Since the 1960s, various models have been proposed to explain cross-linguistic variation of the transitive clause. A persistent problem has been the difficulty of making correct generalisations to account for asymmetries in the empirical data. In the absence of a clearly formulated null hypothesis, higher-level theorising has lacked a scientific grounding. This paper examines word-order research to uncover the null hypothesis. The simplest explanation of the attested distribution pattern is that the conceptual similarity of subject (S) and object (O), together with their statistical difference, gives rise to the preference of highlighting the secondariness of object in the basic order. As an outcome of primacy highlighting, fronting SO is preferred cross-linguistically. The inverse, fronted OS, is the most infrequent type.