The evolution of Númenórean cartography
Today [this article was first published in December 1999] we have two main divulged types of maps of the entire ensemble of Middle-earth, and they're very different to each other. Of these, the first known by Tolkien's fans was that of the Middle-earth Role Playing (MERP) series.
In it, the shape of Middle-earth follows almost exactly Tolkien's maps with regards to north-western Middle-earth, but the rest of the land masses are nearly random - a logical situation if we take into account that the only references existing when MERP appeared were the words of The Hobbit, The Lord of the Rings and The Silmarillion, and little more.
The second and most "legitimate" maps were Tolkien's own maps of the Ambarkanta, published in The Shaping of Middle-earth and clarified by KW Fonstad's Atlas of Middle-earth.
However, even with these maps we still have the problem of the inexactitude of the "legendary" maps and our actual geography. But, as one of my main interests is to link Tolkien's mythology to actual History, I'd like now to fit all that in the story of a new aspect of the mythology, justifying and adopting the divergences. In this I take the biggest part of the data from Unfinished Tales II.
The first journeys of Númenórean reveal the state of their geographical data. Veantur came to Lindon in the year II 600. Thence the Edain had departed to Númenor, so what they knew was little until the days of King Tar-Aldarion.
The MERP maps could be supposed to be some conserved in Gondor, but deriving from the first Númenórean attempts at cartography, so that the unknown lands were represented following the imagination of the designer.
Tar-Aldarion's curiosity brought him to the western shores of Middle-earth, and he founded many settlements, of which the most fell into abandonment in the time of Tar-Ancalime, his successor. Aldarion began the exploitation of the woods of the sides of River Gwathló, and visited places of future importance and renown, like Umbar, Belfalas (where there were Elves) or the Ras Morthil.
So the next stage of Númenórean cartography would reflect the new discoveries but also would incorporate data from Elven legends. These, then (or derived from them) are the maps that we see in The Shaping of Middle-earth.
I think that's the true sense of Tolkien's world maps of the Ambarkanta.
Though this tale was composed by Tolkien in a time when the association between the "legendarium" (as Christopher Tolkien calls it) and our actual History didn't have all the strength and significance that was achieved later with the development of The Lord of the Rings, it illustrates the words of Rúmil, who we can suppose to have been known by Númenóreans because of their contact with the Elves of Tol Eressëa. And of course the Elves knew things of the far regions of the world, which the Valar told them. So I think that we can't take the maps of the Ambarkanta as geographically exact.
European cartography had a similar period, when for example Atlantis or the magnetic rock of the North Pole appeared in maps. Or again later, with the curious case of the "Island California". And the similarity between the actual European cartography and Númenórean allows us to deduce a new and definitive stage.
After Tar-Ancalime the development of the colonies of Middle-earth was recuperated, perhaps by Tar-Súrion (1394-1556), her grandson, so that Tar-Minastir could defeat Sauron in Eriador sending there an army that had never waged war.
Then comes the golden era of Númenórean exploration, with Prince Ciryatan's journeys (XIX century), already sometimes moved by greed. Númenórean ships came to all the mortal shores (except of course those to the North covered by ice), even to the Gates of Morn in the Extreme East, and expanded their dominions. Colonial kingdoms were created (see Gorgoroth, "Nazgûl", of the MERP series), and Tar-Calmacil forced Sauron to retreat to the inner lands.
In my opinion, this is not only referred to an impulse from the western dominions as Umbar and Pelargir against Mordor and its adjacent countries, but to a closing of the "siege" over Sauron's dominions.
We have here a parallel with European cartography in the XVI and XVII centuries, when the creation and expansion of colonies all over the world resulted in maps being continuously corrected.
In such a way the Númenóreans would have developed almost exact maps, useful to their affairs on Middle-earth and the eastern lands, so that the Ambarkanta would have been perfected or even abandoned because of its inexactitude (and because it was a work of the Elves) in Númenor.
Taking into account the little destruction caused in the western shores of Middle-earth by the Change of the World in comparison with its "geological" importance, it is certain that the lands South and East of Middle-earth, far from the "critical zone" of the Atlantic Ocean, would have changed less.
Then the question is to fit the known north-western lands to the rest of the world ...