Re: GPS to grid reference

The message <5h5pi3F3ghbq7U1@xxxxxxxxxxxxxxxxxx>
from Peter Clinch <p.j.clinch@xxxxxxxxxxxx> contains these words:

I think we discussed this some years ago and the conclusion was that
with a gps the co-ordinates were of a point rather than the SE corner of
a square which the OS apparently thinks is the way to go. That is that
the gps averages the numbers rather than truncating them. I know it is a
bit of a moot point given that the resolution exceeds the discrimination
but I still think it matters.

That the resolution exceeds the discrimination is why I can switch
freely between the two assumptions and not have it affect my navigation
at all, so as you say, very much a moot point. In fact, I can do worse
than that and still get away with it... if something is just inside a
square I'll typically treat it differently at either side of the square,
giving 1s rather than technically correct 0s on one side, and giving
consistently proper 9s on the other side.

Which in practice is part way to how any newcomer to the system would
proceed if they were not in possession of the big secret.

I still haven't found out why the OS switched to the square system from
the point system previously in use by the military who had a national
grid a decade or more before the revised version appeared on civilian
maps.

My guess is OS use a square system as the accuracy and resolution go
hand in hand. A 4 figure ref gives you a 1km square, and you know it's
that because it's a 4 figure reference. If you give a co-ordinate then
it's always a point and you can't tell from a point what sort of
accuracy or resolution you're talking about.

You can tell the accuracy by the length of the co-ordinates given. Using
the square system means that the target progresses consistently SW as
the resolution increases.

Once you get down to 10 figures it's effectively the same thing in any case.

Or indeed why they go to such great lengths to keep the general
public in ignorance of their meaning of a grid reference.

It's described on every map, so it's not /that/ secret.

You may think that, but that is not actually the case. The natural
meaning of "Estimate tenths Eastward ..." is to make the most accurate
guess at the actual distance, not to stop at the invisible grid lines
just prior to the point. That it also once actually meant what it says
but doesn't anymore is the secret the OS shows no desire of sharing. The
wording on OS maps of how to give a Grid Reference has changed very
little since it first appeared on military maps well before civilian
maps had dumped their A-Z type references.

I have a pre-war (1939) copy of the Army "Manual of Map Reading, Photo
Reading, and Field Sketching" which is an updated version of manual
first issued in 1929. This describes 2 Grid Reference systems. The first
"The British System" was in use between 1919 and 1927 when it was
superceded by the "Modified British System". The later system is exactly
like the one we use today other than the position of the false and real
origins of The Grid. Grid references in the earlier system repeated
periodically.

"Points are described by their co-ordinates in kilometres ..."

And on areas:

"As long as it is clear and unmistakable the reference may be given by
the co-ordinates of the S.W. corner of the square - e.g. 3207 or Q.70,
etc., according to scale.

It is often necessary to refer to areas for tactical purposes. The area
of a square may be indicated by the S.W. corner - e.g Bivouac area
squares Q.70 and Q.71 - or squares 7363 and 7463 (according to scale)."

It would have been ridiculously easy to maintain the distinction between
GRs and GSs but the opportunity has been lost. I presume the military
has by now changed its tune, possibly to fall in with the Americans who
have dictated NATO specifications since it was first set up.

And people
manage to use the system because of the nature of mapped data and what
they give grid refs for. If I say "the summit at xxx yyy" folk will go
to more or less xxx yyy and look around intelligently for summits,
rather than take out the verniers and ascertain that actually there is
no summit at that exact location. While this is done on the fly by
human data processing, whether or not you consciously think of squares
or points, it actually fits the square model better: go to an /area/ and
look inside it. It works with a map because it's a highly selective
presentation of data, and often where the symbols take up far more space
than the feature really does on the ground.

Funnily enough I don't see it that way. I go to a point and look around
it. With a 6 figure grid ref the maximum displacement error on the
square system is about 140m while that with the point system is half
that.

It doesn't help that anyone quoting a GR will without question refer to
'the co-ordinates' when strictly speaking they mean the co-ordinates of
a point at the SW corner of their chosen square. Such sloppy thinking
doesn't bode well for accuracy in other quarters.

--
Roger Chapman
Nearest Marilyn still to be visited - Great Orme.
89 miles as the crow flies,
considerably more as the walker drives.
.

Relevant Pages

  • Re: GPS to grid reference
    ... with a gps the co-ordinates were of a point rather than the SE corner of ... the gps averages the numbers rather than truncating them. ... That the resolution exceeds the discrimination is why I can switch freely between the two assumptions and not have it affect my navigation at all, so as you say, very much a moot point. ... if something is just inside a square I'll typically treat it differently at either side of the square, giving 1s rather than technically correct 0s on one side, and giving consistently proper 9s on the other side. ...
    (uk.rec.walking)
  • Re: SUDOKU solver for HP48
    ... if there are no forced choices, I will always pick the square with the ... void addgroup ... int i, j, k, tmp; ... * "unsetbox" is used to remove a number from the grid. ...
    (comp.sys.hp48)
  • [SUMMARY] Pen and Paper (#90)
    ... Make a list of all the possible moves from the current square ... When the quiz mentioned circular solutions, ... subdividing the grid and piecing together multiple smaller ... def self.circular_solutions ...
    (comp.lang.ruby)
  • Re: Dangerous Magic
    ... In game terms this is still one additional square in all ... Either you use a grid or you don't. ... Just tick off boxes until you reach 268. ...
    (rec.games.frp.dnd)
  • Movement on large scale battlefields
    ... For a large scale battle I want to use a grid with squares equalling ... increasing the scale from the default of 5 feet per square ...
    (rec.games.frp.dnd)