recovery
/ rih-KUV-er-ee /
Imagine pouring exactly one litre of water through a leaky funnel and catching only 920 millilitres at the bottom — the rest soaked into the sides. Recovery answers the same question for an analysis: of the analyte you know went in, how much did your method actually get back out at the end and report?
Recovery is the fraction of analyte that survives the whole procedure and is measured, usually given as a percentage of what was truly present. It is found by adding a known amount of analyte to a sample, running it through every step, and comparing the measured amount with the amount added. A recovery of 100% means nothing was lost or gained; 80% means a fifth went missing along the way.
It matters because losses during extraction, filtration or cleanup, or boosts from contamination and interference, quietly bias your final number even if the instrument reads perfectly. The caveat is that good recovery is necessary but not sufficient — a steady 85% recovery can be corrected for, but you must show it is consistent, and recovery alone does not prove the method is free of every other error.
A water sample is split in two; to one half the analyst adds a known 50 µg/L of a herbicide. The spiked half measures 47 µg/L above the unspiked half, giving a recovery of 47/50 = 94%, judged acceptable for the method.
Comparing measured to added analyte to see how much the method gets back.
Recovery is most often assessed by the spike-recovery experiment; an analyte recovered well below 100% usually signals losses during sample preparation, while a recovery above 100% hints at contamination or a positive interference.