More On Population: A Response to Jason Beresford

Gunnar Brolin — Glasmalarvagen 6,122 31 — Enskede, Sweden

The response by Jason Beresford to my article on the population of Gondor and Arnor (OH 3: 4 — 7) was interesting and thought provoking. It gave me a reason to explain to myself the basis of my analysis in a way I hadn’t done before. Let me point out that anything that is said on this subject by myself, Jason Beresford, or anyone else is ultimately a matter of personal judgment. There are no right or wrong.


The statistics on which my (or any one else’s) analysis is based are very hazy to say the least. There are simply no hard facts about population development in Europe or any other part of the world before the 18th century — there are only guesses and estimations based upon the probable trends of population development and upon information from smaller areas.

Tax returns are of good help here, since there are records which have survived from as early as the 9th century. In this context, the Doomsday Book of the Norman conquest is invaluable. Such tax return records are problematic, however, because they are based on households, hearths, or similar collectivities, rather than on individuals.’

This raises the question of the relative size of a normal household. A household of five would result in a population estimate 25% greater than a household of four. And what happens if there are six persons per household instead of four? The only conclusion is that all numbers are very uncertain.

Furthermore, in order to be representative, the average reached must be based on a large selection. To apply an average number to a small area may seem scientific, but it contradicts statistical principles and certainly gives a false security, since it necessarily results in the appearance of very accurate numbers.2

The small size of the areas Jason Beresford has chosen forces him to round his figures off to the nearest thousand (as modern population statistics usually do); by contrast, I had rounded off to the nearest hundred thousand. In his analysis of Arnor, seven out of the fifteen areas Jason has chosen measure 5,000 square kilometers or less (smaller than the state of Delaware), whereas the actual average area in Arnor is about 9,300 square kilometers (or smaller than Connecticut). I’m sorry to say, but the thousand square kilometers of the Lower Annabrith river valley (not much more than an ordinary county) looks slightly ridiculous to me. To extrapolate the population of a small area from an assumed population average is as much of a guess as assigning a number at random, but since it gives the (baseless) impression of scientific research, it is far more misleading.

There are roughly four levels of population density in Europe during the late 13th century. The highest level is present in Italy and the low countries (Belgium and the Netherlands) with an average population density of 40 inhabitants per square kilometer. The next highest is representa-tive of Germany, France and England, with some 25 inhabitants per square kilometer. They are followed by Poland, the Balkans, Asia Minor and the outer areas of the British Isles with a density of about 10 inhabitants per square kilometer. Finally, we have Scandinavia and the European parts of the former Soviet Union with about 2 persons per square kilometer.3

If we look at Europe during the Roman period (say, about 200 AD), the most populous areas are Italy and Greece with more than 20 inhabitants per square kilometer. Most Roman provinces (including the African and Asian ones, as well as non-Roman Germany) had a population density of between 10 and 20. It was only on the British Isles, in Scandinavia, and in Eastern Europe that the population density would have been lower than 10 persons per square kilometer. The conclu-sion would be that in areas within an organized state where subsistence is based on agriculture, the average population density seldom falls below 10. It is only in tribal societies (as the opposite of organized states) and/​or nomadic or semi-nomadic societies that we have numbers much lower than 10 inhabitants per square kilometer.4


As explained above, it is fair to assume that 10 inhabitants per square kilometer could be used for less developed areas within an organized state, especially if we are talking about peak numbers. Looking at Gondor, I could have used the Roman” range of 10 – 20 or the Medieval” range of 10 – 40. Since I found the Roman range too narrow and the Medieval range too wide, I made a compromise and set the most developed area (the Anduin valley) at 30, the intermediate (the peninsula of Belfalas) at 20, and the less developed areas at 10.5

The less developed areas within the organized state of Gondor are Lamedon and Anfalas in the west, the coastland of Harondor in the south, and the habitable area of Calenardhon in the north. Enedwaith is very much a tribal area where royal control or influence is nominal. Accordingly, Enedwaith would have a lower population density than those mentioned above. The interior of Harondor is probably half-desert or at least arid with a small, mostly nomadic population (thus much lower than an agrarian society). Anfalas and Lamedon are very similar in degree of development, society and density of population, and I can find no logical grounds in Tolkien’s writings for there to be any major difference between them.6


My position regarding Arnor differs from Jason’s for very much the same reasons as with Anfalas. His analysis results in a density of less than 5 for the entirety Arnor at the beginning of the Third Age, when the realm ought to have been at its peak (c. 450,000 square kilometers). However, I can’t see why the realm at that time should have a population density which is less than half the average for the less developed areas in Europe under normal circumstances. I won’t quarrel with Jason’s assumption that most people in Eriador live in the river valleys or around Lake Evendim, because I argue exactly this point in my previous article (OH 3:6); but I do not consider it a valid assumption that the population of Elendil’s realm should be on a par with the average density for unorganized tribal societies.

In some ways, even a density of 10 is too low for Arnor at its zenith. I myself had initially considered a higher value, but found that either the population at its nadir (c. TA 2000) would be too great, or its drop from zenith to nadir would unrealistically steep. A point to remember is that all figures for population density represent averages (i.e., 10 persons are not living on each and every square kilometer.).

As I suggest in my article, most of Arnor’s population is concentrated in the west. Rhudaur is the wildest area with the smallest population among the three successor states.

Developments from the split of 861 to the destruction of 1975 must have reinforced this imbalance, as well as the continuing presence of Trolls and Orcs in the Misty Mountains after the fall of Angmar.

Frodo and his friends traveled first through a large woodland area until they reached Bree, and then they moved through the wildest and least inhabited region of Eriador. We would not be allowing the author of The Red Book of Westmarch much poetic license for neglecting to mention such unimportant settlements as this region might have contained. That Tolkien doesn’t say much about Arnor is no proof that Eriador was depopulated most of the time.7


After having analyzed Jason Beresford’s figures, I don’t think there is that much of a difference between us.8 Due to the uncertainty in all these numbers, we must allow for a margin of error of at least +/- 10 – 15% (i.e., 500,000 – 700,000). I would perhaps distribute the figures slightly differently, with less population on the southern side and more on the peninsula.

More significant differences exist between us with respect to the population outside the neighborhood of Umbar. I estimated a population of 100,000 for the area between Rath Annûn and the River Harnen in the north, and the Annabrith, Gondeithel and Cuiviërant valleys in the south. Jason Beresford sets the population of the southern part alone at 130,000 people, which I find to large. In my opinion, he largely overrates both Dûsalan with its surrounding territory and the Annabrith valley.9

Excursus: The Demographic Use of the Term Dúnadan”

The word Dúnadan” may alternately refer to two different groups of people. First, it may have a racial sense denoting those of true Númenórean extraction (the descendants of Elros’ followers). Second, it may designate those who participate in Dúnadan culture without necessarily possessing the genetic characteristics of the Three Houses of the Edain.10

As regards the realms-in-exile, all or most of their inhabitants would fall within the second group if assimilated into Dúnadan culture; by contrast, the Dúnedain in the racial sense would constitute a very small minority. Elendil and his sons arrived in Middle-earth with nine ships following the Akallabêth. Even assuming that these were quite large ships, this group couldn’t have amounted to more than 5,000 – 6,000. If the total population of Gondor amounts to around 9,000,000 people (OH 3: 5), then the racial Dúnedain would make up only about .05% of the total population (i.e., about one for every two thousand).

Even if we add to this number the Númenóreans living in Pelargir and other colonies along the coasts, we are still talking about no more than 100,000 – 200,000 people, or 1 – 2% of the total population — certainly a minority;11 Just because the (racial) Dúnedain were few in Arnor doesn’t mean that the entire population was small. If that had been the case, the population in Gondor must also have been quite minuscule throughout its entire history.


  1. Taxes assessed upon individuals are a rather modern invention.
  2. The larger the base used, the better the average” is a basic rule within the statistical science.
  3. This last must be taken with some caution, as nearly two thirds of Sweden are mountainous or otherwise uninhabitable (given the agricultural techniques of the Middle Ages). Disregarding these marginal areas, the population density rises to about 6 — 8 persons per square kilometer.
  4. And the number of 10 isn’t much — we are talking about one to two households per square kilometer. A village of 500 persons would be the only habitation in a 15 square mile radius.
  5. Remember that we are always talking about the population of Gondor at its largest.
  6. My reason for separating them was purely practical, since this made it easier to measure them on the map.
  7. When looking at his writings, it is clear that Tolkien was more interested in Gondor than in Arnor — just look at how deeply he goes into the history of Gondor in Appendix A,” when compared with his rather sketchy description of Arnor.
  8. Jason Beresford places the population of the coasts of Nen Umbar (as far south as the Annabrith) together with the peninsula at 570,000 inhabitants, while my own estimate comes to 600,000 persons (a difference of 30,000 or about 5%).
  9. By the way, if Dûsalan isn’t a part of the territory of Umbar (as I was going to suggest in the forthcoming Kin-strife module), the most logical alternative would be for it to be a dependent city-state in the same way as most smaller ports in the Adriatic were controlled by Venice during the Middle Ages.
  10. As I understand it, only Elros and his descendants enjoyed the extra longevity of several hundred years.
  11. A useful analogy might be the number of Englishmen in the British colonies of the 19th century or the number of Frenchmen in the French colonies of the same period.


Right-click and choose "Save link target as" for the .markdown files.