Full Idea
Two things can't be vaguely identical, because then a would have an indeterminacy which b lacks (namely, being perfectly identical to b), so by Leibniz's Law they can't be identical.
Gist of Idea
There can't be vague identity; a and b must differ, since a, unlike b, is only vaguely the same as b
Source
report of Gareth Evans (Can there be Vague Objects? [1978], 4.7) by PG - Db (ideas)
Book Reference
Hawley,Katherine: 'How Things Persist' [OUP 2004], p.118
A Reaction
[my summary of Katherine Hawley's summary (2001:118) of Evans] Hawley considers the argument to be valid. I have grave doubts about whether b's identity with b is the sort of property needed for an application of Liebniz's Law.