Wood burning is chaotic, so the particles emitted can be difficult to study in a repeatable way. Here, we addressed this problem by carefully controlling small wood fires in the lab. We saw three burning phases, which could be told apart chemically; we also saw evidence of these in measurements of wood burning in London in 2012. Controlled experiments like this help us to understand why emissions are so variable and to recognise burning conditions just from the particles seen in the atmosphere.