Let R be a Seifert surface obtained by applying Seifert's algorithm to a connected diagram D for a link L. In this paper, letting D be almost alternating, we give a practical algorithm to determine whether L is a fibered link and R is a fiber surface. We further show that L is a fibered link and R is a fiber surface for L if and only if R is a Hopf plumbing, that is, a successive plumbing of a finite number of Hopf bands. It has been known for some time that this is true if D is alternating, and we show that it is not always true if D is 2-almost alternating. In the appendix, we partially answer C. Adams's open question concerning almost alternating diagrams.

2000 Mathematical Subject Classification: 57M25.

It is an interesting open question when Dehn surgery on a knot in the 3-sphere S3 can produce a lens space (see [10, 12]). Some studies have been made for special knots; in particular, the question is completely solved for torus knots [21] and satellite knots [3, 29, 31]. It is known that there are many examples of hyperbolic knots which admit Dehn surgeries yielding lens spaces. For example, Fintushel and Stern [8] have shown that 18- and 19-surgeries on the (−2, 3, 7)-pretzel knot give lens spaces L(18, 5) and L(19, 7), respectively. However, there seems to be no essential progress on hyperbolic knots. It might be a reason that some famous classes of hyperbolic knots, such as 2-bridge knots [26], alternating knots [5], admit no surgery yielding lens spaces.

In this paper we focus on the genera of knots to treat the present condition methodically and show that there is a constraint on the order of the fundamental group of the resulting lens space obtained by Dehn surgery on a hyperbolic knot. Also, this new standpoint enables us to present a conjecture concerning such a constraint, which holds for all known examples.