Male adoptees are using consumer DNA tests to predict the surnames carried by their biological fathers, the BBC has learned.
They are using the fact that men who share a surname sometimes have genetic likenesses too.
By searching DNA databases for other males with genetic markers matching their own, adoptees can check if these men also share a last name.
This can provide the likely surname of an adoptee's biological father.
The genetic similarities between men who share surnames occur on the Y chromosome, a package of genetic material passed on, more or less unchanged, from father to son - just like a last name.
Because of this pattern of inheritance, men with the same surname may also share a similar complement of genetic markers on the Y chromosome.
At least 30 men registered with US consumer genetic testing firm Family Tree DNA have found their "biological surname" in this way, the company's chief executive told BBC News. The company has an online database called Ysearch containing genetic information from 125,000 men, along with surnames and other genealogical data.
Bennett Greenspan explained: "We now have a growing number of people who are adopted, who have tested with us and have matched several individuals with a particular surname, and maybe they haven't matched anyone else with a different surname.
"From that, they can get the idea that they have at least found the surname they need to start looking for in the town in which they were born."
The tests can "read" up to 67 genetic markers on the Y chromosome. Mr Greenspan said that, for some adoptees, discovering the surname of their birth father in any other way might be extremely difficult, or even impossible.
"That's the real miracle of the DNA test. [The Y chromosome] can act in a sense like a silver bullet." he said.
A little light
Mark Jobling, professor of genetics at the University of Leicester, UK, who is unconnected with Family Tree DNA said: "If you have a surname which is reasonably rare, but not so rare that the chances of another person being typed and going into that database are infinitesimal, then you could be in luck.
"There's a big gamble in doing it, but people sometimes say that if you're in a dark room then even a little light can be useful.
Chandler Barber, a 37-year-old advertising copywriter from Dallas, who was adopted at birth, said he had learned about the possibility of discovering his surname from a magazine article about consumer DNA testing.
Of six people in the Ysearch database who were close genetic matches, all had variants of the surname Ritchie, including one US-based Ruetschi who was a very close match.
"It was pretty concrete evidence," Mr Barber told me.
"It's a quick and effortless way to at least find some nugget about your history. I am sure there are people who have been searching for their birth parents on foot, with pen and paper, for years - and have got nowhere.
"You start to wonder to yourself - if I do this, am I letting my family down? I told my mother: I really don't want to find my birth family. I just want to know where I'm from. But she told me that she had expected me to do this a long time ago."
Edward Cerullo, 48, a computer programmer from Norway, knew his birth father's surname - Page - before testing his DNA.
"When the results came back, of the 22 names they sent back who matched my DNA 11 were Page or Paige. That's statistically pretty hard to argue against," he explained.
The database allowed him to see how his own line of descent fits into the wider family tree for this surname.
The link between last name and likeness on the Y chromosome gets stronger, the rarer the surname is. But, said Mark Jobling: "Even in reasonably common surnames you see 'descent clusters'.
"In a name like Jefferson, for example, which is quite a common name, you find lots of these little descent clusters. There is identity within those clusters but there are many of them.
"In a name like Attenborough, there is just one great descent cluster, and a few people who don't fit into it. There's a spectacular common ancestry for that name."
But he cautioned that these general patterns might differ from country to country. Also, some rare markers ran across two or more surnames, which might cause false matches.
Such false matches might also arise from technical issues with DNA databases. For instance, genetic testing companies sometimes used different naming methods for genetic markers. Confusion might arise when customers whose DNA had been tested by different companies uploaded their own genetic information into the same database.
Mark Jobling said tests offering better resolution on the whole genome should be able to solve other familial puzzles. In the first half of the 20th Century, when a child was born out of wedlock, grandparents would sometimes raise the child as their own.
Professor Jobling said he knew of one man who suspected this had been the situation with his own immediate family. An "older sister", this individual believed, had actually been his mother. Unfortunately, the putative sister and parents were now deceased.
"If there is another relative, such as an acknowledged grandchild of this grandparental/parental couple, you can set up a hypothesis whereby you say: 'if they were his parents, how much of his DNA should he share with this cousin'," the University of Leicester geneticist explained.
"If they were his grandparents, he should share a certain lesser proportion of his DNA with his cousin. You can distinguish the two scenarios."
Professor Jobling said that falling costs of sequencing entire genomes offered the promise of finding genetic variants that were specific to one surname - with no room for ambiguity.