History of Atomic Frequency Standards: A Trip Through 20th Century Physics
Proceedings of the 1996 IEEE International Frequency Control Symposium (pp. 1225-1241) © 1996 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Arthur O. McCoubrey
Arroyo Grande, California
The opening year of the 20th century brought with it Planck’s bold announcement of the quantum nature of radiation energy interacting with matter; an avalanche of unprecedented insights into the physics of atoms and molecules followed during the years leading up to World War II and continued during the period after the war. These remarkable scientific advances, in combination with extraordinary advances in technologies accelerated by war and defense related programs provided a broad foundation for the development of atomic and molecular frequency standards with a number of approaches reflecting different scientific research methodologies. This paper will relate the history of atomic and molecular frequency standards development in a framework that includes linkages with scientific advances that underlie the progress.
As the century draws to a close, the rate of investment in the further refinement of atomic and molecular frequency standards and the related practical research is much reduced; however, ongoing scientific advances continue to provide a very high level of potential for improvements well into the next century. But perhaps it is even more important to note that rapidly expanding applications in areas that affect larger and more widely distributed user populations will provide the most important driving forces affecting future demands for this complex technology.
It is an overstatement to suggest that the history of atomic frequency standards is linked with 20th century physics in any complete sense. But, there is a very remarkable connection between the evolution of atomic frequency standards and the foundation of what has come to be known as modern physics. This foundation was laid down during the first decades of the present century and stunning successes in explaining detailed behavior of atoms and molecules soon led directly to concepts for practical frequency standards. The advent and the venue of atomic frequency standards are both very closely coupled with scientific research that provided the basis for their development Thus, most of the key ideas, technical principles, methodologies and many important details of practical frequency standard design originated in a number of academic laboratories. Professors engaged in scientific research also contributed practical insights and many of their students became leaders in the demonstration of practical atomic frequency standards and their refinement in the development laboratories of government and industry.
Even before the beginning of the 20th century James Maxwell and Lord Kelvin recognized  that measurement units for length and time based on atomic reference standards should provide important advantages because of their reproducible properties and availability throughout the World. Their suggestions were based on advancing knowledge of optical spectroscopy and, in the case of the unit of length, the practical use of wavelength associated with optical radiation emitted by atoms was clearly realistic. However, there was no possibility for linking the extremely high frequencies of optical radiations with radio frequencies or any of the clock related technologies; therefore, at that time, concepts for atomic frequency standards were, at best, fanciful or idealistic.
The opening year of the 20th century brought with it Planck’s bold announcement of the quantum nature of radiation energy interacting with matter; an explosion of new insights into the physics of atoms and molecules followed and during the early years of the century modern physics expanded to include ideas that were to be critical to the advent of atomic clocks:
- electromagnetic radiation consists of discrete quanta (Plank 1900);
- atoms are structures consisting of nuclei and electrons in orbits, but only certain orbits are allowed as “stable” orbits associated with discrete energy states (Bohr’s theory of the hydrogen atom , 1913);
– changes in the energy states of orbital electrons are accompanied by emission or absorption of electromagnetic quanta;
– Bohr model accounted for main features of the optical spectrum of hydrogen;
- no two electrons in an atom can exist in the same state (Pauli’s exclusion principle, 1925);
- the time interval during which a state of energy can exist and the energy of that state cannot be simultaneously determined with arbitrary precision (Heisenberg’s uncertainty principle, 1927);
- relativistic effects (Sommerfeld, 1916) in combination with the effects of electron spin and the associated magnetism (Goudschmidt and Uhlenbeck, 1925) explained fine structure details of atomic spectra ;
- presence of hyperfine structure in atomic spectra identified with interactions of nuclear magnetism and electronic magnetism. Additional forces associated with nuclear structure accounted for hyperfine structure in the radiation from molecules.
Hyperfine structure in the spectra of atomic and molecular radiations were particularly relevant to possibilities for atomic frequency standards because the very small energy differences corresponded to frequencies in the radio spectrum.
Until the end of the first third of the present century the scientific observations that inspired progress in modern physics were based on optical spectroscopy and the extremely high frequencies of the electromagnetic radiation provided no possibilities for the creation of frequency standards in the practical radio spectrum. However, in 1934 this barrier was overcome when C.E. Cleaton and N.H. Williams at the University of Michigan made a direct observation, for the first time, of a radio frequency change of state in the ammonia molecule . This achievement opened the whole new field of microwave spectroscopy and the possibility of a frequency standard based on internal properties of atoms or molecules took a giant step forward.
In 1934 the sources and detectors of microwave radiation that were available were extremely limited and difficult to work with. For this reason, additional progress in microwave spectroscopy was delayed until the end of World War II in 1945. By that time, advances in radar technology had provided excellent sources and detectors and a period of rapid progress followed in the field of microwave spectroscopy , including extensive studies of the microwave spectrum of ammonia. In the course of this work a number of investigators reported the experimental stabilization of the frequency of klystron oscillators using the ammonia absorption line at 24 GHz (1.25 cm; K-band).
The rapid advances in microwave spectroscopy immediately following World War II removed remaining barriers standing in the way of using atomic or molecular phenomena for the standardization of frequency. Accordingly, early in 1948 Harold Lyons at the National Bureau of Standards initiated an experimental study of a frequency standard based on the 24 GHz absorption frequency of ammonia and in August of the same year he and his colleagues operated the World’s first atomic clock with a quartz crystal oscillator in a feedback control loop with a frequency multiplier and synthesizer chain to drive an ammonia absorption cell. This ammonia atomic clock, illustrated in Figure 1, was announced to the news media in January 1949, attracting considerable attention. In its original form, the ammonia clock had a stability of about one part in 107 and an improved model was stated to be about five times better. The performance of the ammonia clock was limited by Doppler broadening and pressure broadening of the ammonia absorption line.
Additional improvements in the performance of the ammonia clock were certainly possible and, indeed, it was also evident that the use of different absorbing gases would be have advantages. However, by the time that the ammonia clock had been placed in operation, it was already clear that the greatest potential for improved performance would be realized through he use of radio frequency resonances in beams of atomic particles.
Before leaving the discussion of ammonia frequency standards it is, perhaps, important to mention briefly the ammonia maser invented at Columbia University by J.P. Gordon, H.J. Zeiger and Charles H. Townes in 1954 . The ammonia maser is relevant because it functioned as an active source based on stimulated emission of 24 GHz radiation from a population of excited ammonia molecules. While the possibility of using the ammonia maser as a frequency standard was studied, it became clear that beams of atomic particles offered better possibilities for higher levels of stability.
The most important role of the ammonia maser is almost certainly linked with the striking demonstration of stimulated emission of radiation as predicted by Einstein (1916). It also set the stage for the invention of its optical counterpart, the laser, in 1960.
In the early 1920’s Stern and Gerlach used a beam of atoms passing through a strongly diverging magnetic field to study magnetic properties of atoms and associated quantum effects. The result of their famous experiment provided critical support for the idea that electrons are spinning particles with associated magnetic moments. In addition, their experiment demonstrated the means for separating atoms in different magnetic energy states that could be selected for special studies or other purposes.
In the 1930’s, Professor I.I. Rabi and his students at Columbia University refined the molecular beam method and in 1938 they applied radio frequency fields to the particles in beams and the molecular beam magnetic resonance method was invented  as a powerful tool to study changes in the energy states of atoms and molecules and the associated frequencies of electromagnetic radiations that were either emitted or absorbed. That is to say, the molecular beam magnetic resonance method provided a new form of a high precision spectrometer that permitted the direct observation of energy transitions involving radio frequencies.
Rabi clearly recognized the possibility of using the molecular beam magnetic resonance apparatus as a frequency standard and as early as 1940 he identified a hyperfine structure microwave transition of Cs133 at 9.l9 GHz as the resonance of choice  for frequency reference purposes. There is clear evidence that the potential for unprecedented frequency stability was fully appreciated and the possibility of observing the relativistic gravitational red shift was even discussed.
However, in 1940 microwave technologies were still not widely available and the priorities of World War II had begun to affect laboratory research programs. Thus, the consideration of atomic frequency standards had to be delayed; however, additional advances in the molecular beam resonance method continued to set the stage.
The resolving power or sharpness of response of a molecular beam magnetic resonance apparatus depends upon the length of time that atomic particles in the beam are actually under the influence of the electromagnetic field. This time interval is directly proportional to the physical length of the space in which the atomic particles are exposed to the electromagnetic field and inversely proportional to their velocities in the beam. In the early days of molecular beam magnetic resonance spectrometers, there were very limited possibilities for controlling velocities of particles in beams because they were obtained by evaporation form thermal sources. Therefore, it was generally desirable to make the interaction space as long as possible taking into consideration all the factors involved. These factors included the need for a very uniform magnetic field in the interaction space and the need to control the electrical phase of the applied electromagnetic radiation. Professor Norman Ramsey addressed these factors in 1949 and invented the separated oscillatory field method of molecular beam magnetic resonance spectroscopy .
In 1989 Norman Ramsey was awarded the Nobel Prize in recognition of his work on the separated oscillatory field method and his work on the atomic hydrogen maser.
Even during the construction and operation of the ammonia absorption cell frequency standard at the National Bureau of Standards it was recognized that atomic beam methods would provide much greater potential for stability. Accordingly, in September 1948 Dr. Lyons engaged Professor Polykarp Kusch  of Professor Rabi’s laboratory at Columbia University as a consultant to assist in the design of an atomic beam frequency standard. Professor Kusch described the Some design considerations of an atomic clock using atomic beam techniques at a meeting of the American Physical Society in Washington on April 30,1949.
The construction of the first cesium atomic beam frequency standard at NBS began in the summer of 1949 and the apparatus, illustrated in Figure 2, was first operated early in 1951. The operating results from this atomic frequency standard were the basis for reporting, early in 1952, the first direct measurements of cesium hyperfine frequencies.
In 1954 the Central Radio Propagation Laboratory of NBS was moved to new quarters in Boulder, Colorado. The cesium beam frequency standard was reassembled there and, later, it became known as NBS-I Atomic Frequency Standard, the first in a series that has now advanced to number 7. NBS-I, with improvements from time to time, was operated as a part of the NBS atomic frequency standard program until 1966; during part of this time, it was designated as the United States Frequency Standard (USFS) until it was replaced by a more advanced cesium atomic beam frequency standard, NBS-II .
Dr. Louis Essen of the National Physical Laboratory (NPL) in England followed closely the frequency standard developments at NBS and the collaboration with Professor Kusch and, in the summer of 1953, Essen and J.V.L. Parry undertook the construction of a cesium atomic beam frequency standard at their laboratory in Teddington.
The NPL atomic frequency standard was placed in operation in May 1955 and a new value for the cesium hyperfine transition frequency was measured based on the uniform time scale maintained at the Royal Greenwich Observatory. There was a significant difference between this time scale and the Ephemeris Time Scale established by Dr. William Markowitz, Director of Time Service at the U.S. Naval Observatory. Essen and Markowitz collaborated on the determination of a best value for the cesium hyperfine frequency and in 1958 they reported a value of 9 192 631 770 Hz. This value was adopted by the 13th General Conference for Weights and Measures (CGPM) in 1967 as the basis for defining the Sl unit of time (second); this action replaced the traditional definition of the second based upon astronomical observations.
While the first cesium beam atomic frequency standard was being operated at NBS and Dr. Essen was exploring possibilities for constructing an atomic frequency standard in England, Professor Jerrold R. Zacharias at MIT became specially interested in possibilities for a cesium atomic beam frequency standard that could be manufactured on a commercial basis. In pursuit of this objective in 1953 Zacharias and his associates undertook the construction of a feasibility demonstration model in the Molecular Beam Laboratory. Many practical problems were addressed, including:
- development of an efficient cesium beam source;
- development of the “Stabilivac” titanium evapor-ion pump to permit the operation of the cesium atomic beam apparatus without a cumbersome mechanical pump in combination with a diffusion pump;
- development of electronic circuits to permit operation with a quartz oscillator “locked” to the cesium resonance.
In August 1954 the cesium atomic beam frequency standard at MIT was operated for the first time; it was unique in the sense that it was not only quite compact considering its time in history, but it was also the first atomic beam frequency standard to operate in a “closed loop” mode with an external oscillator locked to the cesium hyperfine resonance using a servo control system. The cesium physics package for the MIT frequency standard is illustrated in Figure 3. This photo was taken in the Smithsonian Museum of American History.
With the MIT demonstration apparatus as a starting point, the National Company of Malden, Massachusetts, in the summer of 1954, undertook the design of a cesium atomic beam frequency standard suitable for industrial production. Dr. Richard Daly, a student of Zacharias, had worked on the MIT project and he joined National Company with responsibility for the physics package which, by then, had become known as the cesium beam tube. Dr. Joe Holloway also joined National to work with Daly. The industrial program at National was partially supported by funds provided by National Company; in addition there were substantial contracts from the U.S. Army Signal Corps, Office of Naval Research, Airforce Cambridge Research Laboratories, and Rome Air Development Center. The first National Atomic Frequency Standard, a unit of Model NC-1001, was delivered to the Naval Research Laboratory in September 1956. The name Atomichron was suggested by Dr. Daly and it was formally registered by National Company as a trade name. During the five year period from 1956 through 1960 about 50 NC-1001 Atomichrons were manufactured and most of them were delivered to government agencies. The stability of the Atomichron, about one part in 1010, was substantially better than the original objectives at the time. The Atomichron is illustrated in Figure 4.
Following the delivery of the first commercial Atomichrons, additional development was undertaken to harden them for operation in military environments in ground based installations and aboard ships. A compact model was also developed under contract for the Wright Paterson Airforce Base for installation in military aircraft. This unit was also displayed in the Smithsonian Museum of American History and it is visible in Figure 3.
During the period from 1960-1965 the efforts to improve cesium atomic beam frequency standards were particularly intensive. Joseph Holloway, working at Varian Associates in Beverly, Massachusetts led an effort to make a major reduction in the size of the cesium resonator while maintaining a performance objective of one part in 1011; this level of performance had been established for equipment the size of the original Atomichron. At the same time, Joe Holloway and Richard Lacey, with the collaboration of Norman Ramsey made a systematic study  of all the factors that affect the reproducibility and stability of the cesium hyperfine resonance; on this basis they also made design provisions to optimize performance. The result of this work, a cesium atomic beam tube that was sixteen inches long, was offered, as a component, for sale to industrial firms interested in designing the complete cesium atomic beam frequency standard. Somewhat later, the sixteen inch cesium beam tube was modified to meet special requirements of Hewlett-Packard for a twelve inch configuration. The resulting cesium tube, illustrated in Figure 5, was incorporated into the design of the HP 5060 Cesium Atomic Beam Frequency Standard by Dr. Leonard Cutler and his associates. In 1967 Hewlett Packard acquired the manufacturing rights to the Varian cesium beam tube designs. In the years that followed the HP 5060, illustrated in Figure 6, rapidly became the most widely used cesium beam atomic frequency standard throughout the world.
Additional improvements in cesium atomic beam frequency standard technology have continued throughout the 1970’s and 1980’s with particular emphasis on requirements for space applications. These requirements include:
- additional hardening to survive launch vehicle conditions an function in the space environment;
- measures to extend lifetime;
- weight reduction and operating power reduction;
- quality assurance provisions consistent with requirements for space applications.
These relatively recent developments have been carried out at Frequency and Time Systems, Inc. (FTS) and KERNCO, Inc., both in Danvers, Massachusetts and at Frequency Electronics, Inc., Long Island, New York. Two models of space qualified cesium atomic beam frequency standards are illustrated in Figure 7. A more complete discussion of space qualified atomic frequency standards is the subject of another paper at this meeting.
The performance of every atomic frequency standard is critically dependent upon the length of time that is available for the essential atoms to interact with electromagnetic radiation near their resonance frequency. The longest possible time interval for this interaction is most desirable. In atomic beam frequency standards this time interval is the time of flight through the electromagnetic structure; in this connection the physical length of the structure is the limiting factor.
In 1953 Professor Robert H. Dicke described a method  for extending the time of interaction of atomic particles with the electromagnetic field by the use of a selected gas to provide a medium in which diffusion limits the rate of movement of the active atomic particles. The selected gas, known as a buffer gas, is usually one of the noble gases; in any case, it must be a gas without magnetic properties that would affect the magnetic hyperfine energy states of the active atomic particles. In practice the principle works very well and it is possible to observe very sharp magnetic hyperfine resonances for alkali atoms in buffer gases.
Additional scientific advances provided a very efficient means, in the form of optical pumping, to populate useful energy states of the resonant atoms. An optical method was also developed to detect the resonance signal. The pioneering work on these methods, accomplished in France  and United States  , is described in more detail by Ramsey .
The advantage of the buffer gas cell method is compactness and light weight. However, buffer gas collisions cause a relatively large offset from the undisturbed hyperfine frequency of the active atoms. This pressure dependent frequency offset typically amounts to a few parts in 107. If a single buffer gas is used, the offset is also temperature dependent; fortuitously, it is possible to use mixtures of buffer gases having opposite temperature coefficients and, thereby, compensate the dependence of frequency upon temperature.
Buffer gas cells do not have the intrinsic reproducibility that is possible in atomic beam frequency standards; purity and a high degree of long term stability of the gas mixtures is very difficult to achieve. However, they generally have excellent short term stability and, in this respect, they are usually better than atomic beam devices.
A single isotope of rubidium, Rb87, is usually the active atom of choice in buffer gas cell atomic frequency standards. The physics of the magnetic hyperfine transitions is essentially the same as the corresponding physics of cesium atomic beam frequency standards.
Early studies of buffer gas cell resonance devices were focused on very high resolution spectroscopy and it is now difficult to identify the first proposal to apply these devices to the control of frequency. However, it is clear that the early collaboration of the National Bureau of Standards and the United States Naval Research Laboratory was directed toward the realization of advanced frequency standards .
The first commercial rubidium buffer gas cell frequency standards were produced in the early 1960’s by a number of United States industrial firms, including ITT, RCA, Space Technology Corporation, Varian Associates, Hewlett Packard and Clauser Technology Corporation; in Europe they were first manufactured by Rohde and Schwarz and later by Efratom. A very significant development appeared in the early 1970’s when the Efratom Company in Munich, Germany produced a very compact model configured as a cube about four inches on a side. Several of these rubidium atomic frequency standards were sold in the United States and two of the units were specially prepared by the naval Research Laboratory and flown in a Timation satellite in 1974. The compact rubidium atomic frequency standard units were later manufactured by Efratom in the United States.
Rubidium gas cell frequency standards have been intensively developed for space applications and a review of recent proceedings for this conference as well as the PTTI Applications and Planning Meeting seems to indicate an increasing number of organizations around the world are attracted by the possibilities for small size, low power requirement, fast warm up and low noise characteristics.
In order to increase the effective interaction time for atomic particles in a molecular beam apparatus, Norman Ramsey conceived the idea of placing a box with specially coated walls in the path of the particles at the half way point of their passage through apparatus with two oscillatory electromagnetic fields. After bouncing around in the box for an extended period of time the concept provided for the particles to exit and pass through the remainder of the apparatus. Teflon was considered to be a possible wall coating that would permit wall collisions without disturbing magnetic hyperfine energy states. The experiment, referred to as a broken atomic beam resonance experiment, was performed by Daniel Kleppner who was then a graduate student of Professor Ramsey. The experiment, using cesium as the atomic particles was a partial success and it was found that paraffine as a wall coating was better than Teflon, at least in the case of cesium.
In an effort to improve the experiment, Ramsey and his associates decided to use an active atom having lower mass and lower electric polarizability in order to reduce the undesirable effects of wall collisions. Atomic hydrogen was an attractive possibility but it is very difficult to detect in an atomic beam apparatus. Therefore, Ramsey and his students decided to try to observe the direct radiation from the trapped atoms in a maser experiment following the experience of Townes with ammonia. The result was a new signal source with unprecedented spectral purity and stability.
It should be noted here that coated wall resonators have also been used in place of buffer gas cells in an optically pumped active rubidium maser .
The first atomic hydrogen maser was constructed by Goldenberg, Kleppner and Ramsey and its successful operation at Harvard as a free running oscillator was reported  in 1960. The hyperfine transition frequency for hydrogen was measured to be 1 420 405 751.7667 +/-0.0009 Hz and it is in good agreement with quantum electromagnetic theory. On the average, a hydrogen atom is resident in the coated wall storage vessel for 10 seconds and the sharpness of the associated resonance is very high and the noise level is correspondingly extremely low. The stability of the hydrogen maser over a period of several hours is better than 1 X 10-15. The wall collisions in the hydrogen maser cause a frequency shift of about 1 X 10-11; however, this wall shift is relatively stable in time and, if necessary, they can be determined by special experiments.
In 1962 Robert Vessot and his associates, H. Peters and J. Vanier at Varian Associates in Beverly, Massachusetts undertook the development of a commercial hydrogen maser with the collaboration of Professor Ramsey and Dr. Kleppner. Test results were reported  at the 17th Annual Frequency Control Symposium is 1963. The first units of the Varian Model H-10 Hydrogen Maser Frequency Standard, illustrated in Figure 8, were delivered in 1965.
Commercial hydrogen masers have also been produced by Oscilloquartz, S.A. in Switzerland and, in recent years, the most widely applied hydrogen masers have been manufactured by Sigma Tau Standards Corporation in the United States.
The former Soviet Union produced a large number of hydrogen masers for use in a nation wide network of frequency standards. Commercial units produced by the KVARZ Institute of Electronic Measurements in Russia are now available through a cooperating instrument firm in England.
Hydrogen masers have also been constructed by several research organizations throughout the world. Non-commercial as well as commercial hydrogen masers are used in an increasing number of very demanding applications, including:
- very long base line interferometers (VLBI) for radio astronomy;
- NASA deep space network;
- laboratory reference standards for measurements of high spectral purity.
Improvements of hydrogen masers have continued for a number of special applications. Perhaps the most interesting development was carried out by Robert Vessot and his associates at the Smithsonian Astrophysical Laboratory in Cambridge, Massachusetts. The purpose of this project was a critical test of the special and general theories of relativity by a sub orbital flight of a rocket borne hydrogen maser to an altitude of 6 000 miles. The experiment was very successful and the resulting changes in the rate of the clock agreed with the predictions within a very small uncertainty .
Scientific advances during the closing decades of the 20th century continue to provide splendid opportunities for additional improvements in atomic frequency standards. These advances include:
- improvements in laser sources of optical radiation, particularly diode lasers;
- laser radiation cooling of atomic and molecular particles to microkelvin temperatures;
- methods for trapping atomic and molecular particles within small volumes of space for extended periods of time.
These scientific advances have already been introduced into contemporary atomic frequency standards, including, for example:
- The most recent realization of the national frequency standard of the United States is NIST-7, an advanced cesium atomic beam frequency standard in which optical pumping, utilizing diode lasers, has replaced magnetic state selection. Very complete theoretical analysis of the performance of NIST-7 has determined its uncertainty to be 5 X 10-15 with a prospect for additional improvement by as much as a factor of five . The operation of NIST-7 is automated by a digital servo control system that locks the microwave source to the cesium resonance under computer control with provisions for automated evaluation of factors that contribute to uncertainty.
- A cesium fountain atomic frequency standard has been constructed and operated by the Laboratoire Primaire du Temps et des Frequences (LPTF) in France. In this frequency standard cesium atoms are cooled, using laser radiation, to temperatures of a few microkelvins and allowed to drift upward through a microwave cavity and fall back under the influence of gravity . The successful operation of the fountain demonstrates an idea first explored without a positive result by Zacharias in 1952 . The uncertainty value achieved by operation of the cesium fountain is estimated to be 2 X 10-15 and improvements are anticipated.
- A number of laboratories have constructed frequency standards based upon ions trapped in a radio frequency field. The first practical frequency standard of this type, based on mercury ions and constructed by Hewlett Packard has been in operation at the Naval Observatory for ten years . Mercury trapped ion frequency standards have also been studied extensively at NIST and it is considered possible that they may provide the next opportunity for a major advance in the accuracy of atomic frequency standards .
To conclude this review of the history of atomic frequency standards I want to acknowledge the main sources of historical information that I have used to refresh my own memories of direct experiences and to fill in the gaps where my experiences have not provided insights. In particular, I want to acknowledge:
- Paul Foreman, The First Atomic Clock; Program: NBS, 1947-1954, Proceedings of the 17th Annual PTTI Applications and Planning Meeting, Washington, D.C., 1985.
- Paul Foreman, Atomichron: The Atomic Clock from Concept to Commercial Product, Proc. IEEE, Vol. 73, No. 7, July 1985.
- Wilbert F. Snyder and Charles L. Bragaw, Achievement in Radio, NBS Special Publication 555, October 1986.
- Norman F. Ramsey, History of Atomic Clocks, Journal of Research of the National Bureau of Standards, Vol. 88, p. 301, 1983.
The article cited above is an updated version of an earlier publication by the same author:
- Norman F. Ramsey, History of Atomic and Molecular Standards of Frequency and Time, IEEE Transactions on Instrumentation and Measurement, Vol. IM-21, p. 90,1972.
These accounts of history are all outstanding.