Someone called?
A few comments:
First, it is correct that fuel does not burn, it is the mixture of fuel vapour and the oxygen in air that burns. However, I believe that fuel tanks are not usually sealed from the atmosphere but have "breathers" to link them to the air; as fuel is used up, the empty space will be filled with a vapour which is a mixture of air and fuel particles, which can explode if ignited. This risk was always considered high, and it was not unusual for arrangements to be made to replace the fuel used up in tanks with a gas other than air to prevent it happening. I know that the Russians even went to the trouble of pumping exhaust gas into the fuel tanks of some of their combat planes for this very reason.
It is also correct that a steel-cored AP bullet can create sparks when it hits some suitable hard material, so these could in theory set off the fuel tanks. However, the impression I have is that this happened very rarely.
Tracers would stand a higher chance of igniting fuel, but again practical experience in WW2 showed that this didn't often happen. Tracers were not all the same: in the BoB the RAF made some use of the .303 B Mk IV "Incendiary Tracer". This was an incendiary which ignited on firing, leaving a smoke trail behind. But by then, most incendiaries ignited on impact with the target, such as the .303 B Mk VI "De Wilde", which was copied in a simplified form for the US .30 and .50 incendiaries.
The British tested RAF .303 and Lufwaffe 7.92 incendiary ammunition, as described in
Flying Guns – World War 2: Development of Aircraft Guns, Ammunition and Installations 1933-45:
"Comparative British tests of British .303" and German 7.92 mm incendiary ammunition held in 1941 against the self-sealing wing tanks in a Blenheim bomber, fired from 200 yards (183m) astern, revealed that the .303" B. Mk IV (based on the First World War Buckingham design) and the 7.92 mm were about equal, each setting the tanks alight with about one in ten shots fired. The B. Mk VI, which contained 0.5 grams of SR 365 (a composition including barium nitrate) was twice as effective as these, scoring one in five. The 7.92 mm API completely failed to ignite the tanks."I expect that a .50 calibre incendiary would have done better than this due to the larger quantitiy of incendiary material carried.
Tony Williams: Military gun and ammunition
website and discussion
forum