My presentation from the Mythgard Mid-Atlantic Speculative Fiction Symposium 2016. It begins with the paper from this summer, drops the math, and investigates what happens if we try to apply the same methods to other fantasy works.

Primary Geography of Sub-created worlds


A subcreated world is derivative of the primary world, so fantastical geography is going to be derivative of primary-world geography.  This work began last winter when I stumbled across Oliver O’Brien’s public-access database that lets you type in a name and see where in the UK people are called that. Geographers have done extensive work on matching names with places all over Europe, it turns out. Like hobbits, English people move around only slowly, so names are strongly associated with places. When I was given the opportunity to map a family name in Great Britain, the second name I tried was “Baggins”. The distribution of that name is centered near Birmingham almost exactly over top of J.R.R. Tolkien’s boyhood home.

It turned out that this is not a coincidence.  Tom Hillman pointed out that the connection wasn’t dumb, and is worth pursuing. So let’s see what we can learn about fantasy novels by using real-world geography. This talk will tell you about three of them. The title has the word “fragments” because they don’t fit together into a solid piece of china.

The Lord of the Rings

Alone among the denizens of Middle Earth, hobbits have family names. They’re  conscious of their difference.  They effortlessly slide into “Peregrin son of Paladin” phrasing when they encounter Men.  As Tolkien wrote in letter 25, “[hobbit] family names remain for the most part as well known and justly respected in this island as they were in Hobbiton and Bywater.”

map of Baggins

There had always been a Baggins at Bag End. Computed by Oliver O’Brien.

I took this comment, along with another from Letter 211,  as explicit permission to look at The Lord of the Rings with real-world geography.  Hobbits are from different parts of England, according to their role in the story, and we know this is intentional because the story was stretched to fit them, if they didn’t match.


When we start looking at names, we can identify three categories of hobbits. Some people you keep close to you. They could be family, or they could be just friends.

JRRT was from the West Midlands; the close-by hobbits have names from that region. The Tooks and the Bagginses, despite their considerable differences in temperament, are quite close.

map of Took

The Tooks of Birmingham

This shows up on the map as surely as it does in the text. The Cottons are nearby, too, as is Hob Hayward, whom Sam picks out of the group of Shirrifs as being someone friendly. The readers Tolkien had in mind would have heard “Hayward” as going perfectly well with Cottons or Bagginses.


The second group of hobbits we find are neighbors who aren’t exactly bad, but they frequently seem to act in a way that interferes with others getting on with their lives. It’s good practice to keep them at arm’s length. They might be an unwisely-chosen in-law, or a cousin who borrows books and doesn’t return them. For example a Sackville-Baggins, rather distant from Birmingham to start with, can be led further from the fold by marriage to a Bracegirdle, who’s practically a foreigner in hobbitish terms.

The last category is the “liminal” hobbit-names. Hobbits in Buckland and the

Marish have them. Their family names in our world are either concentrated far from Birmingham, or they don’t appear in the UK at all,

underhill map

Underhills in the UK

like the Maggots or Brandybucks don’t. Frodo uses the name “Underhill” when he’s in Bree because Gandalf knows that someone who hears it won’t think of the bearer as living anywhere near the Shire. You can see here how well Gandalf knows his business. The Hornblowers are in this group, too. They’d “hardly ever set foot in Hobbiton before.” Although “Brandybuck” isn’t a name found in England, “Buckland” is. It’s on the extreme east outside Dover, that is, almost in France.


I mentioned that Tolkien would stretch the story, if a name didn’t fit.  Sam Gamgee is the conspicuous example. We know from Letters 72 & 144 that “Gaffer Gamgee” was a name Tolkien made up to amuse his children. It was a pun on “cotton”, like the Cotton family, and it occurred to him because there was a Dr. Gamgee at the University, but — it turns out that the name “Gamgee” doesn’t come from the West Midlands.

map of Gamgee

Serving the Bagginses may have involved commuting.

How then can Tolkien make Sam, whose presence in the story is essential, fit the pattern? The solution he found to this problem explains something I wondered about since the first time I read the Appendices.

Why is it worth half a page of Appendix F to write a lengthy discourse on the subject of how “Gamgee” isn’t really a family name, but more an epithet derived from the town of Gamwich whither their ancestors had decamped several generations back? This is why – Tolkien needed to get his hero within shouting distance of Birmingham. A London name wouldn’t do at all, but a nickname isn’t definitive.

Map of Gardner

Mayor Sam Gardner’s family

The final resolution of the story is in Appendix C. We learn that upon his accession to Mayor, Samwise changed his family name to “Gardner”.


That name has a much closer association with the West Midlands, though it’s not primarily in Birmingham. Evidently, three generations of residence plus saving the whole world from evil is almost sufficient to get one accepted into Hobbiton society.

The match of names isn’t perfect, but it works. As Tolkien says in the Prologue, “By the time of this history these names were no longer found only in their proper folklands.” Of course there will be some leakage across the boundaries for artistic purposes, and perhaps some because geographic information systems in the 1950s were less accessible.


At this point I went off looking for another novel I could subject to the same scrutiny. First consideration, I needed a world that was linguistically integrated. Most fantasy novelists don’t take the time to do that. They just pick names that sound cool. Unfortunately, this gives a geographical critic nothing to work with.

Tigana, by Guy Gavriel Kay, seemed ideal. First, most of the names sound Italian. Second, the author makes a grateful mention of the cartographer in the acknowledgements. He mentions that having clear maps guided him to fix some problems with the story. There are five maps in the TOR paperback edition, so it was reasonable to suppose that geography is important.

Tigana isn’t as famous as The Lord of the Rings, so here’s what it’s about in a few sentences. Two warlord/sorcerers are contending for the Peninsula of the Palm.

Frontispiece map from Tigana.

The Peninsula of the Palm

The Palm is divided into a small number of pseudo-feudal states. In the process of conquering the left side of the Palm, one warlord lost his son in battle. To punish the country that he was fighting at the time, he devised what may be the most absolute revenge ever conceived: The cities were razed, the populations were dispersed, and he cast a spell so that nobody can remember or even say the name of the country any more. Except, some of the people who were born there have formed a resistance movement, and when they induct another Tiganan (by birth) into their cell, they teach him or her the name. Covert action is only half of their subversion; the greater crime is memory. In the second part, a Tiganan woman has insinuated herself into the king’s seraglio to assassinate him, but ended up falling in love instead. Her love of her country gets opposed to her love of Brandin. In a nutshell, these are Kay’s two themes: love and memory.

So this is where I went looking for geographical significance in the text. What I found was several early mentions of the shape of the peninsula. Its shape is important to the characters in the first few chapters. After that, though, maps and geography are almost never mentioned. What’s going on?

Kay wants us to read this as an upside-down Italy. In case an inattentive reader didn’t notice the place names are all Italian (though the character names aren’t), he includes an epigraph from Dante. There are plenty of other inversions, once you start to look for them: Italy looks like a foot, the Palm looks like a hand. Italy points south towards a desert, the Palm points north towards a desert. Lower Corte, the new name given to the lost land of Tigana, is actually uphill from Corte. The queen of the empire to the east is named Dorotea, flipping the Byzantine Empress Theodora.

Putting the book in an inverted Italy has two purposes. First, this book is about memory. The former Tiganans, even if they don’t know that’s what they are, have a nagging feeling of uncertainty about their past. The spell isn’t perfect, so they have fleeting half-memories that leaked around its edges. I confess that’s a pretty good description for my own education in history – I never know if I’m remembering something correctly. For example, I had to hit up Wikipedia to see if there actually was a Queen Dorotea somewhere, to whom Kay was alluding. So this reader (at least) has the same sort of feeling about the past that the heroes do.

The second purpose is that, rarely among fantasy novels, Tigana fits into single volume. Using a mirror-image of our own world lets Kay focus his efforts on constructing his meticulously-crafted moral dilemmas. Kay uses this technique in many of his books. He invites us readers to use medieval Italy as a touchstone, any time we feel like we need a detail that didn’t interest the author. There’s no danger of taking us out of the story when he does this, because when we supply details from our own memory, they feel right automatically. But it’s not historical fiction, which brings us to the third book.

A Song of Ice and Fire

George R.R. Martin told January Magazine, “the only problem I have with historical fiction is that I know too much history. So I always know what’s going to happen. … With this sort of thing you can take people by surprise. It reads like historical fiction. It feels like historical fiction but you don’t know how it’s going to come out.”

It’s obvious that Martin isn’t interested in writing about anything but political history. He’s said as much, in fact. Perhaps out of respect for Tolkien’s precedent, there’s a map in the front of A Game of Thrones, but we can get an idea of its importance to the story by noticing that Chapter 2 takes place in a city that’s not on the map.

Westeros is the British Isles

Westeros, by KingofFairview, via  Brilliant Maps

When Martin does take the time to describe a landscape, in the very next sentence he’s back to politics. Here are two consecutive sentences. Catelyn is riding through the lands where she grew up. She hasn’t seen them in a long time.

North of here the Kingsroad ran along the Green Fork of the Trident, through fertile valleys and green woodlands, past thriving towns and stout holdfasts and the castles of the river lords. Catelyn knew them all: the Blackwoods and the Brackens ever enemies, whose quarrels her father was obliged to settle; Lady Whent …; Lord Frey ….

A Game of Thrones, P. 241

The first sentence jumped out at me because it’s almost a pastoral lyric, by GRRM standards. I’ve elided about 40 words of aristocratic political detail with those ellipses. It’s clear that the political actors are the important part.  Martin uses the land as a board to hold the pieces in his chess game, and that’s all. His text doesn’t encourage geographic analysis.

His fans, though, seem to have overruled him.  By now, twenty years after the first volume was published, Westeros has been invaded by amateur geographers like Adam Whitehead, professional demographers like Lyman Stone, and what appears to be a large detachment from the Stanford University Department of Geology.

The geologists support Martin fairly well. Mapping what must be happening in his world from beneath, they can find analogues to the earth’s crust. Nothing they found in the book is obviously absurd.

The geographers and demographers, on the other hand, are pretty brutal. They find only one way to make sense of the population densities, the city sizes, the number of ethnic groups, the patterns of dynasties, and so forth. They conclude that A Song of Ice and Fire has an unreliable narrator. If you assume that city populations are a factor of 10 too large, that all the distances are 5 to 10 times too broad, and that spans of time are 2 to 5 times too long, then Westeros at least is a self-consistent world. In a sense, Martin has reproduced a familiar problem – all the way back to Herodotus, sizes of armies and all the casualty figures are inflated by anyone’s-guess-how-much. Martin has actually expanded that mis-feature of ancient historiography into new areas.

At this point, I was all set to declare that my objective – using real-world geography as an approach to understand fantasy stories – must be limited to only a few, extremely thorough authors. But then I read a post by Adam Whitehead. Martin is listening to his fans, so these criticisms are feeding back into the series. In A Feast for Crows, Whitehead notes that the characters have begun mentioning that you can’t exactly trust things you hear from the maesters at the Citadel. The characters now know the lore is wrong.


I’ll conclude, therefore, by saying that Tolkien, and the writers who followed him, do to a large extent “keep their feet on their own mother-earth,” so this approach can be useful.

But subcreation is becoming a new and different thing.   World-building need not be an author’s solitary pursuit. It can be a collaborative effort, drawing on a range of knowledge that is limited only by the breadth of the author’s fan base. And not only “can be”, soon we may find that it must be. Once an author has a fan base, he may find that a world will be built whether he likes it or not. And literary criticism via the sciences is no longer a passive pursuit. In some cases, it appears to be able to affect the work it studies.