The length of a day on Earth is not constant. The planet’s rotation has been slowing for billions of years, primarily because tidal friction transfers rotational energy from Earth to the Moon, which slowly moves outward to a higher orbit. The current rate of slowdown is small but measurable. According to the Wikipedia summary of the relevant atomic-clock and historical records, the day has lengthened by roughly 1.7 milliseconds per century in recent decades, and by about 2.3 milliseconds per century averaged over the past 2,700 years of astronomical observations. The cumulative effect of these tiny shifts is large. A day in the Mesozoic was shorter than a day now, a day in the Paleozoic shorter still, and a day in the Precambrian shorter again. The unusual thing is how we know the specific numbers.
The cleanest piece of evidence comes from fossils. In 1963, the paleontologist John W. Wells, then a professor at Cornell University, published a two-page paper in Nature titled “Coral Growth and Geochronometry.” Wells showed that modern corals deposit a thin daily growth layer on the outside of their skeletons, with finer daily bands grouped into broader monthly and annual cycles, in the same way that trees produce daily fluctuations within annual rings. Counting the daily bands within one annual cycle of a fossil coral gives the number of days the coral experienced in a year during the period when it was alive. If the year length has been roughly constant, but the day has been getting longer, then ancient corals should show more days per year than modern ones.
What Wells’s corals showed
Wells counted the growth lines on Middle Devonian coral specimens, which by independent radiometric dating were known to be roughly 385 million years old. According to the Paleontological Research Institution’s account of his work, Wells found that his Devonian corals carried between 385 and 410 daily growth lines per annual cycle, averaging about 400 days per year. Modern corals, growing under the same kind of seasonal cycles, deposit about 365.
The arithmetic gives the day length. A year is essentially the time the Earth takes to orbit the Sun, and on geological timescales that figure is stable. The year was about the same number of hours in the Devonian as it is now, roughly 8,766. Dividing 8,766 hours by 400 days yields a Devonian day of about 21.9 hours, slightly under 22. Wells’s result was the first direct fossil confirmation of the long-suspected slowdown in Earth’s rotation, and it was published the same year that geophysicists were starting to measure the present-day slowdown with the first generation of atomic clocks. The two lines of evidence agreed.
Wells later extended the technique to corals from the Pennsylvanian period, which began about 320 million years ago. The Pennsylvanian counts came out at about 385 to 390 days per year, implying a day of around 22.4 hours, slightly longer than the Devonian figure but still shorter than today’s. The pattern across the two datasets is the one tidal physics predicts. The further back you go, the more days per year, and the shorter each day was.
How far back the slowdown goes
Coral did not exist for most of Earth’s history, so the fossil method has a natural limit. For older times, scientists rely on different proxies, particularly the patterns of sedimentary layers laid down under tidal conditions and the cyclic signatures of Earth’s orbital variations preserved in deep rock. These techniques are less direct than counting coral bands, but they extend the rotational history back into the Precambrian, the long period before complex multicellular life appeared.
The Precambrian numbers are striking. According to a 2023 paper in Nature Geoscience by Ross Mitchell and Uwe Kirscher, the day on Earth may have stalled at about 19 hours for roughly a billion years during the mid-Proterozoic, between about 2 billion and 1 billion years ago. The proposed mechanism is a balance between two opposing tidal torques. The Moon’s gravity, acting on the oceans, was slowing Earth’s rotation. The Sun’s heating of the atmosphere, acting on a kind of atmospheric tide, was speeding it up. For about a billion years, the two effects appear to have cancelled, holding the day at roughly 19 hours until some climatic change broke the resonance. Earlier models had proposed a similar stall at 21 hours during the late Precambrian, broken roughly 600 million years ago around the time of the Marinoan or Sturtian global glaciations.
Whatever the exact figure for the Proterozoic, the broader trajectory holds. The Earth was rotating faster in the past, and faster still further back. According to the Marine Science Institute at the University of Texas, the coral and shellfish growth-band record suggests that 500 million years ago, around the time complex animal life was diversifying, the day was roughly 21 hours long, with a year of about 420 such days. The first dinosaurs would not appear for another 270 million years.
Why the slowdown happens
The underlying mechanism is straightforward in principle. Earth’s gravity pulls the Moon, and the Moon’s gravity pulls Earth’s oceans into a slight bulge facing the Moon. Because Earth rotates faster than the Moon orbits, the rotating ocean carries this tidal bulge slightly ahead of the Moon’s position in the sky. The bulge’s gravity exerts a small forward pull on the Moon, accelerating it into a higher orbit. By conservation of angular momentum, Earth’s rotation must slow to compensate. Lunar laser ranging experiments, using corner-cube reflectors left on the Moon’s surface by the Apollo astronauts, measure the Moon’s recession at approximately 3.8 centimetres per year.
The friction is not the only factor. Earth’s day length is also affected by mass redistribution within the planet itself, including post-glacial rebound, changes in the polar ice sheets, and even atmospheric and oceanic circulation. Some of these short-term effects can temporarily speed Earth up rather than slow it down. According to the Wikipedia summary, Earth’s rotation actually accelerated slightly during the years around 2020, and on 29 June 2022 the rotation completed in 1.59 milliseconds less than the standard 24 hours, setting a record for the atomic-clock era. The long-term trend, however, remains one of gradual slowing.
What this changes about how we see deep time
The practical takeaway is not particularly large. A 22-hour Devonian day or a 19-hour Proterozoic day does not change the calendar in any useful sense, since neither civilisation nor the calendar existed at the time. What it changes is a small, intuitive picture of geological time. Familiar reference points, the position of the Sun, the rhythm of day and night, the count of days in a year, were genuinely different in the distant past. A trilobite scuttling across a Cambrian seabed was experiencing a sun that crossed the sky in roughly 21 hours, not 24. The familiar planet underneath was rotating noticeably faster than it does now.
The reason we know any of this is a piece of natural bookkeeping. Coral, like the rings in trees and the layers in glaciers, is one of the materials that records its own time. Wells’s contribution in 1963 was noticing that the records were there, and counting them.
