@ Ben Davidson
On small acreage surveys, there is very little difference between 1:20,000 and 1:75,000.
I remember one particularly defiant setup from the late 1980s that no one of the half dozen that tried to double an angle and none had satisfactory results, averaged error was ?ñ20". It was at a poopnot hill along a rural road and the only place to setup and see both directions. An enormus oak was on the east side of the road and a root ran under the setup and its natural state of rocking to and fro was causing the error.
Bottom line, not all locations are suitable for setting the instrument. The tell is in the results and not how hard you try.
That sort of error sounds like a distance constant error. If your angles are doubling and add uo to what the figure dictates that would be the likely culprit.
Look up 'ABC distance meter check' in your search engine. It's a simple routine that will tell you what sort of constant error you are dealing with.