As I've been adding grid references to Articles without grid references, I've noticed quite a few of the outstanding ones are on Jersey.

I haven't a clue what grid reference system they use, but its definitely too far south to use the GB one and definitely too far east to use the Irish one. Any ideas?

That'll be why they are without grid references - they can't easily be given them!

May I suggest some lat/lng fields for these and other roads outside Britain and Ireland to allow them to have the same treatment on SABRE maps?

|lat_from =
|lng_from =
|lat_to =
|lng_to =

They don't have to show up on the page, and it causes problems with using SABRE maps to easily find the references (though if you give me the references I can do Jersey with ease, as I know where all the roads go, having done almost all their lengths).

There's my two cents...

PS - don't forget that the British grid reference system covers the whole of Ireland, half of France, some of Iceland and so on...

You read my mind! Easy to implement, can perhaps show in a similar way to the grid references, but the link showing as (map) rather than showing the actual coord.

Quick and easy way of getting coords... http://itouchmap.com/latlong.html

This page has some suggestions on how to handle grid references in Jersey, suggesting a set based on UTM30U.

I think having lat / lon for roads as well as grid references will definitely be worthwhile anyway now that Maps has OSM route traces. Since a 3 digit grid reference can only approximate a particular point, while the traces uses lat / lon with far higher accuracy, meaning that the marker for the start of the road, and the start of the trace don't line up.

Of course, then we'll need a maintenance category "Articles without lat / lon", and have to start plugging those in...

I think if we introduce lat/lng for all roads, as well as grid references (as much as I'd love to (I generally work in lat/lng rather than grid refs myself)) our collective todo list is going to get unmanageable.

Perhaps its something to consider for the longer term, but for now I think we should focus on getting the existing missing gridrefs/counties/authorities/mileages in first (all can be found under Category:Wiki Maintenance Categories, generally beginning with "Articles without...").

I have (thanks Nic's box of stuff} a Jersey map that has OS grid on it (which I didn't have for working out lengths), so I can work out grid references off that (which would be easier for me than playing about with software). There's nothing stopping us having both, though lat/lng would be easier to obtain for Jersey.

I'm very certain a maintenance category for 'Route articles without grid references or lat/lng coordinates' would be easy enough to set up. That way we can have data for all roads - one, the other or both - and can have the added accuracy/provision for international roads. We wouldn't have to (though we could) add lat/lng for every GB&I road, just the ones we want to - maps would still work with just gridrefs.

On a related point, some crossings, namely a lot of ferries and the Channel Tunnel, are rather long. May I propose adding to {{Infobox}}, etc the gridref and lat/lng from/to fields. This way, we don't just point to a bit in the middle of the sea! I'm definitely not saying we ought to have this for all, or even any Bridges and Tunnels (though Dartford Crossing, Humber Bridge, Dublin Port Tunnel, Severn Bridges and other long crossings may desire these fields).

The Channel Islands are certainly covered by the National Grid.

I've worked out (I have far too much time on my hands ) that St Helier is at about XD3222.

This assumes that one degree of latitude or longitude is about 111km - a major assumption I know. And of course my maths could be wrong!

Not that any of this helps, of course!

Jersey is outside the official boundary of the grid. Places more than 11km south of the Lizard have a negative northing. (On an OS map look at the digit printed in small type before the two digits in big type. The northing of the south edge of Landranger 203 is [0]11km.)

If you don't mind that, you can of course extrapolate the grid southward to give references beginning X or Y.

111 km is right (to the nearest km) for a degree of latitude, but the equivalent in km of a degree of longitude varies with latitude, from a similar amount at the equator to zero at the poles. It looks to me as though you haven't allowed for that. St Helier seems to be roughly due south of Wareham, so your easting looks wrong.

The website Ritchie linked to had Jersey as being in the middle of WV, so it's SW of that point. However WV as a square is in the bottom-most row of squares. ZZ is northern Italy, so that can't be right!

XD fits better (might be XC). The true origin is not very far South of the Island (in line with the east of centre - in line with Gorey). The false origin is SV000000 and is 400km west and 100km north of 49N 2W. Lets head south 100m to begin to make it easier - that drops us into XA000999 - add 400km, and you get XE000999, drop 99.9km and you get XE000000 (or 49N, 2E). Therefore, we're looking at XD references for most of Jersey, and a few XE ones for the east coast.

http://www.fieldenmaps.info/cconv/cconv_gb.html covers the channel islands according to the other place.

This article shows how the national grid fits most of western Europe from Iceland to Italy.

It makes for an entertaining read too!

Jersey is definitely all in XD. It's entirely west of the line of longitude 2 degrees west. This line is the "central meridian" of the Transverse Mercator projection used by the OS, and is the only line of latitude or longitude that coincides exactly with a grid line. It has an easting of [4]00 along its whole length. It marks the boundary between XD and XE, also between SY and SZ, and so on northwards up to HO and HP.

rechecking my map, all of mainland Jersey is west of 2W (just), only about 5 or 6 outling outcrops are east of it.

The map I linked to uses a different origin to the GB National Grid, so the grid square will be different.

Extending the grid south of the "false" origin (but still north of the "true" one) to fit in the Channel Islands is certainly do-able, but I'd be reluctant to do this without some evidence that it's used in the real world.

Can you enter a couple of grid references from your map, Si, and I can use those as test cases. Alternatively, what happens if you try and plug a Channel Islands grid reference into GeoHack?

Very funny but it's wrong about making Scotland seem smaller. The scale is the same all along the central meridian, but gets larger as you get further from it. That's why this projection is used, because Britain doesn't extend very far in an EW direction. Since the places in GB furthest from the central meridian are in the Western Isles, those places are magnified most.

The article is right, though, about the weird coincidence that the junction of squares N, O, S and T is incredibly close to the coast. The tiny bit of beach in square O isn't shown on any Landranger, unless a blister has been added to recent editions of 94 or 101. There even seems to be an ultra-tiny bit of land above high tide that isn't covered. I once checked to see if it was shown on any 1:25000 sheet, but I can't remember whether it was or not.

My copy of sheet 94 does show the part of square OV above the low-tide mark. It's not particularly obvious, however.

Certainly when 1:25000 maps were green, square OV wasn't mapped (or at least wasn't easy to lay your hands on). Now they've gone orange, that part of the national grid is shown. Of course, use any online mapping service and you'll see the small amount of sand.

OpenData and Land-Line Tile selector (tile OV0000) appear to show that there is some land above HWM in square OV.

here's the A5 - 662478 (west end) 705479 (east end). The false origin of the grid is 0N 3W and we're looking at the 500kmE, 5400kmN 100 square. it's the Universal Transverse Mercator Grid Zone 30.

If it is, its no more than the thickness of the gridline - a few metres at most. Given the rate of erosion along that coast, it may have gone by now anyway!

As for whether Scotland looks smaller: it does, but only beacuse the familiar Mercator projection, in which lines of longitude are parallel, exaggerates the size of areas at higher latitudes. The grid keeps areas equal - all 1km squares are, by definition, the same size. Consequently the grid, by making Scotland its true size, does make it look smaller in relation to England than a Mercator projection would.

Not true. Grid lines are defined to be exactly straight, exactly parallel or perpendicular, and exactly equally spaced as they appear on (an idealised version of) the map. Grid squares on the ground get slightly smaller as you go further E or W of the line 2 deg W, so by appearing the same size on the map, the scale gets larger further away from that line. If you continued the grid until near to a point on the equator at 92 deg W or 88 deg E, nominally 1km grid squares on the ground would be microscopic.

The projection used preserves small shapes accurately. No projection can do this and also preserve areas accurately.

As stated in my previous post, the projection used by OS is a transverse Mercator. Transverse means that the whole mechanism of a "familiar" Mercator is rotated through 90 degrees. Instead of exaggerating areas further from the equator, it exaggerates areas further from the central meridian.

In a transverse Merc, no line of latitude or longitude appears exactly straight on the map, except the central meridian (and, if the map extends far enough north or south to show them, the equator and the meridian 180 degrees away from the CM (178degE in the case of OSGB)).

Have I understood correctly? Would a transverse Mercator based on 2 deg W smear the areas on the equator near the Galapagos (92W) and Sumatra (88E) out in the way a conventional Merc does to the Poles?

If the grid squares are not 1km, does this mean grid squares are different sizes (not all 2cm) on the map, or does the scale itself change?