To test whether religion might have emerged as a way of improving group co-operation while reducing the need to keep an eye out for free-riders, Dr Sosis drew on a catalogue of 19th-century American communes published in 1988 by Yaacov Oved of Tel Aviv University. Dr Sosis picked 200 of these for his analysis; 88 were religious and 112 were secular. Dr Oved's data include the span of each commune's existence and Dr Sosis found that communes whose ideology was secular were up to four times as likely as religious ones to dissolve in any given year.

A follow-up study that Dr Sosis conducted in collaboration with Eric Bressler of McMaster University in Canada focused on 83 of these communes (30 religious, 53 secular) to see if the amount of time they survived correlated with the strictures and expectations they imposed on the behaviour of their members. The two researchers examined things like food consumption, attitudes to material possessions, rules about communication, rituals and taboos, and rules about marriage and sexual relationships.

As they expected, they found that the more constraints a religious commune placed on its members, the longer it lasted (one is still going, at the grand old age of 149). But the same did not hold true of secular communes, where the oldest was 40. Dr Sosis therefore concludes that ritual constraints are not by themselves enough to sustain co-operation in a community-what is needed in addition is a belief that those constraints are sanctified.

Dr Sosis has also studied modern secular and religious kibbutzim in Israel. Because a kibbutz, by its nature, depends on group co-operation, the principal difference between the two is the use of religious ritual. Within religious communities, men are expected to pray three times daily in groups of at least ten, while women are not. It should, therefore, be possible to observe whether group rituals do improve co-operation, based on the behaviour of men and women.

To do so, Dr Sosis teamed up with Bradley Ruffle, an economist at Ben-Gurion University, in Israel. They devised a game to be played by two members of a kibbutz. This was a variant of what is known to economists as the common-pool-resource dilemma, which involves two people trying to divide a pot of money without knowing how much the other is asking for. In the version of the game devised by Dr Sosis and Dr Ruffle, each participant was told that there was an envelope with 100 shekels in it (between 1/6th and 1/8th of normal monthly income). Both players could request money from the envelope, but if the sum of their requests exceeded its contents, neither got any cash. If, however, their request equalled, or was less than, the 100 shekels, not only did they keep the money, but the amount left was increased by 50% and split between them.

Dr Sosis and Dr Ruffle picked the common-pool-resource dilemma because the communal lives of kibbutz members mean they often face similar dilemmas over things such as communal food, power and cars. The researchers' hypothesis was that in religious kibbutzim men would be better collaborators (and thus would take less) than women, while in secular kibbutzim men and women would take about the same. And that was exactly what happened.