When timber buildings are excavated, the post holes must be connected in a disciplined manner, so that the reliability of the patterns that are selected may be tested. Two cases from bronze age Britain are discussed. In the first example, the original analysis of the post holes at aldermaston wharf was repeated using an explicit statistical procedure, and the resulting configurations tested for conformity to the original model. In the second, a blind test was applied to the post-hole distributions inside a similar enclosure at south lodge camp, where complementary evidence for the position of the house floors was known only to the investigators. The statistical procedures are explained, together with construction of the reference model. This approach indicates the value of putting normal procedures of grouping post-holes on a more secure footing. Diagrams, histograms, references.