Tuesday, March 22, 2022

Worldbuilding Archive: Settlements

Part 1

I'll start with how I create settlements.

I start with a map from my 1952 Colliers Encyclopedia.  The only way I have of copying it at the moment is to take a picture with my phone, so please forgive the shadow and the slight fuzziness.  The map shown corresponds closely to this area of map that I've published.

I copy all the cities on the map into an excel file, with their corresponding 1952 population.  Then, one by one, I research them to find out when they were founded, how many times they were destroyed, defenestrated, plundered or suffered a plague, among other similar disasters that arise.  I find their latitude and longitude, and their elevation ... and with so many cities, which often have the same names as other places, I make mistakes.  That's why the "Bistrita" in my game world is located near Ramnicu-Valpes and not at the top, north of Reghin.  Most of the time, by the time I learn of the mistake, I decide not the fix it.  As I've said, it's "my world," not the real world.  Slight differences are interesting ... and, as it happens, helpful in a copyright dispute.

If a place hasn't been founded prior to 1650, I discount it.  Sineia and Slanic-Prahova, which are near Brasov/Stalin (1952, remember), are examples.  Mind you, this applies according to when I originally searched for the place.  Wikipedia adds data and changes founding dates all the time, so that searching for a place in 2005 might say that it was founded post 1650, whereas an article today might say it was founded in 1161.  I'm not going to check every location every day, so the existence of the settlement depends on when I happen to research it.  Many places, especially in Romania, were searched before Wikipedia was expansive.  I once used several textbooks and encyclopedias as my primary source.  Not that it matters at all how accurate I am.  If I found no data for when a place was founded, I left it in; I only removed places where the founding date was definitely listed as post 1650.

Next I would create a list like the one shown for each region.  This one is for Moldavia.  The upper list in white are places existing in 1650, according to my research.  The lower part in black with white lettering were founded after 1650.  The numbers are their population according to the Colliers Encyclopedia.  "TOTAL" means the total population of all cities.  "Region Pop. 1952" is the population of Moldavia as given by Colliers.

For some regions, occasionally, there would be no population figure.  In those cases, I would determine the average population of those cities below the median.  That is, of that the lower half of cities.  I'd then generate a random number between 1,000 and the bottom-median average.  I chose "1,000" because there were hardly any places on any map appearing in the entire lexicon of Collier's maps.  Of a place appeared on a map, about 99.99% would be listed as 1,000 people or more.  I tend to think those listed with less were typos.

Incidentally, letting the encyclopedia decide what cities were the most important seemed the best course of action.  Someone somewhere that was an expert in their field decided that these were the places in Romania that deserved representation.  I was willing to go along with that.

Next, I used a formula to calculate the 1650 populations for the existing places from the 1952 population.  Take Botosani; it was founded in 1350.  For it, I used the formula (1650 minus date founded) divided by 4000, multiplied by the 1952 population.  I chose "4000" since that represented the year 2,350 BCE.  There were cities that were founded prior to that date, but most of those had been deserted, abandoned or utterly destroyed, sometimes moved, up to or prior to that date.  For example, Jericho was founded somewhere around 8000 BCE.  It had also been destroyed more than a dozen times before Joshua was supposed to have destroyed it.  So, 4000 sounded like a nice easy number.

Botosani had 29,145 people in 1952.  (1650-1350)/4000*29145 = 2,186.

If a city had been razed or plundered or suffered a terrible earthquake of some kind, I reduced this population by 10%.  If it was plundered twice, then the total was multipled by 0.9 x 0.9.  If the city was destroyed, I reduced its population by half.

Here are the populations for Moldavian settlements using this formula. As can be seen, Bacau has more people than Botosani in 1952, but because it was razed twice, and founded slightly later, it's game population is reduced to 1,689.

During my research, I also look to find what name it was called in 1650.  That's why "Iasi" on the left is renamed "Jassy" in my game world.  If I couldn't find an alternate name, I went with the one I had.  "Piatra-Neamt" is probably two places, but I was unable to find when they were merged, so I left it be.

Calculating a settlement's population demands the founding date; so if I had no founding date, I took the average of those I could find.  For example, the average of the above is 1396.  I then divided the difference of this number from 1650 by 36 (which would be 1650-1396 =254, divided by 36 = 7.1).  I then rolled 6d6, multiplied by the product.  Thus, for example, if I rolled a 22 on 6d6, that's 22 x 7.1 = 156, +1396 equals 1552.  This would mean that places for which I couldn't find a founded date would always tend to be more recent (but not always) than settlements with confirmed founding dates.

I am discussing this at length to show the lengths I will go to establish a universal system that applies to EVERY settlement, everywhere.  And as it happens, it works very well.  If a huge city, like Manhattan, is founded in 1626, it gets a population of 11,748 — not far off the real number in 1650.

The next step is to take the total urban population of this last list, 17 908, and compare it with the total urban population of 1952, 402 761.  Notice I include those places that don't exist at all, since their population influences the total population of Moldavia in 1952.  We divide 17,908 into 402,761, and multiply that number by 2,782,182.  This gives us a population for Moldavia of 123,707.

Now, when I research each place, I also search for who the place belonged to in 1650.  All of Moldavia at that time was within the Ottoman Empire; as such, it was under the authority of an Emir.

Part 2

Now we can play some games.  Counting only the world that I've actually designed, using 1952 figures, the largest cities are, showing thousands of population, Berlin (4339), Moscow (4137), London (3390), Paris (2850), Cairo (2100), Vienna (1731), Hamburg (1712), Rome (1583) and Rangoon (1502).  Keep in mind that these are city populations, not metro, and that the way things were counted back then was slightly different from today, particularly with non-European places.  For example, I have a population figure of 1,490 thousand for Bombay, or Mumbai ... which surely has to have left a lot of people off the rolls.  Nonetheless, it's these figures I used to calculate my populations.

The largest cities of my game world are as follows, again in thousands of population.  I'll list the top 20.  Remember, these are according to my calculating system, so don't expect a perfect alignment with the real world in 1650.  I only include this for interest sake:

Paris (941), Vienna (930), London (838), Napoli (590), Barcelona (534), Lisbon (505), Lahore (467), Constantinople/Istanbul (394), Casablanca (392), Cairo (358), Baku (354), Glasgow (326), Rome (314), Moscow (307), Turin (299), Milan (293), Birmingham (288), Bombay (279), Palermo (275), Tashkent (250)

Casablanca, Glasgow, Birmingham and Palermo are the most off, I think ... but really, it is just a game.  English population figures are always quite high, causing me to wonder somewhat about the original figures.  Casablanca was founded in 650 BCE, however, and was the leading city of the large Saadi Empire at the time of my world.  Baku was a major center of the Safavid Empire.  Italy was a heavily urbanised part of the world in 1650, even though it had lost its grip on incoming world trade by then.  All in all, the figures work ... and I leave them here as a suggestion for how large the reader might want to make their own cities.

Because Jericho was an empty field in 1650, it doesn't appear on the list of "oldest" cities in my game world.  Here's a list of the oldest twenty.  All founding dates are BCE:

Amman (8500), Edessa (8000), Gafsa (8000), Corinth (6000), Coleraine [Eire] (5935), Damascus (5500), Hamah (5500), Anantnag [Kashmir] (5000), Jubayl [Arabia] (4992), Qom (4500), Larnaka [Cyprus] (4000), Rodosto [Thrace] (4000), Sidon [Lebanon] (4000), Aqaba [Arabia] (4000), Chur [Switzerland] (3500), Paterno [Sicily] (3500), Qatif [Arabia] (3500), Muttura [India] (3228), Dvaraka [India] (3102) and Asyut [Egypt] (3100).

Not a very spectacular list.  Seemingly, anything really old just didn't catch on, long-term.  

As a last bit of play, counting only cities with an adjusted population above 5,000, here are my world's newest cities, date only (the largest of these is Bucharest, with 24,873).  Again, might as well include the top 20.  All dates are AD:

Hyderabad (1591), Tsaritsyn [Stalingrad] (1589), Samara (1586), Saratov (1586), Voronezh (1585), Amritsar (1574), Ivanovo (1561), Khortytsia [Zaporizhzhia] (1556), Liverpool (1550), Helsingfors (1550), Colombo (1517), Ahmadnagar (1494), Rawalpindi (1493), Surat (1490), Bikaner (1486), Las Palmas (1478), Hacibey [Odessa] (1466), Sarajevo (1461), Bucuresti (1459) and Jodhpur (1459)

Waste of time, really, since I can't say for sure that any of these founding dates are accurate ... except that they are absolute fact in my game world.  All I can do is hope this sort of information provides some idea of how old or young your game cities might be.

Do remember, all these lists are for parts of the world that I have mapped and calculated regions for.  Manhattan doesn't appear on the list because although I have sort of sketched out the area, I haven't yet incorporated it into the game world.

Part 3 

One problem the reader will have in designating settlements is to answer, "How many?"  This isn't easy to answer.  The world as a whole is inconsistent on this matter.  The best I can do is to offer a series of examples and hope this imparts some of my experience to the reader.

All I can do is give examples:

Andorra, an existing principality and one of the smallest countries in the world, exists in my game world also.  It has one settlement, Andorra la Vella.  As I count my areas in 20-mile hexes, not square miles or kilometers, my Andorra covers 1.5 hexes.  The real Andorra is half that size.

On the other hand, the Duchy of Apuania in Italy has two settlements, Carrara and Massa, which are very close together; the duchy has an area of 1.3 hexes.

Astrakhan, on the other hand, is a large very poor soiled region of 117 hexes in area.  It has two settlements: Astrakhan the city, with 12,841 people, and Langan, an outpost with only 182 people.

Buraydah is an Emirate in the heart of Arabia, 200 hundred miles or so south of Riyadh.  It has two settlements, oases, Buraydah and Qusaybah; the region covers 67 hexes.  It is a large desert area.

Sumi, the same city in Ukraine discussed in the news, is an independent het of 3 hexes; Sumi is the only city.

Nord-Trondelag in north central Norway covers 8 hexes and has two settlements, Levanger and Stjordalshasen.

Tyan-Shan is a monastical entity in upper modern-day Kirghizia; it has one settlement for its 44 hexes.  It is a barren frozen country some 12 to 14 thousand feet above sea level.

I guess we could say that regardless of the area, civilised places with a single settlement, or even just two, will be very tiny in size ... whereas large uninhabitable places will also have a minimum of settlements, just one or two.  All of Ireland in my world is carved into tiny pieces, with each individual settlement representing its own clan and, effectively, it's own country.  These pieces gang together to attack larger enemies, but then fall back into disputing every rock and rill that lays on their perceived boundaries.

The counties of England are relatively small, with between 2 and 10 hexes.  Cheshire, Guildford and Nottinghamshire each have 4 settlements.  Monmouthshire, Lothian, Hertfordshire, Elgin and Dorsetshire each have 3.  Angusshire, Fifeshire, Glamorganshire, Hampshire and Somerset have 5.  But many English counties are stuffed full of settlements.  Yorkshire has 16.  Sussex has 9.  Lincolnshire has 10.  Lanarkshire has 11.  These last are much larger, are heavily populated and industrialised by 1952 standards ... so the encyclopedia crowd wanted to include many more of these cities, cramming them into the map.

A different problem occurs with France.   Because the departments were established after the Revolution in 1689, I used the original provinces ... which are giant areas compared with English counties.  Thus, Languedoc in the south of France has 51 settlements scattered over 38 hexes.  The Ile de France, including Paris, has 24 settlements pounded into an area of 13 hexes.  Forez, including Lyons, squeezes 21 settlements into 9.3 hexes.  And Brittany, a sprawling backwater area of 35 hexes, has so many ports stuffed around it's coastline that it nonetheless has 52 settlements.  And 15 of these are described as "market" towns.  Which I adhered to with my trade tables, giving each the ability to assign its own local price.

It's slightly better in Asia, but this is because the population is predominantly rural.  Hindustan is one massive region under the Moghuls, of 187 hexes ... but it still has 35 settlements - and an adjusted population of 14,144,861 ... which means many, many villages that don't appear on the map.  Odisha has 40 settlements scattered into 281.4 hexes.  The Punjab has 40 distributed into 290 hexes.  Kabolistan, the region around modern Kabul in Afghanistan, has 17 settlements withiin 114 hexes.  

It's difficult for me to say there should be such and such many settlements per hex, since the number for me is dependent on too many variables: the number of settlements actually included on the maps, for instance, or the tendency of a region to urbanise.  Or the size of the regions in different parts of the world.  The reader simply has to guess at an earth-region according to its climate, vegetation, culture, etcetera, count the cities in a square of latitude and longitude and decide that this is the right number of settlements to throw in.  But there is no right or wrong answer.

I can say that one settlement produces a very dull infrastructure map.  Two settlements or three help create shadows and dense places.  Ten are even better, especially if there are about one settlement per three hexes.  But that shouldn't be every place in the world, or the distribution will end up being the same everywhere.  Some places ought to have one settlement per 30 hexes; and others should have four settlements in one hex.  That's just how it should play.


  1. There are few things as exciting as watching the infrastructure populate a hex map. The little roads and trails spreading out, thinking about why this cluster is unpopulated and why that one is, and what that means in relation to this feature here or whatever. Inventing problems to be solved, solving problems you invented.

    The ideas that percolate during this process always make me really excited to run a game.

    The settlements are difficult to engineer for a non-Earth world because the whole system is taking the snapshot of the game world at that exact moment in time and then inventing the explanation for it after the fact. Constantinople is a settlement because Constantinople is a settlement, and the infrastructure merely measures what that means for the area surrounding it at this moment.

    Why are these areas settlements? Because they are. They're the nodes by which the whole system operates, and aren't produced or affected by anything inside it. If you want non-Earth nodes, you have to find non-Earth reasons for them to be nodes.

    It makes an awesome world to play in and develop, but damn if it isn't hard to duplicate the efforts in other worlds.

  2. I grant you that, Pandred. This is the reason I run the earth.

  3. Wonderful. Now I just have to plan yet another long dip in databases of cities and dates and such ^^ .

    Alas, it's indeed a sad thing that there is no easy way to place settlements in a non Earth world, but quite understandable.

  4. Would the population of a settlement (which is admittedly small compared to the cities) be taken out of the rural population of that hex? Eg, if the rural population of a hex is 10,000, and a village of 350 is generated, are there now 9,650 rural population? Or is it a separate number entirely?

  5. Careful, Shelby. A "settlement" can be an enormous city. Paris and London are settlements.

    Note that the system carefully fails to establish the population of a hex in any way. I have the population of a settlement and the population of the whole province, but not of any specific hex.

    Therefore, yes, the population of the settlement is included in the hex's population, whatever that might be. With a really big settlement, like Paris, the city extends across many 6-mile hexes, most of which will be type-1.

    But what would knowing the exact population contribute? We have the wealth of the hex, which determines its taxable income. I'm working on a measure of its facilities based on hammers. We have a comparative measure for produced food. I suppose we need to know how much food mouths consume, but since there's no standard for how many calories everyone in the population lives upon, a population number alone wouldn't help with that. Logically, we'd have to base food consumption on hex type and hammers, not actual people.

  6. And, for that matter, why would the players of the game ever have to calculate food supply unless they were personally responsible for its distribution? That's the only possible figure the DM ever has to provide - the NPCs under the player's authority. We don't need a number for everywhere in the world.

  7. Hey, sorry, this isn't relevant to the post but I can't find a better place to ask:

    How, roughly, do you get coastlines? I've started doing a version of Earth too, starting in Estonia circa 1230, but have realized that the method of screengrabbing from Google Maps, overlaying a hex grid, and eyeballing from there is doomed to failure past the first arbitrarily-selected region. Do you just find places along the coast, then use their latitude and longitude to find where it's meant to go from there?

  8. Have a look at this:

    Coastline Making

    May be a bit dull, but this is how I make coastlines.

  9. Thanks heaps for the link! It'll be good to finally be able to move up into Finland, since I've got a good while before the game is expected to hit the table.


If you wish to leave a comment on this blog, contact alexiss1@telus.net with a direct message. Comments, agreed upon by reader and author, are published every Saturday.

Note: Only a member of this blog may post a comment.