Standard views about the factors that determine verb position and subject position predict that there should be Subject-Verb-Object languages in which tense and aspect are indicated by a particle or auxiliary that comes before the subject. Julien's (2000) large-scale survey of the languages of the world, however, indicates that this word order is never found. This striking gap suggests that the theory of how verbs are related to tense needs to be rethought. I suggest that the gap can be explained by abandoning Chomsky's (1993, 1995) checking theory, in which the relationship between the T node and the inflected verb can be established abstractly. The correct word order typology follows if the computational system of human language can combine tense and verb only by overt head movement (Baker 1988, Pollock 1989) or by the PF merger of morphemes under adjacency (Marantz 1988, Bobaljik 1994).