Hypothesis: That's when something triggered inside of Google and it started tracking the trend (based on a build-up of volume? news? sitting in the queue for a while), and the steep initial slope is not real but a relic of the graph going from 0 to whatever their first recorded data point is.

shrug

Collatz is the last name of a high school athlete who made track and field headlines at around that time.

It will get a spike from today's XKCD, which will sadly probably be higher than the spike it receives when the thing is finally solved.

Given Lothar Collatz was German and he died in September of 1990 I'd assume something about his death (a remembrance of some kind) triggered it. Most of the searches are from Germany along with neighboring Switzerland and Denmark.

Also note someone posted a picture of him on his Wikipedia page in Sept. '09 which is the first time the page had been touched by a non-bot in 15 months (http://en.wikipedia.org/w/index.php?title=Lothar_Collatz&...)

(He's also mentioned in the Sept. 2009 issue of the European Mathematical Society Newsletter for what it's worth)

I'm not sure if I'm more interested in the explanation for this data or in how you came across it.

Today's XKCD, in part, but also because I'd been having a discussion about it with several friends, caused by a tweet to a blog post about it: http://jpcameron.com/blog/?p=47

I'd heard about the conjecture once as a Freshman in college and then thrice in the past month - hence, the Google trends search.

Prompted by today's XKCD, most likely.

Here's some info from wikipedia in case you're wondering what the XKCD strip is about: http://en.wikipedia.org/wiki/Collatz_conjecture