Assume $\chi$ is some sufficiently large regular cardinal, and $\langle H(\chi),\in, <_\chi\rangle$ is as usual.

An elementary substructure $M$ of $H(\chi)$ is

*$\kappa$-presentable*for a regular cardinal $\kappa$ if $M=\bigcup_{i<\kappa} M_i$ for some sequence $\langle M_i:i<\lambda\rangle$ where- Each $M_i$ is an elementary substructure of $H(\chi)$,
- $i<j<\lambda\Longrightarrow M_i\subseteq M_j$,
- if $\delta<\lambda$ is a limit, then $M_\delta=\bigcup_{i<\delta} M_i$,
- $M$ has cardinality $\kappa$,
- $\kappa+1\subseteq M$, and
- $M_i\in M_{i+1}$ for each $i<\kappa$ (so $M_i\in M_j$ whenever $i<j<\kappa$)

No assumption is made on the cardinality of $M_i$ for $i<\kappa$, so each $M_i$ could have cardinality $\kappa$, or cardinality less than $\kappa$.

This is clearly related to the notion of a structure being internally approachable of length $\kappa$; this concept suffices for the [AM] presentation, but I'm not certain it suffices for what Shelah is doing in [Sh:371]. This is one of the details that I'd like to clear up.