DRAFT 10th January 2012 © John Howells & Ron Katznelson The Management of Overlapping Vacuum Tube Pioneer Patents in the Development of Radio – How the Legal Trajectories of Fleming’s Diode and De Forest’s Triode Patents Influenced Radio Commercialization By John Howells and Ron D Katznelson Abstract It has been proposed that in the scenario when multiple, independently-owned and perhaps over-lapping patents must be licensed for legal technology development, the difficulty of negotiating cross-licenses may hold-up or deter that development. Radio development in 1905-1920 was characterised by numerous independently-owned pioneer patents and we find numerous references in the literature to alleged impasse in bargaining the necessary patent rights for radio development among the owners or assignees of pioneer patents on vacuum tube technology. Accordingly, this paper re-examines the legal trajectories and exploitation of Fleming’s diode and De Forest’s triode1 patents on vacuum tube technology by means of the relevant historical record, patent claims, litigation records and other relevant law. We marshal substantial evidence and analyse with new precision how the overlapping patent rights of the Fleming diode and De Forest triode were resolved by the courts, by the immunity of suppliers to the Government from patent infringement liability, and we trace the cross-licensing agreements between the different radio interests that enabled state-of-the-art radio development. We find that contrary to various secondary source assertions, robust radio technology development ensued and licensing was chosen over hold-up even when the path to a full state of the art cross-licensing agreement was complex. 1 De Forest’s term for his device was the “audion” and “triode” was a later term that captures the distinguishing physical feature of the device that it had three electrodes enclosed within its glass envelope: triode is the term used in this paper. DRAFT 10th December 2012 © John Howells & Ron Katznelson 1 Introduction – The Proposition that Licensing becomes difficult in the scenario of ‘fragmented invention’ and may delay development This paper has the goal to present an accurate account of the role of pioneer patents in radio development with particular precision regarding patents alleged to overlap and patents alleged not to have been licensed or licensed with delay such that radio development was retarded. The scenario where development requires access to the technology controlled by multiple, independently-owned patents has attracted attention in recent years as it appears to be commonplace in the contemporary information and biotechnology fields. It has been proposed that in this scenario of ‘fragmented invention’ (Hall and Harhoff, 2012, p11) variously described as an ‘anticommons’ in biotechnology and a ‘patent thicket’ in software development (Heller and Eisenberg, 1998, Shapiro, 2001) there may occur an ‘impasse’ ie. difficulty in negotiating multiple licenses for access to necessary technology that may delay or deter development. Shapiro identifies an associated problem of ‘overlapping’ patents and in a recent review of the role of patents in innovation in the economic historical literature Moser describes these occurring when patent examiners issue patents that cover “overlapping areas of technology space” so that each of two firms can block the other’s access to the same technology leading to litigation and the discouragement of innovation (Moser, 2013, p33). According to the US constitution, patents should ‘promote… the progress of the useful arts’. In other words they should promote the progress of technology. If it can be shown instead that there occur impasses in the coordination of rights that ‘hold-up…’2 the progress of the useful arts, then some kind of remedial policy, a change in the patent law, may be appropriate. What matters is the evidence on firm and entrepreneurial behaviour: whether it can be shown that an impasse occurs in the negotiation of licenses between patentees and if it does, whether as a result hold-up in development of the technology occurs. Whatever the scenario is called and whether its description stresses the fragmentation or overlap of rights as the root cause, it is the difficulty of negotiation of patent licenses or the refusal to grant a license that is the proposed direct cause of development hold-up. The arguments made for this scenario in the biotechnology and software industries are theoretical and those who have attempted to counter them argue that there is no evidence of development hold-up while there is evidence of vigorous growth 2 ‘Block’ and ’hold-up’ have the essential same meaning in this paper, a delay in the process of development caused by patents. 1 DRAFT 10th December 2012 © John Howells & Ron Katznelson in investment, research and development and patenting activity in these industries (Buckley, 2007, Mann, 2004). It is in the economic historical literature that examples of this phenomenon have been proposed for a number of single pioneer patents. For example, it has been alleged that the Watt patent on steam engine technology held-up development (Boldrin and Levine, 2008) and it is known that Boulton and Watt did not license the Watt patent (Dickinson and Jenkins, 1927, p306). In this case there is a refutation of the assertions of hold-up based on an analysis of this patent’s claims, litigation history and later steam engine developments (Selgin and Turner, 2006, Selgin and Turner, 2009, Selgin and Turner, 2010). Another is Edison’s patent on the incandescent lamp (Leibowitz, 2002) which also was not licensed, but here it has been shown that the enforcement of the patent stimulated inventionaround the patent claims, so this patent stimulated technology development (Katznelson and Howells, 2012). These cases do show that when the pioneer patent holder does not license others directly to develop the technology it does not necessarily mean that development of the technology is blocked. In our review of the economic historical literature we were able to construct a prima facie case that there was both an anticommons scenario in radio and a significant case of overlapping pioneer patents for vacuum tube technology. Three features of radio development attracted our attention: first, there was a point in time, post-World War One when radio technology had diffused widely but patent licenses had not been negotiated; they now had to be negotiated for all patented radio technology for legal supply of radio equipment to commercial markets. Second, there are known to have been many pioneer patents, independently-invented and held and with no close substitutes.3 These included: Fleming’s patent on the diode and De Forest’s patents on the triode vacuum tube technologies; General Electric’s patents on the Alexanderson Alternator, crucial early long-distance broadcasting technology; Fessenden’s heterodyne and Armstrong’s superheterodyne pioneer patents on radio receiving circuits. Third and last, there were widespread statements that appeared to describe pioneer patents on vacuum tube technology as overlapping and further statements that licensing had proved so difficult that development of the revolutionary new amplification property of the triode had been delayed. 3 The invention and genesis of these technologies is told in the principle US radio histories H.G. Aitken. 1985. The Continuous Wave: Technology and American Radio, 1900 - 1932. Princeton University Press, Princeton, New Jersey, G.L. Archer. 1938. History of Radio to 1926. American Historical Society, New York, L.S. Howeth. 1963. History of Communications-Electronics in the United States Navy. US Government Printing Office, Washington, W. Maclaurin. 1971. Invention and Innovation in the Radio Industry. Arno Press and the New York Times, New York. 2 DRAFT 10th December 2012 © John Howells & Ron Katznelson For narrative purposes we take the last of these first. The allegations derive from attempts to interpret a 1916 court decision, the result of an infringement action by the Marconi Company against De Forest (Marconi Wireless Tel. Co of America v. De Forest Radio Tel. and Tel. Co., 236 F. 942, 946 (S.D.N.Y. 1916) hereafter (Marconi v. De Forest 1916)). This is widely-credited with a) giving the diode patent control over the physical properties of the triode and b) with the consequence that licensing was difficult or delayed. Indeed, this interpretation is an embedded ‘truth’ in the literature: one example of many4 is from the radio historian Maclaurin; ‘The court decided that De Forest had infringed the two-element Fleming vacuum tube, while Marconi had infringed the three-element De Forest patent. Neither company could manufacture the triode’ (Maclaurin, 1971, p85). We note at this point that Maclaurin’s statement is not specific about which claims of the diode patent were infringed. Furthermore, as we shall see, the claims of the Fleming diode patent were drawn such that they encompassed the De Forest triode patents: this appears to be a true and litigated case of ‘overlapping patents’. 4 Three other post-1945 historians of radio cite Maclaurin’s interpretation of the 1916 decision rather than the decision itself: Douglas cites Aitken, and Aitken and Reich cite Maclaurin H.G. Aitken. 1985. The Continuous Wave: Technology and American Radio, 1900 - 1932. Princeton University Press, Princeton, New Jersey, S.J. Douglas. 1987. Inventing American Broadcasting, 1899 - 1922. Johns Hopkins University Press, Baltimore and London, L.S. Reich. 1977. Research, Patents, and the Struggle to Control Radio: A Study of Big Business and the Uses of Industrial Research. Business History Review. 51(2) 208-235..4 A fifth, the Naval communications historian Howeth, cites the decision directly but expresses the same interpretation, ‘Neither Marconi nor De Forest nor any other company could legally manufacture the badly needed three-element tube without the mutual consent of the two interested parties’ L.S. Howeth. 1963. History of Communications-Electronics in the United States Navy. US Government Printing Office, Washington.. The most explicit version of this interpretation can be found in an early FTC (Federal Trade Commission) report on radio, ‘the court holding that the Fleming patent dominated the use of a vacuum tube either as a detector, repeater, amplifier, or oscillator. As a result of this decision the De Forest Co. ceased manufacturing vacuum tubes in their improved form Federal Trade Commission. 1924. Report of the Federal Trade Commission on the Radio Industry In Response to House Resolution 548, 67th Congress, 4th Session. Government Printing Office, Washington.. Finally, legal scholars have accepted and cited to these early errors: Kitch cites Maclaurin’s interpretation of the decision E.W. Kitch. 1977. The Nature and Function of the Patent System. Journal of Law and Economics. 20 265-290. and Merges and Nelson cite the FTC report R.P. Merges, Nelson, R.R. 1990. On the Complex Economics of Patent Scope. Columbia Law Review. 90(4) 839-916. 3 DRAFT 10th December 2012 © John Howells & Ron Katznelson The next section develops focal questions to structure the investigation into the primary legal sources derived from analysis of the interpretation quoted above and other elements of our review. 2 Focal Questions The primary focus of this article is to provide a detailed answer to the following question: Question (1): Given their initial distribution and subsequent litigation how were pioneer patent rights managed to coordinate development? Much of the answer will concern how the Marconi Company and De Forest sought to enforce their vacuum tube technology rights with respect to one another. Given the widespread interpretation of the 1916 court decision such as that cited above, we also ask; Question (2): Whether the commercial evidence for development of vacuum tube technology supports the idea of an impasse in negotiations between the rights-holders. After World War One (WW1) radio technology had become widely-distributed without permission of the many pioneer patent right owners and yet when civil markets for radio reopened, permission would once again be necessary to avoid an infringement suit. This would appear to be a unique situation and we ask; Question (3): What was the legal context which enabled the scenario of widespread diffusion of radio technology without licensing of rights? Even Brady’s 1922 account of radio patents and litigation uses the word ‘deadlock’ to describe the postwar situation in radio patents (Brady, 1922, p277). The prospective return to civil radio markets can be seen as an ‘empirical-historical’ test of the prediction that the ‘anticommons’ or ‘patent thicket’ scenario is too complex for timely negotiation of appropriate licensing contracts. RCA was created as a solution to the post-War radio rights distribution but we ask: Question (4): In this unique situation is there evidence that it was difficult to cross-license pioneer patents to enable state of the art radio operation when it was first commercially feasible to operate? 4 DRAFT 10th December 2012 © John Howells & Ron Katznelson In practice we: search for reviews of the technology at issue and published close in time to the events of interest;5 we seek primary source information contained in patent specifications and especially their claims and relevant court decisions. Furthermore, secondary accounts often neglect information on licensing and such information is known to be often difficult to collect because licensing contracts are privately negotiated (Kitch, 2000). We therefore conduct a wide search for evidence of licensing of the pioneer patents at issue and do not assume an absence of evidence in otherwise authoritative monographs is evidence that licensing did not occur. To further focus our account on specific patents and events, we conduct a preliminary and selective review of assertions in the secondary literature that licensing of pioneer patents proved difficult or failed. 2 Development of the Triode and Licensing of the Diode Patent 2.1 Development of the diode and triode vacuum tube technologies as radio detectors 1905 to 1918 Inspection of the Marconi Company’s Fleming vacuum tube diode patent, U.S. Pat. 803,684 reveals that it was issued November 7th, 1905, fully ten years before suit was brought against De Forest’s commercial triode product while De Forest’s vacuum tube triode patent U.S. 841,387 on the amplification property of the triode issued January 1907, over eight years earlier than the commencement of Marconi’s suit in 1915. The answer to question 1, above, should begin with an account of vacuum tube technology development and the exercise of these patent rights in the years leading up to the suit. We note that court documents can be important records of a technology’s market significance and that when the US Government was sued for compensation for its unlicensed use of the Fleming diode in the Court of Claims, the Court, as part of its assessment of the technological significance of this patent made an analysis of the market for the Fleming vacuum tube detector. It found that the Fleming diode tube ‘did not go into any extended commercial use, only about 700 Fleming valves [vacuum tubes] having been shipped by plaintiff to anyone between 1910 and 1915’. Moreover, the Court found that some of the many crystal detectors then in use were ‘much more effective than the Fleming apparatus’ (Marconi Wireless co. v. US, 81 Ct. Cl. 671, 783 (1935)). This statement probably refers to the Dunwoody carborundum crystal detector, for according to Morse’s review, the Dunwoody detector was the best detector before the advent of the effective triode technology. Furthermore, it is telling that it 5 What was understood as the ‘state of the art’ in the past may be more relevant for understanding past events than the current state of the art. 5 DRAFT 10th December 2012 © John Howells & Ron Katznelson was the Marconi Company that used these crystal detectors and exclusively for transatlantic services and radio-equipped ships in preference to its patented diode vacuum tube detector technology; the crystal detectors remained in use into the 1920s (Morse, 1925, p36, Morse, 1943, p636). Marriott’s 1917 paper reviewing diffusion of different radio technologies in the US to that year, confirms this picture. Marriott charts the growth and decline of several detector technologies and describes them,7 but the Fleming vacuum tube does not feature (Marriott, 1917, p184). The Fleming diode detector was a minor technology in competition with superior substitutes and not even favoured for important uses by its owner, the Marconi Company. Marriott does describe the late rise in importance of De Forest’s triode detector (the “Audion”): it was used ‘in very small numbers until about 1912’ (Marriott, 1917, p184). De-Forest Triode Sales Units 18,000 All sales To U.S. Government 16,000 14,000 12,000 10,000 8,000 ? 6,000 4,000 2,000 ? 0 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 YEAR Figure 1. Sales of De Forest-manufactured Triodes 1909-1918. Sources: ‘All sales’ from Tyne’s figures for McCandless’ (De Forest’s contract manufacturer) supplies of triodes to De Forest (Tyne, 1977, p108-111, 131) (data for 1915 estimates the last quarter’s sales by the average monthly sale of the first nine months); ‘To US Government’ based on De Forest’s sales to the US Government until February 1919, as reported in aggregate over multiple year blocks in (Tyne, 1977, p130-131) and estimated for each year by a proportional allocation for each triode type. 7 Marriott also reports on patent litigation in radio before 1915 and on the commercial activities that ensued under patent rights concerns at page 93, R. Marriott. 1917. United States Radio Development. Proceedings of the Institute of Radio Engineers. 5(3) 179-195. He details how in response to patent assertions, non-infringing solutions were developed, such as the crystal detectors that replaced electrolytic detectors, the loop antenna replacing other patented antenna. 6 DRAFT 10th December 2012 © John Howells & Ron Katznelson In figure 1 we chart De Forest triode sales data from tables available in Tyne’s monograph on vacuum tube technology (Tyne, 1977, p110-111, p131) to more readily identify when the triode detector became commercially important. The inflection in triode production from 1912 is evident.11 Important evidence bearing on the question when vacuum tube technology became commercially significant can be found through inspection of the U.S. Fleming diode patent issued in November, 1905. This reveals a peculiar feature: Fleming added a disclaimer to limit the scope of Claims 1-6 and 10-15 in November 1915, no less than ten years after issue. The disclaimer limited the application of the Fleming diode to a detector of “high frequency alternating electric currents or electric oscillations of the order employed in Hertzian wave transmission” – in other words, it retrospectively limited the claims to a radio frequency detector. Claim 1 of the US Fleming patent originally read, ‘The combination of a vacuous vessel, two conductors adjacent to but not touching each other in the vessel, means for heating one of the conductors, and a circuit outside the vessel connecting the two conductors.’ The district court accepted the late disclaimer and construed Fleming’s claim as limited to radio frequency detection, but at the same time observed that the Edison tube was prior art encompassed by the scope of the original Fleming claim (Marconi v. De Forest 1916). Such prior art was Edison’s patent 307,031, as also observed later by the Court of Claims Marconi Wireless Co. v. US, 81 Ct. Cl. 671, 732 (1935).13 This disclaimer limiting the scope of the Fleming diode patent and avoiding Edison’s tube as prior art was therefore correctly judged to be necessary if litigation were to proceed and it was indeed added to the Fleming patent just months before Marconi’s assertion of the patent against De Forest. But most important is that it was found necessary, yet added ten years ‘late’; this tells us that the Fleming diode was only seriously examined for its potential to serve in litigation ten years into its term. It is reasonable 11 This growth matches the market share growth of “Audions” that Marriott reports in Item 5 in Chart 2 R. Marriott. 1917. United States Radio Development. Proceedings of the Institute of Radio Engineers. 5(3) 179-195.. 13 Years later, the Supreme Court would hold that the 1916 district court had erred in accepting such a late disclaimer and that it therefore followed that by reason of the Edison prior art that the Fleming diode patent was invalid (Marconi Wireless Telegraph Co. of America v. US, 320 US 1 (1943)). Such a late decision had importance in deciding the retrospective compensation owed the legatee owners of the Fleming patent for that patent’s unlicensed use by the US Government (the issue before these Courts), but it is the 1916 court decision that matters for the issue of radio technology development. 7 DRAFT 10th December 2012 © John Howells & Ron Katznelson to infer that litigation had not been considered seriously before. This conclusion is consistent with the commercial facts outlined previously that tell us vacuum tube technology was of little commercial significance until some years after 1912. Indeed, this appears to be a nice example of how a patent is not necessarily asserted against a product believed to be infringing unless it is commercially worthwhile to do so. The Marconi Company would win its suit for infringement of the Fleming diode patent against De Forest’s commercial triode radio detector device in 1916 (Marconi v. De Forest 1916), affirmed 243 F 560 (2nd Cir. 1917), hereafter (Marconi v. De Forest 1917). Figure 1, above, confirms that the triode was in uninterrupted development before, during and after this suit. This contradicts the widespread interpretation of the 1916 decision referred to in the introduction and so it will be necessary to consider two legal events that enabled this uninterrupted development; first, the 1910 law providing immunity to infringement for suppliers of patented goods to the government and second, the narrow scope of the 1916 decision in Marconi v. De Forest. 2.2 The Legal Background to Radio Development During Wartime and the 1916 Decision Congress passed the June 25, 1910 Act enabling patentees to sue the Government in the Court of Claims for just compensation when the Government used infringing products ordered from contractors.14 By 1914 the prevailing case law including the Crozier Supreme Court decision in 1912 construed the 1910 Act as precluding patentees from having any patent infringement cause of action against equipment suppliers to the Government and held that only the Government could be sued to recover lost royalties, although it could not be enjoined from procuring and using infringing products.15 Owing to this law, there had been many, including numerous new manufacturers who had, at the Government's request, supplied infringing products to the Government with effective immunity from 14 15 Public Law 61-305, Ch. 423, 36 Stat. 851, (June 25, 1910). Crozier v. Fried. Krupp Aktiengesellschaft, 224 U.S. 290, 301 (1912) (Ruling the district court's dismissal as proper and instructing the plaintiff to file a complaint directly against the U.S. government in the Court of Claims to recover infringement damages); See also Marconi Wireless Telegraph Co. of America v. Simon 227 F. 906 (S.D.N.Y. 1915), aff’d, 231 F. 1021 (2d.Cir. 1916), rev’d, 246 U.S. 46 (1918); International Curtis Marine Turbine Co v. William Cramp & Sons Ship & Engine Bldg Co. 232 F. 166 (D.C.Pa. 1916) (Under Act June 25, 1910, authorizing the United States to appropriate a license to use any patented invention, a contractor, who makes a patented device for use of the government, is not liable for infringement), rev’d, 246 U.S. 28 (1918). 8 DRAFT 10th December 2012 © John Howells & Ron Katznelson infringement suits (Howeth 1963, p374). This state of the law remained until March 1918.16 Until then, federal district courts would dismiss any patentee’s suit against radio equipment manufacturers, directing the patentee to file instead suit against the U.S. Government in the Court of Claims to recover any infringement damages (Crozier, 224 U.S. 301). Moreover, because the Government could not be enjoined from procuring infringing products (Crozier, 224 U.S. 308), no radio supplier to the Government was ever at risk of having their radio sales halted on account of patent rights. The Fleming diode patent could not have controlled production of the triode by any supplier to the Government and as shown in Figure 1, the Government was a source of demand far greater than the pre-war civilian markets for radio. Figure 1 prompts us to question what was the nature and scope of the 1916 court’s decision and what did it mean for the Marconi Company’s control of the revolutionary properties of the triode? It is clear from the text of the 1916 decision that the Marconi suit against De Forest concerned his sale of his ‘PN Type Audion [triode] De Forest Detector’ (emphasis added) – in other words his implementation of the triode as a radio detector product and that the Court duly found to have infringed claims 1 and 37 of the Fleming diode patent17 (Marconi v. De Forest, 1916 at 955). This decision had nothing to do with De Forest’s control of the radical, amplification property of the triode – this is made especially evident in the text supporting the decision where the Marconi Company admits infringement against claims in two of De Forest’s US triode patents no. 841,387 and 879,532 that claimed the triode’s ability to amplify (Marconi v. De Forest, 1916 at 942). This decision did not render the Fleming diode patent controlling for triode technology implemented for amplification, oscillation or for any of the other revolutionary properties of the triode. 16 On 4 March 1918, the Supreme Court decided two cases, Cramp v. Int’l Curtis Marine Turbine Co., 246 U.S. 28 (1918) and Marconi v. Simon, 246 U.S. 46 (1918), that reversed the decisions previously rendered by the lower courts and held that patentee’s right to recover damages from the U.S. Government does not absolve a directlyinfringing manufacturer from liability. Only then did the Government find itself confronted by manufacturers’ threats to cease production unless the Government provided indemnity, which it promptly did by way of specific indemnification letters sent to its suppliers. Congress, by an act of July 1, 1918 (Ch. 114, 40 Stat. 705), reiterated the intentions contained in the act of 1910, amended it and codified in statute the Government indemnification given to suppliers. See (Howeth 1963, 373-375). 17 De Forest’s attempt to counter-sue for infringement of a range of his audion patents was dismissed, 9 DRAFT 10th December 2012 © John Howells & Ron Katznelson On the other hand precisely because the Marconi Company admitted infringing De Forest’s patents on the revolutionary amplification property of the triode, we know they deployed the triode without permission but were now volunteering this fact and were prepared to settle with De Forest. We could not find evidence that De Forest thereafter licensed the Marconi Company to use the triode following the 1916 decision, but we note that as stated above De Forest, when selling the important rights of the triode to AT&T, specifically retained the non-transferable, personal right to license the Marconi Company to use the triode (Tyne, 1977, p115) – we suggest that he did so. To return to Maclaurin’s (and other historians’) rendering of the decision quoted in section 2, his assertions are a misreading of the 1916 court decision because, first, they are not specific about which claims were infringed; the Marconi Company admitted infringement of the amplification claims in De Forest’s patents and won its suit on the basis that De Forest’s commercial device infringed the diode patent claims concerning rectification. Only for this purpose could De Forest be precluded from making triodes; De Forest was not precluded from making triodes per se. Second, whatever the scope of the decision, it does not in itself preclude manufacture; licensing would enable manufacture and the absence of evidence of licensing in these sources does not mean licensing did not occur. Indeed, we show below that the companies subsequently decided to cross-license to enable the manufacture of triodes. This mistake in the secondary literature is necessary has not been pointed out before but it matters because it has helped mislead legal analysis of the decision (we make no criticism of other aspects of these historians’ work and of course the error is not intentional). This has had the dismaying consequence that erroneous legal analysis is being fed back into the patent law policy process.18 It may nevertheless be asked what use the Marconi Company made of the 1916 court decision upholding a narrow scope interpretation of the diode. In principle the injunction granted, but suspended pending appeal in 1916 was rendered enforceable from May 191719 – but the US had entered the First 18 There is insufficient space here to document fully how this mistake by Maclaurin has influenced later scholarship; but the influence is wider than the histories cited here eg. Sabety argues nanotechnology patents today resemble the scenario in radio patents described by Maclaurin and allegedly holding back radio development: therefore Sabety advocates a liberal public licensing policy for foundational patents generated by publicly funded nanotechnology research today T. Sabety. 2005. Nanotechnology Innovation and the Patent Thicket: Which IP Policies Promote Growth? Albany Law Journal of Science and Technology. 15 477-515.. 19 The 1917 decision refers to the existence of an injunction, but the Westlaw version of the 1916 decision does not contain the addendum which details that in September 1916 an injunction was granted. The addendum can be found attached to a copy of the trial notes in an appendix to Fleming’s book ‘The Thermionic Valve’ J. Fleming. 1919. The Thermionic Valve and its Developments in Radiotelegraphy and Telephony. Wireless Press, London. 10 DRAFT 10th December 2012 © John Howells & Ron Katznelson World War (WW1) in April 1917. The Government became the only significant customer for triodes and as described above, the prevailing law pertaining to infringement suits against suppliers to the Government protected De Forest from infringement liability. Only with the Post-War resumption of civil markets for radio could the Marconi Company possibly have enforced the diode patent on De Forest’s triode implemented as a radio detector product. 2.3 Evidence that the revolutionary properties of the triode were in uninterrupted development A difference in the terms of the Fleming diode patent in the US and UK allows us to make an important control on the role of the Fleming diode patent in development by comparing US with UK radio development. Once the US legal situation had been settled in its favour the Marconi Company made the same disclaimer to its British diode patent prior to an application to extend the term of that patent. The application was unsuccessful and the Marconi Company’s control of the diode ended in Britain in 1918 (Morse, 1925, p38). There was therefore no presumed-valid, pioneer diode patent in Great Britain from this date and the diode technology would be freely available in Great Britain.20 If US radio development had been retarded by the refusal of the Marconi Company and De Forest to license each other or by the difficulty in negotiating a license for the Fleming diode patent, we should expect British development to be ahead of the US. Of the many candidates for a marker of relative ‘advance’ in radio development with which to compare these two countries, we may take the dates for first continuous-voice broadcast entertainment in Britain and the US. Here we find the US arguably one year ahead of Britain which tells us that the Fleming diode patent, licensed or not, did not retard US radio development.21 As a matter of logic, if the diode effectively controlled the technology of the triode it would not have been possible to sell licenses to develop the triode without a license from the owner of the diode. But open copyright pdf version available through Google books. 20 Of secondary importance, De Forest’s British patent on his triode detector (No. 1,427/08) had lapsed in Great Britain in 1911/1912 when he failed to pay the renewal fee, whereas the US equivalent patent would expire in February 1925 A. Morse. 1925. Radio: Beam and Broadcast - Its Story and Patents. Ernest Benn Ltd, London.. 21 The first British (and European) transatlantic continuous voice broadcast was in March 1919, but AT&T had achieved this in 1915, although with over 300 vacuum tubes compared to the British three vacuum tubes in 1919 (but US vacuum tubes had by the later date also evolved greatly and RCA began commercial transatlantic broadcasts in 1920, see below) W. Baker. 1970. A History of the Marconi Company. Methuen, London. The first entertainment broadcasts occurred in the US in 1919 and in Britain 1920 ibid. and no US lag behind the UK is evident. 11 DRAFT 10th December 2012 © John Howells & Ron Katznelson after 1912 when the advantageous properties of the triode were recognised it was licensed for development and without mention of a corresponding license from the Marconi Company: AT&T bought the non-exclusive rights to use the triode in telephony in 1913 for $50,000, then as they developed it and understood its potential, the radio rights in 1914 for $90,000 and finally in 1917 AT&T would pay $250,000 for outstanding rights to De Forest triode patents, including the pending feedback and ‘regenerative’ patents crucial for the development of vacuum-tube broadcasting technology (Archer, 1938, p135). There is evidence that the prospective legal outcome of an infringement suit was not deterrent: AT&T’s legal department had assessed the risk that De Forest’s product would be subject to infringement suit by the American Marconi Company at the time AT&T purchased radio rights to De Forest’s patent in 1914: but the company proceeded anyway (Aitken, 1985, p247, Hijiya, 1992, p91). By 1915 AT&T had implemented the first transcontinental telephone lines with the aid of the improved triode acting as a wire amplifier. So the ‘revolutionary’ properties of the triode were in commercial use in the form of a telephone wire amplifier before, during and after the 1916 and 1917 court decisions. While such significant development occurred outside De Forest’s company, in his license agreements with AT&T De Forest had retained a personal, non-transferable or assignable right to manufacture his triode device for sale direct to amateur users and to the US Government - and the right to license the Marconi Company to use the triode (Tyne, 1977, p115). It is evident that the triode was being sold to and used by the US Government, amateur users and AT&T in the periods up to and after the court decisions of 1916 and 1917 and it is sales to these customers that are captured in figure 1. Government also intervened directly in radio by Executive Order following the declaration of War in April 1917. President Wilson ordered the Navy Department to take over ‘such radio stations within the jurisdiction of the US as might be required by the Naval Communication System’ and to close all stations not necessary to the Government (Howeth, 1963, p237). The US Navy also controlled amateur reception and there was therefore no legal civilian market in radio equipment during the War years and therefore no issue of infringement in the supply of such markets. It is ironic, since this is an investigation into the role of patents in development, that it was a Government-mandated closure of amateur stations that eliminated the amateur operators who had provided the major civil market for De Forest’s triode detector and who had supported the growth of his business until the War (Carneal, 1930, p273). The significance of these interventions for our inquiry is that the issue of how patents control development can be considered suspended during the War, while technology diffusion and development 12 DRAFT 10th December 2012 © John Howells & Ron Katznelson proceeded apace. The War saw the US Navy set and meet ambitious development targets for the supply of vacuum tubes, establishing long production runs and component standardisation (Howeth, 1963). The corporate R&D laboratories of AT&T and General Electric were important contributors and Tyne recounts multiple avenues of investigation into the tube’s functions which radically altered it from the primitive, erratic craft product with short life span made for De Forest by his manufacturer McCandless, into a reliable component for advanced applications (Tyne, 1977, p86-88). Many firms had exploited their freedom to use patented technology for supply to the Government during the War (Howeth, 1963, p374) and at its end in November 1918 they were left in possession of much state-of-the-art radio technology that would again be subject to infringement suits by the relevant patents should they now supply civil markets. This unique situation might be considered a ‘natural experiment’ on the behaviour of firms that find themselves in a patent thicket or ‘anticommons’: would there be an explosion of infringement suits; bilateral licensing negotiations so complicated that agreement could not be reached; or would there prove to be several paths available to attain a crosslicensing agreement to enable legal commercial operation? 2.3 Post-War Resumption of Civil Radio Markets We are now in a position to answer question 4; how did the Marconi Company use the diode patent once it became enforceable on the field and how was the transition managed from the widespread distribution of patented technology at the end of the War to the legal supply of civilian markets? In the answer to both questions it matters that the US Navy relinquished control of radio stations in stages to private operators and much later than the end of the War in November 1918.28 2.31 Marconi Company Licensing of the Diode Patent and Continued Development of the Triode The amateur markets upon which De Forest had depended prior to the outbreak of the War were made legal again when the US Navy allowed amateur radio reception to resume in April 1919 and in September 1919 when amateur transmission upon application and receipt of a license was allowed (Douglas, 1987, p298). The earliest date at which the Marconi Company might choose to enforce injunctions against De Forest’s company would therefore be after April 1919. This never happened because immediately after the War on 30th November 1918 the Marconi Company granted a 28 The reason was that Secretary of the Navy Daniels was engaged in an (ultimately futile) attempt to have Congress pass a bill granting the Navy permanent ownership of radio assets and was reluctant to return these assets to private hands G.L. Archer. 1938. History of Radio to 1926. American Historical Society, New York.. 13 DRAFT 10th December 2012 © John Howells & Ron Katznelson nonexclusive license ‘to make and sell equipment under the Fleming patent to amateurs and any Government’ to Moorhead Laboratories (Tyne, 1977, p171). Moorhead Laboratories was the most successful of the new entrant, ‘independent’ manufacturers of triodes with production during the War for the Allied Governments having reached 30,000 per month (Electrical World, 1919, p1349). In return for the diode license, Moorhead admitted infringement of the diode patent, paid damages and agreed to pay future royalties to the Marconi Company (Tyne, 1977, p171-172). Moorhead now began negotiations with De Forest for a license to use the triode patents which resulted in an agreement signed April 30th 1919 and intended to run to 1925, the agreement being that Moorhead would manufacture tubes for De Forest under his personal license, ‘Then De Forest was to sell all such tubes to the Marconi Company, which agreed to act as distributing agent and sell tubes back to the De Forest Radio Telephone and Telegraph Company and to other purchasers. These tubes were to be sold to the public for amateur and experimental use in radio reception and amplification’ (Tyne, 1977, p172). So months before the Marconi Company was able to use its power of injunction against De Forest, it chose to negotiate a complex licensing agreement that exploited the manufacturing capacity of a third party to once again produce the triode deployed as a radio wave detector for amateur markets.29 The object of the Marconi Company appears clear enough: to establish a royalty income. 2.32 Licensing Pioneer Patents for State-of-the-Art Broadcasting through RCA and the Disruptive Role of the US Navy The fourth question also concerns whether the licensing that resulted in the creation of RCA can be considered difficult and to have delayed development: an important control on this idea is to ask whether RCA, the eventual result of the cross-licensing arrangement, was able to engage in commercial civil radio broadcasts on the first date that this became possible after the War. The US Navy returned its Alexanderson Alternator ‘high power stations’ to their private owners on 1 st March 1920 (Archer, 1938, p188) - and this was the date that RCA began its first commercial radio broadcasts from the US to England with this equipment. As for the resumption of triode radio detectors for the civilian market, the US Navy, not patents, were the ‘block’ on the start of civilian transatlantic continuous voice broadcasts 29 This agreement was cancelled effective July 30th 1920, but De Forest and Moorhead continued manufacturing triode tubes through various arrangements and companies G. Tyne. 1977. Saga of the Vacuum Tube. Howard Sams and Co. Inc., Indianapolis, Indiana.. 14 DRAFT 10th December 2012 © John Howells & Ron Katznelson by the state-of-the-art Alexanderson Alternator. Whatever licensing negotiations were necessary had been achieved prior to the first possible date for commercial long distance broadcasting. There is also the issue of whether the patent situation prior to the start of long-distance broadcasting could be considered an ‘impasse’ in the sense that it was so complicated that it could not be resolved by private parties, but required the intervention of the US Navy. The US Navy did intervene, but to prevent a licensing deal being struck: when the US Navy heard that GE had resumed its pre-War negotiations to sell its Alexanderson Alternator patent rights to the British-owned Marconi Co. in late February 1919, it persuaded GE to break off negotiations (Aitken, 1985, p323, Archer, 1938, p160-165); it feared that this technology would help consolidate the British company’s monopoly in world communications and so its motivation was ‘American-national,’ not commercial and nothing to do with patent ‘logjams’ (Howeth, 1963, p353-358). Aitken, using GE company archive data, writes that GE was willing to abandon negotiations with the Marconi Co. only because it came to believe that the US Navy had promised the creation of a national monopoly radio company which would become the alternative buyer to the Marconi Co. of GE’s Alexanderson Alternator (Aitken, 1985, p322-354). This story is not yet complete because strictly speaking, the US Navy had no role in the specific design and creation of RCA: having broken off the GE - Marconi negotiations, by May 1919 it no longer gave clear directions or any plan of action to GE for how a Government-mandated American national monopoly radio developer company was to be created. There was now a delay of months – a US Navyinduced delay - until GE, growing impatient, took the lead in the negotiation of a series of patent crosslicensing deals that created RCA (Howeth, 1963, p358). The consequence was that RCA would be of GE’s design. The first action GE took met the US Navy’s earlier expressed concern that the technology be kept in US hands: GE and the American Marconi Company persuaded the British Marconi Company to sell its shareholding in the American Marconi Company to GE in September 1919 (Archer, 1938, p178). The Radio Corporation of America was granted a charter in October 1919 and by November 1919 GE had arranged for the American Marconi Company and RCA to be officially merged to create an American-owned radio operating company (Archer, 1938, p180). A patent cross-licensing agreement of November 1919 allowed RCA access to current GE patents including the Fleming diode patent and rights in exchange for GE access to current and future RCA patents and rights. At this point the complexity of this series of actions necessary to meet the US Navy expressed ‘need’ for a US national radio company might usefully be compared with the relatively simple bilateral 15 DRAFT 10th December 2012 © John Howells & Ron Katznelson licensing negotiations between GE and the Marconi Company which the US Navy had interrupted. It appears to us that the US Navy’s non-commercial, nationalist intervention rendered more complicated the path to an effective cross-licensing agreement. Finally, a continuation of the above account should briefly include how the other listed pioneer patents (see introduction) were licensed. The most important of these were the rights to the triode for radio that De Forest had licensed to AT&T and AT&T’s subsequent patents on triode development. GE only secured the agreement of AT&T to join the RCA cross-licensing arrangement on 1st July 1920 (Aitken, 1985, p445) after RCA had begun commercial broadcasts. Negotiation could be described as ‘difficult’, but it did not hinder commercial operation of the technology. Indeed, Aitken writes that these companies had signed a provisional agreement to create a jointly-owned radio company with exclusive rights to all vacuum tube patents as early as 29th May 1918 (Aitken, 1985, p434), which suggests that these companies would have arranged some kind of cross-licensing deal on vacuum tube technology patents, if RCA had not been the means by which cross-licensing was in fact arranged. On the other hand there is some irony when we observe that while the 1916 decision has been widely and incorrectly attributed with giving the Fleming diode patent blocking power over the development of important physical properties of the triode, an entirely unremarked July 1919 court decision did find De Forest’s ‘Oscillion’ to infringe the Fleming diode patent and so did give the diode patent that theoretical ‘blocking power’ over the important oscillation property of the triode,36 a property fundamental to the future development of radio transmission and reception37 (Marconi Co., v. De Forest Radio Telephone and Telegraph Co, 261 F. 393 (SDNY 1919). Without access to the records of negotiation between GE and AT&T we can nevertheless guess that what the 1919 decision achieved in practice was an increase 36 For example, Armstrong’s regenerative receiver circuit exploited the oscillation property of the triode. Circuit diagrams of the regenerative receiver circuit including a triode operating as an oscillator can be viewed at http://www.electronixandmore.com/articles/oscillators.html. On January 20th 1922 another infringement suit would end with the reversal of the July 1919 decision so that the diode patent controlled only the detector application of vacuum tube technology again Radio Corporation of America v. Radio Audion Co. 1922, 278 F. 628 37 However such commercial applications of vacuum tube technology for radio were still under development. Neither AT&T nor GE had yet achieved development of the power vacuum tube to the point where it could substitute for the Alexanderson alternator as a superior broadcasting technology. GE would first develop prototype broadcasting equipment based upon its 20kW pliotron vacuum tubes in 1922 J. Brittain. 1992. Alexanderson. Johns Hopkins University Press, Baltimore, Maryland. and the Fleming diode patent scope was reduced in scope in January 1922, expiring in April 1922. 16 DRAFT 10th December 2012 © John Howells & Ron Katznelson in the strength of GE’s negotiating position as it sought to bring AT&T and its advanced triode technology into the RCA cross-licensing agreement. The last of the important pioneer patents cited in the introduction as forming part of the potential ‘anticommons’ in radio are the heterodyne and superheterodyne receiving circuit patents. When RCA began to broadcast continuous voice radio in 1920 it was point to point, it worked, but RCA had not understood the value of the receiving circuit technologies and Westinghouse pre-empted RCA in acquiring the heterodyne patents and an option on Armstrong’s superheterodyne circuit patent, issued to Armstrong in June 1920 (Aitken, 1985, p467-468).38 RCA acquired the necessary licenses to this fundamental receiving circuit technology in exchange, once again, for the admission of Westinghouse to the GE-RCA-AT&T patent cross-licensing agreement (Aitken, 1985, p478-479). 5 Discussion and Conclusion – The Fleming Diode Patent was Licensed to Facilitate Continued Production and Development In answer to the questions posed in the introduction: first, the 1916 decision in Marconi v. De Forest found the De Forest radio detector product to infringe the Fleming diode patent and the Marconi Company could not control the revolutionary properties of the triode. Indeed, the Marconi Company admitted infringing De Forest’s patents on the amplification property of the triode. Before this decision and in the early years of vacuum tube technology there was so little commercial value in this technology implemented as radio detector products in a field where better, substitute detector technologies were traded that it was not worth bargaining to license or to litigate. The Marconi Company brought suit against De Forest late in the term of the diode and triode patents but when commercial returns to De Forest’s triode product were becoming significant. When civil markets were in prospect post-1918 and so there was a prospect of enforcing the diode patent, the Marconi Company chose instead to facilitate a three-party cross-licensing arrangement for the continued production of the triode for the amateur radio market. Second, commercial evidence of De Forest’s triode production shows an unbroken rise in annual production and does not support the notion of development hold-up. Third, the complex legal situation post-War was a result of the US Navy having closed civil radio markets and suppliers of patented radio equipment to Government being immune from infringement 38 The superheterodyne circuit was a means of stabilising high amplification radio receiving circuits and would prove basic to all future receiving technology in the broadcast-to-home receiver model of radio development H.G. Aitken. 1985. The Continuous Wave: Technology and American Radio, 1900 - 1932. Princeton University Press, Princeton, New Jersey.. 17 DRAFT 10th December 2012 © John Howells & Ron Katznelson suits under the 1910 law. Radio technology under Wartime conditions became widespread without being licensed. Fourth, in the period between the end of the War and the return of radio assets by the US Navy, the essential cross-licensing deals were worked out that enabled legal operation of state-ofthe-art civil radio services at the earliest possible dates. The achievement of a cross-licensing deal was arguably complicated by the US Navy because it urged the end of bilateral licensing negotiations between GE and the Marconi Company and expressed the ‘national’ need to keep GE’s Alexanderson Alternator technology US-owned. The remaining important pioneer patent rights of the triode owned by AT&T and the receiving circuit patents owned by Westinghouse were acquired after RCA had already begun broadcasting; AT&T preferred to negotiate rather than litigate while RCA was slower than Westinghouse in understanding the pioneer status of the receiver circuit patents necessary for the business model of broadcasting to home receiver sets. In each case GE-RCA acquired the necessary patent licenses in exchange for membership of the RCA ownership and cross-licensing arrangement. This paper presented new facts from mostly primary sources that have contributed to a thorough reevaluation of the role of pioneer patents and especially the Fleming diode patent in US radio development. First, we used court decision records and data from neglected technological reviews to assess the scale of market for and nature of substitute technologies for the early years of vacuum tube technology. Second, we used the information in the ten year late disclaimer added to the diode patent to infer that it had not been evaluated for litigation for ten years after its issue. Third, we noted that the unremarked early expiry of the British diode patent enabled a comparison of radio development between the US and UK to be a control on assertions that through not being licensed, the US patent had retarded US radio development. Fourth, by citing to the decision itself, we showed that there was a widespread error of interpretation of the scope and significance of the 1916 decision in Marconi v. De Forest. Fifth, we noted that the unremarked July 1919 decision in Marconi v. De Forest did broaden the scope of the diode patent to prospectively control valuable properties of the triode. In conclusion, court decisions on the validity and scope of the diode patent altered the scope of its control of technology several times; the result was that although who controlled what changed, development continued through licensing arrangements in every scenario we examined. While the situation in radio patents resembles what has been called an anti-commons it was resolved through private interests negotiating an original cross-licensing arrangement, despite, rather than because of the intervention of the US Navy. We would argue that licensing negotiations were complex, but precisely because they did not hinder development or operation of state-of-the-art radio technology this case rather illustrates the determination of patentees to develop business from their patent assets from which they can draw an income. With the historical example of radio as a guide, we would propose that 18 DRAFT 10th December 2012 © John Howells & Ron Katznelson licensing arrangements that facilitate continuing development will be the likely reality in contemporary examples of the ‘anticommons’ in contemporary technologies such as biotechnology and software. References H.G. Aitken. 1985. The Continuous Wave: Technology and American Radio, 1900 - 1932. Princeton University Press, Princeton, New Jersey. G.L. Archer. 1938. History of Radio to 1926. American Historical Society, New York. W. Baker. 1970. A History of the Marconi Company. Methuen, London. M. Boldrin, Levine, D.K. 2008. Against Intellectual Property. J. Brady. 1922. The Relation of Patents to Radio Manufacturing. Journal of the Patent Office Society. 5 268-285. J. Brittain. 1992. Alexanderson. Johns Hopkins University Press, Baltimore, Maryland. T. Buckley. 2007. The Myth of the Anticommons. Biotechnology Industry Organization, Washington DC. G. Carneal. 1930. Conqueror of Space: The Life of Lee DeForest. Horace Liveright, New York. H.W. Dickinson, Jenkins, R. 1927. James Watt and the Steam Engine. Moorland Publishing, Oxford S.J. Douglas. 1987. Inventing American Broadcasting, 1899 - 1922. Johns Hopkins University Press, Baltimore and London. Electrical World. 1919. Radio Vacuum-Tube Litigation is Settled. Electrical World. 73(25). Federal Trade Commission. 1924. Report of the Federal Trade Commission on the Radio Industry In Response to House Resolution 548, 67th Congress, 4th Session. Government Printing Office, Washington. J. Fleming. 1919. The Thermionic Valve and its Developments in Radiotelegraphy and Telephony. Wireless Press, London. H.B. Hall, Harhoff, D. 2012. Recent Research on the Economics of Patents. National Bureau of Economic Research, Cambridge, MA 02138. M. Heller, Eisenberg, R. 1998. Can Patents Deter Innovation? The Anticommons in Biomedical Research. Science. 280(5364) 698-701. 19 DRAFT 10th December 2012 © John Howells & Ron Katznelson J. Hijiya. 1992. Lee De Forest and the Fatherhood of Radio. Lehigh University Press, London and Toronto. L.S. Howeth. 1963. History of Communications-Electronics in the United States Navy. US Government Printing Office, Washington. R.D. Katznelson, Howells, J. 2012. Inventing Around Edison's Incandescent Lamp Patent: Evidence of the Role of Patents in Stimulating Downstream Development. To appear. E.W. Kitch. 1977. The Nature and Function of the Patent System. Journal of Law and Economics. 20 265-290. E.W. Kitch. 2000. Elementary and Persistent Errors in the Economic Analysis of Intellectual Property. Vanderbilt Law Review. 53 1727-1741. J.S. Leibowitz. 2002. Inventing a Nonexclusive Patent System. Yale Law Review. 111(8) 22512287. W. Maclaurin. 1971. Invention and Innovation in the Radio Industry. Arno Press and the New York Times, New York. R. Mann. 2004. The Myth of the Software Patent Thicket. Bepress. R. Marriott. 1917. United States Radio Development. Proceedings of the Institute of Radio Engineers. 5(3) 179-195. R.P. Merges, Nelson, R.R. 1990. On the Complex Economics of Patent Scope. Columbia Law Review. 90(4) 839-916. A. Morse. 1925. Radio: Beam and Broadcast - Its Story and Patents. Ernest Benn Ltd, London. A. Morse. 1943. Fleming and De Forest. Instruments. 16 636. P. Moser. 2013. Patents And Innovation: Evidence from Economic History. Journal of Economic Perspectives. 27(1) 23-44. L.S. Reich. 1977. Research, Patents, and the Struggle to Control Radio: A Study of Big Business and the Uses of Industrial Research. Business History Review. 51(2) 208-235. T. Sabety. 2005. Nanotechnology Innovation and the Patent Thicket: Which IP Policies Promote Growth? Albany Law Journal of Science and Technology. 15 477-515. G. Selgin, Turner, J.L. 2006. James Watt as Intellectual monopolist: Comment on Boldrin and Levine. International Economic Review. 47(4) 1341 - 1348. 20 DRAFT 10th December 2012 © John Howells & Ron Katznelson G. Selgin, Turner, J.L. 2009. Watt again? Boldrin and Levine Still Exaggerate the Adverse Effect of Patents on the Progress of Steam Power. Review of Law and Economics. 5(3) 1101-1113. G. Selgin, Turner, J.L. 2010. Strong Steam, Weak Patents, or, the Myth of Watt's InnovationBlocking Monopoly, Exploded. C. Shapiro. 2001. Navigating the Patent Thicket: Cross Licenses, Patent Pools and Standard Setting. MIT Press for the National Bureau of Economic Research, Cambridge and London. G. Tyne. 1977. Saga of the Vacuum Tube. Howard Sams and Co. Inc., Indianapolis, Indiana. 21