OWS GPS Accuracy

Not that I swim with my 945, but I know that swimming is hard since the watch picks up the GPS in the short time the hand is in the air before it goes back into the water. To rule out that your swimming techniques are very different, have you tried swapping watches to see what results you will get?

Another thought - since the times are different, you haven't enabled auto-pause for her watch?

I don't think that a 100 yard difference is a big deviation, but that's just my opinion.

I read your formulation that you set the older watch, the 920, as the reference.
So the difference between the 920 and the 945 is only 96 yards or 3.6%. I would say, 96 yards (88m) is ~3.5 laps in a 25m pool. 
I don't get how you can get to 200 yards. I also don't understand the problem with 3.6%.

You don't tell us wich stroke you use, but I assume freestyle. half of that time the watch was under water. What do you expect?
Distance is only calculated from GPS signal wich is known for its inability to travel through water.

If you need it more precise, do it like DC Rainmaker and put the watch an a buoy you pull behind yourself. But I don't know if that's worth the effort for 96 yards of 2650 yards.

Actually, the 945 is possibly the right one.

My triathlon coach and one training partner have the 920.
Me and others training partners have the 935 and 945.

We do a lot of OWS together, at least one time a week during the entire year.
After weeks of complaining about the accuracy of the 945, I find out that the older watches don't catch the round portions of the path properly.
When we do point-to-point swimmings, which are virtually in a straight line, the distance and the pace between 920, 935, and 945 are almost spot on.

Per your screenshot, the first auto lap is almost identical in both maps.
Then, the second one shows a very sharp line going into the land.
After that, the laps differ between the two maps.
Between laps 3 and 4, there is another sharp line.

Also, even if you and your wife are swimming together and at the same pace, one or another can have higher stroke hate, impacting the satellites the watch can connect when your wrist goes above the water surface.

To better measure the differences between the accuracy of both watches, you can swim with one in each arm at a continuous pace. If you go to round tracks, you probably will see sharps lines on the 920 maps.

Simple maths will tell you the difference is 96 yards (2650-2554). Note also the straighter lines from the 920 suggest fewer satellite 'grabs' and more approximation of distance.

Your watches also suggest a time difference of 1:24s. That could indicate slightly further distance before the watch was stopped. It's also possible to see a difference between the start and stop positions for the watches too. Not only were they not stopped at the same time, the start/stop positions were also different.

it is not possible for you both to swim exactly the same course. as noted...

TwinLabs13 said:

swim virtually the same (similar route).

The best way to get close to the true distance for your swim is, as already noted) pace one of the watches in a swim float towed behind you.

There is always likely to be slight differences even between the same watches. The watches do not always grab the same GPS signal at the same instant thus the 'magic algorithm' Garmin uses to approximate swim distance and direction will always have different information to work with.

If this is all you have to complain about with your OWS results, perhaps read some other posts where others have very different problems with much bigger discrepancies.

Seriously, both tracks look fine for an OWS. Half the time the watch is under water, so there's now way this can be super accurate. Also, FR945 looks more accurate in the top bend. That's not really an issue you're describing here.

It's simply to understand that the right track is the 945 track (if it's second image), more fluid and with more precision. In the first image I see only some straight lines and I don't think that is your right way...