Explanation of the digital exhibition 'Bos atlases: mapping the world (1877-1939)'

The digital exhibition Bos atlases: Mapping the world (1877-1939) provides a magnificent overview of all the pre-war editions of the famous Bosatlas, the most important school atlas in the Netherlands for the past 150 years. The aim of this exhibition is to show the user how the world has changed, how the representation of the world has changed, and how the editors of the atlas believed that the representation of the world had to change. The demonstrable differences between consecutive maps created in that period say something about both actual changes taking place in the world and the interpretation of these changes by atlas editors, who were obviously influenced by opinions in society.

All editions of De Bosatlas and many associated maps are accompanied by a written commentary, which should help to identify and explain the changes to the user. This background story forms a cartographic evaluation of those comments. The various ways in which this website can be used, as well as the necessary functionalities, have been outlined in a paper presented in Seoul in May 2016 (Ormeling, 2016). A particular aspect of De Bosatlas, namely the way in which the editors dealt with colonialism, was presented at a seminar organised by the ICA Commission on the History of Cartography in Dubrovnik, in October 2016.


Comments were added to atlas editions and maps from all Bosatlas editions by photographing all the maps and adding them to a specific file for each edition – a method which is now outdated because of the complete digitising. Subsequently, the same maps which appear in multiple editions were grouped together so that, for example, all the maps of France or all world maps of sea currents were in the same place. The same maps from subsequent editions were then displayed next to each other on the computer screen for comparison. It is possible to zoom in on both images. It would have been even better if they had been superimposed for comparison. This is therefore only a preliminary inventory of the changes; with the new tool – Utrecht University’s synchronized viewer – developed over the course of the project, the maps can and should be re-evaluated, because it is possible that there are still many more changes waiting to be discovered (Figure 1).

Ontwikkeling havenwerken Amsterdam en Rotterdam in vergelijkingsviewer Bosatlas
Figure 1: Example of the synchronized viewer, with the map of Rotterdam in two subsequent editions.

‘Gold rushes’ in 'De Bosatlas'

In those instances where changes in the map imagery were identified, the author tried to discover the reasons behind them. Research was conducted to find out what sources were used for the editorial work. Initially, this involved trying to uncover the story behind the atlas (the ‘narrative’). What did the editor want to achieve with the atlas? To do this, the prefaces of the various editions were closely examined. Issues of the Tijdschrift van het Aardrijkskundig Genootschap (Royal Dutch Geographical Society Journal), for example, were used as sources for the map content. If comparing two editions of a map turned up a new place, this was then researched on Wikipedia. For striking place names, borders and symbols used on the maps, the edition was determined in which they first appeared and Wikipedia was used to try to find out why that might have happened. In doing so, a whole host of possible 'gold rushes' were found. Apparently, editor Bos in particular was very sensitive to this aspect. He included many sites where precious metals were found, the importance of which was negligible for geography teaching. Within a few years these locations would degenerate into ‘ghost towns’, and they certainly did not fit in with Von Humboldt’s concept that only the most important issues should be incorporated into a school atlas. It may well have been his penchant for highlighting recent events and drama, but in the process he confused short-term success (demonstrating how up-to-date the atlas was) with a long-term effect (a clear overview of the most important aspects). It also indicates that the person charged with editing the atlas had a significant influence on the content and narrative of the atlas.

Different map updating rates

The difference in the rate of updating geographical information lies mainly in the fact that, for the Netherlands and its colonies, changes were implemented immediately. This process was much slower for neighbouring European countries as well as other more distant parts of the world. For the Netherlands, all changes in the landscape were scrupulously tracked, so that maps of the provinces in De Bosatlas in particular are an excellent source of information for landscape reconstruction. Depicted with painstaking detail on these maps of the provinces are soil types (from the 12th edition), the character of the settlements (from the 13th edition), the expansion of urban areas (from the 3rd edition), railway and tram lines (from the 6th edition), shipping channels and areas of cultivated peat lands (from the 13th edition), and the dikes along the sea and rivers until, for the sake of readability, communities and building symbols were removed in the 32nd edition released in 1929.

Maps of other countries were updated at a slower pace; the process of implementing changes was much slower for countries outside Europe than European countries, with the exception of our colonies. The slow pace of revision was particularly evident in themed inset maps. In the 19th edition (1910), the 1910 iron ore and coal mining production figures are given for Britain. The figures were subsequently removed, but proportional symbols for this information were included in the map in the 36th edition (1939) based on the figures for 1910!

For the colonies, the process of updating information worked a lot better: the preface often included an apology if an administrative change (which happened every other minute!) such as a border change or a change in status of a residence was not immediately reflected in the next edition of the map.


Keeping the maps up-to-date, however, ran counter to ‘anticipating’ developments, whereby planned objects that were not yet complete were also included in the maps. Editor Niermeyer in particular had a tendency to do this. Take, for example, the Hedzjas railway, which was planned around 1900 between Damascus and Mecca. In 1908 the line was completed from Damascus to Medina, but the section from Medina to Mecca would never be realised. Despite this, the railway line was shown in the 17th edition of De Bosatlas as if it were already fully completed (Figure 2). The same is true for the railway line on the other side of the Red Sea, along the Nile between Wadi Halfa and Assuan (now Aswan). Due to the challenging terrain, this track would never be built. Only in the 23th edition did the Medina-Mecca railway line disappear from the atlas.

Detail kaart Afrika zeventiende editie Bosatlas (1906)
Figure 2: Detail from the 17th edition (1906).

Island cartography

‘Island cartography’ is the term given to the phenomenon that only a certain part of the map is completely filled with data, while other parts, which relate to areas not covered in the map title, are not. For example, take the Austrian-German-Russian border area near Kraków in 1885 (Figure 3). On the map of Germany, the railway line runs through the Austrian province of Galicia between the important (but not featured on the map fragments) centres of Lemberg (Lvov) and Prague (or between Krakow and Troppau/Olmütz) via the German cities of Königshütte and Ratibor; on the map of Austria-Hungary, the railway line runs west of Krakow, south of the Vistula (Wisla), and thus remains completely in Austrian territory.

On the map of Asia from the 2nd edition (Figure 4), the consequence of island cartography is evident: areas outside Asia are left white and the objects situated in that area are left unnamed. This makes it difficult to explain relationships between objects in (or on the border of) Asia and in Europe or Africa. This island cartography phenomenon remains a feature of De Bosatlas almost until the end of the period being dealt with here; the latest criticisms about island cartography are included in the various editions during the transition from the 33rd to the 34th edition. These comments state that it is confusing to use white or lines to display parts of political maps that do not form part of the area covered by the title, since after the World War I these lines were also used to indicate dominions.

Detail kaart van Duitsland 7e editie Bosatlas
Detail kaart Oostenrijk-Hongarije 7e editie Bosatlas
Figure 3a and 3b: Two maps from the 7th edition of De Bosatlas; above, the map of Germany; below, the map of Austria-Hungary.
Staatkundige kaart van Azië 2e editie Bosatlas
Staatkundige kaart van Azië 18e editie Bosatlas
Figure 4a, and 4b: The political map of Asia from the 2nd (above) and 18th (below) editions of De Bosatlas. Island cartography features in both.

Changes in structure

The geographical proportions used for map sheets in the 1st edition of De Bosatlas are: Netherlands 8%, Europe excluding the Netherlands 52%, world 8%, other parts of the world (excluding Europe) 32%. Bos obviously attempted to use the same scale (1:3.7 million) for all major European territories. The proportion used for colonies (a map of the Indian archipelago is included in editions after 1878) is 3%.

As demonstrated in Table 1, the map proportions for the Netherlands increase gradually from 8 to 21%. Conversely, for Europe and other European countries this proportion decreases gradually from 52 to 32%; the proportion for other parts of the world increases from 32 to 38%. However, this increase can be explained by the increase in maps of the colonies (from 3 to 14%!).

Verandering in de structuur van de Bosatlas: aantal en aandeel der atlas-spreads 1877-1939
Table 1: Changes in the structure of the atlas: number and share of atlas spreads between 1877 and 1939. *Black and white maps are counted as half a point.

In addition to the proportions allocated to the various areas, the order in which these areas are presented also changes (Figure 5). This occurs as follows: within Europe: in 1919, after World War I, Britain was moved forward and now immediately followed France. In 1939, possibly as a response to the Spanish Civil War but more likely because Scandinavia was closer to our hearts than Russia, the Balkans and Southern Europe, the order changed radically: the United Kingdom was now followed by Scandinavia, then Russia, the Balkans, Italy and finally Spain.

Volgordes kaarten in de Bosatlas, 1881, 1919 en 1939
Figure 5a, 5b, and 5c: Order in 1881 (left), order in 1919 (centre), order in 1939 (right).

As far as continents are concerned, the experiment to put Australia/Oceania immediately after Asia was already abandoned in the 3rd edition. For the other continents, the order changed as follows: neighbouring Asia and Africa came first, followed by North and South America and finally the most distant continent, Australia.

Black and white maps were gradually added to the reverse sides of the colour-printed atlas spreads. Colour was not used because a correct colour application could not be guaranteed at the time. This did not become a reality until 1960, after many internal disputes between the then editor and the head of the printing company.

Geographical names

Research into differences between the editions also included the different spellings of place names. Tichelaar’s article (2002) proved to be very helpful in this regard. Unfortunately, nowhere do the editors explain their reasons for displaying geographical names in a particular way. For example, they might have indicated whether they chose to use local official versions (endonyms) or the Dutch versions (exonyms). Moreover, when adopting names from countries that use writing systems other than our Latin alphabet, they could have indicated whether they had opted for transliteration (focusing on the correct conversion of the characters) or transcription (focusing on the correct conversion of sounds). Editor Kwast in particular, but also Niermeyer, appear to have employed the transcription method, but not in a very systematic way. The current editor of the atlas, Tjeerd Tichelaar (2002), examined their working methods for the map of the Balkans.

Toponyms in the Balkans

The map of the Balkans in the 1st edition from 1877 contains Turkish, Slavic, and Italian names, as well as exonyms derived from Byzantine antiquity. Greek names are transliterated, Cyrillic and Arabic names are transcribed. In the 11th edition of the atlas, many Italian names were Hellenised; Bos opted for the Katarevoussa version rather than the Dimotiki version (i.e., a literary version influenced by ancient Greek rather than the vernacular). Moreover, Latin spellings lose out to Greek versions: ‘Kykladen’, ‘Mykene’, ‘Granikos’ (rather than ‘Cycladen’, ‘Mycenae’, ‘Granicus’). In Bulgaria, Bulgarian names replace Turkish ones. Most double names were reduced to single names.

In the 14th edition, the remaining Turkish names in Bulgaria were also written in Slavic; the caron (š,Č) also makes its first appearance in the transliterations. Croatian-Latin script was introduced for Serbia and Bosnia. The transcription of Turkish names in Turkey was completed with the help of linguistic expertise. For Greece, classical Greek transliteration is no longer used in the 14th edition of the atlas, but rather the modern transliteration of the Katarevoussa version i.e., ‘Limnos’ and ‘Levkas’ instead of ‘Lemnos’ and ‘Leukas’.

Exonyms and endonyms

Bos set a trend to replace exonyms and transcriptions with endonyms and internationally accepted transliterations. His successor, Niermeyer, was a revisionist when it came to names. He replaced many new Greek names with the classical versions: ‘Sparti’ became ‘Sparta’ and ‘Eubea’ became ‘Euboea’. It was only in 1925 that editor Kwast started to look more closely at this issue, but even that did not occur systematically. Eibergen, who helped Kwast edit the atlas after 1929, was the first to add some level of structure to the nomenclature. As a result, the 32nd edition included many changes. However, it is not possible to identify a general approach. Apparently the necessary expertise was no longer available. For the 36th edition (1939), Eibergen tried to revert to using exonyms and transcriptions - except for Greek. All Slavic names, both written in Croatian using our Latin alphabet and in Serbian using the Cyrillic alphabet, were transcribed: The example of the Serbian town 'Kruševac’ illustrates this point: ‘Kroesjewats’ (1877), Krusevac (14th edition, here Bos chose to use the endonym), ‘Kruševac’ (15th edition, endonym with correct spelling), ‘Kroesjewats’ (36th edition, Eibergen reverts to using the exonym with correct Dutch pronunciation).


‘Protrusions’ of the mapped area outside the frame of the map feature in De Bosatlas more than in any other school atlas. This happens because, when using a particular (usually nicely rounded) scale, an area doesn’t quite fit within the frame, or because a depicted phenomenon (the course of a river, an archipelago, a railway line, a telegraph cable or a coastline) doesn’t quite fit within that frame. Cases of ‘protrusions’ are often the result of growing insight: for example, the political map of the Netherlands had been included in the atlas for years when, in the 18th edition, the edge of the map was breached to incorporate the lower reaches of the river Ems. A protrusion appeared on the map of Asia following the discovery of Severnaya Zemlya. The map of the Balkans also required a similar protrusion once Bessarabia had become part of the Kingdom or Romania after World War I, as did the map of the Pacific Ocean (34-45) when telegraph cables had to be added. When the decision was made to indicate language borders, protrusions were considered necessary for the maps of Belgium and northern France (to represent the Dutch-speaking region). The map of Italy also needed to be extended due to the newly drawn borders in the aftermath of World War I, namely to include the addition of South Tyrol.

Bosatlas kaart 26-27, Italië 1922.
Figure 6: Bosatlas map 26-27, Italy 1922.
Bosatlas 35ste editie kaart 45, Australië (1936) met telegraaflijnen.
Figure 7: Bosatlas 35th edition map 45, Australia (1936) with telegraph lines.
Frans-Vlaanderen in de Bosatlas, 1908, 1910, 1914
Figure 8: French Flanders in the 18th (1908), 19th (1910) and 21st editions (1914) of De Bosatlas. The bright red line indicates the language borders.

The atlas as an ongoing project

Due to the amount of staff working at the lithographers, De Wijer – the company charged with preparing De Bosatlas on stone – only a limited amount of time could be spent on each new edition of the atlas. If a change needed to be made, it would take several editions before it was implemented in all of the maps. Geographical coordinates are a good example. As indicated below, it was not until the 15th edition that these were based on the Greenwich prime meridian - a transition which had already begun in the 8th edition. Other examples include the adding of legends and the introduction of the metric system: initially, everything was still measured in miles and feet. The transition to kilometres took place only gradually, especially for the inset maps and the maps of the colonies. And yet, in the 36th edition, the scales used for all maps of the world and European countries are still expressed in geographical miles, while the survey maps of the Netherlands are given in hours of walking distance.

Invoering van de nulmeridiaan van Greenwich in de eerste 16 edities.
Table 2: Introduction of the Greenwich prime meridian in the first 16 editions. Green is Ferro, yellow Amsterdam and orange Greenwich. The corresponding maps are indicated in the left-hand column.

Table 2 demonstrates that, apart from the maps of the hemispheres, the switch from the Ferro prime meridian (green) to the Greenwich prime meridian (orange) began in the 8th edition. With the exception of the maps of the Netherlands highlighted in yellow, which in the 1st and from the 9th until the 11th editions used the Amsterdam prime meridian, the Greenwich prime meridian was used for an increasing number of maps. This process would be fully completed by the 15th edition. The maps of the colonies were the first to switch to the international prime meridian; maps of Britain, Spain and North America held on to the Ferro meridian the longest. In total, it took seven editions to fully switch from one meridian to the other!

Population legends

A similar picture emerges from the table (Table 3) regarding the introduction of a ‘population legend’, detailing the population size of a particular place depending on the symbol it was allocated. Although the maps of Dutch provinces classify places according to population from the 3rd edition onwards, Bos started to introduce population legends systematically in the 6th edition, at a time when the maps of Great Britain and Spain were being redesigned and the model for other new maps was being developed: in the 7th edition for the map of the Alps, in the 8th edition for the European, African and North American continents, in the 9th edition for the survey maps of the Netherlands, in the 10th edition for Germany, Asia and South America, the 11th edition for the United States, the 12th edition for Australia, the 13th edition for Russia, the 14th edition for Belgium and Switzerland, and finally, the 15th edition for France, Austria-Hungary, Italy, the Balkans and Southeast Asia. For a long time, places in the Dutch East Indies continued to be labelled with the symbol for their administrative status rather than their population. For Sumatra, this continued to be the case until 1939.

Invoering van plaatsstippenlegenda’s per editie op de verschillende kaarten.
Table 3: Introduction of population legends in different maps, per edition

Depth contours and elevation legends

That there was another way of doing things, that changes could be implemented all at once, is demonstrated by the introduction of other shades of depth throughout the atlas. One example is the introduction of depth contours of 1,000 and 3,000 fathoms in the 6th edition, which were simplified to show only the depth of 1,000 fathoms in the 7th edition, while depth contours of 200, 2,000 and 4,000 metres were introduced in all maps of continents in the 12th edition.

Similarly for the introduction of legends detailing information about elevation it seems that changes could be implemented in just one step. Table 4 shows that, apart from the Netherlands and Java, where elevations and corresponding legends had already been included since the 1st edition, such a legend was introduced for virtually all other geographical survey maps in the 8th edition in one go.

Invoering van hoogtelegenda’s per editie op de verschillende kaarten.
Table 4: Introduction of elevation legends in different maps, per edition.

Later, outside the 1877-1939 time frame being dealt with here, the transition from lines to a ring of shading for representing relief, the transition from one type of script to another and the introduction of hydrography printed in blue would also take many editions to complete. The limited capacity of the graphic design office therefore offers an explanation for the mismatches between different maps of the same area in a single edition. In the 7th edition, the depths on a map of the Dutch East Indies are already stated in meters, while other maps still use fathoms. By the 12th edition, meters are used universally.


However, we can also apportion a certain amount of blame to the atlas editors for errors that cannot be excused by the limited capacity of the graphic design office. For example, there is a lack of consistency for the representation of place names: on the physical map of North America, Salt Lake City is labelled as ‘Z.M.St‘ from the 8th to the 14th editions and as ‘Zoutmeerst.’ from the 15th to the 34th editions (thereafter, place names were no longer labelled on this map). On the political map of North America, the city is labelled as ‘Gr. Zoutmeerstad’ from the 1st until the 7th editions; as 'Salt Lake City' (‘Zoutmeerstad') from the 8th to the 14th editions; as ‘Zoutmeerstad’ from the 15th to the 33th editions, and as ‘Salt Lake City’ from the 34th to the 36th edition. On the map of the United States, the city is labelled as ‘Groote Zoutmeerstad’ from the 1st until the 10th editions; as ‘Great Salt Lake City’ from the 11th until the 23rd editions, and as ‘Salt Lake City' from the 24th to the 36th editions! Of course, this could have been standardised much earlier.

Obvious errors

In some cases it also takes many editions for obvious errors to be detected and corrected, highlighting the fact that the editors did not review the atlas critically enough. The 'East Anglican Heights’ included in the 2nd edition was not corrected until the 6th edition (‘East Anglian Heights’); the incorrect place name ‘Faikbaai’ on the isle of Curaçao used in the 6th edition was not corrected until the 17th edition (‘Fuikbaai’). On the language inset map for the map of Great Britain, an unfortunate choice was made to make areas with a Gaelic-speaking minority darker than those areas where there was a majority. This wasn’t rectified until at least the 14th edition. It is astounding that no one had ever criticised such a basic mistake.

Mount Everest, the highest mountain in the world, wasn’t even included in the atlas from the 10th to the 15th editions. In the 10th edition, the railway line between Santos and Sao Paulo was accidentally omitted from the map; a mistake that was discovered only eight years later, and the railway line was finally put back on the map in the 13th edition of 1897. In the 12th edition, the coastal town of Portland in the Australian state of Victoria was confused with Warnambool. This error wasn’t rectified until the 17th edition! In the 13th edition, Herkenbosch near Roermond is labelled on the map as ‘Berkenbosch’. This was corrected in the 15th edition. In the 13th edition, the name of the seaport town of Aqaba at the northern tip of the Red Sea was mistakenly changed to ‘Ikaba’. This error remained uncorrected until the 28th edition of 1923.

On the map of Canada in the 15th edition, Fort Good Hope is positioned along the Mackenzie River. Although it was located in the right place in the previous two editions, approximately at the point where the Mackenzie River and the Arctic Circle intersect, it was now labelled as being where Fort Norman should be. Why Bos positioned Fort Good Hope correctly in the two previous editions and then labelled Fort Norman as being Fort Good Hope we will never know. This was only discovered in the 33rd edition!
When designing the 17th edition, the editors discovered, after ten editions, that the scale used for the map of Curaçao was wrong: it was stated as 1: 100,000 and it should actually have been 1:1 million.

Flemish minority or French minority?

Another example of carelessness is the language map which was included as an insert for the map of France (Figure 9): in the 29th to the 31st editions, it was suggested that there was a French minority living in north-western Belgium instead of a Flemish minority living in northern France. This mistake has a history going back a long time. On the map included in the 18th edition, the Dutch-speaking region in northern France is still indicated by a bright red colour. In the subsequent 19th edition (second map), the Dutch or Flemish-speaking region is left white, while the Walloon region is given a distinctive orange colour, differing from the vertical white/light orange shading for the other French dialects.

Especially when the French political border becomes quite dominant, as in the third map, it is difficult to see that a different language is spoken in the little corner bordering Belgium. This problem comes home to roost in the fourth map (25th edition): when the line used for demarcating the state border returns to normal proportions, it is no longer clear that a dialect other than French is spoken in that corner and it is given the white/light orange shading of a French dialect. In the next, fifth step, the black dotted line of the dialect border becomes confused with the state border - we know that there is a minority dialect in this area, and if both are French dialects, then the minority must be in Belgium: this results in the fifth map (29th edition), in which the Belgian state border extends beyond Dunkirk and within this expanded Belgium there is a French-speaking minority, besides the already mentioned Walloons.

No one picked up on this until preparations were being made for the 32nd edition - five years later - and only then was the mistake corrected! And this at a time when the language border was a hot topic in Belgium between 1924 and 1929. Did nobody check the atlas properly? In the last map, the linguistic situation is once again presented correctly.

Weergave van de Vlaamse minderheid in Noord-Frankrijk in opeenvolgende Bosatlassen
Figure 9: Presentation of the Flemish minority in northern France in successive Bosatlases.


In addition to a number of careless mistakes there are fortunately lots of wonderful maps hidden in old editions of De Bosatlas, which are ideal for making comparisons with maps included in more recent editions of the atlas. For example, a map of the Mississippi Delta from the 6th edition (1884) and the 54th edition (2012) in Figure 10.

De Mississippi-delta in 1884.
De Mississippi-delta in 2012
Figure 10: Mississippi delta in 1884 (above) and 2012 (below).


Em. Prof Ferjan Ormeling, 2016