Jacquards' Web (31 page)

‘probably as old as the science of arithmetic itself’.

After brief references to various early and unsatisfactory attempts to mechanize calculation, including the devices developed by Pascal and Leibniz, Aiken discusses the slide rule before turning to the work of Charles Babbage.

It is important to emphasize that Aiken did not have access to the detailed technical knowledge of Babbage’s machines we have today. This knowledge has only been available since the
1970
s, when painstaking research among Babbage’s original papers unearthed it. Aiken’s main source of information appears to have been Babbage’s autobiography,
Passages from the Life of
a Philosopher
. While extremely illuminating on the matters of Babbage’s personality and outlook, this book is weak on the technical features of his two Engines. It would have been quite impossible to have built a Babbage machine from the material in it, for example. Still, Aiken would at least have learned from
Passages
that there were two basic types of Babbage machine—

the Difference Engine and the Analytical Engine. He was also aware that the Analytical Engine (as he put it)

pointed the way … to the punched-card-type of calculating machine … it was intended to use perforated cards for its control, similar to those used in the Jacquard loom.

Which is, of course, exactly the point.

In the proposal, Aiken moves on to discuss Hollerith’s invention of punched-card ‘tabulating, counting, sorting and arithmetical machinery’. Aiken takes pains to emphasize the historical continuity of the ideas that originate with Babbage. As he explains, the ideas then proceed through the work of Herman Hollerith and result in the proposal itself. The link he suggests between 221

Jacquard’s Web

Babbage’s work and Hollerith’s is misleading, however. There is no evidence that Hollerith knew of Babbage’s ambitions. The real link is from Hollerith back to the work of Jacquard.

Aiken ends the retrospective part of the proposal by observing that the machines ‘manufactured by the International Business Machines Company’ have made it possible to achieve daily in the accounting offices of industrial enterprises around the world everything that Babbage wished to achieve. The somewhat fairly crude flattery here may, perhaps, be excused on the grounds that Aiken was doing everything he could to give his proposal the best chance of succeeding when presented to IBM. Aiken was planting in the reader’s mind the notion that his new type of machine might also have important commercial applications in the offices of industrial enterprises.

Of course, Aiken’s comment was in any case an exaggeration; IBM’s tabulators, highly sophisticated as they were, were ultimately merely extremely fast card-readers and card-counters.

They were not computers in the modern sense of the word, and it
was
a modern computer (technically a digital computer) that Babbage had been trying to build. And so was Aiken.

The historical material Aiken provides is an important link which in effect summarizes much of the story of Jacquard’s Web. It furnishes a clear and unequivocal series of links that connect the Jacquard loom to the modern computer via Charles Babbage and Herman Hollerith.

Aiken’s proposal states the case for constructing a machine capable of providing ‘more powerful calculating methods in the mathematical and physical sciences’. He emphasizes that ‘many recent scientific developments’ are based on highly complex phenomena that cannot be properly studied until some means is found to make the abstruse calculations that would be necessary in order to track the phenomena, understand them, and predict them. There are problems beyond our capacity to solve, he 222

IBM and the Harvard Mark 1

writes, ‘not because of theoretical difficulty, but because of insufficient means of mechanical computation’. This is essentially a twentieth-century version of Charles Babbage’s lament at the lack of a reliable way of calculating mathematical tables.

Aiken specifies four design features that differentiate ordinary punched-card accounting machinery from the kind of machine he has in mind. The brilliance of his overall conception is shown nowhere better than here, where he lays down what is nothing less than a specification for the modern computer. This specification, along with the discursive material that precedes it, gives Aiken’s proposal a claim to being one of the most important documents in the intellectual history of mankind.

The first essential design feature, Aiken points out, is that the machine must be able to handle both positive and negative quantities, whereas the punched-card machines could only deal with positive numbers.

The second design feature is that the machine must be able to handle all types of complex mathematical functions such as trigonometric functions. These describe different relationships between the angles and the length of the sides of a triangle and have always been of crucial importance in calculations relating to navigation and engineering. Aiken emphasizes that the machine must be a computing device in every sense of the word; that it needs to be able to do much more than basic addition, subtraction, multiplication, and division.

Thirdly, he stipulates that when the machine is carrying out a mathematical calculation, the calculation process must be ‘fully automatic in its operation once a process is established’. This was a vital matter, very likely referring directly to Babbage’s Analytical Engine.

Fourthly, Aiken specifies that the machine must be able to handle calculations that involve adding up figures horizontally in rows rather than only vertically in columns, which was all that the tabulation machines could do. Aiken pointed out that this feature was especially important for differential equations, where 223

Jacquard’s Web

the computation of a value frequently depended on existing values.

Aiken goes on to assert that the adoption of these four features is all that is required to convert existing punched-card calculating machines (‘such as those manufactured by the International Business Machines Company’) into a new type of machine that would be, as he puts it, ‘specially adapted for scientific purposes’. Aiken concedes that the number of components would have to be increased compared with existing machines. Yet it seems clear from the proposal that he envisaged the machine being constructed from the same components that IBM was already using. In fact, as things turned out, the world’s first computer not only needed to be built from new types of components but also had to be operated in a manner radically different from even the most sophisticated automatic tabulation machine.

The proposal also specifies sixteen mathematical operations that would need to be built into the machine. These include the four fundamentals of addition, subtraction, multiplication, and division, logarithms and antilogarithms of base ten and other bases, as well as trigonometric functions. Like Babbage, Aiken emphasizes the importance of being able to
print out
the results so as to avoid the danger of human error in copying the results from a ‘window’ in the machine. The proposal ends with a general statement of confidence that, for IBM, the job of building the machine is merely a further step along a road already trod-den:

Suffice it to say that all the operations … can be accomplished by these existing machines [i.e. IBM’s latest generation of automatic tabulation machines] when equipped with suitable controls, and assembled in sufficient number. The whole problem of design of an automatic calculating machine suitable for mathematical operations is thus reduced to a problem of suitable control design, and even this problem has been solved for simple arithmetical operations.

224

IBM and the Harvard Mark 1

Aiken delivered his proposal for the world’s first computer to James Wares Bryce in early November
1937
. Thanks to the enthusiastic support of Bryce, Thomas Watson, and other senior people at IBM, work soon got underway. At Bryce’s suggestion, Aiken attended an IBM training school, where his existing detailed knowledge of IBM machines was brought completely up-to-date and where he gained important experience in actually using the machines.

Then, early in
1938
, Aiken visited one of IBM’s principal research and development laboratories at Endicott in New York.

There, he started to work on the machine in conjunction with a team of Bryce’s most trusted and able senior engineers. Bryce placed in charge of the project an extremely capable male engineer with the unusual name (yet another one) of Clair D. Lake.

Lake was one of IBM’s longest-serving engineer inventors. He had designed the first automatic printing tabulator in
1919
, a formidable feat that involved linking, with great precision, the rapid punched-card counting mechanism with a series of electromagnetic relays—that is, on/off switches—which controlled magnetic hammer-type print heads. Aiken could hardly have asked for a better project leader, nor for a better team of engineers.

Needless to say, engineering of this calibre cost serious money, much more than Aiken had envisaged. IBM’s initial estimate of $
15 000
to build the machine was quickly raised to $
100 000
(about $
1
.
5
million at modern prices) once the full extent of the true complexity of the task was understood by IBM’s engineers.

Early in
1939
, IBM’s board gave its formal approval for the construction of what was by now called an ‘Automatic Computing Plant’. Once the project had been authorized, and the engineering team had started its work, Aiken’s role became primarily that of a consultant and adviser rather than a full-time member of the engineering team itself. This was not due to any lack of enthusiasm on his part, but because during the early years 225

Jacquard’s Web

of the work his knowledge of the technical issues was not extensive enough to allow him to assemble the machine itself. Later he had no time to work on the machine because during the war years he had full-time responsibilities within the armed forces.

During the war, Aiken’s project became just one of several special projects that IBM was carrying out for the US Army. It was inevitable that it should become a military project in a time of war, although it was not immediately clear what practical military applications the machine would have. The most likely application would be calculations to do with ballistics and battle-field information. As these contain numerous variables, they could not be handled by IBM’s most complex punched-card machines, let alone by individual clerks.

In the meantime, Aiken had joined the United States Naval Reserve, where he was assigned to the Naval Mine Warfare School in Yorktown, Virginia. With the rank of Lieutenant Commander, he acted as a senior instructor to young officers on the subject of electricity and electronics. Aiken appears to have enjoyed being in the Navy, but he was disappointed at being deprived by the war of the chance to be intimately involved with the work at IBM. By this point the actual construction of the machine was mainly a matter of building the components and assembling them, and IBM’s own engineers were more than competent to do this. Aiken’s moment of glory, however, was to come.

The war, coupled with the technical difficulties of the project, caused many delays in the schedule. IBM was not ready to run its first test problem on the computer until January
1943
. It was a secret operation, further evidence that the US military was hoping the machine would offer useful war-time applications.

The complete machine was a wonderful piece of engineering. Fifty-one feet long, eight feet high, and two feet wide, it weighed more than five tons. Altogether it contained more than
750 000
parts and hundreds of miles of wiring. In operation, the 226

IBM and the Harvard Mark 1

Howard Aiken as Lieutenant Commander.

machine was described by one commentator as making the sound of ‘a roomful of ladies knitting’. All the basic calculating units were lined up and driven by a fifty-foot shaft powered by a five-horsepower electric motor (not steam, though). The machine could store up to seventy-two numbers in its memory. It was capable of three additions or subtractions a second. A complex multiplication took six seconds, while calculating a logarithmic or a trigonometrical function required about a minute.

The machine is today regarded as the first automatic digital calculating device ever to be successfully constructed. As such, it deserves to be regarded as the world’s first computer. It was programmed by a sequence of instructions coded on punched paper tape. The word ‘program’, meaning a set of instructions for a mechanical device, was entering general currency at around this 227

Jacquard’s Web

time. Aiken made an important contribution to the proliferation of the word in this sense, but he does not seem to have originated it.

Aiken specified that the actual data were to be inputted on punched cards, with the result of the calculations either being recorded on such cards or by an electric printing mechanism not dissimilar to the printing mechanism of the automatic tabulators.