Any measurement repeated produces a distribution rather than a single value. Two laboratories drawing from that distribution will differ, and the size of the expected difference is a property of the method.
Any measurement repeated produces a distribution rather than a single value. Two laboratories drawing from that distribution will differ, and the size of the expected difference is a property of the method. Treating any disagreement as error misunderstands what a measurement is.
Different separation techniques, detectors and sample preparation produce genuinely different numbers for the same material. Comparing results across methods without saying so is the most common source of apparent contradiction in published figures.
If two laboratories received different portions of a batch, they measured different material. Heterogeneity within a production run can exceed the difference between laboratories, and the comparison then says nothing about either.
A percentage is measured against a reference of known content. Laboratories using different reference materials, or references of differing quality, will differ systematically rather than randomly, and the difference will persist across repeat tests.
Some laboratories report to one decimal place, some to two, some cap a result at a ceiling figure. A difference that looks substantial can partly be a reporting convention, which is visible only if both reports state their practice.
Differences far larger than the method's expected spread, on the same portion of the same batch, measured by the same technique, are worth investigating. That combination is rare and is the case where one result is probably wrong.
Laboratories participate in schemes where everyone measures identical distributed samples and results are compared. Participation is a meaningful credential and is the direct evidence that a laboratory's numbers track everyone else's.
Check whether the method, the batch portion and the reference standard were the same before treating the difference as meaningful. Most published disagreements dissolve at one of those three.
Laboratories participate in programmes where everyone measures identical distributed samples and results are compared. Participation is a meaningful credential and the most direct evidence that a laboratory's numbers track everyone else's.
Whether the method, the portion of the batch and the reference standard were the same. Most published disagreements dissolve at one of those three before any question of error arises.
The claims on this page about different results on identical material, without either being wrong can be checked against sources outside this site: