# Dyatlov Pass: what killed the nine, 1–2 February 1959

Published report: https://historical-mysteries.netlify.app/dyatlov-pass/report
Local copy: `dyatlov-pass-report.html` (figures in `figs/`; HTML fragments, stylesheet and build script in `report-src/`).

## The mystery, the sides, and how it was judged

What happened. In late January 1959 nine ski hikers, eight students or recent graduates of the Ural Polytechnic Institute in Sverdlovsk (now Yekaterinburg) led by 23-year-old Igor Dyatlov, and a 37-year-old ski instructor, Semyon Zolotaryov, set out across the northern Urals. On 1 February they pitched their tent on the open east slope of Kholat Syakhl, a treeless mountain, 1.5 km above the forest. That night all nine cut their way out through the tent wall and walked downhill without boots or coats in the dark, with the air falling toward minus 25 °C. Searchers found the tent on 26 February with everything inside; five bodies, dead of cold, between the tent and a cedar at the forest edge over the following week; and the last four in May, in a ravine stream bed under metres of snow, three of them with crushed chests or a crushed skull and no wound on the skin. The investigator who ran the case, Lev Ivanov of the regional prosecutor's office, closed it on 28 May 1959 with the finding that they had died of «стихийная сила», an elemental force; the resolution does not say what the force was.

Why it is a mystery, and the sides. Nothing in the file explains why nine experienced people left their only shelter at once or how three of them were crushed 1.5 km from the tent. The explanations fall into five camps: snow (a small avalanche or a collapse of snow onto the tent, the view of the Russian prosecutors' 2019–20 review and of the snow scientist Johan Gaume and the geotechnical engineer Alexander Puzrin, whose 2021 paper gave it a physical model; opposed by people who have examined the slope and its snow in winter, the aviation engineer Vladimir Borzenkov and the 1959 searchers Vladislav Karelin and Sergey Sogrin); wind (a violent fall wind, the Swedish archaeologist Richard Holmgren, 2019); panic from low-frequency sound (the American writer Donnie Eichar, 2013); people (Mansi hunters, criminals, or a KGB operation, the last from the pseudonymous author Alexei Rakitin, 2013); and weapons (a rocket, a secret weapon or a toxic release, fed by fireball reports and radioactive traces on the clothing). The engineer Evgeny Buyanov (2011) holds the avalanche view but puts the crushing injuries at the tent.

How it was judged. The 1959 case file (392 sheets) was read in Russian and 459 facts extracted with their sheet numbers; the slope, darkness and cold were computed from public data; the Gaume–Puzrin model was reimplemented from its published equations; the 2018 examination of Zolotaryov's exhumed bones was added; and every explanation was scored against 28 documented facts.

## Verdict

Cold, for six of the nine; crushing under a mass of snow, most probably in the ravine, for three. No other people, no weapon, no explosion, no vehicle and no poison is recorded on or around the bodies. The group left the tent because its uphill half came down on them under snow in the dark, most probably a small slab avalanche released from the stepped slope above the platform they had cut, less probably drifted snow collapsing the roof; they went down to the forest expecting more to follow and could not find their way back.

Confidence. Every figure is a judgement that depends on the nine explanations scored in the matrix and on the weight given to each of the 28 facts. None is a frequency, and a different list of rivals or different weights would give different numbers. The sensitivity ranges are this report's own re-weightings of the facts, not data:
- Natural causes, no human or technological agent: about 90% (sensitivity 80–95, on the weight given to the missing chemical sheets). The residual is for what the file cannot check: the chemical and histological analyses referred to in the 1959 autopsy acts are missing from the file, and only one body (Zolotaryov, 2018) has been examined since 1959.
- Snow on the tent as the trigger: about 65% (sensitivity 50–80), of which a small slab avalanche of the kind Gaume and Puzrin modelled about 45% (sensitivity 20–65, on the unmeasured snow above the 1959 position and the fitted delay) and drifted snow collapsing the roof about 20%. Wind alone about 15%; something not in the record about 20%.
- The three crushing injuries sustained in the ravine rather than at the tent: about 75% (sensitivity 60–85, on the examiner's survival-time statements against the claim by Eduard Tumanov, a Moscow forensic expert, that the rib fractures run on different lines). This rests on the mobility and survival statements of the 1959 forensic examiner, Boris Vozrozhdenny (sheets 381–383), and on the 2019 reading of Zolotaryov's bones by Sergey Nikitin, the forensic expert who examined them after their exhumation in 2018. The positive case for a slab at the tent does not by itself place the injuries at the tent, and the objections to the slab do not by themselves place them in the ravine; the 75% depends on those statements; the slab argument does not bear on it either way.
- Evidence labels: the evidence for natural causes is strong; the trigger mechanism has a leading candidate, the slab, among several that remain possible; the ravine as the place of the injuries is probable. Ivanov's resolution of 28 May 1959 («стихийная сила») stands; what is added, as the most probable answers rather than as established facts, is the force (snow) and the place of the crushing injuries (the ravine).

The Gaume–Puzrin 2021 slab model: internally sound and reproduced here to the published precision; physically possible on this slope. Its terrain requirement, a step steeper than 28° above the tent, is met on the authors' 9 cm drone survey of September 2021 (Commun. Earth Environ. 3:63, 2022), which shows 4–6 m steps over 28–30° "continuous" above the candidate tent positions. Its snow requirement, a buried weak layer on that step in February 1959, is unmeasured (pits in 2010, 2014, 2015 and 2019 found none); its release delay is fitted to the forensic clock rather than predicted; its injury calculation cannot tell the tent from the ravine.

The skull. Thibeaux-Brignolle's depressed fracture (9 × 7 cm, a 3 × 3.5 cm bone piece driven onto the dura, sheet 353) needs 2.5–10 kN (kilonewtons; one kilonewton is roughly the weight of 100 kg) in cadaver tests (Allsop et al. 1991; Yoganandan & Pintar 2004). Snow can transmit at most its crushing pressure times the contact area: the 2.5 kN floor needs about 400 kPa over the whole depression, 500 kPa over the 50 cm² test plate, 2.4 MPa over the bone piece alone. At the 100 kPa yield pressure Gaume and Puzrin assign their wind slab, a face on the whole depression gives 0.6 kN; Malcolm Mellor's 1975 review of snow mechanics puts laboratory snow at the site densities at roughly 300–950 kPa under rapid loading. So a soft or medium slab is excluded as the sole loader; a hard, well-sintered slab or crust bearing on most of the depression with the head backed, or a hard object between snow and bone (a stone or ice edge in the stream, or the camera that Buyanov and the paper's supplement put under his head), is not. This constrains the loader; it does not exclude snow, it does not discriminate the tent from the ravine, and it does not change any of the percentages. Arithmetic: `code/physics/skull_constraint.py`.

## How this stands to earlier views

The 1959 investigation closed with "an elemental force the people were unable to overcome". The 2019–20 review by the Sverdlovsk regional prosecutor's office named a small slab plus hypothermia, with the ravine injuries from a collapse of the undercut snow bank; the Prosecutor General disowned the press conference (10 Aug 2020), and the conclusions were never published or tested in court. Gaume and Puzrin 2021/2022 supplied a physical mechanism (delayed wind-slab release on a buried weak layer at a 28° step) and field evidence that such steps and such slabs exist nearby. Buyanov 2011 puts the crushing injuries at the tent.

This finding agrees with the 1959 verdict of a natural cause and with the prosecutors' 2020 placement of the injuries in the ravine (site badge: confirms). It agrees with Gaume and Puzrin that the slab mechanism is possible on this slope and treats it as the leading single trigger candidate. It reads the model's two quantitative claims differently: the delay is a fit rather than a prediction, and the injury physics is indifferent to location. It rejects Buyanov's injuries-at-the-tent version on the examiner's mobility statements, and it closes the fireball, Mansi, KGB, infrasound, katabatic-wind-as-hurricane and minor theories on cited sheets. Nothing here overturns what historians already think.

## What this adds to earlier work

None of it is new primary evidence. Every item is a new analysis of evidence already in print (the 1959 file, public elevation, astronomical and launch data, the published model and its peer-review file, the 2018 exhumation reports). A historically plausible conclusion does not validate the computation that reached it; the code and inputs are saved so each computation can be checked on its own.

- A from-scratch reimplementation of the Gaume–Puzrin release model with sensitivity and regime maps: delayed release exists only between 24.7° and 29.7°; the forensic 9.5–13.5 h window is met only at 26–28.5°; cohesion of 300/440/450 Pa gives failure at the cut / the published delay / no release; the delay is proportional to an unmeasured wind-deposition rate (Dieter Issler, the avalanche scientist who reviewed the paper for the journal, called the agreement "coincidental"; the regime map bears him out). `code/physics/`.
- Injury physics that makes the examiner's statements quantitative: any 100–300 kg of hard snow at 1–2 m/s, or 3–4 m of static snow on a person against a firm base, gives the rib pattern without breaking skin, so chest injuries cannot distinguish tent from ravine; for Thibeaux-Brignolle's 9 × 7 cm depressed temporal fracture, the 2.5–10 kN of the cadaver tests over the depression needs a crushing pressure of 400 kPa or more, which excludes a soft or medium slab as the sole loader but not a hard slab, nor a hard object under the snow.
- A 28-fact × 9-explanation matrix with the contradicting sheet named in every red cell, keeping the three slab versions (Gaume–Puzrin, prosecutors 2020, Buyanov 2011) separate because they place the injuries differently.
- The darkness and the clock from open data: moon below the horizon 11:27 on 1 Feb to 04:15 on 2 Feb; full night from 19:40; tent site in mountain shadow from 14:40; walking time to the cedar 35–85 min; Burmantovo readings falling from −8 to −21 °C between 15:00 and 03:00. `code/geo/`.
- The fireballs closed against the General Catalog of Artificial Space Objects (GCAT), the astronomer Jonathan McDowell's launch list: 17 Feb 06:46 local = R-7 launch 01:46 UTC; 31 Mar 03:53 local = R-7 failure 30 Mar 22:53 UTC; nothing capable of reaching the northern Urals on 1–2 Feb.
- A catalogue of 41 places where the standard English transcription of the autopsy acts and interrogations differs from the Russian on load-bearing points (e.g. "bomb" for «воздушной взрывной волне», an air blast wave; "first hour" for «первые часы»; knuckles rendered as wrist), plus 16 internal contradictions and 16 gaps, and 17 further translation differences in the scene and search documents.
- The legal status of the 2020 conclusions and the state of the case to September 2026 (relatives' demand for a new criminal case, 2 June 2026, unanswered).
- A bound on the *average* slope from a public 30 m digital elevation model (DEM) (15–16° at the cairn-marked tent position, no candidate above 19.5°), which does not support the 2021 paper's 23° ± 2° average but says nothing about the 4–6 m steps.

## Where this falls short

- No field measurement of the snow at the site; every snow observation is other people's (1959, 2010–2019, 2021–2023) and none is a pit line up the step above the fixed 1959 tent position.
- The ravine collapse, the least secure step of the reconstruction, is treated by static and quasi-static calculation only; nobody, including this report, has run a dynamic model of it.
- Only volume 1 of the case file and the published radiograms were read; volume 2, the unpublished 2020 expert reports and the original 1959 negatives were not. The model paper's Zenodo code and the cadaver-test primaries were not reachable; the reproduction used the published equations and the peer-review file.
- Weather for the night is reconstructed from three daily stations 100–200 km away, Maslennikov's notes of the Burmantovo readings and the 2020 microclimate expertise as reported; no hourly 1959 record for the site was obtained.
- The hour of leaving the tent is inferred within a 2.5-hour window from a digestion estimate written identically for five bodies.

## What would overturn or move the verdict

1. A snow profile up the step above the 1959 tent position: a buried weak layer under a wind slab → slab to ~65%, snow trigger to ~80%; repeated hard, well-bonded snow to the ground → slab toward 20%.
2. The 1959 tent position (prosecutors' 2019 photogrammetry) overlaid on the 2021 drone model, naming the step above the cut and its angle and length.
3. The missing 1959 chemical and histological sheets (a toxin would cut the natural-cause figure sharply).
4. A full exhumation with modern imaging (fracture geometry: broad compression vs localised blows).
5. A range or telemetry record of any launch over 61.8° N 59.4° E on 1–2 Feb 1959 (none in published lists).
6. The hourly 1959 synoptic record for Burmantovo, Ivdel and Nyaksimvol (a recorded fall-wind event would raise the wind-only trigger).
7. The written conclusions and nine expert reports of the 2019–20 check (Sverdlovsk state archive, restricted).
8. The original 26–28 Feb 1959 negatives of the tent and the slope above it (a crown line or debris lobe).
9. A dynamic model of the ravine bank collapse.

## Files

- `dyatlov-pass-report.html`, `figs/`, `report-src/` — the report, its four figures, and its source (fragments, `styles.css`, `build.py`, `gen_rivals.py`).
- Figures (`figs/`), all made for this report: Figure 1, the east slope of Kholat Syakhl (`map_dem.png`), and Figure 2, the tent-to-cedar profile (`profile_tent_cedar.png`), both from the Copernicus GLO-30 elevation model by `code/geo/dem_slope.py`; Figure 3, time to release against weak-layer inclination (`fig_B1_delay_vs_slope.png`), and Figure 4, the regime map (`fig_B1_regime_map.png`), both from the reimplemented Gaume–Puzrin equations by `code/physics/slab_release.py`.
- `research/case-file-scene-facts.md` (246 facts, sheet-cited) and `case-file-scene-conflicts.md`; `research/case-file-forensic-facts.md` (213 facts) and `case-file-forensic-conflicts.md` (A1–A41 translation errors, B1–B16 contradictions, C1–C16 gaps).
- `research/geo-timeline-weather.md` (with a note on the DEM's scale), `research/physics-assessment.md`, `research/rival-theories.md` (full matrix with reasons), `research/reinvestigation-2019-2020.md`, `research/zolotaryov-exhumation-2018.md`.
- `code/geo/` (`run_all.sh` reproduces the DEM, astronomy, weather and walking outputs in `code/geo/out/`) and `code/physics/` (`gp_model.py`, `slab_release.py`, `slab_impact.py`, `hypothermia.py`, results and figures).
- `code/physics/skull_constraint.py` and `skull_constraint_results.txt` — the force-cap arithmetic for the temporal fracture.
- `data/` — saved sources: Maslennikov's notebooks and the Burmantovo readings CSV, Borzenkov 2022, Karelin, Louhi et al., the Puzrin–Gaume 2022 Comment (PDF and text) and the dyatlovpass.com interview, `exhumation-2018/` (KP and other pages on the 2018 exhumation), Mellor 1975 (`mellor-1975-snow-mechanics.pdf`, text, and Fig. 17 as PNG).

## Conventions

Sheets are cited as «л.д. N» of volume 1 of the 1959 case, read at https://dyatlovpass.com/case-files-N?rbid=17743 (EN) and https://dyatlovpass.com/case-files-N-ru?rbid=17747 (RU). Times are local Sverdlovsk time (UTC+5).