1: \begin{abstract}

2: \noindent

3: The phase diagram of a one-dimensional tight-binding model with a

4: pair-hopping term (amplitude $V$) has been the subject of some

5:  controvery. Using

6: two-loop renormalization group equations and

7: the density matrix renormalization group with

8: lengths $L\leq 60$, we argue that no spin-gap transition occurs at

9: half-filling for positive $V$, contrary to recent claims.  However, we

10: point out that away from half-filling, a {\it phase-separation}

11:  transition occurs at finite $V$.  This transition and the spin-gap

12: transition occuring at half-filling and {\it negative} $V$ are analyzed

13: numerically.

14: \end{abstract}

15: