> > A cleaner version is:
> > 
> >   marc.245$a = "Smith"
> >     prox/unit=element/distance=0
> >   marc.245i1 = "4"
> > 

I think that probably is the right way of doing it but not the most
obvious of syntaxes. I wouldn't say "cleaner" as I think the relational
modifier approach whilst it might be "wrong" in the CQL sense, is
definitely "cleaner" on the eye (and slightly more intuitive).

The other problem with the above is that someone (e.g. me) is going to
ask what other values of unit and distance are permisable in the above
and what they mean

E.g. prox/unit=sentence/distance=0, prox/unit=paragraph/distance=0

Do we for instance interpret paragraph to mean within any 400 field if
400 is repeatable

Indeed is element the right unit here (to me I'd argue that element
meant subfields whereas sentence meant the whole field?

proc/unit=element/distance=1

Would distance in this case mean nearby fields within the record as
present in the database (in which the fields might not be ordered
sequentially) or within fields with closeby MARC tag names? Is there a
case for being able to do either?

Matthew

(I strongly disagree with Mike's earlier comment that a MARC context set
is trivial!)