For more than a century, a small, leather-bound book has sat collecting dust and attracting little attention in a gray stone library on the corner of Brown University’s Main Green in Providence, R.I. In a library full of old and obscure texts, the 234-page quarto was older and more obscure than most. Its brown, battered leather cover was blank, its title page missing, and its author unknown. Inside, a series of inscrutable symbols filled every inch of the book’s margins: Scrawled in black ink were what looked like a combination of Greek, Hebrew, and some wholly invented characters. Who wrote them? And what do they say?
The only hint came from an unsigned note attached to the book and dated Nov. 11, 1817. It read, in part, “The margin is filled with Short Hand Characters, Dates, Names of places &c. &c. by Roger Williams or it appears to be his hand Writing…. brot me from Widow Tweedy by Nicholas Brown Jr.”
Despite this intriguing reference to the man who founded Rhode Island and brought the idea of religious liberty to the New World, the book languished—until an offhand remark at a small lecture in 2010 led a team of Brown scholars and undergraduates to crack the code, confirm it was written in Roger Williams’ hand, and discover, this summer, his last known work of theology.
On a late fall day in 2010, Ted Widmer, then-director of Brown’s John Carter Brown Library, was giving a talk on Williams’ life and legacy to about 20 members of the Pembroke Club, a group of Brown alums. The attendees were “mostly people with either gray hair or no hair,” says Bill Twaddell, a retired diplomat and member of the library’s board of governors who sat in on the lecture.
At one point, Widmer (an occasional contributor to Slate) mentioned the book and the suspicion that Williams had authored the code in its margins. Twaddell’s ears perked up. Why not scan the code and let computers attempt to crack it?
With the help of Kim Nusco, the JCB’s manuscript librarian, Widmer and Twaddell began recruiting faculty from the fields of mathematics, computer science, comparative literature, and history. But 17th century penmanship triumphed over 21st century technology: The writing was simply too messy for a computer to make sense of.
Manually cracking the code was going to be a slog. The project didn’t neatly fall into any one field, and the academics recruited by Widmer, Twaddell, and Nusco had work of their own to worry about. They decided to give Brown’s undergrads a shot.
In the fall of 2011, two juniors and two seniors—students of history, American studies, and mathematics—signed on to tackle the code during the spring semester. But over winter break, one team member, Lucas Mason-Brown, decided to get a head start. (Full disclosure: Mason-Brown is a friend.)
Mason-Brown studied math and had a particular interest in cryptography. He began with a basic technique called frequency analysis. It relies on statistical truths—for example that e is the most common letter in the English language—to crack the kinds of simple ciphers one would expect from the Colonial era.
It yielded nothing. He then turned to co-occurrence analysis, a more sophisticated technique based on the frequency with which certain letters tend to succeed or precede others. For example, the most commonly occurring pair of letters, or bigram, in English is t and h.
Still nothing. The cipher might have been more complex than expected. Or it could combine elements of the six other languages Williams knew. If that were the case, the uselessness of frequency analysis would be the least of the group’s problems.
Setting his equations aside, Mason-Brown turned to books on early modern shorthand. His reading pointed him to the system set out by John Willis’ 1602 Art of Stenographie. The system was popular in England at the same time a young Williams worked as a stenographer for the noted jurist Edward Coke at the Star Chamber Court.
In the system, symbols stood in for most consonant sounds, and most vowels were left out. The symbols were simple, generally requiring a single pen stroke, making them easier to write than the consonants they stood for. The relative position of the consonant symbols indicated the vowel in between them. For example, placing the symbol for g to the bottom left of the b symbol encoded bag. Move the g symbol to the upper right and you instead had bog.
Guessing that this might be the system Williams used, Mason-Brown tried tweaking his frequency analysis so it applied to consonants only. Thus modified, the analysis allowed him to tentatively match several symbols to English letters.
Next, he turned to the longhand “flags” that dotted the first of three sections into which he had preliminarily divided the code. These were words, mostly place names, that the author had written entirely or partially in longhand English.
If a word was partially written in longhand and partially written in shorthand, he could guess at the entire word and then determine what the symbols in the shorthand portion of the word stood for. For example, Mason-Brown encountered “Meso” written out in longhand followed by a string of symbols. Given the prevalence of ancient place names among the longhand flags, he guessed that the word was Mesopotamia. The string of short-hand symbols standing for potamia gave him a set of precious potential correspondences.
Applying these two tactics in tandem, Mason-Brown was able to build a tentative key. When applying the key to paragraph-length passages yielded intelligible English, he knew his hunch about the Willis system had been correct.
By the start of the spring semester, he had built and confirmed a key of the 28 most common symbols and the letters or sounds to which they corresponded. But this was far less than half the battle. The code was filled with “defectives,” long words given idiosyncratic abbreviations that did not follow the normal rules of the encoding scheme. It was also liberally peppered with pictograms. “Friendship” was rendered with the symbol for F followed by a sketch of a ship. And the handwriting was, as Mason-Brown describes it, “atrocious.”
But the longhand flags offered another possible shortcut. The flags in the first section—geographical terms like “Nov Belgium,” “Mutina,” and “Paphlygonia,”—suggested an expertise beyond Williams’ main fields of interest. The flags in the third section—medical terms including “Hermaphrodite” and “Eunouch” —suggested the same (as well as an unexplained interest in sexual aberration). All looked like keywords. It was possible, the students and scholars hypothesized, that these sections were copied down from reference texts.
With the help of historian Tim Harris, the undergrads matched the first section to Peter Heylyn’s Cosmographie in Four Books: Containing the Chorographie and Historie of the Whole World, and All the Principal Kingdoms, Provinces, Seas and Illes thereof, a sort of 16th-century geographical encyclopedia. With the help of medical historian Hal Cook, they matched the third section to Bartholinus’ Anatomy, a popular 17th-century medical reference. Translating these sections would now be a simple matter of matching the shorthand to the passages from which it was copied.
Meanwhile, in early March, the Rhode Island Historical Society furnished two letters in Williams’ hand that included brief snippets of shorthand. By comparing the idiosyncratic defectives in the samples, Mason-Brown was able to confirm Williams’ authorship of the shorthand in the mystery book, and the academics attached to the project concurred with his conclusion.
By the end of the spring, Mason-Brown had confirmed Williams’ authorship, cracked the code, and translated most of its contents. It was a gratifying, if relatively minor, accomplishment. But the undeciphered shorthand from the middle section of the mystery book still beckoned. It lacked those helpful longhand flags, making it harder to translate, but also offered a glimmer of hope that it contained a different sort of text: original writing by Williams.
This summer, Mason-Brown moved to the Lower East Side of Manhattan with the single goal of translating the rest of the code. He would likely have to work through the remaining shorthand word by word. It was a daunting task: 24 pages of uninterrupted shorthand. The only words written out in English were two names, “Eliot” and “Norcott.”
But in July, the first passage emerged: “[Here is a] a brief reply to a small book written by John Eliot called, “an Answer to John Norcot Against Infant Baptism,” a plea to the parents of the children of Christ. [Argued] from “Acts” and “John” and other [scriptures], written with love.”
Then a week of work yielded this:
The words of the Great King enjoin us to protect the gospel, whose written word [refutes] John Eliot and whose word must prevail over the book of John Eliot. [I hope a] beam of light will appear to you by my labor. I [shall] not weary the reader with a large and onerous discourse. I shall not [let it so that] principles themselves prevail over the written word of God.
It was original theological writing by Roger Williams. And it was previously unknown to history—the first such discovery in decades.
In 1676, John Norcott wrote a treatise attacking infant baptism, the accepted practice throughout the majority of Christendom. John Eliot, a missionary to the American Indians, penned a retort three years later. The text in the middle section of the mystery book was a rebuke of Eliot in defense of Norcott and adult baptism.
Williams also touched on conversion of American Indians, another hot topic in 17th-century theology: “[As to] the conversion of the Indians by the gospel: it would be cause of great joy if they were feeling true, but [in many cases] they are converted by treachery and [coercion] and not by the wisdom of the gospel of Christ as [Eliot’s] treatise doth declare.”
Both of these stances stemmed from Williams’ radical (for the time) belief in religious freedom. He held that people could only become true followers of Christ by consciously accepting him and that no one else could make such a decision on their behalf. Most of the rest of the work, which has not yet been fully translated, consists of citations of scripture in support of his views.
According to Brown historian Linford Fisher, a specialist in early America, there is good reason to believe Williams may have intended to publish the writing. The reference to a generic “reader” in the introduction indicates an intention to write for a general audience. The work’s structure mirrors that of the Eliot treatise to which it systematically responds, indicating a concerted intellectual effort rather than scattered notes. And its tone is reminiscent of Williams’ published polemical works, including 1644’s The Bloudy Tenent of Persecution.
Based on the publication date of Eliot’s treatise, Williams was writing in 1679, at the earliest. He died in 1683. The timing helps explain the use of space-saving shorthand in the mystery book’s margins. Paper had always been scarce in Rhode Island, and in 1676, as King Philip’s War raged, American Indians burned much of Providence, including Williams’ home, to the ground. He lost most of his possessions and was forced to move in with his son Joseph, further contributing to a need for frugality.
Though the work does not significantly alter our understanding of Williams, according to J. Stanley Lemons, a professor emeritus at Rhode Island College and a Williams specialist, it’s now Williams’ last known work of theology and confirmation that the radical theologian remained staunch in his convictions into the twilight of his life.
Brown’s Fisher also sees the work as testament to the remarkable consistency of Williams’ views. He notes that despite the personal trauma wrought by King Philip’s War, Williams continued to ruminate on the salvation of Indians, and his position remained in line with that expressed in earlier writings.
“I’m not sure that what we’ll learn inside the text will be terribly important,” says Widmer, who along with Mason-Brown, Lemons, and Fisher will contribute to a scholarly book on the project. “I think the chase is as important as the result.”