If you can navigate with a hand held compass to an accuracy of less than 5 degrees I'd be amazed. In my youth I was a national level orienteer, and did lots of detailed navigation in the Army, 5 degrees was always the quoted accuracy variation.
I'm not sure what others were talking about but the variation I was referring to was between grid north and magnetic north, true north is never considered when map reading.
I agree with you that when walking on a heading, 5 degrees is about as good as it gets, and that corresponds well with the inherent capabilities of a normal baseplate compass. That's enough for "dead reckoning" navigation but when it comes to establishing a fix (which every so often we must), we tend to rely heavily on terrain association and a good map to get more accuracy. That works well for most of us, most of the time. In more challenging environments, such as woodland or, conversely, feature-poor moors we may have to rely on accurate back-bearings or resection using features quite far away. For those situations, we need a better sighting compass, IMHO. A mirror compass can give us 2 degrees accuracy with care. A prismatic compass is twice as good as that, but you'll have to go to a chandlers to find a good one, these days, at a reasonable price!
The term "variation" was (AFIK) defined long ago, by mariners to mean the difference (E or W) between magnetic north and true north. The term "declination" is never used, at sea, as a synonym for "variation" because it is used instead to describe a completely different kind of angle in astro-navigation. The two norths for mariners are true north and magnetic north. The lines on a marine chart are lines of latitude and longitude. The latter are aligned N-S (true) and the chart projections commonly used (Mercator and Polar) cause lines of longitude always to be straight lines. In the former projection, lines of latitude are also straight lines; in the latter they are concentric circles. Marine charts do not utilise a regular grid and there is therefore no such thing as grid north.
On land, we use gridded maps because we can (there are technical reasons why regular square grids are feasible only for relatively small areas and won't work on oceanic scales) and because they are much more convenient. Consequently, it is grid north which is most useful. As I said in an earlier post, grid north (be it UTM, OS or whatever) cannot be the same as true north, except at one designated meridian. Therefore, the angle between magnetic north and grid north is nearly always different from the "variation" described above. This is why I chose to use the term Grid Magnetic Angle (GMA) in my previous post.
I've looked up "magnetic declination" and in every definition I've seen it is identical to "variation" i.e. it is related to true north, not whatever grid north one might happen to be using. Misuse of these terms is confusing and if this applies to military training (which I hope it doesn't) it's positively dangerous. Imagine what might happen if an SAS soldier were to call in a missile from a ship using a bearing which the soldier and the sailor interpreted differently!