Whatever their faith, historiansof religion must be professional atheists, argues Patrick Collinson. Around 1600, in London's already sprawling suburb of Southwark, virtually the entire population of many thousands received communion according to the rite of the Book of Common Prayer, once a year, round about Easter. (The few exceptions included a poet and dramatist called William Shakespeare.) Why was nearly everyone there? What did they suppose they were doing?

Max Weber, the German social scientist, thought that "the modern man is in general, with the best will, unable to give religious ideas a significance for culture and national character which they deserve".

Unless we are stupidly arrogant, that inability stays with us. And yet, paradoxically, the modern historian is disposed to give more weight to religion as a motive historical factor than ever before. In the world beyond academe, religious belief declines. But interest in the subject, measured by book sales and media ratings, increases. Religion too has become a spectator sport.

But how is the modern historian to recognise and define religion, this factor as often as not alien to his experience? An African peasant is repeating a ritual form of words as he sows his seed. Does he suppose that if he fails to sing this refrain the crops will not grow? So is this a "religious" (or magical) act? Or is his song like a sea shanty, a custom and a solace? When the ancient Aegean people poured liquid food into the mouths of their dead, was that a religious act, different from what little girls do when they feed their dolls? There is a particular problem for the religious historian of early modern Europe or of any society experiencing cultural change. We cannot assume that everyone living at the same time shared the same mind-set. In such circumstances there may exist a detritus of vestigial, or at least unarticulated belief, the bits and pieces left behind by a monster snowball when it melts.

ÖйúAƬ

These are the questions about religion, and there are no others. What is it? What does it consist of? Where does it come from? What does it do?

Here is my own answer to what, considered as a question in social history, religion does. Religion acts as a precipitant, provoking action by individuals and groups which would not have been taken without it. Religion serves as a bond, not merely affirming and strengthening social groups as diverse as churches, sects and nations, but actually inventing, "imagining" them. And religion provides legitimation, licensing as just causes, conditions and acts (sometimes atrocious acts) which would otherwise be deemed unlawful.

ÖйúAƬ

All this begs the question what religion is, which most historians duck as requiring no generalised answer within the particularities of the various small universes, "my period", which they inhabit. The wider world of social science has provided only two answers. Either religion is a belief in divine beings, which is to say, a commitment to faith and participation to activities conventionally deemed to be religious; or it is a glimpse of a reality beyond reality, the ultimate sacredness of things.

According to the first definition, only churches, in the broadest sense, are religious bodies. According to the second, a Nazi rally at Nuremberg, or a football match, or a rave, are essentially religious, or at least quasi-religious, events, which gives the religious historian rather more scope than the historian of churches, the conventional ecclesiastical historian.

Where religion comes from and what it does constitute in reality but one question - origins determine function, or vice versa, at least for the 19th century, which defined things with respect to their origins. To be sure religion, every religion, has its own answers to such questions, devout answers, requiring faith. The social historian of religion, whatever his personal faith, must suspend belief (and disbelief) and subscribe to the formal, methodological atheism premised by the founding fathers of social sciences for whom religion mattered - Karl Marx, Emile Durkheim and Max Weber. They believed different things about the matter but all assumed that religion was an understandable mistake.

For Marx, religion was a false consciousness, but one which he appreciated as one might appreciate a great symphony: not merely an opiate but "the sigh of the afflicted creature, the soul of a soulless world . . . the spirit of spiritless conditions". The Marxist account of religion will not do, not because it explains religion as a social contrivance but because it understands "social" as merely a material system of distributive economics. There are many other senses in which we are social beings.

ÖйúAƬ

Durkheim will not do either. For Durkheim, religion is society affirming itself: "Belief and practice which unites in a single moral community called a Church all those who adhere to them." The object of religion is society. But this is tautology, and a false tautology, since religion and society appear to be the same thing by different names. And often they are not.

Weber, for all that he has been disregarded by some historians, points a way through the ice floes. Religion is a powerful, transcendent idea or "ethic" originating in circumstances which are inaccessible to the social scientist. The prophet and his message cannot be explained, still less explained away. But once they enter the materiality of human affairs, prophets and prophetic messages interact with society in ways which are always intricate, and are never determined absolutely by either material or spiritual forces. Human history is a railroad, but a railroad with points, and those points are ideas.

Since Weber was but one atom in the universe of human endeavour, and inclined by his own particular background and formation of Wilhelmine Germany to emphasise the individual and the ideal, he has left us not with assent but with much dissent, and with a debate which outlived him by 70 years. But no more serviceable tool for the understanding of religion in society has ever been devised than Weber's tool of "elective affinity". People, for reasons which may be unknowable, but which deserve to be investigated, select and make use of elements of a religious idea, the genesis of which may also be an unknown. But out of this comes something like a natural law. "Material without ideal interests are empty, but ideals without material interests are impotent. " Patrick Collinson is regius profesor of modern history, Trinity College, Cambridge.