Recent work on gas cell frequency standards has led to a clearer understanding of the capabilities of such systems. The physics of the quantum mechanical system allows the designer to start with a linewidth of approximately 1 part in 10 8 and with a signal-to-noise ratio, in a one-second noise bandwidth, in excess of 10 4 . Carefully engineered systems have been demonstrated to have high stability with high reliability and small weight and volume. Frequency stability, which is the important quality of a standard, has been studied over a wide range of averaging and measurement times. A plot of standard deviation of frequency stability σ shows a broad minimum of σ= 2.5×10 -12 for a run of 40 hours; a σ= 2×10 -11 for a run of 15 minutes (¼ second averaging time); and σ = 4×10 -11 for a run of 330 days.
Numerous measurements of the speed of light published during the last 30 years lead to widely divergent results as compared with the assigned experimental uncertainties. Because of wide diversity in the methods employed in the measurements, all of the data may not be combined effectively in a grand average. Sufficient data had been obtained by the geodimeter method to group them and derive a statistical estimate of the uncertainty in the speed of light by this method. This result, and conclusions reached from careful examination of several experiments, leads to the conclusion that the value 299,792.5 km which has been internationally adopted for use in radio propagation and geodetic work is very close to the best value and not likely to be in error by as much as one part in one million.
Ammonia masers are studied concerning their use as frequency standards. They oscillate on the J = K = 3 inversion line of N 15 H 3 . Single beam masers allow the realization of a frequency standard with a stability of 2-3.10 -11 , and an accuracy of about 10 -9 . Experiments on double beam masers suggest that their use makes it possible to obtain a stability of the order of 10 -12 and an accuracy better than 10 -10 .
Following a proposal by Kusch a thallium atomic beam frequency standard has been constructed and put into operation. The main characteristic is that its frequency of 21,310 Mc is only very weakly affected by inhomogeneities of the C-field. The construction, which is similar to conventional cesium standards, differs from these in the following main points: a high (1000°K) oven temperature, high (14,000 gauss) A-and B-fields, an oxidized tungsten filament detector. The interaction length is 90 cm; the Q factor of the resonance is 160 million. The measurements show that the uncertainty in the Zeeman-shift is negligible, while the error arising from an unsymmetrical cavity is the same as in the case of cesium.
The present nearly universally accepted frequency standard is the atomic cesium resonance at 9.192631770 Gc. However, for many reasons it is not practical to use a cesium device for continuous frequency generation. Secondary frequency standards such as quartz-crystal controlled oscillators are commonly used for this purpose. Recent improvements in oscillator, oven and crystal unit design have produced crystal oscillators which require infrequent adjustment to maintain their frequency constant to a few parts in 10 11 (pp 10 11 ). This paper will present the performance of several of these oscillators and the state of the art in crystal unit fabrication and temperature control. The average drift rate and cyclic variations of 2.5 and 5 Mc oscillators will be given. The megacycle oscillators appear to have an average drift rate of less than 1 pp 10 10 /day while that of the 2.5-Mc oscillators is less than 1 pp 10 11 /day. The short-time stability (down to 10-msec period) of both types of units when measured in the kilomegacycle range appears to be of the order of a few parts in 10 10 . The effect of oven cycling and oscillator shutoff on the average drift rate has been measured. The period required for the frequency to stabilize after oven shutdown depends on the oven off-time. The oscillator frequency will return to a normal drift rate within 96 hours after a 24-hour shutdown, while it requires 5 to 6 days to return to a normal drift rate after a week shutdown.
The atomic time scale A.1 is based on an assumed frequency of 9 192 631 770 cps of Ephemeris time, and is derived from the operation of cesium resonators at 9 laboratories. Coordinated transmissions of time and frequency make the atomic time scale available. The International Committee of Weights and Measures may redefine the unit of time interval, the second, by an atomic transition in 1966.
This paper presents some methods for making comparisons between standards and items undergoing calibration. These methods may be used in a variety of measurements. The purpose is to accumulate data that provide objective estimates of the precision and that are also useful in detecting sources of systematic errors. This purpose is achieved in using some standard statistical designs in the scheduling of the work program. The problems of stating the uncertainty and of combining the uncertainties in a chain of calibrations are discussed.
A new type of microcalorimeter has been developed at the Electrotechnical Laboratory as the national millimeter-wave power standard in Japan. This is the single load type calorimeter using semiconductor Peltier thermoelements and a standard bolometer mount. The experiment has been successful in the frequency region of 35 kMc, and this method is considered to be suitable for millimeter-wave regions. By this method and equipment, the effective efficiency of the standards mount is determined precisely. The errors in this microcalorimetric technique are investigated with the aid of heat flow analysis and auxiliary experiments. As a result of this, the absolute accuracy of the power measurement in the millimeter wave region (35 kMc) is evaluated with high accuracy (better than ± 0.5 per cent).
Using the Curie-Cheneveau principle, A. Thorpe and F. E. Senftle have developed an absolute yet simple force method for measuring the magnetic susceptibilities of very small samples. In our laboratory an apparatus was constructed so that absolute susceptibilities could be determined by both the Thorpe-Senftle and the Gouy methods down to liquid helium temperatures. Measurements were made on powdered samples of the same batch of (NH 4 ) 2 OsBr 6 , HgCo(SCN) 4 , and Fe(NH 4 ) 2 (SO 4 ) 2 ·6H 2 O. Comparable accuracies were obtained by the two methods; however, about 3 ×10 -2 as much sample is required to do a Thorpe-Senftle measurement.
Experiments have shown that in addition to the well-known radiation pressure, another more efficient thermal mechanism results in the production of an elastic stress wave when an energy pulse is absorbed at the surface of an elastic body. This mechanism permits measurement of the peak-power density of pulses of electromagnetic energy by the use of an energy-absorbing surface, an elastic wave sensor and an indicating device. Analysis and experiments will be described to illustrate the operation of this measurement system when it is actuated by microwave pulses, pulses of light from a laser or a stroboscopic lamp and by a pulsed beam of electrons in an evacuated chamber.
The selection of a material suitable for use as a standard of dielectric properties at microwave frequencies is discussed, and tests are described which indicate that a glass and a glass ceramic are satisfactory for such standards. The probable accuracy of measurement of the real part of the dielectric constant is estimated at ± 0.3 per cent. Loss measurements are discussed. A correction is developed for the error resulting from the small airgap often present around the sample in transmission-line measurements. The effects of humidity and temperature variations are examined, and preliminary results of measurements to 800°C are given.
The design and construction of a relatively compact device for use as a frequency standard is discussed. Hydrogen scavenging is done by titanium sorption pumps having a continuous operating lifetime of one year. The magnetic and thermal shielding of the cavity is described, and frequency errors due to various effects are discussed. Preliminary data of measurements of relative stability between two masers is presented and a means for measuring short- and long-term stability is discussed.
An adjustment-free peak-to-dc transfer technique is described for making precision comparisons between audio frequency ac voltages and a reference dc. Limitations and estimated systematic errors are discussed and justifications for an over-all ac-dc transfer worst case error of 50 ppm presented. Photographs of output oscilloscope patterns are shown for various conditions of ac-dc transfer agreement for several different ac sources. Experimental over-all comparison against a modification of the Smith-Clothier peak calibration technique is described with agreement on the order of 30 ppm.
The characteristics of a double-beam maser of N 14 H 3 3-2 line were investigated. In the experiment, Zeeman effect is used for the determination of the reference frequency. The shift of the reference frequency due to a variation of 30 per cent in the ammonia pressure, that of 20 per cent in the focuser voltage or that of 10 per cent in the intensity of the perturbing magnetic field is less than one part in 10 10 . The variation in the traveling-wave effect by change of the effective beam intensity, both in single-beam and double-beam types, were calculated for practical use. The phase variation and noise in the frequency multiplier and the quartz oscillator were preliminarily investigated in order to improve the precision of frequency comparison between maser and quartz crystal oscillator.
The increasing demand for higher accuracy in coaxial line measurements and standards, particularly for the VHF and UHF bands, has led to the development of cylindrical metal film resistors of uniform surface resistivity in which the main characteristics have been considerably improved when compared to the cracked-carbon film resistors which have hitherto been used in precision coaxial resistor mounts employing a tractorial outer conductor. A description is given of an alternative design employing a conical resistor element with a cylindrical outer conductor which results in a much simpler mechanical design with fewer discontinuities than the tractorial configuration. It is estimated that with the new resistor element it should be possible to achieve a resistive termination of VSWR 0.996 in the range 0-4 Gc with a ¾ inch OD system. The principle of operation is explained in terms of electromagnetic wave propagation over a resistive film surface in which the tilt angle θ of the electric field from the normal to the film surface is given by sin θ= ρ/Z m where ρ is the surface resistivity and Z m the wave impedance of the medium. A simple explanation is given of the generation of the tractorial and conical systems by transformations from the simple parallel-plate line terminated in a resistive film sheet. Second-order effects caused by field penetration of the resistive film are described and formulas given for their evaluation and compensation.
Some basic properties of a proton maser making use of the dynamic Overhauser effect for population inversion and enhancement are reviewed. The long proton spin relaxation times of liquid samples, in the order of seconds, facilitate the production of self-sustained continuous oscillation with high spectral purity. Theoretically the proton maser oscillator should have a spectral purity of some parts in 10 14 for an inhomogeneously broadened proton resonance absorption linewidth of 100 cps at a frequency of 15 Mc. A proton maser was operated at this frequency in a temperature stabilized permanent magnet. In preliminary experiments a frequency drift in the order of 1 part in 10 9 /sec was measured. The spectral purity was tested by beating together the frequencies of two proton maser oscillators operating within the same magnetic field. Preliminary results yielded frequency deviations of less than 6 parts in 10 10 . The sources introducing frequency instability are discussed and their relative contributions evaluated.
Two systems, in which the frequency of a high quality quartz crystal oscillator can be controlled by a servo system employing as a reference frequency the (F = 4, m F = 0)↔(F = 3, m F = 0) transition in the ground electronic state of cesium 133 , have been in operation for about one year at the National Bureau of Standards. These systems are presently used in conjunction with the United States Frequency Standard, NBS II, and the alternate standard, NBS I, for measuring the frequencies of the United States Working Frequency Standards on a regular basis. The dependability, precision, and accuracy of the servo-derived measurements have been compared with the corresponding figures for the more direct manual method. Although both measurement systems have been found to be highly dependable, the servo method has significant advantages with respect to convenience of operation and measurement precision. These advantages can be utilized with no sacrifice of accuracy. Typical servo measurement precision is 2×10 -12 for a 30-minute averaging time, while the measurement accuracy for both methods is 1.1×10 -11 . For longer measurement periods of 12-14 hours, precisions and reproducibilities of 2×10 -13 have been observed.