The point is that it is impossible to know all variables for any complex system, in general, until tests approach real world conditions.
Empirical testing with unknown, uncontrolled variables has merit because it tells you if your hypothetical model accounts for all relevant variables.
Case in point, the circumspect testing results prompted more digging, revealing contradictory reporting of efficiency results from the factory.
The manufacturer has discrepancies in its efficiency reporting and/or between skus and/or over time. The validity of ISO 4548 is irrelevant if the results are misreported. The above is for p/n PBL10241, used in the questioned empirical test, showing 99% at 20 micron. Could it be that this particular sku has better efficiency than others?
The merit of such empirical test was therefore in prompting further investigation to explain unexpected results. Your explanation was a test error… certainly a good potential explanation, but not proven and based on the assumption that all Boss filter SKUs have had the same efficiency since 2021.
Other potential explanations include:
- A change in the product.
- A misreported test result from the manufacturer.
- Differences between skus within the same product line.
- Differential decreases in efficiency between filters as the test progressed.
My opinion is that other recent empirical testing would be ideal, as it would reduce the effect of misreporting or product changes, the which is why I was asking if it was available.