The causes and types of continuous spectra emitted by flames are discussed and their importance stressed. It is shown that the yellow-green continuous spectrum emitted by some flames containing oxides of nitrogen is probably identical with the spectrum of the air after-glow and is therefore due to a reaction between nitric oxide and atomic oxygen. It thus becomes possible to test for the presence of atomic oxygen in a flame by admitting nitric oxide and observing if a yellow-green emission results. For the carbon monoxide flame there appears to be a high concentration of atomic oxygen, both for the dry and moist flame. The combustion mechanism is discussed in detail using this knowledge. For the hydrogen flame a little atomic oxygen is present, but results do not permit of definite conclusions. For hydrocarbon flames there is no sign of atomic oxygen in the inner cone, and this is taken as strong evidence in favour of a peroxidation rather than a hydroxylation mechanism.