Not a lot of time over the past couple of days to work on this, but I’ve managed to tweak some of the landforms a little. The planet resembles a mirrored Earth a bit less now. I’ve also started painting altitude contours on the map. So far, just the very highest peaks, the Andes- or Himalayas-equivalents, but the next few layers should cover all of the land areas with colored zones to indicate altitude.
A few hours of work this evening, while I had Wonder Woman playing in the background, and I ended up with a decent set of land-masses for my world map.
I seem to have reinvented an Earth, although flipped east-to-west. That not-North-America stands out in particular, and all those island arcs in the far western not-Asia are kind of reminiscent too. It makes sense, I suppose, since plausible plate tectonics aren’t going to generate completely arbitrary shapes.
There are differences too, of course. The pseudo-Atlantic ocean is a bit wider, and the continents are in general separated by stretches of sea. There’s a narrow gap between the not-Americas, and instead of a Mediterranean Sea there’s an open ocean between the not-Africa and the not-Asia. That’s going to do some interesting things to ocean currents, I think.
No matter. The actual stories I intend to write are going to be on a much smaller scale, so if the layout of the continents looks a little derivative, that won’t be obvious to my eventual audience. What’s important right now is that I’m reasonably satisfied with this layout, so I can move on to the next steps again.
One of the major stumbling blocks with world-building, at least for me, is that even when I’m momentarily satisfied with the outcome of a task, it doesn’t take much to rob me of that satisfaction. In this case, while staring at my world map draft in progress, I began to compare it to both the real world and to other world-builders’ efforts, and found it lacking. Too crude.
So I’ve gone back to first principles and started over, this time rebuilding a map of tectonic plates without pre-designing any of the continental land masses. This time I strove to come up with something to resemble the general pattern of tectonic plates on the real Earth, at least as far as the number of major and minor plates was concerned. I also paid attention to the way plate boundaries are arranged – whether they tend to be convex or concave, and how they form seams and three-way intersections.
One thing I found useful was to simply mark off the polar regions and ignore those. One of the things that was giving me fits was the transformation from a flat projection to the globe and back, and that switch always introduces the most distortion close to the poles. By assuming there will be no major polar land masses, I can gloss over how any plate boundaries might be laid out in the arctic or antarctic regions.
The result (equirectangular projection only) is as follows. So far, so good. I haven’t marked continental plates yet, but there will be five major continents and a few minor land-masses and island arcs.
Next step will be mark out the relative movement of plates at each boundary, and then sketch land-forms to match.
Just a quick report today: progress on my world maps for the Curse of Steel project. After tinkering a bit and learning how to build and use layer masks in Photoshop, I managed to paint mountain belts in their own layer on my map, with the following results:
Here, the deep-red belts are “young” mountains, the result of recent orogeny at the site of plate collisions or subduction. Think the Andes, Rockies, or Himalayas. The narrow, golden-brown belts are “old” mountains, the eroded remains of ranges that formed many millions of years ago in previous orogenic periods. Think the Appalachians or Atlas range.
One thing strikes me: the big continents to the east are going to have really big rain-shadow deserts, since those young, high mountains are going to block any kind of monsoon climate from moving too far inland. I’ll have to figure out the air circulation patterns next to know for sure. It makes sense, though, since large continents tend to have big arid zones anyway.
Next, it will be time to work out those climate patterns. I’ve been reading up on techniques for that all week, and the long weekend coming up should be a good time to work out the details.
Had the day off sick today, so in between bouts of ick I got a bit more work done on the world map for The Curse of Steel. Mostly this involved refining the landforms, using a much finer pencil stroke to create crinkly coastlines and islands. I’m fairly happy with the results. Here’s the equirectangular base map:
Much better continental shapes, not so cartoonish now, and clear island arcs. Another view, in the Mollweide projection for variety:
Next step will be to lay out mountain ranges, in accordance with the underlying map of tectonic plates. Once that’s done, I’ll need to work out air and ocean circulation patterns, and then lay out climate zones. Then it will be time to drill down to the regional scale and build the maps I’ll need to support the story.
Okay, given my level of frustration over the weekend, I’m rather happy with today’s developments. I’ve managed to produce a very rough draft of my world map, using Photoshop, the GPlates software, and GProjector. By no means is this as detailed as a good map of Earth yet, but I’m reasonably satisfied with the realism of the planetary geology involved.
Here’s a flat map in equirectangular projection:
This planet is in the middle stages of the breakup of a supercontinent. An Atlantic-like ocean has opened up, breaking off the equatorial continent and sending it south and west, creating a nice long chain of island arcs along the edges of two subduction zones as well.
The big continent that covers the north polar region is actually made up of three major continental plates. The piece covering the polar region itself is one plate, then a second is in the process of breaking away and heading southward, with a rift valley and a newly opening ocean basin dividing them. The third piece, down in the southern hemisphere, is actually a separate plate that started out attached to “Equatoria” but found itself divided from it by the new mid-ocean ridge. It’s currently being driven east and north, and is probably forming a blocked-off sea basin or an impressive range of mountains (or both) along the point of contact with the larger land mass.
The blot of land in the middle of the pseudo-Atlantic is my equivalent of Atlantis (or Númenor), the home of the most advanced human culture on the planet, one which is just starting a period of sea-borne exploration. The land-form is basically a super-Iceland, an exposed piece of the mid-ocean ridge that has a magma plume under it. Lots of volcanism and hot springs, and the inhabitants are feeling crowded enough that they’re ready to sail away and find primitive lands to colonize.
For variety, here’s a two-hemisphere orthographic map, produced using GProjector:
I did mention that this is a very rough draft map, right? I think I may produce a somewhat more detailed version of this map with Photoshop first, so I can add mountains and other major land-forms, then work out ocean currents and climate zones. Then it will be time to drill down to the specific region(s) that will appear in the story, and use Photoshop or Campaign Cartographer to put together finely detailed maps for those.
How did I get through this in just a few hours, after struggling all weekend? As often happens in world-building, the secret is finding the right workflow.
For a couple of days, I was using the GPlates software to try to draw features on the sphere. Problem is, although GPlates is perfectly good for that, that’s not what the software is actually designed for: it’s a very sophisticated plate-tectonics simulator. So by using it just to sketch features, I’m ignoring 99% of the thing’s functionality – and some of that functionality very much gets in the way. I was spending most of my time juggling multiple raster files, and fighting the very elaborate system GPlates uses to save projects, and getting frustrated with the results.
So today I switched my workflow around. Rather than do any drawing in GPlates, I did all of it in a Photoshop document with three layers (one each for ocean, tectonic boundaries, and land-masses). I would draw a few features, then save the result as a PNG image and import that into GPlates, purely to see how it looked on the sphere. More often than not, I would spot absurdities on the sphere that weren’t obvious on the flat map – so I would go back to Photoshop, fiddle with a few lines, and then re-import the result back into GPlates. I never tried to save anything in GPlates, so I never had to deal with its weird file-management system. Fifteen or twenty iterations later, I finally had the planet divided into a reasonable set of major tectonic plates, I knew where the major mid-ocean ridges and subduction zones were, and I was ready to finish the sketch map here.
I’ll take my progress where I find it.
For the record, trying to develop a fictional planet’s geography from the plate tectonics up is a royal pain in the nether regions. I begin to see why most people just call up a noise-driven fractal map generator and call it a day.
I will persist. At the very least, I’m learning a great deal about how plate tectonics actually work. I think I may cave in and go see if anyone has developed a more detailed work-flow to make sense out of this.
A short note, since it’s been a few days since I last posted anything here. I’ve been up to my eyebrows at the day job, teaching a course on risk management and cybersecurity. After a full day on the platform I’m rarely in good condition to get a lot of creative work done in the evening. Still, my brain has been percolating along on the Curse of Steel project.
I’m currently beginning work on some maps, to give the story some structure. The overall plot of the novel is very much in the “heroic quest” vein, with Kráva and a few companions going on a long journey across unexplored and dangerous countryside to reach an objective. So I need to at least sketch out the geography.
This, as usually happens with me, turns out to be more complicated than it might appear at first glance. Knowing too much about world-building often means you can’t be satisfied with the simple or naïve approach to any problem.
In this case, my brain got stuck on the question of how to draw regional and world maps on a sphere. I keep thinking back to the classic Baynes-Tolkien poster map of Middle-earth, which has been the inspiration for a hundred thousand fantasy-world maps since then. It’s a beautiful map, but the big unspoken problem with it is that it’s flat. The map legend indicates both constant directions and a constant distance scale, and that just cannot be done with any flat projection of a spherical surface. That’s a subtle flaw in the world-building for Middle-earth, especially if (as I suspect) Tolkien did his meticulous measurements of distance and travel times on a similarly flat map.
So, since this piece at least of my world-building is decidedly in the same mold, I want to draw a similar map – but I want to envision my world as a sphere and do my regional map-making on that basis. Which means I need to expand my cartographic tool set.
I usually do map-building with Photoshop, but it’s a challenge to draw on a sphere with that tool, and there’s no way to easily do the standard map projections. However, I’ve recently come across one of the superb world-building YouTube videos produced by Artifexian, in which he discusses a work-flow he’s developed to do just this kind of thing. Here’s a link to the specific video I’m talking about.
So I’ve gotten started on this piece of the project by downloading a couple of freeware tools (GPlates and G.Projector), and will be sketching out global and regional maps over the next few days. I’ll post some of the interim results here.
One of the occasional pitfalls I see in genre writing is the awkward use of constructed vocabulary, usually in the production of names, sometimes in the development of bits of exotic dialogue. This is usually to suggest the living language of a fantastic culture. Unfortunately, many authors are careless about this and seem to come up with their constructed vocabulary at random, so we end up with “Qadgop the Mercotan” or something equally silly. (Five kudos to anyone who recognizes the source of that name, which did in fact appear in a piece of genre fiction. At least in that case the author was trying to be silly.)
The world-building challenge is to produce an actual constructed language from which names and bits of vocabulary can emerge organically. There’s something aesthetically pleasing about this when it’s well done. The human brain seems to recognize the internal logic of a well-constructed language, even if we’re not fluent in it. J. R. R. Tolkien, of course, was the past master at this, but a lot of other authors (and hobbyists) have had a crack at it over the years.
For The Curse of Steel, I’ve decided to build at least one constructed language, mostly for naming purposes. Since I tend to insist on doing things the hard way, I’m actually building an “ur-language” and producing my primary language by applying a consistent set of sound-changes. In the back of my mind, I have half a thought that I may need a second constructed language, one that feels related to the first, rather as (e.g.) Greek and Latin are both members of the Indo-European language family. If and when I go that far, I can generate words in the second language by applying a different set of sound-changes to the ur-language roots I’ve developed.
The past few days have been fairly productive in this area. I seem to have finally developed a work-flow that actually functions, without getting me snarled up in unnecessary details of semantics, grammar, or phonology. In particular, I decided to write some text in English and “translate” that, developing new vocabulary and bits of grammar as needed. At the moment, I have about sixty words of vocabulary, several rules of inflection and word morphology, and about a page of notes on semantic structure. Enough to produce an actual paragraph of text:
Esi degra tremárakai múr kresdan. Esi kráva degraka bendír. Augrinír tan esa nekám velka devam. Enkorír skátoi taino. Antekrír skátoi tainmuro, dún begrír tan múr bákha. Vóki degra velka kresdani, dún tarthámi da skátoi. Verti kráva ked saka kó márai. Asgáni skátokai kestan, dún verti dó an atrethen degra. Rethi kráva arekhton saka padír, dún verti sa múr skáto. Dághi kráva aspera rethen skátoka klávo; esi dó kresdághen, dún esi dó degraka danpreta.
A rough back-translation into English would read something like this:
Lion was a great warrior of the Mighty People. Raven was Lion’s daughter. One night they visited the Wolf-clan. Orcs attacked the hill-fort. The orcs broke into the stockade and threatened to do great harm. Lion summoned the Wolf warriors, and opposed the orcs. Raven slew many with her bow. A chieftain of the orcs came forth, and slew Lion in single combat. Raven fought to avenge her father, and slew the great orc. After the battle, Raven took the orc’s sword, as a spoil of war and as Lion’s weregild.
You’ll recognize that as a one-paragraph summary, in pseudo-epic style, of the first chapter of The Curse of Steel, posted a few days ago here.
A few notes:
The convention in this language is to tell stories in the present tense, which is how the untranslated passage is written. In English, of course, narrative is normally framed in past tense.
The language has a very strict verb-subject-object (VSO) sentence structure. VSO languages are uncommon, although not unheard of; notably, many of the Celtic languages use that structure. It seemed appropriate, since I have a sense that Kráva’s people resemble the ancient Celts in many respects. Using a very strict word order helps with the design, since strongly positional languages don’t need quite as elaborate a system of noun or verb inflections.
I’m using a system of word roots very similar to the reconstructed Proto-Indo-European vocabulary, although in most cases I’m deliberately selecting different roots. The result should be a language that sounds as if it would be at home in the Indo-European family, without actually bearing more than a superficial resemblance to any one IE language.
A few pieces of vocabulary I’m rather pleased with:
skáto “orc” is from a word root that means “to hate,” with a noun suffix that implies a “thing” rather than a living creature or human being. Essentially, a skáto is a “thing that hates,” and notably not a person that hates. Yes, Kráva’s people really don’t like orcs.
There’s a whole vocabulary around the word kresa “war,” including kresdan “warrior” (or literally “war-man”) and kresdághen (“plunder, spoils,” literally “war-taking”). Some cultures have a hundred words for snow, but I suspect Kráva’s people may have dozens of words for armed conflict.
arekht- literally means “to set straight,” but it also carries the meanings of “to make right,” “to carry out justice,” and “to avenge.” Which probably is another clue about this culture. Related to that is the word danpreta “man-price,” or more appropriately “weregild.”
Now that I’ve been able to produce one paragraph, I can probably develop more as needed, hanging more bits of vocabulary and syntax onto the partial framework I have. I think the next piece of this project will be to start assembling a map for the story, and coming up with names for terrain features and settlements. Not sure whether I’ll do that immediately, or get back to working on Architect of Worlds again . . .
This is the last step in the design sequence for star systems – once the user has finished this step, she should know how many stars are in the system, what their current properties are, and how their orbital paths are arranged.
At this point, I’ve finished the current rewrite of the “Designing Star Systems” section of the book. I don’t plan to make any further mechanical changes to that section, except to correct any errors that might pop up. The instructions and other text might get revised again before the project is complete. A PDF of the current version of this section is now available at the Sharrukin’s Archive site under the Architect of Worlds project heading.
This step determines the orbital parameters of components of a multiple star system. This step may be skipped if the star system is not multiple (i.e., the primary star is the only star in the system).
The procedure for determining the orbital parameters of a multiple star system will vary, depending on the multiplicity of the system.
The important quantities for any stellar orbit are the minimum distance, average distance, and maximum distance between the two components, and the eccentricity of their orbital path. Distances will be measured in astronomical units (AU). Eccentricity is a number between 0 and 1, which acts as a measure of how far an orbital path deviates from a perfect circle. Eccentricity of 0 means that the orbital paths follow a perfect circle, while eccentricities increasing toward 1 indicate elliptical orbital paths that are increasingly long and narrow.
To begin, select an average distance between the two stars of the binary pair.
To determine the average distance at random, roll 3d6 on the Stellar Separation Table.
| Stellar Separation Table | ||
| Roll (3d6) | Separation | Base Distance |
| 3 or less | Extremely Close | 0.015 AU |
| 4-5 | Very Close | 0.15 AU |
| 6-8 | Close | 1.5 AU |
| 9-12 | Moderate | 15 AU |
| 13-15 | Wide | 150 AU |
| 16 or more | Very Wide | 1,500 AU |
To determine the exact average distance, roll d% and treat the result as a number between 0 and 1. Multiply the Base Distance by 10 raised to the power of the d% result. The result will be the average distance of the pair in AU.
Feel free to adjust the result by up to 2% in either direction. You may wish to round the result off to three significant figures.
Next, select an eccentricity for the binary pair’s orbital path. Most binary stars have orbits with moderate eccentricity, averaging around 0.4 to 0.5, but cases with much larger or smaller eccentricities are known.
To determine an eccentricity at random, roll 3d6 on the Stellar Orbital Eccentricity Table. If the binary pair is at Extremely Close separation, modify the roll by -8. If at Very Close separation, modify the roll by -6. If at Close separation, modify the roll by -4. If at Moderate separation, modify the roll by -2. Feel free to adjust the eccentricity by up to 0.05 in either direction, although eccentricity cannot be less than 0.
| Stellar Orbital Eccentricity Table | |
| Roll (3d6) | Eccentricity |
| 3 or less | 0 |
| 4 | 0.1 |
| 5-6 | 0.2 |
| 7-8 | 0.3 |
| 9-11 | 0.4 |
| 12-13 | 0.5 |
| 14-15 | 0.6 |
| 16 | 0.7 |
| 17 | 0.8 |
| 18 | 0.9 |
Once the average distance and eccentricity have been established, the minimum distance and maximum distance can be computed. Let R be the average distance between the two stars in AU, and let E be the eccentricity of their orbital path. Then:
Here, Rmin is the minimum distance between the two stars, and Rmax is the maximum distance.
Whichever arrangement is selected, design the closely bound pair first as if it were a binary star system (see above). This binary pair is unlikely to have Wide separation, and will almost never have Very Wide separation. Select an average distance for the pair accordingly. If selecting an average distance at random, modify the 3d6 roll by -3. Select an orbital eccentricity normally, and compute the minimum and maximum distance for the binary pair.
Once the binary pair has been designed, determine the orbital path for the pair (considered as a unit) and the single component of the star system. The minimum distance for the pair and single components must be at least three times the maximum distance for the binary pair, otherwise the configuration will not be stable over long periods of time.
If selecting an average distance for the pair and single component at random, use the Stellar Separation Table normally. If the result indicates a separation in the same category as the binary pair (or a lower one), then set the separation to the next higher category. For example, if the binary pair is at Close separation, and the random roll produces Extremely Close, Very Close, or Close separation for the pair and single component, then set the separation for the pair and single component at Moderate and proceed.
Select an orbital eccentricity for the pair and single component normally, then compute the minimum distance and maximum distance. If the minimum distance for the pair and single component is not at least three times the maximum distance for the binary pair, increase the average distance for the pair and single component to fit the restriction.
As in a trinary star system, design the closely bound pairs first. Each binary pair is unlikely to have Wide separation, and will almost never have Very Wide or Distant separation. Select an average distance for each pair accordingly. If selecting an average distance at random, modify the 3d6 roll by -3. Select an orbital eccentricity, and compute the minimum distance and maximum distance, for each binary pair normally.
Once the binary pairs have been designed, determine the orbital path for the two pairs around each other. The minimum distance for the two pairs must be at least three times the maximum distance for either binary pair, otherwise the configuration will not be stable.
If selecting an average distance for the two pairs at random, use the Stellar Separation Table normally. If either result indicates a separation in the same category as either binary pair (or a lower one), then set the separation to the next higher category. For example, if the two binary pairs are at Close and Moderate separation, and the random roll produces Moderate or lower separation for the two pairs, then set the separation for the two pairs at Wide and proceed.
Select an orbital eccentricity for the two pairs normally, then compute the minimum distance and maximum distance. If the minimum distance for the two pairs is not at least three times the maximum distance for both binary pairs, increase the average distance for the two pairs to fit the restriction.
Each binary pair in a multiple star system will circle in its own orbital period. The pair and singleton of a trinary system will also orbit around each other with a specific period (probably much longer). Likewise, the two pairs of a quaternary system will orbit around each other with a specific period.
Let R be the average distance between two components of the system in AU, and let M be the total mass in solar masses of all stars in both components. Then:
Here, P is the orbital period for the components, measured in years. Multiply by 365.26 to get the orbital period in days.
Most binary pairs are detached binaries. In such cases, the two stars orbit at a great enough distance that they do not physically interact with each other, and evolve independently. However, if two stars orbit very closely, it’s possible for one of them to fill its Roche lobe, the region in which its own gravitation dominates. A star which is larger than its own Roche lobe will tend to lose mass to its partner, giving rise to a semi-detached binary. More extreme cases give rise to contact binaries, in which both stars have filled their Roche lobes and are freely exchanging mass in a common gaseous envelope.
This situation is only possible for two main-sequence stars that have Extremely Close separation, or in cases where a subgiant or red giant star has a companion at Very Close or Close separation. If a binary pair being considered does not fit these criteria, there is no need to apply the following test.
For each star in the pair, approximate the radius of its Roche lobe as follows. Let D be the minimum distance between the two stars in AU, let M be the mass of the star being checked in solar masses, and let M´ be the mass of the other star in the pair. Then:
Here, R is the radius of the star’s Roche lobe at the point of closest approach to its binary companion, measured in AU. Compare this to the radius of the star itself, as computed earlier. If the star is larger than its Roche lobe, then the pair is at least a semi-detached binary. If both stars in the pair are larger than their Roche lobes, then the pair is a contact binary.
The evolution of such close binary pairs is much more complicated than that of a singleton star or a detached binary. Mass will transfer from one star to the other, altering their orbital path and period, profoundly affecting the evolution of both. Predicting how such a pair will evolve goes well beyond the (relatively simple) models applied throughout this book. We suggest treating such binary pairs as simple astronomical curiosities, special cases on the galactic map that are extremely unlikely to give rise to native life or invite outside settlement. Fortunately, these cases are quite rare except among the very young, hot, massive stars found in OB associations.
One specific case that is of interest involves a semi-detached binary in which one star is a white dwarf. Hydrogen plasma will be stripped away from the other star’s outer layers, falling onto the surface of the white dwarf. Once enough hydrogen accumulates, fusion ignition takes place, triggering a massive explosion and ejecting much of the accumulated material into space. For a brief period, the white dwarf may shine with hundreds or even thousands of times the luminosity of the Sun. This is the famous phenomenon known as a nova.
Most novae are believed to be recurrent, flaring up again and again so long as the white dwarf continues to gather matter from its companion. However, for most novae the period of recurrence is very long – hundreds or thousands of years – so nova events from any given white dwarf in a close binary pair will be very rare. Astronomers estimate that a few dozen novae occur each year in our Galaxy as a whole.
Arcadia: Alice has already decided that the Arcadia star system has only the primary star, so she skips this step entirely.
Beta Nine: Bob knows that the Beta Nine system is a double star. Proceeding entirely at random, he rolls 3d6 on the Stellar Separation Table and gets a result of 7. The two components of the system are at Close separation. He takes a Base Distance of 1.5 AU and rolls d% for a result of 22. The average distance between the two stars in the system is:
Bob rounds this off a bit and accepts an average distance of exactly 2.50 AU. He rolls 3d6 on the Stellar Orbital Eccentricity Table, subtracts 4 from the result since the stars are at Close separation, and gets a final total of 5. The orbital path of the two stars has a moderate eccentricity of 0.2. Bob computes that the minimum distance between the two stars will be 2.0 AU, and the maximum distance will be 3.0 AU.
Bob can now compute the orbital period of the two stars:
The two stars in the Beta Nine system circle one another with a period of a little more than eight years.
The two components of the Beta Nine system form a binary pair, with a minimum separation of 2.0 AU. There is no possibility of the pair forming anything but a detached binary, so Bob does not bother to estimate the size of either component’s Roche lobe.
The paper by Duchêne and Kraus cited earlier describes the best available models for the distribution of separation in binary pairs. The period of a binary star appears to show a log-normal distribution with known mode and standard deviation. Generating a log-normal distribution with dice is a challenge without requiring exponentiation at some point, hence the unusual procedure for estimating separation used here.
