The twenties. An era of Prohibition (and gangsters)…jazz…flappers…The Great Gatsby…and an accelerating stock market. I thought it might be fun to take a look at the state of technology as it stood a century ago, in 1925. This third post of the series is focused on calculating and information management systems. (The first post focused on communications and entertainment and the second post on transportation)
Devices to assist human computation go back a long way, The abacus first appeared around 2400 BC. and notched tally sticks have been found from as long ago as 20,000 BC. The first true mechanical calculator was developed by Blaise Pascal in 1642. (It was apparently developed to assist with tax calculations!) It was improved by Leibnitz in 1673, but the first commercially-successful calculating machine was the Arithmometer, introduced in 1820!
The growth of large organizations–government and business–drove the need for more computation, as did the expansion of scientific research. By 1925, machines for addition and subtraction were well-developed, with many being electrified to reduce operator fatigue. Multiplication was a harder problem, and multiplying numbers on most of these machines involved a rather klutzy multistep process. If your process involved a lot of multiplication, probably the best option in 1925 was still the Millionaire, introduced in 1893 and featuring direct multiplication–the multiplication table was actually mechanically built into the machine, rather than requiring multiple additions for each multiplier digit. These devices were priced at $475 to $1100 in the early 1900s–for comparison, in 1909, a new Oldsmobile Runabout automobile cost approximately $650, and the average annual wage was under $750. I haven’t found any sources for the Millionaire price in 1925, but given the mechanical complexity of the system, I doubt that it had gotten any cheaper.
The cash register first appeared in 1886 and was dubbed The Incorruptible Cashier, reflecting its primary purpose–preventing employee fraud. Some history.
Bookkeeping machines (also called accounting machines), which evolved from cash registers, were basically adding machines which could maintain multiple totals and print them when required. Here’s a video about the NCR Class 2000 Accounting Machine, which was introduced in 1921 and marketed through 1955. These machines were used for a range of applications, notably in hotels and banks. They were fiendishly expensive..the video quotes a 1940s machine at $2000, which would be around $40,000 in today’s money…but businesses seemed to feel that they got a good return on their investment from them.
When great precision was not required, slide rules were employed: they were common in science and engineering. They also found some use in business for such things as profit margin calculations, but they couldn’t be used for accounting purposes where precise balancing was needed.
Typewriters had become common in offices. One major advantage they offered, in addition to improved legibility, was the ability to make multiple copies via carbon paper–Xerox machines were still a long way in the future. (It’s generally claimed that it was the introduction of typewriters that brought women in large numbers into offices, although it’s not obvious to me why they couldn’t have performed equally well making copies and doing other functions in pre-typewriter offices)
Mathematical tables were important as a way to minimize burdensome calculations. Celestial navigation was one area in which great effort had been developed to creating tables and making them as accurate and easy-to-use as possible. Logarithms were extensively used to shortcut the multiplication and division of multi-digit numbers.
Punched card systems were a significant technology in 1925, though not as important as they would later become: just one year earlier, the predominant company in the field had changed its name to International Business Machines from its previous one, the Computing-Tabulating-Recording Company. (“That little outfit?” thought young Tom Watson Jr when his father announced the change, picturing the company’s rather random-seeming collection of products, which included time clocks, coffee grinders, and scales, and the “cigar-chomping guys” who sold them)
Punched card information processing had been developed circa 1900 to assist with the US Census effort..by the early 1900s, these systems were used in a range of commercial as well as government applications. The primary machines available were the keypunch, which transcribed information into punched cars, the sorter, which could rapidly place cards in a particular pocket based on their contents, and the tabulator, which could accumulate and print totals. However, there was not yet a punched card machine that could multiply–the IBM 601 multiplying punch would not be introduced until 1931–and the collator, which merged two streams of punched cards into a single combined stream, did not make its appearance until 1937.
IBM was not the only company in the punched card business; their primary rival was the Powers Accounting Machine Company. There is remarkably little information about this company and its products available on-line, but one interesting point about their systems is that the sensing of holes in punched cards, and the subsequent calculating logic, was done entirely mechanically rather than electro-mechanically as in the case of IBM. The company was ultimately acquired by Remington Rand.
Mechanical Analog Computers represented numbers via the continuous motion of levers and shafts rather than as digits. In the 1870s, James Thompson had developed a mechanical device to perform the calculus function of integration, and his brother William (Lord Kelvin) had applied it and other mechanisms to create a mechanical tide-prediction system. (Some of these systems were still in use in the early 1970s!) Other than the tide prediction system and the slide rule (which is a very simple example of an analog computer), I haven’t run across any examples of mechanical analog computers in 1925, with the very important exception of military–especially naval–fire control computers. These devices had been developed to a rather remarkable level of sophistication, and the subject is worthy of its own post. A general purpose mechanical analog computer, capable of being reprogrammed for a wide range of problems, wasn’t introduced until 1931. The reprogramming involved wrenches and screwdrivers, see my post here.
The Prescience of Torres Quevedo. Researching this post, I ran across some references to the remarkable Spanish inventor Leonardo Torres Quevedo. He was trained as a civil engineer and among other things he designed the cable car called the Spanish Aerocar, installed at Niagara Falls in 1916 and still in operation. What is relevant here, though, is his work on calculating machines. His 1913 “Essays on Automatics” referenced the work of Charles Babbage and pointed out that Charles Babbage’s ideas could be much more easily implemented using the electromechanical components (relays, etc) which were available in 1913 than with Babbage’s strictly-mechanical approach. He sketched out a design for a machine that would use conductive points on a rotating disk to control a program…and which could perform conditional execution of parts of the program…and also suggested the use of what would later be called floating-point arithmetic. While he never attempted to build a Babbage-style Analytical Engine, he did build and demonstrate in 1920 a simpler machine using some of the same concepts describe in his 1913 essay. Torres Quevado also deserves his own post, which I will try to put together as time permits, although detailed information on his work in English translation seems to be limited.
A slide rule is a modernized version of Napier’s Bones, invented in 1617 by the discoverer of logarithms, John Napier.
https://www.famousscientists.org/john-napier/
Mechanical multiplication was a cinch compared to division. Division was by serial subtraction, largely the way computers do it today. I remember you would put your numbers in and hit the = key, the machine would often grind away for several minutes before printing the answer. Dividing calculators were very expensive, rather delicate and rare. It was faster to use a slide rule to assist long division on paper if you needed better than 3.5 digit precision.
Logarithms are the original computation hack. Lots of different tables of logarithms of different functions, not just trigonometric. Still useful for many things if you remember how to use them.
Remember the first micro controller, the Intel 4004, was invented to power a desk calculator. I still remember using an HP35 scientific calculator that cost about $375 in the early ’70’s if memory serves. A couple of years later, I bought a TI-30 for about $35.
Back in 1969 (I think, might have been 1970) my high school was visited by a rep from Wang business machines who demonstrated their first electronic calculator. The processing unit was the size of a desk, with 4 8-character display keypads connected by cables. The price was $4,000.
MCS, I still have my HP-35, bought in 1973, and it still works just fine. It’s on the desk a foot away as I type this. As for slide rules, well, my two favorites are a 1920s Nestler 23 (Rietz pattern with the “railroad track” scales) and a Hemmi 259 log-log model made October 1957 (date code HJ).
And then there’s the Chinese and Japanese abaci elsewhere in the house….
I’m enjoying this series —
Bill Laughlin — I learned Reverse Polish Notation for arithmetical computation on Wang 360 calculators in the late 1960s. A couple of years later that experience made HP’s RPN calculators simple to use right out of the box.
Not sure which Wang model you saw, but the one in the university I was at had a couple of terminals linked to a briefcase-size CPU. Each terminal had a number keypad, specific function keys, and a Nixie tube numerical display.
See: https://www.wangmuseum.nl/desktop-calculators/ and
https://www.oldcalculatormuseum.com/wang360.html
I’d love to have one now — but only for a day or so. It would be like using a Model T as an everyday car today.
There were also edge-notched cards, which are an entirely different thing from mechanically-processable punched cards as I discussed above. These were useful in categorizing information and being able to actually find it later. Equipment required was minimal: a punch to create the notches and a knitting needle to run through them…the cards that had a notch punched would fall off the needle, those with no notch would stay attached, and either batch could be subject to later selection operations.
https://www.youtube.com/watch?v=v8qHPXPnQps
This video says that Doug Englebart used notched cards to organize his research notes while writing his 1962 paper on computer augmentation of the human intellect, and that these cards helped inspire some of the ideas that led to hypertext and the world wide web.
https://www.youtube.com/watch?v=EwR481vcQSU
In 1968 my dad bought a Friden calculator. It had a two line 10-12 digit crt with four usable registers and a memory. The registers were accessed in a stack. It only had addition, subtraction, multiplication and division and cost $1200. I remember discussing with him that a square root key was another $600 and whether it was worth it. When it eventually died, I took it apart and found about four boards, all with lots of low scale integrated circuits and a mechanical keyboard using magnetic reed switches. It was about 16″ square and maybe 6″ high with a horizontal layout, the roughly 4″ square screen was above the keyboard. It was in a die cast aluminum enclosure and weighed about 30 lbs.
https://en.wikipedia.org/wiki/Friden,_Inc.
We got around the square root problem using Newton’s Method of approximating roots. It would normally converge in five iterations. I remember using it on micro controllers to save the memory overhead of floating point routines. Now even $5 chips come with 256K or more memory so it’s like the tool I still have somewhere, maybe, to install distributor points in early ’50’s GM engines.
David Foster: I have been looking for the keywords that would lead me to “edge notched cards” for some time. Thank you.
I worked at a public library branch from 1981-3, part of a system serving an entire metropolitan area (Buffalo NY). The checkout cards for the entire system used (what I now know are called) edge notched cards. Records of who checked out what were on microfilm.
If I am recalling correctly, the purpose of the edge notched card was that it was the record of the book that the library retained on checkout (they were kept in a pocket in the back of the book). They were microfilmed with your card, and you took the book.
I do not recall what happened after that. I do know the cards had to be sorted, and somehow they ended back in the returned books, which were handed to me for shelving after they were processed.
What I do recall is that there was a class of employees that were not librarians. And a main skill that they had was using the needles to process stacks of cards. Ordering them somehow I presume.
I did some research for a paper last year on the official ranking of NFL teams for playoff seeding. (The working paper isn’t public yet, but if you’re curious shoot me an email and I can send you one when I finish the revision).
The NFL has an odd and problematic method that developed in steps from the 1920’s through the late 1970’s.
Researching that led to some rabbit holes about how MLB ranks teams. Percentages were barely known and rarely used in the 19th century (the technique entered American textbooks around 1840). But sportswriters and leagues began to use them widely by the 1880’s. I imagine that for many people this was their first exposure to them.
I do not know this. But I would imagine that there were tables showing percentages for various combinations of wins and losses. I wonder if any still exist. (I can’t imagine newswires employing someone to redo the long division for every day’s league table).
Wow, this was a trip down memory lane!
My father managed a five-and-dime during my childhood, and used some kind of machine for calculating the totals he needed for the accounting books, which involved pressing keys to enter numbers and select the operation, then pulling a lever on the side to make the calculation.
No idea what it was called.
I started college in 1970, and all of the science & engineering students had slide rules attached to their belts (female SEs were rare) in nice leather (?) holders. We had learned how to use them in junior high math class, but didn’t do much with them in my high school classes. When I came back after the summer, in 1971, the slide rules had been universally replaced by simple four-function calculators. IIRC the Texas Instruments may have been a favored brand. At one time I had a calculator that used Reverse Polish, which I rather liked.
We still have the slide rules used by AesopSpouse’s father, who was a USAF fighter pilot and earned an MS in Mechanical Engineering during his career.
The university-built computer, magnetic core and all, had just been retired in 1970 and replaced by a Burroughs 5500 which was later replaced by an IBM 360/370. I managed to learn Algol, FORTRAN, and PL-1 in my 4 years. I didn’t actually learn to code BASIC until I taught programming around 4 years after graduating, and learned COBOL for the same reason.
(Took a detour to get a master’s in political science, which was fun but ultimately not very useful for earning a living.) I still did a lot of programming for statistical and data base applications during grad school, and earned a living that way, including programming what was essentially a high-level calculator, not a real computer, to do cost-accounting.
Good times, good times.
Burroughs was a leader in the mechanical devices field, and in early computers, but fell before the IBM juggernaut.
https://en.wikipedia.org/wiki/Burroughs_Corporation
AesopFan…”programming what was essentially a high-level calculator, not a real computer, to do cost-accounting”…do you remember what it was called?
@ David – that job was in 1977, and I don’t recall the name of the machine. My boss wrote a simple operating system with a very stripped-down command set (less than BASIC), and we used floppy disks for data storage. It was for the Utah Power and Light company to keep the accounts for a new plant they were building south of Price in Castle Dale.
Our company was named Mini Business Systems; it consisted of the boss/owner, me, another programmer, and a secretary.
Continuing the Retrotech theme, MBS moved on from the programmable calculator (which was rather limited in its operating-system and memory capacity) to a venture in selling real computers, from a then-viable competitor within the industry which sank without a trace: Cincinnati Milacron.
We sold a few machines, but IBM and the other big players didn’t leave much of a market share for the little guys.
I found some information, in the company’s history in a compendium and in an encyclopedia, and one of them is a copy of the other.
These might be a useful resource for your stories about other companies. Much better than Wikipedia, in this case at least.
https://www.company-histories.com/Milacron-Inc-Company-History.html
https://www.encyclopedia.com/books/politics-and-business-magazines/milacron-inc
“Also in the 1970s, the Mill intensified its focus on electronically operated machines, venturing into the markets for minicomputers and semiconductors. While its efforts to manufacture and market minicomputers proved disappointing and shortlived, the company had more success with semiconductors, specifically in the development and manufacture of the silicon wafers on which semiconductors were built.”