So I measured the track sections (curved and straight) in both systems, and both are correct, using their own measurement tools within the software. Perhaps one system is adding some distance in making the joints? As I mentioned, the actual track sections don't always butt/join perfectly, sometimes there is a very small gap detectable looking closely. But so small its hard to imagine that this could accumulate into a measurable difference.
The gap I measure on the track loop on AnyRail is 7.227".
So it makes sense that in AnyRail, 7.26 will bridge the 7.227" gap. And on SCARM, 7" will span the 7.1 gap.
Likewise, it is sensible that 7" will not span the AnyRail 7.227" gap. And 7.26 is too big for the 7.1" gap on SCARM.
Interesting, but does not tell us where the variations originate, or which is the most accurate. I look forward to the results of your experiments with the actual track.
Both SW systems use math to represent the track sections, and there must be a precision (number of decimals) limit in each system. The rounding errors might accumulate a difference. It doesn't seem that the location on the screen/grid makes a difference.
BTW, I started a conversation with the author of AnyRail, on their forum, about measurements of track sections and precision. The author is indicating that the track dimensions and length are accurate to 2 decimal points in millimeters, using the data from the manufacturer. NO, using actual track measurements from batches of track!
