Jeff,
I hope you follow up and tell us more about this problem.
From your description, it sounds like you are not using the
GPS map feature at all. BUT, would a UTM reading implicitly
be using some map data? I wish I knew.
If you take
only the UTM coordinates from the GPS,
and plot them on the USGS paper map, does it intersect
in the right place? As suggested earlier, what if you take
only the Long/Lat from the GPS and plot it on the USGS map?
... If the road or landmark was added to a revised map it could have been added in with some "artist's license". Try to find a benchmark in the area of the map and see what your GPS says.
Jeff and zman308,
I think this is the right idea, find out how general, and how widespread the
problem is.
Test more data points in an expanding distance from the problem.
Many "benchmarks" are also triangulation points but not all;
and you want triangulation points, which are specifically horizontal control.
On USGS maps, black script was better established than brown script.
Sometimes you can reproduce the problem in a more convenient place,
i.e. using an old map near your home, rather than traveling to the original
sight each time you test.