The Book Nobody Read

by Owen Gingerich

  The estimated bid was very low, but I asked Quaritches to go higher on my behalf if necessary, because, as I explained, Maestlin's Ephemerides is so rare that I had never seen a copy pass through the market. Shortly after the auction Rick Watson called from Quaritches to say that there was good news and bad news. I had won the ephemerides for less than my bidding limit, but it was a rat-nibbled and incomplete stack of leaves with a problem. "There is something else mixed in, about astronomical hypotheses," he informed me.

  "That's very interesting," I responded. "Maestlin held forth in a Tubin­gen University disputation on astronomical hypotheses, but that little book is really rare. It would be fantastic to get it."

  I suppose such occasional public disputations provided the chief intellectual entertainment for university students in those centuries. The printed versions, often not particularly attractive typographically, border on ephemera, and very few copies generally survive. Rick explained that it didn't look like a disputation, and that in any event there seemed to be a letter from Kepler dated 1595 in it. He couldn't see my jaw drop. I told him that when Kepler was a not-yet-famous high school teacher in Graz, he had written a fan letter to Nicolaus Raimerus Ursus regarding his book on the fundamentals of astronomy, and that Ursus subsequently printed it in his De astronomicis hypothesibus. "The book was such a fierce attack that Tycho brought legal action to have the copies banned and burned," I added, "so it's terribly rare."

  "That must be it," Rick said. "The title page is missing, but Ursus' name is right here at the front."

  The staff at Quaritches hadn't checked out the lot in advance and so were completely blindsided by my extraordinary good luck. Clearly, I had ruined the week for them. But Kepler's naive enthusiasm, which had generated the admiring letter to Ursus, had pretty well ruined a year for him. He had desperately needed a new position after the Catholic rulers of southern Austria expelled all the Lutheran teachers from Graz, leaving as his best (and perhaps only) possibility to go to work for Tycho Brahe, who by 1599 had moved from Uraniborg to Prague under the patronage of Rudolf II. In fact, Kepler didn't know that Ursus had published his letter—he hadn't even kept a copy of it, and Tycho took the opportunity to give Kepler his comeuppance. The arrogant nobleman was not about to stoop to replying to a former swineherd, so instead he assigned the task to young Kepler.

  Kepler took up the challenge in a typically Keplerian way. He didn't simply parry and thrust each of Ursus' jabs. Instead, he probed much deeper, into the meaning of astronomical hypotheses and the basis for deciding between them. The result was a serious philosophical treatise, unlike the polemical tracts that characterized many of the Renaissance controversies; in fact, as the translator and commentator Nicholas Jar-dine put it, this was the birth of the history and philosophy of science. Kepler framed his account according to the classical rhetorical rules for a judicial oration, as he had learned at the university. Within these rigid constraints he argued that the astronomer must seek hypotheses that not only predict the phenomena accurately but are also physically plausible. These principles served well to guide his own brilliant researches.

  Significant as Kepler's response was, he never saw it published. Ursus had died in 1600, shortly after Tycho had arrived in Prague and started his lawsuit. And by the end of 1601 Tycho, too, was dead. So Kepler simply stashed away the four chapters comprising his tract, where they remained among his legacy until finally, in 1858, they were printed in a comprehensive multivolume edition of his works. But there they continued to sleep for another century until Jardine took a careful look at them, and with a quotient of sideline cheerleading from me, eventually produced a scholarly analysis and English translation. While he was at it, he wrestled with the Bear's somewhat coarse Latin and included the relevant passages in his commentary.

  Because of Nick Jardine's involvement with Ursus' text, I couldn't resist phoning him to share the excitement of my astonishing trophy from the Bloomsbury Auction. "My God!" he exclaimed. "You've just acquired the third known copy!"

  "No, it's rare, but it can't be that rare," I protested. "There have got to be more copies than that."

  It was more than ten years before another copy appeared on the market. This time Rick Watson was determined to capture it, and when he did, he researched the number of extant copies. He was able to locate only eight other copies, two in the United States (including mine) and six in Europe. I've been able to locate only three more. Tycho's lawsuit had been amazingly effective.

  NORMALLY, BOOKS don't disappear so dramatically. Galileo's Dialogo, the book that got him in trouble with the Inquisition, was published in an edition of a thousand, and despite the ban by the Inquisitors, it remains one of the most common of the great scientific classics. Apparently, its listing in the Index of Prohibited Books simply made it more apt to be preserved in the seventeenth century. By the same token, Kepler was worried about sales in Catholic countries when his Epitome of Copernican Astronomy was placed on the Index, but a correspondent from Venice assured him that his book would be all the more sought after.

  One of the most spectacular attritions concerns a famous English translation of Copernicus' cosmological chapters—famous because its version of the heliocentric blueprint has been so widely reproduced. In 1576 the English astronomer Thomas Digges took over his father's perpetual almanac, A Prognostication Everlasting, which had already come out in six previous editions, and added to it an English version of chapters 9 to 11 of the first book of De revolutionibus. A large folding diagram of the heliocentric system showed the starry frame not in a spherical shell but scattered out in all directions. The caption stated, "This orbe of stars fixed infinitely up extendeth hitself in altitude spherically and therefore immovable," which was in itself a remarkable argument for the fixity of the stellar matrix. Of course this was bad news for the empyrean, the home of the blessed, for traditionally heaven had been placed right outside the shell of fixed stars. Digges's solution was ingenious. To the caption he added, "The palace of foelicitie garnished with perpetuall shininge glorious lights innumerable . . . the very court of celestial angells, devoid of greife and replenished with perfite endless joye, the habitacle for the elect."

  Digges expressly stated that he had included the Copernican excerpt in the almanac "so that Englishmen might not be deprived of so noble a theory." Eventually, I stumbled onto his own copy of De revolutionibus, which turned up, rather unexpectedly, in the Geneva University library in the course of my systematic survey of Swiss libraries. Digges had scarcely annotated it, but he penned a telling remark on the title page: "Vulgi opinio Error" (the common opinion errs) (plate 7b). His comments thus enroll him among a handful of sixteenth-century readers who accepted the heliocentric doctrine.*

  Thomas Digges's diagram of the heliocentric system, with the stars spread out toward infinity from his A Prognostication Everlasting (London, 1592), author's collection.

  Thomas Digges's version of A Prognostication Everlasting proved very popular and came out in eight known editions between 1576 and 1626. Typically, only two or three examples from each edition survive, so there could easily have been other editions with no known copies. Just looking over the list, I would guess that there were editions around 1581 and 1588 that have no survivors. For such popular works the production of a thousand copies seems typical, so we can estimate that about 10,000 copies were printed by 1626, yet fewer than 40 examples exist today. This is a survival rate of less than half a percent.

  What happened to the 99.5 percent of those copies? Some pages were no doubt used to polish boots or candlesticks, but if unvarnished truth be told, the majority probably literally ended up as toilet paper.

  Copernicus' book, on the other hand, was comparatively expensive and became famous early on, so it is unlikely that many copies were deliberately destroyed. In fact, I know of only one specific case. The large Gian Vincenzo Pinelli library, which contained a copy of De revolutionibus, was around 1604 being shipped by sea from Venice to Naples when
it was attacked by Turkish pirates. The robbers, so disgusted at finding only books, dumped thirty-three chests overboard. Twenty-two of them were recovered and later purchased by Cardinal Federico Borromeo for his Ambrosiana Library in Milan. Because today the Ambrosiana doesn't have a copy of Copernicus' book with a Pinelli provenance, we can only conclude that Pinelli's copy went to a watery grave.

  WHEN I FINALLY published my An Annotated Census of Copernicus' De revolutionibus in 2002, it described 276 copies of the first edition and 325 of the second. I am often asked how many copies were printed in the first place. Since no publisher's records remain from Nuremberg or Basel, the best I can do is make an informed guess, starting with the maximum number the Nuremberg printer Johannes Petreius could have printed.

  Some years ago I learned that as a rule of thumb, a single sixteenth-century press could print both sides of a ream of paper in a day, that is, 480 sheets. Each sheet in De revolutionibus contained four pages, and there are just over 400 pages in the book. That would mean at high speed a single press could have printed 480 copies of the book in a hundred days or just under four months. The printing started in the spring of 1542, but was far from complete in the autumn of that year, perhaps because there was some delay in getting the 142 woodblock diagrams cut. The printing was finished around mid-April 1543. Thus more than seven or eight months elapsed for the main part of the printing. Printing a ream of paper a day could have produced as many as a thousand copies during those months. But is the "ream-a-day" rule reliable?

  By and by I became curious about the "ream-a-day" report. Could the press really print both sides of the sheet? Wouldn't the wet ink on the one side create a mess? I consulted with veterans of hand-printing techniques, and I examined the vivid description of early printing given in Philip Gaskell's A New Introduction to Bibliography. Gaskell had been librarian at Trinity College, Cambridge, in the early days of my Copernicus chase, and I shared various news with him. When I demonstrated that the earliest copy of De revolutionibus owned by Trinity, and probably the very copy that Isaac Newton might have read when he was a fellow there, had in the meantime been sold as an imperfect duplicate, Gaskell went after it. He traded another, perfect copy with the University of Leeds, where I had found the oldest Trinity copy. It reminded me of the merchant in the Thousand and One Nights who offered new lamps for old!

  Between Gaskell's book and the advice of the veteran printers, I discovered that the "ream-a-day" estimate was much too low. I learned that the ink had two principal components, the pigment (lampblack or soot) and the vehicle or varnish (such as linseed oil). The paper needed to be damp when printed to get the best impression, or "bite," of the type. In preparation the sheets were individually dipped in water the previous evening and stacked in a heap, generally 250 sheets. One side would be printed in the morning, and the other side in the afternoon before the sheets could dry out. Otherwise, the sheets had to be dampened again before printing on the verso, and the shrinkage could cause problems with dimensional stability. Various precautions, such as wiping off the parchment backing that touched the partially dry ink, prevented offsetting.*

  Two pressmen operated the heavy press, one to manage the handheld ink balls with which he would ink up the type before the sheet of paper was put in place, and the other to slide the carriage with the type and overlying sheet of paper into the press proper and to turn down the screw to make the impression. In a really efficient operation an apprentice would place the fresh paper on the tympan (part of the folding apparatus that positioned the paper over the type), and another would remove the printed sheet and carefully stack it in the growing pile. Later the sheets would be hung up to dry. A well-run printing operation obviously had to have a capacious drying room. With a four-man operation 250 sheets could be printed on a press in an hour, about one every fifteen seconds, so 1,500 double-sided sheets could be printed in a twelve-hour day if the stamina of the pressmen held up.

  Petreius no doubt had several presses, since during 1543 he produced more than twenty other titles (including three sermons by Andreas Osiander, his Copernican proofreader), and probably his typesetters alternated between projects to keep the presses at work even while a particular project had a pause for proofreading. If Petreius had used two presses, he could have printed a still larger press run in the same elapsed time, but we know he didn't do that. Petreius printed two pages side by side on each sheet, and then two more on the opposite side. Two folded sheets, one inside the other, were used to make a signature of eight pages. Petreius had a single alphabet of fancy initials used for the first letter of each chapter in De revolutionibus, a set cut by the Nuremberg artist Hans Sebald Beham. Only once did he need the same large letter twice on the same sheet and there the second came from a different set of letters. But if he needed the same letter in two successive signatures, as he did for the A in the four signatures P through S, the same fancy letter was used each time. In other words, his compositors distributed the type from each signature before they set the next one.* Had two signatures been running simultaneously on two presses, he would not have had the duplicate fancy letter for the second signature. Because of the proofreading, there would necessarily have been a little lag between breaking down one frame of type and getting the next sheet ready to print. Thus it seems quite unlikely that he could have printed day after day at top speed. In seven or eight months of concentrated effort he could have printed as many as 3,000 copies, but in fact the edition was substantially smaller than this.

  Printers working with a sixteenth-century press. The pressman is inking the type while his assistant places a sheet onto the tympan and prepares to fold the frisket over it.

  Johannes Petreius, printer of the first edition of De revolutionibus, from J. G. Doppelmayr's Historische Nachricht von den niirnbergischen Mathematicis und Kiinstlern (Nuremberg, 1730).

  Although we have no records from Petreius' shop, excellent records come down to us from the sixteenth-century Plantin-Moretus Press in Antwerp. Although it was a larger establishment than Petreius' (or at least it produced more titles), it must have been comparable in many other aspects. At the well-documented Antwerp shop, the press runs ranged from two hundred (for subsidized, special editions) to about 2,500. For popular works such as liturgical texts or herbals, 1,250 was Plantin's favorite number. Since paper was one of the most expensive parts of a printing operation, Petreius would not have wanted to overestimate his sales. Clearly, a work as large and technical as De revolutionibus would have required a considerably smaller print run than 1,250. But how much smaller?

  There are a couple of ways to use the number of surviving copies to estimate how many he actually printed, first, by comparing the number of surviving copies of De revolutionibus with some similar book where it is known from the records how many were printed. For example, a huge number—a thousand copies—were printed of Galileo's Dialogo, and a substantial number survive, so that the price of his first edition is much, much lower than for Copernicus' book. The big computer database, the OCLC,* gave eighteen copies of Copernicus' first edition, but thirty-three of the Dialogo. This suggests that about half as many copies of De revolutionibus were printed, that is, a press run of around five hundred.

  Matching up numbers of books in the OCLC sounds easy, but in fact it's not, because that database is so corrupted on this kind of thing. Untrained students all over the country were hired by various colleges to enter the thousands of volumes into the database, and many smaller schools simply have facsimiles of these rare books, not the expensive treasures themselves. As a result, about a third of the purported sixteenth-century copies of Copernicus' book turn out to be twentieth-century reprints. I proposed to compare four early science titles whose press runs were known with four that were not, including the first- and second-edition De revolutionibus. In the end my staff had to undertake a marathon phone campaign to several score libraries in order to purge the list of the false entries.

  My OCLC survey came up with three surprises.
Kepler had collected part of his back salary from the emperor in paper, enough for a thousand copies of his Rudolphine Tables. But the scarcity of that title today suggests that only around 550 copies were actually printed (in which case Kepler sold the rest of his paper to make up for his back salary), or else that a substantial part of the press run remained unsold and was eventually pulped.

  The second surprise concerned the print run of the first edition of Newton's Principia. We know that 750 copies were printed of the second edition, but the best guess for the first edition had been 400 copies. Perhaps I shouldn't have been so astonished to find that there seem to be more copies of the first edition than the second, because the Principia has always sold for only about a third as much as a De revolutionibus. It's clearly a more common book. There is no escaping the fact that more than 600 copies of the original, 1687 edition of Newton's Principia were printed, possibly as many as 750 copies. By the same reasoning, substantially fewer copies of De revolutionibus came off the press.

  Finally, the biggest anomaly of the survey was the extraordinary rarity of Galileo's first astronomical treatise, Sidereus nundus or Sidereal Messenger, which announced his spectacular telescopic discoveries. In a letter of March 1610 to Cosimo de' Medici's personal secretary, Galileo mentioned that 550 copies had been printed. The OCLC survey picked up only five copies, compared with twenty-one for the second edition of Copernicus' book, which probably had about the same number printed. I can only ascribe this paucity to what I call Stoddard's Law: "Bigger books linger longer." (It takes half a dozen Sidereus nundus copies to fill the space of one Dialogo or De revolutionibus.) I remember trying to persuade Roger Stoddard, one of Harvard's most knowledgeable librarians, to bid for a copy of De nive sexangula (The Six-Cornered Snowflake), a booklet that Kepler had put out as a New Year's gift for a distinguished friend, and which is now regarded as a seminal treatise on mineralogy. "It can't be too expensive," I opined, "since it's so thin." Roger groaned and retorted, "Everything you say about it makes it more expensive. Other things being equal, thin books are much harder to find than thick ones."*