How many people could fit in ancient public buildings? A lit review and some problems

I’ve been revising an article manuscript based on my thesis work looking at a way of modeling the number of people that might have occupied the earliest non-dwelling buildings in the Near East.  This means that I have been reviewing the literature on the amount of space required per person in houses. I struggle with this issue as the amount of space calculated to be required per person are based on cross cultural comparisons of household size and dwelling size, which becomes a problem when translating that to a non-domestic context. People don’t occupy public and social spaces in the same way they do houses, and any analysis of these public spaces do not make much sense if you use information about dwelling spaces.

Most of this literature on the calculation of the amount of space required by each person has been based on cross-cultural analysis, looking at ethnographic accounts to determine the size of houses and number of people living in them.  This started with Raoul Naroll (1962), who calculated that people needed 10.2 m² each, based on the amount of roofed space and household size.  His research lead to a flurry of refinements, criticisms, and tweaks, which are capably summarized by Brown (1987).  Irritatingly, the most problematic critique raised completely escaped me due to my lack of understanding of statistical analysis.

Now, Naroll in his 1962 article determined that house floor area and household size varied allometrically.  As I had no idea what allometric meant, I continued reading, as is my wont, assuming I would get the gist of it at the end.  I assumed that as Naroll had used the term, that he would then effectively account for it’s significance.

Apparently not.

Allometric, and it’s alternative isometric, refer to ways in which two variables might relate to one another.  For example, variables that vary isometrically, vary in proportion to one another.  So, if one variable becomes twice as big, the other variable grows to twice as big as well.  Alternatively, variables that are allometrically related do not have a proportional relationship.  So, if one grows to twice as big, the other might grow to 8 times, or 15, or 75 times as big (though only one of those, and the relationship is stable and predictable, just not proportional).

So, what Naroll determined was that as settlements became larger, gradually less space would be allocated to an individual person.  However, when it came time to put together his conclusions, he chose a constant that was at the extreme edge of of his data without emphasizing the fact that it would change based on settlement size.

Not as simple as 10.2 m² per person then.

The other problem people have discussed with Naroll and subsequent researcher’s work is the fact that it is difficult to reconcile the fact that they only used roofed space with the concept of dwelling space.  Some communities spend much of their time outside, and this raises the question as to whether only analysing roofed space is really relevant in these communities.  On the flip side, some communities have all sorts of non-dwelling activities that take place indoors, such as storage and animal penning.  Would this space really help us determine how many people were living there as well?  And of course, almost all of these studies immediately remove non-dwelling structures from their statistical analyses – exactly the type of spaces I am interested in as public spaces!

All of these problems cause a bit of a headache, but most scholars come to the conclusion that if the requirement for roofed protection from the elements is a cross-cultural need for all humans, then it should be possible to determine some ratio between the amount of roofed space and the number of people using it.  More sophisticated statistical analysis seems to show that in general, people require about 6m² each.  In fact, Brown (1987) showed that if you fix some silly mistakes on Naroll’s part (recording feet as meters, and his statistical…oversights) then his results would have been much closer to 6m² than 10.2 m².

Interestingly, recent refinement of this work has shown that if you control for settlement type, degree of mobility, and even marital residence pattern, then some telling patterns emerge.

The variation within the average house floor area (not the average space per person, but total house area) seems to be correlated with whether a society was matrilocal or patrilocal (Porcic 2010).  On average, communities with larger houses tended to be matrilocal, while communities with smaller houses tended to be patrilocal.  While it is difficult to predict whether a community was matrilocal or patrilocal based solely on house size (particularly difficult for mobile communities), it is interesting to speculate why matrilocal communities would have larger houses than patrilocal.  Some scholars have suggested that matrilocal communities tend to have larger houses because adult sisters live together cooperatively after the marriage under the same roof as they have grown up sharing work.

Additionally, it appears that more mobile communities allocate less space per person than more sedentary communities (Porcic 2012).  This means that if we know the residence strategy of a particular community, we can use a more refined ratio of house floor area to household size and calculate a more nuanced population estimate.

What does this mean for public and non-dwelling buildings then?  Public buildings were clearly built to hold people – menstrual huts, temples, meeting rooms, sweat lodges, sun shades, and any of the other myriad of meeting spaces were built by various cultures to hold people performing non-dwelling activities.  But how do we estimate how many?  As these spaces are non-dwelling, we can’t base this off a basic human need for shelter, so where do we go?

I could find almost no literature on this.

Trust me, I’ve looked.

Nobody seems interested in reporting how many people attended events in public buildings ethnographically (if you know of some, please let me know).  And nobody seems to know how to handle it archaeologically.

My research uses AutoCAD modelling to explore various scenarios of how people might have occupied public structures, but this fits poorly with the rest of the literature.

Modern public buildings also have little description of how they determined how many people could occupy them.  Sacred Destinations says that St. Peter’s Basilica in Vatican city covers 23 000 m², and could hold 60 000 people.  That’s only 0.38 m² per person!  Big difference from the dwelling ratios found in the cross-cultural research.

Anyway, I will let you know if my article ever makes it past peer review and you can see how I have tackled this sticky problem of dealing with non-domestic and public spaces!

Further Reading

Naroll, R. (1962). “Floor Area and Settlement Population.” American Antiquity 27(4): 587-589.

Brown, B. M. (1987). “Population Estimation From Floor Area: a Restudy of “Naroll’s Constant”.” Cross-Cultural Research 21: 1-49.

Porcic, M. (2010). “House Floor Area as a Correlate of Marital Residence Pattern: A Logistic Regression Approach.” Cross-Cultural Research 44(4): 405-424.

Porcic, M. (2012). “Effects of Residential Mobility on the Ratio of Average House Floor Area to Average Household Size: Implications for Demographic Reconstructions in Archaeology.” Cross-Cultural Research 46(1): 72-86.