|
|
|
by
James J. Gillogly
This paper orinally appeared in Cryptologia April 1987;Volume XI, Number 2.
ABSTRACT: William Coleman wrote a letter in longhand and shorthand dated 14 Jun 1796. This paper describes the cryptanalysis of the shorthand portions of the letter and the subsequent identification of the system as Byrom's Shorthand.
KEYWORDS: Shorthand; Coleman, William; Byrom's Shorthand Writing.
Cracking an unknown shorthand system is much like dealing with a simple substitution cipher, but presents three unique problems. The first is determining the units of meaning: where does each letter or sound begin? The second problem is the spelling: shorthand systems are typically phonetic, rendering standard pattern-word lists useless and making word recognition more difficult. The third and potentially most serious problem is abbreviation: each shorthand system defines short strokes or sequences for common words and affixes. Worse, the shorthand writer may create his own idiosyncratic abbreviations.
In January 1986 Cryptologia Louis Kruh asked for help reading a letter written partly in shorthand by William Coleman (1766-1829) [1]. The longhand portions of the letter instruct "Tim" to liquidate some of Coleman's assets. The shorthand portions are reproduced in Figure 1, with some of the cleartext context around them. The lines of shorthand are lettered A through I for reference. For clarity I will call the longhand portions "cleartext", the shorthand portions "ciphertext", and the decrypted shorthand "plaintext." Following the Gregg shorthand definition, the shorthand "words" (which may include more than one word) will be called "outlines."
Despite the fairly small amount of material, the presence of several "cribs" and the mixing of cleartext with ciphertext give a plausible entry into the system. Near the beginning of line A an outline has a word above it that appears to say "numbered." I took this to be the work of the letter's recipient, deciphering a difficult outline. The cleartext "Virginia" and "150000" on line C appear to tie in with the second section of cleartext discussing his large land holdings. The crossed-out outline at the end of line D might well be an attempt at "September", which appears as cleartext at the beginning of line D. Line D has a fairly clear "repay" near the beginning. Two rather obscure decipherments appear over words in line E. The first might be "hazy" or "crazy", and the second might be "resumé" or something else. On line F we have cleartext "Tim" and probably "groped", and on line G the cleartext "Military" suggests the "military warrants" of the first cleartext section, while "Virginia" may refer again to the land to be sold. Line I has the cleartext word "gulph" following a scratched-out outline or word.
After a false start trying to get "hazy" to work in line F, I tried the word m "crazy." There are three or four different strokes in it, depending on whether the top part consisting of a circle and line are one or two strokes. In any case, there are too few to be C-R-A-Z-Y, so something is being left out. And where is the R? I finally decided that the R could go either up or down, as long as it was a diagonal sloping right. This is unlike Gregg shorthand, where all letters go the same way; however, it's rather useful for keeping the outlines running along the same line. In any case, if R is the second stroke, then the top piece (loop above horizontal line) must be a C and/or K. We're left with a single horizontal line on the bottom, which is more likely to be the Z sound than all of AZY. This means that the vowels have apparently been left out of this word completely -- an unexpected source of ambiguity.
We can now look at "resumé" on line F and decide that it really isn't. We have R-Z-R-Z, with a dot at the bottom between the first R-Z. Adding in vowels, this could be RESOURCE, and the crib above it can be made to look like "resource" also. The dot near the bottom could be the E or the O vowel in RESOURCE, since it follows the first R and precedes the second. The horizontal line can now be used for either S or Z. Moving to line G, the guess that the outline after "Military" is "warrants" seems to be confirmed: it includes an R near the beginning and an S on the end. This gives us W as a counter-clockwise loop followed by a vertical downward stroke (assuming that doesn't include the A of warrants), N as a concave-upward stroke, and T as a vertical stroke, which we will assume can go either up or down like the R.
We have now identified several letters and the general system, so the next step is to fill in the known strokes throughout the ciphertext and use them to deduce more letters. The result of this effort is shown in Figure 2. At this point much of the letter can now be read and the sense is clear: in the cleartext Coleman is giving business-like instructions for his property disposal, and in the shorthand portions he is expressing urgency and his fears that he will be financially ruined.
The Britannica article gave several pages of examples, including all of the brief forms used by Coleman. The completed solution can now be read:
[Cleartext... TILL IT ENDS IN THE BANKRUPTCY OF MANY PERSONS OF
APPARENT PROPERTY.] And God knows whether I should not be numbered
amongst them, for my debtors fail to make my payments which have become due,
and I am depressed I assure you, to a very great degree, and I have not enjoyed
myself a moment for these 5 weeks. Besides, to add to my embarrassment I feel
myself oppressed with too great quantity of [cleartext VIRGINIA] land which I
now have in hand unsold, amounting to 150000 acres, for which I have to advance
next [cleartext SEPTEMBER] 8000$ and also to repay 5000$ more which I have
borrowed for the purpose of making the purchase. All [?] I can do with all this
heaven only knows, I declare to you Tim (and this must go no further) and I am
almost crazy, and have no resources by to turn from the [situ? sight?] of my
situation. So we turn away from the prospect of death. Now [cleartext TIM], in
this circumstance, I do not feel myself justified in proceeding further, till I have
[cleartext GROPED] a little out of this horrid darkness. I will sell my [cleartext
MILITARY] warrants, I will sell my [cleartext VIRGINIA], if I can, at a very small
profit, and the particular purpose of this letter [cleartext IS TO REQUEST YOU
ASSISTANCE ...]
[Cleartext NOW TIM IT IS ABSOLUTELY NECESSARY I SHOULD GET RID
OF] this tract of land. Otherwise it will sink me down into a bottomless [cleartext
GULPH]. For heaven's sake, write me a letter a soon as this reaches you and
[cleartext DIRECT TO ME AT GREENFIELD].
In many places in this letter Coleman wrote a stroke upside down, with the loop on the
wrong side. This was confusing, and suggested that he was making occasional mistakes in his
writing. However, the Britannica article considered this a "feature" rather than a "bug": if a
letter connection is clumsy, either letter may be replaced by its mirror image. The shorthand
system is not strictly phonetic, either: in the first line the word "knows" is written with the
K stroke (upside-down) first. The experiences described in this letter did not improve Coleman's grasp of business. After his bankruptcy he went to New York and started a newspaper with the help of Alexander Hamilton and other Federalist friends. William Cullen Bryant wrote, "He had no skill as a manager of property; he took little thought for the morrow; when he happened to have any money, it was spent freely, or given away, or somebody who would never return it contrived to borrow it. In a short time the finances of the Evening Post became greatly confused and embarrassed." Coleman's friends gave financial control of the paper to another man, and the Evening Post and Coleman soon got out of trouble and became prosperous. Coleman was quite successful as a journalist, and remained the editor until his death in 1829 [3].
A later edition of the Britannica [4] provides more perspective on Byrom's Short-hand System and its inventor, John Byrom (1692-1763):
To secure linearity in the writing and facility in consonantal joinings he [Byrom]
provided two forms for b, h, j, w, x, sh, th, and three for l. A, e, i, o, u he
represented by a dot in five positions with respect to a consonant. Practically it
is impossible to observe more than three (beginning, middle and end). With all
its merits, the system lacks rapidity, the continual recurrence of the loop seriously
retarding the pen.
Despite these perceived problems, Byrom used his shorthand to take down speeches in the
House of Commons over the objections of some of the members, and Coleman later used it to
transcribe articles Hamilton dictated for the Evening Post.
Despite the difficulties of vowel elimination, vowel placement and letter reflection, it was possible to arrive at an almost complete and unambiguous decryption of the letter, The process was greatly sped up by the existence of cribs written in above the outlines, and the final result was refined when the system was identified as Byrom's Short-hand Writing.
1. Kruh, L. 1986. l8th Century Shorthand Expert Needed. Cryptologia. 10: 60-62.
2. A Society of Gentlemen. 1771. Short-hand Writing. Encyclopaedia Britannica (First Edition). 3: 586-599.
3. Bryant, W. C. 1902. New York Evening Post. One Hundredth Anniversary.
4. Chisholm, H. (ed.) 1910. Byrom, John. Encyclopaedia Britannica (Eleventh Edition). 4: 897.
James J. Gillogly designs and writes software, mainly for UNIX machines and personal computers, and enjoys most the jobs where he gets paid to do cryptography. He received his PhD from Carnegie-Mellon University in Computer Science, specializing in Artificial Intelligence. He has written programs to solve or help solve dozens of the cryptosystems used by the American Cryptogram Association, but still has not figured out the Voynich Manuscript.
Mail Jim Gillogly
Converted to hypertext by Joe Peschel October, 2000.
