"git add -p" versus "git add -i", followed by "p"

  testing adding by patch for the very first time (i've just never
needed this), and reading the "progit" book and reading the man page,
and the impression i'm getting is that running "git add -p" (going
straight to patch mode) is supposed to be equivalent to running "git
add -i", then typing "p" to switch to patch mode.

  that is most emphatically not what i'm seeing. if i run "git add
-p", then i get to what i expect -- the patch subsystem:

  $ git add -p
  diff --git a/README.asc b/README.asc
  index fa40bad..840e85b 100644
  --- a/README.asc
  +++ b/README.asc
  @@ -1,3 +1,9 @@
  +change 1
  +
  +
  +
  +
  +
   = Pro Git, Second Edition

   Welcome to the second edition of the Pro Git book.
  Stage this hunk [y,n,q,a,d,j,J,g,/,e,?]?

but if i start with "git add -i", there seems to be no way to get to patch mode -- certainly "p" doesn't do it. am i stupidly missing something trivial? is the explanation misleading or inncomplete?

rday