You have seen the picture even if you have never studied a day of biology. A hunched, knuckle-dragging ape shuffles along on the left. Moving rightward across the page, each figure straightens a little more, loses a little more hair, until on the far right a tall, upright, spear-carrying man strides confidently into the future. It is called, appropriately, "The March of Progress," and it has been reprinted, parodied, and half-remembered by more people than have ever read a paleoanthropology textbook.

It is also wrong, in a way that matters.

The March of Progress compresses millions of years of branching, overlapping, occasionally colliding lineages into a single confident line, as if evolution were a relay race with one runner at a time carrying the baton toward "modern human." For most of the twentieth century, this was more or less how professionals pictured it too, even if the professional versions came with more caveats and fewer spears. Different regional fossils were slotted into a single trunk, and the differences between them were treated as regional variation on one theme.

Then the field got a new set of eyes, and everything got more complicated in the way that science gets more interesting: not by producing a simpler story, but a more accurate one.

Starting in the mid-2000s, researchers learned to pull usable genetic sequences out of bones tens of thousands of years old — bones that had been sitting in museum drawers and cave sediments for far longer than DNA was supposed to survive. This is a genuinely difficult technical feat. Ancient DNA is degraded, contaminated with the DNA of every bacterium and human handler that has touched the specimen since, and present in vanishingly small quantities. The techniques that made it possible to extract, sequence, and confidently interpret this kind of material earned Svante Pääbo the Nobel Prize in Physiology or Medicine in 2022, specifically for his work sequencing the Neanderthal genome and identifying the previously unknown hominin group we now call the Denisovans, from a single finger bone found in a Siberian cave.

What that genetic evidence showed was not a straight line. It was a network — separate branches that split from a common ancestor, went their own way for hundreds of thousands of years, and then, in several different places, at several different times, briefly reconnected. When two lineages reconnect and produce fertile offspring, geneticists call it admixture. You and almost everyone you have ever met carry the genetic residue of admixture events that happened tens of thousands of years ago, on the other side of the world, between your ancestors and people who were not quite the same species as them — or who may not have been a different species at all, depending on how you draw the line.

This book is about that story: not the March of Progress, but what has replaced it. It centers on a framework called the Multi-Regional Ancestral Synthesis, or MAS, which organizes a great deal of already-published genetic research around a simple, testable idea — that different human populations picked up their archaic ancestry from different sources, in different places, at different times, and in different amounts. It is not one clean pulse of mixing. It is a mosaic. And, as we will see in the book's final chapters, it may be an illustration of something even bigger: that hybridization itself, not just mutation and selection working alone, has been a genuine engine of evolutionary change.

Before we get to the evidence, one thing needs saying plainly, because it will matter for how you read everything that follows: this book is not announcing a discovery. Every individual finding it draws on — the Neanderthal admixture, the two waves of Denisovan mixing, the ghost lineages hiding in African genomes — was published by other scientists, in other papers, well before this synthesis existed. What MAS offers is a name, an organizing lens, and a testable causal claim about why the pattern looks the way it does. That is a real contribution, but it is a different kind of contribution than a new dataset, and the distinction is worth respecting rather than blurring.

With that said, let's meet the cousins.