Laboratory duplicates (lab dup) are necessary and typical quality control sample used for all analyses.
A lab dup in a batch of 20 samples or less is randomly selected, so the reported Relative Percent Difference (RPD) values reported on the Certificate of Analysis may represent submission in the batch.
For soil samples, the lab duplicate is taken from the same container as the sample (i.e. a second duplicate). Soil sample are notoriously heterogenious, so RPD values that fall outside acceptance criteria are simply a representation of the matrix.
For water samples, depending on the analysis, the lab duplicate may actually be representative of a field duplicate (i.e. for organics analysis when whole container is used for extraction and analysis); in this case, a second container collected in the field is really considered as a separate sample for which RPDs falling outside control limits may again be representative of the matrix.
For water analyses where the volume provided allows for reanalysis, a second aliquot analyzed as a laboratory duplicate that falls outside control limits may be representative of an inconsistent matrix if sediment or suspended particulate matter is present.
It is recommended that whenever you review data falling outside the suggested or required limits, follow up with the laboratory to request an explanation or commentary.