The infant in its nocturnal habitat

A baby stirs in his sleep, his hand seeking his mother who lies next to him. Still asleep herself, she calms his batting motion by patting him gently. Another night, another mother and baby. They have dozed off to the sound of the World Service on the radio. Each time "God Save the Queen" is played, at midnight and at dawn, the baby changes sleep state, moving in and out of his dreams to the musical cue.

These are scenes from a sleep lab, a new British project to research the interactions between mothers and babies sleeping together and apart. The Sebastian Diamond Mother and Baby Sleep Laboratory, named after Anne Diamond's baby who died of sudden infant death syndrome (SIDS), is a unique investigation into the way babies really sleep. Officially opened by Anne herself last month, the lab is part of the Foundation for the Study of Infant Deaths research unit in Bristol, and its work may eventually shed light on Britain's annual toll of around 500 "cot deaths" - still the largest single cause of infant fatality.

Professor Peter Fleming, the consultant paediatrician who heads the unit, was responsible for the research that led to Britain's Back to Sleep campaign of 1991 which advised parents to lay babies in the supine position or on their sides. But while this advice has been phenomenally successful (previously cot death claimed between 1,500 and 2,000 infant lives every year) controversy rages about another possible factor contributing to cot death.

The question is: should the baby be allowed to sleep in the parents' bed? In November 1991, headlines told us this was a Bad Thing. Researchers from New Zealand had not only discovered that babies were safer on their backs, but had investigated the Maori population where many babies died in the family bed. The New Zealand government was already issuing public health warnings not to take baby into bed - it was presumed that overheating was the problem. Unfortunately, the advice was premature. When the New Zealand statistics were corrected for other factors, it was found that co-sleeping was only a problem in combination with smoking and alcohol consumption.

In a much less publicised paper in 1993, the New Zealand researchers came back to say: "... the ethnically specific relative risks associated with this factor do not appear to be homogenous," which is academia-speak for "don't draw any cross-cultural conclusions from our statistics".

Nevertheless, public health policy in New Zealand has not changed, and over here the Foundation for the Study of Infant Deaths also advises extreme caution. They are concerned about the risk of babies becoming too hot under an adult duvet and strongly recommend that parents keep baby close at night, but in a separate cot. Yet, according to Professor Fleming, there is no justification for offering advice either way.

"New Zealand researchers are doing scientific gymnastics in order to justify an earlier over-simplification of their data," he says. "I find it quite extraordinary. For 20 years we made the mistake of giving advice to lay babies on their tummies - we drew conclusions from studies of pre-term babies and we were wrong to do so. We killed thousands of babies with that advice, and if the medical establishment can't learn from a catastrophe of that proportion, I don't know when we will learn.

"We do not recommend that parents sleep with their babies, but equally we do not recommend that they should not do so. We do not yet have clear evidence either way. Where we do not have good science upon which to base our advice, we should be honest and accept that parents' instincts and 'what feels right' may be a good substitute."

Professor Fleming is in the process of analysing some important data on cot death, due for release in the new year. The Confidential Enquiry into Still-births and Deaths in Infancy (CESDI) has for two years been gathering information from 40 per cent of the British population. Bereaved families had to answer 600 questions and four control families did the same.

This, the largest population study ever into cot death, is expected to show that babies do die in their parents' beds as well as in cots, but Professor Fleming questions whether the bed-sharing arrangement itself is to blame. He suggests that contributory factors such as smoking, drinking, poverty and a careless attitude are sometimes connected with the absence of a cot. Such practices may have little to do with co-sleeping as practised in a healthy, safety-conscious household.

Moreover, the Maoris appear to be an exception to the epidemiological rule. Studies from populations all over the world demonstrate that where babies routinely sleep alongside their mothers, the syndrome of cot death is either very rare or non-existent. Asians and Bangladeshis in Britain, the Hong Kong Chinese, black and Latino populations in America, and the Japanese, all demonstrate a curious lack of SIDS.

"We have realised that you have to rewrite the whole of our understanding of infants' sleep," says Professor Fleming, "taking into account where they really sleep and the interactions they have in the night. We wouldn't dream of trying to study monkeys by putting them in a cage, so why have we only considered babies sleeping alone in a cot? Our intention is to look at infants in the wild."

This field of research is too young yet to allow video cameras into real bedrooms, so for the moment the "wild" is a small room at St Michael's Hospital in Bristol, containing a double bed with a flowery duvet and pretty curtains. There are copies of Cosmopolitan, a radio and a wash basin. On the wall is a painting by an Australian artist, depicting the flowers which lay on his grandson's coffin after cot death.

Volunteer babies are connected to an umbilical cord of wires, which measures temperature, brainwaves, breathing and heartbeat. Mothers are allowed complete mobility, so they can care for their babies as they would at home. One night the baby will sleep with them, the next night in a cot nearby.

"We are trying to discover what happens in a normal bedroom at night," says clinical research fellow Dr Andrew Sawczenko. "How babies communicate their needs, how mothers respond, how often babies wake up and why. Is there a sixth sense going on? From our preliminary studies, it seems the mother is in tune with her infant."

In one video highlight, a mother responds to a baby's barely audible squeak by giving him more space and fanning him with her bedclothes. She drops instantly back to sleep. "When we checked the data, we found that the baby was getting very hot and even we didn't know," says Dr Sawczenko. "The mother's actions cooled him down straight away."

This reminds me of an astounding moment with my first baby when I woke to find myself holding her out of the bed to be sick. Somehow, in my sleep, I had sensed the almost-imperceptible heaving of her stomach and acted automatically. Not all mothers are this responsive, however. "One baby lay crying next to mum and mum was totally blotto," says Dr Sawczenko. "The baby's crying went on for 20 minutes."

Noreen Hart and her 12-week-old baby, Fintan, are two of the volunteers helping to cast new light on the ancient art of co-sleeping (this is, of course, the context in which the human species has evolved). They will visit the sleep lab for two nights every month until Fintan is five months old, and attempt to provide a living model of mother-baby interactions. Noreen has four boys, but did not consider co-sleeping with the first two.

"I was more relaxed about my two younger children," she says. "I realised it was so ridiculous getting up in the night to feed the baby in a cold chair, sitting there freezing at 3am. For me, sleeping with the baby is very pleasurable. My husband is happy about it, too. In the morning, Fintan wakes and looks at us with a gummy smile."

It is unlikely the Bristol research will be able to offer a categoric answer to the question of the family bed. Clearly, the practice is not going to suit everyone, but good information, properly researched, is desperately needed to stop statisticians taking pot-shots at caring, careful parents. Already, it seems clear that bedsharing with a baby should be avoided if parents smoke or drink or take drugs. What we can now hope for are some fascinating insights into the nocturnal world of the newborn human infant, in his natural habitat.

"We have so little understanding and yet so many pompous statements on the subject," says Professor Fleming. "We have to admit how little we really know and start from there."

The Sebastian Diamond Mother and Baby Sleep Laboratory is an ongoing project which is in need of more funding and volunteers. If you are able to help, contact research nurse Jeanine Young on 0117 928 5676.

Deborah Jackson is the author of 'Three in a Bed - Why You Should Sleep With Your Baby' (Bloomsbury, pounds 5.99).