Α Β · Β ΚΒ · Γ Ι · Δ Λ · Ε ΙΗ · Ϛ Ζ · Ζ ΚΖyear 1: 2 April · 2: 22 March · 3: 10 April · 4: 30 March · 5: 18 April · 6: 7 April · 7: 27 March

Byzantine Gospel lectionary · Greek letter-numerals · 11th or 12th century manuscript, addition undated · not a cipher

The numerals under Ioannes the reader’s note in a Greek lectionary: a table of the Paschal full moon

Go to the full text ↓

Three lines of Greek numerals under a reader’s note, catalogued as a cryptogram that might hide a name, turn out to list the date of the Paschal full moon for each year of the 19-year lunar cycle.

Daniel Bourdeau · posted

Method: not a cipher

Contributed by Brice Dudout (solution)

At a glanceNot a cipher · 19 of 19 values match, three of them read from doubtful ink

The document
Three lines of capital letter-numerals at the foot of a page of a Greek Gospel lectionary, below a note in which a reader named Ioannes salutes his teacher.
Where
Cambridge, Trinity College, MS O.4.22, f. 248v, online at the Cambridge Digital Library. Its record calls the lines an unresolved numerical cryptogram.
What it is
A table, not a cipher: the numbers 1 to 19, each followed by a day of the month. The day is the Paschal full moon for that year of the lunar cycle.
How it was identified
Brice Dudout, working with a Claude agent, took the numerals from the images as 19 pairs and matched the values to the Paschal full moons of the Julian Easter rule: 19 of 19, three of them read from doubtful ink. A project session reached the same result separately the same day.
What it says
Year 1, 2 April; year 2, 22 March; year 3, 10 April; and so on to year 19, 13 April. No name is hidden in it.
Still open
Letter shapes at three places (years 5, 14 and 18) are doubtful in the faint ink; the table fixes their values. Who wrote the table, and when, was not examined. A printed Byzantine table was not consulted.
Credit
Identified by Brice Dudout, working with a Claude agent (pull request 690, 16 November 2026). The library’s transcription, used for comparison, is Christopher Wright’s.

01 The lines on the page

Trinity College MS O.4.22 is a Greek Gospel lectionary. The Cambridge Digital Library dates it to the 11th century; the Trinity catalogue says 12th. On f. 248v, the library’s record says, another hand added a short excerpt from Mark after the last lection, and an early user then added a note that it prints as Ἐγὼ Ἰωάννης ἀναγνώστης χεριτιζω τὸν διδάσκαλόν μου· ἀρχιερέα· καὶ ἐξαδελφόν μου: “I, Ioannes, reader, salute my teacher, arch-priest and cousin” (the record gives “cousin or nephew” for the last word).

Under the note, across both columns, run three lines of capital letters, each letter or pair of letters under a short overline. The record, by Dr Christopher Wright, says the text “appears to have been encoded using a numerical cryptographic system”, that it may include the name of the person saluted, and that it “is not, however, decipherable by any of the more common Greek cryptographic systems, and remains to be resolved”. It transcribes the lines as:

ΑΒ· Β· Κ· Β· Γ Ι· Δ ΛΕ ΙΗ· Ϛ ζ: Ζ ΚΖ· Η ΙΕ Θ Δ ΙΚ Δ. ΙΑ· ΙΒ ΙΒ· Α ΙΓ ΚΑ· ΙΑ Θ ΙΕ ΚΘ ΙϚ ΙΖ· ΙΖ Ε ΙΗ. ΚΕ· ΙΘ ΙΓ

No image of the lines is shown here: the library licenses its images CC BY-NC 4.0, which this site does not reproduce. They are on image 503 of the Cambridge Digital Library viewer.

02 Nineteen pairs

An overline is the usual mark of a Greek numeral, and the letters here are numerals: Α is 1, Β 2, Ι 10, Κ 20, Λ 30, and so on, with the sign Ϛ for 6. Read as numbers, the lines are 38 numerals in 19 pairs, a grouping the project’s catalogue entry had suggested. The first number of each pair counts from 1 to 19; the second is a value.

Brice Dudout, working with a Claude agent, transcribed the lines from full-size crops of the image and compared them with the library’s transcription: the same 38 numerals. The grouping changes, and one letter (the count of year 14); the project’s check found that the overlines on the image give the grouping. Β· Κ· Β· is Β, then ΚΒ (2, then 22). Δ ΛΕ ΙΗ is Δ Λ, then Ε ΙΗ (4 and 30, 5 and 18): Λ and Ε have separate overlines. ΙΚ Δ is Ι, then ΚΔ (10, then 24). The count transcribed ΙΑ before Θ must be ΙΔ, 14; on the image that letter has a triangular body and could be Α or Δ, so this one letter is taken from the sequence and not from the image alone. Dudout notes two more doubtful places: the value of year 5, where the ink shows Ι and two Η-shaped signs, and the pair of year 18, where a small unclear group with a tilde stands after the Η and the Ι of the count is uncertain. The small minuscule words at the right ends of lines 1 and 2, which the library’s transcription leaves out, are the end of the note (φον and μου), not numerals.

03 The Paschal full moon

From one year to the next the written value falls by 11 or 12, or rises by 20; as a date, it comes 11 days earlier (once 12) or 19 days later. That is how a lunar date moves through a solar calendar: twelve lunar months are about 11 days shorter than a year, so the date comes 11 days earlier each year until a thirteenth month is added and it jumps forward.

The values are the day of the month of the Paschal full moon, the “legal Passover” of the Greek Easter reckoning, for the 19 years of the cycle of the moon. Easter is the Sunday after that date. The month is not written: the full moon always falls between 21 March and 18 April, so a value of 21 or more is in March and a smaller one is in April.

Year of the cycleAs writtenValuePaschal full moon, computed
1Α Β22 April
2Β ΚΒ2222 March
3Γ Ι1010 April
4Δ Λ3030 March
5Ε ΙΗ1818 April
6Ϛ Ζ77 April
7Ζ ΚΖ2727 March
8Η ΙΕ1515 April
9Θ Δ44 April
10Ι ΚΔ2424 March
11ΙΑ ΙΒ1212 April
12ΙΒ Α11 April
13ΙΓ ΚΑ2121 March
14ΙΔ Θ99 April
15ΙΕ ΚΘ2929 March
16ΙϚ ΙΖ1717 April
17ΙΖ Ε55 April
18ΙΗ ΚΕ2525 March
19ΙΘ ΙΓ1313 April

The last column is computed from the standard rule of the Julian Easter reckoning (21 March plus (19 × (year mod 19) + 15) mod 30 days), with the years numbered the Byzantine way, by the year of the world: (AD year + 5508) mod 19. All 19 values equal the computed day, and the counts run from 1 to 19 without a gap. The one step of 12 days falls between years 16 and 17. With the western numbering of the same cycle (the golden number) the same dates stand three places later and would not fit.

The measuring script was fixed (by checksum) before it was run. Beside the 19 of 19, it shows that 200 random shuffles of the values over the counts give at most 4 matches of 19, and that each of the 18 other ways of numbering the cycle gives none. The pattern was seen in the same numerals that confirm it; there is no further text to test it on.

A project session, given the catalogue page the same day, reached the same identification from the library’s transcription before it saw his pull request, and checked the grouping on the image and the 19 values with a script of its own. The check ran on the 19 pairs of pairs.tsv, not on a fresh transcription, so the two derivations agree on the values and the cycle numbering; the project did not re-read the doubtful places at counts 5 and 18.

04 What this means for the note

The lines do not hide the teacher’s name, or any name. Why it stands in a lectionary was not examined. The table carries no date of its own: the same 19 values serve every cycle.

The library’s record assigns the numerals to Ioannes. Whether the table and the note are by one hand, and which was written first, was not examined; the last words of the note stand at the right ends of the first two numeral lines.

05 What remains uncertain

Full texts

The transcription and the decryption of every document on this page, so that the reading can be used and checked. Short files are shown here; all are in the repository.

O.4.22 f. 248v, the 19 pairs not prose

Table of the legal Passover (Paschal full moon) for each year 1-19 of the Byzantine lunar cycle: 1: 2 April, 2: 22 March, 3: 10 April, ... 19: 13 April.

Transcription

pairs.tsv

# Trinity College Cambridge MS O.4.22, f. 248v: the 38 letter-numerals read as 19 (count, value) pairs.
# count	count_ms	value_ms	value	byz_lunar_cycle_GN	paschal_full_moon(Julian)	doubt
1	Α	Β	2	4	2 April	
2	Β	Κ·Β	22	5	22 March	ms has a point between Κ and Β
3	Γ	Ι	10	6	10 April	
4	Δ	Λ	30	7	30 March	CUDL groups "ΛΕ" (Λ value + Ε next count)
5	Ε	ΙΗ	18	8	18 April	image shows Ι + two Η-shaped glyphs (dittography?)
6	Ϛ	Ζ	7	9	7 April	
7	Ζ	ΚΖ	27	10	27 March	
8	Η	ΙΕ	15	11	15 April	
9	Θ	Δ	4	12	4 April	
10	Ι	ΚΔ	24	13	24 March	CUDL groups "ΙΚ Δ"
11	ΙΑ	ΙΒ	12	14	12 April	ΙΑ written as a ligature at the end of line 1
12	ΙΒ	Α	1	15	1 April	
13	ΙΓ	ΚΑ	21	16	21 March	
14	ΙΔ	Θ	9	17	9 April	CUDL "ΙΑ"; Δ and Α look alike in this hand
15	ΙΕ	ΚΘ	29	18	29 March	
16	ΙϚ	ΙΖ	17	19	17 April	
17	ΙΖ	Ε	5	1	5 April	
18	ΙΗ	ΚΕ	25	2	25 March	Η followed by a small unclear group with a tilde; Ι before Κ under an ink blot
19	ΙΘ	ΙΓ	13	3	13 April	line 3
Reading

pairs.tsv

# Trinity College Cambridge MS O.4.22, f. 248v: the 38 letter-numerals read as 19 (count, value) pairs.
# count	count_ms	value_ms	value	byz_lunar_cycle_GN	paschal_full_moon(Julian)	doubt
1	Α	Β	2	4	2 April	
2	Β	Κ·Β	22	5	22 March	ms has a point between Κ and Β
3	Γ	Ι	10	6	10 April	
4	Δ	Λ	30	7	30 March	CUDL groups "ΛΕ" (Λ value + Ε next count)
5	Ε	ΙΗ	18	8	18 April	image shows Ι + two Η-shaped glyphs (dittography?)
6	Ϛ	Ζ	7	9	7 April	
7	Ζ	ΚΖ	27	10	27 March	
8	Η	ΙΕ	15	11	15 April	
9	Θ	Δ	4	12	4 April	
10	Ι	ΚΔ	24	13	24 March	CUDL groups "ΙΚ Δ"
11	ΙΑ	ΙΒ	12	14	12 April	ΙΑ written as a ligature at the end of line 1
12	ΙΒ	Α	1	15	1 April	
13	ΙΓ	ΚΑ	21	16	21 March	
14	ΙΔ	Θ	9	17	9 April	CUDL "ΙΑ"; Δ and Α look alike in this hand
15	ΙΕ	ΚΘ	29	18	29 March	
16	ΙϚ	ΙΖ	17	19	17 April	
17	ΙΖ	Ε	5	1	5 April	
18	ΙΗ	ΚΕ	25	2	25 March	Η followed by a small unclear group with a tilde; Ι before Κ under an ink blot
19	ΙΘ	ΙΓ	13	3	13 April	line 3

06 Sources

Notes and files: targets/trinityo422/

Cite this page

Brice Dudout and Daniel Bourdeau, “The numerals under Ioannes the reader’s note in a Greek lectionary (Trinity College O.4.22): a table of the Paschal full moon”, Cyphersolver, posted 17 November 2026, https://cyphersolver.com/trinityo422.html.