This one is invisible on paper and consequential everywhere else. The "µ" in µg/dL is not one character. It is two, and they look the same.
The two characters
U+00B5 is named MICRO SIGN in the Unicode Character Database. It is a legacy compatibility character, inherited from older character sets that needed a micro symbol.
U+03BC is GREEK SMALL LETTER MU — the actual Greek letter.
They are formally related: U+00B5 carries a compatibility decomposition to U+03BC, and its case-folding and NFKC_Casefold properties both map it to U+03BC. Unicode also lists them as confusables — visually similar, distinct code points.
For a reader, they are the same symbol. For software comparing two strings byte by byte, they are not equal.
Why lab reports contain both
Because reports come from many systems. A laboratory information system, a PDF generator, a clinician typing a unit by hand, and an instrument's own firmware may each emit a different one, and nothing in the printing pipeline normalises them. Two reports from the same laboratory can carry different code points in what reads as the same unit.
Nothing about the sample or the measurement differs. Only the byte does.
What breaks
Any system that decides whether two results are comparable by matching their unit strings. µg/dL
and μg/dL are the same unit to a person and different strings to a computer, so a naive
comparison concludes it does not recognise the second one.
The failure is quiet, which is what makes it expensive. Nothing errors. A value is simply treated as having an unknown unit, and drops out of a chart or a comparison without announcing why.
Lowercasing does not fix it — both characters are already lowercase, and they stay distinct.
How Arkalife handles it
Every unit comparison goes through one function that normalises these characters before deciding whether two units are the same, rather than comparing the printed strings directly.
This is not a hypothetical tidy-up. When the unit-matching in Arkalife was audited against real reports, most of the unresolved unit pairs turned out to be this single distinction — not genuinely unknown units, and not conversions that were missing, but the same unit written with two different code points. No conversion factor could have fixed them, because there was nothing to convert.
Why it is worth knowing
If you export lab data, paste it into a spreadsheet, or compare readings across systems, this is a plausible reason two identical-looking units refuse to match. Searching for one form will not find the other.
