[Benjamin Kalish]

Yes, it’s not an easy problem! Perhaps an approach like follows might make sense:

• Create new fields (to store these values)
• For all existing records, and for each new record, as it is entered, do some text analysis to populate the extent fields whenever the physical description fits one of a number of predictable patterns
• allow library staff to update these fields in case the automatic procedure can’t identify the extent or gets it wrong

I think that automatic procedure could be made to get it right in a lot of cases. And I am certainly fine with the information being missing from some records. Having it on most records would be more than worth it.