When the fossil is gone, every copy becomes a different kind of evidence

A replica of a Peking Man skull at the Paleozoological Museum of China in Beijing. It makes the central distinction visible: a copy can carry anatomical information without becoming the vanished original. Photograph by Yan Li, 2009, CC BY-SA 3.0, via Wikimedia Commons.[9]

The skull looks convincingly old. Its brow forms a heavy bar, its low vault catches a warm museum light, and every brown patch appears to promise direct contact with the Pleistocene. Yet the object in the photograph is a replica. It is a human-made reconstruction of Peking Man, the name long used for Homo erectus fossils from Zhoukoudian, near Beijing.[9]

Calling it “only a copy” would miss its scientific force. Casts of the Zhoukoudian material preserved three-dimensional form after the bulk of the famous collection vanished during the Second World War. Calling it equivalent to the originals would miss something larger. A cast keeps the surface selected by the casting process. A photograph keeps one view under one light. A drawing keeps the features an observer chose to emphasize. A table keeps measurements, but not the bone from which another measurement might be taken.

These records are not lesser versions of one interchangeable thing. Each is a different transformation of the specimen, with a different capacity to answer questions. Four histories—Peking Man, Spinosaurus, an early ichthyosaur, and fossil ostracods from Brazil’s Museu Nacional—show why paleontology can continue after catastrophic loss, and why even superb documentation cannot make that loss disappear.

A cast keeps a surface, not a specimen

The copy network around Peking Man began before anyone knew it would become a rescue system. In 1932, a brief report in Nature noted that anatomist Davidson Black had made casts of five jaw fragments, the skull found in 1929, and an endocranial cast before taking reproductions abroad. During the later 1930s, Franz Weidenreich’s studies added detailed descriptions and drawings to that record. The main group of fossils disappeared in 1941 during the Second World War; its final fate remains unknown.[1][2][3]

The familiar shorthand that “all the originals were lost” is too broad. Four early Zhoukoudian teeth survive at Uppsala University, one of them rediscovered in a long-unopened crate in 2011, and postwar excavations in China recovered additional original hominin material.[2][3] The distinction matters. Paleontology does not inherit a clean before-and-after rupture. It inherits a patchwork: first-generation casts of missing skulls and jaws, published descriptions and drawings, a few early teeth elsewhere, and new fossils collected from the site after 1949.

For decades, the casts and publications kept the missing animals available for comparison. Researchers could examine cranial proportions, visible sutures, tooth positions and other external anatomy. What they could not do was ask the vanished material to meet a method invented later.

A 2018 study of six original teeth recovered at Zhoukoudian after the war made that boundary unusually clear. Song Xing and colleagues noted that the loss of most earlier fossils had confined much subsequent work to casts, drawings and descriptions, preventing micro-computed tomography of those specimens. The surviving teeth could be scanned. Their volumes revealed the enamel–dentine junction and pulp cavity beneath the crown, including a highly crenulated internal surface that is invisible on an ordinary exterior cast.[3]

The lesson is not that the cast failed. It succeeded at the job for which it was made: carrying visible form across institutions and time. Internal dental anatomy was never inside that copy. Nor are original mineral composition, adhering sediment, microscopic tool marks or molecules available merely because the replica looks complete. The cast preserves an encounter with the fossil; the fossil preserves the possibility of another encounter.

That makes the cover photograph doubly honest. It records a physical museum object, not a digital fantasy, and the object itself records a scientific reconstruction. The brow and vault can still teach. The smooth brown surface also marks where the evidence changes category.

A camera can correct a drawing—and record a reconstruction

The Spinosaurus aegyptiacus holotype had a different afterlife. Fossil collector Richard Markgraf excavated the partial skeleton from the Bahariya Formation of western Egypt in 1912. Ernst Stromer described it in 1915: jaw material, teeth and vertebrae with the extraordinary neural spines that gave the animal its name. The rocks are early Cenomanian, roughly 97 million years old. The specimen, BSP 1912 VIII 19, was destroyed with most of Stromer’s Bahariya collection during an air raid on the night of 24–25 April 1944.[4]

For half a century, Stromer’s published descriptions and illustrations carried most of what researchers could know about the name-bearing skeleton. Then another record surfaced. Two prewar photographs, donated by Stromer’s son in 1995 and recognized in the museum archive in 2000, were analyzed by Joshua Smith and colleagues in 2006.[4]

The comparison vindicated Stromer and corrected him at the same time. His drawing of the right lower jaw was broadly faithful, but the photograph suggested that its front margin had been drawn too squarely: the researchers measured an angle of about 61 degrees in the photograph and about 71 degrees in the illustration. Some cracks and openings were schematic rather than exact. The wider museum view exposed a different filter. Three vertebral centra visible in the display were almost certainly reconstructions, and the order of some tall-spined vertebrae reflected a mount arrangement that Stromer himself later revised.[4]

Here a photograph does more than substitute for an absent bone. It audits a drawing. It separates anatomy from an illustrator’s decisions and, in the second view, fossil from exhibition restoration. Yet the photograph has its own hard frame. Perspective can change an angle. Lighting can hide a surface. Only the exposed side is available, and no researcher can rotate the jaw, inspect its broken edge, sample it or scan its interior.

This is why “the old image shows” should always be followed by a description of which image. A photograph of an isolated element, a photograph of a prepared mount and a line drawing are three observations with overlapping—not identical—content. With the original gone, their disagreements are not annoying noise. They are one of the few ways to detect where preparation and interpretation entered the record.

Two plaster skeletons can recover one lost animal

An early ichthyosaur supplies an even cleaner test. In 1819, the surgeon Everard Home introduced what was then the first complete ichthyosaur skeleton known to science. The Lyme Regis specimen, probably collected by Mary Anning, eventually entered the Royal College of Surgeons in London and was almost certainly destroyed in a Second World War bombing raid. Its published illustration survived.[5]

So, unexpectedly, did two plaster casts. Dean Lomax and Judy Massare identified them in the collections of Yale’s Peabody Museum and Berlin’s Museum für Naturkunde and compared them with the nineteenth-century drawing. The casts confirmed much of the illustration and clarified several bones. They also exposed discrepancies, especially in the forefins and hindfins. The evidence was sufficient to place the skeleton in Ichthyosaurus, but not to determine its species.[5]

That result is more useful than a triumphant “lost fossil found” headline. The original was not found. Two copies narrowed the taxonomic problem and tested the drawing, while preserving a limit that a new name could not honestly cross. They also showed why copies themselves belong in research collections. A cast may retain contours no illustration captured, but it can include a sculpted repair, a misplaced piece or detail softened by the molding technique. A second cast gives those alterations something against which to be checked.

Redundancy works best when it is heterogeneous. Ten copies made from the same processed file may share one invisible error. A cast, a photograph, a drawing and a written measurement fail differently. When they converge, confidence rises. When they conflict, the conflict points toward the step where information changed.

A microscopic photo can keep a diagnosis alive

On 2 September 2018, fire devastated the Museu Nacional in Rio de Janeiro. Among the losses were name-bearing fossils: the individual specimens against which zoological names are objectively anchored. For 23 species and subspecies of tiny fossil crustaceans called ostracods, however, researchers had another layer of evidence. In 2011, the museum’s imaging laboratory had photographed holotypes and paratypes with a scanning electron microscope.[6]

Those images became the basis of a 2025 reassessment. They show shells from Lower Cretaceous units in Brazil; the same taxa also occur in central West Africa, and some are useful in correlating rock layers across sedimentary basins. Scale bars allowed measurements; multiple views preserved ridges, nodes, grooves and outlines used in identification. Lucas Antonietto and colleagues could update several taxonomic combinations and publish a practical reference for recognizing the species without direct access to the destroyed types. Importantly, some corresponding paratypes remained in a separate Petrobras collection, so the physical evidence was damaged unevenly rather than erased everywhere.[6]

The SEM archive is more anatomically precise than an ordinary photograph and less open-ended than the specimen. Its resolution, viewing angle, detector settings and sample preparation were fixed in 2011. A future researcher can remeasure what the image contains. They cannot turn the missing shell to expose an unphotographed side, change the electron-beam conditions, test its chemistry or determine whether a subtle feature came from coating or the fossil surface unless the surviving metadata and other specimens resolve the issue.

Museu Nacional’s recovery has therefore joined the safeguarding of physical collections to digital curation. In May 2026, Rio de Janeiro’s state research foundation reported that the museum had captured about 22,000 collection images representing roughly 3,800 species, with type material representing about 650 of those species. A cloud-based collection-management system and a dedicated digital-collections laboratory were also part of the effort.[8] The point is not to choose between cabinets and servers. It is to prevent one building, one database or one finished rendering from becoming the only place where a scientific relationship survives.

Loss does not automatically create a replacement type

If a holotype is destroyed, why not simply select another specimen and continue? Zoological nomenclature allows that move, but deliberately makes it difficult. Under Article 75 of the International Code of Zoological Nomenclature, a neotype is not a routine replacement for a missing object. Its designation requires an exceptional need to clarify a taxon. The author must explain why the original type is believed lost, describe the proposed replacement, show consistency with what is known of the old type, provide evidence that it came as nearly as practicable from the original locality and geological horizon, and place it in an accessible research institution.[7]

Those conditions reveal what a type specimen does. It is not a prize for the most complete individual, and it is not an average body for the species. It fixes a name to a material reference. Replacing it can alter the practical meaning of that name, especially when a later fossil comes from another place, another layer or a population whose relationship to the original cannot be tested directly.

The surviving copy record is crucial here. Descriptions, photographs and casts can show whether a proposed neotype agrees with the lost material. They can also show that agreement cannot be demonstrated. The Code even anticipates the original returning: a rediscovered former type generally resumes its name-bearing role and the neotype is set aside, unless the Commission rules otherwise to preserve stability.[7]

Documentation therefore does not make replacement automatic. It can make restraint possible. Sometimes the strongest scientific act is to acknowledge that the missing holotype still anchors the old name, use its surviving documentation to bound comparisons, assign new material cautiously and wait for a better fossil or a clearer taxonomic problem.

Preserve the raw encounter

The four histories suggest a practical anatomy of copying. Written descriptions preserve interpreted characters and vocabulary. Measurements preserve chosen dimensions. Drawings can clarify edges that photographs obscure, but they select and regularize. Photographs preserve light from a viewpoint, often with preparation history visible. Casts carry three-dimensional surface geometry while potentially carrying restoration too. SEM and CT datasets preserve structure at specified scales and settings. Catalog records preserve identity, locality, stratigraphy and the chain that connects object to claim.

None is universally “best.” The useful question is narrower: best for what observation? Peking Man casts support external comparison but cannot disclose an enamel–dentine junction that was never molded. A Spinosaurus photograph lets researchers test a jaw angle rendered differently in a drawing while trapping the perspective from which the photographed angle was seen. Ichthyosaur casts recover morphology and still stop short of a species. Ostracod micrographs preserve diagnostic surfaces but cannot be rerun with tomorrow’s instrument.[3][4][5][6]

Good documentation should therefore keep more than the polished endpoint. It should retain scale, orientation, specimen number, locality and geological context; record which portions are original, restored or mirrored; preserve raw scans alongside processed models; and distribute copies and metadata beyond one room. The physical specimen remains central because future questions are unpredictable. The copies remain central because physical specimens are vulnerable.[6][8]

This is not an argument that fossils are immortal once digitized. It is an argument for making loss scientifically legible. When records state what they captured, how they were made and what they omit, later researchers can use them at full strength without asking them to impersonate the missing object.

The replica skull in the photograph is neither fake evidence nor a recovered fossil. It is a bounded witness. It carries a surface assembled from earlier observations, a history of scientific labor and an unmistakable absence. That is enough to matter.

Every copy preserves particular recorded features. An original leaves more ways to ask another question.

cronfeed.work