In the 1940s, the electronic digital computer was a new, largely unproven machine developed in response to specific needs like the code-breaking requirements of Bletchley Park or the ballistics calculations of the Aberdeen Proving Grounds. Once these early computers proved their worth, projects like the Manchester Mark 1, EDVAC, and EDSAC implemented a stored program concept that allowed digital computers to become useful for a wide variety of scientific and business tasks. In the early 1950s, several for-profit corporations built on this work to introduce mass-produced computers and offered them to businesses, universities, and government organizations around the world. As previously discussed, Ferranti in the United Kingdom introduced the first such computer by taking the Manchester Mark 1 design, increasing the speed and storage capacity of the machine, and releasing it as the Ferranti Mark 1 in February 1952. This would be one of the few times that the United Kingdom led the way in computing over the next several decades, however, as demand remained muted among the country’s conservative businesses, allowing companies in the larger U.S. market to grow rapidly and achieve world dominance in computing.
Note: This is the third of four posts in a series of “historical interludes” summarizing the evolution of computer technology between 1830 and 1960. The information in this post is largely drawn from Computer: A History of the Information Machine by Martin Campbell-Kelly and William Aspray, The Maverick and His Machine: Thomas Watson, Sr. and the Making of IBM by Kevin Maney, A History of Modern Computing by Paul Ceruzzi, Computers and Commerce: A Study of Technology and Management at Eckert-Mauchly Computer Company, Engineering Research Associates, and Remington Rand, 1946-1957 by Arthur Norberg, and IBM’s Early Computers by Charles Bashe, Lyle Johnson, John Palmer, and Emerson Pugh.
The UNIVAC I, the first commercially available computer in the United States
For a brief period from 1943 to 1946, the Moore School in Philadelphia was the center of the computer world as John Mauchly and J. Presper Eckert developed ENIAC and initiated the EDVAC project. Unlike the more accommodating MIT and Stanford, however, which nurtured the Route 128 tech corridor and Silicon Valley respectively by encouraging professors and students to apply technologies developed in academia to the private sector, the Moore School believed commercial interests had no place in an academic institution and decided to quash them entirely. In early 1946 the entire staff of the school was ordered to sign release forms giving up the rights to all patent royalties from inventions pioneered at the school. This was intolerable to both Eckert and Mauchly, who formally resigned on March 31, 1946 to pursue commercial opportunities.
While still at the Moore School, Mauchly met with several organizations that might be interested in the new EDVAC computer. One of these was the Census Bureau, which once again needed to migrate to new technologies as tabulating machines were no longer sufficient to count the U.S. population in a timely manner. After leaving the school, Eckert and Mauchly attended a series of meetings with the Census Bureau and the National Bureau of Standards (NBS) between March and May devoted to the possibility of replacing tabulating machines with computers. After further study, the NBS entered into an agreement with Eckert and Mauchly on September 25, 1946, for them to develop a computer for the Census Bureau in return for $300,000, which Eckert and Mauchly naively believed would cover a large portion of their R&D cost.
Census contract aside, Eckert and Mauchly experienced great difficulty attempting to fund the world’s first for-profit electronic computer company. Efforts to raise capital commenced in the summer of 1946, but Philadelphia-area investors were focused on the older industries of steel and electric power that had driven the region for decades. In New York, there was funding available for going electronic concerns, but the concept of venture capital did not yet exist and no investment houses were willing to take a chance on a startup. The duo were finally forced to turn to friends and family, who provided enough capital in combination with the Census contract for Eckert and Mauchly to establish a partnership called the Electric Control Company in October 1946, which later incorporated as the Eckert-Mauchly Computer Corporation (EMCC) in December 1948.
As work began on the EDVAC II computer at the new Philadelphia offices of the Electric Control Company, the founders continued to seek new contracts to alleviate chronic undercapitalization. In early 1947 Prudential, a forward-thinking company that had a reputation as an early adopter of new technology, agreed to pay the duo $20,000 to serve as consultants, but refused to commit to ordering a computer until it was completed. Market research film A.C. Nielsen placed an order in spring 1948 and Prudential changed its mind and followed suit late in the year, but both deals were for $150,000 as Eckert and Mauchly continued to underestimate the cost of building their computers. To keep the company solvent, the duo completed a $100,000 deal with Northrop Aircraft in October 1947 for a smaller scientific computer called the Binary Automatic Computer (BINAC) for use in developing a new unmanned bomber. Meanwhile, with contracts coming in Eckert and Mauchly realized that they needed a new name for their computer to avoid confusion with the EDVAC project at the Moore School and settled on UNIVAC, which stood for Universal Automatic Computer.
EMCC appeared to finally turn a corner in August 1948 when it received a $500,000 investment from the American Totalisator Company. The automatic totalisator was a specialized counting machine originally invented by New Zealander George Julius in the early twentieth century to tally election votes and divide them properly among the candidates. When the government rejected the device, he adapted it for use at the race track, where it could run a pari-mutual betting system by totaling all bets and assigning odds to each horse. American Totalisator came to dominate this market after one of its founders, Henry Strauss, invented and patented an electro-mechanical totalisator first used in 1933. Strauss realized that electronic computing was the logical next step in the totalisator field, so he convinced the company board to invest $500,000 in EMCC in return for a 40% stake in the company. With the funding from American Totalisator, EMCC completed BINAC and delivered it to Northrop in September 1949. Although it never worked properly, BINAC was the first commercially sold computer in the world. Work continued on UNIVAC as well, but disaster struck on October 25, 1949, when Henry Strauss died in a plane crash. With EMCC’s chief backer at American Totalisator gone, the company withdrew its support and demanded that its loans be repaid. Eckert and Mauchly therefore began looking for a buyer for their company.
On February 15, 1950, office equipment giant Remington Rand purchased EMCC for $100,000 while also paying off the $438,000 owed to American Totalisator. James Rand, Jr., the president of the company, had become enamored with the scientific advances achieved during World War II and was in the midst of a post-war expansion plan centered on high technology and electronic products. In 1946, Rand constructed a new high-tech R&D lab in Norwalk, Connecticut, to explore products as varied as microfilm readers, xerographic copiers, and industrial television systems. In late 1947, he hired Leslie Groves, the general who oversaw the Manhattan Project, to run the operation. EMCC therefore fit perfectly into Rand’s plans. Though Eckert and Mauchly were required to give up their ownership stakes and take salaries as regular employees of Remington Rand, Groves allowed them to remain in Philadelphia and generally let them run their own affairs without interference.
With Remignton Rand sorting out its financial problems, EMCC was finally able to complete its computer. First accepted by the U.S. Census Bureau on March 31, 1951, the UNIVAC I contained 5,200 vacuum tubes and could perform 1,905 operations a second at a clock speed of 2.25 MHz. Like the EDVAC and EDSAC, the UNIVAC I used delay line memory as its primary method of storing information, but it also pioneered the use of magnetic tape storage as a secondary memory, which was capable of storing up to a million characters. The Census Bureau resisted attempts by Remington Rand to renegotiate the purchase price of the computer and spent only the $300,000 previously agreed upon, while both A.C. Nielsen and Prudential ultimately cancelled their orders when Remington Rand threatened to tie up delivery through a lawsuit to avoid selling the computers for $150,00 dollars; future customers were forced to pay a million dollars or more for a complete UNIVAC I.
By 1954, nineteen UNIVAC computers had been purchased and installed at such diverse organizations as the Pentagon, U.S. Steel, and General Electric. Most of these organizations took advantage of the computer’s large tape storage capacity to employ the computer for data processing rather than calculations, where it competed with the tabulating machines that had brought IBM to prominence.
The UNIVAC 1101, Remington Rand’s first scientific computer
To serve the scientific community, Remington Rand turned to another early computer startup, Engineering Research Associates (ERA). ERA grew out of the code-breaking activities of the United States Navy during World War II, which were carried out primarily through an organization called the Communications Supplementary Activity – Washington (CSAW). Like Bletchley Park in the United Kingdom, CSAW constructed a number of sophisticated electronic devices to aid in codebreaking, and the Navy wanted to maintain this technological capability after the war. Military budget cuts made this impractical, however, so to avoid losing the assembly of talent at CSAW, the Navy helped establish ERA in St. Paul, Minnesota, in January 1946 as a private corporation. The company was led by John Parker, a former Navy lieutenant who had become intimately involved in the airline industry in the late 1930s and 1940s while working for the D.C. investment firm Auchincloss, Parker, and Redpath, and drew most of its important technical personnel from CSAW.
Unlike EMCC, which focused on building a machine for corporate data processing, ERA devoted its activities to intelligence analysis work for the United States Navy. Like Eckert and Mauchly, the founders of ERA realized the greatest impediment to building a useful electronic computing device was the lack of suitable storage technology, so in its first two years of existence, the company concentrated on solving this problem, ultimately settling on magnetic drum memory, a technology invented by Austrian Gustav Tauchek in 1932 in which a large metal cylinder is coated with a ferromagnetic magnetic material. As the drum is rotated, stationary write heads can generate an electrical pulse to change the magnetic orientation on any part of the surface of the drum, while a read head can detect the orientation and recognize it in binary as either a “1” or a “0,” therefore making it suitable for computer memory. A series of specialized cryptoanalytic machines followed with names like Goldberg and Demon, but these machines tended to become obsolete quickly since they were targeted at specific codes and were not programmable to take on new tasks. Meanwhile, as both ERA and the Navy learned more about developments at the Moore School, they decided a general purpose computer would be a better method of addressing the Navy’s needs than specialized equipment and therefore initiated Task 13 in 1947 to build a stored program computer called Atlas. Completed in December 1950, the Atlas contained 2,700 vacuum tubes and a drum memory that could hold just over 16,000 24-bit words. The computer was delivered to the National Security Agency (NSA) for code-breaking operations, and the agency was so pleased with the computer that it accepted a second unit in 1953. In December 1951, a modified version was made available as the ERA 1101 — a play on the original project name as “1101” is “13” in binary — but ERA did not furnish any manuals, so no businesses purchased the machine.
The same month ERA announced the 1101, it was purchased by Remington Rand. ERA president John Parker realized that fully entering the commercial world would require a significant influx of capital that the company would be unlikely to raise. Furthermore, the close relationship between ERA and the Navy had piqued the interest of government auditors and threatened the company’s ability to secure future government contracts. Therefore, Parker saw the Remington Rand purchase as essential to ERA’s continued survival. Remington Rand, meanwhile, gained a foothold in a new segment of the computer market. The company began marketing an improved version of ERA’s first computer as the UNIVAC 1103 in October 1953 and ultimately installed roughly twenty of them, mostly within the military-industrial complex.
In 1952, the American public was introduced to the UNIVAC in dramatic fashion when Mauchly developed a program to predict the results of the general election between Dwight Eisenhower and Adlai Stevenson based on the returns from the previous two elections. The results were to be aired publicly on CBS, but UNIVAC predicted a massive landslide for Eisenhower in opposition to Gallup polls that indicated a close race. CBS refused to deliver the results, opting instead to state that the computer predicted a close victory for Eisenhower. When it became clear that Eisenhower would actually win in a landslide, the network owned up to its deception and aired the true results, which were within just a few electoral votes of the actual total. Before long, the term “UNIVAC” became a generic word for all computers in the same way “Kleenex” has become synonymous with tissue paper and “Xerox” with photocopying. For a time, it appeared that Remington Rand would be the clear winner in the new field of electronic computers, but only until IBM finally hit its stride.
IBM Enters the Computer Industry
Tom Watson, Sr. sits at the console of an IBM 701, the company’s first commercial computer
There is a story, oft-repeated, about Tom Watson, Sr. that claims he saw no value in computers. According to this story, the aging president of IBM scoffed that there would never be a market for more than five computers and neglected to bring IBM into the new field. Only after the debut of the UNIVAC I did IBM realize its mistake and hastily enter the computer market. While there are elements of truth to this version of events, there is no truth to the claim that IBM was completely ignoring the computer market in the late 1940s. Indeed, the company developed several electronic calculators and had no fewer than three computer projects underway when the UNIVAC I hit the market.
As previously discussed, IBM’s involvement with computers began when the company joined with Howard Aiken to develop the Automatic Sequence Controlled Calculator (ASCC). That machine was first unveiled publicly on August 6, 1944, and Tom Watson traveled to Cambridge, Massachusetts, to speak at the dedication. At the Boston train station, Watson was irked that no one from Harvard was there to welcome him. Irritation turned to rage when he perused the Boston Post and saw that Harvard had not only issued a press release about the ASCC without consulting him, but also gave sole credit to Howard Aiken for inventing the machine. When an angry and humiliated Watson returned to IBM, he ordered James Bryce and Clair Lake to develop a new machine that would make Aiken’s ASCC look like a toy. Watson wanted to show the world that IBM could build computers without help from anyone else and to get revenge on the men he felt wronged him.
With IBM seriously engaged in war work, Bryce and Lake felt they would be unable to achieve the breakthroughs in the lab necessary to best Aiken in a reasonable time frame, so instead argued for a simpler goal of creating the world’s first automatic calculator. To that end, an electronics enthusiast in the company named Haley Dickinson was ordered to convert the company’s electro-mechanical Model 601 Multiplying Punch into a tube-based machine. Unveiled in September 1946 as the IBM 603 Electronic Multiplier, the machine contained only 300 vacuum tubes and no storage, but it could multiply ten times faster than existing tabulating machines and soon became a sensation. Embarrassed by the limitations of the machine, however, Watson halted production at 100 units and ordered his engineers to develop an improved model. Ralph Palmer, an electronics expert that joined IBM in 1932 and was recently returned from a stint in the Navy, was asked to form a new laboratory in Poughkeepsie, New York, dedicated solely to electronics. Palmer’s group delivered the IBM 604 Electronic Calculating Punch in 1948, which contained 1,400 tubes and could be programmed to solve simple equations. Over the next ten years, the company leased 5,600 604’s to customers, and Watson came to realize that the future of IBM’s business lay in electronics.
Meanwhile, as World War II neared its conclusion, Watson’s mandate to best Aiken’s ASCC gained momentum. The man responsible for this project was Wallace Eckert (no relation to the ENIAC co-inventor), who as an astronomy professor at Columbia in the 1920s and 1930s had been one of the main beneficiaries of Watson’s relationship with the university in those years. After directing the Nautical Almanac of the United States Naval Observatory during much of World War II, Eckert accepted an invitation from Watson in March 1945 to head a new division within IBM specifically concerned with the computational needs of the scientific community called the Pure Science Department.
Eckert remained at headquarters in New York while Frank Hamilton, who had been a project leader on the ASCC, took charge of defining the Aiken-beating machine’s capabilities in Endicott. In summer 1945, Eckert made new hire Rex Seeber his personal representative to the project. A Harvard graduate, Seeber had worked with Aiken, but fell out with him when he refused to implement the stored program concept in his forthcoming update of the ASCC. Seeber’s knowledge of computer theory and electronics perfectly complemented Hamilton’s electrical engineering skills and resulted in the completion of the Selective Sequence Electronic Calculator (SSEC) in 1947. The SSEC was the first machine in the world to successfully implement the stored program concept, although it is often classified as a calculator rather than a stored program computer due to its limited memory and reliance on paper tape for program control. The majority of the calculator remained electromechanical, but the arithmetic unit, adapted from the 603, operated at electronic speeds. Built with 21,400 relays and 12,500 vacuum tubes and assembled at a cost of $950,000, the SSEC was a strange hybrid that exerted no influence over the future of computing, but it did accomplish IBM’s objectives by operating 250 times faster than the Harvard ASCC while also gaining significant publicity for IBM’s computing endeavors by operating while on display to the public on the ground floor of the company’s corporate headquarters from 1948 to 1952.
Tom Watson, Jr., son and successor of Tom Watson, Sr.
The success of the IBM 603 and 604 showed Watson that IBM needed to embrace electronics, but he remained cautious regarding electronic computing. Indeed, when given the chance to bring Eckert and Mauchly into the IBM fold in mid-1946 after they left the Moore School, Watson ultimately turned them down not because he saw no value in their work but because he did not want to meet the price they demanded to buy out their business. When he learned that the duo’s computer company was garnering interest from the National Bureau of Standards and Prudential in 1947, he told his engineers they should explore a competing design, but he was thinking in terms of a machine tailored to the needs of specific clients rather than a general-purpose computing device. By now Watson was in his seventies and set in his ways, and while there is no evidence that he ever uttered the famous line about world demand reaching only five computers, he could simply not envision a world in which electronic computers replaced tabulating machines entirely. As a result, the push for computing within the company came instead from his son and heir apparent, Tom Watson, Jr.
Thomas J. Watson, Jr. was born in Dayton, Ohio, in 1914, the same year his father accepted the general manager position at C-T-R. His relationship with his father was strained for most of his life, as the elder Watson was prone to both controlling behavior and ferocious bursts of temper. While incredibly bright, Watson suffered from anxiety and crippling depression as a child and felt incapable of living up to his father’s standards or of succeeding him at IBM one day, which he sensed was his father’s wish. As a result, he rebelled and performed poorly in school, only gaining admittance to Brown University as a favor to his father. After graduating with a degree in business in 1937, he became a salesman at IBM, but grew to hate working there due to the special treatment he received as the CEO’s son and the cult of personality that had grown up around his father. Desperate for a way out, he joined the Air National Guard shortly before the United States entered World War II and became aide-de-camp to First Air Force Commander Major General Follett Bradley in 1942. He had no intention of ever returning to IBM.
Working for General Bradley, Watson finally realized his own potential. He became the general’s most trusted subordinate and gained experience managing teams undertaking difficult tasks. With the encouragement of Bradley, his inner charisma surfaced for the first time, as did a remarkable ability to focus on and explain complex problems. Near the end of the war, Bradley asked Watson about his plans for the future and was shocked when Watson said he might become a commercial pilot and would certainly never rejoin IBM. Bradley stated that he always assumed Watson would return to run the company. In that moment, Watson realized he was avoiding the company because he feared he would fail, but that his war experiences had prepared him to succeed his father. On the first business day of 1946, he returned to the fold.
Tom Jr. was not promoted to a leadership position right away. Instead, Tom Sr. appointed him personal assistant to Charley Kirk, the executive vice president of the company and Tom Sr.’s most trusted subordinate. Kirk generously took Tom Jr. under his wing, but he also appeared to be first in line to take over the company upon Tom Sr.’s retirement, which Tom Jr. resented. A potential power struggle was avoided when Kirk suffered a massive heart attack and died in 1947. Tom Sr. did not feel his son was quite ready to assume the executive vice president position, but Tom Jr. did assume many of Kirk’s responsibilities while an older loyal Watson supporter named George Phillips took on the executive VP role on a short-term basis. In 1952, Tom Sr. finally named Tom Jr. president of IBM.
The IBM 650, IBM’s most successful early computer
Tom Jr. first learned of the advances being made in computing in 1946 when he and Kirk traveled to the Moore School to see the ENIAC. He became a staunch supporter of electronics and computing from that day forward. While there was no formal division of responsibilities drawn up between father and son, it was understood from the late forties until Tom Jr. succeeded his father as IBM CEO in 1956 that Tom Jr. would be given free reign to develop IBM’s electronics and computing businesses, while Tom Sr. concentrated on the traditional tabulating machine business. In this capacity, Tom Jr. played a significant role in overcoming bias within IBM’s engineering, sales, and future demands divisions towards new technologies and brought IBM fully into the computer age.
By 1950, IBM had two computer projects in progress. The first had been started in 1948 when Tom Watson, Sr. ordered his engineers to adapt the SSEC into something cheaper that could be mass produced and sold to IBM’s business customers. With James Bryce incapacitated — he would die the next year — the responsibility of shaping the new machine fell to Wallace Eckert, Frank Hamilton, and John McPherson, an IBM vice president that had been instrumental in constructing two powerful relay calculators for the Aberdeen Proving Grounds during World War II. The trio decided to create a machine focused on scientific and engineering applications, both because this was their primary area of expertise and because with the dawn of the Cold War the United States government was funding over a dozen scientific computing projects to maintain the technological edge it had built during World War II. There was a real fear that if IBM did not stay relevant in this area, one of these projects could birth a company capable of challenging IBM’s dominant position in business machines.
Hamilton acted as the chief engineer on the project and chose to increase the system’s memory capacity by incorporating magnetic drum storage, thus leading to the machine’s designation as the Magnetic Drum Calculator (MDC). While the MDC began life as a calculator essentially pairing an IBM 603 with a magnetic drum, the realization that drum memory was expansive enough that a paper tape reader could be discarded entirely and instructions could be read and modified directly from the drum itself caused the project to morph into a full-fledged computer. By early 1950, engineering work had commenced on the MDC, but development soon stalled as it became the focus of fights between multiple engineering teams as well as the sales and future demands departments over its specifications, target audience, and potential commercial performance.
While work continued on the MDC in Endicott, several IBM engineers in the electronics laboratory in Poughkeepsie initiated their own experiments related to computer technology. In 1948, an engineer named Philip Fox began studying alternate solutions to vacuum tube memory that would allow for a stored-program computer. Learning of the Williams Tube in 1948, he decided to focus his attention on CRT memory. Fox created a machine called the Test Assembly on which he worked to improve on the reliability of existing CRT memory solutions. Meanwhile, in early 1949, a new employee named Nathaniel Rochester who was dismayed that IBM did not already have a stored-program computer in production began researching the capabilities of magnetic tape as a storage medium. These disparate threads came together in October 1949 when a decision was made to focus on the development of a tape machine to challenge the UNIVAC, which appeared poised to grab a share of IBM’s data processing business. By March 1950, Rochester and Werner Buchholz had completed a technical outline of the Tape Processing Machine (TPM), which would incorporate both CRT and tape memory. As with the MDC, however, sales and future demands’ inability to clearly define a market for the computer hindered its development.
A breakthrough in the stalemate between sales and engineering finally occurred with the outbreak of the Korean War. As he had when the United States entered World War II, Tom Watson, Sr. placed the full capabilities of the company at the disposal of the United States government. The United States Air Force quickly responded that it wanted help developing a new electro-mechanical bombsight for the B-47 Bomber, but Tom Watson, Jr., who already believed IBM was not embracing electronics fast enough, felt working on electro-mechanical projects to be a giant step backwards for the company. Instead, he proposed developing an electronic computer suitable for scientific computation by government organizations and contractors.
Initially, IBM considered adapting the TPM for its new scientific computer project, but quickly abandoned the idea. To save on cost, the engineering team of the TPM had decided to design the computer to process numbers serially rather than in parallel, which was sufficient for data processing, but made the machine too slow to meet the computational needs of the government. Therefore, in September 1950 Ralph Palmer’s engineers drew up preliminary plans for a floating-point decimal computer hooked up to an array of tape readers and other auxiliary devices that would be capable of well over 10,000 operations a second and of storing 2000 thirteen-digit words in Williams Tube memory. Watson Jr. approved this project in January 1950 under the moniker “Defense Calculator.” With a tight deadline of Spring 1952 in place for the Defense Calculator so it would be operational in time to contribute to the war effort, Palmer realized the engineering team, led by Nathaniel Rochester and Jerrier Haddad, could not afford to start from scratch on the design of the new computer, so they decided to base the architecture on von Neumann’s IAS Machine.
The IBM 702, IBM’s first computer targeted at businesses
On April 29, 1952, Tom Watson, Sr. announced the existence of the Defense Calculator to IBM’s shareholders at the company’s annual meeting. In December, the first completed model was installed at IBM headquarters in the berth occupied until then by the SSEC. On April 7, 1953, the company staged a public unveiling of the Defense Calculator under the name IBM 701 Electronic Data Processing Machine four days after the first production model had been delivered to the Los Alamos National Laboratory in New Mexico. By April 1955, when production ceased, IBM had completed nineteen installations of the 701 — mostly at government organizations and defense contractors like Boeing and Lockheed — at a rental cost of $15,000 a month.
The success of the 701 finally broke the computing logjam at IBM. The TPM, which had been on the back burner as the Defense Calculator project gained steam, was redesigned for faster operation and announced in September 1953 as the IBM 702, although the first model was not installed until July 1955. Unlike the 701, which borrowed the binary numeral system from the IAS Machine, the 702 used the decimal system as befit its descent from the 603 and 604 electronic calculators. It also shipped with a newly developed high speed printer capable of outputting 1,000 lines per minute. IBM positioned the 702 as a business machine to compete with the UNIVAC I and ultimately installed fourteen of them. Meanwhile, IBM also reinstated the MDC project — which had stalled almost completely — in November 1952, which saw release in 1954 as the IBM 650. While the drum memory used in the 650 was slower than the Williams Tube memory of the 701 and 702, it was also more reliable and cheaper, allowing IBM to lease the 650 at the relatively low cost of $3,250 a month. As a result, it became IBM’s first breakout success in the computer field, with nearly 2,000 installed by the time the last one rolled off the assembly line in 1962.
IBM’s 700 series computers enjoyed several distinct advantages over the UNIVAC I and UNIVAC 1103 computers marketed by Remington Rand. Technologically, Williams Tube memory was both more reliable and significantly faster than the mercury delay line memory and drum memory used in the UNIVAC machines, while the magnetic tape system developed by IBM was also superior to the one used by Remington Rand. Furthermore, IBM designed its computers to be modular, making them far easier to ship and install than the monolithic UNIVAC system. Finally, IBM had built one of the finest sales and product servicing organizations in the world, making it difficult for Remington Rand to compete for customers. While UNIVAC models held a small 30 to 24 install base edge over the 700 series computers as late as August 1955, IBM continued to improve the 700 line through newly emerging technologies and just a year later moved into the lead with 66 700 series installations versus 46 UNIVAC installations. Meanwhile, installations of the 650 far eclipsed any comparable model, giving IBM control of the low end of the computer market as well. The company would remain the number one computer maker in the world throughout the mainframe era.