It all started some 370 million years ago with a fish that had fins, a tail and a dream.

In those days, the world's oceans were swarming with weird and diverse creatures: fish with armored bodies and bony jaws, primitive sharks, crablike creatures called trilobites whose skeletons now litter the fossil record. But the land was relatively empty, colonized only by plants and a few bold invertebrates.

Our distant ancestors wanted to be part of that world. And in the absence of a sea witch to turn their flippers into legs, they decided to make do with what they had: their thick, muscled fins and tails, which they could use to drag themselves out of the water and onto the sand.

So instead of looking like this ...

... that early landlubber may have looked something like this:

At least, that's according to a new study published in the journal Science this week, arguing that a tail could have come in handy for a creature learning to walk on land (Ariel might have avoided a lot of trouble, had someone bothered to tell her). Using three models — one mathematical, one robotic and the third an animal that lives today, the authors illustrate that fish might have used their tails to propel themselves up sandy slopes of ancient beaches when they first crawled onto shore.

It's inelegant, acknowledged Daniel Goldman, a physicist at the Georgia Institute of Technology and an author of the study — a far cry from the gazelles and horses and cheetahs whose graceful gait we so admire.

"But they all descend from some critter that was hauling itself out on land," Goldman said. "How that happened, to me, was a fascinating scientific question."

Goldman has been obsessed with odd creatures for most of his life; he was the kind of kid who was constantly getting into trouble running after frogs and digging for bugs. "Somehow," he says, he wound up getting a PhD in physics (usually a fairly critter-less field). But he couldn't quite let go of his interest in animals, so he joined a lab that studies biomechanics — how living things move.

Physics is about "core principles of the natural world, right?" he said. "And the evolution of locomotion problems is as core as anything."

One of Goldman's graduate students, Benjamin McInroe, had built a robot to study how modern sea turtles move on sand when a colleague pointed out that the model could be helpful for paleontologists. Much like sea turtles, which must crawl over the beach to reach the water on flippers built for swimming, early terrestrial vertebrates would have been navigating solid ground with their waterlogged fins. Scientists don't know how these extinct creatures might have moved, but based on the fossil record they know the animals had tails.

"So Ben went to the 3D printer and printed out a tail" for the robot, Goldman said.

Loosely based on Ichthyostega, an extinct, salamander-like creature with an alligator's squat physique, the robot allowed Goldman and his colleagues to model various methods of motion the creatures might have used. But when they modeled it walking on slippery and sandy surfaces, as Ichthyostega and his friends would have done, it was clear that the form wasn't well-suited for life on land. Past research suggested that the creature's hind legs were basically useless. And in these experiments, the model had particular trouble trudging up sandy slopes — every time it moved one of its limbs, the sliding sand would carry it back down. It's the same thing that happens when a human tries to climb up a dune.

An improbable little creature called a mudskipper offered a way around that. These funny-looking fish swim in the ocean, but they also clamber around on land, using their fins to push themselves forward — a behavior called "crutching." When they need to ascend inclines, they use their powerful tails to launch themselves up and control their motion.

When Goldman tested this behavior on his robot and mathematical models, he found that it was far more effective than the simple dragging motion most paleontologists had been imagining when they envisioned early tetrapods.

"The tail kind of helps erase any kind of poor control in your limbs," Goldman said. "That was a big 'aha' moment because that says if you use your tail approximately it allows you to kind of buffer against poor limb structure control, and we wonder how well-controlled the earliest land locomotors could have controlled their limbs given that they were relatively primitive."

In an accompanying perspective published by Science, biologist John Nyakatura of the Humboldt University of Berlin noted that interdisciplinary efforts like this one offer a way to solve seemingly unanswerable problems from evolutionary biology. When organisms haven't walked for a hundred million years or more, paleontologists can only theorize about how they might have moved. Robots and mathematical models — two things not often seen in evolutionary biology labs, allow "new hypotheses to be generated and existing ones to be tested in a transparent and reproducible manner," he wrote.

But he cautioned that the findings from Goldman's robot models don't mean that early tetrapods definitely behaved as modern mudskippers do. To the extent that question ever can be resolved, it will be done by paleontologists comparing the anatomies of both creatures.

And Goldman acknowledges that fact.

"It's just a possible mechanism for how they might have moved effectively," he said.

But he believes that this method of using physics-style models to test theories has a place in paleontology. It lets scientists see functions they'd only imagined. It also gives them a much greater appreciation of the sophisticated behaviors they're attempting to recreate.

"It was an exercise in humility," Goldman said of building the robot model. "We had to combine all these discipline and spend all these years to describe how a little fish can crutch up a sandy hill. And the fish just does it."

Correction: An earlier version of this post misreported Daniel Goldman's first name.  

Read more: