A recent report from Circle of Blue, credited to Alissa de Chassey and dated October 23, 2026, turns on a condition that sits upstream of most PFAS headlines. Its summary begins with the case where "When PFAS concentrations in drinking water fall below laboratory reporting limits" the number on the page stops describing the water in the same way a measured value does. The summary does not give the study's methods, so this explainer does not assume them. It addresses the bookkeeping question that any reader of a PFAS table has to answer: what does a result look like when the instrument could not resolve it?
Three numbers that look like one
A PFAS report usually carries several thresholds, and they are easy to confuse. The method detection limit is the lowest concentration the method can distinguish from background with stated confidence. The reporting limit is the lowest concentration the laboratory will report as a quantified value, and it is often set above the detection limit. A detected result sits above the reporting limit. A result below it is commonly flagged with a qualifier, reported as not detected, or given as an estimated value. Labels differ between laboratories and programs, so the qualifier code and the limits printed for that sample batch control the reading, not the column heading.
The substitution rule is a decision
When a result is non-detect, someone has to choose what number, if any, goes into an average. The common options are to treat the value as zero, to treat it as the reporting limit, to treat it as half the reporting limit, or to leave it out. Each produces a different mean, and none is neutral.
Zero gives a lower bound. It is correct only if the compound is truly absent, and a non-detect does not establish that. The reporting limit gives an upper bound, which overstates a mean built from many results that sit well below the limit. Half the limit is a midpoint convention that is easy to defend in a table but has no physical basis. Excluding non-detects can push the mean upward, because the removed values are the lowest ones. A file that reports detects only describes detects only.
The choice should follow the purpose. A compliance comparison, a public summary, and a research dataset may each call for a different rule, and the rule should be fixed before anyone looks at the direction of the result. Choosing the convention after seeing which one makes the number look better or worse is the failure to avoid.
Write the record before the number
A defensible PFAS summary usually carries the following entries alongside the mean:
The count of total results, detected results, and non-detect results for each compound and each sampling event. The reporting limit for each result, since limits can differ between samples because of dilution or sample matrix. The qualifier code as the laboratory issued it. The substitution rule used, stated in plain words. A lower-bound and upper-bound pair, with zero and the reporting limit respectively, whenever non-detects are a meaningful share of the set. When the bounds are far apart, the bounds are the finding, and a single midpoint hides that.
Where the uncertainty sits
A non-detect is a statement about the method at a given threshold. It is not a statement that the compound is absent from the water. Laboratories may lower reporting limits as methods change, which can turn an earlier non-detect into a detect in a later round, and that shift can read as a trend when it is partly a change in measurement capacity. A comparison across years should note whether the limits were the same.
None of this settles how a particular study or utility handled its non-detects. The reader's task is narrower: confirm what the table says, identify which rule produced the average, and keep the bounds visible until the purpose of the number is clear. That discipline costs little at the file stage and avoids a correction later.