82.1°F

Content note: this article discusses a fatal family case and a severe suicide attempt. Optional exhibit views contain blood and post-treatment debris; they stay closed unless selected.

Neurogenic shock can make a patient cool faster. That does not tell us whether it supplies nine missing degrees in 81 minutes. The mechanism has two thermally relevant parts: less heat made by shivering and more heat carried from core to skin. Both can be put in watts.

This article was written while the trial was still running. The jury retired on August 27th, reported itself deadlocked on September 1st and was sent back to deliberate further. It argues neither that Lindsay Clancy is guilty nor that she is not, and asks only whether one reported temperature can tell the competing accounts apart.1

The public record reviewed here does not show the reading's exact time, measuring site, device or repeat curve. A doctor remembered active warming but could not recall whether the hypothermia was present on arrival or developed during treatment. Later testimony gave 95.2°F. If serial readings show that core temperature was already rising at 82.1°F, an earlier minimum was lower; active warming alone does not establish that curve.2 A reported GCS of 10 establishes serious impairment, but not whether drugs, trauma, shock or hypothermia caused it—and not when the cooling began.3

The event timeline—not a made-up starting clock—sets the test. The prosecution treats a 5:38:33pm iPhone stair event as Lindsay going upstairs. Patrick says he found her outside shortly after a recorded 6:09pm unanswered call. Responders later found her outdoors, and Pembroke moved her into an ambulance. Giving that courtroom chronology its longest cooling-favourable edges produces an 81-minute model: 52 minutes on the ground, six outdoors on a backboard, eighteen in transport and five in early hospital care. Every pre-responder link remains conditional.

If 82.1°F was early and reliable · extrapolated model

Use the longest cooling-favourable version of the courtroom timeline. The coldest of four baseline 81-minute runs is still about 9.0°F too warm.

Reported hospital value

82.1°F

Four baseline moisture runs

91.1–92.5°F

Closest baseline result 91.1°F about 9.0°F too warm

Reading metadataUnknown

The exact time, site, device and serial temperature curve are not in the acquired public record.

Earlier cooling163–196 min

Ordinary boundary runs need this much total model time, 82–115 minutes beyond the 81-minute stress-test clock.

Extreme water stress test35–43 min

Uninterrupted broad direct-skin flow—not wet clothes or one soaking. The scene record does not establish it.

Show the four 81-minute endpoints and limits

Dry92.5°FSnow-wet91.7°FSaturated after wetting91.1°FFive-minute declared flowing-water run91.1°F

The 52-minute ground phase assumes that the 5:38 phone event was Lindsay going upstairs, that she exited immediately afterward, and that direct contact continued to the latest working board time. Six boarded minutes use the latest working ambulance edge. Those are cooling-favourable choices, not observed durations. All four endpoints are below 35°C and therefore extrapolated; none is below 32°C.

The limited conclusion: the named snow-contact and retained-wetness runs do not reproduce 82.1°F even at the cooling-favourable upper edge of the courtroom chronology. If the phone event or Patrick's sequence is wrong, this clock disappears; the model does not prove flowing water, indoor cooling, a particular history or a culprit.

Model thermal-fixed 0.1.0 · method notes 7–8 and 14–18 · download inputs, code and tests

Give the courtroom account its longest clock

The record independently puts Lindsay outside only when responders arrive, at about 6:14–6:21pm. Everything earlier depends on attribution.4

When did Lindsay go out the window? The record doesn't say.

Without Patrick or Lindsay's accounts: the earliest exit remains unknown. Responders independently place her outside by about 6:14–6:21pm.

Conditional courtroom window: if the 5:38 stair attribution and Patrick's discovery account are accepted, the working board range makes direct ground contact roughly 9–52 minutes and total outdoor time no more than about 58 minutes. That is a reconstruction, not a measurement.

5:38:33pm: Phone record. A stair event is recorded; the carrier is not.
6:09pm: Phone record. Lindsay's phone is not answered.
About 6:09–6:11pm: Patrick's account. He supplies the search and discovery sequence.
About 6:14–6:21pm: Responder estimate. First independent exterior confirmation.
Roughly 6:20–6:36pm: Derived ranges. Long-board placement and movement into the ambulance.

The 9–52-minute interval is a reconstruction, not a measurement. Without Patrick or Lindsay's accounts, responders establish only that she was outside by about 6:14–6:21pm; the earliest exit remains unknown. Sources and limits: notes 2–5.

A long backboard reduces snow contact; an ambulance changes air movement, radiant surroundings and insulation. Core temperature can keep falling after rescue as cold tissue draws heat inward, but ground contact, board placement and entry into a vehicle are not thermally equivalent.5

How the clock was builtResponder sequence, the neighbour clue and the weather proxy

Duxbury's first ambulance crew treated Lindsay first and placed her on a long backboard before or around Pembroke's arrival. Pembroke found her outdoors, bandaged and collared on the board, then reported roughly seven minutes on scene and eighteen minutes in transport. Public testimony does not fix when she entered the ambulance.

The chronology has different kinds of evidence. The phones and watch record activity without proving the operator or wearer. Retail video independently fixes Patrick at CVS and ThreeV. The 5:34 call is recorded, but Patrick supplies the speaker's identity, words and apparent condition. The 5:38 stair event is recorded, but the prosecution supplies its meaning. Patrick supplies the discovery sequence after 6:09. Those distinctions are part of the timeline, not footnote trivia.

A second-hand neighbour account may place a woman in the yard at about 5:50pm. If authenticated and shown to concern Lindsay, it would establish that she was outside at that moment. It would not establish when she first went out or prove continuous exposure until police arrived. Without its call log, interview or sworn statement, it remains a clue.

Nearby airport observations put the evening air around 1–2°C with light wind. Those are regional anchors, not backyard measurements of air, radiant temperature, wind at body height or the snow-contact surface. The simulator exposes each quantity instead of treating a regional report as a backyard measurement.6

Severe cold exposure moves heat in hundreds of watts

The closest direct human watt measurement found for this comparison is severe cold-water immersion. In six healthy men, the highest reported 30-minute group means were 558W crossing the skin and 369W net loss of stored body heat after ongoing heat production was counted. Those numbers cannot simply be copied onto an acutely injured patient. They establish the scale: severe cooling moves heat in hundreds of watts. Kulkarni et al. (2019).

A core reading is not a whole-body temperature. Skin and limbs cool first, so 98.0°F to 82.1°F cannot simply be multiplied by the full body mass. Final skin temperature is missing. The figure starts with the smallest bill in the displayed sensitivity—mean skin set equal to the reported core—and puts the colder-skin cases in the audit drawer. Even that smallest version needs about 373W leaving the body on average across the 81 minutes.7

Core-plus-skin heat-content sensitivity

Healthy-human scale: in six healthy men immersed to the sternum in stirred water cooling to 8°C, the highest reported 30-minute group means were 369±195W net stored-heat loss and 558±35W cutaneous loss (mean±SD).

Declared final mean skin	Stored heat missing	Net rate over 81min	Outward rate with declared heat inputs
27.8°C, as warm as core	1.38MJ	285W	373W
25°C	1.57MJ	324W	412W
20°C	1.91MJ	392W	481W
15°C	2.24MJ	461W	549W

The rates change during cooling. In the saturated 52-minute stress test using the declared regional-flow proxy and 0W shivering, mean shell temperature falls from 32.9°C to 18.3°C. With the displayed boundary conditions held fixed, total body-boundary loss—including about 9.7W of direct respiratory loss—falls from 553W to 287W while core-to-shell transfer rises from 63W to 228W. The changing shell pathways drive the fall.

The calculation uses the published relation mean-body temperature = 0.64×core + 0.36×mean skin, with a declared 98°F core and 32.95°C mean-skin start. Final mean skin is not in the public record. The study values are not a case analogue, sustained ceiling or case confidence interval.

Under the displayed mean-skin ≤ core assumption, the equal-temperature endpoint is the smallest bill shown: about 1.38MJ. It is not a patient lower bound. Across the full 81-minute upper clock, that is 285W of net stored-heat loss. Adding the declared 70W metabolism and 300W of external warming during the final five minutes makes the average outward requirement about 373W. The saturated reference run loses only 0.80MJ across those 81 minutes and ends at 91.1°F—about 0.59MJ short of even this displayed minimum. Every number is conditional on the declared start, mass, tissue properties, treatment inputs and reading metadata. Equation and limitations: note 7; healthy-human watt comparison: note 17.

Neurogenic shock still has to supply watts

Trial testimony described two ways the spinal injury could accelerate cooling: impaired shivering and dilated skin vessels. It supplied no patient cooling curve, blood-flow measurement or multiplier. The model therefore grants the first effect completely—shivering is 0W in every row—then widens the core-to-skin link from a regional proxy through the source model's ceiling and, finally, to the impossible limit of no internal resistance. The declared regional proxy adds 17W. Reaching 150W takes the source model's own input ceiling; 179W takes deleting internal resistance altogether, a mathematical limit rather than a physiological state.8

Theoretical spinal-injury watt sensitivity

Core-to-skin coupling test	Average net stored-heat loss during the 52-minute ground phase	Change from low-flow comparison	Core after the full 81-minute sequence
Low-flow comparison	283W	—	93.0°F
Regional-flow proxy	300W	+17W	91.1°F
Both regions at neutral flow	324W	+41W	89.0°F
Source-model input ceiling	432W	+150W	83.9°F; 83.3°F low
No internal bottleneck, K→∞	462W	+179W	83.3°F; 82.23°F low

Every row uses 0W shivering. The last two are boundary tests, not a patient-plausible neurogenic-shock range.

Relative to the 0.5/0.5 low-flow comparison, the declared 0.5/6.3 regional proxy adds 17W. Both regions at neutral flow add 41W; the source-model input ceiling adds 150W; K→∞ adds 179W. The 17W difference is about 53kJ over the 52-minute phase. These are model sensitivities, not measured acute-SCI watts. K→∞ also starts with a 1.34°F fall in displayed core caused by energy-preserving mixing with the declared cooler shell; it is not a like-for-like physiological starting state. Mechanism, human comparisons and limits: notes 8 and 14.

Engineered vascular heat sink

A palm cooler reached a maximum reported group mean of 42±12W

One cooled palm · maximum reported group mean42W

Displayed 81-minute outward ruler · conditional373W

Ten healthy men were heated to an oesophageal temperature of 38.8°C, kept in 42.2°C air and given a 10°C plate against one palm for 50 minutes. Maximum heat extraction averaged 42±12W and declined with time; core fell about 0.38±0.21°C. Adding suction changed neither result. Kuennen et al. (2010).

Use it for scale, not as a ceiling or a like-for-like comparison. The 42W value is a reported maximum group mean; 373W is an 81-minute conditional average, so the bars do not share an averaging window. A palm is also a small specialised surface, not a wet body or a hose. The useful point is narrower: opening a vascular route helps only when an external sink carries the heat away, and this purpose-built route produced a measured rate in the tens of watts.

What is left after the watt test?

If 82.1°F and the short clock both stand, the cooling has to start earlier or become much harsher

1 · Check the reading

Was 82.1°F an early central value?

The exact time, site, device and repeat curve could change the target or the clock. Those records remain missing.

2 · Check the starting state

Was she already cold at 5:38?

Pre-window cooling changes the starting temperature instead of asking the spinal injury to pay the whole bill after the fall. A drug-impaired period in a cold bedroom is one testable version, not a result of this model.

3 · Check the boundary

Was there uninterrupted cold-water flow?

The only short water run that reaches 82.1°F requires 35–43 minutes of broad direct-skin flow. One soaking does not do it. The photographs show a hose reel, not use.

4 · Check the physiology

Was coupling near a mathematical limit?

The source-model ceiling still finishes 1.8°F warm. Only zero internal resistance nearly reaches the guide, and that is not a physiological neurogenic-shock estimate.

The alleged 5:50pm neighbour sighting matters if it can be authenticated. If the observer saw Lindsay lying outside without active water, it cuts the extreme-flow window to roughly eleven minutes after the 5:38 stair event—far short of the model's 35–43 minutes. With the reading accepted and that water branch removed, pre-window cooling becomes the main untested thermal history. It still does not identify a drug, a room temperature, an actor or a cause.

Audit exactly how each study is usedEquations, borrowed numbers, comparisons and limits

Which sources actually change the model?

Source-to-calculation map

No paper gives a cooling rate for Lindsay. The papers supply formulas, an estimate of whole-body heat capacity and one mean retained-water mass. The volunteer cooling rates appear only for comparison. We did not tune the model until it matched them.

Most case-specific inputs are ours. We chose the body and starting state; the timing, clothing and ground contact; the backyard and water conditions; and the packaging and warming. None is a measurement of Lindsay or a confidence interval. We use standard engineering values for water properties and emissivity.

Changes the calculation

Used in model · equationGagge + ASHRAECore, skin, air, clothing and evaporation formulasExact use and limit

Source supplies: Gagge supplies the core-to-skin energy balance and blood-flow conductance. From ASHRAE’s D4 code we borrow the air-speed formula, 0.155 clo conversion, clothing-area factor, sea-level evaporation relation, complete breathing formulas 0.0014M(34−T_a) and 0.0173M(5.87−P_a), 0.06 skin-diffusion term and 0.5/6.3/90L/h/m² minimum, neutral and maximum-flow settings. The second breathing formula is converted from D4’s Torr units. We adapt: those pieces into one fixed core and two fixed outer regions.

Does not establish: Neither source tests that rewrite in acute trauma or below 35°C (95°F). The 90 setting is an algorithm ceiling, not the maximum human skin flow. We implicitly fix air pressure at one atmosphere instead of exposing ASHRAE’s pressure correction. Combining settings across body regions is our choice, not a patient estimate. Gagge, Stolwijk & Nishi (1971; linked 1972 reprint); ASHRAE Handbook two-node model; ASHRAE 55 Addendum g (approved 2022).

Used in model · numberXu, Rioux & Castellani (2022)2.980kJ/kg/K body heat capacityExact use and limit

Source supplies: We use 2.980kJ/kg/K as the estimated heat needed to cool one kilogram of body tissue by one degree. We rerun the model at 2.443 and 3.339 to show how sensitive the result is to the paper’s tissue-property assumptions.

Does not establish: Those values are not a measured range for Lindsay, a confidence interval or a conversion from core temperature to average body temperature. Read the study.

Used in scale check · equationLenhardt & Sessler (2006)Mean body = 0.64×core + 0.36×mean skinExact use and limit

Source supplies: The Burton relation for estimating mass-weighted mean-body temperature from core and mean-skin temperatures. The validation compared it with measured tissue heat content across major thermal perturbations.

Does not establish: It supplies no skin temperature or start state for Lindsay. The authors reported poorer performance during rapid cooling and at the coldest conditions; their ±0.42°C validation spread is not a case confidence interval. Read the study.

Used in model · equationDu Bois + Buck + NASASurface area, air moisture and a water-flow safety limitExact use and limit

Source supplies: Du Bois estimates body surface area. Buck estimates how much water vapour air can hold. The NASA heat-exchanger formula prevents the water pathway from removing more heat than the contact surface or captured water flow could carry.

Does not establish: These sources supply no body measurements, wet area, water coefficient, flow, duration or evidence water was used. Du Bois; Buck CR-1A manual; NASA report.

Used in model · one study meanHenriksson et al. (2015)Mean 1.232kg retained water (SD 0.091kg)Exact use and limit

Source supplies: An underwear ensemble with an approximate 575g dry weight retained a mean 1.232kg of water (SD 0.091kg). We choose: that mean as the saturated and renewed-water presets’ moisture ceiling, assume the entire amount could evaporate and assign 45% to the upper and 55% to the lower region. We show the study’s temperature curve separately for comparison; it never changes a run.

Does not establish: The study did not test Lindsay’s garments or measure their water distribution. It does not validate this model’s temperature predictions. Read the study.

Comparison or mechanism only

Not used in model · comparisonHelland et al. (2025)0.82°C/h (1.5°F/h) wet-air comparisonWhy it is not a fit target

Displayed only as: Volunteers in wet cotton and cold air whose shivering was reduced cooled at 0.82°C/h (1.5°F/h). We show that rate beside the model; it never changes a run.

Does not establish: This was not trauma, snow contact or validation near 27.8°C (82.1°F). Read the study.

Not used in model · comparisonKulkarni et al. (2019)369W highest 30-minute net mean in stirred-water immersionWhy it is not a fit target

Displayed only as: During stirred-water immersion to the sternum, the highest reported 30-minute group means were 369W net stored-heat loss (SD 195W) and 558W crossing the skin (SD 35W). Net loss accounts for heat the participants were producing; skin loss does not. The final-ten-minute core slope was −2.4°C/h (SD 1.7). None changes a model run.

Does not establish: A highest 30-minute group mean is not an 81-minute sustained average; participant SDs are not case error bars; sternal-depth immersion is not partial runoff. Read the study.

Not used in model · comparisonKuennen et al. (2010)Maximum reported group mean: 42±12W from one palmWhy it is not a ceiling

Displayed only as: In ten healthy hyperthermic men, a 10°C palm plate reached a maximum reported group mean of 42±12W and core fell 0.38±0.21°C over fifty minutes. Mild cyclic vacuum added no measurable improvement. None of those values changes a run.

Does not establish: The reported maximum is not an 81-minute sustained average. One palm is not broad water contact or an upper bound on hosing. The participants began hyperthermic in 42.2°C air and were not cooled through normothermia into hypothermia. Read the study.

Not used in model · comparisonGrissom et al. (2004)0.7–1.2°C/h (1.3–2.2°F/h) snow-burial comparisonWhy it is not a fit target

Displayed only as: Healthy volunteers buried under snow while insulated cooled at 0.7°C/h (1.3°F/h) with normal blood CO₂ and 1.2°C/h (2.2°F/h) with elevated CO₂. We show those rates only for comparison; they never change a model run.

Does not establish: Full insulated burial in healthy volunteers is not partial ground contact with acute injury. Read the study.

Not used in model · mechanismAcute SCI guidance + autonomic standardsWhy blood flow and shivering must be tested separatelyExact use and limit

Supported mechanism: High thoracic SCI can impair vasomotor temperature control below the neurological lesion; shivering above it may remain. The international standard says the degree depends on measured neurological level and completeness—not a vertebral label alone. Acute guideline; ISAFSCI standard; Downey et al..

Does not establish: The acute guideline found no acute-phase thermoregulation study. These sources supply no multiplier, watts, injured-skin flow or patient cooling rate. They justify exposing two controls; no number is imported.

Not used in model · comparisonThree chronic-SCI cold studiesMeasured direction and scale—not an acute T5–T6 rateWhy they are not fit targets

Displayed only as: In a controlled 18°C test, seven men with chronic C3–C7 tetraplegia lost 1.2±0.12°C rectally over 109±15.9 minutes on average; seven controls had no core decline. Older chamber data varied with lesion level. In an uncontrolled −15°C blizzard, one C8-incomplete skier's sublingual reading fell 3.9°C in 70 minutes. Handrakis et al.; Guttmann et al.; Aguilera et al..

Does not establish: These were chronic injuries, mostly cervical, with different temperatures, activity and measurement sites. They do not validate an acute trauma model or define case error bars.

Not used in model · definitionsDitunno + Summers“Spinal shock” is not one measured vasodilation stateExact use and limit

Terminology: Spinal shock describes evolving depression and recovery of spinal reflexes. Neurogenic shock is a circulatory syndrome from autonomic disruption. A nine-patient acute series found several different hemodynamic patterns rather than one uniform state. Ditunno et al.; Summers et al..

Does not establish: Neither term supplies a heat-transfer coefficient. Hemorrhage, fluids, treatment and changing circulation remain separate physiological questions.

The spinal-injury evidence is now a stress test, not a shrug. The studies support impaired vasomotor control and show faster core decline in some chronic high-SCI cold exposures. They do not fix the direction or size of acute whole-body heat loss for this injury. The section below therefore shows model sensitivities, not a correction factor.

Only one study result enters a preset: the mean 1.232kg retained-water mass. Its regional split and evaporation cap are ours. Open a row for the transfer and limit; titles, URLs and code hooks are in the complete research pack. Equation derivations and limits: notes 7–8 and 14–18.

Show the inverse time and extreme-water solutionsExact model inputs and afterdrop audit

What would reach the target?

Mathematical solutions are not equally plausible histories

Sensitivity tests—not estimates of what happened.

Keep the 81-minute setup. Add ground time.

Total model time, including the same 29 board, transport and early-hospital minutes.

Longest courtroom-chronology run shown above 52 ground + 6 board + 18 transport + 5 early hospital81 min total

Dry
196 min
Snow-wet
175 min
Saturated after wetting
163 min
5-min water run
163 min

These are inverse model solutions, not estimated exposure times or confidence intervals.

Change several inputs

Use extreme continuous water

0°C water35 min 4°C water39 min 8°C water43 min

What that takes: continuous direct-skin flow over 42–52%; 20L/min captured flow; maximum displayed transfer and skin blood flow; no shivering; no rescue phases.

This proves only that the declared input is sufficient in the model. It does not establish that water was used. If the reported 5:50pm sighting is authentic and showed no active flow, this uninterrupted branch does not fit the chronology.

Show every assumption needed for this result

Continuous direct-skin water over 42 per cent of the upper region and 52 per cent of the lower.
Maximum displayed water transfer: 80W/m²/K and 20L/min captured flow.
Maximum displayed skin blood flow: 15L/h/m² in both regions.
No shivering: zero modelled shivering heat.
Faster core–shell coupling choice: 30 per cent of heat capacity assigned to the shell; this is an uncalibrated compartment choice, not evidence of impaired defence.

This is broad, immersion-like coupling with no rescue phases, not ordinary wet clothing. The severe-hypothermia range remains an extrapolation.

Reproduce the 81-minute upper scenario and afterdrop resultExact inputs, endpoints and energy ledger

The upper scenario assumes a 98.0°F core, no shivering, 52 minutes on the ground, six minutes outdoors on a board, eighteen minutes in transport and five minutes of early hospital care. The four endpoints are 92.5°F dry, 91.7°F snow-wet, 91.1°F saturated and 91.1°F with the declared five-minute flowing-water phase.

Afterdrop does not close the gap in the saturated run. Stored body heat falls at an average 300W on the ground, then rises at about 42W in the ambulance even while core temperature falls another 1.3°F. The core is lagging behind the body's overall warming. Afterdrop can move the lowest core reading later; it cannot create another environmental heat sink.

What follows: the longest named courtroom-chronology runs do not reproduce 82.1°F. What does not: flowing-water use, indoor cooling or who caused anything.

Wet clothes are not flowing water

“Wet” hides three different physical problems. Snow-wet fabric loses some insulation. A one-off soaking also removes heat while the trapped cold water warms. A stream is stronger because it keeps replacing water the body has already warmed.

Snow-wet clothing Heat escapes faster

Some insulation is lost, but no new cold water arrives.

One-off soaking One dose of cold water

The first water takes heat. Then it warms, drains or evaporates.

Flowing water New cold water keeps arriving

Fresh water replaces warmed water, so rapid cooling can continue.

What each wetness model includesFinite moisture is not continuous flow

The retained-moisture presets begin after wetting. They include reduced insulation and finite evaporation, but omit an unknown first cold-water pulse and the changing heat capacity of water held in the fabric. The extreme-water test is different again: it imposes continuous direct flow.

The photographs fix two locations. Exhibits 37 and 38 show a turquoise wound hose reel beside the deck; responder testimony places Lindsay's treatment area by the far-right window well. A later close view shows the ground only after treatment and removal.9 10

Exhibit 38, an uncropped front-on night photograph of the rear elevation: a turquoise wound hose reel stands beside the deck at far left and a window well is at far right — **Exhibit 38.** 1 visible hose reel at the deck 2 reported treatment area by the far-right well

Exhibit 37, an uncropped oblique night photograph showing snow across the rear yard, the deck and turquoise wound hose reel, and the side of the house — **Exhibit 38.** 1 visible hose reel at the deck 2 reported treatment area by the far-right well

The testimony does not establish how wet her clothes were. Dr Elizabeth Laposata referred to wet clothing when explaining faster cooling, but did not identify the source of that observation. Located first responders described jeans, a tank top and socks without calling them dry, damp or soaked. Their silence does not show the clothes were dry; it leaves the moisture state unresolved.12

Can the bare patch measure heat or water?No—not without missing scene measurements

Water at 8°C is cold relative to skin and warm relative to snow. A stream could cool a body while melting or moving snow. But the post-treatment bare patch is not a second thermometer: without scale, capture order, snow properties and a measured water source, pixels cannot be converted into litres or joules. The image cannot separate thaw and drainage from body contact, blood, treatment fluid, trampling or directed water.

Separate question: were the clothes rinsed?Inspect the photographs, limits and a testable protocol

Maffeo said he saw no blood on the displayed item or top and did not recall any on the jeans; counsel then proposed that blood had seeped through to the bra, and he agreed. Lawler later described only “some” or “a little bit” of staining before identifying it from experience as blood. The public testimony reviewed here names no garment-level confirmatory blood or DNA result.13

The full public frames show diffuse mottled discoloration without an obvious concentrated outer-garment deposit. That is not proof of absence. The fabrics are dark, the wounds were uneven, the garments were cut and handled during care, and their drying history is missing. A photograph cannot establish blood, clot, crust or rinsing by sight.

Water-altered blood is plausible: dilution or washing can spread or blur a deposit while latent traces persist. General mottling is not a diagnostic rinse pattern or a validated garden-hose signature. A Day 6 bloodstain supervisor said red-brown staining requires confirmation. A Lindsay DNA profile on her own clothes would also be expected; even a profile mapped to a blood-positive area would not by itself prove rinsing, deposition mechanism or actor.

Exhibit 116, an uncropped photograph of cut blue jeans showing broad pale and blue mottled areas across the fabric — **Diffuse mottling is visible; no concentrated deposit is obvious in these reproductions.** Partial rinsing can reduce visible blood while leaving latent traces, but neither the discoloration nor the lack of an obvious deposit confirms blood or supplies a unique rinse signature. Admitted exhibits, source and preservation record: note 13.

Exhibit 117, an uncropped photograph of a cut grey and black lace bra with uneven dark and pale areas across the textured fabric — **Diffuse mottling is visible; no concentrated deposit is obvious in these reproductions.** Partial rinsing can reduce visible blood while leaving latent traces, but neither the discoloration nor the lack of an obvious deposit confirms blood or supplies a unique rinse signature. Admitted exhibits, source and preservation record: note 13.

A useful test would predeclare the expected difference between exposed and protected areas, photograph both faces, seams, folds and overlaps, map presumptive reactions, confirm human blood independently and tie any DNA result to the same mapped sample. Matched clothing should then undergo blood-then-water, snow contact, EMS cutting and handling. Rinsing loses support if the discoloration is not blood, the predicted spatial difference does not appear, or competing sequences reproduce the result as well.

Audit the spinal mechanism and medical evidenceAnimated heat circuit, chronic-SCI comparisons and equations

Spinal-injury stress test

How much cooling does the spinal injury buy?

Keep the same 81-minute clock. Change only the two proposed defences.

It could matter. A high spinal-cord injury can leave the body less able to close the radiator—constrict skin blood vessels—or switch on the furnace—shivering. But it does not create a third heat-loss pathway. Laposata described those mechanisms; she supplied no watts, multiplier or patient cooling curve.

Trial descriptionT5–T6 cord injury Possible effectless constriction and less shivering Not measured hereneurological level, completeness, skin flow or shivering watts

Already granted in every displayed 81-minute run 0 W shivering plus weaker flow control in a 55% lower-region proxy

Zero is the lowest possible shivering setting. The 45/55 body split and the 0.5/6.3L/h/m² blood-flow anchors are declared stress-test choices, not measurements or an anatomical map of a T5–T6 lesion.

Average heat circuit during the saturated 52-minute ground phase

Blood-flow test	Core to shell	Air + radiation	Snow + ground	Evaporation	Net body loss	Final core after 81 min
Both low-flow, 0.5 / 0.5	142W	176W	103W	64W	283W	93.0°F
Declared regional proxy, 0.5 / 6.3	186W	185W	109W	67W	300W	91.1°F
ASHRAE/Pierce ceiling, 90 / 90	423W	249W	145W	99W	432W	83.9°F
Zero internal resistance, K→∞	429W constraint transfer	262W	153W	107W	462W	83.3°F; 82.2°F low

Vasodilation widens the core-to-shell link; it does not remove the external bottleneck. The circuit uses the saturated preset and averages each displayed pathway over its 52-minute direct-ground phase. Breathing is drawn separately because it removes heat directly from the core; the other losses cross the skin or clothing boundary. Final core includes the same board, ambulance and early-ED phases. Arrow width and motion encode watts. Shell heat storage also changes, so internal and external arrows need not balance one another. Model equations and the source-model input ceiling: note 14.

More core-to-shell coupling keeps the shell warmerSustaining environmental loss · final core range across four moisture presets

Both regions low-flow 0.5 / 0.5

93.0–94.1°F

Declared regional proxy 0.5 / 6.3

91.1–92.5°F

Both regions higher-flow 6.3 / 6.3

89.0–90.8°F

Source-model ceiling 90 / 90

83.9–87.0°F

Zero internal resistance K → ∞; not physiology

83.3–86.7°F

The finite source-model settings stay warm; deleting the internal bottleneck gets close. The declared regional proxy makes the endpoint 1.6–2.0°F colder than the both-low-flow comparison. At the Pierce/ASHRAE model's 90-unit input ceiling across the whole model body, the saturated run bottoms at 83.3°F and finishes at 83.9°F. A second, deliberately impossible test forces core and shell to one temperature with zero internal resistance. It bottoms at 82.23°F—only 0.13°F above the guide—and finishes at 83.3°F after rescue warming. That is not a robust exclusion. It shows the stack required to nearly reproduce the reading in this model: zero shivering, broad saturation, the 81-minute upper-edge clock and internal transport pushed beyond any patient-calibrated setting.

Published run0 Wwhole-body shivering

The furnace is already completely off.

Round sensitivity100 W × 52 min= 0.312MJ returned

In the saturated run, the endpoint rises from 91.1°F to 93.7°F. The 100W is arithmetic, not a patient estimate.

Controlled, but not acute18°C dry air · chronic C3–C71.2°C in 109 min

Seven men with tetraplegia cooled about 0.66°C/h on average; seven controls had no core decline. Rectal measurement.

Extreme field case−15°C blizzard · chronic C8 incomplete3.9°C in 70 min

One active skier's sublingual reading. Uncontrolled and not a central-temperature calibration.

Plain English: human evidence supports failed vasoconstriction and faster core decline in some chronic high-SCI cold exposures. It does not establish acute whole-body watts or a universal “spinal-shock multiplier.” The 90-unit row is a ceiling imposed by a thermal-comfort model, not a measured physiological maximum or an acute-SCI range. The zero-resistance row is not “more vasodilation”; it is a mathematical boundary in which circulation can no longer be the bottleneck.

Show the exact translation into the modelConductance, shivering and terminology

Skin blood flow changes core-to-shell conductance through K = area × (5.28 + 1.163 × flow). Across the model body, the four finite-flow rows correspond to 9.86, 16.10, 21.20 and 184.93W/K. The displayed differences are core endpoints, not the same number of degrees of extra whole-body heat loss: in the saturated run, the declared regional proxy is 1.96°F colder at the core than the both-low-flow comparison, but 0.73°F colder as a heat-capacity-weighted mean and has lost about 75kJ more to the environment. The rest is faster redistribution between core and shell. The flow setting stays fixed through every phase, so it cannot mimic changing hemorrhage, shock, fluids, vasopressors or treatment.

The zero-resistance branch is solved separately, not by typing an enormous flow into the finite-step model. Before its clock starts, the declared 98.0°F core and cooler shells are mixed to one heat-capacity-weighted 96.66°F temperature with no change in stored energy. The solver then holds all three nodes at one temperature and advances (Ccore + Cupper + Clower) × dT/dt = net external power. This produces the mathematical K→∞ boundary while preserving each external pathway and the moisture ledger. The immediate 1.34°F core change is redistribution, not cooling. If every node instead starts at 98.0°F, adding initial stored heat, the saturated branch bottoms at 83.22°F and finishes at 84.24°F. That sensitivity is why starting shell temperature remains an exposed assumption.

“Spinal shock” is the evolving depression of spinal reflexes; “neurogenic shock” is a circulatory syndrome from autonomic disruption. Neither word is a heat-transfer coefficient. A single 57bpm watch sample cannot diagnose that circulatory syndrome: on the courtroom fall account it predates the cord injury, and independently the public record does not establish wearer, posture, paired blood pressure or exclusion of low-volume shock.

What does GCS 10 tell us?Marked impairment, not a cooling clock

A reported Glasgow Coma Scale score of 10 supports markedly depressed consciousness. It does not time the cooling or distinguish hypothermia from drugs, shock, trauma or brain injury. The public record lacks the component scores and exact assessment times.

How the model was checked—and where it stopsCode verification is not patient validation

The model divides the body into a core and two outer regions, then tracks metabolism, breathing, air and radiation, snow contact, retained moisture, evaporation and flowing water separately. Water transfer cannot exceed either the surface-contact limit or the heat the captured flow can carry away.14

Every published run uses one-second steps. Halving the step changed the endpoint by less than 0.01°C, and the energy ledger closes within its declared tolerance. Those checks test code and arithmetic, not patient assumptions. The human benchmarks stop around mild hypothermia; outputs below 35°C are extrapolations and below 32°C are stronger extrapolations.15 16 17

The code can check

units and conversions
energy in versus energy out
whether named inputs reproduce 82.1°F
how the answer changes when one input moves

The code cannot establish

the true starting temperature or exit time
how wet the clothing was
whether flowing water was used
which history occurred or who caused it

The controls name physical inputs, not causes. “Pre-exposure state” does not assert drugging; “renewed water” does not assert hose use; “reduced defence” does not diagnose shock. Every slider is an assumption because the case has not fixed it.18

No case measurement fixes how much heat capacity belongs to these simplified shell nodes. The lab therefore shows fixed 10, 20 and 30 per cent shell shares as engineering stress tests, not a physiological confidence interval. Changing that share while holding starting node temperatures fixed also changes the implied initial whole-body heat content.

The spinal-injury section above shows the current 81-minute regional-flow and shivering sensitivities. Wet presets begin after wetting. Warming their 1.232kg retained-water inventory from 0°C to 20°C would absorb about 0.103MJ against the 1.63MJ whole-body-equivalent scale check.

Audit the model yourselfChange every assumption, inspect every watt and download the complete research pack

Cooling-path comparison

The interactive model needs JavaScript. Its four reference configurations are dry snow contact, snow-wetted clothing, saturated clothing and a renewed-water branch. No configuration is a case reconstruction.

The full model is here for challenge, not persuasion by complexity. Every preset is a counterfactual. Per-run downloads contain the active configuration and output; the complete research pack adds sources, provenance, model code and tests. Equation provenance, benchmarks and limits: notes 7–8 and 14–18.

82.1°F can rule stories out. It cannot tell us which remaining story happened.

More time, a lower starting temperature, continuous cold water, weaker heat production or a later measurement could all end at 82.1°F. The number can show that some short cooling stories are too weak. It cannot choose among the stories that still fit.

Different histories

One reported endpoint

More timeColder startContinuous waterSeveral changes together

82.1°Fcannot identify its own path

A model can show that a set of assumptions reaches 82.1°F. It cannot show that those assumptions were real. Model context and limits: note 18.

Forensic science needs a way to be wrong

A mechanism becomes useful in court when it makes a prediction that can fail. “Wet clothing increases cooling” is a mechanism. A stated moisture condition, weather and interval producing a stated temperature range is a test. The National Academy of Sciences and PCAST made the same broader demand of forensic methods: expose validity, reliability and empirical failure.19

Separate example: when an expert inherits a premisePatrick's report, the bedside glass and the missing chemistry

Patrick reported powder or residue in a bedside glass. After being directed to a photograph on that basis, an expert perceived powder-like material and eventually described medication as “actually crushed”. The photograph is independent; the identification of its contents still traces back to Patrick's report. Toxicology proves drug exposure, not what was in that cup, when it was taken or by whom.20

Repeating one person's premise through a photograph, report and expert does not create three independent observations. Cup chemistry, reconciled pill counts, gastric findings and specimen-level draw times could strengthen or weaken the chain. Repetition cannot.

The records that would decide it

The missing evidence is concrete:

Temperature: Exhibit 138, the time, site and device for each reading, repeats and timestamped GCS components.
Rescue: Duxbury and Pembroke ePCRs, CAD, GPS, loading time and serial temperatures.
Wetness and scene: contemporaneous garment notes, native front-and-back photographs, drying and laboratory records, mapped blood testing, the warming and fluid sequence, native scene images and any hose inspection or water-use record.
Drugs and cup: cup custody and chemistry, pill counts, gastric findings, medication administration, serial concentrations, tube labels, draw times and accession mapping.

Until those records are produced, the thermal history remains unresolved but physically constrained. Within this model, the named snow-contact and retained-wetness runs do not reproduce an early central reading of 82.1°F on either the 63-minute lab reference or the 81-minute upper courtroom scenario. A short clock needs a much stronger sustained sink; the displayed boundary conditions need much more time.

The model cannot identify which history occurred or who caused it. The chart, ePCRs, garment tests and scene records are the evidence that can.

Sources and notes

Deliberations began on the afternoon of Thursday August 27th 2026. By Tuesday September 1st the jury had deliberated about 23 hours across four days; roughly an hour into that day's session it told Judge William Sullivan that it was deadlocked, and he asked it to “go back out … and to continue your deliberations at this time”. It sat for several more hours, stopped for the day at about 3:30pm and returned on Wednesday September 2nd. Published reports differ on whether a simple not-guilty option was given alongside first-degree murder, second-degree murder, manslaughter and not guilty by lack of criminal responsibility, so no count of the verdict options is stated here. A deadlock is a statement about the jury's progress toward a verdict on criminal responsibility. It is not a finding about the hospital temperature, the timeline, or any other item of physical evidence discussed in this article. Status as of September 2nd 2026. Boston Globe live coverage, September 1st; NBC News report on the deadlock. ↩
Day 5 testimony introduced the South Shore record and described 82.1°F with active warming; the physician could not recall whether the reading was present on arrival or developed during treatment. On Day 6, Christina Carpio agreed that the temperature later came up to 95.2°F. The acquired public record does not yet contain the relevant Exhibit 138 pages, exact time, site, device or repeat sequence. Massachusetts v. Clancy, Day 5 transcript locator; Day 6 transcript locator. ↩
The Day 14 records-review testimony describes GCS 10 on South Shore arrival and the later intubation decision. The Day 5 treating physician separately recalled eyes open, nonverbal, response to pain and spontaneous breathing before intubation. A Day 5 psychiatrist placed GCS 10 around transfer to Brigham and was not present for either assessment. The admitted South Shore chart, component scores and assessment times have not been acquired, so the relationship between those descriptions remains unresolved. Day 5 testimony; Day 14 testimony. ↩
The displayed event timeline separates clocks from interpretations. The phone extraction records the 5:10 ThreeV call, 5:13 unlock, 5:15 text, 5:34 callback, 5:38 stair event and 6:09 unanswered call; the watch records a 5:23:52 heart-rate sample; retail records independently place Patrick at CVS and ThreeV. Those records do not identify every phone operator, the watch wearer or the meaning of the stair event. Patrick supplies Lindsay's identity and apparent state on the 5:34 call, the return/search sequence and the discovery before the successful report. The 9–52-minute ground-contact range therefore exists only on that courtroom chronology: its short edge combines a roughly 6:11 discovery with the earliest working 6:20 board time, while its long edge combines the 5:38 stair event with the latest working 6:30 board time. An official release places police receipt of the successful report at about 6:11pm; responders separately placed dispatch at about the same time. Josephine's 3–4-minute estimate and the first Duxbury ambulance crew's 4–5-minute estimate cluster at 6:14–6:16. Hall said he arrived with Josephine but estimated 7–10 minutes, producing the full 6:14–6:21 witness envelope. The board, Pembroke and hospital ranges combine that envelope with a 5–8-minute mutual-aid response, about seven minutes on scene and about eighteen minutes in transport. A separate police timeline places Lindsay at South Shore by 7:35pm. These later ranges are transparent reconstructions, not CAD, GPS or ePCR times. The roughly 5:50pm neighbour account is second-hand; no call log, police interview or sworn statement currently authenticates its time, subject or continuity. Day 2 Patrick and ThreeV testimony; Plymouth County District Attorney release; Day 3 responder testimony locator; Day 12 South Shore upper-bound locator; neighbour-report locator. ↩
Massachusetts protocol expected insulation from the ground, shelter from wind, wet-clothing management and warm packaging. Protocol establishes expected doctrine, not the steps actually performed in this case. Core afterdrop can continue after exposure is reduced because cold peripheral tissues remain a sink; its size depends on state and handling. Massachusetts EMS protocols, version 2022.1; Wilderness Medical Society hypothermia guideline. ↩
Nearby KGHG observations reported about 1–2°C and light wind across the relevant period; fresh snow appears in the exhibits and nearby station records. Airport weather is a regional proxy, not a backyard air, radiant, ground or wind measurement. The simulator therefore labels those quantities as scenario inputs rather than case measurements. Iowa Environmental Mesonet KGHG station archive. ↩
Stored body heat follows mass-weighted mean-body temperature, not core temperature alone. Lenhardt and Sessler tested the Burton estimate Tmean = 0.64×Tcore + 0.36×Tskin against measured tissue heat content and reported a pooled mean estimation error of −0.09±0.42°C. They also found poorer performance during rapid cooling and at the coldest conditions. This article therefore exposes final mean-skin temperature as a sensitivity rather than inferring it from the reported 82.1°F hospital value, provisionally treated here as core. The heat-capacity term uses 2.98kJ/kg·K, a 2022 mass-weighted estimate from tissue-property databases and four virtual human models; 2.44–3.34kJ/kg·K is a source-property sensitivity, not a population or patient interval. Core-temperature interpretation separately depends on anatomical site and sensor response. None of these sources supplies Lindsay's skin temperatures, patient-specific heat capacity or thermometer metadata. Mean-body temperature validation; whole-body specific-heat analysis; core-temperature measurement review. ↩
On Day 15, Laposata described a T5–T6 thoracic cord injury and said lost temperature control and vessel dilation below it would accelerate heat loss. She gave no patient cooling rate, watts or calculation. Acute SCI guidance supports impaired vasomotor thermoregulation above T6, then states that its review found no acute-phase thermoregulation studies. ISAFSCI says impairment varies with measured neurological level and completeness; the public material reviewed here supplies no ISNCSCI level, AIS grade or autonomic examination. In Handrakis et al.'s controlled test, seven men with chronic C3–C7 tetraplegia in dry 18°C air lost 1.2±0.12°C rectally over 109±15.9 minutes on average, while seven controls showed no core decline over 120 minutes. Guttmann's small older chamber study varied markedly by lesion level. Aguilera's extreme field case recorded a 3.9°C sublingual fall over 70 minutes in one active C8-incomplete skier during a −15°C blizzard. Those are chronic comparison scales, not acute T5–T6 error bars. Downey supports possible shivering above a lesion. Ditunno defines spinal shock as evolving reflex depression; Summers found heterogeneous circulatory profiles in nine acute neurogenic-shock patients. Neither term supplies a heat-transfer coefficient. Kuennen et al. provide a separate intuition check: a purpose-built 10°C one-palm exchanger in ten healthy hyperthermic men extracted a maximum reported mean 42±12W, core fell 0.38±0.21°C over fifty minutes, and cyclic vacuum added no measurable benefit. The heat extraction declined over time. That small treated surface is not an upper bound on broad water exposure; it shows that vascular access still needs an external sink and supplies measurable, finite watts. The 5:23:52pm 57bpm watch sample cannot diagnose neurogenic shock: on the courtroom fall account it predates the cord injury, and independently lacks a verified wearer, posture, paired blood pressure and exclusion of low-volume causes. Day 15 testimony locator; acute SCI guideline; ISAFSCI standard; Handrakis et al. (2015); Guttmann et al. (1958); Aguilera et al. (1987); Downey et al. (1969); Ditunno et al. (2004); Summers et al. (2013); Kuennen et al. (2010). ↩
Exhibits 37 and 38 visibly show a turquoise wound hose reel at the deck end of the rear elevation. Witness testimony places the broad treatment position by the far-right window well. The public material reviewed does not establish hose connection, nozzle or valve state, operability, deployment, flow, use or operator. Court TV admitted-exhibit gallery. ↩
Exhibit 103 is post-treatment and post-removal. The visible exposed ground, snow, debris and staining cannot be converted into original body position or snowmelt without capture order, scale, snow state and a source-specific energy balance. The page reproduces the full image behind a content warning. Court TV admitted-exhibit gallery. ↩
The source JPEGs are 1600×1236 rehosted exhibit copies. They were converted to WebP at the same aspect ratio with no crop, resize, overlay or tonal edit. Source SHA-256: Exhibit 37 e14faa4373ac…; Exhibit 38 d02f75c9727…; Exhibit 103 25ee2a9649ee…. The full hashes and local acquisition trail are retained in the publication record. Court TV admitted-exhibit gallery; publication provenance record. ↩
Laposata referred to wet clothing while explaining heat loss, but the underlying observation was not exposed in the testimony reviewed. A later hospital image shows clumped glossy hair after transport and treatment; it cannot establish pre-rescue clothing moisture. Located frontline testimony does not affirm wetness. That is an unresolved premise, not a dry finding. Massachusetts v. Clancy, Day 15 transcript locator. ↩
Day 4 testimony first records no visible blood on the displayed item and top and no recollection on the jeans; the through-clothing seepage mechanism appears in counsel's subsequent leading questions. Day 12 describes unspecified staining and a later experience-based agreement that it was blood. Full public frames of Exhibits 116–119 show diffuse mottling but no visually obvious concentrated outer-garment deposit. Under the stated substantial-undiluted-deposit hypothetical, failure to see a concentrated burden raises a conditional testing question; the photographs cannot establish an absence because the materials are dark and the wound, cutting, handling, packaging and drying histories matter. Day 6 supplies the laboratory standard: visual red-brown staining cannot be called blood without confirmation. The public testimony reviewed here does not identify a garment-level blood or DNA result. A controlled study of 880 textile stains found derivation-related morphology across mixed blood/water, water-then-blood and blood-then-water sequences; it did not validate general mottling as a rinse marker or simulate a garden hose on layered clothing. Laundering and hand-washing studies show that naked-eye blood can disappear while latent reactions and DNA remain, and that washing can blur morphology or cross-transfer blood. Wearer DNA is expected on a person's own clothing; even a profile from a mapped blood-positive area would not by itself prove its body-fluid source, rinsing, deposition mechanism or actor. The UK Forensic Science Regulator discusses enhancement where cleaning-related dilution is suspected and cautions about dark and textured fabrics and false reactions. The mapped sampling and matched reconstruction proposed in the article are an experimental design, not a quoted regulatory requirement. These sources make the rinse hypothesis testable; they do not establish rinsing here. Court TV admitted-exhibit gallery; Day 4 testimony; Day 6 testimony; Day 12 testimony; van den Berge et al. (2019); Edler et al. (2017); Hofmann et al. (2019); Nakanishi et al. (2020); wearer-DNA study; UK Forensic Science Regulator guidance. ↩
The fixed-node model adapts the energy pathways and conductance terms of Gagge, Stolwijk and Nishi and the later Pierce/ASHRAE implementation. From ASHRAE’s Appendix D4 code it borrows the mixed-convection formula, 0.155m²K/W-per-clo conversion, clothing-area factor, sea-level Lewis relation, 0.0014 sensible-respiration factor, the 0.0023/Torr latent term converted and rounded to 0.0173/kPa, the 0.06 diffusion-wettedness term and the 0.5/6.3/90L/h/m² minimum, neutral and maximum-flow settings. The regional 0.5/6.3 combination is this article’s scenario choice. The 90 setting is the Pierce/ASHRAE algorithm’s ceiling, not a physiological maximum. The code implicitly fixes atmospheric pressure at one atmosphere instead of carrying D4’s pressure correction. Pierce changes an effective skin-mass parameter with blood flow. This reduction instead holds total and regional node capacities fixed within a run so a physiology switch cannot create enthalpy by remapping node mass; its displayed 10/20/30 per cent shell shares are uncalibrated stress tests. The separate K→∞ branch is an energy-preserving mathematical boundary: it mixes the declared nodes to one heat-capacity-weighted starting temperature, then advances their combined capacity using net external power. It is not a blood-flow setting. Du Bois supplies only the body-surface-area formula. The vapour-pressure function uses the updated coefficients in the Buck Research CR-1A manual, not the original 1981 numerical coefficient set. Renewed-water transfer uses the standard constant-surface-temperature effectiveness relation Q=ĊΔT(1−e^−UA/Ċ), with water-flow heat-capacity rate Ċ=ṁc_p. None of these sources supplies case geometry, water conditions or patient calibration. Gagge, Stolwijk and Nishi; ASHRAE Handbook two-node model; ANSI/ASHRAE 55-2020 Addendum g; Du Bois and Du Bois; Buck Research CR-1A manual, Appendix 1; NASA effectiveness–NTU equations 10–12. ↩
Helland and colleagues exposed healthy volunteers in hand-wrung wet cotton to a median 5.9°C, 88.4 per cent humidity and 1.8m/s air while pharmacologically reducing, but not eliminating, shivering. The drug branch cooled at 0.82°C/h; ten of fifteen reached 35°C. It is the closest wet-air benchmark, not validation of trauma, snow contact or 27.8°C. Helland et al. (2025). ↩
Henriksson and colleagues placed eight wet-clothed volunteers at −18.5°C on a spineboard plus sleeping mattress, then compared packaging. Their underwear ensemble had an approximate 575g dry weight and retained a mean 1.232kg of water (SD 0.091kg). The model borrows only that mean as a starting finite-moisture scale, then makes two additional choices: it treats the full retained mass as potentially evaporable and splits it 45/55 between regions. The inventory remains partly present at the published endpoints, so the exact cap does not determine those temperatures. Core stayed near baseline during twenty minutes of exposure while mean skin fell from 33.5°C to 24.9°C; the next thirty minutes produced about 0.3–0.4°C core decline across branches. Those trajectories are comparison-only. Garments, shivering and rescue geometry differ from this case. Henriksson et al. (2015). ↩
Kulkarni and colleagues immersed six healthy men to the sternal notch while a stirred bath fell from 21°C to 8°C. Their highest reported 30-minute group means were 369±195W net stored-heat loss and 558±35W cutaneous loss (mean±SD); the final-ten-minute esophageal-temperature slope averaged −2.4±1.7°C/h. Net heat gain was defined as metabolic heat production minus estimated respiratory heat loss and total cutaneous heat flux. The subjects averaged 84±18kg, shivered through most of the immersion, received meperidine during the final ten minutes and remained only mildly hypothermic. A highest 30-minute group mean is not an 81-minute sustained average or a physiological ceiling, and participant SD is not a case error bar. Sternal-depth immersion is an adjacent scale anchor and the wrong geometry for partial runoff. Grissom's fully buried, insulated snow volunteers cooled at 0.7°C/h under normocapnia and 1.2°C/h under hypercapnia; that protocol is not partial ground contact. Kulkarni et al. (2019); Grissom et al. (2004). ↩
Model release thermal-fixed-0.1.0 uses SI units and frozen scenario manifests. Tests cover unit conversions, zero-flux and constant-flux limits, deterministic replay, one-versus-split phases, timestep refinement, water-limit switching, finite moisture, monotonic sensitivities, published readout regression, cumulative energy closure and the zero-resistance branch’s initial-energy preservation, node constraint and timestep convergence. The declared 70W metabolism is fixed rather than reduced by a deep-hypothermia Q10 relation. Retained water supplies a finite latent inventory but neither its initial sensible-cooling pulse nor its heat capacity is placed in a dynamic clothing node; the published runs retain water at the endpoint, so their air-side latent path does not terminate from inventory exhaustion. Adjacent benchmark observations are shown, but the model has not been validated against those protocols. Outputs below 35°C are extrapolations; below 32°C, stronger extrapolations. The source code is reproducibility, not independent warrant. external equation source; model source; frozen manifests and source registry; tests embedded in the research pack. ↩
The 2009 National Academy of Sciences report described major gaps in the scientific basis, validation and reliability of several forensic disciplines and called for research and standardisation. PCAST's 2016 report framed foundational validity for subjective feature-comparison methods around empirical studies of repeatability, reproducibility, accuracy and error rates. These institutional findings concern forensic practice generally, not the truth of any proposition in this case; the article applies their testability logic by analogy to a quantitative heat-transfer claim. National Academies, Strengthening Forensic Science in the United States; PCAST, Forensic Science in Criminal Courts. ↩
Patrick reported powder or residue in a bedside glass. On Day 20, Saathoff described returning to a photograph after that report focused his attention on it, perceiving powder-like material and later referring to medication that was actually crushed. The photograph is an independent object; the identification and event meaning are interpretations with a shared upstream prompt. The public record reviewed here contains no cup-residue chemistry, pill-count reconciliation or gastric bridge establishing composition, dose, route, timing or actor. Separately, Day 7 toxicology testimony identified prescribed drugs and described most reported values as therapeutic with quetiapine elevated but nonlethal; the earlier arraignment proffer attributed a roughly two-hour ingestion clock to an unidentified doctor. The source analysis and tube-to-laboratory mapping needed to reproduce that clock have not been acquired. Day 20 courtroom video; Day 7 toxicology transcript; 2023 arraignment video. ↩