The ADJUST: command is the central tool of Bayes linear methodology as embodied in [B/D]. We focus on the three particular usages of the command. These are (1) the adjustment of one collection of elements by a number of others; (2) The partial adjustment of a collection by further collections, given others; and (3) The extraction of collections from an adjustment.
Adjustments may be performed with or without actual observations to hand: we do not distinguish between these two kinds of adjustment, although more aspects of the problem are examined when data is present. In the former case we also investigate the implications of our belief specifications, whilst the latter typically enable diagnostic comparisons between what was expected and what we observed.
[B/D] is designed to handle exchangeable situations. For these circumstances, the adjustments need to be organised in a particular way to exploit any exchangeability. Consequently we deal with this also below. In this chapter we will tend to refer to both elements and bases as information sources.