Note, I did not take splits manually. I just hand picked points nearest to the start line. Thats why they are so nicely close together. But timing may not be as accurate.
There is tall building close the track Ross were running. That may have had some effect. But I must admit my logger is usually very accurate in open terrain. The problem is orienteering is done in forest and there my forerunner (and globalsat unit) is better.
What it comes to forerunner and hand swinging, I dount think the swinging is the problem. More likely it is the fact your body shadow is blocking different satellites depending on what part of the track you are running. For example you are runinng 180 degrees other direction on straights, so your body is blocking satellites at opposite sides. Thats why device end up using different satellites for position calculations. And that means the error may be to opposite direction, so gps end up drifting out and back all over again. If held on head the device will see same satellites all the time, so it will end up drifting a lot less. On shoulder there is head's shadow but that may not effect that much. And in open terrain holux's chipset is not as sensitive to that problem, like I said it keeps using all possible satellites longer than sirf III (even if it sometimes should not). Holux has the same problem (but worse) in forest.
take a look at this:
West part of the east X. It looks like I would have used two paths, on for running out and an other for coming back. Both have same effect, but Fore is better. I thnk this is because I turn around 180 degrees and my body shadows different satellites. But I haven't confirmed this with cap gps setup.
Note also, holux's tracks are close together on the open hilltop, Fore is not. Usually Holux is very accurete in open land, like now on track. It looks like Forerunner is almost as accurate in forest than in open terrain.