Recently a number of philosophers (e.g. Feyerabend and Kuhn) have maintained that the meanings of terms in a scientific language are “theory-laden” or determined by the theory in which they occur, and thus that if the same term (e.g.; ‘mass’) occurs in different theories, it will take on different meanings in the different theories; so the theories are incommensurable. An often-stated corollary to this doctrine is the claim that possessors of different theories cannot express or possess the same facts since they attach different meanings to the terms used to give expression to the facts. Various attacks against this extreme doctrine on the relativity of facts have been mounted. Some of them consist in showing defective the argument advanced in support of this doctrine; but such criticisms at best show that the defenses offered for the doctrine are defective, not that the doctrine itself is defective.

The positivistic Received View construed scientific theories syntactically as axiomatic calculi where theoretical terms were given a partial semantic interpretation via correspondence rules connecting them to observation statements. This paper assesses what, with hindsight, seem the most important defects in the Received View; surveys the main proposed successor analyses to the Received View—various Semantic Conception versions and the Structuralist Analysis; evaluates how well they avoid those defects; examines what new problems they face and where the most promising require further development or leave unanswered questions; explores implications of recent work on models for understanding theories; and rebuts the few criticisms of the Semantic Conception that have surfaced.

Scientific articles exemplify standard functional units constraining argumentative structures. Severe space limitations demand every paragraph and illustration contribute to establishing the paper's claims. Philosophical testing and confirmation models should take into account each paragraph, table, and illustration. Hypothetico-Deductive, Bayesian Inductive, and Inference-to-the-Best-Explanation models do not, garbling the logic of papers. Micro-analysis of the fundamental paper in plate tectonics reveals an argumentative structure commonplace in science but ignored by standard philosophical accounts that cannot be dismissed as mere rhetorical embellishment. Papers with illustrations often display a second argumentative structure differing from the text's. Constraints on adequate testing and confirmation analyses are adduced.

Achinstein, Putnam, and others have urged the rejection of the received view on theories (which construes theories as axiomatic calculi where theoretical terms are given partial observational interpretations by correspondence rules) because (i) the notion of partial interpretation cannot be given precise formulation, and (ii) the observational-theoretical distinction cannot be drawn satisfactorily. I try to show that these are the wrong reasons for rejecting the received view since (i) is false and it is virtually impossible to demonstrate the truth of (ii). Nonetheless, the received view should be rejected because it obscures a number of epistemologically important features of scientific theorizing. I show this by sketching an alternative analysis which reveals some of these features and gives a more faithful picture of scientific theorizing.