A Century of Time Measurement:
From Pendulum to Optical Clocks
Michael Lombardi
NIST Time and Frequency Division
NCSLI 2011
Introduction
For most of recorded history, the most accurate measurements of time
involved dividing the period of a day into smaller parts. For example, the
solar second was measured by dividing the solar day in 86,400 parts. The
rotational rate of the Earth was known to be the best clock of all, the
ultimate reference for timekeeping.
This began to change when mechanical clocks first appeared in the 14th
century. Mechanical clocks measured time by counting the oscillations of a
repetitive event, such as the swings of a pendulum or balance wheel. This
was a fundamental change in timekeeping, because instead of dividing
days to get seconds, they multiplied seconds to get days. Early
mechanical clocks were not very accurate, and for about 600 years,
mechanical clocks were calibrated with astronomical clocks. About a
century ago, technology finally improved to the point where man made
clocks were known to be more stable and accurate than the Earth’s
rotation.
This presentation begins at that point, and takes a very brief look at
developments in clock technology during the past 100 years.
Outline
• Pendulum Clocks
• Quartz Clocks
• Atomic Clocks and the Redefinition of the Second
• Optical Clocks
• Radio Controlled Clocks
Riefler Pendulum, 1904

Manufactured by the Clemens
Riefler Company of Munich,
Germany.

Served as the U. S. national
standard for time interval from
1904 to 1929.

Accurate to tens of
milliseconds per day (parts in
107).
Shortt Pendulum, 1921

A mainstay of astronomical
observatories in the 1920s and
1930s, the Shortt pendulum was
advertised as “The Perfect Clock”.

Designed by the British railroad
engineer William H. Shortt.

Used two pendulums. The master
pendulum was disturbed as little as
possible.

Accurate to 1 s per year (a few parts
in 108). It was so accurate that it
suggested for the first time that the
Earth was not a perfect timekeeper.

Briefly used as the U. S. national
standard for time interval, during
part of 1929.
Quartz Clocks

The Curie Brothers demonstrated
the piezoelectric effect in quartz
and other crystals in 1881. It
remained a scientific curiosity for
years.

The first application of
piezoelectricity was detecting
enemy submarines in World War
I, with independent work
conducted in France and the U. S.

Walter Cady, an American
physicist, worked on submarine
detection systems during the war.
After the war he focused on
building a standard for radio
frequency, and patented the first
quartz oscillator circuit in 1920.
General Radio Type 275, 1924

The first commercially
available quartz oscillator,
it sold for $145.

Cady and George Pierce
(who improved upon
Cady’s basic circuits) were
involved in the design.

Used by radio engineers to
calibrate transmitters, and
was soon followed by
more accurate quartz
standards.
U. S. National Frequency Standard, 1929

A group of four 100kHz quartz oscillators
manufactured by Bell
Telephone
Laboratories.

Accurate to about
1 × 10-7. Even though
this was no better
than the pendulum
clocks it replaced, it
could serve as a
standard for both
radio frequency and
time interval,
something that the
pendulum could not
do.
First Quartz Clock, 1927
• Designed by William Marrison and Joseph Horton of Bell Telephone Laboratories
• Used a 100 kHz oscillator and perhaps the first frequency divider, which reduced the frequency to 1 kHz
• The 1 kHz frequency controlled the speed of a synchronous motor that moved the hands
• Accurate to about 2 parts per million, slightly better than a typical quartz watch of today
Rohde & Schwarz Model CFQ, 1938
Hamilton Electric 500, 1957

The first battery powered
watch, it had the unique
styling of a 1950s
automobile.

Ran at the same frequency
as a mechanical watch, 5 Hz,
but it derived its frequency
from tiny electrical contacts
that opened and closed five
times per second.

Was not particularly reliable
or accurate, but was very
popular, showing the large
demand for a watch that
never needed to be “wound”.
Bulova Accutron Spaceview, 1962

Designed by the Swiss
engineer Max Hetzel,
the Bulova Accutron
was first introduced in
1960

Its tuning fork oscillator
ran at 360 Hz, as
opposed to 5 Hz for
mechanical watches

Accurate to 2 seconds
per day (2 × 10-5), a
factor of 10
improvement over the
best mechanical
watches

“Hummed” instead of
“Ticked”
First Quartz Watch Oscillator Circuit, 1966
• Designed by Armin Frei of the Centre Electronique Horloger (CEH) in Switzerland in response to the
threat posed to the Swiss watch industry by the Bulova Accutron.
• Ran at 8192 Hz. Thirteen binary flip-flops divided the frequency to 1 Hz.
• Was included in the first quartz watch prototype (July 1967), but was never used in a commercially
available watch.
Seiko Astron, 1969

Introduced on
Christmas day in 1969,
beating the Swiss
watchmakers to the
market.

Sold for $1250, about
the same price as an
economy car. It had
nearly 200 analog parts
that had to be hand
soldered (the Swiss had
designed ICs).

Ran at 8192 Hz with 13
binary divider stages.
By 1972, nearly all
quartz watches used 15
divider stages and ran
at 32768 Hz, which
became the standard.
Atomic Clocks

The Scottish physicist James Clerk Maxwell
suggested that atoms could be used to keep
time as early as 1879.

The first atomic clock experiments took place
some 60 years later at Columbia University in
New York, conducted by a team led by Isidor
Rabi.

Rabi publicly discussed his plans for an
atomic clock during a lecture at Columbia in
1945, and the New York Times (left) covered
the story.

The concept was actually simple. Because all
atoms of a specific element are identical,
they should produce the exact same
frequency when they absorb or release
energy. An atom, then, could potentially
serve as a “perfect” oscillator.
First Atomic Clock, 1949
• Based on the ammonia molecule, it was unveiled in January 1949, designed by a
team led by Harold Lyons at the National Bureau of Standards.
• It never worked well enough to be used as a standard or reference. Its best
reported uncertainty was about 2 x 10-8, less accurate than the quartz oscillators
then used as the national frequency standard. But it provided a glimpse of what
the future would bring ……
NBS-1, Cesium Prototype, 1952

The NBS team, led by Harold
Lyons and Jesse Sherwood, had a
large early lead in the race to
build the first cesium clock. They
began work in 1950 and reported
their first results in 1952.

NBS interrupted the program in
1953, for budgetary and other
reasons. By 1955, both Lyons and
Sherwood had left NBS.

The clock was taken apart and
moved to the new NBS labs in
Boulder, Colorado where it was
reassembled. It finally became
the national frequency standard in
1959, but by then NPL in England
had operated a cesium standard
for several years.
First Cesium Time Standard, NPL, 1955
The Atomic Second, 1967

In 1956, the second was defined as 1/31,556,925.9747 of the tropical year 1900.
The ephemeris second was nearly impossible to use as a time reference and of
little use to metrologists or engineers.

Ephemeris time was determined by measuring the position of the Moon with
respect to several surrounding stars. The best Moon observations had been
recorded at the United States Naval Observatory (USNO) in Washington, DC by
the astronomer William Markowitz.

Louis Essen and Jack Parry of NPL compared their new cesium clock to a quartz
clock steered to ephemeris time at the USNO. Because the two clocks were
located across the Atlantic from each other, they simultaneously compared each
clock to radio signals that could be received at both laboratories, a measurement
technique now known as common-view time transfer.

Four different solutions were made to determine the effects of using different
data. The final result was the average of the four solutions, and was published
as 9 192 631 770 cycles/s in August 1958. In 1967, the second was finally
redefined as:
“the duration of 9 192 631 770 periods of the radiation corresponding to the
transition between the two hyperfine levels of the ground state of the caesium 133
atom."
NBS/NIST constructed seven Cesium Beam Primary Frequency
Standards from 1959 to 1998.
NBS-6
NIST-7
NBS-5
NBS-4
NBS-1
NBS-2
NBS-3
NIST-F1 Atomic Fountain Clock
A cesium fountain frequency
standard that provides the best
possible realization of the SI
second.
Current accuracy (uncertainty):
NIST-F1
laser-cooled
fountain
standard
“atomic
clock”
• 3 x 10-16
• 26 trillionths of a second per day.
• 1 second in 105 million years.
Equivalent to measuring distance from earth to sun (1.5 x 1011 m or 93
million miles) to uncertainty of about 45 mm (less than thickness of
human hair).
Improvements in Primary Frequency Standards at NBS/NIST
-9
-9
10
-10
10
NBS-1
NBS-2
-11
Frequency Uncertainty
10
-12
10-13
10
-11
10
NBS-4
NBS-3
10
-10
More than 50
Years of Progress
in Atomic Clocks
10
10-12
10-13
NBS-5
NBS-6
10-14
-14
NIST-7
10
-15
-15
10
10
NIST-F1
Initial
-16
10
NIST-F1
Today
-16
10
10
-17
10-17
10-18
10-18
1940
1950
1960
1970
1980
Year
1990
2000
2010
2020
Optical Clocks: The Next Generation of Primary Standards
Optical clocks “tick faster” than microwave clocks. For example, the mercury ion
clock resonates at a frequency more than 100,000 times higher than a cesium
clock. This is comparable to using a second, rather than a day, as the base unit of
time.
In principle, faster “ticks” means better accuracy and stability.
15
f 0 optical
10
5
 10  10
f 0 microwave 10
Al+
Hg+
Yb
Ca
Cs
1124 THz (1124 x 1012 Hz)
1064 THz
520 THz
456 THz
Optical
0.0092 THz
Microwave
Improvements in Primary Frequency Standards: Optical Clocks
• Optical clocks have the
potential for accuracy at
the 10-18 level, >100 times
better than NIST-F1.
• Likely to take many years
to realize that potential.
Laser-cooled
calcium atoms.
Ytterbium atoms
in optical lattice.
Single
mercury ion.
Single mercury ion trap.
Improvements in Primary Frequency Standards: Optical Clocks
-9
-9
10
10
NBS-1
-10
10
-10
10
-11
-11
10
10
-12
10-12
10-13
10-13
10-14
10
Frequency Uncertainty
10
-14
NIST-F1
-15
10
-16
-16
10
10
NIST-F2
-17
10
Optical
Standards
-18
10
1940
-15
10
1950
1960
1970
1980
Year
1990
2000
2010
10-17
10-18
2020
Junghans MEGA 1, 1990

The first radio controlled
wristwatch, manufactured in
Germany. The antenna was
hidden inside the wrist
band.

Synchronized to time signals
broadcast by radio station
DCF77 on 77.5 kHz. This
station was synchronized to
the German time standard
maintained by PTB.
Low Frequency (LF) Radio Controlled Clocks

Low frequency time
signal stations operate
at frequencies ranging
from about 40 to 80
kHz.

The pictured watch can
synchronize to
transmitters in the
United States, England,
Germany, Japan, or
China.

The U. S. transmitter is
radio station WWVB,
operated by NIST. More
than 50 million WWVB
radio controlled clocks
are believed to be in
operation.
GOES Satellite Clock, 1976

The first clocks controlled
by satellites received NBS
time signals from the GOES
geostationary satellites.
These clocks appeared
about three years before
the launch of the first GPS
satellite.

Designed by a team led by
Dick Davis, these clocks
could remove most of the
path delay between the
clock and the satellite and
were accurate to less than
100 microseconds. The
picture shows a GOES
clock built to
commemorate the U. S.
bicentennial.
GPS Clocks

Best known as a positioning
and navigation system, GPS is
also the main system used to
distribute accurate time and
frequency worldwide.

A constellation of as many as
32 satellites can deliver time
accurate to less than 1
microsecond (typically 100 ns)
anywhere on Earth. This has
revolutionized timekeeping,
and made many new
technologies possible.

Many metrology labs use GPS
disciplined oscillators as their
standard for frequency.
Mobile Phones are Radio Controlled Clocks

The clocks on mobile
phones are usually very
accurate. Many phones
synchronize to GPS clocks
located at cellular base
stations, with only a few
microseconds of additional
delay added.

Unlike LF radio controlled
clocks, mobile phones
automatically correct when
you change time zones.

A recent study indicates
that 1 out of 7 people have
stopped wearing watches,
mostly because of mobile
phones. That figure is
twice as high among 15 to
24 year olds.
Summary
• The uncertainty of time measurements has improved by about 10
orders of magnitude during the past century, from parts in 106 to parts
in 1016. Optical clocks should further reduce uncertainties by at least
two more orders of magnitude.
• In everyday life, time-of-day clocks that are accurate to within 1
second will become more common, and should eventually be the
norm rather than the exception. Many technologies could contribute
to this trend, including LF radio signals, satellite signals, mobile
phone signals, Internet time codes, and miniature atomic clocks.
•
If you are interested in reading more about the history of time measurements,
see the five-part series now being published in IEEE Instrumentation and
Measurement Magazine (the first installment is in the August 2011 issue).