Authors: James Essinger
The actual calculations were to be carried out by electromechanical relays; that is, mechanical switches operated by electrically activated magnets.
With this formidable array of hardware and software, the machine could be instructed to carry out any number of complex calculations by means of its paper tape program being left to run until the calculation was completed. Once it was set in operation, the machine would often run for hours or even days, doing what Aiken liked to describe, with a sort of affected rusticity, as
‘makin’ numbers’.
Many programs were repetitive in their basic nature—computing the successive values of a mathematical table was a typical example. A repetitive program was usually inputted into the machine in the form of a loop of paper tape with the ends glued together. A short program might contain a hundred or so instructions along the paper tape, with the whole loop repeating about once every minute. A longer program would take correspondingly longer to run, and might involve a loop rotation of many hours. Because a long program involved a very lengthy loop of tape, special racks and pulleys had to be built into the machine to take up the slack of the loops of tape and to ensure that the tape was at the right level of tension when it entered the tape-reading unit. The machine might take as long as half a day to calculate one or two pages of a volume of mathematical tables. This length of time seems ludicrous to us today, but at the time represented an enormous advance from manual calculation machines.
Above all, Aiken’s computer was remarkably reliable; there was no feeling generated among its users that its results needed 228
IBM and the Harvard Mark 1
to be continually checked. The machine consequently at last did away with the psychological insecurity that had infested the calculation of mathematical tables since Babbage’s day.
By the time Aiken’s computer was reaching completion, it was becoming known as the Harvard Mark
1
. Aiken favoured this name, although IBM preferred the appellation of Automatic Sequence Controlled Calculator. This minor disagreement over the name indicated a wider and growing lack of agreement between Aiken and IBM as the world’s first true computer started to become a reality. Aiken was arguably more at fault here than IBM; he was inclined to see the machine as entirely his own invention, while paying insufficient homage to IBM’s enormous investment of time and money in the actual building of it.
This eventually led to serious problems in his relationship with Thomas Watson.
In many respects the computer which Aiken designed and IBM built did indeed represent the fulfilment of Babbage’s dreams, given that the fulfilment used a different technology to the purely mechanical cogwheels Babbage had envisaged. In fact, in one important respect Aiken’s machine fell seriously
short
of Babbage’s plans. Aiken’s computer was incapable of carrying out what is now referred to as a ‘conditional branch’—that is, a change in the progress of a program according to the results of a previous computation. The Harvard Mark
1
could not change the direction of a computation in this way, which made complex programs very long and slowed down the machine significantly.
This was in spite of the fact that the technology would certainly have made a conditional branch system possible. As Martin Campbell-Kelly and William Aspray point out in their book on the history of the modern computer:
If Aiken had studied Babbage’s—and especially Ada Lovelace’s—writings more closely, he would have discovered that Babbage had already arrived at the concept of a conditional branch. In this respect, the Harvard Mark I was much less 229
Jacquard’s Web
impressive than Babbage’s Analytical Engine, designed a century earlier.
But the Harvard Mark
1 had
been completed to fully working order, and the Analytical Engine had not.
On
7
August
1944
, by which time there was no longer any real doubt that the Allies would win the war, the press office of Harvard University issued the following announcement:
World’s Greatest Mathematical Calculator
The world’s greatest mathematical calculating machine, a revolutionary new electrical device of major importance to the war effort, will be presented today to Harvard University by the International Business Machines Corporation to be used by the Navy for the duration.
The apparatus will explore vast fields in pure mathematics and in all sciences previously barred by excessively intricate and time-consuming calculations, for it will automatically, rapidly and accurately produce the answer to innumerable problems that have defied calculation.
The machine is completely new in principle, unlike any calculator previously built. An algebraic super-brain employing a unique automatic sequence control, it will solve practically any known problem in applied mathematics. When a problem is presented to the sequence control in coded tape form it will carry out solutions accurate to twenty-three significant figures, consulting logarithmic and other functional tables, lying in the machine or coded on tapes. Its powers are not strictly limited since its use will suggest further developments of the mechanism incorporated.
The press release also made much of the physical dimensions of the machine and its enormous number of components. As the announcement proudly explained:
230
IBM and the Harvard Mark 1
The machine is of light-weight, trim appearance: a steel frame, fifty-one feet long and eight feet high … bearing an interlock-ing panel of small gears, counters, switches and control circuits. There are
500
miles of wire,
3 000 000
wire connections,
3500
multiple relays with
35 000
contacts,
2225
counters,
1484
ten-pole switches and tiers of
72
adding machines.
The formal presentation of the Automatic Sequence Controlled Calculator to Harvard University, which lent it to the US
Navy for the remainder of the War, took place in the Faculty Room of Harvard’s University Hall on Monday
7
August
1944
. It was a major event. The Governor of Massachusetts was there, as were four admirals, numerous other officers of the armed forces, faculty members, deans, and members of the Harvard governing boards, plus three of the IBM engineers who had constructed the Harvard Mark
1
and members of the machine’s operating staff.
Aiken, the hero of the hour, made a speech that not only summed up much of his life’s work but can also be seen as a summing-up of much of the story of Jacquard’s Web.
I should say that the purpose for which we are forgathered here this afternoon is as old as civilisation itself. Our purpose is to consider a device designed to assist in the solution of mathematical problems, and to derive numerical results, because the record is clear that those who invented the fundamental processes of arithmetic were themselves the first to feel the need of mechanical aids.
The first mechanical aid to be used was, of course, the ten fingers of the hands. It is for that reason that the number system we still use at the present time is based on the numeral
10
. After the invention of zero, and the extension of the numbers system, the fingers no longer sufficed for counting, and pebbles were used assembled in piles on sheets for the purpose. It was from this that the invention of the abacus came, wherein beads strung on wires took the place of the pebbles, and the wires facilitated their easy movement.
231
Jacquard’s Web
Aiken went on to provide a fulsome and expansive tribute to Babbage. He explained how Babbage attempted
to build first what he termed a Difference Engine, and then what he called an Analytical Engine, after he had failed with his Difference Engine in the first place. I say Babbage failed, but I should like to make it especially clear that he failed because he lacked machine tools and electric circuits and metal alloys, but through no fault of his own. Babbage’s failure was due solely to one fact: he was a hundred years ahead of his time.
Aiken’s remarks about Babbage—the father of the modern computer talking about the achievement of someone who might reasonably be described as the father of the entire concept of computing—set the scene for the way Babbage was seen by posterity during the second half of the twentieth century. Of course, blaming Babbage’s failure on a lack of
1940
s technology rather tends to exalt that technology, which may have been the idea. In fact, as we have seen, there is compelling evidence that under the right circumstances Charles Babbage might indeed have been able to build a working Difference Engine in his own century. The idea that he failed simply because he was ahead of his time is really an oversimplification, though not an unreasonable one.
Aiken continued:
But after machine tools had been developed, and all the assets of modern manufacturing became theirs, the problem again was opened up by a variety of different manufacturers. Still, the problem remained in the field of the four processes, the four fundamental processes of arithmetic. And then there came one of the fundamental inventions of all times in the art of computation, the use of punched cards for the storing of numbers and for the rapid distribution of those numbers into counters for carrying on numerical processes. It was this 232
IBM and the Harvard Mark 1
invention, developed by the International Business Machines Corporation, and all the associated parts of mechanisms for speeding and using such cards, that has brought the possibilities of scientific calculating machinery again into a position where one could look at the situation with hopes of success.
This was the situation then, as we saw it, compared with the situation that Babbage saw one hundred and more years ago.
Aiken unfortunately skates over the enormously important contribution of Herman Hollerith as a pioneer in punched-card information storage and also as a key link between IBM and the Jacquard loom. Indeed, Jacquard was also missing from this concise account of the history of mechanical aids to calculation in general, and the history of the punched card in particular. But Aiken had mentioned Jacquard by name in the proposal which he prepared for IBM, so there is no doubt that he was aware of the importance of Jacquard’s work as a precursor to Babbage’s use of punched cards. In fact, as Aiken was so heavily influenced by Babbage, and as Babbage freely borrowed Jacquard’s ideas, it is entirely reasonable to say that Aiken was in a very real sense influenced by Jacquard.
In
1946
Aiken was co-author of an important work entitled
A
Manual of Operation for the Automatic Sequence Controlled Calculator.
It is clear to any reader of the
Manual of Operation
that Aiken considered Babbage to be his intellectual father. The very first chapter starts with the quotation from Babbage’s
Passages from the
Life of a Philosopher
that heads this chapter, a quotation in which Babbage speaks with uncanny foresight to whoever will be the inheritor of his mantle.
Aiken proceeds to look at the most ambitious aspect of Babbage’s work:
Having been unable to complete the difference engine, Babbage embarked upon the creation of a far more ambitious concept, an ‘analytical engine’. Though the terms of the prob-233
Jacquard’s Web
lem proposed were enough to stagger the contemporary imagination, he attempted to design a machine capable of carrying out not just a single arithmetical operation, but a whole series of such operations without the intervention of an operator. The numbers in the first part of the machine, called the ‘store’, were to be operated upon by the second part of the machine, called the ‘mill’. A succession of selected operations was to be executed mechanically at the command of a ‘sequence mechanism’ (a term unknown to Babbage).
For this latter, he intended to use a variation of the Jacquard cards.
These cards, the precursors of Hollerith’s punched cards, were used by the Jacquard weavers to control the looms to produce and reproduce the patterns designed by the artists.
The designs were first sketched as they were to appear in the finished product, transferred to squared paper and used as guides for punching cards. The cards allowed certain needles to be extended through the punched holes, thereby controlling hooks which, in turn, raised particular warp threads to produce the desired pattern. In order to continue the weaving of the same design, the cards were interlaced with twine in an endless sequence so that one card was brought into position immediately after another was used. Holes were punched for the lacings as well as for the pegs which guided the cards over a cylinder.
In adapting these cards for use in his machine, Babbage required two decks: one of variable cards and one of opera-tional cards. The first set was designed to select the particular numbers to be operated upon from the store; the second set, to select the operation to be performed by the mill. The deck of operation cards therefore represented the solution of a mathematical situation independent of the values of the parameters and variables involved. Thus the analytical engine was to have been completely general as regards algebraic operations.
234
IBM and the Harvard Mark 1
Aiken’s account of Babbage’s work is concise and fairly accurate, even in the light of the very considerable new information about it which has since come to light.
A review of the
Manual of Operation
that appeared in the British scientific journal
Nature
was published under the title
‘Babbage’s Dream Comes True’. The start of the enormous surge of renewed interest in Babbage’s work—an interest that is continuing to increase—dates from around this time. The idea that a brilliant and unjustly neglected Victorian gentleman had anticipated the development of the modern computer by more than a century, and even provided a blueprint for its development, was too romantic, exciting, and interesting to be overlooked. Yet one must avoid being too carried away by it. As I. Bernard Cohen, Aiken’s biographer and a leading computer historian, wrote in
1999
: