Somebody reads you a position over the radio, or hands you one on a scrap of paper, and you type it into your device. That is the single most error-prone action in this entire course, and none of the errors announce themselves. A transposed pair of digits produces a coordinate that is perfectly well formed and lands on real ground. A digit dropped from the end silently coarsens your precision by a factor of ten. The device accepts all of it without complaint.

Lesson 03 dealt with the two settings-level failures — wrong datum, wrong format. This is the third failure, and it is human rather than technical. The defense is a fixed sequence you run every time, including the times you are certain you do not need it.

Step 3 is the one people skip, and it is the one that works. The instinct is to check your typing against the paper you copied from — but if you misread the paper, you will misread it the same way twice and the check passes. Reading off the device and confirming against the sender closes the loop at both ends. It is the same reason a radio operator reads a message back rather than saying “got it.”

Step 4 is nearly free and catches the errors that survive step 3. If a point should be two kilometers northeast and the device says eleven kilometers southwest, something is wrong even though every digit was confirmed — and it is usually format or datum rather than typing.

2. Going to a stored position

You select the point, tell the device to navigate to it, and get a pointer, a bearing, and a distance. All three are simpler than they look, and one of them is routinely misread.

The bearing and the pointer aim at the destination from wherever you are standing right now. They do not describe the route you planned, and they do not know about the terrain in between. Every time you move, the bearing updates, so following the pointer traces a curve rather than a straight line if you drift at all.

The distance is straight-line distance to the destination — not distance along your route, and certainly not walking distance. On flat open ground the difference is small. In a drainage system, on a road network, or anywhere you cannot walk through what is in front of you, it can be off by a factor of two or more. A member reporting “eight hundred meters out” from a device is reporting how far away the point is, not how far they still have to walk.

Figure 4.1 Cross-track error is the number that tells you how far you have drifted from the line you chose. The pointer will not mention it, so if the route mattered, that is the figure to put on screen.

Most devices can also show your relationship to the original course line rather than just the direction to the destination. That distance to the side of your planned line is cross-track error, and it is the number that tells you whether you have drifted off a route you chose deliberately. If the route mattered — because of terrain, because of a hazard, because somebody else is walking it too — cross-track is the number to watch, not bearing.

3. Recording a position worth keeping

There are two different reasons to record a position, and they deserve different amounts of effort.

Marking on the move is what you do when something matters and you are not stopping: a vehicle, a blocked route, a casualty collection point, a track junction. One press, keep walking, fix the name later. The position is worth whatever the fix was worth at that instant, which is usually fine.

Recording a fixed site is different. A cache, a rally point, a water source, a helicopter landing site — somewhere a person will have to find in the dark, possibly on somebody else’s directions, possibly months later. That deserves a better position than one press gives you, and the tool for it is position averaging.

What averaging actually does, and what it needs

Averaging takes repeated fixes at the same spot and combines them, on the reasoning that random error falls on all sides of the truth and cancels out with enough samples. That reasoning is sound. The catch is the word random.

Two fixes taken thirty seconds apart are computed from very nearly the same satellites in very nearly the same positions in the sky, with the same atmosphere in the way and, if you are near a rock face, the same reflection arriving. Their errors are not independent — they are almost the same error twice. Average fourteen of them and you have carefully computed the average of one error, which is that error.

Figure 4.2 Fourteen samples taken inside a minute share the same satellite geometry, so they share the same error and their average lands on the bias rather than the truth. Fourteen taken over several hours see the geometry change, fall on different sides, and converge. Averaging only removes error that differs between samples.

Getting genuinely independent samples means letting the geometry change, and the geometry changes on the satellites’ schedule rather than yours. Manufacturer guidance is consistent on this and more demanding than most people expect: collect four to eight samples, with at least ninety minutes between them. That turns a cache position into an afternoon rather than a minute — which is the honest cost, and worth paying exactly once for a site you will rely on for years.

4. Reading the accuracy figure in the field

Lesson 01 established what the accuracy number is: the receiver’s estimate of its own work, honest within its own terms and blind to the errors it cannot detect. Here is how that translates into action.

Use it as a relative signal, not an absolute one. Its most useful property is how it changes. A figure that climbs as you enter a drainage is telling you something real about the drainage. A figure that is unchanged while your plotted position jumps around is telling you the device does not understand what is happening to it — which is more informative than the number itself.

Compare it against what you are about to do with the position. Marking a vehicle you will walk back to in daylight, almost any figure will do. Passing a grid that somebody will fly a helicopter to, or recording a cache you will look for at night, and the same figure is not remotely good enough. The number does not change; the standard you hold it to does.

Never let it substitute for looking at the ground. A healthy accuracy figure and a wrong position coexist comfortably, and every one of the settings failures from Lesson 03 produces exactly that combination. The figure is one input. The terrain in front of you is the other, and Lesson 06 is where the two get formally reconciled.

5. Naming, briefly

Every position you save gets a name, and the device will happily supply one like 017. That is fine for a mark you will use in the next twenty minutes and useless for anything else — particularly for anything another member has to find in a list on a dark screen.

A workable name says what the thing is and where, in a form that sorts sensibly: purpose first, then a short locator. The full naming convention is a group decision and it belongs with routes, so Lesson 05 sets it properly. For now, the rule is simply that a default number is not a name, and renaming is easier standing still now than searching later.