Principles and laws that apply to nonorthogonal multiphase experiments are developed and illustrated using examples that are nonorthogonal but structure-balanced, not structure, but first-order, balanced or unbalanced, thus exposing the differences between the different design types. The design of such experiments using standard designs, a catalogue of designs and computer searches is exemplified. Factor-allocation diagrams are employed to depict the allocations in the examples, and used in producing the anatomies of designs or, when possible, the related skeleton-analysis-of-variance tables, to assess the properties of designs. The formulation of mixed models based on them is also described. Tools used for structure-balanced experiments are also shown to be applicable to those experiments that are not. Example 1. A small wheat experiment with a Youden square design in the laboratory phase. A simpler example of a nonorthogonal multiphase experiment is used to establish concepts and introduce terminology associated with them.Design identification. A two-phase experiment begins with a field phase in which seven lines of wheat are compared, with each line replicated six times. Discussion with the researcher establishes that a soil fertility trend in one direction is likely, but not in the perpendicular direction. That is, the anticipated model is Lines | Blocks + Blocks∧Plots, where the term to the left of the vertical line (|) is assumed fixed and those to the right are assumed random. The underlined term is an identity term. A randomized complete-block design is A-optimal for this model.In the second phase, a single sample of grain is taken from each of the 42 plots and the produce from the 42 plots is allocated to 42 oven spots for determining the moisture content using the standard oven drying method. The oven can hold seven samples and so six drying runs are needed. It is believed that drying could vary between spots in the oven in a manner that is consistent from one run to the next. Extra variation is also likely between the runs. Main effects would capture these two behaviours and so the anticipated model for the second-phase units is the random model Runs + Spots + Runs∧Spots and a row-column design is required for the allocation of plots to spots. However, each first-phase unit factor cannot be allocated to a second-phase unit factor; in particular, it is impossible to allocate the 42 levels of Blocks∧Plots to the seven levels of Spots. So the allocation to Runs∧Spots needs to consider the allocation of Lines, to avoid an undesirable amount of Lines information being confounded with Spots. When the formulation of the allocation to second-phase units must account for the factors in both first-phase tiers and the allocations are randomizations, the randomizations are termed randomized-inclusive (Brien & Bailey 2006. For the example, both the lines and plots tiers must be accounted for in formulating the randomization to the spots tier. This is in contrast to composed randomizations, in which the firstphase a...