Skip to content

The Pocket Calculator

Abstract

For 350 years the slide rule was the engineer’s portable computer. Between 1971 and 1976 it was retired by a shirt-pocket gadget. The electronic pocket calculator arrived from Japan, was made possible by the same silicon that produced the first microprocessor, and set off a price war so fast that four-function models fell from hundreds of dollars to a few dollars in five years. Hewlett-Packard’s HP-35 killed the slide rule for scientists in 1972; Texas Instruments then killed the high price. Most of the companies that started the race, including the one that gave America its first pocket calculator, did not survive it.

Electronics on the Desk

Electronic calculation reached the office a decade before it reached the pocket. The ANITA, built by the Bell Punch Company of Uxbridge near London, was the first electronic desktop calculator. Its chief designer, Norbert Kitz, had built a small valve computer as a student at Birkbeck College and worked on the DEUCE at the National Physical Laboratory; Bell Punch filed the first patents in September 1956, and a prototype ran in December 1958. The ANITA used about 180 cold-cathode tubes, which needed no heating, and showed its result on Nixie tubes. It was shown in London and at an office-equipment fair in Cologne in October 1961, delivered from 1962 at £355 (DM 4,200 in West Germany, the price of the cheapest Volkswagen Beetle), and from 1964 sold around 10,000 a year.

In 1964 transistor desktops followed from Italy, Japan and the United States, among them Canon’s Canola and Friden’s EC-130, which took formulas in reverse Polish notation. A small Massachusetts start-up, Mathatronics, delivered its first programmable Mathatron in July 1963: 36 kg, a paper-tape printer, floating-point arithmetic and parenthesised formulas, for about $3,500. Hewlett-Packard later bought the inventors’ rights. Olivetti’s Programma 101 and HP’s own 9100A of 1968 belonged to the same class of machine, sold as calculators to avoid the word computer (see Hewlett-Packard).

The Prototype Nobody Could Sell Yet

The pocket calculator began at Texas Instruments as a demonstration that a calculator could be small. In 1965 TI’s president Patrick Haggerty was looking for a product to sell integrated circuits into, and settled on a calculator the size of a paperback. In 1967 a team led by Jack Kilby, co-inventor of the integrated circuit, built the “Cal-Tech” prototype: a battery-powered handheld that could add, subtract, multiply, and divide, printing results on a strip of thermal paper. It was handed to Haggerty on March 29, 1967: an aluminium case weighing about 2.5 pounds, holding four integrated circuits and a battery, and accepting numbers of up to six digits. It was heavy, it was slow, and TI had no intention of selling it; the prototype is now at the Smithsonian. The point was the patents. TI filed for master patents on portable electronic calculators and then waited for the components to get cheap enough to matter.

The first products to reach shops came from Japan. Sanyo’s ICC-82D and Sharp’s QT-8B “micro Compet” were on sale by early summer 1970, and Dictaphone sold the Sanyo in the United States from September 1970, probably the first handheld calculator an American could buy. Canon announced the Pocketronic in April 1970, built on the TI Cal-Tech design with three TI chips; it reached Japanese shops in the autumn and American ones in spring 1971 at just under $400, and printed on thermal paper rather than showing digits on a display. Sharp’s EL-8, shrunk to 16 by 10 cm, followed before the end of 1970 at $345. Most of these were still too large and too expensive for a pocket in any honest sense.

The First One That Fit

The machine that earned the name went on sale in Tokyo in January 1971. The Busicom LE-120A “HANDY” was the first calculator small enough to carry in a pocket, the first to use a light-emitting-diode display, and the first to run on a single integrated circuit and replaceable AA batteries. Busicom is the more consequential name in this story for a second reason: it was Busicom that had commissioned Intel to build a set of calculator chips, and Intel’s engineers answered by putting the logic on one general-purpose part, the Intel 4004. The microprocessor was born as a way to build calculators (see Federico Faggin and the Microprocessor). Busicom, short of cash, traded away its exclusive rights to the chip in exchange for lower prices, and the 1974 recession in Japan made it the first major Japanese calculator maker to fail. Intel kept the microprocessor.

The HANDY measured 7.2 by 12.4 by 2.4 cm, showed twelve digits and ran on a Mostek calculator chip; it cost ¥89,800 in Japan and just under $400 in the United States. One of them, brought back from Japan in early 1972 by the East German electronics minister Otfried Steger, became the model for the GDR’s first pocket calculator, the minirex 73, presented at the Leipzig Spring Fair in 1973 (see Computing in East Germany).

America’s first pocket calculator came from an Indiana instrument maker. Bowmar put the 901B, nicknamed the “Bowmar Brain,” on sale in autumn 1971: eight red LED digits, four functions, $240. It sold well enough that Bowmar could not keep up, which turned out to be the wrong problem to have.

The Slide Rule Killer

None of these could do the work an engineer actually needed. They added and multiplied; they could not take a sine or a logarithm. For that an engineer still reached for the slide rule, the two-scale device William Oughtred had assembled around 1622 and that had sized bridges and flown to the Moon on Apollo (see Before the Computer: Mechanical Calculation).

Bill Hewlett changed that on a whim. He had watched customers buy HP’s desktop 9100A and wanted the same power in a shirt pocket, so he measured his own shirt pocket and told his engineers to build a calculator that fit. The idea dated from March 1968, when Hewlett demonstrated the 9100A to the calculator maker An Wang in New York and afterwards told the 9100A’s designer, Tom Osborne, that the next machine needed a tenth of the price and size at ten times the speed. Low-power chips were the obstacle until project leader David Cochran found suitable PMOS parts at Fairchild in autumn 1970. Hewlett presented the HP-35, named for its 35 keys, on January 4, 1972, at $395 (DM 1,987 in West Germany). It used reverse Polish notation, and its algorithms came from the 9100A. It was the first handheld calculator with trigonometric and exponential functions, the first that a scientist could carry instead of a slide rule (see Hewlett-Packard). The market research said not to build it: a Stanford Research Institute survey recommended against the product, and the most optimistic forecast ran around 1,000 units a month (the HNF gives a forecast of 50,000 units in all). HP soon sold more than 10,000 a month and kept waiting lists. It sold 100,000 in the first year and more than 300,000 by the time it was discontinued in 1975. HP-35s flew on the Skylab 3 and Skylab 4 missions.

The effect on the slide rule was quick and total. Colleges stopped teaching it; slide-rule makers stopped making it. A tool that had lasted three and a half centuries was gone from engineering practice inside a decade. Dennert & Pape in Hamburg, West Germany’s leading maker under the Aristo brand, stopped making slide rules in 1976 (another HNF post gives 1978).

The bug HP told everyone about

An early HP-35 had a flaw in its transcendental-function code: certain calculations, such as raising e to the logarithm of 2.02, returned 2 instead of 2.02. HP had already sold some 25,000 units when the bug surfaced. In the meeting about what to do, someone suggested not telling; Dave Packard’s pencil snapped, and HP wrote to the owners it could reach, explained the error, and offered a replacement to anyone who wanted one. Many customers, told their expensive calculator was flawed, chose to keep it anyway.

The Price War

Texas Instruments had the patents, the chips, and a reason to stop merely licensing to everyone else. In September 1972 it entered the market directly with the TI-2500 Datamath, a four-function LED calculator at $119.95, and then did to the calculator what it had spent years preparing to do: drive the price down until competitors could not follow. Because TI made its own integrated circuits, it could undercut companies that had to buy theirs.

The numbers fell at a speed the industry had never seen. In Britain, Clive Sinclair shipped the Sinclair Executive in September 1972 as the first genuinely slim pocket calculator, then the Sinclair Cambridge in August 1973 at £29.95 plus VAT, or £24.95 as a build-it-yourself kit; by late 1974 the mail-order price had fallen to £16.95. Casio pushed volume from Japan. By 1976 the cheapest four-function calculator cost a few dollars, about one-twentieth of the 1971 price. The device that had been a status symbol in 1971 was a giveaway by 1976.

The companies that had started the race were the ones it crushed. Bowmar, which had given America its first pocket calculator and depended on buying chips from TI, was squeezed between rising demand and falling prices and filed for bankruptcy in 1976. Commodore survived the calculator business only by buying its own chip supplier, MOS Technology, whose 6502 processor would later run the personal computers of the next decade. The calculator war killed more than the slide rule. It selected, brutally, for the handful of firms that owned their silicon.

Made in Europe

European assembly began in 1972, almost always around an American chip, usually TI’s TMS0105. In Yugoslavia, Digitron of Buje in Istria put its db 800 on sale on January 15, 1972, probably the first pocket calculator made in Europe (see Yugoslavia’s Computing Industry). West German firms followed the same year: Helec of Huntlosen near Oldenburg with the GT 530, also sold by the mail-order house Quelle at DM 248; FG-Elektrotechnik of Rückersdorf with the Multiplus kit from September 1972 at DM 339; and Walter Dittel of Landsberg with the W200. The slide-rule maker Aristo bought the rights to a design by the Berlin vocational teacher Hilmar Bentert and sold it from autumn 1972 as the Aristo M27 at DM 466; its own M 64 followed in 1973. Aristo gave up calculators in early 1979.

The Model That Ended the Argument

Texas Instruments closed the scientific-calculator market the same way it had closed the four-function one. In 1976 it released the TI-30, a scientific calculator with the functions of an HP-35, for about $25 when HP’s cheapest scientific model still cost well over $100. It sold an estimated 15 million units between 1976 and 1983, which made it the best-selling calculator of its era and put a machine capable of trigonometry into the hands of essentially every science student. What had been a $395 professional instrument in 1972 was, four years later, a classroom staple that cost less than the textbook next to it.

Programmable and Graphing Calculators

At the top of the market HP and TI played leapfrog for a decade. TI’s SR-50 of January 15, 1974 matched the HP-35 for $149.95 and took input in algebraic order. Four days later HP announced the HP-65, the first programmable pocket calculator, which stored programs on small magnetic cards and cost $795. TI answered on September 16, 1975 with the SR-52 at $395, which allowed 224 program steps against the HP-65’s 100; HP caught up with the HP-67 in 1976. The TI-59 of May 24, 1977 added plug-in ROM modules of ready-made programs, cost $299.95 and reportedly sold three million. HP’s HP-41C of July 1, 1979 brought an alphanumeric LCD and user-assignable keys.

The graphing calculator came from Japan. Casio’s fx-7000G, advertised in October 1985 at £79.95, drew curves and lines on a 95 by 63 pixel display and ran programs of up to 422 steps. Sharp followed in 1986 with the EL-5200 and HP in 1987 with the HP 28C, which could differentiate and integrate symbolically and was programmed in HP’s RPL language; the HP 48 series appeared in 1990. TI’s TI-81 of 1990 cost $110 and started the line that made TI the classroom standard: TI had sold 25 million graphing calculators by 2003.

📚 Sources