The use of the sea pattern nomenclature was a provisional tactic, a fi rst pass at describing the animations and theorizing the potential range of kinetic pattern. Clearly, there are defi ciencies in the taxonomy, as there were a signifi cant number of animations not ascribed within the range of classes. Despite the use of a rigorous approach where six basic types were defi ned, from which the theoretical number of hybrids could be generated, a signifi cant number were considered to be outside the taxonomy. As discussed at the end of the previous chapter, the non-ascribable animations fell into three groups. The fi rst was located in relation to those animations generated by noise algorithms, where what was described as a ‘counter movement’ occurred. Two patterns moving in opposite directions overlap, and rather than appearing as a hybrid, the two are distinct, creating a unique self-cancelling outcome. The other two typical non-ascribable outcomes occurred in relation to the series generated by cellular automata (CA) scripts. The variants of CA used produce patterns that are under continual reconfi guration: patterns form in seemingly random locations; their relative proportions shift; they may be relatively static; incrementally changing direction or dramatically re-orienting the direction of the pattern shift. However, on close analysis, two distinct transformations recur: life-like CA, as implemented in

this study, produce large-scale spatial shifts over rapid time periods; those based on cyclic CA were also continually reforming, but these occurred as an incremental interweaving along the edges of spatial zones; while the fl ocking script exhibited both these characteristics.