The hypothesis of "pyrophoric carbon," that cooking causes chemical enhancement of self-heating properties, is inconsistent with measurements of the self-heating parameters of cooked and uncooked wood chips. These tests employed a recently-introduced experimental method that offers advantages over traditional iterative techniques. Predicted critical sizes and temperatures are higher for the cooked material than the uncooked. The additional effects of cooking-induced reductions in the thermal inertia, due mainly to lowered density, are quantified experimentally. The cause of delayed ignition of whole wood is shown to be diffusion-controlled self-heating facilitated by cooking. The delay is caused by long-term effects of cooking on the physical structure of wood cells, which consist of long, slender tubes. This microstructure, when undamaged, accounts for high axial permeability but low transverse permeability, with ratios on the order of 104: 1 for the softwoods typically used as structural lumber. Cooking eventually allows oxygen access deep into whole wood via greatly enhanced transverse permeability caused by disruption to the microstructure in the form of transverse cracks. Lack of oxygen diffusion prior to cooking prevents self-heating in whole wood except near end-grain surfaces. These processes are demonstrated experimentally.