Imagine you are trying to piece together a jigsaw puzzle, but someone has told you that ninety-five per cent of the pieces are missing — and then, crucially, has told you that the missing pieces were not selected at random. The pieces that survive are those that were made of harder material, stored in drier rooms, and considered more valuable by the people who moved house five hundred years ago. You have no way of knowing, from the surviving pieces alone, what the missing pieces showed.
This is the situation ancient historians face. Not only is the evidence incomplete — it is incomplete in patterned, non-random ways that the evidence itself cannot fully reveal. We know enough to know that we are missing something. We do not always know enough to know how much, or which parts, or what they would have changed.
This generates a question that goes to the heart of what ancient history can claim to know:
Consider the Library of Alexandria — the most famous emblem of ancient knowledge lost. Popular imagination pictures a single catastrophic fire consuming hundreds of thousands of irreplaceable scrolls in a single night. The reality, as historians have established, is more complicated and more instructive. There was no single burning. The Library suffered multiple periods of damage across several centuries, beginning with Julius Caesar's fire in 48 BCE and continuing under successive rulers. By the time the Arab conquest of Egypt in 642 CE is sometimes blamed for a final destruction, most scholars believe the great collection had already been depleted across centuries of neglect, underfunding, and political instability.
The Library of Alexandria is itself a fragmentary record — we know about it mainly through references in ancient texts, most of which postdate its supposed peak. What it actually held, how much survived different periods of damage, and precisely when it ceased to function as a functioning scholarly institution are all contested. The greatest symbol of ancient knowledge lost is itself partially unknowable. This is the character of the fragmentary record: it applies even to our knowledge of the fragmentation.
By the end of this article, you will understand not just that ancient evidence is incomplete, but why it is incomplete in the specific ways it is — and what analytical strategies exist for working honestly and productively with that incompleteness.
The scale of the loss
Before examining why the ancient record is fragmentary, it helps to appreciate the scale. The numbers, where we can estimate them, are humbling. Ancient catalogues, cross-references in surviving texts, and the testimony of ancient scholars allow us to calculate rough survival rates for some of the most celebrated bodies of ancient writing.
Aeschylus, Sophocles, Euripides
Nine books of lyric verse
Published dialogues, praised in antiquity
Cicero alone delivered hundreds
142 books originally written
Popular stories, wisdom texts, spells
Three causes of fragmentation
The loss of ancient evidence did not happen in a single catastrophe. It was the result of at least three distinct and overlapping processes, each with its own logic and its own implications for the pattern of what survives.
The transmission bottleneck
Most surviving ancient texts did not travel directly from antiquity to modern times. They passed through one or more transmission bottlenecks — periods in which a small number of manuscripts were copied and the rest were lost. The most important bottlenecks for Western classical literature were the Byzantine scholarly tradition (which preserved most of our Greek texts), the Carolingian Renaissance of the 9th century (which revived interest in Latin texts and prompted a wave of monastery copying), and the Islamic Golden Age (which preserved and transmitted Greek scientific and philosophical works that had largely disappeared in the Latin West).
Each of these bottlenecks had its own priorities. Byzantine scholars valued literary prestige and philosophical utility. Carolingian monks valued texts useful for Christian education and rhetoric. Islamic scholars valued Aristotle, Plato, Galen, and Euclid — philosophy and science — and were less interested in Latin poetry or Roman historical writing. What emerged from each bottleneck was a different selection, shaped by the values and needs of the copyists, not by any principle of comprehensive preservation.
This means that the ancient texts we read today are not the ancient texts — they are the subset of ancient texts that happened to be valued by the specific institutions and cultures that transmitted them across the medieval period. This is a profound limitation that shapes every claim made about ancient literature, philosophy, and ideas.
What the pattern of survival hides
The systematic character of the selection means that the fragmentary record is biased in specific, identifiable directions. Understanding these directions is not merely an exercise in historical humility — it is an active analytical tool. If we know that the selection process systematically excluded certain kinds of text, we can make careful inferences about what the missing material likely contained.
Élite over popular: Literary texts valued by educated copyists survived; popular entertainment (mime, farce, popular song) was rarely copied and is almost entirely lost.
Male over female: Women's writing was systematically undervalued by the institutions — monasteries, Byzantine academies — that controlled copying. Sappho is the extraordinary exception.
Centre over periphery: Rome's literary culture is far better represented than provincial literature; Athens dominates over other Greek city-states; the Nile valley over the rest of Egypt.
Dry over wet: Climate determines material survival. Egypt's papyri exist in abundance; Britain's are vanishingly rare except in waterlogged anaerobic conditions like Vindolanda.
Three case studies follow, each illuminating a different dimension of the fragmentary record: what the selection process cut out, what deliberate destruction erased, and what was never recorded to begin with. Together they demonstrate that the fragmentary record is not a uniform condition — it has specific causes, specific consequences, and specific implications for specific historical questions.
Sappho of Lesbos (c. 630–570 BCE) was regarded throughout antiquity as one of the greatest lyric poets who ever wrote in Greek. Plato called her the tenth Muse. She commanded the highest fees for her poems in the ancient book trade. Ancient catalogues record nine complete books of her verse — somewhere between five hundred and a thousand poems.
We have fewer than 650 surviving lines. One complete poem — the Hymn to Aphrodite — and a handful of near-complete poems; the rest are fragments of one, two, five, ten lines, pieced together from quotations in other authors' works and from scraps of papyrus recovered from Egyptian rubbish heaps since the 19th century. New fragments still emerge: a significant new Sappho poem was identified on a papyrus in 2004 that had been sitting in a University of Cologne collection, unexamined, for decades.
What happened to the rest? Medieval monastic scriptoria — the institutions that determined which ancient texts would be copied and preserved — were ambivalent about Sappho. Her frank treatment of desire, including desire between women, made her texts uncomfortable for communities dedicated to Christian chastity. Some medieval commentators mention her explicitly as an example of moral vice. The manuscripts that might have preserved her nine books were not copied. They were not burned — they were simply not considered worth the considerable labour and expense of reproduction. They survived into late antiquity, then decayed when no institution replaced the copies.
In 1073 CE, Pope Gregory VII is alleged to have ordered Sappho's works burned in Rome and Constantinople, though the historical evidence for this specific act is thin. Whether the burning happened or not, the effect had already been achieved by three centuries of non-copying. Neglect accomplished what destruction might have done more dramatically.
On 12 July 1562, in the town of Maní on the Yucatán Peninsula of Mexico, the Franciscan friar Diego de Landa oversaw the burning of dozens of Maya books — codices made from bark paper and animal skin — along with five thousand cult images and numerous other objects he identified as idolatrous. De Landa recorded the event himself: he noted that the Maya "regretted this greatly and it caused them much pain." He had destroyed, in an afternoon, texts that encoded centuries of Maya astronomical observation, ritual calendars, genealogy, mythology, and historical memory.
Only four pre-Columbian Maya codices are known to survive today, preserved in European libraries where they had been sent as curiosities before de Landa's burning. The Dresden Codex (now in Germany), the Madrid Codex (Spain), the Paris Codex (France), and the Grolier Codex (Mexico) together constitute a tiny fraction of what the Maya intellectual tradition had produced across more than a millennium.
This is an extreme case, but not a unique one. The Spanish destruction of Aztec records following the conquest of Tenochtitlán in 1521 similarly eliminated vast bodies of Central American historical and scientific knowledge. The burning of Carthage's libraries by Rome in 146 BCE — if the ancient tradition is accurate — destroyed whatever Punic literature and records existed. In each case, deliberate destruction served a political and cultural purpose: eliminating the intellectual infrastructure of a conquered civilisation made that civilisation easier to reinterpret, assimilate, or dismiss.
Some of what we don't know about the ancient world was never recorded at all. The enslaved populations of ancient Greece and Rome — estimated at between 30 and 40 per cent of the total population in some periods and regions — left almost no direct written testimony of their experience. This is not because their records were lost or destroyed. It is because literacy itself was largely controlled by the free population, because enslaved people rarely had access to writing materials or the time and safety to produce personal documents, and because no ancient institution had any interest in systematically recording slave experience for posterity.
What we know about ancient slavery comes almost entirely from the perspective of slave-owners: legal texts governing the purchase and manumission of slaves; philosophical discussions of whether slavery was natural (Aristotle) or conventional (Stoics); comedy and satire featuring stock slave characters; inscriptions recording the names and occupations of freed slaves; occasional anecdote in historical narratives. From all of this we can infer a great deal about the institution of slavery. We can recover, very partially, the experience of individual enslaved people — particularly those who were eventually freed and left epitaphs, or those (like Epictetus, the Stoic philosopher) whose later lives were recorded.
What we cannot do is hear the enslaved speak for themselves, in their own words, about their own experience. The silence is not an accident or a catastrophe. It is the product of a social structure that systematically denied enslaved people the conditions for self-documentation. This is the most difficult kind of absence for the historian: not something that was lost, but something that was never created.
The argument from silence — the most treacherous move in ancient history
The argument from silence (argumentum ex silentio) is the move of inferring something from the absence of evidence. It is one of the most powerful and one of the most dangerous tools in the ancient historian's kit. Used carefully, it can establish genuine historical knowledge. Used carelessly, it produces confident claims with no evidential basis whatsoever.
The key is understanding that there are at least three different kinds of silence in the ancient record, and they require quite different analytical responses.
Writing productively about what we don't know
The fragmentary record is not only an obstacle — it is itself historical evidence, and can be incorporated productively into analytical responses. The key is to treat gaps in the record as objects of analysis rather than as confessions of ignorance. Rather than writing "we cannot know what slaves thought because there are no sources," write "the systematic absence of enslaved voices from the literary record is itself evidence of how ancient literacy was controlled and who it served — and can be supplemented, if partially, by reading the legal and comic traditions against the grain."
This move — from absence-as-obstacle to absence-as-evidence — is one of the most sophisticated things a student of ancient history can do. It signals genuine methodological awareness and earns marks at the highest band levels across all Australian curricula.
Are we creating a new dark age?
The ancient historian's experience of fragmentation has an urgent contemporary parallel. Digital formats become obsolete faster than parchment decays. The floppy disks of the 1980s are already largely unreadable. The word-processing formats of the early 1990s are degrading. NASA has lost original high-resolution footage from the Apollo missions — recorded on magnetic tape that has partly deteriorated, in formats that no surviving machine can easily read. The documents produced on early government word processors in the 1980s and 1990s are, in many cases, inaccessible.
A future historian trying to reconstruct the early 21st century may find that the period is better documented by physical newspapers than by the digital archives of the same newspapers — because the newspapers were printed on acid-free paper designed to last centuries, while the digital files require continuous hardware, software, and institutional maintenance to remain readable. This is the digital preservation paradox: we are producing more information than at any point in human history, but may be preserving less of it durably than any previous civilisation.
The ancient historian's habits — awareness of what survives and why, sensitivity to the gap between the record and the reality, scepticism about assuming that what survives is representative — are exactly the habits a responsible archivist, journalist, or public administrator should develop for the digital present.
The destruction of heritage as a war crime
The deliberate destruction of cultural heritage — as practised by the Islamic State in Palmyra (2015), the Taliban at the Bamiyan Buddhas (2001), and in countless other conflicts — is now recognised as a war crime under international law. The Rome Statute of the International Criminal Court defines intentional attacks on cultural heritage as a war crime when directed against civilian objects. Ahmad al-Faqi al-Mahdi was convicted by the ICC in 2016 partly for the destruction of religious and historical buildings in Timbuktu, Mali.
This legal development reflects a historical insight: cultural heritage is not merely ornamental. It is the material record through which a community understands its own history, identity, and continuity. Destroying it — as de Landa understood when he burned the Maya codices, and as the Islamic State understood when it bulldozed Nimrud — is an act of historical erasure. It attacks not just buildings and objects but the capacity of a people to know their own past. The ancient historian's sensitivity to what is lost when records are destroyed is not antiquarian sentiment — it is directly relevant to some of the most contested legal and ethical questions of the contemporary world.
New tools, new recoveries
The fragmentary record, while permanent in its losses, is not entirely static. New discoveries continue to recover material that seemed lost beyond retrieval. The Herculaneum papyri — carbonised scrolls buried by Vesuvius in 79 CE — are now being virtually unrolled using X-ray phase contrast tomography, allowing scholars to read texts that have been sealed and unreadable since the eruption. The first substantial text recovered by this method, announced in 2023, proved to be a previously unknown philosophical work.
Ancient DNA analysis is recovering biological information about ancient populations that no written source preserves. Ground-penetrating radar and LiDAR are revealing buried structures that no ancient text describes. The Oxyrhynchus papyri, recovered from an Egyptian rubbish heap beginning in 1897, are still being edited and published over a century later — each new publication bringing fragments of Sappho, Menander, early Christian texts, and countless private documents back into the light.
The fragmentary record is not a fixed object. It changes shape as technology advances and as new excavations are undertaken. The ancient historian's work is never finished — which is one of the reasons it remains endlessly generative as a scholarly discipline.
Looking ahead: from fragmentation to historiography
Article A5 takes the implications of the fragmentary record into historiography — the study of how historians have interpreted the ancient world differently across time. If the record is always fragmentary, then every generation of historians builds its picture of the ancient world partly from evidence and partly from the assumptions, values, and questions that their own era brings to that evidence. A5 asks: what does it mean that our understanding of the ancient world changes, not just because new evidence is found, but because the questions we ask of the same evidence change?