HSNS3803_03 7/10/08 5:58 PM Page 405 E D WA R D J O N E S - I M H OT E P ∗ Icons and Electronics A B S T R ACT In the late 1950s, a wide-ranging debate erupted over the seemingly innocuous question of how transistors—the revolutionary new electronic devices—should be drawn. By forcing a break in the long-standing traditions of electronic drawing, transistors generated a crisis in the ontology of circuit diagrams, forcing a choice between representations that emphasized form and those that stressed function. This paper explores what was at stake in that mid-century debate over visual culture. It tracks one function-based symbol through concerns about auto-comprehension, visual communication, and electronic reliability to see how transistor symbols formed crucial sites for articulating the meanings of material devices and their relationship to the wider populations of electronic entities, especially vacuum tubes. In doing so, the article shifts the emphasis in the history of electronics from material to visual culture, recasting our understanding of postwar electronics as a history of drawings as well as devices. Electronics, visual culture, transistors, symbols, vacuum tubes, historiography, schematic diagrams KEY WOR DS: I N T R O D U CT I O N In early October 1957, an émigré British electrical engineer named Norman Moody addressed a small audience of physicists and engineers gathered at CERN, the newly built high-energy physics laboratory straddling the French-Swiss *Science and Technology Studies, York University, Bethune College 218, 4700 Keele Street, Toronto ON M3J 1P3, Canada; imhotep@yorku.ca. The following abbreviations are used: DRTE, Defence Research Telecommunications Establishment; IRE, Institute of Radio Engineers; NAC, National Archives of Canada, Ottawa, Canada; RAF, Royal Air Force; RCA, Radio Corporation of America; RG, Record Group; RPL, Radio Physics Laboratory; TP, Phillip Thompson Personal Files, Ottawa, Canada; TRE, Telecommunications Research Establishment; WW, Wireless World. Historical Studies in the Natural Sciences, Vol. 38, Number 3, pps. 405–450. ISSN 1939-1811, electronic ISSN 1939-182X. © 2008 by the Regents of the University of California. All rights reserved. Please direct all requests for permission to photocopy or reproduce article content through the University of California Press’s Rights and Permissions website, http://www.ucpressjournals.com/reprintinfo. asp. DOI: 10.1525/hsns.2008.38.3.405. | 405 HSNS3803_03 7/10/08 406 | 5:58 PM Page 406 J O N E S-I M H OTE P border. Moody had been invited as part of a small group of specialists to discuss the best methods for generating data from the increasingly gargantuan machines of high-energy physics. With the material culture of particle physics split between the image-based traditions of the bubble chamber on the one hand, and the logic-based tradition of coincidence circuits and statistical analyses on the other, Moody’s allegiances must have seemed clear. His own expertise was in pulse triggers and counting circuits, developed as he rose (with no university degrees or formal training) through Depression-era television design, wartime radar circuitry, and postwar atomic weapons and nuclear engineering. Now the head of a defense electronics laboratory outside Ottawa, he had come to address an altogether different divide in twentieth-century material culture—one with implications stretching far beyond the image- or logic-based machines of high-energy physics. Cutting across the material divisions between bubble chambers and coincidence circuits had been an engineering culture built around evacuated tubes— Geiger-Müller counters and photomultiplier tubes, which formed the basis of counting circuits, but also the ubiquitous vacuum tubes, which drove the output of logic-based devices and helped automate the analysis of bubble chamber images.1 Moody was there to propose how engineers and physicists might begin breaking their long-standing relationship to vacuum tubes and start cultivating a new one in its place. His talk, “The Present State of the Transistor and its Associated Circuit Art,” began with a warning. If deployed properly—artfully—the transistor could prove to be more reliable than the vacuum tube. Deployed incorrectly, however, the results could be disastrous. As if to enliven the language of art in his presentation, Moody placed iconography at its core. In a move riddled with irony (given his audience), the electrical engineer began a talk consumed by logic circuits with an assertion about the primacy of images: the first step toward making transistors reliable, Moody suggested, was to change the way they were drawn.2 It is that remarkable assertion—that making transistors reliable meant drawing them differently—that motivates this paper. Moody’s comment is startling because of the way it makes the proper functioning of transistors depend not on the materiality of the devices, but on their 1. For more on image- and logic-based traditions, see Peter Galison, Image and Logic: A Material Culture of Microphysics (Chicago: University of Chicago Press, 1997), esp. chaps. 1, 2, and 6. 2. N. F. Moody, “The Present State of the Transistor and its Associated Circuit Art,” Nuclear Instruments 2 (1958): 182–92, on 182–83. HSNS3803_03 7/10/08 5:58 PM Page 407 ICONS AN D E LECTRON ICS | 407 representation. That shift in emphasis—from the details of devices to the intricacies of drawings—destabilizes a view of transistors (and electronics generally) that already enveloped the devices in the late 1950s, and which has dominated historical writing about them ever since.3 On this view, making transistors work was ultimately about getting the material details just right—tapping crystal “hot spots,” managing impurities, slicing gold foil just thin enough, growing oxide layers just thick enough to transform elusive quantum-mechanical phenomena into robust semiconductor actions. “Dirt effects,” Wolfgang Pauli declared them, as if to certify their grubby materiality. The consequences were dramatic. By making transistor action reliable, the story goes, those details and their refinements made possible the defining material break in twentieth-century electronics—the central organizing rupture, as machine after machine gradually saw the fragile evacuated glass of the vacuum tube yield to rugged semi-conducting crystals, ultimately 3. One approach to the history of transistors has stressed the material transformations needed to create and recreate the devices, as a corrective to the emphasis on conceptual breakthroughs. See, for example, Christophe Lécuyer, Making Silicon Valley: Innovation and the Growth of High Tech (Cambridge, MA: MIT Press, 2005); Ross Knox Bassett, To the Digital Age: Research Labs, Start-up Companies, and the Rise of MOS Technology (Baltimore, MD: Johns Hopkins University Press, 2002); Hyungsub Choi, “The Boundaries of Industrial Research: Making Transistors at RCA, 1948–1960,” Technology and Culture 48 (2007): 758–82. For a fascinating discussion of the problems of production, and the role of workers, see Stuart W. Leslie, “Blue Collar Science: Bringing the Transistor to Life in the Lehigh Valley,” Historical Studies in the Physical and Biological Sciences 32, no. 1 (2001): 71–113. For the emphasis on conceptual breakthroughs, see John Peter Collet, “The History of Electronics: From Vacuum Tubes to Transistors,” in Science in the Twentieth Century, ed. John Krige and Dominique Pestre (London: Routledge, 1997), 253. For this view in relation to the transistor, see Lillian Hoddeson, “Research on Crystal Rectifiers During World War II and the Invention of the Transistor,” History and Technology 11 (1994): 121–30; Lillian Hoddeson, “The Discovery of the Point-Contact Transistor,” Historical Studies in the Physical Sciences 12 (1981): 41–76; Lillian Hoddeson and Michael Riordan, Crystal Fire: The Birth of the Information Age (New York: W. W. Norton & Company, 1997); Ernest Braun, “Selected Topics from the History of Semiconductor Physics and its Applications,” in Out of the Crystal Maze: Chapters in the History of Solid State Physics, ed. Lillian Hoddeson, Ernest Braun, Jürgen Teichmann, and Spencer Weart (New York: Oxford University Press, 1992), 443–88; Charles Weiner, “How the Transistor Emerged,” IEEE Spectrum 10 (1973): 24–33. Riordan and Hoddeson have also suggested that material devices gave “operational reality” to theoretical entities, like the electron or the hole. See Michael Riordan and Lillian Hoddeson, “The Electron, the Hole, and the Transistor,” in Histories of the Electron: The Birth of Microphysics, ed. Jed Z. Buchwald and Andrew Warwick (Cambridge: MIT Press, 2001), 327–38. For a contrasting view on the development of the transistor, see M. Gibbons and C. Johnson, “Science, Technology and the Development of the Transistor,” in Science in Context: Readings in the Sociology of Science, ed. Barry Barnes and David Edge (Cambridge: MIT Press, 1982), 177–85. HSNS3803_03 7/10/08 408 | 5:58 PM Page 408 J O N E S-I M H OTE P enabling the micro-miniature, high-speed, and highly dependable electronics of today.4 That focus on materials and devices has been crucial in helping us understand the origins of the transistor, its deep moorings in materials science and the quantum theory of metals, and its role in transforming electronic material culture at mid-twentieth century. If taken too far, however, the approach threatens to obscure a more complex and, I think, more interesting history—one rooted in a key irony. Throughout the 1950s, technologists were able to agree wholly on the material details of transistors and on the physics that drove the devices, and yet disagree simultaneously and fiercely about what transistors were and how they functioned.5 By collapsing those multiple meanings of the devices into a unitary material identity, materialist histories obscure the sources and the stakes of those disagreements. They ultimately tell us little about how and why technologists learned to think differently about transistors, to understand them as electronic entities, and therefore to put them to work.6 4. On the role of production techniques in making the transistor reliable, see Christophe Lécuyer, “Silicon for Industry: Component Design, Mass Production, and the Move to Commercial Markets at Fairchild Semiconductor, 1960–1967,” History and Technology 16 (1999): 179–216, esp. 188–91; Ernest Braun and Stuart Macdonald, Revolution in Miniature: The History and Impact of Semiconductor Electronics, 2nd ed. [1978], (Cambridge: Cambridge University Press, 1982); Riordan and Hoddeson, Crystal Fire (ref. 3), esp. chap. 10; and Ross Knox Bassett, To the Digital Age: Research Labs, Start-up Companies and the Rise of MOS Technology (Baltimore, MD: Johns Hopkins University Press, 2007). On the role of the military, particularly in the context of miniaturization, see Thomas J. Misa, “Military Needs, Commercial Realities, and the Development of the Transistor, 1948–1958,” in Military Enterprise and Technological Change: Perspectives on the American Experience, ed. Merrit Roe Smith (Cambridge: MIT Press, 1985), 253–88. The use of the transistor as a periodization tool is common. See, for instance, Collet, “History of Electronics” (ref. 3), 254; on its role in the shift from modern mass culture to postmodern information culture, see Hoddeson and Riordan, “Electron” (ref. 3), 336. 5. Trevor Pinch and Wiebe Bjiker have discussed the multiple meanings of singular artifacts through their concept of “relevant social groups”; see Pinch and Bijker, “The Social Construction of Facts and Artifacts: or How the Sociology of Science and the Sociology of Technology Might Benefit Each Other,” Social Studies of Science 14 (1984): 399–441. I use symbols to explore those multiple meanings without suggesting that users of a given symbol formed a monolithic group. 6. Bruce Hevly has suggested that learning to use the devices ought to be part of any history of the transistor; Hevly, review of Revolution in Miniature, by Braun and Macdonald (ref. 4), Technology and Culture 25 (1984): 882–83. More generally, Ruth Schwartz Cowan has focused on the users of artifacts as a way of arriving at “properly constituted” histories of technology; Ruth Schwartz Cowan, “The Consumption Junction: A Proposal for Research Strategies in the Sociology of Technology,” in The Social Construction of Technological Systems: New Directions in the Sociology and History of Technology, ed. Wiebe Bjiker, Thomas Hughes, and Trevor Pinch (Cambridge, MA: MIT Press, 1987), 261–80. HSNS3803_03 7/10/08 5:58 PM Page 409 ICONS AN D E LECTRON ICS | 409 Moody, working in the thick of these debates, points us to that more complicated and more complete history. Speaking at the tail-end of a soon-to-beforgotten debate over drawing, he suggested that the key to making transistors work lay not in conduction bands or advanced doping techniques, but in symbols. It is that move from the bulky material devices of mid-century electronics to their abstracted line drawings—from material culture to visual culture—that I explore here. The symbol that animated Moody’s discussion was part of an explosion of competing schemes for transistors: in the mid-1950s at least fifteen public proposals and countless more private initiatives swept across electronics journals, engineering schematics, sketch-pads, and blackboards.7 Driving that explosion was a seemingly simple question: how should transistors be drawn? The act of drawing and the place of diagrams has been a lively topic in science studies for two decades now.8 Within the history of technology, it has been central to histories of mechanical engineering, highlighting, for instance, connections between uniform production, mechanical drawing, and the emergence of the modern state; or detailing how the rise of drafting helped draw the boundary between the modern and pre-modern worlds.9 Those issues, however, have been almost completely absent from the history of electronics and electrical engineering. Historical and technical writings alike have treated competing circuit 7. I am grateful to David Hounshell for bringing to my attention the existence of these private schemes in the U.S. 8. For a fascinating discussion of diagrammic representation in scientific pedagogy and training, see David Kaiser, Drawing Theories Apart: The Dispersion of Feynman Diagrams in Postwar Physics (Chicago: University of Chicago Press, 2005); and a more general discussion of pedagogy in David Kaiser, ed., Pedagogy and the Practice of Science: Historical and Contemporary Perspectives (Cambridge, MA: MIT Press, 2005). More generally, see Bruno Latour and Steve Woolgar, Laboratory Life: The Construction of Scientific Facts, 2nd ed. [1979] (Princeton: Princeton University Press, 1986), 51, 89, and 245; Michael Lynch and Steve Woolgar, eds., Representation in Scientific Practice (Cambridge: MIT Press, 1990); Martin Rudwick, “The Emergence of a Visual Language for Geological Science, 1760–1840,” History of Science 14 (1976): 149–95; Caroline Jones and Peter Galison, eds., Picturing Science, Producing Art (New York: Routledge, 1998); Brian Baigrie, Picturing Knowledge: Historical and Philosophical Problems Concerning the Use of Art in Science (Toronto: University of Toronto Press, 1996); Luc Pauwels, ed., Visual Cultures of Science: Rethinking Representational Practices in Knowledge Building and Science Communication (Lebanon, NH: University Press of New England, 2005). 9. Ken Alder, “Making Things the Same: Representation, Tolerance and the End of the Ancien Régime in France,” Social Studies of Science 28 (1998): 499–545; David McGee, “From Craftsmanship to Draftsmanship: Naval Architecture and the Three Traditions of Design,” Technology and Culture 40 (1999): 209–36. For a more general discussion of “visual thinking” in engineering, see Eugene Ferguson, Engineering and the Mind’s Eye (Cambridge, MA: MIT Press, 1992). HSNS3803_03 7/10/08 410 | 5:58 PM Page 410 J O N E S-I M H OTE P symbols, if they have treated them at all, as deviations and curiosities to be swept away by an inevitable standardization.10 In the process, they have ignored issues of representation when writing the history of a discipline that, like few others, uses icons in order to think. That oversight masks broader issues. For technologists in the 1950s, the seemingly simple question of how transistors should be drawn entailed other, deeper concerns: How should workers be commanded? What dangers arose as line drawings morphed into dense, recalcitrant artifacts? What knowledge, if any, should symbols presume, and what should they embody? How were transistors related to the wider population of electronic devices, and how might symbols capture those relationships? And, not least, could engineers be trusted? Positioned at the very core of the visual culture of electronics, circuit symbols made those questions both pressing and public. They formed the contested ground in debates over what transistors were, what drawings ought to be, and what practices might make both reliable. This paper explores what was at stake in drawing transistors in the mid1950s. It argues that for Moody and his contemporaries, transistor symbols were instruments crucial to making sense of the new devices. These symbols helped express what kinds of electronic entities transistors were, and, in particular, what relationship they bore to vacuum tubes. We take that relationship largely for granted now, but in the 1950s it was still in flux, and symbols formed one of the primary means of articulating it. This paper focuses on how a specific form of transistor, the junction transistor, sparked a crisis in the nature of schematic diagrams. By severing the long-standing link between macrophysical form and electronic function in these drawings—a link vouchsafed by the vacuum tube—the new transistors gave rise to two competing views of what transistor icons should represent, each slightly incongruent with established forms of drawing electronics. A material-structuralist view sought to emphasize material composition and structure within a tradition of representation where materials did not otherwise matter; a competing functionalist approach aimed to suggest how transistors worked in a graphic universe dominated by suggestions of physical form. This, in turn, caused 10. The literature on the history of transistors is almost completely silent on the issue of symbols. Riordan and Hoddeson’s Crystal Fire (ref. 3) makes no mention of competing schemes. Linda Petiot goes further than most in acknowledging the existence of different conventions in her excellent study of Moody’s computer research, but she does not discuss its significance. See Petiot, “Dirty Gertie: The DRTE Computer,” IEEE Annals of the History of Computing 16 (1994): 43–52. HSNS3803_03 7/10/08 5:58 PM Page 411 ICONS AN D E LECTRON ICS | 411 an explosion of possible symbols, as the idioms of classical design strained under the tension between structure and function. The symbols at the core of Moody’s CERN presentation initially emerged from that functionalist impulse. As they were appropriated by different members of Moody’s group, their meaning shifted, shaped each time by different visions of how schematic diagrams functioned: as taxonomical expressions, as social texts, or as heuristic devices. The various sections of the paper are organized according to those understandings and the context in which they took shape. The paper begins with the symbols’ initial development by Phillip Thompson, a physicist working in the Transistor Section of Moody’s laboratory. The symbols emerged from Thompson’s attempt to capture visually the essence of transistor action by drawing on the existing symbols for crystal diodes. The new diode-based symbols soon became the basis of a graphical nomenclature, a taxonomy of semiconductor symbols that embodied functional relations through iconography. The paper goes on to locate the symbols within a changing ontology of electronic drawings and the emerging conflict between structural and functional representation. Under attack from a competing approach that saw vacuum tubes as the functional analogue of transistors, the symbols were quickly reinterpreted as part of a circuit grammar that saw schematic diagrams as communicative devices. Here, the building blocks of the original symbol became syntactical elements governed by rules of composition that ensured the final diagrams’ coherence and intelligibility. The second section of the paper examines this new role of the symbols and their importance in the changing modes of producing electronics during the Cold War. It then turns from the functioning of diagrams to the functioning of machines. It situates the final interpretation of the symbols within a broader crisis in electronic reliability during the 1950s, fueled by concerns over the agency and influence of engineers and circuit designers. Fearing that conceptual and visual analogies between the transistor and the vacuum tube would lead to costly and even fatal errors, Moody reinterpreted the diode-based symbol as a way to discipline and guide the otherwise fallible mind of the engineer. The transistor symbol that initially emerged from concerns over comprehension and taxonomy now became a way to break the misguided and dangerous links between vacuum tubes and transistors, links that clouded the reasoning of new engineers and plagued the electronic schematics of the early Cold War. The paper ends with some conclusions about what visual culture can tell us about the history of electronics during the mid-twentieth century. HSNS3803_03 7/10/08 412 | 5:58 PM Page 412 J O N E S-I M H OTE P U N D E R S TA N D I N G T R A N S I S TO R S Moody’s comments at CERN in the fall of 1957 had their origins in a series of curious drawings that began appearing in the Electronics Laboratory of the Canadian Defence Research Telecommunications Establishment (DRTE) in early 1954. Formed just months after the start of the Korean War, DRTE brought together two existing laboratories—one in radio physics, the other in electronics—to create an organization capable of addressing the growing telecommunications needs of the Cold War, including radio communication to the strategically crucial Canadian North. While the Radio Physics Laboratory launched investigations of ionospheric storms and high-frequency radio blackouts, the Electronics Laboratory provided electronic support, initially designing and constructing radio physics instruments, and later mounting what officials termed “anticipatory research” into military electronics—investigations of advanced communications, electronic countermeasures, missile guidance, digital computing, and, eventually, space technology.11 In 1954, the Laboratory created a special Transistor Section, headed by Norman Moody, to investigate the newly developed junction transistor, the more useful successor to the original point-contact transistor invented at Bell Laboratories in late 1947. The new section’s aim was to develop circuit techniques that built on the transistor’s unique properties—rather than using transistors “as if they were another kind of vacuum tube.”12 The section’s first recruit was Philip Thompson. Trained as a physicist at Cambridge University in the early 1940s, Thompson had passed the Natural Science Tripos at the end of the war, then moved to Liverpool’s Automatic Telephone and Electric, Salford Electrical Instruments (where he designed a radio altimeter for the Royal Air Force), and finally the Radio Section of the Cavendish Laboratory, where he designed instruments for the Laboratory’s research in ionospheric physics. Thompson had continued this work in 1950 when he joined the Electronics Laboratory and headed a team of technologists designing instrumentation for radio propagation studies in the Canadian Arctic. As his focus shifted to transistors in 1954, Thompson saw the problem of grasping the transistor’s unique circuit characteristics as a problem of understanding the physics that drove them. 11. Frank T. Davies, “Actual Lecture Memos” (n.d.), Frank T. Davies Papers, Communications Research Centre, Ottawa, Canada, p. 4. 12. Philip Thompson, correspondence with the author, 10 Jun 2007. HSNS3803_03 7/10/08 5:58 PM Page 413 ICONS AN D E LECTRON ICS | 413 Junction transistors had been conceived as alternating layers of P- and N-type semiconductors suitably arranged so that they created the microelectronic conditions that made current amplification possible. Thompson quickly saw that the key to transistor action did not lie in the monolithic semiconductor layers themselves, but in the micromillemetric border regions where the different materials met. Focusing on these junctions, Thompson observed that individual P-N junctions acted like rectifiers, restricting current flow in one direction and privileging it in the other. (Indeed, a semiconductor rectifier, or diode, was simply a single P-N junction.) With sufficient voltage applied to the intermediate region, two such junctions placed close enough to each other would set up electrical potentials that allowed electrons to spill over one junction and then pour over the electrical potential at the next, creating amplification. Seizing on junctions in this way, Thompson believed he could reduce the action of the transistor to the arrangement and action of its constituent junctions. Preferring to picture physical processes in his work, Thompson casually shifted his attention from the workings of the transistor to its symbol.13 The most commonly used symbol for junction transistors had been borrowed from the original point-contact devices. These suggested nothing about the existence of junctions, much less their action. Thompson now set out to find a new way of drawing transistors that made the junctions central and explicit. He approached the problem of representation through a simple graphic synecdoche: from the point of view of its physics, a semiconductor device was its P-N junctions, so the problem of representing a transistor reduced to the problem of representing its specific arrangement of junctions. But how must the junction itself be represented? The simplest of all semiconductor devices relying on P-N junctions was the diode—a single junction used to “rectify” current flow. By transforming the traditional icon for the diode into a symbol for a P-N junction, Thompson could combine diode symbols to represent not only the backto-back arrangement of junctions in transistors, but any arrangement of P-N junctions used in semiconductor devices (Fig. 1).14 In a memorandum explaining the system to his colleagues on the Transistor Panel, Thompson elaborated the scheme: “As a diode consists of a single P-N junction in a semiconductor, a junction transistor two such junctions, and a compound transistor several, these 13. Ibid. 14. An icon is a signifier, the form of which directly reflects the thing it signifies. A symbol, by contrast, relies not on direct visual resemblance, but on suggestion or conventional relation. The icon for the semiconductor diode was in fact a schematized drawing of the device itself. HSNS3803_03 7/10/08 414 | 5:58 PM Page 414 J O N E S-I M H OTE P The Bell symbol and the diode-based symbol. At left, the traditional Bell symbols for PNP (top) and NPN transistors (bottom). Thompson’s symbols, at right, instead built on the standard symbol for the diode in order to represent combinations of P-N junctions. Source: TP. Reprinted with permission from IEEE Standard 57 IRE S3, IRE Standards on Graphical Symbols for Semiconductor Devices, Copyright 1957, by IEEE. Diode-based symbol reproduced with the permission of P. M. Thompson. FIG. 1 devices are represented as combinations of these junctions,” each symbolized by the diode symbol.15 For Thompson, the significance of the system was that it not only captured the inner workings of transistors, but also laid the foundations for a graphical nomenclature of semiconductor symbols—a logical and systematic method of choosing symbols for the entire range of present and future semiconductor devices.16 It was not only diodes and transistors whose symbols emerged naturally from this scheme. The system was, in principle, infinitely expandable. “Any semiconductor device,” Thompson explained, “which consists [of ] points, junctions, [or] both, will have a logical symbol within this system.”17 In its emphasis on the junction, the system was a statement about the semiconductor components it represented, telling its user “what the device was” and what connections it bore to the wider populations of electronic devices.18 15. P. M. Thompson, “The Graphic Nomenclature of Transistors,” EL Memorandum no. 505950A, Defence Research Board Telecommunications Establishment, Electronics Laboratory, Nov 1954, 1, NAC, RG24, Accession 89-90/205, Box 91, file 170–180/T8, part 1. 16. The modifier “graphical” distinguished the system from the assortment of letter symbols (such as “V” for valves or “T” for transformers) used to identify electronic parts. (This was also a source of controversy at the time.) 17. Thompson, “Graphic Nomenclature” (ref. 15), 1. 18. Thompson, correspondence (ref. 12). HSNS3803_03 7/10/08 5:58 PM Page 415 ICONS AN D E LECTRON ICS | 415 Through Thompson’s scheme, circuit designers would peer past the surface appearance of the devices to the fundamental P-N junctions that governed their behavior, determined their circuit characteristics, and defined them as electronic parts. F O R M S A N D F U N CT I O N S : D R AW I N G S A N D T H E O N TO LO G I E S O F E L E CT R O N I C S Thompson and his group became the principal users of transistors in Canada in the early 1950s, and the diode-based symbol quickly became the accepted standard for drawing the devices in circuit diagrams throughout the Canadian defense establishment. But already in late 1954, as the new symbols were beginning to delineate secret amplifier circuits and radar systems at the Electronics Lab, rival schemes began appearing publicly in lectures, conference presentations, classrooms, and trade publications. At least two sets of concerns (one general, one specific) seem to have driven the proliferation. The general problem was a growing anxiety over circuit diagrams, also known as schematics— What were they? How did they carry out their functions? What norms should they follow? I will deal with these questions in the next section. The second, more specific, concern emerged from a break that transistors caused in the idioms of classical circuit drawing itself. Much has been written about how transistors transformed the material culture of postwar electronics.19 Much less has been written about how technologists’ engagement with transistors shaped norms of representation and practices of circuit design and electrical engineering. In his classic study of mechanical engineering in ancien régime France, Ken Alder speaks of the “thickness” of devices, by which he means two things: a “thickness” of the artifacts, the difficulty of shaping recalcitrant matter into a working device or into physically identical copies; and a “thickness” in their representations, “the related challenge of assimilating ordinary artifacts to any idealized representation in such a way that their qualities can be captured in their entirety.”20 19. See, for instance, Hoddeson and Riordan, Crystal Fire (ref. 3); Braun and Macdonald, Revolution in Miniature (ref. 4); Misa, “Military Needs” (ref. 4). As Ross Bassett notes, the term “revolution” is pervasive in discussions about the impact of the transistor. See Bassett, “Digital Age” (ref. 4), 2. 20. Alder, “Making Things the Same” (ref. 9), 503. HSNS3803_03 7/10/08 416 | 5:58 PM Page 416 J O N E S-I M H OTE P Electronics diagrams shared this problem of assimilation. But unlike mechanical diagrams, which had developed sophisticated ways of rendering the physical layout of machines through isometric drawings and projective geometry, circuit diagrams were meant to show functional relationships rather than spatial relations.21 The symbols that populated diagrams in the 1950s seem to have evolved as visual abstractions of material devices, derived from either drawings of experimental setups or from patent applications in which devices were represented in full, alongside abstracted icons (Fig. 2). The point of the abstraction was to suggest both the physical form of the devices and, through a set of mechanical analogies and hydrodynamical metaphors (such as potential and flow), to tie that macroscopic physical form to the actions of electrons and therefore to electronic function. The symbol for the vacuum tube, for instance, stylized the elements of the device—cathode, anode, grid, envelope—and then analogized them to a valve, restricting or permitting the flow of electricity through portions of the circuit. Virtually every icon used in electronics drawings in the 1950s, including both the original transistor icon and the crystal diode symbol that Thompson used to challenge it, had their origins in this mix of abstracted form and suggested function.22 The transistor severed this traditional link between macroscopic physical structure and microphysical action. Whereas the bulky grid, cathode and anode of the vacuum tube could suggest a valve, restricting or yielding to electron flow, it was unclear how existing analogies could bind the large-scale structure of the transistor straightforwardly and intuitively to the micro-actions of the electrons inside. Furthermore, the new devices created problems for the ontology implications of schematic diagrams—the (often implicit) statements about what could be said to exist within the drawings. Classical diagrams had been built on a relatively simple diagrammatic ontology of electrons (implied), undifferentiated conductors (drawn), and insulators (blank space). In their place, the transistor proposed a world of dual-charge carriers (electrons and holes), their respective conductors (P- and N-type semiconductors), and insulators, all of which had to be distinguished within the two-tone scheme of the diagram. In the face of these breaks in idioms and ontology, a central question 21. For an overview of mechanical drawings and their emphases, see Peter Jeffrey Booker, A History of Engineering Drawing (London: Chatto and Windus, 1963). 22. For the patent origins of the point-contact diode symbol, for instance, see G. W. Pickard, “Means for Receiving Intelligence Communicated by Electric Waves,” U.S. Patent No. 836, 531 (filed 30 Aug 1906, issued 20 Nov 1906). HSNS3803_03 7/10/08 5:58 PM Page 417 ICONS AN D E LECTRON ICS | 417 Patent drawings for de Forest’s vacuum tube (top) and Bardeen and Brattain’s point-contact transistor (bottom). Circuit symbols emerged as schematized versions of experimental diagrams or patent drawings like these. Note the similarity between the figure at bottom left in the transistor patent and the traditional transistor symbol in Fig. 1. Note also the use of existing circuit symbols, like capacitors and resistors, alongside the less abstract drawings of the new devices. Source: Lee de Forest, U.S. Patent 0,879,532; John Bardeen and Walter Brattain, U.S. Patent 2,524,035. Courtesy of the United States Patent and Trademark Office. FIG. 2 emerged: what essential qualities should symbols embody, and how might they be drawn? The question was particularly vexing when applied to junction transistors. The existing Bell symbol had been a schematized drawing of the original experimental setup built by John Bardeen and Walter Brattain at Bell Labs in HSNS3803_03 7/10/08 418 | 5:58 PM Page 418 J O N E S-I M H OTE P 1947 for the point-contact transistor. Although the junction transistor had originally used the same icon, many observers believed it had been orphaned by the Bell symbol, which spoke to neither its form nor its functioning. Attempting to capture the details of the new transistor graphically, proposals for a new symbol split along two principal lines. One approach embraced the new ontology by emphasizing material composition and physical form; the other rejected questions of form in favor of symbols that primarily suggested function. Much of the resulting debate was intensely local, taking place in laboratory offices and seminar rooms, or within the hermetic Symbols Committee of the Institute of Radio Engineers (IRE), predecessor of today’s IEEE (Institute of Electrical and Electronics Engineers). When the IRE Symbols Committee issued its standards for electronic symbols in 1950, for instance, it was strangely mute on how transistors and semiconductor devices ought to be drawn. But in July 1952, possibly as a reaction to the growing profusion of transistor symbols, the committee’s former chair, Allen Pomeroy (himself a member of the technical staff at Bell Labs), revealed some of its anxieties, pleading for a single “symbol language” in place of what he saw as the emerging dystopia of organization-specific symbols that distinguished schematics at Bell Labs, for instance, from those at RCA.23 When the IRE finally adopted the Bell symbol in 1957, the compromise language of the Committee reflected the wider debates. In a remarkable and unusual foreword, the Committee set out the criteria for its choices: that symbols should “reflect the past,” “look to the future,” and “indicate physical properties . . . without overcomplication.”24 In the years leading up to that decision, however, the symbols debate went public as rival symbols sprang up in conference presentations, blackboard lectures and trade publications. In those challenges the inadequacies of the Bell symbol stood out, even among those who agreed with the emphasis on structure and form. One such challenge appeared in April 1954, when Henry Morgan submitted a one-paragraph note, along with a proposed transistor symbol, to the journal Wireless World. The standard Bell symbol, Morgan claimed, was fine in its original context, “but let us have a new symbol for junction transistors, 23. Allen F. Pomeroy, “Universal Engineering Shorthand,” Proceedings of the IRE 40 (1952): 771. 24. IRE Standards on Graphical Symbols for Semiconductor Devices,” Proceedings of the IRE 45 (1957): 1612–17, on 1612 (emphasis added). HSNS3803_03 7/10/08 5:58 PM Page 419 ICONS AN D E LECTRON ICS | 419 and one which will give the maximum information with the least work for the printer and drawing office.” In a move that would be echoed in a dozen similar schemes, Morgan located the “maximum information” in representations of material structure. Bearing a striking resemblance to the descriptive drawings of mechanical engineering, his symbol traced the physical form of the diffused junction transistor—its two odd-sized half circles, interrupted by the rectangular base region—but then added the key element in wider concerns over structure: an indication of the alternating materials making up the device (Fig. 3). In place of the simple universe of the classical circuit diagram, Morgan sought to indicate a more complex universe of dual-charge carriers and alternating semiconductors that he and others saw as constituting the physical and symbolic core of the devices. For the uninitiated, a mnemonic device decoded the new ontology: “black—dense with electrons—negative— n-type,” or “white—full of holes—positive—p-type.”25 Over the next three years, nearly a dozen symbols would compete publicly with Thompson’s scheme; a host of others would be used privately by circuit designers and engineers in lectures and courses. The majority would mirror Morgan’s materialstructuralism, emphasizing that the essence of transistors could be found in their alternating P and N regions, and that a proper symbol ought to represent these. In their eagerness to depict material structure, Morgan’s scheme and its defenders were vulnerable to at least one damning criticism: a novice engineer or technician approaching transistors for the first time would see nothing in the symbols that suggested how they actually worked. Morgan’s symbol, in particular, placed issues of function at one remove. Even after decoding the order of the semiconductor layers (NPN or PNP), readers still had to translate the structure forward, as it were, through depletion zones and conduction bands, in order to gain an understanding of how the devices operated. Function, some observers argued, demanded immediate accessibility. Thompson’s own symbol emerged as one of the earliest public examples of a functionalist alternative.26 Responding directly to Morgan’s proposal in 25. Henry Morgan, letter to the editor, WW 60 (1954): 178. 26. In using the term “functionalist,” I am not suggesting that Thompson’s symbol made no reference to structure, since the arrangement of P-N junctions could be seen as indicating structure. The term instead points to the principal aim of the symbol, which was not to show material structure or composition, but rather to exhibit an electronic function that could, in principle, be produced in a number of ways. 7/10/08 420 | 5:58 PM FIG. 3 Material-structuralist proposals for transistor symbols. The left column shows alternative symbols, including Henry Morgan’s proposal (top). The right column shows the mechanical and patent drawings for the same devices, including Shockley’s original patent drawing for the junction transistor (bottom). Source: Henry Morgan, letter to the editor, WW 60 (1954): 178; W. Gosling, “Electronics,” Physics Education 14 (1979): 402. Reproduced with the permission of Electronics World. Patent drawing courtesy of the United States Patent and Trademark Office, William Shockley, U.S. Patent 2,569,347. HSNS3803_03 Page 420 J O N E S-I M H OTE P HSNS3803_03 7/10/08 5:58 PM Page 421 ICONS AN D E LECTRON ICS | 421 June 1954, Thompson submitted his own diode-based symbol to Wireless World. His main criticism targeted Morgan’s “excessive concern with mechanical form.” Juxtaposing his own symbol against Morgan’s, Thompson explained, “You will note that our convention has been developed from the electrical function of transistors rather than their mechanical form.”27 But Thompson’s scheme itself rested on a particular understanding of what “function” meant. In the details of his scheme, he had located “electrical function” in circumscribed questions about mechanism. Here, the rectifying junction acted as the key functional unit, the black box. Put differently, what mattered for Thompson was not so much what transistors were made of or how their materials were arranged, but rather that their operations were based on a rectifying action that might, in principle, be achieved in a number of ways. The diode symbol was then introduced to symbolize the junction and shift the workings of the transistors onto familiar visual and conceptual grounds. In endorsing Thompson’s symbol, the editors of Wireless World called this the “historical” approach, since “the user is going from the known to the unknown.”28 But there were other possible “histories.” One particularly compelling one rested on a wholly different understanding of what it meant to demonstrate “function.” It involved stepping further back, shedding the quantum physics of electrons and semiconductors altogether, and even the operationalized picture of junctions, to see the entire device as the black box. Jerome Kurshan, for instance, one of RCA Laboratories’ first transistor researchers and head of the Tubeless Amplifier group, initially approached the transistor by shelving solid-state theory, focusing instead on measurements of input, output, and ratio of amplification. He would ultimately define the transistor functionally, as a device that amplified electric current without vacuum tubes.29 Under Kurshan and others, function became a question, not of mechanism, but of behavior. The key question—how did transistors work?— was really another question: what work did transistors do? How did they operate within the context of circuits and electrical variations? As Kurshan’s operationalization suggested, one dominant answer pointed away from rectifiers and solid-state electronics altogether, and toward the larger category of 27. Philip M. Thompson, “Transistor Symbols,” WW 60 (1954): 325 (emphasis added). 28. Editorial, “Transistor Symbols,” WW 61 (1955): 151. 29. Choi, “Boundaries of Industrial Research” (ref. 3), 763. HSNS3803_03 7/10/08 422 | 5:58 PM Page 422 J O N E S-I M H OTE P three-contact devices known as “triodes.” Viewed in this way, with the transistor itself as monolith, the natural analogies were no longer with semiconductor diodes, as Thompson argued, but with the triodes that dominated the material and design cultures of electronics in the mid-twentieth century— vacuum tubes.30 Functionalism relied on emphasizing function through visual analogies with existing electronic parts; but exactly which devices served as the proper referent hinged on one’s view of the meaning of “function.” The debate over electronic representation was an occasion for disagreements over what transistors really were—semiconducting analogues of the vacuum tube or evolved solid-state diodes. Because functional symbols argued for how transistors worked, and therefore shaped understandings of how they should figure in questions of design, the stakes over how to show function were enormous. It is on this debate among the proponents of functional representation, and particularly their disagreements over the links to vacuum tubes, that the rest of this paper focuses. U N D E R S TA N D I N G T R A N S I S TO R S I I The view that the transistor represented an analogue of the vacuum tube had begun with its inventors. In 1936, Bell Laboratories’ Director of Research, Mervin Kelly, launched the search to replace the electromechanical relays that he believed formed a bottleneck for Bell’s expanding telephonic empire. Vacuum tubes, which, like all amplifiers, could also be used as switches, were too bulky, power-hungry, and unreliable. So Kelly, a former director of vacuum tube development, suggested semiconductor amplifiers might furnish the ideal alternative.31 From its inception, the research program that produced the transistor aimed to produce a solid-state version of the vacuum tube.32 That understanding drove the image 30. Indeed, because of the analogy with the vacuum tube, the staff of Bell Labs considered a number of possible names—“semiconductor triode,” “surface states triode,” “crystal triode”— before agreeing on “transistor.” See John R. Pierce, “The Naming of the Transistor,” Proceedings of the IEEE 86 (1998): 37–45. 31. Jack Morton, Organizing for Innovation: A Systems Approach to Technical Management (New York: McGraw-Hill, 1971), 46–48. 32. That understanding was so strong, in fact, that William Shockley initially tried to place a copper grid inside a semiconductor crystal. See Riordan and Hoddeson, “Electron” (ref. 3), 5; Gerald L. Pearson and William H. Brattain, “History of Semiconductor Research,” Proceedings HSNS3803_03 7/10/08 5:58 PM Page 423 ICONS AN D E LECTRON ICS | 423 of the device after its invention. In their first published description, entitled “The Transistor, A Semiconductor Triode,” Bardeen and Brattain described their invention as a “three-element electronic device,” a triode that could be employed “as an amplifier, oscillator, and for any other purposes for which vacuum tubes are ordinarily used.”33 Bell’s publicity campaign quickly reinforced the view. In 1948, it demonstrated to its own staff a radio whose vacuum tubes had been entirely replaced by transistors. That summer, the New York Times reported on the new device, “called a transistor, which has several applications in radio where a vacuum tube ordinarily is employed.”34 Symbolic connections soon spilled onto the shop floor as manufacturers like Westinghouse, RCA, and Bell placed the production and development of transistors in their tube divisions.35 Writing in Scientific American, Louis Ridenour gave the substitution moral force, suggesting that there was nothing wrong with electronics that the elimination of the vacuum tube could not fix.36 In early 1957, those connections made their way into the debate over transistor symbols. They were most prominent in the writing of Marcus Scroggie, a former radar instructor with the British Royal Air Force, whose textbook, Foundations of Wireless, would become a classic. Born in Scotland and trained at the University of Edinburgh, Scroggie worked for various radio and electronics companies in the 1930s before serving as a radar operator for the RAF. He eventually became an instructor at the RAF Radar School, where the Air Ministry put him in charge of its secret publications on radio and radar. In the early 1950s he was best known for his regular columns in Wireless World, written under the pseudonym “Cathode Ray.” Starting in July 1956, Scroggie published a series of brilliant articles introducing readers to semiconductors and transistors. He began by noting that transistors generally received one of two opposing treatments in the literature. Some authors, claiming that one could use transistors without knowing how of the IRE 43 (1955): 1794–806; Walter Brattain, “Genesis of the Transistor,” Physics Teacher 6, no. 3 (1968): 109–14. For a general account of this view of the transistor, see Stuart Macdonald and Ernest Braun, “The Transistor and Attitude Towards Change,” American Journal of Physics 45, no. 11 (1977): 1062–63. 33. J. Bardeen and W. H. Brattain, “The Transistor, A Semiconductor Triode,” Physical Review 74 (1948): 230–31, on 230. 34. “The News of Radio,” New York Times, 1 Jul 1948. 35. See, for example, Leslie, “Blue Collar Science” (ref. 3), 89–90. 36. Louis N. Ridenour, “A Revolution in Electronics,” Scientific American 185 (1951): 13–17. HSNS3803_03 7/10/08 424 | 5:58 PM Page 424 J O N E S-I M H OTE P they worked, portrayed the devices through their “black box” representations as current-source-equivalent circuits. Others, seizing on the detailed physics behind transistor action, launched into Fermi-Dirac statistics, wave mechanics, and “all that” to explain the actions of electrons in solids. Unimpressed in equal measure by pragmatic operationalism and high-flown theory, Scroggie ventured a guess about his readers: “I suspect there are a large number of people who would like a simple picture of what goes on in transistors, to hang up in their minds alongside the picture they already have of what goes on in valves— the nice clear picture of electrons boiling off from the hot cathode and streaming across to the attractive anode, to an extent controlled by raising or lowering the potential of an intervening grid.”37 This gesture towards understanding transistors as the counterpart of vacuum tubes would inform Scroggie’s criticism of Thompson’s scheme and shape his vision of an appropriate icon. In launching into transistors, Scroggie realized that the “nice clear picture” he was after lay obscured in the confusing details of electron behavior in solids. He went on to negotiate a careful middle ground for his readers: “One thing we shall see is that the complicated picture of some of the actions [in transistors] can be replaced for routine purposes by a more condensed description, the details being, as it were, ‘understood.’” As if presaging a confrontation with Thompson, Scroggie added, “I only hope that the necessary simplifications will not make the physicists squirm too much.” With this, Scroggie began a discussion of electrical currents in solids. Vacuum tubes, he explained, were so easy to understand because the electrons released from the cathode traveled through a near vacuum rather than the “jungle” of fixed atoms that made up semiconductors, whose detailed behavior depended on advanced chemistry and physics.38 Avoiding that advanced discussion altogether, Scroggie moved on to the structure of atoms and explanations of shared electrons (both thermally and photo-electrically released), intrinsic conductivity, P- and N-type doping, and finally P-N junctions. Noting the tendency of junctions to privilege current flow in one direction and summarizing his material so far, he concluded (just as Thompson had two years before), “In brief, a P-N junction is a rectifier.”39 37. Marcus Scroggie [Cathode Ray, pseud.], “Semi-conductors—1,” WW 62 (1956): 317–21, on 317. 38. Ibid., 317–18. 39. Marcus Scroggie [Cathode Ray, pseud.], “Semi-conductors—2: Another Step on the Way to Transistors,” WW 62 (1956): 394–98, on 397. HSNS3803_03 7/10/08 5:58 PM Page 425 ICONS AN D E LECTRON ICS | 425 Here the two approaches diverged. Thompson had used the insight about P-N junctions to link transistors to solid-state diodes and to the wider world of semiconductor devices. Scroggie’s connections cut orthogonally. In the last of his expository essays, published in August 1956, Scroggie sought to link the transistor not to semiconductor diodes—the solid-state rectifiers that composed transistors in the form of P-N junctions—but rather to vacuum tubes. Beginning with a caution, Scroggie explained that in order to obtain “a transfer ticket from valves to transistors,” one had to recognize certain key differences: that vacuum tubes emitted only electrons from their cathode, whereas transistors employed both electrons and holes; or that electrons emitted from a valve’s cathode moved through a vacuum, whereas electrons in transistors made their way through a germanium or silicon crystal lattice.40 But the caution only underlined Scroggie’s principal aim of presenting the transistor as a functional analogue of the valve. Thompson had grouped devices according to their junctions, with all their untidy physics; Scroggie grouped them according to operational homologies. Viewed en bloc, all diodes functioned the same way, privileging current flow in a single direction, regardless of specific make-up. Triodes—three-contact devices, such as vacuum tubes or transistors—might differ in their details, but not in their basic electronic operation and structure. In a series of examples, Scroggie explicitly and visually translated semiconductor behavior and circuits into their vacuum-tube equivalents (Fig. 4). Current flow in diodes moved from disembodied charges to the interchangeable forms of vacuum tubes and their crystal analogues; a discussion of basic transistor arrangements erected solid-state circuits on a foundation of thermionic counterparts. Again and again throughout his insightful discussion, Scroggie drew his lines of relation, his “histories,” visually from transistors to valves and back again. To judge from the drawings, nothing functioned so much like a transistor as did a vacuum tube. Drawing Connections In April 1957, his introduction to transistors complete, Scroggie used his column in Wireless World as an opportunity to wade into the debate over symbols. Surveying most of the proposals that had come into the journal, Scroggie observed that with the exception of Thompson’s, all used the original Bell 40. Marcus Scroggie [Cathode Ray, pseud.], “Transistors: An Introduction for Beginners,” WW 62 (1956): 449–53, on 449. HSNS3803_03 7/10/08 426 | 5:58 PM Page 426 J O N E S-I M H OTE P Scroggie’s depiction of the relationship between transistors and valves. Current flow in diodes (top) moves interchangeably from disembodied charges to vacuum tubes and semiconductors. Below, basic transistor circuits built on their vacuum tube analogues. Source: Cathode Ray, “Transistors: An Introduction for Beginners,” WW 62 (1956): 449. Reproduced with the permission of Electronics World. FIG. 4 symbol to denote the point-contact transistor. This, Scroggie suggested, was as it should be. He had no quarrel with the incredible popularity of the Bell symbol in its original context. It did everything a symbol ought to: “it strongly suggests the thing it represents; it is easy to draw; it fits in easily to circuit diagrams; and it has become generally accepted. So, say I, as Glasgow says about itself, let it flourish.” Employing the same symbol for the junction transistor, however, was doubly problematic. First, it did not suggest the physical form of a junction transistor. “If the Editor will pardon my saying so,” Scroggie pointed out, “this applies especially to the Wireless World version, in which a great thick slab, like a foundation stone, is used to represent the thinnest possible layer of solid that modern technique can contrive.” Scroggie then went on to invert the problem—it was not just that the symbol was inaccurate; it was misleading, HSNS3803_03 7/10/08 5:58 PM Page 427 ICONS AN D E LECTRON ICS | 427 FIG. 5 The RCA symbol. The arrow at the bottom represents the emitter, the center line is the base, and the T-shaped portion at the top is the collector. Note the similarities with the common vacuum tube symbol, shown on the right, from which the enclosing circle is borrowed. Source: Cathode Ray, “Transistor Graphic Symbols: A Critical Analysis of Existing Ideas and Conventions,” WW 63 (1957): 194–98. Reprinted with the permission of Electronics World. Vacuum tube symbol reprinted with permission from IEEE Standard 54 IRE 21 S1, IRE Standards on Graphical Symbols, Copyright 1954, by IEEE. since it suggested that junction transistors were point-contact devices. “So, looking at a transistor circuit diagram, one wastes time searching for information on which kind of transistor is meant. If point-contacts become completely obsolete that objection will disappear, but we will still be left with an absurdly inappropriate symbol.” One by one, Scroggie dispatched the competing symbols through a series of criteria that any good representation should meet: it should “suggest” a junction transistor (presumably through a reference to form); it should be easy to draw on paper or blackboard; and it should be easy to distinguish from all other symbols. Symbols that required shading of some sort violated ease of drawing, and had to go. “Life these days is just too short,” Scroggie quipped. A remaining symbol belonging to E. Aisberg, editor of Toute la radio, showed the emitter and collector on the same side, a violation of the demand that a representation suggest physical form. “So all go.”41 Instead, Scroggie turned to one of the most popular early textbooks on the subject. Transistor Electronics had been co-written by five members of RCA who had developed their own transistor symbol (Fig. 5). Scroggie had already been using the symbol for some time, privately and in his teachings. He liked that it emphasized the thinness of the base and that it put collector and emitter on opposite sides. “Accepted practice rightly favors circuit symbols that primarily 41. Marcus Scroggie [Cathode Ray, pseud.], “Transistor Graphic Symbols: A Critical Analysis of Existing Ideas and Conventions,” WW 63 (1957): 194–98, on 194–95. HSNS3803_03 7/10/08 428 | 5:58 PM Page 428 J O N E S-I M H OTE P suggest function, with only a very general hint of outward appearance—liable to change in detail.”42 The RCA symbol was better for quick sketching and formal drawing. It made the distinction between point-contact and junction transistors. But the key factor was its crucial homology with the vacuum tube: At the same time it clearly points the analogy between the junction transistor and the valve. In fact, the only criticism I can imagine as having weight (though it has none with me) is that this analogy is pointed too clearly. There is, I believe, a school of thought that deprecates likening a transistor to a valve. Personally I hold that transistors have so much in common with valves, as regards function, methods of use, and to some extent internal workings, that it is futile not to note the similarities.43 In private Scroggie was more gracious. He had written to Thompson in midMay 1957, expressing three specific concerns with Thompson’s symbol: it took too long to draw, it could be confused with a simple rectifier, and it went against the British Standard symbol for rectifiers. In his reply Thompson dealt with the criticisms in turn. First, the British Standards symbol was an obsolete requirement. Second, his own symbols were being sketched at the Electronics Laboratory in two or three “strokes of the pencil”—far less, in any case, than those required for valves, which were seen as unproblematic. Finally, on the crucial suggestion “that transistors may be confused with rectifiers,” Thompson explained, “it is the nature of the system to illustrate the similarities and the differences between devices which contain one or two P-N junctions. There is a parallel situation in the system for valves, and I find no one is confused in this one either.” In a sketch that initially mirrored Scroggie’s comparison of transistor and valve circuits, Thompson laid out the sequence of vacuum tube and crystal devices—diodes, triodes, pentodes—but stressed the horizontal, or sequential, connections between them, emphasizing how the more complex depictions emerged from the simpler forms (Fig. 6). Just as the vacuum diode formed the basis for higher-order devices, so too did the semiconductor diode provide the basis for the emerging class of semiconductor devices, including transistors, based on the fundamental P-N junctions. Here the transistor emerged, not as the successor to the vacuum tube, but as an evolution of the diode. “In conclusion,” Thompson asserted, “I would like to say that I do not propose a special system of symbols, as much as a philosophy of an integrated 42. Ibid., 195. 43. Ibid. HSNS3803_03 7/10/08 5:58 PM Page 429 ICONS AN D E LECTRON ICS | 429 Thompson’s genealogy of transistors and vacuum tubes. Connections here are orthogonal to those in Fig. 4. The depiction notes the analogy between semiconductors and valves, but only as a way of stressing the serial connection of devices to their to simpler forms. Source: TP. Reproduced with the permission of P. M. Thompson. FIG. 6 and self consistent system. Being a bit of a traditionalist I have based mine on the accepted rectifier symbol, and adapted it to fit the new field.”44 There was irony in Thompson’s horizontal emphasis. Almost a decade earlier, the future head of RCA’s transistor program, Edward Herold, had laid out precisely this genealogy. Reading the New York Times announcement in the summer of 1948, he had written in his laboratory notebook, “For many years I have been interested in the possibility of making a three-element amplifying crystal which bears the same relation to the silicon and germanium diode as does the triode to the Fleming thermionic diode.”45 The symbol that RCA ultimately adopted for the transistor would nevertheless push aside those sequential connections—from diode to triode—in favor of the material and conceptual connections that the company was making between transistors and valves. Transistor projects at RCA would be housed in the Radio Tube Research Laboratory, and specifically in the “Tubeless Amplifier” group; the RCA symposia in 44. P. M. Thompson to Marcus Scroggie [Cathode Ray, pseud.], 31 May 1957, 2, TP. 45. “Solid-state Crystal Amplifiers,” E. W. Herold Notebooks, P–2758, in the David Sarnoff Library, Princeton, NJ, as cited in Choi, “Boundaries of Industrial Research” (ref. 3), 761, added emphasis. HSNS3803_03 7/10/08 430 | 5:58 PM Page 430 J O N E S-I M H OTE P 1952, which presented transistors and their applications to possible clients, drew direct analogies between the emerging junction transistor and the vacuum tube. Ultimately, the symbol developed by the largest tube manufacturer in the United States would reinforce those connections, rather than the genealogy elaborated in Herold’s notebook or Thompson’s iconography.46 The two symbols, RCA’s and Thompson’s, would eventually confront each other directly in September 1957; but not before their meanings had shifted away from concerns about essence and isolated, individual acts of comprehension to anxieties about the wider social world that drawings inhabited and the mediating role of symbols within it. C I R C U I T SY N TA X A N D T H E R E L I A B L E D I AG R A M The disagreements over transistor symbols were ultimately possible because of an initial consensus about schematic diagrams—the larger depictions of circuits that individual symbols populated. For all their differences about how to draw transistors, Thompson and Scroggie both saw schematics as extensions of the mental apparatus of the engineer. Standing between thought and artifact, they were the sites where machines were first speculatively assembled, operated, taken apart, and reconfigured. It was because of the way schematics helped engineers to think through the work of designing electronics that symbols had to capture the defining qualities of devices. In December 1948 Scroggie had argued for one view of schematics by focusing on a critical feature: their heuristic power for designers and engineers at work. Likening them to the system of Arabic numerals, Scroggie called them “a practically indispensable aid to thought; simple and effective.” He summoned the image of circuit designers leaning intently over drawings, rather than material circuits, when a piece of circuitry behaved unexpectedly. In fact, Scroggie pressed on, schematics were so effective that they created “the danger of handing over to them too much of our reasoning powers,” possibly fooling designers into believing that whatever existed in the diagram represented the totality of what happened in the circuit.47 The British physicist L. H. Bainbridge-Bell echoed Scroggie’s emphasis on circuit diagrams as specialized thought-tools. They should not be 46. “Tentative Presentations for Licensee Transistor Meetings, October 30, 1952,” 1, as cited in Choi, “Boundaries of Industrial Research” (ref. 3), 771. 47. Marcus Scroggie [Cathode Ray, pseud.], “Invisible Components: What the Circuit Diagram Fails to Show,” WW 54 (1948): 459–62, on 459. HSNS3803_03 7/10/08 5:58 PM Page 431 ICONS AN D E LECTRON ICS | 431 confused or combined with layout drawings, for instance, which gave exact spatial relations, useful for a purpose entirely different.48 In the hands of Scroggie and Bainbridge-Bell, the circuit diagram was a crucial “paper tool,” used to think through the workings of material artifacts before they took physical form and after things went awry.49 Another view, however, saw schematics not so much as the thought-things of individual engineers and designers, but as public artifacts meant to communicate and coordinate among the various groups implicated in the construction of artifacts. Throughout the spring of 1957, that debate over symbols had been characterized by an anxiety about the individual engineer, struggling (like Thompson had) to make sense of transistors. Over the summer of 1957, however, the focus of the discussion shifted from the heuristic value of symbols to their communicative function. The key figure in that transformation was a former British RAF signals officer and senior research technician at the Radio Physics Laboratory named Jack Bateson. Throughout his years at RPL, Bateson distinguished himself not only through his remarkable skills as a technician and teacher, but also through his commitment to pedagogy, especially clear and proper exposition. At yearly symposia Bateson would make careful, often exhaustive, stylistic commentaries on his colleagues’ technical lectures— how they handled slides and overhead transparencies, the cadence of their speech, their use of grammar, and their expository style.50 Circuit diagrams intersected Bateson’s concerns about language through at least two points. One was syntax: like the constructions of natural language, circuit diagrams were compositions, built up from more fundamental symbols (resistors, capacitors, vacuum tubes, inductors, transformers, and so on).51 Just as the rules of syntax shaped proper expression in natural language, so the rules 48. L. H. Bainbridge-Bell, “Circuit Symbols: Notes on the New British Standard,” WW 54 (1948): 437–38, on 437. 49. Here I borrow the term “paper tool” from Ursula Klein and David Kaiser. See Ursula Klein, “Techniques of Modeling and Paper-tools in Classical Chemistry,” in Models as Mediators: Perspectives on Natural and Social Science, ed. Mary Morgan and Margaret Morrison (New York: Cambridge University Press, 1999), 146–67; Ursula Klein, “Paper Tools in Experimental Cultures,” Studies in the History and Philosophy of Science 32 (2001): 265–302; Ursula Klein, Experiments, Models and Paper Tools: Cultures of Organic Chemistry in the Nineteenth Century (Palo Alto, CA: Stanford University Press, 2003); and David Kaiser, Drawing Theories Apart (ref. 8), chap. 1. 50. J. Bateson, “The Defence Research Board fifth symposium,” 3 Dec 1953, 2, TP. 51. Bateson’s views unconsciously built on a long and fascinating history of technical artifacts as language systems. The eighteenth-century Swedish technologist Christopher Polhem constructed a “mechanical alphabet” in which the “five powers”—levers, wedges, screws, pulleys, and HSNS3803_03 7/10/08 432 | 5:58 PM Page 432 J O N E S-I M H OTE P of electronic design governed the proper graphical articulation of circuits. The placement of power supplies, the direction of signal travel, and graphic conventions for crossing wires were more than simple matters of etiquette for Bateson; they were what made circuits clear and meaningful, structuring how drawings should be read and understood. Here diagrams provided a second point of intersection with concerns about language. Whether appearing in engineering journals or spread across the workbench, circuit diagrams were instruments of effective communication; they were directed toward an audience. In Bateson’s own Radio Physics Laboratory, a set of drawings might move from engineer to technician and back a dozen times, with individual circuits and elements contributed by just as many people. Breakdowns of communication through diagrams jeopardized the effectiveness of radars and communication circuits. As Bateson’s colleague Philip Thompson would later explain, “circuits will be contributed by a variety of individuals, and they will be compatible only if there is some central guidance in a few fundamentals.”52 Rules of composition ensured compatibility in this production system. Bateson soon brought these concerns over syntax and communication to bear on the debate over symbols. He had eagerly followed the discussion of transistor symbols in Wireless World. Philip Thompson was a colleague at RPL’s sibling, the Electronics Laboratory, and Bateson had long followed Marcus Scroggie’s columns, where he admired Scroggie’s expository abilities, particularly the “ruthless scorn” to which Scroggie subjected misguided thinking in electronics. He had even drawn on Scroggie’s style to make his own colleagues better “platform expositors.”53 But Scroggie’s discussion of transistor symbols represented a breach. “I feel bound to protest,” Bateson wrote to Scroggie directly. “I rather suspect that Mr. P. M. Thompson, with whom I occasionally establish contact across the 30 miles that separate us by means of another of Bell’s inventions, will protest also.”54 For Bateson, Scroggie’s criticisms of the diode-based symbol hinged on a fundamental misunderstanding of what circuit symbols were. Like the larger schematics to which they belonged, symbols aimed at clear communication. winches—functioned as the vowels, while the other elements provided the consonants. He suggested that the proper metaphor for technological development was composition, rather than invention. 52. Philip Thompson to Seymour Schwartz, 1 Aug 1958, TP. 53. Jack Bateson to Marcus Scroggie [Cathode Ray, pseud.], 15 Apr 1957, TP. 54. Ibid. HSNS3803_03 7/10/08 5:58 PM Page 433 ICONS AN D E LECTRON ICS | 433 As he would later explain to a colleague, “whatever symbols or conventions one puts forward their basis ought to be clarity of understanding.”55 Also, like the larger diagrams, symbols gained power and coherence through their modularity. That modularity was obvious in the way the diagrams were put together or assembled, much like the artifacts themselves, with individual component symbols providing the discrete building blocks of the larger circuits and equipment. But for Bateson, symbols were also modular, built up from more basic graphical elements. Writing to Scroggie, Bateson explained how the vacuum-tube symbol had obscured this modularity, leading to a misguided view of the proper icon for transistors: It seems to me that familiarity with the vacuum-tube symbol has led you to regard it as a symbol for a vacuum-tube, rather than a symbolism of cathodes, grids, envelopes, gas fillings, etc., all of which may be combined suitably to represent some device which makes use of these in a practical way. This, in turn, has led you to look for a symbol [that] will represent a complete transistor. . . . Thompson’s suggestion is not a series of symbols for transistors; it is a symbolism of P-N junctions (and envelopes) based upon a symbol [that] is already the internationally accepted standard for a P-N junction, or semiconductor diode, or “nonthermionic rectifier.” The danger of thinking in terms of transistor symbols, rather than semiconductor-device symbolism, is that one is left to a revival of the present controversy every time the fertile field of semiconductor research produces another device based on P-N junctions (e.g., photo-diode, Zener diode, etc.).56 In a remarkable memorandum produced in the late summer of 1957, Bateson and Thompson (the latter now head of the Airborne and Allied Applications Group of Basic Circuits) joined forces. The document began with a historical defense of functionalism in electronics drawings. Citing the evolution of the vacuum tube symbol, the authors explained how the “early and also pictorially derived symbols for valves gave way to symbols which could be helpful in explaining the action of a valve.”57 They then went on to combine Thompson’s heuristic concerns with Bateson’s emphasis on communication, elaborating the diode-based symbol into a full-blown iconographic system (Fig. 7). Two principles dominated the scheme. The first was Thompson’s view 55. Jack Bateson to Editor of WW 25 (1962): 2, TP. 56. Jack Bateson to Marcus Scroggie [Cathode Ray, pseud.], 15 Apr 1957, TP. 57. Philip M. Thompson and Jack Bateson, “Semiconductor Symbols: Logical System for Diodes, Transistors, and Other Junction Devices,” WW 63 (1957): 525–28, on 525. HSNS3803_03 7/10/08 434 | 5:58 PM Page 434 J O N E S-I M H OTE P The expanded diode-based system. Here Thompson and Bateson elaborated both the system for constructing symbols and the rules for incorporating them into circuit diagrams. Source: Philip M. Thompson and Jack Bateson, “Semiconductor Symbols: Logical System for Diodes, Transistors and Other Junction Devices,” WW 63 (1957): 525. Reproduced with the permission of Electronics World. FIG. 7 HSNS3803_03 7/10/08 5:58 PM Page 435 ICONS AN D E LECTRON ICS | 435 of a “logical, self-consistent, graphical nomenclature” for semiconductors, based on the diode symbol, and capable of depicting any device that depended on P-N junctions. On the genetic link to the diode symbol, the authors were emphatic: “The connection between junction diodes and junction transistors was a logical one,” they explained, “and not easily broken, and there was no real reason for trying to break it.”58 The second key feature was the emphasis on the ability of the symbols, and the schematics containing them, to communicate reliably the intentions of the designer to other engineers and technicians. “Schematic circuit diagrams are line expositions of circuits,” the authors explained, in Bateson’s characteristic language. “These line pictures should be as easy to read as the pictures or descriptive texts that go with them.”59 They continued, “In general, it should be remembered that the contribution made by the eye when reading a diagram is greater than that made by words. The appearance of the diagram, therefore, is not unimportant.”60 In place of the existing patchwork of semiconductor symbols, the authors stressed principles and logic, rules, and fundamentals. Together with a set of guidelines— connections to positive supplies should be upward; connection to negative supplies should be downward with ground in between; signals should travel from left to right, and so on—the new system, its authors argued, would reduce the drawing and reading of circuits to a set of logical and straightforward principles. But even as Thompson’s symbols entered the broader debates over semiotics and circuit diagrams, their meanings were shifting again. Thompson’s initial intervention had been motivated by the individual act of comprehension—the lone engineer unlocking the workings of the transistor through a revelatory symbol. Through his debate with Scroggie and collaboration with Bateson, the symbols became instruments directed outward, away from the comprehension of circuit designers and engineers themselves, and toward communication and pedagogy. And in the hands of Thompson’s supervisor, Norman Moody, they would finally turn inward again. No longer mere instruments for communication, they became a way to discipline the minds of the engineer and the circuit designer, and to save Cold War electronics from itself. 58. Ibid. 59. Ibid., 4. 60. Ibid. HSNS3803_03 7/10/08 436 | 5:58 PM Page 436 J O N E S-I M H OTE P F R O M P H I LO S O P H Y TO A R T : SY M B O L S A N D T H E R E L I A B L E D E S I G N E R This last transformation in the meaning of Thompson’s symbols was rooted in a deep ambivalence about electronics that ran through the 1950s. Writing on the impact of the transistor in early 1958, Jack Morton at Bell Laboratories expressed this ambivalence well. Stressing the transistor’s transformative power, Morton laid out a broad cybernetic vision where electronics extended the visual, mental, and tactile capacities of “electronic man,” projecting the human mind even more forcefully than nuclear weapons projected human strength. Those capacities, however, came at a price. Morton cautioned, “all such functions suffer from what has been called ‘the tyranny of numbers.’”61 The complexity of circuits carrying out those advanced functions required hundreds, thousands, even tens of thousands of electron devices. And as the number of circuit components increased, so did the likelihood of circuit failure.62 One had only to look to the beginning of the decade and the Korean War, when electronic failures had grounded a squadron of Marine Panther jets only months before malfunctions in dense fog sent them into a mountainside.63 Electronics would later be partly blamed for the disastrous failure of tactical missions. Barely two weeks after the armistice, one influential report suggested that Korea was only the tip of the iceberg: two-thirds of U.S. military equipment was likely malfunctioning.64 By the time Morton made his comments, concerns had spread to the sprawling integrated systems of radar, targeting, and guided missiles that secured continental defense and nuclear deterrence. The threat that electronic failure posed, not only to limited tactical missions but also to the broader strategic aims of the Cold War itself, created what contemporaries called the 61. J. A. Morton and W. J. Pietenpol, “The Technological Impact of Transistors,” Proceedings of the IRE 46 (1958): 955–59, on 955. 62. Morton saw this as an economic effect, in that the cost of making vacuum tubes reliable had held back the development of advanced electronic systems. The corollary was, of course, that with existing vacuum tubes, complex electronics were dangerously prone to failure. 63. On the proportion of malfunctioning equipment to functioning equipment, see Keith Henney, ed., Reliability Factors for Ground Electronic Equipment (New York: McGraw-Hill, 1956), 1–1. For reference to Korea, see Chester Soucy, “Present Unreliability,” briefing to the Department of National Defence and Services, 30 Dec 1954, 2, NAC, RG24, vol. 2484, file 801–E447–1, part 1. 64. R. R. Carhart, “A Survey of the Current Status of the Electronic Reliability Problem,” U.S. Air Force Project RAND Research Memorandum RM–1131 (Santa Monica, CA: RAND Corporation, 1953), 66. HSNS3803_03 7/10/08 5:58 PM Page 437 ICONS AN D E LECTRON ICS | 437 “reliability crisis,” launching the most wide-ranging and comprehensive examination of electronic reliability ever mounted. Morton’s now-classic invocation of the “tyranny of numbers” captured the deep anxiety about electronic complexity in the late 1950s. On the most common reading, Morton was merely crystallizing a widespread understanding that reliability in complex circuits was a problem of interconnections: as systems included more and more parts, the fallible solder joints that connected them multiplied, generating site after site of potentially imminent failure. In this way, he was simply articulating the famous “interconnection problem,” which the integrated circuit would ultimately solve.65 But that interpretation of Morton’s comments collapses the complexity of the reliability problem into a prehistory of the integrated circuit, a misreading that obscures Morton’s real concerns. Closer attention to Morton’s comments in the spring of 1958 suggests that they were not at all about interconnections. For him, the “tyranny of numbers” expressed a problem of electronic parts, specifically vacuum tubes, not their interconnections. As complex systems incorporated more and more tubes, reliability plummeted. The power-hungry and inherently unreliable vacuum tubes therefore placed limits on circuit complexity, stifling the enormous potential of electronics to extend human capabilities. Tubes could be made more reliable, but at prohibitive costs. Morton’s solution—the transistor—was still bulky and discrete; like any other part, it still had to be soldered and could not possibly be a solution for multiplying interconnections. Instead, the transistor escaped the “tyranny of numbers” because it was efficient, small compared to the vacuum tube (though not to the integrated circuit), and inherently more reliable. In this way, Morton’s comments formed part of one of the most pervasive, but increasingly challenged, views of reliability in the 1950s.66 Built on a philosophy that transformed the atomic quality of the electronic part—“an item which cannot be 65. There seem to be a number of reasons for this interpretation. Morton, for instance, would later use the same language of “tyranny” in the mid-1960s to describe the problem of interconnections. Also, one of the original inventors of the integrated circuit, Jack Kilby, used Morton’s expression to express the same problem. Finally, the later importance of the integrated circuit has encouraged scholars to write the history of reliability as a history of the problems which the integrated circuit solved. See, for instance, T. R. Reid, The Chip: How Two Americans Invented the Chip and Launched a Revolution (New York: Random House, 2001), 16. For Morton’s later use of the term “tyranny,” see Jack A. Morton, “The Microelectronics Dilemma,” International Science and Technology 55 (1966): 35–4, on 36. For Kilby’s quotation of Morton, see Reid, “Chip” (above), 72. 66. Carhart, “Survey of the Current Status” (ref. 64), 6. HSNS3803_03 7/10/08 438 | 5:58 PM Page 438 J O N E S-I M H OTE P disassembled without destroying its identity”—into an atomized understanding of reliability, in which the reliability of a system was simply the product of the individual part reliabilities, the view held that making systems reliable meant improving the individual devices that composed them.67 It was that view that dominated understandings of the reliability crisis throughout the 1950s, guided failure analyses in the U.S. ballistic missile program, and drove massive military and industrial initiatives to improve parts throughout the decade.68 As Morton’s comments show, that view of reliability as a problem of parts was tenacious. In his foundational work on packet-switching in the early 1960s, for example, Paul Baran at the RAND Corporation could still criticize Bell Telephone’s pervasive attempts to increase the reliability of the telephone system by making each component as reliable as possible.69 Baran instead favored using more ordinary components, but implementing redundancy or rerouting to make the system insensitive to their weaknesses.70 Baran’s views are significant here because they grew out of a competing, but essentially complementary, view that saw electronic reliability as a property of the system—a problem not only of parts, but of the way they were arranged, either physically or conceptually, into larger assemblies. Even given the most reliable components, features of the way those components were organized and assembled could still generate unreliability. In the drive to secure the entire system, every one of its aspects was up for questioning—manufacturing procedures, maintenance practices, even soldering and assembly techniques, including those that led to the integrated circuit. At the level of electronic engineering and design, however, this systemic understanding of the causes of failure targeted both electronic parts and the larger circuits in which parts operated.71 Even the most robust parts could be misapplied, subjected to wild voltage or current fluctuations, or other electrical stresses. Only imaginative and careful design, like the kind Baran would advocate, could 67. F. N. Halio, “Improving Electronic Reliability,” IRE Transactions on Military Electronics MIL–5 (1961): 12. This view was reinforced by reports that electronic parts were “responsible” for sixty percent of ballistic missile failures. See Chester Soucy, “Reliability Control of Electronic Equipment in Aircraft and Weapons Systems: General and Management Aspects,” Canadian Aeronautics Journal 3 (1957): 222–31, on 227. 68. See Carhart, “Survey of the Current Status” (ref. 64), 7. 69. J. Abbate, “Cold War and White Heat: The Origins and Meanings of Packet Switching,” in The Social Shaping of Technology, ed. Donald MacKenzie and Judy Wajcman, 2nd ed. (Philadelphia: Open University Press, 1999), 351–71, on 357. 70. Ibid., 356. 71. For a broader discussion of these competing understandings of reliability, see Edward JonesImhotep, “Disciplining Technology: Electronic Reliability, Cold-War Military Culture, and the Topside Ionogram,” History and Technology 17 (2001): 125–75. HSNS3803_03 7/10/08 5:58 PM Page 439 ICONS AN D E LECTRON ICS | 439 reshape the electrical environments in which individual parts operated, shielding even ordinary components from the conditions that drove them to fail. Throughout the mid-1950s, the Electronics Laboratory, where Thompson’s symbols emerged, became one of the most prominent sites for this systems view of electronic reliability. Formed at the outbreak of the Korean War, the Laboratory’s early research centered on the stream of new devices then pouring out of the electronics industry. To focus its efforts, the Laboratory’s director, Frank Davies, authorized the creation in early 1954 of a Components Section supervised by Frank Simpson. Subdivided into the major sections of Solid-State Research, Magnetic Research, and Component Techniques, Simpson’s team investigated the characteristics of individual electronics parts with a view to improving their reliability. One group of components, however, escaped the scrutiny of Simpson’s team. Transistors instead fell under a second section of the Electronics Laboratory, also created in 1954—and led by Norman Moody, who began our story. Smaller, lighter, and physically more rugged than vacuum tubes, transistors held out the hope of smaller, faster, more robust electronics. In their early years, however, transistors were still problematic devices. Early point-contact transistors had limited frequency performance and little power; they were noisy and vulnerable to power surges, radiation, and high temperatures.72 Even the germanium junction transistors introduced in 1951 and commonly used through the mid-1950s were highly temperature-dependent (failing to work altogether at about 75°C) and suffered from relatively large leakage currents, which caused problems for the switching circuits required by digital electronics.73 The new devices had to be made reliable. Moody’s group worked on carefully linking, through precise measurements, transistor parameters (like base-emitter voltages or input impedance) to circuit parameters (such as current at the collector), with the goal of understanding how properties of the individual part varied with the larger electrical environment of the circuit. They then worked to devise transistor-specific design solutions that would both keep the devices within tight electrical parameters and make the circuits employing them relatively insensitive to variations in their performance. Using Thompson’s symbol in all its design work, Moody’s section quickly grew to include several subdivisions—Transistor Measurements, Radio Frequency Receivers, Digital Techniques, and Advanced Circuit Research. 72. Macdonald and Braun, “Transistor and Attitude Towards Change” (ref. 32), 1062. 73. Riordan and Hoddeson, Crystal Fire (ref. 3), 207, 221. HSNS3803_03 7/10/08 440 | 5:58 PM Page 440 J O N E S-I M H OTE P The foundational work that went on there was recognized in the new title given to the section—Basic Circuits. Moody had come to Basic Circuits through an unlikely path. The Great Depression had kept him from attaining a formal education in electrical engineering. He began his career as a radio repair technician at a large London store, worked as a junior engineer designing radio receivers in the mid-1930s, and eventually rose to the position of senior television designer at Halcyon Radio. Here, in the design of televisions, he developed his abilities in advanced circuit design—at least one colleague would call it “artistry”—that led to his wartime recruitment by the British Telecommunications Research Establishment (TRE), the main research group for RAF radar applications. At TRE, Moody was attached to a small circuit group under the direction of Frederic Williams, the English engineer reputed to have developed the first operational amplifier and, with the physicist P. M. S. Blackett, an automatic curve follower. Williams was famed at TRE for his emphasis on “designability”—the property of circuits to reveal their precision, reliability, and even producibility through design alone. The emphasis on conceptual rigor and elegance ran strong in Williams’s work. As J. R. Whitehead, radiophysicist and wartime colleague, explained, Williams’s “sole concern was to see the desired electronic function performed elegantly, efficiently and reliably. . . . He was notorious for his tangled breadboard circuits, which often drooped over the edge of the bench towards the floor—a unique combination of conceptual elegance and material chaos.”74 For six years Williams’s group deployed his principle of designability to develop a host of circuits for radar, airborne interception, and IFF (Identification Friend or Foe) systems, which enabled radar operators to distinguish between their own planes and enemy aircraft. Moody’s main job involved collaborating with Williams on timing and pulse circuits like the “phantastron,” a monostable pentode circuit used to generate sharp timed pulses after being triggered, or the “sanatron,” a linear but variable time-delay circuit, so-called because of Williams’s fondness for referring to well-behaved circuits as “sanitary.” Moody would take that work on robust, “designable” pulse circuits through a succession of projects after the war—to Chalk River, developing instrumentation for Canada’s nuclear laboratories; to the British atomic weapons project, designing high-speed circuits to measure gamma ray fluxes from atomic bombs 74. J. R. Whitehead, quoted in T. Kilburn and L. S. Piggott, “Frederic Calland Williams, 26 June 1911–11 August 1977,” Biographical Memoirs of Fellows of the Royal Society 24 (1978): 583–604, on 590. HSNS3803_03 7/10/08 5:58 PM Page 441 ICONS AN D E LECTRON ICS | 441 before the electronics were destroyed by neutron bombardment; and to the Basic Circuits group, after returning to Canada in early 1952, where his interests in the months after his arrival focused on transistor triggers. He gradually gathered around him a group of engineers and technicians dedicated solely to investigating the emerging field of transistors and their circuits. No project better embodied Moody’s attitudes toward transistor design than his work with bistable multi-vibrators, also known as “flip-flop” circuits. Invented in 1919 by William Eccles and Frank Jordan, the flip-flop circuit originally employed symmetrically arranged vacuum tubes to produce a binary switching circuit in which one tube was always conducting while the other was off. A trigger pulse would then switch or “flip” the tubes’ state.75 Since each state was stable indefinitely, the circuit had the property of “remembering” input pulses and could be used to provide one “bit” of memory. These same properties made the circuits useful as counters, and therefore interesting for both digital computing and nuclear instrumentation.76 It was in both these capacities, as counters and pulsed memory circuits, that flip-flops engaged Moody’s interests. His wartime design of a monostable trigger circuit for radar detection had won him recruitment to Williams’s group at TRE; his postwar research in flux measurements for radioactive particle emissions drove his interest in faster and more reliable trigger circuits, including variations of the flip-flop.77 Previous efforts to design solid-state versions of the flip-flop circuit had involved straightforward material substitutions, trading vacuum tubes for transistors but maintaining the same structural symmetry. The arrangement, however, meant that half the circuit was always conducting, and therefore continually drawing, current, which contributed to the heat and material degradations that weakened transistor reliability in the first place. Instead of viewing the transistor as a substitute for the tube, Moody approached the problem by exploiting a unique feature of transistors—their polarity. By coupling two transistors of opposite polarities, PNP with NPN, Moody created a circuit that was either fully conducting or not (Fig. 8). The circuit’s operational stability, however, still depended on its design details. Moody now used the diode analogy to translate 75. W. H. Eccles and F. W. Jordan, “A Trigger Relay Utilizing Three Element Thermionic Vacuum Tubes,” Radio Review 1 (1919): 143–46. 76. John Mauchly, co-designer of the ENIAC computer, based his work with computer flip-flops on the circuits used to count pulses from Geiger-Müller tubes. See R. K. Richards, Electronic Digital Components and Circuits (Princeton: Van Nostrand, 1967), 4. 77. Norman Moody interview, 1985, National Museum of Science and Technology, Ottawa, Canada. HSNS3803_03 7/10/08 442 | 5:58 PM Page 442 J O N E S-I M H OTE P The flip-flop circuit. The original Eccles-Jordan circuit (left) used a symmetrical arrangement of identical vacuum tubes, shown with their enclosing envelopes, so that one-half of the circuit was always conducting. Detail from one of Moody’s circuits (right) shows his transistorbased flip-flop, which employed two transistors of opposite polarities to produce an “off-state” that drew no current. The core of the flip-flop is the set of complementary transistors, T1 and T2, drawn using Thompson’s symbol. Source: W. H. Eccles and F. W. Jordan, “A Trigger Relay Utilizing Three Element Thermionic Vacuum Tubes,” Radio Review 1 (1919): 143–46. Reproduced with the permission of Electronics World. Circuit drawing, Allouette files, Canada Science and Technology Museum (CSTM) Supplementary files that accompany the artifacts in the collection. Moody illustration courtesy of the Canada Science and Technology Museum. FIG. 8 the transistor’s physical parameters into composite design parameters—the frequency cutoff after which the transistor gain was cut in half; the leakage current caused by “collector diode leakage,” which affected the switching ability of the devices and therefore their usefulness as counters or gates; and the input impedance, which would have to be matched to the surrounding circuits for stable operation (Fig. 9). Here the suggestive power of the new symbol became clear. The diode analogy, captured in Thompson’s symbol, led directly to the crucial circuit parameters that allowed reliable integration of the transistor into the circuit. Dispensing with the vacuum tube analogy was what allowed the novel antisymmetries of the circuit; embracing the diode analogy was what made it stable and robust.78 Working with fellow engineer David Florida, Moody then expanded the design to create a “family of circuits that thrives on heavy load 78. In large logic applications—where a group of circuits might all switch from on to off— the circuit could cause large current fluctuations, which would compromise reliability. For the simple trigger circuits Moody was interested in, however, the circuit showed remarkable stability. HSNS3803_03 7/10/08 5:58 PM Page 443 ICONS AN D E LECTRON ICS | 443 Diode-based design parameters. Note how Moody used the diode properties (center) to translate physical parameters into design parameters. These then structured the details of the flip-flop circuit, giving it its robustness under heavy load and extreme electrical conditions. Source: Moody, “Present State” (ref. 2). Reprinted from Nuclear Instruments. Copyright (1958), with permission from Elsevier. FIG. 9 currents, often to the limit set by transistor destruction.”79 Singling out one circuit from their family of configurations, Moody and Florida went on to stress its crucial properties. “This circuit,” they claimed, “though relatively complex in itself, yields an excellent performance in the face of wide transistor and component variations.”80 With its emphasis on more complex, but ultimately more reliable, designs that exploited the special features of the transistor, the PNPN flip-flop became symbolic of the approach of Basic Circuits to the field of transistor electronics. Designing circuits in this way was no neutral act. First of all, the spaces in which novel design was possible within the Canadian military were tightly restricted, ever since the move to standardization in NATO in the mid-1950s 79. N. F. Moody and C. D. Florida, “Some New Transistor Bistable Elements for Heavy Duty Operation,” IRE Transactions on Circuit Theory CT–4 (1957): 241–61, on 241. 80. Ibid. HSNS3803_03 7/10/08 444 | 5:58 PM Page 444 J O N E S-I M H OTE P forced Canada to adopt existing U.S. designs for its equipment.81 Second, the act of design itself had come under attack as part of the broader program that saw human thought and action as the ultimate causes of electronic failures.82 Some, like Robert Lusser, the émigré German rocket engineer, located the problem in the moral failings of designers who needed to take a more responsible “life-and-death” approach to their work.83 Others, however, saw the problem as a wider, more fundamental issue of the engagement between humans and electronic devices. Criticism of electronics technicians and operators, for instance, had partly been an attack on the reliability of human senses and the functioning of the body when engaging electronics. The Reliability Bulletins, the series of publications issued by the Royal Canadian Air Force to “indoctrinate” personnel on reliability issues, argued that the human senses were “badly designed” to monitor electronic devices, since electronics ideally obeyed linear laws (according to which doubling input would double output, for instance), whereas human senses performed logarithmically.84 Moody himself suggested that errors in design stemmed not so much from lapses in judgment as from failures of intellect in the demanding and imaginative task of circuit design. Paired with concerns over the reliability of the senses, Moody’s anxieties about reason formed the other half of a growing pyrrhonism in relation to electronics, where neither the senses nor the intellect (and therefore neither technicians nor designers) could be wholly trusted. In no area were engineers more prone to error, nowhere could their faculties be more easily misled, Moody reasoned, than in moving from vacuum tubes to transistors. The source of error lay precisely in the constant, often 81. Chester Soucy, “Industry Comments and Queries from Management and Engineers Relative to Talks on Reliability Control by Mr. C. I. Soucy, AMC/STelO presented Apr–Oct 1957,” Appendix D, 2, NAC, RG24, accession 1983–84/049, vol. 1664, file 150–23–7. 82. M. L. Card, for instance, explained to the Institute of Radio Engineers Convention in Toronto that “the ultimate cause of unreliability is some form of human error,” and that the need, then, was to “ensure that the reliability requirement is clearly defined and policed at every stage.” See Card, “Electronics for Defence,” address to Institute of Radio Engineers Canadian Convention, Toronto, Canada, 1–3 Oct 1956, 4, NAC, RG24, acc. 1983–84/049, vol. 1664, file 1950-123-7, part 2. 83. Robert Lusser, “The Notorious Unreliability of Complex Equipment: A Condensed Presentation for Management Use” [technical report] (Redstone Arsenal, 1956); see also Soucy, “Reliability Control” (ref. 67), 222. 84. Royal Canadian Air Force, Reliability Bulletin, nos. 1 and 2 (1958): i; NAC, RG24, accession 83–84/049, vol. 1664, file 1950–123–7, part 2. HSNS3803_03 7/10/08 5:58 PM Page 445 ICONS AN D E LECTRON ICS | 445 relentless, connections drawn between the two devices. That connection had turned on what, to many technologists, was a critical functional similarity— the ability of both components to produce power amplification. Scroggie likely had precisely this similarity in mind when he argued, in text and illustrations, for the transistor as ersatz vacuum tube. For Moody, the property was trivial, but its implications dire. “To designers who are familiar with vacuum tubes,” Moody explained, “this trivial resemblance [of power amplification] . . . has been a trap. Many of their failures are strewn along the path of progress.” Basic Circuits had been created precisely out of the belief that the circuit techniques culled from experience with vacuum tubes were not simply transposed onto transistors.85 Moody’s flip-flop circuits were the best example. For Moody and his team at Basic Circuits, the key to the transistor’s robustness and performance lay in developing its own “associated circuit art.” Designers had to approach the devices fresh, learning entirely new circuit techniques, avoiding tube-by-tube replacement by transistors, redesigning existing circuits on a functional basis, and ultimately spending at least a year laboring with the new devices in order to master their techniques. If they could rise to the task, designers could wring from the transistor two unparalleled advantages: they could make it achieve a reliability that the vacuum tube could not even approach, and they could produce electronic effects forbidden to its thermionic cousin. If, however, they continued to design in error, by which Moody meant seeing transistors as tube analogues, they would produce only “the most dismal results.”86 Indeed, they already had. Insofar as the transistor behaved unreliably, and insofar as it generated failures, it did so because the design practices surrounding it were not equal to the task. The device had weaknesses, but few that robust designs could not overcome. The challenge of making the emerging electronics trustworthy was, therefore, the challenge of making engineers and circuit designers trustworthy by making them think in more reliable ways. Here Moody seized on Thompson’s symbol for the transistor. Moody was initially drawn to it because of the connections it made. Steeped in his work 85. Others shared this belief that the transistor was not a mere substitute for the valve. See Emerick Toth and William N. Keller, “Principles of Transistor Application and System Design,” in Symposium on the Application of Transistors to Military Equipment (New Haven, CT: Yale University Press, 1953), 360. 86. Moody, “Present State” (ref. 2), 182. HSNS3803_03 7/10/08 446 | 5:58 PM Page 446 J O N E S-I M H OTE P on pulse circuits in 1954, Moody had seen in Thompson’s symbol a crucial tool that he could use to think through the functioning of circuits, including asymmetrical flip-flop gates, leakage currents, and so on—things with no counterpart in vacuum tube electronics. He had personally waded into the controversy over symbols in February 1957, when he presented his innovative bi-stable circuits at the annual Transistor Circuits Conference, held in Philadelphia. Speaking alongside members from his team in Basic Circuits, Moody had sketched his innovative designs using Thompson’s symbols, one of five different symbols used at the conference. When engineers in the audience had complained “with intensity” about the lack of a standard representation, the guest editor of the proceedings, Howard Tompkins of the University of Pennsylvania’s Moore School of Electrical Engineering, who had developed his own alternate symbol, explained that the variety represented “either the last stand of rugged individualists in the face of oppressive conformity, or the last gasp of deluded non-conformists in the face of orderly progress, depending on your point of view.”87 Initiated into the controversy in February, in the midst of a battle at home over the place of design, Moody used the podium at the Nuclear Instruments Conference at CERN in early fall 1957 to emphasize the centrality of Thompson’s symbol in the transistor’s circuit art. The symbol had two key features that secured its primacy. First, it broke definitively with the vacuum tube. That is, it made a clear statement about what the transistor was not. No one leaning over the schematics produced by Basic Circuits could ever see a visual, and therefore a functional, similarity between transistors and valves. Severing that connection again and again using Thompson’s symbol, Moody contrasted the two devices throughout his presentation, pointing out crucial differences in gain and impedance or in the switching parameters so central to his own work. The second key feature of the symbol was that it made an equally emphatic statement about what the transistor was. Of all the symbols put forward for the transistor, Moody observed, only Thompson’s seemed to suggest the two key properties so crucial to good circuit design: that its action was based on rectifier-type junctions (like those in semiconductor diodes) which are crucial in pulse circuitry; and that it had a built-in time delay to all signals that traversed it. One could, of course, glean these insights from other sources; but the diode analogy led most directly to those realizations. In a remarkable synecdoche 87. Howard E. Tompkins, “Foreword to Transistor Papers,” IRE Transactions on Circuit Theory CT–4 (1957): 173. HSNS3803_03 7/10/08 5:58 PM Page 447 ICONS AN D E LECTRON ICS | 447 Moody’s depiction of the transistor as a diode assembly (top). The diagram (left) decomposes the NPN transistor into an equivalent circuit, reducing its workings to the action of resistors, capacitors, and, crucially, diodes. An alternate presentation from E. Baldinger (bottom) would draw incompatible connections to the vacuum tube. Source: Moody, “Present State” (ref. 2); E. Baldinger, “Transistors in Nuclear Instruments,” Nuclear Instruments 2 (1958): 193–202. Reprinted from Nuclear Instruments. Copyright (1958), with permission from Elsevier. FIG. 10 that echoed Thompson’s initial insight, Moody questioned his audience, “If a transistor is a diode, why not draw it as one?”88 In a testament to the multiplicity of meanings found in circuit symbols, Moody now broke the transistor apart visually, rendering the device as its own circuit: a combination of diodes, resistors, and capacitors that explained its functional characteristics (Fig. 10). (In the very next talk, his colleague Ernst Baldinger would give his own, completely incompatible rendering, drawing similarities to vacuum tubes.) Moody went on to demonstrate the key elements of the transistor captured by the diode analogy—resistance, current gain, and transit time. Having extracted the key parameters from the symbol, Moody 88. Moody, “Present State” (ref. 2), 183. HSNS3803_03 7/10/08 448 | 5:58 PM Page 448 J O N E S-I M H OTE P then showed how these features were especially central to pulse circuitry, precisely the kinds of applications he had encountered early on with atomic weapons and nuclear instrumentation, applications which he had continued working on at Basic Circuits and which formed the core of his audience’s interests. Getting transistors to work properly involved recognizing key features that were obscured by the analogy with valves. Lumping both together as triodes and active elements carried no weight with him. As he would later explain in his textbook on semiconductors and their circuits, “sound design” meant going beyond “the gross terminal properties of the active elements.”89 In the critical applications at the core of Cold-War science and technology, there was simply no substitute for the diode-based symbol. Only this symbol could hold the transistor’s essential characteristics constantly before the mind’s eye, guiding the act of design and therefore securing the reliability of everything from Doppler radar to particle detectors. The best way to make the transistor reliable was by adopting a reformed symbolism, by drawing it differently. CONCLUS ION In concluding his talk at CERN, Moody was clear: his presentation, launched and backed by icons, was meant to help particle physicists and engineers avoid the bypaths leading away from “the haven of safe, and reliable design.”90 For Moody, drawing similarities between transistors and vacuum tubes was not just inaccurate; it was dangerous. Wartime trigger circuits and nuclear instrumentation had taught him that the delicate lines of schematic diagrams ultimately governed the bulk and brawn of the emerging machines of Cold-War technology and science. Materials mattered—but getting the iconography of the transistor’s circuit art just right meant the difference between the catastrophic failures of Korea or the “reliability crisis” on one hand, and, on the other, the well-behaved invisibility of transistor circuits as they illuminated enemy aircraft or clicking away particle counts. In that appeal to symbols, Moody’s comment unsettles two views of transistors that have been at the center of our understandings of the devices—that making sense of them meant simply 89. N. F. Moody, Semiconductors and Their Circuits (London: English Universities Press, Ltd., 1966), vii. 90. Moody, “Present State” (ref. 2), 191. HSNS3803_03 7/10/08 5:58 PM Page 449 ICONS AN D E LECTRON ICS | 449 grasping their materiality, and that they substituted for vacuum tubes in the defining material rupture of twentieth-century electronics. Throughout the 1950s, even as technologists agreed to the letter on the structure of transistors and the materials that shaped their behavior, they could still disagree fiercely about what transistors were and how they functioned. As transistors replaced vacuum tubes in device after device, there were still spheres where that substitution was problematic and deeply contested. Diagrams and symbols, I have argued, were one of those sites, revealing the crucial tensions in transistors’ mid-century meaning and reality. The symbol that emerged from Basic Circuits grew precisely out of an attempt to embody graphically what Thompson saw as the transistor’s defining qualities—its core rectifying junctions and therefore its natural links to crystal diodes and to the larger family of semiconductor devices. Under attack from those who instead drew analogies with valves, Thompson and Bateson redeployed the symbol to show the parallel, but separate, graphical evolution that linked the devices to one another even as it kept their symbols distinct. For Moody, working in the midst of deep suspicions about the act of design, even those parallels cut too close. Fearing that conceptual and visual analogies between transistors and valves would lead engineers to commit costly and even fatal errors, Moody turned to the diode-based symbol as a way to discipline and guide the otherwise fallible mind of the engineer, and so make the transistor reliable. Drawing transistors in this way, tracing these particular platonic abstractions of thick material devices, became the first step along a path away from the “unsanitary,” insecure world of fighter jets smoldering in heaps on Korean mountainsides, or defense computers seizing under the burden of critical computations. Drawings were not just sites where concepts and ideas were worked out—they were instruments through which materials and objects gained meaning and acted back upon the drawings themselves. Debates over symbols were about devices; debates over devices returned once again to symbols and diagrams. They forced articulations of what kind of things electronic drawings ought to be: for Thompson, taxonomical expressions; for Bateson, syntactical constructions; for Moody, heuristic devices. Transistor symbols proliferated and clashed in the 1950s because of disagreements over what drawings ought to reveal—but also because they formed the leading edge in disputes over how the new devices ought to be understood. Because those debates were about symbols, we are inclined to see them as mere disagreements over conventions. But precisely because it was a battle over symbols, the terms of the debate ran toward the perceived essence of transistors and HSNS3803_03 7/10/08 450 | 5:58 PM Page 450 J O N E S-I M H OTE P how to capture it in graphical form. Drawing different transistor symbols entailed drawing specific genealogies through the populations of electronic devices—to and from diodes, and back and forth between vacuum tubes. To choose a specific symbol and to trace it in the late 1950s was to take sides in a debate over the nature of electronic devices. It meant situating oneself on either side of a divide between a view that saw transistors and vacuum tubes as functional kin, and one that divided those two devices by a rupture as radical, as deep and as disorienting, as the one between the classical and modern physics by which each was understood. Thompson’s symbols eventually fell out of use as the number of technologists using transistors in Canada expanded, and as their prior experience with the vacuum tube waned. The schematic diagrams that survive from those years remain as the sites of now-forgotten debates. Drawings were the places where electronics were first (imaginatively) assembled, erased, operated, and reconfigured, and therefore a key site for contests over the meaning of the devices that went into them. Exploring those drawings, alongside the objects they sought to represent and express, can help return the social study of electronics to the full substance and sweep of technical work by illuminating not just the stable material meanings given to the devices initially, but the multiplication of different, often incompatible, understandings on the blackboards, sketchpads, and drafting tables of the electronic arts. They give us a perspective perceived only darkly in the machines themselves, where those discussions had already ended—where transistors assertively replaced vacuum tubes, and where perspectives and solder hardened of a piece rather than remaining contested and fluid. AC K N OW LE D G M E N T S I would like to thank Michael Gordin, Yves Gingras, Richard White, Chen-Pang Yeang, W. Patrick McCray, and an anonymous referee for Historical Studies in the Natural Sciences for their very helpful and perceptive comments on earlier drafts of this article. I am also grateful to Chris Hume, Max Kaploun, and particularly Umberto De Gaetano for their research assistance, and to Philip Thompson for invaluable discussions, material, and insight. Part of the research for this article was completed with the support of grants from the Junior Faculty Fund of York University and the Social Sciences and Humanities Research Council of Canada.