5f8e45ce25dc32d7.tex
1: \begin{definition} If $\pi\colon (\mathcal{Y}, \mathcal{D})\to \Delta$ is a Type III degeneration, we define the \textsl{monodromy invariant of $\mathcal{Y}$} to be the class $\lambda\in \mathcal{B}_{\text{\rm{gen}}}\cap \Lambda$ modulo the action of $\Gamma(Y,D)$, the group of admissible isometries. The monodromy invariant only depends on the central fiber $(Y_0,D)$, and so we shall also use the above to define the monodromy invariant of a $d$-semistable Type III anticanonical pair. 
2: 
3: A Type III degeneration is \textsl{of $\Q$-type $\lambda$} if its  monodromy invariant is of the form $r\lambda$, $r\in \Q^+$.
4: The $\Q$-type only depends on the family $(\mathcal{Y}^*, \mathcal{D}^*) \to \Delta^*$ and is preserved by base change. In particular, let $\overline{\pi}\colon (\overline{\mathcal{Y}}, \mathcal{D}) \to \Delta$  a one parameter family  whose central fiber is the contraction of the cusp $D'$ in $V_0$ and whose general fiber has only RDP singularities. By the argument of \cite[Theorem 2.5]{FriedmanMiranda}, possibly after a base change, the family $\overline{\pi}\colon (\overline{\mathcal{Y}}, \mathcal{D}) \to \Delta$ is birational to a Type III degeneration $\pi\colon (\mathcal{Y}, \mathcal{D})\to \Delta$ via  a birational isomorphism which is a minimal resolution on the fibers away from $0$. Thus, it makes sense to say $\overline{\pi}\colon \overline{\mathcal{Y}} \to \Delta$ is of $\Q$-type $\lambda$.
5: \end{definition}
6: