The First Eyes
The second example is the eye, and it comes from the very start of the animal fossil record.
Trilobites, armored sea creatures distantly related to today's crabs and insects, are among the first animals with hard parts to appear in the rocks. The oldest are about 521 million years old, and they already had eyes. Not light-sensitive patches, but compound eyes, the design still found today in bees and dragonflies: a dense array of separate units, each pointing in a slightly different direction, together forming an image.
There is a simpler eye on record, and it deserves its place here. In 2017, researchers described a trilobite from Estonia, about 530 million years old, whose eye has about a hundred units spaced well apart and no lenses at all. Its discoverers suggested it may be the earliest eye anyone will ever be able to find, because the rock layers beneath it hold only faint traces of soft-bodied animals that didn't fossilize. But even this eye is already a compound eye, built on the same basic plan as a modern insect's. Its discoverers concluded that compound eyes have changed little in half a billion years.
What came soon after was far more sophisticated. Trilobite lenses were made of calcite, a hard mineral, and a study published in Science in 1973 confirmed that they were mineral in life, not turned to stone after death. Because the material survives essentially unchanged, researchers have been able to work out how well these lenses focused, the way an optometrist would describe a lens today. And by about 515 million years ago, rocks in South Australia preserve the eyes of a large predator called Anomalocaris, each with more than sixteen thousand lenses. A housefly has about three thousand. Only a dragonfly, with about thirty thousand, is known to have more. The researchers who described them concluded that the compound eye arose very early in the arthropod line, earlier than some features usually thought of as basic to it.
Couldn't a lens be the kind of thing that either works or doesn't? It's tempting to think so, but physics says otherwise. A slightly curved surface focuses slightly, and a light-sensitive patch with no lens at all still tells an animal which direction the light is coming from. In 1994, two Swedish researchers, Dan-Eric Nilsson and Susanne Pelger, worked out a series of small steps from a flat patch of light-sensitive cells to a focusing eye, each step a slight improvement on the last. Using deliberately cautious assumptions, they estimated that it could take only a few hundred thousand generations. This is the standard answer to anyone who says eyes evolved too fast, and it deserves a fair hearing.
It also deserves a careful reading. The model starts with light-sensitive cells already in place; the authors explicitly set aside how those cells arose. It models the shape of the tissue, not the genes that build it, the nerves that connect it, or the brain that has to make sense of an image. In that respect it's like the Weasel program from Chapter 8: it shows that a path exists if every step is available and every step gets kept, not that the necessary changes would actually turn up in time. And an eye is worthless without something to read it. A lens that forms an image for no brain to interpret is just a window.
Last revised 2026-11-19