Dinosaurs were covered in scales, not feathers.

The illustration appeared in textbooks for most of the twentieth century: a Tyrannosaurus rex with thick, grey-green scales, its skin bare and reptilian under the Cretaceous sun. The image was not based on direct fossil evidence of integument, soft tissue preservation in dinosaur fossils was rare enough that most reconstructors simply assumed an analogy with modern reptiles and drew scales accordingly. The assumption was rarely examined. Dinosaurs were classified as reptiles. Reptiles have scales. The question seemed settled before it was ever asked.

The first serious crack came from an unexpected source. In August 1996, a specimen was unearthed in Liaoning Province, in northeastern China, near the town of Beipiao. The farmer Li Yumin recognized its significance and sold it, in two slabs, to separate institutions: the National Geological Museum in Beijing and a museum in Nanjing. Ji Qiang, the Beijing museum's director, noticed that the fossil was surrounded by an impression of fine filaments along the neck and back, not scales, but something more like a fringe of simple feathers. Ji published a brief description in Chinese later that year. The creature would be named Sinosauropteryx prima, "first Chinese lizard wing."

Sinosauropteryx was a small compsognathid, not a bird. Its skeletal anatomy placed it clearly among the theropod dinosaurs. But the filaments preserved around its body were structural, not artifacts of decay, not bacterial mats, not aberrant scale formations. Subsequent studies using scanning electron microscopy found preserved melanosomes within the filaments, the microscopic pigment-bearing organelles that give feathers their color. Analysis of those melanosomes in 2010 allowed researchers to reconstruct Sinosauropteryx's coloration: reddish-brown with alternating bands along its tail, making it the first non-avian dinosaur to have its color known from the fossil record.

Sinosauropteryx opened a flood. Liaoning's lake-bed formations turned out to be extraordinarily productive, their fine-grained sediments preserving soft tissue that ordinary stone would destroy. Protarchaeopteryx and Caudipteryx appeared in 1998, both with clearly pennaceous feathers, feathers with a central shaft and branching barbs, structurally identical to the flight and body feathers of modern birds, arranged on animals whose skeletons made their theropod identity unmistakable. Sinornithosaurus and Microraptor followed. Anchiornis, described in 2009, was covered in feathers from head to toe, its plumage reconstructed in detail from melanosome analysis: black and white with chestnut facial patches.

The implications spread beyond Liaoning. Specimens long held in museum collections were reexamined. Analyses of Velociraptor forelimb bones revealed quill knobs, the roughened attachment points along the ulna where flight feathers anchor in modern birds. Quill knobs meant not just that Velociraptor had feathers, but that it had large, structured feathers along its arms: the anatomy of a feathered animal that could not fly.

By the mid-2000s, Mark Norell and Xing Xu of the American Museum of Natural History had compiled in a landmark Annual Review of Earth and Planetary Sciences paper the accumulating evidence that feathers had evolved well before flight, distributed across a wide range of theropod lineages. The developmental logic followed: feathers began as simple filaments, perhaps for insulation or display, and became elaborated over time in multiple lineages, with powered flight emerging only once, in the ancestors of modern birds. The scaly dinosaur of the textbook had never been reconstructed from evidence. It was an assumption transferred wholesale from the classification of dinosaurs as reptiles, and the assumption was wrong.