--- canonical/LM_Modal_Bridge_Report.tex
+++ derived/LM_Modal_Bridge_Report.tex
@@ -43,6 +43,8 @@
 \author{Research synthesis and independent finite verification}
 \date{18 September 2026}
 
+% Reading-copy typography only: narrower table padding and breakable paths.
+\setlength{\tabcolsep}{4pt}
 \begin{document}
 \maketitle
 
@@ -85,17 +87,19 @@
 
 The central question is whether the schematic chain
 \[
+\begin{gathered}
 \text{formula-valued LM}
 \longrightarrow
 \text{Boolean valuation}
 \longrightarrow
-\text{CM/coefficient representation}
+\text{CM/coefficient representation}\\
 \longrightarrow
 \text{state/resource}
 \longrightarrow
 \text{effect/measurement}
 \longrightarrow
 \text{logical or modal outcome}
+\end{gathered}
 \]
 is a genuine mathematical semantics, rather than a collection of later constructions sharing matrix
 notation.
@@ -119,14 +123,14 @@
 
 Thus the clean architecture is
 \[
-\boxed{
+\boxed{\begin{gathered}
 \text{Boolean symbolic LM calculus}
 \longrightarrow
-\text{\(A\)-enriched symbolic calculus}
+\text{\(A\)-enriched symbolic calculus}\\
 \overset{\text{valuation}}{\longrightarrow}
 \text{\(A\)-valued states/effects}
 \overset{\text{support contract}}{\longrightarrow}
-\text{modal outcomes}.}
+\text{modal outcomes}.\end{gathered}}
 \]
 
 The missing structure consists of at least three independent ingredients:
@@ -459,7 +463,7 @@
 So an enriched symbolic scalar may be read simply as an \(A\)-valued function of the Boolean
 valuation. This is the finite Boolean-power viewpoint.
 
-\section{The strongest bridge theorem: symbolic support realizes modal possibility}
+\section{The strongest bridge theorem:\\ symbolic support realizes modal possibility}
 
 Let
 \[
@@ -909,7 +913,7 @@
 \end{center}
 It uses only the Python standard library. The exact machine-readable output is included as
 \begin{center}
-\texttt{tests/lm\_modal\_bridge\_checks\_output.json}.
+\path{tests/lm_modal_bridge_checks_output.json}.
 \end{center}
 
 The expanded run verified:
@@ -959,25 +963,27 @@
 The best placement is a central section of the broader Boolean--modal synthesis paper. A strong
 narrative is:
 \[
+\begin{gathered}
 \text{formula LM}
 \to
 \text{Boolean-ring contraction}
 \to
-\text{valuation naturality}
+\text{valuation naturality}\\
 \to
 \boxed{\text{idempotent obstruction}}
 \to
 \text{\(A\)-enrichment}
 \to
-\boxed{\text{symbolic-support theorem}}
+\boxed{\text{symbolic-support theorem}}\\
 \to
 \text{modal effects/bases}
 \to
-\boxed{\text{extra support contract}}
+\boxed{\text{extra support contract}}\\
 \to
 \text{division-free branch update}
 \to
 \boxed{\text{tensor/update closure obstruction}}.
+\end{gathered}
 \]
 
 This structure has three advantages:
@@ -1034,11 +1040,11 @@
 \begin{lstlisting}[language=bash]
 python tests/lm_modal_bridge_checks.py
 \end{lstlisting}
-The output should match \texttt{tests/lm\_modal\_bridge\_checks\_output.json} modulo the environment
+The output should match \path{tests/lm_modal_bridge_checks_output.json} modulo the environment
 string. The script contains no network calls and depends only on the Python standard library.
 
 The prior adversarial-review archive is preserved under \texttt{inputs/} and unpacked under
-\texttt{prior\_adversarial\_review/}. The package manifest records SHA-256 hashes for every included
+\path{prior_adversarial_review/}. The package manifest records SHA-256 hashes for every included
 file.
 
 \section{Search strategy}