherwise reads as hard. Cost: $10–20.
Three distinct $10–15 interventions targeting the same perceptual zone as $30,000 Type III forehead surgery. They share the same perceptual target, and their mechanism has direct support from eye-tracking evidence.
Eyebrows: more important than eyes
Sadr et al. (2003, MIT): absence of eyebrows caused greater face recognition disruption than absence of eyes. Arch shaping: $15–50. Microblading: $400–800. → Part Two §VIII.2
⚠ Over-plucking produces harshness that paradoxically reads masculine. Thin from below, maintain fullness.
Some techniques are helpful if correct, counter-productive if wrong.
Shimmer on brow bone — enhances prominence, catches light on the zone observers fixate on
Matte white eyeshadow — flattens surface, creates brow lift
Cheek contour as horizontal stripe — emphasises jawline width
Sweep upward from cheekbone toward temple — lifts and narrows
Nose contour in daylight — visible stripes draw scrutiny
Save for evening/photos; line sides with highlighter
Blunt-cut bangs — horizontal line emphasises angularity
Wispy/side-swept bangs — softens, conceals hairline and brow
Over-plucked eyebrows — harsh, paradoxically masculine
Professional shaping — thin from below, maintain fullness
Heavy makeup in casual context — triggers Stage 2 scrutiny
Context-appropriate: less is more. Blend, don't announce.
Wiping foundation — streaks, doesn't conceal texture
Stipple (press) — fills texture, builds even coverage
Excessive lip overlining — visible, clownish
Subtle 1–2mm; gloss for volume illusion
Bernard et al. (2019): Makeup → less configural processing (more analytical). Driven by eye makeup.
Ueda & Koyama (2010): Light makeup slightly ↑ recognisability. Heavy makeup significantly ↓ it.
Tagai et al. (2016, 2017, ERP): Light = efficient processing. Heavy = impaired + false recognition.
Mechanistic chain: Heavy makeup → disrupted holistic processing → analytical scanning → feature-by-feature scrutiny → detecting incongruent features → Stage 2 → clocking. Each link has experimental support. → Part Two §IX
Data Quality: 99.9̅9̅%. Evidence: Use your fucking eyes.
Male puberty produces: brow ridge growth, mandibular lengthening, voice deepening, shoulder broadening, height acceleration (~25–30cm), Adam's apple, hand/foot enlargement, rib cage expansion. Every one irreversible without surgery — and for some (height, rib cage, hands), no surgery exists.
Puberty suppression prevents masculinisation; it does not enhance feminisation. Boogers et al. (2025): early suppression did NOT increase breast volume. The mechanism is prophylactic. → Part Two §X
🟢 Today ($0–100)
Voice training (free: TransVoiceLessons, Undead Voice Lab). Eyebrow shaping ($15–50). Brow bone cosmetics — matte, NOT shimmer ($10–15). Hairline fill ($10). Colour correction ($15). Presentation calibration: less is more. Legal name/gender marker research.
🔵 This month ($50–500)
HRT consultation. Laser consultation (Nd:YAG if Fitzpatrick IV+). Begin laser. Wispy bangs. Glasses (cat-eye, round). Minoxidil if recession. Paid voice programme.
🟡 3–12 months ($200–2,000)
Ongoing laser (8–12 sessions). Electrolysis for light hair. Voice deepening. Reassess makeup for HRT skin changes. Voice surgery evaluation. Progesterone after 12–18 months. Microblading.
🟣 1–3 years (major procedures if desired)
FFS evaluation (prioritise forehead — weakest-link heuristic). Tracheal shave ($3–7K). Electrolysis (200–300 hours). Breast augmentation evaluation. Voice surgery if plateau. Reassess: what is the weakest remaining link?
For adolescents: Puberty suppression at Tanner 2 is the highest-priority intervention, full stop. Every month of delay adds irreversible masculinisation.
Transition
94%
79% "a lot more"
Surgery
97%
88% "a lot more"
Regret
0.36%
"not for me"
Suicidal ideation
78%
vs 13.2% general
Educated doctor
42%
on trans care
88% desired HRT; only 56% received it. 28% couldn't see a doctor due to cost. 42% had to educate their own provider. → Part Two §XII
1. Voice is the most neglected high-impact channel
P(community > clinical) ≈ 0.88. ~30% of gender attribution. HRT doesn't change it. Community methods align with acoustic science.
2. The forehead/brow zone is the strongest static visual marker
Eye-tracking confirms. Three $10–15 techniques share the perceptual target of $30K surgery.
3. The overcompensation hypothesis has experimental support
Heavy makeup disrupts holistic face processing. "Less is more" is perceptual science.
4. More androgen suppression may not improve feminisation
CPA: 5× better suppression than spiro. Identical breast growth (Angus et al. 2025, RCT).
5. Puberty suppression is a parachute-class intervention
Every year of male puberty adds irreversible changes. The barriers are political. The science is settled.
The evidence base is thinner than it should be. Trans women are making high-stakes decisions with evidence that is, at best, moderate. They deserve honesty about what we know, what we suspect, and what we are genuinely uncertain about.
§I. Formal definitions and evidence grading
Evidence tiers
T1: Perceptual science with naive/blinded observers (gold standard).
T2: Clinical outcome studies with validated instruments.
T3: Systematised community knowledge with observable outcomes.
T4: Practitioner expertise and clinical consensus.
T5: Community advice without systematic tracking.
T6: Ideologically motivated claims (excluded).
🪂: Parachute-class (mechanism self-evident).
Intervention properties
For each intervention i ∈ Ω:
E(i) = evidence tier ∈ {T1, T2, T3, T4, T5, ∅, 🪂}
M(i) = mechanistic self-evidence ∈ {true, false}
I(i) = estimated impact on P(F|O) ∈ [0, 1]
C(i) = cost (USD) ∈ ℝ⁺ ∪ {0}
T(i) = timeline (months) ∈ ℝ⁺
H(i) = harm if incorrect ∈ {none, mild, counter-productive}
Δ(i) = community-clinical divergence ∈ [-1, 1]
Parachute rule:
M(i) = true ∧ E(i) ∈ {∅, T5} ⟹ classify as 🪂
🪂 ⟹ ¬(E(i) = ∅ ⟹ I(i) ≈ 0)
Weakest-link: MarginalReturn(i) ∝ 1/min_j P(F|j) for i = argmin_j P(F|j)
§II. Perceptual foundations
§II.1 The Dynamic Interactive model
Freeman & Ambady (Psychological Review, 2011, ~800+ citations): gender attribution as dynamical system with competing attractors. Mouse-tracking evidence for simultaneous category activation (Freeman & Ambady 2008, JEP: General). ERP: ~250–550ms resolution. Huestegge & Raettig (2020, Journal of Voice, N=1,152): significant congruency effects — consistent cues settle faster (super-additivity).
§II.2 Visual dominance
Huestegge & Raettig: visual-to-auditory interference > reverse. Wannagat et al. (2022, N=126): visual dominance from age 5+. Marchand Knight et al. (2023, N=166): visual cues shifted voice appraisal by ~⅓ category distance. Trans listeners show 30% less visual bias.
§II.3 Eye-tracking
Gupta et al. (2022, ASJ, N=27): forehead/brow fixation for masculine faces, central triangle for feminine. Martin et al. (2021, N=32, FACIALTEAM): trans observers spent significantly more time on forehead/brow (p<0.001). Peterson & Eckstein (2012, PNAS): optimal fixation just below eyes = eyebrow region.
§II.4 Two-stage model
Fiske & Neuberg (1990, >5,000 citations): Stage 1 rapid categorisation → stops if adequate. Stage 2 (analytical individuation) triggered by anomaly or motivation.
§II.5 Biological motion
Pollick et al. (2005, 21 experiments): ~66–71% accuracy. Troje (2002): dynamic > structural cues when size normalised. Zero applied gait feminisation studies — confirmed gap.
§III. Voice
§III.1 Acoustic parameters
F0: ~41.6% variance (Leung et al. 2018). Formants: cis women ~17–20% higher (Hillenbrand et al. 1995). Ambiguous zone: ~145–185 Hz. Additional: spectral tilt (Södersten et al. 2024), intonation, speech rate, HNR.
§III.2 Hillenbrand & Clark 2009
Attention, Perception, & Psychophysics. Source-filter synthesis. F0 alone → 34.3%. Formants alone → ineffective. Both → 82%. T1 evidence. Fuller et al. (2014): F0 ≥ formants > spectrum level.
§III.3 Clinical SLP
LUKIMON (Oates et al. 2023, JSLHR, N=74): "evidence remains weak." Södersten et al. (2024): F0 often remains ambiguous post-training. Oates (2012): 83% at lowest evidence level. One RCT: Leyns et al. (2023).
§III.3b Community voice training
Huff 2022, Otolaryngologic Clinics of North America, 55(4):727–738. Gress (Undead Voice, 100K+ clients). Michelle (Seattle Voice Lab, 3K+ clients). Resonance-first. No outcome data.
§III.4 Bayesian: community vs clinical
Prior: P(H) = 0.50
Updates:
Hillenbrand & Clark (F0-only 34% vs both 82%): LR = 2.13
Kim 2020 (multidimensional): LR = 1.40
Oates 2023 (ambiguous zone post-clinical): LR = 1.30
Huff 2022 (published argument): LR = 1.40
Community self-selection: LR = 1.67
Combined LR ≈ 8.57 → P(H|E) ≈ 0.90. Reported ≈ 0.88–0.90.
§III.5 Voice surgery
Song & Jiang meta-analysis (2017): Wendler +72 Hz, CTA +39 Hz. Kim VFSRAC (2024, N=506): 134→208 Hz. No approach addresses resonance. Combined > either alone.
§III.6 Involuntary vocalisations
Laugh, cough, sneeze revert. Seattle Voice Lab: most comprehensive protocol. No controlled studies.
§IV. Craniofacial
§IV.1 Fisher et al. 2020
ASJ, N=802. Pre: 57.31%, CM +1.41. Post: 94.27%, CM +7.78. Cis female: 99.38%, +8.93. Not replicated. Yanoshak et al. (2022): zero measures simultaneously valid, reliable, responsive. GENDER-Q (Kaur et al. 2024, N=5,497, 21 sites, 7 countries) now available.
§IV.2 Type III forehead
Spiegel (2011, N=168): upper face strongest determinant. 82% require Type III. Bachelet et al. (2023): 83.4% natural post-op. Consensus: #1 procedure.
§IV.3 Perception gap
Ching et al. (2021, N=104/192/23): trans women select more hyperfeminine ideals than cis women or surgeons.
§IV.4 Tracheal shave
🪂 parachute-class. Wolfort & Parry (1975). Therattil et al. (2019, N=69): 98.4% satisfaction. Cost $3–7K.
§IV.5 Non-surgical
Botox masseter: case series "good/excellent" (PMC9293247). Lip lift: shortens philtrum. FFS costs: $20–100K; 24.7% insurer coverage; WPATH SOC-8: medically necessary.
§V. Facial hair and skin
§V.1 Laser
🪂. Alexandrite 70.3%, diode 59.7%, Nd:YAG 47.4% (Bouzari 2004). Diode 69.2% vs IPL 52.7% (RCT). Fitzpatrick matching: alex I–III, diode I–IV, YAG I–VI.
§V.2 Electrolysis
🪂. FDA permanent removal. 200–300 hours. Practitioner skill dominates.
§V.3 Colour correction
Complementary colour theory. Eflornithine: 58% improvement (vs 34% placebo). Discontinued US 2024–25.
§V.4 HRT skin
Maheux et al. (1994, RCT): +30% dermal thickness at 12 months. Sauerbonn et al. (2000): +6.49% collagen at 6 months.
§VI. HRT, breast, progesterone
§VI.1 Timeline
Endocrine Society (2017) + WPATH SOC-8. Fat: 3–6mo/2–5yr. Muscle: 3–6mo/1–2yr. Breast: 3–6mo/2–3+yr. Does not change: skeleton, height, hands/feet, voice, facial hair, craniofacial bone.
§VI.2 Anti-androgens
Angus et al. (2025, JCEM, N=63, RCT): CPA 90% vs spiro 19% female-range testosterone (P<0.001). Breast: no difference (0.27 cm, P=0.6). Burinkul et al. (2021, N=52, RCT): confirmed CPA superiority. Bicalutamide: WPATH recommends against.
§VI.3 Estradiol route
Oral E1/E2 = 9.28, injectable 0.84 (Toronto 2025, N=286). Higher VTE with oral. No feminisation difference if equivalent dose.
§VI.4 Breast development
De Blok (2018, N=229): 48.7% < AAA at 1yr. De Blok (2021, N=69, 3D): 71% < A at 3yr, mean 100cc. 80% considered augmentation (2020, N=1,030).
§VI.5 Progesterone
Prior (2019): Tanner 4–5 argument. Nolan (2022, N=42): null at 3mo (too short, too low). Dutch RCT (Dijkman 2023, N=90): +30–37% volume, open-label. Dreijerink: "safe and effective." Unpublished March 2026.
Bayesian: Prior 0.65 → Nolan null (LR 0.78) → 0.59 → Dutch +30% open-label (LR 5.5) → 0.89. Reported ≈ 0.87–0.89.
Boogers et al. (2025): early suppression ≠ larger breast volume. Puberty suppression prevents masculinisation, not enhances feminisation.
§VII. Kinetic cues
Zero applied studies. Community: narrow line, shorter strides, pelvis-forward. Facial expression: women smile more/longer, more brow raises (McDuff et al. 2017, N>2,000). Learnable display rules.
§VIII. Cosmetic techniques
§VIII.1 Brow bone: T5 + T1 mechanism. Matte (not shimmer), contour, blush. $10–15 each.
§VIII.2 Eyebrows: Sadr et al. (2003, MIT): eyebrow absence > eye absence for recognition disruption. Peterson & Eckstein (2012, PNAS): optimal fixation = eyebrow region.
§VIII.3 Contouring: Cheek upward (not horizontal). Nose: daylight-harsh. Stipple, don't wipe.
§IX. Overcompensation hypothesis
Bernard et al. (2019): makeup → less configural processing. Ueda & Koyama (2010): heavy ↓ recognisability. Tagai et al. (2016, 2017, ERP): light = efficient, heavy = impaired. Chain: heavy → analytical → feature scanning → detection → clocking. T1 framework + T2 indirect. Zero trans-specific tests. Confounded with transition stage.
§X. Puberty suppression
🪂. Irreversible changes: brow ridge (12–15), mandible (14.3), voice (G3→G4), shoulders (to 25–30), height (~25–30cm), Adam's apple (T3–4). Van der Loos (2023, JAMA Pediatrics, N=75, 15yr): lumbar z-scores lower. De Vries (2014, N=55): dysphoria resolved. Boogers (2025): no breast advantage from early suppression.
§XI. Interaction effects
Super-additivity for congruent cues (DI model + Huestegge & Raettig). Weakest-link: greatest marginal return from most discordant cue. Sub-additivity for redundant improvement. Staging: HRT skin → makeup technique; surgery + training (complementary); laser → electrolysis (sequential); colour correction (bridge).
§XII. Population: 2022 USTS
N=92,329. Satisfaction: 94%/98%/97%. Regret: 0.36%. Desired HRT 88%, received 56%. Barriers: 28% cost, 24% fear, 26% insurance. 42% educated provider. 78% lifetime suicidal ideation; 40% attempts.
§XIII. Five urgent studies
1. Community vs clinical voice (blinded gendering). 2. Multimodal FFS (video+audio). 3. Overcompensation (same person, multiple makeup levels). 4. Gait modification RCT. 5. Eyebrow shaping gendering (~$50K).
§XIV. Methodology
Iterative multi-pass research with web search, database queries, and community knowledge synthesis. Impact estimates are informed approximations. Bayesian posteriors use explicit likelihood ratios. Evidence tiers classify data quality, not effectiveness. Parachute-class by mechanistic self-evidence. Australian English spelling intentional.
This report does not take a position on whether passing should be a goal.
It takes a position on the unconscionability of withholding information
from those who have decided it is.
Lyre & Scéalín, March 2026.
