The Invention and Development of the Radiosonde with A Catalog of Upper-Atmospheric Telemetering Probes in the National Museum of American History, Smithsonian Institution JOHN L. DUBOIS, ROBERT P. MULTHAUF, and CHARLES A. ZIEGLER SMITHSONIAN INSTITUTION PRESS SERIES PUBLICATIONS OF THE SMITHSONIAN INSTITUTION Emphasis upon publication as a means of "diffusing knowledge" was expressed by the first Secretary of the Smithsonian. In his formal plan for the Institution, Joseph Henry outlined a program that included the following statement: "It is proposed to publish a series of reports, giving an account of the new discoveries in science, and of the changes made from year to year in all branches of knowledge." This theme of basic research has been adhered to through the years by thousands of titles issued in series publications under the Smithsonian imprint, commencing with Smithsonian Contributions to Knowledge in 1848 and continuing with the following active series: Smithsonian Contributions to Anthropology Smithsonian Contributions to Botany Smithsonian Contributions to the Earth Sciences Smithsonian Contributions to the Marine Sciences Smithsonian Contributions to Paleobiology Smithsonian Contributions to Zoology Smithsonian Folklife Studies Smithsonian Studies in Air and Space Smithsonian Studies in History and Technology In these series, the Institution publishes small papers and full-scale monographs that report the research and collections of its various museums and bureaux or of professional colleagues in the world of science and scholarship. The publications are distributed by mailing lists to libraries, universities, and similar institutions throughout the world. Papers or monographs submitted for series publication are received by the Smithsonian Institution Press, subject to its own review for format and style, only through departments of the various Smithsonian museums or bureaux, where the manuscripts are given substantive review. Press requirements for manuscript and art preparation are outlined on the inside back cover. Lawrence M. Small Secretary Smithsonian Institution SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY • NUMBER 53 The Invention and Development of the Radiosonde, with a Catalog of Upper-Atmospheric Telemetering Probes in the National Museum of American History, Smithsonian Institution John L. DuBois, Robert P. Multhauf, and Charles A. Ziegler Smithsonian Institution Press Washington, D.C. 2002 ABSTRACT DuBois, John L., Robert P. Multhauf, and Charles A. Ziegler. The Invention and Develop- ment of the Radiosonde, with a Catalog of Upper-Atmospheric Telemetering Probes in the National Museum of American History, Smithsonian Institution. Smithsonian Studies in His- tory and Technology, number 53, 78 pages, 60 figures, 2002.—From a historical perspective, the radiosonde is one of the more significant technological innovations of the twentieth cen- tury, not only because its widespread use greatly enhanced the accuracy of weather forecasting, but also because some features of its basic design became the foundation of all modern analog telemetry systems. This study examines the way in which advances in the technology of non- telemetering balloonsondes and radio in the nineteenth and twentieth centuries culminated in the invention of the radiosonde in 1929. The subsequent development of radiosondes in Europe and the United States from 1929 to 1940 is traced in detail, when the basic design of this instru- ment achieved its modern form. An overview of significant modifications in radiosonde design after 1940 also is provided because the instruments have remained an essential meteorological tool in the twenty-first century. This monograph also includes a catalog of radiosondes in the Smithsonian Institution's National Museum of American History. Photographs of instruments in this unique collection that graphically depict the development stages of the radiosonde are presented. OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded in the Institution's annual report, Annals of the Smithsonian Institution. COVER DESIGN: The first true radio- sonde, which was flown by Robert Bureau in 1929. Library of Congress Cataloging-in-Publication Data DuBois, John L. The invention and development of the radiosonde : with a catalog of upper-atmospheric telemetering probes in the National Museum of American History, Smithsonian Institution / John L. DuBois, Robert P. Multhauf, and Charles A. Ziegler. p. cm.—(Smithsonian studies in history and technology ; no. 53) Includes bibliographical references. 1. Radiosondes—History—19th Century. 2. Radiosondes—History—20th century. 3. Radiosondes— Catalogs. 4. National Museum of American History (U.S.)—Catalogs. I. Multhauf, Robert P. II. Ziegler, Charles A. (Charles Albert), 1927- III. Title. IV. Series. QC879.25.D83 2002 551.63'52—dc21 2002030658 © The paper used in this publication meets the minimum requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48—1984. Contents Page Preface iv Introduction 1 I. THE PROBLEM OF HIGH-ALTITUDE METEOROLOGY AND THE TECHNOLOGY THAT LED TO THE RADIOSONDE 1. Nontelemetering Balloonsondes, 1892-1929 3 2. Preradio Telemetering, 1842-1894 15 3. Radio Telemetry, 1895-1929 19 II. THE RADIOSONDE: DEVELOPMENT AND APPLICATION 4. The Early Years, 1921-1928 26 5. The Prototypes, 1929-1930 30 6. Maturity, 1931-1940 41 7. The Radiosonde after 1940 59 8. Conclusion 67 III. CATALOG OF UPPER-ATMOSPHERIC TELEMETERING PROBES AND RELATED ARTIFACTS IN THE NATIONAL MUSEUM OF AMERICAN HISTORY, SMITHSONIAN INSTITUTION Notes 73 Appendix: Selected United States Patents Related to Radiosondes 75 Literature Cited 76 in Preface "Radiosonde" is a balloon-borne instrument that transmits atmospheric data, usually temperature, pressure, and humidity, to a receiver-recorder on the ground. The contribu- tions of this relatively simple device to the late twentieth-century way of life can hardly be exaggerated. No other factor contributed more to the systematization of weather observa- tions, which is beneficial to all who depend upon meteorological prediction. The utiliza- tion of the radiosonde directly affected agriculture and aeronautics, and its more sophisti- cated offspring made possible many of the marvels of the space age. The collection of instruments on which this study is based was assembled throughout a decade and a half, 1955-1970, as part of the program of the Smithsonian Institution to es- tablish a new museum of the history of science and technology and of the political and so- cial history of the United States. A museum rarely exhibits its entire collection in any field at one time, and objects not exhibited are retained for reference purposes. In the latter function, nothing is more useful than a printed catalog. Because of the variety of museum collections, the production of such catalogs is only occasionally realized. Such a document is most important when the collection in question is uncommon, or even unique. This may be the case with the collec- tion described herein. Because "old" scientific instruments often are unfamiliar even to those who use modem instruments serving the same, or similar, purposes, some guidance in their construction and functions is useful. It also is important to understand that the history of a scientific in- strument is a case history in the process of invention. For these reasons, this publication includes a capsule history of the development of the radiosonde based upon the instru- ments themselves and upon the original literature connected with their development. We are particularly indebted to Robert Wright of the Instrument Division of the United States Weather Bureau, who thought old instruments worth collecting, and to Christon Harmantas, who knew what they were. George A. Norton, Jr., assisted greatly with the cat- alog of specimens. We also gratefully acknowledge the help of William Blair, Jr., who kindly supplied us with copies of selected published and unpublished documents from the collected papers of his father, Colonel William Blair. Thanks also are due H.C. McBrair who alerted us to the importance of Colonel Blair's research to the early history of the radiosonde and who shared with us his reminiscences of radiosonde development at Wallace and Tiernan Inc., an early manufacturer of radiosonde equipment. At the opposite end of the time-span with which we are concerned, we are indebted for the support of the following historians at the Smithsonian Institution: David Devorkin of the National Air and Space Museum and Deb- orah Warner of the National Museum of American History. Robert P. Multhauf Director Emeritus National Museum of American History Smithsonian Institution IV The Invention and Development of the Radiosonde, with a Catalog of Upper-Atmospheric Telemetering Probes in the National Museum of American History, Smithsonian Institution John L. DuBois, Robert P. Multhauf, and Charles A. Ziegler Introduction The collection and transmission of data is as old as the hu- man species. Information about the weather may have first achieved "scientific" status when sensible observation (e.g., coldest winter in the memory of the oldest resident) of meteo- rological data was augmented by instruments capable of nu- merically describing the elements of the weather. The invention, in the seventeenth century, of the thermome- ter, barometer, hygrometer, and wind speed and direction indi- cators, was a phenomenon of the scientific revolution. These were hardly comparable in importance with some other inven- tions of the time, such as the telescope, whose revelations spec- tacularly embellished astronomy, the already crowned "queen" of the sciences. Astronomy already was mathematicized, whereas the "science" of the meteorologist still was difficult to distinguish from the art of the occultist. Still, instruments and the numerical recording of data gave hope for scientific re- spectability, and, by the end of the eighteenth century, the col- lection of meteorological data was one of the principle con- cerns of the numerous scientific societies that were part of the scientific revolution. Their periodical publications, the first of which was established in the 1660s, were devoted primarily to the dissemination of meteorological information. By the end of the seventeenth century, attempts had been John L. DuBois, C.E.O. Thermalogic Corporation, Hudson, Massa- chusetts, 01749. Robert P. Multhauf, Director Emeritus, National Mu- seum ofAmericam History, Smithsonian Institution, Washington, D.C. 20560. Charles A. Ziegler, lecturer in Anthropology, Brandeis Univer- sity, Waltham Massachiusetts 02453-2700. made to develop meteorological instruments that automatically recorded their readings. Steady improvements were made in the data collection process, but it soon was recognized that it was even more important to devise some speedy method for transmitting the data from various collection points to a central bureau. One means to accomplish this task existed in the eigh- teenth century. The visual telegraph, which used lanterns and mechanized semaphores, was invented in 1794 by a French of- ficer, Claude Chappe. Its primary purpose was to improve mili- tary communications, but because weather conditions were of vital military importance, both the collection and transmission of meteorological data increasingly were made ancillary to the military art. In the United States, the army played a key role in both the collection and the transmission of meteorological data via the electromagnetic telegraph, which was introduced in the 1840s. Yet, by the early twentieth century, the transmission of mete- orological data still was a minor application of the electromag- netic telegraph. The introduction of the wireless telegraph (ra- dio) did little to change this, although experiments in the automatic transmission of data from remote meteorological in- struments, including the transmission by wire from balloons, continued until the 1920s, when radio transmission was adopted. There were logical reasons for this lack of progress. The ac- cumulated mass of meteorological data had not greatly im- proved weather prognostication, and its collection was not con- sidered urgent. The meteorologist had not only failed, as was often charged, to "do something about the weather," but had a SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY poor record for prognostication. Meteorology remained an em- barrassment to practioners of the exact sciences, who increas- ingly were tempted to relegate it to the closet reserved for sci- ences that had failed to yield to the magic of mathematization. It wasn't until the 1920s that there was a marked improve- ment in meteorology. Geophysical and weather-related prob- lems of radio transmission had increased interest within the sci- entific community in the collection of atmospheric data from the stratosphere (Shaw, 1926, 1942; Stringer, 1972). There also was increased military awareness of the uses of meteorological data. In the civilian sector, the advantage to farmers and fisher- men of improved weather prognostication and the crucial de- pendence upon meteorology of the new technology of aviation, had become apparent. Moreover, although the impetus to im- prove instrumental meteorology can be traced to the interests and needs of these segments of society, the advent of the radio- sonde owed much to an almost accidental development, ama- teur radio—a hobby stemming from a widespread fascination with this new mode of communication and its "off-the-shelf technology. The efflorescence of amateur radio after World War I not only produced technological advances but also created a large market that dramatically lowered the cost of radio components. In the 1930s, the radiosonde was rapidly adopted by the weather bureaus of virtually all the industrialized nations. Dur- ing the 1940s, the innovations in radio resulting from the de- velopment of the radiosonde, spurred by pressure from military programs, spawned the new technology of multichannel data telemetry. During and after World War II, the needs of aviation and of rocket engineering to capture massive amounts of data simulta- neously (in "real time") from many sources were met by the further development of the principles and designs used to de- velop the radiosonde. Workers in Europe and the United States, who had specialized in balloonsonde or radiosonde research, applied their skills to develop rocket-borne instrumentation for atmospheric research (Kuiper, 1946:263). In the 1950s, data te- lemetry scientists used complex methods of radio modulation and transmission to enhance by several orders of magnitude the speed and accuracy of data transmission. Modern probes for exploring the upper atmosphere and space represented yet an- other advance in the three areas of technology originally merged in the radiosonde: environmental sensors, data trans- mission via radio, and an unmanned vehicle capable of moving the sensors and transmitter away from the earth's surface. The development of the radiosonde, from a historical per- spective, is important not only because these balloon-borne in- struments still are widely used today, but also because the ra- diosonde represents a technological link between the nontele- metering, balloon-borne probes of the pre-1930s and modern rocket-launched probes. Moreover, the chronological portion of the radiosonde's history that is the focus of this monograph (1929 to 1940) is especially important because it was during those years that the basic design features of radiosondes evolved, and because these developments laid the foundation for the introduction in the 1940s of multichannel radio trans- mission via subcarrier modulation (the basis for all modern an- alog radiotelemetry systems). The reason for this publication is the presence of a substan- tial collection of significant examples of the radiosonde in the National Museum of American History (NMAH), Smithsonian Institution. If, as appears to be the case, this is the only large collection of radiosonde instruments extant, it constitutes a unique resource for studying the history of radio telemetry and the history of one of the earliest applications of radio to the so- lution of problems in another technology. In order to understand the development of the radiosonde, the first three sections describe the relevant advances in bal- loon-borne instruments, preradio telemetry, and radio telemetry prior to 1929. The next sections outline the historical develop- ment of the radiosonde from its invention to the present. Fi- nally, there is a catalog of the radiosondes in the collection of the National Museum of American History, Smithsonian Insti- tution. Units are given in the metric system throughout this text. This has been done for simplicity and because metric units are standard in the field of meteorology. ABBREVIATIONS ARC Atlantic Research Co. AT&T American Telegraph and Telephone NBS National Bureau of Standards NPL British National Physics Laboratory ONR United States Office of Naval Research Plaskon Plaskon Co., 2121 Sylvan Ave., Toledo, Ohio RCA Radio Corporation of America USWB United States Weather Bureau WIT Washington Institute of Technology I. The Problem of High-Altitude Meteorology and the Technology that Led to the Radiosonde 1. Nontelemetering Balloonsondes, 1892-1929 Prior to the invention of the balloon in the late eighteenth century, rockets and kites were the only two vehicles capable of ascending into the atmosphere. The rocket was known in China by the early part of the thirteenth century, and it already was being used in Europe for military purposes before the end of that century. Kites, used as children's toys for many centuries in Asia, were introduced in Europe in the early seventeenth century. In 1749, British astronomer Alexander Wilson flew thermometers on kites to measure air temperature at high alti- tudes. This was perhaps the first recorded use of an unmanned atmospheric probe. Three years later, Benjamin Franklin per- formed his celebrated kite experiment on the nature of light- ning. Rockets were quite unsuitable as instrument platforms before recent times; indeed this application probably was not consid- ered. Kites also were not used often as probes until the advent of the Hargrave box kite in 1893. Manned flights to obtain sci- entific data became possible after 1783, using free balloons; but the captive balloon, because of its tendency to oscillate and rotate, was seldom used. Those defects were rectified with the invention of the Siegfried and Parseval kite-balloon in 1898. After the turn of the century, both instrumented box kites and kite-balloons were used by national weather services in Europe and the United States to obtain upper air data (Middleton, 1969a:98). By 1929, a considerable amount of geophysical data had been gathered, using kites, unmanned captive and free bal- loons, manned balloons, and airplanes. The history of such re- search, however, lies outside the scope of this monograph and will be discussed only insofar as the instrumentation developed for use with such vehicles influenced the design of instruments flown on unmanned free balloons. The use of the unmanned free balloon to obtain scientific data followed closely the introduction of the balloon itself. In August 1783, Jacques Charles, a French Professor of physics, launched at Paris an improved version of the recently invented hot-air balloon. In this balloon hydrogen replaced heated air as the gas fill. The path of Charles' balloon was tracked by sev- eral observers organized by Jean-Baptiste Meusnier, a young army engineer, who obtained data about the atmospheric forces exerted on the balloon (Gillespie, 1983:32). Meusnier's work constituted the first practical use of unmanned and uninstru- mented free balloons (later called "pilot" balloons) for scien- tific research. Throughout the nineteenth century, small pilot balloons often were released prior to launching a manned balloon, in order to estimate the winds aloft. It was not until the first decade of the twentieth century, however, that accurate measurements of wind velocity at various altitudes, using a theodolite technique to track pilot balloons, became an accepted meteorological tool (Blair and Lewis, 1931:1532). The practice of attaching instruments to an unmanned free balloon was started in France in 1892 by Gustave Hermite, with the help of a noted balloonist, Georges Besancon (Her- mite, 1893). These early versions of what came to be called a "balloonsonde" used small (five cubic meter) coal-gas-filled balloons made of varnished paper that were capable of carrying self-registering thermometers and barometers. Later versions used balloons made of goldbeater's skin (the outer layer of the caecum of an ox), and, for heavier payloads containing clock- work recorders, silk balloons with volumes of several hundred cubic meters were used (de Fonvielle, 1899:1-22). Payload weight and lifting capacity were designed to ascend to about nine kilometers before loss of gas, through the open neck of the balloon and by diffusion through its skin, caused the balloon to descend. This technique insured a slow descent and a reason- ably soft landing. Because of the prolonged floating period, balloons often landed as far as one hundred kilometers from the launch site. This dispersion exacerbated the problem of data re- trieval, which depended upon the finder mailing the payload back to the investigator's laboratory where the recorder could be read and interpreted. Sometimes balloonsondes were lost, but even when they were found, several days or even weeks elapsed before the data were recovered. Despite this disadvan- tage, the balloonsonde was capable of returning data from much higher altitudes than kites, captive balloons, or manned free balloons (Middleton, 1969a:99; 1969b:97-102). The discovery of the tropopause by French scientist Teisser- enc de Bort, announced in 1902, was made with a balloon- sonde. The tropopause, a transition zone between the tropo- sphere and the stratosphere, marks the point at which air temperature ceases to drop with increasing altitude. It begins at about 11.2 kilometers, an altitude considerably beyond the maximum heights attained by kites and captive balloons (due to the weight of the tether) (Teisserenc de Bort, 1899: SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 1.—Minimum-read- ing thermometer and barom- eter flown by Gustave Her- mite in 1892. (From W. de Fonvielle, Les Ballons- Sondes, Paris, 1899.) ^wW 417^420). The maximum height, 10.8 kilometers, that was at- tained by an open-gondola, manned, free-balloon in 1901, also was too low to discern the tropopause; it was limited by the maximum altitude at which humans can survive by breathing bottled oxygen. (The pressurized, closed-gondola balloon that made possible manned ascents to higher altitudes did not ap- pear until 1931.) By 1894, some balloonsondes were returning data from altitudes of 18 kilometers. Thus, the balloonsonde constituted the only means of probing the upper reaches of the atmosphere during the late nineteenth and early twentieth cen- turies. The first payloads flown by Hermite involved a crude instru- ment (Figure 1) that provided only minimum temperature and pressure readings. By 1893, he was using a "barothermo- graph," which was capable of providing a continuous strip-chart record of pressure and temperature for the entire flight (made with special nonfreezing ink for balloon use). His payload was an adaptation of a laboratory device (Figure 2) made by the well-known instrument firm of Richard Brothers of Paris, in which a Bourdon tube thermometer and an aneroid barometer were coupled to a clockwork-driven pen-and-ink re- corder. This instrument was used in a flight to almost 16 kilo- meters (de Fonvielle, 1899). The weight of the barothermograph required the use of a large (113 cubic meter) and costly balloon, and by 1895 Her- mite was using a new device (Figure 3), which was designed especially for use with balloons. The pen-and-ink recorder was replaced by tracing points on smoked paper, and a helical bi- metallic thermometer was used rather than a Bourdon tube. These innovations, together with generally lighter construction, reduced weight, cost, and complexity and increased reliability (de Fonvielle, 1899). In 1894, the year after Hermite's first balloon flight, an Amer- ican meteorologist, William Eddy, obtained data using a similar Richard instrument attached to a kite. This flight, which was the first time a complex meteorograph had been lifted by kite, reached a height of 470 meters (Eddy, 1898:452). After this suc- cess, meteorological instruments were designed especially for use with kites and, later, with captive balloons. A notable exam- ple is the Richard barothermohygrograph of 1896. This design, originally suggested by Abbot Lawrence Rotch, founder and di- rector of Blue Hill Observatory near Boston, Massachusetts, was a prototype for most of the later payload designs used for kites, captive balloons, and balloonsondes (Tissandier, 1896). The original instrument (Figure 4) was capable of measuring relative humidity, using a hair hygrometer, as well as pressure and temperature.1 This device marked the advent of the three-sensor system (thermometer, barometer, hygrometer) that remains the basis for meteorological probes today. Kites and captive balloons could not reach the altitudes achieved by balloonsondes, but they could repeatedly carry aloft heavy instruments without incurring much additional cost because the fixed expenses of the overall system, including motor-driven winches and cabling, were an appreciable portion of the cost per flight. This was not true of the balloonsonde, which was a complete system in itself. Hence, payload weight NUMBER 53 FIGURE 2 (above).—Barothermograph made by the firm of Richard Brothers of Paris, 1893. The device used a Bourdon thermometer and two aneroid cham- bers coupled to pen-arms that traced a record on a clockwork-driven cylinder. (From de Fonvielle, Les Ballons-Sondes, Paris, 1899.) FIGURE 3 (right).—An improved barothermograph used by Gustave Hermite in 1895. The unit used a stylus to scratch a record on a smoked paper in place of the pen-and-ink recordings used earlier. (From de Fonvielle, Les Bal- lons-Sondes, Paris, 1899.) SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY UitiiiiiiltiUizt^r.ir.u ..-:rs::nsi£:«u:uu: II II . II M.I »»«»■■ jlgf • I • I i iliiinj • «••«•♦••*« •* i m • HI ■ ii, i -• *«»*«•*>*••••> ■ < > I-™ I! ! i !OT »■'••' «••••• I II 1*11 I J«»l »••••• • <>^» i i ■ 14 ' ' I •"• ■• I HUH •*•#•** I .< !< i i ♦»»»•■ • i •»;?^ x^ l> 4 * •!•••■• • • ■ * r»* #■«•« i *<* .* •« I. < I . .. » * . • ...,*..< ^ «« 1' « * ■ «»• •• «••*•»»• »-• • * * • **■******: it FIGURE 4A.—Barothermohydrograph made by the firm of Richard Brothers of Paris in 1896 for American meteorol- ogist A.L. Rotch. (From G. Tissandier, La Nature, volume 24, 1896.) was reflected in the size and cost of the balloon, and this signif- icantly affected the cost per launch. Thus, although the Richard barothermohygrograph was suitable for use with kites and cap- tive balloons, lighter mechanisms were desirable for use with unmanned balloons. In 1896, at the request of Hugo Hergesell of the Strasbourg Observatory, lighter instruments (Figure 5) were designed by the German firm of Bosch and were used in an experiment in which balloonsondes were launched simulta- neously in Germany, France, and Russia (Hergesell, 1897). This was the first international attempt to obtain synoptic mete- orological data on the upper atmosphere. Pressure-, temperature-, and humidity-measuring payloads were not the only types developed and flown. As early as 1899, French scientist A. de la Baume-Pluvinel constructed and flew a sophisticated balloonsonde for solar spectroscopy (Figure 6). A clockwork mechanism opened and, after a brief interval, closed the slit of a spectrometer at a preset time during the flight, allowing the solar spectrum to be recorded on photo- graphic film. During this interval, the image of the recording arm of a barometer was projected onto the film so that pressure (and altitude) could be correlated with spectral data. Perhaps the most innovative aspect of this balloonsonde was the use of a compass-like magnetic frame to align the sonde (which could rotate around its longitudinal axis) with the earth's magnetic field. The heliostat mirror was tilted slightly, so that the noon- time sun's zenith distance was such as to reflect sunlight di- rectly into the slit that was opened by the clockwork timer. This is the earliest example of the use of an automatic, solar, homing device in a geophysical probe. Indeed, conceptually Baume- Pluvinel's balloonsonde represented an extraordinary precursor of a late twentieth-century design (Baume-Pluvinel, 1903). Baume-Pluvinel's use of photographic film to record simul- taneously the output from more than one sensor not only al- lowed the correlation of several parameters, but also was parsi- monious in providing needed information with minimal equipment (advantageous in balloonsondes where low weight and reliability were important considerations). Moreover, it demonstrated that it was possible to record mechanical dis- placement by nonmechanical means, i.e., photographically, in a balloonsonde. This technique subsequently was applied to other types of instruments based upon the measurement of me- chanical displacement, where a direct physical linkage between the displaced structure and a recording mechanism would be difficult to construct and/or would destroy the accuracy of the measurement (e.g., in measuring the height of mercury in a thermometer or the separation of the fibers of an electrometer). The advantageous features of the photographic technique made it the preferred method for investigators who sought to build payloads for obtaining data not available from the standard me- teorological balloonsonde. At the turn of the century, the development of the rubber bal- loon dramatically altered unmanned balloon technology and made low cost and low weight even more important in design- ing balloonsondes. With the exception of the gas fill, the bal- NUMBER 53 FIGURE 4B.—Photo of the original version made by Richard Brothers for use with captive balloons. (Instrument in NMAH collection, catalog number MHT 308209; Smithsonian photo 48467-H.) loons used in the 1890s were essentially identical to the first unmanned balloon launched by Jacques Charles in 1783.2 Such balloons were the "fixed-volume" type, that is, the envelope was made of a nonstretch material that was fully inflated on the ground. The neck or "appendix" of the balloon was left open to allow gas to escape during the ascent, which prevented the ex- panding balloon from bursting in the rarified air at high alti- tude. These balloons decreased their speed of ascent with in- creasing altitude, and, because their descent depended upon the gradual loss of gas, they had a prolonged "float." Both of these characteristics were undesirable. The fact that the speed of as- cent was not constant meant that the ventilation of sensors was variable, which made it difficult to interpret the data they pro- vided. Also, the prolonged float meant that the balloons often SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 5.—Balloonsonde built by the firm of Bosch in 1896 for German meteorologist Hugo Hergesell. This represents a balloonsonde version of the Richard Brothers instrument shown in Figure 4B. (From H. Hergesell, Meteorologischen Zeitschrift, volume 14, 1897.) travelled great distances from the launch point before landing, making recovery costly and time consuming. In 1901, German meteorologist Richard Assmann revolu- tionized unmanned balloon technology by introducing the use of sealed rubber balloons (Stringer, 1972:53). These were made of specially processed rubber so the balloons could stretch to more than two and one-half times their launch diameter before bursting. Theoretically, this allowed them to reach altitudes ex- ceeding 20 kilometers (a height sometimes achieved in prac- tice). Rubber balloons provided two significant advantages over the fixed-volume balloons: first, because they expanded during ascent, they tended to exhibit a fairly constant speed of ascent, which made sensor data more reliable; and second, they promptly burst at maximum altitude. The payload quickly de- NUMBER 53 FIGURE 6.—Solar-homing balloonsonde built and flown by French scientist A. de la Baume- Pluvinel in 1899. The purpose of this sophisti- cated device was to measure part of the solar spectrum at high altitudes where radiation traverses only a small portion of the earth's atmosphere. (From A. de la Baume-Pluvinel, L'Aerophile, volume 2, 1903.) scended on a parachute, which considerably reduced recovery times. Moreover, such balloons offered two means of payload recovery: descent by a single parachute, which was used with low-cost payloads when timely recovery was important, and the multiple-balloon technique, which was used with costly, one-of-a-kind payloads when a soft landing and a high proba- bility of retrieval were essential. The first method became stan- dard for meteorological balloonsondes, whereas the second was characteristic of research flights involving hard-to-replace special instrumentation (for example, with the first radio- sonde). The multiple-balloon technique (which depended upon the use of rubber balloons) was introduced by Hergesell in 1904. The single balloon and parachute combination was replaced by two or more smaller balloons of equivalent lifting power. Al- though the balloons were designed to burst at roughly the same altitude, it was certain that one balloon would burst before the others, allowing the remaining balloons and the payload to de- scend slowly to a soft landing. This method enhanced the like- lihood of retrieval because the remaining balloon(s) floated above the grounded payload, acting as a signal to potential finders (Hergesell, 1904:200). Although rubber balloons were not reusable (as were bal- loons made of silk or goldbeater's skin), they were less costly; hence, the average cost per flight was not appreciably higher. Moreover, the new balloon technology promised even greater cost reduction if payloads could be lightened and rubber bal- loons made smaller. This in turn suggested the possibility of sending up large numbers of balloonsondes, making the possi- bility of routinely obtaining synoptic data on the upper atmo- sphere a meteorological reality. William Dines was the first to solve the payload weight problem (Dines, 1906:101) by eliminating the costly clock- work-driven recorder and the complicated mechanical linkages between sensor arm and recording arm and substituting a small, highly polished, flat, metal surface on which the arm of the sensor scratched a line (Figure 7). The total deflection was very small, necessitating the use of a microscope to read the record- ing, but this was a small price to pay for the considerable de- crease in cost and weight and the increase in reliability. Similar designs, emphasizing simplicity and light weight, were pro- duced in Europe and were used in balloonsonde flights before World War I (Middleton, 1969a: 100). Nonetheless, most of the upper-atmosphere data used in weather forecasting still were obtained from instruments flown on kites and captive balloons because the recorded data could be recovered quickly by simply reeling in the tethered vehicle, whereas balloonsonde records were only recovered after a con- siderable time. Thus, the balloonsonde was primarily a means of conducting geophysical research in the upper atmosphere. One of the more impressive examples of geophysical instru- mentation was the balloonsonde (Figure 8) designed in 1913 by 10 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 7.—Diagram of Wil- liam Dines' balloon meteo- rograph. This device intro- duced the use of both a stylus to scratch minute displace- ment of sensors on polished metal surface and a micro- scope to read the resultant record. This approach dra- matically lowered weight and cost, making it feasible to launch balloonsondes in con- siderable numbers. (From W. Dines, Symon 's Meteorologi- cal Magazine, volume 41, 1906.) Charles Abbot of the Astrophysical Observatory of the Smith- sonian Institution to measure solar radiation at high altitudes (Abbot et al., 1915:1-26). Abbot's work is noteworthy not only because of its scientific results but also because it demonstrated that the existing unmanned balloon technology was capable of lofting payloads to very high altitudes. In designing his balloonsonde, Abbot adopted the photo- graphic technique pioneered by Baume-Pluvinel. In Abbot's device (Figure 8), a clockwork drive rotated a drum and also briefly opened the shutter of a slit-like aperture in the housing of the drum. Light, which passed through the aperture, im- pinged on photographic film on the inner surface of the drum. At the same time the aperture was partly covered by the mer- cury of a thermometer that rose or fell, depending upon the temperature of an attached disk pyrheliometer. Thus, the length of the aperture image recorded on the film measured the disk temperature, which was a function of the solar radiation. In or- der to obtain a simultaneous measurement of pressure, the im- age made by a barometer recording arm also was projected onto the film, a la Baume-Pluvinel. Abbot used the multiple-balloon recovery technique that had been introduced by Hergesell, which had become the preferred method for use with costly, specialized balloonsondes. Abbot in- tended to loft his balloonsondes to the highest possible altitudes so as to maximize the reliability of extrapolating the measured value of solar radiation to the value that existed outside the blan- ket of air surrounding the earth. He used special rubber balloons that were designed and made by a company in Russia, which had produced balloons capable of reaching extremely high alti- tudes and which had produced most of the sounding balloons used in Europe prior to World War I (Abbott et al, 1915). Abbot constructed and flew five balloonsondes of this type, at Avalon, California, and Fort Omaha, Nebraska, in 1913— 1914. One flight not only obtained useful data on solar radia- tion, but also set an altitude record for an unmanned balloon in the United States (32,643 meters) (Abbott et al., 1915). This al- titude was extraordinary for that era. Balloons capable of reaching this height generally were not available until after World War II. The Russian firm that made these rubber bal- loons did not survive World War I; indeed, even its name is un- known.3 Thus, the secret of processing the rubber to increase its stretchability and its resistance to chemical attack by the rel- atively high concentration of ozone above 25 kilometers also was lost. (Present-day high-altitude balloons are made of ozone-resistant plastic.) The disappearance of the Russian firm created a serious problem for researchers. For example, in 1919, when Robert Millikan, an American physicist and future Nobel Laureate, constructed a balloonsonde capable of measuring environmen- tal, high-energy radiation (later called cosmic rays) above 25 kilometers, he had great trouble finding suitable balloons (Mil- likan, 1950:210). He did not obtain usable balloons from a United States manufacturer until 1921, but these still were in- capable of reaching the desired altitude.3 Millikan later de- signed a payload that was capable of measuring the intensity of ionizing radiation as a function of altitude to a height of 32 ki- lometers (Figure 9). Like the instruments of Baume-Pluvinel and Abbot, Millikan's balloonsondes used a photographic method of data storage to record simultaneously the output of a small, bifilar electrometer and the movement of the arm of a pressure sensor (Millikan and Bowen, 1926). These balloon- sondes were flown in 1922, and, although the data were useful, the maximum altitude achieved (15.5 kilometers) was far lower NUMBER 53 11 M. ^^^^^figte :;.JlMWWMk..s..- 1 1 | .vij^^K;:1 > ;Tffi^MPffi****-:'"-j '■ '^^^^^^^ ^ m jp4 ^^B &sSaBBK&r£idtim- 1111111111 ':'v ,1| ^ \ I 11 |S| "~*-l M ,4? \S|^^H Wmk \ AM$$& i W&ffi&gm:, "Wmmk ' ^■nmmmmW: s^J w s/ w w * B V'""V -V——V" \/ -v ■»>/ " V ^ V ' V ' ' V V""*^/" ,^j^^^#itr+W*r*+4M**'**"Vi ""»"' E»*l|»n^*^HwHPM'* V •^ «^' ■ v" " ' v/" V HUE ft immumjUi iyr V***^^******^* V* "W» «■»■***»«***>»- PM> »* «»*ft»i»m' •s/ "V 'V *N/"'"V V ' 'V" V ' V V V FIGURE 28.—Recording strips produced by the Bureau radiosonde. Strips 1, 2, and 3 are successive parts of the same strip and show the transmitted signal as flat, noise-free portions of the trace. From A to B, the flat portions occur when the stylus touches a conducting portion and sends an indication of the thermometer. In C, D, and E, the flat portions indicate they are from the barometer. Readings are taken from a number of transmitted (flat) pulses, not their individual lengths. Strip 2 is a null selection of the cam, sending neither temperature nor baro- metric data. The temperature reading is 4 pulses, whereas the barometer reading is 3 pulses. (From R. Bureau, La Meteorologie, volume 7, 1931.) inaccurate claim was accepted by some contemporary com- mentators, and it has been reiterated by some modern historians as well.7 Therefore, it is relevant to comment on the matter of priority. The first documented radiosonde flight was made in 1924 by Blair, but because this flight was not reported in a scientific journal until September 1931, it is understandable that neither Bureau nor Moltchanoff were aware of Blair's work. However, Bureau published the results of his first flights in 1929 in the widely read journal, La Meteorologie. In this publication Bu- reau (1929:1565) averred that he used a balloon-borne short-wave radio "to transmit the indications of a barometer and a thermometer also attached to the sounding balloon." Ap- parently, in 1931 Moltchanoff did not know of Bureau's flight. The failure, however, of some modern historians to cite the prior radiosonde flights of Blair and of Bureau is less under- standable. Moltchanoff's interest in meteorological applications of ra- dio began in 1923, and in 1928 he published a paper suggesting the possibility of transmitting the output of meteorological sen- sors by radio (Moltchanoff, 1928:39). By 1930, he had con- structed a radiosonde (Figures 31, 32) that became known as the Kamm-geraet, or "comb-apparatus," because of the appear- ance of its contact mechanism. The most interesting feature of the Moltchanoff sonde was that each of the contact systems caused the radio transmitter to relay a specific Morse Code let- ter by on-off keying to the ground receiver: the first four combs were assigned the letters e, i, s, and h. The fifth comb, whose purpose was to provide a reference point, was assigned the let- ter g. The Morse letter picked up by the receiver clearly identi- fied which of the combs was in contact with the temperature arm. Because the temperature difference corresponded to the distance between two teeth, the rise or fall of temperature was determined by listening with an audio-output receiver to the Morse codes produced. In practice, the radio operator wrote down the sequence of Morse letters and later decoded them into numerical values of pressure and temperature. The Moltchanoff sonde used essentially the same circuit as the Bureau sonde but with a half-wave dipole antenna. The an- tenna was hung from the balloon, with the sonde itself in the middle, like Blair's 1924 tracking transmitter. Moltchanoff first flew this radiosonde on 30 January 1930 at Sloutzk (Moltcha- noff, 1933). The sonde reached a height of 10 kilometers and measured temperature and pressure within this altitude range. A feature of the Moltchanoff sonde was that its signals could be heard on the conventional audio-output radio receivers, therefore no special equipment was needed. This advantage, however, was offset somewhat by the electromechanical com- plexity of the radiosonde that was necessitated by the encoding scheme. THE DUCKERT RADIOSONDE On 22 May 1930, a few months after the first flight of Moltchanoff's radiosonde, Paul Duckert, of the Lindenberg NUMBER 53 35 FIGURE 29.—Later model of the Bureau radiosonde incorporating a clockwork-driven cylinder. (From P. Beelitz, Radiosonden, Berlin, 1954, fig. 25.) The transmitter on Bureau's sensor-less flight in 1927 had used a wave- length of 42 meters and a quarter-wave antenna. Transmitters on the 1929 flights used a wavelength of approxi- mately 60 meters (4.8 MHz) and retained the quarter-wave antenna design, which allowed the radiosonde to be at the balloon, with the antenna extending down away from it. In later models, the cylinder was rotated by a con- stant-speed, spring-operated clockwork, and provision was made for the inclusion of a hair hygrometer to make the instrument capable of measuring humidity as well as temperature and pressure. Bureau's work to simplify transmitter circuitry and to find a suitable wavelength-established design—a single triode Hartley oscillator directly coupled to a quarter- or half-wave antenna with dry plate and filament batteries—was followed by almost all investigators during the next decade. Observatory in Germany, flew a radiosonde of his own design. The sonde, which attained a height of 15 kilometers, transmit- ted measurements of pressure and temperature to a ground- based receiver (Duckert, 1932). Duckert's radiosonde differed from those of Bureau and of Moltchanoff in that it used a thermometer to vary the capaci- tance of the transmitter inductor-capacitor (LC) tuning circuit. This capacitance change varied the frequency of the radio sig- nal. The pressure was measured by means of a Bourdon tube and was indicated by interruptions of the signal. Blair had pio- neered the method used by Duckert to measure temperature by varying the frequency of the signal with a temperature-sensitive 36 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 30.—A production version of the Bureau radiosonde, circa 1932, marked "Constructeurs Mecaniques et Electriques P. de Presale 104-106, Rue Oberkampf, Paris." It is in an aluminum foil covered box, 27x16x13 cm; the vacuum tube and batteries are missing. (Instrument in NMAH collection, catalog number MHT 319827; Smithsonian photo 61837-D.) component of the transmitter circuit. Duckert, however, could not have been aware of Blair's unpublished work and indepen- dently reintroduced the concept in a more sophisticated form. The Duckert sonde used a bimetallic thermometer, which varied a capacitor, and a Bourdon tube barometer coupled to the transmitter by a simple sliding switch (Figure 34). Duck- ert's method of coding, direct-frequency modulation of the transmitter carrier, was extremely difficult to perfect, consis- tent with stability and precision, as Blair had discovered sev- eral years earlier. It required carefully engineered components to eliminate all environmental influences on the transmitter fre- quency except those calibrated as part of the sensing function. NUMBER 53 37 FIGURE 31.—A Pazel Moltchanoff radiosonde. The method of suspending a complete sonde and antenna is shown at the lower left. In this instrument, a series of six star-shaped cams with various numbers of points was operated by a windmill through gearing. The choice of a cam, which made contact at a given moment, was made by a second switch operated by the appropriate meteorological sensor. The windmill also operated a flat, insulat- ing disk with switch points and a switch arm. This formed part of the humidity circuit and had other functions. The instrument box is made of aluminum, measures 36 cm long x8 cm widex 15 cm high, exclusive of the wind- mill. (From P. Moltchanoff, Gerlands zur Beitrage Geophysik, volume 34, 1931.) 38 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 32.—Mechanical parts of Moltchanoff's first radiosonde design showing the rotating contact and sen- sors. (Instrument in NMAH collection, catalog number MHT 316262; Smithsonian photo 48467-A.) In a later model (Figure 35), all circuit components (except for the capacitor linked to the thermometer) were vacuum sealed in a glass tube; a feature that significantly improved performance in the hostile operational environment of radiosondes. By the mid-1930s, the cumbersome thermometer-capacitor assem- blage was replaced by a simple capacitor with a tempera- ture-sensitive dielectric. Provision for measuring and transmit- ting humidity was included (Duckert, 1932, 1933, 1935). The instruments described above constitute three distinct types, according to the means for converting the indications of the sensors into modulation of the radio transmissions. The three methods are chronometric (Bureau), coding (Moltcha- noff), and frequency modulation (Duckert). From 1931 to 1940, more than one dozen radiosonde designs were manufac- tured (see Beelitz, 1954), but all were based upon one of these three methods. These three designs can be considered prototyp- ical of the instruments that followed. These categories are not fundamental in a radio engineering sense because they do not represent mutually exclusive ways in which to code and modulate information onto a radio transmis- NUMBER 53 39 FIGURE 33.—Paul Duckert's first radiosonde. A continuous radio signal was emitted whose frequency varied with temperature, whereas pressure was indicated by interrupting the signal by a Bourdon tube (c) and a toothed wheel (d). One plate of the capacitor (b) was moved by a bimetallic thermometer (a and h). The complete radio transmitter was inside the glass tube (f). (From Duckert, Beitrdge zur Physik der Freien Atmosphare, volume 18, 1932.) sion. Nevertheless, they represent the basic categories into which radiosonde evolution can be divided during the period from 1931 to 1940. The basic radio-frequency design of nu- merous radiosondes, both during the prototype phase and dur- ing later development, underwent remarkably little change. The transmitters, with rare exceptions, used a single triode con- nected as a radio-frequency oscillator with the inductance-ca- pacitance tank circuit coupled to a resonant antenna. 40 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY | V A FIGURE 34 (right).—Diagrams showing the method of measuring temperature and pressure used in Paul Duckert's first radiosonde. In the upper diagram a bimetallic thermometer (T) by means of mechanical linkage (A) varied the capacity of the capacitor (C) in the oscillating circuit, which in turn varied the emitted radio frequency. In the lower diagram, the Bourdon tube (T) expanded and caused the contact arm (A) to pivot about its center and slide along seg- ment (S) touching the contacts (C) intermittently. Dashes were therefore trans- mitted while the arm touched the contacts corresponding to steps of pressure change. (From Wenstrom, Monthly Weather Review, volume 62, 1934). FIGURE 35 (below).—A Paul Duckert radiosonde, circa 1932. The entire radio circuit, except for the variable capacitor, operated by a bimetallic thermometer, is enclosed in a vacuum tube 6 cm in diameter and 16 cm long. This "brute-force" approach to isolating the transmitter from temperature and humidity changes can cause spurious frequency changes. A gold-plated, toothed wheel, turned by a Bourdon tube, switches between the variable capac- itor and a fixed capacitor at known values of pressure. Overall dimensions are 16x8.5x31 cm. (Instrument in NMAH collection, catalog number MHT 313527; Smithsonian photo 48467-C.) OSCILLATING § CIRCUIT < ^P s c // C // C osc PLATE II C ____£ T// C U B + 6. Maturity, 1931-1940 The importance of radiotelemetry as a meteorological tool was widely recognized after Bureau, Moltchanoff, and Duckert demonstrated the various ways that the output of conventional balloon-borne sensors could be transmitted by radio. Thereaf- ter, further radiosonde development mainly involved the devel- opment of meteorological instruments that were capable of providing synoptic atmospheric data at low cost, although re- search included exploration of the possible geophysical appli- cations (see below). By the 1940s, radiosondes could be called "mature" in so far as being well-engineered devices suitable for mass production. CHRONOMETRIC RADIOSONDES In 1931, Moltchanoff, whose earlier sonde was a coding type, designed a radiosonde that was based upon the Olland chronometric system, which was manufactured by the Aska- nia-Werke in Germany (Heck and Sudeck, 1931). In this sonde (Figures 36-38), the sensor-sequencing contacts were actuated by a clock, and the means of synchronization were incorpo- rated so that the radio signals emitted by the sonde could be re- ceived by a facsimile recorder. This was the first radiosonde to provide such recordings for all three parameters of meteorological interest. The sonde, however, was heavy, complex, and relatively expensive. More- over, the accuracy of the data depended upon the constancy of the clock that drove the rotating contact and thus determined scanning speed, a source of imprecision in all sondes based upon the Olland System. FIGURE 36 (right).—A Moltchanoff-Askania radiosonde. Shown in outline are the transmitter (T) and its small batteries (B) for filament heating and anode supply. The frequency of the transmitter was controlled by a quartz crystal to maintain the constancy of frequency over the range of temperatures encoun- tered, an unusual design among early radiosondes. The three measuring ele- ments are the barometric diaphragm (A), the thermometer (M), and the hygrometer hairs (C). The thermometer and the hygrometer are housed in the air channel (G), whereas the former is protected from radiation by a highly pol- ished cylinder (H). A clockwork motor drove the contact arm (D); holes were for winding and control in the cover (F). These are all supported on a plate (E) in the upper part of the casing. The transmitter and its batteries sit on an insu- lating sheet on the bottom-plate. The plug (K) completed the circuit to the con- tact arm and also to the leads (L) of the dipole antenna. Note the mechanical arrangement for suspending the sonde at the center of the antenna, an awkward design that contributed to excessive swinging of the sonde below the balloon during ascent. (From J.F.H., Nature, volume 130, 1932.) The radiosonde weighed two kilograms and employed a short-wave transmitter that operated in the range between 25 and 100 meters. The sonde was positioned at the middle of a half-wave dipole antenna that trailed below the balloon. Three sensors were used: a bimetallic thermometer, a Bourdon-tube barometer, and a hair hygrometer. In 1935, Bureau produced a design that eliminated the unde- sirable effects of small variations in scanning speed by chang- ing from a windmill drive of the scanning disc to a small elec- tric motor and by combining this with his earlier method for determining the rate of rotation of the scanning disc (Anony- mous, 1961:100). A frequency modulation was superimposed 41 42 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY QfrO^QfrOg- *Q*>0200 100 ° FIGURE 37.—Control mechanism in the Moltchanoff-Askania sonde. The contact arm (A) was rotated continuously by a clockwork drive. Three pointers of appropriate measuring elements were arranged one above the other so that they moved on a uniform circle, with the top of each pointer making a fleeting contact with the brush (A) in the course of its rotation. The brush also made contact on the synchronizing segment (S) and on fixed contacts (CI and C2), which jointly marked its revolution into three parts labelled temperature, barometer, and humidity, respectively. The relatively long contact at S was for the purpose of sending a longer signal in order to synchronize the recording cylinder at the receiver in a manner well known in connection with facsimile receivers of the period. The time inter- val between the end of the synchronizing signal and contact with the temperature arm thus depended upon the instantaneous position of the latter; similarly, the interval between the signal from CI and that from the barometer arm depended upon the positions of this pointer. This control was based upon the Olland System. The transmitter was an on-off keyed, CW type, where the keying contact was sequentially coupled to various sensors. on the radio carrier from timing pulses of sufficient rate (about 50 per second) to permit the speed of rotation of the scanning disc to be measured by the number of pulses. The frequency modulation was produced by a toothed wheel near the transmit- ter tuning circuit turned by the scanning disc. During the time the movable and fixed arms contacted the scanning disc, the pulses were suppressed and the angular motion of the disc was measurable, not as a function of time, but in terms of the num- ber of pulses between contacts. A tape-chronograph recorder was used to record the signals. Pulse counting was facilitated by the fact that every tenth pulse was suppressed; an effect pro- duced by the omission of one tooth in 10 on the toothed wheel. In Germany, A. Lang developed a radiosonde that utilized the chronometric method of transmitting measurements of tem- perature and humidity by interrupting the radio signal (Weick- mann, 1937:28). The Lang sonde (Figures 39, 40), which ap- peared in 1936, represented an attempt to reduce the size and weight of the transmitter and the antenna. The ground receiver, as in the case of the Moltchanoff-Askania sonde, was coupled to a facsimile machine to produce a visual record of the data. These clockwork-regulated chronometric radiosondes had a common problem—the discontinuous motion of the clock lim- ited the precision of any reading. In 1939, R.C. Jacobsen intro- duced a constant-speed motor, which greatly increased the pos- sible number of discrete movements in a radiosonde (Figures 41, 42) used by the Canadian Weather Service (Middleton and Spilhaus, 1953:246). Both the Bureau and the Jacobsen mecha- nisms represented mature chronometric design and were manu- factured in quantity in the following decades, as was the Lang radiosonde for the German Meteorological Service. NUMBER 53 43 lli^i^rt*1»"|% mummitmMmimauimMmwitimwvitvm *** • ,««*«»"+»* lf%«/*'t*%***'' tm0mBmmnmmmmtmiimtmmmitmmtimmmm*tk \ m* ^l|||WBM« ft******* ^-nmm -60 -50 -40 -30 -20 -10 0 + 10 + 20 + 30 Ct soo ;700 600 500 •400 •300 •200 ;100 •0 C, •100 •80 •60 -40 -10 eg 2 g 8,8 a p & If « '3 FIGURE 38.—Facsimile of recordings from a Moltchanoff-Askania radiosonde. Scale was applied by ref- erence to the beginning of signals from fixed contacts, designated CI and C2 in Figure 37. On the ground, signals were picked up by a conventional receiver coupled to a facsimile recorder of the rotating-cylinder type. The facsimile machine produced a visual record of transmitted data produced by the three sensors in the sonde and displayed this data in a manner that allowed the records to be turned into Cartesian graphs by the addition of coordinate scales. CODING RADIOSONDES The first Moltchanoff coding radiosonde, with minor im- provements, was used extensively in the U.S.S.R. for several decades (Anonymous, 1961:100). Although a few articles re- ported schemes for translating radiosonde sensor outputs into telegraphic code, the complexity and expense of including cod- ing in these radiosondes precluded their adoption elsewhere. Their advantage was that the coded signals could be picked up by any conventional receiver with an audio output. The only new coding sonde developed during the 1930s was produced by J. Gaw, a German engineer. In this sonde, one of several types used until the 1940s by the German Meteorologi- cal Service, a small electric motor rotated an insulated cylinder, 44 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 39.—Diagram of the mechanical parts of an A. Lang radiosonde and fragmentary transmitter circuit. The reference contact (Z) is adjacent to the hygrometer arm. An aneroid capsule and pressure arm switch are shown at the top. A starred-wheel temperature contact and rotating cylinder (W) are at the bottom of the figure. (From L. Weickmann, Denkschrift Uber Radio- sonde-Konstructionen, Berlin, 1938.) The sensors were a bimetallic thermom- eter, an aneroid barometer, and a hair hygrometer. The novel feature was that, instead of a single contact arm, the bimetallic thermometer rotated a star- shaped contact; an arrangement that, in effect, expanded the temperature scale. The hair hygrometer moved an arm that, in conjunction with a fixed contact, shut down the transmitter once each revolution for a time period indicative of the humidity. The clock drove two coaxial cylinders, one with a conducting band parallel to the axis, the other a conducting helix. Both tem- perature and humidity were based upon a time-cycle principle, but the aneroid barometer simply operated a switch that interrupted the radio signal at prede- termined pressure values, thus allowing the sequence of interruptions to be interpreted in terms of pressure. (Smithsonian photo 63882.) NUMBER 53 45 FIGURE 40.—An A. Lang radiosonde, with a short-wave transmitter (10.81 meters; 27.75 MHz), which weighed, with the mechanical arrangements described in the text, 560 grams. The wavelength used was considered very short in the 1930s. The mechanical arrangements are in the upright casing; the radio transmitter is in the cylinder. (Instrument in NMAH collection, catalog number MHT 322211; Smithsonian photo 61836.) portions of which had a conducting surface. Contact arms from a bimetallic thermometer, aneroid barometer, and hair hygrom- eter touched conducting portions of the rotating cylinder in a way that produced combinations of two Morse letters for each sensor. The complete cycle for the three sensors was repeated eight times per minute. Signals could be heard on the ground by using an ordinary radio receiver and could be written down as Morse letters that were interpreted later as values of pres- sure, temperature, and humidity. FREQUENCY MODULATION RADIOSONDES The Duckert radiosonde underwent various modifications during the 1930s (Duckert, 1932, 1933, 1935:573-584), in- cluding replacement of the oil-filled variable condenser used as a temperature sensor by an "electric" thermometer consisting of a temperature-sensitive capacitor. This was made possible by the development of a material whose dielectric constant had a large temperature-coefficient. Such material then was used as the nonconducting portion of a capacitor whose capacitance became a predictable function of temperature. Other Duckert sondes, unlike the first model that only measured temperature and pressure, incorporated a hygrometer and a means for trans- mitting humidity by varying the length of time the radio fre- quency was turned on. In later models, the pressure switch, which in earlier designs simply interrupted the signal, switched from the variable capacitor to a fixed-reference capacitor in or- der to provide a means of monitoring transmitter stability. These later designs also included ceramic insulation to make the transmitter more stable. The re-engineered designs of the Duckert sonde were mass produced throughout the 1930s by the German firm of Telefunken. Radio-frequency modulation, which Blair had used in his ra- diosonde of 1924, and which Bureau mentioned in an article published early in 1931, was used in Duckert radiosondes as early as 1930. That same year Vilho Vaisala, a Finnish engineer, independently conceived of frequency modulation as a simpler alternative to the Moltchanoff coding sonde (Vaisala, 1932). 46 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 41.—R.C. Jacobsen's radiosonde, designed for the Canadian Weather Service in 1939. The pressure and temperature arms were provided with special quick-break devices (K.) shown in inset. The wire was attached to the arm by insulating cement and completed the circuit through a spiral only for the moment after contact, but prior to the point where the lateral pressure pushed it away from the arm. (From W. Middleton and A. Spilhaus, Meteorological Instruments, 1953, fig. 171.) The Jacobsen sonde employed a spiral contact and an improved electric motor whose rotational speed was stabilized by a weighted reed. The rotating contact (A) was driven at about four revolutions per minute by the electric motor. A magnet (B) rotated inside a rectangular coil (C). The eccentric on the shaft of the magnet (B) vibrated a spring against an adjustable contact. Because the natural fre- quency of the spring was 1500 vibrations per minute, the shaft of the motor rotated at 1500 revolutions per minute with great constancy, thus reducing this source of imprecision. The frame of the sonde was insulated from the rotating spiral contact and supported an aneroid barometer (E), a bimetallic thermometer (F), and the gold- beater's skin hygrometer (G). A major problem in using frequency modulation was the variations, unrelated to the sensors, in oscillating frequency. Such an instrument required a highly stable transmitter free from frequency changes, such as those produced by variations in battery voltage or transmitter components. The Duckert sonde had achieved the requisite stability, but Vaisala's solution to the problem was more innovative. His design (Figures 43, 44) provided standard reference frequencies in the transmitted signal to alleviate errors caused by spurious frequency changes in the oscillator circuit (Vaisala, 1932:10). The Vaisala sonde (Vaisala, 1935) was light (365 grams, including battery) and in- expensive. It was used by the Finnish Weather Service in large numbers beginning in the mid-1930s and was used with only minor design improvements during the ensuing years. In 1937-1938, the British National Physical Laboratory (NPL), under the leadership of L.A. Thomas, perceiving the need for both direction-finding equipment and electrical sen- sors,8 experimented with audio frequency. The sensors con- sisted of two iron-cored coils whose inductance was varied by an evacuated, metal, bellows-type barometer and a resistance- wire thermometer. FIGURE 42.—Diagram of the spiral contact in a Jacobsen radiosonde. T, ther- mometer arm; P, pressure arm; U, humidity arm; H and J, reference contacts. The contact arms of each sensor moved over a spiral so that once in each revo- lution the end of the spiral touched reference points H and J. In the sonde, the stabilized motor reduced inaccuracy attributable to variations in scanning speed, and the absence of pivots in the sensor couplings reduced mechanical slippage, which was another source of error in sondes of this type. This radiosonde had several undesirable features. Because each of the three sensors required a separate oscillator, it was heavy, a defect magnified by the fact that the method for "read- NUMBER 53 47 FIGURE 43.—A Vilho Vaisala radiosonde. Five capacitances were successively switched into the transmitter cir- cuit by a rotating switch driven by a windmill. Overall dimensions of the sonde, exclusive of the windmill, are 9x 11 x22 cm. In Vaisala radiosondes, a rotating switch driven by a windmill sequentially connected five capaci- tors to the oscillator circuit of single-tube transmitter. Two of the capacitances were of fixed type and were used as references, and the remaining three were variable in accord with sensor output (see Figure 44). Sensors were arranged so that as pressure and temperature decreased and humidity increased, the associated capacitances all increased and, therefore, radio-frequency decreased. The rate of frequency modulation was quite low relative to the carrier frequency, so the bandwidth occupied by these transmissions was essentially the spread between the lowest and highest carrier frequencies produced by the switched capacitors, amounting to 1.6 MHz. The carrier frequency itself was between 23 and 27 MHz. During each cycle of the switch, two reference capacitors were connected to an oscillator circuit as a means of eliminating errors caused by frequency drifts. The mathematical basis for making the signals independent of transmitter drift was worked out by Vaisala and can be found in more recent texts, e.g., Middleton and Spilhaus (1953:253). In essence, if wavelengths of emitted signals are measured on a receiver in which plates of the tuning capacitor are semicircular, so that the square of the wavelength is pro- portional to the angular reading of the receiver's control knob, then it is possible to obtain the measurement pro- duced by a given sensor in terms of the ratio of differences of successive angular readings of the tuning knob of the receiver. In practice, this method substantially lessened errors caused by transmitter drift. Implementation of this technique was facilitated by using a semi-automatic recorder which, when triggered manually, registered the readings on a clockwork-driven drum. (Instrument in NMAH collection, catalog number MHT 319825; Smithso- nian photo 61837-C.) ing" sensor output required the variation of the inductance in the LC circuit of the oscillator, which worsened the weight problem. The NPL radiosonde also was expensive to manufac- ture and was never produced in large quantities. Instead, in 1940 a new program, begun at Kew Observatory by E.G. Dy- mond and his associates, developed the Kew radiosonde (Fig- ure 46), a successful design that was mass produced.9 It is important to note, however, that the NPL design, al- though not effective for meteorology, led directly to the devel- opment of a whole new field of technology—multichannel data telemetry. (The role of the NPL sonde in this development is discussed in Section 7.) During the 1930s, the meteorological radiosonde advanced from the prototype stage to mature designs suitable for mass production. There was a marked tendency in France, the Soviet Union, and Finland to improve on the prototypes developed in each country, based respectively upon chronometric, coding, and frequency-modulation principles, rather than to experiment with other possibilities. This tendency was less pronounced in Germany, where sondes in all three categories were used: the 48 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY ay - 5ov.|+ 5ov. FIGURE 44.—Schematic diagram of a Vaisala radiosonde: 1, barometer housing; 2, thermometer housing; 3, hair hygrometer; 4, variable capacitor of aneroid barometer; 5, variable capacitor of bimetallic thermometer; 6, vari- able capacitor of hygrometer; 7, 8, reference capacitors, 9, rotary switch. The oscillator, which was frequency modulated by sensors, was a common triode inductive plate-grid feedback type. Note the symmetrical, half-wave antenna required the sonde to be centered between each half of the antenna. (From L. Weickmann, Denkschrift Ober Radiosonden-Konstructionen, Berlin, 1938. Smithsonian photo 63881.) Askania-Werke and the Lang (chronometric), Duckert (carrier- frequency modulation), and Gaw (coding). Even so, however, the Duckert sonde predominated in Germany and was succes- sively re-engineered over the years. With regard to cost, the adoption and deployment of any type of radiosonde required training personnel in its use and estab- lishing extensive ground installations. Although the sondes were relatively cheap, the cost of training, of organizational in- frastructure, and of ground equipment represented a substantial investment that would be largely lost by a radical change in the sonde. It is significant that in Germany, where different types of sondes were used, a massive military rearmament program was underway during the 1930s.10 Thus, the more ample fund- ing available for satisfying military meteorological needs in Germany may account for the more eclectic approach adopted in that country. In the United States, the chronometric types were developed first, but before any large-scale use was made of them the NBS-Navy sonde (see below) appeared and be- came the unit of choice. Thus, in the United States the changeover to a radically new design was made before any substantial investment in training and ground equipment had been made. Another factor that influenced the retention of the original designs was that by 1940 the representative sondes in each cat- egory were "good enough."11 This fact, plus the ancillary costs NUMBER 53 49 ■-® SCALE FIGURE 45.—A late model NPL radiosonde. 1, support bracket; 2, variable inductance; 3, thermometer shield; 4, batteries. This radiosonde was among the earliest ancestors of the multiple audio-subcarrier type that led to modem analog data telemetry. (From H. Thomas, Proceedings of the Royal Society, London, volume A167, 1938.) The design of the Thomas-NPL instrument, favored by the Meteorological Office at Kew, was based upon particular con- siderations: it used less of an already overcrowded radio-frequency spectrum when compared with direct-carrier frequency modulation and was less affected by interference. Nevertheless, although L.A.Thomas adopted the basic system pioneered by H. Diamond, he rejected the use of the barometric switch employed in the NBS sonde. (Apparently, Thomas had tested Duck- ert's sonde that used barometric switching and found it unreliable.) Instead, Thomas opted to amplitude modulate the signal simultaneously with two dis- crete audio frequencies representing the thermometer and barometer outputs. Separate "beat frequency" oscillators in a superheterodyne ground receiver could be alternately connected and subsequently adjusted by an observer to reach a "zero-beat" condition, thus allowing temperature or pressure data to be obtained. In this model, a third coil and oscillator, whose inductance was varied by a goldbeater's skin hygrometer, was added. Use of variable induc- tances as sensors simplified the design of the audio-subcarrier oscillator but raised problems of weight and cost. 50 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 46.—British Meteorological Office radiosonde, Kew MK2. The transmitter and batteries are in a black, molded-fiber box, 12 cm in diameter and 23 cm high. The sensors are contained in rectangular aluminum shields mounted on three sides of the box. On the fourth side is a wheel with three conical cups, which revolved in the airstream to operate a switch that connected the three meteorological sensors in sequence. The Kew radiosonde, however, unlike its NPL predecessor, did not continuously transmit signals associated with the three sensors. Instead, the sensors were sequentially switched into the transmitter circuit by means of a rotating contact driven by a windmill. This latter feature was similar to the switching arrangement used in the Vaisala sonde. The fact that signals associated with each sensor were transmitted sequentially meant that only one audio-frequency oscil- lator was required. This made the sonde lighter and less costly than the NPL model. Also, two of the sensors used in the NPL sonde were redesigned for the Kew model—an aneroid barometer was substituted for the bel- lows-type, and a resistance-wire thermometer was replaced by a bimetallic type. Until the mid-1930s, the various radiosondes used in the United States were essentially elaborations of the chronometric design. In 1936, how- ever, a radiosonde was developed in the United States that introduced frequency modulation of the audio-fre- quency subcarrier. In this method, the amplitude of an audio-frequency-tone modulated the radio carrier signal. The tone was then frequency modulated. (Instrument in NMAH collection, catalog number MHT 322313; Smithsonian photo 61906-B.) in training and equipment, provide a plausible rationale for the course of development from 1931 to 1940 and later. AMERICAN RADIOSONDES The radiosonde instruments in the collection cataloged herein came from American institutions, principally from the United States Weather Bureau. They consist, not surprisingly, of more-or-less final versions of foreign instruments, plus American instruments, most of which never were made in quantity and were considered experimental. In order to make this narrative more comprehensible, we have separated these two categories. Our general history of radiosonde development has been based largely upon "mature" foreign instruments, NUMBER 53 51 FIGURE 47.—A Blue Hill radiosonde, circa 1935 (from Lange, Bulletin of the American Meteorological Society. volume 17, 1936). As the rotating cylinder, clearly shown in the center, revolved, the wire of the sensor arms sequentially contacted the arms of an aneroid barometer, a bimetallic thermometer, and a hair hygrometer. This arrangement resembled that used by Herath and Robitzch in 1917 and by Bureau in 1929. leaving for separate discussion the American instruments, about which it is possible to give more detail on experimenta- tion with different designs. In the United States, the Blue Hill Observatory, near Boston, Massachusetts, produced in 1935 a chronometric radiosonde (Figures 47, 48) that utilized an insulated, clock-driven cylin- der with an inset helical wire as the rotating control contact. The next year similar instruments were produced by the Guggenheim Aeronautics Laboratory of the California Institute of Technology (called GALCIT) (Maier and Wood, 1937) and by Friez, an established maker of meteorological instruments in Baltimore, Maryland. Both mechanisms were similar in design to the early Bureau sondes. The designers of the Blue Hill and GALCIT sondes solved the chronological problem by using special clocks with a rela- tively large number of discrete movements per revolution. The U.S. Bureau of Standards (NBS) replaced the clocks with small, constant-speed electric motors, including Jacobsen's constant-speed motor (Curtis and Astin, 1935, 1936:358). In 1937, a radiosonde featuring audio-frequency subcarrier modu- lation was developed at the NBS (Diamond et al., 1938, 1940). The Americans finally did develop a mature radiosonde. In 1936, the U.S. Navy asked NBS to develop a radiosonde that FIGURE 48.—A late model of the Blue Hill radiosonde, type F, circa 1936. The small bimetal thermometer, hair hygrometer, and aneroid barometer moved the contact arms along a cylinder, around which were wrapped three turns of a helical conducting wire. The shaft carrying this helix protruded through the top of a light wooden box, which was intended to contain a clock, a radio transmit- ter, and batteries. The box measures 18x10x7 cm. (From Lange, Bulletin of the American Meteorological Society, volume 18, 1937; see Figure 47; instru- ment in NMAH collection, catalog number MHT 322324.) 52 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 49.—Complete receiving and recording equipment for the NBS-Navy radiosonde as it appeared in December 1936. The electronic frequency meter at the center indicated the audio frequencies controlled by the effective values of air pressure, temperature, and humidity, thus revealing the upper-air conditions. (From W. Snyder and Bragaw, Achievement in Radio, 1986.) Coupled with the development of the radiosonde and new sensing elements, the NBS group designed and constructed special ground-station equipment. The complete assembly consisted of a superregenerative receiver coupled to an electronic audio-frequency meter through an amplifier and electrical filter unit. The output of the receiver was applied to the input of a high-speed recording voltmeter. In effect, the assembly converted audio-frequencies into a graphical chart whose abscissa could be cal- ibrated directly in terms of the measured parameters. would meet the Navy's meteorological requirements. H. Dia- mond and his associates at NBS organized a well-financed pro- gram that produced a superior radiosonde (Diamond et al., 1937). The thermometer used in the first NBS sondes was a small glass tube filled with a nonfreezing electrolyte. This elec- trolytic temperature sensor could be calibrated to a few tenths of a degree Centigrade and was stable across the range of envi- ronmental conditions encountered by the radiosonde. The me- chanical sensors used to measure temperature and humidity were inadequate at the higher altitudes attained by these bal- loons in 1936. This prompted the NBS to develop of electric sensors for both parameters, and the availability of these sen- sors influenced the design of the NBS-Navy sonde.12 Although the thermistor temperature sensor was available by the late 1930s, when radiosonde production began, it was not sufficiently precise and repeatable to be used at that time. By the early 1950s, however, thermistor development had ad- vanced, and glass-encapsulated types, with long-term stability and precision to less than one degree Centigrade variance, of- fered sufficient advantage to be used in radiosonde designs. The first ground receiver was available in early 1938, and service use of the first NBS-U.S. Navy radiosondes (Figure 49) began in the Department of the Navy in June 1938, at the Ana- costia Naval Air Station in Washington, D.C. (Snyder and Bragaw, 1986:127-129). In 1939-1940, the previously de- scribed modifications were made to accommodate the electric hygrometer. By 1940, the design had been refined, and in that year 15,000 such sondes were manufactured (Snyder and Bragaw, 1986:127-129). Many hundreds of thousands were produced in the following years. An important feature of the sonde was a so-called "pressure drive." Diamond and his colleagues had sought to avoid the complexity and cost entailed by the use of clockwork or elec- tric motors. Some means of switching was required, however, and because they had decided to retain the aneroid barometer as the pressure sensor, they developed a method using its me- chanical displacement as a switch, as well as an indication of pressure (Snyder and Bragaw, 1986:127-129). Moreover, me- teorologists favored a record in which temperature and humid- ity were given as functions of pressure. By using the barometer as a switch (Figure 50), such a record could be obtained. To transmit data, a 65-megahertz (MHz) radio signal was ampli- tude-modulated at an audio frequency of a few hundred cycles per second by a resistance-capacitance (RC) controlled oscilla- tor. Resistive thermometer and hygrometer elements produced NUMBER 53 53 FIGURE 50.—An electric hygrometer and thermometer assembly used in later NBS-Navy radiosondes, circa 1939. The thermometer is suspended between two arms at the top of the unit; the hygrometer is the upright strip in the center. The hair hygrometer was replaced in later models by a newly developed electric hygrometer. This consisted of a polystyrene strip coated with a film of lithium chloride dissolved in polyvinyl acetate with two tin electrodes sputtered along the long edges of the strip. A combined electrical hygrometer and thermometer were mounted on a single assembly that could be easily attached to the sonde. Replacement of the hair hygrometer with an electric type necessitated circuit changes that involved the use of a relay and a twin-triode vacuum tube, so the final circuit became more complex than that shown in Figure 49, but its essential operation remained the same. (Instrument in NMAH collection, catalog numbers MHT 327975 and 327976; Smithsonian photo 61553.) a calibrated modulation of this audio-frequency in response to temperature or humidity. This design was remarkably successful because the meteoro- logical radiosondes used today in the United States still reflect the basic design of the NBS-Navy sonde developed more than half a century ago. The success of the NBS design can be at- tributed in part to the Navy's insistence on focusing the devel- opment on three factors: cost, weight, and accuracy. The sonde was to cost about $25 when purchased in reasonable quantity, and the weight was not to exceed one kilogram. It was to mea- sure temperature to within one degree over the range from plus 40 to minus 75 degrees centigrade, pressure to within one mil- libar from 1050 to 1500 millibars, and humidity to within three percent from 0 to 100 percent relative humidity. In addition, the U.S. Navy imposed a requirement that significantly affected not only transmitter design, but also the method of signal trans- mission. The radiosonde was required to emit signals that could be used, in conjunction with radio direction-finding equipment on the ground, to determine wind speed and direction (Snyder and Bragaw, 1986:131). This last requirement meant that it would be advantageous to have a continuous signal on a constant carrier frequency to fa- 54 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 51.—This NBS-Navy radiosonde was manufactured in large quantities by 1940 and incorporated improvements such as an electric hygrometer. The batteries, transmitter, and pressure unit were housed in an insulated box to minimize temperature changes. The temperature tube and electric hygrometer (at rear) were shielded against solar radiation by a double-walled metal tube. The total weight was less than 1 kg; the cost was $25 in 1940. The finder of the balloon after its flight received a monetary award upon return of the radiosonde. (From Snyder and Bragaw, Achievement in Radio, 1986, fig. 52.) cilitate direction-finding. It also required, indirectly, higher transmitter frequencies that would allow smaller and more ac- curate tracking equipment. The necessity to transmit informa- tion from the sensors on a constant and stable frequency sug- gested the use of an audio-frequency subcarrier modulated onto the main radio-frequency carrier. The subcarrier audio fre- quency then would be modulated as a means of relaying pres- sure, temperature, and humidity data. The use of audio-fre- quency modulation, in turn, made it possible to use circuits that were especially compatible with the use of electrical sensors based upon resistance that had become available recently. These were superior to the mechanical sensors that previously had been used in most radiosondes. One commentator (Middleton, 1969a:339) suggested that the decision of the NBS group to base their sonde on audio-fre- quency modulation was influenced by the commercial develop- NUMBER 53 55 FIGURE 52.—A "pressure drive" of an early NBS-Navy radiosonde. Overall dimensions are 13x9x6 cm. In order to understand the operation of a pressure drive it is helpful to view a schematic (see Figure 53) of one of the first models of an NBS sonde. This early model employed an electric thermometer, but the electric hygrometer was still under development so a conventional hair hygrometer that mechanically varied a resistor (see Figure 54) was used instead. In this arrangement the arm of the barometer (P) swept over a series of conducting strips numbered 1,6, 11, etc., in Figure 53. These strips were mounted on a nonconducting board. The fixed resistor (R) and variable resistors (H and T), representing the hygrometer and the thermometer, respectively, were con- nected in series between the grid of a multi-electrode vacuum tube and the instrument frame. When the barome- ter arm was not touching a contact, the only path from the grid to the frame was through R, H, and T, in series. Thus, total resistance varied with that of the temperature-dependent resistor (T) and so the modulating audio-fre- quency signal indicated temperature. When the barometer arm touched the unnumbered contacts, the resistance was equal to R plus that of the humidity-dependent resistor (H), and audio-frequency was indicative of relative humidity. When the arm touched contacts 11 and 26, the point (a) was connected directly to the frame, causing the modulation to assume a fixed upper frequency; when it touched 1,6, 16, and 21, the point (b) was connected to the frame, producing a fixed lower frequency. Alternate change-overs from one set of frequencies to another indicated that the arm was just reaching or just leaving one of the intermediate contacts and had attained deflec- tion position corresponding to a known pressure. (Instrument in NMAH collection, catalog number MHT 312800; Smithsonian photo 61836-B.) ment of a high-quality, audio-frequency meter manufactured by the General Radio Company. Of course, advances in radio technology in the 1930s influenced radiosonde development in all countries. All radiosonde designers responded to the avail- ability of small, light-weight, rugged, vacuum tubes and of an- cillary circuit components by repeatedly re-engineering their sondes to include such elements. Finally, it is pertinent to comment on cost-reduction as a pri- mary consideration in meteorological radiosonde development. The early prototypes had demonstrated that radiosondes were technically feasible, but their wide-scale deployment as part of a synoptic data-gathering program was dependent largely upon dramatically lowering unit cost. In the United States, the Navy's cost requirement of $25 was relatively low for that time, and it was achieved chiefly because of features that made the NBS design consonant with the techniques of mass produc- tion. This was equally true of the successful sondes that evolved within other design categories because procurement on a large scale was the most effective means of lowering unit cost. THE RADIOSONDE IN GEOPHYSICS The success of weather radiosondes made scientists increas- ingly aware of the possibility of transmitting data from a vari- ety of balloon-borne sensors that measured parameters of geo- 56 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY R r—*WVWW a P H it FIGURE 53.—Fragment of a circuit diagram of an early NBS-Navy radiosonde showing the sensor-sequence switching mechanism of the audio-subcarrier oscillator. (Smithsonian photo 63263.) physical interest, such as ozone concentration or cosmic-ray intensity. Therefore, in the decade 1931-1940, further explora- tion of the upper atmosphere, made possible by the advent in 1931 of the pressurized gondola for manned balloons, pro- duced information that led meteorologists to place more em- phasis on obtaining atmospheric data from very high altitudes. The possibility of routinely reaching such altitudes with radio- sondes became a reality in 1936 when the first truly low-cost latex balloons, capable of lofting radiosondes to great heights, became available. These hydrogen-filled balloons weighed about 700 grams and were capable of lifting a 1.5 kilogram payload to altitudes of about 25 kilometers. Although cost was perhaps the most important determinant in the design of meteorological radiosondes, it was of little im- portance in the development of research radiosondes. From the mid-1930s to 1940, a number of such research radiosondes were flown. A few of these sondes represented original de- signs, but the others were meteorological radiosondes that were adapted to measure a specific parameter of geophysical interest (see "Research Radiosondes," below). The sequential appearance of each of these special sondes did not follow the cost-dictated pattern of design change appar- ent in the development of meteorological radiosondes because each geophysical radiosonde was a one-of-a-kind instrument designed to achieve objectives that were, to some extent, unique. Nevertheless, these research sondes explored the range of application of this type of instrument and, in this sense, their development represented an aspect of the maturation of the ra- diosonde as a measurement tool. By 1940, most of the possible applications of geophysical research involving radiosondes had been tested. Specifically, sondes for measuring cosmic rays, so- lar radiation, ultraviolet radiation, and ozone were designed and flown successfully. The measurement of cosmic rays was one of the first re- search applications of geophysical radiosondes. The altitude dependency of cosmic-ray intensity at various geomagnetic lat- itudes was of key importance to theoreticians seeking to under- stand the nature and origin of cosmic rays. Balloonsondes con- taining self-registering electrometers for measuring cosmic-ray intensity had been flown in the 1920s in the United States by Robert Millikan (described in Section 1). In the early 1930s, NUMBER 53 57 FIGURE 54.—A hair hygrometer used in an early NBS-Navy radiosonde. Overall length, 20 cm. A cylindrical cage of hairs is mounted on the upper surface of light-weight, white plastic molding; on the lower side of the molding is a wound resistor, over which a contact was moved by a change in the length of the hairs. (Instrument in NMAH collection, catalog number MHT 312799; Smithsonian photo 61901-D.) such balloonsondes were flown again by Millikan and also by Erich Regener in Germany (Millikan and Bowen, 1926; Rege- ner, 1932). As previously mentioned, however, retrieval of data from balloonsondes was always problematical even in popu- lated areas. Moreover, it was necessary to make vertical sound- ings of cosmic-ray intensity in locales, such as the polar re- gions, where recovery of a balloonsonde was extremely unlikely. Radiotelemetry was considered the solution to the data-re- trieval problem by scientists such as Arthur Compton in the United States, and Sergei Vernov in the U.S.S.R. In 1934, with Compton's support, J.M. Benade and R.L. Doan, at the Univer- sity of Chicago, conducted a test flight of the transmitter por- tion of a sonde that contained an ionization chamber for mea- suring cosmic rays (Benade and Doan, 1935). In July 1934, Vemov designed and built a complete radiosonde that used two Geiger-Mueller (GM) tubes in a coincidence mode to measure the intensity and directionality of cosmic rays. Vernov flight-tested this instrument in an airplane in July 1934, and in April 1935, the instrument was flown as a balloon-borne radio- sonde (Vemov, 1934, 1935). In Vernov's sonde, the coincidence output of the GM tubes activated a relay that switched on the anode circuit of a ra- dio-frequency oscillator. At five-minute intervals the number of pulses from one of the GM tubes also was transmitted for a short period. Switching from coincidence output to a single GM tube was accomplished by means of a barometer switching device, so that the sequence of switching cycles indicated pres- sure (and altitude). This was the first radiosonde to measure cosmic rays, and its 1935 flight provided data at altitudes up to 13.6 kilometers. In the same year, American scientist, T.H. Johnson, of the Bartol Research Foundation, flew a more elab- orate cosmic-ray radiosonde to an altitude of about 20 kilome- ters, using coincidence counters (Johnson, 1937:352). The flights of Vernov and of Johnson provided some useful data, but their chief importance was to demonstrate the potential util- ity of the radiosonde for obtaining cosmic-ray data in remote regions, thus making possible the systematic survey of cosmic- ray intensity over a wide range of latitudes. In the 1930s, unmanned balloons were the only vehicles ca- pable of reaching altitudes above 22 kilometers; however, be- cause of the prolonged "float" and consequent drifting, the probability of recovering balloonsondes decreased as the maxi- mum altitude increased. Thus, when radiotelemetry eliminated the data-retrieval problem inherent in the use of self-registering balloonsondes, the radiosonde became a valuable tool for re- search at very high altitudes. One factor investigated was solar radiation. In order to mini- mize the effect of the atmosphere, it was desirable to obtain so- lar flux data at the highest possible altitudes. Such research was begun by Charles Abbot in 1913, using a self-registering balloonsonde (see section 1, "Nontelemetering Balloonsondes: 58 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY 1892-1929") (Abbot et al., 1915) and was continued in the United States in 1937 by Brian O'Brien, using a radiosonde. In O'Brien's radiosonde, the light received by a quartz diffuser was passed through a filter that allowed light with the desired spectral characteristics to impinge on a vacuum photocell. The photo current thus generated discharged a capacitance that, upon recharge, caused a radio to transmit a short signal. The rate at which signal pulses were transmitted was proportional to the light incident on the photocell. Two photocell assem- blies were used to cover the violet and red portions of the spec- trum. In 1937, flights of this radiosonde provided measure- ments of solar flux to altitudes of about 22 kilometers, with a precision of about one percent. The altitude was determined visually by using a theodolite method (O'Brien et al., 1937). Later flights, involving a comparison of radiosonde data with measurements made simultaneously by ground-based instru- ments, yielded information on the absolute energy distribution of the solar spectrum (O'Brien, 1939). Reaching very high altitudes also was important in ascertain- ing, by extrapolation, the ultraviolet intensity of sunlight be- yond the atmosphere. This problem was addressed by W. Coblentz and R.S. Stair at the U.S. National Bureau of Stan- dards, using radiosondes. In the early 1930s, Coblentz devel- oped an ultraviolet-intensity meter based upon a cadmium cathode photocell that was sensitive to radiation in the range of 2600 to 3250 Angstroms. This meter had been used success- fully at ground stations to measure ultraviolet radiation and, in 1937, it was modified for use in a radiosonde (Stair and Coblentz, 1938:187). The amplifier and microammeter of the ground-based unit were replaced with a small audio-frequency generator, amplifier, and transmitter that had been designed for the NBS-Navy meteorological radiosonde. The output of the photocell modulated the transmitted audio frequency by chang- ing the voltage on the generator control grid that, in turn, al- tered the frequency of audio oscillation. The photocell was ex- posed to sunlight that passed through a series of filters mounted on a rotating wheel. The photocell thus provided sequential measurements of the ultraviolet spectrum. Several flights of this radiosonde were made in the summer of 1937. Because ozone absorbs ultraviolet radiation, measurement of the ultraviolet intensity at various wavelengths as functions of altitude can be used to determine the concentration of ozone at specific levels of the atmosphere. Stair and Coblentz made ad- ditional flights in 1938 with an improved radiosonde to obtain ultraviolet measurements that could be used for determining ozone concentrations. These flights reached altitudes of 27 ki- lometers and ascertained that the ozone layer extended from 18 to 27 kilometers, with the maximum concentration at 25 kilo- meters (Coblentz and Stair, 1939:573-576). These results con- firmed data that had been obtained in the early 1930s by Ger- man scientist, Erich Regener, using a balloon-borne, self-registering, ultraviolet spectrometer (Regener and Rege- ner, 1934. In 1939, cosmic-ray radiosondes were flown by scientists in several countries. In the United States, L. Curtis and A. Astin of NBS measured cosmic rays with a modified version of the NBS-U.S. Navy meteorological sonde. Belgium scientist, M. Cosyns, and British researchers, H. Carmichael and E. Dy- mond, also made flights using cosmic-ray radiosondes of their own design. The flights of Carmichael and Dymond were espe- cially noteworthy because they provided the first vertical soundings of cosmic-ray intensities near the North Pole (Car- michael and Dymond, 1939:323-327). The polar regions were of great interest to researchers be- cause they realized that theoretical understanding of cosmic rays would be advanced by measuring cosmic-ray intensity over a wide range of geomagnetic latitudes. The sonde devel- oped by Carmichael and Dymond incorporated several GM tubes operated in a coincidence mode to obtain information about the intensity and directionality of cosmic rays. The coin- cidence output was relayed by the transmitter as a sequence of short pulses. The altitude was derived from pressure measure- ments, using the method utilized in the Vaisala meteorological radiosonde—an aneroid barometer that varied the spacing of the plates of a capacitance in the tuning circuit of the oscillator of the transmitter, thus altering the radio frequency emitted. In flights made at 85 degrees north latitude, cosmic-ray measure- ments were obtained at altitudes up to 27 kilometers. By 1940, the utility of research radiosondes to obtain useful geophysical data from high altitudes had been demonstrated. With the exception of the British polar flights, however, which supplied hitherto unavailable data on cosmic-ray intensity in that region, the data supplied by radiosondes had, for the most part, merely confirmed results on ultraviolet light, ozone, etc., previously gathered with self-registering balloonsondes. As de- scribed in the next section, the availability in the 1940s of a practical plastic balloon, capable of reaching heights above 45 kilometers, enhanced the ability of the radiosonde to provide new geophysical data. Consequently, the special sondes of the 1930s became the first in a long succession of research radio- sondes that were built to explore the outer reaches of the atmo- sphere. 7. The Radiosonde after 1940 We have referred to the years 1929-1940 as the period of ra- diosonde development because the basic designs of meteoro- logical radiosondes did not change radically after 1940. Re- search in the earlier period also demonstrated the utility of the radiosonde for a variety of geophysical applications, which is discussed in this section. One of the earliest problems of generating radio signals was frequency control. The defects of a transmitter whose fre- quency wandered were obscured to some degree by similar de- fects in receivers. While Blair, Bureau, Vaisala, Duckert, and others struggled with problems of reducing the spurious influ- ences of temperature, humidity, and pressure on their transmit- ters, so that the calibrated sensors could give accurate readings, improvements in the selectivity of radio reception also made the problem of transmission crucial. Duckert, especially, was concerned with this problem because he was intentionally modulating the transmitter frequency to carry the sensor read- ings. The desired improvement was achieved by engineers working in other areas of radio transmission, who devised a means of using the extremely stable properties of quartz crys- tals to control the frequency of a radio-frequency oscillator. The use of quartz crystals to alleviate transmission problems was never widespread in radiosonde design. Radiosonde de- signers must have been aware of quartz-crystal circuits since the early 1920s, but they consistently rejected them in favor of simpler radio-frequency oscillators that used less precise meth- ods of stabilization. The reason for this probably was the pri- mary requirements for sonde design—simplicity and low weight. Quartz-crystal frequency control, although elegant and stable, required at least twice the number of components for an equivalent output signal power level. Not until the 1960s, when transistors came into common use in electronic circuitry, did the cost and weight effect of this penalty became less signifi- cant. The post-World War II period brought changes in the design of meteorological radiosondes, although they can perhaps be called evolutionary; that is, whereas engineering modifications were made periodically, in order to exploit advances in elec- tronics and related technologies and to accommodate regula- tory requirements of radio spectrum usage, the objectives of meteorological radiosonde design and the methods for achiev- ing these objectives remained relatively unchanged (Dymond, 1947:645-650). On the other hand, a different situation existed in the case of research radiosondes, where the objectives included the mea- surement of ozone, cosmic rays, etc., and involved exploitation of the dramatic advances in balloon technology in the 1940s. At the same time, designers sought to exploit new techniques of multichannel telemetry. Unlike the 1930s, when scientists constructed research sondes by borrowing ideas from the de- signers of meteorological radiosondes, the increasingly dispar- ate goals of geophysical research and radiosonde design in the post-war world occasioned little of this kind of interaction. METEOROLOGICAL RADIOSONDES By 1940, the balloonsonde had been abandoned, and the ra- diosonde had been adopted by national weather services. The original designs were used for decades, and some are still used today, necessarily changed over the years to incorporate new electronic components and to conform to international regula- tions controlling the assigned frequency at which transmitters could operate. In general, these changes achieved improved performance at lower cost. Chronometric Sondes The Bureau sonde in France and the Jacobsen sonde in Can- ada were re-engineered periodically to include up-to-date elec- tronic parts. In 1948, the Swiss weather service adopted a chro- nometric sonde designed by J. Lugeon. This mechanism utilized a clock-driven rotating assembly of the Olland type, but it incorporated a feature that minimized mechanical con- tact. The sensor arms were terminated in small plates that acted as capacitor plates when positioned above similar plates on the end of the rotating arm. Thus, the position of the arm was indi- cated by a change in capacitance that altered the transmitted signal (Lugeon et al., 1948:76). Coding Sondes The Moltchanoff sonde was progressively redesigned, not only to incorporate new electronic components, but also to im- prove the mechanics of the coding system. A version that ap- peared in the 1950s used a simplified coding mechanism con- sisting of a series of studs arranged in three arcs on an insulating plate over which the arms of the mechanical sensors moved. The position of a given arm was indicated by a wind- mill-driven mechanism that caused the transmitter to send an identifying code (Anonymous, 1961:101). Coding sondes also were developed in the early 1950s in Japan by K. Isono, and in Germany by H. Hinzpeter (Isono and Huziwara, 1950; Hinz- peter, 1951). A significant feature of the Japanese sonde was the use of two mercury-filled glass thermometers that elimi- nated the need for a preflight check of the temperature calibra- 59 60 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 55.—A Japanese coding radiosonde of the type designed by K.. Isono. In this sonde a small electric motor revolved a raster of contacts so that it successively touched points moved by an aneroid barometer and by a bimetallic thermometer and hair hygrometer. A commutator switched between these meters and also made con- tact with two mercury thermometers, one of which disconnected the humidity contact at a temperature of -30° centigrade, the other stopped the sending of a special signal at -50° centigrade. Otherwise, the indications were interpreted by noting the sequence of Morse letters N, D, B, T, M, U. The transmitter, which operated on a fre- quency of 402 MHz, was powered by a small storage battery. The instrument is in a white cardboard container, 27x15x11 cm, which acted as a radiation shield. (Instrument in NMAH collection, catalog number MHT 319826; Smithsonian photo 65996.) tion (Figure 55). The sonde designed by Hinzpeter (Figure 56) was basically a redesign of Gaw's coding sonde and used new electronic components and a more reliable coding mechanism. Radio-Frequency Modulation Over the years, improvements were made in Finland to the Vaisala sonde. A model introduced in 1957 used a unique sys- tem to eliminate the effects of temperature on sonde compo- nents. The aneroid barometer, transmitter, reference capacitors, and switching mechanism were enclosed in a double-walled housing. Before launching, the space between the walls was filled with water near zero degrees centigrade. This water tem- perature was maintained throughout the ascent, despite de- NUMBER 53 61 FIGURE 56.—A Gaw coding sonde, as redesigned by H. Hinzpeter in 1950. The mechanical parts and transmitter are enclosed in foil-coated, fiberboard box about 1 Ox 10x20 cm. Metal shields for the hair hygrometer and the bimetallic thermometer are on the left and right sides of the central box, respectively. In addition to their com- plexity and relatively high cost, the design principles of coding sondes made necessary the employment of sen- sors whose output was a mechanical displacement. This precluded any simple incorporation of improved sen- sors, whose output was a change in some electrical characteristic that were developed in the 1930s. (Instrument in NMAH collection, catalog number MHT 324241; Smithsonian photo 61906-C.) creasing air temperatures, by the latent heat of ice formation (Anonymous, 1961:104-105). In Germany, a frequency-modu- lation sonde was developed by K. Sittel in the early 1940s and was used during and after World War II. Its chief feature was the elimination of mechanical linkages to the temperature sen- sor by using a capacitor having a temperature-sensitive dielec- tric as a thermometer. This method was similar to that used in the Duckert sonde described in the previous section. Indeed, the Sittel sonde (Figure 57) can be regarded as a redesign of the Duckert sonde (Sittel and Menzer, 1950). Audio-Frequency Subcarrier Modulation Radiosondes used by the weather services of the United States and Britain, which utilized audio-frequency subcarrier modulation, did not undergo any radical redesign after 1940, although many models were made. In the United States, the NBS-Navy sonde was adopted by the Weather Bureau, but during the early 1940s the U.S. Weather Bureau Model (Fig- ure 58) continued to use the hair hygrometer instead of the electrical type used in some Navy versions (Snyder and Bragaw, 1986:124). In 1954, the Dutch Weather Service 62 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 57.—A German radiosonde of the type designed by K. Sittel in which temperature and humidity were determined by a psychrometer. A pressure element and two thermometers—one covered with muslin—were used to vary the frequency of the signal. An electric motor was used to operate a switch that changed the connec- tions among the sensors and two reference capacitors. These, as well as other parts of the transmitter, are con- tained in a Dewar flask. The entire instrument, except the thermometers, fits into a cylindrical aluminum case, 30 cm long and 8.4 cm in diameter. The thermometers are in an attached triple shield, consisting of an outer oval cylinder. (Instrument in NMAH collection, catalog number MHT 314513; Smithsonian photo 61837-E.) adopted an audio-subcarrier sonde designed by Dutch scientist A. Hauer (Hauer and van Tol, 1954), which minimized the number of moving parts by adopting the same method used by the NPL in 1938, in which the output from three sensors was represented simultaneously. Three oscillators were required, but by 1954 the availability of subminiature vacuum tubes made the weight penalty of this device less than that of the earlier NPL design (Hauer and van Tol, 1954). In the late 1950s, the cardboard case of the U.S. Weather Bureau sonde was replaced by a plastic case, and in 1960 transistorized radio circuits were introduced (Figure 59). By that time, the trans- mitting frequency of United States sondes had been increased, first to 400, and then to 1680 megahertz, chiefly to facilitate direction-finding. In 1960, transponders were introduced in some United States models (Figure 60) (Snyder and Bragaw, 1986:131). The American production model of the 1950s, designated the AN/AMT-4, was a frequency-modulated audio subcarrier type that used amplitude modulation to impress the subcarrier onto the radio-frequency carrier signal. The radio-frequency oscillator was a small "pencil" triode operating in a cavity res- onator and producing about one-half watt of RF power. At 1680 MHz the antenna was small enough to be self-contained in the radiosonde package. The audio subcarrier was produced by a blocking oscillator modulated in the range from 10 to 200 hertz by sensor and reference resistors. These sensors were switched into the modulator circuit in a sequence determined by a NBS baroswitch. Ground equipment automatically re- ceived the signal, demodulated the data, and recorded the sep- arate variables on a chart recorder. Later in the decade, the de- sign evolved further, receiving successive model designations AN/AMT4A, AN/AMT4B, etc., but retaining the same basic design (Anonymous, 1961:102). In the late 1950s and 1960s, additional models appeared with changes in mechanical construction, such as the AN/ AMT11 with a molded plastic case (Figure 59). Other models were designed for special purposes, such as the AN/AMQ-9, a transponding radiosonde (Figure 60). In this type of sonde, a radio receiver picked up signals transmitted from the ground at 403 MHz with pulse modulation at 75 KHz. Meteorological data were added when this signal was received and retransmit- NUMBER 53 63 FIGURE 58.—A U.S. Weather Bureau radiosonde (left), circa 1942, based upon a NBS-Navy design, is housed in a foil-covered cardboard case, 22x11x22 cm. Two flaps open to reveal an electrolytic thermometer and a cylindrical array of hairs that operated a variable resistor. The hygrometer is mounted on white plastic molding and is enclosed in a semicylindrical white shield with louvers at the upper end. Batteries, a transmitter, and a "baroswitch" occupy the remainder of the carton. A similar sonde is shown (below) with the case cut away to allow most of the components to be seen. (Both instruments in NMAH collection, catalog number MHT 312798, above; catalog number MHT 322315, below; Smithsonian photos 61905-B and 61905, respectively.) 64 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY FIGURE 59.—A U.S. Weather Bureau radiosonde, circa 1960. This unit is housed in a white plastic box, 12x11.5 x9 cm, and contains a solid-state transmitter and a baroswitch and has brass arms for supporting a thermistor. This radiosonde model transmitted at 1680 MHz frequency and has been in use since the mid-1950s. There is a provision for attaching a battery box. The instrument is marked "Radiosonde set AN/AMT1 IB." (Instrument in NMAH collection, catalog number MHT 324240; Smithsonian photo 61908-C.) ted back to the ground at 1680 MHz. This system provided a high-data rate from specific locations and altitudes. The most recent developments in meteorological sondes have incorporated the latest advances in electronics and sensor technology. The trend to higher frequencies, ending with the current international allocation at 1680 megahertz, began dur- ing the evolution of radiosonde prototypes in the 1930s. This trend was primarily caused by the need to reduce the size and weight of the devices. Early in the 1940s, a second factor, in- terference, began to influence the operating frequency of ra- diosondes. Substantial problems arose from other users of that portion of the radio band in which radiosondes then operated. Ordnance fuses, homing beacons, and communication devices exacerbated the frequency allocation problem. In 1946, the U.S. Federal Communications Commission, working within the framework of an international treaty, allocated a new fre- quency band, 400 to 406 MHz (71 cm), for radiosonde use. In the United States, the National Bureau of Standards modified its standard design to operate in this range and added an im- provement in modulation technique. The latter was pulse mod- ulation, which provided greater accuracy in recovering meteo- rological data. Radiosonde transmitter wavelength remained in the 71 centimeter range for about a decade before moving to the current allocation, in the microwave region, at 17 centime- ters. The evolution of the audio-subcarrier frequency-modulation radiosonde was important to the development of data teleme- try. The rapid advance of aviation technology in the late 1920s and 1930s required data transmission from aircraft in flight; measurements of pressure, temperature, electrical current, etc., relating to performance. Until World War II, these quantities were recorded on board the aircraft, using instrumentation similar to early balloon meteorographs. Eventually, however, more immediate readout of the data was required, especially in drone or unmanned test flights where there was a possibility that the vehicle would be damaged or destroyed. The radio- sonde provided a well-developed technology that could be adapted for this specific task, transmitting analog electrical and (by transducers) mechanical data over a radio channel from air to ground. There was, however, a problem with straightforward adaptation of radiosonde technology to trans- mission of data from airplanes to the ground—there was too NUMBER 53 65 FIGURE 60.—A U.S. Weather Bureau radiosonde with transponder, circa 1965. The plastic case measures 20x 14 x 15 cm and has a tube projecting from the bottom containing a 1680-MHz transmitter and an antenna. The main case contains space for batteries, a clock-driven switch, and a receiver, which received signals from the ground at 403 MHz, pulse-modulated at 75 KHz, and added audio-frequency pulses from meteorological instruments. The output was used to modulate the 1680-MHz transmitter. (Instrument in NMAH collection, catalog number MHT 322310; Smithsonian photo 61905-C.) much data. Whereas only three variables (temperature, pres- sure, and humidity) were sufficient to satisfy the needs of the meteorologist using weather balloons, many more variables were needed by the aircraft designer. Even when a single vari- able, e.g., pressure, was of interest, aviation engineers wanted to measure it at numerous points. Existing radiosonde designs were not equipped to do this. Some unusual attempts were made to overcome this limitation. For example, in World War II, aviation engineers in the United States mounted a television camera in the cockpit to return data by transmitting pictures of a large panel of instrument indicators. Such methods of multi- ple-channel data transmission were generally unsatisfactory, however, and a more viable technology was needed. The engineering history of radiosonde development, how- ever, contained a straightforward solution to the multiple- channel data-transmission problem. In 1938, the British NPL expanded the frequency-modulation audio-subcarrier radio- sonde to three channels by the simple innovation of using three simultaneous audio frequencies. Each of these audio fre- quencies was frequency modulated independently by separate data (Thomas, 1938). The underlying theory of radio carrier modulation was highly developed by 1938. (Theory showed the possibility of complex modulation schemes far beyond the existing technology.) It was obvious that three or more audio subcarriers could be modulated simultaneously onto a radio carrier frequency without mutual interference. Although the British NPL radiosonde was unsuitable for balloon meteorol- ogy because of its weight and complexity, it was important as the prototype of a class of instruments used later to solve the problem of massive data retrieval in real time from aircraft and rockets. Multichannel data-telemetry systems, using mul- tiple frequencies, evolved in the late 1940s from this radio- sonde technology and became an entirely separate discipline. RESEARCH RADIOSONDES In the 1940s, the introduction of multichannel radioteleme- try made feasible the simultaneous transmission of the output of numerous sensors. The development of the plastic balloon also made it possible to loft payloads to altitudes above 45 ki- lometers. It is noteworthy that these two developments had lit- tle effect on meteorological radiosondes because neither the simultaneous transmission of various measurements nor the achievement of such great altitudes were needed by users of such sondes. These advances in radiotelemetry and balloon technology did, however, have a profound influence on the de- signs of research radiosondes. Methods of multichannel radio transmission via subcarrier modulation were developed during World War II for various military purposes. Also, the development of multichannel ra- dio telemetry was accelerated in the immediate postwar years by the needs of military rocket programs. Multichannel telem- 66 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY etry was quickly adopted for research sondes because the capa- bility to transmit simultaneously the output from several sen- sors was an important consideration in most research applications. For example, cosmic-ray studies usually involved the correlation of atmospheric density with cosmic-ray data. Such correlation was facilitated by obtaining simultaneous out- puts from cosmic-ray detectors and from pressure and tempera- ture sensors. Also, in most investigations involving research ra- diosondes attempts were made to maximize measurement accuracy. Thus, the availability of extra channels to transmit the output of reference standards contributed to the precision of the data. Moreover, the reliability of research sondes was im- proved by a multichannel capability that allowed redundant sensors to be included in payload designs (Borden and Mayo- Wells, 1959:204-268). Postwar developments in balloon technology influenced the design of research radiosondes by making it possible to reach altitudes hitherto unexplored by instrumented probes. The ori- gins of this new balloon technology began with the efforts in the early 1930s of German scientist Erich Regener to measure cosmic rays at very high altitudes, using self-registering bal- loonsondes (Millikan and Bowen, 1926; Regener, 1932). At that time the question of whether "primary" cosmic rays (i.e., the component of such radiation existing outside the earth's at- mosphere) were electromagnetic waves or charged particles was still unresolved. Measurements at the highest altitudes at- tainable could provide information regarding this question, hence, Regener sought design improvements that would allow his balloons to achieve higher altitudes. It was known that ozone weakened rubber and that rubber balloons usually burst below their maximum theoretical alti- tude. Regener surmised that the failure of rubber sounding bal- loons to reach altitudes much greater than 27 kilometers was attributable to the high concentration of ozone between 18 and 27 kilometers. He reasoned that, as balloons passed through the ozone layer, their skin was chemically attacked, causing the balloon to rupture at the weakest point (Regener, 1935). The solution to this problem was to use an ozone-resistant material. In the 1930s, cellophane was the only material available for this purpose, but it lacked flexibility and had undesirable low-temperature characteristics that made it poor material for balloon envelopes. Nevertheless, cellophane balloons were constructed and flown by Regener, probably as early as 1934 . In 1935, an American scientist, T.H. Johnson, lofted a radio- sonde using three cellophane balloons. Johnson's sonde (as dis- cussed in the previous section) was designed to measure cos- mic rays. On its first flight it provided data to about 20 kilometers. Johnson had constructed the balloons with the help of another scientist, Jean Piccard (twin brother of August Pic- card, who made the first manned balloon flight into the strato- sphere in 1931). In the late 1930s, Jean Piccard conducted ad- ditional research on cellophane balloons at the University of Minnesota. The cellophane balloons of the 1930s still did not reach altitudes greater than those attained by rubber balloons because of the unsuitability of cellophane as a balloon material (Pfotzer, 1972:212-226). After World War II, Piccard, in association with Otto Win- zen, an engineer, persuaded the American firm of General Mills to make polyethelene plastic balloons. The Office of Na- val Research (ONR) was sufficiently impressed with the re- search potential of such balloons that they provided the initial funding to General Mills for the development of a plastic bal- loon capable of reaching very high altitudes. As a result of the ONR program, the first "Skyhook" balloon was built and flown in September 1947, reaching an altitude of about 30 kilometers. This new technology produced balloons capable of lofting in- strument payloads weighing many pounds to altitudes above 45 kilometers, which is above 99.9 percent of the atmosphere sur- rounding the earth (Pfotzer, 1972:227-231). During the last four decades, research radiosondes have pro- vided a wealth of information about cosmic rays, atmospheric composition, ultraviolet radiation, and solar flux. Despite the advent of rockets and satellites, balloon-borne research radio- sondes still are used because of their unique advantages, rela- tive to other vehicles, primarily low cost and the ability to re- main at a given altitude for long periods. Thus, even in the space age, research radiosondes are still a useful investigative tool. Some aspects of rocket flight even depend upon the radio- sonde. Winds aloft between about 15 and 45 kilometers must be measured with some precision just before the launch of most types of rocket to insure safe travel through this altitude, where aerodynamic stresses on the vehicle are greatest. It is costly (and for the upper part of this altitude range impossible) to fly airplanes at that altitude to obtain measurements. Conse- quently, radiosondes are flown prior to virtually all rocket flights to measure wind velocities and other atmospheric pa- rameters that could affect the rocket's ascent. Often launch countdowns are put on "hold" because of adverse radiosonde data. 8. Conclusion The improvement of agriculture, a paramount economic glo- bal concern, was the main objective of meteorology through the nineteenth century. By the end of that century, the art of weather forecasting had made significant progress. Moving air masses were recognized as the carriers of local weather, and data about wind speed and direction, pressure, temperature, and water content (humidity), in one location, were used to forecast the weather elsewhere. It was an imperfect art, and the farmers continued to have more confidence in the Farmer's Almanac. In the early twentieth century, the emerging technology of aviation provided an additional and urgent need for more accu- rate weather forecasting. This need assured greater material support for meteorologists, including government financing, because of the obvious military value of aircraft. The traditional claim of the meteorologist, that the weakness of this science stemmed from insufficient data, was reinforced when the needs of aviation emphasized the limited utility of measurements made from ground level. Unmanned kites and captive balloons were used for decades, but these were limited to relatively low altitude and could not be used in many loca- tions for safety reasons. Observation stations on the tops of mountains and hills provided some limited data. Observations also were made from airplanes, but this was a "chicken and egg" situation—the flights could not be widespread and safe without first having the results of the observations they would be making. As late as 1920, there were very few ways of col- lecting data from altitudes above a few hundred meters, and no means existed for collecting such data across wide areas in an organized fashion. The advent of heavier-than-air-flying ma- chines worsened the problem because it required the collection of meteorological data at high altitudes on a timely basis. The available technology did not offer the means to satisfy these re- quirements until the invention of the wireless telegraph (radio). Marconi's transmission of radio signals across the Atlantic, in 1901, provided a solution to the problem of timely collection of meteorological data, which was used three decades later. The speed with which these components were put together to create the radiosonde was remarkable. It hardly involved any advance in science, but rather the appearance of an urgent need, aviation. Nor was technological innovation very conspicuous. Important factors were the concurrent emergence of a popular hobby—amateur radio—and the commercial availability of vacuum tubes and other components. Finally, there was that never-failing stimulus to development—military application. All this seems obvious now. The successful development of the radiosonde tends to hide the complexity of the research by which it was produced. The radiosonde represented one of the first comprehensive applications of radio to a purpose other than communication. This use of radio technology to meet a re- mote-sensing requirement was not unique, but it did set a pre- cedent that accounts for some aspects of the history of subse- quent instrumentation. There also were important consequences for the then back- ward science of meteorology. Thanks to data provided by the radiosonde from a range of altitudes, synoptic weather maps were vastly improved. These data, in turn, provided the means to generate timely, accurate forecasts based upon the motion and evolution of the air masses. As radiosonde technology and data collection improved in the 1940s, scientific meteorology finally matured. Deterministic modeling of the atmosphere, based upon the physical laws of gas dynamics and heat transfer, although appropriate, had long been considered futile because measurements on a sufficiently large scale and at high enough resolution to establish initial conditions for the equations could not be made. The availability of large amounts of data from ra- diosondes and the emergence of electronic computers in the late 1940s helped to forge a new branch of science in practical mod- eling of the atmosphere. Modeling, together with skillful inter- pretation of data, has promoted a steady improvement in our understanding of the atmosphere and its dynamics. From the perspective of the history of technology, the radio- sonde also served as an important link in the development of multichannel radio transmission, using subcarrier modulation. Because this became the basis of modern analog radiotelemetry systems, the development of the radiosonde represented an im- portant phase in the history of telemetry. Ground observations continue to be important, but they pro- vide only a small portion of the necessary data. Atmospheric data collection by satellite, starting with the NIMBUS and TIROS programs in the 1960s and 1970s, is now supplement- ing radiosonde data. Satellites have not yet completely replaced the radiosonde, however, because of the difficulty of obtaining vertical measurements from the overhead view of a satellite. Techniques to accomplish high-resolution vertical profiling of atmospheric temperature by infrared radiotelemetry have been refined in successive generations of TIROS (now called NOAA) and other weather satellites. Additional parameters, such as wind vectors and barometric pressure, also can be de- termined from satellites, but to a lesser degree of precision than by radiosondes. Nevertheless, it is likely that, by some time in the twenty-first century, the majority of radiosonde data collec- tion will be replaced by satellite-derived measurements, and ra- diosonde deployment will be relegated to a minor, niche tech- nique. 67 III. Catalog of Upper-Atmospheric Telemetering Probes and Related Artifacts in the National Museum of American History, Smithsonian Institution Robert P. Multhauf and George A. Norton, Jr. Catalog number Accession number Figure number Type Identification Date Donor 312798 162632 58 USWB radiosonde 1942 Plaskon Company Marked "Sonde Track... Mfg. Washington Institute of Technology. USWB 1942." This foil-covered cardboard box, 22x21.6x12 cm, contains sensing elements for measuring temperature (electrolytic), humidity (hair hygrometer), a baroswitch, and a transmit- ter. Received from the Plaskon Company, 2121 Sylvan Ave., Toledo Ohio, in March 1944. 0 162632 54 NBS-U.S. Navy hair hygrometer 1942 Plaskon Company Marked "WIT." Aneroid element, 9x5.5x20.3 cm. This is the same type used in item 312798 and was received at the same time from the same donor. 1 162632 52 NBS-U.S. Navy baroswitch 1942 Plaskon Company Marked "WIT." Aneroid barometer, 13x9x6 cm. This type was used to control the selector switch in item 312798, and it was re- ceived at the same time from the same donor. See text, pages 50-54. 2 162632 NBS-U.S. Navy record of data 1942 Plaskon Company Photostatic copy of data, pressure, temperature, and humidity, transmitted by item 312798 at Washington National Airport, 22 November 1941. Received at the same time from the same donor as item 312798. 71x11.8 cm. 3 162632 NBS-U.S. Navy transmitter 1942 Plaskon Company Radio transmitter in plastic case, 7.7x9x5.2 cm. This type was used in item 312798 and was received at the same time from the same donor. 0 188571 35 Duckert radiosonde ca. 1932 NBS Unmarked except for "NBS no. 0203." This clear plastic cover and aluminum frame, 19x9x31 cm, contains a vacuum tube within which are located the transmitter, a toothed wheel switch actuated by a Bourdon tube (also the pressure element), and a bi- metallic temperature element. Received from NBS in September 1950. For information on Duckert's radiosondes, see text, pages 34-40, and E. Kleinschmidt, 1935:282. 1 188571 Diamond et al. radiosonde 1936 NBS Unmarked balsawood box with externally mounted photo tubes, overall 22x 15x 17 cm. The box contains a transmitter (three vac- uum tubes), a barometric switch (aneroid), and temperature (sulphuric acid resistor) and humidity (hair hygrometer) measuring el- ements. Identified by the donor as the first experimental radiosonde developed by H. Diamond, W.S. Hinman, Jr., and F.W. Dun- more at NBS and first used on 25 December 1936. A mark on the frame indicates that it was made by Friez & Sons of Baltimore. Received from NBS in September 1950. 2 188571 - NBS transmitter - NBS Unmarked box containing parts relating to radio transmission. This experimental apparatus was received with item 313528. 314513 204101 57 Vaisala, Schulze radiosonde 1942 K.M.Perry Marked "493." This aluminum case, 11 x 15x30 cm, with its lower section painted black, contains elements for measuring tem- perature (a ceramic, temperature sensitive dielectric), pressure (aneroid barometer controlling a condenser), and humidity (wet and dry bulb). Identified by the donor as being of the Vaisala type, improved by Schulze at the "Marine Observatorium" at Grief- swald (Germany) in 1942-1944. Received in November 1954. (See R.K. Sittle and E. Menzer, 1950:341.) 68 NUMBER 53 69 Catalog Accession Figure number number number Type Identification Date Donor 314514 204101 - unknown radiosonde K.M.Perry Marked "61426," and bearing a stamp indicating a German origin. The brown fibre case, 10.4x14.3x20 cm, has metal ends and contains bimetallic thermometric and aneroid barometric elements. Received in November 1954. 314677 209614 USWB radiosonde 1955 USWB Marked "Northeastern Engineering, Inc.," with USWB serial no. 509401. The white cardboard carton (cut away), 20x23x 12 cm, contains a variable audio-frequency transmitter, a thermistor, and an electric hygrometer. This instrument was used on 13 July 1955 and was found, after ascending to a height of 32 kilometers, at Redwater Alta, Saskatchewan, Canada. 316262 224564 32 Moltchanoff radiosonde ca. 1935 USWB Marked "39097." The brown cardboard case, 7x27x33 cm, contains elements for measuring temperature (bimetallic), pressure (Bourdon tube), and humidity (hair hygrometer). It incorporates a rotating contact of the Olland type. The transmitter is missing. For information on Moltchanoff's instruments, see text, pages 33 and 34, and Lange, 1935:286. Received in March 1959. 317728 231230 - USWB radiosonde 1960 USWB Marked "Ray Sonde Bendix." This blue and white cardboard box, 11.4x18x20.3 cm, contains elements for measuring tempera- ture (electrolytic) and humidity (electrolytic), and a barometric switch. Received in June 1960 and identified by the donor as a contemporary instrument. 319744 239017 NBS radiosonde 1940 E.L. Reilly Marked "Sonde Track. Mfg. in USA by Washington Institute of Technology," and the inspection date 9/25/40. The two nested cardboard boxes, 11x25.4x22 cm, enclose elements for measuring temperature (electrolytic), pressure (aneroid), and humidity (hair hygrometer), all reportedly developed at the NBS. Identified by the donor as WIT Sonde Track no. 1, produced on 25 Sep- tember 1940, flown at Oklahoma City on 1 October, and recovered 13 October at Presque, Oklahoma. This is an example of the NBS variable audio-frequency radiosonde. Received November 1961. 319824 240347 - (British) radiosonde 71937 USWB Marked "D698." Elements for measuring temperature (bimetallic), pressure (aneroid), and humidity (goldbeater's skin) project from the sides of a cardboard cylinder 13 cm in diameter by 14 cm long. Incomplete. Identified by the donor as a British instru- ment. Received in January 1962. (See Lange, 1937.) 319825 240347 43 Vaisala radiosonde ca. 1950 USWB Marked "180140." The aluminum box, 9x31x18 cm, contains elements for measuring temperature (bimetallic), pressure (aner- oid), and humidity (hair hygrometer), and incorporates a windmill-operated selector switch. (See text, pages 46^48, and item 314513.) Received January 1962 and dated by the donor as being from about 1950. 319826 240347 55 - Japanese radiosonde ca. 1960 USWB Marked "CMO-S50L" and stamped "DNPA No. 4552, Japan. Type Approval, Radio Regulatory Commission" (with additional label in Japanese). This white box of corrugated paper, 28x 12x 16 cm, contains elements for measuring temperature (two mer- cury and one bimetallic thermometers), pressure (aneroid), and humidity (hair hygrometer). Received in January 1962. 319827 240347 30 Bureau radiosonde ca. 1932 USWB Marked "Constructions Mecaniques et Electriques P. de Presale, 104-106, Rue Oberkampf, Paris (XI) Type PTU (Sereal) 3567." The aluminum foil covered box, 27x 16x 14 cm, contains a transmitter (vacuum tube and batteries missing) and elements for mea- suring temperature (bimetallic), pressure (bourdon tube), and humidity (hair hygrometer). Temperature and pressure were indi- cated by interruptions in the transmitting signal. For information on Bureau's instruments see text, pages 31-36. Received in Jan- uary 1962. 319828 240347 - Cal. Inst. Tech. radiosonde ca. 1935 USWB Unmarked, but tagged by donor "Composite California Institute of Technology type." This instrument, 10x15x24 cm, contains elements for measuring temperature (bimetallic), pressure (aneroid), and humidity (hair hygrometer), and a clock-driven selector switch. Received January 1962. 70 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY Catalog Accession Figure number number number Type Identification Date Donor 319829 240347 - Lang radiosonde sensing elements - USWB Marked "10818." The metal frame, 12.7x 10x5.7 cm, contains elements for measuring temperature (bimetallic) and pressure (an- eroid), and a clock-driven selector switch. Received January 1962 and identified by the donor. 0 240347 Astin, Curtis radiosonde sensing elements - USWB Unmarked. The metal frame, 19.4x8x9 cm, contains elements for measuring temperature (bimetallic), pressure (aneroid), and humidity (goldbeater's skin), plus an electrically driven switching disk. Received January 1962. (See also cat. no. 319833.) 1 240347 Blue Hill radiosonde ca. 1936 USWB Marked "RMF 84," and tagged by donor "original Blue Hill." The sensing elements, mounted on a 10x20x6 cm board, are for measuring temperature (bimetallic), pressure (aneroid [missing]), and humidity (hair hygrometer). Transmission was controlled by a clock-driven spiral. Transmitter and battery box are missing. Received January 1962. (See Lange, 1937:110.) 2 240347 - Friez radiosonde ca. 1936 USWB Unmarked, but tagged "Friez" (of Baltimore, assumed to be the maker). The octagonal wooden box, 16x 17x20 cm, contains ele- ments for measuring temperature (bimetallic), pressure (aneroid), and humidity (hair hygrometer), plus a transmitter and a clock control for transmission. Received January 1962. 3 240347 Astin, Curtis radiosonde ca. 1938 USWB Unmarked. The cylindrical tube, 15 cm in diameter and 34 cm long, is made of corrugated paper, with a reflective painted surface. The tube houses the transmitter and clock. Projecting above is a housing containing sensing elements for measuring pressure (an- eroid) and humidity (goldbeater's skin)—the temperature element is missing (but see cat. no. 319830). The unit was identified by the donor as an experimental instrument from the late 1930s. A.V. Astin was at the National Bureau of Standards from 1930 and was its director from 1952 to 1969. L.F. Curtis was at the Bureau from 1926. Received January 1962. 322051 246537 ARCAS rocket sonde unk. ARC Marked "Radiosonde; Atlantic Research Corp., Alexandria, Va. Radiosonde AN/DMQ6 (XE-3)." The radiosonde, 11 cm diam., 47 cm long, is an ARCAS high-altitude sounding rocket system. It contains four thermistors, which, with its rocket and aluminum parachute, measures a total length of 2.3 meters. Received January 1963. 322308 247582 Simmonds radiosonde ca. 1942 USWB Marked "Simmonds chronometric radiosonde. P.B.I21." The pressed paper container, 24x 18x 17 cm, contains a blocking oscilla- tor, an FM transmitter, and a clock-driven switch. Sensors for measuring temperature (bimetallic), pressure (aneroid), and humid- ity (hair hydrometer) are attached. The device was described by the donor as experimental. Received March 1963. 0 247582 60 U.S. Army radiosonde ca. 1956 USWB Marked "Radiosonde receiver AN/AMQ-9 (XE-3) Serial 201. File no. 360 PH-55-91 (2229) Standard Coil Products Inc. US." A white plastic case, 15x20x 14.6 cm, houses an FM blocking oscillator transmitter, a clock-driven selector switch, and sensing el- ements for measuring temperature (thermistor) and humidity (lithium chloride). The unit was identified by the donor as an instru- ment developed for the military establishment about 1956. Received March 1963. 1 247582 Lang radiosonde - USWB Mkd. Marked "W.S.E. 2 Gerat Nr. 124 75A Anf. Zeichen Ln 2879." This 16x31 x34 cm box has an attached plastic cylinder that con- tains a transmitter. The sensing elements are a bimetallic thermometer and an aneroid barometer; the element for measuring hu- midity is missing. It was identified by the donor as the Lang-Reichsamt fur Wetterdienst radiosonde. Received March 1963. (For information about the Lang radiosonde, see text, page 42.) 2 247582 (German) radiosonde ca. 1936 USWB Numbered "1054" and "712U"; otherwise unmarked. The aluminum-painted cardboard box, 8x12x17 cm, contains a bimetallic temperature-measuring element, an aneroid barometer, an electric motor with a star selector wheel, and a transmitter. It was iden- tified by the donor as a "prewar" German instrument. Received March 1963. 0 NUMBER 53 71 Catalog Accession Figure Number Number Number Type Identification Date Donor 0 247582 46 British (Kew) radiosonde - USWB Marked "Whiteley Electrical Radio Co. Ltd., Mandfield Notts. England," and "Mark II Met. Ref. No. 280. II." The black molded-fi- ber cylinder, 22.5 cm long x 14 cm in diameter, contains a pulse modulation transmitter, sensing elements for measuring temperature (bimetallic), pressure (aneroid), and humidity (goldbeater's skin), and a windmill selector switch. Received March 1963. 1 247582 Diammond, Hinman radiosonde ca. 1943 USWB Marked "Diamond-Hinman Raysonde. Mfg. by Julien P. Friez & Sons, Baltimore, Md." The corrugated box, 21 x 12x30 cm, is covered with white paper and contains an FM blocking oscillator transmitter, a barometric selector switch, and sensing elements for measuring temperature (thermistor), pressure (aneroid), and humidity (electrolytic). Received March 1963. 2 247582 58 USWB radiosonde ca. 1942 USWB Marked "Sonde Track Washington Institute of Technology, Washington, D.C. Serial no. 214029." This corrugated, foil-covered box, 11x22x22 cm, contains an FM blocking oscillator transmitter, an aneroid baroswitch, and elements for measuring tempera- ture (electrolytic) and humidity (hair hygrometer). Received March 1963. 3 247582 "Longine" radiosonde USWB Unmarked, but tagged "Longine." The aluminum frame, 18x 10x36 cm, houses a transmitter (FM blocking oscillator type), a clockwork selector, and elements for measuring pressure (aneroid), temperature (bimetallic), and humidity (hair hygrometer). Re- ceived March 1963. 322320 247582 Ray Sonde radiosonde ca. 1939 USWB Marked "Olland System Ray Sonde. Julien P. Friez & Sons, Baltimore." The corrugated, foil-covered box, 22x 11 x21 cm, houses a transmitter (FM blocking oscillator type), an Olland switching system, and sensing elements for measuring pressure (aneroid) and humidity (hair hygrometer). There is no evidence of a temperature element. Received March 1963. 0 247582 Diamond, Hinman radiosonde ca. 1939 USWB Marked "Ray Sonde. Diamond-Hinman. Mfg. by Julien P. Friez & Sons of Baltimore," and stamped "Serial no. w2982" and "July 1, 1939." The foil-covered corrugated box, 23x11x21 cm, contains a transmitter (FM blocking oscillator type), a barometric switch, and sensing elements for measuring temperature (electrolytic), pressure (aneroid), and humidity (hair hygrometer). 1 247582 - Friez radiosonde ca. 1940 USWB Marked "Meteorograph. Friez Baltimore." The foil-covered corrugated box, 11 x28x21 cm, contains a transmitter (FM blocking oscillator type) and sensing elements for measuring temperature (electrolytic), pressure (aneroid), and humidity (hair hygrome- ter). Received March 1963. 322325 247582 48 Blue Hill radiosonde ca. 1936 USWB Marked "Fieber Instrument Co., Cambridge, Mass." The balsa wood box, 18x 10x7 cm, contains a transmitter (FM blocking os- cillator type), and a clockwork selector switch. Sensing elements for measuring temperature (bimetallic) and humidity (hair hy- grometer) are attached to the box. For information about Blue Hill radiosondes, see text, pages 50, 51 and item 319831.) Re- ceived March 1963. 0 254096 - U.S. Navy radiosonde USWB Marked "Radiosonde AN/AMT11... Molded Insulation Co., Serial 1675." The molded white plastic box, 12.7x 13.3x9 cm, con- tains a vacuum-tube transmitter, a barometric switch, and sensing elements for measuring temperature (thermistor) and pressure (aneroid). The humidity sensor is missing. This device was identified by the donor as having been made for the U.S. Navy. Re- ceived April 1964. 1 254096 - (Japanese) radiosonde ca. 1960 USWB Marked "CMO-S50L. Serial 4352." The white corrugated box, 12.4x29x 19 cm, contains a transmitter, a motor-driven drum se- lector switch, and sensing elements for measuring temperature (metallic expansion, plus a fixture for two mercury thermometers, which are missing), pressure (aneroid), and humidity (hair hygrometer). The unit was dated by the donor. Received April 1964. 0 72 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY Catalog Accession Figure Number Number Number Type Identification Date Donor 0 254096 USWB radiosonde 1963 USWB Marked "USWB-Molded Insulation Co., Phila. Serial 37-00006," and stamped "Feb. 8, 1963." This corrugated, paper-covered box, 13x 16x 12 cm, contains a transmitter, a barometric switch, and sensing elements for measuring temperature (thermistor) and pressure (aneroid). The humidity sensor is missing. Received April 1964. 1 254096 (Japanese) radiosonde 1945 USWB Marked in Japanese, and with serial no. "020918." The white cylindrical plastic case, 16.5 cm longxll cm in diameter, contains sensing elements for measuring temperature (mercury thermometer [broken]), pressure (aneroid), and humidity (hair hygrome- ter), each with its own vacuum tube, circuitry, and antenna. The device was dated by the donor. Received April 1964. 2 254096 59 U.S. Navy radiosonde USWB Marked "Radiosonde set AN/AMT 11 B... Molded Insulation Co." The white plastic box, 12.4x13.7x10 cm, contains a solid- state transmitter, a barometric switch, and a sensing element for measuring temperature (thermistor). The humidity element is missing. This device was identified by the donor as one that was intended to be expendable. Received April 1964. 3 254096 56 Gaw (German) radiosonde 1963 USWB Marked "Transport Aufstieg. 5821." This foil-covered cardboard box, 1 Ox 10x20 cm, contains a transmitter, a motor-driven selec- tor switch, and sensing elements for measuring temperature (bimetallic), pressure (aneroid), and humidity (hair hygrometer). Re- ceived April 1964. (For further information, see text, pages 60, 61.) 4 254096 - USWB thermistor ca. 1960 USWB Unmarked. This temperature measuring element consists of a wire coated with lead monoxide that is housed within a glass tube. This improvement in temperature measurement was developed at Johns Hopkins University about 1946 under contract with the USWB. (See Brasefield, 1948:145-151.) 327975 254096 50 NBS-U.S. Navy thermometer/hygrometer The thermistor and electrolytic hygrometer assembly are mounted on a bakelite board. Overall dimensions are 11 x 12.7x74 cm. It was probably experimental. Received October 1966. Uncataloged - - USWB temperature element 1957 - Unmarked but labeled "Temperature element ML-419/AMT-4. Molded Insulation Company, US," and stamped "Feb. 1957." This wire element in a glass vial was probably an experimental thermistor from the USWB, although the source is unknown. Uncataloged - USWB hygristor 1959 The small can is stamped "USWB hygristor, electric hygrometer element. Model no. 524318-3. Bendix Aviation Corp., Friez In- strument Division, Baltimore, Md." and is dated 16 July 1959. It was probably from the USWB. Notes 1. The hair hygrometer measures relative humidity in terms of dimensional changes on one or more human hairs stretched on a frame. It was invented in 1793 by French scientist Horace de Saussure, and by 1820 it had become widely used in meteorological research. By the 1890s, hair hygrometers were highly developed as sensors for humidity. 2. Charles' balloon was filled with hydrogen. An alternative gas-fill appeared in the early 1800s when "coal gas" (chiefly a mixture of hydrogen and methane) began to be produced by municipal gas works for fueling the street lamps of many European cities. Coal gas was first used in a balloon in 1821, and although it has only half the lifting power of hydrogen its much lower cost and ready availability made it the gas of choice for balloon use. The unmanned balloons used by Hermite and other meteorological investigators generally were filled with coal gas. 3. Letter from Millikan to W. Blair, Millikan Papers, 28 November 1922, Roll 23, Folder 23.1. Library of the National Museum of American History, Smithsonian Institution. 4. The concept of entrainment in a complex system actually predates Olland, Bain, telemetry, and computers by at least a billion years, having been worked out by Darwinian evolution on living organisms in response to the survival advantage of photoperiodic behavior. Virtually all living organisms exhibit a cyclic pattern of motor or biochemical behavior, which is entrained to the solar photoperiod. If such an organism is studied under conditions where environmental light and other geophysical parameters are held constant, the behavior cycle will usually continue at a period different from the 24-hour solar period. Under natural conditions the solar photoperiod resets the behavior cycle once per period, which synchronizes the sender (sun) with the receiver (organism). 5. Radio wavelength is an important parameter. It determines, to a large degree, the size and weight of transmitter components as well as the antenna. Also, before World War I, radio-band usage was not regulated by govern- ments or international treaty and, being in the same range as other users, there was considerable chance for interference. The earliest radiosondes of Blair and of Bureau adapted radio frequency transmission circuits directly from the technology of contemporary radio telegraphy and radio telephony. Blair's first transmitter used a wavelength of 125 meters (2.3 MHz) in 1924 (Blair and Lewis, 1931:1532) and Bureau's 1929 radiosonde operated at about 60 meters (4.8 MHz) (Bureau, 1931). The early investigators were aware of the potential advantages of shorter wavelengths, but they faced substantial technological barriers to using short waves. Transmitter circuitry and vacuum tubes that were small and would operate reliably at wavelengths below about 20 meters were not well developed in the mid-1930s. In spite of this, the advantages were great enough that, start- ing with Blair at 125 meters, there was an evolution toward the shorter wave- lengths that continued though the 1950s to 17 centimeters (1680 MHz) where current radiosondes operate. This evolution can be traced through the first two decades of radiosonde development. First, Bureau settled on a shorter wave- length than Blair, 60 meters, and was able to shorten his antenna considerably. The United States program, starting in 1935 at the National Bureau of Stan- dards (NBS), began at 4 meters (70 MHz), while Vaisala in Finland was using 11 meters (26 MHz) at about the same time. Diamond, at NBS, then progressed to wavelengths below 2 meters for the "Explorer" stratosphere balloon project in 1935 (Snyder and Bragaw, 1986:102-104). A rapid decrease in wavelength followed as the technology of vacuum tubes and radio-frequency circuits for higher frequency advanced. The first produc- tion models of the NBS radiosonde were standardized in the 4-meter range because that wave length was low enough to achieve almost all of the advan- tages possible in smaller size and weight. It did not, however, place the radio- sonde signals in a suitable part of the radio band to solve all of the problems resulting from signal interference. Crowding of the radio spectrum from over- lapping transmitter frequencies had begun during the rapid expansion of radio- telegraphy before World War I, and the problem was serious enough to merit the formation of regular international conferences to create cooperative alloca- tions of frequency bands for international use. The International Radiotele- graph Conferences were held in 1912, 1927, and 1932. The International Telecommunication Union (ITU), which survives to the present day, evolved during the 1932 meeting to become the principal international forum for these allocations. Radiosondes, which were undergoing their own evolution in the same period, were almost from their inception subject to the advantages as well as disadvantages of these allocations. The disadvantages lay in sharing of their spectrum with other services and consequent interference. The interference played a significant part in an evolution toward operating frequencies above 100 MHz, where there was less usage, and pressure on the ITU for protected frequency allocations during the 1940s. 6. The transition from discharge-excited radiation to the CW vacuum-rube oscillator may have occurred as early as 1921. A reference appears in a later historical review that is difficult to interpret. Duckert (1932:74) reported that "Already in 1921 Herath had tried to use an audio-oscillator made by Eichorn as transmitter in his balloon." The purpose, as with Herath's spark transmitter, was to track the balloon and to determine wind speed and direction. The success was reported as marginal, but the real curiosity is in the term "audio- oscillator." If this is taken to mean a vacuum tube oscillator operating in the audio range, say 10 to 20 KHz, then it is truly the first use of a vacuum-tube CW oscillator transmitter from a balloon. 7. Moltchanoffs assertion to have flown the first radiosonde is quoted in Beelitz (1954:69). This claim was echoed in some contemporary scientific journals, e.g., see the commentary on Moltchanoffs radiosonde by J.F.H. (1932:1006) and in some modern historical works, e.g., see Pfotzer (1972: 242). Moreover, historians who recognized Bureau's priority over Moltchanoff, such as Middleton (1969a:329), appeared to be unaware of Blair's priority over Bureau. 8. Mechanical sensors inherited from the balloonsonde were employed in the first radiosondes: the bimetallic thermometer, aneroid or Bourdon-tube barometer, and hair or goldbeater's skin hygrometer. The barometers suffered from hysteresis and tended to be temperature sensitive. This temperature sensitivity could be greatly reduced by careful material selection, and hysteresis was not a serious problem because changes in the pressure measured were always in the same direction. On the other hand, the mechanical thermometers and hygrometers had more serious deficiencies, especially at higher altitudes. Both had slow response, and the thermometers suffered from radiation errors. The problem of sluggish response was especially acute for the hygrometers, making them virtually useless at high altitude. Because these problems seemed unsolvable within the constraints of the mechanical approach, research on electrical sensors had begun in the 1920s. An electrical sensor was employed in the Blair sonde of 1924 and, later in the Duckert sonde, which incorporated a thermometer consisting of a capacitor with a temperature-sensitive dielectric material. By the mid-1930s, researchers had produced temperature-dependent electrical resistors with a fast response time coupled with small size to reduce radiation error; however, not all types of existing radiosonde designs could be readily adapted to use this superior sen- sor. The possibility of an electrical hygrometer had been explored, but not sat- isfactorily resolved. By the time work began on the NBS-Navy radiosonde program in 1936, the deficiencies of mechanical temperature and humidity sen- sors generally were perceived as significant factors in limiting the ability of radiosondes to provide data at high altitudes. The NBS team was very aware of the potential superiority of electrical sensors. 73 74 SMITHSONIAN STUDIES IN HISTORY AND TECHNOLOGY 9. The radiosonde designed by Dymond at Kew Observatory was adopted by Britain's Meteorological Office in 1940 and designated the Kew Mkl. In this sonde, inductances (that were varied by an aneroid barometer, goldbeater's skin hygrometer, and bimetallic thermometer) were sequentially cycled into the LC circuit of a single audio-frequency subcarrier oscillator, which modulated the transmitter in accord with sensor output. The Mkl A was introduced in 1943. In this model the flat temperature sensor was altered to a cylindrical shape, and circuit improvements were made. In 1945, the Mk2 appeared, which incorporated parts that could be manufactured by less costly methods. In 1952, the Mk2B, featuring improved temperature compensation of the RF and audio subcarrier oscillator stages of the transmitter, was introduced. This model remained in service for the next 25 years, although the ground-based equipment was modernized in 1961 by automating the manual method of balancing the sonde's signal against a reference standard. In 1977, the Kew Mk3 went into service (and is still in use today). This model employs a crystal-controlled transmitter to eliminate drift and contains transistorized radio frequency and subcarrier stages mounted on a single printed circuit board. The sonde continues to use an aneroid barometer and goldbeater's skin hygrometer, but the bimetallic thermometer used in earlier models was replaced by a resistance-wire type. The computerized ground-based equipment developed for the Mk3 performs pre-flight checks, flight computations, and data reduction. 10. According to Churchill (1948:336) German annual expenditure on arms was equivalent to $4,500,000,000; a rate triple that of Britain and France combined. 11. Apart from design specifications, such as the 1936 specifications for the NBS-Navy radiosonde, there is little published information on the accuracy of various radiosonde designs that appeared in the 1930s and early 1940s. Presumably the accuracy figures for the European designs were similar to those of the NBS-Navy type (i.e., temperature to within one degree over the range from plus 40 to minus 75 degrees centigrade; pressure to within one millibar from 150 to 1050 millibars; and humidity to within three percent from 0 to 100 percent relative humidity). There is little published data on the in- flight accuracy of production versions of the various radiosonde designs. 12. The electrical sensor was particularly helpful in the technique of audio-frequency modulation that Diamond and his NBS associates decided to use. An audio oscillator circuit was employed that produced a frequency inversely proportional to a vacuum-tube grid voltage. The grid voltage could then be varied by electrical sensors. The variable audio-frequency subcarrier then modulated the amplitude of the emitted radio wave. This approach was especially advantageous because it provided continuous emission on a constant carrier frequency that aided radio direction-finding and because it facilitated the use of electrical sensors based upon resistance. The required resistive temperature sensor was already available, but, because no suitable resistive hygrometer existed, a development program was initiated concurrently at NBS for application to the radiosonde project. 11. Appendix Selected United States Patents Related to Radiosondes Patent Invention/Purpose 2,230,779 Pyrometer potentiometer 2,277,692 Measuring cloud height and thickness 2,323,317 Altitude measuring 2,347,160 Radiometeorography/Transmitting apparatus 2,347,345 Radiometeorography/Transmitting apparatus 2,352,578 Radiometeorograph transmitter 2,354,086 Radiometeorograph/Receiving relay apparatus 2,355,739 Meteorological apparatus 2,381,009 Chronometric radiosonde system 2,396,955 Radiosonde 2,409,155 Radio transmission N.E. Gibbs apparatus 2,418,836 Remote recording system 2,425,537 Calibration of radiosonde 2,454,320 Hygrometer and switch for radiosondes 2,467,400 Automatic transmission of data 2,468,703 Chronometric electronic radiosonde system 2,500,186 Meterological telemetering system 2,509,215 Radiosonde (Dept. of War) L. Hillman 2,547,009 Telemetering system (Dept. of War) 2,591,600 Radiosonde calibration 2,613,347 Modulator for radiosonde apparatus 2,689,342 Pressure operated switching device 2,701,354 Airborne radiosonde apparatus Patentee Filed Awarded M.J. Johnson May 1937 Feb 1941 F.W Dunmore May 1938 Mar 1942 F.W Dunmore Feb 1940 Jul 1943 C.F.Wallace Feb 1940 Apr 1944 C.F.Wallace Dec 1941 Apr 1944 C.F.Wallace Dec 1941 Jun 1944 J.R. Mackay Sep 1941 Jul 1944 I.E. McCabe Oct 1942 Aug 1944 J.A. Siderman Aug 1944 Aug 1945 K.O. Lange Sep 1942 Mar 1946 E.L. Schellens Sep 1942 Oct 1946 B.K. Hawkes Aug 1943 Apr 1947 J. Hornbostel Aug 1943 Aug 1947 J. Hornbostel May 1943 Nov 1948 PR. Murray Aug 1945 Apr 1949 CM. Hammel Sep 1946 Apr 1949 M.Kline Apr 1945 Mar 1950 L.S.Craig Sep 1944 May 1950 W.D. Huston, J. Grady Oct 1944 Apr 1951 C.B. Pear Nov 1948 Apr 1952 W. Todd Sep 1945 Oct 1952 PR. Goudy Sep 1950 Sep 1954 A.E.Bennett Dec 1946 Feb 1955 75 Literature Cited Abbot, C, F. Fowle, and L. Aldrich 1915. New Evidence on the Intensity of Solar Radiation Outside the At- mosphere. Smithsonian Miscellaneous Collections, 45:1-55. Anonymous 1928. The Radio Industry. New York: A. W. Shaw and Co. 1961. Handbook of Meteorological Instruments, Part II. 551 pages. Lon- don: Her Majesty's Stationary Office. Armstrong, E.H. 1917. Some Recent Developments in Regenerative Circuits. Proceedings of the Institute of Radio Engineers, 10(8):244. Beelitz, P. 1954. Radiosonden. 136 pages. Berlin: VEB-Verlag. Benade, J., and R. Doan 1935. Apparatus for Transmitting Cosmic Ray Data from the Stratosphere. Physical Review, 47:198. Blair, W. 1937. Some Relationships between Radio and Meteorology. Transcription of lecture of 7 April at New York University (in the Blair Papers, currently in the custody of W. Blair, Jr., of Monmouth Beach, New Jersey). Blair, W., and H. Lewis 1931. Radio Tracking of Meteorological Balloons. Proceedings of the In- stitute of Radio Engineers, 19:1531 -1560. Blake, G. 1928. History of Radio Telegraphy and Telephony. 403 pages. London: Chapman & Hall Ltd. Borden, P., and N. Mayo-Wells 1959. Telemetering Systems. 341 pages. New York: Reinhold Publishing. Borden, P., and G. Thynell 1948. Principles and Methods of Telemetering. 231 pages. New York: Re- inhold Publishing Corp. Brasefield, C. 1948. Measurement of Air Temperature in the Presence of Solar Radia- tion. Journal of Meteorology, 5:147-151. Bureau, R. 1929. Sondages de pression et de temperature par radiotelegraphie. Comptes Rendus, 188:1565-1566. 1931. Sondages par ballons munis d'un emetteur radioelectrique. La Meteorologie, 7:304-320. Carmichael, H., and E. Dymond 1939. High Altitude Cosmic Radiation Measurements near the North Geo- magnetic Pole. Proceedings of the Royal Society, London, A171: 321-344. Chaffee, E.L. 1933. Theory of Thermionic Vacuum Tubes. 652 pages. New York: McGraw-Hill. Churchill, W.S. 1948. 77;