How to Construct Valid Geographical Entities

Those Who Ignore Geography
Shall Shiver in the Dark

After many years of creating my little private world and (paradoxically) sharing it with others, there are still a few little details that I have never quite managed to make clear. One of the most outstanding of these is, of course, why do I do what I do.

It’s certainly not the money; there seem to be quite a few people who have made more money from my original work than I have. It’s not the fame or critical acclaim, after all, strangely enough, there are a considerable number of people who, while they may have heard of my little world, have never heard of me… they may even think my world was designed by some corporate committee somewhere… go figure.

There are even a few folk out there who think my work is public domain and they are free to copy it, amend it or even share it with others! Why I keep creating and refining my little world in the face of considerable adversity and relatively little appreciation, is one of the “little details” that I have never made clear even to myself. The best I can do with this sort of thing is put it aside for later (on the assumption that there might not be a ‘later’).

Another little detail is ‘what it’s all for’, or if you prefer ‘what are we supposed to do with it?’ This one is easier. My basic position is that HârnWorld is a very personal work of art: you experience it. You can use it for roleplaying if you like, but this is certainly not a requirement. I’m perfectly happy if you buy it and simply ‘admire it’ or, at least, appreciate it. Of course, that’s the biggest ‘little detail’ of them all, or at least, I hope the easiest to explain.

From talking to fans of my work over some thirty odd years, I get the impression that an awful lot of these good folk are missing a lot of what’s in there. When it comes to environment, I am, fundamentally, a fantasy cartographer. It is my unshakable belief that geography must precede everything else. This is more complicated than it sounds, but I’ll try to explain it now.

The first premise is that we want things to work ‘realistically’ in the sense that we are not going to rewrite the laws of physics just because we don’t know how the real world works. In other words, unless we have a sufficient, credible, and fully explained reason we are not going to rewrite any fundamental, universal laws; to do so would be an insult to any intelligent reader and demonstrate our own ignorance. Unfortunately, this leaves a lot of excellent fantasy writers in need of geographical or ecological advice. On the other hand, what are these fundamental laws?

Well they vary a great deal. Physical laws involving gravity, conservation of mass and energy, entropy, things like that, are pretty well standard (unless you change them for some very good reason and explain why). To be fair, magic systems pretty well have to suspend physical laws (or at least sneak around behind their backs) but the effect of magic cannot be pervasive and can rarely be permanent or even enduring. This might in fact be considered a ‘law’ of magic… something about magic being defined as an exception to the norm. In this narrow sense, miracles can be lumped in with spells (there are very few cases where this is true). So if we come across a river that flows uphill, or some streams that cross each other, there are two possible conclusions: (1) there is magic or divine intervention involved and, if this is the case, whoever brought about the anomaly must have had a bloody good reason… and let’s hear it right now, or (2) the cartographer was a bit thick. To be fair, lots of these cases occur when someone who didn’t understand the basic ‘laws’ was called upon to complete a task beyond her/his capacity. Such things can be considered ‘forgivable mistakes’ but they should be corrected as opportunities arise.

Obviously, with projects like my original Hârn Map the scale and design is such that it’s difficult to commit ‘thick’ geography crimes/errors. I carefully left out contour lines (because they are useless to GMs on this scale) so it’s hard to argue that such and such a river is flowing uphill. It’s when someone adds contour lines or develops the maps to a larger scale to add streams, smaller villages, and the like that they run into trouble. It’s not as easy as it looks. Let’s look at a few ‘geographical’ laws that often seem to be ignored by cartographers who are ‘inventing’ terrain or ‘filling in details’. Some of these are physical geography, but most involve human geography.

People and animals drink water. No matter how bad an idea this might be, and no matter how many people convert the water into something more stimulating first, water is the most essential requirement for human settlement. No one is going to found a settlement that doesn’t have a secure source of water. In most cases, a well doesn’t count. You have to dig a well and that takes time. You don’t know ahead of time whether the well will succeed and you are not going to risk the lives of your children or even your livestock on the off chance that you’ll be able to discover hidden water. (There are exceptions of course, but at least 99% of all settlements are going to be on streams or rivers if there are streams or rivers available.) Cartography that puts a settlement off a stream when there is a choice, is ‘thick’ cartography.

By the way, the principal that settlements will be adjacent to streams may be particularly true on the coast. Near the sea, well-water tends to be brackish, unpleasant or even dangerous.

The amount of water on the ground should reflect the amount of water in the air. Well, of course it should! In general, there are going to be more streams and rivers on the windward side of a major ridge near the water than on the leeward side, and somewhat less on the steppes in the middle of the continent.

The presence of a settlement presumes its survivability. This seems obvious, but it’s amazing how often this rule seems to be ignored. All it means is that if a settlement exists somewhere, then it has survived any and all previous local threats.

Settlement spreads according to geographical and political factors. This should also be obvious. This includes such factors as the availability of and security of fertile land as well as social pressure to emigrate from established settlements. This may be so complex as to daunt any cartographer, but must not be ignored.

Natural Vegetation is determined by a topography and climate. Obviously, but again one sees some alarmingly distressing cases where this is ignored. The problem mainly comes up when a region is ‘detail mapped’. South facing vegetation may differ from north facing, windward from leeward and so on. Some vegetation may seem arbitrary, but such cases should be the exception rather than the rule.

Settlements tend to first appear on fertile flat land. Obviously, no one is going to try ploughing a mountain or a desert when there are freely available fertile flatlands. There’s a lot more to it than that, of course, and the cartographer has to be aware of the why and where if s/he is going to break the rule.

"Cute" is a rare treat. Every fantasy cartographer has at least considered the idea of a lake in the middle of an island in the middle of a lake in the middle of an island, but any cartographer worth even half her/his salt knows that such geographic ‘escapades’ must be rare if they are to be properly enjoyed. How rare? Well never is a safe bet, more than this is a bit of a risk.

Lead us not into temptation. Just because we suddenly discover a new type of topography doesn’t mean we have to use it every time we make a map.

If you can’t do it right, don’t even try. I despair at the number of times I have seen things like contour lines abused and mis-deployed beyond all reason. To be fair, contour lines are difficult. They can be very useful at certain scales of map, depending on the topography, but ultimately, over a large, topographically ‘interesting’ region (say something as big and hilly as Wales) there is no way to select a reasonable contour interval. This is why so many real cartographers use one contour interval in the flatlands and another on rough terrain. On most of the fantasy cartography I’ve seen the contour lines have been ‘unconvincing’ at best and impossible at worst. Even I, with 40 years of experience in fantasy cartography, am hesitant to include contour lines. The basic principle here is that if you don’t really know how to do something, get some help or leave it out; your map will be better off.

Because I get a lot of questions about contours, I’ll provide a couple of maps of the area around Aléath to show the difference between a contour map and a relief map.

On the left is a vegetation & contour map of the area using a 1000’ contour interval. Note that there are no contours on the map since none of the land is above 600’. If I used a 500’ interval there would be one contour. If someone were making a contour map of Hârn and wanted to use a uniform contour interval over the whole island, s/he would pretty well be stuck with a 500’ or 1000’ interval. The interval is dictated not by the area around Aléath, but by the mountainous regions. With a 1000’ interval there would be 9 contours and with a 500’ interval there would be 19 contours, but in most settled regions there would be one or two contours at most.

Even on the (contourless) contour map there is a little bit of relief (because I am using textured vegetation patterns).

On the relief map (on the right) a clear ridge can clearly be seen running across the middle of the map from west to east. It is also possible to make out some interesting smaller ridges in the heaths in the west and northeast as well as along the central large ridge. None of these features are apparent on the contour map. While making a relief map may require somewhat more expertise than simply adding contours, ultimately it is a far more intuitive process and easier to execute. Ultimately, for the viewer, such systems provide far more detail and are much easier to read.

N. Robin Crossby

Keith Mann
Keith Mann's picture

Great insights...thanks!

Thanks for taking the time to write this, Robin. As someone who was drawn to Harn (25! years ago) by a love of maps, it's wonderful to read your thoughts on this issues.

Peter the skald
Peter the skald's picture


Hi Robin,

It is clear from reading this that I do not know what a contour map is! I thought a contour map was a map with contour lines on it!

The two maps you provide as comparisons look very similiar to me at first glance; but on closer inspec there are subtle differences as you say. What exactly is a contour map and a relief map..I have already paraded my ignorance on contour maps, so here goes with it where there is shading to indicate height?

Yours awaiting enlightenment, Peter.

Robin's picture

Contourless Contour map

Sorry, I was being 'clever'. The map on the left is has very little relief and no contours.

The reason for this is that the contour interval is either 500 or 1000 feet, and there is no land above 500 feet anywhere on the map.

I was trying to illustrate the point that, if you select a 'reasonable' contour interval for the whole island, then you can barely show any topographical detail at all, whereas, with relief mapping you can show a great deal of detail regardless of the local elevation.

Sorry about the confusion :)

I suppose part of the problem is that I have designed several completely different mapping systems now, and only one or two of my maps (local maps) have ever shown any contours ;)

Peter the skald
Peter the skald's picture

Hi Robin again, This contour

Hi Robin again,

This contour interval of 500 or 1000 feet...where is this from? On my Ordnace survey maps the interval is 10 meteres...although they would be a different scale.

I would like to see a comparison of the same area done contour style versus relief style..even if only to show that a smaller contour interval was, well, unreasonable.

I agree that if the choice is no contours versus relief; relief is better.

Robin's picture

Ordinance Survey Maps

I grew up on OS maps... I loved them as a child growing up...

There were 1":mile and 6":mile. Wonderfully detailed.

Yes, at these scales you use much tighter contour intervals. Hmmm... 10 metres sounds a lot like 33' ?

Did they redo all the OS maps in metric measurements? That must have been a task and a half.

Let me think... the grid map of the Aleath district is um... 62.5 miles across and um the square (at full atlas scale) would be 9.8 inches across... that's 0.1568 inches to the mile?

I'm beginning to forget where I'm going with this, but it seems like you couldn't really do a meaningful OS map at these scales?

So we have to look elsewhere for contoured maps at this scale... (and, frankly, they are hard to find).

The assumption I made about contours is this: that for mapping all of Harn at this scale you first consider how high you have to go... that's about 10,000'. Then you think about how many contours you can cram into the mountainous bits (the only bits where you'll need to use all the contours in close proximity). So it's pretty hard to have more than 10 or twenty contours so show a few mountains (in fact even at 10 or 20, you could end up with solid brown bits).

In order to show a hill in the 'flatlands' you need to add contours. The problem is that every time you add a contour to the flat bits, you also have to add it to the mountainous bits too. (Although some maps have taken to using a different contour interval in the mountainous bits than the flat bits... these tend to be overwhelmingly confusing, since the 'moderately' flat bits may have more contours than the mountains.)

In other words, if you want a 33' contour (for example) in the flat areas (like around Aleath), that means you need to have more than 300 contours in the mountains (well, the biggest mountains anyway)... and that won't work at all.

Even with a 33' contour interval, I can *still* show more detail with relief!

And BTW, while it is theoretically possible to demonstrate the sheer folly of this many contours, it's a lot less work to leave it to your imagination ;)

Peter the skald
Peter the skald's picture

Hi Robin, Yes, the OS maps

Hi Robin,

Yes, the OS maps are all redone in metres/KM now...I agree, a monumental task but no doubt computers made it...well possible if not easy!

My OS map of the plymouth area (UK) is 1:50 000; which is 2cm per kilometre, about 1.2 inches per mile. It opens out to about the size of a broad sheet newspaper and covers an area of about 24 miles by 30 miles.

This means it covers about half the area of the maps you presented I guess...

The contours on this scale work very well, even on my welsh maps (on which they laugh at 300 contours being too much!!!!) which cram them in as you would expect. I am of the opinion that to double the area covered would not create a folly of presentation (one could use 20m increments and it would solve any further cramming); but it would present a folly of endeavour!!!!!!!!!!!!!

In essence I dispute that relief maps show more than contour maps on the scales presented.

However, on much larger scales I think relief maps are the only way to present things; much like in the much loved world atlasses of my youth. To put contours on 'nation' or continent depictions would be folly...

Robin's picture

The Sheer Folly of Hopeless Endeavours

You certainly struck a chord of reason.

To actually draw a map with 300 contours would be a task of such horror that I would not consider it, not even for obscenely huge piles of cash. To undertake such a task 'by hand' even using one's computer to do all the 'work' would take thousands of man hours... and I'm quite good at drawing contours when I want to be. For someone who is not as 'familiar' with the concept, the task could well be fatal (to him or, more likely to the map).

And this is why, in the end, together with the other reasons, I'm not including contours on these maps.


Keith Mann
Keith Mann's picture


I can't imagine anyone drawing such a map from scratch either, Robin. Terran cartographers have the advantage of a physical planet to work from, which makes the task possible, but even then I wonder how some pre-computer topo maps ever got drawn.

I have, however, toyed with the idea of writing some sort of algorithm to generate digital elevation models given a set of spot heights and certain things like watersheds to work from. In fact, didn't Auran have something of the sort in the works at one time?

Robin's picture

Auran Elevation mapping

Auran createda fairly low res 256 greyscale elevation map (from which they rendered those nice landscape postcards... although there was a substantial exageration on their mountains... several miles high and whatnot).

Greg had a 1 metre / pixel (and probably more than 256 level) map planned... but, sadly, never actually got it done.

I have experimented with 'shading/elevation' software packages, but, frankly, when working from scratch, I have never found a 'usable' interface.

One possibility, to which you allude, is some clever kind of scanner that can read my V&R (vegetation and relief) maps *as if* they were satellite photos and produce shaded gradient maps (and eventually full 3D renders and topographic maps).

Another possibility is software that could read the pixel density on my various layers and calculte an approximation of elevation from that.

But, these seem like dreams for a future that is growing shorter.

Here's the good news. I have completed the V&R mapping for all of Hârn all of Shôrkýne, about half of Ivínia and a sixth of Tríerzòn. If I push hard enough I might get a few more regions at least mapped out as far as I can...

Keith Mann
Keith Mann's picture

DEM from V&R

I'll hack a bit and see if I can come up with something to show you.

Jack's picture

Where's the download?

The text is certainly nice and I'm really looking forward to the completion of Chelemby (and I'd love to learn more about Hurisea), but I'd buy the V&R maps as stand-alone products. I can make details up as I go along, but despite my love of cartography, I can't make a decent map to save my life . . .

More releases along the lines of the Atlas Kelestia Folio 1 (E5 and 6 through H5 and 6) would be great!


Old style heraldry: Sable, the pale argent.

New style heraldry: Oreo, resting on edge.

Neil's picture

6" to 1 mile


As Peter has said, the old 1” to 1 mile (1:63360 scale) maps were replaced by the new (metric) 1:50000 series. The 6” to 1 mile (1:10560 scale) maps were replaced by 1:10000 scale maps. In the past 10 years these maps have been digitised and are regularly updated by the O.S. using satellite surveys. The map I’m looking at (at work) shows a housing development which was only approved 3 years ago and shows the partially finished roads and completed dwellings.


- "Pardon me for living, I'm sure."

-- (Terry Pratchett, Mort)

User login

Recent comments

Keléstia Connect

Who's online?

There are currently 1 user and 17 guests online.

Online users

  • Alex Greene

© 2014 Keléstia Productions Ltd. and N. Robin Crossby (1954-2008).
The opinions expressed on this website are those of their respective owners and do not necessarily reflect the views of Keléstia Productions Ltd.
Trademarks are the property of their respective owners.