The Silmarillion

The Silmarillion is a single volume assembled out of five texts of markedly unequal weight. Two short pieces open it — a cosmogonic account of the world's making, and a brief descriptive catalogue of the Valar and the chief Maiar — after which the Quenta Silmarillion proper occupies the bulk of the book in numbered chapters. Two further narratives, both far shorter, close the volume by carrying the matter out of the Elder Days into the Second and Third Ages. Around this Christopher Tolkien set a working apparatus: a foreword accounting for how the text was constituted, genealogical tables, a map of Beleriand, a note on pronunciation, an appendix on the elements that recur in Quenya and Sindarin names, and a full index of persons and places. The book is therefore as much a work of reference as a book of tales, and it is usually consulted in both ways.

The manner of telling is chronicle rather than novel. The voice is remote and summary, moving through centuries in a sentence and treating whole wars as a single turn of events; direct speech is sparse, interior thought almost absent, and the reader is never addressed. Where the compression relaxes, it does so for a handful of long tales — the story of Beren and Lúthien and that of Túrin most conspicuously — which are given something closer to scene and pace. The diction is deliberately elevated and archaic in cast, and the text presents itself as compiled and transmitted tradition rather than as eyewitness report. Tolkien himself described the cycle as mythology and legend of graduated kinds, with the earlier matter left more distant and less circumstantial than what follows.

In the shape of the legendarium the book stands behind The Lord of the Rings rather than after it, and it addresses readers who have met its names in passing there and want the frame they belong to. Elbereth invoked in prayer, Eärendil in Bilbo's verse, Gil-galad and the older wars named at Rivendell, the descent of Elrond and of Aragorn set out in the appendices — all of these point back to matter the later book does not itself relate. That later book also supplies the fiction by which such a compilation could exist at all, in the note describing Bilbo's three volumes of translations from Elvish sources made at Rivendell. Read in that light, The Silmarillion is less a companion to the novel than the body of tradition the novel keeps quoting.

The matter of the book was begun in the First World War and never finished. Tolkien dated the earliest of the tales to 1916 and 1917, written while he was invalided out of the army, and named the story of the fall of Gondolin as the first of them to be fully set down; the group of narratives he composed in those years already contained, in an early and very different form, several of the tales that would stand in the published volume. He described the undertaking afterwards as an attempt to make a body of legend of his own, connected to the languages he was inventing and loosely attached to England. From that beginning he returned to the material at intervals for the rest of his life, so that its composition spans close to sixty years and produced not one text but a long succession of them, differing in scale, in style and often in substance.

Between the wars the legends were recast more than once: parts were worked into long poems that were abandoned unfinished, and the whole was several times compressed into continuous prose accounts, each rewriting altering names, chronology and motive. When The Hobbit was accepted in 1937 Tolkien offered his publishers this older and quite unrelated body of work, hoping it might be issued in some form. It was not taken; what the publisher wanted was a further book about hobbits. He began that sequel almost at once, and the mythology was set aside at the point it had then reached.

The interruption lasted far longer than anyone expected. The new story was begun soon after The Hobbit appeared, occupied Tolkien through the war years with long pauses, and was not brought to an end until about 1949, after which further delay preceded publication in three parts in 1954 and 1955. During those years he was unwilling to treat the two works separately, holding that the later book was unintelligible without the earlier one and that each required the other; the long account of the whole cycle he wrote for a prospective publisher around 1951 was made expressly to argue that they be issued together. That plan came to nothing, and when the sequel was at last published it went out alone, its debts to the older matter visible only in appendices and passing allusion.

Tolkien returned to the legends after 1955 and worked at them for much of the time remaining to him, but the late labour tended toward revision, annals and philosophical enquiry rather than toward completion, and some of it would have required rebuilding the cosmology of the older narratives from the ground up. At his death in 1973 no finished text existed, and for a number of the central chapters the latest full version was decades older than the latest thinking about them. Christopher Tolkien, as literary executor, undertook to produce a single connected narrative out of this, assisted in the earlier stages by Guy Gavriel Kay. The work meant choosing between versions of widely different date, reconciling names and dates across them, and in places composing linking or concluding passages where no late text existed at all.

The resulting volume was published in 1977 by George Allen & Unwin in London and by Houghton Mifflin in Boston, four years after the author's death; a German translation by Wolfgang Krege followed in 1978, and subsequent printings have carried corrections to text and index. Its editor was candid from the outset that the book was a construction and not a manuscript, and he continued that accounting afterwards: the collection of related material he brought out in 1980 was deliberately presented in its unresolved state, with commentary showing where the published narrative had smoothed over divergence, and it set out explicitly how that narrative stood to the papers behind it. The distinction matters for citation. The 1977 volume is a coherent and authorised text, but it is an edited one, and where it can be checked against the drafts it sometimes represents a stage of thought its author had already left behind.

The two opening pieces are short and carry subtitles that declare their function: the first announces itself as the music of the Ainur, the second as an account of the Valar and Maiar drawn from Eldarin lore. The Valaquenta is the more consultable of the two, since it is subdivided into a survey of the Valar, a shorter one of the Maiar, and a brief closing part on the enemies, so that a reader meeting a divine name elsewhere in the volume can generally place it here without recourse to the index. The long central text follows under its own title, the history of the Silmarils, and it is the only part of the book divided into numbered chapters — twenty-four of them, nearly all headed by the same formula, the preposition 'Of' followed by the person, people or event treated. The table of contents therefore doubles as a finding list, and it is the quickest route into the volume.

Those twenty-four chapters fall into three unequal movements. The first eleven or so remain largely in the West and treat the ordering of the world, the awakening of the Elves and their journey, the making of the jewels and the departure of the Noldor. From the return to Middle-earth the narrative reorganises itself around territory and battle, and one chapter here is not narrative at all but a descriptive survey of Beleriand and its several realms, best read against the map rather than in sequence. The final third gives the long tales their own chapters — Beren and Lúthien, the fifth battle, Túrin, the ruin of Doriath, the fall of Gondolin — and closes with a single chapter carrying the age to its end. Several chapters are very short and function as notices rather than stories, among them the brief ones on the Sindar and on the coming of Men. Matter once concluded is rarely reopened, so an episode is normally found in one place only.

The two closing texts stand outside that numbering and outside the Elder Days. The Akallabêth bears an Adûnaic title and adopts a Númenórean vantage on the Second Age; the last piece is the shortest in the volume and compresses the remainder of that age and the whole of the Third into a summary that ends at the departure over the Sea, meeting the close of The Lord of the Rings. The back matter is designed to be used rather than read. The genealogical tables set out the princely house of the Noldor, the kindred of Thingol and Olwë, and the houses of the Edain, and are the fastest way to fix a relationship the narrative states only in passing; the note on pronunciation and the appendix on the elements recurring in Quenya and Sindarin names together allow a name absent from the index to be broken down and construed; and the index itself glosses many entries and cross-refers the alternative names under which a single figure or place appears in different chapters.

Chronologically the volume holds ground that no other book published in Tolkien's lifetime occupies. Everything before the drowning of Númenor exists elsewhere only as allusion: a few names in the appendices to The Lord of the Rings, a genealogy of the Half-elven, the brief opening remarks on the origin of the Númenóreans and their debt to the Elder Days. For the awakening of the Elves, the wars in Beleriand and the fate of the jewels themselves there is no competing continuous narrative, and the encyclopedia depends on this text alone for that whole span. The balance reverses for the later ages. Where the Akallabêth and the closing piece pass over the Second and Third Ages in summary, the appendices to the earlier-published novel give dates, regnal lists and connected accounts in far greater detail, so for those centuries The Silmarillion is the thinner witness and should be read against them rather than in place of them.

Beside it stand the later posthumous volumes, which occupy the same period but on quite different editorial principles. Unfinished Tales prints long texts in the state its editor found them, with notes and unresolved alternatives left standing; the single-tale books devoted to Túrin, to Beren and Lúthien and to the fall of Gondolin restore at length what the compiled volume gives in a chapter, and two of them set successive versions side by side instead of choosing between them. The relation is therefore not one of supplement but of scale: the same episodes appear in a compressed harmonised form here and in fuller, more provisional forms there. Where the two disagree in substance, the disagreement is usually a matter of date of composition rather than of authority.

Several such divergences are well marked. The published text makes Celeborn a kinsman of Thingol who never left Middle-earth, whereas the material gathered in Unfinished Tales preserves a later conception in which he was of Alqualondë and came east with Galadriel, together with more than one account of why she departed from Aman and whether she was permitted to return. The Túrin chapter likewise abridges a much longer narrative, and the abridgement removes motive and incident rather than merely words. Nothing in the compiled volume flags these choices; they are recoverable only from the commentary printed with the fuller texts, which is why a citation to a chapter here is not always a citation to Tolkien's final thought.

For this encyclopedia the book carries the K2 layer and defines it: authorised, coherent, and edited from papers its author did not settle. The practical rule follows from that. Where the volume and the works published in Tolkien's lifetime disagree, the lifetime text governs; where it and the other posthumous volumes disagree, the divergence is recorded rather than resolved. Tolkien's own description of the cycle, in the long letter written to interest a publisher in it, treats the three ages as one connected body of legend of which the tale of the Rings is the last part, and that framing is the one the arrangement of the book follows.

The clearest documented measure of how the volume was taken up lies in what its editor did next. Christopher Tolkien set out in the introduction to the 1980 collection of related material that he had been uncertain whether more of his father's papers should be printed at all, and that the questions the compiled narrative left open bore on that decision. The answer proved to be a series rather than a single supplement: between 1983 and 1996 twelve volumes of drafts, annals and commentary appeared, presenting in sequence the manuscripts from which the 1977 text had been constituted. A programme of publication sustained over that span is the plainest evidence of the demand the first volume created.

The compressed manner of the central narrative was the point to which the editor returned most often in later prefaces. Introducing the separate Túrin volume of 2007, he observed that readers who came to the book from The Lord of the Rings met a mode of telling far removed from it, and he gave that as part of his reason for setting the tale out at full length as one connected text. Two further tales were issued on the same principle, Beren and Lúthien in 2017 and the fall of Gondolin in 2018, the latter presented by its editor as the close of the work he had begun four decades earlier. None of the three displaced the compiled volume, which has remained in print in its original arrangement throughout.

The text has been stable in print, but its surroundings have not. Later editions add as a preface the long letter of about 1951 in which Tolkien described the whole cycle to a prospective publisher, so that his own account of its scope now travels with the book itself. Illustrated editions have also appeared, among them one with paintings by Ted Nasmith and a later one drawing on Tolkien's own drawings, designs and heraldic devices. Translation began within a year of first publication, the German version being among the earliest, and has continued widely since; the apparatus of tables, appendix and index has generally been carried over with the narrative rather than dropped from it.