Fred Suppe claims that the refereed journal article is an appropriate unit of scientific debate for philosophical analysis. He also claims that when we regiment scientific papers correctly, we can see that the hypothetico-deductive method, Baysian induction, and inference to the best explanation fail to capture the structure of scientific articles adequately. In what follows I demonstrate that the coding scheme Suppe used for uncovering the structure of a scientific paper is not appropriate under all circumstances, illustrate alternative structures found in various scientific articles, and show that the hypothetico-deductive method can accommodate the alternative structures I find. My conclusions are that the article that Suppe analyzed is not paradigmatic of published scientific articles, that different papers have different structures, that the structure depends upon the rhetorical goals of the article, and that, because of the different structures and different goals, no one philosophical account of testing is going to suffice.