LaTeX defines \vec as
\DeclareMathAccent{\vec}{\mathord}{letters}{"7E}
which eventually is expanded to
\mathaccent"017E\relax
whereas plainTeX defines it as
\def\vec{\mathaccent"017E }
replacing \relax with a space.
Is there any practical difference?
The syntax for \mathaccent is (TeXbook, p. 291)
\mathaccent⟨15-bit number⟩⟨math field⟩
A ⟨15-bit number⟩ is an integer (in any denotation allowed by LaTeX) in the range 0–215. What's a ⟨math field⟩? We find it on page 289
⟨math field⟩ → ⟨filler⟩⟨math symbol⟩ | ⟨filler⟩
{⟨math mode material⟩}
A ⟨filler⟩ is any sequence of space tokens and \relax tokens which are ignored (doing expansion) when processing the input stream looking for something that fits the syntax rule.
Thus in LaTeX the \relax will be gobbled as a ⟨filler⟩, whereas in plain TeX the space terminates the specification for the number and is ignored as well.
Why does LaTeX use \relax? The answer is in the definition of \DeclareMathAccent that internally uses \set@mathaccent:
% latex.ltx, line 3596:
\def\set@mathaccent#1#2#3#4{%
\xdef#2{\mathaccent"\mathchar@type#3\hexnumber@#1#4\relax}}
The choice might depend on the fact that when writing to an auxiliary file \relax is better suited to the job.
