Some commentators on Wittgenstein's Tractatus (e.g., Anscombe and Sellars) have argued that Tractarian names stand for Fregean objects, that there are no predicates in a Tractarian elementary proposition. Others have argued in response (e.g., Stenius, Hintikka and Hintikka and Hacker) that Tractarian names include monadic and relational predicates: that is, that Wittgenstein more or less shares with Russell a conception of the elementary proposition. I and others have argued elsewhere that both of these camps are wrong: Wittgenstein presupposes no determinate conception of the categorial structure of elementary propositions, and indeed even insists that the logician must not take a stand on this question, that elementary propositional forms cannot be given a priori. In the present paper I propose to show how this reading makes better sense also of Wittgenstein's transition to the next (fleeting) stage in his development, the paper "Some remarks on logical form." I will argue that it makes better sense of the argument he presents in that paper to take the categorially indeterminate reading of the Tractatus as its background, than to take the background conception of elementary propositional form to be "Russellian".