Development Paths of the Optical Institute

S. I. Vavilov, Leningrad

Two reports at the present session of the Academy—the communication by Academician D. S. Rozhdestvenskii and my own—have the common task of characterizing, on the basis of the work of the State Optical Institute, the state of our research work in optics. Of course, the Institute by no means exhausts all research in the field of optics in the Union; on the other hand, our reports concern only selected problems of the Optical Institute. Therefore, from the very beginning it is necessary to warn of the selective and incomplete nature of the characterizations contained in our communications.

Light Technology at the Optical Institute

The Applied Character of Modern Optics

Two branches of physics that practically almost exhaust its content—the science of matter and optics, the science of light—are far from equal either in scope or in practical significance. Optics is undoubtedly more limited and specialized, and its technical role has hitherto been extremely narrowed, because the energetic properties of light have only minimal application. Despite the fact that sunlight is the principal source of the earth’s energy, modern technology continues to make use, as energy resources, of secondary sources: coal, oil, water, wind, etc. We do not doubt that such a state of affairs is only temporary; that sooner or later—and most likely sooner—it will be necessary to turn to the primary source, i.e., to the sun. Then roles will change, and light technology will take the principal place among the other branches of technology. But up to now the practical role of light has been determined not by its energetic properties, but by entirely different ones. Therefore even the highest practical achievements of optics—its microscopes, rangefinders, photographic and motion-picture cameras—still could not change human life to the same extent as steam engines and dynamos have done.

Despite this limitation of the immediate technical prospects of optics, it is precisely the practical applications of certain

properties of light determined its development. The dimensions of light waves, very small in comparison with the scales familiar to us and, conversely, large relative to molecules and atoms, make it possible to obtain images of luminous and illuminated bodies, and to direct and concentrate light beams. This is the basis of so-called optical engineering. Until now it has been the principal practical driving force of the science of light.

The monstrously effective action of light on the eye determined another important practical branch of the science of light—illumination engineering, concerned with questions of lighting and sources of light.

Finally, the quantum, corpuscular properties of light, as we now know, explain the special character of photoelectric and photochemical processes and underlie a special branch of optics that may be called phototechnics.

Along the three broad roads of optical engineering, illumination engineering, and phototechnics, modern light technology is chiefly advancing. In addition, however, the applications of the science of light are practically innumerable in various methods of control that have found use in laboratories and in factories.

For the present state of optics, however, it is characteristic that not only its technical side, but also the development of scientific research proceed almost exclusively along the line of applying laws long since firmly established, concentrated in wave and geometrical theory and in the elementary versions of the quantum theory.

This applied, in the broad sense of the word, character of modern optics does not at all mean, however, that we possess a completed and wholly faultless theory of light. On the contrary, the imperfection of the existing theory is clearly felt by physicists. We shall have to return to the experimental side of this question in the second part of the communication and to clarify why the unsatisfactory state of the theory has practically no effect on the development of optics, technical and scientific, including spectroscopy, which constitutes the theme of the report by Academician D. S. Rozhdestvensky.

The Emergence of the Optical Institute

In the history of optics there have been two epochs of sharp acceleration in its development. From the moment when, in the sixteenth century, the telescope in Galileo’s hands proved to be a miraculous instrument of astronomy and navigation, optics acquired real rights to existence, and optical engineering arose. From that time the entire cultured world occupied itself with the manufacture and calculation of mirrors and optical glasses; on this soil grew the optics of Newton, and later Fresnel’s wave theory. From this impulse, given by Galileo and supported by practical demands, in the Petersburg Academy of Sciences in the eighteenth century there flourished the optics of Euler, Lomonosov, and Kulibin, which left us such remarka-

remarkable monuments as Euler’s three-volume Dioptrics, achromatic telescopes and microscopes. The fruits of Petersburg optics were reaped in Western Europe, while in Russia this artificially cultivated field of knowledge and technology proved to have no soil and dragged out an unnoticed existence almost until the era of the revolution. The second powerful impetus to the development of optics throughout the world was the World War. The practical importance of optical instruments—binoculars, rangefinders, sights, and so on—in the modern army became plainly evident. One of Germany’s important technical advantages during the war was its highly developed optics. Feverish attempts in France, England, and America to fill the optical gap did not soon lead to success. Soon after the end of the war it became clear everywhere that the only way to solve the difficult task was to create special research and educational optical institutes. Such institutes arose almost simultaneously in our country, in France, and in Italy. Our Optical Institute was founded earlier than the other institutes, already in December 1918. It was prompted by the war and brought into being by the October Revolution, which flared up during the war and existed during its first years under the conditions of civil war and intervention. Therefore one may say with full justification that our optics is the child of war and revolution. Optical institutes in our country and throughout the world became necessary because the realization of the construction of optical instruments proved to be a very far from simple matter, requiring not only knowledge of the physical foundations of the theory of light and workshops of precision mechanics; a school for calculating optical systems was needed, where, besides theory, art, skill, and great experience were required in still greater measure; the necessity became clear of being able to prepare various sorts of optical glass; delicate methods for testing optical instruments were required. It turned out that optics as a branch of technology had by no means yet emerged from under the guardianship of physics and chemistry. Many countries had to learn optics and, by way of long and difficult experience, to discover its delicate aspects.

For us this was especially difficult: before the revolution one sixth of the globe, with respect to optics, was tabula rasa. From time to time, in the Academy and in the universities, individual, sometimes brilliant, studies appeared on questions of physical optics, but there were almost no people capable of calculating optical systems, designing and inspecting optical instruments, and preparing optical glass. The very modest beginnings of lighting engineering huddled in the backyards of electrical engineering. Photographic technology likewise had neither personnel nor a base. To revive this desert, to master the technology of light, and on this basis to help the optical industry—this task guided the initiator of the Optical Institute, D. S. Rozhdestvensky, and to a significant degree determined the paths of development of the institute.

Structure of the Optical Institute

The Institute understood its purpose very broadly. Any optical problem, scientific or technical, that merits investigation can and even must be studied at the Institute. Optics must be covered practically in its full scope. This tendency was realized gradually, year by year, in the form of the Institute’s modern, rather complex structure, with its numerous sectors and laboratories. In each such highly independent scientific unit, a fairly narrow circle of optical problems is systematically studied. But taken as a whole, almost everything is encompassed. Represented here are the principal questions of scientific optics, concentrated mainly around the problem of the structure of matter; very diverse applications of physical optics, for example interferometry, applications of invisible infrared and ultraviolet radiation; optical engineering in the broadest sense, beginning with the calculation of optical systems, designs, and methods of control, and ending with details of the technology of manufacturing optical glass.

The illuminating engineering of the Optical Institute embraces various problems of subjective and objective photometry, calculations of natural and artificial illumination, and questions of long-range projectors and beacon lights. Photographic technology has developed into a large sector of scientific photography, cutting through the entire depth of the problems of this field, beginning with the profoundly theoretical question of the nature of the latent photographic image and extending to concrete technological questions of producing scales, color photographs and cinema, manufacturing photographic gelatin, standardizing emulsions, and so forth. The theoretical foundations of photochemistry are being laid step by step in the photochemical laboratory of the Institute. Diverse and practically very necessary questions of physiological optics—relating, for example, to color and stereoscopic vision—are also represented rather broadly in the Institute. If we add that in recent years fairly large experimental mechanical and optical workshops have developed at the Institute; that in the laboratory of astronomical optics the Institute manufactures the most difficult optical parts for astronomical observatories; that there is no optical instrument or part of one that the Institute would not be able to manufacture in its own laboratories or workshops, then the picture of the Institute’s “optical completeness” becomes sufficiently clear. Of course, this complexity and “completeness” of the Institute have been purchased at the cost of a very impressive scale, and at times also of insufficient development of certain important laboratories; and therefore the question of the expediency and necessity of this, undoubtedly cumbersome, structure is entirely appropriate. Some explanations are needed here, taking into account the particular features of optics in general and, in our country, in particular. The Institute arose amid the “optical desert” of which we have already had occasion to speak, and in the country, in the words of Prostakova, there was no

of those optical “carriers” that would make it possible “not to know geography.”

The Optical Institute had to know everything—or, more precisely, to learn everything. Optical engineering, for example, required subtle interference methods for determining refractive indices and dispersion, methods for the photometry of optical instruments, knowledge of physiological optics, and the ability to make photographic plates that fully utilized the resolving power of a good photographic objective. Thus, comprehensiveness arose of itself, naturally and organically; it was impossible to assign one or another sector to someone else. Not a single laboratory of the institute (I have in mind its tasks) arose artificially, “according to a scheme”; the reason for its organization was always the presence of certain blank spots in our optics that had to be filled. There is also another basis that allows the institute to be complex and comprehensive: the unity of optical methods in the most varied tasks. The methods are often difficult and very delicate, but they are highly general. The Michelson interferometer served the optician both for checking the foundations of the theory of relativity and for measuring aberrations of objectives and defects of polished surfaces.

The spectrograph, the spectrophotometer, the microscope, polarizing prisms, photometers—these instruments unite opticians of all specialties. Removed from the institute’s general organization, each of its laboratories would undoubtedly lose an extraordinary amount and might simply become powerless. The institute’s saturated optical atmosphere, on the contrary, leads to the deep penetration of physical, optical methods into areas that seemed very remote from such a possibility. It would scarcely have been possible, for example, to connect the phenomena of combinational scattering of light and elliptical polarization upon reflection at the Brewster angle with questions of the manufacture of optical glass in any way other than within the comprehensive setting of the Optical Institute.

We thus arrive at the conclusion that the institute’s comprehensiveness is inevitable and is its great advantage until such time, at least, as the country has new, sufficiently strong centers of optical research. Any attempt at a mechanical division of the large Optical Institute into specialized institutes would, in our opinion, be clearly harmful. The Institute is not the arithmetical sum of separate laboratories, but an organic whole whose significance is many times greater than such a sum.

On Methods of Work

If in the field of science it is always possible to build the new immediately at the existing world level, then in technology the difficult and lengthy path of mastering what has already been achieved is inevitable. This applies to optics to an especially strong degree. Optical engineering is very artisanal

both in its theoretical part, i.e., in the methods of calculating systems, and in its execution, which often requires enormous precision. A successful calculation is sometimes achieved only by the intuition of a very experienced computer; the exact embodiment of this calculation is the work of the hands of a master-artist. The technique of such a handicraft type is mastered with particular difficulty and slowness. This was the path that the Optical Institute had to follow, and with it the gradually developing industry. But fortunately, in many cases the institute managed not to repeat the road already traveled by Western optics and to find its own paths. This originality of methods served as a good school for the institute, developing in it a living theoretical initiative that determines its independence and distinctiveness. I can explain this only by examples.

A very instructive example is the manufacture of objectives for microscopes. In this matter the operation of centering and fastening the lenses is especially responsible and delicate. In Germany, according to old tradition, it is entrusted to especially experienced masters, and our industry, moving naturally along the line of least resistance, turned simply to the aid of these foreign masters. In this form the task would plainly have fallen outside the domain of technique, turning into an art. V. P. Linnik at the Optical Institute, however, succeeded in turning it into a technical task. The operation of centering by means of a simple optical device on the lathe was reduced to establishing the coincidence of two “bunnies” from the front and rear surfaces of the lens, accessible to an inexperienced person. Next, the metal mount, inside which the lens centered in the indicated manner, is being held, is turned on the same lathe chuck with the same setting. The usual barbarous method of fastening the front lens in the mount was replaced by V. P. Linnik with electrolytic copper plating. In this form the assembly of a micro-objective became a simple technical operation accessible to a low-skilled worker. Now V. P. Linnik’s method has entered production.

Another example is from an entirely different field. There has long existed a need for light filters for the ultraviolet region of the spectrum. To this day we do not know substances that would make it possible to isolate arbitrary, sufficiently narrow regions of ultraviolet rays. This is an old problem, of interest to the optician as well as to the chemist and biologist. At the Optical Institute E. M. Brumberg gave a new and very simple solution to this problem. Curiously, this method returns us to one forgotten experiment in Newton’s Opticks. The experiment consisted in a solar ray falling on a system of two prisms of total internal reflection. In the thin air layer between the two prisms the light beam (Fig. 1) underwent total internal reflection, while the long-wave part of the spectrum passed through. Brumberg’s method extends and generalizes Newton’s method. Rectangular glass prisms are replaced by quartz ones; the air between them is replaced by a pure liquid or solution, the dispers—

whose dispersion increases steeply in the ultraviolet part of the spectrum. Under such conditions a light beam incident on a system of two quartz prisms with a liquid layer between them undergoes total internal reflection at the boundary; to the eye such a system is completely opaque, while the ultraviolet part of the spectrum, beginning with a certain wavelength, passes through almost without attenuation. By changing the liquid, this boundary can be obtained in different parts of the ultraviolet spectrum. For example, a solution of common salt in water makes it possible, with this system, to isolate the region of the spectrum between 2000 and 2100 Å. The principle of this light filter is entirely different from that of ordinary absorption filters. In it, not absorption but dispersion is employed; therefore it does not depend, for example, on the thickness of the liquid layer. At the same time it does not have the temperature sensitivity characteristic of the Christiansen light filter, which is very complex in practical implementation.

Fig. 1.

The institute had to follow its own special path in setting up the production of optical glass at the Leningrad LenZOS plant. In this matter the usual trivial beginning was the purchase, still during the war, before the revolution, of a secret technological formula from the English firm Chance Brothers. Slow but very effective work at first, and then, under the pressure of failures and of the growing demand for optical glass, a sharp change in the technological process, a change that quadrupled its speed and improved the quality of the glass. From the moment when the plant and the institute, in the joint work of physicists, chemists, and engineers, set off on new, independent rails, there began an unbroken line of growth and improvement of Soviet optical glassmaking. Optical glass was taken up seriously and in a new way. The investigations of the physicists A. A. Lebedev and A. I. Stozharov first revealed the chemical essence of the process of annealing optical glass. It turned out that annealing corresponds to the chemical equilibrium of two molecular modifications. At temperatures above the annealing temperature one of these modifications is in excess; at temperatures below annealing, the other. The new viewpoint in practical application led to an improvement in the quality of the glass in the sense of its homogeneity and made possible the organization of serial production of critical optical parts. Ahead lies the next major task—the search for a new method of melting optical glass to replace the old method of melting in furnaces, associated with breaking the pot and a very small yield of usable material. The goal is a transition to melting by electric current. For this it is necessary

conduct long systematic studies on the catalysis of chemical reactions in the charge, making it possible to lower the temperature of subsequent melting, and on the electrical conductivity of glass. As a result of many years of work by Academician I. V. Grebenshchikov, the enormous role of chemical processes in the polishing of glass, as well as of metals, is becoming clear. On this basis new polishing methods are being created, distinguished by their speed and by the excellent quality of the polished surfaces. Gradually it has become possible, in general, to master the surface of optical glass. However paradoxical this may sound to an optical ear accustomed to Fresnel’s laws, it is possible, for example, within wide limits to change the reflection coefficient of glass without changing its mean refractive index. Here Fresnel is practically corrected by Newton, by interference phenomena in the surface film of the glass, which has optical constants different from those of the entire block. The character of such a film can be changed by chemical treatment. This result, apart from applications in laboratory practice, is important for increasing the effective aperture ratio of optical systems with a large number of reflecting surfaces.

I shall confine myself to just one more example to clarify the thesis concerning the originality of the institute’s technical paths. This example is on quite a different scale from the preceding one, but it may be of special interest to the physicist-optician. There exist many interference methods for determining aberrations and errors of optical systems, for example photographic objectives. Known, for example, is Twyman’s bulky and very expensive apparatus, which is in essence a large Michelson interferometer, where a regular undistorted wave is compared with a wave passed through the objective and distorted by aberrations and errors. V. P. Linnik proposed a new principle for the interferometric investigation of optical systems. A converging light beam from a point source, passing through the objective under study, falls on a semitransparent plate in which there is a very small aperture. This aperture serves as the center of elementary Huygens waves, free from imperfections or deformations in the incident wave. Through the remaining regions of the semitransparent plate passes a wave that has imprinted the defects of the objective. This wave and the ideal Huygens wave emerging from the small aperture interfere; the interference pattern is viewed in a telescope, and from the character of this pattern one can judge, by the usual method, the quality of the optics under investigation.

It is not difficult to see that the method described is only a special case of a new general interference method.

New, original technical paths are a characteristic feature of the work of the Optical Institute. This assertion is difficult to prove otherwise than by examples. I, of course, have been able to present only a few here. Meanwhile it is precisely this feature which, it seems to us, deserves special attention.

WHAT THE INSTITUTE HAS GIVEN TO TECHNOLOGY

Has the institute justified its task? I believe it has. At present we no longer have grounds to speak of an “optical tabula rasa.” The country has optics, opticians, an optical industry, and, in general, an optical culture in the broad sense of the word.

There is no doubt that the Optical Institute has played no small part in creating this culture. The institute has indisputable merits before industry, but it also has no less indisputable debts. The optico-mechanical industry of our country has now reached very great dimensions; it grew together with the institute and sometimes with substantial assistance from it. On the initiative of the institute, and through joint work by industry and the institute, the country now produces its own optical glass—good glass, of all grades and in any quantity. Suffice it to say that from our glass we manufacture enormous meter-sized disks for astronomical objectives; that we have all types of colored optical glass; that in the technology of optical glass we have, in some areas, undoubtedly outstripped foreign countries. The institute has, unquestionably, helped in the development of computation in the country. Scarcely less than a good half of the optical parts of optical instruments in our industry have been calculated by the institute. The testing instruments of our optical plants, the instruments by which products or individual phases of assembly are examined, have in many cases been devised and realized at the institute. The various instruments invented, designed, and produced in prototypes by the institute number in the many dozens.

The manufacture of first-class optical parts for astronomical instruments has been advanced extraordinarily. Craftsmanship, combined with new methods of testing, has made it possible in some cases to obtain entirely unique specimens, far surpassing existing models in perfection.

Within the walls of the institute Soviet crystal optics arose, with new technological methods and new types of polarizing prisms. The most complex forms of prisms can now be manufactured better and more cheaply than abroad. These are the institute’s quite tangible achievements, which, if desired, can also be expressed in rubles.

But alongside this, the institute did not in time take account of one important circumstance—the growth of the optico-mechanical industry and its relatively enormous scale. For such an industry, the handicraft methods founded on the enormous skill of the worker, which by tradition are cultivated in the West, are often unacceptable.

For our industry, entirely new, as yet unseen technological processes are needed, adapted to the low-skilled worker; in this respect the institute has helped industry little and could hardly have truly helped, being basically a physical institute with leading staff—physicists.

Precision mechanics, which in both volume and difficulty constitutes no less than 60% of optico-mechanical production, lay almost entirely outside the Institute’s competence, and this, of course, affected the quality of production. Another defect, for which both industry and the Institute are simultaneously to blame, is the slight concern shown for leading engineering personnel for industry. Allow me here to make a small digression.

Last summer I had occasion to visit optical institutes in Paris and in Florence. An examination of these interesting institutions, which have acquired great importance in their own countries, was very instructive for me above all as a living scale by which to judge our Optical Institute. Both the French and the Italian institutes are still very small in comparison with ours; their activity is almost confined to optotechnics, only slightly touching upon certain questions of physiological optics needed for optotechnics. But the Paris Optical Institute is closely connected with a separate Institute of Scientific Photography and with the physics and astronomy of the Sorbonne, while the National Optical Institute in Florence this year is tripling its size, developing departments of illuminating engineering, photometry, and photography. The latter institute is, moreover, linked in the closest possible way with the Florentine Institute of Borosilicates, which is essentially a small experimental optical-glass factory, though, to be sure, it is still only taking its first steps in this field. The immediate neighbor of the Florentine Optical Institute—the Arcetri Astronomical Observatory, situated nearby on the same hill—is connected with the Optical Institute by various ties of an instrumental character. In both institutes, therefore, one must note the same tendency toward optical fullness and comprehensiveness that has long since been realized in our country; they give the impression of shoots which must develop into forms analogous to the Leningrad Optical Institute.

It is very difficult to give a comparative assessment of the scientific and technical output of the three institutes. There is no doubt that all three institutions have generally recognized results of great significance. In Paris I had occasion to see work on interferometry and optical glass based on the results of work by our Institute, just as we often make use of the methods and results of Parisian and Florentine opticians. Quantitatively, in accordance with our scale, we of course produce more. But the small institutes in Paris and in Florence have one very important advantage over us: they unquestionably exert a greater influence on industry than we do. Paradoxical as it may seem, the small Florentine institute, belonging to the Ministry of Public Education, in a country where optico-mechanical factories are private competing enterprises, plays the role of scientific and technical conductor for the entire industry—a very effective role, as I was able to convince myself when visiting the principal Italian optical ...

plants. This is accomplished above all by the fact that the optical institutes both in Florence and in Paris are just as much research institutions as they are educational institutions. In our terminology these institutes have an extensive postgraduate program, but a peculiar kind of postgraduate program, intended not for scientific work but for the plants. Physicists and engineers graduating from universities and higher technical schools study for two years in optical institutes; they are trained in research work, attend special courses, and undergo computational and laboratory practice. This two-year stay in the atmosphere of an optical institute produces a distinctive type of optical engineer, combining the technical concreteness of the engineer with the breadth and flexibility of the physicist.

I had occasion to meet several such former postgraduates of the optical institutes in Florence and in Paris. As a rule, they occupy the posts of technical directors of plants or play a leading role in the shops, in laboratories, in the computational and design bureaus of plants. They proved to be reformers of the optical industry, which in Italy and France has achieved considerable flourishing in recent years. Through their former students the optical institutes also keep in their hands the scientific and technical threads of the life of the plants, carry out there their proposals and their methods, direct the plant laboratories, and remain in constant living contact with the plants.

It seems to me that in this respect we should learn from the West: industry must meet halfway, sending its engineers to the institute for a long time for training. This is one of the best ways of linking science and industry. If we could add to the impressive list of works for industry a continuous conveyor of engineering personnel, we would undoubtedly help industry much more than we have done up to now.

In speaking of the institute’s technical work, I had chiefly in mind the optical-mechanical industry. Indeed, the institute’s connections with it are especially old and strong, and recently, by order of Comrade Ordzhonikidze, the institute has in general been transferred into the system of the VOOMP. We hope that this new phase in the life of the institute will make it possible, to an even greater degree than before, to embody the institute’s creative thought in real work, very necessary for our country.

In my characterization of the technical role of the institute I was compelled to confine myself to a schematic outline instead of a concrete exposition: I have said nothing at all about the institute’s work in illumination engineering, photography, color science, etc. Here too the institute has often acted as a pioneer and a practical helper in questions of lighting, the organization of the photochemical industry, the glass industry, and other practical branches. But I am afraid to burden your attention with these details; I can only refer to the collection

materials from our reports, published for the session, where more concrete information may be found, and I shall turn to the second part of my communication.

THE NATURE OF LIGHT

The Modern Theory of Light

Our session is devoted to problems of physics. Among physicists there exist different points of view as to what should be understood by the words “physical problem.” One often hears the opinion that in physics today, in essence, there remains only one fundamentally unsolved problem: the structure of the atomic nucleus and the questions of relativistic quantum mechanics and electrodynamics connected with it. Modern optics, and the Optical Institute in particular, from this point of view, in the main, along its principal lines, does not deal with “physical problems.” As I have already said at the beginning, the work of an optician in almost all cases comes down to the application of firmly established principles of physical optics to concrete scientific and technical tasks.

It is hardly worth arguing, however, about words and definitions and about the degree of importance connected with the word “problem.” What is beyond dispute is that the tasks before the institute are difficult and very necessary, and in most cases their solution can be provided only by a physicist. From this broader point of view we are also presenting the work of the Optical Institute for discussion by the session. There is, however, one sector of the front in optics—practically a very small and inconspicuous one, but for the physicist himself fundamental and deserving the title of a “physical problem” even in the aforementioned narrow sense as something still more principled, unclear, and unresolved. I have in mind the centuries-old question of the nature of light. M. V. Lomonosov spoke from the academic rostrum on this problem exactly 180 years ago in his “Discourse on the Origin of Light”; this same question was the subject of an academic competition in 1807 and was raised many more times in the Academy, but we have to return to it even now as to an important and unresolved theme. As a result of a truly fierce struggle between the corpuscular and wave views on the essence of light, physics, under the pressure of facts, took theoretical positions on this question which we are unable to explain by means of a clear mechanical picture or model. In the new theory the conception of waves and particles has merged in a dialectical synthesis inaccessible to mechanical interpretation: the motion of light is neither wave-like, nor corpuscular, nor a mechanical superposition of the one and the other. The essence of light is such that, within certain limits, both wave and corpuscular properties can easily be traced in it, but our model thinking, which has grown up on the material of everyday images and impressions, is not in a position to form a concrete model of the phenomenon, understandable in the usual sense of the word. The formal mathematical theory of light, based-

...based on the principles of quantum mechanics and developed by Dirac, is constructed according to the following scheme. In a closed space containing matter and light, everything together is regarded as a single system, whose energy may be divided into three parts: the energy of matter, the energy of the electromagnetic field, and the energy of the coupling between the two. For this system the classical Hamilton equations are set up. Then a procedure is repeated analogous to the transition from the equations of Hamiltonian mechanics to the wave equation, as used by Schrödinger in deriving his famous fundamental equation of quantum mechanics.

In other words, the classical “ray” Hamilton equation is arbitrarily generalized into a quantum wave equation. The resulting outcome replaces the equation of the classical theory and encompasses, in the main, all the known properties of light, in accordance with the fact that both wave and quantum laws were used in the derivation. This device can hardly be called a theory in the usual sense of the word, i.e. an unambiguous derivation from certain more general and fully justified principles; rather, what we have before us is a heuristic path for finding a new principle that agrees well with experimental data. However, even from the formal side Dirac’s theory is connected with difficulties, still not overcome, concerning the structure of the electron—for the self-energy of the electron infinitely large values are obtained.

The new theory leaves practically in force all the conclusions of the wave theory of light concerning the laws of propagation, if one abstracts from processes accompanied by a change in wavelength, i.e. by a loss of energy and momentum (for example, Compton scattering and combination scattering), and has in mind sufficiently powerful light fluxes corresponding to a large number of photons. Therefore, optotechnics and lighting engineering can develop calmly, without being affected at all by the difficulties of the theory of light.

On the other hand, for understanding energy relations in photochemical and photoelectric processes, an elementary corpuscular conception of light as a stream of light quanta, photons, is sufficient. Therefore the difficulties of the theory of light also had little influence on the development of phototechnics.

The theory threw a formal bridge between the wave and quantum properties of light. However, the imperfection of the theory itself and its extreme abstractness require experimental support. This part of my report is devoted chiefly to the experimental investigation of light under such conditions where, theoretically, one should expect sharp deviations from both wave and corpuscular laws.

Visual observation of light fluctuations

A necessary condition for experiments of this kind is the extremely weak intensity of the light source under study. The theoretical ex...

deviations from the classical laws of propagation are possible only with a small number of photons falling on the receptive surface. Consequently, an extremely sensitive receiver for light is necessary. If one speaks of visible light, then the solution, one may say unequivocally, is that only the eye can be such a receiver. In fact, its usual competitor—the photographic plate—in the present case proves powerless; enormous exposures, measured in days, are needed in order to obtain on the plate traces of an image at the required intensities. At the same time the statistical deviations from the classical laws of propagation required by the theory must be averaged out and become imperceptible. Indeed, numerous experiments on the interference of light, carried out by the photographic method during the last 20 years, have always only confirmed the conclusions of the classical wave theory. The second competitor of the eye, a more recent one—the photoelectric counter—has insufficient sensitivity even in the ultraviolet spectrum, not to mention the visible region. Meanwhile the astonishing sensitivity of the human eye, adapted to darkness, as a simple calculation shows, is theoretically quite sufficient for observing quantum fluctuations in the propagation of light. This sensitivity of the eye, combined with the presence of a sharp threshold of visual sensation, made it possible to hope for carrying out a test of the conclusions of the modern theory of light simply with the aid of the eye. Some six years ago, guided by these considerations, I proposed to V. I. Fedorova and S. V. Kravkov at the Institute of P. P. Lazarev that they carry out comparative measurements of statistical deviations in establishing the threshold of visual sensation in different regions of the spectrum. Owing to the preferential sensitivity of the adapted eye to the blue-green region of the spectrum, it could be expected that the statistical deviations in this region, corresponding to a small number of photons, would be especially large. The results confirmed the expectation, but owing to unaccounted physiological factors they could not be considered definite. In 1932, Czerny and Barnes in Berlin, on the basis of the same ideas, carried out several qualitative experiments, though still less convincing for the same reason—the failure to take into account unavoidable physiological factors.

In the same year, 1932, we resumed experiments on the observation of light fluctuations at the Optical Institute on the basis of a new methodology, which I shall set forth below. The first report on the results was made by me in the Physico-Mathematical Group of the Academy in 1933. Our experiments continue to the present time, and today I can supplement the first report with a number of new data.

The essence of our method of observation amounts to the following. There is a small luminous area, which I shall hereafter call simply a luminous point, although its dimensions must be such that the image on the retina of the eye covers a sufficiently large number of receptive elements—rods. This point is observed by the eye peripherally, in other words,

the eye is fixed on another, so-called fixation point, situated at some angle to the first. The image of the radiating point is obtained, consequently, not at the center of the eye but at the periphery, where the sensitivity of the retina, for example for the blue-green part of the spectrum, is considerably greater than at the center. With a high brightness of the point and a constant regime maintaining this brightness, we observe an unchanging picture of a luminous point of constant brightness. Suppose now that the brightness of the point under investigation has been weakened so much that we have come close to the threshold of visual sensation of the eye. If this threshold corresponds to a small number of photons per second, then, on the basis of the new theory of light, sharp oscillations, fluctuations of brightness, should be observed, since the photons in the light flux can be distributed only in a statistically disordered manner. But if the number of photons is less than the number corresponding to the threshold value, then no visual sensation arises, and thus our point should appear to the eye not as a constant source, but as a flickering one. In such a simple form, however, the experiment cannot be carried out, because the eye has the property of retaining a visual impression for several tenths of a second, and consequently the objective fluctuations are averaged, smoothed out, and cease to be noticeable. To eliminate this difficulty we place between the eye and the point a rotating sector with a cutout, which exposes the point to the eye for 0.1 sec. In this way the practically instantaneous state of brightness of the source is observed. After weakening the mean brightness of the source to the threshold value, it should turn out—if, of course, our theoretical expectations are correct—that each passage of the cutout past the luminous point will correspond to a visible flash.

Qualitative observations of this type, made by many persons in our laboratory, in all cases confirmed the expectation: beginning with a certain brightness of the source, omissions are observed when the cutout passes, i.e. absence of flashes—and they are the more frequent the more the brightness is weakened. The dependence of the frequency of flashes on the mean brightness of the source makes it possible to pass from qualitative observations to quantitative measurements.

Let \(n_0\) denote the number of monochromatic quanta exactly corresponding to the threshold of visual sensation from an irritation lasting 0.1 sec. Let \(n\) be the mean number of photons reaching the retina during the time of the flash; denote the ratio \(\dfrac{n}{n_0}\) by \(x\). Then, using probability theory, one can derive the following expression for the probability of observing a flash:

\[ w=\frac{1}{2}-\frac{1}{2}\sqrt{\frac{n}{2}}\,\frac{1-x}{\sqrt{x}}. \tag{1} \]

This formula is approximate and is valid only for the following limits of variation of the probability:

\[ 0.15<w<0.8 \tag{2} \]

and under the condition that \(n_0\) is considerably greater than unity. Numerical calculation shows, however, that formula (1) is practically valid for any values of \(n_0\), down to \(n_0 = 1\), provided only that the limits (2) are observed.

Formula (1) also serves as the basis of our quantitative method. If one varies the mean value of the luminous flux, i.e. the quantity \(n\), then, within the indicated limits of variation, there must exist a linear dependence of the probability on the expression \(\dfrac{1-x}{\sqrt{x}}\).

To obtain the absolute value of \(x\), one may use that property of formula (1) that for \(x = 1\), i.e. \(n = n_0\), the probability \(w\) must acquire the value \(1/2\). Knowing the relative values of \(x\), which is experimentally very simple to accomplish, and finding the relative value of \(x\) corresponding to the probability \(1/2\), one can determine the absolute values of \(x\). The slope of the straight line connecting the probability \(w\) and the quantity \(\dfrac{1-x}{\sqrt{x}}\) gives directly the number of photons corresponding to the threshold:

\[ \frac{dw}{d\left(\dfrac{1-x}{\sqrt{x}}\right)} = -\frac{1}{2}\sqrt{\frac{n_0}{2}}. \]

I shall allow myself to omit the description of the details of the experimental apparatus, noting only that the probability is determined in the following way.

Each revolution of the disk is automatically marked on one edge of the paper tape of the chronograph. When the observer notices a flash as the cutout in the disk passes, he presses by hand a key that actuates the second pen of the chronograph—on the other edge of the paper tape a second mark appears. The ratio of the number of marks on this edge of the tape to the number of marks on the other edge, which records the number of revolutions of the disk, gives the probability \(w\). I shall also note that observations of fluctuations require considerable preliminary training of the eye. In our laboratory these measurements were carried out by four observers.

Up to the present time, hundreds of measurements have been made which have confirmed the linear dependence between the probability of flashes and the quantity \(\dfrac{1-x}{\sqrt{x}}\) required by the theory. In Fig. 2 examples of such straight lines are given. From the slope of the straight line, as I have already said, the number of photons corresponding to the threshold is determined. This number \(n_0\) must vary, as is known from physiological data, depending on the angular distance between the center of the retina and the place of observation on the retina, on the spectral composition of the emitted light, and on the individual properties of the observer’s eye. I shall dwell first of all on the spectral dependence of the fluctuations, since its study gives a decisive argument in favor of the physical, quantum explanation of the phenomenon.

Physiologists in physics have repeatedly measured the so-called “summary” curve of the spectral sensitivity of the eye. In Fig. 3 the data of Abney and Watson are given, and in Fig. 4—the data of Hecht; moreover, in Fig. 3 the logarithms of the minimal values of the energy producing a visual sensation at threshold are given. These curves were obtained by ordinary energy observations. In our method no measurements of energy are made; we count only fluctuations and from them, in the manner already explained, derive the number of photons corresponding to the threshold; the product of \(n_0\) by the photon energy \(h\nu\) gives, obviously, the energy value of the threshold for the given wavelength of light. Our data thus have an absolute value, which may, for example, be expressed in ergs. Unfortunately, the physiological data are only relative; therefore one can consider only the relation of the quantities obtained by us to these physiological measurements.

Fig. 2.

Let us proceed to the analysis of the results. In summary Table 1 are given the absolute values of the minimum energy necessary for obtaining a visual sensation for various wavelengths. These values are calculated from fluctuations and refer to the observations of K. B. Panshin. In these experiments the distance between the center of the eye and the observed place of the retina was about \(10^\circ\). The maximum of sensitivity lies in the region of \(5100\ \text{Å}\). Under these conditions

Fig. 3.

(Visible labels in the graph: “Brumberg,” “Panishin”; other curve labels are [[unclear: handwritten abbreviations]]. The horizontal axis is marked from 420 to 630 mμ.)

Fig. 4.

(The horizontal axis is marked from 360 to 660 mμ.)

the eye of the indicated observer detected 8 more photons \(h\nu\). On the logarithmic graph of Abney and Watson, for comparison, the values obtained by the fluctuation method by K. B. Panishin and E. M. Brumberg are plotted. The vertical displacements of the curves relative to one another have no significance, since the data of Abney and Watson

relative and different for different observers. From the comparison of the energy and fluctuation curves it is clear that both have the same course. Substantial discrepancies are noticeable only in the orange-red part of the spectrum.

In this region the measurements are in general especially difficult and strained. The sensitivity of the retina in this case rapidly decreases from

TABLE 1

Measurements by K. B. Panshin

Fig. 5.

the center to the periphery, and it may be supposed that even in the presence of a fixation point the eye tends to turn toward the place of greater sensitivity. In addition, the experimental conditions are undoubtedly made somewhat more complicated by the presence in the field of view of a fixation point of the same color as the radiation being studied. Finally, the values obtained in the orange-red region cannot claim great accuracy because of insufficient monochromatization by means of a single dispersion. Even insignificant admixtures, for example of green rays, in this case can strongly distort the result.

The agreement obtained by us between the energy and fluctuation spectral curves may be regarded, as we think, as a decisive argument in favor of the physical, quantum interpretation of the observed fluctuations. In this connection it is of interest to note several episodic observations by K. B. Panshin which, unfortunately, were repeated an insufficient number of times, at considerable peripheral angles. Orienting experiments in the region of maximum sensitivity (5000 Å) showed that the sensitivity of the retina

continuously increases up to about \(26^\circ\). But, beginning from here, the sensitivity falls with a further increase of the peripheral angle. Thus, for this part of the spectrum there exists a maximum of retinal sensitivity. In the region of this maximum fluctuation measurements were made; their results are given graphically in Fig. 5. As always, a straight line is obtained, but with a very small slope. Both the exact and the approximate calculation give in this case a value close to 1 photon (1.4), i.e. the value of the maximum physically conceivable sensitivity. Here we have one more extremely convincing argument for the physical nature of the phenomenon.

The conclusions to which the study of spectral fluctuation curves leads us, however, are not limited to this. One cannot expect complete coincidence of the fluctuation and energy curves in those regions of the spectrum where the principal media in front of the retina possess strong absorption. The fluctuation method, as is quite evident from the derivation of the formulas on which it is based, makes it possible to determine only the sensitivity of the final receptive system, i.e. of the retina, whereas in the energy methods the sensitivity of the eye as a whole is determined.

In this connection, observations in the ultraviolet region of the spectrum are very interesting. It has long been known that the eye can see ultraviolet rays, but the sensitivity of the eye here is very small. We made fluctuation measurements in the region of the ultraviolet spectrum nearest to the visible end.

As in comparison with the data of Abney and Watson, we obtain satisfactory agreement of the curves in the yellow, green, and blue parts. In the extreme violet and ultraviolet regions, however, a sharp divergence is found. The fluctuation sensitivity curve, i.e. the curve of retinal sensitivity, here rises sharply and, for one of the observers, soon reaches a maximum. This unexpected result can, however, be explained qualitatively if one takes into account the data on the extremely strong absorption of ultraviolet rays in the crystalline lens of the eye.

According to physiological and medical data, the absorption of the crystalline lens increases extremely sharply, beginning approximately at \(4000\ \text{Å}\); quantitative measurements, unfortunately, are lacking. From this it obviously follows that the sensitivity of the retina in the ultraviolet region must probably be tens of times greater than the sensitivity of the eye as a whole. In support of this one may also cite Widmark’s old experiments, showing that in patients in whom the crystalline lens had been removed during an operation, the boundary of visibility of the spectrum shifts at least from \(4000\ \text{Å}\) to \(3100\ \text{Å}\), which is in complete agreement with our observations. It may therefore be asserted that the eye does not see, or sees very poorly, ultraviolet rays only because, on the path between

between the light and the retina there is a very strong light filter—the crystalline lens. It therefore performs in the eye a dual function—of an optical lens and of a light filter.

The totality of the facts obtained, namely:

1) the linear dependence, required by theory, of the probability of fluctuations \(w\) on the quantity \(\dfrac{1-x}{\sqrt{x}}\);

2) the correct magnitude of the slope of these straight lines, which, when combined with energy measurements, makes it possible to determine visually the value of the quantum constant \(h\);

3) the correct form of the twilight spectral curve obtained by the fluctuation method;

4) the detection, by fluctuations, of the sensitivity of the retina in the ultraviolet region,

seems to us to resolve quite unambiguously the question of the physical quantum nature of the observed fluctuations. We are continuing these investigations, since, apart from their physical interest, they are important for the study of the properties of the eye, constituting a new method of physiological optics.

I shall allow myself to leave aside these physiological questions and return to the main theme—the problem of light.

The Quantum Character of Interference Phenomena

Possessing a visual method for measuring light fluctuations, we were able to proceed to testing the most paradoxical consequence of the modern theory of light. From the point of view of this theory, the strongest argument for the wave view—the phenomenon of interference—must have a quantum character; it is not a collection of light rays that interferes, but, as the theoreticians put it, a ray must interfere “with itself.” Of course, the paradoxical nature of this assertion is connected with the fact that we operate with imprecise figurative concepts of “ray” and “interference,” which in the theory are replaced by a single whole inaccessible to pictorial modeling.

I shall not dwell on the details of our interference experiments. Their difference from the described measurements of fluctuations consisted in the fact that, instead of one luminous point, we observed two adjacent coherent points obtained by splitting the primary ray with a Fresnel biprism. In wave theory the phase relation in these coherent rays, converging at one point, determines the result of the interference. Indeed, the observation of such rays at very low intensities showed that both rays fluctuate in a completely disordered manner and independently of one another. Random coincidences of flashes at both points may be extremely rare and obey ordinary statistics, as we verified by direct measurements. If the mean brightness of the two points is the same, then for coincidence

fluctuations, a linear relation is obtained only by substituting the square root of the probability instead of the probability, which follows directly from the theory.

We also made observations directly in the interference field. These observations prove quite clearly the statistical and, at the same time, to a certain degree wave-like character of the phenomenon. The bright interference fringes flash and fluctuate quite independently of one another, but the dark fringes remain dark all the time. The inadequacy of both the wave and the corpuscular interpretation is here literally evident. Light simultaneously fluctuates and interferes, but the interference of each ray is independent of the others.

The experiments described seem to confirm with complete persuasiveness the conclusions of the quantum theory of light. But, as always, no single experiment can be an unappealable experimentum crucis. An experiment may with certainty refute a theory, but it is not in a position to confirm it with the same certainty. Recently theoretical physics has been undergoing another crisis, and doubts have arisen as to the correctness of the existing quantum electrodynamics and, consequently, of the theory of light. I have already pointed out the imperfection of this theory and its internal contradictions. It is beyond doubt that wave mechanics explains atomic physical phenomena very accurately in the case of velocities not very large in comparison with the velocity of light; but when this condition is violated, the correctness of its conclusions becomes doubtful. Meanwhile optics, light phenomena, belong to this domain. One of the experimental blows that have shaken the foundations of quantum electrodynamics was the recently published experiments of Shankland. These measurements were a check on the famous Bothe and Geiger experiment, which had seemed to prove that the elementary processes of photon scattering and electron recoil in atoms occur simultaneously in accordance with the laws of conservation of energy and momentum. Shankland, repeating this experiment in a modified arrangement and with harder rays, refutes the conclusion of Bothe and Geiger and finds that photon scattering and electron recoil occur independently and are not linked in time. On the basis of the published data it is still difficult to judge the degree of conclusiveness of these experiments, but it is highly characteristic of the state of contemporary theoretical physics that this, perhaps fictitious, blow forces the theorists quickly to clear the positions they occupy.

To save the situation, Dirac proposes returning to the old positions of the theory of Bohr, Kramers, and Slater, which were abandoned precisely under the pressure of the results of the Bothe and Geiger experiment. This theory is attractive in that it returns to the old “good” electromagnetic theory of light, transferring all quantum difficulties to matter, to the atom, in which abrupt quantum changes of states in elementary acts must occur. However, this agreeable aspect is obtained at the cost of renouncing the conservation of energy and

momentum. From this point of view, the conservation laws acquire a strictly statistical meaning.

One can hardly see any fundamental, still less philosophical, impossibility in attempts to abandon conservation laws for elementary processes. The concepts of elementary particles and processes are in themselves, undoubtedly, an abstraction—very useful and necessary, but hardly completely correct. The world is continuous, and changes in its parts can and probably always must find an echo in the surroundings. From this point of view, violations of conservation laws in elementary processes may prove to be just as permissible as the disappearance of energy in a cooling stove. Moreover, the typically mechanical concepts of energy and momentum may turn out to be replaced by other, more general concepts. A physicist feels great embarrassment at attempts to abandon conservation laws in elementary processes mainly because he knows of no other laws that should be introduced in their place.

Let us therefore suppose for a moment that Dirac is right and that it is better to return to Bohr—Kramers and Slater and, consequently, to the old electromagnetic theory of light. In that case photons turn out to be a fiction; they do not exist: there are old electromagnetic waves and quantum fluctuations in the receiving apparatus. From this point of view, our measurements turn out to be measurements of quantum fluctuations in the retina of the eye. A ray interferes “with itself” not because this is so in reality, but because at the point of interference in matter the law of conservation of energy is statistically violated now in one direction, now in the other. From this point of view, our experiments are not yet in themselves capable of deciding the question of the reality of photons; they only confirm the quantum character of the actions of light. True, this time, in order to get away from the photon, one has to abandon the conservation laws.

Properties of Photons

Can one prove by some other method the objective existence of photons and decide the dispute between this last and the penultimate word in the theory of light? Theory has hitherto pointed chiefly to only two specific properties of photons—their energy \(h\nu\) and momentum \(\frac{h\nu}{c}\). These properties have been confirmed, but they can also be explained without photons if the quantum discontinuities are referred to matter, assuming a violation of the conservation laws in elementary processes. Are there other properties of photons, indicated by theory, that would make it possible to resolve in principle the emerging alternative? Einstein’s analysis of the laws of equilibrium temperature radiation, carried out in 1909, showed that the statistics obeyed by the “photon gas” inside the cavity of a black body are, generally speaking, different from the classical one. At low temperatures there is

classical statistics, gradually passing at very high temperatures into the so-called Bose—Einstein statistics. This should manifest itself in the fact that the magnitude of the fluctuations, for example, at \(30\,000^\circ\) should be 1.3 times greater than for a temperature of \(3000^\circ\). Such a difference in the statistical laws of light fluctuations for different radiators acquires special interest, since fluctuations are measurable. Indeed, if it is possible to transfer Einstein’s conclusions not only to the interior of the cavity of a black body, but also to the beam of light emerging from it, then by measuring fluctuations in monochromatized light it is in principle possible to determine the temperature of the source. An experiment of this kind is impracticable under laboratory conditions, but one may resort to astronomical objects—stars of various temperatures. In doing so, of course, the influence of the twinkling of the stars must be eliminated, which, apparently, is feasible. Such an experiment, in the event of a positive result, would not, however, resolve the question of the existence of photons; it would prove only that the observed fluctuations are in part of external origin from the source, and are not entirely due to fluctuations of the receiver. Using the visual method of measuring fluctuations, we made a comparison of the fluctuations of a thermal and a luminescent radiator (an incandescent lamp and the light of a fluorescing solution) under conditions of equal intensity and approximately identical spectral composition, which was achieved by the use of light filters and a monochromator. The apparatus was modified in such a way that the flashes of luminescent and thermal radiation followed in succession one after another, and the observer had no possibility of distinguishing one flash from the other. The separation of the marks on the chronograph tape for the one and for the other case was carried out automatically with the aid of the corresponding electrical connection of the rotating disk, the chronograph pens, and the key. Numerous experiments revealed no noticeable difference in the law of fluctuations for the thermal and the luminescent radiator. Unfortunately, we have no theory of the fluctuations of nonequilibrium radiation by fluorescence; therefore the result found has, for the time being, only empirical significance.

Dirac’s theory of light leads to the conclusion that there exists an effect of self-scattering of light at the intersection of two light beams. The phenomenon is analogous to what, in Lomonosov’s words, is “interference in rays,” which should be observed if light has a corpuscular nature. If such a phenomenon existed, we would have before us the rarest case of the action of light not on matter, but on light itself, and would obtain direct proof of the quantum-photon nature of light. According to the calculations of Euler and Kockel, however, the effect should be extraordinarily small: the effective photon cross section for visible light can be only approximately \(10^{-70}\ \text{cm}^2\). To observe such scattering under laboratory conditions is impossible. More favorable is the space around the sun, where extraordinarily powerful beams of light intersect,

and, moreover, the effect of their mutual interaction is concealed from the terrestrial observer. If we had any grounds for ascribing the inner solar corona to the effect of self-scattering of light, then the mean effective photon glow for visible light would be \(10^{-30}—10^{-40}\ \mathrm{cm}^{2}\). Even this quantity, however, is too small for it to make sense to undertake experiments under laboratory conditions.

Thus there is hardly any possibility of proving the existence of photons in space by means of a direct optical experiment. Only a careful verification of the fulfillment, or, conversely, the violation, of the conservation laws in elementary acts of interaction between light and matter can decide the question of the real existence of photons.

I have come to the end of my report, having outlined in very schematic terms the extremes between which the work of the Optical Institute proceeds: large-scale technology and a very small area in which the experimental study of the nature of light is concentrated, and where the Institute seems to depart far from its basic practical tasks. Yet scientific research is rarely practically fruitless; in the present case, experiments with fluctuations of light gave rise to a new method for measuring extremely weak brightnesses, inaccessible to other methods. The numerous applications of the threshold photometric method in recent years have made it possible to discover new phenomena—for example, the remarkable case of the visible glow of all pure liquids under the action of \(\gamma\)-rays, the laws of the intrinsic glow of the sky, and the fluorescence of solutions of platinum-cyanide salts. The new method has made it possible to develop quantitative, extremely sensitive luminescent analysis of ozone and oxygen, and to clarify the laws governing the decay of phosphorescent substances, etc. An unbroken line from profoundly scientific to concrete technical problems, linking the riddles of quantum electrodynamics with the difficulties in the technology of the fireclay pot in which optical glass is melted—this line has been, and, in our opinion, should remain, the axis of the Optical Institute. We firmly hope that it will be preserved in the new phase of the Institute’s development, when it has entered the system of the Unified Optical-Mechanical Industry.