In his posting 'New method of locating when adding mines' (http://www.aditnow.co.uk/community/viewtopic.aspx?p=80969), Simon says
... grid references are only accurate to 100m.
I'm afraid that with my pedant's hat on :smartass: I have to say that Simon is talking about the
precision of a grid reference (in this case a six-figure grid reference), not the
accuracy.
The
precision is implied by the number of figures given in the grid reference (which may be of the usual form of two letters and up to ten numbers and which for a ten-figure reference is precise to the metre), or else a pair of co-ordinates, which can have decimal points and offer millimetric precision (though I've only seen these used in archaeological surveying).
The
accuracy of a grid reference is how well it corresponds to the location of the feature on the ground. For instance, if the initial letters of an alpha-numberic grid reference are incorrect, the grid reference will be inaccurate by perhaps 100s of kilometres.
It is perhaps also worth mentioning that a grid reference designates the south-west corner of a square (of the appropriate size - 100m for 6 figures refs, 10m for 8 figures etc.) which contains the feature that is referenced. Many people take the nearest grid reference, which is incorrect. Hence a feature at SH12394569, if given as a six-figure reference, will be SH123456 rather than SH124457. (A number of the map applications on the web are incorrect in this respect.)
Finally, it is also incorrect to use eight-figure references with the last digits being 0 or 5 in an attempt to provide a precision of 50m. (I'm aware of at least one professional organisation that does that.) The eight figures imply a precision of 10m.
More (much more) information on this topic can be found in the Ordnance Survey publications.
If you have read this far, thanks for your attention. We now return you to your normal programme ...
Dave