Calculators
Calculators assist people with arithmetic. As science, industry, commerce, shipping, finance and insurance grew in the 19th century accurate calculation became imperative. Steam engines would explode, bridges collapse and ships sink without proper calculation. Furthermore shopkeepers invoices wouldn't get paid.
Technical progress in the 19th and 20th centuries was intricately linked to calculation. Good government required censuses, but a census creates vast amount of data that needed calculation. Maps were made by triangulation, so even the shape of the world was unknown without calculation. Electrical, chemical, and aircraft design in particular are reliant on calculation.
Computers are based on calculators. Historically a ‘computer’ was a person skilled in arithmetic who could perform calculations quickly and accurately, a skill comparable to playing a musical instrument. Great numbers of calculations were needed to produce the log tables used in science and navigation and the annuity tables for insurance as well as the basic arithmetic needed for commerce and science. The problem was (and remains) that although a few people can perform arithmetic quickly and accurately most cannot. Throw a ball at a boy and he can learn to hit it with a bat, so he can effortlessly calculate the ballistic curve of an object and a response simultaneously. Show a man the trajectory as algebra and several minutes of thinking later you might get a result.
Modern computers are primarily word-processing and communication machines but those tasks are done by a combination of memory and arithmetic. The core of today's computers is a binary calculator, combined with memory and program control.
Calculators are mostly low-cost convenience items these days, often used on a desk that also has a computer on it. It is just fractionally more convenient to grab the calculator than to clutter the computer screen. It wasn't always like this: well into the 1960s businesses would pay a years wages for a calculator - it made people much more productive and there was no alternative. So calculators along with teletypes and tabulators are one of the underpinnings of Information Technology and were a major industry until the 1980s.
Navigation, accounting and scientific calculation are now usually dealt with by computers so arithmetic problems are less visible than they once were. Simulating the behaviour of weather systems, electrons or the economy or to mimicking a thinking process are the new problems. Underlying it all is still a mass of calculation.
Abacus
An abacus is more of an aid to thinking than a calculating device although is skilled hands it can be very impressive.
The Greeks, Persians and Romans all had abacuses and the idea may be older.
The Chinese Suanpan remains in use today, there are techniques to perform the four basic arithmetic operations as well as square and cube root operations very quickly. The abacus doesn't seem to give rise to any automatic method of calculation, although the Jevons Logic Piano built in 1869 might refute that remark.
Mechanical Calculators
Mechanical calculators have been reinvented several times.
The Antikythera mechanism, thought to date back to 100BC, was a complicated astronomical clock. Its true purpose and complexity are uncertain because the machine was discovered in 1901 in an ancient shipwreck. The remains are encrusted with marine deposits and parts may be missing. The Antikythera mechanism looks like a hoax, since a two thousand years old clockwork computer is unexpected. It seems not to be.
Clockwork parts appear in Islamic civilization in the 13th century and in Europe somewhere between 1280 and 1320. Church records make a number of references to clocks and horologues. A clock rather implies the possibility of a calculator, but it wasn't until the 17th Century that this seems to have been realised.
Logarithms and Slide Rules
The idea of a slide rule is to perform arithmetic by adding or subtracting lengths. If the rulers are marked linearly with (say) numbers 1 to 100 people can add any numbers up to 200 by setting two rulers alongside one another. Since most people can add numbers this size in their heads that isn't much use. Rulers to add four or five digit numbers might be useful but too long.
Rulers marked with logarithmic scales are more interesting. Adding and subtracting logarithms has the effect of multiplying and dividing, which are more troublesome things to do. We need multiplication and division to work out areas, weights, quantities, distances travelled and probabilities. Logarithms were invented by John Napier around 1614. The slide-rule as a single logarithmic scale was invented around 1620 by Edmund Gunter. William Oughtred combined two Gunter rules to make something like a modern slide-rule.
Henry Briggs devised ways to calculate logarithms. (the pseudo-division/pseudo-multiplication method subsequently used in calculators).
Slide rules and logarithmic tables became the basis of scientific and engineering calculation for more than 3 centuries until electronic devices became commonplace.
Slide rules are precise enough for a lot of engineering and science work, but the scale is difficult to read and sometimes not quite accurate enough for finance. Basically if you buy 9s-11p worth of goods and give the shopkeeper £1 you want 10s-1p change - not to be told 3 digit accuracy is unnecessary.
The Shickard Calculator
Wilhelm Schickard designed a Calculating Clock in 1623. It had the numbered cog wheels and digit carry wheels that would be used on future devices. Unfortunately it was destroyed in a fire before he could complete it and the drawings were only rediscovered in 1957. Inventions sometimes have no influence at all. Schickard's achievements were as a cartographer.
Pascaline
Credit for inventing the mechanical calculator therefore belongs to Blaise Pascal, whose "Pascaline" of 1642 became the inspiration for other mechanical calculators. The principle used in the pascaline is a set of numbered cogwheels between which are carry wheels.
Leibniz
Gottfried Wilhelm Leibniz invented the "Stepped Reckoner"around 1672. It had the advantage that it could do all four arithmetic operations. Liebniz wheels are cylindrical wheels with teeth of varying length. Engage the wheel at one point and it sends one "vibration" into successive cogs, at another 2, 3, 4 through to nine. The idea of carry and stepped cogs based on the Pascaline and Stepped Reckoner became the basis of mechanical adding machines that were made all the way through to the 1970s. Leibniz actual machines weren't all that good, metalworking technology wasn't up to producing large numbers of cogs with the accuracy needed .
Thomas's Arithmometer
Thomas de Colmar invented the arithmometer sometime around 1815 whilst serving in the French Army where he rose to be inspector of supply. Control of supplies naturally requires a lot of arithmetic and inspired the invention. On leaving the army he co-founded an insurance company, but left over disagreements about his many new ideas. In 1829 he founded the fire insurance company "Le Soleil" and grew that by mergers and acquisitions. He then founded "L'Aigle Incendie", together they formed the largest insurance business in France.
The Arithmometer uses Leibniz cylinders to perform arithmetic and has vertical sliding knobs to enter the numbers and a horizontal sliding carriage containing the accumulator that is moved to perform multiplication and division. A few were built around 1822 but then production halted, possibly Colmar became aware of Babbage's work. Development restarted in 1848 and by 1872 some 1,000 had been made. Production continued until 1915 but by then the design was overshadowed by the Comptometer and Burroughs' adding machines. The Arithmometer was, nevertheless, the first calculating device to find any sort of popular market. A bank, insurance or government office might have one.
Charles Babbage
Charles Babbage is often noted as the inventor of computers a hundred years before their time.
What Babbage was trying to achieve needs a bit of context.
Babbage's Difference Engine was originally conceived as a way to produce tables of logs and trigonometric functions. These tables were essential, not only for mathematicians and astronomers but also for navigation at sea - ships might be wrecked because the tables were wrong. Producing these tables was a major exercise. In 1784 the French government decided new tables were needed and six distinguished mathematicians devised the methods, seven or so human computers served as foremen and another 70 or so performed the bulk of the calculation. The result was two handwritten volumes held in a library.
Published copies of log tables had to run the further gauntlet of typesetting - it would be a nightmare print job. Stan Augarten says "In 1835, an informal survey of one scientist's library turned up 140 books of tables, and an examination of only some of the figures in forty of the books uncovered 3,700 inaccuracies. Even the British Nautical Almanac - the navigator's bible, was sprinkled with mistakes, and more than one ship was said to have run aground or been lost at sea as a result of the miscalculations. Maddeningly enough, some of the slip-ups were even deliberate, inserted by publishers as traps for would-be plagiarisers."
Babbage originally intended to build a machine capable of calculating and printing log and trigonometric tables by the method of differences. Any consistent numerical progression can be generated by a series of differences, although the calculation may take several stages or "orders". Babbage first realized that such a thing was possible whilst he was University student around 1812. He built a small part of the engine between 1822 and 1832, but the the manufacture of such things stretched the metalwork capabilities of the time. Babbage's chief engineer grew greedy and the project came to a halt. The government had been part- financing efforts and finally cancelled in 1842. The project came to a halt, the ideas did not.
Sometime around 1836 Babbage thought of an Analytical engine, more complicated but more capable than the Difference Engine. It is difficult to sum up the design of the Analytical Engine - Babbage kept redesigning it - but the general idea was a difference engine built in a circle and a memory device alongside with punch cards for data and program entry. The prescient feature was that it included conditional branching; results such as zero in the data could be sensed and the program change its action accordingly. Conditional branching makes the Analytical engine into a mechanical computer. Ada Lovelace suggested how some problems might be worked out using the machine so she is often regarded as the “patron saint” of programming. A part of the Analytical Engine was built by Henry Babbage in 1910.
Georg and Edvard Scheutz
Pehr Georg and Edvard Scheutz read about Babbage's machine and were inspired. Rather than asking for information they built a small model and then came up with their own design and built a working Difference Engine in 1853. They brought a version to London in 1854. Babbage was thrilled with the demonstration. Several of the Scheutz's machines were made. In the UK the Register General had one built and used it to calculate lifetime, annuity and premium tables for the insurance industry amongst other things.
Historical Impact
Babbage's Difference and Analytical Engines were quite famous at the time, written about in magazines and discussed in scientific circles. Babbage undoubtedly advanced the engineering and toolmaking capabilities of his time but they just weren't good enough to make tens of thousands of components to the standards needed.
The Difference engine wasn't completed in Babbage's lifetime. The Science Museum in London constructed a working machine in 1991 to celebrate Babbage's 200th birthday and computer entrepreneur Nathan Myhrvold had the output mechanism and a second complete engine built.
If Babbage's Analytical engine had been built the world might be different. Marine navigation would have been improved, fractionally and sooner. But the big uses for calculation in electricity distribution, electronics and aircraft design were unknown at the time. Many early engineers like the Stephensons and scientists like Faraday worked more from intuition than mathematics.
As Babbage was experimenting so were Telegraph pioneers like Davy, Cooke, Wheatstone and Morse
If Babbage and Wheatstone had realised that their projects were compatible, then at the very least we would have achieved a telegraph switching center like ‘Plan 55-A’ a century early. If Babbage had used relays instead of push-rods computers would have been slow but possible in 1840. Perhaps if they had all realised that information could be described according to Peter Mark Roget's newly devised thesaurus then we might have had computers and indeed the ‘semantic web’ at least a hundred years earlier. Speculation on these amusing themes is ‘steampunk’.
Whether they knew of Babbage's machines or not for most of the next hundred years calculator inventors stuck to improving adding machines. The inventors of proto- computer devices mostly knew nothing of Babbage's achievements. And it seems we can prove that the ideas were simply before their time. The Scheutz machines don't seem to have had much influence either, even though there were practical working examples.
Progress became a matter of improving the Arithmometer, which might now seem a backward step. In 1856 Thomas de Colmar estimated that he had spent 300,000 francs of his own money perfecting the invention. A 12 digit Arithmometer sold for 300 francs
Hill's Arithmometer
Thomas Hill invented an improved Arithmometer in 1857. Hill's device uses number wheels as an accumulator, mounting them on a shaft at the back of the machine. In front of each wheel is a column of keys numbered 1 to 9 (or 1 to 0) mounted over a lever with the 1 at the top and higher digits below. When a key is depressed the lever underneath moves. When a 1 is pressed the lever travels a short way and a ratchet and pawl pulls the number wheel one digit. When a 6 is pressed the lever travels sufficiently for the ratchet and pawl to pull the number wheel 6 positions. A carry mechanism is arranged at the back.
The arithmetic principle is similar to a Pascaline.
An advantage is that the number wheels are now worked by buttons like those that would later be used on typewriters. Jabbing a key is a quicker action than turning a knob or working a slider.
Hill solved the problem of subtraction by having smaller numbers printed on the wheels and the keys in reverse order so that subtraction and division are performed by 9s complement arithmetic.
Comptometer
Comptometers were the definitive office calculating machine for most of the 20th century. Despite electronic calculators becoming cheaper and easier to use a few comptometers remained in use into the 1980s because a skilled operator could use the comptometers numeric column keyboard faster than an ordinary calculator. The special design has disappeared but numerical data entry would presumably be quicker if it made a comeback.
Dorr E. Felt invented the Comptometer in 1886. It is basically an arithmometer but the keyboard is an improved version of that proposed by Hill. Each arithmetic wheel is set by numbered keys. Drive from the keys to the number wheels is improved by using a segment-rack to drive the wheels accurately and reliably. Felt made a model of his machine using a macaroni box, bits of wood, staples and rubber bands. On the basis of this Robert Tarrant who owned a machine-shop in Chicago paid him to develop the idea.
Felt and Tarrant formed a company to make comptometers, and a couple of years later they developed a printing machine they called the Comptograph.
The name Comptometer became generic for this kind of machine. The idea remained current for nearly a century.
Comptometer operators were still paid more than ordinary office clerks into the 1970's. Even in their fully developed form Comptometers weren't always very easy to use and could need a bit of arithmetic agility to work them. For instance it was often quicker to type "4" and then "5" than to stretch the hand down to press "9". Speedy but accurate operation was a valuable skill.
Burroughs
William Seward Burroughs developed his idea of an arithmometer also in 1886. Burroughs was the son of a mechanic and he worked with machines in his childhood, but father was determined his son should have a gentleman's occupation and so he became a bank clerk. He disliked the work, and long hours spent on bank ledgers apparently wrecked his health but they also gave him an interest in arithmetic machines. The arithmometer machines the bank had were not very reliable. At the end of seven years a doctor advised him to move to somewhere warmer, so he moved to St. Louis, Missouri and took a job in a machine shop. There he designed and perfected his adding machine
Burroughs formed the American Arithmometer Company in 1886. His calculating machines use a more complicated but generally similar mechanism to that of the comptometer to not only add but print a result. Burroughs machines usually had one extra digit compared with a Comptometer at the same price point. Within ten years the company had grown to 68 people. However Burroughs himself died in 1898 and the company moved to Detroit in 1904 where it built a 70,000 square foot plant on what had been a cornfield.
Burroughs Adding Machine Company then had 1,200 people and 7,804 machines were sold in 1905. By 1935 the product line included 450 standard models of manual and electric calculators, bookkeeping machines and typewriters.
After World War II Burroughs began to develop electronic products, notably the Sensimatic accounting machine. This led to a very successful development of computers. In 1986 Burroughs merged with Sperry UNIVAC to become UNISYS but they still trace their heritage back to the American Arithmometer Company.
In 2010 UNISYS sold it's Payment Systems Division to a private investment firm so the name has re-emerged as Burroughs Payment Systems. William Burroughs didn't entirely enjoy being a bank clerk but his familiarity with the procedures made his machine popular in banking - a position Burroughs has continued to hold. That history still echoes through the company a century and more later.
Victor
Victor started in 1918. The story is that Carl Buehler gave a salesman $100 but due to a misunderstanding bought 10 shares in the company instead of the expected adding machine. To protect his investment he became a director and then company president. The first adding machine was produced in 1919. The company was successful and by the early 1950s had sold more than a million machines. In 1953 Victor bought the McCaskey Register Company and entered the cash register business. and in 1961 they merged with Comptometer Corporation.
By the early 1970s the company was making electronic calculators like the 1800 series and they apparently sold dot-matrix printers as an OEM product. The company was acquired by Kidde Inc in 1977. Sirius Systems (financed by Walter Kidde ) bought Victor in the early 1980s but although Sirius computers were initially very successful and sold in the UK as the ACT Sirius 1 they went into Chapter 11 protection in 1984. After several years of ownership and product changes they have returned to the office equipment market with some rather entertaining and useful calculators
Burroughs successfully transformed itself from an adding machine to a computer company, as did NCR and IBM. Comptometer and Victor stumbled.
Odhner
The Odhner Arithmometer was invented in 1873 by Willgodt Theophil Odhner, a Swede living in Russia. He worked in Ludvig Nobel's mechanical factory and then joined a paper mill and printing house. Whilst continuing to work in the paper mill he started his own engineering company making printing presses, cigarette making machines and scientific instruments.
Odhner had been asked to repair one of Thomas de Colmar's Arithmometers in 1871. He felt that the heavy Leibniz cylinders were a problem and came up with the pinwheel idea. The pinwheel is a cog with retractable teeth, the number of teeth exposed is controlled by a lever alongside. Odhner developed the first version in 1873 but then spent some time perfecting it. He made 14 machines for Ludvig Nobel in 1877 and developed an improved machine that went into production in 1890. In 1891 he opened a factory in Germany but found it impractical to have a factory so far away and sold it to Grimme, Natalis & Co. who started production under the name Brunsviga. Odhner died in 1905 and his factory closed in 1918 after the Russian revolution. The family went back to Sweden and restarted production under their own name. The Russian government moved production to Moscow and produced machines under the name Felix Arithmometer.
The pinwheel idea made a very neat little unit. The idea was adopted by many other manufacturers and became one of the main mechanisms used in mechanical calculation. Imitators included Triumphator, Thales, Walther, Facit and Toshiba
Many Producers
A great many companies began to produce adding machines and comptometers in the early 20th century. Very few accounting offices were without one, whilst engineering offices were generally distinguished by use of slide-rules.
The Millionaire was the first commercially successful calculator that could multiply numbers together directly without repeat operations.
Dalton made the first ten key adding machine. In the USA Friden, Marchant and Monroe were the main makers of rotary calculators with carriages. Remington Rand, Mercedes-Euklid and Olivetti were other noted producers.
IBM
IBM were primarily makers of tabulating machines. Tabulators start as quite distinct from calculators but the idea ultimately merges. They could be seen as calculators with automatic input and output; close relatives of the computers IBM builds now.
Herman Hollerith began making tabulators in the 1880s constructing punch-card machines to process information for the US Census.
Prior to this census data had been processed manually. Someone sat and counted the marks on returned ledgers. Generating statistics from this was slow, so that in the 1880s it could be foreseen that the 1890 census would overrun into the new century and the next census. Since US law used the census to apportion congressional districts and federal spending the problem was urgent. Hollerith was persuaded to devise counting machines.
Hollerith had seen ticket collectors on trains use a punch to record a few details of passengers. He realised that a punched card could be used for census data and that the punched holes could be counted automatically with electrically operated clock dials. Hollerith described the system in ‘ An Electric Tabulating System ’ Submitted to Colombia University as his doctoral thesis in 1889. He also developed a patent portfolio. With a contract from the Census Office he built machines which tabulated the 1890 census in a year. On the strength of this Hollerith started a business supplying machines and cards to censuses around the world and to insurance companies.
A dominant customer can be difficult. Hollerith and the director did not get on and they took the contract elsewhere. Hollerith developed some new markets, control of railway stockyards for instance.
In 1911 Hollerith, in failing health, sold the Tabulating Machine Company to Charles Ranlett Flint for $2.3 million (he got $1.2 million). Flint got together with owners of some other businesses in fields that seemed related and proposed a merger. Computing Scale Corporation, International Time Recording and Bundy Manufacturing Company who were also in time recording came together to form the Computing Tabulating Recording Corporation. The new company made everything from tabulators and clocking-in machines to automatic meat slicers. Flint was something of an enthusiast for industrial combinations; he also formed US Rubber, American Wool and American Chicle. Something of a character his activities earned him the name "Father of Trusts".
AT&T president Theodore N. Vail had a similar faith that Trusts were the right way forward. See the Kingsbury Commitment . IBM was to come to blows with AT&T over monopolies in the 1960s.
Flint was approached by Thomas J. Watson, recently fired by NCR owner John Henry Patterson after courts had found their business practices illegal. Watson set about building morale in a sales-force of 400 people and improving the product line. Watson focused on tabulating, whether through prescience or because this was most similar to the cash register field he was familiar with from NCR.
Watson insisted on dark-suited well-groomed salesmen and company pride and loyalty in every worker. There were generous sales incentives for salesmen, who became highly professional and knowledgeable about what the company could deliver. Customer focus was important, equipment was rented because that meant the salesman had to stay in touch and satisfy customer needs. Watson also focused on providing large-scale custom built tabulator solutions and left the market for small office products like calculators to others.
Watson also introduced his favourite slogan “ THINK ”. He sponsored employee sports teams, outings and a company band - as well as company songs. All managers were to have an Open Door Policy encouraging any employee to approach him with any complaint. In 1924 the company was expanding in Canada and Europe and Watson renamed it International Business Machines - IBM.
IBM pioneered the idea of a corporate culture. Its newspaper ‘ Business Machines ’ covered all the company's activities. The Quarter Century Club honoured its established salesmen who had worked for the founding companies. A One Hundred Percent Club rewarded salesmen who met their quotas and Suggestion Plan gave cash rewards for new ideas.
Revenues doubled within four years of Watson taking over.
Rather daringly Watson stared the 1930s recession in the face and kept the factories working, not only keeping his people in jobs but creating a big inventory of unused equipment.
One effect of the recession was that the US introduced a Social Security Act in 1935. Recording details for 26 million workers was an immense task. Thanks to the surplus equipment IBM won the contract, and was able to fulfill it laying the ground for further government work.
IBM introduced a:
- printing tabulator
- Type IV, a tabulator that could subtract
- 80-column punched card
- IBM 600 Multiplying Punch
- Alphabetic Tabulator Model B
- Columbia Difference Tabulator
- Purchased the Electromatic Typewriter Co.
The multiplying punch is an interesting machine. A multiplying punch is useful for things like payroll and stock valuation where a count needs to be multiplied by a rate to get the answer - so hours worked time payrate or stock number times purchase price. The early multiplying punches are electromechanical and have multiplication tables as a mechanical memory. In 1948 IBM introduced a model 604 valve base multiplying punch - it is only just short of being a computer.
Big Calculators
In the Second World War a great deal of the precision engineering equipment in Britain, the US, Canada Australia and Russia (on one side) and of Germany, Japan, Italy and the conquered countries on the other was put to military use.
In the 1930s and '40s the big calculating machines that gave rise to the mainframe computer industry were built. They were not all wartime secret projects although ballistics, cyphers and the atom bomb were all concerns.
Relay Calculators
George R. Stibitz realised that relays could be used as calculating devices in 1937. He made a relay based binary device he called the Model-K because it was assembled on a kitchen table. On the basis of this his bosses at AT&T Bell Labs authorised a research program.
The Complex Number Calculator subsequently known as the Model 1 Relay Calculator was completed in 1940. Built by a phone company it used about 450 phone systems relays and some crossbar switches. It worked in binary (a number scheme called "plus3 BCD" which uses fewer relays). Input and output were from a teletype, so a demonstration could be given at a mathematical conference in New Hampshire in 1940 to the machine in New York.
Several more relay calculators were built; the Model II was for aiming anti-aircraft guns, Model III and IV were both Ballistic Computers and the Model V was a general purpose computer using relays.
Bell labs also built the SIGSALY digital speech scrambler, which remained secret for some time. It was probably influential in moving their research towards binary design.
Harry Nyquist, Ralph Hartley, Richard Hamming and Claude Shannon were all based in Bell labs. This was an age when discoveries in fundamental physics, mathematics and information theory were seen as directly relevant to the interests of a phone company.
Very notably from the computing point of view Bell Labs developed the transistor. Transistors are solid-state switches capable of very fast action. Great numbers of transistors can be made at once in the form of ICs and miniaturized -(today's transistors are smaller than wavelengths of light).
Curiously enough, Bell labs were not the first to make transistor computers, but they subsequently made many contributions to computing such as the invention of UNIX and the C programming language around 1970 and the development of twisted-pair network cable in the 1980s.
Many of the fundamental innovations of the 20th Century originated in Bell Labs but their contribution to computers was hampered in some ways by their monopoly of telecoms. Political opposition to their gaining a similar position in computing was maintained by IBM. In settlement of an antitrust case in 1958 AT&T was forbidden from entering the computer market and was even required to license it's non-telephone technology to anyone who asked.
In some ways the Bell Labs story parallels that of Xerox PARC, a corporate research centre that created some of the most significant innovations of the 20th Century, but whose parent companies failed to capitalise on them. However this is drifting away from the calculator story.
Harvard Mark I
Howard H. Aiken devised the Harvard Mark 1 computer also known as the IBM Automatic Sequence Controlled Calculator (ASCC). It was mainly built for US Navy work
The machine was built at IBM and tested in 1943 then installed at Harvard in 1944. It was literally a calculator, or more properly 72 adding machines each with 23 significant digits. There were also 3,500 relays, 2,225 counters and 1,464 ten pole switches arranged on a panel 51 feet long and 8 feet high. Altogether it contained 765,000 components. The calculating units were driven by a 4 KW electric motor. It read instructions from a 24 channel punched tape. What it lacked was conditional branching, so it was literally what it says, a sequence controlled calculator, not a computer.
IBM sponsored the machine from a combination of interest and good publicity. To make the banks of relays and counters look impressive Thomas J. Watson had a futuristic case designed for it. However Aiken put out a press release listing himself a sole inventor and made no mention of IBM. Aiken went on to build three more Harvard machines. IBM built the SSEC instead.
Parts of the Mark 1 survive in the Harvard Science Center. It really does look like the future. But its essentially mechanical operation meant it was slow and it was designed to perform arithmetic sequences, there isn't the conditional branching ability based on a program held in memory.
The Mark 1 shared with Stibitz Model 1 the distinction of being very reliable. Electromechanical relays operate slowly, however, so whilst they can be used in a calculator they aren't a good basis for computer which by nature gets its power by doing rather dumb actions very quickly.
The ABC
John Vincent Atanasoff and his graduate student Clifford Berry built what is now called the Atanasoff–Berry Computer or ‘ ABC ’ starting in 1939 with a prototype 16 bit adder.
Atanasoff was inspired by the drudgery of using a mechanical Monroe calculator whilst working on his thesis on the Dielectric Constant of Helium. He then invented an analog calculator for analysing surface geometry and this showed the limits of analog methods.
The ABC had a special purpose processor for solving simultaneous linear equations but used binary mathematics and regenerative capacitor memory. It used vacuum tubes for arithmetic so whilst it was still essentially a calculator it could be fast.
Atanasoff's ABC proved highly influential because it provided some of the inspiration for ENIAC. Atanasoff himself was called to Washington DC to do physics research for the US Navy.
ENIAC
ENIAC was built by a team around John Mauchly and J. Presper Eckert at the Moore School of Electrical Engineering, University of Pennsylvania. Project PX as it was known was intended to calculate artillery firing tables for the US Army's Ballistic Research Laboratory. Work began in 1943.
The Moore School had two Bush Differential Analysers, (analog computers) which were useful for calculating shell trajectories, so the US Ballistic Research Lab had an outpost there under Lt Herman H. Goldstine. The idea of an electronic computer had already been suggested, and was taken up.
ENIAC was big. It had 20 accumulators which could each hold a ten digit decimal number. The accumulators were essentially ring counters made from ten dual triodes acting as flip-flops with a carry generated every time a counter wrapped around. Together with the gate logic each digit had 36 valves. The design is basically a fully electronic realisation of a mechanical adding machine, the difference being that use of electronic valves makes it 1,000 times faster.
Adder / accumulators could be wired together to give double precision arithmetic. ENIAC also had a Multiplier unit that could perform 385 operations per second and a Divider/square-root unit capable of 40 divides or 3 square-root operations per second. There were also constant-transmitters and three function tables. The master programming unit controlled loop-sequencing and crucially was capable of branching making the machine Turing Complete. However there was a design-freeze imposed to get the thing completed so there was no program memory, rather the components were wired together to do a job. Improvements such as mercury delay lines were added after 1948.
Input was from IBM card readers and output to a card punch. Printed output or further processing could be achieved with IBM tabulation machines.
A team of six specially trained programmers supervised most of the machines programming. Most computing tasks were done by women in those days so the programmers were women.
The machine contained 17,468 vaccuum tubes, 7,200 crystal diodes, 1,500 relays, 70,000 resistors, 10,000 cpacitors. It was altogether something like 100 feet long, 8 feet high and 3 feet deep and weighed about 27 tonnes. Power consumption was 150 KW. Since most of the failures happened during warm-up and cool-down periods some measure of reliability was achieved by rarely turning it off.
John von Neumann, a mathematician working on the hydrogen bomb at Los Alamos, became aware of the project in 1944. Los Alamos made heavy use of the machine and von Neumann later said the work would have been impossible without it. He also wrote a set of notes “ First Draft of a Report on the EDVAC ” which ENIAC administrator Herman Goldstine distributed to several institutions; it spurred great excitement.
The project was naturally a classified secret, but on Feb 14th 1946 the US government declassified ENIAC. Dianne Martin's paper “ ENIAC: The Press Conference That Shook the World ” analyses the way the machine was achieved huge press fame as an "Electronic Brain". ENIAC became the foundation of UNIVAC, Sperry UNIVAC and now UNISYS.
The Moore School Lectures ‘Theory and Techniques for Design of Electronic Digital Computers’ were held from 8th July to 30th August 1946 to disseminate ideas developed for EDVAC. Discussion of ENIAC was discouraged because the successor would be a very different machine, but the student petitioned to see a demonstration.
ENIAC co-inventors Eckert and Mauchly had departed the Moore School amidst a patent rights dispute and founded Electronic Control Company but were attracted back by lecture fees. 28 students were invited to attend and the lecturers attended one-anothers talks.
Zuse Machines
Konrad Zuse worked as a design engineer at the Henschel aircraft factory in Berlin in 1936. He found the calculations involved boring and thought they could be performed by machine.
Working on his own in his parents flat Zuse built the Z1, a binary floating-point calculator mainly using mechanical parts. He applied for patents in 1937 and finished the machine in 1938. Both the machine and the blueprints were destroyed during World War II.
In 1939 Zuse began constructing the Z2, using telephone relays. On the basis of the Z2 he started a company Zuse Apparatebau to make the machines. The Z3 was a binary 22 bit floating point calculator and allowed programs with loops but without conditional branching. Zuse was working on the Z4 when the company and the computers were destroyed in a bombing raid. The partially completed Z4 had been moved to safety, but work could not resume until 1949.
Zuse knew nothing of developments in America and Wartime conditions would have made contact impossible. The Z4, finished in 1950 was delivered to ETH Zurich in September 1950 making it one of the first computers ever sold. IBM took an option on Zuse's patents.
Zuse did not know of Stibitz or Atanasoff, and could not have known of ENIAC or Collosus because those projects were top secret. The other computer pioneers didn't need to know much of one-another to see the possibilities. (Likewise with the invention of the telegraph, electric motor, light bulb or television). In Zuse's case we have a lone inventor working at the same time as others on a similar project - an argument for the right time bringing the man?
Colossus
Colossus machines were built for code breaking at Bletchley Park by Tommy Flowers at the Post Office Research Centre in Dollis Hill. Altogether 11 were built and operational at the end of World War. They were high speed statistical and pattern matching machines, not full computers, even though Alan Turing worked at Bletchley. All but two of the machines and most blueprints were destroyed on the direct orders of Churchill so that the techniques they embodied would not become widely kown. Two were moved to GCHQ and dismantled in 1960. Colossus nevertheless created a lot of skilled and knowledgeable people who knew what was possible.
Colossus was reconstructed recently from 1994 onwards using the few remaining diagrams, notes and the memory of people who had worked on it. Since it had been largely built from standard Post Office parts the reconstruction is thought to be accurate.
Influence
Computing is still not ubiquitous in today's world: 2 billion people use the Internet and something like 4 billion have a mobile phone, but that is less than half of the population. Even in the
Computers were not entirely creations of war. People like Alan Turing had the idea. Stibitz and Zuse created relay based machines and Atanasoff built the ABC more or less out of interest, to work through a problem. What war did was allow fast, expensive machines like ENIAC and Colossus to be constructed, they showed that part of a machines power is sheer speed.
There is no way to un-invent things. Once the atomic bomb and then the H-bomb had been demonstrated governments faced the stark choice of competing or capitulating in the Cold-war that was breaking out.
The USSR began it's own secret computer project, the Automatic Digital Computer M-1. The M-1 was working by 1951. The first nuclear weapon was tested in 1949.
The Manchester Small-Scale Experimental Machine (SSEM), nicknamed Baby was intended as a testbed for Williams tube memories devised by Frederic C. Williams. It was designed by Williams and Tom Kilburn to be the simplest possible machine, with 32 words of 32 bit memory and the only instructions available in hardware being subtraction and negation. As it happens it was the first machine to actually run with a stored program in June 1948. The Manchester Mark 1 was built a year later. Then the Ferranti Mark 1 commercial computers were built on the strength of those.
The SSEM contained 250 Pentodes, 300 diodes, and consumed 3.5 KW of power. It had one Williams tube giving 32 words of 32 bits for RAM, a second with a 32 bit accumulator and a third acting as an instruction register and address register. A fourth CRT acted as a display
The Museum of Science and Industry in Manchester has a working replica of the SSEM
Digital60.org - 60 Years of the Modern Computer. Notes on Moore School Lectures 34 to 48.Australia's CSIRO Sydney Radiophysics Laboratory built the CSIR Mk 1 which ran its first program in 1949. At the time of building it was the fifth computer worldwide. It was decomissioned in 1955 but recommisioned in Melbourne as CSIRAC where it was used for academic projects until 1964. CSIRAC is one of about four first generation computers to survive intact, it is in the Melbourne Museum.
EDSAC
The Electronic Delay Storage Automatic Calculator (EDSAC) was built by Maurice Wilkes and a team at Cambridge Mathematic Laboratory. Wilkes had worked on radar at TRE during the War. Soon after his appointment he was given sight of Von Neumann's report on the EDVAC and realized that Cambridge must have a machine to remain at the forefront of research. He travelled to the US to enroll in the Moore School Lectures. He visited other US computers and became very familiar with ENIAC. On his return he decided his task was not to invent a better computer, but to make a computer available to the University, so his approach was relentlessly practical. The result was the first practical stored program computer to be completed, operating in May 1949. (Baby worked, but wasn't very practical)
Some big businesses began to experiment with computers. J. Lyons & Co, best known for their chain of tea-shops they were so taken with the idea of computers that they helped pay for EDSAC and then built their own. ‘ LEO ’ was remarkably succesful and Lyons set up a company to make more of them. LEO seems to have been the worlds first commercial computer, it ran its first business application in 1951. That a tea-shop chain built the world's first commercial computer is extraordinary.
These big machines were influential and futuristic but LEO apart they did nothing to change the average office environment. Most businesses remained wedded to the typewriter and adding machine in small operations, with counting or even multiplying tabulators to deal with large volumes of information. Most actual processing was done by a growing army of clerical workers.
Thomas Watson, Chairman of IBM 1943. I think there is a world market for maybe five computersBy 1960 there were thought to be about 5,000 in use worldwide. Big operations like banks opened computer centres, Barclays was the first in the UK installing an EMIDEC 1100 at their specially built London computer centre in 1961. Barclays was partly motivated by their innability to get enough good quality staff in London at that time.
Sumlock ANITA
The world's first desktop electronic calculator was the Sumlock ANITA, launched in 1961
Bell Punch Co made the ticket punches used by tram and bus conductors, the company was established int 1878; it aquired the UK rights to some American patents. A ticket punch made it possible for a bus or tram conductor to determine if a passenger was overstaying what they paid as a fare. The company then developed small rotary printers, used by conductors to issue tickets at any stage in the journey. In 1891 Bell-Punch began supplying ticket machines for the London General Omnibus Company and went on supplying equipment to London Transport. Bell-Punch was a fairly succesful operation in early "Information Technology"; it had its own paper mill, Isaac Warwick and Co., Limited, of Wraysbury. The importance of ticketing is often overlooked in discussions of IT history.
Bell-Punch diversified over time; they developed ticketing machines for Cinemas and theatres; horse race totalisators and taxi-meters. Adding machines were developed from an idea it bought from Petters of Yeovil (which turned into Westland Helictopters).
Bell Punch were making key driven Comptometers under the names ‘Plus’ and ‘Sumlock’ in the 1930s and through to the the 1960s. The ‘Plus’ machines had abbreviated keyboards sometimes needing two sets of keypresses to enter a number. The Sumlock machine had full columns of keys and more columns of figures. Both machines could be built for British currency of pounds, shillings and pence.
Bell punch even had it's own Sumlock School teaching the techniques and awarding a Sumlock Diploma badge. With products from paper to training the company was very much the value-add operation people aim for today.
It may seem odd that the makers of bus conductor's ticket machines made the world's first electronic calculators. Innovation doesn't always come from expected places. Sumlock had already developed a range of products including navigation instruments for the RAF. A community of design techniques was being used to make instruments long before microprocessors came along.
Bell Punch already had the idea that electronics would be important and were developing the SAM (Semi-Automatic Multiplier). This project was a complex electro-mechanical device, led by Mr C.F.Webb, the companies chief designer, who was nearing retirement.
Norman Kitz had worked on the Pilot ACE computer project and by chance met some of directors of Control Systems Ltd who owned Bell Punch. Whilst looking at the mechanical calculators exhibited in the British Museum he had realised that future calculators would be electronic.
Bell Punch set up an electronic development department in 1957, made Kitz project leader and he recruited John Lloyd and Hugh Mansford. The project name was an acronym for “ A New Inspiration To Accounting ” ANITA.
ANITA machines had to be priced very differently from computers - the cheapest computer cost £50,000 and the target price for the ANITA range was around £350.
To make such a device they needed a cheap, reliable technology. Initially they chose Dekatron counter tubes made by Ericson Telephones at Beeston, Nottingham as the counting element and triode valves with solid state diodes to make AND and OR gates. A prototype was ready and working in about a year. The problem was that if the Dekatrons were left in one position for a long time metal would sputter from the lit cathode and damage it's counting abilities. To overcome this they needed a redesign.
All but one of the Dekatrons were replaced with a ring counter made from cold-cathode tube triodes (one was retained because it acted as a keyboard scanner clock, wasn't static and didn't have the problem. Cold cathode tubes don't have heaters (hence the name) and they have a long life compared to ordinary vacuum tubes.
In 1961 they launched the first all-electronic desktop calculators the "ANITA" Mark VII and Mark VIII
The machines were launched in 1961 and by 1964 were selling 10,000 units per year, not many by the standards of today's mass produced computers but a great number for a business product at the time.Many companies started to produce electronic calculators. Initially their main selling points were quietness, reliability and some had the ability to perform the trancendental functions wanted in scientific calculation without resort to log tables or slide rule.
Some operators thought electronic comptometers too quiet and they lacked the tactile feedback of the mechanical machines. But of course electronic devices can be mass produced more cheaply than mechanical things and the price of integrated circuit functionality would fall, whilst that of labour intensive mechanical components would rise.
Within a couple of years Sumlock had competition from Friden and Sharp using transistors. Canon's Canola 130 a ten-key electronic calculator was introduced in 1964. Wang and Hewlett Packard were making scientific calculators. Initially transistors gave no great advantage because they weren't much smaller or more efficient than the cold cathode devices and decatrons used in the ANITA. Subsequent ANITA models used transistors, then integrated circuits. In 1972 ‘New Scientist’ reported "Considering non-programmable calculators, Sumlock Comptometer was easily the largest UK supplier in 1971, with sales estimated at about £4.5 million." (a 1972 pound was worth about 10 times as much).
Sumlock ANITA machines sold as something between a calculator and a computer; it spanned both markets. It had something in common with minimalist computers like the BCL SADIE.
Sumlock Anita were wrong footed by a single big misjudgement.
US defence contractor and chipmaker Rockwell was keen to diversify. Rockwell was a manufacturing conglomerate with a base making bearings and truck axles but had steered this into aerospace making the P-51 Mustang and B1-bomber. Rockwell built the Apollo spacecraft. In the early 70's recession was reducing NASA's spend in the wake of the moonshots and B-1 production was cancelled. Rockwell needed to diversify and bought Collins Radio, the main supplier of avionics radios and Admiral Radio and TV a consumer TV company.
Rockwell had developed their own desktop calculator based on MOSFET chips for their own engineers in 1967 and set up Rockwell Semiconductor to make them.
Rockwell offered to put ANITAs circuits on a chip and the two companies developed a prototype pocket calculator called the ANITA 800. When Norman Kitz showed the device to his M.D. he saw no market for it. Kitz was annoyed and put the prototype in his drawer.
A few months later Clive Sinclair's ‘ Sinclair Radionics ’ brought out the ‘Executive’ pocket calculator, selling for £79.95 each and making £1.8 million profit in its first year. This was a surprise, Sinclair were known for HiFi kits, calculators were a new line. For several years Sinclair concentrated on a succession of lower priced and more functional calculators.
Sinclair in turn was taken down by the ‘ Black Watch ’ fiasco in 1975-6 which sold well but didn't work properly. This was compounded by Microvision, an early hand-held flat screen television which didn't sell very well.
After a couple of name changes Sinclair re-emerged as Sinclair Research Ltd which produced a series of succesful British microcomputers in the early 1980s. Sinclair's contract manufacturer for the ZX-Spectum was Timex, formerly a watch factory. By the mid 1970s calculators, clocks, HiFi and TV were beginning their slow convergence into the computer industry.
Other competitors for the ANITA 800 were the Busicom LE-120A Handy - probably the very first handheld and HP the HP-35 scientific calculator which sold 100,000 in its first year.
In 1973 Rockwell, the main supplier of ICs took the Sumlock division of Bell-Punch over. They had also aquired a Californian company called Unicom. By this time Japanese companies like Sharp, Canon and Casio as well as Texas Instruments and National Semiconductor were making calculators. Sumlock were unlucky several times over. Recovering and readying the ANITA 800 into production Rockwell closed Sumlock in 1976 as the price of calculators plummeted to the £5-£10 mark. they sold the 33 nationwide service depots off to management buyouts. Most of the buyouts turned to selling computers, typically the Commodore PET.Bell Punch was sold
Busicom - Calculators to Computers
Busicom started as Nippon Calculating Machine Corp in 1945, originally making the popular Odhner style of pinwheel machine. It changed its name to Business Computer Corporation or Busicom in 1967.
Like most calculator makers Busicom originally used multiple chips in their machines, but the number needed fell, particularly if they used application specific circuits. Busicom engineers came up with a design that used just 12 ICs to make a printing calculator in 1969 and asked Intel to make them. Intel had been founded a year earlier in 1968 with the primary aim of making memory chips, then quite a new idea. At the time Intel had difficulty persuading mainframe manufacturers who were the main users of memory to switch away from core stores, so they were keeping the production lines running with contract work.
Intel assigned the work to Ted Hoff and Frederico Faggin, together with with Masatoshi Shima of Busicom. Hoff studied the design and realized that it could be done more efficiently with just four ICs - a 4 bit microprocessor, a ROM, a RAM and an IO device. Busicom agreed to this and used it to make the Busicom 141-PF. It was also sold as the NCR 18-36 and the Unicom 141P.
Busicom had exclusive rights to the chip but with the price of calculators continuing to fall they asked for a renegotiation of the contract. Busicom gave up exclusive rights to the chip for a price reduction. Intel launched the MSC-4 microprocessor chipset a few months later.
Busicom went on innovating, using Mostek MK6010 to make less expensive machines. They were reportedly the first company to use an LED display.
Busicom's calculator business and a global recession in the early 1970s closed Busicom in 1974. The business name was bought by Broughtons of Bristol, a business machine company.
Microprocessor
Intel's intended market, MOS memory devices, began to really take off in the early 1970s as the performance overtook that of core stores. So did a new market for products such as green-screen terminals and small printers based on microprocessors. Other companies fairly quickly began to produce microprocessors. Texas instruments may have made a single chip computer at much the same time as Intel but seems to have been less sure of its market potential. Motorola (6800)and National Semiconductor (IMP-8) both developed products quickly.
Most recent equipment is based on microprocessors, but whilst they are now seen as a breakthrough the initial reaction to them was warm but they weren't widely seen as revolutionary. It was amazing to cram a computer into something so small, but early microprocessors weren't much of a computer compared to an IBM 370. And of course the truly big thing about a computer was not its hardware but the information it held. There were no networks, Ethernet had just been conceived at Xerox PARC and the ARPANET (forerunner of the Internet) first spanned the US in 1970 but had fewer than 20 nodes. The newly founded Compu-Serv Network Inc was selling timeshare on DEC PDP-10 computers primarily used by Golden United Life Insurance. So a microprocessor on its own wouldn't have much data to work with.
The main aim for microprocessors was primarily devices that had previously used mechanical or dedicated logic circuits. Taxi meters, weighing machines, cash registers coin counters, ticketing machines and every kind of measurement instrument were all potential applications. So were all kinds of office equipment: phone switches, typewriters and copiers. Early microprocessors made copiers and faxes much more affordable byt replacing dozens of special-purpose circuits.
Peripherals for computer systems was a big market. In the early 1970s getting a job run on a computer was typically done in batch mode, the user submitted a set of buff punch-cards holding the data, green cards holding the program and waited an hour or two until the result was delivered on green-lined printout paper. Since most programs didn't work first time writing a program could be a rather frustrating experience. Moving from batch to timeshare mode to make the computer interactive meant a more complicated CPU with drum or disk store and use of terminals. Terminals themselves needed enough memory to hold the pattern on the screen and counters for the cursor position as well as for the transmit and recive buffers, so they were quite complicated. Using a microprocessor typically cut what had been four boards of logic in a terminal to just one, making them more affordable.
A 4 bit microprocessor is a weak device. The Moore's law progress that made the microprocessor possible allowed more powerful processors in succeeding generations. In fairly quick succession Intel created the 8008, 4040, 8080 and 8085. Competitors improved the power of their offerings as well. Zilog produced the more powerful and less costly Z80. MOS Technology recruited the engineering team who had created the Motorola 6800 with a simpler and less expensive design, the 6502. The battle on computer features, speed and price that continues to this day was beginning.
Early microprocessors lacked computer power but by the early 1980s processors like the Motorola 68000, Zilog Z8000, TI 9000 and Intel 80386 were equal to or outperformed minicomputers and mainframes. What they lacked were the kind of peripherals that those devices had, multi-megabyte hard disks and tape drives. Ultimately this proved unimportant because small hard-disks quickly outclassed anything the mainframe disk could do.
Personal Computers
In retrospect the puzzle is that neither the computer or microprocessor companies developed any idea of personal computers. They had internal proposals from engineers but seeing no market they turned the idea down.
Stan Augarten cites Jonathan Titus as one of the founders of personal computer age. He ordered an Intel 8008 for $120 and built a prototype computer. US magazine Radio-Electronics published an article on the machine and made a booklet on how to build it available for $5.50. Ten thousand people bought the booklet, and a quarter of them bought the set of circuit boards to go with it. It was becomming evident that there was a serious market for home computers.
Edward Roberts of MITS took a similar approach with an article publihed in Popular Electronics six months later. The Altair 8800 was based on the more powerful Intel 8080 and it sold as a complete kit for $395 or assembled for $650. MITS had done well with its ‘816’ and ‘1440’calculator kit in 1972, moving to bigger premises and buying a wave-soldering machine. By 1974 the semiconductor manufacturers like Texas Instruments and National Semiconductor were producing their own calculators at low prices and MITS owed the bank $300,000. Roberts was an Air Force captain, and with two of his colleagues they devised the Altair. The bank gave a further loan mostly in the hope of preserving their first investment; Roberts said he thought they would sell 800 kits, the loan officer thought 200 might be more realistic.
In the first month MITS took 1,000 orders, they had to hire more people just to answer the phone. By August 1975 they had 5,000 orders. As shipped, the machine could do little more than accept a program and flash lights but it had a 16 slot backplane for expansion boards.
The expansion bus was one of the big successes. It did mean that competitors like IMS could make compatible boards and even machines such as the IMSAI 8080. Processor Technology made a more reliable 4K memory board and a CRT display card as well as its own machine the Sol-20.
The Altair is noteworthy not just as the first great success of a micro-computer but also because Roberts was called by two students who promised to write a BASIC interpreter for the machine. Harvard students Paul Allen and William Gates created an 8080 emulation on Harvard's PDP-10, wrote a basic interpreter and delivered a paper tape a few weeks later. Paul Allen became a vice president for software. The pair started Micro-Soft on the strength of it.
Ed Roberts sold MITS to Pertec for $6.5 million in stock in 1977. (He retired to Georgia and became a farmer and doctor). Pertec made peripheral devices like 14 inch disk and 9 track tape drives but they apparently thought they had bought the rights to BASIC - Roberts had paid for PDP-10 time for Gates and Allen but didn't know how to create new products so MITS collapsed in 1979, as did competitor IMSAI. Pertec itself dissapeared around 1987.
Altair gave rise to many imitators including Apple, Commodore, Cromemco, North Star and Vector Graphic. Apple are the great survivor from this list as are their rivals Microsoft.
Most of the traditional calculator companies in the 1970s had a dilema. They had to get into electronics because neither the comptometer or pinwheel mechanism could possibly provide the same performance. But they either stuggled with the technology (Smith Corona Marchant) or with the marketing for instance Altair ). Altair's llustrates the dilema
IBM launched the 5100 portable computer in 1975. With a base cost of $8,975 rising to $19,975 with a full 64KB of memory it was a competitor to the Wang 2200 and HP 9830. These machines appealed to research laboratories that couldn't get sufficient access to a mainframe.
The blockbuster market for microprocessors proved to be small computers what we now call a PC. and then the 16 bit 8086. The 8086 was functionally equivalent to many minicomputers like the DEC PDP-11 or Data General Nova. IBM chose a low cost implementation of the 8086 with an 8 bit bus as the basis for their PC in 1981. Most of today's PCs are inheritors of that chip series.
Pocket Calculators
In the early 1970s the calculator market fundamentally changed. Machines like the Busicom and the Sumlock ANITA were sidelined by the likes of Sharp, Canon and CASIO making little pocket calculators. One of the companies that rode the change and emerged as a winner was Hewlett Packard.
HP were a scientific instrument maker primarily known for audio oscillators, oscilloscopes and digital voltmeters. They had created and sold the 9100A calculator partly because a couple of people suggested the idea and partly because laboratories often needed a calculator alongside other equipment.
HP were not known for computers. Some of their more elaborate instruments were connected to computers but in that case they used Digital Equipment Coropration minicomputers. In the early 1970s depressed world market meant HP were all on short-time working, laid off and not paid for one day per fortnight. They prefered that to 10% redundacies and it was seen as a sign of HPs enlightened thinking.
The HP-35 Scientific Pocket Calculator was Bill Hewlett's vision. He wanted a pocket version of the 9100A and went so far as to have the design department create an outline design - in the hopes that the new chip technology would be able to meet the needs.
Dave Cochran worked on transistorising the oscillator first devised by Bill Hewlett. Then he worked on a digital voltmeter. The 9100A project was their first calculator, it was just something that might be sold to go alongside other instruments - although it has claims to be one of the first mass-produced computing devices. The 9100A had transistorised electronics and algorithms for transcendental functions. Cochran remembers getting a letter from Wang Labs saying HP had infringed their patent. He sent back a note back quoting Henry Briggs reference from 1624 in Latin and saying "it looks like prior art to me".
It can be argued that calculators and not mainframe computers ultimately gave rise to the modern computing industry. The mainframe computer makers saw little purpose in personal computers although IBM and DEC both had projects proposed. Xerox not only built the Alto, widely regarded as the earliest personal copmputer but invented the laser printer, Ethernet and graphical computing to make use of it.
The truth is probably a bit more complicated; the mainframe computer industry saw no great need for smaller processors but someone in the semiconductor industry was going to realise that it was now possible to put the processor parts of a computer on one chip.
Rebirth of the Calculator
By 1980 low cost calculators had been reduced to a single chip device with an LCD display. Competition was focused on things like solar power so that the machine never needed new batteries, and on providing scientific and programmable calculators - really pocket computers. With advent of Windows PCs and Apple Macs one of the demo programs given free was a calculator. Wyse used to provide a calculator in the ROM on some of their terminals. In theory the calculator might have disappeared.
In practice many offices still have a few calculators around. It may be that every PC and smartphone has a calculator function, but it isn't readily to hand, or it takes up screen area and gets in the way of the primary task, or the buttons on the calculator are more friendly.
Mechanical Calculators Now
Mechanical calculators were made in great numbers; Victor made a million machines between 1929 and 1953. At the time these were fairly expensive pieces of machinery and needed some investment of time to learn to use them. However only a few people had the skill of adding up the 40 or so 4 or 5 digit lines in a ledger manually so the machines were worth the trouble and expense.
When the electronic calculator and computer came into use great numbers of mechanical calculators went into skips. Luckily a few found their way into museums. Like anything no longer made the mechanical calculator is of interest.
The "Curta" hand calculator designed by Curt Herzstark was perhaps the last and certainly the most elegant mechanical calculator.
The modern calculator is often really a microprocessor - its a computer running a fixed program that acts as a calculator. Solar-power calculators are likely to be ASICs configured as calculators, or even dedicated chips in order to minimise the power consumption.
Some mechanical calculators are things of beauty but they are not computers. A person sets up the problem to be solved, winds a handle, cogs turn and the machine gives an answer.
---