Only Connect

The first telephones [Previous Part] were point-to-point devices, connecting a single pair of stations. As early as 1877, however, Alexander Graham Bell envisioned a grand, interconnected system. Bell wrote in a prospectus for potential inventors that, just as municipal gas and water systems connected homes and offices throughout major cities to central distribution centers,¹

…it is conceivable that cables of telephone wires could be laid underground, or suspended overhead, communicating by branch wires with private dwellings, country houses, shops, manufactories, etc., etc., uniting them through the main cable with a central office where the wires could be connected as desired, establishing direct communication between any two places in the city… Not only so, but I believe, in the future, wires will unite the head offices of the Telephone Company in different cities, and a man in one part of the country may communicate by word of mouth with another in a distant place.

But neither he, nor any of his contemporaries, possessed the technical means to actually achieve this vision. It would take decades and the exercise of a great deal of ingenuity and labor to turn the telephone into the most vast and intricate machine yet known to mankind, spanning continents and, eventually oceans, to allow any telephone station in the world to connect to any other.

This transformation was enabled, among other things, by the development of the exchange – a central office with equipment to route a call from the line of the caller to that of the callee. The automation of these exchanges brought about a vast increase in the complexity of relay circuits, with important implications for the computer.

The First Exchanges

In the early days of the telephone, no one knew exactly what it was for. There was precedent enough for the long-distance delivery of written messages to know that the telegraph would have useful commercial and military applications. But no prior art illuminated the purpose for the long-distance transmission of sound. Was it a business instrument like the telegraph? A medium for social intercourse? A means for entertainment and edification, such as the broadcasting of music or political speeches?

Gardiner Hubbard, among Alexander Bell’s primary financial backers, did find one useful analogy. Enterprising telegraphers had built, over the previous decades, a number of local district telegraph companies. Wealthy individuals or small businesses rented a dedicated telegraph line to the company’s central dispatch office. By  sending a telegraph they could request a cab, have a message boy sent out to a client or friend, or summon the police. Hubbard believed the telephone could replace the telegraph in these businesses. It was much easier to use, and the ability to hold a conversation would provide speedier service and reduce misunderstandings. And so he encouraged the formation of exactly this sort of company, offering to rent Bell telephone equipment to district telephone companies, whether newly formed or converting from the telegraph.

A manager at one of these early district telephone companies might have noticed that he needed twenty phones to talk to twenty different customers. And in some cases one customer may have wanted to send a message to another: for example a doctor submitting a prescription to a pharmacist. Why not simply let them talk directly to one another?²

Bell himself was another potential source for this idea. He spent much of 1877 circumnavigating the lecture circuit to promote the telephone. George Coy attended such a lecture in New Haven, Connecticut, in which Bell expounded on his grand vision for a central telephone office. Suitably inspired, Coy formed the District Telephone Company of New Haven, acquired a license from the Bell Company and found his first subscribers. By January 1878 he had interconnected twenty-one of them via the first public telephone switchboard, a makeshift contraption of second-hand wire and teapot lid handles.³

Within the year, similar bespoke devices for interconnecting local telephone subscribers were spreading across the country. The world’s mental model for how the telephone would be used began to crystallize around these nexuses of local conversation – among merchants and suppliers, businessmen and clients, doctors and pharmacists. Even among friends and acquaintances, for those wealthy enough to afford such a luxury. Alternative visions of the telephone (e.g., as a broadcast medium) fell by the wayside.

Within a few years, these telephone offices also converged on a template for switchboard equipment that would remain stable for decades to come: an array of sockets that could be inter-connected by plugged cables wielded by the operator. They converged, too, on the ideal sex of that operator. At first the telephone companies, many of which evolved from telegraph companies, hired from the existing labor pool of young, male telegraph clerks and messenger boys. But customers complained of their rudeness while managers despaired at their rowdy behavior. It was not long before they were entirely replaced by polite, respectable young women.

The course of the further evolution of these central switching offices would be determined by a contest for dominance over the landscape of telephony: between the incumbent Goliath of the Bell system and the emerging challengers known as the independents.

Bell and the Independents

The holder of Bell’s 1876 patent 174,465 on “Improvement in Telegraphy”, the American Bell Telephone Company, benefited immensely from the broad scope granted to that patent by the American courts. By ruling that it covered not just the specific instruments that Bell described, but the very principle of transmitting sound via an undulating current, the judicial system granted American Bell an effective monopoly on the telephone in the United States until 1893, the end of the patent’s seventeen-year term.

The company’s leaders used this time wisely. Most notable among them were President William Forbes and Theodore Vail. Forbes was a Boston brahmin and leading member of the coterie of investors who took control of the company when the capital of Bell’s early partners ran dry. Vail, great-nephew of Samuel Morse’s partner Alfred Vail, presided over Bell’s most important operating company, Metropolitan Telephone in New York City, and acted as general manager for American Bell. Vail had proved his managerial mettle as superintendent of the Railway Mail Service, which sorted mail in railway cars enroute to its destination, and was considered among the most impressive logistic feats of the age.

Forbes and Vail focused especially on building up Bell’s presence in major cities, and on interconnecting those cities with long-distance lines. Given that the most valuable asset of a telephone company was its existing subscriber base, they expected that the unparalleled access to existing customers in Bell’s network would give them an unassailable competitive advantage in acquiring new customers after the expiration of their patents.

Bell generally expanded into new cities not as American Bell per se, but by licensing its patent portfolio to a local operating company, and buying a controlling interest in that company as part of the deal. To further the expansion of “long lines” to interconnect these urban offices, they founded yet another company, American Telephone and Telegraph (AT&T) in 1885. Vail added the presidency of this new company to his already substantial duties. Perhaps the most crucial investment in Bell’s portfolio, though, was the 1881 purchase of a controlling interest in Chicago electrical equipment maker Western Electric: originally co-founded by Bell rival Elisha Gray, later the primary equipment supplier for Western Union, now to become the manufacturing arm of the Bell system.

Only in the early 1890s, with Bell’s legal monopoly on the verge of expiration, did independent telephone companies began to emerge from the hiding places into which Bell had driven them with the truncheon known as U.S. Patent 174,465. For the next twenty years or so the independents remained a serious competitive threat to Bell, and both sides expanded rapidly in a battle for territory and subscribers. In order to fuel their expansion, Bell, in a flourish of legal and financial legerdemain, inverted their organizational structure to transform AT&T from held company to holding company. American Bell was organized under Massachusetts law, which still hewed to the older notion of a corporation as a limited public charter – therefore American Bell had to petition the legislature of that state in order to take on new capital. AT&T, organized under the liberalized corporate laws of New York, had no such requirement.

AT&T expanded its networks and founded or bought new companies to consolidate and defend its hold over the major urban centers, while weaving its ever-growing network of long lines across the country. The independents meanwhile claimed new territory as fast as possible, especially in the smaller towns where AT&T did not yet have presence.

During this period of intense competition, the number of telephones in service grew at an astonishing pace. By 1900 there were roughly 1.4 million telephones in the United States, as against 800,000 in Europe and 100,000 in the rest of the world together. This amounted to one phone for every sixty Americans. Outside the U.S., only Sweden and Switzerland came anywhere close to this density.  Of the 1.4 million American phone lines, 800,000 were Bell subscribers, with the rest belonging to the independents. Just three years later those numbers had grown to 3.3 million and 1.3 million, respectively, with tens of thousands of exchanges.⁴

The growing size of those exchanges placed ever greater strain on the central switching offices. In response, the telephone industry developed new switching technology along two main branches: one, which Bell favored, remained operator-assisted. The other, taken up by the independents, used electro-mechanical devices to eliminate the operator altogether.

For convenience, we will call this a split between manual and automatic switching. But it’s good to keep in mind that this terminology is a bit misleading. Much like “automated” grocery checkout stations, the electro-mechanical switches, especially in their earliest incarnations, put new burdens on the customer. They automated away labor costs from the point of view of the telephone company, but from a systemic point of view, they merely offloaded labor from a paid operator to the (unpaid) customer.

An Operator Is Standing By

During this era of competition, Chicago was a primary center of innovation in the Bell System. Angus Hibbard, general manager of Chicago Telephone, pushed the boundaries of telephony in order to expand its offerings to a wider customer base, in ways that made AT&T headquarters uncomfortable. But because of the loose affiliation between AT&T and its operating companies, they had no direct control over his actions, and could only watch and squirm.

Up to this time, most Bell customers were merchants, business leaders, doctors, or lawyers, who paid a flat yearly amount for unlimited telephone use. Few others could afford the $125 annual fee, equivalent to several thousand dollars today. To expand its service to a wider clientele, Chicago Telephone introduced three new offerings in the 1890s, of decreasing cost and service level.  The first tier was metered service on a multiple-party line, which cost the user a fee per call in addition to a much lower base fee (because it was shared among others on the same line). The operator recorded each customer’s use on paper: not until after World War I was the first automatic meter installed in Chicago. The second tier was the neighborhood exchange, which offered unlimited calling within a few blocks, but fewer operators per customer (and therefore longer average connection times) than full-service exchanges. Finally there was the “nickel-in-the-slot,” a pay phone installed in a customer’s home or office. A five-cent fee sufficed to call anywhere within the city, with a five minute limit enforced by  the operator. This was the first telephone service truly accessible to the middle-class, and by 1906 40,000 of the 120,000 phones in Chicago were nickel-in-the-slots.

In order to serve his rapidly growing subscriber base, Hibbard worked closely with Western Electric, whose main plant also lay in Chicago, and especially with Charles Scribner, its chief engineer. Though he has since faded into obscurity, Scribner, author of several hundred patents over the course of his career, was in his time a renowned inventor and engineer. Among his first achievements was the development of the standard switchboard for the Bell system, including a connector for the operator’s cord known as the “jack-knife” for its resemblance to a pocket-knife, later shortened to “jack”.

Scribner, Hibbard, and their teams re-engineered the central telephone office to enable operators to serve calls with ever increasing efficiency. The busy signal and “howler” (indicating that a phone was off the hook) relieved operators of the need to inform callers about these error conditions. Tiny electric lamps to indicate active calls replaced shutters which had to be manually reset by the operator. The operator’s greeting of “Hello,” which invited conversation, was replaced by “Number please,” which allowed only one possible response. Due to these and other changes, the average connection time for local calls in Chicago decreased from 45 seconds in 1887 to 6.2 seconds in 1900.⁵

While Chicago Telephone, Western Electric, and other arms of the Bell octopus worked to make operator connections as fast and efficient as possible, however, others were trying to dispense with the operator altogether.

Strowger

Devices for connecting telephones without human intervention were patented, exhibited, and put into service as early as 1879, by inventors in the United States, France, the United Kingdom, Sweden, Italy, Russia, and Hungary. In the U.S. alone, twenty-seven patents were filed for automatic telephone switchboards by 1889.⁶ Yet, as so often in our story, the credit for automatic switching has accrued disproportionately to a single name: Almon Strowger. This is not entirely unjust; those who preceded him built one-offs or private curiosities, had the misfortune to reside in small and slow-growing telephone markets, or simply never successfully exploited their ideas. Strowger’s machine was the first deployed at industrial scale. Though to call it Strowger’s machine is also an elision, for he never did build the thing himself.

Strowger, a fifty-year-old Kanas City schoolmaster turned undertaker, was an unlikely innovator in an era of increasing technical specialization. The tale of how he conceived of his switch has been variously told, and seems to belong more to the realm of myth than fact. All of the stories, however, revolve around Strowger’s frustration that the operator or operators at his local telephone exchange were diverting clients to a rival undertaker. It’s unclear, and at this point unknowable, whether Strowger was really the victim of such a conspiracy. But it seems more likely that he was not so good of an undertaker as he liked to believe. In any case, from this (possibly fevered) imagination emerged the idea for a “girl-less” telephone.

His 1889 patent described how it would work, with a rigid mechanical arm to replace the gracile one of the telephone operator. Rather than a corded jack, it held a metal contact point that would sweep through an arc in order to select among up to 100 different subscriber lines (either in a single plane, or, in the “two-motion” design, ten stacked planes of ten lines each).

The caller controlled the arm with two telegraph keys, one for the tens digit, the other for the ones digit. To connect to subscriber 57, the caller pressed the tens key five times to move the arm to the correct group of ten subscribers, then pressed the ones key seven times to move to the correct subscriber in that group, then pressed a final key to make the connection. On an operator-assisted phone, by contrast, the caller would simply lift the transmitter, wait for the operator to acknowledge, say ’57,’ and wait to be connected.

In addition to being laborious, the system was equipment-intensive: it required five wires from the subscriber station to the central office and two local batteries (one for controlling the switch, and one for talking). Bell by this time was already moving to a central battery system, so their newest subscriber stations had no battery and a single pair of wires.⁷

Strowger reportedly built the first model of his switch from from pins inserted into a stack of stiff, starched collars. In order to realize a practical instrument he required financial and technical assistance from several important partners: notably businessman Joseph Harris and engineer Alexander Keith. Harris provided Strowger with funding and oversaw the creation of the Strowger Automatic Telephone Exchange Company to manufacture switches. He wisely chose to site the company not in Kansas City, but in his own home of Chicago. Due to the presence of Western Electric, Chicago was the bustling hub of telephone engineering. Among the first engineers recruited was Keith, who crossed over from the world of electrical power generation, and became technical director of Strowger Automatic. Keith, with the help of other skilled engineers, transformed Strowger’s rough concept into a precision instrument ready for mass production and use, and oversaw all of the major technical improvements to that instrument over the next twenty years.

Two of those improvements were of special importance: first, the replacement of multiple keys with a single dial that automatically generated the pulses needed to move the switch into position, as well as the connection signal. This greatly simplified the subscriber equipment, and became the default mechanism for controlling automatic switches until Bell’s introduction of Touch-Tone in the 1960s. The automatic telephone became synonymous with the dial telephone. Second was the development of two-tiered switching systems that allowed first 1,000, then later 10,000, customers to connect to one another by dialing three or four numbers. The first-level switch selected among ten or 100 switches in the second level, and then that second-level switch selected among 100 subscriber lines. This made it possible for automatic switching to compete in larger towns and cities with thousands of subscribers.

Strowger Automatic installed its first commercial switch in La Porte, Indiana in 1892, serving the eighty subscribers of the independent Cushman Telephone Company. The former Bell affiliate in town had conveniently been swept out of the way after losing a patent dispute with AT&T, giving Cushman and Strowger a perfect opportunity to step in and sweep up its former customers. Five years later, Keith oversaw the first two-level installation at Augusta, Georgia, serving 900 lines.

By that time Strowger himself had retired to Florida, where he died several years later. His name was dropped from the company, now the Automatic Telephone Company, more commonly known as Autelco. Autelco was the dominant supplier of electro-mechanical switching equipment in the United States, and in much of Europe. By 1910, automatic switches served 200,000 American subscribers from 131 exchanges, almost all built by Autelco. Every one was owned by an independent telephone company.⁸ 200,000 was still only a small fraction, however, of the millions of American telephone subscribers. Even among the independents most still followed the lead of Bell, and Bell had yet to seriously consider replacing its operators.

Common Control

The Bell system’s opponents tried to attribute its commitment to operator-assisted switching to some kind of nefarious motive, but it is hard to find any of their insinuations convincing. There were several good reasons, and another that seemed sensible but appears specious in hindsight, that Bell resisted switching to automatic systems.

The first problem for Bell was to develop its own switching system. AT&T had no desire to pay Autelco to fit out its switching centers. Luckily, in 1903, it had acquired the patent to a device developed by the Lorimer brothers of Brantford, Ontario. This was the very town where Alexander Bell’s parents had settled after leaving Scotland, and where the notion of a telephone had first congealed in his imagination, during a visit in the summer of 1874. Unlike the Strowger switch, the Lorimer device used revertive pulses to move the arm of its selector – that is, the electric pulses originated at the switch, with each pulse triggering a relay in the subscriber equipment, causing it to count down the number set by subscriber on a lever until it hit zero.

In 1906, Western Electric tasked two separate teams to develop switches based on the core Lorimer patent, and the systems they created – panel and rotary switching – formed a second generation of automatic switching devices. Both replaced the Lorimers’ lever with a standard dial, removeing the revertive pulse receiver to the interior of the central office.

More important for our purposes than the mechanics of Western Electric’s switching equipment – carefully recorded by historians of the telephone in loving, not to say excruciating, detail – were the relay circuits used to control them – sadly neglected by those same historians, who briefly acknowledge their existence before passing on to the true objects of their devotion.

This is doubly unfortunate, because those relay control circuits have two important consequences for our story. In the long term, they inspired the realization that combinations of switches could be built to represent arbitrary arithmetic and logical operations. That realization will be the subject of our next installment. More immediately, they solved the last major engineering obstacle to adoption of automatic switching: the ability to scale it to serve the large urban areas where Bell had many thousands of subscribers.

The means by which Alexander Keith scaled the Strowger switch to 10,000 lines could not be stretched much further. Continuing to multiply switching levels simply required too much equipment to dedicate to each call. Bell engineers called the alternative scaling mechanism that they devised a sender. It stored the number dialed by the caller into a register, then translated that number into arbitrary (usually non-decimal) codes to control the switching machinery. This allowed for much more flexible switching arrangements – for example, calls between exchanges could be routed via a central office (which corresponded to no digit in the dialed number), rather than having to directly connect each exchange in a city to every other one.

It seems that Edward C. Molina, a research engineer at the AT&T Traffic Division, first conceived of the sender. Molina had made his mark with novel studies that applied the mathematics of probability to the study of telephone traffic. These studies lead him the realization, around 1905, that if call routing could be decoupled from the decimal number dialed by the user, much more efficient automated use could be made of the lines.

Molina had demonstrated mathematically that spreading calls over larger groups of lines allowed a switch to support a larger call volume while maintaining the same probability of a busy signal. Strowger selectors, however, were limited to 100 lines, selected by two dialed digits. 1,000 line selectors based on three digits had proved impractical. The movements of a selector controlled by a sender, however, did not have to correspond to decimal digits entered by the caller. Such a selector could choose from 200, or even 500, lines, as in fact the rotary and panel systems, respectively, did. Molina proposed a register and translation device built from a mix of relays and ratcheted wheels, but by the time AT&T was actually ready to deploy panel and rotary systems, other engineers had concocted faster senders, made solely from relay circuits.

It was a short step from the concept of the sender to the concept of common control. There was no need, the Western Electric teams realized, to have a sender for each subscriber line, or even for each active call. Instead, a small number of these control devices could be shared among all lines. When a call came in, a sender would engage briefly to record the dialed digits, talk to the switching equipment to route the call, then disengage to make itself ready for another call.  With the panel switch⁹, sender, and common control, AT&T had in hand a flexible and scalable system that could address even the needs of the massive New York and Chicago telephone networks.

Despite its engineers having swept away all possibly technical objections to operator-less telephony, however, executives at AT&T were still not sold. They were not convinced that users could reliably dial the six- or seven-digit numbers that would be required to enable automatic switching across large metropolitan areas. At the time, callers reached subscribers at other local exchanges by giving the operator two pieces of information: the exchange they wished to reach, and the (typically four-digit) number they wished to reach there. For example, a customer in Pasadena might reach a friend in Burbank by asking for “Burbank, 5553”.  Bell leadership believed that replacing “Burbank” with an arbitrary two- or three-digit code would lead to a high frequency of mis-dialed numbers, frustrating users and degrading the company’s quality of service.

In 1917, William Blauvelt, an employee of AT&T (role unknown), proposed a means to mollify this concern. When manufacturing the subscriber station, Western Electric could print two or three letters next to each number of the dial. The telephone directory, meanwhile, would show the first few letters of each exchange, corresponding to its dial code, in bold, for example Burbank. Instead of having to remember an arbitrary numerical code for the destination exchange, the caller would then simply dial the letters directly: BUR-5553.

Even with no remaining objections to its adoption of automatic switching, however, AT&T still had no compelling technical or operational reason to change its very successful method of connecting calls. Its hand was forced only by the advent of the Great War.  Massively increased demand for industrial production drove manufacturing wages ever upwards: in the U.S. they more than doubled between 1914 and 1919¹⁰, pulling wages in other sectors along in their wake. Suddenly the key point of comparison between operator-assisted and automatic switching became not technical or operational, but financial. Given the growing cost of employing operators, AT&T decided by 1920 that it could not afford not to mechanize, and gave the order to begin installing automatic offices.

The first such office, using a panel switch in Omaha, Nebraska, came online in 1921, followed by a New York City exchange in October 1922. By 1928 twenty percent of AT&T offices were automatic; by 1934, fifty percent; 1960, ninety-seven percent. Bell decommissioned its last operator-assisted exchange, in Maine, in 1978. Operators still had a role to play for the foreseeable future, however, in connecting long-distance calls, where they did not begin to be replaced by machines until after World War II.

It might be natural to assume, given the stories about technology and business that tend to have currency in our culture, that a lumbering AT&T had barely escaped destruction from nimble, innovative upstarts – the independents – by finally adopting the obviously superior technology that they had pioneered. But in fact AT&T had put paid to the threat of the independents a decade before it began seriously automating its switching centers.

Bell Triumphant

Two events in the decade before 1910 convinced most in the business community that no challenger would ever topple the Bell System. First was the failure of the United States Independent Telephone Company of Rochester, New York. United States Independent sought, for the first time, to build a competing long-distance network to AT&T’s. But they suffered financial ruin after failing to gain entry to the crucial New York City market. Second was the collapse of the independent Illinois Telephone and Telegraph’s effort to penetrate the Chicago market. Not only could no other company compete with AT&T’s long lines, it seemed that none could challenge Bell operating companies in the major urban markets either.

Moreover, the 1907 rechartering by the city of Chicago of Bell’s operating company there (Hibbard’s Chicago Telephone) signaled that city governments would not try to foster competition in the telephone business. The new economic concept of natural monopoly had taken hold – the belief that for certain kinds of public services, convergence on a single provider was both beneficial and the natural result of market forces. The correct response to monopoly, under this theory, was public regulation, not forced competition.¹¹

The 1913 “Kingsbury commitment” added the imprimatur of the federal government to Bell’s position. At first it seemed that the incoming progressive Wilson administration, deeply skeptical of massive corporate combinations, might break up the Bell system or otherwise curtail its dominance. So it certainly appeared when Wilson’s Attorney General, James McReynolds, immediately re-opened an anti-Bell lawsuit under the Sherman Anti-Trust Act that had been tabled by his predecessor. But AT&T and the government soon came to a brokered settlement, signed by company Vice President Nathan Kingsbury. AT&T agreed to divest Western Union (in which it had purchased a controlling interest several years earlier), to stop buying independent telephone companies, and to interconnect independents into its long-distance network at reasonable rates.

At face value, it may seem that AT&T had suffered a significant check to its ambitions. But the ultimate effect of the Kingsbury Commitment was to confirm it as the national power in telephony. Cities and states had already signaled that they would not try to force an end to monopoly in telephony, now the federal government had done the same. Moreover, the fact that independents could now get access to AT&T’s long distance network ensured that it would be the only such network of note in the United States until the advent of microwave transmission half a century later.

The independents became part of a vast machine with Bell at its center. In fact, the ban on acquisition of independent companies was lifted in 1921, because so many of those companies, eager to sell, petitioned the government to be allowed to do so. That being said, many independents did survive, and even thrive, notably General Telephone & Electric (GTE), which acquired Autelco as its counterpart to Western Electric and held its own collection of local operating companies. But they all felt the gravitational pull of the Bell star around which they circled.

Despite their now-comfortable situation, Bell’s leaders had no intention of staying idle. In order to promote the innovations in telephony that would ensure its continued dominance, AT&T President Walter Gifford formed the Bell Telephone Laboratories in 1925, with some 4,000 employees. Bell also soon developed a third-generation automatic switching system, the crossbar switch, controlled by the most complex relay circuits yet known. These two developments would lead two men, George Stibitz and Claude Shannon, to ponder the curious analogies between circuits of switches and systems of mathematical logic and computation. [Next Part]

Sources

Christopher Beauchamp, Invented by Law: Alexander Graham Bell and the Patent That Changed America (2015)

John Brooks, Telephone: The First Hundred Years (1975)

Robert J. Chapuis, One Hundred Years of Telephone Switching, vol. 1 (1982)

M.D. Fagen, ed., A History of Engineering and Science in the Bell System: The Early Years (1875-1925) (1975)

Anton A. Huurdeman, The Worldwide History of Telecommunications (2003)

Richard R. John, Network Nation (2010)

Oscar Myers, “Common Control Telephone Switching Systems,” The Bell System Technical Journal (November 1952)