Any attempt at the unitary reconstruction of science as a whole, or at least of a set of sciences presumed to be representative of it, pursues a certain ideal of unity of scientific knowledge. One is interested here in the ideal of logical unity (alias, logical universalism), aiming at the reconstruction of all the sciences concerned within the same logical framework, and in the more demanding ideal of a systematic unity (alias, systematic universalism), aiming at the reconstruction of all of these sciences within one and the same system. In his famous Wahrheitsbegriff of 1935 (postscript not included), Tarski, the inventor of semantics as a science, proposed a criterion, the so-called convention T, which any definition of the notion of truth should meet in order to be adequate, and he constructed a semantic definition of this notion meeting convention T for some explicitly given systems taken as examples, fitting into a logical framework which he considered universal, viz. the simple extensional theory of types. But, he also brought to light the limits of the exercise by demonstrating negative theorems, which implied that, for no system fitting into the logical framework and including it, is a definition of the notion of truth meeting convention T possible—be this within this system itself or, more broadly, within the chosen logical framework—, without the system in question falling, in one way or another, prey to the Liar. And, moreover, there was every reason to think that one could not be content with an axiomatic explication of the notion of truth. Under these conditions, taking for granted that the Tarskian explication of the notion of truth for science should be on the reconstruction agenda, it would be all over for logical universalism (and a fortiori for any universalism more demanding than it), and Tarski in fact gave up on it, though without saying so, in his postscript to the Wahrheitsbegriff. In the present article, it is preferred to go back to the program of the Wahrheitsbegriff with a new logical framework, viz. Zermelo-Fraenkel set theory with axiom of choice, and to challenge convention T in favor of new conventions on adequacy. It then turns out to be possible to save logical universalism and to take a few steps forward on the path to systematic universalism.