The Selby NOACK, I've posted to a few of Selby's papers before.
The variance with whatever technique is used is still significant.
The important words are
Quote:
In an effort to ake D5800 more useful for licencing and auditing engine oils, the D02B0.07 Bench Test Surveillance Panel implemented a new strategy for LTMS monitoring. Using Exponentially weighted moving averages of monthly calibration runs, instrument that exhibit a consistently severe bias receive a severity adjustment, while inconsistent instruments are excluded from the system. While a common approach in engine testing, this marks the first time this monitoring technique has been applied to bench testing.
It's a precision and bias thing...
Machines that repeatedly give the same answer, but at one end of the spectrum will be given a "handicap"...those that produce random variance of sufficient magnitude get excluded.