The birthdate tool works out when someone might have been born, given their age.

That doesn’t sound very hard, but turns out to be very tricky. If you find a bug, please leave a post here telling me what dates have gone wrong, and I’ll be very grateful.

Sometime in in March 2004, I saw the new baby. “Goodness, she’s so beautiful! How old is she?”, I said.

“She’s two months old”, came the reply.

OK, when was the baby born?

I know I asked the question in March 2004. So, on some day between 1 Mar 2004 and 31 Mar 2004, the baby was 2 months old.

If I had asked the question on 31 Mar 2004, the baby must have been born on or before 31 Jan 2004. If she’d been born on 1 Feb 2004, she’d not be 2 months old on 31 Mar. She’d be “Very nearly 2 months”, or something like that. So the latest possible birthdate for her is 31 Jan 2004.

If I had asked the question on 1 Mar 2004, the earliest she could have been born is 2 Dec 2003. If she’d been born on 1 Dec 2003, she’d have been 3 months old, not 2, on 1 Mar 2004.

So, there are two uncertainties, a vague age – “two months”, not “two months and 4 days” – and a vague date we asked the question, sometime in March. These two uncertainties combine to produce a range of possible birthdates, from earliest to latest. So the program produces two dates, an earliest possible birthdate, and a latest possible birthdate.

All this is of more than academic interest. I researched my family tree some while back, and was forever adding up ages and dates. I wrote a not-very-good spreadsheet to help. This web tool is intended to replace my spreadsheet, to be a more precise and general tool, and to be useful to others.

Much more on this later! Meanwhile, try out the tool at http://www.eversholt.org.uk/birthdate