Maybe an extreme example; I once burned in, in the intended circuit, about 20 Soviet 6J9P-E for a differential preamp looking for pairs. The behaviour was absolutely wild to say the least. Pairs with several volts DC between the anodes to start with could drift to more than 20 and this did not take place over a matter of a few minutes but over more than 1 hour. They did finally settle down and subsequent measurement showed them to be quite stable so I have continued using this excellent valve. The serious looking military rhombuses on the envelope suggested they had already been burned in, albeit not at my operating point.