Herbert R.J. Grosch
ISBN 0-88733-085-1
Copyright © 1991, 2003, by Herbert R.J. Grosch
First edition book design by Ted Pedersen
All Rights Reserved
THIRD EDITION
For Mabel and Dorothy and Elizabeth and Joyce
—and especially for Nancy
[ Contents ] [ Chapter 01 ] [ Chronology ] [ Index ] [ Biography ]
Computer: Bit Slices from a Life was converted to HTML for the Web by Frank da Cruz in May 2003 for the Columbia University Computing History Project with permission and collaboration of Dr. Grosch. This is a manuscript of the 3rd edition, a work in progress sponsored by the US National Science Foundation. The first edition was published by Third Millenium Books, Novato, California, in 1991 and as noted below, copyright was reassigned to the author upon breakup of that company. As to conditions for use, Dr. Grosch says "make sure it's as generous as possible. The only restriction I really want is, no alterations, or elisions which change [my] intentions." To this I would add that the definitive, complete, and up-to-date Web copy of this work is to be found in the following location; copies at other sites might be dated, incomplete, or altered:
http://www.columbia.edu/acis/history/computer.html
This is a simple, monolothic, and validated HTML file, 1.7MB in size, representing over 500 printed pages, self-contained, fully searchable, and displayable by any browser. The only cautions for viewing, other than sheer size, are the presence of several tables, which browsers such as Lynx might not format correctly, and the fact that it contains a fair number of words in German, French, and other languages containing accented characters, coded in ISO 8859-1 Latin Alphabet 1, properly announced.
The Index (which applies only through Chapter 25) is fully linked, and page anchors have been added through the end of Chapter 25, and correspond to the printed pages of the first edition (which has 24 chapters and no index). Page numbers appear inline, representing the beginning of the corresponding printed page, as [-xx-] (in red if your browser and computer permit), where xx is the page number. You can search for any particular page (through 270) using this format. New chapters will be added as they arrive.
| Comments to: | fdc@columbia.edu | ||||||||||||||||||
| Chapters: | 56 | ||||||||||||||||||
| Change Log: |
|
[not yet finalized]
—Las Vegas, Nevada [2003]
EIGHT YEARS AGO I sent the first 23 chapters of COMPUTER to press. I wanted to go on; I had toured Europe in the early Sixties, and had other stories from those days that historians and veterans would want to hear. But as I began again, a family tragedy intervened, and I set the work aside.
When I returned to the United States I brought with me the remains of my huge original archive: my 1962 call reports to Control Data, my flight logs, my appointment (and hotel/restaurant) lists, some of my expense account carbons. Cautiously I began to reconstruct what had happened to me and to the burgeoning world of computers from mid-1959 on. It went slowly.
Then I was awarded a National Science Foundation grant. It was to me as an individual investigator (unusual), and it permitted me to move from New Mexico to Washington, where I had the resources of the Library of Congress and of the Smithsonian within walking distance. The publishers of the first edition had broken up, and courteously relinquished the copyright. I decided to continue the material more or less seamlessly, and to call the result a second edition of COMPUTER rather than a separate book.
What follows is the result. There are no pictures this time, although I have many more to display. The pagination of the bound volume is preserved. Errors that I am aware of in the first edition have been corrected, although I am sure there are others, and not a few in the new material. The table of contents and the chronology have been extended. And at the very end I have appended the index to the first 23 chapters, which Underwood-Miller did not print.
Providing this edition on CD/ROM, uncoded and uncompressed, means that chapters or sections of the book can be downloaded and subjected to full-text search. This is a poor substitute for a careful index, and I apologize. If I ever complete the 1967-87 years, which is possible but not too likely, I will then undertake a full index.
In a few months Y2K will be upon us. To look back from that cusp to a younger and happier time, when anything seemed possible and the world of computing was just opening out, is a great privilege, and a very great pleasure.
—Washington, D.C. [1999]
THIS BOOK was to be an autobiography. I was made into a computer fifty years ago. I was the second scientist ever hired by IBM, and I watched the Watsons on Olympus, and Bill Norris and Ken Olsen and Gene Amdahl, and a thousand great commercial and academic figures. Later, for three amusing years, I was the top Federal computer honcho, and escaped to be editor of the major trade newspaper. I was a charter member of the world's first and largest professional computer society, and the first national president ever elected by membership petition. I worked in Monaco and Switzerland and the Netherlands when I was too controversial to be employable in the U.S., and I consulted for the largest and best Japanese computer company.
And I had four wonderful wives, and a hundred lovely supporters, and terrific friends on five continents - and some very lively enemies. Perhaps I could do for the computer trade what Casanova did for Venetian diplomacy? Only problem was, his story ran to twelve volumes!
Or I could do a history: a history of the computer times I had lived through. I had mountains of books and papers and clippings to help me crosscheck against the increasingly divergent recollections of others, and could certainly make a useful contribution to the unfolding story of that intelligence amplifier everybody now calls the computer. Well, it looked like a lot of work, and not much fun for me or for the readers.
For a while I thought about doing an ALICE. Lewis Carroll is a favorite preceptor for just about everybody in the computer wonderland, and the thought of Watson Senior shouting "Off with her head!" at every married IBM woman [1948] was very appealing. I could see myself saying "Good-bye, feet!" as I grew into a manager - oh, the parallels were evocative. But I had to admit the idea was pretty far-fetched.
Then I wondered about PLAYER PIANO. Kurt Vonnegut did his satire on automation so well; I had gone to GE's Association Island [1955] and had sat under the elm that was transformed into a ruined oak at the end of his yarn. As Schenectady became Ilion, so could Endicott become, say, Watsonville, and the shoe workers could parade past the IBM factory shouting "While you're thinking, we're drinking!" just as they had in real life. But the themes would have to be spare, or the satire would be blunted. What I wanted to do was to weave an exceedingly rich and complex fabric, with a warp of computer history but a filling, a woof, of wives and friends and travels. No, Vonnegut didn't look much easier than Carroll.
Think of my magnificently complicated life as a huge multi-dimensional data bank. The totality is the autobiography - millions and millions of bytes. Slice it along the time axis and you have histories: big computers, software, the evolution of standards. Slice it another way and you have applications: science, or banking, or air defense; another, and you have organizations. What I have finally chosen to do is to slice it so as to feature wonderful people: some of them famous, like the great Vons - von Braun, von Karman, von Neumann; some of them tremendously significant but not so well known, like L.J. Comrie and Wallace Eckert; some of them just vivid and exciting, and important to me because they were wonderfully human if not historic.
[-ix-] I aspire to Magic Realism - a picture somewhat larger and much clearer than life. But let me reassure readers, and future thesis perpetrators: I have not given "...to airy nothing/ A local habitation and a name." All the things in my stories actually happened. The Old Man really did order the support columns removed from the SSEC room. The television link between the jet engine test cells in Lynn and the GE 704 really did click only a few moments before the generals and the airline presidents arrived. Tom Junior really did walk through the sacred halls of Armonk carrying a whisky carton, and with a duck in it!
I have not sharpened the joys and the sorrows; there was genuine gold and genuine guilt. And when I praise Grace Hopper and Gerhard Neumann, or criticize Vin Learson, it is from the heart. I lived in vivid times.
Finally, there is no point to disclaimers. The characters in my story are real, every one of them. Some are not identified precisely. That is not an accident. And some, some of the best ones like Comrie and Watson Senior and Wallie Eckert, have gone beyond my praise. As with all the others, wives and lovers and pets, friends and enemies and quizzical bystanders, I've drawn them as I remember them, with no fear and much, much favor.
—Mies, Switzerland [1991]
In Chapter 01 you will encounter
(in order of appearance):
Watson Senior [Thomas J. Watson] the Imperial Presence
IBM [International Business Machines Corporation] the empire
The SSEC [Selective Sequence Electronic Calculator] 12,000 tubes
Betsy Stewart mistress of the SSEC console
The ENIAC [Electronic Numerical Integrator And Calculator] 18,000 tubes
The ASCC [Automatic Sequence Controlled Calculator] first of the giants
Wallace Eckert first scientist ever hired by IBM
John McPherson IBM engineering director
Pete Luhn inventor of relay calculators and much else
Pres Eckert [no relation to Wallace] builder of the ENIAC
John Mauchly designer of the ENIAC
Howard Aiken he upstaged Watson Senior at the ASCC dedication
Grace Hopper everybody's favorite naval person, and mother of COBOL
The Watson Lab [Watson Scientific Computing Laboratory] at Columbia
Rex Seeber master of the SSEC
Hilleth Thomas mathematical physicist
The NORC [Navy Ordnance Research Calculator] first supercomputer
Frank Hamilton builder of the SSEC
George Richter Gray Eminence of IBM deliveries
L.J. Comrie [Leslie John Comrie] great English table-maker and computer
Becky Jones the Moon on a Friden
John Backus father of FORTRAN
FORTRAN [FORmula TRANslation] longlived technical computer language
Ted Codd lawgiver of relational data bases
Richard Courant famous NYU mathematician
I.I. Rabi top physicist at Columbia
Dwayne Orton editor of THINK
THINK Magazine IBM prestige publication
Dorothy [Dorothy Carlson Grosch] Mount Wilson's loss was my gain
The Waldorf Astoria Hotel scene of many IBM parties
NOTE: In later lists, there will be entries without comments. Instead, I give the number of the chapter where that entity was first introduced.
[-1-] He would be 74 next month, although no one in IBM dared to mention the number to him. The family, the large and devoted family, would of course celebrate - very privately, and without emphasizing his age. In his sharp but nevertheless sentimental way he was looking forward to the dedication next week as a major present - a recognition by devoted employees of his foresight, his vigor, his determination to make "the IBM", as he called his empire, an even more important leader in American and world business.
This was Thomas John Watson, "Mr. Watson" when addressed directly, "T.J." in most third-person mentions inside the company and nearby, "The Old Man" informally and pejoratively. The use of "Watson Senior", to distinguish the emperor from the heir apparent, would not be in frequent use for another five or six years; indeed, it was really a coined label for the convenience of media people and historians rather than something used in IBM.
It was Thursday, January 22, 1948. Watson and his entourage stood inside the world's second giant electronic calculator (we say "computer" today, but more about that a little later). Three months ago the space had contained The French Bootery, a handsome and quite successful shoe shop catering to 57th Street ladies and would-be ladies. From the ground floor of the building adjoining IBM World Headquarters, a building owned or at least controlled by IBM, the Bootery had been summarily moved across the street - these things were much easier in 1947 - and the cavern remaining had been miraculously transformed, with walls of plate glass and stainless steel shielding some 12,000 full-sized vacuum tubes and 20,000 special relays: the SSEC.
Watson, his vice presidents and sycophants, and senior technical people nervously awaiting judgement, were clustered around the free-standing units on the raised floor of the lovely room. A little further back, a gaggle of less senior engineers and scientists was tensely watching the action, available to answer questions or open and close panels and doors. The giant control unit [-2-] covered with hundreds of identical but handsomely labeled switches, was "manned" by Betsy Stewart, appearing quite at ease and in command of the situation. Behind her were attractively cased line printers, and fairly close were repackaged punched card readers and gang punches, and some special units.
Questions, many questions, and answers; none about money, for T.J. assumed that others were seeing to it that enough had been provided and none wasted. Not much about the press; T.J. knew that they would flock when he beckoned. Would the demonstration impress the important men and women who had been invited for the dedication? Would the IBM creation outshine the ENIAC? How would it compare to the machine Mr. Watson had donated to ungrateful Harvard, back before electronics?
There was a pause. Those of us away from the center of Watsonian attention, standing near the inscription which offered the huge device to Science - over the famous Watson signature, of course - began to relax a bit. T.J. had not appeared to notice that the machine was not doing anything important, although those of us who had been close to him on other occasions could not be sure. He had eyes in the back of his head, and the memory of an elephant. True, he was not supposed to understand much about electronics - on the other hand, somebody had supported a good deal of experimentation before the war, and was authorizing the hiring of bright young guys from the radar shops of MIT. The recent upsurge, including the giant machine we stood inside of, was arguably the brainchild of Wallace Eckert and John McPherson. But the pre-war patents, now embodied inside the walls of the new calculator, had been funded by The Old Man.
Anyhow, no embarrassing comments. He assumed everything would be ready by next Tuesday - after all, he had ordered that it be so. We were not so sure, but we were all committed, seniors and juniors alike.
The emperor looked around one last time. The whole session, although vitally important to the fantastic future of IBM, had taken less than an hour. "A great credit to the IBM," he said, and all the vice presidential heads nodded in unison. "Science will move ahead more rapidly. Our universities and our government will surely benefit, and world peace and world trade will be fostered as people from other countries visit and perhaps use the calculator."
"There is just one thing," he said somewhat off-handedly. "The sweep of this room is hindered by those large black columns down the center. Have them removed before the ceremony." And out he marched, tailed by his quivering followers!
Well, we had four days, counting the weekend. Also - slight complication - those columns concealed the steel uprights that held up the twenty-story building. Even for Mr. Watson, no one could remove them. Now, that elephant memory was selective. Later that year I watched him order one of his cleverest inventors, Pete Luhn, to build terminals and connect them to a big relay calculator "on a lower floor", so that the "research area" would be more attractive - not bad, in 1948, and from a 74-year-old! I asked Pete later what he would do. Would he propose a hundred-man effort, abandon his current projects, and set out after The Old Man's casual dream?
"No," said he, "he'll never mention it again unless I bring it up. Of course, [-3-] if he gets mad at me about something else, even ten years from now, he'll remember he told me to do it - and that I didn't. And tell everybody nearby I disobeyed a direct order."
The cooler heads assumed that T.J. would forget the whole thing. Cynics like me figured he just wanted to keep the troops on their toes; he knew very well those columns held the building up. On the other hand, non-tech types who had watched the old boy in action reminded us newcomers that he might use noncompliance as an excuse to erase some poor soul who had "failed" him before, perhaps - looking hard at me - by omitting to invite someone to one of Mr. Watson's celebrations (I had just done the scientific invitations for the dedication).
Inspiration! Wasn't mine; I seem to remember a chap named Macnamara, who was doing the goodies: programs and table favors and souvenirs for the luncheon, and discreet corsages for the two or three lady guests. There was to be a fancy four-color brochure about the machine, and it was to feature a big centerfold view of the main area. We junked the whole stack (how I now wish I had had a copy!) and, over a frantic weekend, printed a revised edition. Alas, it had to be in sepia; even for IBM the colored option was out of the question. The centerfold photograph was carefully retouched, and all traces of the offending columns were removed.
I loved it; the semantics boys were right - there is Reality, and then there are representations of reality, and layer upon layer of more remote imagery. The columns were still there on opening day, and they stayed there for the life of the machine, and on into the golden days when a 701 lived there. A major Saturday Evening Post story published later in the year has a fine colored illustration of the area, with the columns front and center. But the history books show a beautiful room without columns - the room Watson ordered!
What was I doing in the back of that room, and why had I been the one to work up the scientific invitation list for the unveiling of the SSEC, the Selective Sequence Electronic Calculator? It was the third giant machine of the Babbage line, and like the first, the ill-fated Automatic Sequence Controlled Calculator which Watson had had built to Howard Aiken's specifications four years before, it was to be dedicated to the greater glory of Science and - unlike the ASCC, now universally called the Harvard Mark I - to the glory also of the International Business Machines Corporation. The second giant, and the first to use primitive Forties electronics, had been the ENIAC, cobbled together by Pres Eckert and John Mauchly in the Moore School of Electrical Engineering of the University of Pennsylvania, in dumb downtown Philadelphia.
ENIAC added nothing at all to IBM prestige, although the builders had reluctantly been forced to use IBM gang punches as input/output devices. What it had done, by appearing on the front page of the New York Times - the first such device ever to do so - was to make The Old Man grind his teeth so ferociously that people, dozens of people, shivered as far away as Endicott, home of the IBM "research" laboratory.
I was the second scientist ever to be hired by IBM, and an anomaly in a dozen other ways. In later chapters I'll tell the details of how it happened; the important points are that I had been a computer since the autumn of 1935, [-4-] was an authentic member of the same tribe of scientists as Wallace Eckert, and had responded instantly to the announcement in SCIENCE that he, Eckert - no relation to Pres Eckert, by the way - had been hired in March 1945 to head a new Department of Pure Science in IBM, and to establish the Watson Scientific Computing Laboratory at Columbia University.
Remember the name ENIAC? That acronym stood for Electronic Numerical Integrator And Calculator. Almost every current article about that incredible machine, with its 18,000 highly-fallible vacuum tubes, and the thousandfold increase in speed it brought to major calculations, says "Computer". But it wasn't so; from the mid-Nineteenth century until about 1952, a computer was a human being who did computations (and associated technical tasks in laboratories and observatories). He - or more often, she (I married one) - used printed math tables and calculating devices. The people who didn't do it full time, or who used slide rules and planimeters and such, usually didn't use the label. I did, and was quite proud to have been made a computer by a more experienced member of the clan, in the mid-Thirties. All the Cs in ENIAC and EDVAC and EDSAC and UNIVAC stood for Calculator, which is what we computers used, along with Marchants and Fridens and Monroes and Brunsvigas, until the Defense Calculator became the IBM Type 701.
Eckert hired me in May of 1945, and the Watson Lab had been under way for some months - and both the European and the Pacific wars deliriously concluded - before the ENIAC announcement. Although I had been busy shaping up and running the computing side of the enterprise, I had helped Wallace interview his second senior employee, a relieved refugee from Howard Aiken's sweat shop. He was one of the two or three people in the whole world who had managed a giant machine, the one at Harvard. Aiken, the irascible boss, and Grace Hopper, nowadays everybody's favorite computer person but then known principally as the co-author of the handsome Harvard Mark I operating manual, were his only counterparts.
Eckert wanted Rex - Robert R. Seeber - because he expected to build another giant. Watson had told him IBM intended to do so, and had indicated vigorously how unhappy he was with Aiken and Harvard. Now, Eckert had good lines into MIT, and although not himself at ease with even old-fashioned electronics, knew that the new radar techniques developed so quickly there (and in Britain, and at the Bell Labs, neither of which were readily accessible to Watson and his business-machines-oriented crew) would be the building blocks of future calculators.
In parallel to his recruiting Seeber, and IBM's third scientist, Hilleth Thomas of Thomas-Fermi statistics fame, Eckert set out to pick up three or four youngsters, hot off the radar griddle, to work on long range projects at the Watson Lab rather than in IBM shops in Endicott and Poughkeepsie. He was not at all an empire builder - it would have been good for IBM, and certainly for me, if he had been. He simply planned to stockpile some advanced technology against the time when Mr. Watson might need it. When ENIAC hit the front pages, he was ready with Seeber, but the pulse-electronics boys were not tuned up, and the SSEC had to be built with big bottles and old circuits. Out of that tiny specialized group finally came NORC, the Naval [-5-] Ordnance Research Calculator - the very first supercomputer. It just took a while!
Watson pushed the "start" button. He picked Frank Hamilton, who had been Number Two on the ASCC. "Put every IBM resource to work, Mr. Hamilton. I want the new machine to be faster than that one in Pennsylvania, more capable than the one we gave to Harvard. I want it to be installed here at World Headquarters, in one year. And I want it to be available to the scientific world, not hidden away at a selfish university or at a military installation."
Frank realized how horrendous the task would be. But he was a big, handsome, forceful engineer - a natural leader, with a team in place in the Endicott laboratories and a clear idea of what other projects he could dip into. It was a great challenge. Besides, you did not say no to the boss - not when he gave you an, ahem, opportunity like that.
Although I did not yet see it clearly, this was my first exposure to the IBM Contract. During good behavior, you had a life job, at non-decreasing pay, and with great perquisites, ranging from membership in lovely dollar-a-year country clubs to suites on the Queen Mary, provided:
Good behavior included not growing hair on your face, not wearing colored shirts, not taking a drink in public (one very senior executive, who frequently occupied those Queen Mary suites, was to die two years later from a hobnailed liver), and never contradicting The Old Man or his direct representatives. I did all four in my first two years in the company, and for a while lived to tell the tale. What got to me in the end was c); I kissed the whip very poorly indeed.
Hamilton and Eckert and Seeber settled the specs in just a few days. There had been a series of memos and meetings in 1945, and not just desultory ones. John McPherson, then Director of Engineering and approximately Eckert's boss, had made heavy contributions - intellectual and technical ones, not managerial; you will see much more of him in later chapters. The old-timers who had built the Harvard machine and who held the primitive but valuable vacuum-tube-circuit patents had joined in. What was lacking was money and people and priorities. Watson supplied them.
The SSEC was built in Endicott, under the toughest kind of forced draft, out of standard pieces and simple circuitry. Only IBM could have done it. No other outfit had stuff on the shelf - well, except the phone company, and it was too hidebound; had no Watson, and wanted none. By today's standards the cost was low; not much over a million dollars. But T.J. brooked no nonsense from beancounters. There was a story in the trade that when Claude Shannon and his lovely wife were moonlighting at the Murray Hill Bell Labs, and "borrowing" components from the stockroom to build a maze mouse, accountants disguised as [-6-] janitors were recording every peculation and taking notes for the patent attorneys besides.
None of that in 1946/47 IBM. The million dollars or so was off the top; the myriad services which Hamilton drew on from all the rest of the company, ranging from the time of Eckert and McPherson down to that of the Endicott janitors, were never charged. No one knows - in fact, no one ever wasted time trying to find out - what share of IBM's total resources over say 18 months went into the SSEC. I'd bet it was four or five per cent, at a time when total annual sales worldwide were well under $200 million. Ah, Watson!
I said McPherson was approximately Wallace's boss. There were no organization charts in IBM, and if there were any in 1958 when I was invited back to the much larger company - T.J. died in 1956 - I never saw them. Tasks in the Forties came almost entirely from Mr. Watson. If John had needed to force something on Wallace - and he almost certainly never did, at least for many years - he would have had to somehow arrange that his needs reached T.J. The latter, if he agreed, would have given Eckert the word, and in no uncertain terms. The idea of formal, announced channels of authority so prevalent in General Electric in the Fifties when I worked there, was unheard of in IBM. Mr. Watson ran everything.
By extension, there was no budget. Oh, there were accountants; there were taxes due in many countries, an annual report to produce, invoices to send and bills to pay. And some very strange Watson charities and beneficences, we understood. But in general, people downstairs proposed and Mr. Watson disposed. You - well, not me, but Eckert for instance - you sat outside The Old Man's office until someone shoehorned you in between the man from the Metropolitan (Opera or Museum; he did both) and the regional manager he was "promoting" to run the Anchorage office. You asked briskly for what you needed, answered amazingly pointed questions, and got your money or your refusal, or something in between. Very quickly!
I have often said Watson kept the money in a shoebox, and handed it out to petitioners as he saw fit. That's amusing, and somewhat true, but it doesn't allow for two things: first, the box never ran empty; he had a magnificent feel for the inflows as well as a running score on recent disbursements. Second, there were secret forces at work, like those Shannon-watchers; he built the French plant about that time, and there must have been many cross checks. He could expand Poughkeepsie off the cuff, but something more sophisticated must have been coming over the wires from World Trade in Europe. And he had more than sources of information; he had hidden controls. Watson could speed up or slow down the factories, and deliveries, and hence rental income - there was always an eager waiting line of customers, especially at war end. All it took was a short call or note to George Richter, the Gray Eminence Of Scheduling, two floors away.
Even in the Forties, IBM was something special. Small compared to, say, General Electric, it was already recognized for its amazing growth, its cash flows (for years to come, the punched card machines and the early computers would only be rented), its lovely profits, and above all, its Maximum Leader. Watson was into everything: not just the opera and the art world, but as the most powerful trustee of Columbia, the head of the International Chamber of Commerce, and in a hundred power-related civic tasks and charities. A whole floor of World Headquarters, "590" [590 Madison Avenue], could barely hold the secretaries and assistants that surrounded this power source. Where an AT&T or an RCA would have had to go through carefully announced organizational maneuvers, and much-discussed budget reallocations, [-7-] to start up a Hamilton group, T.J. did it in minutes. And with the Endicott and World Headquarters grapevine carrying the word about Watson's heavy support, Hamilton in turn could cut through personnel and procurement problems almost instantly. And these were serious; remember, this was 1946 - even IBM was having trouble with people and supplies. Frank's engineers had expediters up and down Cortlandt Street looking for surplus electronic gear, and the major factor in choosing vacuum tube types for the buffer memory and arithmetic unit of the SSEC was availability ("We need ten thousand right now, Sol, and they gotta work!").
In parallel with the electronic and mechanical work - system design, we call it now - the physical layout and handsome presentation of the huge machine had to go ahead. As soon as The French Bootery was known to be extrudable, plans for a big power supply and a unique air conditioning system were set in train; each unprecedented in central Manhattan and in IBM experience. No replays would be possible.
The machine took form in Endicott. A year, in spite of the Watsonian directive, was not enough. But in that year, all of the electronics, the sixteen-foot-high racks of special Lake relays, and the incredible tape units, were designed and built and tested. "The Soul of a New Machine" tells about a tiny child's circus, and with thirty years of industry experience to draw on; this was center ring in The Big Top: "Ladies and gentlemen, a feat never before attempted on any continent!!".
On 57th Street, the heavy machinery went in. Also there was to be a raised floor - another world first. Watson would not have his guests, his customers, his stockholders tripping over snarls of cable. The inscription on the wall had to be written, by Wallace Eckert I think, and approved (and quite possibly changed) by The Old Man, and engraved, and approved again. But that, as we all know today, is only half the story. How about the software?
No problem about languages - there weren't any yet. Whatever problem we chose would be written, and debugged, and run, in machine language, and in absolute: two four-address instructions per line of input tape. Eckert and I talked it over: one very short meeting. It would be from astronomy, of course; that was our trade. The very first automated scientific calculations in the world had been done by the great L.J. Comrie at Greenwich in the early Thirties, and had drawn both of us into the game. And if we redid them, more carefully, and from the original math rather than from intermediate tables, as Comrie had had to do with his rented Hollerith equipment, the output would have actual immediate value to the national almanac offices around the world. So - the SSEC would calculate the positions of the Moon!
Eckert had insisted from the beginning on a strong table-lookup capability: a design offshoot of the big tape drives, as it turned out. Its capabilities were set. For the demonstration problem, each date put in would require looking up 1870 sines and cosines, to eight-figure accuracy. I had begun to develop a powerful method for minimizing the size of such tables (later it came into heavy use among operators of the first IBM mass-production, um, minicomputer, which had only a few dozen words of storage - and the technique disappeared completely a year or two later, overtaken by Big Memory).
[-8-] The method wasn't quite ready, but no matter; off I went to design the table anyhow. Like everything else, it had to be ready on time - or else! Youngsters in corners at World Headquarters were feverishly writing code, without knowing the first thing about lunar theory. That too had to be finished and checked out by opening day. Eckert was the authority (I had a substantial computing shop to run at the Watson Lab, and classes to teach - and some of the youngsters were in them). Somehow it got put together.
It is worth noting that the check calculation, done on a Friden desk calculator by Eckert's assistant Becky Jones, took more than three months - for one date! And even then, we did not have time to exactly duplicate the table lookup and the arithmetic operations to the full precision of the machine, which was fourteen decimal digits; other tests of the pertinent units, and my off-line experiments on the sine/cosine table, had to suffice. I often wondered in later years, when lunar landers were putting down on an object whose coordinates came out of an ephemeris based on those 1947 programs, whether anyone had ever gone back and checked out the less-significant digits!
Incidentally, among the youngsters were John Backus, the father of FORTRAN, and Ted Codd, the father of relational databases.
The programs, and data, and my table material, were punched at the Watson Lab on standard IBM cards. These were converted to input tapes on a free-standing punch called The Prancing Stallion (from its profile), which later was put in its final position on the SSEC raised floor. Some program testing was done under Rex Seeber while the machine was partly assembled in Endicott, but most of it had to be deferred until just before the ceremony. This partially explains why I was Cool-Hand Luke years later at GE; after that kind of pressure, the jet engine business looked pretty peaceful!
About those tapes: the card plant in Endicott got enormous rolls of card stock from the paper mills. For regular card manufacturing they slit the rolls to three-inch width (card height). For the SSEC they furnished rolls eight inches wide (card length). The resulting rolls weighed 400 pounds, and had to be hoisted onto the SSEC with a thoughtfully-provided chain fall! For the Stallion, we pushed the rolls up a ramp.
The punch stations, slightly modified from standard IBM reproducer components, punched two round sprocket holes at the edges, and 78(!) regular IBM rectangular holes in between. The sprockets drove the tape one line at a time, and drives under separate program control fed the fresh or pre-punched tape under ten 78-brush reading stations. The tapes hanging down could lengthen and shorten, and for program tapes and the table lookup unit we cemented the tape end-to-end into short loops (yes, someone had had to provide the jig). There were three of these monsters at the end of the machine room.
Up to 36 of the fixed-length tape loops could be mounted on the separate table-lookup unit, which in later years was also sometimes used for program reading. For the lunar calculation, I used 24 loops to make lookup time as short as possible, and we got programming from the main tape readers.
Behind the scenes were racks and racks and racks of superfast small relays, which were used for intermediate storage. The tapes were slow [-9-] storage, and of course we had infinite storage on standard punched card decks, which could be read by the SSEC directly (and punched). The two printers behind the operator had hidden plugboards for producing handsome output formats.
On the famous Thursday when The Old Man ordered us to remove the offending columns, Rex and his crew had had all of these units buzzing and punching and looping. But the lunar program had not yet run. I was smugly aware that the sine/cosine table was checked out and working, so I was somewhat relaxed. But the strains were literally palpable. It was to be a long weekend!
Two months before, Hamilton had committed to an opening date: Tuesday January 27, 1948. He had done his fantastic task - designed and built the huge system, assembled and partially tested it in Endicott, torn it down and shipped it to Manhattan, fended off the unions who demanded that non-union IBM let them reassemble the monster, provided the support systems, and prettied up the joint - all in a little over 16 months. And, to jump ahead a little, on opening day, everything ran! Yes, the hardware, and the program, and the air conditioning, and the people - one of the very greatest feats of nearly a century of data crunching!
On the great day, the multitude assembled. It was not the first major Watson charivari I had attended, but it was a first for Rex and his senior crew. Everyone looked very cool and confident: T.J. himself because he assumed everything was as ordered, the rest of us because he was watching. The scientific contingent turned out in full force; I hosted one table at lunch, and landed Richard Courant, whom I greatly admired and had never met. I unblushingly told him I had done the invitation list - well, I had proposed, even if Wallace and John McPherson and a whole floorful of Watson aides had disposed. He was more interested in my stories about the punched card work we had done for Los Alamos than in the SSEC; like most of the seniors present he assumed that there would never be more than two or three machines like the one being dedicated, and that only very special science would ever be allowed near them. Today the NYU center named after Courant mounts several hundred thousand times as much computing power as the SSEC, and complains about the constriction!
Mr. Watson - you couldn't call him anything else at that moment - made a sententious little speech and posed for photographs with guests, including Rabi of Columbia and other scientists. One could often figure out who wrote the draft of a Watson speech; I could hear Eckert in this one, but the overall effect was preachier - probably from Dwayne Orton, editor of THINK. They had done the internal photos, the poses with Hamilton and Eckert and Seeber, the previous Thursday; this was a day for his guests and personal friends; another Watson triumph. He had many every year.
Six months later my wife Dorothy and I received a telegram, delivered to our apartment in Chelsea, although addressed to the Watson Lab at Columbia (T.J. had special relations with Western Union, too!) It invited us to an IBM Family Weekend, at the Waldorf Astoria no less, to thank the men who had worked so hard on the magnificent Selective Sequence Electronic Calculator, [-10-] now at work on important scientific problems at IBM World Headquarters - and especially to thank their wives, who had been so supportive.
Turned out that the last time The Old Man had done this (also without warning), he had organized a private train and taken a hive of senior executives and their wives across country to see the IBM exhibit at the San Francisco World's Fair. It had been a very dry trip, even with Pullman porters fanning out at every stop looking for liquid relief. This event had benefited from that experience; the Waldorf had been found to list bar tabs and such as "restaurant" on the final bill!
You can imagine the scurry for babysitters and new hats in Endicott. Well, come the great day Mr. and Mrs. Watson had a little luncheon for us (and a hundred or so Headquarters executives and their wives) in the Wedgewood Room, and told the honored if somewhat flustered ladies that the weekend was theirs. "Go to restaurants and theaters and the opera, as IBM's guests. Enjoy the shops, and the spectacle of New York. We at World Headquarters thank you for the support you and your families gave your husbands, which in turn enabled them to produce this spectacular machine." There were tours of the SSEC, by the way, which many of the wives had not seen - a nice touch.
Asides: of well over two hundred ladies present at the luncheon, only two, my wife Dorothy and Hilleth Thomas' wife Naomi, did not wear hats. Hmmmf! Intellectuals! And the next night the Watson Lab couples, who after all lived in the city, had a fancy dinner together, in the hotel. As a matter of convenience, sitting at one end of the table, I signed the check: about $200 (it would be ten times that today). Six weeks later Eckert got a phone call from a hushed-voice little beancounter at 590. "About this, ah, rather large restaurant charge..." Eckert told him it was quite all right; he had been in the party. There were accountants in IBM; they just didn't come out of hiding very often!
Watson got his investments back many-fold: the price of the SSEC, and the price of that Family Weekend. The one placed him in front in the Big Science sweepstakes, at a time when a dozen larger hi-tech companies should have towered over him. The other confirmed the feeling in two hundred engineering and executive families that Mr. Watson's IBM was a wonderful place to belong to. It was a good time to be an emperor.
In Chapter 02 you will encounter
(in order of appearance):
The Naval Observatory where Eckert did the Air Almanac
Pearl Harbor you could see the smoke from the Japanese Embassy garden
Dorothy 01
NBS [The National Bureau of Standards] still on Conn Avenue in D.C.
William Meggers NBS spectrographer
Uranium undoubtedly for the Stagg Field pile
The Watson Lab 01
The American Air Almanac to be done on IBM machinery
John Willis officemate at the Naval Observatory
The Ritchey-Chrétien [reflecting telescope] notorious plate breaker
J VIII [Jupiter's eighth satellite] the lost is found - and a Ph.D. too
Harvard summer conferences optical design skills, and a fiancée
Leland Cunningham my opposite number at Harvard
Aberdeen Proving Ground where Cunningham did his war work
Army Air Corps geodetic surveying in Alaska
BuOrd [Navy Bureau of Ordnance] exit pupils and boresights
Stanley Ballard he traded up to Re4e from pineapples
American Institute of Physics optics and a lot more
Russell Banker a link to the pre-war Gun Factory
MIT Radiation Lab the word "radar" was still secret
Japanese optics captured gunsights were different from Zeiss'
NDRC [National Defense Research Committee] a "torrent of novelty"
The Alabama [new battleship] made us all forget BuOrd bureaucracy
Doc Draper [C. Stark Draper] his gunsight gave 20mm cannons more punch
Keuffel & Esser stereoscopic rangefinders from Hoboken
OSA [Optical Society of America] my second professional society
Foldboating Dorothy and I preferred it to spelunking
Sperry [Sperry Gyroscope Company] mechanism yes, radar yes, optics no
Suzy [cat] many litters in a Hempstead coal bin
Interchemical Corporation spectrography and glossy black paint
The B-29 [bomber under design] fancy optics instead of a tail gunner
Eastman Kodak my first view of industrial research and development
Rudolf Kingslake Kodak's master lens designer
Harold Bennett wanted to automate ray tracing - on a Friden!
The Mark 14 Draper's gunsight went into mass production at Lake Success
George Bentley thousands of Mark 14s for the Pacific
Jack Varick [Sperry buyer] "it took more than scientists to win a war"
Perkin-Elmer from astronomical telescopes to war work
The Mark 15 [gun director] we put optics and radar into a full box
C.L. Farrand his factory needed a second optical designer
Sy Rosin Farrand Optical Company designer
[-13-] There was of course no such thing in World War II as a civilian draft; at least, not in the United States. There was a military draft, and millions of youngsters had been involved with it. A fair percentage escaped from the process. I was one. Yet when I talk in the next chapter about how I came to IBM and the Watson Lab, and overnight to be in complete charge of one of the most powerful computing installations in the world, I usually say,"I was drafted" - that is, drafted into a civilian job.
I clung to the Navy for a while after the war started. I was at the U.S. Naval Observatory, just up the street from the Japanese Embassy on Massachusetts Avenue in Washington, at Pearl Harbor time. My wife Dorothy, having helped me finish my doctoral dissertation, and fired with California patriotism - a somewhat more vigorous form than the District of Columbia variety - had drawn on her Mount Wilson connections to obtain a laboratory job at the Bureau of Standards. We were surrounded by the excitements of the time.
Dorothy's work turned out three years later to have been involved with the atomic projects, although we didn't know it then. Her boss was a famous spectrographer, William Meggers, who was an authority in metals analysis, and Dorothy found herself obtaining and measuring (she was new to the former but very experienced at the latter) spark spectrograms of the rare element uranium. We talked about it freely; there was no security, no hint of anything mysterious - except that, aside from very specialized uses such as coloring glass, and from its unusual density, uranium wasn't exactly an everyday market item.
Both of us, as astronomers, knew of the Hahn-Meitner experiments on the fission of uranium under laboratory conditions, and I had been an avid reader of science fiction for over thirteen years, so we wondered idly about whether there might be a connection. But neither of us had been close to that kind of physics. I was doing the driest and most remote kind of celestial mechanics; [-14-] she had been doing observational astrophysics at Mount Wilson. We didn't have day to day contact with the quantum mechanics people, although I had had courses from them (Goudsmit, for one) two or three years back.
Incidentally, that kind of security, out of fashion for decades, is very effective - no obvious security at all, that is. What you keep secret is that there is anything to keep secret. If Dorothy had had to take a lie detector test, and list all of her addresses from conception, and all her Swedish third cousins, we would have known something important was up. As it was, we didn't pick at it, and tended to talk about her problems in preparing samples!
A great deal of the work at the Naval Observatory had value in the war effort. There was the time service, obviously, and the continuing annual preparation of the astronomical ephemeris and marine navigation almanac. Right across the hall from me was the first permanent rent-paying scientific IBM installation, hard at work on the brand new Air Almanac. But these activities tended to be staffed by older men and women, and such younger men as were involved were much married and bechildered. Also, turnover in those barely-post-depression days was low. It didn't look as though I could offer help in those areas. Of course, the more purely astronomical tasks such as my own were going to continue; the observatory had weathered earlier wars and intended to keep its key researches alive. An astronomical observation is like a modern airline seat; you have to take it when available - tomorrow will be too late.
I was rather surprised, for instance, to discover that some solar observations that were collected in my division were routinely being sent to Italy for international compilation. Italy was at war, so the stuff had to be routed through Switzerland. But science was to be served!
I had some ideas about improving the reflecting telescope on the grounds. Because of the bright lights of the city, completely surrounding the minor hill on which the Naval Observatory stood, not much "modern" (i.e. astrophysical) use had been made of this strange beast. In fact, when I arrived it was used only a fraction of a night per week, as against every clear night for the huge old fashioned refractor - the one with which the tiny satellites of Mars had been discovered many decades ago. As an important part of my thesis calculations, I did a prediction of where to look near Jupiter for its eighth, extremely faint, satellite. John Willis, my office mate, had said there was no chance of photographing it locally. But when the city lights were shut down or dimmed as a wartime precaution - Japanese submarines coming up the Potomac, no doubt - we decided to give it a try. And there, right in the middle of the big photographic plate, after a ninety-minute exposure that could not have been made six months before, was my Ph.D. insurance, J VIII, later christened Pasiphae to please the space amateurs and the Carl Sagans.
This gave me the unpleasant experience of using the reflector, called a Ritchey-Chrétien type. The problem was that you had to use very thin glass plates, and bend them slightly to a spherical contour in special plate holders. Many broke in the process, and a few even broke afterwards when they were being released for development, wasting the human and telescope time invested. I was fortunate; nevertheless, I had held my breath.
[-15-] It occurred to me to use my skills in computing higher optical aberrations, which I had acquired (along with a fiancée and a first taste for fancy Chinese food) at a summer course at Harvard in 1940. I would see if I could design a field-flattening lens so that ordinary photographic plates and plate holders could replace the special ones. Willis and other senior figures were startled at the idea. But no special permissions were required in those days; you just worked longer hours so as to add the new project to the assigned ones.
While I was happily at work on the idea, I got drawn into a typical informal just-after-Pearl-Harbor project on improving the maps of Alaska. The Army Air Force command involved needed to locate ground stations with geodetic accuracy. They wanted to use a very carefully levelled zenith camera, and were asking us astronomers to help.
I couldn't do much better for them than a minute of arc - say a mile on the ground. The officers were impressed, but we agreed it wouldn't be good enough. I suggested that if they took two exposures on each film, with the camera turned about the vertical axis in between, much less accurate measurements would show where the center of the field was. They looked unhappy, presumably because of the time needed to modify the cameras yet again, and went away.
I had had fun. I had been caught up in the Real War for a few weeks. The fly boys had come to me because of my optical knowledge. Could I use it elsewhere? Not at the observatory, probably. I cast about. In retrospect, I should have looked farther afield, and certainly taken more time. My opposite number at Harvard, a strange plump creature named Leland Cunningham, ended up running an IBM shop at Aberdeen Proving Ground and writing the sample-problem specs for ENIAC. But it took him a year or so to get the job, and I wanted to do something right away - a trait that will surface many times in these stories!
Also Dorothy was doing very well indeed in her new job at the Bureau of Standards, and having uprooted her only the year before from lovely Pasadena, I wanted to stay in Washington if I could. Simple inquiries revealed that the Navy had need of optical experts in the Bureau of Ordnance. Presto! I was a P-2 Assistant Physicist in BuOrd Re4, Fire Control Research and Development. And at a pleasant salary increase besides: $2600 a year.
Doesn't sound like much today, but on the strength of such wealth Dorothy and I bought our first house. It was brand new, located in Kensington, Maryland, on an unpaved street - and cost $8500. It is probably still there (they paved the street after the war), and I'll bet it now goes for ten times our price.
Most of BuOrd was housed in horrid temporary structures left over from World War I, and I was put to work in a gigantic drafting room at the very back of the buildings. My routine, six days a week of course, was ten minutes walk to the Kensington Jitney, twenty minutes ride to Chevy Chase Circle, thirty minutes more downtown. Plus connecting waits, plus the Ordnance Amble from the front door and the guards to the sign-up sheet; call it an hour and a half. Dorothy was in a five-way car pool; gasoline rationing had begun.
I worked for a civilian named Mike Goldberg, an expert on mechanical linkages, a curious component of the analog gun directors that are now one [-16-] with the dinosaur and the dodo. He passed on assignments from the young Re4e officers, who worked in somewhat more pleasant quarters down the hall. Their boss was a two-striper named Stanley Ballard, who had been an insignificant physics professor in Hawaii when the war started. He fit the job well, and afterwards used his connections to rise quite high in the American Institute of Physics.
Another physics type was Urner Liddell, who was seconded much later to work on the postwar nuclear tests in the Pacific. And there were several others, all quite serious about their uniforms but none very knowledgeable about old-fashioned lens-and-mirror optics, which was a backwater of 1942 physics in the same way that my orbit computing specialty had been a backwater of 1936 astronomy. So I was welcome, although on my side I knew too much about the subject rather than too little. That is, I knew about higher aberrations of fancy astronomical telescopes and Schmidt cameras, and not much about gunsights!
Anyhow, Goldberg sat me down next to a career civil servant, an engineer from the Naval Gun Factory named Russell Banker, who knew plenty about gunsights but very little about theory. We made a good team, and in only a few days I knew about parallax and exit pupils and such. Also, Re4f (next door to Re4e) turned out to be the boys who did radar, so secret still that the name was not used in public, and through them I first heard of the gigantic buildup at MIT, which was for disguise called the "Radiation Laboratory". So there were lots of new work tools on the one hand, and war vistas far beyond those at the Naval Observatory, on the other.
About the difference between gunsight optics and fancy aberration theory: when I came up for my thesis defense in Ann Arbor, a few weeks after moving to BuOrd, I tried to explain to the physics prof on my committee what I was doing. He was, by coincidence, the optics man at Michigan, and was so put off by my plebeian use of his elegant science that he sat on the suggestion I get a cum laude for recovering J VIII. Sigh!
I had a nice Friden calculator - such gadgets were already under wartime allocation - and was put to work on donated British and captured Japanese gunsights. The latter were tested and disassembled at the Gun Factory, the lens parameters measured, and the dope sent over to Re4e for analysis and comment. I was startled to be told that, while Japanese optics were supposedly copied from the Germans, the Navy patent files showed nothing similar from Zeiss or thereabouts. This was the first indication I personally had had that Japanese technology was not just duplication and adaptation. It was confirmed several times for me during the war; first hand in optical engineering, anecdotally in nearby fields like fire control and radio. So I was not as surprised as the media and the public when the Japanese cameras and consumer electronics came on strong two decades later.
The load increased. Somebody produced an assistant for me, a young New Yorker who yearned audibly for the fleshpots of the Grand Concourse but found working in BuOrd preferable to boot camp. Indeed, after I left Washington things got tighter in the Bronx draft boards, and they took him away. His output was low, and supervising him took more time than he saved me - another lesson to store away for Watson Lab futures. But there was [-17-] indeed some small pleasure in command, or at least I found it an interesting departure from being commanded.
The pressures of a Real War were fascinating. Major projects, considerable excursions from old ways of doing things, could be set up in days. I was accustomed as a budding scientist to "keeping up with the literature", so the flood of reports and recommendations and proposals didn't faze me the way it did the older civil servants like Banker and the drafting room types. I began to realize that having young physicists in the Re4 offices was working better than their lack of practical experience might have promised.
For instance, there was a horrendous shortage of fancy rangefinders, and of experienced operators on shipboard. Some years back a decision had been made to adopt stereo rangefinders rather than split-field. In a peacetime service the Navy could take the time to select and train sailors with keen depth perception. Now they needed ten times as many, instanter. Re4e beat the bushes for psychologists, and also for firms to design and build training gadgets. The bright people under Ned Land at Polaroid were among those who surfaced.
I wasn't directly involved, but had lots of ideas and advice to offer anyhow - much of it unwelcome, because there was so little time to argue. Decide! Act! Try something else; we just lost another carrier! In my drafting room men were worrying about the torpedoes that unexpectedly failed to work. Was it the fire control equipment they had designed, or manufacturing problems, or poor training, or something unforeseen? There were horror stories from the European theaters too. I still remember one about the escape hatches on one type of bomber being just a little too small for a gunner with his parachute to squeeze through. Solution: select very small gunners!
There was a weird project to increase the accuracy of those stereo rangefinders by filling them with helium instead of dry nitrogen, and that involved a Princeton professor named Merrill Flood who surfaced thirty years later in Management Information Systems - another helium-filled instrument! The NDRC, National Defence Research Committee, under the aegis of Vannevar Bush, produced such projects, and academics to staff them, in an amazing torrent of novelty. But I learned little about radar, and heard nothing about atomic weapons.
With all this excitement and hard work, there were still strange interludes. The battleship Alabama was commissioned and before going off to first station, anchored in Chesapeake Bay for a week or so. Annapolis got a day, and us Navy civilians got a day; we were bussed over, put aboard via glamorous launches, and allowed to clamber all over the huge ship. My gang looked at fire control gear: gun turrets, gun directors, rangefinders, radar equipment. Others looked at propulsion, or ammunition hoists, or whatever. Great idea; it charged us all up for weeks; we even quit complaining about the (lack of) air conditioning!
Another: one day a very academic type from the University of Chicago showed up at BuOrd and walked off with a best-design submarine periscope. Since periscopes were even scarcer right then than submarines, we were all flabbergasted. Of course, three years later I realized he had been outfitting [-18-] the Stagg Field pile, for Fermi and Co.! It was the second clue Dorothy and I had had about the bomb, and we missed it as we missed the first.
Antiaircraft guns were increasingly important in the Pacific, although kamikaze attacks were still rare. These cannon were then directed by gunsights attached to the guns. There were few skilled gunners; the old open sights that we still see in World War II movies didn't let novices knock down many attackers. So, technology - in fact, Massachusetts Institute of Technology - came to the rescue. Doc Draper, whose big postwar enterprise was to draw student and flowerpower protest in Cambridge in the Sixties, was designing an analog computer, later to be called the Mark 14 Gunsight, to replace the low-tech ring sights. The latter cost maybe $50; the Mark 14s, even in huge quantity, cost nearly $10,000 - but the warships the kamikazes were damaging cost a lot more!
I came in contact with that effort in a roundabout way. The 20mm cannon involved had to be boresighted on shipboard. That is, you had to make sure the cannon and the gunsight were lined up. The sight was no problem, but when a gunner looked through the bore of this skinny little gun his field of view was so tiny he often wasted much time finding the boresighting target. I had the idea of putting a plug with a negative lens in it, in the gun muzzle, and another, with a positive lens, and a mirror to make things convenient, in the breech. A reverse telephoto or reverse opera glass, in effect; what it did was increase the field of view fourfold. The demagnification of the target didn't matter; if anybody cared, a bigger one could be used. Well, much to my pleasure, a small company was found to make the little kit, and in some quantity. I still have the drawings, but nobody ever sent me a sample. Ah, wartime!
This sounds like a lifetime. Actually, I had come down to BuOrd from the Observatory in May [1942], had gotten my Ph.D. in June, and all the excitements I've been reciting had taken about six months. Some of it was the war, some of it was biological; the clock runs differently when you are 24 - and the gonads too, for sure! I did some travelling; notably, flew up to the rangefinder production operation at Keuffel & Esser in Hoboken, in a two-seater Navy trainer with a nice Re5 flyboy who did barrel rolls until I informed him grimly that I was about to throw up all over his airplane! I joined the OSA, Optical Society of America, and because I had lots of surplus energy - and it didn't hurt any that I was sponsored by Dorothy's boss Meggers, an OSA bigwig - began rising in that hitherto-conservative outfit.
Also at the Bureau of Standards Dorothy had encountered an unusual outdoor group. They called themselves spelunkers (cavers, in Britain). They specialized in exploring Schoolhouse Cave in nearby West Virginia, longer drives being impossible under gas rationing. We were not much into caves - darkness, and narrow places, and a lot of mud - but two dozen of the bunch were foldboaters. They had learned about the sport on pre-Hitler European travels, brought the folding rubber-skinned kayaks back with them and stored them in their Washington closets, and went out on the upper Potomac and such, when the water was white and the weather not too grim. I'll have more to tell about the sport, which was the first of many high-thrill adventures I have enjoyed after emerging from my soft [-19-] childhood; anyhow, Dorothy and I bought one of the last prewar foldboats available, and tried to learn by doing.
There wasn't much other social life, especially for newcomers. The Naval Observatory crew was pretty inbred, and the BuOrd civilians were the same. Ballard and his young officers were learning to be Navy - Navy Reserve, we said cuttingly - and had special privileges we mortals lacked. The Meggers family had a big house out toward the Bureau, which was then on Connecticut Avenue, and invited us several times. They ran to collections of many kinds, and indeed in postwar the oldest daughter became a well-known anthropologist, which is sort of people-collecting. We liked them and admired Meggers himself very much; he was a cross between a senior scientist and a Will Rogers, and had a comic feud going with George Harrison of MIT, the world-famous designer of ruling engines (on which the diffraction gratings for Dorothy's spectroscopes were made). Dorothy's part of NBS, in fact, looked very much like the Harvard astronomy milieu out of which I had plucked my optical theory and my marriage.
We were isolated, besides. I had been a part of the Ann Arbor scene, and Dorothy, of the Cal Tech/Mount Wilson scene, for six years. Washington was very different, and especially in wartime. We had a house, and problems with furniture, and not much money, and our personal relationships to work out. So what time was left over from pretty heavy work commitments got used up fast. Neither of us were family-dependent; still, it didn't help any that mine were in Michigan and Dorothy's in California.
I was restless. I had enjoyed the little burst of creativity while I was doing the boresight design. And I was more than a little jealous of Navy officer privilege; you didn't actually have to bow when a four-striper swept down the hall, but you felt that nobody would be surprised if you did! The wartime selection process put some great guys in the Pacific (or so the papers told us, between accounts of the many U.S. disasters) but the ones left behind on Constipation Avenue, as we frequently called it, didn't turn me on.
Seemed like I ought to be able to help more, and be more creative, and still not rile my Michigan draft board. Dorothy was willing, if not eager. We had decided not to try pregnancy, let alone parenthood, for a while yet, so we were portable. I cast about, and almost immediately tripped over the Mark 14, which was being boresighted with my gadget. The Draper Gunsight, as it was still called, was going into mass production at Sperry Gyroscope up on Long Island; meanwhile, would I consider a position at their Garden City laboratories as a fire control scientist? I certainly would!
The house was simple: we told the real estate agency we were leaving, they kept the payments we had made so far, the papers got torn up, and somebody moved in as we left - same day, in fact. Washington! We called a mover and our very minor possessions were collected at Sperry expense, and reappeared a few days later (after all, it was wartime, and a very small part load besides) in Garden City.
We had found another house, in nearby Hempstead, after about two days. We knew no more would be built; the remaining potato fields would survive until we left in 1948, as it turned out. In fact, our little place had been too late for a gas hot water heater; there was a gas stove, but hot water came via pea [-20-] coal and a peculiar pot stove in the basement, which went out several times a week. Our cat Suzy had her many litters in the big crate we kept the special coal in, next to the regular coal bin beside the hot-air furnace. The house cost $11,000 but the down payment, although not as minuscule as in Kensington, was only $1500. And we had a garage; useful on Long Island, where clearly there weren't going to be any new cars for a while, or even new tires. Our neighbors were not war types; on one side, a barber; on the other, a lead-burner (he said "boiner") who did things like chemical sinks, and industrial plumbing in general.
Dorothy answered ads for a while, and found a job in Manhattan as an industrial spectroscopist at the labs of the Interchemical Corporation, which until the war had been mostly making printing inks and special paints. In fact, they were about to get a contract for a surprising black paint for the Black Widow night fighter. The surprise was that it was not dull, dull black but high-gloss; turned out one of those NDRC professors had shown that the returns when a searchlight found such a plane were less for specular reflection!
She had general work to do, metals and pigments and such, but also was trying to apply spark technology to organic materials like amino acids. Hasn't survived; things like infrared spectroscopy and chromatography are easier and more general. But she had some fun, and felt useful. Main trouble was that she had to commute on the notorious Long Island Railroad, plus a short hop on the Eighth Avenue subway, which wreaked havoc in our sex life. And six days a week!
Meanwhile, between attempts to keep the pot stove in our basement going, I was entering a new world of optical complexity. Of course the K&E Navy rangefinders were fantastic, but those designs had been frozen before Pearl Harbor. My Sperry project was in the design stage, or at least in design revision after prototyping. It was a double-ended aircraft periscope, to stick out of the top and bottom of the secret new B-29 bomber, mid-fuselage. There were small plastic domes through which the optics looked at attacking aircraft. When the gunner tracked a plane across the zero-elevation "equator", a prism flipped the eyepiece field from the upper to the lower periscope or vice versa. The junction case also contained an analog computer featuring three-dimensional cams, rate gyros, and a hundred pounds of ancillary gear. Outputs from the system were to control remote machine gun turrets on the top and bottom of the plane, and at the tail. One human advantage was that there would be no poor devil as tail gunner. The system had a major competitor from General Electric, which won out in the end, but that was some years in the future - in fact, so was the B-29, which was still under design and test at Boeing Seattle. And it too had competitors in 1943.
The periscope and all its complex optics had been designed, and would be built, by the Hawkeye Works of Eastman Kodak, in Rochester. Sperry Gyroscope furnished the central box and the entire gun turret complex, with much hydraulics from Vickers, another division of Sperry Corporation. My major job was to keep track of Kodak progress, check proposed design improvements, approve the optical tests at Hawkeye, and generally represent the Garden City crew in a completely novel arena. They knew fire control, [-21-] radar (the Varian brothers were big guns on the staff), and hydraulics. As far as lenses and prisms were concerned, well, they knew they were made of glass. Usually!
It was fascinating. There were all kinds of new ideas. The plastic domes on the ends of the periscopes introduced a systematic error in elevation; should we put a mechanical correction cam in the computer or make the inner and outer surfaces of the domes non-concentric? Would new eyepiece designs (like the ones in today's most expensive binoculars, and actually a copy of a Zeiss patent) really improve gunner performance - wider field, yes, but more distortion?
Main thing for me, though, was meeting an individual. As I began to make frequent trips up to Kodak, usually by overcrowded train, I had more and more to do with a real lens design expert, the head of the Hawkeye group. He was an Englishman, Rudolf Kingslake, world famous in his very special field, and also the son-in-law of a chap named Conrady, the author of the treatise on lens design from which I was studying the art. There was supposed to be a second, unpublished volume of this opus, with much reference to photographic lenses, and Rudie, as his bevy of female computers called him, had the only copy. Great story, and it may even have been true; more likely, Kingslake knew enough to have written the volume himself!
He had a chap (Harold Bennett) working for him as a senior lens designer, who wanted to mechanize ray tracing calculations, which Kodak did with Fridens and Marchants, having stepped up from five-place logarithms in the Thirties. You could tell an optical computer by the fact that his or her trig tables were much dirtier - that is, much more frequently used - at the front (small angles), while astronomy or physics computers used all angles impartially! Today a programmable hand-held calculator would be much, much better, but such computations are now deeply imbedded in fancy overall design packages (expert systems) and no longer are performed out where designers can see them.
Having watched the punched card operations at the Naval Observatory, and having heard about all the fancy electronics in the MIT and Sperry radar equipments, I was quite sure attaching funny little servos to the keyboard and twiddlers of a Friden was not the way to go. But my obvious admiration for Kodak (and Kingslake) design skills, and my activities in the Optical Society, made up for my criticisms.
Some years later the prototype machine was demonstrated. It was so far behind what could be done with truly automatic equipment, and so far out of step with the promises of ENIAC and radar electronics, that it died. Compared to the room-sized machines of the late Forties, you could call it an early "micro"computer!
About this time the Mark 14 began to come off the assembly lines, and I was asked to move to the new Lake Success plant on Long Island (where the nascent UN activities lived later, while the Manhattan headquarters was being built). The work was entirely different, and I would have preferred to stay with the Garden City/Rochester project. But the war in the Pacific was central to everything, and the kamikazes were increasingly dangerous. The [-22-] Mark 14 was needed, and in vast quantities - one to every 20mm cannon, if possible.
The optics were exceedingly primitive. The gunner looked through the box, which was mounted right on the gun. He tracked the target plane through two very thin unsilvered glass plates, off which an image of bright crosshairs was reflected. The plates were wiggled by the innards of the gunsight - rate gyros working against springs, whose tensions were set by a range knob on the front of the sight. So they had to be light.
Sperry needed sources for tens of thousands of sets of these optics, and then to inspect them for adequacy and assemble them into the sights, and test the whole box on special calibration rigs. The set was two thick windows, two thin reflecting plates, a collimating lens (no problem), a special light bulb (ditto), and a reticle with a very fine transparent pattern on an opaque background. Parenthetically, the company we found which could etch the reticle pattern through a thin metal substrate is today active in chip fabrication - Buckbee Mears, it was called then.
My boss was a George Bentley, who had gotten a doctorate from Draper at MIT and then become director of research for Hamilton Watch. He teamed me with a buyer named Jack Varick, and sent us out to hunt glass and reticles. The thin plates were going to be difficult: optical glass, bubble and stria free, quite accurately dimensioned, and with the reflecting surfaces very flat and very parallel. Kodak could make them on their camera-lens line for a thousand bucks a pair, and I wasn't convinced mass quantities were possible. The shop tested them for flatness before uncementing them from the tool, and most of the plates sprang unflat after separation. Post-separation testing was tricky - expert hand work - and I was going to have to set up methods back at Sperry for girls to use, so I was very worried.
Jack remembered that quartz crystal plates, which he bought for the radar boys, had to be pretty fancy too, and to extremely close thickness tolerances besides. He found a company that could make what I needed in unlimited quantities, and for less than $50 a pair. It took more than scientists to win a war!
Buyers and travelling salesmen had other capabilities. Jack was an expert at finding the "action" - good wartime restaurants in Rochester, bars with stocks of Scotch in Chicago. He wasn't as interested as I in women, but was considerably more expert at dealing with them (I was a great judge of striptease - a much more passive skill). He learned a little about optical parts from me; I learned a lot about life on the road from him.
Back at the ranch I had to set up inspection methods. Sperry had used almost no optics; the regulars could handle the reticles and the light bulbs, but I had to do the rest. My astronomical antecedents got me into Perkin Elmer, which was so loaded down with high-priority work even Jack Varick and Sperry were not welcome. They were using a simple, sturdy interferometer for several inspection tasks, and I persuaded them to make me a couple. These became the heart of the Lake Success department. Wages were frozen, but I got a small raise anyhow.
One of the small pleasures of life was to walk down the long rows of girls and women on the assembly floor, and have them whistle at you. Men were [-23-] very scarce; even my pre-IBM goatee was acceptable. Dorothy was warily amused.
Bentley And Co. were now engaged in a much more ambitious project - to redesign and build the next Draper gunsight. This was to control quad-mounted 40mm cannon and even larger guns, would be free-standing, and needed to reach out much further for its targets. It therefore was to have a 5-power telescope, with lots of difficult optical goodies that are of little interest against today's Star Wars electronics.
I was asked to design - yes, design, not just approve, or inspect - an auxiliary optical system to superimpose a radar image from a tiny internal CRT, on the gunner's field of view. Trouble was, the box was full! I learned to greatly respect the skills of the senior designer/draftsmen, who moved things around for me. This came in handy later, in IBM and GE, when most of my confreres were carried away with exotic electronics. I still marvel today at the innards of the magnificent laser printers, knowing how demanding the mechanical parts of the system are, and therefore understanding that such machinery will remain expensive even with one-dollar chips, and will always require skilled maintenance.
There were procurement problems with this gunsight also (it was really a minidirector), notably to find a zero-thickness beamsplitter flat to less than a wavelength of light. Jack and I did it, but it wasn't as much fun the second time around. I didn't look forward to new inspection technology.
So in the fall of 1944 I began looking for a real design job. I knew I was not in the Kingslake class, but how would I ever get to that eminence beating the Utica bushes with Jack Varick? What I needed was a job at Perkin Elmer or Keuffel & Esser or Polaroid, if not at Bausch and Lomb or Kodak. And I found the Farrand Optical Company, in the very far Bronx.
Clare Farrand held valuable patents on the dynamic loudspeaker, and had answered the call of WW II by establishing a remarkable optical house, mostly with his own money, to tackle the toughest level of the technology. For instance, Farrand was building a very complex large rangefinder. That was the province of Bausch and Lomb in Rochester and K & E in Hoboken - and Zeiss in Jena, of course. Kodak's Hawkeye works had tried to build a similar instrument, and it was giving them fits. Farrand had mastered that, and was looking for more difficult projects. They had a designer named Seymour Rosin, and needed another. It looked promising.
In Chapter 03 you will encounter
(in order of appearance):
Sy Rosin 02
Bob Tripp chief engineer at Farrand, and ex-Disney
The Flight Briefing Trainer "dodge-em cars and science fiction lighting"
Dorothy 01
Optical patents the mass of ray tracing reminded me of Comrie and Eckert
Wallace Eckert 01
Department of Pure Science the new IBM venture at Columbia
The Watson Lab 01
Columbia University Eckert's old Morningside Heights stamping ground
C.L. Farrand 02
The Manhattan Project Los Alamos wanted me taken me away
IBM World Headquarters later I christened it "Galactic Headquarters"
John von Neumann a major intellectual force on a dozen wartime fronts
I.I. Rabi 01
Los Alamos still ostensibly a post office box in Santa Fe, New Mexico
ASTOUNDING STORIES several hundred copies vanished into thin air
Lillian [Lillian Feinstein Hausman] hired by Eckert back in 1938
Stan Frankel and Nick Metropolis ran the special 601s at Los Alamos
Watson Senior 01
John McPherson 01
Thomas J. Watson Astronomical Computing Bureau doing war work upstairs
Pupin Physics Lab I got half of the tenth floor, and a big safe
Hans Bethe and Roy Marshak bomb physicists until Hiroshima
Maria and Joe Mayer the same; like Feynman, she later won a Nobel prize
Dick Feynman "like a torrent from a fire hose"
S. Chandrasekhar mathematical astrophysicist
Ev Yowell worked on the upstairs project with Lillian
Marj [Marjorie Severy Herrick] she ran the IBM machine room for me
Spherical explosion a simulation 50,400,000 times as slow as Alamagordo
Liz Ward a bluestocking rides the shock wave
"von Neumann ripples" "a nice dissertation topic", said Johnnie
Hiroshima now we knew what we had been working on
[-25-] No one else at Farrand lived on Long Island, let alone near Hempstead. And moving in wartime was almost impossible. So a major task when I accepted the job there was to convince my local rationing board that I had to drive over twenty miles a day each way, six days a week. It was actually harder than persuading my draft board, which by now was used to having my various employers claim that the entire war effort revolved around my fantastic scientific capabilities. But with a little help from my new personnel department both projects succeeded.
My poor little 1937 Plymouth soon edged past the 100,000 mark, but actually benefited from its new duties, since it was soon allowed a new set of tires and a better battery. Meanwhile, diligent Dorothy continued to suffer on the Lonn Guyland, as the veteran commuters called it. Even she got some pleasure from my gas coupons, however, since occasional little trips which we could not have done on an A-ration book (into Manhattan and back for a concert, for instance) became possible with a C sticker. And to my amazement, Farrand made me an allowance for gasoline and bridge tolls.
Sy Rosin and I reported to a man named Bob Tripp, who had been chief engineer for Disney, and was a fountain of weird ideas. Sy was conservative and I was still rather astronomical, so Tripp was a bridge to wilder shores. I have to get back to stories about the computer, but a short sample of the games we played may be fun, if only to show how differently we had to approach problems like flight simulation before we had computer graphics and heroic software.
Tripp had agreed to explore building a Flight Briefing Trainer, to simulate invasion approaches to the Pacific islands and the Japanese coastlines. He planned a large room with an absolutely flat terrazzo floor (in my cynical way [-26-] I claimed that this would be the hardest part of the whole project, and I was right). A powered cart with analog computing gear would run around on this floor with the same freedoms as a dodge-em car at a carnival, and would carry a simulated cockpit and controls, and the trainee. He would be looking into the eyepieces of a periscope, which stuck up through the roof of the cart and could be raised or lowered by the cockpit controls. On the ceiling was a gorgeous relief map of the target area, and the head of the periscope could come "down" close to the terrain. What a monstrosity!
I was charged with designing the head prism and top end of the optics, which had to replace the upper part of a specified Navy periscope. Tripp had already solved the problem of depth of focus, which is serious in all such model devices, by planning to synchronize rapidly changing focal length of my top lens with selective illumination of the map, on a 30-cycle-a-second basis. Good grief - dodge-em cars and science fiction lighting!
Most of it was built, I heard later, but fortunately the war ended before full operation was needed. I'd love to have seen it, and ridden it.
There was a great Christmas party [1944], where us upstairs types got a long-awaited chance to, ah, mingle with the all-female assembly and test crew from the rangefinder area. Dorothy had been hoarding red stamps to buy a Christmas Eve roast, which was ruined long before I staggered in from my long drive and other exhausting efforts, and collapsed on the sofa. It was a very cold Christmas, it was. A once-a-year fall from grace might have been forgiven, but coupled with other adventures which I had not concealed very well, and with those wasted ration points...
I finished the initial periscope head design and began on the eye end. The idea was to split the light into two eyepieces to simulate infinite-distance binocular vision, and also to feed in a view of the instrument panel. Then Tripp said one day that he and Mr. Farrand wanted to build optics for projection television after the war, and would I put a few hours a month on a library and patent search, since I knew about Schmidt and similar high aperture systems, and since the shop downstairs was making deep-dish mirrors on a secret contract? I quickly decided that to check out even a few patents would require hundreds of hours of ray-tracing - even with a machine like the one planned at Kodak!
Unlike an initial design, where you trace a ray or two and then change something, and trace a few more, in the assessment of finished designs you could select an assortment of rays in advance and trace them all through the first surface of the lens, or even through the first surface of each of several lenses under investigation. Then you could repeat through the second surface or surfaces, and then the third, and so on. I was of course reminded of Wallace Eckert's punched card shop at the Naval Observatory, and of Comrie's work at Greenwich, where calculations for many dates were carried ahead one step at a time.
Today we would immediately think of parallel computing, one of the major current big-machine enthusiasms. In a machine with say 1024 processors we could trace a hundred rays through each of ten lens designs (at a millisecond a surface),and finish all ten analyses in less than a second. In 1945 you could only have one "processor", and it would be a whole room full of [-27-] electromechanical punched card machines, and with two or three operators. Ten complex lenses might well take a week to run - but by hand it would take months.
One very nice thing about working at Farrand was that you were encouraged to keep up with the literature. And not just by subscribing to it yourself; the company took all the useful journals, and suggestions for additions to the list were welcome. These were circulated briskly, and Rosin and I were right behind Mr. Farrand and Bob Tripp in priority. In early April a very short article appeared in SCIENCE to the effect that the International Business Machines Corporation had appointed Dr. Wallace J. Eckert, former director of the Nautical Almanac Office at the Naval Observatory, as head of a new Department of Pure Science, and that he would be establishing a Watson Scientific Computing Laboratory at Columbia University.
The note said that the laboratory would have a full complement of IBM gear, which would be used for instruction and research. It mentioned Eckert's former connection with Columbia as chairman of the astronomy department, and somewhat sotto voce that the chief executive of IBM was a prominent trustee of the university. I saw the piece at the beginning of May, and immediately wrote Eckert a letter explaining my idea about mechanizing large-scale optical calculations. I suggested he might let me come over at night when things got rolling, and experiment a little. And to sugar-coat things I mentioned that my employer, "Farrand Optical in the Bronx", would have commercial use for such an application after the war.
Ten days later a little man showed up in Farrand's office. He introduced himself as representing a Manhattan Engineer District, a secret agency of the Army with offices in the Empire State Building. I never knew exactly how he approached it, but somehow he explained to Farrand, who of course had very high level personal security clearances, that something much more important than optical design was going on out in New Mexico. He produced my letter to Eckert, which Farrand had never seen, and said he wanted me for this operation at Columbia. Coooo!
They had me in. I was dumbfounded. I protested that I had not talked to Eckert, that I had had no idea of going to work for this new laboratory, that I liked it where I was. "No matter," said the little man. "Report to this Mr. Eckert tomorrow, and he will arrange everything. His office is at 590 Madison Avenue." I protested that no such impressment was possible - I remember using the term "shanghai" - and that my Michigan draft board would have to be consulted. But down inside, a great feeling of excitement was growing. I was going to get my hands on punched card machines at last, and do some massive computing.
The next days were a blur. I went in to Manhattan with Dorothy on the train the next day, and discovered that 590 Madison was the headquarters building of IBM. In later years I invented the name "Galactic Headquarters", but the Watsonites then called it merely "World Headquarters", capitalizing on the IBM phrase "World Peace Through World Trade" which T.J. had adopted in the Thirties. In typical IBM fashion, I was expected (although the reception desk was somewhat disconcerted by my beard and sport jacket). Eckert was indeed upstairs, and was delighted to see me.
Within a few minutes I knew that there was to be an atomic bomb. It was [-28-] mid-May, two months before the Alamagordo burst, but Eckert had not been cleared for such information. What he knew, and passed on to me, was that a secret laboratory had been established "near Santa Fe", that very senior scientists like his old friend I. I. Rabi of the Columbia physics department, and the famous John von Neumann, were prime movers, and that the group was harnessing the fission of uranium to make a fantastic weapon.
This laboratory, which I was to hear called Los Alamos a few days later, was officially just a post office box in Santa Fe (which among other curious mail got several hundred copies each month of ASTOUNDING STORIES!). Eckert told me that there was a major punched card installation there, with something like four IBM 601 multiplying punches, all of which could allow for algebraic signs and two of which could actually do division. There were supporting machines: tabulators and sorters and reproducers and such. The shop was run by Stan Frankel and Nick Metropolis (and, I found out many years later, Doc Nelson), who supervised a mixed bunch of young civilians and dragooned enlisted men; it was working around the clock and still falling behind, and there was neither floor space nor personnel to expand the operation.
"They came to IBM for help," Eckert said. "Mr. Watson and John McPherson (whom I was about to meet) thought immediately of the Astronomical Bureau at Columbia, but it is heavily engaged in fairly high priority work for another part of the Army, and really has no room for physical expansion anyhow. It has only two 601s and an old 285 fixed-plugboard tabulator, and there is hardly room to move."
What I had not thought of when I read the story in SCIENCE was that IBM itself had to have wartime priorities - and very high ones at that - to retain any of the machines that were coming down its overstressed production lines. Even the cards were under strict allocation. If T.J. wanted to help the war effort, and at the same time position his company for postwar scientific initiatives, here was the perfect opportunity. He had his all-seeing eye on Eckert, on the Naval Observatory shop, on Columbia of course. He was still outraged at Harvard. So he decided that there should be a new venture on Morningside Heights, told McPherson, who was then director of engineering at 590 Madison, to see to it, and turned to his next imperial task.
Eckert was offered a chance to go back to his beloved Columbia in a very prestigious role, a chance to direct a far more powerful computing facility than he had ever had before, a chance to build a better machine than Aiken's - and, as I found out sixteen years later, at two and a half times the rather good salary he was getting from Uncle Sam!
He had set up the little astronomy shop a decade ago with his own hands, but it had been run for the last five or six years by a vigorous woman he had selected and trained, named Lillian Hausman. The newer enterprise in Washington had been run for him by Jack Belzer, about whom more in later stories. Eckert needed another subordinate, and was therefore delighted to get my letter. I was well known to him, a fellow astronomer of the same celestial mechanics persuasion, a Ph.D. acceptable to his snooty Columbia buddies, an expert computer and numerical analyst - scarce as hen's teeth, especially in wartime - and I lived right next door! And also I came cheap, although it was [-29-] a year or two before I sadly recognized the fact. But I couldn't have said no, with that little man telling Farrand where to send me, could I?
Eckert and McPherson had worked out a list of machines, and the Columbia administration had found space on the tenth floor of Pupin, the physics building which contained the astronomy department and the Thomas J. Watson Astronomical Computing Bureau. There was even an old teaching telescope up topside, for undergraduate use on the three clear Manhattan nights a year! The Los Alamos priorities began to function; crates of IBM equipment, chairs and tables and file cabinets (and a very big safe), cartons of punched cards and tabulator paper, all appeared as if by magic. What fancy priorities couldn't provide, IBM money and prestige - and Wallace's old Columbia connections - always could.
All I had to do was put it together. McPherson said he would send some bodies. I went upstairs and borrowed operating manuals from Lillian and her people, and asked naive questions about plugboards (Wallace had not ordered nearly enough). We all knelt down prayerfully in front of the hoary old 285 for my first wiring lessons - wasted, because it turned out we were getting a much fancier machine if the Pupin elevator was up to it. You have to understand that I had never even run a sorter!
Also I had summer flu, and Dorothy's parents arrived from California for a long visit. And the cat was pregnant again, but not urgently.
The intellectual excitement was enormous. In two weeks we had visits from Johnnie von Neumann, Hans Bethe, Roy Marshak, Maria and Joe Mayer, and - like a torrent from a firehose - one Richard Feynman. Three of them were to be Nobelists, and Johnnie was supposed to be the world's greatest applied mathematician (I was a Chandrasekhar rooter myself). The punched card boys were too busy to come, but Feynman seemed to know all about the machines (and everything else besides). For instance, he showed me how to wire a chain of selectors on my new 405 tabulator, a technique which in those days was too esoteric to be in an operating manual. He showed me once, and pretty briskly; you didn't get a second shot with Feynman! Lillian didn't know such tricks, Eckert didn't know, McPherson knew (he had edited "IBM Pointers" for a couple of years) but was far out of my reach. So I had to learn from a future Nobel prizewinner!
World Headquarters had me fill out a form or two, and took over the struggle with my draft board. I still have my original ID card, almost certainly the first one ever issued at 590 with a beard on it. There was no physical; a year or so later the personnel department ginned one up, and told me to get my tonsils out - no mention of removing the plaid lining in my sport jacket. I kept my tonsils, and have them today; they are healthier than the rest of me, having been exercised much more vigorously.
An expert machine operator? One appeared. Machine repairs? A very good man was on call. A receptionist? A secretary for Eckert? More plugwires? Coming right up! Ah, IBM!
The operator was an oldtimer named Cliff, who had the machine room culture down pat. He could joggle a vast wodge of cards, carried a sorting needle, knew how to stick chips back in a card for a quick fix. The customer engineer (IBM for "maintenance expert") showed me where the machine [-30-] blueprints were hidden. Lillian knew these things as well, but thought astronomers should be above such details; when I got tired of imitating Cliff I would sidle upstairs and talk to Everett Yowell, who was running the military project in the Astronomical Bureau for her (years later he was scientific marketing manager for NCR). He didn't have his Columbia degree yet, and had to be affable.
One day an attractive young blonde - sharp features, summer freckles, clear voice - came up from 590. Her name was Marjorie Severy, and she had just graduated from Wellesley with a math major. The personnel people thought she would make a possible machine room supervisor. She was just what we needed.
By this time Eckert and I knew that the unit of temperature in my calculations was a million degrees Kelvin, and so on. We kept the equations I was working with, and my translation into computable form (approved, with a warning, by the great Johnnie on his first visit), and the starting values, in the big safe. What Marjorie and Cliff and the others saw was a long shelf of 28 IBM plugboards, which when cycled through the various machines produced a messy and unlabeled printout from the 405 tabulator. While I bundled this up and mailed it with my own hands, registered, to von Neumann - later, to Marshak - at the anonymous Santa Fe box number, the machine room team began to repeat the cycle. We did two a day; each predicted the temperature and pressure up to the (moving) shock wave after one more time step - a millisecond. Allowing for the fact that we did not work Sundays, that was 50,400,000 times as slow as the real atomic explosion! A Cray 2 today would be ten million times faster - and give you Saturday off besides!
About von Neumann's warning: he instantly realized that Eckert and I were experts only on numerical solutions of systems of ordinary differential equations, which are central to all the problems of celestial mechanics. I had had a good course in quantum theory in graduate school, but no practice in solving those kinds of partial differential equations numerically. Nor had Eckert; in fact, nobody had - solutions took much more computing power than the physicists were accustomed to in 1940.
The little group of Giant Brains at Los Alamos, especially Feynman and the card pushers, had learned from Johnnie and from sad experience that if a certain ratio of distance interval to time interval was exceeded, the numbers went wild. I was told to check after each time step to make sure that ratio was still all right - a minor desk calculator operation. One of our new hires, a bluestocking named Liz Ward, did our desk calculator work. After figuring out where the shock wave had gotten to, she did the little ratio calculation carefully, but neglected to report to Marjorie or to me when the ratio went bad. And I didn't look over her shoulder: mea culpa!
So I had to get on the telephone to Johnnie and confess we had blown two cycles - a day's work. "Dr. von Neumann, suppose I run a smoothing operation to get rid of the ripples? Will it distort what follows?" "That will make a nice dissertation topic in applied mathematics after the war, Grosch. Meanwhile, to be safe, do the last three time steps over." And there went Sunday!
[-31-] Eckert and I had clearances, and access to that safe. McPherson had even higher clearances - IBM was involved in cryptography, for instance - but no need to know. The machine room gang had none, although they did know they were doing something very important. Oh, and the listings I mailed to Los Alamos had no hand-written labels, but matched a labeled master copy that had been carried back by Feynman: primitive but exceedingly effective security.
Then it was August 6, and the radio and the newspapers told us about Hiroshima. We knew what we had been working on, and that tens of thousands of Japanese civilians had been incinerated. Before the moral pressures could mount, events swept us away. The war in the Pacific ended, the war in Europe ended. All our perspectives lengthened overnight - from a few months "to the end of the war," to the long reaches of peace.
In Chapter 04 you will encounter
(in order of appearance):
The Great Depression dominated the American scene until Pearl Harbor
Parents [Mabel and Bert Grosch] produced a hatchling in Saskatchewan
Saskatoon the Central Hospital got an incubator in early 1918
Andrew Booth built a drum computer at Birkbeck College, London
Canadian schools skipped the bright kids ahead
Mrs. Linhart we started the Science Club together in Toledo
"Benjamin Franklin" [play] turned out I was a born ham
Early heroes: Ozma, Dr. Doolittle and Captain Nemo no computers yet
AMAZING STORIES Buck Rogers and his flying belt, on a 1928 cover
H.G. Wells I knew at twelve why "cavorite" was impossible
Edgar Rice Burroughs but I had hopes for Deja Thoris
The Toledo Museum of Art the collections were better than the classes
Sex even the juvenile form was great stuff
High school mathematics flying solo in calculus was too difficult
High school debating a big thing in Michigan and the Middle West
High school chemistry Mr. Erickson had an unofficial lab assistant
The Detroit Public Library an adult permit for a persuasive 13-year-old
Graduation [Royal Oak Senior High School] at fifteen, but not a monster
Albion College offered a debating [!] scholarship
University of Michigan [Ann Arbor] offered a tuition scholarship
Going steady "meager sensual pleasures"
[-33-] After almost five years of war, peace was a stranger. And I did not remember it all that fondly, especially when I thought back before my college years. The Great Depression had been pretty awful for a youngster, especially near Detroit, where the clang of closing factory doors was nastily punctuated by the splat! of defenestrated bankers.
My parents, Bessie Mabel Adams ("Mabel") and Reuben John Grosch ("Bert", since his father had also been Reuben), had emigrated from England to Western Canada. They were both Londoners - Walthamstovians, and from large families. Dad had been apprenticed as a cabinetmaker, and retained the skills all his life - made very handsome violins as a hobby during the Evil Thirties. The rather Germanic name stretched back through a line of furniture makers, perhaps to the Hanover accession with its train of tradesmen - always in the London suburbs. Dad broke the pattern; came out to Saskatchewan as a construction carpenter, and sent for Mabel as soon as he had his feet under him. They were married in 1912. I was born in Saskatoon in 1918 in Caesarean style, and survived only because the Central Hospital had just added an incubator to its equipment. Not surprisingly, I remained the only child.
We moved steadily eastward in the next years. When I finally got back to my birthplace fifty years later, I asked my hosts to drive me past the hospital to see if there was a plaque to their first (and presumably most famous) incubator baby; alas! they had torn the place down. My host, Andy Booth, ex-Birkbeck College, London, and builder of the first British drum computer, claimed they had sown salt in the ashes!
I went to school in Ontario; first in Chatham, then in Windsor. Canadian schools did not, at least in those less-U.S.-influenced days, let a student sit [-34-] idle; I spent one semester in each grade and ended up three years ahead of myself. I remember only one teacher clearly. The provincial government was instituting French classes as I was zooming through the sixth grade [1928], and I can still sing a few phrases about the bridge of Avignon.
My father had been inside superintendent and senior detailer for a Detroit fine-woodworking company, commuting across the river each day. He was offered a better job in Toledo, Ohio, and worked there for a few months while coming home to Windsor on weekends - via interurban trolley, today only a hobbyist's dream. He persuaded Mabel to put away her English preferences and emigrate yet again.
Parenthetically, he soon became a U.S. citizen, and conferred the same citizenship on me by the derivative route, but my mother remained a determined British subject. George V and George VI were much more real to her than our all-enveloping FDR. Yet although she corresponded copiously with relatives and friends in England from 1912 until her death in 1962, she never showed the slightest interest in going back to visit the country that seemed to hold her loyalties.
I started the seventh grade at Alexander Hamilton Junior High School in Toledo, a few days before my tenth birthday. The teachers were dubious, especially when they found that I believed the Canadians had won the War of 1812! Within a month or two the science teacher, Mrs. Linhart, and I had started the school's first Science Club. A few weeks more and I was writing for the school paper, usually the province of the eighth graders. My English teacher, Miss Morgan, reported to Mother that she wanted to hug me, and to shake my head off, on alternate days - an emotion shared often in later years by a long string of wives, managers and friends.
In the spring rehearsals began for the Annual Play, customarily dominated by the older grade. But this year the staff had chosen "Benjamin Franklin", which required a smallish boy to be on stage continuously, gabbling away the while. Guess who was the only boy in AH that could memorize the part?
I loved it - loved being up there in front of those kids and teachers and parents, the center of all attention. It turned out that in spite of my glasses, and my reading, and my science club, and my good grades, I was a Born Ham.
The previous August I had persuaded my dubious father to buy me my first AMAZING STORIES, the pioneer Gernsback science fiction magazine, which I read without missing an issue for over a dozen years. Twenty five cents, it cost, and had a drawing of Jupiter as seen from Ganymede on the cover, and the original Buck Rogers story ("Armageddon, 2419 A.D.") inside. I had started with the Oz books at five, and Dr. Doolittle; then Verne; then a heavy dose of mythology - Greek, Roman, Norse, even Hindu. Well before my tenth birthday I was off into space!
AMAZING led me to Wells and Burroughs, and a year or two later the first issues of ASTOUNDING appeared on the newsstands, with science as well as fiction. Perhaps more importantly, I graduated from the primarily juvenile library collection in the basement of the Toledo school, to an adult public library, and later to the large main Detroit Public Library. Reading had to be the prime resource of a boy three years younger and a head shorter than his classmates.
One of my teachers got me enrolled in a Saturday class at the very good Toledo Museum of Art, and I liked it. But I really wasn't much at fine arts, and enjoyed wandering among the collections more than being creative with pen and brush. A modern color graphics terminal would have helped, of course.
And I discovered sex. Given my social disabilities, it didn't involve girls or women, or even boy friends [-35-] - just what the Victorians rightly called "solitary" and wrongly called "vice". I was barely ten. I thought it was terrific. Not one day has passed since that first wonderful discovery without my wishing for or planning for - or on good days, experiencing - the manifold pleasures of sex.
The company Dad had worked for in Detroit, a private outfit named Moynes, which did fancy churches and Grosse Pointe mansions and lots of circular staircases, wanted him back, and came down to Toledo and said so. This was 1930; in the Fifties and Sixties, when I was asked back for second hitches by IBM and GE and Uncle Sam, I often thought of his experience. He was flattered, as I was later; he accepted. We moved to a convenient Detroit suburb, Royal Oak, and I entered the ninth grade at the senior high school. I was barely twelve. The Great Depression was only a few months away. I don't remember much about mathematics in Windsor or Toledo. Skipping ahead as I did, I missed whole chunks of arithmetic and only noticed it once, when a review examination revealed I had never had decimals! I did the whole exam on the assumption that 6.28 meant 6 and 1/28th, got an abysmal grade, and discovered the right system by myself, in a week or two. The teacher never knew.
At Royal Oak I took considerable pleasure in algebra and geometry, and was so expert that in my senior year I sometimes sat in as instructor for the freshman algebra teacher's last class while he coached football. At the end of my second year my geometry teacher, Miss Gibson, who was to teach junior-year algebra, gave me the next textbook to work over during the long summer vacation, graded my problems and administered a mini-final examination in October, and excused me from classes. In my senior year we did solid geometry in the fall and trigonometry in the spring; the latter introduced me to logarithms, and I spent many hours in class - while the others sweated - building a better seven-place table from a curious book I got from the town library which had accurate logs, but of prime numbers only.
From that same library - adult side, but they had lots of Jules Verne on the juvenile side! - I got the Granville Smith and Longley calculus text which turned out next year to be used in my college class. I tried to master it on my own, but stuck at the approach to the limit idea. I asked my trig teacher, Miss Kirk, for help, only to learn she had never had calculus and taught her "advanced" classes by rote. I was disillusioned.
On the Ham side, I soon discovered debating, a big thing in that part of the Middle West, with a statewide competition among high schools each year and even college scholarships similar to but less plush than those for football and basketball stars. But it was the public-appearance thing I enjoyed, especially when it involved a trip to another high school, and a fresh audience. I soon became the expert rebutter and "closer", much improved the Royal Oak standing in the state, and captained the squad my senior year.
On the science side, I became unofficial lab assistant in chemistry, had the [-36-] complete run of the shop for two years while the teacher, Mr. Erickson, very good indeed but also busy coaching track, watched my struggles to do primitive qualitative analysis without a good balance (I remember making standard solutions with constant-boiling hydrochloric acid and lots of pipette/burette work).
My father cut into a gas line at home and inserted the appropriate fixture so I could have a Bunsen burner instead of an alcohol lamp. I did the usual dangerous flashlight powder/ thermite/ gunpowder experiments in our back yard, and decided to be a research chemist. All this was amplified by a steady flow of books from the big Detroit Public Library, for which I managed to promote an adult permit on the strength of Dad's employment a few blocks away. I was then just thirteen, but persuasive!
Physics was a bust. Astronomy I studied out of AMAZING and ASTOUNDING, plus dull books from my three libraries (there was a rather good library in the high school also). I didn't confuse the sources; I already understood at twelve or thirteen that H.G. Wells' "cavorite", which took his confused heroes to the Moon, was an impossibility - and why.
This is a good place to explain why I flourished so early, and yet didn't turn into some kind of a monster. I was immersed in a rather ordinary environment - good schools, good teachers, but nothing like the Bronx High School of Science, or the Cambridge or Pasadena hotbeds. And everything was seriously damped by the Depression, which was raging with special virulence in the Detroit area. That meant no travel, very little spending money, very few purchased books, and at school very limited equipment.
On the other hand, my parents and all my teachers were enormously supportive. The former never really understood what I did after I finished high school, but they were proud of my honors and my Ph.D. and my mysterious jobs. They watched me as Benjamin Franklin and as debating captain, came to all the school events - but shunned the PTA. My teachers were no great minds, but they worked far harder, especially on me and the other bright kids, than they would today; teaching was still honored in Canada, and a stable and fairly well-paid job in Toledo and Royal Oak - and the Depression sharpened their appreciation.
Of course it seemed like a struggle at the time, and I had occasional setbacks and social disfunctions, but compared to most youngsters in those Depression years I had a red-carpet path. My debating coach, Miss Moore, got me the offer of a scholarship at Albion College; my Latin teacher, Mrs. Land, was an active alumna of the University of Michigan, and helped me apply for and win a tuition scholarship there - and that, I accepted.
If I had been surrounded by really fancy teachers, or if my parents had been intellectuals, I would have been steered to Harvard or Cal Tech or whatever, and with much more generous scholarship support - I saw this quite clearly after only a couple of years at Ann Arbor. And I could have upgraded myself, but by then I was doing published research - at seventeen! - and was far too engrossed to make the effort.
About the high school social thing: I yearned for girls and cars and sophistication just like other boys, but I was too small for athletics, too young to drive or work (and there were no jobs, anyhow), and too intellectual to be [-37-] a smoothie. I managed a year of "going steady" when I was a junior, and only fourteen, but was dissatisfied with the meager sensual pleasures extended by my lovely but cautious girl friend, and regressed to less frequent dating as a senior.
In Chapter 05 you will encounter
(in order of appearance):
Alumni Scholars welcomed to Ann Arbor during Freshman Week
President Ruthven of the University of Michigan
The Michigan Catalogue familiar terrain for outmanoeuvering an advisor
Astronomy courses not for freshmen, but after six years of AMAZING ...
Maxwell [Allan Douglas Maxwell] he made me a computer at seventeen
James Craig Watson anciently director of the Michigan observatory
Computing forms the software of desk calculation
The Jellyroll Hectograph you could pull forty copies, and in color
A.O. Leuschner the link between Oppolzer and Maxwell
The Delporte Object an asteroid that came almost as close as the Moon
"Real Research" like a tonsured scientist or an ASTOUNDING STORIES hero
Parents 04
The Director [Heber Doust Curtis] welcomed me to the Observatory family
Scholarships and fellowships Curtis and Maxwell kept me going
The Suitcases a necktie semanier and a two-year supply of scratch paper
The Marchant ACT-10M Maxwell and I preferred it to a Monroe or a Friden
Oberth and Hohmann German authors of serious treatises on space flight
The AAS [American Astronomical Society] Maxwell nominated me at twenty
L.J. Comrie 01
S. Chandrasekhar 03
Wallace Eckert 01
Seth Nicholson [Mount Wilson] discovered J IX, J X and J XI
New satellites of Jupiter [J X and J XI] they needed the 100-inch
A Rackham Predoctoral Fellowship my first bicycle, and summer at Harvard
Harvard College Observatory a major center of world astronomy
Harvard summer conferences 02
Whipple and Cunningham the Harvard competition for Maxwell and Grosch
Paul Herget the Cincinnati competition for Maxwell
J VIII 02
N. Boeva the woman astronomer at Leningrad who had last done J VIII
The Royal Society elects Comrie, its first computer, in 1950
The Naval Observatory 02
[-39-] I arrived in Ann Arbor on the 16th of September [1934], just three days after my birthday. Freshmen had to be sixteen on registration day, which was the 25th for Names Starting With G; I made it with 12 days to spare! The new Alumni Scholars were welcomed during Freshman Week by the president, a herpetologist named Ruthven. I next shook hands with him when I received my doctoral diploma in 1942 (when I got my bachelor's, he handed it to me at diploma length). It was a big place, not at all like Royal Oak Senior High School, and I was excited and challenged.
There were a thousand rules and regulations about course selection. I had studied the catalogue for months, and knew it much better than my advisor, but was unable to find a way into the "professional" introductory chemistry class; the department wanted you to have calculus along with it, and calculus was normally a sophomore subject. The best I could do with math was to choose the "professional" introduction, which got you ready for differential equations as a sophomore, but clearly didn't do calculus for many weeks. Physics was also impossible, and with more justification; the budding chemists didn't use calculus at all, even in the second semester, and the budding physicists certainly could have used it - but usually didn't.
I wiggled on the hook. In a careless moment the advisor allowed me to sign up for the entry astronomy course, which really was also for sophomores. It was a snap for someone who had been reading and rereading Verne, Wells, AMAZING STORIES and ASTOUNDING STORIES for six enthralled years. By the time an assistant dean called me on the carpet for being in a sophomore subject, I was leading the class, and the laboratory class as well. He could see I had been a knowing infractor - but he let me stay!
I had been depressed that my trigonometry teacher had not known calculus. Now I was in a different league; my math professor was a famous mathematician, who took the advanced freshman section to encourage entry [-40-] into the field. My astronomy lecturer was a real live researcher who became director of the Dominion Astrophysical Observatory in Victoria a few years later. Even my English class was taught by a publishing poet, who was also proselytizing (it too was an advanced section).
And I made the freshman debating squad.
The Depression was fierce. Dad managed to get me a little summer work as a helper, building a temporary keg-washing plant for the Stroh Brewery in Detroit - alas, I had not yet learned to drink beer, pails of which I brought out for the delighted carpenters. I suspected it was a non-standard arrangement; there were unions about, and I was still only sixteen. And it was heavy work - but how I needed the money!
Ann Arbor was a relief that fall. With my tiny earnings I found a single room in a nicer boarding house: great landlady and good housemates (I lived there contentedly until I left for Washington six years later). I signed up for an advanced astronomy course, and unknowingly set all the rest of my career in train. I had been drawn to the astronomers of my freshman year, and chose physics in the new term - yes, the fancy course - instead of chemistry, and to take an extra fifth subject, as a consequence. But in September [1935] I was also thinking about being a mathematician or a physicist; by Christmas, those options had pretty well faded.
The fifth course was Astronomy 101, Practical Astronomy - the old-fashioned stuff that had been the entry to classical astronomy for centuries, and was then withering under the fires of relativity and quantum mechanics and astrophysics. There were only ten students. Most of the others were seniors and graduate students majoring in other subjects (surveying and navigation and such), and there were two or three apprentice astrophysicists. I was the first student who ever elected the course before choosing a major, and was five or six years younger than the rest.
We were to do the old precise-measurement kind of astronomy: use of the meridian circle and the chronometer and the filar micrometer. No photography. No spectroscopy. Very accurate. Lots of spherical trigonometry; in fact, it was the only course in the university which taught that variety of math heavily.
The professor was one Allan Douglas Maxwell, unhappy with the way that his kind of astronomy was being brushed aside by the astrophysicists, and hungry for a disciple. He had been at Ann Arbor for nearly ten years and had not had a single doctoral candidate. His father had been a streetcar conductor in Woodland, California, and by his mathematical aptitude Maxwell had won scholarships and prizes at UC Berkeley. He became the final disciple of Armin Leuschner, then chairman of the department and head of the Student Observatory (not to be confused with Lick, on a mountaintop behind San Jose, where the Real Research was done).
Leuschner had spent much of his career reformulating old methods of computing the orbits of asteroids and comets for "modern" calculating techniques. He discouraged the use of six-place addition-subtraction logarithm tables, used a hand-cranked German Brunsviga desk calculator and tables of natural trig functions, and looked forward to the days when many computers - remember, computers were still people in those days - when many computers would have access to calculators with (gasp!) electrical motors.
[-41-] Maxwell found this immensely satisfying. He did a thesis, helped Leuschner write and publish a major text on orbit computing methods, corresponded with the other dozen or so specialists elsewhere in the U.S. He was ready to go out into the world. The University of Michigan had a tradition in the field and in the specialty: there had been an active observatory in Ann Arbor for over a century, and one former director was James Craig Watson, whose book Theoretical Astronomy had been the standard American text for (logarithmic) orbit computing since 1867. Even the astrophysicists conceded that Ann Arbor should have another orbit man, and Maxwell had been their choice.
He taught by demonstration. At the blackboard he would derive the appropriate expressions for computation, and then distribute beautifully prepared computing sheets, with an example filled in on one page or set. We called them computing forms, and they were the equivalent of programs - the term "software" would not be invented for more than twenty years. They directed the computer, Maxwell himself or a student, step by step through the complex calculation. The sample showed how many decimal places to carry, neither too few nor too many, and he would justify the choice and recommend calculating tools and tricks as we went along. Just as in today's software, bugs sometimes showed up, or (much more frequently) Maxwell would see a better way to do the job. Then the forms would be revised or replaced.
Instead of a laser page printer he had his personal hectograph machine. It was a long roll of jelly on a cloth backing. He would draw a form in hectograph ink on bond paper, roll out a fresh foot of reproducing surface, press the original face down for just the right number of minutes, peel it carefully off, and then draw off duplicate forms on plain paper until the copies became too faint or blurry (copiers were also twenty years in the future!). When the takeup roll was full he would go back to the beginning, where the six-month-old image had completely absorbed, and start over. I thought such modern technology was terrific. Why, you could even use more than one color!
Maxwell did all his calculations in ink, and admonished us to do likewise. I did; the others tried, and reverted to pencil. He told us that engineers had to do many calculations in bound laboratory books, for patent and similar reasons, and that they worked in ink and crossed out but never obliterated errors. "In practical astronomy and especially in orbit calculations, which you will do in Astronomy 201, the calculations are so voluminous that the sheets would be a mess, so I use ink eradicator on my mistakes, and write again in ink after the spot is nice and dry" - much shifting in their seats by the poor victims, who were mostly resolving never, never to take 201. I adopted his tactics enthusiastically, and in later years helped him experiment on what combination of paper, eradicator and fountain pen ink worked best (one of the Hammermill bonds, Clorox [!!], and blue Washable Quink). I made very few mistakes; indeed, I didn't dare!
One of the young astrophysikers dropped out. The surveyors and would-be navigators grumbled. Then, metaphorically at least, the heavens opened. An astronomer named Delporte, at the Uccle Observatory near Brussels, photographed a very rapidly moving asteroid. Even assuming the motion was [-42-] entirely lateral, and against the grain of the solar system, it could not move so fast unless it was closer than any previously observed body - except meteoroids, of course. The world orbit-computing experts were agog, at all eight or ten sites: Harvard, Leningrad, Berkeley, Greenwich, Cincinnati, Ann Arbor et al.
Observations were sent by specially-coded telegrams from a central bureau. Maxwell started in. But he needed a check computer, who would work twenty or thirty hours straight and not make many more mistakes - hopefully, different ones - than he himself. The forms were ready, although like today's software, "a few small changes" were required. Ignoring the child labor laws, and the fact that I had an examination in Physics 35 scheduled for the next day, he drafted me!
I still have the sheets. Faded now, and not at all elegant because I had never done anything that serious and that urgent before, they mark my initiation. I was barely seventeen. I loved the whole thing; I was doing research, just like a real astronomer or an AMAZING STORIES hero. And in the end I got an A in the physics course anyhow. I had become a computer.
Maxwell was my leader for almost six years. I took my Ph.D. from him, but much more important, learned numerical analysis and practical computing as his apprentice. Neither of us doubted, from early 1936, that I would follow the line, through Oppolzer the eclipse computer and Leuschner the asteroid computer (he had been Oppolzer's disciple in Germany) and Maxwell the comet computer, to become Grosch the (as it turned out) satellite computer.
There were obstacles, mostly financial. I was literally undernourished. My parents managed five dollars a week my freshman year, out of which I paid room rent of $1.25 [!], surely the lowest in Ann Arbor, saved toward next semester's textbooks, and bought AMAZING and ASTOUNDING of course. A wonderful thick malted milk was ten cents in those days, and a large hamburger the same, but it was several years before I could afford breakfast. It will give some idea of my priorities to note that I always bought brand new textbooks, even when used ones were available, and that after dozens of transcontinental and intercontinental moves I still have the best of them - no longer pristine - on my office bookshelves!
The sophomore year started out only a little better; my earnings as a summer worker were minuscule. But I was quickly adopted by the observatory "family". The University of Michigan Observatory sat on a knoll across from the big teaching hospital, which was unfortunate for the observing astrophysicists because of light pollution, but good for me because it was within walking distance of the main campus and my boarding house. The director, a wonderful warm old guy named Heber Curtis, lived in an attached residence which went with the job, and used to walk through the intervening halls and library at all hours, often finding Maxwell and his prize catch computing away, independent of bad observing weather.
Maxwell used to take me out for late snacks after such sessions, and observed my scrawny state. I ate everything the all-night Greek cafe could produce, including messes I couldn't look at today after decades of gourmet [-43-] dining all over the world. He reported this to Curtis, who first produced a $50 "assistantship" (it was too late in the budget cycle to do more), then in my third year somehow awarded me half of a graduate astronomy fellowship, and later heavily supported my applications for larger scholarships and fellowships. Also, fearing that I might collapse under the 8-by-10 beams of my construction job, Curtis gave me summer observing assistantships in 1936 and 1937, ostensibly "to show you what most astronomers do at night", but really to provide eating money. I was nurtured by Maxwell and Curtis, and intellectually by the other astronomers as well: nice, nice memories.
The office that I was to share for more than five years was at the end of the main block, and butted on to the circular wall of the main reflecting telescope. This was convenient during thunderstorms, since the dome was metal, and Maxwell, who was deathly afraid of lightning, could unobtrusively slip out of his quarters and be safe. The observatory carpenter had built a set of inexpensive shelves along another wall, and on these shelves sat sixty or so identical cardboard suitcases, the size of a large attache case. Almost everything Maxwell owned was in these cases or elsewhere in the office; he kept his meager wardrobe and linen in his nearby apartment, but his professional and private belongings were "at work".
An unwilling bachelor of 36 or so, he belonged to a faculty eating club called "The Apostles". He would rise early - when he had not been up all night computing or, on clear nights, observing - put on a clean shirt and one of his two somber suits, and go to his club for breakfast. He would then go to the office, where in my thesis years he would find me banging away after an all-night computing session, and take down Suitcase 14. Opening it at his desk, he would take out his electric razor (the first one I had ever seen; they were quite new) and a small round magnifying mirror, and shave. Putting these tools carefully away, he would next take out a large paper envelope containing one empty and six laden smaller ones, labeled for the days of the week. Slipping the noose of yesterday's necktie over his head, he would put it in the empty envelope. Taking today's equally dull necktie from its envelope, he would put it on, draw the loop tight, and be ready for the day. Except for infrequent trips to astronomical meetings, he did this every single day of the year. He was never ill.
I have mentioned that the computing forms were often replaced by improved ones. Obsolete forms were cut in half and saved, since their blank back sides made admirable scratch paper for calculations, first drafts of letters, and such. When I was made Keeper Of The Suitcases in 1937 - or to be more precise, Indexer - "we" had three suitcases of scratch paper, and the number grew to five by the time I left for the Naval Observatory. Maxwell's rather small technical library was in university-issue bookcases; his professional correspondence was in suitcases, his tiny family mail likewise; his skimpy loveletters and miniature diaries were in a locked tin box in his desk (but the spare key was in a carefully labeled envelope in Suitcase 1).
Our days revolved around two other possessions, however. When he made me a computer in late 1935, he had two Monroe calculators on trial. One was full size, and had electric shifting - you held down the plus or minus bar for the six or less revolutions of a multiplication (for seven, say, you [-44-] "shortcut" by doing one forward, shift, three backward; shortcutting saved 27 percent compared to just holding down the motor bar and gritting your teeth at the nines), then shifted with an adjacent bar. The other was, ahem, a mini, and you shifted the carriage with a knob (left hand) while you held down one motor bar or the other (right hand). It had never occurred to me until I wrote these lines that there were no left-handed desk calculators, manual or electric!
I did the check computing on the Delporte Object with the mini, and Maxwell used the bigger and faster machine. We agreed the other way around would have evened out our speeds, but he had additional chores to do while I was plugging away, such as decoding the observations coming in by wire. Oh, and teaching his classes!
Next year [1936] he tried the equivalent Friden, which shortcuts automatically when you push the multiplier digit on a separate keypad, and the top-of-the-line Marchant, which had multiplier keys up the right edge of the main keyboard, and a higher RPM. In the end he bought the mini-Monroe and the Marchant, Model ACT-10M. And the observatory had a communal Heavy Old Monroe, which sat in the conference room. It had been retrofitted with a giant external electric motor, and experiment revealed that if you pushed down all of the keys on the keyboard - that is, ten in each of the ten columns - held down the motor bar, and switched on the motor, the entire sixty-pound contraption would leap at least half an inch vertically.
It drew only a few users, and most of the staff used logarithms - the astrophysicists, Maxwell and I would say scathingly, only needed six-inch pocket sliderules.
My esteemed preceptor - and, needless to say, good friend, although I always called him "Dr. Maxwell" until I left his tender care - had three goals. He wanted to get married (I considered that a euphemism, but bluntness was not appreciated), he wanted to have a nice car, and he wanted the best calculator available. But he only earned $3500 a year, so he settled for buying his own machines (the Marchant was over $600, even with a professorial discount) and a drab second-hand Chevrolet, and yearning after the pretty young girls he saw going to the classes of the other Apostles. Alas, he complained, no such lovelies ever elected Astronomy 101, let alone 201!
He made no effort to steer my course selections in the following years, mostly because he approved of my choices. I took all the astronomy courses, a great deal of mathematics - not too pure - and as little other than sciences as the regulations permitted. But I was deeply interested in the outside world; as my finances improved I joined a local rental library, brooded over Mussolini and Hitler, and for broader themes managed to get a stack permit for the huge Main Library, where I read mountaineering and city planning and fine bookbinding, and Rider Haggard. The Observatory, physics and math libraries were very good, and Curtis would approve special purchases for me (and other apprentices); I still remember his permitting me to get the Oberth text on rocketry, and another serious German book by Hohmann about space travel trajectories.
Maxwell had not been infected with academic greed. Where the poor physics grad students were doing research in the name of their advisors, or [-45-] helping build the Michigan cyclotron - not only without credit but mostly even without pay - my guy was routinely making me co-author of "our" publications and even nominating me for membership in the American Astronomical Society - at twenty! Sure, he yearned for a disciple, and wanted me to stay around, but he was naturally generous - and disapproved of the way other youngsters were exploited.
What has stayed with me, and what (mostly through me) he contributed to the philosophy, the morality if you will, of modern computing, were the ideas of economy and elegance. The first lessons were elementary - don't work to seven places if five will do, use the special qualities of the machine at your disposal, make your computing forms carefully and revise them continually. Then he showed me the ways of using higher differences in tablemaking, so as to minimize the size of complex calculations, and introduced me to the publications of one L.J. Comrie, then head of the Almanac Office at Greenwich. The economy he had had to learn because he worked alone so much; the pleasures of accuracy and neatness and elegance were natural to him.
It wasn't all sweetness and light. I got uppity, and poked fun at some of his idiosyncrasies, and he sat on me firmly. For nearly a year [1938/39] I drifted away from him toward what you might call computational astrophysics: the use of fancy applied math to explore stellar interiors, following the work of an Indian at Yerkes named Subramanyan Chandrasekhar. I got excited about mechanized computing - punched cards - after reading Comrie and Wallace Eckert, and was unhappy when Maxwell would not introduce me to the local expert in Angell Hall, who did much administrative stuff and a little statistics, and years later published one of the first books on matrix arithmetic.
What drew me back in the end, along with the old friendship and his continuing support, was another discovery. An astronomer at Mount Wilson named Seth Nicholson, who was allowed to use the 100-inch for old-fashioned direct photography, announced two new satellites of Jupiter, X and XI, and the recovery of VIII and IX, which had not been identified for several years. I was intrigued with the odd problem of preliminary satellite orbits, which didn't seem to be in the textbooks. I was about ready to start thesis research, having decided to pass up a master's degree as not useful in astronomy (holders seemed to end up as planetarium lecturers, and NASA and the missile boys were two decades ahead). How about the new Jupiter moons?
Maxwell was willing, but pointed out that the observational material was pretty skimpy for a definitive orbit, and that out at the edge of Jupiter's gravitational dominion the solar perturbations would be enormous, and therefore hard to calculate. I set to work to investigate. Much to my surprise, I found that multiple solutions were possible. On reflection, you realize that an observed direction would be the same if the satellite were "in front" or "behind" Jupiter - that is, less or more distant from the observer than its primary - and that four combinations of early and late observations were possible. Three of these fit the mid-range data only for very strange orbital parameters; the fourth was probably right, but you could only be sure after the new satellite had been observed for a near-revolution.
I had won a super fellowship for my first full graduate year, 1938-39: one [-46-] of the new Rackhams, a gigantic $1000. I got my teeth fixed, bought a better radio and a three-speed bicycle, and still had money left over. I decided to attend an all-summer conference at Harvard Observatory, where Fred Whipple and Leland Cunningham, contenders with Maxwell and Grosch for the orbit computing championship, held forth. I gave a conference talk about the multiple solution thing, which was well received, but found out from Whipple that Paul Herget of Cincinnati had sort of staked out the two new satellites, with the concurrence of Nicholson.
I wrote Nicholson, and he replied promptly that Herget was really being a little ambitious because of observational deficiencies, that IX seemed to be lost again, but that there was a wealth of data going back thirty years for VIII. Moreover, VIII seemed to be the most highly perturbed satellite in the solar system, and had already been lost (and recovered) three times. "A really good job would not only keep that from happening again, Mr. Grosch, but one might later make a very valuable determination of the mass of Jupiter" (relative to the Sun, that is). I announced to a waiting world - about six people, including not only Maxwell and Curtis and Nicholson and Whipple and Herget but a woman in Leningrad named Boeva, who had done the best orbit so far - that I would massage VIII to a fare-thee-well.
It was probably more than I should have tackled. A reasonable thesis topic would have given me my Ph.D. in mid-1941, at 22. The usual age for the degree in those simpler times, at least for astronomers, was 27 or 28 if you worked right through (unusual even then). As it was, the huge mass of calculations took 2500 hours of all-night Marchant work - I kept meticulous logs, as Maxwell would have done - up to May 1941, when I left Ann Arbor for the Naval Observatory job, and many hours more in Washington. Still, I got done with everything except the writing by Pearl Harbor time, and was able to take my doctorate at the 1942 Commencement.
This chapter is not really about my doctorate, but about an early and effective computer now almost forgotten. There were dozens like him, and not just in the United States. Some were innovators, like Wallace Eckert. Some were caught up in the early machine revolution, like Paul Herget. Comrie, by far the greatest of the clan, was elected to the Royal Society for his contributions, but died before the Big Boom. Maxwell, having helped me so much, dipped a toe in the waters and pulled back.
He wrote me about his discouragement with astrophysicists. Should he follow me to the Naval Observatory (I was well into IBM by then), and would his preference for and his expertise in hand computation be valuable? "Allan," I said, "what you do is basic to any kind of calculation, and vital for astronomy. But these places are going to get electronic machines. Can you still work with youngsters using them, who need to be schooled in numerical analysis, and encouraged to do better than just grind away?"
He went. The Almanac Office got an IBM 650. He returned to teaching, in Washington, and told me the last time I saw him that he had more or less retired his desk calculators, and reverted to logarithms. He found it dull to sit there and watch the machines - even the Marchant - "do all the work".
But I was delighted with him that day, and for a very intimate reason. While all this was going on, he had finally married. Mona was an Ann Arbor [-47-] town girl, proud to be the partner of such a clever and remarkable man. After all those years, he no longer needed fancy desk calculators. He had a better car. And he had found a wife.
He was the best computer I ever worked with. He died some years ago. I haven't forgotten him.
In Chapter 06 you will encounter
(in order of appearance):
Maxwell 05
Leland Cunningham 02
Fred Whipple 05
Harvard College Observatory 05
J VIII 02
Dirk Brouwer leading celestial mechanics researcher, from Yale
James Baker already a superlative optical designer
The Ritchey-Chrétien 02
Harlow Shapley director of the Harvard Observatory
Dowse Institute lectureship amazingly, they paid me to talk
Frank Edmondson he drew the Packard but I drew a fiancée
Kate Gordon she let me drive!
Dorothy 01
Chinese food Boston's Chinatown started me off, and with Mandarin
L.J. Comrie 01
The Smithsonian Astrophysical not astrophysics, but all Whipple
OSA 02
Tom Lehrer entertaining in a Winchester basement
The Harvard Society of Fellows fostered Baker's optics magnificently
Howard Aiken 01
The ASCC 01
Skew rays grim Aiken let Baker trace just one
Rudolf Kingslake 02
Perkin-Elmer 02
The New York World's Fair I took notes on Gypsy Rose Lee for Maxwell
Parents 04
The Junior Astronomer examination Civil Service paid better than Academe
FORTRAN 01
Steam power love letters to Pasadena behind 60-inch driving wheels
Mount Wilson visiting astronomers stayed in Kapteyn Cottage
Griffith Planetarium one lecturer was E.C. Bower
Palomar the dome was ready but the mirror was still at Cal Tech
Engagement it looked like a long wait, until the Navy came through
[-49-] Maxwell and I more or less assumed I would follow in his academic footsteps, although certainly not at Michigan or Harvard or Berkeley or Cincinnati, where there was an ample supply of orbit computers - that is, one at each institution. We talked idly about the circumstances of Leland Cunningham, who was an assistant to Fred Whipple, Maxwell's counterpart at Harvard. We decided that, as Cunningham did not seem to be getting a doctorate and did not seem interested in marrying, he probably could stay on at Harvard indefinitely, but at bicycle rather than used-car (let alone new-car) wages. That didn't attract me.
My orbit of Jupiter VIII progressed forward from the 1938 observations through 1939, where I gleefully discovered that Nicholson's single published observation at the 100-inch was of something else, and was able to tell him where to look on his plates to find a genuine image. I was beginning to run the calculations the other way, to pick up and use the 1935 data, when the summer of 1940 rolled around.
I had gone to the conferences at Harvard the year before, at my own expense, and the 1940 version promised to be even better - a course by Dirk Brouwer, undisputed international champion of fancy celestial mechanics, and an unusual offering by a Harvard Society of Fellows chap named James Baker. The latter was giving lectures - perhaps the only ones he ever gave, since the war carried him upstairs the next year - on his unusual optical design methods. Some pages back I mentioned my attempt to design a field-flattening lens for the peculiar Naval Observatory reflector - well, this was the methodology. Baker turned out later be the very best lens designer in the whole world, and the father of the Baker-Nunn cameras NASA used so enthusiastically in the Sixties (and of their classified variants orbiting overhead).
Trouble was, I was not as flush as I had been in 1939. I still had a good [-50-] fellowship, thanks to Curtis and Maxwell, but not the lavish Rackham. So I wrote Harlow Shapley, the famous head of the Harvard Observatory, and asked his help. He produced a Summer School tuition scholarship, and a public-night Dowse Institute lectureship. I talked about the satellites of the solar system, and got paid $100 - the very first money I ever earned by my speaking talents. Ham that I already was, I loved it - both the lecturing and the pay!
When Maxwell shipped me off he counseled me to be alert for attractive women the first few days. "I have been to many such conferences, and taught summer classes, and the pretty ones get taken early", he said mournfully. There had been two or three possibilities in 1939, and I had indeed been too late to pick one off. Years of all-male graduate science classes had blunted what little negotiating skills I had had in high school, and the four dates I had managed with a lovely but wary creature from Michigan State that spring had not re-sharpened them all that much; nevertheless, I promised to do my best, and report. We had shared many a Saturday night expedition to the Detroit burlesque theaters, but it didn't seem likely I would attract anything that spectacular.
A few days later I found myself and a reasonably attractive young professor named Frank Edmondson paired off at Harvard with the two prettiest female attendees. By the luck of the draw - who was standing closest to whom - Frank went with Katherine Gordon, who later turned out to have driven up from the family apartment on Park Avenue [!] in her Packard convertible [!!]. My partner was a Dorothy Carlson, from the Mount Wilson office in Pasadena - yes, she knew Nicholson well - who had been saving for two years to see "the East", and meet some younger astronomers. She was delighted with the arrangement, since it turned out Frank was married. I was delighted because, in order to share the rumble seat [!!!] with Frank, Kate let me drive the convertible on all the double dates. Quelle exstase!
Dorothy and I had a wonderful time together. Because it was my second year at the conference, because I had been helped very substantially by the ineffable Shapley, and because a nice young couple was easier to accommodate socially than awkward male singles, we were invited everywhere. The Shapley family had a complete set of Gilbert and Sullivan, album after album of heavy shellac 78s, and we went to their "evenings". Afterwards we would wander around the dark grounds of the observatory, and perhaps lie on the opened roof of the patrol camera building and help watch for meteors. Much handholding, much kissing, much talk of futures (my future, mostly, as I remember).
We went with the younger crowd to Boston's Chinatown. I had barely mastered middlewestern chop suey, and only with a fork. This bunch was full of experts. Dorothy was already more used to the Chinese cuisines than I, having lived in Pasadena for years and gotten her degree from Berkeley, but managed not to let me see it too often. Mandarin was considered advanced in those days; Szechuan and such had not arrived. Still, I learned a lot.
There were trips to the country, notably a lightly chaperoned one to the main observing station in a darker part of Massachusetts. And we shared many good evenings with Kate and Frank (and the Packard). I reported most of this [-51-] to Maxwell, emphasizing my charming companion's openness (and Mount Wilson connections) and not her willowy figure, which would have disappointed his stripteaser expectations.
I also reported on intellectual matters. I was able to observe Whipple and Cunningham close up, and discovered that Cunningham filled more than the assistant role; in fact, was the real force in the computing work. Neither of them had Maxwell's passion for neatness, nor his interest in the performance of desk calculators, nor in Comrie. But it was obvious even to me that Freddie was already a skilled politician, worthy of his Harvard tenure. The calculating work done during many later years at his so-called Smithsonian Astrophysical installation confirmed my 1940 assessment; it was not astrophysics, it was at Harvard rather than at the Smithsonian, it got major press attention - and it was Whipple through and through.
Baker was harder to fathom. I was enormously impressed with what he had done, none of it reported in the literature. He didn't hold back on his methods; on the other hand, he talked equations rather than practical computing - how I wished he would hand out forms, like Maxwell, or tell us how many trials he had to make to finish one of his super-Schmidts! He had a Friden in his office, but clearly did most of his work at home. Later, in the Fifties, he had a major long-throw optical testing laboratory in the huge basement of his Winchester house. (He threw a party there for Optical Society people, and hired Tom Lehrer, still a Harvard student, to entertain). He was by then the design genius behind Perkin-Elmer, and the source of the very best aerial camera lenses. He was elected president of the OSA. But somehow he never hit the publicity levels of, say, von Neumann or von Karman.
Two more stories about Baker, before I go back to the mating game. He had a curious problem at the end of the war: the Society of Fellows, of which he had been a member for the years while he developed his methods and began to demonstrate them, did not permit course credits. You were supposed to do research and think beautiful thoughts, and some day drop back into routine Academe for your degree. Jim did all this magnificent war work, and came back (after being fawned on for several years by two-star generals and company chairmen) to the realities of astronomy: no Ph.D., no instructorship.
Harvard was great; they called his secret NDRC reports and super lens designs his "thesis", and awarded him a doctorate sort of by acclamation. The astronomers, however, were unforgiving - jealous of his reputation, which was spread vigorously by the Perkin-Elmer organization to boost its already excellent market position - and relegated him to a trial untenured slot out West. "Do some real astronomy", they said, "and then we'll see". He took it for a while, and finally went back to optical design; that was better for everybody, astronomers and optickers and the whole world of technology. But it was still a sad story.
The other yarn is more technical. When Watson gave Harvard the ASCC, Baker went to see the irrascible Aiken. High-aperture lens designs were known to suffer from a computing lacuna. Desk calculation was too slow and expensive to permit the tracing of skew rays. These were rays which did not stay in one plane - the plane of the lens cross section. They would have taken six or eight times as much work per surface as the ones usually traced; [-52-] moreover, the designers, since they never traced skew rays, had also never worked out how to use the results.
You designed the lens in two dimensions, and tested a sample on the bench (the effect of skew rays could be examined by using appropriate masks). The folklore of lens design, which was the stock in trade of Kingslake and his father-in-law, much less familiar to Baker, and hardly known to me, said that certain types of fast lenses, notably the Zeiss Sonnar family, suffered less from wandering rays than most. Some of Baker's methods bypassed this difficulty, but he wanted to explore the problem in detail.
Aiken reluctantly let him write a skew ray program for the big machine, and trace one ray. Losing patience with the whole thing - perhaps partly because the ASCC was by this time the, ahem, Navy MARK I, and Jim's wonderful designs were for the Army Air Force - he ran the protesting young man off the premises. When I reported tracing thousands of skew rays on the SSEC in 1949 (and, not accidentally, through a Sonnar type), Baker told me he liked my formulation, and had used a less satisfactory equation "on Aiken's machine", which is how I first heard the story. When I pressed him to say whether he would use the new machines that were on the drawing boards, he put on his Perkin Elmer hat. Of course he and his assistants did in the end.
Well, Baker's course and Brouwer's course, and the ones Dorothy was taking, came to their ends. I pressed her to come down to the New York World's Fair with me (absolutely no chaperones) but she had family visits to make. In the end I went alone, took notes on Gypsy Rose Lee and her competitors for Maxwell, and got back in time to take Dorothy off the train in Detroit the following week. She met my parents, who were a little reserved - they feared for that doctorate. We went out to Ann Arbor and saw Maxwell, who approved thoroughly. That evening I put her on the train to Chicago and California. We would meet at Christmas, in Pasadena, we agreed.
I ran Jupiter VIII back toward 1935 enthusiastically. But I also resumed talking to Maxwell about a job. He thought things were rather quiet in all the astronomical fields at the moment; there would be post-doctoral fellowships at a few places like Yerkes and McDonald (in very remote Texas), "but not for orbit men". I cast about. There was an announcement in SCIENCE of an examination for openings at the U.S.Naval Observatory: Junior Astronomer, $2000 a year. Now, the starting salary in 1940 for an unmarried instructor was around $1600 - maybe $1800 if married, but that was not considered too sensible, at just 28 - so in spite of the high cost of living in Washington, the stipend sounded pretty attractive. And it would be my kind of astronomy, not an appendage to astrophysics.
This was a competitive written technical examination, administered at the Ann Arbor Post Office [!] by a Civil Service official. Different from 1986: instead of writing a FORTRAN program (or describing in poor English how experienced you were at writing FORTRAN programs), you sat down with curious logarithm tables furnished by the examiner and actually did sample calculations. It was easy for me; the budding astrophysicists across the nation must have sweat gallons - well, liters!
I was surprised to run second in the country, of over a hundred aspirants; I had assumed I would be an easy victor. In fact, the victor, whom I knew slightly, was named Victor, which probably accounted for his victory. He [-53-] ultimately did not take up the post, and in the early spring [1941], after a marvelous Christmas, I got a stodgy telegram from the Secretary of the Navy - well, that's what it said - offering me the job. Needless to say, I packed up Jupiter VIII and went. What I found across the hall from my new office is described in the following chapter.
About that Christmas: Dorothy and I corresponded four or five times a week all through the fall. I would bicycle down to the Ann Arbor station on my way to an all-night session with the Marchant, and post my love letter in the slot on the side of the Chicago-bound mail car as it came through just before midnight. Behind steam, big beautiful 60-inch-driver steam - what a thrill to see your increasingly urgent literary efforts go ka-chuff, KA-CHUFF, KA-CHUFF away to California. A diesel would have seemed second rate!
We arranged I would come out to sunny Pasadena by Greyhound - three days, five busses - as soon as school shut down. That got me to Dorothy on Christmas Eve, in time for Swedish Lutheran festivities and family celebrations. I was to stay a week, meet Nicholson of course, spend one night "up on the mountain" - that is, in Kapteyn Cottage (for visiting astronomers) on Mount Wilson. We were to visit Griffith Planetarium, the Cal Tech workshop where the 200-inch mirror for Palomar was being ground, and go down to Palomar itself. John Carlson donated the family car for much of this, and Dorothy's mother Ida and brother Bob were fairly happy with me also.
Southern California was still smog-free. There were lots of orange groves. Each evening we parked up among them in Altadena, and shared a wonderful private world - after the windows steamed up, that is! We were definitely engaged. We would marry when I found a job. We would not "spoil things" by endangering my doctorate, however, so it looked like a rather long wait (I had not yet heard that the Naval Observatory job was still unfilled). The war in Europe worried both of us.
There was a small problem about ages; Dorothy was six years older than I, and nervous about my parents, and criticism. I pointed out that I was now running five years ahead of the pack in every other aspect of my life, and that she could not just, ah, rob me of my precious virginity and run off into the bushes (a very unsuccessful joke)! We agreed most people would assume I was her age - and that I would still be the youngest wolf in the pack.
Decades later I looked around and found I was the oldest wolf in a very, very much larger pack. I tried to remember the crossover but found I couldn't. The biological clock keeps not only nonlinear but discontinuous time.
In Chapter 07 you will encounter
(in order of appearance):
Maxwell 05
The Delporte Object 05
L.J. Comrie 01
PLANETARY COORDINATES Comrie's Cartesian view of the Solar System
Greenwich Observatory the fundamental meridian, and Comrie to boot
The Apollo Group Delporte found 214 asteroids, but 1936CA was an Apollo
Brunsviga desk calculators you didn't need a wall plug
Crossfooting how bookkeepers added horizontally (they "footed" columns)
The IBM Bank Proof Machine summit of non-automatic data processing
Jacquard's loom A but not DP, in 1801!
Wallace Eckert 01
Hollerith machines what the British called punched card equipment
British Tab where Comrie rented his Hollerith machines
E.W. Brown master of the Moon's motion, and forerunner of Eckert and Brouwer
PUNCHED CARD METHODS IN SCIENTIFIC COMPUTATION my second career signpost
The Naval Observatory 02
Ben Wood Watson Senior gave him an unusual horizontal tabulator
Watson Senior 01
Nicholas Miraculous [Nicholas Murray Butler] president of Columbia
The Thomas J. Watson Astronomical Computing Bureau 03
The sequenced IBM 601 you could change the plugboard without changing the plugboard
The American Air Almanac 02
The American Nautical Almanac for "gnarled fishing boat skippers"
The American Ephemeris the astronomical source for the other two
IBM 01
Lillian 03
Jack Belzer "all about punched cards ... and nothing about astronomy"
Captain Hellweg titular head of the Naval Observatory
John Willis 02
The Millionaire calculator a horrid step down from the Marchant
Dorothy 01
The IBM 405 tabulator a ton of machinery in a huge black box
Proofreading the bane of tablemakers; Eckert rendered it obsolete
Errors in the Air Almanac in tens of millions of digits, not one!
Dick Bennett an IBM customer engineer with superhuman patience
IBM wartime benefits Watson Senior cared about members of the "family"
The IBM table-printing typewriter many number fonts, and elegant spacing
Marriage Dorothy came East, and we did the deed in Washington
Our 1937 Plymouth little had it dreamed of gas rationing
Pearl Harbor 02
The 1942 Michigan commencement Maxwell called me "Dr. Grosch"
[-55-] Before I get into the story of the world's first permanent rent-paying scientific punched card computing laboratory I have to loop back to my earliest days as a computer, in Ann Arbor. Maxwell had set me down in front of the Monroe and explained very briskly indeed - he was, in the inelegant phrasing of today, hot to trot - how I was to check his preliminary orbit calculations. He had taken away his samples, so I would not be tempted to duplicate his undetected mistakes, and left me to it.
Years later, when I was building up my first major staff in unexplored General Electric, I used the same tactics. I called it the swimming pool method: you took the victim over to the edge at the deep end and gave a vigorous push. If you had picked your candidate for initiation correctly, he or she learned to swim very quickly.
After you had a few swimmers, you moved on to the buddy system, which was in vogue through most of the Fifties at 701 and 1103 installations. Fewer drownings!
Maxwell and I did a series of better orbits of the Delporte Object; it had by then been assigned the temporary identification 1936 CA by the international astronomical telegraph bureau which was sending out further observations. Each of these efforts covered a longer stretch, and also utilized data from a gaggle of remote observatories like Algiers which wanted to get into the act. Things were a little calmer in Maxwell's sanctum, and I had time to wonder what was in all those suitcases!
In early 1936 "we" decided to do a definitive orbit. The little rock that was causing all the excitement had now receded too far for even the great telescopes like the Yerkes 40-inch refractor to pick it up any more, and forthcoming observations from strange places could be added at any point in [-56-] the final cycle. The big problem was to allow for the close Earth approach, which clearly must have perturbed the original path around the Sun that we were approximating with the preliminary orbits. Maxwell said, "This will be a nice variation, since we will need to separate the influence of the Earth and the Moon - so we can't use Comrie." Who was Comrie?
"Comrie" turned out to be an oversized but thin purple volume called "Planetary Coordinates", which contained the rectangular coordinates of all the major planets for four decades, usually at 40-day intervals. It was put out by His Majesty's Nautical Almanac Office at Greenwich Observatory - where my father later told me he had installed cabinets when he was a young apprentice in 1904 England.
The author of this crucial tome was the director of the Almanac Office, Leslie John Comrie, whom I regard as one of the giants of the strange computer trade. He was a New Zealander by birth, and computer historians Down Under are now proudly excavating his record. He wasn't much of an astronomer, and Greenwich was one of the great centers of the science - historically the fount of most classical observation and calculation, proud of its association with Newton and Halley, home of the Astronomer Royal. And Comrie was not at all a whiz kid: not a Sagan, or even a Whipple. What he was, was the inheritor and last monarch of the kingdom of mathematical table making. The work of the national almanac offices - Greenwich and Washington and Paris, and the German and Russian and even Argentinian equivalents - was all calculation and table-making; the observational work was done by others. He was the top, and at the top.
Maxwell explained this over the first months of our association. Comrie was "not an orbit man," but he and his people, and to a lesser extent their American opposite numbers, furnished all sorts of tools without which orbit computing would have been almost impossible. Even today, with Crays popping up in every corner and more orbit experts in NASA (let alone DOD, and Europe and Japan, and Russia and China) than in all human history up to Sputnik, the almanac offices produce the fundamental numbers. Their calculations, which in the Thirties were accepted as the most intricate in human experience, now occupy less than the full capacity of a supermini. As Galileo said, things sure do move!
Since the purple masterpiece tabulated, not the position of the Earth, but the position of the center of gravity of the Earth-Moon combination, Maxwell had to work out new ways of using other data from the ephemerides - the annual compendia of astronomical positions which were the main output of the almanac offices - to get the positions of the Earth and Moon separately, and relate those to the figures in "Comrie." I was deeply immersed.
From our best unperturbed orbit we calculated the approximate coordinates of 1936 CA; from the newly produced coordinates of the Earth and Moon we calculated their influences on the asteroid (even though the little rock had come within a million miles or so of us, the pull of the Sun was very much greater). Allowing for these perturbing forces, we recalculated the basic constants of the asteroid orbit. The idea, of course, was to keep track of the object and pick it up several revolutions later, if it came close again. Nobody except a few weirdos who read science fiction [ahem!] were yet doing Lucifer's [-57-] Hammer; the idea was to more closely determine the scale of the solar system (now done directly using radar and laser measurements) by triangulation of close approaches.
Before I get back to Comrie and such, I have to say that we lost the little planetoid. All the efforts since 1936, when Maxwell and I put out our major publication (my first), which have featured much more elaborate perturbations and used modern computers, have failed to predict a recovery. There is now a family name, the Apollo Group, for asteroids that come as close to the Sun as the Earth does, and 1936 CA, permanently christened Adonis by the discoverer, was the attention-getting member of that family. Delporte found 213 other minor planets, but Adonis was his masterpiece.
In addition to many volumes of mathematical tables, culminating in a two-volume set of six-figure material published proudly by Chambers, and all the astronomical and navigational tables from Greenwich, Comrie kept up a remarkable flow of methodology papers in a wide variety of British publications. As an extreme example, he even made suggestions on how to do optical ray tracing, some of which I incorporated into my own efforts during the war.
He refined the use of hand-cranked Brunsviga desk calculators, including unusual variants like a model with two setting barrels that did Ax and Bx when you cranked in x (or even, with a reverse gear, -Bx!). He explored the use of enormous multi-register bookkeeping machines for differencing columns of figures, or to build up and print tables from higher order differences; the ugly term was "cross-footing," which hung on in IBM manuals until the early Fifties - in fact, until IBM quit painting the old clunkers black. In all this, his concern was to use the exactly optimum devices, mathematical and mechanical, for a given task; he was undoubtedly the culmination of his race, and the best we will ever see: a very Tyrannosaurus Rex of non-automatic computing. Yes, the breed is extinct.
But the reason I think so very highly of him is not his stubborn precision, not his expertise in using old- fashioned machinery, but the exact opposite: his vision, his appreciation of novelty. You have to remember that to use a two hundred pound key-driven bookkeeping machine was innovative, in the Twenties. To find optimum devices he had to try new things, and sometimes overshoot. I suppose he encountered more resistance in his (by today's standards) narrow world than Gene Amdahl or Bob Noyce did, in the fantasy universe of microtechnology.
Those clanking monsters, and the IBM Bank Proof Machine of the same era, were the summit of non-automatic information technology - that is, of computing devices that required one human operator per machine. Automatic machinery, not necessarily electronic or even electrical, was being built by Jacquard at the beginning and by Hollerith at the end of the nineteenth century. Bookkeepers and their ilk, and later, statisticians, were exposed to its power. Comrie was the first astronomer, the first scientist in fact, to put it to work. Wallace Eckert, one of the heroes of my first chapter, who was in a dozen ways Comrie's opposite number in the U.S., acknowledges his enterprise repeatedly.
In 1928 Comrie published his experiences in using punched card ["Hollerith"] equipment for table making, and in 1930 he dug down into His [-58-] Majesty's shallow purse and rented a set of machines for the Almanac Office. His young ladies punched up the fantastically complicated double- and triple-entry tables of the motion of the Moon, due to E.W. Brown of Yale (Dirk Brouwer, who drew me to Harvard in 1940, was Brown's successor). The systems design, to use modern terminology, must have been extraordinary. But that was only the beginning; he then made up a punched card for each date for which he wanted a lunar position - every few hours for several decades - and ran the huge pack through step after step after step of intricate multivariate interpolations. There were problems with checking: the results had to be perfect, not only for the astronomers around the world who would use them for the rest of the century, but because lives depended on the navigating tables which were distilled from the Greenwich data.
After many months of the most detailed and demanding effort, the task was done - and Comrie promptly returned the machines to British Tab; he could not justify keeping them for the other work of the office. He published the story in the major British astronomy journal, to which his eminence gave him access (it was not exactly ordinary 1932 research material). It was read bemusedly by the old-timers (which I have to admit included Maxwell), enthusiastically by a young Yale graduate student named Wallace Eckert, and in 1938 by an even younger undergraduate at Michigan whose fingers were coupled to a very non-automatic Marchant!
You will meet Comrie again a little later, when the computer revolution was noticeably under way, and when he was honored in the U.S. But let me go on to another publication which enthused that Michigan student. It was a book called "Punched Card Methods in Scientific Computation," and it was announced in the American astronomical journals as I got ready for my second Harvard summer.
It has recently been reprinted by the MIT Press as the fifth of the Charles Babbage Institute series on the history of computing. I have done a review which recalled its bright orange 1940 cover and labeled it "the second signpost I encountered on the computer road" - the road of the user (the first signpost was the 1932 Comrie article, of course). Even with a dull cover, it makes remarkable reading.
The author was Wallace Eckert, and it is now time to identify him more completely. He was the second man to strongly influence my career, although by the time I went to work for him in 1945 I had my rudder welded firmly in place. It was his book, and what I saw of his work at the Naval Observatory, that had the real effect. Those two things, and my excitement about Comrie's experiments, weaned me away from Maxwell's conservatism and steered me toward punched cards and the automation of computing processes. When the little man from the Manhattan District came for me, I was ready!
Eckert was a graduate student under Brown at Yale. As I intended under Maxwell to become an old-fashioned orbit computer, so he intended to become an old-fashioned celestial mechanics researcher. He was intimately acquainted with the lunar tables when Comrie was mechanizing them, knew the details of the work at Greenwich before final publication, and like me in Ann Arbor dreamed of mechanizing the arduous calculations of classical astronomy - but dreamed of it five or six years before I did.
[-59-] As I later went off to the Naval Observatory before finishing my doctorate, so Eckert went off early to Columbia. He got his degree somewhat before the Comrie article came out, however, and was ensconced in Academe before the Depression struck - and it was less severe in New York than in Detroit. Even so, he saw clearly that a good set of punched card equipment would be far too expensive for the Columbia astronomy department.
Across campus, in the famous but much less purist Teacher's College, Eckert observed a statistical research set-up featuring a slightly modified IBM tabulator (a ton of gears, electro-mechanical counters and circuit breakers driven by a big motor). The proud "owner" was a Ben Wood, who had benefitted from an early dose of Watsonian foresight, or charity (or both). T.J. was an extremely influential Columbia trustee, with direct lines in to the equally famous Columbia president, Nicholas Miraculous [Nicholas Murray Butler].
Eckert enlisted the support not only of his department and the eminent Nicholas but of the American Astronomical Society, and approached The Great Man - the first of many, many times he got to sit outside Mr. Watson's busy office. The result was the Thomas J. Watson Astronomical Computing Bureau I have already mentioned. The work Eckert did there, considerably more advanced than the huge job Comrie had done at Greenwich, culminated in the publication of the Orange Opus.
Where Comrie used standard British machines, and Ben Wood had only a somewhat improved giant adding-subtracting tabulator, Eckert obtained from the IBM laboratories a multiplying punch with added sequencing capabilities. He could change the plugboard without changing plugboards, so to speak. And with it, making my mouth almost literally water as I read, he was able to integrate an asteroid orbit just as I was manually integrating my Jupiter VIII orbit. The details are excruciating, and even at that Eckert left out how he had to use a clothespin on the end of a stick to extend his reach!
He probably couldn't do a time step much faster than I could on Maxwell's Marchant, and I didn't need a clothespin. If you worked out in detail, as Comrie would have done, how much it was costing (including commercial rates for the IBM machines), my stuff was probably five times cheaper. But using the machines was fun, at least for a while, and people like Comrie and Eckert and Grosch and of course McPherson - and, remarkably, Watson himself - expected newer and more powerful machines every year. A better desk calculator might reduce my cycle in Ann Arbor from twenty minutes to fifteen; better IBM equipment would reduce Eckert's cycle from ten minutes to less than a second (on the SSEC, in six or seven years), and to a few microseconds today.
About the time the Orange Opus appeared, a wider door opened for Eckert. Those ephemerides that all the national almanac offices published were what astronomers and geodetic surveyors used. But each year the astronomical product was transmuted into a navigational volume, called in Britain and the U.S. the Nautical Almanac. The data was for fewer objects, and given less accurately, but for more closely spaced dates and times, so that it could be extracted easily by gnarled fishing boat skippers who used marine [-60-] sextants rather than meridian circles. It obviously needed to be more easily readable and useable.
The human computers and calculating machines to do this were in place. But the humans were in short supply; each year some astronomical work had to be deferred because there weren't enough people to do everything. This was partly money; the Civil Service could have scraped up more Victors and conceivably even, more Grosches, if budgets permitted - but not overnight.
For some years the military had wanted the Almanac Office at the Naval Observatory to produce a third book, an Air Almanac, for the special needs of primitive aerial navigation. Bubble sextants, hence less accuracy; cockpits instead of pilot houses, hence even simpler user facilities (today we would say "very user friendly"); only a few stars, and the brightest planets; three or four slim volumes a year, instead of an annual one.
The money had not been forthcoming. Now the war was upon them; bombers were being ferried across oceans; they anticipated long over-water or over-cloud military missions. Money was available at last, but there was no time - and no more trained people. You couldn't reassign the ephemeris personnel, or there would be no fundamental data to boil down. You needed the marine book more urgently than ever.
The senior astronomers, and most of the computing staff in the Almanac Office, knew about Comrie and Eckert and their machine successes. An approach was made. Could the young Columbia professor put aside his academic career, and his IBM-sponsored laboratory, and come down to Washington? At the same time, the Navy told IBM of its problem, and asked for a quick and sophisticated response. It came together quickly; IBM would rent the Observatory an assortment of its best machines, and make sure that exceptional customer engineer service was provided; Eckert would become the director of the Almanac Office, the equivalent of Chief Scientist today; a small cadre of operating personnel would be hired. The Air Almanac would have top priority, with other work to be transferred to the IBM machines when opportunity offered.
Eckert left the Columbia astronomical bureau in the hands of his capable assistant, Lillian Feinstein Hausman. To run the machines at the Naval Observatory - and he understood he would be far too busy to do it himself - he and Lillian looked around New York, and recruited a youngster from the cloak-and-suit arena. He knew all about punched cards, a lot about IBM, and nothing whatsoever about astronomy. His name was Jack Belzer.
The installation was under a full head of steam when I arrived, eight or ten months later. I tried to get the powers that were - not the Secretary of the Navy, for sure, but senior people at the Naval Observatory whose names I knew - to assign me to the almanac operation. I had no advance knowledge of Belzer's shop, unfortunately; the notices of Eckert's apotheosis had carefully not emphasized the novelty or given details of the appointment, although Maxwell and I had nodded sagely to each other and said "punched cards." I wrote Eckert, with whom I had had no earlier direct contact, and a Dr. H.R. Morgan, head of the fundamental meridian circle work and of the internal scientific council which advised the superintendent (a Navy captain of absolutely no distinction, named Hellweg).
[-61-] Eckert replied that there were no vacancies in the Almanac Office - true, in the sense of authorized Civil Service openings, in spite of the military urgencies building up at every scientific institution in the government. In a later chapter I will tell how the same sort of bureaucratism got in Wernher von Braun's way in 1957, when the race to orbit a satellite was heating up.
Morgan told me I would be assigned at first to the Equatorial Division, but that in view of my interest and advanced training in "theoretical" [computational] astronomy the observatory would try to move me over to Eckert's shop "in the future" - that is, when an Assistant Astronomer slot opened up. He then went on to detail the skimpy arrangements for my actual arrival - no travel or relocation expenses in 1941!
I was to share an office in a small, rather new building with John Willis. John was an old hand in the Equatorial Division, and sort of in charge of the unpopular Ritchey-Chrétien reflector (not much of an observing program, and hence not much observing). He did little computing - just what was required to reduce the measurements he made of the strange circular photographic plates. I was therefore allowed prime custody of "our" calculator, which was by a considerable margin the funniest (in both senses of the adjective) I had ever used.
It was a Millionaire, made before the turn of the century in Germany, and examples survive today in collections and museums. It had a gear shift. You cranked a non-ergonomic handle once, while with your left hand you selected one of ten multiplier digits. Clank! Crunch! And the carriage moved to the left, ready for your next assault. But that wasn't all. I had seen such a machine already, perhaps at Harvard. This one, though, was motorized!
Some crazed mechanic, probably from the observatory instrument shop, had slung a ten-pound electric motor underneath, and provided a leather belt up to "the works." There was a neat pushbutton on the top of the crank, and the latter could still be worked by muscle power when the belt broke - which was frequently!
I couldn't believe even Captain Hellweg would want me to live with such a beast, let alone finish my thesis calculations on it. But Willis assured me he preferred it to the little Brunsvigas that others in the building used, although he mentioned wistfully that "the Almanac people" - those who didn't prefer logarithms - had Marchants and Fridens. I reported all this to Dorothy, who was suitably sympathetic, but busy arranging to come East and marry me. She wrote that she was sure she could learn to operate it for me in the long evenings when I would be writing my thesis. I obviously was acquiring a helpmeet as well as a bedmate.
My Millionaire pangs were intensified by what I found across the hall. Because of space limitations in the Main Building, and I suspect also because Eckert was wary of complaints about the noise and heat of the heavier IBM machines, Belzer and his equipment - and, aha! two handsome young ladies - had been put out of the way in my building. I had had only glimpses of the stuff before, and the pictures in Eckert's Orange Opus didn't do it justice. The machines, and especially the big 405 tabulator, were huge. They were obviously expensive (although I now found out from Jack that they were only [-62-] rented, never sold). They were noisy --even the key punch and verifier, which the more nubile of the young ladies operated.
Eckert had a 601 multiplying punch on order (without the special features of his Columbia machine), and the usual sorter, reproducer, interpreter and collator. There may well have been a gang punch attached to the 405, by which accumulated totals could be entered into blank cards. There will be more about these fascinating - to me - black boxes later. The important thing for history buffs is that this was the very first permanent IBM installation in the whole world to pay regular rent and do scientific work. Comrie's Greenwich effort had not been permanent. Eckert's shop on top of Pupin in New York had been one of Watson's many benefactions. There were places all over the world doing fancy statistical work, up to and including actuarial calculations, which were as complicated as what Belzer was doing (although not up to the Comrie and Eckert-Orange-Opus level). But this adventure was unique; I understood that clearly. And I could go across the hall several times a day and watch.
Eckert had a special challenge, and Jack a different but also very important one. Eckert's solution remains a landmark - a completely forgotten landmark. Belzer's efforts were writ on water, but amusing to recall, in a day of page description languages and desktop laser printers. Eckert's Air Almanac was even more demanding than its astronomical and marine-navigational relatives, as regards errors. An error in astronomical data would set back a research program, and waste time and money. An error in nautical data was life-threatening; the ship might be wrecked if its position turned out to be in error. But the tempo was slow, and the sea floated the customer with relative safety.
In the air, things were much tougher. A navigational error meant running out of fuel, or missing a vital landing spot. And the book would have to be used under dreadful conditions - a tiny cockpit, poor lighting, heavy turbulence. Things happened fast.
The men and women who planned the book before Eckert and Belzer arrived did their very best to make it user friendly - simple to use, and fast besides. Belzer was responsible for making it readable. Eckert was to figure out how to make every single digit correct, and he did so. Decades later the Almanac Office could say that not one error had ever been detected in any copy of the hundreds of volumes, tens of thousands of pages, tens of millions of digits, of the Air Almanac. True, cynics like me pointed out that those aviators who found errors never came back to report them! But the claim was undoubtedly valid.
There had long been a tradition of perfection in the computations. Whether done by logarithms, by desk calculators, or (later) by automatic equipment or modern computers, techniques like duplication, with great care not to replicate errors - ah there, Maxwell - and every kind of differencing and consistency checking, made the numbers unexceptionable. But to turn these values into mass printings - that was the rub. Obviously they could be set in type, and the plates made photographically from the masters would be perfect, except for deterioration which careful printers would catch. All that [-63-] had been in train for many years. But how did you make sure the masters were correct? Proofreading: dull, dull, dull - and error-prone.
The most senior and valuable men and women in the Almanac Office did it. Juniors, and clerks and such, simply could not be trusted to not overlook that one incorrect digit in a sea of figures. There were attempts to cross-check; sheets drawn from the master plates were differenced by hand, or at Greenwich on Comrie's bookkeeping monsters. And that gave Eckert his clue.
Sheets pulled from the production plates, the actual plates from which the tables would be printed, were given to Jack's keypuncher. Without having the faintest idea what the numbers meant - and that was good, not bad - she punched them onto IBM cards. When Jack had done his job (we haven't gotten to that yet) the sheets that were photographed had been made from another set of punched cards. The set that young Rubye made was compared with the older set, in a conventional IBM 513 reproducer. Ecco! Completely mechanized proofreading, and by inexpensive and draft-proof help!
In the early stages of the installation, that set of cards from which the plates were made was punched up from the handwritten numbers sent over from the main office. As production neared, however, much of the subtabulation of the marine and astronomical data was being done on the IBM equipment - and all of the work would be transferred when the multiplying punch was checked out and in daily use. Over many years the interface with astronomy would move upstream, so that less and less data (and more and more program material) would be key-entered.
This triumph over human proofreading drudgery and fallibility was complete when I arrived in May [1941], but Jack was in the very middle of his struggle. Eckert had charged him with producing the master sheets that were to be photographed, on the IBM tabulator. Sounds easy: wire the plugboard; feed the cards in; tear off the pages neatly. Check once in a while to make sure the type is clean. But there were a million little glitches.
He and Eckert, and the printers, had had to choose very special paper and work out the forms (very light blue ink), and then persuade some overworked Washington forms manufacturer to produce them, and with better-than-normal precision. It may well have been IBM itself, which had a card plant in the vicinity - I never knew. Then the forms had to track nicely through the 405, which was new but not perfect, The single-use ribbon, also carefully chosen, had to be monitored almost continually; a slip on the last line meant an hour's work to be done over. Experiment had already shown it was not practical to splice or patch or correct the sheets.
But there was worse, much worse. To solve major problems of formatting and presentation, and in view of the photographic reduction that was needed to make pretty pages, a narrower type style was absolutely necessary. And even for Eckert, even for the world's first whatsit, and even for Our Brave Aviators, IBM could not rebuild the typebars and the printer unit on its 405. It would have cost the earth, and there wasn't time. What they did do was accept an order from Eckert for a special printing device to be delivered a year or two up the line - and of course Pearl Harbor intervened - and propose a crazy interim solution, which Jack was trying to make work when I arrived.
IBM put narrow-face slugs in the standard-width type bars, which went [-64-] up and down 150 times a minute for numerical output. They provided a one-of-a-kind platen adjustment which moved the platen and the paper half a typebar width laterally by manual intervention. Jack and the customer engineer and the less-nubile young lady ran the input deck of cards into the feed and printed every other column. Then they collectively held their breaths, and the customer engineer rolled the paper back [!!!] and moved the platen over. A switch changed the columns being read from the cards, the deck was run again, and the missing columns filled in. Today we would say, "You've GOT to be kidding!"
The whole unnatural process revolved around the customer engineer, who had to be ultra-precise and superhumanly patient. IBM had supplied one, as promised. He was Richard Bennett, a quiet but attractive man in his late twenties. To the chagrin of Jack's young ladies, he was married, but this did not keep the draft off his neck; after Pearl Harbor and its follow-ons, and after the arrival of Eckert's table printer, I suppose (I was off into optical adventures by then), they took him away.
IBM had a policy, mistaken in my opinion, of never requesting deferment for its customer engineers, its manufacturing people, or its engineering staff. In some cases a military agency might do so directly - I've always assumed that was what happened in my own case in 1945. More frequently the dragnet, warned of the great value of a trained maintenance man, what with the hundreds of important punched card installations everywhere in the American military environment, captured guys like Bennett as they entered the system.
IBM - or Watson himself, more likely - thought poorly of these youngsters doing the same job they had been doing in civilian life, at a fraction of the pay and without the comforts. Also, like Dick, many of them had families: in The Old Man's view, these, and the families of IBM men in combat, were valued parts of the extended IBM family. So the company, uniquely in the country I think, made substantial supplementary payments to ease the pain. And, as you will see later, it made careful provision to re- absorb the men after the war.
About that table printer: unlike much of the Belzer/Bennett effort (writ on water, I said), there are two or three articles about it in the remote and scanty literature of the time. Eckert was rightly quite proud of it, and told me years later that several copies had been made for other almanac offices and similar institutions. It was a black proportional spacing electric typewriter, which IBM had just begun to make in the Forties, in which most of the alphabetic characters had been replaced by extra numerical fonts: bold face, or small size, or small-size subscripts and superscripts; all designed to be harmonious.
Automatic spacing was crippled, and put instead under external control. This was embodied in a (black) key punch, which read the stack of data cards, rather slowly of course. On such punches there was a rack for a master card, which rode back and forth in synchronism with the detail card. IBM had modified the circuits so that the holes punched in the master card now controlled the spacing of the typewriter. Belzer's shop also supplied other material and controls on the detail (data) cards.
The results were extremely elegant, and the Air Almanacs and marine [-65-] versions printed with the device for decades were also much more readable. It was my knowledge of this antique but extremely successful accomplishment, along with a hundred adjurations from my hero Comrie, that made me such a grouch about character reading standards (MICR and OCR) in later years.
I told the stories earlier about how I left the Naval Observatory in 1942 for war work, and how Eckert was called away in 1945 by IBM to start the Watson Lab. And I have mentioned that Maxwell served there unhappily for a year or two. The Observatory was a seedbed for computational innovation in a way that Comrie could not duplicate. This was partly due to the different kinds of war pressure; Greenwich was seriously bombed, while life went on very quietly indeed at the U.S. Naval Observatory. It was also due to the fruitful relationship between Wallace Eckert and The Great Watson.
If I may drop back into autobiography for a few lines, I can fill in about a very warm and productive marriage. Dorothy was delighted with my telegram from the Navy (even at that crucial moment, by the way, I wrote rather than telephoned - ah, 1941!). She immediately began planning to come East and join me, and we agreed to marry in Washington, and without family; both sides were very short of money, the seniors as well as the Happy Couple. There were showers and feminine festivities in Pasadena, and Dorothy also had to help her bosses find a replacement.
It was more a mark of the times than of my, um, overbearing nature that we assumed she would help me with my astronomical career rather than attempt to keep her own going. She had had one Mount Wilson publication, but she regarded herself as an assistant to the Hubbles and Humasons and Dunhams and Nicholsons, rather than as a junior aspirant. The war made a difference, of course, but the attitude remained.
My parents were a little dubious about my leaving Michigan, but in their usual supportive fashion promised me the family Plymouth - then pushing 75,000 miles but in excellent shape - as our wedding present. I resigned my last fellowship (only two months remained), shook hand with Maxwell fondly - we both knew it would never be the same again, but even the most ardent disciple has to go on the road in the end - and moved six years of belongings back to Royal Oak. Sorting furiously, I packed all the technical stuff and most of the personal - bye bye, bicycle - into the Plymouth and took off for Washington. There were no interstates or autoroutes; it was a long drive.
I found a temporary basement studio apartment near the Swiss Embassy, and on nice days could walk to the Observatory. I had been unable to have a car in Ann Arbor; I could not afford one, of course, but also university regulations prohibited - an excellent leveller, but difficult on today's overextended campuses. By the time Dorothy arrived - to stay chastely at the University Women's Club; it was a different time! - I had found a new but very small unfurnished apartment (with a suitable closet, it later turned out, for our foldboat - but no garage).
We shopped frantically for furniture, had dinner one night with our nice witnesses, and tied the knot in minimum Lutheran style. We could not afford a honeymoon. However, just as I always managed to buy new textbooks in my [-66-] darkest Ann Arbor poverty, so I managed a platinum wedding ring for Dorothy!
Things went very well indeed, although the fine frenzy and the steamed-up windows of the Altadena orange groves did not recur. Indeed, one of our more promising sexual experiments one Sunday afternoon was interrupted by the radio announcing Pearl Harbor, and Dorothy worried for several years whether a wrathful Swedish Lutheran deity had not disapproved of her waywardness!
She indeed learned to run the Millionaire, with and without motor. She was welcome at the Observatory until Pearl Harbor, after which the inimitable Hellweg made it off limits. She typed my thesis, on a rented - yes, black! - typewriter. She waved me off on my first overnight train ride, back to Ann Arbor for my orals, and insisted I have a berth to be fresh for the ordeal.
She came proudly with me to Commencement, and was the first one after Maxwell to call me "Dr. Grosch" - and kissed me besides! The war consumed us; she had only gotten time off to come with me through the intervention of her new boss Dr. Meggers. The uranium was throbbing in the spectrograph, so to speak. But with all the troubles of the time, we were beginning a good marriage.
In Chapter 08 you will encounter
(in order of appearance):
Columbia University 03
Wallace Eckert 01
The Men's Faculty Club in 1945 there was a Women's next door
Selig Hecht he did vision theory; both a scientist and a humanist
Harold Urey already a Nobel prizewinner in chemistry
Maurice Ewing an oceanographer with terrific disciples
Albert Einstein only once, across the dining room
Jan Schilt head of the Astronomy Department; he made me an associate
Rex Seeber 01
Howard Aiken 01
The Harvard Bessel functions Aiken and the Navy missed the whole point
John von Neumann 03
Hilleth Thomas 01
Interior ballistics "explosions and such" rather than trajectories
I.I. Rabi 01
Lee DuBridge back at Cal Tech after running the Radiation Lab at MIT
Robert Jastrow already overbearing, even without a doctorate
The Watson Lab 01
Five floors, two basements and no elevator but we were young!
A major remodelling job from fraternity house to computing lab
Migrating WCs "juggling the johns" hurt an IBM career
John Diebold still a student at the B School
IBM special machines the first at the Watson Lab belonged to Pete Luhn
Dick Bennett 07
The Watson Lab library matching globes and other art - but also books
Mary Noble Smith in charge of the IBM art collections
Watson Senior 01
The IBM scientist portraits Eckert liked Gibbs and I accepted Franklin
The IBM ceramics collection a Syracuse prizewinner from each state
The Newton portrait Mary Noble saw my bid and raised me
THINK signs everybody got one, and often an Imperial Portrait as well
Frank Hamilton 01
Watson Lab adding machines and calculators "Buy an Underwood!" T.J. said
IBM Christmas parties the childless were relegated to the balconies
The Hall of Stuffed Birds a great place to announce an Arabic typewriter
General Groves for his luncheon, even IBM had to serve drinks
Garland Briggs he tried to warn me gently about the "No Booze" rule
The Visit Watson Senior toured a new province of his empire
The automatic measuring engine convinced T.J. about Eckert's priorities
Oil paintings vs. science photos I contradicted him and got away with it
Later Watson Labs the larger one on 115th had photographs in the lobby
[-69-] The looping around is now almost done. To draw you in to an exciting time, and to introduce you to some rather spectacular actors "early on", as the Brits say, the narrative was decidedly non-lineal. To disentangle it, re-order the chapters: 04, "Three Years Ahead Of Myself", was boyhood. Then came college, and Maxwell made a computer of me in 05, "2500 Hours On A Marchant." I spent a Harvard summer and a Pasadena Christmas, and got a wife and a job, in 06, "Mate In Two Moves." The war and the doctorate arrived in 07, "Comrie, Eckert, And A Jack Belzer." Dorothy and I did war work in 02, "In A Glass Darkly," and I was drafted to work at IBM in 03, "A Little Man From The Manhattan District."
Then there was a gap; things that happened a year or two later are described in the first chapter, "Watson In All His Glory," but the transition from von Neumann, and Los Alamos and The Bomb, to relatively peaceful late-Forties IBM, is missing. I'll tell that story now, and when this chapter is done we will jointly be on a non-tangled course - from the Watson Lab days well after the unveiling of IBM's first electronic giant, the SSEC, to the day sixty years later when I finish this book.
The crates of equipment, and that big safe, and the first Giant Brains Of Los Alamos, all came to the tenth floor of the Pupin physics building at the north end of the Columbia campus. In the whirl of getting going, I hardly noticed the relationship to the parent IBM - "590", as I soon learned to call it. Eckert was luxuriating in the familiar and welcoming embrace of his old university and department and friends, which he had missed greatly at the Naval Observatory, and he drew me in to the circle from the beginning. It wasn't necessary; he could have left me outside the door. But I was a fellow astronomer, and had all the other academic credentials, and he wanted me to help him do more at Columbia than what he had time for personally - cooperate in teaching and research, and in nurturing bright youngsters.
He wanted to spread the gospel of mechanized computation, not to make money for IBM, or even to thank The Old Man for his long and substantial support, but because he believed in it and wanted to help others to see the light. He saw me as a fellow preacher. Little did he dream that the little sect he and I were helping found would throw off a torrent of machines and languages, a worldwide glossolalia with Fifth Generation Japanese and born-again spreadsheeters rolling in the aisles and masticating the carpets!
He put me up for the Men's Faculty Club - yes, there was a Women's next [-70-] door, not nearly so well equipped. I began eating lunch nearly every day at the informally arranged round tables. There were famous scientists everywhere, and humanists as well. There were men interested in both sides like Selig Hecht, whose work in the mechanism of color vision I knew from optical references, and Harold Urey the Nobel chemist, who shared a bloody field with me at Cal Tech fourteen years later, and Maurice Ewing the oceanographer. And across the room, once, Einstein.
We astronomers ate part of the time with the physicists, and part with the mathematicians. The latter were pretty pure, and getting purer as the war wound down; Eckert and I were both more comfortable with the former, and I noticed that Jan Schilt, who had succeeded Eckert some years back as head of the astronomy department, was also more at ease with the physicists. Jan was in some ethereal sense also my boss, since I had without even asking been made an associate of his department (Eckert was reassimilated as a professor, although like me without salary).
As I have said, Wallace quickly cast out a net for Rex Seeber, and rescued him from Howard Aiken and the horrors of umpteen-decimal-place Bessel functions, which the Navy was calculating with the help - I have always assumed, the reluctant help - of Grace Hopper. My opinion of the Harvard computations was shared by Comrie, I found out rather quickly; Eckert also thought them a waste but didn't worry much about it. Rex was more at ease with the IBMers downtown and was soon swept away by Hamilton's SSEC project, so we didn't see much of him at the Faculty Club.
Before the Watson Lab left Pupin for its mid-life home on 116th Street, Eckert came to me about another prospect. He had had his name from Higher Authority - I assumed IBM, which was silly of me; I realized years later it must have been von Neumann, who had a finger in nearly every erudite pot in the war effort. This was Hilleth Thomas, a world-famous but exceedingly unworldly theoretical physicist who had been doing interior ballistics - that is, explosions and such, as distinguished from exterior, which was trajectories, and which astronomers and ENIACs worked on. He had been a user of the Aberdeen punched card computing shop, which is why I thought "IBM", but it turned out Great Johnnie had scrimmaged on that playing field also, and at Aberdeen.
Thomas, who had a doctorate from Trinity, Cambridge - the best of all the great prewar scientific colleges in England - had been teaching in the late Thirties at Ohio State. Mirabile dictu! a Welsh fish in midwestern shallows. As a member of the regional astronomy Neighbors' Club, which he often attended, I had met him in Columbus and recognized his great applied mathematical skills. I told Wallace he would be a real catch, and after checking around with our luncheon companions Eckert made him a bewildered IBMer - the company's third scientist. He fit as perfectly into the Columbia ambiance as Wallace, and was a valued member of the physics table from his first luncheon. Sartorially he fitted as poorly at 590 as I, but at least he had no hair on his face!
Eckert had a good relationship with I.I. Rabi, who was still senior man in the Columbia Physics Department, although seconded to the MIT radar laboratory as Number Two to the director, Lee DuBridge. When Rabi returned [-71-] in late 1945, he saw that Thomas was given professorial status. I believe Hilleth had had tenure at Ohio State, but since access to normal academic advancement on Morningside Heights was not a problem for those of us drawing IBM salaries, it didn't really matter. He was the first of the Watson Lab staff to acquire a doctoral candidate: Robert Jastrow, now a major warrior on the space publicity and government grant battlefields, but then only an overbearing graduate student, contemptuous of us plugboard wirers. He came too late to meet Feynman!
Wheels ground swiftly, down at 590. Watson wanted a handsome presence at Columbia for his new Pure Science Department and its Watson Scientific Computing Laboratory. While I was leaping around like a trout on a hook in Pupin, Eckert was drawn in to real estate negotiations downtown. T.J. had found out that a small fraternity had owned an interesting building adjacent to the main Columbia campus: 612 West 116th Street. The building was vacant; the young men were serving elsewhere - many of them in the Pacific. Orders came down to buy it, refurbish it for IBM occupancy, and then transfer it to Columbia ownership. I'm sure those subterranean beancounters at World Headquarters made sure proper tax advantage was taken, but I knew nothing about such matters, nor I suspect did Wallace.
I was swiftly drawn into the planning, which was complicated. The building was only twenty-five feet wide, wedged in between two large apartment buildings. Aside from one very small light well, the only windows were at the front (north) and the back. There was a small rear yard which was never used, but which guaranteed a little breeze; there was of course no air conditioning.
There were five above-grade stories, and two basements. One of these, as was common in such structures, was only slightly down from the street and had a separate direct entrance under the formal front steps, and a rather large front light well. The most serious problem for a computing shop was access; there was no elevator, nor any sensible way of installing one. In the event, heavy equipment had to be brought in through the front lobby for the first floor, over iron railings and through large French doors into the library for the second floor, and down through that awkward light well for the upper basement - or, theoretically, through alleys and the back yard, but we never had to do that while I was in charge.
The stairwell was rebuilt to fire-resistant commercial standards; there was no other vertical trafficway - no back stairs. The kitchen and dining facilities were torn out, and in the process steel I-beams were put in to support my heavy equipment (false floors, which T.J. insisted on at the SSEC, were never even considered). Heavy wiring, entirely different from what the fraternity boys had needed, was installed.
There was an amusing problem with the toilets. First the architect tried to put two toilets on each major floor, on the side lightwell: much too wasteful of space. The unisex arrangements of Europe and of small retail establishments were out of the question for staid IBM. I said, "Male in the first basement, female on One for visitors and a receptionist; male again on Two, female on Three, male on Four. There are gonna be guys on Five and in the sub-basement, but they can hoof it."
Alas for rationality, after the building was done, and as we were moving [-72-] in [1946], an alarmed minor IBM vice president noted that Dr. Eckert (on Three) would be hors concours. Additions were made: urinals and sanitary napkin dispensers, as required. All the labels were switched. But the urinals were not taken out of the new Ladies, prompting one of my more outspoken gals to say that she really appreciated the chance to examine one closely for the first time!
The vice president concerned ended up at Diebold (Safe and Lock; He-Didn't-Invent-The-Word John Diebold was still at Harvard). Indignant surviving IBM vice presidents have heard me tell this story and have denied the culprit was, ah, eliminated because he juggled the johns, but my view is that if that wasn't why they shipped him out, it ought to have been.
The living room of the fraternity house was converted into a lovely reception room, featuring a lovely receptionist who wore lovely tight skirts, lovely high heels, and ran a large and very unlovely switchboard - the kind with lots of cords. More about this room later; it was the scene of one of my greatest triumphs.
In the waist of the building was a space for "special" machines. Pete Luhn's babies lived there first, and later a modified pencil-mark-reading IBM test scoring machine. Still later [1949] I put the prototype 604 there, and it was still working away fourteen hours a day when I left. Then came a Men's (formerly Ladies), and the main machine room: full width, and about fifty feet long.
On the floor below - we usually said "basement" - there was a spare room in front, in which Richard Bennett (yes, the same wonderful customer engineer who had made the Air Almanac printable in 1941 at the Naval Observatory; we will celebrate his Watson Lab advent a little later) was to build an automatic astronomical measuring engine for Eckert. Then came a space for many, many punched card file cabinets, and a restricted number of boxes of blank cards (the main stock was in the sub-basement, along with my foldboat).
In back was the second machine room, also fifty feet long. This was the facility more frequented by our evening-working students and visitors, and the WC immediately became genuinely unisex, even before the offending VP had ceased to encumber us.
In the front of the sub-basement, which had a horrid access chute originally intended to bring in coal, we planned a small machine shop. This sounded wonderful to me, although it was not for my outfit; I had had the run of an experimental machine shop at Sperry, and enjoyed making little gadgets (remember that foldboat in the storage area). In the middle was a brand new oil burner, procured with enormous 1946 difficulty by IBM to replace the coal furnace. And, miraculously, a walk-in vault for the fraternity silver or whatever, whose door I promptly had renovated by a fancy locksmith so I could store our Los Alamos material there. When we moved in, I sent the giant Pupin safe back to the Manhattan District.
Behind all this was a general storeroom. There was no toilet, and the machinist would theoretically have to climb to the main floor. There was a sink in the oil burner room, however. I had had it installed for the janitor, or so I said.
[-73-] There had been two sets of fireplaces, which the architect reluctantly plugged. The rear one left lovely mantelpieces in the two machine rooms (east wall) and went on up to the fourth floor. By no coincidence at all there was a fireplace in my third floor office. The more elaborate set began in the measuring-engine room, flowered in the reception room and the beautiful library on the second floor, became more conservative in Wallace's office on Three, and terminated on Four.
The library was as large as the reception room below, which meant twenty-five by fifty feet. I had it done with walnut shelving all around, to the height of the mantelpiece. To put at the front, where the French doors needed to be kept free for equipment to be rigged in, I had Mary Noble find me a matched pair of globes, terrestrial and celestial; big, expensive, handsome. On top of the shelf cabinetry were items from the IBM American Ceramics collection.
"Mary Noble" was Mary Noble Smith, in charge of the IBM art collections. Her successor in the Fifties, when T.J. put more emphasis on having museums inside his empire, was called a curator, but Mary Noble had not attained that title. For one thing, she did curtains and drapes as well, and furniture, at least for 590 and places like the Endicott Homestead. She appeared even before the alterations on 116th Street were finished, and Wallace shot her over to me with a sigh of relief; I was deeply interested, and he didn't care an iota. He was concerned about books for the library, however, and first got The Old Man's permission to buy some, and then solicited lists from his buddies in the math and physics and statistics departments, which I gladly incorporated into my own.
Mary Noble was in her late fifties, walked rather slowly with a cane, was distressingly barrel-shaped. But she had the proper connections all over New York City to do her multi-faceted job, whether it was to find a huge, gorgeous Oriental for the reception area or merely an ornate gold frame for the big photograph of T.J. which naturally had to go over its fireplace. And, although perpetually harried, she was a nice person, and an eye-opener about The Old Man and the inner reaches of IBM. Mind you, IBM was only a $140 million a year company in 1945 - but it already had lots of inner reaches!
She came to me just after the move, and before there were books in the library. "Dr. Grosch," she said, "perhaps you can intercede with Dr. Eckert. Mr. Watson (she seldom said `IBM', and never `T.J.') has a wonderful collection of portraits of American scientists, and I have asked Dr. Eckert to select some for the library, and his office. There are twenty-five...". "Good grief, Mary Noble," I broke in, "how did T.J. get so many? They aren't all by the same painter, I hope?"
"Mr. Watson sponsored a competition for young portrait artists across the country," she said, "and the judges picked these as the best. Mr. Watson bought the whole collection. Remind me to tell you about our lovely ceramics pieces; he did the same thing recently at Syracuse University, and I have the forty-eight state prizewinners in our warehouse."
"Didn't Eckert like the paintings?" "He didn't want to see them. He just looked at the list and said `We'll take him', and that was that." "Who'd he choose, Mary Noble?"
"It was a Willard Gibbs. Do you know him?" "Yep; the greatest American [-74-] name in thermodynamics - and from Yale besides." She didn't know about Wallace's connections.
"Let me see your list." Short pause. "OK, we'll take Benjamin Franklin; he was quite a natural philosopher even if not a full-time research man." "But Dr. Grosch," she wailed, "how about all the others?" "They're not scientists, dear lady, and the boss obviously won't give `em house room."
She had Edison, and the Wright Brothers, and lots of other inventors. I told her to hang them on the walls of the Endicott engineering lab or in the Education Building next door to it. Mary Noble bristled a little - a very little. "Whom do you call a scientist, Doctor?" "Oh, Isaac Newton, say."
Three weeks later the art van drove up and delivered a big oil painting of the great Sir Isaac. I had them hang it over the mantelpiece in the library, the place of honor. "Mary Noble, the Newton is just great. I hung it over the fireplace. Where did you find it?"
"I had it painted for you," she said placidly. As I said, lots of inner reaches!
One of the curious customs in the IBM Empire in 1946, and for many years thereafter - it probably survives in mutated form even today - was the Portrait Kit. Not to be confused with my Newton or Wallace's Gibbs, it was a package of standard framed photographs of IBM executives which appeared mysteriously whenever a new office opened, and even when someone was promoted into physical surroundings where a current or appropriate-level package had not landed. Everybody got the standard Watson portrait, and a THINK sign; the size and framing of The Old Man's picture of course varied with your importance, or the importance of the new office. Then in proportion, there were portraits of other officers, and big shots in your own line.
Thus Hamilton had not only a large Watson Senior, an early Watson Junior, and several other Olympians, but famous engineers Bryce and Lake and - very naughty of him - a matching but non-standard one of Hollerith. Eckert had a very large Watson Senior over his mantelpiece, and a pile of THINK signs to give to visitors, but hung the Other Faces out in the hallway. The big Watson in the reception area was carefully matted and elaborately framed by Mary Noble, but nevertheless was the standard pose. Some mysterious cubbyhole at 590 dispatched these kits all over the world (the French got RÉFLÉCHISSEZ, with the proper acute accents; the Germans, DENKE).
Over my fireplace I had a handsome French post-Impressionist scene, offered by Mary Noble from her Homestead stock - that is, out of the general IBM art collection, from which paintings and sculptures and such were rotated through the various IBM country clubs and guest houses. I had elegant built-in bookshelves (Thomas had such a mass of books the architect gave up, and let him put in ceiling-high office shelving). And we all had good W.& J. Sloane furniture, but only Eckert had a rug.
Oh, by careful scheming I had a Marchant - yes, the familiar Model ACT-10M - on my side table, and made very good use of it. The two machine rooms had Fridens. Wallace didn't need a calculator, and Hilleth used a slide rule or looked sort of cross-eyed while, like von Neumann and Feynman, he did sums in his head!
I've said that T.J. decided everything. After we had been in business on 116th Street for a while, I asked the boss - I called him "Dr. E." - if I might buy a printing calculator; that is, a fancier adding machine that would do simple multiplications rather clumsily, but print on a tape. In today's world of TI and HP and Casio hand-helds you get that and a great deal more for under fifty bucks; what I wanted was a Remington Rand machine that weighed twenty pounds and cost four hundred 1947 dollars. Eckert by this time was alert to IBM customs and prejudices; he added my request to his list of things to ask Mr. [-75-] Watson. The next time he got to sit outside the throneroom, and get inside - the latter did not by any means always follow the former - he said that someone at his place wanted to buy a RemRand adding machine. The Old Man looked at him sternly and said, "Buy an Underwood!" Next item?
Usually I saw T.J. at a distance, or through the eyes of others. There were ceremonial handshakes, and receiving lines, and he would be paternal at the IBM Children's Christmas party (where Santa Claus brought Mrs. Watson the big present, and the childless were relegated to the Waldorf balcony). I remember a spectacular dinner in the Hall of Stuffed Birds at the Museum of Natural History - he was probably a director of that museum too - when he announced an Arabic typewriter, and Dorothy got the giggles at the incongruities.
After The Bomb was dropped he wanted to celebrate the IBM contribution, and incidentally to re-emphasize the recent creation of the Watson Lab. He gave a magnificent luncheon at the Waldorf, with General Groves (unfortunately) as guest of honor, and with a curious mixture of science types and military men and his personal friends. No Oppenheimer. Eckert had seen that invitations went to von Neumann and the others at Los Alamos, but only a few came - I seem to remember Bethe.
Eckert was at the head table, and I was also introduced. But what I remember is pure IBM - and because of the man involved, nice IBM. For such a gathering, with all that brass, even Watson had to have liquor. The Waldorf, for one of its most valued customers, dug `way down, and produced the best Scotch and bourbon and such - this in the depths of rationing, and when gin was made in Puerto Rico. I was much impressed; it was my very first IBM function, and I already looked forward to many more.
As I stood in the lively crowd, with a double manhattan and a handful of luscious canapes (and not wearing a sport jacket, you may be sure), an older man in IBM costume - very dark blue suit, white shirt, striped tie, conservative black wingtips - came over and introduced himself as a Garland Briggs "from 590." "I envy you people from the Watson Laboratory," he said. "Those of us at World Headquarters don't feel free to have a drink at lunch." "Must be tough," I said sympathetically, setting my empty glass on a tray and reaching for another big manhattan.
Actually I was safe. No one from 590 would dream that the bearded character in the gray suit worked for IBM. Briggs, who had a degree in astronomy [!], knew Eckert and was interested in me, and in what we had been doing for the Bomb Boys. He was trying very unobtrusively to warn me of the No Booze rule, and I realized only a few weeks later that I had been, in a very unusual way, warmly treated.
Briggs had migrated to the headmastership of a fancy boy's school, and tripped over the Depression. A friend had slid him into the financial structure [-76-] at IBM in the late Thirties, and by 1945 he was something like Assistant Treasurer.
Eckert told me his story. A few weeks after he came to 590, and long before he had recovered financially, his son came down with polio. No sparrows were permitted to fall anywhere near IBM without The Old Man hearing. He ordered that "the young man" be given the best medical care, at IBM expense, and as I said earlier, turned to his next imperial task. Briggs would have died for him.
Later in the year of the SSEC dedication [1948], the Watson secretariat at 590 found time for The Great Man to make his first visit to the Watson Lab. Eckert was informed well in advance, and sat down with me to decide what to show him. Lacking an elevator, it had to be the three "main" floors, or less. We decided on the first floor machine room, because of the unusual Aberdeen machines which he had not seen in their latest form, and one or both of Pete Luhn's specials (this was the occasion on which T.J. "ordered" Pete to separate the calculating equipment and the work station).
Wallace wanted to show him the measuring engine Dick Bennett was laboring over. I thought it ill-advised to be too astronomical, but Dr. E. was the boss. And instead of his ascending painfully to Eckert's eyrie, we decided to set him down in the beautiful library for the hardware presentation, and then trot him back to the electronics lab, which you haven't heard about yet. Then back to the waiting limousines and 590.
Mary Noble and her advance men descended on us like furies. Windows had to be washed, slip covers and curtains sent out for dry cleaning. Strange actors from 590 invaded our stage and irritated the junior staff and the Columbia janitors (I was absolutely agog; they could have rolled me up and sent me to the cleaners too, and I would not have protested). Amazingly, a flower fund was installed at the nearest good florist, and the receptionist (who had lovely hay fever as well as lovely everything else) was dragooned into stopping for masses of fresh flowers every day or two. As Mary Noble well knew, Watson might suddenly decide to go to Washington, or Timbuctoo, or even Endicott. The Visit, as we had begun to capitalize it, was iffy.
On D Day, we received minute-by-minute reports. I honestly think little people were stationed in phone booths along Broadway (Manhattan phone booths still worked in those days) to report on the cavalcade. Remember, there were no car telephones or CBs. Yet the switchboard would report, "They're at Columbus Circle." "They're at 79th." "They're at 108th." And the entourage swept in, only a little behind schedule (the secretariat had probably squeezed in the Chilean ambassador at the last moment).
Everything went well. We gathered for farewells in the reception area. Mr. Watson looked around the room, which was panelled in cream and touched off by half a dozen small oil paintings of Paris scenes I had chosen from Mary Noble's best. He looked approvingly at his portrait over the fireplace. He nodded his head. "Everything is quite nice," he said, and the heads of the company all nodded blissfully. "There is just one thing, though." The heads of the 590 people all began to shake doubtfully, and the 116th Streeters began to shake all over.
"These paintings," he said, waving his imperial arm around the room (he [-77-] was a big man). "There should be photographs of galaxies, and atoms, and scientific things. This is a laboratory, after all." Mary Noble turned white.
"Mr. Watson," I said, "our visitors see that sort of thing all the time. What really impresses them is that IBM, a world leader in technology, has collected these beautiful paintings - is a patron of the arts as well as of science." Dead silence. This miserable, bearded nonentity has contradicted Mr. Watson!!
He looked hard at me for a moment. As Pete Luhn said about his own case, if I ever tripped up in the future, The Old Man would remember my intransigence and my beard. He asked a vice president about the limousines; the whole entourage of seniors and sycophants swept out; peace descended like a blessing from on high.
The very next day Mary Noble appeared. "Doctor, about those galaxies and things...". "Forget it, Mary Noble," I said. "Those paintings stay right where they are." "But Dr. Grosch - he specifically said...". "Let's call him up and check. My recollection is that he agreed with me." "Oh God, no!"
When I left for Washington more than two years later the big room still looked lovely. Watson never visited again. But when the laboratory people outgrew the little 116th Street building and everybody moved into much larger quarters nearby, the pictures in the new lobby were not from the art collection, but "galaxies, and atoms, and scientific things." Eckert remembered!
And Mary Noble turned off the flower fund on her way back to 590.
In Chapter 09 you will encounter
(in order of appearance):
The Watson Lab 01
J.C.P. Miller seconded Comrie at the first technical computing bureau
ASTOUNDING STORIES 03
The SSEC 01
David Hill simulated nuclear fission in an SSEC water droplet
General Electric turbine people put the first paying problem on the SSEC
Hans Kraft he struggled for a knowledge of steam flows
George Kimball Columbia member of the ACS punched card committee
Pete Luhn 01
ACS [American Chemical Society] enthusiasts became clients of Pete Luhn
ASIS [American Society for Information Science] later Pete was president
Automatic abstracting Pete practiced on paper tapes from TIME
Beevers-Lipson strips X-ray diffraction calculations before computers
Maurice Ewing 08
Frank Press from 601s he went on to head the National Academy of Science
Astronomisches Rechen-Institut they kept track of over 1500 asteroids
Lillian 03
The 1947 KLEINE PLANETEN Lillian and I did it on the Aberdeens and a 405
Thomas J. Watson Astronomical Computing Bureau 03
The IBM 601 threw its answers petulantly on the floor after midnight
The IBM 602 not withdrawn, you understand, but upgraded!
The IBM 602A at least it used the same new plugwires and cute skipbar
Eric Hankam could make the 602A and even the collator sit up and beg
Paul Herget 05
A private 601 operating manual you plugged some switches both on and off
Edward Teller The Scientist We Love To Hate set Hankam a task
The Survey Computer 150 lbs. of polar-to-rectangular transformations
The IBM 603 heat a whole room with 600 full-sized vacuum tubes
The IBM 604 a salesman's dream, and I had the prototype
Freddie Uffelman "white lightning" and nuclear inventories, at Oak Ridge
Red Dunwell much later, he was head of the STRETCH program
Andrew Salter we went horse racing together on the Watson Lab 604
MTAC [Mathematical Tables and Aids to Computation] where we all hung out
Aberdeen Proving Ground 02
Forest Ray Moulton he set up Aberdeen ballistics computing during WW I
Leslie Simon he did in three weeks what would take two years today
The Bell Labs Model V yeoman duty at Aberdeen after the war
Ben Durfee and Don Piatt they rebuilt the IBM Aberdeen machines
Dahlgren the Navy equivalent of Aberdeen; later, they undertook NORC
IBM dominance the EDSACs and JOHNNIACs were not ready until later
Martin Schwarzschild Columbia astrophysicist strong on stellar interiors
[-79-] Life on 116th Street was not all panelled reception areas and the IBM ceramics collection. The Watson Lab was a nexus of attention for men and women all over the world who had dreamed for years, or had now begun to dream, about doing big technical calculations.
They read about the ASCC - the Harvard Machine, everybody called it, since Aiken's militaristic "Mark I" didn't sound right, postwar. But like a khaki-suited dog in the manger, and acting as if he had one of his migraines even when he didn't, Howard drove the supplicants from the temple. In the days before ENIAC and the SSEC came on stream, he controlled the best-publicized fraction of the world's computer resources. And he was wasting it on a ridiculous project to calculate Bessel functions.
In fact, even Bessel function enthusiasts were upset. Dorothy and I had a famous applied mathematician as house guest - J.C.P. Miller, who was helping Comrie establish the first for-pay scientific computing bureau, in London. He spent much of his leisure time plowing through every page of my 1942-47 collection of ASTOUNDING, from which he had been cut off in wartime England, and what little time was left in "collecting" New York's bridges. Well, Miller, who like Maxwell had spent untold hours bent over a hot desk calculator, grumbled first about unnecessary accuracy and then, reluctantly, admitted that "the new machines" would calculate such things as they went along, so that great tables were obsolete.
So it was no use petitioning at Harvard. A few super-priority jobs got taken on - Los Alamos stuff - via high Navy channels, but in general, no. The announcement of ENIAC raised fewer hopes, because the monster was to be shipped off to Aberdeen Proving Ground as soon as it could stand the journey, and the backlog of Army calculations waiting for it there was known to be enormous. And not nearly as silly as Aiken's Bessel functions; at least the Army [-80-] planned to kill some people outright with their computations, and not just bore them to death!
The SSEC was different. Watson had dedicated it to science - or to Science, to be precise - and there were ways to get access. You could be in the Pure Science Department; Eckert did his Moon calculations, and I did a beautiful optical job. You could be an astronomer, which meant you knew Eckert, and about the older Columbia bureau. You could appeal to the remains of wartime patriotism (and of priorities, which never did die off completely), often through John McPherson. You could come in via Columbia connections. You could be a specially-favored IBM customer, like General Electric.
Eckert and I, with help from Hilleth Thomas and the physics table at the Faculty Club, performed a calculational triage. Some applications went downtown, for Rex Seeber and his crew to fit in on the overloaded SSEC. Some ended up in my machine rooms on 116th Street, either queued up for my youngsters to run - that is, to analyze and wire plugboards and push the cards - or struggled over by the sponsors themselves, when they could get machine time (usually in the evening). And a lot were put in a quiet place to die.
One major job done on the SSEC was a simulation of nuclear fission for David Hill of Los Alamos. He used an analogy based on a drop of liquid, and the resulting equations were unclassified - in fact, his results were published. Wouldn't happen today, but we were at peace in the late Forties.
Another was an unsuccessful attempt to calculate steam flow through the blades and buckets of a giant power turbine, using fancy applied math from an academic consultant and the lovely money of the General Electric Large Steam Turbine Department. That was the first job the SSEC undertook for pay. It introduced me to a pupil of the great Theodor von Karman, an aerodynamics expert from Schenectady named Hans Kraft who became a valued friend. His calculations did not converge satisfactorily, but he did not blame the SSEC - only his consultant and himself. Would that the tens of thousands of customers of such activities who were to follow him had been so civilized!
At Columbia I serviced George Kimball, a chemistry professor who wanted to use computing instead of test tubes. He introduced me to the American Chemical Society committee on punched cards, I introduced the committee to the strange relay machines and European ingenuity of Pete Luhn, and Pete ended up many years later president of ASIS, the American Society for Information Science. ASIS is sort of the computers-for-library-science outfit, and strong on information retrieval.
Because I would almost certainly forget to mention it later on, I'll say here that Pete was also the father of automatic abstracting and indexing, common currency today on PCs and word processors. He had the idea of counting word occurrences in text - he used paper tapes from TIME - discarding a standard list of linking words (and, the, to) and no-content words (I, ouch, today), and ranking the rest by frequency. He assigned weights, went back and counted the sum of the weights of the words in sentences, ranked the sentences, picked out a few of the weightiest, printed them out in their original order and with the garbage words put back in - presto: an abstract. Not bad for a German relay circuit designer, over thirty years ago!
[-81-] Kimball also sent me over some X-ray diffraction experts, who needed a lot of low-precision sums of products. For half a century they had been using Beevers-Lipson strips, sort of cardboard slide rules. Punched card machinery looked good to them. I helped a formidable woman do hemoglobin on our 405, and hoped the Watson Lab was helping medical research too.
Another Columbia figure was Maurice Ewing, an oceanographer who had done important war research on propagation of sound between the thermal layers of the ocean. He mostly sent over graduate students, one of whom, Frank Press, was much later the president of the National Academy of Sciences. Another became a major figure in radio astronomy, which had not yet begun to surge in 1949. And I remember several attractive women.
Early on, before the SSEC was commissioned, I undertook my last serious astronomical project. There had been a famous facility in Germany, the Astronomisches Rechen-Institut, which calculated (by hand, and with enormous diligence) the annual appearances of the 1500 or so named asteroids. Most of them didn't excite anybody very much - my Adonis was an exception, and there were other unusual ones, and the larger objects of course - but you had to eliminate them when you were finding new ones. The task kept a dozen subprofessionals busy around the year; the annual appearance of KLEINE PLANETEN memorialized their efforts.
Well, their institute was rubble in 1946, and the people scattered or dead - and no money to pay them with if they reappeared. The KP was published two years in advance, so 1945 and 1946 had come out before the crunch. There were plans to do the 1948 in Leningrad (the Iron Curtain had not yet descended, and we had occasional Russian visitors at the Watson Lab). The question was, how to produce the 1947 volume, in just a few months. It was an ideal task for Comrie/Eckert punched card techniques; I said we'd try.
We made it with a few weeks to spare. I did the planning, and printed up the final results on hectograph masters (shades of Allan Maxwell's computing forms!). Had to put in some decimal points, and degree symbols and such, by hand; Eckert and I had talked about using the gorgeous table printing typewriter at the Naval Observatory, and sighed over how handsome the result would have been; alas, there wasn't time.
The calculations would have taken many weeks on the regular machines, but we had just received our two Aberdeen machines - correct IBM nomenclature: Pluggable Sequence Relay Calculators. And equally important, I was in process of absorbing the remains of the Thomas J. Watson Astronomical Computing Bureau. And while the machines were pretty standard, I was delighted to find Lillian Hausman in the first crate!
She not only was a superb operator - after all, she was the senior full-time scientific punched card expert in the whole world - but she knew a lot of astronomy. In fact, she probably could have done my part of the job too. Anyhow, she wired the horrific Aberdeen boards, helped run the big decks at what seemed like miraculous speed (6,000 multi-operation cards an hour, or say 0.000005 megaflops).
Another nice benefit was that Lillian was so eager to learn the very difficult new machine, and my supervisor Marjorie Severy was so genuinely admiring of her skills, and we were all so crazy busy with the asteroids, and [-82-] the Columbia night workers, and helping train the youngsters like John Backus and Ted Codd for their coming SSEC duties, that there were no jealousies and no frictions, and we all ended up on the same side. Everybody laughed their heads off at my adventures with Mary Noble and the art collections, and that helped a lot. It was a good, good time.
About equipment: after merging in the better machines from the Pupin attic (I kept the old 285 horizontal tabulator as long as Marjorie had room for it, out of sheer wonder at its clumsiness), we had two machine rooms with sorters and reproducers and collators, an interpreter, a gang punch, and several key punches. The most expensive machine (renting for over $1000 a month with all the bells and whistles I could hang on it) was the huge 405 tabulator, with 80 characters of alphanumeric storage, eating and disgorging 150 80-character cards a minute, printing 80 characters wide a line at a time (the type bars were too long to do this at full speed, but other models could print numbers only, at 150 lines a minute or 200 digits a second; the earth shook!).
The key machines, however, were the multiplying punches, of which I once, for a few weeks, had fourteen. Most were Type 601s, modified to take account of the algebraic signs of the three operands. The typical operation was A times B, with C added or subtracted from the product; it proceeded at the spectacular rate of 600 operations an hour. Usually the operands were read from a card and the answer punched on the same card, but you could get real wizard and hold a factor over from one card to the next, or make it a constant for a while under control of master punches in the deck. Fancy, huh?
The punching mechanism on the 601 was derived from the key punches, so you had to put funny metal or formica strips (you could file the formica ones yourself) called skip bars, on the carriage, to suppress unwanted punch cycles. Every machine room had racks of these stupid things, usually near the key punches, and you planned your output so as not to have to cut a fresh skip bar!
The 601s broke down pretty regularly, often signaling their malaise by throwing their cards petulantly on the floor. It was not uncommon for late workers - my wife Dorothy and I, for instance - to quit only after the last multiplier was busted. You left notes for the IBM customer engineer, who in our case came very early six days a week and fixed as many machines as he could before the troops arrived. (The notes helped him decide which machines would respond most readily.)
Need I say all these machines were black? The day of The Grim Gray Giant, a phrase I coined when the 700-series machines were gray, or of Big Blue, today's equivalent, had not arrived. All the major machines had little locked trapdoors, behind which lived curious stylized blueprints of that machine's current features - the bells and whistles I referred to. The customer engineer kept these prints up to date as he added selectors and emitters and such; when a new engineer appeared he opened the trapdoor and read his patient's chart. All the keys were identical, and the rites of passage for a new supervisor included capturing one of the keys, so as to be able to gaze blankly at the secret documents of "his" machines. Marjorie depended on feminine [-83-] appeal, and someone always produced the prints when she needed them. Lillian had a key.
As the IBM inventors and designers got into the post-war swing, improvements began to emerge. I traded four of my least reliable 601s for a new 602, which had a larger and more rational plugboard, requiring new kinds of plugwires (sigh!) but with a "flexible" skip bar (ah!) you stuck little teeth into. It was more microprogrammable than the 601s, which was admittedly faint praise. And it was a little faster. And the control punches for algebraic signs could be in any of the eighty card columns.
But the 602 was a failure. Our copy worked fine, but of course we had really great maintenance. Most of the machines out in the field seemed to be tender, and didn't stand up nearly as well as planned (the 601s cost the earth to maintain, and the 602s were intended to be much cheaper). Now, T.J. didn't permit IBM to put out fallible machines; there was a whole string of poor jokes built around the idea that customer engineers were supposed to work on the customer! So the 602 was not, ah, withdrawn; it was upgraded to a 602A. True, to anyone but a very loyal IBMer the 602A looked like an entirely different machine, from plugboard to covers. Well, it used those same new plugwires, hmm? Anyhow, I reluctantly traded in my 602 for a 602A; the youngsters learned its capabilities, approved heartily, and in the end asked for more.
About documentation: the poor little wartime manuals were replaced for even the simplest machines by handsomer, well-illustrated, glossy-paper successors. The contents were dull: routine operating dope with a few very simple plugboard diagrams, which most supervisors and operators followed blindly. To tell the truth, so did I when it came to one mysterious machine called the collator. Lillian claimed to understand it, and one of our brightest young men, Eric Hankam, could make it (and the 602, and the 602A) sit up and beg. I suspect they opened those trapdoors and studied, when nobody was around!
One of the rival orbit computing shops back in my days with Maxwell had been at Cincinnati Observatory, and it was run by a Paul Herget. Paul had come out of the Bottoms, which is Cincinnati talk for the poorest part of town, and he dreamed of some day being director of the observatory. The director in the Thirties was Everett Yowell Sr., whose son knelt down in front of the 285 with Lillian and me when I was learning wiring in 1945.
Paul went to the Naval Observatory to help with WW II, arriving after I left but long before Eckert was called away to New York by Watson. He was a great success - probably why Maxwell wanted to go, a little later - and became a leading expert on IBM machines like the 601. I mention it here because in early 1946 he published, at his own expense (those were the days!), a most peculiar private operating manual for that machine. He had discovered that many of the plugboard holes, and the functions they controlled, had extra capabilities. There were things called "bottle plugs" - think of them as pawn-move plugwires, connecting adjacent holes - which were described as switches in the IBM documentation. Paul told us to wire some of those switches both on and off, simultaneously, and what the results would be.
For about three years possession of one of these coverless manuals was the secret mark of a scientific (IBM) computer - again, computers were still [-84-] people in those halcyon days. They were photostated extensively up and down Watson Country, and even overseas - no Export Control Act! When the 602As emerged, the people who had Herget manuals were the first to get the new machines, so the flowering was brief. I had the same experience later with my special kind of punched card mathematical tables, which blossomed nicely but was killed off by cheap storage.
My shop had an advantage for the IBM engineering groups. It was full of bright users, some of them IBM employees and some, outsiders like Ewing's students and Dr. Hemoglobin. One-off machines and prototypes could be tried out "at Columbia" without the hazards of public announcement and 602-type failure. Luhn was our first patron, partly because of The Scientist We Love To Hate, Edward Teller.
The Los Alamos connection wound down rapidly in 1946. I was a charter member of the Federation of Atomic Scientists, which started its third chapter at Columbia, and worked at getting the Manhattan District to take away our classified material; the people we had dealt with were scattering; Eckert wanted to get back to astronomy. Urey came back to Columbia for a while, as did Rabi from the radar labs at MIT. Johnnie von Neumann dived into computers. Feynman disappeared for a while, and emerged with a Nobel, as did a terrific woman named Maria Mayer.
Maria and Joe Mayer, and I think Roy Marshak, brought Teller around to the Watson Lab in late 1945. He had some partial integro-differential equations to solve, and money to solve them with. I never knew how IBM handled my initial calculations; the Old Man might well have done them as an act of patriotism, but somehow I think the beancounters probably sent General Groves a bill. Same for Teller. I didn't yet know about the Super, as the hydrogen bomb was then called at Los Alamos, so I didn't try to avoid the job.
Problem was, the sequences of calculation were much lengthier than the ones we had been doing. Not more difficult - Teller's people had punched card machinery in mind when they did the numerical analysis. But it looked like hundreds of plugboards per cycle, instead of two or three dozen. I reported this to John McPherson.
A few days later an elegantly dressed engineer named H.P. Luhn came down from Endicott (T.J. paid his inventors - in fact, everybody but me - very well). He told Eric Hankam and me about his experimental relay calculators. For the smaller one he had modified a key punch [!!] by adding several movable sets of reading brushes - the holes in IBM cards were read by little wire brushes that contacted a brass roller when holes came along, and the timing of the electrical pulses told the machine what the numbers or letters were. Luhn had his machine punch out initial results to save storage; that result was then read downstream by one of the brush sets and used as one operand in the next addition or multiplication, and so on until you ran off the end of the card. Eric loved it (he was a three-dimensional chess enthusiast).
The arithmetic was done in a box on the back containing almost a thousand Lake relays, which Frank Hamilton was using in vast quantities in the SSEC as intermediate storage. There were the inevitable plugboards, but I seem to remember some of the sequences were hard-wired after Eric's analysis. Oh, when a card was full you reproduced the good stuff onto the [-85-] beginning of a fresh card, so in a sense you could think of a hundred-card string as a single 6,000- or 7,000-column card. As in the earlier work on the Trinity burst, there was to be a card for each space point, and a deck for each time step.
Eric ran the thing himself. He and I both took the transformation of Teller's equation into number crunching as a given. I didn't want to know, Eric really enjoyed the machine more than the mathematics, and we both knew we could call on Hilleth Thomas for help if the calculations ran away. If they did, nobody told us.
Pete had shipped us two machines; the other was a modified 405 tabulator with a separate, much larger box of relays. One of the machines was named Virginia - I think the larger, which was the one T.J. ordered him to split up. The other was called Nancy - also feminine. The smaller one I put in the space next to the disputed first floor WC, so Pete could show it off; the other had to go in the main machine room, much to everyone's annoyance. But one of Pete's young engineers kept it occupied, and gave us a lot of informal intelligence about progress on the SSEC on the side.
Pete also made several copies of a Survey Computer, which was supposed to go out in the field (say, 150 pounds - and it needed 110 volts). It was an electric typewriter hooked up to the usual Luhn box of relays, and it did polar to rectangular coordinates. You typed in X and Y, and a series of distance-azimuth pairs that you had measured with your surveying instruments, and the machine typed out a series of new Xs and Ys. Shades of today's handheld calculators, which will do the same including printing, and three times faster, for less than a hundred dollars!
I produced an extremely compact table of sines and cosines, which were hard-wired into stepper switches in the relay box. Reminded me of the little boresighting gadget I had done aeons ago at Navy Ordnance (ah, six years ago, that is).
After Virginia And Friend disappeared a modified mark sensing machine lived near the WC for a while. IBM made a machine that read soft pencil marks (electrically, not optically), and also sold "special" pencils at a huge markup. The machine ate paper sheets with multiple-choice tests on them, and scored the results on a meter. Analog, by gosh! Anyhow, some engineer wanted to convert the meter reading to digits and punch it on a card. Bob Walker was contemptuous - oh, I haven't gotten to him yet!
The final occupant of that space was the prototype, good old Serial Number Zero, of the enormously successful IBM 604. When Mr. Watson decreed the SSEC, he also set wheels in motion "to use these electronic capabilities in the IBM." The circuits had been patented before the war, mainly by one Halsey Dickinson, whose photo showed up in the SSEC brochure.
At about the same time as the SSEC dedication, IBM let the sales force have a few - twenty, maybe? - Type 603 Electronic Multiplying Punches. These were one-card-feed gang punches that ate a hundred cards a minute, read two six-digit factors off each card, and punched the product on the same [-86-] card. That's all - you couldn't even control rounding - but it ran ten times as fast as a 601. Convair San Diego swore by theirs.
The box of electronics was connected to the punch by a cable over an inch in diameter, and contained about 600 full-sized vacuum tubes - the same as the ones used so lavishly in the SSEC. I sat it up on a window-seat in the downstairs machine room, and we opened the window behind it whenever we used it - great space heater! We soon had better stuff, so it didn't stay too long.
Now, the 604 was a different matter. It was post-war circuitry and miniature tubes, and designed for production. I got mine in early 1949, and the first production version went to Freddie Uffelman at Oak Ridge, and was the Big Event of ACM's annual meeting down there later that year. That was the second ACM meeting outside New York; there had been a major conference at Aberdeen the year before, with Johnnie Von presiding over one session, at which John Mauchly describing the metal tapes of his next machine (which turned out to be the UNIVAC I).
Freddie got Number One because there were still priorities; he used it to keep track of the raw materials and finished products at Oak Ridge, which was still a restricted city. Lots of mason jar white lightning, though!
He was really doing a very simple inventory/accounting job, but with extra-sensitive goods. He wasn't an academic, and his crew were hill girls he had raised from key punchers. But he was sharp. Too sharp for IBM; he complained profanely all the way up to 590 that the #$%&* 604 wouldn't run right. "Fix it," he screamed," or I'll have my boss telephone Old Man Watson."
I went over to see him from the ACM meeting, with Red Dunwell, later to be the head of the STRETCH imbroglio. Red was conciliatory. I christened Master Uffelman "Foul-mouthed Freddie," which tickled the dear boy no end - and his harem, too. In the end I persuaded Red to persuade IBM to replace every single one of those 1500 miniature tubes; they'd gotten cooked somehow, and FMF was quite right. I was awarded the Mason Jar Accolade; not by preference, but all of the hill girls were temporarily taken.
About Uffelman: like many of the best punched card whizzes of the day, he was self-taught; had no degree of any sort. Yet he was married to the daughter of a fancy Berkeley professor (well, actually only a dean). He had had most of his gut removed some years back, and was all crouched over and miserable; no wonder he was foul. He begged the doctors to help him, and they said, "Frederick, this is a Big Deal; when you don't care any more whether you die on the table or not, come back."
Shortly after his 604 got cracking, he went back and said, "I'd rather be dead!" So they took out the rest of his innards, and it worked fine: last time I visited he was playing softball, and his harem was noticeably less horny. His wife Jane stood by him beautifully through all this, left him when he was out of trouble, and ended up a valued member of the Rand Corporation staff in Santa Monica.
The 604 came out of Poughkeepsie, which by 1949 had been transformed from a typewriter plant into an impressive electronic enterprise. This was Ralph Palmer country, and he had great people from MIT and other centers of excellence working for him. The machines came streaming down the floor, [-87-] and customers - not all scientific, either - were lined up outside begging for more.
The punch unit was much more sophisticated than the one on the 603, and the electronics box was five feet high and quite handsome, at least to someone who had had a 285, or racks of skip bars! On one end of the electronic unit was a one-panel plugboard, and it used ordinary plugwires - none of that ENIAC coax. The thing that fascinated the early users was that the machine was what we would now call microprogrammable; you could put together counters of various sizes, perform a completely flexible string of commands pipeline style, carry data over from card to card - and you could divide!
In those days one of the commonest problem areas was matrix arithmetic: sets of simultaneous linear equations, and so on. I had had to do several sets of sixth order, to pretty high accuracy, in my thesis work. On the Marchant that took most of a day, and the time went up as the cube of the order - that is, a set of twelve equations would have taken eight times as long; say a full week. At the Watson Lab this was a common exercise, involving a fairly complex pattern of eight plugboards and careful sorting. On the 601 the time dropped by a factor of about five compared to a desk calculator, which meant that if we had been paying rent for the machines and the building, and so on, hand work would have been about as cheap (Comrie said the same, in London).
The 602 and 602A helped, but not by much. But the 604 ate the problem for breakfast, or so it seemed at the time; cost per hour, if the shop had been run for pay, would have been about the same, and the speed was at least six or seven times the 602A, and nearly ten times our vanishing 601s.
That brings up horse racing - yes, really! I had met a strange little man named Andrew Salter, who was a psychologist or thereabouts, and who worked over an impressive range of Broadway and Hollywood types in his Park Avenue office - Judy Garland, for one. He made the cover of LOOK the year I met him: something about hypnosis. He offered to operate on my wife Dorothy, which sounded good to me but emphatically not, to her.
Andy had a wall of bound Racing Forms and other Triangle publications outside his office. This was decades before Dick Francis, and I didn't know much about racing, or (more important) about betting. But the prospect of supplementing my IBM pay - Andy said we ought to be madly rich in no time - drew me in. I proposed to work out a predictor formula, linear in twenty variables like jockey weight and track muddiness, which would give the most probable speed of the horses in tomorrow's race.
Parenthetically, this is about what most stock market prediction programs do today, on microcomputers. They make it sound much jazzier, of course.
In betting (and I suspect in the stock market play as well), you then have to apportion your money in accord with how others are playing; it does little good to bet only on the winner, and especially if the odds are poor. I had ideas about how to do it, but never got that far. I fit my formula, by least squares, to the horses in a set of a hundred past races which Andy and I carefully picked. The 604 gave me the results in a few hours, versus weeks on a 601.
Playing the predictor back against the original hundred races, I found we did only five percent better than random - and the track and the state were [-88-] taking 17 percent. And of course, results were bound to be less favorable on past races not in our package, and still less on future cases. So I shut down.
It was literally years before I saw that the whole project was unethical - that I had absolutely no right to use the Watson Lab equipment on such stuff. The climate was incredibly different from today. Everything was grist that came to our mill. Indeed, I got a tiny publication out of the experiment, since the timings for 604 procedures on systems of that high an order were news. It wasn't even like, oh, using office supplies to write a letter; you were Doing Something New, something challenging, something good for the art. Today we would call it hacking, and I would be horrified at a senior professional dreaming of such a thing. Then, even Dorothy, who was a distressingly moral person, never thought it questionable.
Before I get to the Aberdeen machines, I need to write a little more about what professors call "the literature." There wasn't any - at least for computers. Stuff about astronomical calculations went in astronomy journals; about optics, in JOSA, the Journal of the Optical Society (which is where I published Grosch's Law, mirabile dictu!); about punched card techniques, in IBM's Pointers. I must mention the exception, MTAC (Mathematical Tables and other Aids to Computation), published quarterly at Brown University with National Research Council money by a Professor Archibald.
This was where Comrie, and Jeff Miller the bridge collector, and their American counterparts hung out, and it was where the very earliest descriptions of gadgets too special to get into the stodgy electrical engineering journals had to be published. Wallace Eckert's card-operated typewriter at the Nautical Almanac Office is there, and the big Bell Labs relay machine - and my Aberdeens. It still appears, under a smoother title: The Mathematics of Computation, or some such. Its early years are a cornucopia of antediluvian hardware history. Some software, too, in a way - Lillian and I put out lists of the punched card math tables in the Watson Lab files, for instance, and people from Aarhus and Aberdeen and Acme Electric wrote for copies. In today's personal computing world, they call it "freeware"!
There began to be specialized technical publications in 1949 and 1950, and DATAMATION started in 1957 under an earlier name; now there are literally thousands of magazines in several dozen languages - human, mostly, although I seem to remember some in APL and LOGO - to satiate the old-timers and bewilder the newcomers. They range from the absurdly erudite to kiddie stuff, and all are eager for material. In the Thirties and Forties computer articles were turned away from engineering and scientific journals unless very well sponsored, and even MTAC was careful: snooty funding, and an Ivy League location.
That having been said, I refer you to MTAC and an article on the IBM Pluggable Sequence Relay Calculator if you want to read more about a really important but forgotten machine. Let me start with the word "Aberdeen"; the Aberdeen Proving Grounds, in Maryland, was where the Army tested its guns and bombs and tanks. From World War One there had been a Ballistic Research Laboratory there, where later Hilleth Thomas and Johnnie von Neumann did interior ballistics (explosions) and Leland Cunningham - yes, [-89-] the one that helped me learn to eat with chopsticks in Boston in 1940 - ran a large punched card shop doing exterior ballistics (trajectories).
Why Cunningham? Well, in WW I a famous celestial mechanicker named Forest Ray Moulton set them up with a hand computing facility, and they remembered. When the Army got into anti-aircraft fire and bomb trajectories in WW II Cunningham ran out of capacity, even with IBM's best 601-type help. The key officer there - higher ones got most of the credit for a while - was a Major Leslie Simon. He let three contracts; in those urgent times, it probably took him two or three weeks. Today with competitive bidding, and EDS and Computer Sciences, and the Big Business press watching, it would take two or three years.
One contract was with the Moore School of Electrical Engineering at the University of Pennsylvania, and resulted in Pres Eckert and John Mauchly and the ENIAC (note that the machine didn't get down to doing ballistic calculations at Aberdeen until late 1947, by which time Cunningham had gone back to orbit computing and Simon was a colonel).
A second was with the Bell Labs, and resulted in a magnificent machine made out of rather slow telephone relays, with error detection and punched paper tape input making it reliable enough to run all night unattended; it would switch to the next problem if it got stuck, or shut down when it ran out. It had pieces of Fry and Shannon in it, but was mostly due to two men named Sam Williams and George Stibitz. And the self-checking features were elevated later to a major part of the computer discipline by Dick Hamming, who came to the Labs from Los Alamos.
Well, the Model V - there had been earlier Bell machines, for smaller tasks - was slow, what with all that checking, and it didn't get to the Proving Grounds until 1946 or so. Mind you, it did yeoman work for years afterwards, and so did ENIAC. But neither, ah, saw duty in WW II. The IBM machines did.
The third contract was with IBM Endicott, to develop two high-speed relay calculators with plugboard sequencing. Two, to run the same problem in parallel (think of them as Maxwell and Grosch); high speed, because IBM had developed the Lake relays, four times as fast as the telephone units, and was building large numbers of them into supersecret cryptographic gear (MAGIC was at least as secret as Los Alamos). The key men for the Aberdeen project were Ben Durfee and Don Piatt, both of whom later contributed to the SSEC - relay storage, and much else. The two machines were finished and shipped to Aberdeen and put to work on actual ballistics before the end of the war; I assume Watson had ordered that it be so!
At the end of the war they were both sent back to Endicott and rebuilt to more advanced system requirements furnished by Cunningham (who had earlier specified the sample calculation with which ENIAC was finally benchmarked). At the same time three more were built to the new specs; one was sold or rented to Dahlgren, the Navy's proving ground in Virginia, and the last two came to the Watson Lab. And by mid-1946; IBM didn't fool around.
Eckert and Lillian and Marjorie and I were delighted. They were to be our powerhouse for internal use for many years; until the lab moved to newer quarters, I believe. Really big jobs went to the SSEC downtown, and our teaching was done on the standard machines, since IBM would build no more [-90-] Aberdeens. But for my minor planets, and Thomas' atomic physics, and the work of our cleverer research visitors, they were the key equipment. I still treasure my dog-eared Ozalid operating manual.
Picture a looming black monster, filling a six-foot cube when closed. The front part was a much-modified 513 reproducer - two hundred-a-minute card feeds and two stackers. There were two double-panel plugboards dripping with wires, on the front, and a panel of numerical switches on the right end. Hinged to the back of the punch was a dour box of thousands of Lake relays, and hinged to that a second box. The covers were usually off for diagnosis, although the un-air-conditioned and otherwise corrupted Morningside Heights atmosphere was full of dirt. There was a special oscilloscope cart nearby, to check timings and look for contact bounce. And lots and lots of thick cables.
Cards were read and punched at either feed under plugboard control: 16,000 decimal digits a minute at full bore. Calculating speed was dependent on the mix of operations: division and square root - yes, square root! - took almost a whole card cycle, while two or three six-digit multiplies or half a dozen faster operations could be fitted into the same. Besides, the Aberdeens could buzz away for several cycles without feeding or punching at all, but the plugboards got awfully complicated. On balance, each machine problem-in-problem-out ran six or seven operations a second.
Mind you, the wonderful 604, when wired by an expert like Lillian or Eric, ran almost that fast. But we had the two Aberdeens for three solid years before I got the prototype 604, and we ran each of them a hundred hours a week. Powerhouse, indeed! The experimental machines that fill the history books didn't even get fired up until the end of the decade, and had very little input or output capacity. Until 1949, it was places like Aberdeen and the Watson Lab - including the SSEC - that were crunching the numbers; the Los Alamos and tin airplane boys didn't have big machines, and the computers being built in Britain and the U.S. weren't ready. From 1944, when Aiken got the ASCC, to the present day, the majority of scientific and engineering computations worldwide have been done on IBM machinery. Commercial data processing too, of course, but that perspective is universally accepted.
One story about the Aberdeens and our night visitors: a Columbia connection Wallace and I really enjoyed was with the astronomers. Eckert was one of the clan all his life, and while I soon realized I wasn't going back, I had friends and admired great figures in the science. One friend at Columbia who obviously was going to be a great figure was Martin Schwarzschild, a nephew or thereabouts of the nineteenth century astronomer/physicist/mathematician. Martin had been at Harvard when Dorothy and I were courting, so his presence close to the Watson Lab pleased both of us. He had married a girl who was helping him with his researches just as my wife was helping me.
Martin, who along with a dozen other contributions was later to become the best-known balloon astronomer, was still doing stellar constitution calculations - the sort of thing that almost drew me away from Maxwell in 1938. Interest was high in the outer world because of the atomic upsurge; people other than physicists and ASTOUNDING readers knew that the Sun was powered by a nuclear furnace. So, although he was too pure to be [-91-] interested in fusion power and such, the public interest made him a prime candidate for computer time.
Lillian wired him a complicated set of plugboards, and he came over with his wife to run his cases. We showed them how to print up the results on a tabulator, how to reproduce fresh card decks, how to run a sorter. He and Barbara were sharp; it was easy.
They made their first long run one weekday evening, late. There was a card jam in the monster's punch unit. Probably not their fault; all it took was a defect on the leading edge of one card. The two of them stood there watching. Why weren't the cards coming out?
An experienced operator would have leapt for the stop button. Even from across the room, the change in the voice of the machine would have meant trouble. But it was all new to Martin and Barbara. Suddenly the covers burst open and a fountain of crumpled cards gushed out. Typically, a card had been accordioned to a cardboard rod eight inches long and about an eighth of an inch square - and there were several hundred!
It took two youngsters and the customer engineer almost an hour to clean out the junk next day. Fortunately reconstructing the input cards was not too difficult; while the two delightfully abashed Schwarzschilds were doing it I asked them what they had felt. "Herb," said Martin,"we thought the machine was storing the data." Wonderful!
In Chapter 10 you will encounter
(in order of appearance):
Interchemical Corporation 02
The Watson Lab 01
Watson Senior 01
Wallace Eckert 01
Lillian 03
Marj 03
L.J. Comrie 01
Gerald Clemence succeeded Eckert at the Nautical Almanac Office
INDEX OF MATHEMATICAL TABLES Comrie was publisher and Miller, an author
Dorothy 01
The Royal Society 05
Jack Kissner he had started making foldboats again
Ted Beckhardt and Chuck Weiss foldboaters who fought at Monte Cassino
Dangerous sports white water, skiing, mountaineering and the Cresta
Camping "in later years [the bag] also accommodated a poodle puppy"
The 1947 Buick Roadmaster convertible "wow! did we have a car!"
The WPA Writers Project we had almost all the state guidebooks
The American West "with very few Americans cluttering it up"
An apartment in Chelsea we liked to entertain the troops
Bill McClelland later we shared 701 times in Poughkeepsie
Aetna Womble a southern belle, with an E.E. degree from Duke!
My erotica everybody enjoyed Harris and Casanova and von Bayros
Shostakovich and Benny Goodman on a hand-made high fi outfit
An SSEC birthday party "Lucky old son, with nothing to do..."
John McPherson 01
The Watson Lab electronics group three men from the Radiation Lab
By Havens he designed and built the NORC, and then moved to France
John Lentz father of the long line of small IBM scientific computers
Bob Walker drew an Atanasoff associate, and Frank Murray, to the Lab
Tony Oettinger a summer assistant who later succeeded Aiken at Harvard
Carl Grosjean a Watson Lab fellow who later had a chair at Ghent
IBM education program blueprint reading, business English, and the tango
IBM 407 we got to study the innards of this magnificent tabulator
Stan Rothman acquiring the expertise for Washington two years later
The AJ [Astronomical Journal] I suggest short courses in astromechanics
Numerical methods unusual then, my course drew interesting students
Joe Harrison had been running the Bell relay computer at Aberdeen
Walt Ramshaw gave me fits when he worked at Pratt and Whitney later
IBM policy on married women it caught Marj but not Ellie Krawitz
Ellie [Eleanor Krawitz Kolchin] she took over the Watson Lab shop for me
True Love even The Emperor could not prevail, at least in New York
[-93-] Dorothy and I had lots of pleasures besides working until four o'clock in the morning in a room full of recalcitrant black boxes. Once the war wound down, we began to use the freedoms of the IBM job and the Columbia Connection and my Astronomical Society and Optical Society activities, and also again to enjoy the outdoors. After some hesitation she decided to give up her spectroscopy at the midtown Interchemical Corporation laboratories, mostly to help me, but also to eliminate the problem of fixed hours for her, and conflicting two-week vacations, and the nasty commute from Long Island.
I had no fixed hours at the Watson Lab, nor did any of the professional staff; Wallace gave us the same liberty we would have had across the street at Columbia. And he got the youngsters off the time clock, even though it was an IBM product, and in spite of stories that The Old Man punched a gold-plated one on the sixteenth floor at World Headquarters!
I had classes to meet, but not every semester. So did Thomas, and Eckert himself, irregularly (Lillian and Marjorie did the Machine Methods lab sessions for him). There was a steady stream of fascinating visitors from all over the world: the assistant director of Pulkovo Observatory near Leningrad, which was being rebuilt; a group of Indian statisticians - and Comrie!
He was to be honored at a meeting of the Institute of Navigation for the beautiful tables he had turned out before and during the war, at His Majesty's Nautical Almanac Office. Dorothy and I had the great pleasure of escorting him to the banquet, where he sat between us and Gerald Clemence, who had succeeded Eckert at the Naval Observatory.
He was not too vigorous physically, but full of new computing projects, and with plans to expand the commercial - well, not very! - scientific computing service he had started on Bedford Square in London. He looked rather Scottish (I did not yet know he was a New Zealander; the current pleasant habit of appending a sentence or two of biography at the end of a technical [-94-] article was not then common). He was quiet until things got going, and then his enthusiasms took over. Can you imagine being excited over which numerical type fonts worked better in a math table? Shades of Don Knuth!
He had just published a magnificent INDEX OF MATHEMATICAL TABLES, one of whose authors was J.C.P. Miller. Comrie intended to update and extend it, and Dorothy and I volunteered to work up a cross-file as a third part of the new edition. This involved her key-punching a good part of the 450-page book; the days of machine-readable text were many years off. The major part was arranged by function, and the second was a bibliography of the books and articles where the tables were to be found; we were to furnish a listing of which tables were in each book or article.
I did the system design and printed up the results; she did the hard part. We sent the cards and listings over to England a year or so later, and Comrie added it to the second edition. It was in some ways a contribution to a dying art, like hardening copper and gold (or designing lenses by hand, or wiring plugboards). But I was still deeply engrossed in 1947 computing techniques, although of course excited about how they were going to change as electronics took over.
Comrie, and Wallace Eckert of course, also saw these changes coming. But I had been reading science fiction for over twenty years, if you count the Oz books. I was more at ease than they were, in the future, although I much enjoyed the present and valued the past. About valued pasts: a year or two after Comrie put out the new INDEX he was made the first member of the Royal Society from our tribe. He wasn't keen on honors, but that was one that really counted. He died in 1951.
As soon as gasoline rationing subsided, Dorothy and I began exploring. We had been limited to Long Island and the area immediately around Manhattan even when I had my Farrand C ration, and our 1937 Plymouth had reached a shuddersome mileage. There was lovely country and new experience within reach. I got in touch with Jack Kissner, who designed and built foldboats in Long Island City after the German Klepper type. Ours had been one of the last civilian models he had made in 1941, and he was just getting into production again. He invited us to join the little gang of enthusiasts who went out on weekends to the Housatonic or the Delaware and enjoyed the white water, and camped away from the roads and the traffic.
Along with Jack and his wife and daughter, our particular friends were Ted Beckhardt and Chuck Weiss, who had served together in the U.S. mountain troops. Ted had been wounded at Monte Cassino, and was not yet up to vigorous skiing. Jack, Ted and Chuck (formerly Wolfgang) were all refugees from Hitler Germany. Ted's father left his Frankfurt department store behind and escaped with his large Meissen china collection. This resulted in a family porcelain shop on Third Avenue, well thought of by serious collectors like my Schenectady friend Hans Kraft.
Foldboating was really my first dangerous sport. Following Ted and Chuck I took up skiing; from skiing and a great deal of enraptured reading and book collecting I started mountaineering, and from mountaineering I graduated to the Cresta, which I rode for decades. From sport cars I got into racing, and from that into my third marriage - also dangerous, and also great fun. But it really [-95-] started with white water running. I perhaps ought to mention that I was the first foldboater ever to do the main chute of the enormous Lachine Rapids above Montreal (true, I did it upside down), and that I don't swim a stroke!
A typical boating weekend would involve a hard drive in traffic out of the city, stopping along the way for a quick snack, and setting up a tiny camp on the river. In the mists of morning we would take the cars and folded boats up to the starting point, and a wife or girl friend would take one car back down to the take-out place (we called this the "shuttle"). The heroes and heroines of the day would set up the boats and take off.
Saturday night we would camp again, either out of the boats or where the shuttle car had stopped. This was the night for singing - usually German - and campfire marshmallows, and loud snoring. Sunday afternoon we would stop early to dry the boats and repack them, becoming dry ourselves in the process. The drivers would be shuttled up to get the other cars. If possible we would then go to a nearby restaurant and have a large and enthusiastic meal; there was a big place called Flo-Jean's in Port Jervis on the Delaware that always fitted us in, along with their fancy Sunday-driver customers, even when one of us was in oilskins and a single espadrille, having lost his all in the river. The owner was a canoeist.
For this my wife and I had to become campers. She had done a lot of early-California auto camping; I had hardly slept outdoors. I applied the usual medicine: went out and bought books about it! Then we found postwar sleeping bags - no down ones available yet - and cooking gear, and a very small olive drab tent. To erect it you used a short piece of wood as a ridgepole, strung up the ridge using the lower branch of a friendly tree, pulled out the side ropes, and behold: room for two good friends. After a little maturing we traded up to down sleeping bags that could be zipped together; it was warmer on cold nights, and had other advantages. In later years the combination also accommodated a poodle puppy.
This weekend action was hard on cars. They got into messy places, needed to start easily after a night on the river, were required to hold large quantities of junk, plus foldboats. I took our wartime savings in hand and went looking for a replacement for the faithful Plymouth. "Not possible," said the Long Island small-car dealers, "we have a two-year waiting list of pre-war customers." But this was New York, where people with friends or money had lived well in spite of red points and blue points and gas coupons.
I thought about it. In Manhattan fancy people were lined up outside the Cadillac dealers, just as hoi polloi were queued for Fords in Hempstead. But how about fancy cars, in the boonies? Jackpot! They took my order for a Roadmaster Buick convertible, with every option and accessory in the catalog. Two years for a Chev, one year for a regular Buick, three months for a Roadmaster; two months if you ordered fog lights. Voila!
Dorothy and I had never even had a car radio. We were living in a cheesebox, and heating our hot water with pea coal. But wow! did we have a car! I had vacation saved up, and besides, we astronomers ought as a matter of professional business to attend the dedication of the 200-inch telescope on Palomar. Eckert didn't like to travel, thank heaven.
Dorothy and I planned the trip for a year. I had maps of all the national [-96-] parks and monuments, most of the national forests, special river maps that showed profiles - where the rapids were steepest. Since marriage we had collected an almost complete set of the WPA Writers Project state guidebooks. Gasoline companies gave away - yes, this was a long, long time ago - beautiful highway maps.
I marked places where no Roadmaster, top up or top down, had ever been: a tiny dirt road along the Colorado into Moab, the Escalante Canyon, Devil's Postpile, the back road down from Lick Observatory. We filled the trunk, filled the back seat area, put in our foldboat and camp gear and the very small tent. And away we went.
We saw the American West at its best, with very few Americans yet cluttering it up. We camped, alone except for a few scrawny sheep, at the Black Canyon of the Gunnison. We drove most of the way around Crater Lake, until blocked by snowdrifts - in July. We cruised for hours on deserted roads at eighty miles an hour. We crossed Death Valley in midsummer with the top down.
And we indeed helped dedicate The Big Eye, along with hundreds of astronomers from all over the world. We saw Dorothy's family. On the way back we stopped in Reno, just before show time (Vegas was primitive by comparison, in 1948). Dorothy ordered a trout; the waiter offered to bone it; she sent him away, proud of her own skills - and the lights went down for the show!
About the eighty miles an hour: I hadn't yet gotten interested in sports cars or racing, and wasn't knowledgeable enough to allow for the fact that the huge Buick straight eight engine was splash lubricated. We burned out the bearings coming up from the Grand Coulee. I sold the lovely beast "as is" in Spokane, where there was a great apple harvest in prospect and no one had yet even dreamed of a Roadmaster convertible, with or without fog lights. We did Glacier by train and Yellowstone and the Tetons by bus, and flew back to New York from Jackson. The Buick dealer in Spokane bundled up our big suitcase of maps, our foldboat, and our very small tent, and expressed them home for us.
No Roadmaster meant problems with our river excursions. I had been in the Blue River above Dillon, and in the Gunnison, in Colorado. And one of the high points of the trip was a day on the Snake, below Teton National Park. So I wasn't exactly hurting for white water. But when the spring runoff started and the sun began to be warm even in Connecticut, we'd need transportation.
It wasn't an immediate problem, because we had moved into town just before we left on the Palomar swing. Dorothy said a reluctant goodbye to her victory garden, I said good riddance to the pea coal supply, and Suzy the cat yowled wistfully for her lost territory. We found an interesting apartment in a Chelsea townhouse, with the owners below to guarantee good heating and garbage disposal. There was little storage space, and the foldboat went to live permanently at the lab, in the storeroom where I used to patch it after the wilder trips.
We could go up to Columbia in twenty minutes on the subway, and that was a real plus; the combination of driving to the Long Island station, changing in Jamaica, battling the crowds in Penn Station, and riding the [-97-] subway to 116th Street had taken an hour and a half each way. I had usually driven, and didn't mind with the older car, but even in the Forties leaving something like the Buick on a New York street all day was asking for trouble. I put a large sign saying "Trunk and glove compartment are EMPTY" in the windshield after the top was slit the first time; I added a note that the doors were unlocked. The police ran me down one day and warned me that unlocked cars were handy privies for bag ladies and such; I left undisturbed the note that the doors were unlocked - but locked `em. Also, although the police were too busy to prevent damage to pretty cars, they were not too busy to hand out parking tickets. Ah, New York!
Another real plus was that Dorothy and I could now entertain - not just visiting Brits, and other friends from out of town, but the youngsters from the Watson Lab and the SSEC. Marjorie was rather busy with her new husband, Jack Herrick, and the natives like Lillian had their own families and friends, but Betsy Stewart, the handsome woman at the SSEC console in the traditional pictures, and Bill McClelland, and a certified Southern Belle named Aetna Womble, and Woody Skillman, and a dozen other bright newcomers used to come to our place, huddle over my von Bayros prints, and Frank Harris and Casanova, listen to one of the first modern high-fi record players in town doing Shostakovich and Benny Goodman, and pet our amiable cat Suzy.
Behind Aetna's charm lay an electrical engineering degree from Duke. Behind my amplifier circuits lay the Watson Lab electronics shop. Behind Frank Harris lay a growing collection of attractive erotica (the Danish pornography revolution was a decade away). Behind the little evenings lay our memories of how observatory people and Cal Tech people used to enjoy each other. The Eckerts and the Thomases lived in a Columbia enclave in Leonia, New Jersey; the IBM executives lived in luxury suburbs like Greenwich. We were central, and liked it.
There were birthday parties for the SSEC. Since it was busy around the clock, and since the No Booze rule was really serious, they were held outside. The second one was in the Dogwood Room on 58th Street, a relief stop for trampled IBM executives not strong enough to face Grand Central. John McPherson and his wife came, much to everyone's pleasure, and I seem to remember Wallace and Dorothy Eckert. Rex Seeber and his wife, who lived up in IBM country, were sort of givers of the feast.
There was a piano player, and much drinking. McPherson pretended not to notice, but could hardly ignore my asking the piano player to do what I christened the Tom Junior Song: "Lucky old sun, with nothing to do but roll around heaven all day." Nevertheless it was a good party.
There was one other social component in our circle; I've hinted at it only slightly. Call it the other half of the Watson Laboratory, or the third part of the IBM Pure Science operation, or call it the electronics group. When Wallace began to hire senior personnel for the lab, he looked also for men who would supplement what he knew as the IBM engineering department. The staff director of engineering was his, ah, monitor, John McPherson; the Hamiltons and the Luhns and the Piatts, and the Lakes and the Dickinsons before them, got their floor space and their people and their paychecks from Endicott [-98-] executives, their detailed technical instructions from John - and their projects from T.J.
Eckert had seen some of this, and I assume was briefed about the 1945 situation in IBM when he came aboard. He realized that the new electronics was not yet represented in Endicott; he was close to Rabi, who was Number Two at the MIT radar laboratories; he had some modern contacts even at the old- fashioned Naval Observatory, and certainly through astronomer friends who were doing war work. One of Dorothy's bosses from Mount Wilson, for instance, turned up doing magnetic mines, and others were in proximity fuses and such. Wallace decided to help.
Through Rabi and other Columbia friends he turned up three men who were ready to leave MIT: By (Byron Luther, a deep secret) Havens, John Lentz, and Robert Walker. To tie them to their later accomplishments in IBM, Havens was the designer and builder of NORC, the Naval Ordnance Research Calculator - the first supercomputer, delivered to Dahlgren in 1954. He later moved to Europe and became a senior manager at the beautiful IBM laboratory up behind Nice.
John Lentz was the father of the early IBM small scientific computers. His work led some years later to the 1620 and its progeny. Bob Walker built an analog linear equation solver, somewhat after the pattern of Clifford Berry, Atanasoff's associate. He also worked with a Columbia math professor named Frank Murray, who was involved with the Cyclone machine (then still analog).
Eckert brought them to Columbia soon after Seeber and Thomas, and set them up as an electronics group; today we would probably say, components group. Walker took over the physical arrangements; there was a major workshop in back of the second floor, with three technicians and an expediter (most supplies came up from Cortlandt Street and the surplus houses, at least in 1945 and 1946). He planned and set up the machine shop in the sub-basement. And later on he helped Havens put a NORC shop up in the Goat Room, a windowless fifth floor space with an elegant floor and finished walls which I claimed was where the fraternity toughs had initiated their new brothers. It was the first space in the Lab to be air conditioned: mid-1947, I'd guess.
There were few links between the Columbia professoriate and the three radar experts, mostly because people like Ray Mindlin had been behind other walls doing other war work. Walker became friendly with Murray, as I said, and saw a little Faculty Club action. Lentz wanted to get his Ph.D., and so in spite of his research record at MIT was in statu pupillari. And Havens was soon too busy to bother - tended to grab a sandwich at Chock Full O'Nuts. Today all three of them would be running work stations: be heavy users of computing. Not in those days! But I was fascinated with what they were doing, and played Mohammed to their mountain for several years, so there was a link.
You have to remember there were no transistors yet. The only semiconductors I had seen were germanium diodes, exotic one-way devices which IBM had not yet dared use behind the plugboards to eliminate back circuits. The SSEC was under construction, using full-sized vacuum tubes. Havens and Lentz were using miniatures. Calculations and storage were decimal; ENIAC used ten or eleven bottles per digit, the SSEC I think seven. Havens was trying [-99-] for four - that is, pure binary coding for each decimal digit. His successes (or initial successes, because it took a lot more than that to eventually build a room-sized NORC) were carried off to Endicott, and later Poughkeepsie, but Havens wanted to put together a whole system, not just furnish ideas or experimental components to the boys upstate.
The electronics shop offered summer assistantships, and early on drew a student from Harvard named Tony Oettinger, who succeeded Howard Aiken when he finally retired, and later still became a president of ACM [1966].
My side of the house provided postdoctoral Watson Laboratory fellowships. The first holder was a young Belgian physicist named Carl Grosjean whom I met again in the late Seventies; he had the chair of theoretical physics at Ghent, and was on the verge of retirement. Good grief! Another was a chap from Stuttgart named Helmuth Sassenfeld, who returned to the U.S. later and went to work for von Braun in Huntsville. And we had Americans, of course.
I used the term "social component." Bob and Mildred Walker found a place near our little Hempstead house, and we saw them frequently. John Lentz and Alice, who was a niece of Justice Burton of the Supreme Court, found a Riverside Drive apartment through Columbia. It had typical Manhattan advantages and disadvantages; Burl Ives lived across the hall, and there were lots of cockroaches. It was partly envy of their location that drew Dorothy and me into town later, and observation of Alice's problem that made us wary of insect life.
John was an audio buff, in the days before this was common. He had a huge amplifier, built while they were at MIT, and was trying to add a variable-gate preamp to cut the horrid scratch and rumble of 78 rpm shellac records. This is what is inside Dolby noise reduction circuitry today. He never really got it right, but like much else being done at the lab, it was interesting, and it was very early.
He designed a later model amplifier for me, which I built at night in the second floor shop, making the sheet metal chassis down in the sub-basement. It used big black metal 6L6 tubes and a ten-pound audio transformer, and even by today's standards was really very good. Dorothy and I found a Hempstead record shop with a big overstock of classical 78s, and arranged to return albums that we played and didn't like - music or recording or too much scratch. Soon we had fifteen feet of solid shellac, although we never went the Shapley Gilbert and Sullivan route.
John was too much of a purist to have a record changer, but we had an early English Garrard. After some 1946 investigation, he advised me to get a GE variable-reluctance pickup, and it did yeoman duty until I retired the 78s fourteen years later. I still have two or three albums, notably the Szigeti Beethoven and Piston's "Incredible Flutist" (and some vintage Goodman), but don't have the machinery to play them now.
The records and the Lentzes led us to Carnegie Hall, where we had a subscription for two years, but the problem of getting home to Long Island was serious. By the time we moved to Chelsea, we were doing Town Hall concerts and Broadway musicals; the "Oklahoma" revolution was in full swing. Senior IBM executives tended toward the Metropolitan Opera because of the Watson involvement, but I was not high enough in the heirarchy to ask [-100-] for seats in the Imperial Box, and Wallace would have died a thousand deaths if made to go.
About one other Watsonian interest: T.J. was a great believer in education. In Endicott it tied to his job enrichment philosophy. Where industrial engineers elsewhere were trying to break bench jobs and assembly line jobs into tiny specialisms - tighten one screw and pass the gizmo to the girl on your right - Watson went the other way. A new employee was put on a single-spindle drill press, say. After a few weeks he or she would be encouraged to sign up for a course - blueprint reading, perhaps.
Next might come promotion to a multi-spindle machine; then a course in machine tool operation; then a simple milling machine; then a course in measurement techniques (and you got to keep the micrometer and the height gauge). Soon there was a little workshop around the employee; he put the parts in appropriate jigs - and could get awards for suggesting tooling improvements - set up his machine or machines, ran the pieces, inspected the work on his little surface plate, and asked for the next batch. Not at all coincidentally, IBM had no unions whatsoever.
Those who didn't take courses didn't rise. And, just as in the sales force word on white shirts and striped ties soon spread among recruits, so shop people soon discovered the advantages of the IBM education program. Out in the field it was more difficult, but even there you could sign up for correspondence courses in Business English or Principles of Accounting Machine Operation. World Headquarters was riddled with courses: released time, after hours, recreational.
Didn't look too apposite for an astronomy Ph.D. But I couldn't ask my young people to do what I wasn't doing, so after much thought I signed up Dorothy and me - yes, families were invited to many courses - for Social Dancing I. We learned to tango in the second-floor IBM cafeteria at 590!
Toward the end of my hitch at the Lab I managed to break the barrier which kept all but aspiring customer engineers out of the serious technical courses which described the innards of the machines. I took a course about the magnificent 407 tabulator, and found it very broadening. One or two of the senior youngsters did the same - Eric Hankam of course, and Stanley Rothman, who was to come down to Washington with me in 1950.
Before I leave these small tents behind, I'm going to write just a little about the courses we gave - that is, at Columbia. These were all part of the regular university curriculum, listed in the appropriate catalogs - we had our own special one also - and open to any student with the prerequisites and the money. We did however encourage our own juniors on 116th Street and at the SSEC to attend as auditors if they did not want to sign up for credit. I vaguely remember that Eckert made arrangements to have IBM pay tuition in the for-credit case, but that there were no takers from downtown; the pressures of SSEC operation were too great. But almost everybody did a little auditing.
Most of our offerings were unusual. Thomas did a very good course in theoretical physics, in which he was a world authority. I did a celestial mechanics course one year; it was really a mélange of spherical trig, practical and theoretical astronomy (meaning time and position determination, and orbit computing), and brief mentions of planetary and satellite mechanics.
[-101-] None of my subtopics were taught anywhere else at Columbia; the astronomy department was solid astrophysics. And they were what was needed for astronomy calculations.
Some years later I was asked to do a short piece for the Astronomical Journal, the very staid professional publication for non-astrophysical astronomers like Brouwer and Clemence and Herget and Eckert. I suggested that with the missile and space programs booming, there was a place for a master's degree level in my kind of astronomy, and jobs going begging. I was delighted to see summer institutes start up almost immediately, and more carefully considered curricula follow at several universities. My once-around Columbia course would have been just right to start those students off.
Most of our value as teachers, however, came from the computing courses. Eckert gave a two-semester machine methods course, which featured hands-on operation under Marjorie, Lillian and Eric; literally the only place in the world where you could learn in the university milieu. He didn't do much about the Aberdeens, because they were not available out in the world. And he didn't know Pete Luhn's two machines apart. But all the regular machines, including the very newest, and prototypes - those were done very well.
I did numerical methods - classical interpolation and matrix arithmetic and integration of differential equations. Most of my examples, and assigned exercises, were at desk calculator level, but I lectured from the point of view of machine operation; in a sense, I did a Comrie. This was one semester, once a year, and Hilleth did an advanced course featuring partial differential equation solutions and error propagation, every other year.
My classes were small; this was a very esoteric discipline indeed in the Forties. But I had interesting students, and not just the auditors like Rothman and McClelland and Backus and Don Quarles. One year I had an experienced computer from Aberdeen, who could tell us stories about running the Bell Labs Model V. His name was Joe Harrison Junior. Twenty years later he was to take over the computer standards work under me at the Bureau of Standards.
Another was Walter Ramshaw, who had worked during the war in naval architecture. At the end, although he was married, he decided to become a scientist instead: an astronomer. He took my two courses, and was beginning to run out of money even though he had hardly started on his Ph.D. Walt was extremely good; I suggested he peel off into computing full time, perhaps using his engineering background as well. He disappeared; six years later he surfaced as the number two computer man at Pratt and Whitney, and gave me a terribly hard time as I struggled to get the first 704s for Generous Electric. He was also one of the dozen or so founders of SHARE.
If By Havens had had the time, or if Lentz had had his doctorate, the electronics courses at Columbia would have gotten a great shot in the arm. But Havens already had the NORC bit in his teeth. So it was my side of the house that carried the teaching. It went on into the Fifties, always as part - but a small part - of the Columbia offerings. The hands-on side of the Machine Methods course was unique, not just because of the equipment but because real use-'em-every-day men and women were running it.
I mentioned how IBM treated its young draftees during the war: attention to the families, supplementary money, and so on. One other thing was that [-102-] they were all promised their jobs back, when things settled down. IBM had always used women, and was quickly able to recruit many more to fill in for the missing males. But it took the precaution to ask new female hires to agree to leave after the war if they were married, to make extra room for the returning heroes. I didn't know this, of course, but woke up to it painfully one day when Marj Herrick told me she had been given her congé.
Wallace and I tried hard to get a waiver, because of course she was in a job that didn't exist when the boys left, and that none of those returning could immediately fill. No soap; it was an edict from On High. Marj wasn't too upset; she and Jack were not very fond of New York, and unencumbered. She was snapped up quickly by Ken Arnold, a statistics professor at the University of Wisconsin who was starting a punched card lab, with a machine methods course contemplated. Jack Herrick found a job in Madison and we staged a great farewell party. IBM made a mistake, and lost not only a good employee but much standing with some important observers.
Two notes: Lillian had retired to married bliss in Brooklyn before Marj was beheaded, and I asked Eleanor Krawitz, her second on the Aberdeens, to take over as supervisor. Ellie was attractive - an understatement, pointed up by the Lab story about how she had stomped a visiting Swede who offered unwanted attentions - and she had a steady boyfriend. After the no-married-women clause was revoked in IBM, there was a grand party to announce their marriage - some years previously!
There was a woman executive in WHQ personnel, also attractive, who often brought her "steady" to company affairs. They were sitting next to us in the Hall of Stuffed Birds when Dorothy got the giggles, for instance. She was part of the beheading action that caught Marj. Many years later, long after the edict was abolished, she died. And of course it turned out she had been married all the time. Even Watson Senior, Emperor Of All He Surveyed, could not prevail over True Love - or New York ingenuity!
In Chapter 11 you will encounter
(in order of appearance):
L.J. Comrie 01
Wallace Eckert 01
R.A. Fisher and T.N.E. Greville statistical and actuarial computers
R.V. Southwell pioneer English engineering computer
Watson Senior 01
IBM Education Department ran hundreds of "classes" for the customers
THINK magazine 01
BUSINESS MACHINES IBM news for internal distribution only
IBM POINTERS machine usage tips for IBMers and advanced customers
IBM computing meetings eight, in Endicott, from 1940 through 1951
WPA Mathematical Tables Project ended up at NBS after the Depression
John McPherson 01
Japan "a very IBM-like country"; I remembered the songbook and such!
The 1948 conference dividing line between pre-war and post-war computing
Bill Bell one of the new breed, from Lockheed and Telecomputing
General Purpose Boards demon board wirers did them for five machines
Ward Beeman put Bill's Telecomputing earnings into peripheral hardware
Maxwell 05
E.C. Bower an astronomer who pioneered unsuccessfully at Douglas
John Lowe he made Santa Monica a leader in tin airplane computation
Chuck Baker "head grip" for Lowe, and still programming 50 years later
The Rand Corporation think tank spun off from Douglas by Hap Arnold
Paul Armer organized computing at Rand, and later helped found SHARE
Cecil Hastings the Great Approximator; in L.A., he never had a car!
Moore School 1946 conference put von Neumann's seal on binary machines
Ed Berkeley he wrote about "Giant Brains" but cared more about people
The ACM [Association for Computing Machinery] a world first in 1947
Grace Hopper 01
The BCS [British Computer Society] a decade later, a much better name
The AIEE [electrical engineers] had a Committee on Computing Devices
The IRE [radio engineers] had an Electronic Computer Committee
Joint Computer Conferences the engineers and ACM started the parade
The DPMA [data processing managers] joined the Joints in the Seventies
The 1949 IBM conference Neumann, Tukey, Mina Rees and Tom Junior
Watson Junior [Thomas J. Watson, Jr.] a power even in the late Forties
John Tukey consultant to the Bell Labs, he ignored the "No Prunes" rule
IBM dominance 09
Alan Turing known much earlier than COLOSSUS; his sad death diminished us
Overseas computer meetings started about 1953
International computer meetings the first one was not until 1959
[-105-] While Dorothy and I were paddling and camping and driving around the country, a new way of transferring knowledge about computing was taking shape. In the Thirties such knowledge was passed literally from hand to hand; almost nothing was published - indeed, by the formal standards of the time, not much was publishable.
There was a little college of astronomical computers, a little college of statistical and actuarial computers, a tiny group of civil and mechanical engineering computers. Comrie and Eckert were deans of the first, R.A. Fisher and T.N.E. Greville were familiar figures in the second, a pioneer named R.V. Southwell had a few disciples in the third. The members of each group knew all the others in the group; corresponded, traded a few reprints. The day of the intercontinental telephone call, let alone of electronic mail, had not dawned. Only an occasional major figure like Comrie knew of the existence of other colleges.
There were people who needed to do a lot of X-ray diffraction calculations, and people who did airplane wing and turbine bucket shapes, and others in special disciplines like lens design. The slide rule manufacturers and the comptometer and desk calculator salesmen did not pull them together; I remember with what sad amusement Maxwell and I flipped through a table of trig functions given us pridefully by one of the latter. But, as in many other ways, IBM was different.
This stemmed more from Watson's enthusiasms than from the nature of his customers. Long before scientific and technical applications were recognized as a business area, T.J. was spreading the gospel of better accounting, better bank operation, more use of his machines in schools and hospitals. If he sponsored business arithmetic and typing courses - yes, and tango classes - for his office people, and insisted that factory hires learn to read blueprints, you could be sure he would impinge on the customers.
[-106-] This was originally done by the Education Department, although the lack of a formal organization chart made it simple for T.J. to make special assignments. The little gang that put out THINK, IBM's prestige publication, and the various internal editions of BUSINESS MACHINES, could print special publications, or something like the retouched SSEC brochure; a young engineer who had realized that customer contact was required for advancement in the company could edit POINTERS; the housekeeping departments could put the Endicott Homestead and the country club facilities at the disposal of a customer group.
By the late Thirties a pattern of customer conferences had developed. I have a rather primitive printed proceedings from a 1940 statistics meeting, for instance, jointly sponsored by the Psychometric Society, and including a paper from the man at Michigan, Paul Dwyer, whose punched card installation Maxwell had ignored.
Not much could be done during the war; everybody was madly busy, and there was travel restriction and rationing. But the machinery was kept oiled and ready. In early 1946 Wallace coolly informed me I should think up something to talk about for a Research Forum, a follow-on to the 1940 conference, to be held in August at what was still called the IBM School, in Endicott. I overshot the mark somewhat - the 1946 meeting was slightly less technical than the 1940 one - and did "Harmonic Analysis by the Use of Progressive Digiting." Wallace described the facilities of the Watson Lab (the SSEC was still under wraps, of course), and was more popular.
There was a 1947 "Educational" Research Forum at which Eckert spoke. The proceedings were a little more attractive, although still in paper covers. Support was not a problem; what our computer friends needed was content.
We looked around. There were two or three dozen punched card shops doing technical calculations. All used IBM equipment; the round-hole boys (Remington Rand) weren't in it, and anyhow didn't put on customer "classes," as the IBM Education Department unfortunately insisted on calling even the most scientific conferences. I called around, and wrote a few letters: the Naval Observatory, the Bureau of Standards (where the WPA Math Tables Project had alighted after the Depression was swept away), Aberdeen and Dahlgren, the Census, two or three tin airplane companies, General Electric; Paul Herget.
Looking back, I'm amused to remember I didn't use the salesman network; these people had all been in direct contact with us at 116th Street. If I had realized then how power flowed inside the company, I would have invited each outfit through its Friendly Local IBM Representative; as it turned out, many of the men and women who came and gave papers, brought their salesmen with them - they knew how the company worked, even if I didn't!
I was sort of the program chairman. Eckert was sort of the general chairman. John McPherson was sort of headquarters representative, although somebody gummy gave the address of welcome. The group - about eighty, including us company types from New York - were put up at the lovely IBM Homestead, at the junction of the two 18-hole IBM golf courses, and not far from the IBM Country Club. Transportation was provided by IBM busses to the IBM Education Building, which had a giant brass THINK sign over the door, featured in the formal portrait of the class taken on the steps by the IBM [-107-] photographer and printed in the next issues of the IBM Endicott and IBM Headquarters BUSINESS MACHINES. There was a tour of the IBM Research Laboratory.
When I first began going to Japan in 1960, I looked at all the company insignias and remembered the Endicott ambience. Japan was (and is) a very IBM-like country. In 1947 in Endicott, the IBM label and the THINK signs and the standard portrait of Mr. Watson were everywhere. And yes, there was a company songbook - but unlike Japan, only the sales classes and the Hundred Percent Club members had to sing. There was an IBM symphony too, which T.J. had commissioned between the inventors' portraits and the ceramics collection, but nobody had to play it.
Even though they were scientists and engineers, the customers were impressed. And they were also impressed by the handsome decor and the excellent meals at the Homestead, and even by the malted milks served instead of nightcaps. They were restive at welcoming speeches, and not too excited by machine descriptions and lab tours; most of them had private lines into one company area or another: Los Alamos and RAND Corporation types directly, GE and airplane company types via their salesmen.
This 1948 conference was the IBM interface between the pre-war and the post-war - the pre-ENIAC and the post-ENIAC - computing experts. Almost all the papers were about the use of standard electromechanical IBM punched card machines, but the ones by oldsters like Gertrude Blanch signaled the last appearance of such people on the scene; they tended to have good pre-war degrees in mathematics, and to do pre-war tasks like table making, or statistics.
The newcomers had degrees in physics or aeronautical engineering, or frequently none at all. They were trying to do hard science, or design airplane structures; they were impatiently waiting for time on the SSEC, or delivery of new IBM equipment, or access to the electronic gear their private lines into Endicott and Poughkeepsie promised them. The old IBM sales apparatus had pretty much ignored the pre-war group; a new apparatus was not yet in place to handle the post-war group, unless you count our amateur efforts at the Watson Lab.
At Lockheed Burbank, there had been a very strong wartime computing shop revolving around an entrepreneurial type named Ward Beeman and a technical type named William D. Bell. Bill came to the Endicott conference, and told us that the group had split off from Lockheed, taken the applicable IBM machines, and formed a technical service company called Telecomputing. They had brought with them a contract from Lockheed to continue doing engineering computations, but were free to take on other work. Because the IBM equipment was rented, no large capital investment was needed, only a certain amount of confidence in their viability by the local IBM sales office. It was the first American enterprise of its sort, and preceded on the wider scene only by Comrie's Scientific Computing Service Limited in London.
Bill Bell was a fine example of the new breed. He was a handsome Californian, with a handsome California wife named Max and two handsome blond California sons (one of whom graduated from Hollywood High, and shortly thereafter turned out not to be able to read). Bill himself had [-108-] no degree; he had worked at an assortment of California-type jobs, including some in or near the movie industry. He was a good swimmer, a fine horseman, and a professional-class photographer.
He had worked himself up through non-scientific punched card operation into the computing group, which was then doing simple stuff like weight and balance control, and getting ready to go into structures calculations; aerodynamics was not yet feasible. He had taught himself the math he needed, including calculus and matrix arithmetic; knew little formal numerical analysis, but all that tin airplanes needed - engineers who had trusted their slide rules worried about loss of accuracy in long strings of IBM calculations, and Bill handled this by perturbing the input data slightly and watching how much the answers were affected.
He became famous among his late-Forties peers as a demon board-wirer. He was the father of the so-called General Purpose Board, which first out-Hergetted Herget on 601s and 602As, then in 1949 was redone for the 604, and in 1950 and 1951 was redone again for the two models of the IBM Card-Programmed Calculator, which I haven't said much about yet. These boards, packed with plugwires and split wires and connectors (and illegal germanium diodes, which had to be concealed from all but the most-trusted customer engineers, whose jobs were at risk if their IBM managers found out), did whole sequences of arithmetic, and in later versions even did primitive floating point.
Telecomputing - at least the business side - didn't want copies made of these confections. But like the atom bomb, the big secret was that anything that complex could be made to work at all. I don't think there was a single man or woman in the regular IBM who had ever wired such things, although the Aberdeen boards at the Watson Lab were just as complicated, and twice as big. The skill spread.
Parenthetically, Telecomputing did very well for several years, until Beeman and others took the profits from Bill Bell's operations and went into the hardware business. In other places I have mentioned a wonderful point plotter that counted the lines on graph paper; they also built a line of chart and film readers which punched digitized readings into cards. In the end that part of the business was bought out (the name survived because it was so attractive), and Bill's computing services became the Mellonics Division of Litton.
At Douglas Santa Monica an even more unusual sequence evolved. Back in 1934 or so, a California astronomer of Maxwell's generation had published some applications of Comrie's subtabulation procedures in, of all places, the Lick Observatory Bulletin. You would expect only the Grosches and Maxwells and Cunninghams and Hergets and Eckerts and Comries to read such stuff. But somehow this man, Dr. Ernest Clare Bower, was brought into wartime Douglas and turned loose in their punched card room. If it had been Cunningham, for instance, remarkable things might have happened; with Bower, it didn't work.
But running a sorter in that shop was a hulking operator named John R. Lowe. He had had a really rough childhood and a terrible education, had been on the road for a while, had had an accident which left him with an artificial leg - and draft-proof. He saw what Bower could be doing with the IBM [-109-] machines if he would forget astronomy and think about the airplane business. By 1947, John was in charge of a special group in Santa Monica almost as large as the one at Lockheed. Like Bill Bell, he taught himself the mathematics he needed, and after the war recruited youngsters who had still more. And, at least as important as his own accomplishments in Santa Monica, he brought along computing facilities at two other Douglas locations, El Segundo and Long Beach, and trained a good number of younger experts who fanned out over the aerospace community. The best known, still active in computing today, was Chuck Baker.
There had been an advanced planning activity at Douglas with connections to General Hap Arnold of the Army Air Force. It was split off and separately funded as the Rand Corporation (Research And No Development, it was rumored). When I first visited them they had moved their IBM machines to the basement of a downtown Santa Monica bank. The key figures were a forceful organizer named Paul Armer, who will appear several times later in this book, and a very strange mathematician named Cecil Hastings. Cecil wrote the first New Computer World hardcover book on numerical analysis, called THE INCOMPLETE APPROXIMATOR: a collection of amazing formulas to replace functions like sines and logarithms and probability functions - they used probabilities a lot at Rand, and still do!
Most of his approximations were what are called rational functions: the ratio of two polynomials. These were computable by IBM machines from the 602 on, whereas looking up the logarithms and such in a big punched card table was a slow process even if you had hundreds of values to pull out each time. Only the SSEC really tackled the problem head on, until the end of the Forties.
You had to be very curly-brained indeed to dream up these strange beasts, and Cecil certainly was that. He did not have an automobile - in Los Angeles! He rode a bicycle to the beach, where he said he did his most creative thinking, undistracted by the pre-Muscle Beach bodies. In the end he married a Chinese-American girl, moved to Hawaii, did a little consulting, and disappeared. As far as I know, he never bought a car.
Armer and Lowe were founders of SHARE. Bill Bell might have been; he would almost certainly have gotten one of the 704s. But as the machines were zooming onto the scene, he contracted multiple sclerosis, the Lou Gehrig disease. He managed nevertheless to produce the first hardcover book on business data processing, which McGraw Hill published. His was the first death among the new computer people, and it was deeply regretted.
About a third of the attenders at the 1948 meeting, which I christened the Scientific Computation Forum (the Education Department called it Customer Administrative Class No. 446) were oldsters; the rest were newcomers. My own case was unusual; I was as young as the new breed, and using the very most advanced equipment, but I also had a decade of desk calculator work and numerical analysis behind me. I was both the youngest of the old breed and a leader of the new.
For the 1948 Forum I did what turned out to be my last numerical analysis paper. I had developed ways of condensing the large traditional printed math tables down to a very few punched cards, starting with reciprocals back in [-110-] 1945 when IBM calculating punches could not divide, then moving to trig functions for the Luhn relay machines and the SSEC, and then to fancier stuff and a general formulation. In the end, each card of my optimum interval tables was a junior version of one of Cecil Hastings' approximations.
The difference was, I substituted routine punched card procedures for his intimate knowledge of the target functions. The paper in the proceeding gave the details, but as I said earlier, the flood of cheap memory made the technique obsolete by, oh, 1954. Cecil's ingenuity was not transferable, and disappeared with him; some parts of what I did survive in a tiny part of present-day numerical analysis called spline theory.
The conference was a big success. I edited the proceedings, and got IBM to put them out in hard cover - a first for the company. I seem to remember the print order was a thousand (copies still survive; not many). Plans were set afoot for an annual event.
When I said this was a new way of transferring knowledge about computing I was referring to repetitive meetings. History buffs know about single meetings of great importance to the art which date back before electronics: the Napier Tercentenary celebration for instance. And the influence of Johnnie von Neumann spread from the Moore School summer conference in 1946. IBM indeed started the pattern of continuing meetings, but as you would expect it was taken over by others, notably the professional societies.
In the fall of 1947 a small group of enthusiasts met in a room of the chemistry building at Columbia. The call was issued by Edmund C. Berkeley on behalf of a half-dozen men he had brought together to start the world's first computer society. Ed was at that time an actuary at Prudential, and had persuaded his company to order an Eckert-Mauchly Computer Corporation machine. He was the author of the first computer book, GIANT BRAINS, and for many years published COMPUTERS AND PEOPLE, the oldest popular journal in our field. He always lived in the future, and he served it well.
Eckert and I attended - he probably had arranged for the room. So did Grace Hopper. There were perhaps sixty persons altogether. We agreed to call the society the Eastern Association for Computing Machinery. In line with what I've been saying, the first order of business was to plan a series of meetings; the 1948 one was to be at Aberdeen, and I have mentioned how Johnnie von Neumann ran one of the sessions. The 1949 one was at Oak Ridge, in Uffelman country.
Grace, at her death Rear Admiral Grace Murray Hopper, U.S.N.R.(ret.), was the most famous and the most respected woman in our field. Of the charter members at that Columbia meeting, she and Ed Berkeley (and I) contributed to the outfit for decades, but only I survive as an active member of what is now universally called ACM. The name is unfortunate; since "Eastern" was dropped repeated efforts have been made to change it to something about people rather than machinery, but it's apparently too late. When the Brits started their outfit a decade later, they called it The British Computer Society, which is better.
The two major engineering groups at the end of the Forties were the Committee on Computing Devices of the AIEE, American Institute of Electrical Engineers, and the Electronic Computers Committee of the IRE, Institute of Radio Engineers. They put on a joint Computer Conference in Philadelphia in [-111-] December 1951, with ACM participation, and from this sprang a long series of conferences and exhibits: the Eastern Joint/Western Joint/Spring Joint/Fall Joint/National Computer Conferences, of which the 55th, the last successful one, was held in Las Vegas in 1986. ACM was a full member from 1952 on, the parent AIEE and IRE societies merged in the Sixties, and DPMA, the Data Processing Management Association, joined in the Seventies.
ACM began its big series of technical publications in the early Fifties, and the other associations and special committees also began to publish. Conference proceedings, usually a good deal less lavish than the IBM versions, began to proliferate. Everybody and his granddaughter wanted to have a meeting; the IBM excitement was immense, and still growing. If you count the 1940 IBM Forum as Number One, my 1948 Forum was Number Four, and four more followed.
The Watsonians reached out expertly. Where Wallace and I relied on our very large circle of acquaintances, the sales people began to make lists; everybody who had IBM equipment and was using it for anything technical, and everybody who was thinking about getting same, was on them. From the other end, the fancy scientific types who had not been interested at the time of the ASCC or ENIAC or SSEC dedications now saw the von Neumanns and Tellers and von Karmans deeply involved, and began to accept invitations to Endicott and Poughkeepsie they had brushed aside earlier.
As an example, the November 1949 meeting in Endicott attracted Herman Kahn, Maria Mayer, Mina Rees, John Tukey, and von Neumann. Tom Watson Junior opened the sessions. The big guns had begun to fire. I remember the meeting for a homely reason. Tukey, who was a world-class statistics theorist, an important consultant to the Bell Labs, and a father of both the Monte Carlo method and the Fast Fourier Transform, liked to sit in the back of the Endicott classroom and eat prunes. Not having as much to do as at the 1948 conference, I sat with him and shared the package.
Next week McPherson, who was now a vice president, had me on the carpet at Galactic Headquarters. Prunes were out, it appeared, even when shared with a customer - the No Booze Rule came to mind. I pointed out rather plaintively that the whole shemozzle sprang from my 1948 Forum triumph. "Herb," said John,"it isn't enough to bat a thousand in IBM. You must also make no errors." I remembered the poor devil who had rejiggered the WCs at the Watson Lab. It was the other side of the shiny IBM coin.
The next Endicott show was even more impressive to us working stiffs. It had what must have been the greatest percentage of the world's senior computer users ever assembled in one place. I have to justify that: I'm talking about men and women who were actually running big problems on big machines at major centers. There were no big installations to speak of outside the United States. Inside the U.S. the one-off machines that were the features of the non-IBM conferences - SEAC, the von Neumann machines, the ones being built at universities and by big amateurish companies like Raytheon and General Electric - were not running even on sample problems. Remington Rand and its Eckert-Mauchly and ERA offshoots were not yet competing.
There were 107 attenders, including six women. There were papers by a dozen men who already figure in my stories: Hamming, Yowell, Lowe, [-112-] Hastings, Herget, Uffelman, Tukey, Bell, Ramshaw's boss, the new president of ACM - and two women. Five referred to my table-making scheme; I was flattered.
Jack Belzer was there, and Marj Herrick's new boss, and two Canadians, and fifteen IBMers - all professionals, not salesmen. Not until the 701 big shots came together in 1954 was there such a powerhouse of IBM users, but by then there were big non-IBM machines working all over the world.
Which reminds me about conferences outside the U.S. There were problems; the British had a cadre of experts, including men who had been deeply inside the American war effort and had contributed to the new electronics. But as the Forties computing scene developed on both sides of the Atlantic, the people actually doing calculations were using desk calculators and differential analyzers and punched card machines (the British said Hollerith). For every Comrie or Hartree or Southwell we had dozens - and a great deal more equipment for them to use.
There were special cases like Turing, and we now know there were amazing early electronic machines like Colossus hidden away. But by and large, men and women from Britain and the Continent who needed to learn from conferences, came to ours. The first good meetings in England and Australia did not begin until about 1953, and the first truly international one was not held until 1959. Many of us regretted it, especially the scientific types: wanted to see our overseas friends functioning on their home territory. In the end we did, of course, but it took a while.
In Chapter 12 you will encounter
(in order of appearance):
John McPherson 01
Watson Senior 01
IBM printer developments four at once in 1947, and they all were used
BINAC a big come-down after ENIAC, and it ran out of money besides
Eckert/Mauchly approach turned down coolly in Watson Lab library
Jim Rand Rem Rand accepted what IBM had refused, and got the UNIVAC I
Bill Bell 11
Northrop wild-eyed airplane engineers coupled the 603 and the 405
Fenn, Woodbury and Toben asked McPherson to do a better job
The CPC I [Card-Programmed Calculator] with 64 words of extra memory!
Brouwer, Clemence and Eckert if there had been a Nobel for astronomy ...
Dick Bennett 07
John Lentz 10
Yale Observatory lent Eckert a giant measuring engine to automate
Hurd [Cuthbert C. Hurd] came into IBM to start Applied Science
John Sheldon Hurd's brightest new hire, he ran a computing bureau
The Watson Lab 01
Laboratory science IBM began its enormous investment
Ken Clark, Becky Jones and George Samson held the fort for Eckert
Jim Birkenstock an interface for Hurd with WHQ
Watson Junior 11
The von Neumann Constant completion: three years from any given moment
The Korean War T.J. fanned out his troops
The Defense Calculator later the 701, it was the key to IBM futures
Poughkeepsie T.J. decided to make it the center for electronic machines
IBM plant location policy it wanted to be a major force in new towns
Ralph Palmer he ran the 701 project, after working on the SSEC
Jerry Haddad Number Two for Palmer, after doing the 604
Rochester, Astrahan, Buchholz and others more Radiation Lab graduates
The RCA labs Rajchman was trying to make an electronic storage tube
ERA [Engineering Research Associates] had better magnetic drums than IBM
The TPM [Tape Processing Machine] put aside, it matured into the 702
The Test Assembly testing electrostatic storage and Rochester's software
The Kenyon Estate bulging with bright young electronics engineers
Bill McClelland 10
"Wilkes, Wheeler and Gill" it was the only hardcover programming text
Copy machines in that dim, dim, dim past, there weren't any!
Binary cards about half the holes in each column got punched
Octal versus octonary Rochester and I, purists, preferred the latter
Floating point everybody except the cryptologists insisted on it
An estimate of Watson Senior "... I shall not look upon his like again"
[-115-] The Endicott engineers didn't worry much about the conferences being held up and down the United States. For one thing, the senior ones like Lake and Hamilton and Luhn - well, Pete was in Poughkeepsie by 1948 - had all they could handle in their own shops. Each of them had a complete facility: office space, a drafting area, a small machine shop and a fair-sized electronics lab, and an assembly-and-test space. Access was severely limited - officially, only the administrative head of the laboratory, a few WHQ executives like McPherson, and of course T.J. himself, could enter.
In reality a good deal of scuttlebutt was exchanged in the cafeteria, and via salesmen and customers who sneaked in, and over iced tea at the Country Club. But in theory, each inventor and/or project was separate. There was a time [1947] when the same Endicott building held four printer projects. One was developing the vertical typebar idea still further, one was doing a wheel printer for what became the 407, one was doing a horizontal chain type which later went into the 1400s, and one was experimenting with better wire matrix componentry - initially for key punches and interpreters. When Watson had a free day he would tour the various facilities and give his verdict: stop, continue, increase the effort, send it over to the factory. Cynics claimed that the verdict sometimes depended on the shape of the casing or the quality of the paint.
My own view was different, and was reinforced by the fact that all of the printer projects came into production. IBM was sales dominated, and the market was ravenous. Almost anything had potential, and the Blue Suit Brigade could sell it even if it didn't. An engineering or production failure like the 602 could occur, but the lab people and the factory people were very good - and the sales force quickly whitewashed the rare blunder.
[-116-] To shift gears slightly, I have sharp memories of two scenes involving John McPherson as a buffer for T.J. I'm sure he did some of the Endicott tours for Watson and made - well, recommended - decisions, but the two I am remembering were in New York City. Very early in the game, Pres Eckert and John Mauchly wiggled out of their Moore School commitments; the story has been told elsewhere, and frequently. They started a small company and began to build a dumb gadget called BINAC, and what was to become the UNIVAC I. In a way that would become very familiar in the Seventies and Eighties, they ran out of money. And their sponsor died.
They came to IBM. I was not directly involved, but Eckert - Wallace, that is - advised McPherson, and I advised Wallace, negatively. Their application certainly went to T.J., but I never heard the details. Anyhow, there was a meeting in the library of the Watson Lab, and John McPherson said no. The two hopefuls went from that session directly to Jim Rand's yacht in Florida, according to folklore, and signed with Remington Rand. The Old Man didn't have a yacht (he used the Queen Mary), but he had a lot more of everything else than Jim Rand. He made a good decision, as usual.
The second scene was more fun. I've mentioned how clever users like Bill Bell wired up "illegal" plugboards which customer engineers were supposed to report, for extirpation. We were always told it was for patent reasons, which I doubted and still doubt. Well, much greater crimes were being committed elsewhere in California. Observing that IBM connected the older machines together, notably summary punches to tabulators, wild-eyed users in the tin airplane business tried connecting the new high-speed electronic boxes to the more flexible older machines.
The first venture to do useful work was a 603/405 combination. Northrop had one of the space-heaters, which you remember did six-figure multiplications only, but very fast. They ran long plugwires between the simple 603 plugboard and the complex 405 tabulator board, taking advantage of the fact that no electronic signals were involved (they were all down inside the 603, and never got to its plugboard). Their customer engineer, who must have been deeply involved, had to pretend to be shocked by all this, and reported it up through his channels. Orders came back down to, ah, terminate with extreme prejudice.
I was called down to McPherson's office. He had visitors from Northrop: George Fenn (who later gave a paper about the combo at my 1948 Forum), Bill Woodbury (who soon came to work for IBM, and is remembered by old-timers as the planner of a similar machine called the Wooden Wheel), and Greg Toben. They were counter-attacking; instead of discontinuing the 603/405 experiment, they asked IBM to build them a better machine using the forthcoming 604 and an improved 403 tabulator, and to provide additional storage. They wanted a few hundred bytes (we said "positions" in those days); today, hungry customers beg for gigabytes - same hunger, larger appetites!
McPherson looked down his nose at these importunates. They should be happy that IBM wasn't about to pull out its machines entirely, for gross violation of the terms of the rental agreement; as for the suggested new combination, certainly not.
Fenn looked grimly amused. He produced a personal letter from Jack [-117-] Northrop to Mr. Watson which pointed out that Northrop had several important military contracts whose rapid completion was of tremendous importance to the national security, and asked IBM to cooperate. I gathered the letter was to be carried upstairs that very day if McPherson's answer was unfavorable.
John got very red in the face. I waited for the outburst. Fenn And Co. had been pretty crude; also I knew from my aerospace connections that Northrop was a long way down on the totem pole compared to Convair, say, let alone Douglas or Lockheed or Boeing. Problem was, all customers were theoretically sacred - and those like Fenn and Uffelman who threatened to put the bite directly on The Old Man were more sacred than most. The local salesman was supposed to see that the bite was never put; no doubt an Anchorage transfer would be in the Los Angeles office in-basket shortly. Meanwhile ...
Like all IBM offices John's had an easel. As his choler faded he stepped over to his and began drawing boxes. In an hour we had the outline of what was to be called the Model I Card-Programmed Calculator: an electronic computing box, a flexible tabulator to read programs and data from cards and print answers, storage in the tabulator and in extra electromechanical boxes. Linked together by tidy cables rather than with plugwires, needless to say.
It was a remarkable conference, and presaged a very popular system. There was another novel aspect; not only had a customer dragooned IBM into building a new production machine, but it was agreed that documentation, and maintenance training, should be furnished. Small scale perhaps, but better than for the 603 - and far better than for the Aberdeen machines. And as for me, I stored away impressions of how to make IBM respond.
I inherited one of the last of the Model Is when I was hired into General Electric [1952], but I was so interested in graduating immediately to a familiar Model II that I hardly noticed the exact rental. Somewhere around $4,000 a month, I suppose, depending on how many boxes of auxiliary storage you had, and on whether you needed to rent support machines or already had them. That means a system would have sold for $150,000 or $200,000 if the option had existed. And, to put it in today's perspective, both speed and storage were less than those of a good end-of-the-Seventies hand-held programmable calculator, selling for $500!
Fenn described his original lash-up, perhaps a little lavishly, as "comparable in programming technique to current large computer design - a poor man's ENIAC." It programmed a lot easier than the original pre-Clippinger ENIAC; he might better have said "a poor man's SSEC." He could card-program double-accuracy multiplication - decimal, of course, and not floating point - in just over two seconds, and six-figure square roots in one minute; trick wiring on the 603 board was not possible. But he was doing vibration calculations and planning on trying some aerodynamics, in 1948!
About this time an important new figure and a new team came onto the field in IBM. I've said that a post-war sales apparatus was not in place to handle newcomers like Bell and Fenn. I should have moved into the gap. I had a steady flow of importunate visitors at the Watson Lab, and was in demand as a speaker for informal groups, who wanted to get into IBM-style computing.
Sometimes an alert salesman was involved, or the potential customer wanted to bypass a dull IBM local office, but the link to the chain of command [-118-] that produced special gadgets or fast delivery was feeble. I could have strengthened it. I wanted very much to rise in the company, but had not purged myself sufficiently of research and teaching. And my wife Dorothy really hoped I would go back to astronomy, or at least be an Eckert and do some major celestial mechanics on the machines I had such wonderful access to.
Hilleth Thomas was delighted to do his physics, and have a Columbia graduate student or two. Eckert had the IBM countryside spread out before him, and chose to ignore it whenever he could. He wanted to do research, and he did; if there had been a Nobel prize in astronomy he and his confreres Dirk Brouwer at Yale and Gerald Clemence at the Naval Observatory would have won it for the tremendous contributions they made to our exact knowledge of the motion of the Moon and the planets, using the SSEC and later IBM equipment.
An example: Wallace rummaged through post-war IBM and found Richard Bennett, the customer engineer who had made Belzer's Air Almanac master sheets possible back in 1941. He brought Dick to the Watson Lab, partly to help with the Aberdeens when they arrived, but mostly to build a photoelectric reader for the huge glass plates that came out of the Yale astrographic program. The star images to be measured were delicate little things by 1946 standards, and John Lentz was hard put to it to devise the electronics. The gigantic measuring engine, with precision screws two inches in diameter, was loaned by Yale; John put a rotating beamsplitter on the reading microscope, picked off X and Y signals, and with them ran delicate servo motors that improved the settings.
The initial setting was made from a punched card deck, one star to a card, and the final settings were punched on the same card. Further work on standard IBM machines at Yale produced the star coordinates for publication. This monster filled a whole room in the first basement at the Lab - I've neglected to say that "huge glass plates" meant about thirty inches square and a third of an inch thick. It was one of the exhibits on T.J.'s tour of the Watson Lab, and I suspect it confirmed his realization that Eckert was an astronomer first and an asset to IBM sales efforts very far down the list.
The new player appeared. He was Dr. Cuthbert C. Hurd, who had come to my 1948 Endicott Forum from Oak Ridge. He had nothing to do with Uffelman; was head of a small statistical group over on the Union Carbide side, and had access to IBM equipment there. Now IBM never hired anyone away from a customer - or so they told people who wanted to be hired away! But men like Eckert and Hurd popped up through special trapdoors.
Hurd was brought in to head a brand new Applied Science Department, which meant that while he did not capture the Watson Lab or the SSEC, he did take over the Endicott conferences. He and the group of youngsters he quickly put together were exactly what was needed to handle the special customers and the special machines like the CPC, as we had begun to call the Card-Programmed Calculator. He himself didn't know much about hotshot computing, but the men and women he acquired from Seeber certainly did, and his new hires included one of Thomas' brightest students, John Sheldon.
Soon that triage that Eckert and I had performed for three years was being done by Hurd and his staff, and very differently. Problems went to the SSEC [-119-] if they were very large, and either had good public relations value for IBM or were submitted by a rich sponsor like GE; the value to world science which had moved Eckert and Thomas, and to a lesser extent Rex Seeber and me, no longer counted heavily.
Customers or potential customers certified by their Friendly Local Sales Representative were encouraged to order CPCs or whatever, and Hurd's senior people would help. If the customer had only a single problem, the Applied Science Department soon had a technical service arm set up on 56th Street to handle the business; it was run at first by Sheldon, and used the youngsters from the SSEC. If the prospects were patted down and found to have no money, the glories of IBM's Watson Laboratory were paraded. And the fairly large cut of problems that formerly was allowed to die in a quiet corner, dwindled to almost nothing. All, or almost all, was grist if it came anywhere near the Applied Science mill.
I was green with envy on Mondays, Wednesdays and Fridays and livid on Tuesdays and Thursdays. Indeed, it was very different from Wallace's lunar theory, and great lunches at the Faculty Club. But it was clearly needed. There were hundreds, and soon thousands, of computing centers waiting to be born. The wave of the future embarrassed Eckert - probably it is not too cruel to say, much as I admired him then and remember him warmly now, that the wave annoyed him. Celestial mechanics moves very slowly; it is ironic that Wallace, wanting to speed it up a little, helped open the floodgates for this enormous torrent of change. He was too sturdy a rock to be swept away, but the flood washed over him and left the little 116th Street Watson Laboratory behind.
After I left [1951] there was a quiet time. Then another force arose in IBM; call it laboratory science - physics and chemistry and such. It also was pure science, but not Pure Science. In the Sixties and Seventies IBM made earthshaking investments: built big research laboratories in several countries. Before that, and especially before Yorktown Heights was planned, the original Watson Lab was moved to a much larger building at 612 West 115th Street, and the instruction at Columbia and for customers, in machine methods and numerical analysis and so on, was subordinated to bench science.
Havens had moved out to nearby quarters to build his NORC. Eckert and his little band of assistants - Becky Jones, Ken Clark, George Samson - were in titular charge. But the novelty of computing was gone, and when the SSEC was torn out of WHQ to make way for a production machine ten times as powerful [1952], the old Watson Lab era ended. The building on 115th Street drew scientists who did not want to work in Westchester, but it was shut down when the post-Watson-Senior rigidities of a giant company dictated. Wallace retired in 1967, kept on with his research of course, and died in 1971. He and Comrie had revolutionized user perspectives, just as electronics had revolutionized the hardware. I remember him fondly.
About the new player: Hurd was a natural executive. He hired excellent men and women, and he delegated. He was an empire builder - the exact opposite of Wallace Eckert. He mixed easily with the McPhersons and the Birkenstocks, and with Tom Watson Junior. He was great with the customers, except for a few real toughies. I've said in a hundred speeches that the supreme art of IBM salesmanship is to convince the customer that the dumb [-120-] hardware and software the salesman is forced to offer him is precisely what he needs; Cuthbert was a master of the art. And he didn't wear a beard!
He was not himself a computer user. He certainly was not a machine designer. He didn't see the wilder shores of the future. But he saw where IBM could go next, and which customers needed more advanced equipment. And he worked far better with the planners and engineers than I had, partly because he was down inside World Headquarters where it all came together, and partly because he had no component, no architecture, no programming method he wanted to push. I wanted - and still want, after more than thirty years - to invent a solution, to convince my peers of its value, to lead the troops to victory. Cuthbert was looking for battles IBM could win - and he was careful not to get badly shot up while looking, or on the battlefield. I tried to warn the airplane customers that floating point was dangerous; Cuthbert argued inside the company for more CPC production.
Out in the world, there was competition. Not much; we were just emerging from the years of the von Neumann Constant: from any given moment to the delivery of a finished computer is three years. All the real work - what in the Eighties we called number-crunching - was being done on 604s and CPCs, both in heavy demand (and the latter in process of revision). But there were orders for Remington Rand machines; none of the big ones had been delivered yet, and the one-off machines like BINAC were dreadful, but T.J. had been alerted.
And there was fighting again, in Korea [1950]. The Old Man had lost a little of his vigor, but none of his patriotism; he started a Military Products Division, and his troops fanned out. The customers were eager. The electronics that had begun at MIT and surfaced in the 604 was blossoming in several places in the company. As the American difficulties in Korea mounted, andWashington began to pump out funding, Hurd and his people took to the road.
They went to the lead customers - the ones who had several CPCs, and were clamoring for better ones. I knew about this from the men they visited in Los Angeles and Schenectady; Hurd was careful not to involve me in the calls. General Electric asked my advice, and I told them to get in line. The airplane boys were jostling for position. Hurd soon had fifty expressions of serious interest, half of them backed by letters of intent. No machine specification. No price. No delivery date. But lots of customers!
Euphemisms abounded, then as now. IBM would build a Defense Calculator, not a Win-In-Korea Calculator, and certainly not a Knock-Out-Univac Calculator. Tom Watson Junior was the key executive; he could go to his father and speak frankly: talk about General Macarthur's problems, and how Remington Rand needed curbing. He did, and came back with a big dollop of money from The Old Man's shoebox. Published accounts say $3 million, but that was just for starters; the internal gossip said $25 million, and was current before the first machines were delivered. Of course both figures may be accurate; remember I said in the chapter about the SSEC that the indirect costs of that machine were never included, or even calculated.
Years later, when IBM was forced to begin selling its machines [1956], it appeared that the selling price was around forty times the monthly rental. The first internal guesses at the monthly rent for the Defense Calculator were $8,000 or so; the amount I paid for Number Six was $15,000; the customers [-121-] who stormed the citadel in 1954 to order 704s were told the Defense Calculator rental should have been $30,000 a month, and would still not have turned a profit. That points to $1.2 million as the sale price and also the cost, or $24 million for the program.
While Eckert was recruiting the three Watson Lab electronics engineers from the MIT Radiation Lab, more conventional methods were bringing others from Cambridge and Los Alamos and such, into major parts of IBM. Watson had decided to develop Poughkeepsie, which was making key punches and electric typewriters, as the company center for electronics development and manufacturing. This was partly because there had been a good deal of munitions production there during the war (Endicott was needed to pump out the larger punched card machinery), and there was available space.
Also T.J. had a philosophy about locations. He liked to go into towns like Poughkeepsie and San Jose and Lexington, Kentucky where IBM would be a major force. Later I was to watch GE do the exact opposite; it had had its bellyful of Schenectadys and Lynns and Eries, and wanted to go to places where it would not yo-yo the community up and down. The Old Man enjoyed powering the upswings, and had ordered that there be no downswings.
When I began going up to the first engineering projects in Poughkeepsie from the Watson Lab, to visit Pete Luhn and also to give informal courses in numerical analysis, the taxi drivers who brought me from the New York Central station invariably asked me how they could get jobs with "the IBM." Which reminds me, there was a clutch of employees who had to go up to Endicott frequently, on the Lackawanna Railroad or via Robinson Airlines, who thought it great that IBM was expanding into Poughkeepsie, because the train service was so much better.
The bright young men there worked for Ralph Palmer, who had been a team leader on the SSEC under Hamilton. The Number Two was Jerry Haddad - Jerrier A., to be precise - who had done the 604. The systems architect was Nat Rochester, who really had wanted to stay near Boston after his Radiation Lab days, but worried about its attractiveness as a nuclear target; in those times no one dreamed that Poughkeepsie would some day become high priority. Then there were Mort Astrahan and Buck Buchholz, I think also from MIT, and Clarence Frizzel from Old IBM, and Max Femmer, and a dozen others.
Most of them were clustered around a project called the Tape-Processing Machine, and a test bed for electrostatic storage called the Test Assembly. You have to understand that no one had today's kind of magnetic tapes working yet - Sam Alexander, who was building SEAC, the Standards Eastern Automatic Calculator, in Washington, was trying to use magnetic wire, and Mauchly had talked about metal tape for what was to be the UNIVAC I, as far back as 1948.
In spite of several years of effort, RCA had not been able to make a working tool out of storage on the face of a special cathode ray tube. Successes were just being reported from a different attempt at MIT's WHIRLWIND. And IBM was behind its smaller competitors like ERA in the art of magnetic drums. So the company was not even sure of the componentry for the Defense Calculator, let alone the system structure.
[-122-] The team and the Test Assembly were in a mansion away from the river and the main plant. It was part of the Kenyon Estate, and IBM had bought the property in order to have room for a new laboratory and more engineering buildings. After the lab went up across the road, the big house was remodelled as IBM guest quarters. But it was bulging at the seams with bright young electronics engineers in 1951. Many of them showed up at a plant classroom to hear me talk about table lookup and the convergence of square root iterations, and my kooky way of looking at number systems like biquinary and signed ternary.
Hurd had assigned Bill McClelland, whom he had recruited from the SSEC the year before, and two or three other Applied Science types, to help with programming development. From reading the first hard-cover book on programming techniques, an English contribution everybody called "Wilkes, Wheeler and Gill," from informal exchanges at conferences, and from the aquifer of brand new ideas that was flowing from MIT and Columbia and the Moore School, Rochester and his crew knew that programming in machine language, and with absolute addresses, was obsolete. The next level of sophistication was being established. The New York contingent pitched in.
I was in "wait" mode - very unusual in those torrential times. In fact, it was a new thing for me; I had whizzed from my thesis to the Naval Observatory to wartime optics to the Watson Lab without a break. I could see that the Defense Calculators were going to go to customers who were mostly very different from the scientists we had seen at the SSEC and Columbia. The Applied Science youngsters understood them very well, but in the role of camp followers of The Blue Suit Brigade, not as equals. That meant, because of the gentility expected of all IBM employees, they weren't practiced at saying no, let alone "Hell no!" I offered to simulate a Fenn-type customer; with a faint shudder, I was accepted. Not a formal reassignment, just frequent several-day visits.
I had to stay at the Nelson House, a hotel whose sole distinction was that the swizzle sticks in the Regatta Room were shaped like sculling oars. Many, ah, software conferences were held there, the Even After Work clause of the No Booze Rule having been relaxed, especially if you had been back to work after supper. At the Kenyon Estate, I shared a tiny office converted from a maid's room, and kept my papers in a file cabinet in the bathtub next door. I made a few design contributions, notably to an argument about whether to retain both positive and negative zero, in the pure-binary number representation (simpler souls wanted to sterilize the minus). Somewhere about this time I also made a major selling-type point, that the IDLE lights on many machines be replaced by READY lights - and not red ones, either.
Nat Rochester had hand-written an operating manual, reproduced by Ozalid machine. It described the operations he hoped Palmer would let Haddad order the engineers to build into the machine. The next version was typed, and had lost most of the TPM, Tape-Processing Machine, operations. This came easy, since these had been business oriented - for sorting, and so on - and the engineers felt a little at sea. The Defense Calculator was to have a fixed length 36-bit word, and extra precision and floating point, dear to the hearts of nuclear physicists and aerodynamics experts, were to be in sub-programs. I helped with that too.
[-123-] Today's youngsters really can't picture how we worked. Not because of the primitive software tools, not because there was no hardware yet to try prototypes programs on, not because nobody had invented terminals - but because there were no copy machines! And if somebody writes a story like this at the turn of the century, I suppose the equivalent remark will be made about word processors. As I said above, Ozalids!
When a gang went out to eat together - there was no dating, in spite of the tempting array of Vassar girls just up the street - each man would keep a mental total of his items. When the bill came, you added tax but not tip and put down your share. If the pile added up perfectly, which it almost always did, you added whatever tip you wanted. If it did not, and postmortem revealed a culprit, he had to pay the total tip. I was the authority on rounding errors. Usually it turned out the waitress had goofed. An innocent pastime - and it took your mind off the food.
There were concentric rings of programming. Right around the machine - or the machine design, at the time I was involved - were small programs written in pure binary, to persuade the ALU (arithmetic and logical unit; the heart of what we now call the CPU) to eat punched cards and understand them. Such bootstrap operations are stored in read-only chips today in millions of small machines, but electrostatic memory was far too small to waste in such fashion in 1951, it woke up empty every day, and it twitched (in the Test Assembly anyhow) when somebody turned on an air conditioner upstairs.
The plan was to start up with a blank memory, by flipping console switches read one card into a corner of memory, transfer control to the first word in that corner, and go. I suppose Nat wrote the first one; the champion of the art was to be FEJ, Floyd E. Johnston, whose meticulous handwork in 1952 was shipped with the production machines. Those so-called binary cards were a far cry from the ones that went into regular punched card machines, or even the CPCs. The Defense Calculator regarded any card as 960 ones [punched holes] and zeros, which on the startup cards were read as 24 binary words plus ignored patterns. What the inner ring of programs did they did with sets of those 24 words, understood as commands. You could tell binary cards easily by the very large number of punches: an average of six per column, where one or two per column was ordinary punched card code.
The next ring of programs got you out of pure binary into octal. Nat and I were purists, and tried to get the gang to say octonary, as in binary, ternary, biquinary (the abacus code). Alas, "decimal" filtered down past nonary and won the day. With octal input, you could not only write programs usefully on coding sheets, which - shades of Maxwell - were being devised and revised and discarded almost daily, but you could punch them on cards with an ordinary keypunch, and reproduce and sort and collate and print with IBM's standard machinery. True to the shoemaker tradition, the crew was very short of such gear, and I invited them all down to my smoothly-running shop on 116th Street; no takers!
One of the binary programs in the first ring taught the Defense Calculator to understand octal, of course. And a small pack of octal program cards taught it to work in assembly language, where the code for operations was [-124-] alphabetic/mnemonic and the locations in memory were relative, and given in decimal. Other second-ring octal programs fired up the tapes and drums, giving faster streams of data in and out. Still others operated the card punch and the line printer, which were also being architected by Rochester And Co. and designed for production by some of Haddad's troops by modifying standard Endicott gear.
To break into the programming story for just a paragraph, this was one of the IBM strengths. While Pres Eckert and his crew were struggling to produce the UNIVAC I (and with a considerable head start) they were having to design everything; not just tape drives, but input typewriters and such. They got no advantage from round-hole cards, partly because there was no Watsonian discipline to force cooperation from the curmudgeons in Norwalk, and partly because the inflexibility of the mechanical Rem Rand equipment made interaction with fancy electronics impossible. The IBM machines, even the older ones, were controlled electrically; sure, you couldn't hook up a sorter to a tape drive, but you certainly could use almost all of the printer unit of a tabulator, and pretty up the results very considerably by retaining the plugboard and the paper controls.
When cards which were punched in octal or in assembly language became re-usable - that is, needed to be run several times before being altered again - a "permanent" first-ring binary card program converted them to binary-card form. These read in about ten times faster than the originals, and all production software was later shipped that way [1953].
Quite a few second-ring programs were working when I began coming to the Kenyon lab. Bill McClelland and the Applied Science men were using assembly language to try out stuff on the Test Assembly (no relation). I wanted to try a few things in the next ring of programming: in my case, multiple precision (I seem to remember someone had already undertaken simple single-precision floating point). These programs would be too big to run on the Test Assembly, which meant they were paper studies. The machine would be modified a hundred times before any of them could hope to be tried. No matter; it showed Rochester where there were lumps in his op codes, and he was not only willing but anxious to be shown. It was understood that everything, including all of the first-ring and second-ring programs, would have to be rewritten for production gear, and modified later besides; the euphemism "maintenance" would not be invented for nearly fifteen years, but the idea was already accepted.
The 36-bit word size was probably unfortunate, and it most certainly stretched the abilities of customer problem analysts and programming experts later. When you chopped off a piece for the exponent (and sign) in floating point - which in spite of my Jeremiads all the customers except maybe the cryptologists insisted on using - there wasn't really enough left for accurate matrix calculations. Or astronomy, but I could already see before the design was fully roughed out that not much of Wallace's stuff was ever going to run on a Defense Calculator! Anyhow, it was too late; looking back from the Eighties, I guessed 48 bits would have been optimal; the 360/370 experience showed us that 32 is not enough, and 64 would have been wasteful.
There was an important difference between the programming crew (and [-125-] me, the simulated hardhat computing customer), the eager MIT electronics group, the Applied Science and Pure Science youngsters back in New York, and all the rest of the IBM apparatus. Every single one of us assumed there was infinite expansion ahead. I saw small waves, and then large waves, and then huge foaming torrents of new science and new technology: science fiction coming to life. The hardware group was perhaps a little more business minded: they saw waves and waves of new machines.
At 590 and in Endicott, and even in the factory reaches of Poughkeepsie, it was not at all obvious. Most of the manufacturing people, most of the planners and factory schedulers like George Richter, most of the salesmen - and, not surprisingly, almost all of those subterranean beancounters - thought the office machines were the way to go. IBM had always followed T.J. into special ventures like the Harvard ASCC with pride, but few of the sales force or the top brass ever thought such stuff could be product lines like time clocks or typewriters, let alone supplement or supplant the punched card equipment.
There was no resistance to electronics, except perhaps from a few discomfited customer engineers; the 604 was a salesman's dream. The CPCs were a different story; if an installation needed help, one of the Applied Science men or women had to be called in. The company had always had a small fleet of systems "engineers" - frequently women, during WW II - who would wire difficult plugboards. But they were part of the local or regional sales teams, and hence floated on the commercial currents. Even after Hurd began to put Applied Science reps out in the field (starting in Los Angeles in 1950 or 1951), they marched to a different drum than the systems bunch, and their services tended to be demanded directly from WHQ by knowledgeable customers.
Everything turned on Mr. Watson. When he said the CPC was a good thing (I suppose he must have), it flourished. If he had said no to the Defense Calculator, as he might well have done if it had been called the Scientific Computer or some such, IBM would not have surged over UNIVAC and ERA and the small specialized companies that were springing up. In the end it would have been a major player, as Rem Rand and National Cash were. But it would never have towered over the information world as it did into the Nineties if it had lingered in those profitable punched cards.
The stories about how the company got into commercial production of computers tend to emphasize how heroic the Hurds and the Birkenstocks were, to have persuaded Tom Junior and Tom Senior to make that $25 million dollar commitment in 1950. For me, though, the hero is The Old Man himself: 76, an imperial figure secure on his throne, with an heir apparent trained and ready. How many men in that position would have taken such a confident stride into the unknown? And it wasn't luck; his track record at National Cash, his insights in the early IBM days, his backing of research in the company and science and the arts outside, his international successes tell us it was something much deeper.
In Chapter 13 you will encounter
(in order of appearance):
Williams and Kilburn two remarkable engineers at Manchester University
Electrostatic memory the Defense Calculator was to have Williams tubes
Max Herzberger Kodak Research Lab expert on optical aberration theory
"Real Research" 05
Hurd 12
The Blue Suit Brigade advancement in IBM was through the sales force
John Sheldon 12
The SSEC 01
A Washington Technical Service Bureau a special market that I knew well
John McPherson 01
The CPC II the hardware was under way, but no one had started software
Ellie 10
Eric Hankam 09
Red LaMotte all Washington was the fiefdom of this tough Old Watsonian
Two 1972 luncheons at the Savoy Univac had me; IBM had the Duke of Kent
Grosch's Law a 1950 afternoon's work and four decades of dividends
Computer speed about all we gave the customer for his money, early on
Ken Knight did a Stanford doctoral thesis featuring my Law
Scientific machines for business processing Lyons in London pioneered
Access to top management at Northrop and Lyons but not at Metropolitan
MONY [Mutual of New York] their building had a great weather display
Watson Junior 11
Horace Post even when Tom was only my salesman, he had a secretary
John Phillips president of IBM when I first visited the Homestead
The Watson Suite the most lavish bathroom I had ever seen
MONY directors my proposal got all the way to the top, worst luck!
Wallace Eckert 01
Frank Hamilton 01
Ernie Hughes helping Frank design what turned out to be the IBM 650
Dorothy 01
Mary Noble Smith 08
The UNIVAC I the first was for Census and the second, for the Pentagon
Don Gamel "a smoothie from the Blue Suit Brigade" who ran Federal
Scottie he had spare space on 19th Street until I showed up
Bachrach same day service for IBM mug shots
A small row house near Q Street, and a big step up from our WW II place
[-127-] I left the Regatta Room and the rest of the Poughkeepsie ambience reluctantly. The whole bunch, programmers and designers and engineers, and even the few executives who impinged, were working together in great style. We all felt certain a huge success was in sight; the very speed of the current - yes, Dorothy and I were still foldboating - was carrying us past the rocks in the stream.
As one example on the hardware side, it was obvious that the simple storage tube circuitry that went into the Test Assembly was out of the question for the production machine. But behind the arras somebody was dickering with two remarkable Mancunians, F.C. Williams and T.M. Kilburn, and when the Defense Calculator emerged as the 701, the Williams tube technology had been mysteriously incorporated.
My waiting period was over, however, and I collected my much-coveted machine from a swarm of hungry IBM salesmen and set out for Washington. To explain the mysteries in that sentence, I have to go back to "green with envy" in the last chapter, and in connection with that to a changing perception of myself. From childhood I had intended to be a researcher. I had married a woman who wanted me to be a researcher. I had been steered by Maxwell, had admired Comrie and Wallace Eckert and two of the three Great Vons. I was surrounded with the apparatus of intellectation: Columbia, the Watson Lab, the SSEC and its probable successors. I had the credentials, the experience, the opportunity. And I was not doing well.
I had worked up Jupiter VIII, blessings on it - and others were building on my small success. I had done the Kleine Planeten, and the little coterie of orbit computers was impressed. I had done a seminal investigation of a famous lens system on the SSEC, told dozens of lens designers [-128-] where their future lay, and turned the results over to Max Herzberger of Kodak Research to exploit.
I was the father of a family of mathematical tables which, while themselves only of temporary importance, could lead to years of approximation theory research and publication, well suited to the rather circumscribed resources at the Watson Lab. And I had a very large and rapidly growing acquaintance among the men and women around the world doing big computations - in fact, I was advising many of them, and enjoying the process.
Note clues in those paragraphs: "told dozens of lens designers" and "advising many of them." I had seen for some time that I was happiest when talking to other people and advising them how to tackle problems, how to set up computing facilities, how to find or train good youngsters. Also I was impatient with the way Eckert turned away from manifold opportunities, more frequent than mine, to do the same - and at the same time to steer IBM through the rapids of technological change. He had done well, but I would have liked him to have done a great deal more.
I was indeed envious of the rapid rise of Cuthbert Hurd. He had taken away "my" Endicott conferences with a twitch of his wrist, hired the best men at the SSEC for Applied Science, cut me out of the Defense Calculator fishing expeditions and from most contact with WHQ. Clearly I had not been ready to do what he was so suavely doing. Not so clearly, but increasingly, I saw I was not likely to be a von Neumann or a Feynman; even in the slower currents of Wallace's celestial mechanics, it didn't look as though I was world-class.
I decided to leave the eddies of numerical analysis and celestial mechanics behind, and paddle up stream toward management. I wanted to stay in IBM; already, although it was barely a billion-dollar corporation, I was completely convinced of its importance in technology, and in the world of affairs. But while I had known for several years that the usual path for advancement in IBM was through sales, I did not want to join the Blue Suit Brigade. I stood to attention very poorly, as my tiny confrontation with Watson Senior on 116th Street showed, and while I had let McPherson chew me out over Tukey's box of prunes, it had rankled for weeks.
I had been managing - well, directing, I guess - a service function at the Watson Lab. I did not have direct hire-and-fire capability, and there were no profits to be made. And, without explicit profit, there was little opportunity for a financial payoff. I could see the large salaries and commissions on every side, and while I was not yet a Big Spender, I had definite interests in becoming one - common in New York City.
John Sheldon and his group were turning away business on 56th Street; the SSEC obviously could be a major money-maker if Eckert would turn a blind eye; the market for computation was booming. Outside IBM, there was the Comrie venture in London, and Bill Bell's Telecomputing shop in Burbank. I decided to start a technical service bureau for IBM, but not under the direct control of the formidable Cuthbert. Location was quite obvious: Washington. Dorothy and I knew it well. IBM was strong down there. There was a huge market for technical and scientific calculation in and near government, and the Feds were moving too slowly to satisfy their needs internally. And although I did not want to be called a salesman, I knew enough about the art to [-129-] realize you had to go out after the customers, and repeatedly, to generate business. Sitting snugly at Columbia or smugly at the SSEC and waiting for government types to beg for your services was not the way to go.
Hurd was shipping work back to his shop as he encountered it, but the latter was already near capacity, and when he wanted to expand he ran into space problems - and into the sales people, who wanted to rent whole CPCs, not just cream off single jobs which would reduce the pressure to start or expand a CPC installation. Also, not a minor point, the big commissions came from sales (that is, rentals). I did not understand the commission plan, but I resolved early on to make sure my new venture paid salesmen handsomely, whether Cuthbert's did or not.
I wrote up some notes, which unfortunately do not seem to have survived, and sent them off to John McPherson. I emphasized the unfulfilled opportunity, especially in Washington, and pointed to the advent of the new model of the Card-Programmed Calculator, CPC II, which was under design at the moment. I volunteered to provide a sophisticated private test of the machine, and suggested that if IBM would give me the prototype, I could not only begin using it profitably (unlike the Watson Lab), but could do so in advance of the, ah, documentation.
Documentation - which was of course not the term I used in early 1950 - was to be not just a simple operations manual, but a description of the tremendously complex plugboards that would have to be wired for the two component machines and shipped to the customers at each installation. I pointed out that no provision had been made in what I knew of the engineering plan, to have these boards laid out and wired, let alone described. I said that the men and women at the SSEC were not interested in stepping down to the CPC (and hinted that the Applied Science activity had kidnaped the likely ones anyhow). I noted that the 56th Street facility which had absorbed some of them was away over its head with profitable business, did not have the floor space to put in an additional machine and play games with it, and could not afford to give up its CPC I.
There were three or four great Pure Science employees at the Watson Lab, I said. Some of them would not want to leave New York - I was thinking wistfully of Ellie Krawitz (now Ellie Kolchin, since the marriage had been happily revealed when IBM's baleful rule had been abrogated). One of them, Eric Hankam, was vital to continued instruction of IBM customers and prospects. But I could easily persuade one or two others to come with me if authorized, and with their help could do the general-purpose boards and write them up - while waiting for a full customer load.
Oh, I was persuasive. And it was all true. Problems were, first, that a dozen CPC I customers were lined up for early delivery of one or more IIs (more than a hundred were delivered before the 701 and especially the 650 shouldered them aside). Second, I was intruding on Hurd's rapidly expanding and well-thought-of department. I had not said I would not work under him, but the tone was obvious; I talked about Pure Science for a starter, and later a new department of computing services to manage a network of similar outfits. I intended to steal away John Sheldon And Co., although I knew better than to say so.
[-130-] And third, Washington, then the fiefdom of a tough Watsonian veteran named Red [Louis H.] LaMotte, showed no signs of wanting such a function, let alone one run by a bearded nonconformist who had once publicly disagreed with The Old Man - yes, such matters soon got around. LaMotte had been king of IBM Federal for some years, and was reputed to go horseback riding most mornings with the Quartermaster General and such. A year or two later he was made Vice President of Sales for the whole IBM - arguably the Number Three job in the company. He retired in the Sixties, but still served on special IBM committees; I saw him last in London two decades ago, and he looked as tough as ever, although more gray than red.
I'll never remember to tell that story in the proper place, so I'll do it here. Sperry Univac was having a fancy customer luncheon at the Savoy [1972], and had shipped me in from the Bureau of Standards to be the guest speaker. IBM was also having a luncheon for their customers, also at the Savoy, also that day. With several company big shots in attendance, they had managed to lay on the Duke of Kent. But just two or three days before, the Duke of Windsor had died, so Kent had to cancel.
My Univac host was delighted - with me too, I think - and sent me off afterwards in his own car. As the chauffeur turned up the road from the Embankment, I saw a straggle of familiar senior IBM executives toiling uphill on foot; leading the pack, straight-backed as ever, Red LaMotte! I stopped my car, leapt out and had a mad embrace with the old devil. It was almost exactly twenty years since he had fired me.
Back in 1950, nothing seemed to happen. No one was angry. No one said no, not even McPherson, who was accustomed to saying no to me. I had told Eckert very little about my plan, just that I was trying to "work out a better way" for the company to handle that kind of work. I couldn't ask him to poke at the system, nor did I think it likely he would be effective. I decided to look at the economics of the service business a little more closely.
One afternoon, sitting comfortably next to my trusty Marchant, I began to lay out some ballpark figures on machine costs. I had the rental prices of the CPC I and the 604, and a pretty good idea of what the rental of the CPC II was going to be. I had the rentals of the 602A, the 602, the 601 for comparison. Making very rough estimates indeed, I converted what I knew of costs of the SSEC, the ASCC at Harvard, and the ENIAC, to monthly figures comparable to IBM rentals. Going further afield, I added unfinished machines like BINAC and the MIT WHIRLWIND, and SEAC in Washington. I had heard little rumors about our Defense Calculator - this was almost a year before I showed up at the Kenyon mansion - and I put in a number based on $10,000 a month rental, later revised to $8,000, and then to $15,000. Two years later I had added guesses at the 650 rental, and about UNIVAC I.
"Well," I thought, "what do we get for all this"? What these gadgets really had to offer was speed; people were still cheap in 1950, and Comrie had told me the economical way for his girls to do a multiplication was on a Brunsviga, punching the result on a card to get back into Hollerith mode (as compared to low utilization on a 601, he explained). Speed for the kind of work I did depended on multiply time; there were only a few divide operations by comparison, and additions and subtractions were much faster. The kind of transfers of control that [-131-] are so important (although even today not very time-consuming) in sophisticated programs were free on four-address machines like the SSEC, subsumed in the plugboard wiring on standard machines, and very fast on WHIRLWIND and that ilk. So I used multiply speed as my measure of performance.
I plotted this all up on a casual piece of log paper, cost versus speed. It looked sort of linear. I extended the baseline: added estimates for desk calculator operation, electric and hand-cranked, and for logarithmic work, and even for Crelle's tables and pencil-and-paper. It still looked linear. I was about to draw a line on the sheet and go back to other work, when I noticed that the probable slope was about one-half. I drew a line with precisely that slope, and it fitted the dozen or so points reasonably well. "Ah," said I,"economy is as the square root of the speed"! Grosch's Law was born.
What I had been looking for was a tool to price computations in a technical service bureau. In order to be practical, especially with lower-end techniques like Crelle, I had added in human costs and computer room inefficiencies. Thus, I got four cents as the cost of a multiplication on my Marchant, assuming a hundred operations an hour, $350 a month for a diligent clerk (1950!), and reasonable benefits and overhead and inefficiencies. My somewhat optimistic guess at the same operation on the Defense-Calculator-to-be was a hundredth of a cent - I assumed good practice would avoid floating point. Wrong!
I went around for several years telling audiences of all sorts, from the popular (Rotary) to the concerned but ill-informed (banking) to the professional, about this observation. When I published it in 1953 I had enough chutzpah to call it a law. It got a little mangled in repetition, and especially when taken up by others. At that time product planners were hungry for guidance; they had new drums and printers and tape drives ready for announcement as well as complete systems, a short list at best of competitive pricing - and IBM was still only renting - and optimistic stories from the engineers about speeds. So Grosch's Law got used as a tool to derive prices from those speed predictions, for computers and for peripherals. A tape drive four times as fast as last year's? Double the rental!
Moreover, when arguments occurred about whether to use the law in a fresh situation, some innocent would plot up the newest announcements on his copy of my curve and marvel at how well it fell on the half-slope line. Confidence in my insight built up - three cheers for self-fulfilling prophecies! In the Sixties Ken Knight did a careful doctoral thesis on performance calculations at Stanford, and hung medals all over The Law. Even in the Eighties, computer science professors who wouldn't know me if I came to their door with a begging bowl, wrote complex papers about how the law holds within classes of equipment, but not any longer across class boundaries (they're wrong; it does).
I loved it - still love it. There have been esoteric attempts to explain the law; there is certainly something down inside the relationship, although nothing deep-deep-deep like Newton or Shannon had. I do an explanation blithely: when the tall ships sailed in the China Trade, the young apprentices had to be taught to go up into the rigging. If they were too frightened, and hung on with both hands, the sails didn't get worked. If they were too eager, [-132-] and took in or let out canvas with both hands, a gust of wind would drop them to the deck or into the sea. So they were told, "One hand for yourself and one for the ship."
My theory is that when you give a programmer or a planner a big new dose of speed or memory capacity, he uses some of it to play little games: to adopt a fancier language or a more wasteful operating system, or to write an internal package that would otherwise be bought outside. And what is left of that speed or capacity he puts at the disposal of his employer; sail does indeed get worked. With friendly audiences I usually concluded with a mildly scatological formulation: "No matter how clever the hardware boys are, the software boys will piss it away"!
That accounts for only an afternoon or two while I was waiting for a response from 590. I had other irons in the fire. I had observed a few intrepid souls begin to go the consulting route, usually after having advised their original conservative managements to order a computer from Remington Rand, as Ed Berkeley had done at Prudential, or to try to use a scientific machine like the CPC for a business task.
Using a scientific machine was perfectly feasible, and the advice was taken at Lockheed and, miraculously, at Lyons Tea Houses in England. In very early 1954 the Naval Aviation Supply Office in Philadelphia put in a 701 to do inventory control, well ahead of the business-oriented Univac that General Electric installed in Louisville to do payroll. The stuff that statisticians or underwriters did in insurance companies was clearly understood and documented, for instance - and simplicity itself compared to Los Alamos or Hilleth Thomas physics, or tin airplane structures calculations. What was lacking in such companies was the kind of channel up to top management that George Fenn and his buddies had obviously found in Northrop, or the mysterious Dr. Thompson had opened in Lyons.
I decided that if I had to leave IBM to start a new venture, I would try the consultant route first, and hope either to convert it to a solid operation or use it as a springboard to someone else's installation. I made a pass at Metropolitan, because of its size, and drew a blank; there was no channel upstairs yet, but a Cerberus type was lying in wait, to pick off intruders who attempted to open one. Next I tried MONY, Mutual of New York, mostly because I liked the imagination they showed in building their new headquarters (... "flashing white means snow").
I met with an assistant vice president who was interested in computers. Yes, he would be glad to look at a proposal, and to carry it further up in management. I went back to the apartment and began writing feverishly - not that I was so anxious to leave IBM, or consult for MONY, but because something was at last moving. My idea was to make an investigation of available equipment that could be matched to appropriate parts of insurance company functions - I was careful to say it was too soon to dive deeply into the policy files or even customer billing, though that would surely follow in a few years. I would recommend either a complete system like the IBM CPC, or assembling components from the various small companies that were offering drums and arithmetic units and the like. I took the slim document down to my contact.
[-133-] About two weeks later I got a call from Tom Junior's male secretary, whom I knew slightly. I was wanted on the 16th floor. Tom was by this time executive vice president, and being sharpened up for the IBM presidency whenever T.J. got around to telling John Phillips, the incumbent, that he should retire. I was curious. The secretary, Horace Post, was quite frank: no, he didn't know what the boss wanted, but he was not happy with me.
Thomas John Watson Junior was not a stranger to me. He had just turned 36, four years older than I, and had not yet developed the lines in his handsome face that would soon appear. He was a big man, like his father - two or three inches over six feet. He was an expert pilot, and had had a genuinely distinguished war record, with a good deal of dangerous flying into places like Moscow. Although I did not know it until later, he was a devoted yachtsman - no, that's too mild; he was an expert and daring sailor. In later years he had been up in the Davis Strait and along Greenland, some of the most dangerous water on earth.
When he got back to IBM [1946] he was assigned a territory in the New York City office, and it included the Watson Laboratory. Yes, during the period I was finishing up the Los Alamos work at the 116th Street building, he was my salesman. Didn't last long; he got promoted right after making his first sales quota, and attending the 1946 Hundred Percent Club. But for a while, if I needed ten boxes of punched cards for the shop, I would call his number and place my order.
There were two differences compared to my previous salesman. First of all, I usually got Horace. Tom indeed made genuine sales calls, and handled orders for new machines and such, but his, ahem, secretary took care of minor items. Second, while cards were still on allocation in early 1946, mine got printed up very rapidly indeed, and if I asked for standard stock, it arrived literally the next day, by taxi!
I also saw Tom a few times at the SSEC and at IBM celebrations, and presumed on our rather remote prior connection to talk a little about the lab and the computing milieu. He was reported to have greatly enjoyed the story, quickly recounted to him by Endicotters, about my adventure with Mr. Phillips.
I had made my first overnight visit to the beautiful Endicott Homestead, the guest house near the Country Club. As was the nice custom, the housekeeper put me in the Watson Suite as a new senior employee. When T.J., often with his wife, came up to Endicott he used this elegant suite, with a porch overlooking one golf course, and the most lavish bathroom I had ever seen in my impoverished life. When it was not so engaged, distinguished visitors or other IBMers were given a chance to stay in it.
On my last day, the housekeeper asked me if I could leave the suite early; she knew I was taking the late sleeper back to New York that night, but John Phillips the IBM president was coming up in the afternoon. "Of course," I said. That evening we all gathered for malted milk nightcaps in the Homestead dining room. I introduced myself to Mr. Phillips, and sat next to him at one of the round tables. Perhaps not choosing the most felicitous phrasing, I told him how much I had enjoyed being in the Watson Suite. "They worked the old hot bed system on us, Mr. Phillips," I said. There was a pregnant silence. "Carries [-134-] me back to the railroad," he said. Nice! He had been a conductor on the Erie when T.J. had spotted him, and hired him as his male secretary.
So I was not too ill at ease when I entered the sacred portals. Didn't last long, though; without preliminaries Tom threw my MONY proposal down in front of me and demanded an explanation. Turned out that it had indeed generated interest at the insurance company, and had been passed around at the Board of Directors - and one director was Tom Junior!!
Oy weh, as my Jewish friends would have said. I told my story. No, I didn't want to leave IBM, but I had been unable to push my idea about the Washington Technical Computing Bureau past the Headquarters staff (that was code for McPherson and Hurd), and was looking for another route. I made it clear I had prepared the proposal outside office hours and so on - no mention was made of when I made visits to the MONY assistant vice president.
"They told me about your Washington proposal yesterday," he said, indicating he had been talking to Eckert and McPherson about whether to fire me out of hand, or just string me up by the ankles and beat me with a sjambok. He was not very angry, but definitely not amused; gossip said he had a violent temper, but in spite of the strains between us then and later, I never saw it. He probably reserved it for failed salesmen, or adversaries over six foot two.
He took back the proposal. I wonder if it is somewhere in the IBM historical archive today; more likely he gave it back to MONY later. I still have a copy somewhere. "Let me know by this time tomorrow whether you want to go down to Washington for IBM. If not, Dr. Eckert will discuss your departure with you." No, no, no, I said. "Just point me in the right direction; I have to arrange for the CPC, and two or three people, and talk to someone down there about space. I wanted to do it - still want to do it." Good, he said, dismissing me.
Before the end of the week I had been promised the prototype CPC II, not yet fully designed in Endicott, three unnamed juniors "from Pure or Applied Science," a special customer engineer like Dick Bennett, and space in Washington Federal "to be arranged with Mr. LaMotte at the proper time."
Indeed, the catch was timing. This was the fall of 1950; I was still teaching my numerical analysis class at Columbia. The crucial decision about the Defense Calculator project was about to be taken (if he had not been thus occupied, Cuthbert Hurd would probably have expunged me single-handed). The CPC had to be finalized, and the prototype put through the shop. Late spring of 1951 was the earliest possible date to ship it to Washington.
This explains "wait mode." I had duties at the Watson Lab, but Wallace and I knew my heart was no longer in them. He was pleasant but disappointed; wanted me to do what was best for me, but a little uncertain how he was to manage the computing rooms and the visitors, and who to plug into my Columbia course in 1951. Fortunately he did not have to deal with Mary Noble and the art collections any more!
So as soon as my Columbia obligations permitted, I began going up to Poughkeepsie to help with the Defense Calculator. Often I would go on to Endicott, to look at progress on the CPC II. It was near Frank Hamilton's engineering suite, although under different auspices; I used to drop in on Frank, who was by now a valued friend, as relief after hassling the CPC crew.
He kidded me a lot about my frequent and unproductive visits, and I explained about how anxious many [-135-] West Coast customers were to steal away my prototype machine. In return he showed me a little of what he and Ernie Hughes, one of the SSEC designers, were doing. They were working on a drum system; today we would call it a minicomputer. When announced in mid-1953 it became the famous IBM 650.
"I may have to come up here and sleep on the joggle board of the tabulator [front shelf], when my baby gets near shipment," I said. "My ticket from Tom Junior is only good for so many rides, and then Cuthbert or North American or whatever will throw me off the train." He said he wished that was all he had to worry about (he was building his system around a big magnetic drum, and hadn't been able to get a working surface on the drum). Frank was an optimist; I was an optimist. Neither of us dreamed that two thousand of his machines would be rented and sold.
Back in our Chelsea apartment Dorothy was facing her first IBM move. We had lots of time, and lots of help. Mary Noble Smith heard I was going down to Washington "for young Mr. Watson"; those things got around World Headquarters like lightning. Why didn't I use the art movers, she said, meaning the special company that had delivered the Syracuse ceramics collection and Isaac Newton to the Watson Lab. I guessed that the charges would be a great deal more than Chelsea Movers, but I was warming up to the idea of travelling first class. Months later, when we asked for them, solemn little men who were accustomed to handling Van Dycks for the Metropolitan Museum came to our place and wrapped Dorothy's jelly glasses. IBM!
I made a trip down to meet the great Red LaMotte. He was abrupt with me - no matter. It was clear that he had never inverted a matrix in his life, and wasn't planning to start in 1952. I worked on him. Yes, he knew about the UNIVAC coming to Census. Yes, he knew about the UNIVAC coming to the Pentagon (he was amazed that a whiskered scientist in a sport jacket had heard about such things). Yes, a little publicity about the new IBM computing facility that was coming to Washington might be helpful.
First he sent me to meet my putative boss, a smoothie from the Blue Suit Brigade who ran the Washington Federal office. There was a commercial office also, and a card manufacturing plant of considerable size, and a Service Bureau; everything reported to LaMotte. Then he sent me to Bachrach for mug shots - for walk-ins, a two-week wait; for IBM, same day appointment. Then he sent me over to the Service Bureau to meet Scottie, a charming but harried ex-Britisher who had just moved his shop into midtown quarters on 19th Street from the card factory. Scottie was luxuriating in his first elbow room since 1940. I told him I would need all his spare space: a separate machine room, a library/conference room, an office nicer than his (which wasn't saying much), and a ladies room for my, ah, staff - he had only men. It had been great the few weeks it lasted, said the poor guy.
Then LaMotte really got up steam. "I have a surgeon friend," he said. "His daughters are getting ready for their debuts, so he has had to take a much larger place for entertaining. His wife has a little row house on Q Street that they plan to come back to in three years or so. Go and look at it; tell the agent [-136-] I sent you." The place was exactly what we needed, and quite a step up from a mud road in wartime Kensington. But of course it was promised.
"No matter," said LaMotte gruffly. Next week the agent mailed me the lease. Then all we had to do was wait.
In Chapter 14 you will encounter
(in order of appearance):
The ARS [American Rocket Society] my third professional affiliation
The BIS [British Interplanetary Society] publications in the Thirties
Billie Slade part-time secretary for the ARS
Willy Ley journalist, author, and former rocket enthusiast in Germany
Ed Pendray publicist for Guggenheim's various aerospace activities
Harry Guggenheim heavy investments at Cal Tech, and the Goddard sponsor
Esther Goddard widow of the pioneer American rocket experimenter
Wernher von Braun stationed at White Sands, he was about to join the ARS
Andy Haley von Karman's lawyer, and hence first president of Aerojet
Theodor von Karman premier aerodynamicist, and Hans Kraft's old teacher
Jimmy Doolittle demonstrated the importance of short take-offs!
JATO [jet-assisted takeoff] built behind Pasadena at what is now JPL
Bill Gore Marine JATO test pilot who later became president of the ARS
Aerojet Andy arranged to sell it to General Tire
John von Neumann 03
"The Great Vons" all three worked eagerly for the military
The V-2 a lethal demonstration of von Braun's engineering leadership
Mars Wernher wanted to go himself - not just send a camera!
Laurence Rockefeller funded Reaction Motors in New Jersey
The ARS presidency I was neutral (an IBMer), and a celestial mechanicker
The AIAA [today's major aerospace society] ARS ate up tin airplanes
Martin Summerfield I made him first editor of the JOURNAL OF THE ARS
The Royal Aeronautical Society Royals didn't get eaten up
Arthur Clarke my friendly counterpart at BIS; not yet world-famous
Red LaMotte 13
Stan Rothman 10
Libby Lindberg joined my Washington venture from the SSEC
John Mayhew a conservative from a Texas wind tunnel, via Applied Science
Carl Southard even Dick Bennett couldn't touch him for diagnostic skill
The Washington Technical Computing Bureau 13
Washington Federal an enclave of seniority in a surround of IBM go-goers
The Air Navigation Development Board pre-FAA, and my first customer
Dorothy 01
Undulant fever my ophthalmologist friend diagnosed it without seeing her
Bob Carlson his arsenic poisoning made a great testbed for BAL
BAL a secret chelating agent to pump out heavy metals
Lovell Lawrence I contacted his secretary when Rockefeller eased him out
Don Gamel 13
Hurd 12
George Petrie he was Applied Science rep in LaMotte's fiefdom
[-139-] While Dorothy and I, and the jelly glasses, are metaphorically en route to our miniature palace off Q Street, I must take time to explain how I had become involved with the other two Vons, von Karman and von Braun. The stories revolve around my third professional society affiliation, the American Rocket Society.
I had been encouraged to join the astronomers in 1939, and the optickers in 1942. And in 1947 I became a charter member of the first computer society, the Eastern Association for Computing Machinery (now ACM). But in 1946, before Brother Berkeley rallied us to the flag of ones and zeros, I had joined the space enthusiasts.
I was in New York, where the ARS was getting free office space from ASME, the mechanical engineers. I was nicely supported by IBM. I was not affiliated with the contending proto-industrial forces in the rocket arena (about which more later). And I had served my time in celestial mechanics, one of the underlying disciplines of space travel; doctorates were scarce, and highly valued as camouflage by the backyard experimenters who were firing midget rockets out on Long Island, and in deepest New Jersey. I had studied the literature - Verne, Wells, AMAZING STORIES, ASTOUNDING STORIES, and Ley and Oberth and Hohmann, and the Goddard reports from New Mexico.
I started going to the monthly meetings, and reading the poor little magazine. To do the rocketeers justice, however, I must point out that there was no regular computer society publication until six or seven years later, while ARS, the British Interplanetary Society, and the pre-war German association all had at least quarterly journals in the Thirties!
Soon I met Billie Slade, the part-time paid secretary of ARS, Willy Ley, [-140-] already a well-known science journalist, Ed Pendray, a public relations man for Harry Guggenheim who had funneled the latter's support funding to Goddard in the Thirties and early Forties, and Goddard's widow Esther. The experimenters included a dozen men who became famous post-Sputnik. Wernher von Braun had arrived in the U.S. and was about to join. The academics, very few in number, were mostly from aeronautical engineering. One unusual figure was an Andrew Haley, a Washington lawyer with considerable personal financial resources and a great fondness for Yellowstone bourbon.
Andy was one of my three routes to von Karman. To explain why requires a little industrial history, leading from Cal Tech to General Tire. Theodor von Karman, the greatest name in theoretical aerodynamics worldwide, had emigrated to the United States from Hungary in 1930, pre-Hitler, and ran the Guggenheim (yes, same money) Aeronautical Laboratory in Pasadena. He was much admired and much trusted by the flyboys in the Army Air Corps, and as part of the Cal Tech war effort was asked to take on the development of jet-assisted takeoff devices - somewhat stimulated by the Doolittle attack on Tokyo, which entirely depended on getting off from a short run, and also by other aircraft carrier applications.
He set up a shop back in the hills above Altadena, and after a certain number of unscheduled explosions began to produce JATO bottles. The first test pilot, Bill Gore, later became an ARS president; it was a very tightly knit group, and small.
There had to be a corporation, and Karman turned to his personal lawyer, who had been doing his will and worrying about the family estate back near Budapest. That was Andy Haley, who became the first president of what is now Aerojet, and later arranged to sell it to General Tire and Rubber (for novices, it might help if I point out that solid rockets, including the ones on the ill-fated "Challenger", use a rubber-based propellant!)
A second route in to the great man was my friend Hans Kraft at GE Schenectady, who had been a favored pupil of von Karman's in Germany. This in turn produced my first meeting with Karman, at the SSEC, when I took him on a tour and explained what the big machine was going to do for Kraft. His continuing interest in fluid mechanics calculation after he returned to Cal Tech in the late Forties encouraged the youngsters at the Jet Propulsion Laboratory (which is what his facility back in the hills had become) to be early and vigorous computer users. And that was my third route in to him.
He was a little gray owl of a man, but kinder than Johnnie von Neumann, whom he knew well of course. Johnnie had big warm liquid eyes, but there was ice down underneath; Karman was more tolerant. Kraft said he had been a wonderful teacher. When I took him through the SSEC in 1947 he was nearing seventy, but still interested in my first impressions of the rocket people. He was not particularly responsive to the science fiction side of rocketry; unlike von Braun, whom I had just met for the first time a few days before, he did not want to ride one of his dragons to the Moon or Mars.
Both von Karman and von Neumann were deep-insight people; two tremendous scientists. Wernher von Braun was a super engineer, one of two I have met in my lifetime, and like the other one, Gerry Neumann, also a terrific leader. The two [-141-] great scientist-mathematicians and the great engineer-leader - the Great Vons, as I like to call them - all put their skills eagerly at the disposal of the military. You thought of this immediately in von Braun's case, because of the V-2s. But I look at it today rather differently. Karman and Neumann had other paths they could have followed; Cal Tech and Princeton could have absorbed them again. Wernher wanted to go to Mars; he had to build monsters to get there, and he built them for Hitler and for Lyndon Johnson, and would have built them for the Russians or the Chinese. All three are dead; from the Nineties, I empathize more with the spaceman.
Haley and von Karman represented a West Coast rocket enterprise. There was strong competition in the East, with a younger entrepreneur instead of the shadowy Harry Guggenheim. This was Laurence Rockefeller, who was backing Reaction Motors in Far Jersey. The two outfits sponsored ARS events, and paid the expenses of ARS volunteers. There were neutrals; I remember a director from Linde Air Products who sported one of the very first synthetic star sapphires, made in the Linde laboratories. And I, as I rose rather rapidly in the tiny organization: IBM supplied both sides, as did the liquid oxygen guys.
I got IBM to take a cheap advertisement in the ARS magazine. I sweated on the board of directors over our painful finances. As a useful writer and speaker, I helped Willie Ley and Ed Pendray on the PR side. All of a sudden, I was president - of about 2500 excited but impoverished enthusiasts. It was a strange feeling. I was 33. I was regarded as a fledgling astronomer. I had already risen to national program chairman for the Optical Society, but in spite of nine years of membership was obviously only at the foot of that rather specialized ladder.
I had made genuine research contributions in astronomy and optical design. I had never seen a rocket fired, let alone designed one. Yet here I was, decked out in my first dinner jacket, welcoming Washington big shots and distinguished foreign visitors to our American Rocket Society annual meeting. And ARS was already much larger than the astronomers, somewhat larger than the optickers, and ten times as large as brand new ACM. My, my - strange world!
Growth was the key, of course. The astronomers are still very small. There are other optical groups now, and all of them together add up to only a few thousand members. The space gang grew and grew, took over the tin airplane society, and today is the American Institute of Aeronautics and Astronautics, with over forty thousand members. And puny little ACM is now up to 80,000 bodies, has a budget of $40 million a year - and I've been president of that too!
Well, ARS wasn't all Boards and banquets. I cast about for ways of legitimizing its wilder talents. Publications looked like a self-supporting possibility; I found an excellent editor at nearby Princeton, Martin Summerfield, and he planned a JOURNAL OF THE AMERICAN ROCKET SOCIETY, the forerunner of dozens of handsome periodicals now issued by the AIAA. Somebody - Ed Pendray, I guess - found a good printer who was used to technical letterpress and to not being paid promptly, and we were off and running.
[-142-] The Brits had an Interplanetary Society, and it too had a publication which wanted to grow into something technical. Too soon - there was no missile program in the U.K., and their tin airplane association, the Royal Aeronautical Society, was soaking up all the advertising and all the industrial support. The president of the BIS the year I was head of the ARS [1951] was Arthur Clarke, now world famous for his science fiction and his movies, and chancellor of a university in Sri Lanka to boot. We are still good friends, and meet once or twice a decade on the lecture circuit - but I have to get back to computing!
I had told some of the rocket story to McPherson, and when I met him, to LaMotte. John found it disagreeable, I think - undisciplined, which it surely was. LaMotte had an eye on the Washington agencies that were putting money into missiles, and was anxious to support my contributions. Perhaps more important, my wife Dorothy thought it was exciting. She felt much more at home with the astronomers and the spectrographers, but she couldn't help liking Andy Haley and Billie Slade, and besides, it was sort of a family enterprise.
So was astronomy, and a worldwide family at that. But it was old and staid, and it took twenty years and a lot of night observing to really become an insider. Heber Curtis, the observatory director at Michigan who had been so kind to me, was just such a one, but it had taken his whole lifetime. Dorothy marvelled that I had been clutched to the ARS bosom in four years.
We were a little ill-at-ease on the banquet circuit, and especially on the dais, but IBM had polished both of us up better than we knew. I was by this time not only an eager ham, but an experienced one. Dorothy bought some less conservative evening dresses.
My little group of workers was assembling. I had brought Stan Rothman from the Watson Lab, and he had begun in Endicott to wire and test the general purpose boards for the CPC II. Stan was still a bachelor; had a good degree in math from Yale, and was nearly finished with a master in mathematical statistics from Columbia. We laughed ruefully over his inability to write well, or even to spell (although he later became author of a major early social-implications book), and I planned to have someone help him when we got to the stage of writing the documentation. He was hoping it would be Libby Lindberg, a very lovely blonde from the SSEC junior crew.
Libby was about to become engaged, and in spite of Stan's dark masculine attractions was not much at risk. I remember faintly that her fiancé was already Washington-bound; anyhow, there was no relocation problem. She thought my struggles to provide her with a WC were ridiculous; was perfectly willing to go unisex. But I remembered that IBM vice president in 1946.
The third recruit was John Mayhew, from one of Hurd's enterprises. He had been in the wind tunnel business in Texas, and found New York a bit much. He didn't mind the dress code or the No Booze Rule, and in fact was extremely conservative by nature. It was the subway hustle and the discourtesies that got to him; he hoped the Washington pace would be more civilized. I figured he would be my analyst while Stan was struggling with the new machines.
I had been very anxious to get a special customer engineer. With a machine so new there were no manuals, for operations or for maintenance, and with Rothman needing all sorts of inside information about timing to [-143-] make his monstrous plugboards work, I was sure we needed someone unusual. Well, one was forthcoming. His name was Carl Southard, and he was already lightly attached to LaMotte's empire when I hove in sight.
I couldn't believe him when we met. An IBMer? A favorite of Red LaMotte's? He was messy; his shirt was indeed white, but only for an hour or two each morning. He pulled his tie loose at the first difficulty. He was ovoid rather than IBM tall-and-lean. And he always had a bedraggled cigarette hanging from the corner of his mouth.
But he was good. Boy, was he good! He would hang over the printer, his eyelids at half mast, ashes falling into the type bars, and listen to Stan or John, or occasionally Libby or me, describe a problem. He would ask dumb questions in a southern, or Southard, accent. You wondered, at first anyhow, if he had had a bad night and was falling asleep. He would mutter to himself. Then he would open up a relay gate, or a rack of tube assemblies in the 605, and twiddle something, or replace a unit. That was it. Oh, he did hit really tough ones once in a while. But even Dick Bennett back at the Watson Lab couldn't touch him for sheer, oh, insight.
Carl didn't want to go upstairs into design engineering. He didn't like the atmosphere, and he was used to Washington. He went to Endicott for me and helped wire the prototype machine, and marked up a private set of blueprints for use later. I was afraid I'd lose him; in fact, toward the end of his Endicott visits, when I saw how good he was, I worried more about losing him than losing the CPC. But it all worked out. The machine was shipped; Carl understood it better than the designers; Stan was well into the plugboards.
Scottie was remodelling the back of the 19th Street building for me. No beancounting; someone was getting cost statements, but it wasn't me. However, when my crew began assembling, he started to hand me payroll printouts for them. I was not satisfied. Inquiry revealed they were actually being generated on his service bureau equipment, in the front of "our" building, from cards punched up at Federal two blocks away. I had Libby punch them instead, from my data, which got my guys (and me) off Scottie's timeclock. I wrote a note to my boss, copy to LaMotte, saying that "I was doing my own payroll, with Scottie's help." Nobody complained.
I tried to order books for the little library. I had put my own rather large collection in there, trusting my troops not to abuse it, and solving the problem of persuading the architect to also put shelves in my private office. Wasn't easy; the general feeling was that sales offices didn't buy books. "Ah," said I, "this is a Technical Computing Bureau," emphasizing the capitals.
"Would you like me to have the librarian at the Watson Laboratory buy them for me, and transfer the costs down here"? Good grief, no, said a beancounter I ran to earth. In the end I got an unsuspecting bookseller in Georgetown to open an account "for IBM on 19th Street," and send a monthly bill. It disappeared into the system; the bookseller never complained either.
I began to get business. There was some interest on the part of the older salesmen, and I was doing a small amount of direct contact myself. The word "older" brings up a difference in Washington Federal. The salesmen I had met in Manhattan, and at the Hundred Percent Club I had attended, were young and snappy. They planned to be local managers in two or three years, [-144-] district managers in two or three more, do a tour at 590, and be Golden Boys or at least on Tom Junior's staff, in their early thirties. Some of them made it; the company was expanding magnificently - twenty percent a year. Quite a few of them disappeared, to surface a year or two later as typewriter or timeclock salesmen elsewhere in IBM, or as tab furniture or paper forms salesmen outside the sacred portals.
It turned out that while banks and insurance companies were impressed with these bright young men, federal employees were resentful. So there was a fair complement of low-key salesmen in Washington Federal (Commercial was like New York - or at least Baltimore) who had been there a long time, or who had been successful local or district managers with a lot of government business and asked for a Washington position. They were mostly salaried, which solved my problem of an attractive commission.
Anyhow, I had prospects. One was the Air Navigation Development Board, a forerunner of the FAA development activity of the Sixties and Seventies. They wanted a simulation of air traffic patterns for various sizes of protection zones around the commercial, private and military aircraft. Two dimensions at first, altitude later - and very simple air lanes. Childlike compared to what we do on micros today, let alone in an arcade game, but hot stuff in 1951. I worked on it a little myself, then John and Libby took over. We got ready to send out our very first bills; exciting.
Also I began to get job applications. One interview I remember was with a serious candidate named Sarahan, who was working on a computer construction project at the Naval Research Lab. I sent him up to Applied Science with some pride. Others tended to be Service Bureau types that I could turn over to Scottie or his deputy.
It was a fascinating time. Our move down from New York had opened some seams in my relations with Dorothy, and we were undertaking repairs with some trepidation. The Rocket Society and ACM were exciting, and I still had an involvement with the Optical Society that I hated to let go; for instance, I had been the arrangements chairman at the meeting where Haloid first demonstrated xerography publicly, and the program chairman when Ned Land demonstrated instant photography - black and white, of course. And I was watching the Applied Science incursions at World Headquarters, and participating defensively in the seminars which Cuthbert Hurd was giving for my buddies out in the world.
About Dorothy: from our first years together she had suffered from nasty little fevers and infections, and from depressions that we attributed partly to the other trouble and partly to the problems of wartime. We had solved the first problem, and had helped with a war problem involving her brother.
I had done a consulting job in 1944 for an ophthalmologist in midtown Manhattan, and we had become friends. He is in his late eighties now, but still in practice; we talked transatlantic a few years back. I had designed a standoff lens for him, to go on the front of a powerful microscope objective and greatly increase the working distance (from much less than a millimeter to about 25). He could not afford to have me do much, even at my low prices, so I could not achromatize the design; he was content to use it with green light [-145-] from a mercury lamp. We tried to get him a patent for several years, which kept the relationship alive.
He was aware of Dorothy's malaise, and wanted to help. Since she did not seem to have eye troubles, we resisted. He said he treated a broad spectrum of patients; all was grist that came to a midtown practice in those days! I did not realize until later that he had a unique medical method; each week he read the Lancet or the AMA Journal, and boned up on one disease or problem. For the rest of that week he treated all his drop-ins for that ailment - not if they had a broken wrist, or just wanted a fresh eyeglasses prescription, of course, but everybody else.
One week he broke the news to me that he had diagnosed Dorothy's problem: she had undulant fever, also known as Malta fever. She should go to a specialist down the street who was an expert in white blood cell analysis and have her opsonocytophagic index measured [after fifty years I think I still remember how to spell it]. He had never examined her himself.
Well, it was true; she did have undulant fever, from drinking raw milk as a child. She had had it for thirty years. There was a specific antibiotic. She took it. The fever went away and never came back. Now that's medicine!
Second story: Dorothy had a younger brother who was an engineer in Pasadena for the phone company. During the war Bob became a radar installation and maintenance engineer for Western Electric; was catapulted off ships at sea to repair jobs; worked out of places like Newfoundland. Not surprisingly, he developed medical problems, including a terrible case of shingles.
To control this the Navy put him on Fowler's Solution, which is arsenic. He broke his medicine dropper and the Navy gave him a big-bore replacement. He came down with arsenic poisoning; the Navy flew him home to New York, where we tried to help. He was on his last legs; they called it drop-foot. Not to worry, they said, implying it wouldn't show in the coffin.
Enter our, ahem, family physician. "I have some doctor friends working on a top-secret project," he said. "They know how to pump arsenic out of patients, and they are all set if the Nazis use poison gas." There was indeed such a chemical, and it was indeed a secret; it was called BAL, British Anti-Lewisite, referring to a variant of WW I mustard gas which poisons the blisters with arsenic. Today such chemicals make it possible to pump all kinds of heavy metals like lead and arsenic and plutonium out of people; they are called chelating agents.
The project people were delighted to have a clinical experiment, and with the Navy and Western Electric watching. They gave Bob the Secret Chemical, presumably with our friend in attendance. It worked. Even the drop-foot went away. Bob went home to Pasadena and resumed the gentle life; even got married.
Dorothy was ecstatic, for a while. But the depressions continued into the postwar years. We found a nasty psychiatrist who was big at Columbia Presbyterian; we were living in Chelsea by this time. He had his own, ah, chelating agent, sodium pentothal; in a few weeks he had pumped some of the heavy metal out of my dear wife. Trouble was, it was only analysis; he told her what her problem was, but not how to cure it. My hypnotist friend Andy [-146-] Salter said he couldn't solve the problem, which revolved around her parents, but he could make her less aware of it. Would it improve our sex life, I asked. "Always," said Andy. "Get another girl," said Dorothy. And at the height of the excitement, I uprooted the poor baby and moved her down to Washington. Indeed, seams opened in the relationship.
I've painted a glowing picture of the Watson Empire. Fantastic things could be done, great rewards were on display, the surroundings - art collections, Waldorf parties, Homesteads, Country Clubs, Kenyon mansions - were lovely. There was a dark side, of which the arbitrary expulsion of married women had angered me most; it was an empire, not a democracy, and the imperial rites were sometimes painful. Two of these darker events came close together just as the little Washington shop was getting off the ground, and in the end drove me to rebellion and expulsion.
The first was by far the uglier, although not so important. I had asked for a secretary "after we get going." There wasn't anyone at Scottie's bureau; his bids were sent out from Federal or Commercial. I had heavy Rocket Society correspondence, with LaMotte's approval, and an assortment of scientific and business paperwork besides. One of my predecessors as president of the ARS was an engineer named Lovell Lawrence. He had been one of the backyard rocket experimenters before the war, had been involved with Goddard, and had ended up in 1950 as head of the rather large Reaction Motors activity in New Jersey. He had a falling out with Rockefeller, and left for another job in the Middle West. His secretary, whom I had dealt with many times on ARS matters, did not want to go to Detroit. Would she consider IBM in Washington, I asked?
She was not exactly Washington Federal, or even Washington federal: a very forthright woman indeed, and vividly handsome. I knew how good she was, and Lawrence had of course confirmed it with a written recommendation. I brought her down for an interview. She was sent to the IBM physician for a physical, and turned down cold. After some very loud bitching on my part about my rights as the prospective superior, I got a hushed-voiced personnel guy to say the doctor "had found traces of sexual activity - and the young lady is not married." What business is that of ours, I said?
Well, I took it to my boss Don Gamel, the head of Washington Federal, and got absolutely nowhere. The doctor was Mr. LaMotte's personal physician; the decision is final, he said. I knew instinctively that I had hit another Watsonism of the dark sort. I had been told at the Watson Lab not to protest the expulsion of my married supervisor: "the law of the Medes and Persians, which altereth not." I should have been a Daniel then, for a loyal and long time employee and a close friend. The Washington affair was also an outrage, but my potential secretary laughed it off, and went away.
No doubt a fair number of the IBMers thought I was involved, and perhaps Dorothy did also; at least she seemed less angry than she had been when Marj Herrick got beheaded. Sadly, I wasn't; the young lady didn't need any amatory assistance from the current ARS president, especially in Washington. She got a good job, but in Buffalo.
The Poughkeepsie excitements were mounting. Bill McClelland and the rest of the Hurd contingent were writing working programs; a new set of [-147-] letters of intent had been received; 701 production was about to commence. Cuthbert began to turn his attention away from the Defense Calculator, and toward defenses. His Washington representative was George Petrie, who while personally very friendly had done almost nothing to help build up my shop or my order book. I was sure my rather threatening progress was being fully reported back at Applied Science.
Suddenly I got an official note from Federal. The business demands on Applied Science and its New York computing bureau were increasing, and they needed to requisition Stan Rothman's services for an indefinite period. Hopefully he could return to Washington in six months, but that would depend on business developments. Deliveries of Model II CPCs were soon to begin, and his knowledge of the machine would be used more effectively than in Washington.
Well, that was my weak spot, and Cuthbert had struck it unerringly. I could not replace Stan quickly. I could not finish up his documentation; I wasn't even sure Carl and John and I could operate with his not-quite-finished plugboards. And I had paying customers coming in the door.
I went to my boss. I said that Stan would be just one more body in Cuthbert's herd, while his departure would destroy my little operation. I said I had talked to Stan about the offer, and that he wanted no part of it. "You can replace him, Herb," he said, "and Mayhew will get you through the Air Navigation Board job while you are looking. As for Rothman himself, I'm sure he understands that advancement in IBM usually requires moving; he came down here with you willingly enough," and more of the same.
What was I thinking? Not about Hurd, not about Tom Junior, not about where to look for another job - but about Marjorie Herrick and my almost-secretary. I had stood up to The Old Man, and felt good about it afterwards. Then, later, I had flinched, twice. If I knuckled under this time, there would be a fourth time, and a fifth, and dark events without end. Somewhere another imperial wizard was plotting against Hurd, as Cuthbert was plotting against me. But I would be erased, or reduced to impotence, long before he was pulled down.
So I went to Red LaMotte and told him how I saw the picture. Nonsense, he said, we all just have to buckle down and do what is best for IBM. "I know how to use Rothman best, for IBM," I said. "If he is not protected, you'll lose him, and probably the rest of the bunch - and me." I didn't try to temporize, and I didn't say I'd appeal to Tom Junior. I put it on the line.
And he fired me. He was not amused.
Stan resigned, and went to work for the Rand Corporation in Santa Monica, where he helped greatly with their new CPC II. Libby Lindberg also resigned, and married her young man in the State Department. Mayhew was transferred back to Applied Science; after all, they had said they had to have a man. Southard shrugged and went back to simpler machines, but some years later left IBM and established a successful nationwide terminal and network maintenance company in Carolina, and got rich.
George Petrie took over the bureau temporarily. I never found out if the air traffic simulation was completed.
In Chapter 15 you will encounter
(in orderof appearance):
Don Gamel 13
Hurd 12
Joint Computer Conferences 11
Jay Forrester made MIT's WHIRLWIND digital, and invented core memory
Sam Alexander building SEAC at NBS, with shuddersome peripherals
Sam Caldwell and Gordon Brown two of the outstanding MIT EE professors
Norbert Wiener disdained digital machinery and almost everything else
Doc Draper 02
Core memory tiny nonlinear ferrites and coincident current circuitry
MIT world's greatest engineering school, according to its people
Signed ternary arithmetic no algebraic signs, and no rounding ambiguity
Infallible computing from highly fallible components: ah, Tandem!
Bob Everett Number Two at WHIRLWIND, and Number One later at Mitre
Irving Reed an intellectual of the logical design persuasion
Charley Adams ran the non-air-defense computing for Forrester
Cape Cod a real experiment with real radars and real fighter planes
Al Perlis and John Carr three future ACM presidents played softball
The Appalachian Mountain Club lots of foldboating
The ASCC 01
Dorothy 01
Havana "you can't learn everything from books", I told her
Déodat [poodle puppy] focus of our attention and our love, for years
E.T. [Charles] Goodwin top computer man at NPL, and Turing's boss
NPL [National Physical Laboratory] British equivalent of our NBS
The Chrysler New Yorker convertible easily the fanciest car at WHIRLWIND
Hans Kraft 09
The Hanford 702 would do multimultidigited decimal multiplications
Bert Sells he mostly did turbine design for jet engines, on a CPC I
GE Evendale was to install a Defense Calculator, near Cincinnati
GE AGT [Aircraft Gas Turbine] Division moving from Lynn to Evendale
Personnel staffing the first big machine in the Midwest would be tough
Paul Herget 05
John the Baptist GE asked for my head on a platter, and Jay said yes
"Generous Electric" and so they were to me, for four years
A Napoleon quote "well positioned to attack in any direction"
Gene Gettel a serious young mechanical engineer; I swept him up
George Richter 01
[-149-] In later years I became quite expert at being fired or uprooted, but this first time found me unprepared. I had no plan for a new job, a new employer, or a new location. I was flexible enough; the problem was that the timing was forced on me by New York. Also the blowup itself took less than a week.
The wild reorganizing and job-hopping of the Fifties and Sixties was just beginning. When something like my Rothman crisis came up in those times, the guy or gal under pressure pretended to accept the unacceptable, put out lots of resumés, slipped away for clandestine interviews, and departed in style - with all the ancillary expenses covered. Gamel gave me a month's salary in lieu of notice, and slammed the door behind me.
Parenthetically, the whole little drama was unfortunate - for me perhaps, and for IBM certainly. Hurd probably brought with him from Carbide, and certainly found at 590, the "win 'em all" philosophy and its attendant rough infighting. He was prepared to overlook Eckert, whose desire to go on with astronomical researches cancelled out his ready access to Watson Senior. But Cuthbert wouldn't hold still for the establishment of a rival computing bureau.
He could have co-opted me with a few words: taken me up on a high hill and spread the world of a technical service empire out before me, under him of course. Trouble was, I had much more confidence in that vision than he did. And he didn't trust me. So the company lost three good people and disturbed others, and a promising new IBM entry into the peculiar Washington scientific/technical environment was set back.
I've already said how important the early computer conferences were, as a substitute for today's technical publication and media coverage. Everybody went to every conference (there were only two or three a year, as against the many hundreds we have now). The second of what would soon be called the [-150-] Eastern Joint Computer Conferences was held in Manhattan right after my firing, and I found myself there in deep conversation with Jay Forrester, who was building the world's fastest computer at MIT.
That meeting featured peripheral equipment, including such weird novelties as magnetic wire input and output for the SEAC [Standards Eastern Automatic Calculator] which Sam Alexander and his crew were building in Washington. Forrester and his project were long on innards but very short on I/O - paper tape and clumsy telex machinery, poor devils - so he had come down from Cambridge to hear about possibilities. Indeed, there were even a few primitive exhibits, harbingers of the giant National Computer Conferences of the early Eighties.
Aside from my need for another job, I had been drawn by the set of papers disclosing, and in a professional sense announcing, the IBM Defense Calculator. I'd been out of touch with Rochester And Company for only six months, but all sorts of exciting new stuff had happened in that time, and speeches were revised up to the evening before presentation in those days - and sometimes at the rostrum!
Jay Wright Forrester was already a major figure in our burgeoning technology, in spite of his youth. He had come out of farm country and the University of Nebraska, and had been a prize graduate student among the great MIT professors: Sam Caldwell and Gordon Brown in electrical engineering, and Norbert Wiener of course. He was tall; handsome in a very reserved way; a driven man always rather uncomfortable to be with. Decades later, when I was president of ACM, I needed to reach him; his secretary said that even for a very old friend, and the head of the professional society he had belonged to for thirty years, the best she could do was to slide a note under his always-closed office door.
He was two months older than I, and had had the same sort of struggles with draft boards early in the war. But these soon diminished under MIT pressures, as he helped found the Servomechanisms Lab, and worked on the analog computers inside those same gunsights I was doing optics for on Long Island. He became associate director under Doc Draper in 1944, and was appointed head of a thing called the Whirlwind Project. This was to be a tremendous land-based analog air defense system (too big to be called a gunsight or a gun director), which would aggregate in some mysterious fashion the signals from a network of radars and vector out fighter planes to intercept the attackers.
It was a measure of Jay's vision that he had seen immediately, in spite of his immersion in analog techniques, that the project would have to be digital. Size was a drawback; the arithmetic and control unit for WHIRLWIND I filled a large room, although today it would be one corner of a chip smaller than a fingernail. And in a very real sense, analog was instantaneous; input was transformed into output without the delays of storage and analysis and computation in my kind of machines.
But the arguments for going digital were overwhelming, even when Forrester made his decision in 1949: accuracy first, perhaps, because it was difficult to go much below one percent in complex analog devices. A [-151-] continent-wide air defense needed, if not astronomical accuracy, at least four [decimal] figures - in modern terms, sixteen bits.
I mentioned aggregation. As soon as you began thinking about selecting signals from a variety of radars, perhaps averaging or otherwise smoothing these inputs, and moving data around under a wide variety of circumstances, the need for storage and variable programming pointed to digital - and did so even when the Whirlwind project was getting underway.
And futures buffs knew that analog techniques were near their limits, and that the digital world was in its earliest infancy.
I had heard that Forrester, drawing on the research of a wide variety of MIT people and laboratories, was well ahead of the IBM storage experts. Palmer had needed to go to Manchester for an electrostatic memory system, and Jan Rajchman at RCA Princeton wasn't having much luck either. But Forrester's crew were building special tubes, wiring them into WHIRLWIND for central storage, and running real-time exercises (much more demanding of reliability than what the SSEC or the CPCs were doing).
And history was in the making. Forrester was becoming famous world-wide as the inventor of core memory, which in the next two decades was to sweep aside tubes and transistors - yes, they had just been invented, and computer possibilities had been described the year before at the first Joint in Philadelphia - and delay lines and fast drums, as main storage for all sizes of digital machinery. I had read his article in the Journal of Applied Physics and understood the basic idea, which in those days was called coincident current technology.
Certain magnetic materials, especially some ferrites being investigated by MIT ceramics experts, had a very nonlinear response to electrical current. A reasonable pulse would cause a tiny magnet made from such stuff to switch polarity, but half that much current had no effect. So Jay had taken a little ferrite ring and passed two wires through the hole. When one wire or the other carried a pulse, nothing happened; when both wires carried a pulse at the same time, the magnetism reversed. In the early embodiments, a 32 by 32 mesh of these rings was woven in a frame, with 32 north-south wires and 32 east-west wires running through the 1024 rings, which from telephone practice were called "cores." In a computer memory, each core stored one binary digit, and 64 switches around the frame enabled you to select any one of its 1024 bits.
A world industry was to grow up around core memory. Forrester assigned his patents to MIT, which got rich from licenses granted to a hundred companies, including IBM. And Jay himself did very well, and richly deserved it. At the time we talked at the JCC, he was considering going to three dimensions, relying on a touchier ratio: that two coincident pulses would cause no response, but three would flip the polarity. I reminded him that the real payoff was how many cores he could pack into a cubic foot, that a 3D structure would be much harder to wire either by hand or mechanically, and that marginal cores would be more frequent.
He already knew all that, of course, but it got us talking. I said what might be genuinely gainful would be to store a ternary digit in each core, and calculate in base-three rather than binary fashion. There were materials - [-152-] some kinds of permalloy, as I remember - that had north, south and neutral stable magnetic states. I told him I had taught my Poughkeepsie evening classes at IBM about a special kind of base-three arithmetic I called "signed ternary," in which zero was in the middle of the number range. In this curious system there was no need for algebraic signs, no problem about the sign of zero, and you rounded perfectly by dropping digits.
Jay being a stiff type, I refrained from calling the ternary digits "tits," a name which had been the source of much boyish amusement in the Poughkeepsie classes.
Twenty years later I was utterly amazed to be asked to comment on a technical letter from two Argentines, who as graduate students at MIT had exhumed the report I had done for Forrester on base-three computing, and were proposing to try it with hardware. It would have been a great idea if something as tiny and as cheap as the core-memory rings could have been made with three stable states; the embodiment with two cores per tit which they proposed was not attractive. And somebody in Kiev had tried it, anyhow.
After I had done strutting about my trick arithmetic I told him I had been extruded by IBM and was, ahem, immediately available. "Come up and see me next week," he said!
In no time at all I found myself a member of the MIT staff - not a professor, but doing something called "industrial cooperation." Jay gave me a salary just slightly, judiciously, below what IBM had paid me in Washington ($10,000 instead of $10,200), and made me head of a new group called advanced logical design, which was to explore ideas like signed ternary arithmetic and variant patterns of core arrangement (yes, I had unwittingly hit two jackpots in our JCC conversation). The main task, though, was to think of practical ways of making an infallible computer out of highly fallible Fifties electronic gear. That infallible computer was to be the heart of the production version of the air defense system; we called it WHIRLWIND II but when it finally appeared it was called SAGE, Semi Automatic Ground Environment, a horrid military/academic acronym.
I reported directly to Jay and his deputy, Bob Everett, who became head of Mitre in later years and a major figure in air defense and related areas - air traffic control, for example. Jay and Bob (and I, in the next few months) were convinced that replication of subsystems, and automatic comparison, and replication of the comparison, and automatic updating of newly-repaired components, would approach infallibility. Shades of Tandem!!
Most of what I did is as dead as mutton today, but it may be amusing to note that my primitive first look at failure rates and reliability requirements led me to recommend five-fold replication. SAGE installations ultimately came out with two machines each, not five, and still filled large buildings. And they weren't infallible, either!
I had a group of three youngsters, and shared a secretary. One of the young men was Irving Reed, who stayed in the logical design business and became an important figure in its more arcane developments. Down the hall was a much larger group run by Charley Adams, which did non-air-defense applications. Most of their customers were MIT professors and graduate students, and it was therefore very close in feeling to my old Watson Lab at [-153-] Columbia. It had to be more closely managed, however, because WHIRLWIND was constantly being improved, and also had to serve as a test bed for future transformations like core memory. And of the time remaining, the air defense work took the lion's share; there was a real-time experiment called Cape Cod with real radars and height finders, and real fighter planes, and real cross-grained military types. So you had to be very careful not to waste WW I time by indiscipline or stupidity.
Somebody brought around a compartmented tray late in the afternoon, and you dropped your little roll of paper tape, held by an approved rubber band, into your assigned compartment. Next morning your run result [teleprinter output] was in your mailbox. Born programmers did quite well; the less gifted took many weeks to get anywhere. Among the more talented paper-tapers were Al Perlis and John Carr, and after the collection we would go out on one of the playing fields and run off a little softball - surely the only time in computer history when three future ACM presidents cooperated that strenuously!
I noticed that, as at Columbia, there were lots of engineering projects in Charley's shop, but no math; chemistry, but not much physics. Inquiry revealed that Norbert Wiener, who had just created the term "cybernetics," never came to our building, and was contemptuous of such foolishness as giant digital computers. I had a couple of disagreements with him at seminars, largely because he kept trying to fit his continuous-math criteria to our discrete-math efforts. I escaped relatively unscathed because I tended to use celestial mechanics examples, and quote Poincaré; also he soon observed I had an uncomfortably frank way of dealing with pomposity. Nevertheless I soon felt impelled to christen him Nasty Norbert. It spread quickly, partly because he was so ugly. What I was referring to, of course, was that other kind of ugliness which was so tough on grad students.
I had known Sam Caldwell, who was somewhat into relay circuitry, from the very earliest post-war days. And I was treated very well indeed by Gordon Brown, then head of the extremely influential EE department; he nominated me for the Appalachian Mountain Club, and gave me good advice about how MIT really worked - not at all the way it looked to most of the industrial types. Forrester knew how, of course, but you didn't talk to Jay about MIT politics (or houses, or sports, or even why Whirlwind didn't get some IBM gear).
I had had a steady flow of visitors from overseas at the Watson Lab, and this had slightly tapered off in Washington. My little shop was not the unique drawing card the SSEC had been, and English friends especially tended to go up to the Bureau of Standards (then still on Connecticut Avenue) and watch the sweat running down Sam Alexander's countenance as he tried to make SEAC work, plan its western successor SWAC, advise the Census people about their always-a-year-in-the-future Univac, and fend off IBM.
Cambridge was again a nexus. Everybody wanted to see Forrester's baby: the military vectored in their NATO buddies, the hardware types wanted to tinker with Jay's core memory, and the academics were accustomed to stop off on their way up to Aiken's shop at Harvard, which was really now a backwater but of course still a great draw in Bulgaria. I knew more of them, [-154-] at least on the user side, than the rest of the project combined, which certified me to the Caldwells and Browns and Everetts. Jay probably didn't even notice.
About living arrangements: I had moved into a graduate dormitory room near the Charles as soon as Forrester hired me, and cast about with MIT help for permanent quarters. Dorothy was getting ready to move. I found a rental house in Belmont, owned by a Boston University professor going on foreign assignment who wanted someone to take good care of his roses (I displayed pictures of my wife's gardens; it was a drawn-out negotiation). There was a small garage for my prize convertible. And the family had had pets, which my Washington lease had prohibited.
Dorothy and I were doing well. She was delighted to see me go back toward research and Academe. She had forgiven me for tearing her away from her psychiatrist in Manhattan, and indeed after many hours of self-analysis based on the new things she had learned about herself, was trying to be a New Woman. She had cut off her long hair, and wore an attractive short hairdo instead of a chignon. She had bought new clothes, and better furniture for our nice little Washington townhouse.
We had had frank discussions about my frequent extramarital adventures, and as a result she had waved me off fairly gaily when I visited a luxurious maison in Havana on a Cuban excursion we took together in 1951, and later listened amusedly to my tales of the novelties I had encountered. Our sex life, never spectacular, had considerably improved. But after ten years of a remarkably good marriage we needed a new focus, and agreed it should be something we both could love: a pet. We missed Suzy the cat, who had entertained us in the dull war years with her lively litters.
Being scientific types, we started with book research, and moved on to Boston dog shows. Soon we came down to miniature poodles; standards were magnificent, but too large for our small-tent camping and our foldboat, and toys were unimpressive - wrong, but you can't learn everything from books, as I had told Dorothy in Havana.
The Putnams, who were top exhibitors of standard poodles in New England, suggested the Morse kennel on the other side of Boston for good miniatures. We chose a gorgeous little black male, took him home to Belmont two weeks later, and named him Diablotin Déodat de Dolomieu. Diablotin was the kennel name, Déodat means "a gift from Heaven," and DdD was the man for whom Dolomite [rock and mountains] was named. We called the little guy Deo, Day-oh, and he was indeed a focus of our attention and our love for the rest of the marriage.
I remember vividly a drive up into the White Mountains with Charles Goodwin, head of the Maths Division at the National Physical Laboratory outside London, who at that instant in time was Alan Turing's boss. Dorothy was happy, Goodwin was delighted - and Deo was ecstatic.
The beautiful but mistreated Buick of the mid-Forties had been replaced by a little Plymouth station wagon which pioneered the idea of folding up the rear seat for extra space - perfect for our camping and boating, but dull to look at. When things were at their best in Washington I had traded the wagon for a superb Chrysler New Yorker, as big as a house (and almost bigger than the Belmont garage). It had the brand new hemi [hemispherical cylinder [-155-] head] V-8 engine, power brakes and steering - then also new - a power-operated convertible top, and on and on. It was by a city mile the fanciest car ever parked outside the Barta Building, where WHIRLWIND I lived. Everett and Adams had very ordinary transportation, the Perlises and Carrs rode bicycles, and nobody knew how Jay got to work. A chariot of fire, the youngsters would have guessed; I would have voted for a deep subterranean passage.
It was a smooth and lovely interlude. Déodat and Dorothy and I were as happy as could be. The work was rewarding, although I hadn't produced anything earth-shaking in my first few months. Gordon Brown had steered me gently toward trying for an academic appointment in 1954, perhaps in his department (mathematics was foreclosed by Nasty Norbert, and there was at that time no prospect for space science or astronomy slots). I saw clearly that MIT policy decisions came from the apparatchiks, who were all professors. And I had a very good Ph.D., unlike Forrester, who had been swept away from one by war work. It might mean a slight cut in salary, and as had been my custom since boyhood, I was spending all I made and then some. But Dorothy wanted to go back to work.
Then the heavens opened. As I have mentioned already, I had advised several close friends about whether their outfits should order Defense Calculators from IBM. That kind of informal contact had continued, especially at conferences and by telephone, while I built up the Washington bureau and after I went with Whirlwind - or the Digital Computer Laboratory, as MIT had begun to call it. In fact, I had had almost all my information about how orders were coming in from such people, rather than from Hurd.
The first paying job that ran on the SSEC had been for Hans Kraft of the General Electric Large Steam Turbine and Generator Department, who was continuing to struggle with Mother Nature. She did not want him to find out how compressible, near-turbulent flows went in a rotating lattice of airfoils [turbine buckets and blades]. He had tried using the giant GE differential analyzer, masses of punched card machines, fancy applied mathematicians, and the SSEC; in later years he would go out to Hanford and try hundred-decimal calculations on their IBM 702. No matter where he pushed in, the problem bulged out somewhere else. Clearly Mother Nature knew how to solve it: squirt lots of steam into actual buckets, and watch very closely. Such experiments cost the earth - more than renting a Defense Calculator, for sure. So I had advised Hans to get in line for one.
General Electric was a valued IBM customer - the largest in the world at that time except for the U.S. government and the Bell System. Kraft had introduced me to one of the other major engineering operations of the company, which designed and built jet engines, and which had turbine problems and compressor problems and combustion problems and control problems. His counterpart was one Bert Sells, whose department was located in Lynn, Massachusetts, and who operated a CPC I in nearby Boston.
Sells had told me the headquarters of the engine division was to move to a huge factory just outside Cincinnati, leaving the test cells and much engineering behind in Lynn. He confirmed that the jet engine operations needed computing power even more than the turbine people in Schenectady, and he and Hans planned to recommend that the two groups share a Defense [-156-] Calculator - by this time scheduled to rent for $15,000 a month for one shift, and $30,000 or more around the clock. That wasn't a trivial expenditure in 1951 even for giant General Electric.
When they told me this I said, "Your big problem will be staffing. You will need dozens of programmers, and clever numerical analysts, and experienced supervision. None of that will come easily in Schenectady or Lynn, and it would be a real wowser in Ohio." Sells was unhappy; he was pushing for Cincinnati, knowing how conservative Schenectady was, and how union-ridden and space-limited Lynn was. Also he had been thinking of, oh, ten people. I showed him the list of Backuses and Codds behind the scenes at the SSEC, and the list of Sheldons and Kubies in Hurd's New York shop. "When a Defense Calculator runs three shifts," I said, "you will need a hundred bods, and you'll have to train most of `em on the job." "Jesus, Herb," he said, "I have trouble keeping my CPC busy one shift." "Sounds like you aren't the right guy to run the new machine," I said in my usual diplomatic way.
I turned to Kraft. "Hans," I said, "you remember Paul Herget from the Endicott seminars. He heads the University of Cincinnati Observatory, has just come back from a hitch at the Naval Observatory, and must be itching for a fancy machine. He'll never get even a CPC through the university. And he's terrific. Only thing is, he suffers from the Wallace Eckert syndrome." "You mean he will want to do a lot of astronomical calculations," muttered Kraft. "Watch him," I said.
I was sitting in my cramped but pleasant little office in the Barta Building one morning, thinking about ternary arithmetic, and also about foldboating with the Appies next weekend, when the phone rang. "How about lunch at the Somerset with Hans and me?" asked Bert Sells. Very fancy for lunch in those days, was the Somerset - and I hadn't talked to Kraft or Sells for months.
"GE has ordered a Defense Caculator, Herb," said Sells. "We expect to get one of the first ones, and we are indeed going to put it in Ohio. The costs will be shared with the turbine division, but my department will run it. We're in a new laboratory building; there's lots of space, and good facilities for engineering-type people. The suburb is called Evendale, and we already have six thousand employees there on the manufacturing side."
"Wonderful!" I said. "Did you ask Herget?" "We sure did," said Kraft, "but he insisted he had to have time on the machine to do asteroid orbit calculations, and fancy celestial mechanics later on. We didn't even try to get that past our bosses. Four or five years ago, maybe - right now we have to show results for all that rent. God knows we have enough problems waiting."
"Herb," said Sells, "you got us into this. You've got to come down to Evendale and run the damn thing." "I'd love to, Bert," I said, "but I've only been here five months. I can't possibly do it." "We've already talked to Forrester," said Sells, "and he says you can go any time." Gulp!!
Well, it turned out I wasn't all that unvaluable to Jay after all. General Electric was funding, from Lynn, a very important gas turbine research facility at MIT, and had been doing so for several years. When Sells persuaded his bosses to ask for my head on a platter, like that other prophet John the Baptist, my small current contributions and possible future academic value were shoved briskly aside.
[-157-] I liked MIT. I liked Boston. Dorothy and Déodat liked Belmont. But I was swept away, and not unwillingly. I would be a line manager in an engineering company, the greatest one in the world. I would be using my contacts in rocketry and in the tin airplane business. And I would have my own great big wonderful computer. Also a little green demon kept whispering in my ear that I would be poking IBM in the eye with a very sharp stick.
I knew where the bodies were buried at 590 and in Poughkeepsie. I was good friends with the Rochesters and the Bill McClellands. I knew more about the Defense Calculator than any other customer. All that was to the advantage of my new employers - and my new employers knew it. They treated me well from the very beginning; I soon learned to call GE "Generous Electric." I got an immediate raise to $12,000 and promises of frequent reviews. I got a great move, much to Dorothy's relief; she had been treated well when we went down to Washington in Mary Noble Smith's museum van, but not so carefully when MIT moved us to Belmont.
On the other hand, I was about to install the very first giant computer in the General Electric Company, the first in the aircraft engine business worldwide, and the first in the American middle west. I knew very little about my new employer, and nothing about Cincinnati. Well, as Napoleon said, I was well positioned to attack in any direction!
I was hired in as a first-level supervisor, and handed the Numerical Analysis Unit of the Aircraft Gas Turbine Development Department. That consisted of a serious young mechanical engineer named Gene Gettel, four attractive women with math degrees who were listed as subprofessionals, and an underutilized Model I Card Programmed Calculator installed in a back room in the office building attached to North Station in downtown Boston.
A few minutes conversation made it clear Gene and the girls were the right stuff. I swept them up - none of them wanted to move to Ohio - bargained briskly with a bemused IBM salesman for two CPC IIs in Evendale (he said the backlog was six months), turned in the Model I, and headed for Evendale. GE had a specialist IBM interface which had handled the big-machine order; I told the people involved on both sides that George Richter would fix up my Model IIs if they could reach him: "If you have trouble I'll call him myself; just let me know." The expression had not been invented yet, but I was On A Roll.
In Chapter 16 you will encounter
(in order of appearance):
Pratt and Whitney almost as dominant in its field as IBM
Westinghouse, Allis Chalmers, Curtiss Wright and GM Allison also-rans
Rolls Royce they made jet engines too, in Britain
GE AGT Division 15
GE research a tradition going back beyond Steinmetz; far deeper than IBM
GE management practices organization and decentralization and a lot more
Watson Senior 01
Evendale Numerical Analysis Unit there were exactly six levels above me
Frank Warner he smiled gently at my predictions and my needs
CPC II 13
IBM 701 there was a new name for the Defense Calculator
Stan Farwell my Very Own Friendly Local IBM Sales Representative
AGT Development Department several buildings, and acres of outdoor gear
Dave Cochran the AGT DD general manager, young and imaginative
Jim LaPierre AGT Division vice president, later Number Three of all GE
Jet engine simulator a big analog, for Wright Field; I was skeptical
Marion Kellogg after Evendale she became GE's first woman vice president
Women computer professionals I broke the ice in GE, with Marion's help
Ralph Cordiner president of GE; I forgot to call him by his first name
Hans Kraft 09
Hopper and Blanch and Maria Mayer, and Dorothy I reserved Marie Curie
Allen Keller the Medium Steam Turbine Department systems expert
Gerhard Neumann right down the hall, the world's best engine designer
Walt Ramshaw 10
Bruno Bruckmann a graduate of PAPERCLIP, he had the Prussian saber scars
Union relations we moved one CPC across town to be ready for a strike
The AGT Fair local TV showed our computer working at the Fair Grounds
Programmer selection "Do you read ASTOUNDING? And play chess? Great!"
Jack Hughes he was to run the GM Research 701, but I was lucky
The basement of Building 300 when the Ohio rose, I wanted to be safe
The open shop when customers wrote the programs, usage skyrocketed
Charge-back I invented my version to make expansion easy, and it worked
John Lowe and Chuck Baker 11
Bemer and Bozak and Amaya they averted a disaster at Lockheed
Harold Smiddy he propounded the GE company philosophy
Cincinnati media I used the ENQUIRER and radio and TV and service clubs
The GE MONOGRAM it reached every manager, and turned up Dan McCracken
The 701 Installation Manual distilled by Palmer, Haddad, Hurd et al
The Secretary of the Air Force his son Don worked for Applied Science
A Magic Pushbroom swept me into GE section managership
[-159-] The jet engine business was wonderful. It put you at the very cutting edge of aerodynamics, combustion theory, mechanical engineering, and controls technology. A host of bright, vigorous men - and a few women - were introducing new ideas and materials, and otherwise increasing the number of design parameters. The market was enormous, and still growing. There were a dozen competitors worldwide, led by Pratt and Whitney in Connecticut, and including such powerhouses as Westinghouse, Allis Chalmers, Curtiss Wright, GM's Allison - and overseas, Rolls Royce. Smaller engines were being built by AiResearch, Allison and others. General Electric was a relative newcomer, but already the challenger all through the spectrum - the tough new kid on the block.
Moreover, the competition was honestly technical. Sure, the salesmen fanned out in the military and across the airline spectrum. There was favoritism; there were deals; no doubt off in the bushes games were being played. But center stage, the outfit that produced the most thrust for a pound of jet fuel got the business. There were considerations of durability, of serviceability, of company resources and speed of delivery - and price counted for a lot. But allowing for those ancillary matters, the best engine was the winner.
I was soon told that Pratt and Whitney and General Electric were swinging six hundred million dollars worth of business back and forth each year, almost entirely because of performance. Move to a better manufacturing facility? Sure! Build a $40 million dollar test rig? Right away! Put in one of those big IBM computers? Why not? Hire a bearded ex-astronomer from MIT? Get two, if they have 'em!
It wasn't chaos. It wasn't war time. GE was a carefully organized and closely managed company, very much larger than IBM and with a long tradition [-160-] of genuine research - Steinmetz decades ago, and Nobel prize winners; far beyond what Watson Senior had been able to do. But the engine boys were hot. They could move much more quickly than the lamp bulb boys or the dishwasher boys or even the steam turbine boys. Company rules said you had to have permission of the board of directors to buy land or build a major structure. That applied to the engine division too - but its requests got approved awfully fast!
I knew how strong GE was. Kraft had shown me the huge turbine shops in Schenectady, where the rough forging for a big generator rotor cost $150,000 and could be ruined by one wrong tool setting, on a machine so big the operator had to climb a tall ladder to look at how the cut was going, and might take his lunch bucket along. I knew the jet engine test cells in Lynn could run only at night because they took most of the power of the area, and that attempts had been made to anchor decrepit naval vessels in the nearby river to supply daytime steam power (and reduce the complaints about noisy post-midnight testing).
Even so, Evendale was a revelation. There were literally square miles of buildings, built at the end of the war for piston engine production. Plant layout was a major operation. Intraplant transport was a major operation. Security was a major operation. Raw material storage was by tens of acres. And they told me most of the division was still up in Massachusetts!
My little band was plopped down in a new-building area across the road from the manufacturing operations. Green grass, nicer cafeteria facilities, snappier guard service to handle the hordes of salesmen calling on the hundreds of buyers; it was industrial but attractive.
There was no art collection, and no Mary Noble Smith to choose curtains and carpets; three years later I put tropical fish and green plants and Herman Miller pretties in my own computer building, but it all disappeared as soon as I left. There was a definite GE culture; I was prepared for it; I was very well treated within it. But The Old Man had been something special, and I often missed him.
I soon found there were organization charts, even if there were no oil paintings. General Electric was an open book, or tried to be. You could trace authority down from the very top through your group, your division, your department, your section, your subsection, your unit - to you. There were tests when you came aboard, entrance interviews, performance (salary) reviews, appeal procedures. And across the road, unions!
One great thing I discovered: the official printed multi-color organization chart (always a little out of date, alas!) said explicitly that while authority flowed down, and via rigidly prescribed channels, information could flow upward and across and leap from block to block. That was why Sells and Kraft could help each other without getting permission from something like five levels of managers above Bert and four levels down to Hans (they were in different groups), or vice versa.
In Olympian IBM it had been much different; you asked yourself, "Would Mr. Watson want me to say this (or do this, or order this)?" If you thought he would, you did it. If somebody upstairs thought he wouldn't, you were in deep trouble, unless someone further upstairs thought he would - in which case Mr. [-161-] Part Way was in the soup instead, like the vice president who guessed wrong about those WCs at the Watson Lab. In GE you could ask without risk - well, with much less risk.
And if an important customer from the American Airlines engineering department wanted to eat prunes in the conference room, you could share them with him unabashed!
Almost anything - I mentioned real estate deals as an exception - could be done with the approval of two levels of management: your direct boss, and his boss. When I piled up a racing car years later at Willow Springs, and woke up debearded and alone in a hospital two hundred miles from my GE troops in Santa Barbara, it took one call from me to my then boss, Tom Paine, and one from him to his division general manager back in Schenectady, to put me into an air-conditioned Cadillac ambulance and on the road to a hospital where the gang could drop in for advice. So the insurance didn't cover it? Two levels of GE management did.
Well, to work! The Numerical Analysis Unit found itself in basement space in a laboratory building. It was cavernous; chosen to hold the Defense Calculator when it came. There were no windows except into a big areaway through which the pieces of the computer would be lowered. We were promised ample office space on the second floor, but there was no elevator. I emphasized that we were going to be one great big huge bunch of guys and gals; my boss, a veteran GE mechanical engineer named Frank Warner, looked at my little huddle of reluctant Bostonians and smiled gently.
His main worry was not floor space, but floor. I told him there would be hundreds of thick, sensitive signal and power cables running between the boxes of the computer when it arrived - in seven months, we thought. The idea of raised flooring was unheard of in our tiny trade, but of course I had seen Watson order it for the beautiful SSEC room in 1946. "No," said Frank, "the ceiling is too low already for the air conditioning you claim you are going to need; a foot less is out of the question." "We'll have to cut trenches with a diamond saw when the layout is finalized," I said. "Where do you want me to put the two CPCs while that is being done? They can't run in cement dust." "Hmmm," said Frank, wincing.
The first of the two CPC II systems got to Evendale before my troops (but not before me; I was indeed on a roll, and would have come by Concorde if the service had been available. Dorothy and Déodat were packing). Six months? Not for a 701 customer who knew George Richter! The power supply wasn't quite right, but the facilities people fixed it in hours; Evendale Plant was on a roll too. I thought we could get along without extra air conditioning, if the ducts were opened up. "Just call us, pal," said the Evendalers.
Note the "701." My first visitor, even before the CPC arrived, was of course my Very Own Friendly Local IBM Sales Representative and one thing he told me was the new name for the Defense Calculator: IBM Type 701 Data Processing System. Coooo!
About my VOFLIBMSR: he was a great piece of luck for me. Not entirely intentionally, perhaps; although Applied Science was trying to place special, um, agents at all the early 701 sites, and I turned out to be Number Six, I was [-162-] treated differently. This chap was not a techie; Cuthbert And Co. claimed they realized I could do without one. My guesses were, a) they were short-handed, b) they were afraid I would steal away or at least pollute one of their innocents, and c) they could stick a sharp stick in an eye as well as anybody.
Anyhow, my man was one Stanley Farwell, nephew of a very famous early-Watson executive who some years later left IBM to become the Number Three at RCA. Fred Farwell, whom I never met, was about at Red LaMotte's level in postwar IBM, and had helped Stan find a place in the sales force when he got out of hospital in the late Forties.
Stan was very tall, lean, a wolf with a gentle smile. He had a glamorous society wife who was a little lost in Cincinnati, and a prewar Lincoln Continental that he was lovingly restoring - twenty coats of hand-rubbed white lacquer, as I remember. He was notorious in his IBM circles for taking leave in the summer to go with the circus, or to a logging camp, to work off his tensions. He took great pains to preserve his Continental while away, but I gathered later, not so much to preserve his marriage.
He got away with this because he was almost the last survivor of a Ranger battalion that was wiped out in Normandy. He had been reconstructed physically and mentally, but a lot of Good IBMers like Watson Junior and Garland Briggs were still watching over him. The assignment to help me put in the 701 was a way of getting him away from the rather grinding sales routine of local offices.
He had great contacts back in Watsonland, but he was irreverent about them. He had heard about George Richter, and was impressed with how my CPCs had appeared out of nowhere. I liked him immediately; in spite of the stories about his problems, which I heard from my VOFLCE, Very Own Friendly Local Customer Engineer, he was always around when we needed him. He brought me all kinds of technical gossip from his trips to 590 and Endicott and Poughkeepsie (and was exceedingly sharp about it for a non-techie), and smoothed the commercial relations with a somewhat bewildered downtown Cincinnati IBM office. And he was a warm if somewhat injured human being.
My new boss was in charge of the laboratories and test facilities of the AGT (Aircraft Gas Turbine) Development Department. That meant he had all of the building we had landed in, several acres of outdoor machinery out back, and a sizeable number of engineers en route from Lynn. He reported to the manager of engineering, Neil Burgess, who reported to the department general manager, Dave Cochran, who reported to the AGT Division vice president, Jim LaPierre. The division had 14,000 employees in Massachusetts and Ohio, plus sales and service reps all over hell and gone: in Korea and on aircraft carriers, at a dozen Air Force bases, and at several airline maintenance centers.
LaPierre had two departments in Lynn; one doing small engines and the other, controls and fuel pumps and such. In Evendale he had a sizeable prototype-engine department, and Cochran's outfit. And divided between the two sites he had a huge manufacturing department with over nine thousand union employees and foremen and managers. The division did nearly a billion 1953 dollars a year, made a useful profit, and was growing like mad. In the [-163-] late Seventies it was a group, and had gotten up to 40,000 people. It must be even larger today.
I'll write more later about the General Electric management philosophy, and how it interacted with the three great challenges of nuclear engineering, jet engine manufacturing and computer marketing. One of its tenets was that each department had its own manufacturing, engineering, finance, marketing and employee/community relations sections - and its own bottom line. In Cochran's case the manufacturing was small and specialized, with emphasis on new materials and processes. And marketing was mostly market and product planning. Engineering, though, had nearly a thousand bods, and cost the earth. Adding a 701 was small potatoes - at least, at first!
Our building, for instance, housed a big electronics lab that was designing, and testing components for, a giant analog jet engine simulator. GE was to build this monster at Wright Field (just a few miles away). At the other end of the ground floor was a suite of bench-science labs doing combustion research, and they also had flame tunnels and such out back. Upstairs was a vast open office space, partly loaned out to a preliminary design group which was not under Frank - another part of the Engineering Section. I was offered an adjacent area, and rather shrewdly said we would need it and a lot more "when our Big Computer comes," but wanted to be near our new-model CPCs in the basement until we got used to them.
I went up to the main building and introduced myself to the personnel crew. Their boss was at subsection level, and tended to resist my romantic yarns about the vast hordes of men and women I was going to recruit. But he reported to a woman, unheard of elsewhere in GE but of course familiar to me from IBM experiences. Her name was Marion Kellogg, and in later years she became General Electric's first woman vice president and made the cover of BUSINESS WEEK.
She saw that my curious ads got printed, that my curious prospective employees got tested and "processed," and th