Actually, IMO these are pretty good comparison numbers.
I used to be a QA manager for an environmental laboratory and control samples usually had a range of +/- 15-20% depending on the compound, the matrix (soil, water, oil, etc.) and the level (how low) at which you are testing at.
I will tell you that any laboratory worth it's weight should be doing quality control samples along side unknown samples in some sort of batch. Typical control samples are:
Laboratory Control Sample - Known sample matrix and level; Ensures that the whole process is correct.
Laboratory Duplicate - Pick a sample in a batch and test it twice. Look for repeatability.
Matrix Spike - Pick a sample in a batch after testing it, add a known amount of compound(s) to it and test it again. Looking for the effect of the matrix (oil...) on the compound(s) of interest.
Bottom line is, the laboratory should have some sort of QA program in place. In addition, usually the test methods that they are using may have some QA requirements as well.
Bill
[ January 29, 2003, 06:43 PM: Message edited by: billman ]