The argument from the good lot

The argument from the bad lot highlights a systematic vulnerability of inferences to the best explanation: inferences will fail if the true hypothesis is not initially under consideration. The present paper develops the natural counter to the bad lot objection: the argument from the good lot. In inferences to causal relevance, scientists routinely rank logical contradictories: "C is a cause of E" and "C is not a cause of E". Thus, they operate within an exhaustive hypothesis space. To demonstrate the conceptual and historical force of this point, I analyze in detail two historical case studies. The first is the discovery of the etiology of puerperal fever in the nineteenth century; the second is the identification of chromosomes and nucleic acids as a material basis of heredity in the early twentieth century. These case studies show that the scientific community recognizes inferences within good lots as decisive, and that the conclusions drawn from such inferences tend to remain stable in the long run. Importantly, good lot inferences are not restricted to observable causes, provided that access to remote entities can be established by suitable detection and intervention procedures. And while we might expect that unknown confounders threaten inferences to causal relevance in the same way as unconceived alternatives threaten inferences to the best explanation, we will see that confounders present a more tractable methodological challenge. On the basis of the argument from the good lot, I outline an account of stability without stasis in biological knowledge.