Recent work on testimony and the norms of assertion considers cases of expert testimony. Thinking about expert testimony clarifies which epistemic goods figure in the expectations placed on experts for their knowledge. Examining the distinctive conditions of expert testimony and the assumptions hearers bring to such conversational contexts can provide broader lessons about how knowledge is represented by speakers, and how it is gained by hearers.Expert testimony is the focus of Jennifer Lackey's (2011, 2013) recent arguments over the norm of assertion. Her cases suggest that experts are plausibly held to a different epistemic standard when speaking as experts, where this standard concerns not only the quantity of an expert's epistemic support for what an expert asserts, but also the quality of that epistemic support. These cases seem to count against knowledge being sufficient for epistemically permissible assertion; and several others have followed Lackey in regarding such cases as important to adjudicating the debate over the norms of assertion and practical reasoning (Carter and Gordon 2011, Coffman 2011, McKinnon 2012, Carter 2014, Gerken 2014, Green 2014.I shall argue in this paper that relying on intuitions about such cases of expert testimony introduces several problems concerning expertise, expert knowledge, and the sharing of such knowledge through testimony. In particular, the cases used thus far are unstable, and refinements are needed to clarify exactly what principles are being tested and exactly what our intuitive judgements are tracking in such cases. But as we shall see, once refined, such cases raise more questions than they answer.The plan of this paper is as follows. §1 introduces the sufficiency principle for epistemically permissible assertion, and notes how it does and does not relate to the knowledge norm of assertion. §2 considers Lackey's notion of isolated second-hand knowledge, clarifying various notions of knowledge as 'isolated' and of seeking an expert's opinion. §3 exposes a significant 492 MATTHEW A. BENTON