By Howard Anderson (modified January 2005 and August 2006)
The trouble with having an interest in the Great War, is that any study leaves more questions than answers. One question that keeps puzzling me is, “can I stand exactly on the site of my Grandfather’s trench?”. After a fair amount of effort, it still puzzles me but I think I am getting closer. I would like to set out the work I have done in trying to answer this question, in the hope that some will find it useful. In the example below, the area covered is very small, but the ideas should apply to the whole Western Front. A desire for accuracy is not driven by any need to understand the battle at any level from a soldier’s right up to the general’s war, but rather like those who touch their relative’s name on a memorial, it is driven by remembrance.
On the face of it, it seems very simple. There is fairly easy access to copies of trench maps and some superb modern IGN maps of France and Belgium. All that is required is to plot the trenches on the modern maps and make the pilgrimage. The trouble is, having done just that, I am left with a doubt. As I stand there at Loos or Polygon Wood, I wonder if am I really standing on the spot or is it over there, near that depression on the ground? Is that a trench or a crater? Did some long forgotten road pass here?
Sadly, I am not a cartographer, so that compounds my difficulty. The business of mapping, land surveying and the accurate representation of ground features is a profession in its own right. Even a little research will show just how large and complex the task is. With the advent of computers, it has now grown into an even larger area of activity known as GIS, Geographical Information Systems. The question then is more refined, “can a non-GIS person, with simple equipment, stand exactly on the site of Grandfather’s trench?” Since nothing is ever completely accurate and to avoid statistical problems with the meaning of accuracy, “can I stand within a Mill’s Bomb lob?” Probably.
Some history is maddenly vague. Take the battle of Loos as an example, September 1915. Just when you think you are about to pin down the exact spot, the place where the 1st Middlesex went over the top only to suffer 454 casualties in a few minutes, the history says something like “D Company attacked in front of Les Briques Farm, their first objective”. The farm is easy to find, a short distance from Auchy, not far from the famous brickstacks of the Cuinchy front line. It even has the good manners to be still there, but what does “in front of” mean! The maps of the Official History show that several battalions were in positions I would describe as “in front of” the farm. One could be forgiven for thinking the job is done if at least one trench map reference can be found, look up the reference and ……..there remains a doubt. Pursuing my example of the 1st Middlesex at Loos, I found that the history says “the front line of the Brigade is expressed as from A.27.b.8.0. to A.21.d.4.9.”Throughout the night of 24th-25th September, final preparations were made, and, at 3.30 a.m. on the latter date, the Brigade reported that all units were in positions of assembly in the trenches south of the La Bassée road from Gun Street to R.1; the Middlesex on the right holding from R.1 to D crater, and the Highlanders from the latter to Gun Street. It should be easy….
Trench maps are fascinating. They are so detailed, so many trenches have amazing names. Any trench map reference is quite easy to find once you have understood the rather odd method used. The battle of Loos was in September 1915, a map of at least approximately of that date is needed. The map in question, sheet 36cNW1, is available as a black and white print from the Association or in colour on the Imperial War Museum’s Trench Map archive CD-ROM. The 1915 edition does not have the British trenches, only the secret editions had these, so immediately, the point in question gets harder. Later editions, like Edition 6 of 1916, do have the British trenches. As the line had not moved much, it should be OK to use those, even if it provides another source of doubt. Close inspection however, shows that the grid on the 1915 Edition 2 map is displaced from the 1916 edition 6 map by a fairly large amount. More detailed inspection still shows that far from being a problem with just these two editions, there are differences between editions 2, 6, 7 and 10, the later from mid 1918. The line still had not moved very much though! The problem is really one of converting trench references or land feature positions to latitude/longitude in order to place them on modern maps.
To demonstrate the differences in the plotting of points, here is an example of a single point, the centre of the road junction at Burbure from four editions of sheet 36cNW1. Yards West and North refer to distances from the bottom left hand corner of square A20d.
Map date Edition Yards West Yards North Trench reference
Nov-15 Provisional 2 330 468 A20d 66.94
Dec-15 6 464 333 A20d 93.67
Jun-16 7a 453 336 A20d 91.67
May-18 10b 450 311 A20d 90.62
The 1915 Edition 2 maps were plotted in a hurry from old French maps. This is described very nicely in the “Topography of Armageddon” by Peter Chasseaud. The old French “cadastrals” (taxation or property maps) were used, but overprinted with the Belgian map grid, using a British reference system. The physical triangulation of ground features represented was not of a high order. This and the displaced grid is bad news for my quest for accuracy. As the trench map references for attack locations of the 1st Middlesex are given from the 1915 grid, using these references on later maps is of no use if the requirement is for precision.
It should be possible to calculate the position of any point on the trench map, having found the latitude/longitude of a known point, usually the bottom left-hand corner of the map. Investigation shows the maps were made using the Bonne* projection which means that lines of latitude and longitude are curves. It seems that on maps of the size and scale of a trench map, this curvature is really too small to see. The 1:10,000 Trench maps have 500 yard squares, so the calculation should not be hard. Metres are often easier to use in such calculations, 500 yards is 500×0.9144=457.2 metres. Each minute of latitude is 1 nautical mile or 1852 metres, each minute of longitude is COS(latitude)x1852 metres. At a latitude of 50 degrees North, this equates to about 1190.4 metres. The trouble is, careful measurement of the grid shows it is not exactly oriented True North/South as one might expect. It is roughly North/South of course but is out enough to cause a problem. Another problem found was that on some of the maps the “squares” are a little rectangular, probably due to distortion of the paper. All this may be correct but my lack of understanding of map projections meant that I was not sufficiently confident to trust the results, so I sought a different method.
A better solution was found to be the plotting of known land features from a modern IGN map then transferring these to the trench map.
This is a laborious process if done by hand, but one made quite straightforward on a PC using some good software. This method also has the advantage of reducing some of the errors in paper maps. Although much good software is available, it is intended for professional use and has a price to match. Much cheaper software is available in the form of Oziexplorer*
The plan was to follow these steps:-
Scan a modern map
Tell the software where calibration points are in terms of latitude/longitude (or UTM* co-ordinates)
Pick some land features like road junctions from the modern map and store them
Scan or load a trench map of the same area
Use the land features to define latitude/longitude positions on the trench map
Mark points and lines of interest
Re-load the modern map to show the points and lines
Some notes on calibrating a trench map can be found by clicking here.
The points <should> then be in the right place!
1&2. The latest IGN sheet for this part of the Battle of Loos, Série Bleue 2405E, has a UTM grid printed in blue. Oziexplorer can use this UTM grid to calibrate or “Geo-reference” the scanned image to latitude/longitude co-ordinates. In Oziexplorer, I used the “Load and Calibrate Map” option, set the map projection (written on the IGN map itself) to Lambert and marked several points on the blue grid lines, typing the UTM positions against each point.
Modern map with calibration points. Click image for a larger version
3. Having done this very carefully and having checked that the displayed latitude/longitude agreed with the grid, land features were plotted as waypoints*.
4. An A4 portion of a trench map was scanned with a normal domestic scanner. (Alternatively, a portion of a map from the Imperial War Museum’s Trench Map Archive CD-ROM can be saved in separate file.)
5. The waypoints showing land features from step 3 were used to calibrate the trench map, using the “Load and Calibrate Map” option. Oziexplorer has a feature to make this process quite simple. The map projection was set to Bonne with the details shown below although this is probably not necessary.
The fixed waypoints are assumed to have been fixed in 1915 and still in the same place. This should make the plotting of points independent of map grids and help to cancel some errors in plotting and map production. Mostly these points were road junctions. As maps show roads as symbols, i.e. not to their true width, road or junction centres were used. (This may be in error somewhat as post war road widening schemes often widen the road on one side only, thereby moving the centre. By using several such points and getting Oziexplorer to average them, this problem should be limited ). Having looked very hard at this problem for some time, it became quite obvious that post-war reconstructors of the roads in the area had made a good attempt at putting the roads back where they had been before the very extensive shelling and mining of the area. For example, the reconstructed D75 road from Auchy to Vermelles has the same “wiggles” as the prewar road, but as you may expect, they are not exact. The N41 between Cambrin to Auchy is now straight, but it was not exactly straight in 1915. Just this observation avoids a possible error of at least 10 metres or so. This problem really requires that any “known points” used to calibrate the map are outside the most heavily attacked areas. Trench maps from late 1918 are helpful here, areas that are not well trenched by then are likely to have many of the old roads in the same place. On the other hand, examination of the before and after photographs of Passendale show that even after extremely intense shelling, roads were still clearly visible, lending some credence to the idea that roads could be re-built in the same places. I shall refrain from using the term “exactly” here as the issue of accuracy is difficult.
Passendale before shelling
And after shelling, “a brick coloured stain in the ground”
The graphic below shows a screen from Oziexplorer with calibration points from the May 1918 edition 10b map. Oziexplorer can use any nine of these calibration points. Point 12 is not on the centre of the road as it was defined using the IGN map, the modern road was re-built straight, the war time road was not. Other points are not exactly in place, the result of the calibration makes a “best fit”.
Calibration points. Click image for a larger version.
6. Once the trench map was calibrated, trench lines were marked as tracks*, points of interest such as craters were marked as more waypoints. Oziexplorer allows you to switch back and forth between maps with the same set of waypoints and tracks loaded, so direct comparisons between the maps are simple. If the process is done with care, an estimated accuracy of plus or minus one Mill’s Bomb Lob can be achieved. I am being careful here, my research into maps and statistics has shown the whole issue of accuracy can cause heated arguments, even between rational people! This gets worse once one embarks upon the accuracy of a “domestic” GPS*. I shall try to stick to lobbing virtual Mill’s Bombs.
Having achieved a good trench map calibration, the interesting part was completed. Trench lines were stored as tracks, points of interest such as mine craters (there were lots of these at Loos) stored as more waypoints. This has the advantage that waypoints and tracks can be loaded into a GPS.
Map showing front line trenches and key points. Click image for a larger version.
British front line from June 1916 in blue
German front line June 1916 in orange
German front line December 1915 in red
1. It is quite possible to achieve an accuracy of one Mill’s Bomb Lob when plotting Great War positions. Apart from any considerations of accuracy of individual points, it is clear the actual line of trenches did move by some metres as a result of re-building them after mining or shelling. Standing on an indicated position therefore will give a high probability that you are standing on a trench line, even if the exact line may from a differnt date.
2. No one map calibration will work across all editions for sheet 36cNW1 and probably for all sheets. For precise calibration, each map will need to be considered on its own. In theory, once the latitude/longitude of the map corners are known, the calibration will work for all editions except those before edition 6 of December 1915. In practice, this is not the case as the plotting of points is slightly different on each edition examined.
3. Trench references for a given action must be refered to the map edition available on the date of the action
4. An attempt was made to calibrate other trench maps after having achieved a good calibration on one. Four points were chosen at square boundaries near each of the four corners and these four points used to calibrate the next map. This was found to have significant errors for reasons outlined, the survey does not exactly match the grid between editions.
Off to the Western Front
Now we need to stand where we want to, on a trench line or crater. We need to decide how to achieve this. You could simply map-read your way there or use compass bearings or a GPS. Each method has it advantages.
Compass bearings. You can take a bearing from several fixed points on the IGN map and establish your position in this traditional way, but how accurate is your resulting postion? Can you assume your fixed points are “known”, for instance was the church tower on the trench map shelled out of existence and re-built in a different place? Once you are sure of your points, how accurate can you measure a bearing? A simple navigation rule of thumb works here, the 1 in 60 rule. For an error of 1 degree in a bearing, anything 60 metres away has a sideways error of about 1 metre. So if the church tower is 600 metres away and you make an error of 1 degree in your bearing, you are about 10 metres away from where you think you are. One degree is quite a demanding accuracy to achieve, even with the use of a hand held military marching compass, 3 degrees is more realistic unless you have it on a very stable base. Modern Silva type compasses offer less accuracy still (but more convenience). So even if your known position is correct, you may still be several Mill’s Bomb Lobs from where you want to be.
GPS. Using positions in a GPS is very convenient but introduces another error. As wonderful as they are, nothing is ever perfect. There are two errors to consider. One is the error in the position value itself caused by all sorts of things such as the original survey, map plotting, distortion of the map and any similar errors in the process described above. The other error is in the GPS position that it reports. The two errors may add together but of course may cancel each other out. The GPS error can be eliminated by careful map reading, i.e. not using the GPS! However, they are so useful that I would always take one with me. GPS is probably more accurate than most compass bearing located positions.
Happy trench walking.
Photograph from 2000′ showing some of the area in the maps above, taken in August 2003, looking approximately West. Click image for a larger version.
Photograph from 2000′ showing Mine Point, taken in August 2003, looking approximately North East. Click image for a larger version.
A waypoint is simply a latitude/longitude position that can be given any useful name. In Oziexplorer, it can be named, coloured, moved etc. to suit. Whole sets of waypoints can be stored to reflect particular themes.
A track, in Oziexplorer and in GPS units, is a set of positions joined in a particular sequence. A typical GPS will record a journey as a track, made by carrying the GPS. This can be displayed on a map in Oziexplorer. Alternatively, a track can be loaded into the GPS and the unit set to follow that track, allowing you to “walk the trench”.
GPS and accuracy
A GPS is the Global Positioning System. It works from satellites and gives a position almost anywhere in the world to a useful accuracy. Many GPS units show a value called the EPE or Estimated Position Error. This often has an optimistic value, but, by multiplying it by 3, one can be fairly sure (better than 97%) the actual and displayed positions are within plus or minus this multiplied value. For example, if the EPE is shown as 26 feet, this is about 8 metres, so as 8×3=24, you can be very confident that you are within 24 metres of your displayed position. Readers more familiar with GPS will recognise this as a common cause of misunderstanding! The deliberate addition of random errors to civilian GPS signals (Selective Availability) has been turned off, so that is no longer an issue.
UTM stands for Universal Transverse Mercator. It is a grid system commonly used around the world and can be converted to latitude and longitude. Apart from any technical issue, it avoids one real bug-bear when using latitude/longitude points. It seems no one in the world sticks to anything even approaching a standard way of writing latitude/longitude co-ordinates. For a latitude of say North 50 degrees, 31 minutes, 12.3 seconds, one could find all these formats and more for the same value:-
N50° 31′ 12.3″
50° 31′ 12.3″N
or the very silly N5031123 where no clue is given that it is not decimal. GPS units can be set to show UTM co-ordinates if required.
Latitude and Longitude, Map datums etc.
All latitude and longitude positions must be measured “from” somewhere, a point North or East must be from a zero position. With maps, there is a dazzling selection to choose from. Every country has at least one, the more ambitious have several. In addition, the world is not quite a sphere, it is flattened. Many of the world’s systems to cope with this flattening are individual to that country, which makes the problem worse. Many people do not realise that if they stand in one place on the world, their position can be described by many different values of latitude and longitude and all are quite correct! Choosing the wrong datum can lead to errors of a kilometre! With the advent of GPS, many people are moving towards using the WGS84 or “World Geodetic System” as a datum. This is only approximately the same as using the Greenwich datum, but the WGS84 zero degrees longitude is roughly 110 metres further east at Greenwich. GPS units will cope with all this but are usually set to use WGS84 when you buy them. Ordnance Survey maps use the OSGB36 datum (the 36 defines the date, 1936), so a latitude/longitude point from an OS will not agree with a GPS unless it has been set to use the OS datum. The UTM grid on modern IGN maps are stated as using WGS84 as the datum.
As a result of the world being curved and maps being flat, a deformation or projection must be used. This means that maps are never quite right, some aspect of the map will be in error, either distance or bearing between points. Trench maps use the Bonne projection as per the original Belgian maps, this preserves bearings so are useful for artillery. The modern IGN maps use a different projection, the Lambert Conformal Conic. This is one reason I chose to use known calibration points on the maps, I did not fully understand the implication of changing the map projection. Oziexplorer will cope with map projections and the system outlined above seems to work.
The Datum for the Belgian Bonne Projection (as used on trench maps) is the Observatoire Royale d’Uccle Datum of 1884, based on the origin of the old Brussels Observatory: N 51º 10′ 06.895″ North, E04º 22′ 05.89 (East of Greenwich), and the ellipsoid of reference was the “Carte de France” (Delambre 1810).
N 51º 10′ 06.895″, E 04º 22′ 05.89 = N 050 10.1149167, E 004 22.098167
or 50.1685819, 4.36830278
If you wish to calibrate an Oziexplorer map using the Boone projection, use N 50º 10.1149167′ as the Latitude of Origin and W4º 22.098167′ as the Central Meridian.
Improvements to accuracy
To further improve accuracy, it is wise to visit the “known points” and record their position with the GPS, noting any errors. If every one is out by the same amount, all you need do is get Oziexplorer to shift your calibration waypoints by that amount. If most are within reasonable limits but one or two are clearly out, eliminate them from your calibration.
Oziexplorer was written as an aid to GPS users and is available from www.oziexplorer.com. It will do all that is required and more, and comes in two versions, free and cheap. The free one has a surprising number of useful features, almost enough for my purposes but the registered edition, at £50, comes with sufficient extra features to make it a real bargain. One feature of note in this context of this problem is that it can cope with distorted maps and various map projections. As I do not properly understand map projections, I rely on telling the software which one to use and letting that do the work!
The April 2001 issue of Stand To! (the journal of the Western front Association) contains an article by Walter Stanners on the conversion of trench map references to absolute references. Walter’s article and his Visual Basic program help to identify points using a GPS, but as I have found, a system that uses calculation only does not give sufficient accuracy to answer my question, can I stand exactly on the site of Grandfather’s trench? I would like to thank Walter for a copy of his program, his help and suggestions.
I have been interested in the Great War for some years. I recently left a job managing a department teaching IT and Computing in a large London Further Education college. Although comfortable using computers, GPS etc. I have no training at all in GIS. Anyone better qualified in this area who is able to point out significant errors will be greatly appreciated.