Norvin Richards. 1999. In Defense of the T-Model. Studies in Modern Grammar 15, 1-21. A great deal of recent work in syntax has concerned the nature of the derivation; should syntax be modeled derivationally? and, if so, what should be the nature of the output of this derivation? In this paper I try to defend one classic set of answers to these questions, sometimes known as the "T-model." The argument is based on the nature of multiple-wh constructions in a number of languages; we see that languages which leave all wh-phrases in situ (like Japanese and Chinese) and languages which move all wh-phrases overtly (like Bulgarian and Serbo-Croatian) behave similarly, and crucially differently from English, with respect to the Path Containment Condition. The T-model is the one model of the syntactic derivation that has been proposed that naturally groups the languages in this way. An account of the distribution of the data is then offered which makes crucial use of the T-model. In the course of developing the argumet, the nature of movement to multiple specifiers more generally is investigated.