Have made many partial discharge tests of oil cooled and dry type transformers to 60 kV RMS. (test equipment limitation)
When testing dry type transformers, the partial discharge start voltage is very repeatable.
When testing oil cooled transformers, if the test voltage is held for any length of time (partial discharge for seconds or less) the corona start voltage decreases drastically. What happens is the partial discharge breaks the oil (but not the solid insulation) into a gas and the gas breaks down at a much lower voltage than the oil. The partial discharge start voltage increases to it's original value if the transformer is not energized for some time. The time required for the transformer to return to it's original partial discharge start voltage depends on how long it takes for the gas to disolve into the oil, which depends on the insulation scheme. This time can be minutes or days.
An interesting question is, if you have a cast type transformer with an inclosed gas pocket, will the partial discharge increase the gas pressure, so that the longer there is partial discharge, the higher the partial discharge start voltage?
(I don't know the answer)