If there is a meaningful concept of scientific fact, it is one that cuts across the more careful standard distinction between scientific observation and scientific theory. That is, there are both empirical scientific facts (water at a pressure of one atmosphere boils at 100 degrees centigrade) and principled general scientific facts (cell theory describes the fundamental structure of living organisms). Both observations and theories may come to be regarded as factual within a scientific community; the sense of 'fact' that applies, though, is not only more rigorous but also more skeptical than factuality as ordinarily understood. To call a fact 'scientific' is not exactly to say that it is a more absolute fact (as common usage implies), but rather that it is a more accountable fact. The rigorous aspect of this accountability is secured by the methodical standards of science: theories must be testable, and be extensively tested; particular observations must meet the criteria of established practice in the relevant field and hence be, at least in principle, reproducible. The skepticism that attaches to the notion of fact in science is just the flip side of the accountability. Scientific theories have to be falsifiable to be testable, and their legitimacy is not a matter of confirmation, strictly speaking, but of demonstrated robustness in the face of efforts at falsification. As the Popperian doctrine of falsificationism has it, scientific facts are not ultimately confirmed, so much as tested beyond reasonable doubt (2002 [1959]).There are pragmatic dimensions to such factuality: theories are accountable within the scope of their applicability, within certain degrees of approximation, and within the limits of the questions they answer, and these considerations are contingent upon the current agenda of research in the relevant fields. A robust theory is one that is too well tested to be worth further testing, but that criterion is relative to a given state of the science. Scientific facts, then, are part of the discourse of knowledge, and there is an irreducible conceptual gap between facts in that sense and the states of affairs with which they are concerned.This point is similarly applicable at the level of scientific observations. They are factual to the extent that they are independent of the circumstan-