Michael Quoc Founder & CEO · Axioms.ai
Vector 1 of 4: Pressure gas and pressure injury

Axiomatic Intelligence · An investigation

The Physics of Diving

The fatal node of a dive is the ascent, not the problem that came before it. Gas held in the lung at depth tears the lung at about the pressure of one metre of water, so the injury happens in the last metres of an ascent from a held breath on compressed gas; the fatality registries have no cell for panic and store it as the emergency ascent, the leading disabling agent in both the fatal and the treated-injury series; and the wound from a shark, a moray or a snake rarely kills a diver, while the startled ascent can. Every headline number met here is a denominator artefact: a diving fatality rate changes by factors of five to twenty-five with the choice of member-years, dives, participants or divers, a trigger percentage is a share of the coded minority of deaths, and shark-bite counts rose while the risk per person fell because ocean users grew faster than bites. Underwater, the diver's own senses fail as instruments: air hunger is the gap between respiratory drive and achieved ventilation, not a reading of carbon dioxide or oxygen, so hypoxia and an oxygen seizure arrive in silence, the carbon-dioxide alarm loses gain with experience and no dry test predicts it, and narcosis, carbon dioxide and hypoxia add up on performance while removing the ability to notice the deficit. On air past about thirty-six metres under load, gas density turns breathing against itself, because past a critical density each extra litre of ventilation makes more carbon dioxide than it clears. Panic is common and real, reported by a quarter to a third of divers across a career, and how often it kills has no instrument on it. Stop, Breathe, Think, Act owns one lawful lever, fewer and larger breaths at a fixed amount of gas, and only while the hazard is flat in time. Four of the eight mechanisms this page set out to test, narcosis, decompression, awe and skill decay, were researched but reached no verified claim, so they carry no axioms yet.

The question this page asks: what actually happens to a human body and mind at depth, and what really ends a dive badly?

“My obsession is sharks. The first time was terrifying. Now it is what I seek most. Whale sharks, giant manta rays, a school of hundreds of barracuda closing around you. Swimming in the presence of massive sea life is a feeling nothing on land replicates.”

“At 100 feet, optimism is fatal. You check your air, your buddy's air, your depth, your time, your ascent rate. Stop. Breathe. Think. Act.”

This page rests on four research vectors and 32 axioms, forged on October 4, 2026 under the current version of the method. Every axiom lists the mechanism, the underlying physics, and the primary sources it rests on. Nothing on this page is taken on faith; it can be read one piece at a time.

How do depth and absolute pressure change gas volume, density, partial pressure and the pressure differentials across a diver's gas-filled spaces, and where exactly does the ten-meter doubling hold?

Pressure gas and pressure injury · eight axioms

The mechanism under test is pressure and the gas laws in the body: volume, density, partial pressure, gas-filled spaces and the ten-meter doubling. The easy answer this vector was told not to settle for: “Pressure doubles every ten meters and everything shrinks”
1.1 Pressure is linear in depth, so absolute pressure doubles once from the surface and then at geometrically spaced depths, never every ten metresVERIFIEDWater weight adds a fixed pressure per metre, about one atmosphere per ten, and a fixed addition halves its proportional effect each time, so doubling happens once.Open this axiomClose this axiom
The axiom in full

Force balance on a still water column gives dP/dh = rho g, so P_abs(h) = P_0 + rho g h. In seawater at sea level (model inputs 101.325 kPa, 1025 kg/m3, 9.80665 m/s2) each metre adds 10.05 kPa and each ten metres about 100.5 kPa, one atmosphere per 10.1 m of seawater or 10.4 m of fresh water. Absolute-pressure ratios across successive ten-metre intervals are 1.99 (0 to 10 m), 1.50 (10 to 20), 1.33 (20 to 30), 1.25 (30 to 40): equal increments, falling ratios. Exact doubling needs a full P_0 of head each time, so it occurs at H(2^n - 1) = 10.1, 30.2, 70.6, 151 m seawater, where H = P_0/(rho g) = 10.08 m (10.33 m fresh). Gauge pressure doubles 10 to 20 m and 20 to 40 m, a different set, and enters no gas law. Bounds: sea level and constant density; at a 1000 m lake with 0.90 atm surface pressure the surface-to-10 m ratio is about 2.08. Evidence class MODEL; population none (water-column physics).

The physics underneath

Hydrostatic equilibrium: the pressure difference across a horizontal slab equals the slab's weight per area, so pressure grows linearly with depth at slope rho g; the surface atmosphere is an additive intercept, and a fixed additive increment is a shrinking proportional increment, so doubling is a one-time event from any starting pressure.

The sources
This is the first axiom on the page.Next axiom: 1.2, Every gas law in diving takes absolute pressure; 'ten metres equals one bar' is exact only because the metre of sea water was redefined as a pressure unit →Link to this axiom: #ax-1-1
1.2 Every gas law in diving takes absolute pressure; 'ten metres equals one bar' is exact only because the metre of sea water was redefined as a pressure unitVERIFIEDGas laws count molecules per volume, which only absolute pressure measures; gauge and msw are conventions, so a slogan exact in msw is a unit definition, not water physics.Open this axiomClose this axiom
The axiom in full

Boyle, the density law and Dalton describe molecular number density and are written in absolute pressure; a gauge reading is referenced to its surroundings and enters no gas law. The metre of sea water (msw) is a pressure unit: one tenth of a bar, or 10.0381 kPa under EN 13319; 1 fsw = 0.30643 msw. So '10 msw = 1 bar' is a definition and 'ten metres doubles the pressure' is exactly true only when the metre means msw and the surface is assigned 1 bar. In real seawater one atmosphere is 10.08 m (model) or 10.1 m (NOAA), one bar 9.95 m; in fresh water one atmosphere is 10.33 m (model) or 10.4 m (NOAA). A depth gauge is a manometer with a density assumed: seawater calibration misreads fresh-water depth by about 2.5 percent while reporting pressure correctly. Bounds: ideal-gas form within one percent below 10 bar ambient; compressibility departs by several percent only at cylinder pressures of 200 to 300 bar; isothermal Boyle holds for body gas, not cylinder transients. Evidence class MODEL plus a standards definition.

The physics underneath

Gas pressure is momentum flux from molecular collisions and depends on molecules per volume and temperature, both absolute quantities; a gauge is a differential instrument, and a unit defined as a fixed pressure makes any depth-pressure sentence in that unit a tautology.

The sources
← Previous axiom: 1.1, Pressure is linear in depth, so absolute pressure doubles once from the surface and then at geometrically spaced depths, never every ten metresNext axiom: 1.3, Density and partial pressure rise linearly with depth while volume and its per-metre change fall hyperbolically, so two opposite hazard gradients run through one dive →Link to this axiom: #ax-1-2
1.3 Density and partial pressure rise linearly with depth while volume and its per-metre change fall hyperbolically, so two opposite hazard gradients run through one diveVERIFIEDCompression packs molecules per litre in proportion to absolute pressure, raising density and every partial pressure linearly, while a trapped volume's change per metre falls with depth.Open this axiomClose this axiom
The axiom in full

At fixed composition and temperature rho_gas = P M/(R T) and p_i = x_i P, both linear in absolute pressure: air at 30 m has about four times the surface oxygen and nitrogen partial pressure (0.84 and 3.2 bar) at the same 21/79 composition, and a breath at 30 m holds about four times the molecules of a surface breath. A sealed isothermal parcel follows V = V_0 H/(H + z), so its fractional change per metre is 1/(H + z): 9.9 percent per metre at the surface, 5.0 at 10 m, 3.3 at 20 m, 2.0 at 40 m (model). Partial-pressure and density hazards worsen steadily with depth; expansion-ratio hazards (lung overexpansion, buoyancy swing) concentrate in the last ten metres of ascent. Measured density bound: in QinetiQ rebreather and open-circuit trial dives the fraction of dives failed for end-tidal CO2 above 8.5 kPa inflects upward near a respired density of 6.0 g/L; the authors recommend 5.2 g/L ideal maximum (air at 31 m) and 6.2 g/L absolute (air at 39 m), noting the trials were not designed for the question and few dives sat at high density; BSAC adopted it and prints 6.3 g/L. Population for the density numbers: working divers in equipment trials. Densities are 0 degree bookkeeping values, about 12 percent above warm humid airway gas. CO2 and panic adjudication belongs to the respiratory load oxygen and carbon dioxide vector.

The physics underneath

Compression raises molecules per litre in proportion to absolute pressure, so density and every partial pressure inherit the linear depth law; a trapped volume's proportional change per metre is the derivative of H/(H + z), falling as depth grows; dense gas raises the pressure drop needed to move it, limiting ventilation and CO2 clearance.

The sources
← Previous axiom: 1.2, Every gas law in diving takes absolute pressure; 'ten metres equals one bar' is exact only because the metre of sea water was redefined as a pressure unitNext axiom: 1.4, Pressure injury is a differential across a wall; a near-rigid cavity gains the same differential per unequalised metre at any depth, with the middle ear hurting near 60 mmHg, locking near 90 mmHg and rupturing above about 0.75 bar →Link to this axiom: #ax-1-3
1.4 Pressure injury is a differential across a wall; a near-rigid cavity gains the same differential per unequalised metre at any depth, with the middle ear hurting near 60 mmHg, locking near 90 mmHg and rupturing above about 0.75 barVERIFIEDA gas cavity that cannot shrink carries the full water-pressure difference on its wall, about one kPa per ten centimetres unequalised, until the ear tube locks near 12 kPa.Open this axiomClose this axiom
The axiom in full

The load on a membrane is delta P = P_inside minus P_outside times area; absolute pressure on both sides cancels. Liquid tissue barely compresses: with water's bulk modulus 2.2 GPa, the 0.3 MPa added by 30 m compresses it about 0.014 percent against a 75 percent volume loss for a sealed gas parcel, so only gas spaces fail. A cavity whose walls cannot move (middle ear, sinus, mask) cannot relieve the Boyle deficit by shrinking, so its differential grows as rho g times unequalised descent, about 1 kPa per 10 cm, the same at 3 m as at 33 m; equalisation adds molecules through the Eustachian tube or sinus ostia. Clinical review thresholds (recreational and clinical divers; review figures, primaries not opened): otalgia from about 60 mm Hg; Eustachian lock at about 90 mm Hg (12 kPa), after which a Valsalva fails; alternobaric vertigo from a 45 mm Hg asymmetry on ascent; eardrum rupture in most divers above 10 psi (0.75 bar). Seawater equivalents about 0.8 m, 1.2 m and 7.5 m of unequalised descent (writer's conversion at 10.05 kPa per metre). Rigid-wall is a worst case; a membrane absorbing 1 to 2 percent of a 2 mL middle ear relieves 1 to 2 kPa at the surface and 4 to 8 kPa at 4 ata (derivation, unsourced). Inner-ear injury from a forced Valsalva against a locked tube is a separate mechanism. Evidence class MODEL for mechanics, CORRELATIONAL for thresholds.

The physics underneath

Net force on a wall is the pressure difference across it; liquids transmit hydrostatic pressure almost without volume change while a gas cavity that cannot shrink must carry the whole difference on its wall, so the wall load tracks unequalised depth change linearly and independently of starting depth.

The sources
← Previous axiom: 1.3, Density and partial pressure rise linearly with depth while volume and its per-metre change fall hyperbolically, so two opposite hazard gradients run through one diveNext axiom: 1.5, The lung is open on a regulator and closed on a held breath; gas trapped at depth reaches its rupture threshold, a transpulmonary pressure of about 95 to 110 cmH2O, within about one metre of ascent, so expansion injury is a shallow-water event →Link to this axiom: #ax-1-4
1.5 The lung is open on a regulator and closed on a held breath; gas trapped at depth reaches its rupture threshold, a transpulmonary pressure of about 95 to 110 cmH2O, within about one metre of ascent, so expansion injury is a shallow-water eventVERIFIEDGas inhaled at depth expands on ascent until the lung wall carries about 10 kPa, one metre of water's weight, so rupture happens shallow, from a held breath.Open this axiomClose this axiom
The axiom in full

Breathing from a regulator delivers gas near ambient pressure, so lung volume stays about normal and the molecule count changes with depth. A held breath or trapped region makes the lung a closed parcel: gas inhaled at 10 m demands twice its volume at the surface, from 30 to 20 m only a third more. The compliant lung tracks volume to its elastic limit: a transpulmonary pressure of 95 to 110 cmH2O (9.3 to 10.8 kPa) disrupts parenchyma and forces gas into the interstitium (cadaver lungs, Schaefer 1958 and Malhotra and Wright 1961, read through the Adir and Bove 2014 review; primaries not opened), the head of 0.93 to 1.07 m of seawater (conversion). Case record: pulmonary barotrauma reported after scuba use at as little as 1 msw (JMVH review); DAN: a breath-hold ascent after inhaling from a tank at 4 fsw may tear alveoli. Escaped gas reaches the mediastinum, pleura or pulmonary veins and arteries (arterial gas embolism). A breath-hold diver who filled at the surface returns to his starting volume; the asymmetry is compressed gas at depth. Risk per metre of ascent is about five times higher at 2 m than at 40 m (1/(H + z)). Bounds: other sources give 60 to 80 mmHg and one 20 to 80 mmHg (unopened), so about one metre is the engineering figure and about 0.3 m the pessimistic bound; obstruction, bullae and asthma lower it by regional trapping. Decompression sickness is dissolved-gas physics and belongs to the tissue loading and decompression vector. Evidence class INTERVENTION (cadaver measurement via review) plus CORRELATIONAL (case series).

The physics underneath

A compliant gas-filled organ equalises by changing volume until its tissue reaches its strain limit; the pressure difference at that limit is about 10 kPa, the hydrostatic head of one metre of seawater, and the expansion per metre of ascent is largest near the surface, so a held breath on compressed gas fails shallow.

The sources
← Previous axiom: 1.4, Pressure injury is a differential across a wall; a near-rigid cavity gains the same differential per unequalised metre at any depth, with the middle ear hurting near 60 mmHg, locking near 90 mmHg and rupturing above about 0.75 barNext axiom: 1.6, Any compressible gas a diver carries makes neutral buoyancy an unstable equilibrium whose stiffness falls as one over (H plus depth) squared, about 25 times stronger at the surface than at 40 m →Link to this axiom: #ax-1-5
1.6 Any compressible gas a diver carries makes neutral buoyancy an unstable equilibrium whose stiffness falls as one over (H plus depth) squared, about 25 times stronger at the surface than at 40 mVERIFIEDGas volume shrinks on descent and grows on ascent, moving displaced water the same way, so any departure from neutral feeds itself, most steeply in shallow water.Open this axiomClose this axiom
The axiom in full

Buoyant force is rho_w g times displaced volume; absolute pressure does not change it by itself. Displacement splits into incompressible tissue and a gas volume (lungs, wing, suit foam) following V_gas,0 H/(H + z), so dB/dz = -rho_w g V_gas,0 H/(H + z)^2, negative at every depth: displaced downward a diver gets heavier and sinks faster, displaced upward lighter and rises faster. Stiffness scales as (H + z)^-2, so for the same gas volume it is (50.08/10.08)^2, about 25 times, larger at the surface than at 40 m. Each litre of added displacement is about 10 N of lift in seawater. Venting changes molecule count and breaks the loop; nothing else does. A runaway ascent couples to the lung axiom: expansion drives ascent, ascent drives expansion, and the one-metre lung threshold is crossed before gas is vented. Bounds: holds for gas whose expansion changes external displacement; gas in a rigid cavity changes nothing; suit-foam magnitudes not quantified. Evidence class MODEL.

The physics underneath

Archimedes makes lift proportional to displaced volume and Boyle makes carried gas volume fall with absolute pressure, so lift decreases on descent and increases on ascent; the restoring force has the wrong sign everywhere, with magnitude set by the hyperbola's slope, steepest near the surface.

The sources
← Previous axiom: 1.5, The lung is open on a regulator and closed on a held breath; gas trapped at depth reaches its rupture threshold, a transpulmonary pressure of about 95 to 110 cmH2O, within about one metre of ascent, so expansion injury is a shallow-water eventNext axiom: 1.7, Henry's law fixes the dissolved-gas destination and perfusion sets the clock: a compartment reaches 98.4 percent of a new equilibrium after six half-times and never exactly 100, and tissue tension is not a mechanical overpressure →Link to this axiom: #ax-1-6
1.7 Henry's law fixes the dissolved-gas destination and perfusion sets the clock: a compartment reaches 98.4 percent of a new equilibrium after six half-times and never exactly 100, and tissue tension is not a mechanical overpressureVERIFIEDPartial pressure sets how much gas a tissue can hold and blood flow sets how fast it gets there, so loading is an exponential approach that never quite finishes.Open this axiomClose this axiom
The axiom in full

At equilibrium and in dilute solution the dissolved concentration is C = k p, k set by gas, solvent and temperature; IUPAC bounds the law to equilibrium and infinite dilution. It says nothing about rate. Uptake needs ventilation, circulation and diffusion; a well-mixed compartment follows dC/dt = k(C_eq - C), approaching C_eq exponentially with half-time ln2/k: 50 percent after one, 75 after two, 98.4 after six half-times. NOAA's slide rounds this to 'theoretically reaches 100%' after six half-times; the exponential never does. Compartments are rates (brain and spine fast, bone slow) and need not map onto organs. Tissue tension C/k is the gas-phase pressure that would equilibrate with the dissolved amount, not a pocket inflating tissue; on ascent, when equilibrium concentration falls faster than gas leaves, the tissue is supersaturated and bubbles may form, a process distinct from stretching an existing gas space. DAN: asymptomatic bubbles clear within about a day; no chronic loading from frequent diving. Bounds: equilibrium values are the pressure gas and pressure injury vector's input to the tissue loading and decompression vector; time-dependent loading, bubble formation and sickness incidence are the tissue loading and decompression vector's; the single exponential is an idealisation. Evidence class MODEL with teaching and practitioner statements.

The physics underneath

Dissolution equilibrium balances molecular flux across the gas-liquid interface, setting concentration proportional to partial pressure; the approach to that equilibrium is a first-order transport process whose rate is blood flow over tissue capacity, so loading is exponential and asymptotic.

The sources
← Previous axiom: 1.6, Any compressible gas a diver carries makes neutral buoyancy an unstable equilibrium whose stiffness falls as one over (H plus depth) squared, about 25 times stronger at the surface than at 40 mNext axiom: 1.8, Ambient pressure has a location on the body →Link to this axiom: #ax-1-7
1.8 Ambient pressure has a location on the bodyVERIFIEDIt differs about 10 kPa per metre of height, so a breathing-gas reference offset of a few decimetres is a static lung load of a few kPa on every breath.Open this axiomClose this axiom
The axiom in full

Water pressure differs across a diver's vertical extent by rho g times height: 30 cm is about 3 kPa, and the NOAA manual puts a six-foot diver's feet almost 3 psi (about 20 kPa) above his head. Gas pressure varies far less over the same height because gas is about a thousand times less dense. The height of the regulator's reference relative to the lung centroid and the diver's posture therefore add a positive or negative static load on every breath, which NOAA lists with gas density and exertion as drivers of increased work of breathing in immersion. Pressure matching at the mouth removes the large static mismatch but not the gradients needed to move gas: viscous losses scale with viscosity, inertial losses with density, and dense gas pushes airway flow toward turbulence, so breathing resistance does not simply multiply by the pressure ratio. Bounds: the static-load numbers are hydrostatics; the physiological consequence (CO2 retention, air hunger) is the respiratory load oxygen and carbon dioxide vector's. Evidence class MODEL with a training-manual figure.

The physics underneath

Hydrostatic pressure is a field that rises linearly with depth at every point, so two points on one body a metre apart differ by 10 kPa; a gas supply equalised at one point is unequalised by that gradient at another, and the lungs feel the difference as a static load.

The sources
← Previous axiom: 1.7, Henry's law fixes the dissolved-gas destination and perfusion sets the clock: a compartment reaches 98.4 percent of a new equilibrium after six half-times and never exactly 100, and tissue tension is not a mechanical overpressureNext axiom: 2.1, Depth multiplies gas density, and density taxes every breath by a square and a square root →Link to this axiom: #ax-1-8

What governs how gas density, breathing resistance, ventilation and oxygen partial pressure produce carbon dioxide retention, air hunger, impairment, panic or oxygen injury in a scuba diver?

Respiratory load oxygen and carbon dioxide · eight axioms

The mechanism under test is oxygen, carbon dioxide and breathing: how respiratory load, gas exchange and breathing pattern drive air hunger, impairment, panic and oxygen injury. The easy answer this vector was told not to settle for: “Slow breathing prevents panic and makes every underwater emergency manageable”
2.1 Depth multiplies gas density, and density taxes every breath by a square and a square rootVERIFIEDAbsolute pressure sets gas density; turbulent loss grows with density times flow squared, so exertion at depth collapses attainable ventilation toward a density-set ceiling.Open this axiomClose this axiom
The axiom in full

For a fixed mixture at near-constant temperature, respired gas density rises in proportion to absolute pressure (equation of state), so air at 1.29 g/L on the surface is about 5.2 g/L at 30 m (4 ATA) and about 6.5 g/L at the 40 m recreational limit. Pressure loss in turbulent and orifice flow scales as density times flow squared, so resistive breathing power scales as density times flow cubed, and the attainable ventilation ceiling falls as density to a negative power between 0.4 and 0.5: maximum voluntary ventilation measured in healthy chamber subjects breathing air at 30 m is about half the sea-level value (within-person chamber intervention); peak expiratory flow falls 23 percent at 33 ft and 39 percent at 100 ft. Boundary: at resting flows the loss is mostly viscous and nearly pressure-independent, so the tax bites with exertion; static lung load depends on the vertical distance between the regulator's pressure reference and the lung centroid (about −25 cmH2O prone on a back-mounted rebreather), so orientation changes the load at unchanged depth; helium lowers density through molar mass; no breathing technique changes density.

The physics underneath

ρ = P·M̄/(RT); ΔP ≈ a·μ·Q + b·ρ·Q|Q|; power = ΔP·Q; at a fixed maximum driving pressure Q_max ∝ ρ^(−1/2). Observation: the MVV halving and PEF decrements are measurements; the cubic power law is inference from the quadratic loss law.

The sources
← Previous axiom: 1.8, Ambient pressure has a location on the bodyNext axiom: 2.2, Carbon dioxide retention is a conservation failure with three attack surfaces, and depth loads all three at once →Link to this axiom: #ax-2-1
2.2 Carbon dioxide retention is a conservation failure with three attack surfaces, and depth loads all three at onceVERIFIEDArterial CO2 equals production over effective alveolar ventilation; depth raises production, caps ventilation and inflates dead space together, so retention is the default under load.Open this axiomClose this axiom
The axiom in full

Arterial CO2 is set by mass balance: PaCO2 ≈ 0.863 × V̇CO2 / [V̇E × (1 − VD/VT)]. Hypercapnia has exactly three causes (production up, minute ventilation down, dead-space fraction up) and compressed-gas diving at depth loads all three simultaneously: work and the oxygen cost of breathing raise production; the density ceiling plus apparatus and elastic loads cap ventilation; dead-space fraction rises with density. In immersed exercising subjects moved from surface immersion to 4.7 ATA (Cherry et al. 2009, within-person chamber intervention, moderate exercise), mean arterial CO2 rose from about 34 to 46 mmHg while ventilation fell and dead-space fraction rose; added external resistance did not produce a further statistically significant rise within that protocol. Breathing pattern at fixed minute volume moves the dead-space term by arithmetic: with 250 mL of dead space and 15 L/min, 30 breaths of 0.5 L clear 7.5 L/min of alveolar gas and 10 breaths of 1.5 L clear 12.5 L/min, a 67 percent difference at identical gas consumption. Boundary: depth does not force PaCO2 up; what fails is the ability or willingness to hold alveolar ventilation against the load. The pattern lever concerns frequency and tidal volume at fixed minute volume, never breath-holding or skip-breathing (which lowers V̇E and raises PaCO2), and it buys nothing once the density ceiling or an inspired-CO2 source (a failed scrubber) binds; larger breaths cost elastic work and shorter breaths raise peak flow, so neither slower nor deeper is optimal under every load.

The physics underneath

Molecular conservation of CO2 plus the equation of state; alveolar ventilation = f × (VT − VD); with inspired CO2 present, PACO2 ≈ PICO2 + 863 × V̇CO2 / V̇A. Observation: the PaCO2 rise and the VD/VT increase are measurements; the three-surface decomposition and pattern arithmetic are inference from the conservation law.

The sources
← Previous axiom: 2.1, Depth multiplies gas density, and density taxes every breath by a square and a square rootNext axiom: 2.3, Above a critical gas density the ventilation loop loses its fixed point: CO2 test failures jump from 3 percent to 41 percent across the 6 g/L knee →Link to this axiom: #ax-2-2
2.3 Above a critical gas density the ventilation loop loses its fixed point: CO2 test failures jump from 3 percent to 41 percent across the 6 g/L kneeVERIFIEDPast a critical gas density, each extra litre of ventilation costs more CO2 than it clears, so the feedback loop inverts and retention runs away.Open this axiomClose this axiom
The axiom in full

Ventilating costs CO2. Once the work of an extra increment of ventilation produces more CO2 than the increment clears, raising ventilation no longer lowers PaCO2 and hypercapnia runs away, while expiratory flow is choked by dynamic airway compression (the Starling-resistor limit: pleural pressure exceeds airway pressure, the airway collapses, more effort yields no more flow). The record shows a knee, not a slope: in the Anthony and Mitchell analysis of rebreather test dives (correlational, a test registry; endpoint end-tidal CO2 above 8.5 kPa, about 64 mmHg, for five consecutive breaths), 15 of 439 tests at 2 to 3 g/L failed for high CO2 (3.4 percent) versus 31 of 66 tests at 6 to 7 g/L (41.3 percent); they proposed 5.2 g/L as the ideal maximum and 6.2 g/L as the hard planning limit, which air reaches at about 31 m and 39 m. Boundary: the denominator is a rebreather test series not designed to find a threshold, the high-density groups are small (66 tests), the transfer of the knee to open-circuit diving rests on unpublished regulator testing reported second-hand and is disputed, and density comparisons need a stated temperature convention. The 5.2/6.2 figures are planning boundaries with a strong mechanical rationale and a narrower empirical base than their adoption implies. The recreational 40 m air limit (6.5 g/L) sits past the knee; 55 m deep air sits at 8.4 g/L.

The physics underneath

The marginal CO2 cost of ventilation rises as ρ·V̇³ while the marginal CO2 cleared per litre is roughly constant, so there is a density beyond which d(PaCO2)/d(V̇E) changes sign; the Starling-resistor flow limit caps V̇E independently. Observation: the 3.4 and 41.3 percent failure fractions are counts from the test record; the sign-change argument is inference.

The sources
← Previous axiom: 2.2, Carbon dioxide retention is a conservation failure with three attack surfaces, and depth loads all three at onceNext axiom: 2.4, Air hunger is a drive-minus-achieved-ventilation signal, not a reading of CO2 or oxygen, so hypoxia is silent and a loaded diver feels suffocation →Link to this axiom: #ax-2-3
2.4 Air hunger is a drive-minus-achieved-ventilation signal, not a reading of CO2 or oxygen, so hypoxia is silent and a loaded diver feels suffocationVERIFIEDAir hunger is the gap between respiratory drive and achieved lung inflation, so constrained ventilation feels like suffocation while silent hypoxia produces no gap.Open this axiomClose this axiom
The axiom in full

Air hunger arises from the mismatch between brainstem respiratory drive (sensed centrally through corollary discharge) and the afferent report of achieved ventilation from lung and chest-wall mechanoreceptors (neuromechanical dissociation). It does not require respiratory muscle work: in four ventilator-dependent quadriplegic subjects held at constant mechanical ventilation (within-person intervention), air hunger appeared at a mean end-tidal CO2 rise of about 10 mmHg above resting, and in healthy subjects the CO2 air-hunger curve is unchanged by neuromuscular paralysis. Tidal inflation relieves it; positive transrespiratory pressure during immersed heavy exercise reduces dyspnoea. Hypoxia with intact CO2 clearance produces little mismatch and so little air hunger: on open circuit a corroded hypoxic cylinder produced immediate loss of consciousness without awareness (case record), and on rebreathers hypoxia is a leading fatal mode because it arrives gradually without a respiratory alarm. Boundary: established in laboratory populations (ventilated patients, healthy volunteers, immersed exercising chamber subjects); the diver-specific inference is that high drive (CO2 retention), constrained output (density and apparatus load) and reduced inflation feedback are the three inputs that maximise the mismatch term. Effort (the sense of work) is a separate sensation; a diver can have high effort with adequate clearance, air hunger with failing clearance, or both; neither measures PaCO2 or PaO2.

The physics underneath

Air hunger ∝ (efferent drive − afferent ventilation feedback); drive is set by chemoreceptors (PaCO2, pH, PaO2); the feedback term is set by achieved tidal volume, which the density ceiling caps. Observation: the 10 mmHg onset and the paralysis result are measurements; the three-input maximisation in the diver is inference.

The sources
← Previous axiom: 2.3, Above a critical gas density the ventilation loop loses its fixed point: CO2 test failures jump from 3 percent to 41 percent across the 6 g/L kneeNext axiom: 2.5, The diver's CO2 alarm is a variable-gain sensor that the exposure itself turns down, and no dry test predicts it →Link to this axiom: #ax-2-4
2.5 The diver's CO2 alarm is a variable-gain sensor that the exposure itself turns down, and no dry test predicts itVERIFIEDSustained breathing load shifts the CO2 controller's set-point and repeated exposure lowers its gain, so the experienced diver retains more CO2 with less warning.Open this axiomClose this axiom
The axiom in full

Hypoventilation at depth is partly chosen by the controller: in immersed exercising chamber subjects, arterial CO2 stayed elevated even when ventilatory limits were not reached and the expected CO2-driven rise in breathing did not appear, so the brain tolerates a higher PaCO2 rather than pay the density-cubed cost of defending it. The controller's gain differs widely between people and is lowered by diving itself: divers show lower ventilatory and occlusion-pressure responses to inhaled CO2 than non-divers (Kerem et al. 1980, cross-sectional); in brainstem fMRI, experienced divers showed the lowest response to 7 percent CO2 of three groups, below healthy controls and far below panic patients (Goossens et al. 2014, cross-sectional). Resting CO2 sensitivity did not correlate with end-tidal CO2 during immersed exercise with inspired CO2 and breathing resistance (Shykoff and Warkander 2012, within-person intervention, US Navy subjects; a retainer there is a subject whose end-tidal PCO2 exceeded 9.2 kPa while inhaling 5.6 kPa CO2). Divers cannot be relied on to feel CO2 excess; some have no early signs. Boundary: chamber and laboratory divers, mostly military and technical; the blunting is of the CO2 response and not the hypoxic response (elite breath-hold divers show the same selective blunting, admitted only as corroboration of controller plasticity, never of apparatus or blackout dynamics). Two hours of thermoneutral head-out immersion at rest raised end-tidal CO2 without blunting peripheral chemosensitivity (Sackett et al. 2017), so immersion alone is not the lever; load and habitual exposure are. Hyperoxia lowered ventilation during immersed exercise at 1.75 ATA O2 (Peacher et al. 2010) but not across 0.7 to 1.3 ATA in the Cherry protocol, so 'more oxygen suppresses breathing' is too broad.

The physics underneath

The chemoreflex is a feedback controller whose set-point drifts under sustained load (trading PaCO2 against the cubic work cost) and whose gain is plastic with repeated exposure; a sensor with drifting gain and no external reference cannot be calibrated by self-report. Observation: the group differences and the non-correlation are measurements; the optimiser framing is inference.

The sources
← Previous axiom: 2.4, Air hunger is a drive-minus-achieved-ventilation signal, not a reading of CO2 or oxygen, so hypoxia is silent and a loaded diver feels suffocationNext axiom: 2.6, Carbon dioxide is a dose-dependent fear stimulus in laboratory populations, with a non-respiratory predictor beside it, and the in-water link from retention to panic is unmeasured →Link to this axiom: #ax-2-5
2.6 Carbon dioxide is a dose-dependent fear stimulus in laboratory populations, with a non-respiratory predictor beside it, and the in-water link from retention to panic is unmeasuredVERIFIEDRising CO2 acidifies the brain and drives an interoceptive alarm whose gain varies by person, so retention can seed panic, but the underwater link is uninstrumented.Open this axiomClose this axiom
The axiom in full

In laboratory CO2 challenge (intervention, between-group): prolonged 5 percent CO2 produced panic in 52 percent of evaluable panic-disorder patients versus 9 percent of healthy controls (Gorman et al. 2001); across 59 panic patients and 39 controls, CO2 was a more potent panic stimulus than voluntary hyperventilation and a rise in respiratory rate preceded the attacks (Papp et al. 1997); in 64 healthy subjects given double breaths of 0, 9, 17.5 and 35 percent CO2, fear and discomfort rose with dose (Griez et al. 2007); 35 percent CO2 produced panic in all three patients with bilateral amygdala damage against three of twelve comparison subjects (Feinstein et al. 2013), so an interoceptive fear pathway exists that does not need an intact amygdala. In divers, trait anxiety predicts panic behaviour in beginning students (correlational), and two retrospective self-report surveys on non-probability samples put lifetime 'panic or near-panic' at 50 percent of men and 64 percent of women (Morgan, n = 245) and 24 percent of men and 37 percent of women (Colvard and Colvard, n = 12,231); the factor-of-two gap is the honest error bar. No located study measures PaCO2 in an open-water recreational diver at the moment of panic. Boundary: the laboratory stimulus is a step and diving hypercapnia a slow ramp, and transfer across that time constant is not demonstrated; ventilatory sensitivity to CO2 and panic susceptibility are different measures (no diagnostic-group difference in ventilatory sensitivity in Gorman 2001); the suffocation-false-alarm theory's specific predictions have failed tests; the amygdala ASIC1a acid-sensing pathway is mouse work and does not make diver panic a chemical inevitability. The one loop that closes without the missing measurement is physical: fear or air hunger → rapid shallow breathing → dead-space fraction and peak-flow cost rise → alveolar ventilation falls while CO2 production rises → drive rises against the density ceiling → the mismatch grows. The 68 percent figure for panic in incident reports noting mental state is a selected denominator and belongs to the accident-record vector.

The physics underneath

CO2 raises brain extracellular acidity, which drives brainstem chemoreceptors and an interoceptive threat response whose gain differs by population; the behavioural output (rapid shallow breathing) feeds back into the conservation equation. Observation: all percentages are laboratory or survey counts with stated denominators; the in-water loop is inference.

The sources
← Previous axiom: 2.5, The diver's CO2 alarm is a variable-gain sensor that the exposure itself turns down, and no dry test predicts itNext axiom: 2.7, Oxygen injures on two clocks with opposite thresholds, and carbon dioxide moves the fast one →Link to this axiom: #ax-2-6
2.7 Oxygen injures on two clocks with opposite thresholds, and carbon dioxide moves the fast oneVERIFIEDBrain oxygen dose scales as time squared times PO2 to about the eleventh power when immersed, and retained CO2 raises cerebral blood flow and so the dose.Open this axiomClose this axiom
The axiom in full

Inspired oxygen pressure is oxygen fraction times absolute pressure. Pulmonary toxicity is slow and low-threshold: it develops at inspired PO2 as low as 0.5 ATA over hours, and repeated six-hour dives at 1.3 ATA produced symptoms and lung-function changes, worse with exercise, some persisting three days (Shykoff and Florian 2018, within-person intervention, Navy subjects). CNS toxicity is fast and high-threshold: its endpoint is a seizure with little or no warning, lethal underwater by drowning. Its dose is a steep power law rather than a cliff: the fitted human model for immersed exposure is K = t² × PO2^10.93 (dry: PO2^12.99), so risk is roughly quadratic in time and about eleventh-power in PO2 when wet, with immersion a separate risk axis (Arieli et al. 2002, a model fitted to chamber and wet exposure records); a 10 percent PO2 excursion is about a 2.8-fold dose multiplier. Operational limits sit inside the curve: 45 minutes at 1.6 ATA, 120 at 1.5, 150 at 1.4 in the 1991 NOAA single-exposure table; the 30 September 2025 guideline supports 240 minutes of working exposure at 1.3 ATA followed by 240 minutes of resting decompression within its stated scope. Hypercapnia shifts the CNS curve: CO2 raises cerebral blood flow and so the delivered cerebral oxygen dose; CO2 retainers have convulsed within the hyperoxic limits, scrubber breakthrough creates real seizure risk at 1.2 to 1.3 bar, and heavy exercise against breathing resistance raises it. Boundary: the exponents are a model fit, no dedicated CNS oxygen-toxicity study has been run since 1993, and 3 percent of divers reported CNS-consistent symptoms at 1.2 bar after 240 minutes in the summarised record; the 1.3 ATA guideline applies to technical and scientific divers under stated conditions and does not license extending higher-PO2 exposures; any PO2 limit is valid only conditional on normocapnia; pulmonary toxicity is rarely limiting on a recreational profile and governs repeated long technical and surface-oriented commercial exposures (DMAC 35 Rev 1, 2025).

The physics underneath

PiO2 = FiO2 × Pabs; CNS dose ∝ t² × PO2^~11 wet; cerebral oxygen delivery ∝ cerebral blood flow × arterial O2 content, and cerebral blood flow rises with PaCO2, so retained CO2 raises the brain's effective oxygen exposure at unchanged inspired PO2. Observation: symptom counts and guideline limits are records; the exponents are a model; the CO2-shifts-the-curve statement is inference from measured cerebral blood flow physiology plus case reports.

The sources
← Previous axiom: 2.6, Carbon dioxide is a dose-dependent fear stimulus in laboratory populations, with a non-respiratory predictor beside it, and the in-water link from retention to panic is unmeasuredNext axiom: 2.8, Narcosis, carbon dioxide and hypoxia impair additively and remove the diver's ability to read his own impairment →Link to this axiom: #ax-2-7
2.8 Narcosis, carbon dioxide and hypoxia impair additively and remove the diver's ability to read his own impairmentVERIFIEDInert-gas narcosis, CO2 and hypoxia each depress the CNS and sum on performance, and because self-assessment is a CNS output it fails alongside.Open this axiomClose this axiom
The axiom in full

In chamber subjects (within-person intervention), high CO2 tensions impaired cognitive and psychomotor performance at 1 ATA and caused further decrements at 6 ATA with no significant N2 × CO2 interaction; the signatures differ, CO2 slowing performance while nitrogen narcosis degrades both speed and accuracy (Fowler et al. 1991). In wet simulated air dives to 6.8 ATA with added breathing resistance, subjects showed severe mental dysfunction with minimal symptoms; severe hypercapnia did not correlate with dyspnoea and dysfunction developed suddenly when high resistance was met (Warkander et al. 1990). In a multitasking simulator across combined N2, O2, CO2 and exercise exposures, some combinations produced incapacitating narcosis with amnesia, and perceived narcosis did not correlate with measured performance (Freiberger et al. 2016). Most divers are significantly impaired breathing air at 60 to 70 m and some at about 30 m; habituation lowers the subjective symptoms of narcosis while performance stays impaired. Hypoxia adds a silent third term: on open circuit it needs a supply error (the corroded cylinder cases) and on rebreathers it is a leading fatal mode because it arrives without a respiratory alarm. Boundary: chamber and wet-pot populations, mostly fit military and technical subjects under controlled exposures; recreational field performance data are not in this record; the additive finding is for N2 and CO2 on performance, while the CO2 to oxygen-seizure interaction is a separate non-additive pathway; 'impaired' is a test score, never a verdict about a person.

The physics underneath

Three independent CNS depressants (inert-gas narcosis by lipid or protein-site action, CO2 by cerebral acidosis and blood-flow change, hypoxia by substrate loss) sum on performance; the interoceptive report of impairment is itself a CNS output, so it degrades with the thing it would report. Observation: the performance decrements and the perception-performance non-correlation are measurements; the three-terms-sum statement is inference across studies.

The sources
← Previous axiom: 2.7, Oxygen injures on two clocks with opposite thresholds, and carbon dioxide moves the fast oneNext axiom: 3.1, The four-node fatality chain →Link to this axiom: #ax-2-8

Which causal chains end scuba dives badly, how often panic appears in them alongside medical and equipment triggers, what rates the records can estimate, and when is Stop Breathe Think Act evidenced?

Accident chains and emergency response · eight axioms

The mechanism under test is human error and the accident record: the causal chains that end dives badly, what rates the records can estimate, and the Stop Breathe Think Act habit. The easy answer this vector was told not to settle for: “Panic kills divers; the original problem almost never does”
3.1 The four-node fatality chainVERIFIEDCause of death is the least informative node, and causal position outranks category.Open this axiomClose this axiom
The axiom in full

In recreational open-circuit scuba fatality records (DAN, 947 deaths, 1992 to 2003; CORRELATIONAL registry), each death is coded as an ordered chain: trigger, disabling agent, disabling injury, cause of death. Drowning was the recorded cause of death in 70 percent of the 814 deaths with one coded and explains nothing; the disabling injury (asphyxia 33 percent, arterial gas embolism 29 percent, cardiac 26 percent, of 590 coded) carries the mechanism, and the investigators themselves ranked it above cause of death. The same category occupies different slots in different chains (insufficient gas: 41 percent of 346 coded triggers; 27 percent of 332 coded disabling agents), so position, not category, carries the information. The same chain template governs the Australian coronial series (126 deaths, 2001 to 2013) and the Royal Netherlands Navy treated-injury series (288 cases, 1966 to 2023). Holds for coded fatality and treated-injury records; it is a grammar of the record, not a per-dive probability.

The physics underneath

An unconscious immersed human drowns whatever disabled them, so the death certificate collapses every upstream pathway into one word; only the ordered chain recovers the mechanism. Observation: the four coded distributions with denominators 346, 332, 590, 814 of 947. Inference: that position dominates category.

The sources
← Previous axiom: 2.8, Narcosis, carbon dioxide and hypoxia impair additively and remove the diver's ability to read his own impairmentNext axiom: 3.2, Published trigger percentages are conditionals on a coded minority, and the uncoded majority is missing in the direction that hides mental-state chains →Link to this axiom: #ax-3-1
3.2 Published trigger percentages are conditionals on a coded minority, and the uncoded majority is missing in the direction that hides mental-state chainsVERIFIEDRetrospective chain coding requires a surviving observer or physical trace, so published trigger shares are conditionals on a coded minority biased toward physically evidenced chains over mental-state chains.Open this axiomClose this axiom
The axiom in full

In the DAN 947-death series (CORRELATIONAL, 1992 to 2003) a trigger was coded in 346 deaths (36.5 percent) and a disabling agent in 332 (35 percent); 'insufficient gas 41 percent' is 41 percent of 346, about 142 deaths or about 15 percent of all 947, and 'emergency ascent 55 percent' is about 19 percent of 947. Assigning the 601 uncoded deaths to the two extremes puts the whole-series gas-trigger share anywhere from about 15 to 78 percent, an identification range, not a confidence interval. Codeability needs a witness, a buddy, a dive computer or a diagnostic autopsy: in the Australian series (126 deaths, 2001 to 2013) the ascent was unwitnessed in 8 deaths including 3 out of gas, and anxiety was coded as a trigger (10 of 126) only where a witness described it; in the Dutch 288-injury series all four steps applied in only 111 cases (38.5 percent). The record is structurally more blind to chains whose decisive link was an unobserved mental state than to chains with a physical trace. Holds for every fatality and injury registry that codes chains retrospectively.

The physics underneath

A chain is codeable only when an observer or a physical trace survives; panic alone and unwitnessed leaves neither, so trigger tables undercount it by construction. Observation: 346, 332, 601, 111 of 288, the witness rule. Inference: the direction of the bias.

The sources
← Previous axiom: 3.1, The four-node fatality chainNext axiom: 3.3, Every diving fatality rate is a denominator choice →Link to this axiom: #ax-3-2
3.3 Every diving fatality rate is a denominator choiceVERIFIEDMember-years, dives, participants and divers are different estimands differing five- to twenty-five-fold.Open this axiomClose this axiom
The axiom in full

Four measured rates exist for recreational scuba death and each estimates a different thing. Per insured member-year: 187 deaths over 1,141,367 DAN insured member-years, 2000 to 2006, 16.4 per 100,000 (95 percent CI 14.2 to 18.9; CORRELATIONAL insured registry). Per dive: 563 deaths over an estimated 306 million dives by US residents, 2006 to 2015, 0.18 per 100,000 dives or 1.8 per million (CORRELATIONAL numerator, survey-weighted MODEL denominator). Per supervised introductory participant: 79 deaths over 7,118,731 PADI Discover Scuba Diving registrations, 1992 to 2019, 2.55 per 100,000 in 1992 to 2005 and 0.87 per 100,000 in 2006 to 2019 (CORRELATIONAL registration count; early under-registration named). Per estimated diver: tank-fill and survey denominators give Japan 8.8 to 33.8 per 100,000 divers and 1.0 to 2.4 per 100,000 dives, Australia 2000 to 2006 3.57 per 100,000 divers and 0.57 per 100,000 dives. A member-year rate is the annual risk of being that kind of diver; a per-dive rate is the risk of one dive; a diver at 100 dives a year and one at 4 share the first and differ 25-fold in the second. The member-year rate converts to 0.7 per 100,000 dives only by assuming 25 dives per diver per year. No rate transfers to a particular deep, overhead, cold or medically loaded dive; experience-stratum denominators are unknown.

The physics underneath

Deaths are counted by surveillance while exposure is estimated by survey, membership, registration or tank fills, so the rate inherits the denominator's construction. Observation: the four rates with numerators, denominators and years. Inference: that cross-country spread is mostly denominator.

The sources
← Previous axiom: 3.2, Published trigger percentages are conditionals on a coded minority, and the uncoded majority is missing in the direction that hides mental-state chainsNext axiom: 3.4, Panic is high-prevalence and low-lethality →Link to this axiom: #ax-3-3
3.4 Panic is high-prevalence and low-lethalityVERIFIEDA witnessed trigger in about one fatal chain in twelve, survived in most reported episodes, and with no identifiable case-fatality rate.Open this axiomClose this axiom
The axiom in full

Three quantities are routinely conflated. Lifetime prevalence among living divers: in a voluntary survey of certified recreational divers (12,087 responses, 76 percent male, September 2000 to January 2001; CORRELATIONAL, self-selected survivors, no exposure denominator), 24 percent of men and 37 percent of women reported at least one panic episode while diving; a pre-diving history of panic roughly doubled it (45 versus 19 percent of men, 57 versus 33 percent of women); 15 percent of first dive panics ended in a rapid or uncontrolled ascent, within which 5 percent of men and 4 percent of women reported decompression-illness symptoms and 1 and 2 percent were recompressed; 81 percent of men and 73 percent of women used their training. Panic coded in fatal chains: in the Australian coronial series (126 scuba deaths, 2001 to 2013; CORRELATIONAL) anxiety was coded as a trigger in 10 deaths (8 percent), all but one a novice, only where a witness described it; panic also appears inside equipment-trigger narratives, so 8 percent is the witnessed-trigger share, not the share with panic anywhere in the chain. Panic-attributable death is a counterfactual needing a per-dive panic incidence that no opened source measures; it is unidentified. Holds for recreational divers; the survey and the coronial series are different populations and denominators.

The physics underneath

Panic is an internal state; in the fatal case the diver is dead and usually unwitnessed, so the record counts it only where a survivor saw it, while the survey counts it only among the living; neither denominator is the other's. Observation: the survey percentages over 12,087; 10 of 126; the witness rule. Inference: that prevalence cannot stand in for attribution.

The sources
← Previous axiom: 3.3, Every diving fatality rate is a denominator choiceNext axiom: 3.5, The record stores panic as its physical output, the emergency ascent, and Boyle's law makes that one action the lethal one →Link to this axiom: #ax-3-4
3.5 The record stores panic as its physical output, the emergency ascent, and Boyle's law makes that one action the lethal oneVERIFIEDFear drives a fast ascent, falling pressure expands trapped lung gas, the lung tears and gas embolises the brain; the registry codes the ascent and the embolism, never the fear.Open this axiomClose this axiom
The axiom in full

The DAN fatality taxonomy (947 deaths, 1992 to 2003; CORRELATIONAL) has no panic category; its disabling agents are emergency ascent, insufficient gas, buoyancy trouble and excessive work of breathing. A frightened diver's decision to surface enters the record as 'emergency ascent', the leading disabling agent (55 percent of 332 coded), and the injury that follows is arterial gas embolism: among deaths with AGE as the disabling injury, 96 percent had an emergency ascent in the chain (odds ratio 30 or more against all other disabling injuries) and 63 percent had insufficient gas. In the nonfatal Royal Netherlands Navy series (288 treated injuries, 1966 to 2023) the ascent is again the most frequent disabling agent (52 percent of 195 coded) and arterial gas emboli were the most lethal mechanism (7 deaths of 288). Boyle's law: gas volume is inversely proportional to absolute pressure, so trapped lung gas expands most per metre in the last ten metres and a breath-held or fast ascent ruptures alveoli and embolises the arteries. The lethal output of panic is one recordable action, not a diffuse state; and the same code contains correct ascents in response to gas loss, so 'emergency ascent 55 percent' is not 'panic 55 percent'. Holds for coded fatal and treated-injury chains.

The physics underneath

Fear selects the surface as the goal; ascent lowers ambient pressure; expanding lung gas injures the lung and embolises the brain; the record sees the ascent and the embolism, never the fear. Observation: the absent category, 55 percent of 332, 96 and 63 percent with OR, 52 percent of 195, 7 of 288. Inference: the ascent code is where mental-state chains surface.

The sources
← Previous axiom: 3.4, Panic is high-prevalence and low-lethalityNext axiom: 3.6, The cardiac arm of the fatality chain has no behavioural node, rises steeply with age, and is unreachable by any in-water habit →Link to this axiom: #ax-3-5
3.6 The cardiac arm of the fatality chain has no behavioural node, rises steeply with age, and is unreachable by any in-water habitVERIFIEDImmersion, cold and exertion load a vulnerable heart until it fails without any decision, so the cardiac quarter of diving deaths has no node a habit can act on.Open this axiomClose this axiom
The axiom in full

In the DAN 947-death series (CORRELATIONAL, 1992 to 2003) cardiac incidents were 26 percent of 590 coded disabling injuries and 13 percent of 814 coded causes of death; cardiovascular disease raised the odds of a cardiac disabling injury 10.5-fold and age over 40 raised them 5.9-fold against all other disabling injuries. Among DAN insured members (187 deaths, 1,141,367 member-years, 2000 to 2006) cardiac-associated death was 12.9 times more likely at age 50 and over, and fatality rates rose with age for both sexes. In the Australian coronial series (126 deaths, 2001 to 2013) health conditions were the leading predisposing factor (59, 47 percent), medical disabling agents were 48 of 121 (38 percent), more than three quarters cardiac, and a cardiac disabling injury was associated with preceding exertion (odds ratio 6.31). In the Dutch treated-injury series health conditions were the most frequent predisposing factor (58 percent of 288). The 2011 DAN workshop summary puts about one quarter of deaths with cardiac events, mostly older divers. A cardiac arrhythmia at depth is a chain with no decision in it; no mnemonic, breathing drill or calm reaches it, and it is addressed only before the dive or by a rescuer. Holds for recreational fatality and injury records; the age gradient is a registry finding, not a per-dive risk.

The physics underneath

Immersion shifts blood centrally, cold and exertion raise cardiac load, and a vulnerable heart fibrillates or infarcts without a preceding decision. Observation: every percentage and odds ratio with its denominator. Inference: that the pathway is behaviour-free and its share grows with the population's age.

The sources
← Previous axiom: 3.5, The record stores panic as its physical output, the emergency ascent, and Boyle's law makes that one action the lethal oneNext axiom: 3.7, Fatal and nonfatal diving records measure different estimands on incommensurable scales, yet the ascent is the leading disabling agent in both →Link to this axiom: #ax-3-6
3.7 Fatal and nonfatal diving records measure different estimands on incommensurable scales, yet the ascent is the leading disabling agent in bothVERIFIEDQuestionnaires, emergency departments, insurance claims and death registries ascertain different slices of the same chains, so their rates cannot be pooled, while the ascent recurs as the shared fork.Open this axiomClose this axiom
The axiom in full

Nonfatal diving harm is counted by instruments with different thresholds. In the control arm of a cluster-randomised trial (426 recreational divers, 2012; self-reported, prompted) major mishaps ran 11.2 and minor mishaps 18.2 per 100 dives. US emergency-department surveillance (2006 to 2015; weighted) estimated 1,394 scuba-injury presentations a year, 80 percent treated and released, and 47 US deaths per 1,000 ED presentations (658 deaths against 13,943 presentations). Decompression-sickness insurance claims among DAN members ran 20.5 per 10,000 member-years, 2000 to 2007. In the Royal Netherlands Navy treated-injury series (288 cases, 1966 to 2023) predisposing factors were identified in 90 percent, triggers in 49 percent (exertion 19, primary diver error 18, faulty equipment 17 percent), disabling agents in 68 percent, and 7 cases (2 percent) died. A per-100-dive mishap rate, a per-1,000-ED-visit death ratio, a per-10,000-member-year claims rate and a fatality series are four scales; pooled into one 'accident rate' they lose their meaning. One fact crosses the scales: the ascent is the leading disabling agent in the fatal series (55 percent of 332 coded) and the nonfatal one (52 percent of 195 coded), so what separates a treated injury from a death is what happens after the ascent begins. Holds for recreational and military-treated populations as measured; the instruments do not transfer to each other.

The physics underneath

Each instrument has its own case definition and ascertainment (a questionnaire prompt, an ED visit, an insurance claim, a death) and counts a different slice of the same chains; the ascent recurs because it is the common path from trigger to injury. Observation: every rate with its instrument and denominator; 52 and 55 percent. Inference: incommensurability, and the ascent node as the survival fork.

The sources
← Previous axiom: 3.6, The cardiac arm of the fatality chain has no behavioural node, rises steeply with age, and is unreachable by any in-water habitNext axiom: 3.8, Stop Breathe Think Act has component evidence but no outcome evidence, and its value is bounded by the hazard clock: it pays where the hazard is flat and the reflex action lethal, and is unavailable where gas, consciousness or time is gone →Link to this axiom: #ax-3-7
3.8 Stop Breathe Think Act has component evidence but no outcome evidence, and its value is bounded by the hazard clock: it pays where the hazard is flat and the reflex action lethal, and is unavailable where gas, consciousness or time is goneVERIFIEDA pause trades margin for action selection, so it pays where the hazard clock is slow and the reflex action lethal, and inverts once gas, time or consciousness is gone.Open this axiomClose this axiom
The axiom in full

Across four independent searches (2026-10-04) no controlled trial, cohort or before-after analysis of the four-step habit on any scuba emergency outcome was located; what exists is instruction (DAN Europe, 2021; ARGUMENT). Component evidence is narrower. INTERVENTION: 44 novice divers randomised to audiotaped mental training (n=15), placebo information (n=15) or training only (n=14); the intervention group reported lower pre-dive cognitive anxiety, lower respiration rate and better bail-out and mask-removal performance, with respiration rate predicting bail-out (Terry 1998). CORRELATIONAL: among 27 divers, high trait anxiety (STAI-T 39 or more) was associated with slower mask clearing, 10.8 versus 7.1 s (P = 0.04), and slower underwater number-Stroop responses, 72.5 versus 64.4 s (P under 0.01), other skills unaffected (Tsai 2020). INTERVENTION, the contrast class: a pre-dive checklist in a cluster-randomised trial (1,043 divers, 2,041 dives, 2012) cut self-reported major mishaps 36 percent and all mishaps 32 percent, showing the field can run outcome trials on safety habits and has not run one on this habit. The boundary follows from the chain: with margin defined as time to disabling deterioration minus time for an effective response, a deliberative pause pays only where the hazard is roughly flat and the arousal-driven action is the lethal one (entanglement with gas, lost buddy, flooded mask, free-flow); it is unavailable where gas is gone (Breathe has no input), where hazard grows with time (buoyant runaway under Boyle's law, immersion pulmonary oedema, hyperoxic seizure), or where the diver is unconscious (cardiac arrest, embolism). Narcosis degrades the Think step at depth. Holds for trained recreational divers who retain gas, consciousness and time; the null result is a dated search finding, not proof of absence.

The physics underneath

A pause buys better action selection at the cost of margin; where margin is large and the reflex action lethal the trade is positive; where margin is near zero or the reflex action is the only route to gas the trade inverts. Observation: the null search (2026-10-04), the three studies with samples and effects. Inference: the margin model and the boundary classes.

The sources
← Previous axiom: 3.7, Fatal and nonfatal diving records measure different estimands on incommensurable scales, yet the ascent is the leading disabling agent in bothNext axiom: 4.1, Shark-bite counts and shark-bite risk are different instruments, and the count moves opposite to the rate →Link to this axiom: #ax-3-8

What governs the measured injury risk to divers from sharks and other large animals, how species, sensory biology and encounter conditions shape observable behavior, and which descriptions, reassuring or fearful, hold?

Animal encounters and measured risk · eight axioms

The mechanism under test is sharks and large animals: the measured injury risk, their sensory biology and how an encounter is structured. The easy answer this vector was told not to settle for: “Sharks are harmless, calm and curious around divers”
4.1 Shark-bite counts and shark-bite risk are different instruments, and the count moves opposite to the rateVERIFIEDParticipation grows faster than bite incidents, so recorded bites rise while per-dive risk falls; a count without its exposure denominator measures population and reporting, not hazard.Open this axiomClose this axiom
The axiom in full

In every exposure-estimated record opened (California white sharks 1950 to 2013, 86 injurious attacks and 13 fatal; Australian scuba 2001 to 2010, about 1.55 million dives a year and one scuba shark fatality), the number of recorded bites rose or held while the per-person or per-dive risk fell by more than 91 percent or sat near 0.065 fatalities per million dives, because the denominator (ocean users, dives) grew faster than the numerator. Population: shark-incident and dive-accident registries (ISAF worldwide, the Australian diver record, the California white shark record). Outcome: a fatality or injury count with its denominator. Range: aggregate historical rates only, never a species- or site-specific estimate. Evidence class: CORRELATIONAL registry counts against survey-derived exposure, with a MODEL component in the California weighting. A bite count with no matched exposure estimates population size and reporting, never hazard. Shark attack is about 3 percent (28 of 926) of recorded Australian diving-related fatalities 1960 to 2017.

The physics underneath

A rate is a ratio of two independently measured quantities; when participation grows faster than incidents the rate falls while the count rises. Ferretti et al. 2015 weighted 86 attacks against coastal population and activity surveys: attacks rose from about 0.9 a year (1950s) to about 1.5 a year (last decade) while individual risk fell more than 91 percent; surfers grew from 7,000 (1950) to 872,000 (2013), scuba divers from about 2,000 (early 1960s) to about 408,000 (2013). Lippmann 2018 reports 1.55 million scuba dives a year 2001 to 2010 and one fatal shark attack on a scuba diver in that decade; the 0.065 per million is this writer's division, not a published rate. Denominators come from participation surveys that Lippmann et al. 2016 call potentially unreliable; one event in a decade carries an interval orders of magnitude wide (zero events in n exposures bounds the rate at about 3/n at 95 percent).

The sources
← Previous axiom: 3.8, Stop Breathe Think Act has component evidence but no outcome evidence, and its value is bounded by the hazard clock: it pays where the hazard is flat and the reflex action lethal, and is unavailable where gas, consciousness or time is goneNext axiom: 4.2, The provoked/unprovoked boundary is a causal classification that moves most diver injuries out of the headline number →Link to this axiom: #ax-4-1
4.2 The provoked/unprovoked boundary is a causal classification that moves most diver injuries out of the headline numberVERIFIEDISAF labels a bite provoked when the human initiated contact, including feeding, spearfishing and touching, so diver bites leave the unprovoked headline and must be read from diver records.Open this axiomClose this axiom
The axiom in full

The ISAF files a bite as provoked when the human initiated the interaction, and its definition names divers harassing or touching sharks, spearfishers, people feeding sharks and people unhooking them; in 2025 the provoked column was 29 of 105 investigated cases (about 28 percent) and the unprovoked activity split was swimming and wading 46 percent, board sports 32, snorkeling and free-diving 15, other 6, with scuba not a listed category. Population: shark-incident registries (ISAF worldwide 2025; Australian divers 1960 to 2017; the Australian Shark-Incident Database 1791 to 2022). Outcome: incident and fatality counts by classification and activity, no denominator. Evidence class: CORRELATIONAL registry. The only opened series isolating scuba (Australia) holds 187 attacks on divers (112 snorkel, 62 scuba, 13 surface-supply), 28 deaths (13, 8, 7), the white shark in 18 of 21 identified fatal events (tiger 2, bull 1), and 21 of 28 divers killed were collecting seafood (spearfishing 9, abalone 5, scallops 4); most would file as provoked. The headline unprovoked total describes swimmers and surfers, not divers.

The physics underneath

Classification is a judgment about who initiated contact, not about the shark's state or the wound; the rule's listed modes (feeding, spearfishing, touching, unhooking) are the modes in which divers meet sharks at contact range, so the diver numerator migrates to the provoked column. A Poisson count with mean near 65 swings about plus or minus 16 a year at one standard deviation, so 65 in 2025 against a 2020 to 2024 average of 61 carries no trend.

The sources
← Previous axiom: 4.1, Shark-bite counts and shark-bite risk are different instruments, and the count moves opposite to the rateNext axiom: 4.3, Where, when and what the diver is doing move exposure-adjusted shark risk by orders of magnitude; species and gear move it far less →Link to this axiom: #ax-4-2
4.3 Where, when and what the diver is doing move exposure-adjusted shark risk by orders of magnitude; species and gear move it far lessVERIFIEDSeason, region and task set the encounter rate and the bite-given-encounter term, so moving a diver in space, time or activity changes risk by orders of magnitude.Open this axiomClose this axiom
The axiom in full

In the one activity-stratified, exposure-adjusted series opened (California white sharks 1950 to 2013, Ferretti et al. 2015, Frontiers in Ecology and the Environment, read through the Stanford release and KQED), season moved a surfer's per-session risk about 24-fold (March versus October and November, Mendocino County), region moved it about 1,566-fold (Los Angeles to San Diego in March versus Mendocino in autumn), and activity moved it about 100-fold (abalone divers about 1 in 1.44 million diving days; surfers about 1 in 17 million; scuba divers about 1 in 136 million; swimmers about 1 in 738 million beach visits; 2013 estimates). Population: Californian ocean users and the California white shark record. Outcome: injurious bites per activity-day or visit, modeled. Range: one species, one coast; transfer elsewhere is argument. Evidence class: MODEL on a CORRELATIONAL registry with participation-survey denominators. The same model put a California scuba diver about 6,897 times more likely to be hospitalized for decompression sickness than bitten by a white shark.

The physics underneath

Risk per session equals encounter rate times bite-given-encounter times injury-given-bite; season and region act on the first term through white shark and pinniped seasonality, activity acts on the first two (abalone divers work the bottom of pinniped habitat, surfers sit in the surf zone as surface silhouettes, scuba divers are submerged, loud and geometrically unlike prey). The Australian diver record agrees in direction: fatal cases cluster on seafood collection and proximity to fishing or seals, and the ISAF provoked rule names handling and feeding as the modes that put a diver in the bitten column.

The sources
← Previous axiom: 4.2, The provoked/unprovoked boundary is a causal classification that moves most diver injuries out of the headline numberNext axiom: 4.4, A shark's olfactory threshold is a concentration, and the folklore converts it into a distance by deleting the transport term →Link to this axiom: #ax-4-3
4.4 A shark's olfactory threshold is a concentration, and the folklore converts it into a distance by deleting the transport termVERIFIEDOdorant must be carried above a measured concentration threshold to the nares by current and mixing, so detection range is set by transport physics, and detection is not attack.Open this axiomClose this axiom
The axiom in full

Measured elasmobranch olfactory thresholds for amino acids run about 10^-9 to 10^-6.9 mol/L across five species (electro-olfactogram, laboratory, 2010), no lower than bony fishes; a bleeding diver is detected only when a water parcel carrying odorant above threshold reaches the nares, which depends on current, mixing and dilution, so 'blood from a mile away' is a threshold converted into a distance with the transport law deleted, and detection is not approach and approach is not a bite. Population: captive elasmobranchs in flume and electrophysiology studies (smooth dogfish, blacktip, bonnethead, nurse, leopard sharks, the five threshold species). Outcome: receptor threshold concentration; orientation and capture in a flume; tracked navigation paths. Dose: laboratory amino-acid stimuli, never whole human blood. Evidence class: INTERVENTION (sensory blocks, controlled stimuli) and MODEL (advection-diffusion transport). No field measurement of blood detection distance exists in the opened record.

The physics underneath

Transport, not sensitivity, sets range: diffusion alone spreads a drop over a pool-sized volume, bulk flow carries a thin, patchy filament downstream, and at 0.1 m/s a parcel takes about 4.5 hours to travel a mile while dilution drives it under threshold (a model figure). Sharks steer by nostril arrival time (smooth dogfish turned toward the first-stimulated nostril even when the later signal was stronger). With senses blocked singly and in combination, odour was the first signal and drove upstream swimming, nurse sharks could not feed without it, and capture switched to vision, lateral line and electroreception, which triggered jaw depression in all three species; electroreception is a sub-metre sense. Anosmic leopard sharks displaced 9 km took more tortuous, less shoreward paths, which is kilometre-scale use of ambient gradients over hours, not instant point-source detection.

The sources
← Previous axiom: 4.3, Where, when and what the diver is doing move exposure-adjusted shark risk by orders of magnitude; species and gear move it far lessNext axiom: 4.5, A shark's observed response to a diver is conditional on species, site, diver movement and distance, the animals seen are a selected sample, and proximity cannot read a mental state →Link to this axiom: #ax-4-4
4.5 A shark's observed response to a diver is conditional on species, site, diver movement and distance, the animals seen are a selected sample, and proximity cannot read a mental stateVERIFIEDShark response is a function of stimulus, state and context, and departing animals leave the sample, so observed proximity describes movement under those conditions and identifies no intent.Open this axiomClose this axiom
The axiom in full

In field observation, shark response category (approach, no visible reaction, alert, evasion) varied with species, location and the diver's behaviour, with direct approaches, sudden movements and distances under about 4 m eliciting stronger responses (five species, two marine reserves, 2011); sharks that leave when divers arrive are under-represented in what divers see (pelagic threshers abruptly changing course when recreational divers arrived at cleaning stations, 1,230 camera hours over 232 days); Caribbean reef sharks approached kneeling divers more often from outside the visual field, the pattern vanishing when a second diver covered the blind side, larger sharks nearer the sea floor, turn-away at about 1.0 to 2.0 body lengths (standardized video, 2015); and an approach alone does not identify its cause (olive sea snakes approached divers in 39 of 58 male and 35 of 100 female encounters across 158 focal observations, a sex and season pattern fitting misdirected courtship, no bites underwater, 2021). Population: free-living reef and pelagic sharks observed by divers or remote cameras; sea snakes as the control case. Outcome: behavioural categories, approach direction and distance, approach counts; no injury rate. Evidence class: CORRELATIONAL field observation, one standardized quasi-experiment. 'Calm', 'curious' and 'precise' are legitimate only as operational descriptions of measured movement (steady speed and tailbeat with no agonistic display over a defined interval; repeat passes at decreasing distance with no contact, against a control; repeatable trajectories and a measured contact-to-approach ratio), never as inferred intent, and each is compatible with effective predation.

The physics underneath

An animal's response is a function of stimulus (the diver's size, movement, bubbles, orientation, distance), state (feeding, reproductive, defensive) and context (provisioning, visibility, escape geometry); the same movement choices are consistent with assessment, caution and stalking, and observational data cannot pick among them. Observation selects animals: if some leave on diver arrival, those remaining are not a random sample, so a collection of comfortable close encounters under-represents avoidance. Even a plausible sensory model (surfer and pinniped silhouettes filtered through juvenile white shark acuity, 2021) does not identify mechanism, as its 2023 critique (ignores other senses; an ontogenetic exploratory alternative) shows.

The sources
← Previous axiom: 4.4, A shark's olfactory threshold is a concentration, and the folklore converts it into a distance by deleting the transport termNext axiom: 4.6, Provisioning raises encounter rate, residency and competition at a known point and conditions animals to hands and food; its safety record splits handlers from guests and transfers to no other water →Link to this axiom: #ax-4-5
4.6 Provisioning raises encounter rate, residency and competition at a known point and conditions animals to hands and food; its safety record splits handlers from guests and transfers to no other waterVERIFIEDFeeding compresses sharks, food and people into one predictable volume, raising encounter rate and food-conditioned contact locally, so a feed-site safety record does not transfer elsewhere.Open this axiomClose this axiom
The axiom in full

At provisioning sites the measured effect of feeding is on presence, residency and learned association, not on temperament: tagged bull sharks at Shark Reef, Fiji (82 individuals, up to 16 receivers, 114,282 detections, 1 to 486 days) visited on feeding and non-feeding days alike but stayed longer on feeding days (2.6 ± 1.2 versus 2.0 ± 1.0 consecutive hours) and still left for days to months (2013); sicklefin lemon shark residency rose over 949 surveys in 44 months at Moorea (36 adults) with intra- and interspecific aggression observed (2010); eight species held site fidelity at Fiji 2003 to 2012 with more individuals of fewer species over time (2014); satellite-tracked tiger sharks kept large-scale movements while locally attracted to chumming (2012); the two Moorea diver bites analysed forensically (2004 hand-feeding tuna heads in a mesh glove; 2008 photographing with hand and strobe movements) name hand-feeding the facilitating factor (2016); and by 1998 more than a dozen bites had occurred on Bahamas feeding expeditions, mostly on handlers, two tourists lightly hurt, with airlifts that did not make the newspapers. Population: provisioned reef sharks (bull, lemon, Caribbean reef, tiger) at commercial feed sites in Fiji, French Polynesia and the Bahamas. Outcome: presence and residency hours, species composition, bite case reports. Evidence class: CORRELATIONAL telemetry and survey time series, case series. No exposure-adjusted bite rate per provisioned dive exists in the opened record; any such rate must separate the handler holding bait from the guest behind the line.

The physics underneath

Provisioning converts a stochastic encounter into a scheduled one: it compresses animals, food and humans into one small volume at a predictable time, raising the encounter term directly and the bite-given-encounter term through competition, food arousal and learned association of hand movement and exposed skin with reward; large-scale movement persists, so the effect is local and conditional. Ascertainment runs against a financial gradient (employee injuries at commercial sites are least likely filed; DAN reports are voluntary). An operator's 'million encounters without one incident' is testimony with no injury definition or exposure; as a bound it says only that the per-encounter fatality rate at a managed site is very low.

The sources
← Previous axiom: 4.5, A shark's observed response to a diver is conditional on species, site, diver movement and distance, the animals seen are a selected sample, and proximity cannot read a mental stateNext axiom: 4.7, Non-shark marine injuries are dominated by accidental contact with a defensive animal and need their own denominators; the crocodile is the contrast class where lethality given attack is high →Link to this axiom: #ax-4-6
4.7 Non-shark marine injuries are dominated by accidental contact with a defensive animal and need their own denominators; the crocodile is the contrast class where lethality given attack is highVERIFIEDMarine injuries arise mostly when a foot or hand meets a camouflaged defensive animal, so patient proportions by species measure contact, not predation, and give no per-encounter diver risk.Open this axiomClose this axiom
The axiom in full

In 4,092 adults treated for marine-animal injury in one US multi-site hospital system 2014 to 2019, stingrays caused 1,196 (29.2 percent), catfish 974 (23.8), other fish 541 (13.2) and jellyfish 421 (10.3); the foot (1,449; 35.4 percent) and hand (1,181; 28.9 percent) were the sites; 98 percent were discharged and 15 (0.4 percent) admitted; no link to scuba was recorded. The case mix records stepping on and handling camouflaged, barbed or spined animals, not predation; a proportion of patients by animal estimates nothing about a diver's risk per encounter. Population: hospital-treated marine-animal injury patients across all water activities; 393 stingray-injured people followed after the sting; Australian crocodile attack victims. Outcome: case counts by animal and site, admission, outcome given sting, survival given attack. Evidence class: CORRELATIONAL clinical series and registry; MODEL for crocodile survival. The one group where lethality given attack is high and size-structured is the saltwater crocodile: 87 unprovoked Australian attacks 1970 to 2014, 27 fatal; a 75 kg person in the water survives with probability about 0.81 against a 300 cm animal, 0.17 against 400 cm, under 0.05 above 450 cm; drowning the main cause of death. No opened source supplies a per-dive or per-encounter rate for any large animal, and cetacean injury studies measure injuries to the animals, so no pooled ranking of whales, dolphins, seals, rays and sharks by diver risk can be built.

The physics underneath

Three estimands hide under 'dangerous animal': encounter rate (habitat overlap, camouflage), contact given encounter (the human's foot or hand meeting the animal), injury given contact (barb, spine, venom, mass ratio). The stingray follow-up estimates outcome given sting, never sting given encounter. The crocodile model shows the third term can be large and governed by mass ratio and position in the water, the opposite structure to sharks, where the first two terms are tiny for a submerged diver. An unprovoked tiger shark bite on a scuba diver in December 2020 with haemorrhagic shock fixes possibility and severity, not frequency.

The sources
← Previous axiom: 4.6, Provisioning raises encounter rate, residency and competition at a known point and conditions animals to hands and food; its safety record splits handlers from guests and transfers to no other waterNext axiom: 4.8, For a scuba diver the lethal channel from an animal is the response it triggers, an uncontrolled ascent or panic cascade, not the wound; shark attack is about 3 percent of the Australian diving fatality record →Link to this axiom: #ax-4-7
4.8 For a scuba diver the lethal channel from an animal is the response it triggers, an uncontrolled ascent or panic cascade, not the wound; shark attack is about 3 percent of the Australian diving fatality recordVERIFIEDAn animal startle at depth triggers breath-hold or rapid ascent, coupling a minor wound to gas-embolism physics, so the diver's response, not the bite, carries the lethal risk.Open this axiomClose this axiom
The axiom in full

Death from shark attack accounts for about 3 percent (28 of 926) of recorded Australian diving-related fatalities 1960 to 2017, while case literature shows the route by which a small animal can kill a diver: a moray that seemed to associate divers with food seized a diver's forearm at 30 m and the diver inflated his buoyancy device and made an emergency uncontrolled ascent (2004 case report), a pathway to arterial gas embolism that a forearm laceration alone does not carry; sea snake researchers name the same channel, that an encounter can imperil divers through inducing panic even if the snake does not bite (2021). Population: recreational scuba divers in dive-accident fatality records (Australia, DAN Asia-Pacific) and published case reports. Outcome: fatality share by cause; case-report ascent behaviour. Evidence class: CORRELATIONAL fatality-registry share plus case reports; the ascent-cascade frequency is measured nowhere in the opened record, which is why the tier claimed is PROBABLE.

The physics underneath

Severity of an animal injury is bounded by anatomy and bite mechanics, but the diver's gas physiology is not: a startle at depth that produces breath-hold, rapid ascent or a lost regulator couples the encounter to the pressure, breathing and panic mechanisms where the accident record's fatal causes live. DAN's reported call mix runs on ear and sinus barotrauma, decompression sickness and marine envenomation, the envenomations typically on entry, exit or handling and rarely life-threatening. The California model's ratio (decompression sickness hospitalization about 6,897 times likelier than a white shark bite for a scuba diver) sets the same scale from the other side.

The sources
← Previous axiom: 4.7, Non-shark marine injuries are dominated by accidental contact with a defensive animal and need their own denominators; the crocodile is the contrast class where lethality given attack is highThis is the last axiom on the page.Link to this axiom: #ax-4-8

The easy answers this run refused

Each vector was told the common saying it must not settle for, and told to rebuild the answer from first principles instead.

Pressure, breath and the animals run on mechanisms, and every mechanism on this page can be checked again.