THE WAVELENGTH OF RADIATION OF ARTIFICIAL MERCURY AS A PRIMARY STANDARD OF LENGTH

W. F. Meggers*)

Since I was able to establish that the idea of using the wavelength of light as an invariable, reproducible standard of length was first expressed one hundred and twenty years ago (in 1827) by the French naturalist Jacques Babinet.[^1]

“If one day a collision of the Earth with a comet were to occur, the axis of rotation of the Earth and its shape would undoubtedly change, and from that moment measurements of the length of a pendulum or of an arc of the meridian would no longer be suitable for finding the metre. Discussing this speculative question one day in one of the scientific societies of Paris, Sir Humphry Davy proposed a standard which, in his opinion, could be restored after severe changes in the shape of the terrestrial globe: its unit of length would be the diameter of a capillary glass tube, in which the rise of water would be exactly equal to this diameter. Considering all the difficulties of the experiment, I, in turn, pointed to measurements of the lengths of light waves in a vacuum, as a method which would more certainly lead to the same goal.”

In 1839 the German astronomer Johann Lamont[^2] noted that the diffraction of sunlight provides an exactly reproducible unit of length in the form of the wavelength of the Fraunhofer line \(D\).

“If someone, for example in America, where obtaining an authentic copy of the metre would involve expense and loss of time, should wish to know the exact length of this measure, he could accomplish this by limiting himself only to the making of a diffraction grating, by putting together two micrometer screws with identical threads, or by transferring the pitch of one micrometer screw onto glass. If he found, for example, that with the aid of this grating the line \(D\) in the first spectrum is deflected by \(4^\circ 30'\), then he could say that \(2223 \tfrac{1}{4}\) turns of his micrometer screw are exactly equal in length to a metre” (p. 209).

“From the theoretical point of view, as a definition of a measure it would be more expedient to choose a length that remains constant in diffraction phenomena, namely—the wavelength of light.”

*) William F. Meggers, Journ. Opt. Soc. Amer., 38, 7 (1948). Speech delivered on October 24, 1947, at the annual meeting of the American Optical Society, on the occasion of Meggers’ receiving the Ives Medal of 1947, awarded to him for his work in atomic spectroscopy. Abridged translation by G. Rozenberg.

If we again take the part of the spectrum corresponding to line \(D\), we obtain:

\[ \begin{aligned} 1\ \text{Paris foot} &= 551\,724\ \text{wavelengths},\\ 1\ \text{meter} &= 1\,698\,45\ \text{wavelengths},\\ 1\ \text{Bavarian foot} &= 495\,712\ \text{wavelengths},\\ 1\ \text{English foot} &= 517\,678\ \text{wavelengths}. \end{aligned} \]

Since, in this case, the wavelength cannot be observed directly, but must be calculated with the aid of the angle of deviation, it seems more convenient “to connect the measure with the angle itself” (p. 212).

In 1870 the English physicist J. C. Maxwell pointed out[^3] that the wavelengths of hydrogen radiation must be invariable and therefore more suitable as a standard of length than the Earth.

“Since the Earth was measured as the basis for a permanent standard of length, and in order to preserve the material from which the standard was made from any changes, all the properties of metals were investigated... Yet the dimensions of our Earth and the time of its rotation, which are quite constant with respect to our modern methods of comparison, are not such by any physical necessity. The Earth may contract as a result of cooling, or increase owing to the settling upon it of meteorites falling onto it; the speed of its rotation may also gradually diminish, and yet it will remain a planet to the same degree as before. But a molecule, say, of hydrogen, if its mass or time of vibration were to change even slightly, would no longer be a molecule of hydrogen. If, therefore, we now wish to obtain standards of length, time, and mass that would be absolutely unchangeable, we must seek them not in the dimensions, or motion, or mass of a planet, but in the wavelength, the period of vibration, and the absolute mass of these indestructible, unchangeable, and perfectly simple molecules.”

In that same year the Danish physicist V. S. M. van der Willigen[^4], discussing the difficulties and errors inherent in the use of the seconds pendulum as a natural standard of length, proposed that the sodium lines be chosen as the standard.

“Under these circumstances it may prove useful to draw attention to another quantity existing in nature—a quantity which is everywhere and always within our reach, which at any time and in any place has the same length and which certainly does not depend on boundaries established on the surface of the Earth, or on astronomical measurements. This quantity is the wavelength of sodium light, or, to speak more precisely, taking into account that the yellow line of the spectrum of this light invariably splits when the dispersion becomes considerable, the length of the middle wave of sodium light, of course in vacuum. At any place on Earth now, or thousands of years hence, so long as it will be possible to have at one’s disposal a grain of common salt, one will be able to determine the wavelengths of these two vibrations, take the mean, and use it for verifying measuring standards” (p. 153).

Van der Willigen was, apparently, the first to arrive at the conclusion that length can be determined in terms of wavelengths by counting Newton’s rings of sodium light, displaced by the slow approach of a plane glass plate to a convex one[^5].

“My first thought was to resort to the method used by M. Fizeau for determining the expansion of glass, quartz, and other analogous materials. I wished, just as this scientist did, to carry out a gradual—

... controlled bringing together of a plane and a convex glass and to count the number of Newton’s rings fitting within homogeneous sodium light. From the known bringing together of the glasses one could find the wavelength, and then, by moving several thousand rings until sufficient friction and a large contact area were obtained, arrive at a material length equal to a known number of wavelengths* (p. 161).

He regretted that this method was at that time practically unfeasible and contented himself with measurements of three plane gratings, from which he derived an approximate number of yellow sodium waves in a meter.

A. A. Michelson and E. W. Morley^5 were the first to devise an effective method of applying the phenomenon of optical interference to measurements of length. In 1887 they outlined a “method of using the wavelength of sodium light as an actual and practical standard of length.” This method, which involved the use of an interferometer, was applied by them in their famous experiments on the relative motion of the Earth and the ether. It consisted of measuring a length and counting the equivalent number of interference fringes. The cited work also contains the first mention of the observation of the doublet character of the red radiation of hydrogen. The disappearance of interference with retardations of 15,000 and 45,000 waves indicated a double line with components separated from one another by approximately one-sixtieth of the distance between the yellow sodium lines.

Two years later, in an article entitled “On the Feasibility of Establishing a Light Wave as the Fundamental Standard of Length,” Michelson and Morley^6 describe the method subsequently used in measurements of the meter, and report the first interferometric values of wavelengths for one sodium line, one lithium line, and three mercury lines. The following quotations contain important remarks concerning mercury radiation.

“In the problem of transforming a light wave into a standard of length, it is desirable to use, as an auxiliary standard, a metallic bar with two plane surfaces, separated from each other by the greatest possible distance at which interference (between two light beams having this path difference) can still be well measured. Preliminary experiments have shown that, in the case of the green mercury wave, this distance may amount to four meters” (p. 181).

“Yellow mercury light is inconvenient, since it is double, and both radiations have almost the same intensity.... The bright green (mercury) line gives very distinct rings, even if the path difference amounts to half a million waves, so that, in all probability, it will be exactly this wavelength that is used as the fundamental standard of length” (p. 183).

In his second paper, “On the Application of the Interference Method to Spectroscopic Measurements,” Michelson^7 gives a “curve of visibility” for the lines of oxygen, sodium, zinc, cadmium, thallium, and mercury. This work marks the discovery of hyperfine struc-

structure of spectral lines and a rejection, at least temporarily, of the green mercury line (5461 Å) as the fundamental standard of length.

“The green mercury line is one of the most complex among those investigated” (p. 292).

We can easily imagine Michelson’s disappointment when he made this discovery, which dethroned the green mercury line as the fundamental standard of length.

“The main purpose of this investigation was to find radiation sufficiently homogeneous for its use as a standard of length. ... The red cadmium line is almost ideally homogeneous and makes it possible easily to estimate the phase difference of the interference fringes to one hundredth of a fringe at a total distance of 20 millimeters, or more than 300,000 waves” (p. 298).

These classical investigations led to Michelson’s invitation to the International Bureau of Weights and Measures, where in 1892 he performed his famous experiment, by means of which he determined the number of red cadmium waves in the meter.

Before Michelson’s classical investigations, all spectral lines had been regarded as monochromatic and invariable, except for the obvious change in wavelength indicated by Doppler in 1842, caused by the relative motion of the light source and the observer along the line of sight. Naturally, the Doppler effect disappears when the light source and the observer are at rest, as is usually the case under laboratory conditions. However, the kinetic energy of individual emitting particles gives rise to Doppler broadening of spectral lines, limiting their monochromaticity. Indeed, at low pressures and in the absence of strong electric fields, the width of spectral lines is completely determined by the Doppler effect caused by the motion of the particles emitting the light. In units of wavelengths this width is proportional to $\lambda \left(\dfrac{T}{M}\right)^{1/2}$, where $\lambda$ is the mean wavelength, $T$ is the absolute temperature, and $M$ is the atomic weight. In other words, the most monochromatic lines are emitted by massive, slowly moving atoms. Since a mercury atom is almost twice as heavy as a cadmium atom, and its radiation can be excited at an absolute temperature less than half that required to excite the luminescence of cadmium, the mercury lines, other conditions being equal, have only half the Doppler width in comparison with the cadmium lines.

The hyperfine structure discovered by Michelson served as a subject of investigation and reflection for forty years before it was finally satisfactorily explained as a consequence of nuclear masses and moments. Since the effective wavelengths of complex lines in interferometric measurements change depending on the phase relations of the various components, and since the relative intensities of the components can vary, it is absolutely necessary, when choosing a natural standard of length, to avoid-

employ lines with hyperfine structure. However, the mere presence of more than one isotope or of one or more nuclei with spin moment does not yet entail the unsuitability of all spectral lines. Thus, although cadmium has eight isotopic forms and two nuclear spins, neither isotopic nor hyperfine structure has been found in Michelson’s red line, though it is clearly observed in the other cadmium lines. Natural mercury is a mixture of seven isotopes, including two with spins, and not one of the lines of ordinary mercury has proved to be free of structure. In 1931 I expressed the idea that sharp lines might be obtained from an isotope of mercury having even mass, if it should prove possible to isolate it in sufficient quantity by means of a mass spectrograph. Nothing came of this proposal at the time, since the corresponding separation of the mercury isotopes would have been too expensive.

In his investigations of the structure of lines Michelson⁷ found that

“the influence of temperature and pressure on visibility can be well explained by the kinetic theory. Indeed, it is almost certain that the chief, if not the only, causes of the broadening of spectral lines and of the corresponding deterioration of visibility are changes in the wavelength of light caused by the motion of the source along the line of sight, and also changes in the period of the source due to frequent collisions” (p. 293).

A later work by J. Humphreys and J. F. Mohler⁸ showed that an increase in pressure not only leads to a broadening of spectral lines, but also causes their asymmetry and a shift toward longer wavelengths.

The only other effective cause of asymmetry and displacement of lines is the Stark effect, discovered in 1913. In contrast to the magnetic splitting of spectral lines (the Zeeman effect), which, generally speaking, is highly symmetric and in ordinary sources is quite imperceptible, electrical splitting or displacement is usually asymmetric and is easily confused with the pressure effect. In light sources, electric fields may be produced both by applied potentials and by the high density of electrons or ions in the discharge.

Although these principal causes of displacement of spectral lines—the pressure effect and the Stark effect—were still unknown in 1892, Michelson intuitively avoided their influence, since his light sources operated at low pressures and with a small current density. Today, fifty-five years later, we stand in admiration before his astonishing skill, displayed in choosing, with the aid of his primitive interferometer, the red cadmium line as the fundamental standard of length. After investigating the entire spectrum with improved apparatus, we must acknowledge that he chose the best line in nature.

The classic experiment of Michelson, by which he determined the ratio between the meter and the wavelength of the red cadmium line, was repeated several times, and its results were fully confirmed1.

In 1907 the International Union for Cooperation in Solar Research established, as the definition of the unit of wavelength—the angstrom—the value of the wavelength of the red cadmium line 6438.4696 Å, observed by Benoît, Fabry, and Perot in 1906. The same value was adopted in 1922 by the International Astronomical Union, the successor of the International Union for Cooperation in Solar Research, which had broken up during the First World War. It is already sufficiently old to be worth recalling that world calamities have always aroused fears that the world prototype meter, kept near Paris, might become the victim of an unfortunate accident. The only way to avoid this consists in replacing the destructible meter by an indestructible, yet accurately and easily reproducible, wavelength as the fundamental standard of length.

In 1927 the National Bureau of Standards recommended to the International Conference of Weights and Measures that the wavelength of the red radiation of cadmium vapor be chosen as the primary standard of wavelength and that the meter be defined in terms of this wavelength2.

“It is recommended that the Conference adopt the wavelength of the red radiation emitted by cadmium vapor, determined by Benoît, Fabry, and Perot, as the fundamental standard of the wavelength of light… The meter should be defined by the relation: 1 meter = 1,553,161.13 wavelengths of the red radiation of cadmium under quite definite normal conditions.”

The ultraconservative Conference considered3 that such a definition of the meter would be a threat to the metric system, and explained that the question was not the establishment of the true ratio between the meter and the wavelength, but only the determination of the metric value of the latter, which could be changed by subsequent experiments. Therefore the decision it adopted specifically established that the value of the meter is provisionally expressed in wavelengths of the red rays of cadmium by the number 1,553,164.13. Presumably, if different values are found in the future, the meter must be taken as unchanged and each of the wavelengths, or the angstrom, must be changed.

Strictly speaking, the world primary standard of length is still the distance, established in the eighteenth century, between two strokes (from 6 to 8 μ in width) on a platinum-iridium bar kept at the International Bureau near Paris, despite the fact that practically all accurate measurements of length in the twentieth century have been made and henceforth will be made with light waves. Who would grieve over the demise of an arbitrary archaic measuring rod, when a better standard, meeting scientific recognition, is available!

Atomic research and nuclear technology have created methods and possibilities for the transmutation and production, on a large scale, of pure elements absent in nature. In particular, it is now possible to make absolutely pure mercury from gold in any desired quantity (contrary to alchemy) and to use it as a material for creating a spectroscopic light source emitting spectral lines twice as sharp as the best of those emitted by natural elements.

The nuclear reaction converting gold was first carried out in 1934 by Fermi and others,^12 who used, as a source of neutrons for irradiating gold, a mixture of beryllium powder and radon. The neutrons are captured by the nuclei of gold atoms and form a heavy radioactive isotope of gold, which rather quickly undergoes β-decay (half-life 2.7 days) and is transformed into a stable isotope of mercury:

\[ {}_{79}\mathrm{Au}^{197} + {}_{0}n^{1} \rightarrow {}_{79}\mathrm{Au}^{198} \rightarrow {}_{80}\mathrm{Hg}^{198} + {}_{-1}\beta^{0}. \]

In 1940, Wien and Alvarez^13 reported that, by bombarding gold with neutrons from a 60-inch cyclotron, they had obtained enough mercury to create a light source emitting spectral lines entirely free of hyperfine structure. Irradiating 31 grams of gold with neutrons for a week, they obtained enough mercury to identify it spectroscopically in tiny electrodeless lamps whose service life was about one minute. A month’s exposure produced an amount of mercury sufficient for operating a lamp for approximately five minutes and obtaining the first Fabry–Perot spectrogram. A ten-month exposure near the cyclotron made it possible to fill a discharge tube that operated for more than thirty minutes.^14 At this ratio, a thousand years would be required to create a single lamp with Hg^198 having the usual service life—about a thousand hours.

At the spectroscopic conference in Chicago in 1942, while discussing the primary wavelength standard,^15 I drew attention to the theoretical advantages of the green mercury line over the red cadmium line, especially if sufficient quantities of Hg^198 became available. In 1942 L. J. Briggs, then director of the National Bureau of Standards, approved the purchase of 1250 g of fine gold and secured the cooperation of the University of California for irradiating this gold over the course of one year or more. The intention was that if 30 g of gold could yield enough mercury to fill one lamp, then 1250 g would provide for the manufacture of forty lamps of the same type, or, possibly, several more efficient lamps. Unfortunately, the Second World War interrupted the experiments, and only submicroscopic quantities of artificial mercury were made. From it, in 1943,

several tiny discharge tubes were made, but in size, brightness, and lifetime they resembled faint fireflies and had no practical significance.

The prospects were quite discouraging when, shortly before the end of the war, rumors appeared of a secret neutron source, thousands of times more effective than the largest cyclotrons. One day a gold ingot was delivered to us, the history of which remained secret. It was only said that the gold contained Hg¹⁹⁸. Subsequent distillations eventually made it possible to create a lamp emitting the spectrum of mercury, whose lines had no hyperfine structure. This lamp was excited by radio waves with a frequency of 110 kilocycles. It was used chiefly for demonstrations and had a service life of about fifty hours before it failed as a result of clean-up, i.e., a decrease in the mercury vapor pressure under the action of the electric discharge.

Experiments with lamps containing micrograms, or even several milligrams, of mercury showed that the clean-up effect seriously limits the service life of such lamps. In electrodeless lamps, under the action of the discharge, mercury is gradually driven into the glass, causing darkening of the walls of the tube. In lamps with electrodes, the electrodes, in the process of sputtering, or as a result of the same process of penetration, tend to absorb mercury until no mercury vapor remains in the lamp at all.

In 1945 the gold of the National Bureau of Standards was transferred from California to Tennessee. The process of treating this gold still remained secret, but a year later we extracted from it by distillation about 60 mg of mercury.

Having only 60 mg of Hg¹⁹⁸ and bearing in mind the numerous requirements imposed on the lamps, it was necessary, as far as possible, to economize Hg¹⁹⁸ and to try to design a satisfactory lamp with minimal losses of the precious material as a result of clean-up.

In July of last year E. U. Condon, now director of the National Bureau of Standards, proposed to the Atomic Energy Commission that a somewhat larger quantity of gold be irradiated with neutrons. The time, place, and method of treatment remain secret, but we hope to obtain, within a year, approximately two grams of artificial mercury. Meanwhile experiments are being carried out with natural mercury and with the 60 mg of Hg¹⁹⁸ previously obtained by us, aimed at developing a spectroscopic light source that would be suitable for careful wavelength measurements.

In designing a lamp with Hg¹⁹⁸ intended to produce radiation suitable as a primary standard of length, it is desirable to achieve the maximum possible with respect to each of the following five characteristics:

1) monochromaticity, 2) reproducibility, 3) intensity, 4) service life, and 5) convenience in use.

It is obvious that some of these requirements contradict one another and, generally speaking, it is necessary to resort to a compromise.

1) Monochromaticity, or sharpness of the spectral lines, depends on the operating conditions of the light source, as well as on certain atomic properties already mentioned. In order to obtain fine lines from atoms of a given mass, it is necessary to reduce Doppler broadening, for which high temperature is responsible; broadening due to the action of pressure and collisions; and also resonance broadening, for which a high density of the gas or vapor is responsible; Stark broadening, produced by large gradients of electric potential; and interatomic Stark broadening, due to the high current density in the discharge. In other words, the source must operate at low temperature, low pressure, moderate voltage, and low current density. If high intensity is necessary, then to some extent one has to sacrifice sharpness of the lines.

2) The reproducibility of the wavelengths emitted by a light source rests on the constancy of the pressure of the gas or vapor, on a low current density, and on the absence of self-reversal and hyperfine structure. Changes in pressure are always accompanied by small changes in wavelength, while the interatomic Stark effect, which appears in the case of a strong current, always causes an asymmetric displacement of the lines. For example, it can be shown that the increase in the wavelength of the red line emitted by an arc in metallic cadmium vapor is, at currents of 2.5 and 5.0 amperes, \(0.0003\) and \(0.0007\) Å, respectively. Self-reversal also often proves to be asymmetric; but even when it is symmetric, it is undesirable, since it reduces the intensity and may lead to an erroneous interpretation of the interference pattern. Self-reversal becomes imperceptible if lines corresponding to transitions to the ground state are avoided, or if the laced discharge is viewed not along but across the luminous column. It is quite evident that the relative intensity of the components of the hyperfine structure depends on the operating conditions of the light source, and it has been sufficiently proved that unresolved fine structure distorts the true values of the wavelengths of spectral lines.

3) The intensity, or brightness, of a lamp filled with metal vapors is mainly a function of the current strength per unit cross section of the discharge. The latter may increase both as a result of a decrease in the resistance of the circuit, entailing an increase in current strength, and as a result of a decrease in the cross section of the discharge when it is constricted. In addition, the intensity can be increased if a long column of luminous vapors is viewed along, and not across, its axis. In general, however, an increase in current density

is accompanied by an increase in temperature and pressure, as well as by an intensification of the interatomic Stark effect, which entails a broadening of the spectral lines. Therefore, when operating a Michelson lamp, in order to obtain the cadmium red standard line the current density is limited to seven milliamperes per square millimeter of the discharge cross-section.

Line broadening caused by collisions with atoms of the same kind as the radiating ones can, in some cases, be greatly reduced by introducing noble gases into the lamp (at a pressure of several millimeters of mercury), since the effect of the pressure of a foreign gas is much weaker than that of the diluted radiating gas. In addition, collisions of the second kind with ions of the noble gas substantially increase the intensity of the emitted light.

The highest intensities are usually obtained with an arc discharge with a hot cathode; but if the current is sufficiently small to make the broadening and displacement of the lines negligible, the advantages in intensity are lost. A similar loss of intensity occurs if, in order to make the lines sharper, the vapor pressure is reduced by strong cooling. High intensity must inevitably be sacrificed when monochromaticity and reproducibility are more important. The observed intensity of some spectral lines is strikingly diminished as a result of partial or complete self-reversal in the light source, if the temperature or excitation is not everywhere uniform.

4) Apart from chance factors, the service life of a lamp containing minute quantities of gas or vapor is determined by the time of clean-up. This is comparatively short in electrodeless lamps excited by a high-frequency discharge, which irreversibly drives atoms of the gas or metal into the glass walls of the vessel, as a result of which the walls of the lamp darken. In lamps with internal electrodes, clean-up likewise occurs either through absorption of charged gases and vapors, or through their capture in the process of sputtering of the material from which the electrodes are made. The mechanism of clean-up is still not fully understood and cannot be predicted for the case of lamps with \(Hg^{198}\), but experiments are being conducted with small quantities of natural mercury and with lamps of various types, made of different materials, in order to determine the conditions that would guarantee a service life of a thousand hours or more. If and when a much larger quantity of artificial mercury is produced, the inevitable losses due to clean-up will become less significant. In my opinion, this is a good argument in favor of using gold rather than cadmium as the control element in designing new boilers with chain reactions. As before, so now, when cadmium is used, no useful by-products have been obtained, as would be the case if gold were transformed into \(Hg^{198}\).

5) Convenience of operation is an important factor determining the possibility of widespread dissemination of spectroscopic sources—

WAVELENGTH OF THE RADIATION OF ARTIFICIAL MERCURY

light sources and their use for precise measurements. In fact, because of the inconvenience of working with a standard cadmium lamp with a heater, this source was used extremely rarely, except for measurements of the meter and determinations of secondary wavelength standards.

Although practically superior monochromaticity is attainable with transverse observation of a directed atomic beam, such a source is too inefficient and inconvenient to be considered in connection with the establishment of a wavelength standard. A sufficiently intense spectrum with very sharp lines is observed in a discharge with a cooled hollow cathode, but this source is so inconvenient to handle that its use is limited almost exclusively to the study of hyperfine structure by the most highly qualified spectroscopists.

An arc with a hot cathode in mercury vapor, operating at the ordinary voltage of a commercial power network, is unsurpassed in convenience, brightness, and lifetime, but it requires more than several milligrams of mercury and sacrifices monochromaticity and reproducibility to intensity[^15]. Such a source cannot be recommended as the primary standard, but it will, of course, be used in those cases where convenience and intensity are the most important requirements.

The greatest simplicity of construction is achieved in electrodeless discharge tubes, which are made by sealing a small quantity of gas or vapor into an evacuated glass or quartz tube. Formerly, excitation was produced by a high-frequency Tesla discharge, but in recent years a simple and convenient radio-frequency generator of very short waves has been used for this purpose. By this method intense radiation of vapors can be obtained at extremely low pressures, but the working pressure requires rather careful selection. A serious shortcoming of this type of source is its relatively rapid deterioration, owing to the penetration of vapor particles into the walls of the tube.

It is clear that the most convenient and practical light source is a Geissler tube specially designed for obtaining spectra of gases at low pressures. Michelson’s cadmium lamp is essentially also a Geissler tube, but the necessity of enclosing it in a heater maintaining a temperature from 300 to 320° C entails undesirable Doppler broadening of the lines, as well as inconvenience in using it. Since mercury has an appreciable vapor pressure even at temperatures below zero degrees Celsius, in a Geissler tube it is possible to excite the radiation of sharp mercury spectral lines. These tubes normally operate with a cold cathode and can, if desired, be cooled with air or water. They require the application of potentials of only a few thousand volts and currents of several milliamperes.

In October 1947 I made an electrodeless lamp of Pyrex, and N. K. Bis made two lamps of Vycor glass with internal

with molybdenum electrodes. Each of these lamps contains 5 milligrams of \(Hg^{198}\) and argon at a pressure of 5 millimeters of mercury. Unfilled lamps of Vycor glass and fused quartz, with zirconium electrodes and a capillary adapted for water cooling and for observation both along and across the luminous column, have also been made. The operating characteristics of these experimental lamps will be studied in order to determine precisely which type of lamp can be recommended as a satisfactory source for the primary standard of length.

Using a Fabry–Perot interferometer with a 25-millimeter etalon, preliminary values were determined for the wavelengths of a dozen \(Hg^{198}\) lines lying in the interval between 3341 and 5791 Å, relative to the primary wavelength standard. Publication of the observed wavelengths of \(Hg^{198}\) will be postponed until final values have been obtained. For the present it is enough to say that the preliminary values, analyzed by means of the combination principle of spectroscopy, proved to be correct to an accuracy of one twenty-millionth, whereas the best measurements made with natural mercury gave deviations of one hundred-thousandth, caused, undoubtedly, by distortion of the wavelengths by unresolved hyperfine structure.

The future availability of satisfactory lamps emitting an excellent standard of length is made possible by the conversion of large quantities of \(Au^{197}\) into \(Hg^{198}\) and by improvements in lamp construction to reduce their sputtering.

It seems beyond doubt that the progressive scientific world will welcome the adoption of the wavelength of the green line of \(Hg^{198}\), 5461 Å (and, possibly, the infrared line 10140 Å for large lengths), as the primary standard of length.

References Cited

1. J. Babinet, Ann. Chim. et Physique, 40, 177 (1829).
2. J. Lamont, Jahrbucht d. König. Sternwarte bei München, 3, 188 (1839).
3. J. C. Maxwell, Rep. Brit. Ass. Nat and Abstr., p. 7 (1870).
4. V. S. M. van der Willigen, Archives Musée Teyler, 3, 142 (1870).
5. A. A. Michelson and E. W. Morley, Am. J. Sci., 34, 427 (1887).
6. A. A. Michelson and E. W. Morley, Am. J. Sci., 38, 181 (1889).
7. A. A. Michelson, Phil. Mag., 34, 280 (1892).
8. W. J. Humphreys and J. F. Mohler, Astrophys. J., 3, 114 (1896).
9. J. E. Sears, Sci. Progr., 31, 209 (1936).
10. Proces. Verbaux Seances Con. Int. Poids Mesures, 12, 64 (1927).
11. Trav. Mem. Bur. Int. Poids Mesures, 18, 53 (1930).
12. E. Fermi, E. Amaldi, O. D’Agostino, F. Rasetti and E. Segré, Proc. Roy. Soc., 146 A, 483 (1934).
13. J. H. Wiens and L. W. Alvarez, Phys. Rev., 58, 1005 (1940).
14. J. H. Wiens, Phys. Rev., 70, 910 (1946).
15. W. F. Meggers, Rev. Mod. Phys., 14, 50 (1942).