Europe embraced Herman Hollerith’s technology for processing general statistics as the technology became stabilized in the 1900s. This contrasted with the lukewarm reception of his first punched-card system in the 1890s. The positive reception was the result of improvements to the technology and the establishment of marketing and maintenance organizations in several countries in Europe.
Foreign marketing could be organized through either independent agents or subsidiary companies. The independents provided the cheaper of the two solutions, while the subsidiaries were easier to control. Recent studies of the internal relations in current multinationals found significant differences between the agendas in a multinational company’s headquarters and in its foreign subsidiaries. In particular, the studies identified distinctions based on the subsidiary’s national business network, its own technological development, and its relations to its national government.1
The Tabulating Machine Company made the inexpensive choice of establishing agencies in London (1902) and Berlin (1910). These agencies were intended as simple agents to distribute equipment produced in the United States. However, the diffusion of the technology proved to be more complex. People in different countries demanded machines with varying capabilities that required them to be adapted. Simultaneously, challengers also emerged in Europe, as the Powers Accounting Machine Company established agencies in Great Britain and Germany. The outcome of this was diverging developments in various European countries, shaped by differing national traditions in private business and public organizations. Particularly, the differences in patent legislation in the various countries played an important role.
The British reaction to punched cards had been negative in the 1890s. The English census office had decided not to use punched cards to process census returns and an attempt failed to establish a British agency in 1895. However, while Herman Hollerith had selected the continental locations for his demonstrations in 1889, journalist Robert Percival Porter selected Great Britain. Porter was an English-born, naturalized United States citizen who had headed the office of United States’ census in 1890.2 In 1894, he arranged the first introduction of Hollerith’s punched-card systems to the British statistical community through a lecture in the Royal Statistical Society.3
Porter moved back to his native England in 1901 and there established contact with Ralegh Phillpotts, a lawyer who was well connected both in politics and in the military. They agreed to establish a Hollerith agency, where Phillpotts would act as general manager and Porter would be chairman. The following year, they obtained an option from Hollerith to organize an agency for the British Empire that would import American-produced machines at cost plus an overhead of 10 percent. To do this, they needed to raise $98,000 (£20,000), of which one half would be paid to the American company for assignment of patents.
A historian observed that this amount was comparable to the total nominal value of the shares at the incorporation of the Tabulating Machine Company in Washington, D.C., in 1897. However, the core of the original arrangement for the British agency was that they would be allowed to take possession of American-produced machines. At any rate, over the next two years they only succeeded in raising one-tenth of the required total amount, and they therefore persuaded the Tabulating Machine Company in the United States to accept payment in installments. On this basis, the British agency was incorporated in 1904 as the Tabulator Limited with an authorized registered capital of $24,000 (£5,000).4
The new company experienced a prolonged start-up, partly due to initial problems with the machines’ ability to handle the shillings and pence of sterling currency. The slow build-up of business was a heavy drain on the finances of the company. More shares were issued, and the agreement with the Tabulating Machine Company was renegotiated for the second time, as the British company could not honor the agreed once-and-for-all payment for the Hollerith patents of £10,000. This caused the American company, still controlled by Hollerith, to introduce tighter control over its first agency.
Although Hollerith originally had agreed to sell his machines outside the United States, the Tabulating Machine Company now retained ownership of the machines and charged the British agency a royalty of 25 percent of their revenues from sorter and tabulator rentals. Consequently, due to the inability of the burgeoning British company to raise the considerable amount required to obtain the full patent assignment, the status of the British agency was reduced from being on equal standing with the American company to a situation comparable to one of the American company’s domestic customers. In 1907, the British agency was renamed the British Tabulating Machine Company (also known by its initials of BTM). Soon the agency’s business grew, and it authorized an increase in the share capital by a factor of 10 to finance machine leases. But the agency once again faced difficulties in raising capital; in fact, the problem of financing machine rentals would remain through the interwar years.5
Shortly after the launch of the British Tabulating Machine Company in 1907, its number of customers grew; the company had only two customers in 1908, rising to about thirty by 1914 and to thirty-four in 1916.6 This progress was reflected in the company finances, which only showed a profit after 1911. Three years later, the shareholders saw their first dividend. In spite of the profits that started in 1911, in 1913 a row erupted with the Tabulating Machine Company over delayed royalty payments. The American company threatened to terminate the supply of machines, and this drastic measure made the British Tabulating Machine Company pay their debt.7
The technical basis for the British business had been established in 1903 by hiring Everard Greene, a graduate in engineering from Cambridge University. As Greene’s job was to facilitate the transfer of the punched-card technology, he spent the first year studying the production of punched-card machines at the Tabulating Machine Company and their applications in the United States.8
Although there had not been much difference between the processing of census statistics in the United States and the statistics applications envisaged for the planned agency in 1895, it proved a challenge to handle sterling currency—an important capability as business statistics was becoming the prime application. Greene had studied this application in the United States, and the British company started to cultivate this market. The machines were designed to accommodate the dollar currency of one hundred cents in a dollar, but in Great Britain at that time there were twenty shillings in one pound, each worth twelve pennies. (In 1971, the currency was decimalized and one pound sterling became worth 100 new pence.)
The sterling currency problem only appears to have been realized during the first trials at the Woolwich Arsenal Ordnance Factory in London and at a plant of Vickers, Sons and Maxims in Sheffield, as both these customers needed to process a large number of calculations involving money. The first tabulator was adapted for sterling currency at the Tabulating Machine Company in Washington, D.C., in 1905, and it used four adding machine wheels to represent shillings and pence. Later, the adaptation of tabulators for sterling currency moved to the British Tabulating Machine Company that, in 1908, started to assemble the American-produced machines and to manufacture cards. The British Tabulating Machine Company devised a simpler system of representing shillings and pence that only used three adding machine wheels.9
Back in 1904, Greene had started to sell punched-card systems, and his first trials had been industrial and railway applications similar to those he had studied in the United States. Further, he approached the Registrar General who was in charge of the British census operations. The first two organizations to give the machines a trial were the Woolwich Arsenal and the Vickers, Sons and Maxims Sheffield plant. Both projects were to tabulate the distribution of costs and wages to each job at the works, and both trials were unsuccessful.
Nearly half a century later, Greene ascribed the rejection at Woolwich to Luddites among the staff: wires in the machines were disconnected repeatedly, and the staff whistled the funeral march when the rejected machines were carried through the office. However, the Woolwich Arsenal staff’s negative attitude was justified by two shortcomings of the machines. The problem of dealing with sterling currency was only realized during the Woolwich and Vickers trials, and a mechanized sorter only became available in Britain later. In fact, Vickers accepted the machines once they had been adapted for sterling and revenues started to flow to the British Tabulating Machine Company. During the years prior to 1916, at least nine similar installations were established at other industrial producers, and two more were acquired to process sales statistics.10
The British Tabulating Machine Company’s third trial was for operational statistics at the Lancashire and North Yorkshire Railway in 1905. The project was to calculate locomotive mileage and consumption of coal and oil, and the Railway approved the punched-card system. During the years prior to 1916, five similar installations were established at other railways. In addition, by 1916, the British Tabulating Machine Company had four insurance statistics customers, and their machines processed public statistics in two towns and produced operational statistics for two gas and electricity utilities.11
Greene’s persistent approaches to the census authorities eventually paid off as, in 1908, he was asked to provide price estimates and trial machines. The Registrar General conceded that the advantage of punched cards over manual card systems was generally recognized, but he considered the price excessive and was worried about relying on a foreign company.12 Census processing only required counting tabulators, but the Tabulating Machine Company in Washington, D.C., by then only produced the more costly adding tabulators as they had terminated their business with the Census Bureau in the United States. However, the British Tabulating Machine Company, so eager to land the census contract, decided to build their own counters to replace the adding units on an American tabulator.
The British Tabulating Machine Company’s engineers built a new counter, but it did not perform satisfactorily during trials at the Census Office and the prototype was scrapped. However, the trials concluded that the most appropriate design would be a tabulator equipped with thirty-six counters, in contrast to the five adding units on a standard tabulator from the American company. In spite of the short time available until the census, an improved counter was built and eight census-tabulators processed the census in England and Wales in 1911.13 Tabulators were also used to process the census in Scotland in 1911. Further, in 1913, the British Tabulating Machine Company negotiated the sale of census-tabulators to Deutsche Hollerith-Maschinen Gesellschaft mit beschränker Haftung (Dehomag) in Germany. However, this contact was discontinued at the outbreak of the First World War.14
A challenger emerged when James Powers went to Europe in late 1913 and demonstrated his prototype punched-card machines in Berlin, notably before he started to produce machines in the United States. While Hollerith—twenty years before—had used census applications as the basis for his first attempt to sell punched-card machines abroad, Powers went directly for business applications. The American Powers Accounting Machine Company paid an individual from each of three British organizations to spend a few days working with Powers’ machines in Berlin, and two of these individuals were impressed by the printing capability of the Powers tabulator. Therefore, two of the three organizations, namely His Majesty’s Stationery Office and the Prudential Assurance Company, became Powers’ first customers in Britain.15
A British Powers agency, owned by the American parent company, was established in 1915 to market American-produced punched-card machines. It was named the Accounting and Tabulating Machine Company of Great Britain Limited. The printing tabulator provided a good sales pitch and the new company’s business grew fast. In late 1916, after only one year’s operation the company had eleven customers, compared with the thirty-five customers of the British Tabulating Machine Company.16
The British Powers agency seems not to have encountered problems with the Tabulating Machine Company’s patent in Britain—in contrast to the situation in the United States and Germany. Hollerith’s patent on the automatic group control facility was an important patent. Filed in the United States in 1914, it caused problems for the Powers companies in the United States and Germany. Hollerith’s equivalent patent in Britain was granted in 1918, one year after Powers’ similar patent had been granted. These patents seem not to have generated any conflict in Britain until Hollerith’s automatic group control patent was granted in the United States in 1931.17
Following this, IBM in New York tried to persuade the British Tabulating Machine Company to enforce Hollerith’s British automatic group control patent on the British Powers company. However, this attempt at enforcement was discontinued in 1933, which a historian explained as being the result of a gentleman’s agreement between the two British companies.18 Gentlemanly behavior might well have been a reason for the absence of patent litigations on the punched-card trade in Britain, but the patent laws provide a more immediate explanation.
Herman Hollerith’s automatic group control patent had been filed in the United States in 1914, but it was only granted in 1931 when it gave rise to IBM’s enforcement attempt in Britain. As patents in the United States were valid for seventeen years from the day they were granted, this patent lasted until 1948—that is, thirty-four years from the date the application was filed.19 In contrast, the equivalent British patent was filed in 1917 and granted in 1918.20 The British patent expired in 1931, as a British patent was valid for fourteen years from the date when the application was filed.21
British patent law resembled the patent laws in Austria, Belgium, and France with respect to protecting British production against imports. Production in Britain of the patented device was required within four years of filing the patent application.22 This requirement explains why the British Tabulating Machine Company never seems to have tried to enforce Hollerith’s sorter patent that had been filed in the United States back in 1901.23
It was first in 1921 that the British Tabulating Machine Company started to assemble machine components imported from the parent company in the United States, and only in 1924 did they commence assembly of the Tabulating Machine Company’s first printing tabulator with automatic group control. This suggests that neither the sorter nor the automatic group control patent could have been enforced in Great Britain. Finally, the British patent laws facilitated enforced licensing of patents, which was encouraged as a private agreement between patent holder and licensee.24 These clauses explain why the British punched-card trade experienced no patent litigation during the 1930s, in contrast to the situations in France, Germany, and the United States.
At the outbreak of the First World War, the British armed forces totaled 733,514; this was a relatively small number compared with the German and French armies. However, the British armed forces were rapidly expanded to 4,231,670 people in December 1916 and to 5,757,457 people in December 1917.25 But establishing an army of more than five million people for the large-scale warfare of the First World War posed extensive demands on production and transportation. The rapidly expanding mobilization and warfare put great pressure on the British offices that were to process more transactions with fewer staff. British historian Martin Campbell-Kelly has investigated office mechanization at two big private and one big public organization, and his study reveals a rather limited level of mechanization in British offices.26
His Majesty’s Stationery Office was responsible for printing and sale of government publications and, for several years, the Stationery Office had shown a keen interest in mechanization. Their records confirm the modest level of office mechanization in government in 1914.27
When war broke out, the Stationery Office had just concluded trials of the Powers punched-card machines, which resulted from an invitation to inspect his machines the previous year in Berlin. The Stationery Office was impressed with the possibilities offered by the printing Powers machines for their accounting and statistical tasks, in contrast to the nonprinting machines from the British Tabulating Machine Company. The war meant that many of the Stationery Office’s experienced clerks were mobilized and, simultaneously, the office was confronted with an enormous increase in their workload because of processing supplies to the military. The Stationery Office discarded their system of accounting by hand and introduced Powers punched-card machines; the printed lists and totals enabled the office to retain their well-established system of lists and receipts for their audit.28
During the war, punched cards became an important tool in producing the statistics needed to monitor the war effort, resulting in rising revenues for the two British punched-card agencies. During the early part of the war, armaments policy was organized by free enterprise, meaning that the government contracted with companies for arms and munitions, and direct control was rare. However, this system did not supply the vast munitions needed for the scale of warfare. In 1915, a separate Ministry of Munitions was established which, gradually, incorporated the country’s industrial capacity into the armament industry.29 A statistics office in the Ministry of Munitions monitored this development and, to do this, they acquired a punched-card installation from the British Tabulating Machine Company.30
Punched-card installations were also acquired, for monitoring purposes, by the National Service Department, Ministry of Labour, the Admiralty (the ministry responsible for the Navy), the War Office (responsible for the Army) and, after the war, this effort was followed by several institutions dealing with demobilization and pensions.31
The Quick Success of Punched Cards in Germany
Herman Hollerith’s agency in Germany was established in 1910 as part of a new strategy to establish agencies in Europe. Whereas the British agency had arisen unsolicited, now Hollerith contracted with American engineer Robert Neil Williams to form companies on the European continent to sell Hollerith’s machines and manage his patents. Williams had engineering offices in Berlin and Paris, and he embarked on his assignment in Germany.
Williams approached Carl Duisberg to obtain money to establish a company. Duisberg was managing director of Farbenfabriken vorm. Friedrich Bayer and Co. (Bayer), which was the ninth biggest company in Germany.32 Duisberg was contacted because he already had visited Hollerith’s company in Washington, D.C., and had ordered a punched-card machine to process detailed statistics on Bayer’s turnover figures. However, Duisberg refused to invest in a German agency, as he thought the German market for punched cards too small. In Duisberg’s opinion, punched cards were only suitable for very big industries and, moreover, only a few of these were interested in systematic management.33 However, Williams got in contact with Willy Heidinger, who was the director of a small company representing the American Elliott-Fisher mechanical office machines in Germany.
Unlike Duisberg, Heidinger saw market potential in establishing an agency in Germany of the Tabulating Machine Company. He raised the necessary funds, $71,000 (marks 300,000), and the German agency was established in 1910 as the Deutsche Hollerith Maschinen Gesellschaft mit beschränkter Haftung (the German Hollerith Machine Limited Liability Company, abbreviated to Dehomag); 91 percent of its capital was held by Heidinger and 9 percent was owned by Williams, who shared the position of CEO with Heidinger. Dehomag gained the right to market machines produced by the Tabulating Machine Company on the same conditions as in the contract with the British Tabulating Machine Company in 1908. Dehomag would pay royalties of 25 percent of the revenues from sorter and tabulator rentals and the Tabulating Machine Company would retain ownership of the machines.
Before the First World War, the royalties consumed 12 percent of Dehomag’s total revenue, 48 percent of which came from sorter and tabulator rentals. The agreement of 1910 granted Dehomag the exclusive rights to sell and lease the American company’s machines and cards in Germany, Switzerland, and Scandinavia. In addition, Dehomag could sell German-produced punched cards, but they were not allowed to manufacture the Hollerith machines.34 To increase the Tabulating Machine Company’s sales in Europe, in 1914 the American company expanded Dehomag’s territory to include Austria-Hungary and the Balkan states.35
In addition to funds, Heidinger brought with him his business network from selling office machines. From this Dehomag soon built up a customer base, both statistical offices of the German Reich, the states (Länder), and industry, where punched-card processing of operational statistics became a big success. During the two first years, Dehomag obtained contracts for equipment to process the census in 1910 in four of the German states, which together comprised 16 percent of the population.36 Further, Dehomag opened a new market segment of municipal statistics processing when the town governments in Berlin and Cologne placed orders for punched cards for their statistics processing.37
Big industries constituted another group of early customers. Williams’ original approach to Duisberg of Bayer shows that Dehomag, from the outset, targeted this market. True, Duisberg refused to invest but, in the following year, his company acquired a punched-card installation for sales statistics in addition to their original application. Punched cards enabled more differentiated statistics than did other methods. Duisberg had reorganized Bayer in the period from 1891 to 1907, and a key element had been to introduce various kinds of statistics to monitor the activities in this large, diversified company. Originally, these business statistics had been compiled in a newly established statistics department using less sophisticated technology than punched cards.38 Punched cards were subsequently introduced to enhance the production of extensive operational statistics. Similarly, punched cards were used to produce operational statistics in other big German companies, including heavy industries.39
Census statistics had been the initial field for punched-card application in the United States, then came railway and insurance statistics. In 1912, Dehomag launched a sales promotion periodical, the Hollerith Mitteilungen (Hollerith Bulletin). The first issue argued for the introduction of punched cards for railway statistics in Germany,40 and the second issue argued for insurance statistics.41 However, railway applications proved to be no immediate success. The state railways in Württemberg and in Prussia considered using punched cards but found that the machines were not suited to handling their freight accounting and statistics. The reason for this was most probably the lack of a printing tabulator in Hollerith’s system. The same applied to trials by several other customers: The Hapag shipping company in Hamburg tried punched cards for bookkeeping tasks and the Berlin Post Office tested it for the calculation of interest, although, in the end both decided not to adopt punched cards for these tasks.42
Dehomag also tried, without success, to attract the Grossbanken (large banks) by making a proposal to audit their vast securities portfolio by use of punched cards.43 In the banking field, too, the absence of a printing capability seems to have been the reason for the lack of interest as banks became punched-card customers in the 1920s when Dehomag started to supply printing tabulators.44
Back in 1913, James Powers demonstrated his tabulating system in Berlin, an event attended by Heidinger and Williams of Dehomag. Powers’ printing tabulator and his horizontal sorter offered major advantages over the equipment from the Tabulating Machine Company. The American Powers company proposed the founding of a common German company to market machines from both producers in the United States. This would cause no technical problems, as both lines of machines were based on the same 45-column punched-card and, in fact, several installations in Europe in the 1920s used a combination of machines from both producers.45 The American Powers company offered to invest in Dehomag to acquire half of the shared capital. Heidinger and Williams could not agree on this proposal. Heidinger wanted to remain exclusively with the Tabulating Machine Company marketing only their machines, while Williams wished to merge the two agencies to market both lines of equipment. Unable to resolve their disagreement, they went their separate ways. Williams established a German Powers agency, Deutschen Gesellschaft für Addier- und Sortiermaschinen mit beschränkter Haftung (German Company for Adding and Sorting Machines with limited liability) in Berlin and became Powers’ first manager for Europe. Williams and Thomas J. Felder provided the required capital of $24,000 (marks 100,000) to establish the new agency. Like Williams, Felder was a citizen of the United States living in Paris.46
Dehomag considered Powers’ printing tabulator a major threat to their business—and for good reason.47 For its part, Powers’ weak point was the dependence on basic Hollerith patents with valid equivalents in Germany. In the United States, the Powers company would enter a license agreement with the Hollerith company in 1914 that required them to pay about 20 percent of gross revenues to the competing company. For Powers, a common Hollerith-Powers company in Germany was a way to avoid having to pay royalties in Germany. Heidinger responded to the creation of the German Powers company by filing a patent infringement suit in 1914. This prevented the Powers company from doing business until the German High Court, in 1916, ordered a compulsory license of 5 percent on machine sales and rentals. Two years of waiting ensured Powers a considerably better arrangement in Germany than in the United States.48 However, only few months would pass before the United States entered the First World War in April 1917, providing Powers with a narrow window of opportunity.
Before the High Court verdict was issued, the German Powers agency had established a test punched-card installation at a major pipe producer, Mannesmannröhrenwerke, in Düsseldorf. Mannesmannröhrenwerke used punched cards in their central bookkeeping department to monitor their current account bookkeeping at the end of every month. For this task, a printing tabulator was essential. The pending patent case prevented the arrangement from becoming commercial, but the German Powers company supplied the installation and cards free of charge, seizing the opportunity to establish a showcase at a well-known company.49 Lack of income as a result of the patent litigation drew heavily on the company’s basic capital, and American financier John Isaac Waterbury bought all shares of the agency in 1914, just as he had acquired the American Powers company the year before.
After the German High Court verdict in 1916, the Powers agency started to receive revenues from its machines at the Mannesmannröhrenwerke, but the United States’ entry into the war in 1917 prevented the agency from getting additional machines and caused it to become subject to German custodianship.50 (Williams was a citizen of the United States, but he had left Germany before the United States entered the war.) In the same year, all Powers’ equipment in Germany was requisitioned to produce war-related statistics, which also provided revenues. However, the cost of running the agency consumed both revenues and capital, and the agency vanished.51 It was not until in 1923 that a new German agency was established and Powers’ business really began in Germany.
At the start of the First World War, the Allies imposed a blockade on the Central Powers, which Germany came to manage for four years. In order to manage the economy under the blockade, a comprehensive system was established for sharing out provisions and raw materials, organized within the Ministry for War and through a large number of special war companies, like the company for the supply of bread grain (Reichsgetreidestelle, Geschäftsabteilung Gesellschaft mbH) and the company supplying sauerkraut (Kriegsgesellschaft für Sauerkraut mbH).52 However, the various companies managed their limited fields without very much overall coordination, in marked contrast to the United States where the economy was controlled through fewer organizations.
At the national company for the supply of bread grain, a small Dehomag punched-card installation produced a variety of statistics on the sale and consumption of bread grain and flour which they called control bookkeeping (Kontrollbuchhaltung). It was based on copies of all the receipts when bread grain and flour was traded and enabled the monitoring of the supply of bread grain, the production of flour, and the use of flour in bakeries and in retail outlets.53 Like in many big German companies, this process was organized through extensive operational statistics and bookkeeping; since manpower was in short supply due to the general mobilization, the process relied heavily on machines. Both Dehomag and Powers equipment was used for this purpose.
Expansion of business was difficult during the First World War, as all American supplies ceased in 1917 following the United States’ entry into the war. But by then Dehomag had a substantial number of punched-card machines in Germany, their revenues rose, and the company increased dividend payments from 4 percent in 1915 to 8 percent in 1916 and to 10 percent in 1917.54 The American entry into the war caused Dehomag to become subject to custodianship, as some of the shares were held by an American citizen, but IBM later acknowledged that the custodianship did not curtail Dehomag’s business.55
Much effort was put into keeping all the machines running, and, in 1918, Dehomag established a small factory in the town of Villingen in the Black Forest to rebuild old machines and produce spare parts. It was a major limitation in the contract with the Tabulating Machine Company that Dehomag had no authorization to produce American machines. Dehomag tried, in vain, to obtain a German government order to break this clause.56 A government order would have relieved the company of the responsibility of breaking their contract with the parent company. Heidinger made the best of the situation during the war to improve his financial position and his standing in relation to the American parent company. With this end, Dehomag established its own technological basis through building its own punched-card machines. This attempt was based on construction work by engineer Heinrich Tolle, hired in 1916, who had experience with calculating machines and punched-card sorter development.57 Tolle’s development work aimed at building independent punched-card machines, although this goal was not reached by the end of the war.58 In the years just after the war, the company finalized the design of two complete punched-card machines: a sorter and a punch. The greatest achievement was a number printing unit that could upgrade the nonprinting tabulators already available from the Tabulating Machine Company as these enabled the company to win customers requiring printing capability.59 However, these inventions never came into production, and Dehomag started to receive American-produced number printing tabulators in 1923.60
Germany proper was untouched by acts of war as hostilities ceased in November 1918, but industry was worn down by four years of war production. Further, five years of crisis followed caused by harsh peace terms with huge reparations and the new German government’s lack of ability to control the economy. The outcome was runaway inflation and soaring exchange rates, which created a catastrophic business climate.
Under these circumstances, Dehomag’s success during the war turned into a liability, as a large part of their revenues had come from tabulator and sorter leasing. As these machines remained the property of the Tabulating Machine Company, Dehomag was required to pay royalties that had to be paid in dollars. During the war, Dehomag was unable to pay their royalties due to currency restrictions, which was serious as the amount in dollars grew in step with the falling exchange rate of the German mark, and additional royalties accrued after the war.
Dehomag’s position became critical in the years immediately after the war. Willy Heidinger tried to postpone payment until the mark regained its strength to save his Dehomag stocks. But the mark continued to fall and, in 1922, the debt rendered Dehomag insolvent.61 At the same time, the Tabulating Machine Company was pressuring to receive their money and, in 1922, Thomas J. Watson arrived in Germany to settle the matter.
The Tabulating Machine Company planned to take advantage of the opportunity to replace the foreign-owned agency with its own subsidiary. However, Dehomag’s success was a strong argument against breaking all ties and starting from scratch in Germany, as IBM later did in Norway in 1935 where a subsidiary succeeded an unsuccessful agency.62 The 1922 settlement made Dehomag a subsidiary of the Tabulating Machine Company, which acquired 90 percent of the shares, while Heidinger kept the remainder. This ensured that he stayed in the company where he kept de facto control as long as his success lasted.63
Assigning operational autonomy to the nationals leading the subsidiaries emerged as a key element in IBM strategy and was found in several European subsidiaries and agencies, in sharp contrast to the limited delegation in IBM in the United States. At the same time, however, Watson closely controlled the Dehomag board.64 This strategy ensured Heidinger remained with Dehomag with his business network and abilities to produce profits which, in turn, yielded revenues to the American company. But Heidinger’s attempt during the First World War to use the German government against the American company indicated that his loyalty could cause problems if conflicts again should arise between Germany and the United States. Probably Watson never learned about Heidinger’s doubledealing, but Heidinger resented Watson’s capture of Dehomag. The elimination of Austria and Hungary (which were separated after the war), as well as Switzerland, Scandinavia, and the Balkan states from the company’s business territory represented major reduction in Dehomag’s scope of business. In accordance with the American company’s new business strategy, Dehomag was reduced to serving the German market.65
The Late Start of Punched Cards in France
Except for Austria, France was the first of the major European countries to introduce punched cards for processing census statistics in 1896, and the last of these countries to introduce them for general statistics in 1921. France shared Austria’s somewhat negative assessment of Herman Hollerith’s first punched-card system created to process census returns in the 1890s. However, while the Austrian census office refrained from applying punched cards for processing the next census, their French counterpart commissioned the building in France of an alternative device to facilitate processing their census returns.
French engineer Lucien March was in charge of the first punched-card application in France that was used to process employment statistics in the census in 1896. The punched-card equipment leased from Hollerith was able to carry out this task, but March was not sufficiently satisfied with it to use the system again. He found the initial transcription of data onto punched cards unnecessary, the system too expensive, and maintenance costs substantial. His assessment of the technical aspects of Hollerith’s simple electromechanical device was based on his training as a mechanical engineer. Subsequently, March built the classi-compteur to facilitate processing population census returns. The classi-compteur applied a different principle to Hollerith’s machine and was exclusively mechanical. This equipment was used to process the French census in 1901.
Lucien March was trained as an engineer at the prestigious École Polytechnique (Technical University) in Paris that gave its students access to senior positions in the civil service.66 The curriculum at École Polytechnique placed great emphasis on theoretical subjects, such as mathematics and physics.67 March followed the French scientific engineering tradition by contributing to a large number of publications. He also contributed to the development of the science of statistics, national statistical education, and international collaborations in statistics.68
March’s classi-compteur had a full keyboard of sixty keys, each connected to a counter that consisted of four ten-digit printing wheels placed in the rows on the lid behind the keyboard, enabling the counting of up to 9,999 entries in each of its sixty positions.69 When the classi-compteur was used to compile a table, each of its keys corresponded to one entry in the table. Processing was accomplished by the operator keying in the relevant information from a form. During this process the key or keys pressed stayed locked in a lower position, and the operator could correct an error by first unlocking the key or keys and then entering the information correctly. When all the information from the form was entered, the operator pulled a handle that made the counters connected to the pressed keys advance one unit, and the operator proceeded to the next form.
When the operator had completed entering the information for a table, she placed a sheet of carbon paper above the counters and tilted the movable frame that held a series of rollers between which paper ran from a large roll. This left an impression of the printing wheels on the white paper. The rollers moved the paper along to a new zone of blank paper, and the printed area was torn off. The subsequent aggregation was accomplished manually or by use of an adding machine. March’s classi-compteur was a mechanical engineer’s response to Hollerith’s simple, electromechanical design, possibly inspired by an Italian design from 1881 that was never applied in actual census work.70
Women working on classi-compteurs in the French national statistics office, c. 1930. (Reference 9183-1, Agence photographique Roger-Violet, Paris, France)
It was simpler to organize the compilation of statistics by use of a classi-compteur than by using punched cards. On a classi-compteur, one person could carry out the whole task of processing the data, one table at a time, using one technical device—but the device had four limitations. First, producing the tables required onerous physical handling of the original forms, which damaged the forms themselves. Second, the sixty keys severely limited the size of a table that could be compiled in one operation. Bigger tables were broken down into part-tables that were later merged manually, for example, the table of age, sex, and marital status for the French census of 1911, which held 1,150 entries for each county (département) as opposed to 166 entries in the similar manually processed tables from the census in 1891.71 Third, it was not possible to verify the compilation of a table except by reprocessing all data. Fourth, operating a classi-compteur was hard work, as the activation of the counters, the printing operation, and the rolling of the paper was done manually. Later the physical strain was relieved by the introduction of electrically powered handling of these operations, a solution resembling the U.S. Census Bureau’s improvement, for the census in 1910, of the reading mechanism on the Hollerith tabulators from 1890.72
As with punched cards, the classi-compteur’s main advantage was to facilitate dividing the work of statistical processing among female machine operators with modest training, while the mainly male statisticians edited and interpreted the results for various publications. The French national statistics bureau (Statistique générale de la France) bought classi-compteurs to process the census in 1901.73 In 1900 the first model of the classicompteur was completed, 157 machines had been build by 1912, and it was still being produced in 1930.
The classi-compteur seems to have been marketed exclusively for processing population censuses. It was applied at the Belgian national statistics office in 1912, at the Dutch national statistics office in the 1920s, and the producer tried, in vain, to sell the machine to the English census office in 1912.74 At the time of its invention in the late 1890s, the classicompteur was comparable as a census processing tool with Hollerith’s punched-card system from 1890. Punched cards only became a superior tool to compile statistics when the Tabulating Machine Company introduced the subsequent improvements: sorting cards by machine, automatic group control, and number printing.
In France, classi-compteurs were used to process the censuses between 1901 and the 1930s. During this period the French national statistics bureau was in a weak position compared with other national statistics offices. While other national statistics offices grew due to the importance ascribed to their role by politicians and civil servants, the French national statistics bureau enjoyed limited political support. The bureau used twenty-five classi-compteurs but did not receive any appropriation for a punched-card installation. In 1921, they were presented with a British or United States Powers sorter and several punches that were used to process statistics relating to the movement of the population.75
Any lack of progress by the French national statistics bureau was not caused by passivity in the statistics community. In 1920, a professional committee proposed improvements at the bureau, which included the acquisition of a punched-card installation. This wish was repeated several times during the following twenty years. Further, three statisticians connected to the French national statistics bureau established the Institute of Statistics at the Université de Paris in 1922 to improve training in statistics in France. The reasons for the lack of modernization are, thus, to be found either in the management of the French national statistics bureau or in its supplier of assignments, the French state. To some extent, the lack of development of statistics production at the French national statistics bureau between 1900 and the 1920s was a manifestation of the weak French state.
France during the seventeenth and eighteenth centuries had experienced strong state intervention in the economy, but the intervention of the state was rather limited during most of the French Third Republic (1870–1940). At that time, the French economy was weak, the country had perpetual trade deficits, and French industry lost market shares abroad.76
The First World War exposed the weaknesses of the French government. Like other belligerent nations, France failed both to anticipate a protracted war and to make adequate preparations for munitions production. During the hostilities, neither the civilian authorities nor the military high command played a prominent role in managing munitions production. In 1916, the government entrusted the federation of employers in the iron and steel industry (Comité des Forges) with the control of the import of pig iron and steel, and only in July 1918 did the government establish centralized control over pig iron production in France.
The war also put pressure on the supply of food. France had been a net exporter of food before the war but after it started French agricultural production declined drastically, due in part to mobilization and, further, to Germany’s occupation of northern and eastern parts of the country. As a result, France became a major importer of food, leading inevitably to price increases. However, the government only introduced price control on grain in October 1915, a measure that was subsequently extended to other basic foods. State monopolies that controlled the entire supply chain from producer to consumer were established for sugar (1916) and grain (1917). But a ration card for bread was not introduced until the summer of 1918.77
Punched-card technology was not used in France for monitoring munitions production or food supply during the First World War: none of the introduced regulations demanded extensive statistics. For example, the import and domestic production of iron and steel could be monitored more easily in other ways, and ration cards were a simple means to distribute the supply of a commodity to the entire population.
It was first in the 1920s that the French state started to become more interventionist to improve the country’s economic performance. From 1923, the government started to extend the telephone network to rural areas and to bring electricity to the countryside as well. From the mid 1920s, investments followed in social housing, transport, and the education system. In addition, the government encouraged mergers between French companies from 1928 to 1932.78
It would be tempting to assume that the weak governments in the Third Republic were the reason for the slow progress at the national statistics bureau. This assumption is substantiated by the absence of any major statistical study of French industrial development between 1860 and 1931.79 However, the management at the national statistics bureau did not take advantage of the situation when the government intervention increased. The French censuses in 1931 and 1936 could have served as an opportunity to extend staff and to improve the equipment used to process the returns. After all, by then the classi-compteurs had been in use for thirty years. Moreover, one government punched-card installation was established in 1926 and another in 1927. Responsibility for the slow progress at the French national statistics bureau should primarily be attributed to the management of the bureau.
In contrast to Germany, Great Britain, and the United States, France had neither significant punched-card applications at the national statistics bureau nor punched-card processing of business statistics during the first two decades of the twentieth century. The first commercial punched-card application in France was established in 1921. The late introduction of this technology can be explained in two ways. First, by the assumption that punched-card processing at the national statistics institutions in other countries served as a basis for the diffusion of the use of punched cards beyond that of population statistics.
The second explanation is based on the combined observation of the late introduction of punched cards for processing business statistics and of the late establishment of large integrated companies in France and French firms’ hesitancy to diversify into new products or processes. The former supply explanation accounts for the late establishment of a subsidiary of the Tabulating Machine Company in France. However, the late start of punched cards can also be explained by the late emergence of demand for equipment to compile business statistics, as the diffusion of punched cards in France coincided with the breakthrough of industrial rationalization. Further, the demand explanation reaches beyond the emergence of punched cards to give reasons for the subsequent use of punched cards for business statistics and bookkeeping in France.
Since 1906, a few companies, like tire producer Michelin and car producer Renault, had introduced time and motion studies of their working routines and processes as a step toward improving productivity, along the lines suggested by the American engineer Frederick Windslow Taylor. But it was only after 1920 that many other companies joined them. Similarly, Michelin and Renault were among the pioneers introducing punched cards to process their operational statistics in the early 1920s. In the 1920s, the rationalization in France was not confined to the factory floor. Also, offices were rationalized, which included the adoption of new accounting methods.80
The Tabulating Machine Company established a French subsidiary in 1920, Société internationale de machines commerciales (SIMC, translated as the International Business Machines Company). It is true that the Time Recording Company, another of the companies in the Computing Tabulating Recording Company, had established its own subsidiary in 1914, but it did not do any punched-card business. SIMC marketed punched-card equipment, while the other subsidiary sold clocks and time recorders. The two companies merged in 1935 to form a new company named Compagnie électro-comptable (CEC, or Electric Bookkeeping Machine Company).81
The Tabulating Machine Company’s subsidiary in France won its first punched-card customer in 1921 at the French subsidiary of SKF (Svenska Kullager Fabrik, the Swedish ball-bearing producer), which leased the punched-card equipment to generate their statistics. Over the next few years, the number of customers grew to twelve in 1923, to twenty in 1925, and to thirty in 1927 with a total of fifty-six tabulators and fifty-one sorters, a growth also reflected in the French IBM’s companies’ aggregate net earnings.
Several large French companies followed the lead of the French subsidiary of SKF during the 1920s. First, there was a group of industrial producers: Renault (1922), Thompson-Houston (1923), Compagnie Continentale des Compteurs (1923), Roger Gallet (1924), Citroën (1925), and Société Kodak-Pathe (1925). Second, SIMC attracted a railway company as a customer in 1923 and two additional companies in 1925.82 Public sector diffusion of punched cards started later in France than in the private business sector. Only in 1926 did the Caisse des Dépôts et Consignations83 acquire the first punched-card installation in the public sector, and four more public institutions introduced the use of punched cards over the next six years, including three ministries.84 All seem to have used their punched-card installations to produce various statistics and sorted lists, for example, as a basis for monitoring and controlling payments.
As early as 1922 progress was so promising that the subsidiaries of the Computing Tabulating Recording Company (the outcome of a merger in 1911 that included the Tabulating Machine Company) bought premises at Vincennes, near Paris, for a machine shop which they opened the following year. The aim was to be able to assemble machine parts and components imported from the United States, as the import of components rather than complete machines was subject to more flexible customs regulations. The intention could also have been to secure French patent protection, which required the patented device to be produced in France within two years of the filing of the patent application.85
This explains why the French IBM companies never took legal proceedings against Powers in France for infringement of the patent—rather like the situation in Great Britain. The French application for Hollerith’s important automatic group control patent had been filed six years before the Vincennes machine shop was established, meaning that this patent was not valid in France.86
France became a battleground between the American Powers company and the British Powers company. In 1918, the British company Morland and Impey acquired the right to market the American Powers machines in France. Morland and Impey was the British importer of the Kalamazoo loose-leaf ledger system from the United States, and they had built up an extensive Kalamazoo business in France. Furthermore, Morland and Impey enjoyed good relations with the British Powers company and managed to obtain an additional agency for the British Powers company’s machines in France.
As all Powers machines during the 1920s were based on Hollerith’s numerical standard punched card from 1907, it was no problem to use a combination of machines from the two Powers producers. The Powers business in France prospered, and it was incorporated as the Société anonyme des machines à statistique (SAMAS, translated as Statistics Machines Company Limited) in 1922. However, the American Powers company cancelled their French agency the following year and established a French subsidiary company, while SAMAS continued to market the British Powers machines. The outcome was a lawsuit filed in France by the American company against SAMAS for patent infringement.87 The American company lost the case, as neither of the companies had any machine production in France that invalidated the French patents.
The link between SAMAS and the British Powers company was strengthened in 1929 when the British Powers company bought a controlling interest in SAMAS. In 1936, the American Powers company merged their businesses in France with SAMAS, so that it again became an agency for both the American and British Powers companies. This arrangement only lasted until 1939 when the American Powers company sold their French interests to the British Powers company.88
SAMAS seems to have done well during the 1920s, and in 1927 the company had thirty customers—the same number as IBM had in France.89 In 1931, the SAMAS customer base included the Finance and Naval Ministries, several banks, four railway companies, and many insurance companies.90 This information apparently conflicts with information from the French IBM company, which claims that during the 1920s the same railway companies and the Ministry of Finance used their equipment. There could be several reasons for this: First, each institution could have had two separate installations. Second, the conflicting information could have been caused by change in supplier. Third, they could have had combined installations, as both suppliers used the same 45-column punched card.
Dynamics of Technology Transfer and Adoption
The extensive demand in Great Britain, Germany, and France for punched cards to process general statistics after 1904 contrasted sharply with the lukewarm reception to Hollerith’s first punched-card system in the 1890s. The lack of demand in the 1890s was caused by the exclusive focus on processing census statistics. This was an area in which Europe had efficient organizational structures—in contrast to the United States. Further, the situation was exacerbated by the absence of organizations for selling and maintaining punched-card equipment in Europe.
Within this short time-frame the situation had changed significantly. Hollerith’s punched-card technology had been improved in the United States to facilitate the processing of general statistics, thus providing access to the much larger market for operational statistics in private companies and public organizations. In addition, organizations for selling and maintaining the installations were established in Great Britain (1904), Germany (1910), France (1920), and several other countries in Europe in the 1920s.
The rate of diffusion varied in Great Britain, Germany, and France. The technology spread faster in Germany than in Great Britain, and France saw the lowest rate of the three countries. The difference between Britain and Germany is clearly reflected in the slower increase in revenues for the British Tabulating Machine Company than for Dehomag. A historian explained this performance as being the result of inadequate leadership of the British Tabulating Machine Company.91 This was endorsed by Dehomag’s rapid establishment of an efficient sales organization within the first few years of its existence, including the publication of a regular sales bulletin. The British Tabulating Machine Company only established a nationwide sales organization in the early 1920s and first started to publish a regular sales bulletin in 1936.
This explanation through leadership in the European subsidiaries is complemented by an explanation based on different demand. In the 1910s and 1920s processing operational statistics became the prime application field of punched cards in all four countries. Alfred D. Chandler compared the forms in big industrial companies in the United States, Great Britain, and Germany.92 He documented the importance of the development of hierarchical organizational forms for the development of big industrial enterprises. He found the United States to have the most complex hierarchies, closely followed by Germany, while Great Britain had smaller hierarchies in their industrial enterprises. This observation is supported by the differences in the speed of punched-card diffusion in the three countries in the 1910s and the 1920s.
Chandler’s analysis did not include France. However, French economist Maurice Lévy-Leboyer compared French companies with businesses in Britain, Germany, and the United States. He found that French companies lacked industrial integration, so that the size of the firm, measured by total assets, was smaller than in other big industrial economies.93 This offers a further explanation for the lower use of punched cards in France in processing operational statistics. However, it cannot explain the total absence of punched-card processed operational statistics in France until 1921. There were companies in France that were rationalizing, diversifying, and introducing new organizational forms before the First World War. In Germany, director Carl Duisberg of Bayer ordered a punched-card installation from the United States before the Dehomag agency was established to process business statistics. A French punched-card installation for business statistics in the 1910s would only have required one comparable enterprising manager—and several such individuals became crucial in the subsequent history of the development of punched cards in France.
It is also essential to look at adaptations, as the later transfer of punched-card technology can be distinguished from the previous phase by the significant modifications to the technology and new facilities. Otto Schäffler in Vienna added plugboard programming to his version of Herman Hollerith’s first punched-card system from 1890. However, the reason for this modification is not documented and could simply result from his experience of telephone equipment rather than being a response to market demand.
The problem of computing sterling currency appeared in the very first business applications in Great Britain. This problem was limited to Britain as this was the only industrialized country using a nondecimal currency. But it exemplifies the problem of tackling national distinctions. A later example was the issue of how to cope with national characters, like ö, œ, ø, and ñ, on alphanumeric punched-card machines and computers. Resolving such issues was essential in diffusing first punched-card and later computer technology.
The ability to print processed numbers and results was considered essential by several punched-card users in Germany and Great Britain and became a crucial competitive advantage for the Powers company in Great Britain. The German agency of the Tabulating Machine Company designed a number printing unit for the machines imported from the United States. A prototype was completed in 1921 but was never produced, as American-produced numeric printing tabulators started to arrive in Germany.
There does not seem to be any record to substantiate that this German and British demand for technical improvement influenced the American Tabulating Machine Company’s introduction of a number printing punched-card machine. But this became the German company’s first experience in developing its own technology, which was continued in the interwar years when Dehomag developed a punched-card technology that was discernibly different from that of IBM in the United States. European users influenced the direction of punched-card technology and its applications in different ways than users and producers in the United States.