Last time I did something of this sort, I added up the number of businesses and multiplied that by four. That was the number of city residents living in the same building as their shop, figuring an average household of four people. Subtracting that from the total population gives the approximate number of people who still need quarters. If your population includes military, you can also subtract those, as they'll be housed in a barracks.
The number of buildings required to house the remaining population depends on average family size, number of families living in a single building, and number of generations living in a single household. Our modern Western idea of a nuclear family is relatively uncommon, so I usually figure on at least eight people per structure.
Size of structures depends greatly on the cost of building materials and the amount of room the community has to spread out in. A walled town will be fairly dense because the wall is expensive to build, maintain and defend. Buildings made of stone will be small because they're expensive. A town near readily available forest with no external threats will have larger dwellings, and they may be spaced further apart, although not so far as to make walking across the town onerous.
I generally base the size of a structure on the length of a bed for an average-sized inhabitant. I build out like you described: inside to outside, until I have a base line for the smallest residence I'd expect to see, then I make most of the buildings bigger than that. I don't know how accurate my end result really is, but it's usually good enough that nobody calls me on it. Unless I make a big measurement error, of course.
And, of course, nothing beats reference. Look around for town and city maps drafted by John Speed. He mapped a great deal of early Renaissance England, and I've found him to be very useful for sanity checking.
Reply With Quote