"The Ruin" is a poem in Old English by an unknown poet, preserved in The Exeter Book. It reflects on the fall of civilizations, and was likely inspired by the ruins of Roman Bath. But the poem is damaged - two passages are lost to a burn mark. Translations are freely available - you can read one on the Wikipedia page - but I'll just illustrate one of the missing passages here:

Still the masonry endures in winds cut down
persisted on__________________
fiercely sharpened________ _________
______________ she shone_________
_____________g skill ancient work _________
_____________g of crusts of mud turned away
spirit mo________yne put together keen-counselled
a quick design in rings, a most intelligent one bound
the wall with wire brace wondrously together.

You will note that not all the words of the missing lines are lost: enough remains for a partial translation of the missing sections.

It's often commentated that, for the modern reader, the missing bits of the poem actually contribute towards its message: The Ruin is, itself, a ruin. That's one reason given why none of the translations I've found have attempted to fill in the missing words. However, given the metrical structure of Old English poems and the partially preserved nature of the lines, it seems to me that it ought to be possible to make a very good educated guess about the nature of the lost text. Yet no-one I've come across has attempted to do so.

Have there been any attempts to use the structure of the poem and the remaining words to fill in the gaps, and if so, is there any scholarly consensus as to what they are? And if not, I would accept as an answer an explanation as to why this is not realistically possible.

  • The second from last line mentions rings, the last line mentions a wall with wire braces. And that seems unusual with Roman or Anglo-Saxon building. Does anyone know what it means. Commented Jun 28 at 4:25
  • @M.A.Golding That would make a good question for literature.se! Or see Klinck's note to lines 19–20. Commented Jun 28 at 7:37

2 Answers 2


A secure or consensus reconstruction is not possible, as too much of the text has been destroyed. Here’s the first part of the poem, on folios 123v and 124r of the Exeter Book. ‘The Ruin’ starts on the last line of folio 123v, with a capital wynn (Ƿ).

You can see that on folio 124r, about half of the text is missing on four manuscript lines (five verse lines), and a quarter of the text on another two lines. To demonstrate the difficulty of reconstruction, let’s take the easiest case, the first affected line, which Wikipedia has transcribed as:

Wunað giet se ...num geheapen

The manuscript says “wonað” [perhaps a variant or mistake for “wanað” = wanes, decays] or “worað” [perhaps a variant or mistake for “woriað” = totters], but most editions prefer the emendation “wunað” [endures, remains], originally proposed by Friedrich Kluge. Anne L. Klinck says that “wonað” or “worað” “would indicate a contrast between the ruin’s present and its former state, whereas giet, ‘yet, still,’ indicates some kind of continuity” (The Old English Elegies, pages 211–212).

The word “se” [the, that, he, she, it] in the Wikipedia transcript is conjectural—it could be a word starting “se‑” or (to my eye at least) “pe‑”. Klinck says, “The tall e after s indicates that a short or descending letter immediately follows. Very likely the word is seo [the; that]” (page 212).

The destroyed portion has space for around eight to twelve letters, and the majority of lines in the poem have three or four alliterations, so there’s a decent chance of another alliterating word, starting with “w‑” to alliterate with “wunað” or “g‑” to alliterate with “geheapen”. So we could conjecture something like this (with twelve destroyed letters):

Wunað giet se gæt   under wolcnum geheapen

The gate yet remains, heaped under the sky

Or perhaps something like this (with only seven destroyed letters, however, so a bit short, also I am not sure that this allterative pattern is allowed):

Wunað giet se wag   eotenum geheapen

The wall yet remains, heaped up by giants

The reconstruction in Wikipedia is credited to Jack Watson:

Still the masonry endures in winds cut down

There are a couple of problems here:

  1. Why “cut down” when the manuscript says “geheapen” [heaped]? This is an emendation, that goes back at least to Levin L. Schücking (1919), Kleines angelsächsisches Dichterbuch, page 33, proposing that the word was originally “geheawen” [hewn], perhaps because the scribe has mis-copied the wynn (ƿ) of the original as a “p”. The reason for conjecturing a mistake here is that “heaped” usually appears in weak forms like “gehypan” or “geheapian”. However, Klinck says, “the MS reading should be retained here as another strong form of a commonly weak verb” (page 212). The heaping up of the stones seems to me at least as plausible as the hewing down of the walls.

  2. “Masonry” and “winds” must be conjectures for the missing portion, but can either of these end in “‑num” as required? “Masonry” could be “wealstanas” [wall-stones] and “winds” could be “winde”, but neither ends with “‑num”, so I don’t know what Watson intended here. (Unfortunately the original source is lost: it was not captured by the Wayback Machine.)

For these reasons, I don’t find Watson’s reconstruction plausible, but no doubt Watson would say the same of mine!

Kluge reconstructed the line as follows (quoted in Krapp & Dobbie, eds. The Exeter Book, page 365), using the “geheawen” emendation, and with eleven destroyed letters:

Wunað giet se wealstan   wæpnum geheawen

The wall-stone yet remains, hewn by weapons

This is most plausible of any of the attempts here, but it requires two emendations to the manuscript, and I’m not keen on singular “wall-stone”—I would prefer the plural “wall-stones”, except that’s not possible due to the singular verb “wunað”. Perhaps Kluge intends us to take the singular as poetically standing for the plural.

The range of possible completions indicates that a secure reconstruction of the line is not possible. And this was the easiest case: the other affected lines have a larger proportion of missing text.

  • 2
    Excellent answer, thank you. I had not reckoned with the additional challenges of trying to decipher a handwritten script in a dead language.
    – Matt Thrower
    Commented Jun 27 at 19:18

I know next to nothing about old english, however I wanted to comment on the statistical aspect of text reconstruction from a partial text.

LM (language model) technology is now at a point where given a large enough corpus of text in a certain language, it is possible to predict which completions under masked regions of text would be more likely. BERT for instance, was one of the first LLM with this feature. This in principle can work even for dead languages, assuming the existing corpus is large enough.

Now, depending on how the inference is performed, you can find many completions of the masked/missing region that would be deemed statistically most likely, but in order to have reasonable certainty that a particular replacement is the correct one, it has to be substantially more likely than neighbouring replacements.

A more or less recent example of success reconstructing ancient greek texts is here. The peer-review article is here.

  • Not having a ChatGPT account, I tried this with Google Bard and it came up with "wundurnum geheapen (heaped with wonders)", where "wundurnum" is not a word—the dative plural of "wundor" is "wundrum"; compare Beowulf 1454. Commented Jun 29 at 14:56
  • 1
    Hi @GarethRees, I wouldn't be surprised, as most commercial and public LLM are trained mostly on english datasets from internet content, for such applications it would be much better to have a separate model trained entirely on the target language without any modern contamination. Commented Jun 29 at 16:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.