For non-pictorial accents \c, \b, and \d, we want just the accented part. The default definitions provide an alternative options. In the last two cases, the \accent command uses faked access number.
<..revised accent definitions..>
\def\:temp#1{{\accent24 #1}}
\let\c:accent=\c \HLet\c:accent=\:temp \let\c=\c:accent
-_-_-
<..revised latex accent definitions..>
\def\:temp#1{{\edef\next{\the\font}\rmfamily\accent0 \next#1}}
\let\b:accent=\b \HLet\b:accent=\:temp \let\b=\b:accent
\def\:temp#1{{\edef\next{\the\font}\rmfamily\accent1 \next#1}}
\let\d:accent=\d \HLet\d:accent=\:temp \let\d=\d:accent
-_-_-
<..revised plain accent definitions..>
\def\:temp#1{{\edef\next{\the\font}\rm\accent0 \next#1}}
\let\b:accent=\b \HLet\b:accent=\:temp \let\b=\b:accent
\def\:temp#1{{\edef\next{\the\font}\rm\accent1 \next#1}}
\let\d:accent=\d \HLet\d:accent=\:temp \let\d=\d:accent
-_-_-
The last two faked fonts, similar to the definition of \t in plain.tex. There, however, we have \the\textfont1 instead of \rm and \rmfamily, which doesn’t work in LaTeX for some reasons (what are the reasons?).