Cryptography

              [Editorial note: [...] indicates use of Coptic or Greek text. Original script is available for viewing in the PDF format of this article.]

(CE:A65a-69a)

CRYPTOGRAPHY. At times the Copts have felt the need to use cryptography in order to hide the contents of certain annotations, formulas, inscriptions, and messages. About thirty examples of this have been recorded. It would, however, be excessive leak here of a “Coptic cryptography,” for, as shall be seen, even in a Coptic context, the scribe used cryptographic systems borrowed from Greek and even preferred Greek over Coptic in formulas disguised. The cryptographic systems employed can be summarized within three types.

First Type

The two principal forms fully merit description as “encoding,” for the Greeks took them not from the classical alphabet but from an archaic repertoire used in writing numbers: the first nine characters of the archaic Greek alphabet were used to express the units one through nine. The next set of nine were employed to express the tens: ten, twenty, thirty, and so on. The next set of nine letters were used to express the hundreds, as Table 1 illustrates. Here one can see three archaic letters that had fallen out of use:  (stigma) for 6, [...] (qoppa) for 90, and [...] or [...] (sampi) for 900.

[See PDF version of this article for Table 1.]

During the Hellenistic period, imitating a Jewish cryptographic process (called atbash), some Greek created a similar encoding based on that division of the former alphabet into three portions (or rows). This system consisted of inverting the letters of each row and replacing the normal row by the inverted row—for example, a b g d e j z q becoming q h z j e d g b a, and normal a being replaced by q, b by h, d by z, and so on (see Gardthausen, 1913, p. 301; Wisse, 1979, pp. 119—20). As this system of inversion had the weakness of not being able to modify [...] (e), [...] (n), and [...] (f), located at the center of each row, special cryptic symbols were fabricated for them. For instance, e was translated by [...] and [...] by [...]. When the Copts borrowed this system, the archaic letter [...], which had the value of 900 and in such texts was written with the letter [...], had fallen into disuse; it was thus rendered with some lack of precision, as if it were the well-known letters [...] bound together. Such was the basis of the cryptography that the Copts mostly used (Table 2). It is noteworthy to remark that in Coptic the Greek [...] is currently confused with [...], by which it is replaced.

[See PDF version of this article for Table 2.]

This system was suitable for encoding a text written in Greek, and in fact, in encoding formulas, the Copts mostly used Greek formulas, even in the body of documents otherwise written completely in Cop-tic. However, when they wanted to hide a truly Coptic formula, they either did not modify the autochthonous graphemes (especially [...], [...], [...], [...], and [...]), or they encoded them by means of conventional signs. Here are some examples of this system and some known variations of it:

1. From the Coptic treatise entitled “Zostrianos” (Nag Hammadi Library, Codex VIII [or IV, according to Doresse’s numbering], p. 132) comes in Greek the following “colophon” at the tractate’s end (first half of fourth century):

[...]

“Words of truth of Zostrianos, the God of truth, words of Zoroaster” (cf. Doresse, 1950; Wisse, 1979, no. 1; with some emendations).

2. A Greek graffito from the sixth century or later found in the Theban mountains (Crum and Evelyn-White, 1926, no. 701) reads thus:

[...]

In lines 1 and 5, the marks in braces {... } are superfluous. In line 1, the first [...], and in line 6, the second and third [...] are ordinary crosses. “Lord and power of the holy prayers of the great [monks], pray for my sins! I am Menas the sinner. Amen, Amen, Amen, Lord! Ind[iction] 7” (cf. Wisse, 1979, no. 2, which strongly improved the reading of this text, in an approximate Greek; formerly published in Crum and Evelyn-White, 1926, pp. 147, 330, 386). The sampi, a rarely used symbol, is sketched in some variable fashions. The siglum [...] for “Amen” is the current Byzantine abbreviation based on the numerical value of the letters of this word: [...] + [...] + [...] + [...] = 1 + 40 + 8 + 50 = 99.

3. Coptic formulas to protect oneself from dogs were not so much for the simple passerby as for lovers or thieves who would fear the dogs’ barking and biting at night; the text shown below was edited by Erman (1895) from a fragment conserved in the British Museum (Or. 1013-A). The reading of the cryptographic formula was specified by Wisse (1979, no. 11). The revelation of this formula is attributed to Isis, an attribution that places it among the most archaic of Coptic formulas. Here are the most essential lines only:

[...]

I bind the dog of [...], the son of the woman [...],who is his
mother!

In this text, the three Coptic letters originated from demotic, [...], [...], and [...], are conserved just as they are, without encoding (cf. Erman, 1895; Kropp, 1930-1931, Vol. 2, no. 5, pp. 14-16, and Vol. 3, no. 249; see Wisse, 1979, no. 11, for review and comment).

4. In the Coptic medical papyrus published by Chassinat (1921), the names of a certain number of drugs are encoded in the same way. The manuscript can be dated from the ninth or the tenth century. Samples: [...] for [...] (or [...]), onion; [...] for [...], [...] (or [...]), bronze; [...] for [...], calf.

5. During the Persian invasion at the beginning of the seventh century, the monks of the monasteries in the Theban mountains (in Dandarah as well as in Dayr al-Bahri) had to withdraw to the surrounding desert. Probably this temporary exile would account for the Coptic graffiti found in a hermitage in the region of Armant in 1947 by Bachatly (cf. Abd alMasih et al., 1965). This graffito was written by a monk who came from the great Monastery of Phoebammon; only the first half is given here:

[...]

In line 2, the first [...] is an ordinary cross. The sign in braces {. . . } is superfluous. In line 3, the scribe wrote the first [...] with the shape of an ankh. “I am Philotheos, the son of Komes, the insignificant dea-con of Terkot. My Lord Jesus Christ, have mercy on me...” The autochthonous coptic letters [...] and [...] remain without encoding (cf. ibid., p. 30; reviewed by Wisse, 1979, no. 12).

6. To disguise the autochthonous Coptic letters, the Copts tried employing conventional signs such as [...] for [...] and [...] for [...]. The following example is a personal invocation inserted by a monk before the title of an epistle on virginity attributed to Saint Athanasius (Bibl. Nat. copte 131, fol. 2r, perhaps ninth or tenth century; cf. van Lantschoot, 1929, Vol. 1, App. 1; reviewed by Wisse, 1979, no. 14):

[...]

Remember me in love, everyone that will read in this book, and pray for me.

7. In a brief message of greeting written on a parchment scrap (B.M. Or. 4720[96]), the Coptic letlets of demotic origin were encoded by Greek letters used as symbols for thousands: [...] for 1,000; [...] for 2000; [...] for 3,000; [...] for 4,000; [...] and for 5,000— this, respectively, for the letters [...], [...], [...], [...], and [...]. These graphemes were conserved in their regular order.

Recto: [...]

Verso: [...]
With God! I greet and salute the health of my pious Abba Paphnouti.

One can see that the name of the addressee, Abba Paphnouti, is written without code on the parchment’s verso (cf. Crum, 1905, no. 669; reviewed by Wisse, 1979, no. 16).

Second Type

A second cryptographic system borowed from Greek also uses the primitive alphabet divided into three rows of characters representing units of tens and hundreds. The units are represented by the letters [...] through [...], the tens by the same [...] through [...] topped by a single dot, the hundreds by [...] through [...] topped by two dots. This system was perhaps borrowed from Arabic (Wisse, 1979, no. 18). Each set of the alphabet is encoded by the signs 1 to 9, 10 to 90, and 100 to 900, respectively, superimposed one on the other without resorting to an inversion, as was the case in the system described above, with the result that the letters [...] to [...] of the genuine text are not modified by this code at all (Table 3).

8. In a Bohairic Gospel book dated from 1327, an invocation is transcribed as follows:

[...]

The wretched Thomas, God be merciful to him! Amen, Amen, Amen.

An original peculiarity: each “Amen” is encoded in a different way. The first is represented by [...], corresponding to the later method illustrated by this invocation. The second is [...] according to the isopsephic system already noted above in example 2. The third appears as [...], according to the cryptographic process described at the beginning of this article (cf. Homer, 1898—1905, p. lxxi; Wisse, 1979, no. 18).

Third Type

A third system substituted for each letter of the normal alphabet the corresponding letter from another normal alphabet written beside it but shifted down by one or more letters, a process called in antiquity “Julius Caesar’s method.” If one shifts the second alphabet by one letter, starting with b replacing [...], one has the following:

[See PDF version of this article Table 3.]

9. Here is an example attested in a Gospel text from the White Monastery dated 1112 AD. The same process was only applied to the five Coptic letters [...] through [...] encoded by their own sequence being shifted by one letter:

[...]

I am the poor Victor, son of Shenoute. Forgive me!

(Cf. Crum, 1905, no. 489; van Lantschoot 1929, no. lxxx—h; Wisse, 1979, no. 19, settled the issue of its interpretation.)

These systems are the most current. The first, transmitted by the Coptic scribes to their Ethiopian colleagues, was even adapted to the Ge’ez language and used under the designation of the “learned lan-guage” or naggara liqciwent (Conti-Rossini, 1927, pp. 524—28; unfortunately notes written very hastily and imprecise).

Other Types

Were there other Coptic processes of encoding? One can suppose this, since several formulas still resist efforts to decode them, unless they are cryptogram imitations devoid of meaning. It must be pointed out that artificial alphabets existed and were used to hide astrological, alchemic, and magical formulas. Indeed, Hellenistic and Byzantine occultism produced many picturesque versions (Doresse, 1950— 1951, pp. 221—26). Furthermore, the Arabs, not merely satisfied to revive such formulas, added a great number of fancies that spread throughout the Mediterranean world (among others, cf. Ibn Wahshiya, “Les Alphabets occultes dévoilés,” in Hammer, 1806).

It is certain that through the Byzantines, the Copts learned the cryptographic method of transliterating the normal alphabet by a “long key”—that is, by a conventionalized phrase embodying all the letters of the alphabet out of alphabetical order, as suggested by repetitions in the artificial sentence

[...]

to be compared with the alphabetic order transcribed as follows: [...]

These two lines were found scribbled on a piece of wood recovered in the ruins of the Theban Monastery of Apa Epiphanius (Crum and Evelyn-White, 1926, no. 616). The grotesque formula

[...]

was so well known that the Palatine Anthology (9.538) had included it. However, there survive no examples of either Coptic or Greek encoded texts employing it as a key. And yet, the cryptographic purpose of this mnemonic device seems to be indis- putable, the normal alphabet being connected with it in the form of three inverted rows (from [...] to [...], from [...] to [...], and from [...] to [...]), as in the first of the systems discussed above.

Purpose

One should ask what purpose these cryptograms served, for it is clear that such systems began virtually at the birth of the Coptic language (examples 1 and 3) and lasted at least until the fourteenth century (example 8). In answer, the limited number of examples so far identified, plus the fact that several cases remain unidentified and that it is impossible in other instances to decide whether the text is written in Greek or Coptic (e.g., the calamus box from Antinoë [ANTINOOPOLIS] in the ancient collection of the Guimet Museum kept today at the Louvre; cf. Doresse, 1951, p1. 1) allow just a few observations rather than true conclusions.

It seems that initially such systems served to hide the entire title of an apocryphal work, to disguise a magic formula or make it more mysterious, or to veil the exact identity of medical drugs from the knowledge of common people (examples 1, 3, and 4). For the rest, it became above all a guileful expedient of some literates to communicate among themselves only: an invocation inserted by a scribe at the end of a manuscript he has copied (examples 6, 8, and 9); a prayer of a monk scribbled on a wall (examples 2 and 5); a brief message, essentially a prayer, to another monk (example 7). All of this was at once naive in its process and impoverished in its content. Apparently, the worthiest things hidden in these cryptograms have been totally lost, except vague memories: “The Thebans tell of an angel giving the science of the mystic language to Pachomius, Cornelius and Syrus in such a way that they ex pressed themselves by means of a special alphabet which concealed the meaning in hidden signs and symbols” (Praefatio ad regulas S. Pachomii, in Migne, PL 23, p. 68). This would confirm a letter from Pachomius to Syrus (ibid., p. 100): “Animadverti enim terminos esse epistolae vestrae Heta et Theta,” where terminos could specify a key to decode the order of those things of which one can only catch a glimpse.

JEAN DORESSE