We actually test all sold tubes on L3-3 testers (sometimes L1-3).
We have more than 10 such devices each ajusted precisely for the specific tasks.
When selecting tubes close in parameter, we re-test on the best (carefully calibrated devices.
It is theoretically impossible to select tubes with a difference of 0% on any device.
From a batch of 100 pieces of any new or vintage number of pairs with a spread of 1-2%, no more than 10.
During testing, any tube changes its parameters by 1-2% in any direction over time.
This information is based on inspections of over 100,000 tubes over many years.
--