Several difficulties arise: (1) A growing banjar might simply divide as another banjar adat association and not incorporate itself as a separate desa. (2) Frequently in organizing a new desa three -temple-cluste r, a set of banjar or factions first erects a death temple nearby; later they might take the more controversial and costly step of building the rest of the kayangan tiga. (3) The process - rarely simply a matter of harmoniously amalgamating a full-fledged free desa - can become politicized and provoke local factions. Moreover, new temples always require approval of officials who must also supervise dedication ceremonies, and a group must be legitimized to provide desa leaders and temple custodians as well.

In the ethnographic literature of this period, the process whereby temples mediate social groups and religious localities is indirectly discussed in analyses of the mother-daughter village pattern:

Such a village-area is economically and politically independent of all others, except for a curious relation of blood. It often happens that various neighboring villages are united by a strong bond into an association of related villages which worship a common original ancestor and with a common temple of 'origin' located in the oldest village of the group, which they recognize as a 'head' or 'mother' village. From this it is supposed that the other villages sprung, and when they grew became independent. Such village associations cooperate with one another by sending offerings and representatives to the temple feasts of the other desas (Covarrubias 193 7).