Introduction

A river has my heart. It’s not the austere, black Thames winding through London, where I was born, but a lazy green one 5,000 miles away, where I spent my adolescence: the Blanco River in Texas. My maternal ancestors have dipped into its waters for generations, as I have on countless summer days.

The Blanco is a tributary of the San Marcos, which flows into the Guadalupe, and on into the Gulf of Mexico. You can probably picture how this looks on a map because all river networks look similar, creeping through the landscape, merging into ever wider and longer channels, downhill to the sea. The pattern resembles twigs on branches that connect to trunks of trees (and the branching of their root systems, too), and it likewise resembles the veins of plant leaves, our own systems of blood vessels, and train and highway networks that feed into cities.

There’s something appealing about this ubiquitous pattern, so appealing to me personally that I have it tattooed on my forearm: the silhouette of a tree, with leafless branches reaching upward and roots burrowing downward, almost in mirror image. “The shapes of rivers and leaf vasculature and so on — branching networks — you can just about grasp the pattern, but it’s still chaotic, so there’s something fascinating with that,” said Chris Paola, a river scientist at the University of Minnesota.

Systems that branch in this way are “transport networks”: They transport some fluid substance (water, blood, traffic) from every place to a single place (the sea, a heart, a city center). Of the various examples, rivers are especially revealing, I think, since they arise from neither biological evolution nor urban planning, but rather chaotic Earth processes. Yet they obey simple, universal laws.

A discovery about river networks in 2026 reignited my curiosity about their universal form and mathematical nature. These were hot topics in the 1980s and ’90s, when rivers were studied as natural examples of “fractals”: mathematical objects whose features repeat in roughly similar forms at many different scales. Geomorphologists, who specialize in the shape (and continual reshaping) of Earth’s surface, have studied the geometry of river networks far longer, since the late 1800s.

Though many details of river behavior are still being actively studied, the existing mountain of research has yielded explanations that add up to a somewhat satisfying basic understanding. The math is elegant, the geophysics is intuitive, and still my sense of wonder is undiminished.

Every square inch of land on Earth’s surface receives precipitation, and much of it drains out, eventually, to an ocean or lake. Rivers are the drainage system.

In 1957, a U.S. Geological Survey scientist named John Hack discovered the most important law of river networks. In rivers and streams in Virginia and Maryland, Hack measured the length of each stream and the area of the land that slopes toward that stream and therefore drains into it, called its basin or drainage area. What he discovered is now known as Hack’s law: Any stream, from the littlest brook to the mightiest river, has a length that’s proportional to its drainage area raised to the power of 0.6. (In symbolic form: L ~ A0.6.) There’s a bit of variance around that 0.6 value — Earth is, after all, a complicated place — but “the general regularity of the relation is nevertheless remarkable,” Hack wrote. “Stream lengths tend to increase proportionally to the 0.6 power of the drainage area, regardless of the geological or structural characteristics of the area.”

NASA/UPI

As more and better data has accrued, especially from satellite imagery, Hack’s law has held worldwide. Why this is the case is the essential mystery geomorphologists have grappled with ever since. “Hack’s law is still the big question,” said Hansjörg Seybold, a geomorphologist at the Institute for Interdisciplinary Mountain Research at the Austrian Academy of Sciences.

It’s not so surprising that the bigger the land area of the basin, the longer the stream that drains it. But in a purely mathematical sense, one might expect that stream length would follow a slightly different power law. Imagine a square patch of land. You might guess that regardless of slope or size, in idealized form, the land would drain into a stream that’s the length of one of its sides — a vertical line down the middle, for example. That length is the square root of the area — or A to the power of 0.5.

Under that circumstance, big river basins would have the same proportions as the small river basins that feed the tributaries within them. Their structure would be the same, regardless of size. But that’s not what Hack’s law reveals.

Addictive Stock Creatives/Alamy

Instead, as a drainage areas get larger, the length of their streams increases faster. “A nice way to phrase it would be that small basins are short and squat, and large basins are long and thin,” said Daniel Rothman, a geophysicist at the Massachusetts Institute of Technology. We unknowingly pick up on this pattern when we look at a network of tributaries on a map; a perfectly self-similar, fractal river network wouldn’t look quite right. Basins and streams become elongated at larger scales, so that river networks have an inherent directionality that stretches toward the sea. One result of that elongation is that neighboring river networks must lie closer together than they would with a 0.5 power law.

But where does this pattern come from?

There are at least two complementary explanations underlying Hack’s law. One theory, proposed in seminal papers by the Italian hydrologist Andrea Rinaldo and colleagues in the early 1990s, says that river networks are elongated at large scales because this is the most energy-efficient way of transporting water downhill.

Imagine designing a drainage system for that square field I mentioned earlier. Rain falls uniformly on the field, and you need to ferry all the water out through a single point. You might try digging many direct channels, leading from every part of the field straight to the outlet. The problem, though, is that this would require an extremely long channel that must be dug (or, if you are trickling water, eroded), an inefficient process.

A better approach is a treelike system of channels with tiny streams merging into larger tributaries and, eventually, one main trunk leading to the outlet. Individual raindrops may travel farther than they did before, but the network as a whole dissipates less energy (through friction) to drain the field, because much more of the infrastructure is shared. When large streams are elongated, more of the shorter tributaries can share those major channels, and this reduces the overall transport cost. The pioneers of this “optimal channel network” theory simulated a bunch of river networks on a computer and showed that those that abide by Hack’s law, with an exponent of 0.6, dissipate the least energy. This scaling law is the most efficient and least resistant way to transport all that rain to the sea.

Implicit in this argument, though, is the assumption that rivers naturally reconfigure themselves to minimize total energy expenditure. But how? That’s where landscape evolution models come in.

Rivers do shift their layouts all the time. In the 1990s, in parallel with the work on optimal channel networks, geomorphologists developed powerful landscape evolution models to capture this constant adjustment and show the mechanism by which Hack’s law etches itself on the landscape. These computer simulations start with water flowing downhill, eroding rock as it goes. Tiny, random irregularities in the topography cause some channels to capture more runoff than others. Those channels in turn erode faster and deepen, which causes them to attract still more water. One streambed might grow toward its neighbor, and thereby intercept some of its runoff. The victorious stream grows longer and carries more water, while the losing stream shrinks or disappears. These sorts of local adjustments like these route water along ever more efficient paths. As the entire drainage network gradually reorganizes over thousands of years or more, it attains and then continues to tweak a configuration that transports water downhill with minimal energy dissipation.

Contains modified Copernicus Sentinel data (2017), processed by ESA

Gravity and friction are the driving forces of this process. Gravity supplies potential energy to flowing water. Friction, the cause of erosion, dissipates that energy. A channel configuration that wastes energy by forcing water along inefficient routes tends to erode rapidly and change. A configuration that routes water more effectively is stabler and therefore more persistent. The network becomes optimal through this dynamic evolution, eventually arriving at a form that adheres to Hack’s law.

That explanation of river network geometry hangs together for me, though geomorphologists still have many questions. Some study rivers that deviate from Hack’s law. Others organize transport networks that follow Hack’s law into one class of optimal transport networks, among a whole family of them. Trees, which branch in three dimensions instead of two, would be in a different class from rivers and follow different optimal scaling laws, for instance.

Now, geomorphologists have a new finding to explain. In April 2026, Tian Dong of the University of Texas, Rio Grande Valley and co-authors made the cover of Science for discovering that Hack’s law holds not only for rivers’ tributary networks, but also for their deltas, the fanlike structures that form where a river meets the sea.

Rivers essentially hit a brick wall when they reach the (nonflowing) ocean. The sudden deceleration of the water causes it to drop the sediments it carries. These pile up to form new land. In the process, the river’s water splits into a different kind of network of channels, which shift locations constantly as sediments build up and wash away.

Scientists told me that they’ve long wondered about the organization of channels in river deltas, but they are hard to study. Unlike the upstream river network, where slope and elevation differences make it easy to calculate the area of land that drains into any given tributary, deltas are flat and especially dynamic. But through a sophisticated analysis of satellite data that allowed them to distinguish land from water, Dong and his collaborators determined that the length of a channel in a river delta scales with the size of its nourishment area — the area that it supplies with sediments — raised to the power of 0.6. Rivers’ tributary networks and distributary networks are opposites — sediments are transported away from one end and deposited at the other — yet they abide by the same math. Geomorphologists are now considering why Hack’s law should apply in this inverse context.

Reflecting on my own question, I think it’s the coexistence of simplicity and determinism with chaos and randomness that makes the optimal structure of rivers so captivating. Natural efficiency is, perhaps, innately appealing to us.

Paola agreed. “There’s something about an artery of water moving through the landscape in this great big channel that we can’t stay away from,” he said. “It has so much structure. … The chaos is sort of what you expect from nature. You could think, water runs around on a landscape and makes puddles and stuff. And the order gives it a feeling of something more than just water and sand. I’m not saying that it actually is anything more than that. I don’t mean to be metaphysical. I think it strikes us that way and draws us in.”