To Tuple or to Map?

[Clark C . Evans]

I've got a question.  In my application I deal with many 
tables (some of them very big) where each row in the table 
has many cells.  I have a question as to how I should model 
cells.  Should they be maps or tuples?   If they are maps, 
then I'd do row['colname'] to access the value and if they 
are tuples, I'd do row[nColName] where nColName is an integer.
Since all of my rows ar relatively standard, I'm leaning
toward using tuples and integers.  But, I must admit, using
a bunch of maps looks a bit cleaner...  So, here is the question:

(a) Does a map have significantly more run-time
    overhead than a tuple?

(b) Is there any way to mark a variable (nColName)
    as a constant so that the "byte compiler" can
    resolve the value to an integer (instead of
    requiring an access to the local namespace)

(c) Which technique would you use?

Thanks!

Clark