The First Commercial Computer Shipped 70 Years Ago Today

Today is the 70th anniversary of a very significant event in all our lives, even if few of us were alive back then. Exactly 70 years ago, March 31, 1951, the first commercial computer shipped.

You might assume that the first computer was shipped by IBM, but you would assume wrong. It was a UNIVAC (UNIversal Automatic Computer) shipped by Remington Rand. Later, Remington Rand would acquire Engineering Research Associates (ERA) and split the company, giving its business computing division the UNIVAC name. It was the U in the BUNCH, the group of companies that tried, largely unsuccessfully, to compete with IBM in the 1960s and '70s (Burroughs, UNIVAC, NCR, CDC, and Honeywell).

Not directly relevant to today's anniversary since the story started a couple of years later, I recommend the book A Few Good Men from UNIVAC as one of the best books on the early days of computing and the larger-than-life engineers who created the systems (especially Seymor Cray). Amazon appears to have it new for $32, which surprises me. It was published in 1987 and is, to say the least, a minority interest topic. Although just the sort of topic that anyone who reads Breakfast Bytes should be interested in. Not only is it a great book about UNIVAC, but it is also one of the best books I've ever read about engineering: Here's the foreword (click to enlarge if you want to read it):

The First Commercial Computer

This first commercial computer was delivered to the census bureau. There is a certain amount of consistency here since a lot of the original punched card infrastructure was developed for the 1920 census. The punched card processing technology was invented by Herman Hollerith, who founded a company that was merged with some others in 1911 to create the Computing-Tabulating-Recording Company. In 1924, this company was renamed to International Business Machines. Yes, that IBM.

Here's a description of some of its capabilities, from the computer history site Bit by Bit:

Since Eckert and Mauchly were most familiar with decimal-based electronic systems, UNIVAC, though a true computer, was based on decimal math. As a result, it used more tubes 5,000—and occupied far more space than a comparable binary machine. The processor was fourteen and a half feet long, seven and a half feet wide, and nine feet tall—ten times larger than the IAS’s binary computer. However, it was quite fast, with an internal drumbeat of 2.25 million pulses a second and the ability to add two twelve-digit numbers in 120 microseconds or multiply them in 1,800 microseconds. The comparable addition time for ENIAC was 200 microseconds. UNIVAC’s memory capacity was the most impressive thing about it. The machine could store 12,000 digits or alphabetical characters in random-access mercury delay lines and millions more in magnetic tape. And it could process ten tapes at a time, each tape storing more than a million characters.

If we take an add operation to be the basic instruction performance, then UNIVAC ran at 0.8 MIPS, impressive for the time (a DEC Vax 11/780 ran at about 1 MIPS about thirty years later). It had 12 KB of memory (or perhaps just 9KB if it used 6-bit characters). Its clock-rate was 2.25MHz. Since it added twelve-digit (decimal) numbers, I assume we would say it was a 48-bit computer. It was based on vacuum tubes, not transistors.

When the UNIVAC I shipped, it was just before the dawn of ferrite core memory. In fact, the UNIVAC II had core memory. But the UNIVAC I still had mercury delay lines. When you first hear about this, it sounds amazing that it actually worked. The memory consisted of tubes of mercury with a transducer (loudspeaker) at one end and a sensor (microphone) at the other. The data was transmitted as sound waves in the mercury. Since the speed of sound in mercury is 1450 meters per second, if the bits were transmitted fast enough then many bits were in transit in the mercury and had not yet reached the other end of the tube. Of course, when a bit reached the end, it had to be recirculated and transmitted again so that the memory didn't lose any data. To read the memory required waiting for the correct bit to come around; to write it, instead of retransmitting the old value, the new value for the bit was inserted. There were seven mercury "tanks" each with 18 pairs of transducers. The picture alongside is one tank.

But the lead of UNIVAC was not to last. As the Bit by Bit site linked to above relates:

[In 1955] Remington Rand’s UNIVAC had an installed base of about thirty big computers to IBM’s four. Meanwhile, UNIVAC’s makers were happily scrambling to fill a substantial backlog of orders. On the basis of these facts, Vickers believed that he had acquired the country’s leading computer manufacturer. But he was in for a shock. In 1954, IBM started taking orders for the 700 series of computers, then on the drawing boards and much superior to UNIVAC’s machines, and UNIVAC soon slipped to second place. By 1956, IBM had seventy-six large computers in the field to UNIVAC’s forty-six, and its backlog was almost three times the size of UNIVAC’s. In medium-sized computers, the fastest-growing segment of the computer market, IBM seemed to be running the race alone. The company had placed seven times as many of these computers (369) as all the other manufacturers combined, and had four times as many (920) on order. Caught unprepared, UNIVAC didn’t even have a medium-sized machine on the market until 1958.

The Computer History Museum is even more brutal in The Rise and Fall of UNIVAC:

By the mid-1950s, UNIVAC’s lead over IBM had evaporated, thanks to poor marketing, delayed products, and new models from IBM.

Learn More

Read A Few Good Men from UNIVAC, the book I recommended above.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.