summary - The founding a-priori catastrophe-bound calculation: shows that however hot a section of atmosphere is driven by a fission/fusion bomb, radiative losses (chiefly Compton scattering of energetic electrons) always outpace the energy gain from nitrogen-nitrogen fusion reactions, so no self-propagating atmospheric chain reaction can occur — even under deliberately stacked “most dangerous” assumptions (e.g. treating every collision as reactive at the full geometric cross-section). The computed safety margin (ratio of energy losses to gains) shrinks to only about 1.6 near a 10 MeV temperature before recovering, and the report explicitly adopts a policy of “exaggerating the dangers at any point which appears at all questionable” rather than using best-guess parameters throughout.
relevance_note - The named methodological analogue for a-priori collider catastrophe bounds: an internal, non-adversarial, theory-only calculation (no experiment could test it beforehand) that turned out correct, but whose reassurance rested on deliberately conservative assumptions and a numerically thin margin at one point in parameter space — exactly the kind of thing Ord et al. (2010) and Kent (2004) argue you must price the failure-probability of.