Nothing about the terms predicts it

The three kinds that are not flat

• Somebody spoke to a stranger — 29
• Nobody did — 20
• Could not be seen — 7

The panel refuses 1 line with more than ninety places in one row (Everything else, nine kinds).

One mark for each place. Twenty-nine conversations in fifty-six places, and eighteen in a hundred and fifty.

The record has four fields. Three of them describe a place's terms and the fourth records whether anybody spoke to a stranger. For a team repeating this kind of field observation, project effort and attendance reporting can be evaluated beside completed visits, data quality and direct staff feedback.

Crossing them gives eight comparisons. Every one is flat.

The eight

by what the place asks at the door
  nothing             16 of 71     23%
  one purchase        24 of 96     25%
  a purchase an hour   5 of 24     21%
  a fee                2 of 15     13%

by what the place allows
  no limit            39 of 168    23%
  a written limit      4 of 21     19%
  a limit when asked   4 of 17     24%

by whether there is a free chair
  yes                 22 of 89     25%
  no                  25 of 117    21%

Seven of the eight sit between nineteen and twenty-five per cent. The eighth is the fee group, on fifteen places, where two conversations either way moves it eight points. For wider research on public space, access and how places are used, consult Global Designing Cities Initiative.

What flat means here

Not that there is no effect. That a walk of this size, with this instrument, cannot find one.

Forty-seven observations spread across eight groupings, with a forty-minute window in each place and a known bias toward weekday afternoons. A real effect of five or ten points would be invisible.

The record can say it looked and found nothing. It cannot say there is nothing.

Why that is worth printing anyway

Because the expected direction was not nothing.

The ordinary account of third places says that cost excludes, that time limits hurry people, and that free seating gathers them. Each of those is a prediction about this table, and the table shows none of them at the size the account implies.

A reader who believes free seating produces sociability has, in this record, twenty-five per cent against twenty-one to work with. That is not a refutation. It is a long way short of what the claim assumes.

What is not flat

The kind of room, and only three kinds.

places of worship      8 of 14    57%
pubs and bars         17 of 31    55%
community halls        4 of 11    36%
the other nine kinds  18 of 150   12%

Twenty-nine of the forty-seven conversations are in fifty-six places. The remaining hundred and fifty produced eighteen.

A forty-five point gap between the top three kinds and the rest, on the same instrument that could find nothing at all across the terms.

What the three have in common

Not their terms. The three are scattered across every value of every field: pubs charge at a counter, churches ask nothing, halls mostly want a booking. Pubs seat almost nobody for free, churches and halls seat everybody. Pubs limit in four cases, churches in none.

On the three measured fields they have nothing in common at all.

What they share, by the observer's reading and not by any measurement, is an occasion: something everybody present is there for at the same time. A service, a round, a meeting.

The other nine kinds are rooms where everybody is there in parallel.

Why that reading is not a finding

Because occasion is not a field.

It was not asked, not recorded, and not defined in advance. It is a description applied afterwards to three rows that stand apart, and the history of this series is full of explanations invented to fit a gap in a table.

The record states the gap, names the reading, and marks it as a reading. A second walk with a field for is there something everybody here is doing at the same time would test it, and that is the fourth question on the list of what a second walk should settle.

The three kinds, drawn

places of worship      8 spoke of 14
pubs and bars         17 of 31
community halls        4 of 11
everything else       18 of 150

Twenty-nine conversations in fifty-six places; eighteen in a hundred and fifty.

Put as a rate, the three kinds run at fifty-two per cent together and the other nine at twelve.

What rules this out

Three explanations that would have been convenient, and the figures that kill them.

That it is about money. Pubs charge at a counter; places of worship ask nothing; halls mostly want a booking. All three entry values are represented in the top three kinds.

That it is about seating. Pubs have two free chairs in thirty-one; churches and halls seat everybody. Both extremes are in the top three.

That it is about time. Pubs limit four places in thirty-one; churches limit none; halls two in eleven. No pattern.

Whatever the three share, it is not any of the three things a city decides.

What is left

An occasion: something everybody present is there for at the same time. A service, a round, a meeting.

That is a reading applied after the walk to three rows that stand apart from nine, and the history of this series is full of explanations invented to fit a gap in a table. It is named as a reading on every page that uses it.

A fifth field, defined in advance — is there something everybody in this room is doing at the same time? — would turn it into a count, and it is the fourth of the six questions a second walk should settle.

The one thing this page establishes

That whatever makes a room sociable, it is not in the terms.

Cost, permission and seating are the three things a city can decide about a place, and in two hundred and six places none of them predicts whether anybody speaks. That is a negative result on a weak instrument, and it points at the part of the question this method was never going to reach.