Stanbury has argued that in those fires sometimes referred to as firestorms in Hamburg, Darmstadt, Kassel and Wuppertal during World War II, the number of buildings initially set alight in the fire area was nearly half the total, whereas for other fires in Germany it was 1 in 8 or less. It is shown here, on the basis of values derived for the likelihood of spread and burn out, that if half the buildings are initially set alight, flames from separate fires might merge over large areas whereas they could never merge generally at any time during a fire starting from a much smaller fraction of ignited buildings. A simplified theoretical model of fire spreading in an urban area from a statistically uniform distribution of ignition centres is described. It is mathematically the same as the well-known epidemic model of Kermack and McKendrik and is basically similar to that given by Albini and Rand, producing similar estimates of the probable damage, but is, however, somewhat more flexible analytically. The paper discusses the relative importance of the initial conditions and the spread parameters on the development of fire and the resulting damage. An approximate modification of the model to describe a steadily advancing fire front is outlined, and used to estimate parameters of fire spread from published correlations of data.

[This publication is referenced in the "Synthesis of knowledge of extreme fire behavior: volume I for fire managers" (Werth et al 2011).]