Audit trail

Where each number comes from, and where the manual stops

This page is the audit trail for the nuclear bomb blast radius table. It gives the source, the exact figure and printed page, the way each value was obtained, and the places where the underlying chart runs out of assumptions. If a number here cannot be traced to a page of the manual, it does not belong on the site.

Two pages consume these values. The nuke simulator shows them around one city at a time; the nuclear bomb map draws the same rings over real ground. Neither page carries a number that is not accounted for below.

The source

Title: The Effects of Nuclear Weapons
Editors: Samuel Glasstone, Philip J. Dolan
Edition: 3rd edition, 1977
Publisher: U.S. Department of Defense / Energy Research and Development Administration
DOI: 10.2172/6852629
Status: U.S. government publication, public domain. No permission is required to reproduce it, and nothing on this site is a derivative model built on top of it beyond the scaling law the manual itself prints.
Page numbers: Every citation uses the printed page number, not the position inside a PDF. The two differ: in the full scan (6852629.pdf) chapter 3 starts at printed page + 9; the chapter files extracted from the same scan differ again. A reader holding the book needs the printed number, so that is what is cited.

How each figure was obtained

Scroll the table sideways →

Source of each row in the blast radius table.
Effect	Source	Printed page	Method
Fireball, air burst	Eq. (2.127.1)	70	Printed equation, R = 110·W^0.4 ft. No chart reading.
Fireball, contact burst	Eq. (2.127.2)	71	Printed equation, R = 145·W^0.4 ft. No chart reading.
5 psi	Fig. 3.73c	115	Read at the peak of each contour, interpolated between 4 and 6 psi, then scaled by W^1/3.
1 psi	Fig. 3.73c	115	Read at the peak of the outermost contour, then scaled by W^1/3.
Thermal, 2nd degree	Fig. 7.42	291	Read directly at each yield column, at 4.5 cal/cm². Not scaled.
Thermal, 3rd degree	Fig. 7.42	291	Read directly at each yield column, at 10 cal/cm². Not scaled.

Height of burst

Height of burst is read off the chart, not chosen in advance. Each overpressure contour has a point where it reaches furthest from ground zero; that point fixes a distance and a burst height together, and both are used. The manual works its own example the same way:

“From Fig. 3.73c, the maximum ground distance to which 4 psi extends for a 1 KT weapon is 2,600 feet. This occurs for a burst height of approximately 1,100 feet.” — p. 114

Reading that same point off the digitised chart gives 2,618 ft, a difference of 0.7%, and the whole family of contours is calibrated from that agreement. Burst heights scale with the same cube-root law as distances, because a contour is only "the same contour" if the geometry is scaled with it.

Contact bursts use height of burst = 0 and are read where each contour meets the bottom edge of the figure, the convention the manual states on p. 106.

Where the chart stops: the high burst heights. Peak burst heights scale to 10,174 ft at 5 psi and 15,750 ft at 1 psi for 1 Mt, and to 7,311 ft for 1 psi at 100 kt. Fig. 3.73c is computed for a homogeneous sea-level atmosphere at burst heights of 5,000 ft or less (§3.73, p. 110). Those cells therefore extrapolate the chart's assumption by up to a factor of three. They are kept because the scaling law is the manual's own and the alternative — silently dropping the largest yield — would hide the limitation instead of stating it. No altitude correction from §3.64–3.66 is applied, and that omission pushes the figures low rather than high. Treat them as the edge of what this chart supports, not as precise distances.

What “psi” means on this site

Fig. 3.73 plots peak overpressure on the ground. That is not the same quantity as the incident (side-on) pressure in free air: the ground reflects the shock front, so the peak value is higher, and for the same psi the distance is larger. A tool that quotes incident overpressure will report a noticeably smaller radius for the same number, and both can be right about different quantities. This site reports the manual's quantity and says so in every cell.

Nothing here converts overpressure into a damage or casualty statement. The manual contains such tables; this site does not reproduce them yet, and does not improvise its own.

Thermal assumptions

Every thermal figure is a slant range, read off a chart computed for air bursts at burst heights of 200·W^0.4 ft with 12-mile visibility (§7.42, p. 290). §12.66 (p. 563) notes that at ranges much greater than the burst height the slant range may be taken as the distance to ground zero, which is how it is reported.

The two columns are thresholds, not a range of outcomes:

4.5 cal/cm² — second degree. The lower edge of the band the manual associates with second-degree burns in unshaded light skin. §12.65 (p. 563) states that at 4.5–6 cal/cm² about 18% of an exposed population receives second-degree burns and the remainder first-degree.
10 cal/cm² — third degree. Read from the same chart and reported as the third-degree column; the burn-severity chart behind the threshold is Fig. 12.64 (p. 564).

Assumptions that move these numbers and are not adjusted for: light skin tone, no clothing, no shelter, no haze or cloud, no evasive action, and a person facing the burst. Darker skin needs more irradiance for the same burn; clothing and shade can remove the exposure entirely. These are the manual's conditions, restated, not this site's preferences.

No thermal rows are given for contact bursts. Fig. 7.42 is an air-burst chart, and the manual handles surface-burst thermal radiation through a different procedure (§7.41, p. 290). Rather than splice two procedures into one column, the contact-burst table reports blast and fireball only.

How the thermal contours were identified

Read this before checking the thermal figures. Fig. 7.42 is a scan with no text layer: the numeric labels printed beside its contour lines cannot be extracted, and they are not reliably legible at the resolution of the available scan. The six thermal values on this site therefore do not come from reading those labels. They come from reading the geometry of the curves and then fixing which curve is which from two independent statements in the manual.

Those two statements are:

§12.66 (p. 563): at 1 Mt, 4.5–6 cal/cm² reaches a slant range of 9–10 miles.
§7.43 (p. 290): at 1 Mt, 8 cal/cm² reaches about 7 miles.

Together they fix the order and the spacing of the contours from the outside in: 4.5, 6, 8, 10, 15, 25 cal/cm². Applying that ordering to the digitised curves reproduces both statements within a few percent, and the third anchor is reproduced as well — the table below shows all three. That agreement is the evidence; the labels are not. Anyone re-checking these values should expect to re-derive the contour order the same way, and should treat a mismatch between our curve order and the printed labels as the first thing to look at.

Cube-root scaling, and what it is not

The chart is drawn for 1 kt. Distances for the other yields come from Eq. (3.61.2), p. 101 — the manual's statement that, for a given overpressure, distance scales as W^1/3 — which is restated in §3.62 (p. 102) and applied to these same figures in §3.73 (p. 110). Height of burst scales the same way.

This is the one place where a number on the site is not a direct transcription. It is a multiplication the manual instructs, not a model this site invented, and it happens once at build time: the JSON ships finished values, and no arithmetic runs in the browser. Each scaled entry also ships the 1 kt reading and the pixel coordinates it was read at, so the multiplication can be re-done by hand. The page label for every such value is scaled from the 1 kt curve (§3.61, p.101).

Cross-checks

Four statements in the manual's prose, used as independent anchors. They are the reason the digitised curves are trusted; they are not a substitute for the citations above.

Scroll the table sideways →

Digitised reading against the sentence in the manual.
Check	Manual	Read here	Difference
1 Mt, 4.5 cal/cm² slant range 9–10 mi band, outer edge (§12.66, p.563)	G1	10.36 mi	+ 3.6%
1 Mt, 6 cal/cm² slant range 9–10 mi band, inner edge (§12.66, p.563)	G2	8.65 mi	-3.9%
1 Mt, 8 cal/cm² slant range about 7 mi (§7.43, p.290)	G3	7.35 mi	+ 5.0%
1 kt, 4 psi maximum ground distance 2,600 ft at about 1,100 ft HOB (p.114 worked example)	G4	2,618 ft	+ 0.7%

Verification status

15 of 24 rows carry an individual sign-off; the rest are marked with a star wherever they appear. The distinction is not decorative:

The fifteen air-burst values carry a sign-off dated 11 November 2026. The twelve headline figures — fireball, 5 psi, 1 psi and second-degree thermal, across three yields — were independently recomputed by the site owner, and five of them were spot-checked against the original figures. The third-degree column was read off the same chart and accepted with them.
The nine contact-burst values reuse the same 1 kt base readings taken from a different edge of the same two charts. They have been through the automated checks but not through an individual human reading, and they are marked with a star rather than quietly promoted.

The automated check (scripts/verify_data.py) enforces structure, provenance, monotonicity, the scaling arithmetic, unit conversions, the interpolation label on every 5 psi row, and agreement between the JSON and the CSV ledger. It currently runs 354 assertions, and a single failure exits non-zero, which fails the build.

Population figures

No population figure is published on the site yet. When city pages are added, the convention is fixed in advance: population is counted within 10 km of the city centre on the WorldPop 1 km grid, 2020 estimates, and the radius, year and dataset version are printed beside the number.

Administrative city boundaries are deliberately not used: every country defines a city differently, so boundary figures from different countries cannot be compared with each other. Population inside an effect ring is a separate figure computed from the same grid and is never merged into the city figure. These are resident-population counts, not footfall, and not casualties.

What this site does not compute

Casualty or fatality estimates.
Damage to specific buildings or structures.
Fallout footprints or dose rates.
Real-time or daytime population.
Underwater, underground or high-altitude bursts.
Multiple warheads or salvo effects.
Terrain shielding, urban canyon effects, or blast propagation through buildings.
Thermal effects on materials, or ignition and fire spread.
Electromagnetic pulse.
Any prediction for a specific event, place or date.

Files and reproduction

src/data/effects.json: Every published value, with citation, burst height, conversion arithmetic, the Chinese reading note taken at the chart, and — for scaled entries — the 1 kt base reading and pixel coordinates.
data-ledger.csv: One row per value: effect, yield, value in km, unit conversion, chapter, figure, printed page, reading method, raw reading. This is the file to check line by line against the book.
scripts/build_effects.py: Regenerates the JSON and the ledger from the recorded readings. No network access.
scripts/verify_data.py: Runs the 354 assertions. Exits non-zero on any failure.

Reproduction recipe for any single value: open the cited figure at the cited printed page, read the distance at the stated burst height, multiply by W^1/3 if the cell says scaled, and convert feet to metres with 0.3048. If the result differs from the table, the reading note in the JSON records the pixel coordinates used, so the disagreement can be located rather than argued about.