The data presented here are from 12 PTs, spanning the period 1998–2011. Table 1 summarises these tests. Up until July 2005 (PT 2134), the robust mean was used to set the assigned value. Although the technique of bump-hunting had been applied to the data since December 2002 (PT 2118), inexperience in searching for modality and how it might be applied to this type of data meant that it was not used to try and set the assigned value. Comparison of the mode, median and robust mean lead to the latter being set as the assigned value in all cases. However, PT 2127 results were issued for information only, i.e. there was a large uncertainty of the assigned value however the data were assessed. It should also be observed that PT 2118 and 2122 specifically required trans retinol to be reported. It might be expected that participants for these two tests used NP HPLC, in order to be able to resolve and report only the trans isomer. However, method returns indicate that the majority used C18 columns. PT 2140 did use the major mode as the assigned value, although the uncertainty was such that z-scores were issued for information only.