GPS accuracy and ancient instruments (Score 1)

Of course surveying by other means can be more accurate than a GPS.

The GPS is limited due to degredation of the signals from the satellites by the atmosphere and from the accuracy of the on-board electronics. The manufacturers cite a 100 m accuracy, though in general it is much better than this. A typical low-cost handheld GPS will often list horizontal errors on the order of 5 meters, but that is an estimate of 1 standard deviation. The vertical error is often two or three times as great. The error is also only an estimate, and can easily be twice that reported. You can improve the GPS with a DGPS (Differential GPS), since the atmospheric errors are largely the same nearby. You can subscribe to a satellite broadcast correction hat can get you to +/- 1 meter, or go for real surveying GPS gear ($10,000+) and get +/- 0.01 meter or better.

Traditional surveying gear (such as a theodolite or a laser range finder) can easily get millimeter accuracy if you measure from a known location. Astronomical surveying can get accurate locations without known points. The latitude is easy, but the longitude requires a properly calibrated set of tables, some serious math, and a good watch.

All of the GPS methods give you coordinates that are in the WGS 84 (World Geodetic System) coordinate system or some simple modification there in. Most topo maps are not in this coordinate system, and, at most, have a few crosshairs printed on them to help the user adjust. The US topo maps are almost exclusively in the NAD 27 coordinate system. Using the lame datum shift in GPS to convert between NAD 27 and WGS 84 can give some serious errors (about 11 meters where I live, as much as 60 in some parts of the US, and more in TROTW). This results in the question of which confluence you want to measure. Differences are huge.

The Degree Confluence Project uses the WGS 84 coordinate system, and so a GPS gives the location easily in the correct system. Using a map and compass or a theodolite and chain or a telescope and lunar table, you not only have to locate the correct point, but you have to figure out where the correct point is located.