As suspected, it was a lack of air that was causing the smoking and the extinction of wood gas.
The existing primary air holes were enlarged but the wood gas was again extinguished after a quarter of the fuel had been burnt.
For test run #4, more air holes were added to the bottom of the outer container.
When the stove pipe was put on top of the stove a forceful draught could be heard, which resulted in better combustion. No smoke was seen to be generated.
The ignited gas jets from the secondary holes at the top of the combustion chamber could clearly be seen.
The fuel used consisted of wood chippings from a garden shredder. This appears to be a better fuel than the wood "pellets" with regards to starting a fire. However, another test with the updated stove will be carried out tomorrow as the pellets are denser than the chippings and so can give off more heat energy per burn.
The pellets could also be kept back and used in some automated refuelling operation for a stove that is already in operation. The reasoning being that the pellets are of uniform size and an auger or similar could administer a given amount of pellets every so often. The stove could then sustain heat for as long as required.
Though the stove pipe improved the draught, the length of the pipe makes the stove less useful. Future experiments will involve finding the minimum pipe length that provides an adequate draught, allows access to the combustion chamber for refuelling and permits a water coil to be so positioned that water can be heated.