Dear Jer:

Your query has two feature that I question:

1.  When you associate two rows on OLD VALUE = NEW VALUE, you presume these
values NEVER REPEAT.  Could they?  Is this a record of changing channels on
a TV?  I can see where this association may not work.

2.  Does the table represent more than one set of sequences, perhaps
identified in the Name column?  What we would call "Groups" in database
terminology, which are to be treated as independent subsets?

What I'm going to suggest is that the sequence in the Date/Time column is
the only true way to find the "previous" row for this comparison.  Now, this
could cause sequencing problems.  Is this value recorded only to the nearest
minute, or is your example just formatted to show only that much resolution?

Specifically, it would be most workable to have Name and Date/Time together
be unique.  As an alternative, we can try to sort this out by also filtering
where the New Value of the previous row is equal to the Old Value of the
subsequent row.  While this may help, it isn't foolproof.  Consider this:

NAME    Time                    OLD VALUE    NEW VALUE    No
W53       3/20/2006 10:45  Station1             Station2               1
W53       3/20/2006 10:45  Station2             Station1               2
W53       3/20/2006 10:45  Station1             Station2               3

Now, in the above, does the No column uniquely number the rows for a given
Name?

Assuming this is the case, here's a trial query:

SELECT *
 FROM Log L, Log L1
 WHERE L1.[Name] = L.[Name]
   AND L1.No = L.No - 1

If the No column is per Name and increments by one every time, then this may
do it.  This gives you a chance to look at it.

I can code to find the previous row by other methods, depending on the
characteristics of your data.  I do not know these characteristics.  I'll
need your help to understand them.

Tom Ellison