How the Master Clock Sets Time For The World
A strontium atomic clock at the National Institute of Standards and Technology is so precise that it will neither lose nor gain one second in about 5 billion years of continuous operation. (Image: Ye Group and Baxley/JILA)


This story is over 5 years old.


How the Master Clock Sets Time For The World

Getting it just right can take a while, and it starts right next to Joe Biden's house.

For the chief scientist charged with keeping America on time, Demetrios Matsakis may seem to possess a surprisingly casual attitude about timeliness. For instance, he doesn't wear a wristwatch.

"The smartest person I knew at the Naval Observatory was never on time. And as a graduate student, one of those scientists I most admired was not only always late but also had an extremely messy desk," he said. "I used to refer to 'excavating it.'"


Time is also very messy, and measuring it is a kind of excavation. Matsakis, a "sixty-something" physicist and astronomer with a tousled shock of silver hair, has been digging a lot. As the chief scientist and former department head of the US Naval Observatory's Time Services department, he has spent an inordinate amount of time with what's called the Master Clock. It is, essentially, America's fifty-year-old grandfather clock, the hidden instrument of Washington's dominion over the world's time.

At its heart, the USNO's Master Clock operates much like grandfather clocks do: with pendulums. More precisely: a set of the most sophisticated pendulums ever built, carefully counting the "swings" of atoms' radiation with a precision unknown anywhere else in the universe.

It is the ticking of these clocks that tells so many of the other clocks—the ones on our phones, on our computers, our webpages, our TV screens, our radios, and our GPS systems—from your geolocation app to your Predator drone—what is the thing called "the time."

In this context, a simple wristwatch ticking according to a quartz crystal might seem almost offensive. Matsakis likes to say he is often frustrated by expressions like "surgical precision" and "like clockwork."

"Anyone who has had an operation can tell you about the former," he said wryly, "and anyone who tries to combine precise clock data at the nanosecond level can tell you about the latter."


Matsakis's interaction with time began, naturally, with an attempt to fix a clock. He arrived at the USNO in 1979 to work as a radio astronomer, measuring the rotation of the Earth. In 1990, a new clock came in, an experimental, mercury-based one, and promptly started to malfunction. Matsakis was the only physicist on the premises, so his supervisor asked him to fix it.

"After about two months I couldn't figure out what was wrong, so I told the director, 'The things I've tried haven't worked out. If you find someone else to do this, I won't be offended.' He didn't, and eventually I figured out what was wrong." His foray into timekeeping—and, eventually, into overseeing the Master Clock—was, he said, "a total coincidence."

The primary aim of timekeeping is to measure the passage of a second. To be precise, that is equivalent to the amount of time it takes for the hyperfine radiation given off by a cesium-133 atom at its ground state as it transitions between energy levels, and its electrons oscillate exactly 9,192,631,770 times.

These days, counting a second depends upon firing a microwave beam at one of these cesium atoms and counting the effect on its electrons. At that scale, the slightest aberration can knock a clock off its count. And then there are the effects of gravity, which Einstein's theory of special relativity showed can shift the pace of time.

"At the nanosecond level—a billionth of a second—every clock has its own personality," Matsakis wrote in an email. Each clock will tick faster or slower at certain times, and generally, scientists can correct for this using software. The trickier part is understanding the rate–sometimes sudden, sometimes slow—at which a clock's ticking may be changing. "We must be on the lookout for deviations from the predicted behavior, and be sure to predict well."


This reporter first encountered the clock in his single-digit years. My father would sometimes raise an eyebrow at his calculator watch and pick up the phone (a landline). He'd punch in some numbers, listen for a moment, then hand the phone to me. Back then—in the '90's—there were few ways to set your watch precisely. There is still no way to get the time like this.

Tick tick tick tick… "U.S. Naval Observatory Master Clock," a voice would announce, in a full sentence, sounding always chipper, punctual. (It belongs, to the 1970s TV actor Fred Covington, whose IMDB entry includes the Master Clock.)

"At the tone, Eastern Standard Time, fifteen hours, nine minutes, exactly."

Pause. BEEP. Tick tick tick tick…

My dad passed the phone number down to me like an heirloom timepiece. I took comfort in the simple certitude of the clock and the authority of its voice. Even in the middle of the night, the Master Clock was there, constant, steady, offering the time to anyone who would listen. A symbol, I thought, amidst the bureaucracies and absurdities of government, of some kind of perfection, responsibility, magnanimity even. The name alone suggests that it could be there for an eternity, steadily ticking, offering a really good time anytime, always at the same number: 202 762-1401.

Call it, and you're connecting to a machine not far from the Master Clock vault, on a hill on the edge of Georgetown in Washington, DC. The Naval Observatory moved here in 1893 from a more central location downtown, because this beautiful spot (adjacent to the verdant estate of Dumbarton Oaks) was then still far from the city's light pollution, making it an ideal place for observing space, a process that's intimately connected to counting time. (Earlier this year, The Atlantic's Katherine Wells visited the Observatory and made a beautiful video, which is embedded at the bottom.)


Founded in 1830—it predates the Smithsonian and the national labs—the Observatory was intended as a depot for the Navy's navigational charts and instruments, with a mission to determine "the positions and motions of celestial bodies, motions of the Earth, and precise time." The observatory still has an active program observing double stars, but has moved most of its serious astronomical observations to a separate location in Flagstaff, Arizona. Since 1974, the place has become more famous for an Earth-bound star: the Vice President sleeps here in a 9,150-square-foot, three-story Victorian mansion.

The clock itself is really a number of clocks which varies, mostly depending upon how many are being calibrated or repaired at any given time. They sit in a few nondescript buildings on the Observatory's campus. Access to the clock areas is tightly controlled, and they have been built to exacting environmental tolerances: the temperatures are be regulated to +/- 0.1°C and the humidities controlled to within 3%.

The most precise clocks consist of cylindrical vacuum chambers, each containing rubidium atoms that are laser-cooled to about the coldest temperature anything can be (a millionth of a degree above absolute zero degrees Kelvin, to be exact). The result is said to be the world's most accurate operational, continuously-running timepiece.

The time-space continuum

The Master Clock is essential for the operation of all kinds of things, from the way advanced science is conducted to when you pay your rent. But don't let the "clock" part fool you. It isn't only about time. It's about space.

Determining your exact position in space, whether you're using a sextant or an iPhone, depends upon knowing the exact time at your location. When you look for yourself on your phone's map, your phone puts the question to a network of GPS satellites up above. The satellites, 32 and counting, were put there by the U.S. military starting in the 1980s; each is outfitted with its own space-hardened atomic clock, which is calibrated every day against the Master Clock.


A 24-satellite GPS constellation. Here, the number of satellites in view from a given point on the Earth's surface changes with time. Gif: Wikipedia


By comparing two times—the time a signal was transmitted by one of those satellites in space with the time it was received by your GPS receiver on Earth—your phone can calculate how far away each satellite is. By doing this with multiple satellites, your GPS is able to triangulate its location on Earth and to compute the time. The more satellites in view—the minimum is four at any given time—the more your GPS receiver can average out errors like random measurement noise and atmospheric delays.

Getting precise location depends upon getting precise time. If one GPS satellite is off by a billionth of a second, your GPS receiver will be a foot off. If the satellite's clock were off by one full second, your location on Google Maps would appear to be about two-thirds of the way to the Moon. If you've ever been told by your phone that you're in the middle of a river when you're standing at a subway station, you may have been the victim of these errors.

There are many other reasons for accurate time, but we civilians can't know them all. The Master Clock also keeps time for a host of other military operations: flying drones, aiming missiles, establishing secure communications, and other secret things. That is to say, not all time is created equal: the more critical you are to America's national security (say, you're a Navy SEAL, or a drone missile), the more accurately you might need to know the time or your location.


"From the onset of locating a threat, to placing a weapon on target, and subsequently evaluating the success of this engagement, all are impacted by the precision of time," John G. Grimes, the Navy's Chief Information Officer, said in 2008. And this is precisely why Europe, Russia, and China have for years been assembling GPS-like systems of their own—Galileo, GLONASS and Beidou, respectively—which are said to be cheaper than GPS and don't rely on increasingly suspect American technology.

For now, however, GPS, with its Master Clock-defined time, is central to time telling even in Moscow and Beijing. Many people worldwide get their time from GPS satellite signals. If, somehow, GPS time dissemination stopped working, "a lot of things would break down," from cell phones to financial markets, Matsakis said.

There have been some [attacks]… we think about it all the time.

Conversely, in the event that the Master Clock (and an Alternate Master Clock, which is located in Boulder, Colorado), were destroyed in, say, a terrorist attack, GPS could serve as a backup system. "Especially in the context of nuclear war: if a whole bunch of cities get destroyed, at least GPS would work for awhile."

Still, without the Master Clock to calibrate them, the clocks on the GPS satellites would gradually drift, per the law of special relativity. "If the Master Clock broke for 24 hours, the world would probably be okay. But I wouldn't do that for a week."


The Master Clock has failed on Matsakis's watch at three unfortunate times: twice while he was aboard an airplane, and once while at his son's wedding. In any event, the network time signal was maintained by backup clocks, and while some users experienced "outages" they were able to cope until the Master Clock was restored, about six hours later.

As a piece of military infrastructure, the Naval Observatory's timepieces are secured accordingly, he added. Security isn't just physical: there is a "very real" concern that the clock could be the target of a cyberattack.

In 1997, "a time back in the 20th century," Matsakis recalled, he set up a public webpage. A bug found its way into the page's root directory. "Someone took over my computer. We think they were looking for nuclear secrets."

More of a concern, he said, are denial of service attacks (DDoS) that could rough up the country's method for distributing time to Wall Street, not to mention any system that relies upon the internet for its time. "They're a fact of life for everybody," Matsakis said of DDoS attacks. "There have been some. But we have ways to deal with it."

One recent attack began at around 3:30pm EST on November 29, 2011. The NTP servers that distribute time to the Internet began receiving a rapid influx of requests, many tens of thousands of packets per second.

"Someone is 'at war' with USNO NTP service," Rich Schmidt of the USNO's Time Services Dept, wrote to the NTP mailing list. "They could be students, who knows? But all of the offending addresses traced to Chinese sites."


Generally, Dept. of Defense servers are kept protected inside a secure network known as NIPRNET. To provide time to the world, the Naval Observatory's NTP servers live outside that boundary. At approximately 11pm EST, as the attacks continued, the Navy Cyber Defense Operations Command ordered USNO to take its NTP servers offline. For 3 million clients, time, as the Master Clock knows it, stopped. "This is the first time in 17 years that we have ceased NTP operations," Schmidt wrote that night.

The Observatory's network admins restored the server the following day, but not before actively denying requests to specific IPs within China, a process that "requires considerable horsepower." The decision, Schmidt wrote, wasn't easy. "When it comes to making a choice between staying online and denying USNO NTP to China, we must unfortunately make the more secure choice."

"The DOD is very aware of the dangers," Matsakis told me. The Master Clock itself "has never really been attacked," he said, "but we think about it all the time."

A Brief History of (Disseminating) Time

Security fears help illustrate why the ability to keep time is only as good as its ability to disseminate it to the world; like time and space, time keeping and time dissemination are inextricably connected. For decades, the Observatory's time signal was a "time ball" that sat atop its telescope's dome—cutting edge technology for 1845. By dropping the ball every day precisely at noon, the inhabitants of Washington could set their timepieces, while ships in the Potomac River could set their clocks before putting to sea. A re-creation of that ball still sits atop the Observatory's main building, next to one of its telescopes.

In 1865, the Observatory began broadcasting a time signal via telegraph lines to the Navy Department, and eventually, via Western Union lines, to railroads across the nation. It was the railroad lobby who by 1885 would succeed in standardizing US time into time zones; up until then, the time depended on the city you were in: New York was five minutes ahead of Philadelphia, which was seven minutes ahead of Washington.


By law, today the USNO shares the responsibilities for measuring and disseminating time with the Time and Frequency department of the National Institute of Standards and Technology (NIST), which sits under the US Dept. of Commerce. The USNO sets time for GPS and navigational systems and the Dept. of Defense, while NIST sets the standard for the financial sector and other civilian applications. (NIST receives several billion computer requests per day for this service, and broadcasts time to over 50 million radio clocks, wristwatches, and other clocks with radio receivers. You can also call up NIST for Coordinated Universal Time at 303-499-7111; you'll also get a space weather report.)

While there is a lively cooperation between the two agencies charged with telling the time—and the occasional competition over talented PhDs—they mostly operate in different domains: NIST performs most of the cutting-edge research, while USNO focuses on counting and disseminating the time to the military, as a matter of national security.

The Master Clock's time is also sent through the internet via the Network Time Protocol, or NTP, which is one of the internet's longest running applications; the USNO receives about half a billion individual time requests per day for this service. Both USNO and NIST also transfer time by satellite to military and commercial users through a process called two way satellite time transfer, which relies on geostationary satellites in order to precisely cancel out communication delays. The USNO also maintains an experimental time service based on the world's digital television signals, and a telephone line that supports ancient 1200-baud dial-up modems.


And then there's the Master Clock's soothing, time-telling hotline. It's often used by certain broadcasters to coordinate network feeds, and it still gets about four million calls from people like my dad (and me) every year.

Between the Master Clock's contribution to Universal time—it makes up about 1/3 of the average—and the dominance of GPS, the U.S. has become the world's dominant timekeeper

Time wars: Greenwich drift

Like the mechanism of our modern clocks, which echo that of the pendulum clocks first developed in the 10th century, the definition of a second is also a sort of hand-me-down.

It was around 1000 AD when the Persian scholar al-Biruni began using seconds—a word derived from the second division by 60 of an hour—to measure the times of new moons. Based on the solar definition of a second—1/86,400 of each day—the Paris-based group called the International Bureau of Weights and Measures (BIPM), which is itself part of the U.N, came up with a standard based on the more stable measurements of atomic physics: when the electrons of an undisturbed cesium atom transition from one energy level to the next, they are known to emit radiation that oscillates 9,192,631,770 times per second. The time that takes, the BIPM determined, was closest to the previous definition of a second: "the fraction 1/31,556,925.9747 of the tropical year for 1900 January 0 at 12 hours ephemeris time."


Atomic clocks quickly went into use after their creation in 1955. The cesium second was defined in 1967 by the International Bureau of Weights and Measures, or BIPM; now there are over two hundred cesium atomic clocks spread across some fifty national laboratories worldwide. With specialized timescale algorithms—computer software designed to take advantage of the strong points of each type of clock, and take into account the noise that comes from comparing them—the world's timekeepers send their time to the BIPM, which averages them and publishes them on a monthly basis in its journal, Circular T. This number is what's known as the Coordinated Universal Time, or UTC.

Before then, the United Kingdom's Greenwich Mean Time (GMT) was the world's most widely-used standard, originally meant for coordinating time across the ships and the railroads of the world's largest empire. But as soon as atomic time went into use, GMT's days were numbered. Scientists already knew that atomic time was far more stable than time determined by the rate of the Earth's rotation.

On January 1, 1972, several international groups agreed to replace the solar-based GMT with the atomically-determined UTC. (Especially throughout the Commonwealth, GMT is still widely and erroneously used to refer to UTC.) Between the Master Clock's contribution to UTC—it makes up about a third of the average—and the dominance of GPS, the U.S. has become the world's dominant timekeeper.


Perhaps the only real problem with having a single "universal" time is getting everyone, everywhere, to agree to using it. But fortunately, even as space and cyberspace become ever contested terrain, time has remained a peaceful matter, relatively speaking.

Barring a war between any of the world's time powers, Mataskis envisions a future with more equitable ways of keeping universal time. The decision by other nations to field GPS-equivalents suggests that de-facto time may eventually become a fairer average of various countries' master clocks. Already the European GPS-equivalent, Galileo, is being designed to be an average of the times of several European labs.

"I can imagine a time—I would guess in 50 years—when the world is so interconnected that the clocks are merged instantly in a real time situation," he told me. "That way, there's no one country in the lead anymore. I could see a giant melting pot of time."

First, however, the world must agree for just one leap second. The problem of the leap second has mainly to do with the moon. Modern astronomy has shown that the moon's gravitational effect is dragging on the Earth, causing the rate of the Earth's rotation around the Sun to very slowly decrease. That increases the length of our solar days. Five hundred million years ago, before the dinosaurs, days were a mere 21 modern hours long. Around 1820, a day averaged 24 hours on the dot, give or take a few milliseconds; today, one day averages about 1 millisecond longer than 24 hours.


As soon as atomic time and universal time were synchronized at the start of 1958, they began to move apart. The markings of the sun, our oldest "clock," were drifting, very slowly, from the time set by atoms. When the world's timekeepers switched from Greenwich Mean Time to Coordinated Universal Time, after great debate they agreed on another new rule meant to solve the problem: the addition of "leap seconds," which would be added to Universal Time at irregular intervals as needed to bring atomic time in line with solar time. The last leap second was on June 30, 2012 (when the last minute of the day actually had 61 full seconds); 25 leap seconds have been added since 1972.

This, however, poses a problem for eternity. The Earth's slowing rotation means we'll need to add increasingly more leap seconds to the clock. Some time in the next century, a leap second will be required every year. By the time the year 2600 rolls around, three or four leap seconds will be required each year; by 4300, the Earth's time keepers could be adding leap seconds every month.

While leap seconds are an annoyance for timekeepers like Matsakis, they can cause havoc for ever-more crucial electronic navigation and computer systems. On the 2012 leap second, servers running Linux, including those of Reddit, Mozilla, Qantas Airlines, Gawker and others, went haywire. "It's really annoying," Linus Torvalds, Linux's creator, told Wired, "because it's a classic case of code that is basically never run, and thus not tested by users under their normal conditions."


The International Telecommunication Union, a UN agency that traces its lineage to 1865, is now considering a proposal that would do away with leap seconds once and for all. The proposal would simply keep UTC as it is, 35 seconds behind atomic time, with no more leap seconds added. By the end of the 21st century, this would mean the time on our clocks and the solar time would be off by a little under a minute.

Opponents have called the abolishment of the leap second a costly hassle for civil timekeeping, requiring a massive inspection of all the world's computer code for astronomy, for instance, and insisted that it would mean a major disruption to Earthlings' perception of time. Leap second abolitionists insist that humans would be able to cope just fine, because they would naturally adjust for this drift accordingly.

After consulting with agencies like the Naval Observatory, NIST, NASA, the general public, and companies like Garmin, which build GPS receivers, the U.S. State Department recently announced its support for the ITU's proposals, joining other anti-leap seconders like France, Italy, Japan, and Mexico.

They're up against Russia and the UK, who have rallied behind the leap second. Others, such as Germany, remain undecided, while China has switched from supporting leap seconds to supporting their abolishment. A decision on the question has been postponed twice already and is expected to be addressed—finally, maybe—at the next World Radiocommunication Conference in Geneva in 2015.


Russia has said it worries the loss of new leap seconds could hurt its space assets and its GPS-like GLONASS system. The United Kingdom's opposition is rooted in concerns that "without leap seconds we will eventually lose the link between time and people's everyday experience of day and night," as Minister for Science David Willetts has said. Because a new time standard would slowly diverge from Greenwich Mean Time, some have suggested that the introduction of a new time standard would also be a blow to Britain's national pride.

The issue belies the notion of time keepers as an orderly, diplomatic community, Matsakis said. "With some people, I don't want to talk to them about the leap second, because then 'the speech' comes out and I have to spend ten minutes listening to it."

Matsakis admits that the cost of conversion away from the leap second "is a serious one, and that's a very strong argument against it." Still, he thinks those concerns are likely exaggerated, and thinks that the cost would be akin to those spent dealing with every new leap second addition.

A room at the USNO's time vault, home to the Master Clock. Image: Richard M. Hambly

The worries about conversion reminded me of the hysteria over Y2K, when the world worried the turn of the Millennium would ruin computer systems that never knew years that were more than two digits long.

On midnight, December 31, 1999, Matsakis was at his desk, he said, but he wasn't overly concerned. In advance of the changeover, the Naval Observatory purchased an alternate time-averaging computer for the occasion, months before the New Year, and tested it by pushing its clocks forward. Everything worked fine.


The problem was another computer: the one that ran the website. "We isolated it from everything so we just forgot about it. There was a line of code in a language called Perl, so the display went wrong. It said 'January 1, 1900,' or something like that. Nobody knew Perl, so I called my son at MIT. 'How do I compile Perl?' 'You change the code right in there,' he told me. So I went in there and I changed the code and fixed it."

The time on the USNO's homepage was a century off for just 45 minutes, which was apparently enough time for a reporter at the Washington Post to notice and write a story for the next day's paper.

Matsakis points to another upcoming Y2K-like issue: systems that use UNIX might fail come 3:14:07 UTC on January 19, 2038, as the operating system isn't designed for dates beyond that. Matsakis isn't overly worried; by then, he expects to be long since retired.

"But maybe my life support will need it."

For now, only the oscillations of a cesium atom's radiation are part of the official definition of time, but more precise clocks are already in use, hinting at possible future definitions. To better calibrate the Master Clock—and to keep up with new standards for accuracy, like that of the new GPS III protocol—Matsakis and his colleagues now also use a clock called a hydrogen maser, a 500-pound machine that looks like a stout utility robot from Star Wars and employs hydrogen atoms, excited by microwaves, to tell the time. (Maser stands for "microwave amplification by stimulated emission of radiation," as opposed to a laser, which uses visible light.) Both were invented by Nobel Laureate Charles Townes, who, coincidentally, was Matsakis's thesis advisor.


In addition to about 80 cesium clocks, the USNO has about 40 masers and four clocks that use rubidium. The Observatory's newest cesium clock dates from the year 2000; about half of its masers are from this century. "The clocks themselves can live forever, but they need repairs," said Matsakis.

By averaging out the readings on all of their clocks, the time keepers at the Naval Observatory can produce a time whose rate, says its website "does not change by more than about 100 picoseconds (0.000 000 000 1 seconds) per day from day to day." (One nanosecond is one billionth of a second, while a picosecond is about one trillionth.) Only the most stable clocks are included in the average, based upon current performance. If you're on the internet, you can find out just how many of the Observatory's clocks are being weighted at this very minute.

Cesiums. Strontiums. Rubidiums. Ytterbiums.

People are constantly working on making better clocks, and many promising designs have worked well in test mode. In 2011, a cesium clock in Britain was presented as the world's most accurate long-term time keeper, after scientists determined that the clock would only lose or gain less than a second in 138 million years. And yet, these clocks are already outdated. While clocks that use cesium atoms are known to be reliable over long periods of time, they're not as accurate in the short term as timepieces that make use of other elements like hydrogen, mercury, and calcium, and a relatively new mechanism called, poetically, a "fountain."

In this apparatus, a laser traps and cools atoms until they are so cold that their random motions essentially cease. Then they're launched straight up into a microwave cavity, where their natural resonance frequency can be measured. In April, the National Institute of Standards and Technology (NIST) in Boulder debuted a new cesium fountain clock, the NIST-F2, which is designed to not gain or lose 1 second in 300 million years. (See video showing how it works below.)


One reason to build better clocks is simply to maintain a standard the rest of the world's clocks can follow.

"The clocks they sell are as good as the clocks we could possibly build in the laboratory 20 years ago, and they're being used all over the place," said Steven Jefferts, a NIST physicist and lead designer of NIST-F2. "And so we have to stay out in front of that, if for no other reason than we have to be able to calibrate them."

There are other, as-yet-unknown reasons too. Precise clocks like these may not change how we tell the time tomorrow, but like a lot of exotic scientific instruments, they promise innovation. Jefferts predicted that the new clock "is going to be the progenitor for something that really is important ten years from now."

For example, by measuring the environmental factors that cause slight variations in the ticking of an atomic clock, scientists can use these timepieces to map magnetic and gravitational fields. Someday, clocks could be used for a variety of precision-based jobs, including mining, volcano and earthquake prediction, heart- and brain-imaging, and even highly-detailed passive surveillance systems.

In May of last year, DARPA-funded researchers at NIST unveiled an "optical lattice clock" that uses lasers to excite atoms of the rare earth element ytterbium, whose oscillations can help to divide periods of time into finer and finer intervals. The clock is said to be ten times better than its predecessors; if it were to have started ticking at the Big Bang—about 13.8 billion years ago—by now, it would be only off by one second. The team behind it is dreaming even farther, according to the lead researcher at NIST, Dr. Jun Ye: "Our aim is that we'll have a clock that, during the entire age of the universe, would not have lost a second."


"We can see requirements ten years down the road that are going to need better clocks," Matsakis said, specifically "space-related" technologies and more accurate GPS. "That's the reason the Navy funds us to build better clocks."

For now at least, the Master Clock is the most precise continuously-operating system ever constructed to measure anything.

In February, the Naval Observatory in Colorado upgraded the Master Clock with several brand new fountain clocks that use the element rubidium. (Counting the oscillations of rubidium could boost the timing reference for GPS by 10-fold, from 1 to 2 nanoseconds down to 300 picoseconds.) The clock is so delicate that installing it required an "airsled hover lifter"—essentially, a hovercraft—in order to ensure that it didn't come into contact with the floor or the walls.

The Master Clock now relies on up to four rubidium fountains, which, Matsakis said makes it, for now at least, "the most precise continuously-operating system ever constructed to measure anything."

Increasing precision is increasingly costly. A good cesium clock can now be purchased for as little as $75,000, a maser can cost around $250,000, and a rubidium fountain simply cannot be bought. Its parts can cost as much as $600,000, but that doesn't include the salaries of the half-dozen PhDs who've worked for a decade developing these fountains. In total, the Observatory runs on less than $20 million a year, a modest budget for the Department of Defense. (The Observatory is also allocated funds by NASA, the Air Force, and other agencies for special projects that are performed at cost.)


There is a display of historical clocks at the Observatory, and when he walks past it, Matsakis is able to put the costs of his newest clocks in perspective. "We have a Cummins clock from the mid-1800s in there. It says that in today's dollars, it cost just as much as a maser does today. We have over a hundred clocks now. In the Song dynasty the Emperor used a water clock. It had a staff about as large as my department just to maintain it."

Researchers are also developing atomic clocks that require even fewer people and take up much less space. The most accurate cesium clocks are the size of a compact car and draw about a kilowatt of power; now there are efforts to get atomic clocks down to the size of a computer chip, enabling GPS devices far more accurate than what we use today.

Wait a second: what is time?

At some point, the big question about time came up. What is it?

"Once I had this definition of time, that it's a coordinate that you can measure the evolution of in a closed system," Matsakis said. "Now, I think of time as something that, stripped down to its essence, is a measure of interactions," an idea based on Einstein's theory of relativity, which pins time and space to the relative motion of objects. "It's an intriguing thought: if you don't have interactions, time is irrelevant."

He offers an example that begins at the end of time. "One way that time could stop is if the universe could reach a cold death. If our universe expands forever, and the suns die out, and they become black holes which evaporate over eons, what's left is a rarefied gas, a cold gas that's uniform across the universe. With everything the same, how can you have time? There'd be nothing to measure time. Time would stop, and not with a bang. It would just peter out."

The relative interactions that govern the movement of time explain why events in the universe don't easily fit along a timeline. "Imagine that people witness Al Capone robbing a bank in 1930," he said. "Then, a supernova a thousand light-years away is observed somewhere on Earth in 1987. Did the star explode first, or was the bank robbery first? It depends on the observer."

In an email later, I mentioned Nietsche's proposal of eternal return. Matsakis shot back: "It is very hard to define fundamental things. I haven't tried to define 'place.' Socrates spent years trying to define justice. Maybe they were right about him corrupting the youth."

It was Aristotle's contemporaries who first mastered the calculation of the passage of time, or chronos, and it was they who also recognized another kind of time, kairos—the moments that define our pleasures and our pains and our deepest feelings and thoughts.In other words, the kind of time that can't be metered by any clock.

Tensions linger between this sense of "time," which Henri Bergson would later describe as "duration," and the tick-tocks of "the time" overseen by the timekeepers. Now more than ever before, argues Douglas Rushkoff in Present Shock, distractions keep the latter version of time in a kind of tug of war with the former.

"We spent centuries thinking of hours and seconds as portions of the day," Rushkoff told David Pescovitz last year, "But a digital second is less a part of a greater minute, and more an absolute duration, hanging there like the number flap on an old digital clock." The rush of the present and its seemingly infinite, hyperlinked possibilities means "a diminishment of everything that isn't happening right now—and the onslaught of everything that supposedly is."

But according to time, not everything is happening at once, as Wheeler joked. Could the Master Clock, I wondered, with its steady, orderly pace, remind us that the world isn't moving any faster?

But not even this ticking will be the same in the future. With new clocks, the way that time is counted will change. And in time, the definitions of time will change too, if not the questions that endlessly circle it.

"What I tell people is, I can't tell you what time is," Matsakis said, "but I can tell you what a second is."

At least for now.

See also:

A Short History of Long-Term Thinking, for Our Fifty Thousand Year Time Capsule

If We Want Anyone to Remember Humanity, We Need to Talk About Time Capsules

Chasing the Elusive Arrow of Time with Computer Algorithms

"Where Time Comes From": The Atlantic Video