Discussion: Could be...

accuracy of the satellites acquired?

More data needed. Do it again. :-)

OK, since you commented, I plotted the GPS distances of this same route over the past year (click to enlarge). The data are a bit noisy, but there seems to be a trend.

I got this watch in February. Before I started using it, I had traced the route in gmap-pedometer.com, and it was a little over 4 miles. The first few runs were less than a tenth of a mile short, so OK. But lately they've been 0.2 mi short, and I can't really understand that.

I've had terrible luck with GPS watches. This is the 3rd one I've owned. The first two were Garmins, and they were both awful. The second one was particularly frustrating, because it would take 15 minutes to locate satellites, and then time out and lose them while I was walking to the start of an event.

I gave up on the whole idea of a GPS watch for years. I got this TomTom Runner 2 for $50 earlier this year through a tip on an AP thread. It's been great so far, instantly locks in, easy controls, easy download. I liked it so much I bought Florence a Runner 3 (more expensive but more capable). She's not using it much so maybe I should borrow it and compare. If this distance trend is real, it's strange and troubling. But it's possible it's just insufficient data.

Okay, maybe it *is* the watch.

I found this site with watch gps accuracy analyses, which doesn't give me a lot of confidence in my (ordered but yet to arrive) Garmin 235. Hmmmm.

I also took a look at the tracks on your last few runs on this route, and the track cuts a few more corners than typical, so that would make a difference too.

Good point about the corner-cutting. I hadn't looked that closely. That may be a sufficient explanation. I didn't cut those corners, but the GPS evidently thought I did. Here's a case where the GPS was obviously off.