So far as I know the sun compass was in use in North Africa only. The compass was very specific, it was equipped with the plate engraved with the so called Weir Diagram which consists of concentric ellipses. The diagram's main purpose is determination of azimuth with few simple actions. Accuracy is about 0.3 degree.
I think the main reason to use them in the desert was simplicity and reliability as most days are sunny there. Some soldiers said the plate was annoying because after a time of looking at it it resembled wild animal eyes. Using not compensated magnetic compass in a tank (who would compensate them in combat conditions) is fine for short distances of say few kilometers, from this forest to that village. This was not the case in the desert where navigation sometimes was like at sea and orientation points were very few. I sailed few times in steel yachts (btw, very good and easy to maintain in spite of all appearances) and magnetic compasses required special care. You had to make error table for different courses few times a year at least as the magnetic field of the yacht changed. You could not use hand bearing compasses because you did not know their errors in different places on the yacht and errors could be up to 20 degrees. For taking bearings the main compass mounted on deck was used. There was another compass for helmsman. Both compasses had separate deviation tables for different courses. Theoretically when compass error (called also deviation and not to be confused with variation (declination)) exceeds 10 degrees, compensation should be done. Once we took a proffessional guy (from big ships) to do this. He did his job very well and it took him few hours to place little magnets in the cockpit around helm compass. Yet next day at sea we removed all magnets because they worked fine only when there was no heel. Yachts spend most of their time heeling however, and the heel changed magnets positions with respect to the compass. Compass was more or less always level thanks to Cardan suspension but the magnets were attached to the cockpit and moved up and down and changed their distances too. On steel yachts there is also heel deviation which on courses close to N-S may be up to few degrees. Most magnetic fields of the hull work in horizontal plane. But yachts (and ships) spending most of their time on the Northern hemisphere become also veritcal magnets with S pole at the keel (most keels are of steel, lead is too expensive). This field does not affect the compass when the yacht is level. When the yacht heels, the poles of this magnet are left and right of the compass and start working. The effect obviously is biggest on N-S courses. Still I am sure that John (Pickard) is successful when using his compass inside the car. And there is no need to fight GPS. Our old methods are fine, GPS is fine too. We are proud that our methods work by candle light (logarithmic tables), they are proud because GPS works instantaneously. And they have spare batteries of course. Few dialists doubted usefullness of magnetic compasses because of all those errors. You simply have to gain practice with compasses and they become useful, even in the city. I measured very accurately deviation in few places around my house using my new lovely hand-bearing compass (French Mini Vion 2000). The deviation caused by metal parts of the house was up to 2 degress. Much bigger was the one caused by electric torch when it was on and placed by my feet. I did not need it for the compass but to look at the watch to read time at night. John, I often have this compass in my car too and take readings on highways. Do not worry, I do it carefully. All best Slawek Slawek Grzechnik 32 57.4'N 117 08.8'W http://home.san.rr.com/slawek