^{note (1) page 40} Principally radio and television receiving sets.

^{note (1) page 40} This list is shown in full in Appendix I. It is the classi fication which was used by the Ministry of Aviation in collecting UK statistics.

The principal products which are included in the definition used here fall into seven main groups :

• (i) Electronic data processing equipment, including digital and analogue computers and their peripheral equipment, but excluding computers which form part of other systems, as shown below in groups (ii), (vi) and (vii).

• (ii) Ground, marine and airborne radar and electronic navigational aids and guidance systems, including the associated computer systems. This includes missile and satellite tracking and detection equipment, and sonar.

• (iii) Radio communication equipment, other than public broadcasting equipment, e.g. for ships, aircraft, police, ambulances.

• (iv) Public radio and television broadcasting equipment and the associated studio equipment.

• (v) Electronic and nucleonic measuring and testing equip ment, including such instruments as oscilloscopes, signal generators, spectrometers and radiation detectors.

• (vi) Industrial electronic control equipment, including machine tool control, process control, and the associated computing and data logging equipment.

• (vii) Other (specified) electronic capital goods, including electronic heating and welding equipment, medical equip ment, X-ray equipment, electronic teaching aids and simulators.

This list excludes telegraph and telephone equipment and conventional instruments for measurement and control. In both of these industries electronic firms are heavily involved and electronic applications are growing fast. Other firms are ‘becoming’ electronic even if they were not previously involved in electronics.

It may soon be useful to classify them with the electronics industry altogether and some authorities already do so. Because of the close assocation between the two industries, some figures for telephone and telegraph equipment are shown in the tables, as in the most recent statistics published by the Ministry of Aviation.

^{note (1) page 41} Including computers of all types and not only those used for commercial data processing.

^{note (2) page 41} Counting the associated computer equipment within the group.

^{note (1) page 42} This does not mean that technical progress is synonymous with the introduction of electronic capital goods, or with ‘automation’, nor that electronic techniques are the only ones which should be employed in’ automation’.

^{note (2) page 42} It is true that many firms which have installed electronic equipment have found that it takes some time before they get a satisfactory return on their investment; as with all major innovations, there is a learning process for individuals, firms and communities.

^{note (3) page 42} Some of these are discussed in Appendix I.

^{note (1) page 45} Excluding the Communist countries. The United States' share was about 82 per cent for the five principal manufacturing countries. The output of other countries (Italy, Belgium, Netherlands, Canada and Sweden) is probably equal to some 4-7 per cent of this total.

^{note (2) page 45} About 45 per cent of the French and 40 per cent of the British from 1961 to 1964. In West Germany the proportion is lower and the figures used here are believed to exclude military sales.

^{note (3) page 45} For example, IBM accounts for a major part of the computer exports of both France and West Germany. The division of labour between IBM factories and the interchange of parts and peripherals make it particularly difficult to interpret export statistics in this category.

^{note (4) page 45} See also Appendix IV, page 91, for a comprehensive list of computers installed in the United States, Britain and West Germany, classified by type of installation and by manufacturers.

^{note (1) page 46} Appendix IV, page 91.

^{note (1) page 47} Analogue computers are based on the continuous measurement of a physical quantity which is analogous to the numbers in the problem under consideration, while digital computers are based on the numerical counting of discrete data.

^{note (2) page 47} There is a considerable American financial particpation in this firm.

^{note (3) page 47} Comparisons of this type are often distorted by the inclusion of some categories of computers in one country but not in another. Although the American monthly magazine, Computers and Automation appears to be one of the best published sources for the number and type of United States computers, its international figures seem less accurate. For the purpose of this article the smaller keyboard ‘desk-calculators’, such as the NCR 390, the Burroughs E101 and E2100, and the Monrobot machines, have been omitted from the figures for each country, but the Univac 1004 has been included.

^{note (1) page 48} In March 1964 in a Survey by Metalworking Production, 17th June, 1964, it was estimated that 395 had been delivered and 152 were on order.

^{note (2) page 48} The National Physical Laboratory and National Engin eering Laboratory also made important contributions. [7]

^{note (3) page 48} International Telephone and Telegraph Corporation (ITT) employed 185 thousand people in 1964, of whom 128 thousand were in Europe. ITT subsidiaries are among the largest electronic and telephone equipment firms in most European countries, including Standard Telephone and Cables (STC) in Britain and Standard Elektrik Lorenz (SEL) in Germany.

^{note (1) page 49} STC is an ITT subsidiary.

^{note (2) page 49} Raytheon now own Cossor in Britain and Selenia in Italy. The gyro compass exports of the British Sperry Gyroscope Company greatly exceed those of the parent company.

^{note (3) page 49} Not all of this turnover is in electronic measuring equipment.

^{note (4) page 49} A high proportion of British ‘imports’ of radar equip ment and radio communication equipment consists of sets returned for repair and maintenance. If these are deducted from the total, then computers account for over half of British imports in 1964.

^{note (1) page 51} United States salaries are between two and three times as high as those in most European countries. The total of bought-in materials and components varies enormously with the degree of vertical integration, but for the British industry as a whole, it was estimated at 46 per cent of gross output in 1964.[17] Most American companies for which figures are available show a figure lower than this, but above one- third of gross output.[18]

^{note (2) page 51} The degree of vertical integration varies and this will of course affect the capital intensity of individual firms.

^{note (3) page 51} See table 5 of The United States Electronics Industry in International Trade, page 96 of this Review.

^{note (4) page 51} See The United States Electronics Industry in International Trade, page 92 of this Review. The United States' share of world exports of radio and television sets fell from about 32 per cent in 1937 to 12 per cent in 1958 and 3 per cent in 1964.[22]

^{note (5) page 51} Measured by the ratio of R and D expenditure to sales.

^{note (1) page 52} The Marconi American subsidiary was also the largest manufacturer of radio in America up to 1914.[24]

^{note (1) page 53} Owen Young's account states:—

‘When Admiral Bullard arrived in my office, he said that the President, whom he had just seen in Paris, was con cerned about the postwar international position of the United States and had concluded that three of the key areas on which international influence would be based were shipping, petroleum and radio. But in radio, the British were now dominant and the United States, with her technical proficiency, had an opportunity to achieve at least a position of equality.’

W. R. Maclaurin, Invention and Innovation in the Radio Industry, Macmillan Co., New York, 1949, page 101.

^{note (2) page 53} CSF, the leading French electronics firm, was also formed partly from a former Marconi subsidiary. The French industry made rapid advances in the First World War and the 1920s, but also had difficulties with Government.

^{note (3) page 53} Exports from the Netherlands were about as big as those of the United States in 1937.

^{note (1) page 54} This outline deals only with electronic systems and not with mechanical.

^{note (2) page 54} Until 1930 Zworykin worked at the Westinghouse Laboratories, but in close association with RCA. In 1930 the GE and Westinghouse radio and TV development work was transferred to RCA Laboratories.

^{note (3) page 54} Broadcasts using the less satisfactory Baird mechanical system had begun as early as 1929.

^{note (4) page 54} There were financial links between EMI and RCA before this and EMI had the advantage of knowledge of some of Zworykin's work through their connections with RCA. Sarnoff was a Director of EMI.

^{note (5) page 54} But Farnsworth spent a million dollars on television R and D from 1929 to 1938.

^{note (1) page 55} Before his tragic death in an air crash, Blumlein led the development work on the most sophisticated type of reconnaissance radar at EMI (H₂S).

^{note (1) page 56} For example, the Lorenz blind approach beam system used to guide bombers and its successor, X-Gerät, both need continuous wave systems rather than pulse generation. These were less accurate and more easily jammed. Pulse techniques were not used until 1944. For similar reasons the German ground chain, Freya, using Würzburg sets (53 cms) was virtually immobilised by British bombers in 1943. [39]

^{note (2) page 56} But the reservations in the footnote to the table should be taken into account in interpreting the patent statistics.

^{note (3) page 56} For the period 1952-61 the percentage of total patents taken out by United States applicants in London was 18.4 per cent. But for most electronic categories it was far higher, and consistently around 50 per cent among the 30 leading firms in calculating and computing machinery from 1931 through to 1962. In radar and navigational aids this proportion fell from about 45 per cent in 1947-54 to about 30 per cent in 1955-62.

^{note (1) page 59} Then known as the Computing-Tabulating-Recording Co. The name was changed to IBM in 1924.

^{note (2) page 59} One of his associates wrote: ‘It required a great deal of courage to authorise the tremendous expense for this develop ment and there were many of us who seriously doubted that the customers would stand for the increased rental necessary for the increased complication of the machines; but it is now evident that Mr. Watson had correctly estimated the final result.’ It took four years to develop the machine.[49]

^{note (3) page 59} ‘We had some new machines and ideas to give our salesmen … If we had to depend on the line we had five years ago, it would have been a different story.’ [50]

^{note (4) page 59} Gross income grew from $116 million in 1946 to $696 million in 1955, and employment from 22 thousand to 41 thousand. By 1964 gross income reached $3,239 million and employment was 149 thousand.[51] Growth of income was 22 per cent per annum to 1955 and over 18 per cent since.

^{note (1) page 60} ‘One of the most exciting chapters in IBM's post-war history has to do with large-scale electronic computers and data processing. Many very large engineering computational jobs and a fair number of accounting applications were being hampered by the slowness of the calculating machines available in the later 1940s. However, Drs. Eckert and Mauchly of the Moore School at the University of Pennsylvania had built a large electronic computer—the Eniac—for the Army to make ballistic curve calculations. Many of us in our industry, including me, had seen the machine, but none of us could foresee its capabilities. Even after Eckert and Mauchly left the Moore School and began privately to manufacture a civilian counterpart to the Eniac—the Univac, few saw the potential.

The Company was finally absorbed by Remington Rand in 1950, and soon had installed several machines in the US Government, including one in the Census Bureau which replaced a number of IBM machines.

During these really earth-shaking developments in the accounting machine industry, IBM slept soundly. We had put the first electronically-operated punched card calculator on the market in 1947. We clearly knew that electronic computing even in those days was so fast that the machine waited 9/10 of every card cycle for the mechanical portions of the machine to feed the next card. In spite of this, we didn't jump to the obvious conclusion that if we could feed data more rapidly, we could increase speeds by 900 per cent. Remington Rand and Univac drew this conclusion and were off to the races.

Finally we awoke and began to act. We took one of our most competent operating executives with a reputation for fearlessness and competence and put him in charge of all phases of the development of an IBM large-scale electronic computer. He and we were successful.

How did we come from behind? First, we had enough cash to carry the loads of engineering, research and production, which were heavy. Second, we had a sales force which enabled us to tailor our machine very closely to the market. Finally, and most important—we had good company morale. All concerned realised that this was a mutual challenge to us as an industry leader. We had to respond with all that we had to win, and we did.

By 1956, it became clear that to respond rapidly to challenge, we needed a new organisation concept. Prior to the mid- nineteen fifties the company was run essentially by one man— T. J. Watson, Sr. He had a terrific team around him, but he made the decisions. If we had organisation charts, there would have been a fascinating number of lines—perhaps thirty—running into T. J. Watson.

In the early 1950s the demands of an increasing economic pace and the Korean War were calling for more rapid action by IBM at all levels than our monolithic structure was able to respond to adequately. Increasing customer pressures— plus a few more missed boats of lesser consequence than the Univac situation—forced us to decide on a new and vastly decentralised organisation.

Here, we hope we responded a little more rapidly than we did in the case of Univac. The new organisation was 180 degrees opposed to the old in fundamental concept, but we made the move.

In late 1956, after months of planning, we called the top one hundred or so people in the business to a three-day meeting in Williamsburg, Virginia. We came away from that meeting decentralised.’

Thomas J. Watson, Junior, ‘Meeting the Challenge of Growth’. McKinsey Foundation Lecture, No. 2.

^{note (2) page 60} As with colour television, RCA had to lay out over $100 million before computers began to be profitable in 1964.[55]

^{note (1) page 61} There was additional expenditure on orders to firms.

^{note (2) page 61} Some observers believe that the weaknesses of some of the British firms were due to a tendency to regard ‘EDP’ custo mers as though they were ‘scientific’ customers.

^{note (1) page 62} See page 65.

^{note (1) page 63} This is also true of many of those products which have not been described for lack of space, such as the cathode ray oscilloscope, spectrometers, signal generators, the electron microscope, X-ray apparatus, sonar and various navigational aids.

^{note (2) page 63} Taking manufacturing firms only—that is, excluding those which operate exclusively as sales agents or importers.

^{note (3) page 63} The process known as ‘Anglicisation’, that is, the manu facture in Britain of American designs with some (or all) British components, can involve a considerable development effort. Several firms came to the conclusion that it would have been cheaper and quicker in some cases to design a fresh product from the start.

^{note (4) page 63} This does not necessarily mean that there must be vertical integration. As in the car industry, specialist component makers may often have some economic advantages.

^{note (5) page 63} For example, the pentode valve was developed at Philips and the hexode by Steimel at Telefunken.

^{note (1) page 65} Total R and D expenditure of six of the leading companies are estimated as follows :

These estimates cover all R and D, some of which is non-electronic, the proportion varying within each company. They also include both private venture and government- financed R and D.

^{note (2) page 65} Between 1952 and 1963 the Western Electric Co. (Bell) received over £9 million in total income for royalties from companies all over the world, excluding cross-license benefits, and of this the Company estimated that over £3 million was attributable to transistors. By far the greater part of this income came from companies outside the United States (£556,000 from licensees having their principal office in the UK). United States concerns are able to use Bell patents prior to the consent decree of 1956 royalty-free, but pay royalties (varying in scale with the individual agreement) on patents issued since the decree.

The Company estimated total R and D expenditure at Bell Labs. on transistors and transistorised equipment at £2.7 million up to 1953, at £28 million up to the end of 1960, and £57 million up to September 1964. Total expenditure on the negotiation and administration of license agreements was £6.3 million from 1952 to September 1964, of which £1 million was for transistors and transistorised equipment. Cross-license benefits in respect of agreements involving the basic transistor patent were estimated at £2.6 million. Only E4,000 of this was attributable to UK companies.

As a result of their technical lead, other American com ponent firms, in addition to Bell, enjoy a considerable royalty and know-how income from the licensing agreements which they have concluded with European firms. The planar technology patents of Fairchild are particularly important, and in addition to European firms, leading American firms, such as Texas Instruments and ITT, made license agreements with Fairchild to obtain this technology. There are, it is true, a few component developments in Europe which have been licensed to the USA: for example, the Lucas development work in industrial semi-conductors which began in 1954 has resulted in the successful development of high voltage devices for ignition systems, which have been licensed to Delco in the USA. Siemens licensed Westinghouse for their ultra-pure silicon process. But in total American companies have a substantial positive balance in ‘technical payments' in the component field; this reflects their lead in most areas of new component development since the war. Only in the older components, such as valves, is there a real two-way traffic.

^{note (1) page 66} ‘Due to its considerable interest in semi-conductors and particularly in transistors, the government has throughout the 1950s tried to stimulate the development of improved types. Around the middle of 1950 they were convinced transistors were needed for future military equipment so they accelerated the production refinement to provide developmental and production facilities for making certain types which were considered to be desirable for future military electronic equipment. Thus, whereas during most years in that decade government funding for research, development and industrial preparedness work ran at the rate of four to eight million dollars each year, in 1956 a major additional appropriation became available and was channeled into industrial pre paredness studies on transistors. Resulting in a $14 million additional expenditure, the contracts granted extended over the next two or three following years.

The contracts for a total of thirty different types of german ium and silicon transistors were placed with about one dozen of the major semi-conductor companies, and this helped some of these to gain a foothold in the industry. In many cases, this investment was matched by similar amounts of capital equipment or plant space supplied by the contracting companies. Requirements of the contract included the delivery of about three thousand transistors of each type to be made on production lines capable of producing that many per month. Thus a total potential capacity of over a million transistors a year was created. Since at that time transistors were manufactured at yields as low as 5 per cent to 15 per cent, it is not unrealistic to assume that a productive capacity eventually capable of tens of millions of transistors was created by these contracts, and this at a time when the total unit shipments of finished transistors by the industry ranged from 14 million units in 1956 to 28 million in 1957. Here, although the individual companies certainly paid for the plant space, almost all engineering design and development was paid for by the governmnent.

An additional impetus to the industry resulted also from the clear delineation of the specifications required which predicted certain types that were likely to be in major military usage a few years hence, and this stimulated many companies including a number of those which had not participated in this earlier windfall to develop types at least as good, preferably better, than the ones under the government contract in order to gain back this potential business. Possibly the present overcapacity within the industry was created during this period, which resulted in major price declines beginning in 1960.’

US Department of Commerce, Patterns and Problems of Technical Innovation in American Industry, A Government Research Report to the NSF, prepared by A. D. Little, 1963, page 163.

^{note (2) page 66} ‘In order to get around the complexity factor, and arrive at a common denominator which can be used in talking about such things as numbers of circuits, price and cost of circuits, and so forth, various ways of counting circuits have come into use. Perhaps the one most frequently used is the concept of the active element group’, or AEG, … Basically, it is defined as the active elements in a circuit together with their associated components of other types … By 1968 we expect some 80 million AEGs out of 280 million in industrial elec tronics equipment of all types to be in the form of integrated circuits. By 1973, nearly three-fourths of all industrial AEGs may be in integrated form … The consumer market presents quite a different picture. It has taken quite a long time even for the transistor to penetrate this market. Although transistorized TV sets are beginning to appear in some quantities in 1964, I expect it will be several more years before they begin to seriously reduce the consumption of vacuum tubes. The consumer goods field is one in which cost is by all odds the strongest determining factor in the choice of circuit techniques, with performance and reliability in second and third place and such things as size and power drain of relatively little importance … Considerable design work remains to be done to make integrated circuits attractive from a cost standpoint; therefore, we believe that, although there may be some gradually expanding use in specialised consumer applications over the next 5 years, it will be in the period beyond 1968 that consumer applications become fairly general. By 1973, perhaps half of all AEGs in consumer equipment may be integrated.’

Sprague Technical Paper, no. TP-64-9.

^{note (1) page 67} Plessey has been the fastest growing large firm in the British industry and now manufactures a wide range of capital goods as well as components. It is also one of the largest manufacturers and exporters of telephone equipment.

^{note (2) page 67} An interesting discussion of it is by Mr. Yorke-Saville, Managing Director of British Jeffrey-Diamond, and the classification used here is largely based on his treatment of development lead times for mining machinery. [70]

^{note (1) page 68} Several firms estimated that it cost about £1 thousand in advertising and other expenditures to recruit a good develop ment engineer. The recruitment of enough system analysts and programmers for the computer firms and computer users is even more difficult. But one firm mentioned that a cheap and quick way was an advertisement for maths teachers in the Times Educational Supplement.

^{note (2) page 68} In this context competition in the market defines the ‘product’. A process control computer is a different ‘product’ from an EDP computer, because it requires different development, design and marketing processes, and similarly the oscilloscope market is divided between several different’ products’, which are not directly competitive.

^{note (1) page 69} It is extremely difficult to get accurate figures because the available R and D data are for multi-product firms which have both a good deal of non-electronic output, and output of electronic consumer goods.

^{note (1) page 70} This assumes that success in this market depends on the capacity to supply a range of data processing equipment. Total annual expenditure on R and D in the radio and com munication industry in Norway is not more than £1 million per annum.[74] Norwegian firms and the responsible Government authorities have in fact sensibly decided to concentrate their limited resources on developing automated systems for the design and production of plates for shipyards (the Autokon system) including numerical control equipment for flame- cutting machines. [75]

^{note (2) page 70} The marketing threshold is far higher for commercial data processing computers than for scientific computers.

^{note (3) page 70} The contracts are expected to be worth about £100 million and the principal members of the consortia are :—

1. (i)

   (i) Hughes, Marconi, Telefunken, CFTH and Selenia;

2. (ii)

   (ii) ITT, Elliott-Automation, CSF, AEI, Litton;

3. (iii)

   (iii) Westinghouse, Plessey, IBM, and Siemens.

^{note (1) page 71} The aircraft industry has an even higher ratio of government-funded R and D than the electronics industry. Many firms in Britain, France and America have more than 95 per cent of their R and D under government contracts. But two of the most consistently successful firms, Rolls-Royce and Boeing, have a private venture R and D which is well above the average for the industry. Boeing also has a special division to make use of research discoveries which have potential commercial applications.

^{note (2) page 71} That is, hardware sales plus Government contract R and D.

^{note (1) page 72} ‘The extent to which firms may benefit from DOD R and D contracts is indicated by certain contracts awarded by the Signal Corps of the Department of the Army. The Signal Corps has informed the sub-committee that it is currently employing 27 commercial contractors on research programs for semi-conductor devices; 40 commercial firms on electronic parts and materials projects, and 13 commercial firms on quartz crystal units. With respect to the semi-conductor programs, the Signal Corps has stated that the advancement of the state of the art has opened the door to the creation of a new industry with tremendous growth potential. Many civilian applications have been made possible in a period of time estimated by the Signal Corps to be perhaps 75 per cent shorter than would have been possible without Government support.’

Patent Practices of the Department of Defense, US Government Printing Office, Washington, 1961, page 37.

^{note (1) page 73} This is not to under-rate the purely military achievements of these programmes, either in Britain or in the United States:

^{note (1) page 75} For example, the Ferranti electronic controls for machine tools and Elliott Automation ‘Arch’ systems of on-line process controls and, in rather a different field, Radyne industrial heating and welding equipment.

^{note (2) page 75} As in the scheme just announced for loans of up to £2 million from NRDC for Elliott Automation projects. (Times, 15 November 1965).

^{note (3) page 75} Those who argue against the development of such grids stress the organisational and social problems involved in programming very large computers. They also point to the problems of private security and the lack of adequate resources in the telephone equipment industry. On these grounds some people suggest that small computers may well be more important than large ones.

^{note (1) page 76} One field in which such a Computer Grid could very usefully operate would be the data processing (real-time and otherwise) required for social security and revenue purposes. Two large government departments (the Board of Inland Revenue and the Ministry of Pensions), each with a national network of offices, are separately considering various EDP applications (for example, maintenance of taxpayers' records and payment of sickness and unemployment benefit). Large economies would probably follow from joint use of a Computer Grid; the concentration of programming effort alone would be a worthwhile gain. In the absence of such a joint plan, the least that should be agreed would be the purchase of compatible equipment and a common approach to software requirements and usage. There would probably be similar advantages in national medical records.

^{note (2) page 76} It is not without interest that in 1964 IBM acquired an American company, Science Research Associates, which has been a pioneer in teaching aids.

^{note (1) page 77} This is in line with the findings of the Survey of the Stanford Research Institute on the research activities of American firms with subsidiaries in Europe. [83]