Redefined but still alive

In section 4 of Starting Forth tutorial, the answer to the 5th question is:

: STAR   [CHAR] * EMIT ;
: STARS  ( #stars -- )  0 ?DO  STAR  LOOP ;
: STARS  ( n -- )  ?DUP IF  STARS  THEN ;

The third definition redefines STARS word using previous definition.

Interpreter says:

redefined STARS   ok

and as I type see:

SEE STARS
: STARS  
  ?dup 
  IF     STARS 
  THEN ; ok

Why the former definition is still alive as the word STARS has been replaced in the words dictionary? And is it possible to access/see the former definition?

Solution

The second definition of STARS hasn't replaced the first definition. It has hidden it. The second definition calls the first definition.

words 
STARS STARS STAR mov-regv-Iv mov-reg8-Ib xchg-ax jcc-short conditions...
\ Stars appears twice in the words listing

The forth interpreter searches for the latest definition of a word in the dictionary. In this case once it finds STARS it stops looking any further.

: STAR   [CHAR] * EMIT ;
: (STARS)  ( #stars -- )  0 ?DO  STAR  LOOP ;
: STARS  ( n -- )  ?DUP IF  (STARS)  THEN ;
see stars 
: STARS  
  ?dup 
  IF     (STARS) 
  THEN ; ok
see (stars) 
: (STARS)  
  0 
  ?DO    STAR 
  LOOP
  ; ok
words 
STARS (STARS) STAR mov-regv-Iv mov-reg8-Ib ...

By naming the first STARS as something else, here (STARS), the first definition can still be accessed.

This hiding can be useful with tables of constant data. I think in 'Starting Forth' there's an example of a sine lookup table. The table is called sine and the word to access the table is also called sine. This means nobody can accidentally write into the sine table after the second sine is defined.