Experiments on samples of iron and rock held at immense pressures have led to new ideas of how Earth's core formed.
Scientists from Stanford University have shown that iron metal will flow through rocks 1,000km beneath our feet.
Using sophisticated X-ray imaging, they watched molten metal moving through rocks, squeezed to huge pressures between the tips of pairs of diamonds.
Their results suggest that Earth's core did not form in a single step, but grew in a complicated sequence over time.
The depths of Earth are complex and multi-layered.
At the surface, the rocks forming the foundations of our cities, the stones that we build our lives upon, also provide the raw materials for society - metals, fuel, water and nutrients.
These are no more than a thin geological veneer on the planet. In many respects, the deep Earth remains as much of a mystery as Jupiter or Mars.
But new research in the journal Nature Geosciences gives new clues about how Earth may have taken shape and built its core.
A group of scientists, led by Stanford's Prof Wendy Mao, have shown how metallic iron may be squeezed out of rocky silicates more than 1,000km beneath the surface to form a metallic core.
If you were to follow Jules Verne on a journey to the centre of the Earth, you would find a chemistry dominated by just three elements, until you got almost half the way to the centre - that's the first 3,000km of your journey.
Oxygen, silicon and magnesium (plus a little bit of iron) make up more than 90% of Earth's blanketing "ceramic" mantle.
Electrically and thermally insulating, the mantle is like a rock-wool blanket around the core. The minerals of the mantle are the stony part of the planet. But as you delve deeper on this "thought field trip", things suddenly and drastically change.
With more than half your journey ahead of you, you cross a boundary from the stony mantle into the metallic core. It is initially liquid in its upper stretches, and then solid right the way to the centre of the Earth.
The chemistry changes too, with iron forming almost all of the core, segregated into Earth's dense inner sphere.
The boundary between the metallic core and rocky mantle is a place of extremes. Physically, Earth's metallic liquid outer core is as different to the rocky mantle that overlies it as the seas are from the ocean floor here near Earth's surface.
One might (just about) imagine an inverted world of storms and currents of flowing red-hot metal in the molten outer core, pulsing through channels and inverted "ocean" floors at the base of the mantle.
The flowing of metal in the outer part of the core gives Earth its magnetic field, protects us from bombarding solar storms, and allows life to thrive.
How Earth's core came about has puzzled Earth Scientists for many years. Experiments on mixtures of silicate minerals and iron, cooked up in the laboratory, show that iron sits in tiny isolated lumps within the rock, remaining trapped and pinned at the junctions between the mineral grains.
Droplets of iron
This observation has led to the view that iron only segregates very early in the life of the planet, when the upper part of the rocky mantle was in fact super-hot and molten.
It is thought that droplets of iron rained down through the red-hot magma ocean to settle at its base, resting on the solid deeper mantle, then sinking as large "diapirs" driven by gravity through the solid mantle to eventually form a core.
The paper by Crystal Shi and Wendy Mao begins to paint a different picture.
"We know that Earth today has a core and a mantle that are differentiated. With improving technology, we can look at different mechanisms of how this came to be in a new light," said Prof Mao.
Using intense X-rays to probe samples held at extreme pressure and temperature squeezed between the tips of diamond crystals, the researchers find that when pressure increases deep into the mantle, iron liquid begins to wet the surfaces of the silicate mineral grains.
This means that threads of iron can join up and begin to flow in rivulets through the solid mantle - a process called percolation.
It also means that iron can begin to segregate if the rocks are deep enough, even when the mantle is not a molten magma ocean.
"In order for percolation to be efficient, the molten iron needs to be able to form continuous channels through the solid," Prof Mao explained.
"Scientists had said this theory wasn't possible, but now we're saying - under certain conditions that we know exist in the planet - it could happen. So, this brings back another possibility for how the core might have formed."
Commenting on the results, Geoffrey Bromiley, of the University of Edinburgh, UK, who was not involved in the study, told the BBC: "This new data suggests that we cannot assume that core formation is a simple, single-stage event. Core formation was a complex, multi-stage process that must have had an equally complex influence on the subsequent chemistry of the Earth.
"Their deep percolation model implies that early core formation can only be initiated in large planets. As a result, the chemistry of the Earth may have been 'reset' by core formation in a markedly different way from smaller planets and asteroids.
"As such, we might not be able to use geochemical data from meteorites to constrain the bulk composition of the Earth. This is currently an important assumption pervading Earth Science."
The results were reliant on recent advances in 3D imaging of minuscule samples using powerful synchrotron electron accelerators that generate intense beams of X-rays.
Similar to medical imaging, these sorts of experiments are revealing the nanoscale properties of minerals and melts. But they are also leading to new understanding of how huge objects like planets form and evolve.
Dr Bromiley and his colleagues are now investigating the influence of other factors, like the deformation that asteroids and other bodies might have experienced on their chaotic pathways through the early Solar System, on their formation.
He added: "The challenge now lies in finding a way to model the numerous processes of core formation to understand their timing and subsequent influence on the chemistry of not just the Earth, but also the other rocky bodies of the inner Solar System.
"We are increasingly observing metallic cores in bodies much smaller than the Earth. What process might have aided core formation in bodies that were never large enough to permit percolation of core forming melts at great depths?"