I view it as a hierarchy.

If I only have the street, I  create a place for the street. If I then have a number, I'll create another location the specifies the number.  This usually occurs with states and towns.  I'll see that somebody was born in Maryland from say a US Census, but no idea where in that state; thus I have a location of Maryland.  When I find them having been born in Baltimore, MD from a different record I'll up the birth location to the new location of Baltimore, MD.  At least I'll have a place to start looking for more info rather than a blank location for their birth.

GRAMPS also has the ability to have historical locations.  So if the name of the city changes I can change the location to the current city name, and record the old city name as a historical info for the same physical location.


From: paul womack <pwomack@papermule.co.uk>

Second - more technical - how best to handle
locations of varying degrees of precision

For example, I have some data about 3 families
living in a single street.

But in some of the records the only location
information they give is the street name - no house number.

Similarly, Whilst I sometimes have a "full" address
for an event, other data (e.g. from FreeBMD) only
has the "containing" area.

Q2) Are there any good conventions for handling this varying precision?