Interference by leachates from plastic components, which has been mentioned occasionally in the past, has recently been indentified as a cause of difficulty with several receptor systems. This may perhaps have distracted attention from a very common problem that has been known to analysts for 50 years or more: at low concentrations, the measured concentration of a compound tends to decrease more rapidly than the dilution. Examples I recall from the 1960s and early 1970s concerned the determination of steroids, and pharmacokinetic studies.

Frequently there is a fairly sharp threshold below which a compound just practically disappears. When this happens to a receptor ligand, one would obtain the expected sigmoid response, but the apparent IC50 would be a function of the loss threshold and not of the affinity of the receptor for the ligand.

"Saturable" losses might be due to traces of reactive impurities in solvents. Often, they are attributed to retention on the surface of glass or plastic pipettes, containers or stoppers. Losses could be adsorptive or by ion exchange. When I was running a GC/MS laboratory 2-3 decades ago, we would try a method out before taking the decision to silanise the glassware, a time consuming procedure. Sometimes, a suitable amount of a stable isotopically-labelled internal standard or an unlabelled close analogue was sufficient.

In liquid chromatography attention is paid to the materiels used for the septa of glass sample vials. Some workers simply avoid letting the sample solution coming into contact with the septum.

Saturable loss is usually easy to detect in the analytical field, because most kinds of calibration curve ought normally to pass through the origin.

In my experience, receptor pharmacologists would rarely ask for analytical support in order to verify the concentrations of in vitro media. Analytical method development used to be quite long and difficult. In any case, good practice is to repeat critical experiments on different days; this would likely catch serious problems over losses, which tend to be very variable. Nowadays, it could be argued that important data should be backed up by LC/MS determinations, for which "generic" rapid methods are available.