There is a bug in MID$() in FB 3.1.5 beta.  Contrary to the behaviour in
FBII (and contrary to the Reference Manual's statement that "If startPos  is
greater than the length of string$ , then a null (zero-length) string is
returned.", the following code fragment correctly gives a length of 0 and a
null string in FBII, but gives the curious string "ibcdefghij...." of length
248 in FB^3 (Lite or FBII emulation)

LOCAL
DIM i,63 txt$,255 a$
LOCAL FN Tester
  txt$ = ""
  FOR i=_"a" TO _"z"
    txt$ = txt$+CHR$(i)
  NEXT
  txt$ = ""
  a$ = MID$(txt$,9)
  PRINT LEN(a$),a$
END FN

This illustrates it some more. Note that a negative numChars parameter
doesn't inhibit it:

LOCAL
DIM i,63 txt$,255 a$, b$
LOCAL FN Tester
  txt$ = ""
  'fill txt$ with location-specific characters:
  FOR i=_"a" TO _"z"
    txt$ = txt$+CHR$(i)
  NEXT
  txt$ = LEFT$(txt$,5)
  FOR i=3 TO 8
    a$ = MID$(txt$,i)
    PRINT i,LEN(a$),a$
    b$ = MID$(txt$,i,LEN(txt$)-i+1)
    PRINT ,LEN(b$),b$
  NEXT
  b$ = MID$(txt$,7,-2)
  PRINT "*",LEN(b$),b$
END FN

-- 
Robin
====================================================
Genesearch Pty Ltd
E-mail: robinc@...
WWW:    http://www.genesearch.com.au
====================================================