Who dares (to get married) wins

What does 'family' mean to you? It depends whether you think of SAS 'widow' Anna Homsi as a 'Ms' or 'Miss'

'Family" sounds like a safe, comforting haven. But it's become less and less clear-cut what it means, and where it's located on the modern social map. These past few days, we've seen a classic example. The Ministry of Defence is up the creek of family law, and paddling. Anna Homsi has threatened to sue because she wasn't granted a war pension after the death in Sierra Leone last year of her child's SAS father, Brad Tinnion, with whom she lived for seven years, but to whom she was not married. The ministry gave the daughter £2,000 a year; the mother got a one-off £20,000. But the rules – which the Defence Secretary is under pressure to review – said "no war pension".

A fog is beginning to enwrap that easy old word, family. In the liberal press, Anna Homsi was Brad Tinnion's "partner" and was called "Ms". In the conservative press, she was his "girl-friend" and remained firmly "Miss". In his book The Transformation of Intimacy the sociologist Anthony Giddens – director of the London School of Economics and head guru of Tony Blair's Third Way – said that "marriage in the traditional sense is disappearing".

But the blunt fact is that it isn't. As you read this, about 80 per cent of children in Britain are living in a family headed by a couple, who are usually married to each other. You might not always think so, from the way debate recurrently focuses on unmarried or lone parents. But much public policy is still geared to this huge preponderance of marriage and family "in the traditional sense", as Anna Homsi has been discovering. If, for example, a couple live together unmarried, and one of them dies, the survivor isn't entitled to the state bereavement benefits which might help with child care.

In interviews, couples who are just living together – especially, perhaps, the women – often say there's "no difference" whether you are married or not.

At one end of the income scale, this may be due to the perverse effect of official rules. If the Department of Social Security decides you're cohabiting, you'll be treated as a quasi-married couple, getting less benefit than if you were two single people. But it does not follow that to cohabit gives you anything like the range of duties, protections and problems of marriage.

As in the Homsi case, the awkward truths usually emerge when it's too late – when the couple are separated by death or disputation. Most cohabitations are shorter-lived than marriages.

In a recent national study, Jonathan Gershuny and Richard Berthoud, of the University of Essex, found that, out of every 20 cohabiting couples, 11 eventually marry each other (often when a child is wanted, due or born), eight split up without marrying, and only one couple will still be living together after 10 years without marrying.

Under the law, the mother has most of the powers over the children. An unmarried father may find that a casualty ward won't accept his word if a surgeon needs legal authority for a child's urgent operation. If the couple separate, he may have no say if the mother decides to put the child up for adoption. On the other hand, he does remain liable for paying child support. The penalties for the woman, however, are often financial.

There are some slow, slight changes. An Adoption and Children Bill is expected to grant extra rights if the father's name is put on the birth certificate. At present, this is mainly a psychological and statistical statement, with few legal consequences.

One of the startling things about cohabitation is how little its precise implications are known. The law itself is a blur. There's no legal definition of cohabitation. The DSS relies on its bureaucratic rules of thumb, backed up by neighbourly nosiness.

About 10 per cent of the adult population cohabits at any one time; but only one in 20 of these is living together long-term. If this is a deliberate, thought-through choice, you could argue that the penalties have been knowingly traded off against other perceived gains. Then, so be it.

Often, however, the knowledge isn't there. When the implications come home to roost, it's often not only too late but also one-sided. The bereft man or woman wants the back-dated benefits of marriage, but not the obligations.

It's tempting to suggest it would be better to get people in the mood to marry, rather than to keep trying to chip away at an already fragmented law, and end up with more and more anomalies. At the very least, there should be a much wider spread of information about the real differences between marriage and non-marriage. These include the awkward truth that, on average, the children of the married fare better. All that matters, in the end, is the children. On this, both the married and the unmarried can surely agree.

Paul Barker is a senior research fellow at the Institute of Community Studies