Columbia University Computing History   

L.J. Comrie

Leslie J. Comrie

Leslie J. Comrie, Ph.D, (1893-1950): astronomer and pioneer in mechanical computation, born in Pukekohe, New Zealand, and educated at Auckland University College, University College London, and Cambridge University, where he received a Ph.D. in astronomy [3]. Photo: 1932, from the Comrie obituary in Biographical Memoirs of Fellows of the Royal Society of London. Like his American contemporary, Columbia University Astronomy Professor Wallace Eckert, Comrie was a pioneer in the application of punched-card machinery to astronomical calculations and the production of astronomical and mathematical tables: the first scientific use of these machines, which had been designed purely for business use.

Comrie pioneered the teaching of numerical analysis in 1923-25 at Swarthmore College and Northwestern University in the USA, and then joined His Majesty's Nautical Almanac Office in Greenwich. England, in 1926 -- the same year Eckert joined Columbia's Astronomy faculty. Comrie was Superintendent of the Almanac Office from 1930 to 1936 (where, when Comrie arrived in 1925, "the Nautical Almanac was computed by retired Cornish clergymen with long white beards, using dog-eared 7-figure logarithm tables" [Tee 1989]); Eckert was Director of the US Naval Observatory's Nautical Almanac Office from 1940 to 1945.

Comrie was the first to turn punched-card equipment to scientific use: computation of astronomical tables by the method of finite differences, as envisioned by Babbage 100 years earlier for his Difference Engine, but not practical until the appearance of reliable calculating machines. Differencing was used because the early punched-card machines were capable only of addition and subtraction; not division, multiplication, exponentiation, roots, trigonometric functions, etc. Comrie had already been using this method for many years with hand-operated desktop calculators; an excellent review and history of differencing and its application to various types of calculating machines is given in Comrie's 1928 Brunsviga-Dupla article (publication list below).

Comrie at work, Photo: [103].

Comrie's April, 1928, article On the Construction of Tables by Interpolation describes at length a laborious and error-prone procedure for interpolating numeric table values by desktop calculator (right) and then, in the last few pages, describes how to do the same thing using punched cards, a tabulating machine, a duplicating punch, and a sorter with far greater speed, but then goes on to conclude that since Hollerith equipment is so expensive, it is rarely practical -- as is often the case with new technology. Nevertheless, as he says in his 1946 MTAC article:

The first scientific application of the Hollerith—in 1928—was to the summation of harmonic terms, or, in other words, Fourier synthesis. In this way the principal terms in the motion of the Moon from 1935 to 2000 were computed. The ordinates of the various harmonics to be compounded were taken from E.W. Brown's Tables of the Moon, and punched on to half a million cards. These were brought together in ever-differing groups; because the periods of the different harmonics are incommensurable, no card ever had the same partners twice, although it may have been used 20 times in the 65 years covered. Something like 100 million figures were added in groups and group totals printed in the course of seven months. I showed this to Brown in the summer of 1928; he had done a great deal of this synthesis himself by hand, and I shall ever remember his ecstasies of rapture as he saw his figures being added at the rate of 20 or 30 a second. I think I am right in saying that the enthusiasm with which he described the process on his return first stimulated W.J. Eckert, the leading American pioneer of these machines for scientific calculation.

In his May, 1932, article, The Application of the Hollerith Tabulating Machine to Brown's Tables of the Moon, Comrie describes his punched-card methods in far greater detail, with paragraphs given over the function and operation of each component -- card, punch, sorter, tabulator, printer -- plus plates showing each. In this article, for the first time, we have a description of "programming":

Automatic Control.—This feature is of the utmost importance in the present application. When the cards corresponding to any one date have been added, the feeding must cease while the total is being printed, the counters must be cleared and then the feeding must be resumed. This sequence is performed automatically, without any attention whatsoever on the part of the operator. (Comrie goes on to describe how the tabulator plugboard has been wired to "compare" a key field on a pair of cards and handle the sums and reset the counters when they differ.)

Further on he says:

A Hollerith installation was used in H.M. Nautical Almanac Office in 1929; actually punching was started six months before the arrival of the sorter and tabulator, as it was necessary to punch 20,000,000 holes in half a million cards. ... The sorter and tabulator, which may only be hired, cost about £2 a day. The cost of doing by hand what was done on the machines has been estimated at £6000; this estimate probably errs on the low side. The cost with the machines was certainly less than £1500.

Thus in the four years since his 1928 article, punched-card computation had become cost effective. In the following years (as can be seen from his publication list, below) Comrie became strong advocate of the punched-card method. Like Eckert, he was preoccupied with saving steps, particularly error-prone and time-consuming ones like copying the results of one calculation to the input of the next. Comrie, unlike Eckert, was happy to use later-model desktop calculators (rather than the more expensive punched-card machines) when they acquired the ability to store results in registers. Of course this was purely a matter of finance; Comrie did not have Thomas J. Watson as a patron, eager to shower him with expensive equipment at no cost!

Comrie was also a strong proponent of using standard, "off-the-shelf", unmodified hardware (calculators or punched-card equipment), rather than modified, one-of-a-kind, or special-purpose machines because he wanted the methods he developed to be useful to others. Eckert, on the other hand, was constantly prodding IBM to modify its designs and to produce special one-off machines for his lab, and IBM was happy to comply. Comrie expressed his feelings in 1946 (same MTAC article), the age of ENIAC, BINAC, EDVAC, EDSAC, ORDVAC, MANIAC, et al.:

The adaptation age has overlapped the beginning of another—namely that of the construction of special machines. I have sometimes felt that physicists and engineers are too prone to ask themselves "What physical, mechanical or electrical analogue can I find to the equation I have to solve?" and rush to the drawing board and lathe before enquiring whether any of the many machines that can be purchased over the counter will not do the job. ... What would not be right in India, where subsistence requirements may almost be measured in handfuls of rice, can easily be justified if computers cost $2000 to $3000 a year plus supervision, administration and costly accommodation. In other words look at what has been provided by and for the industry before you leap to the adventure of a special design.

Eckert was of the opposite persuasion and, through his close contacts with IBM, relentlessly drove modifications and improvements to existing machines to adapt them to scientific use. In 1933-34 he took Comrie's method to the next level by designing a calculation control switch for the machines in his laboratory, thus automating complex series of calculations with a minimum of card handling: this was the first realization of automatic sequencing (the aim of Babbage's Analytical Engine), and a giant step towards modern computing, a step built upon the foundation laid by Comrie.

Comrie and Eckert corresponded for decades and collaborated on many projects [88]. Shortly after the War, Comrie visited Watson Lab.

L.J. Comrie was elected Fellow of the Royal Society shortly before his death in 1950. A lunar crater (23.3N 112.7W) and an asteroid bear his name, as does a computer lab at his alma mater, the University of Auckland (left).

Garry Tee of the Department of Mathematics of the University of Auckland adds, "Immediately after the end of World War 2, LJC went to the USA for 3 months, examining developments in computing technology. (The fact that he was able then to get transport so promptly proves that he had very influential friends). He presided at a conference at M.I.T. where some of the wartime computing developments in the UK and USA were described. He gave to Howard Aiken a copy of Charles Babbage's fascinating memoir Passages From The Life Of A Philosopher (London, 1864), inscribed 'From one admirer of Babbage, to another'. But when LJC examined the Harvard Mark 1, he asked where the if-statement was! Aiken gulped and blushed: it was nearly a year later before Aiken got hardware for the if-statement working! The Harvard Mark 1 was thereby converted from a very large calculator, to a very very small computer.

"Herb Grosch invited LJC to attend the 1945 Fall Meeting of the USA Institute of Navigation. I have posted to you some of my papers about LJC, with a copy of the photograph which Herb gave to me of the conference banquet in the Hotel New Yorker on October 26 1945, with LJC seated alongside Herb:"

Photo courtesy of Garry Tee. Click to enlarge.

Comrie and Grosch are near the right edge of the photo, towards the top. Grosch is the one with the goatee; Comrie is to his right. In Grosch's words: "I can identify ten by name, several more by generalities. Around Comrie are [Grosch's wife] Dorothy, goatee, LJC, [Gerald] Clemence, Ralph Haupt (co-author of [Eckert's] c[ard]-o[operated]-typewriter paper [77]), another [Naval Observatory] guy. Third from left on the dais is Sam Herrick. Moonface near camera is Charles Smiley of Brown, editor of Sky and Telescope. There are Hahvahds, some in uniform. No [Fred] Whipple. No Dorrit Hoffleit. Famous Cdr. Weems is on dais also."

Grateful thanks to Garry Tee of the Mathematics Department of the University of Auckland for his many contributions to this page, and to Lindsay Rollo of Wellington, NZ.