M-94

Background

The wheel cypher device has been reinvented a number of time in history. Thomas Jefferson (1790) with 36 disks, commandant Bazeries (1891)with 20 disks, Parker Hitt (1914) as a strip device and later in 1917 Joseph 0. Mauborgne of the U.S. Army signal corps came up with substantially the same idea. As late as the mid 1960s the Navy employed a version.

My friend Bryan Locks made a working replica of the Army M-94 in 1989. It uses 25 Bakelite disks with the alphabets taped around the outer edge. A central rod clamps the disks. Each disk has a label in a pocket on the flat face. The Navy referred to this unit as the CSP-488.

Disk number 17 (has the letters AR... = ARMYOFTHEUSZJXDPCWGQIBKLNV shows the Army origin of the M-94 Thomas Jefferson (President of the U.S.) also independently invented this device as well as others. The disks are ordered by finding the letter after the letter "A", In the case of the letter R it is the seventeenth disk (there is no disk with two "A"s).

Experiment using StarBand for image hosting:
disks are ordered based on the letter below "A", there is no disk with two "A"s so the first disk is AB, then AC etc. Since there are 25 disks and 26 letters in the alphabet the last disk is AZ.

Photo with Cryptologia M-94 issues

After the plain text (pt) message has been set by rotating the disks (that have already been installed in the key order) the operator selects one of the remaining 25 different messages at random. Some instructions said to not select offsets that were close to the pt message.

When the receiver has set the cypher text (ct) along the ruler he then must look at all of the rows to find the one with the plain text. This aspect of the system makes it much harder to analyze than systems that have only a single choice.

Test Messages and my best guess at reading the plaintext as shown in SRH-366 US Strip Ciphers - starting at 149/310 pages.

Following are the disk layouts:

Disk
ID Letters Around The Rim
B 1 ABCEIGDJFVUYMHTQKZOLRXSPWN
C 2 ACDEHFIJKTLMOUVYGZNPQXRWSB
D 3 ADKOMJUBGEPHSCZINXFYQRTVWL
E 4 AEDCBIFGJHLKMRUOQVPTNWYXZS
F 5 AFNQUKDOPITJBRHCYSLWEMZVXG
G 6 AGPOCIXLURNDYZHWBJSQFKVMET
H 7 AHXJEZBNIKPVROGSYDULCFMQTW <fixed typo 6/23/01>
I 8 AIHPJOBWKCVFZLQERYNSUMGTDX
J 9 AJDSKQOIVTZEFHGYUNLPMBXWCR
K 10 AKELBDFJGHONMTPRQSVZUXYWIC
L 11 ALTMSXVQPNOHUWDIZYCGKRFBEJ
M 12 AMNFLHQGCUJTBYPZKXISRDVEWO
N 13 ANCJILDHBMKGXUZTSWQYVORPFE
O 14 AODWPKJVIUQHZCTXBLEGNYRSMF
P 15 APBVHIYKSGUENTCXOWFQDRLJZM
Q 16 AQJNUBTGIMWZRVLXCSHDEOKFPY
R 17 ARMYOFTHEUSZJXDPCWGQIBKLNV
S 18 ASDMCNEQBOZPLGVJRKYTFUIWXH
T 19 ATOJYLFXNGWHVCMIRBSEKUPDZQ
U 20 AUTRZXQLYIOVBPESNHJWMDGFCK
V 21 AVNKHRGOXEYBFSJMUDQCLZWTIP
W 22 AWVSFDLIEBHKNRJQZGMXPUCOTY
X 23 AXKWREVDTUFOYHMLSIQNJCPGBZ
Y 24 AYJPXMVKBQWUGLOSTECHNZFRID
Z 25 AZDNBUHYFWJLVGRCQMPSOEXTKI

Variants

Thomas Jefferson Wheel Cypher, 1790 or 1793. to have 36 wheels,
Monticello Research Department - Thomas Jefferson's Wheel Cipher -
Tales of the Encrypted - an online version of the Jefferson wheels from the National Maritime Muesum (must use IE not Netscape)
The applet appears to be an M-94.

Bazeries

M-138 Army Strip Cypher

A strip cypher is made using stiff strips each of which has two alphabets, for example the (B1) strip would have:
ABCEIGDJFVUYMHTQKZOLRXSPWNABCEIGDJFVUYMHTQKZOLRXSPWNThis way the center 25 letters will always be overlapping. The order of the strips in the frame is how the key is set.

USS Pampanio (SS-383) ( Pier 45 at Fisherman's Wharf S.F.):

NAVY CSP (Code & Signal Publications) CSP-845 Navy & M-138 Army Strip Cypher
CSP-1088 CSP-845 with CSP-488 compatible strip alphabet No. 488
CSP-845 M-138A = CSP-1088 = VENUS (strip cipher,
CSP-847 Instructions for use of CSP-845 strip cipher.
CSP-488 Navy Disk Cypher (duplicate of the M-94) and disk #17 R = "ARMYOFTHEUSZJXDPCWGQIBKLNV"
So you can see that the "army" on a disk does not mean that the device is of Army origion.

Reproduction Jefferson Wheel - made of plastic, photo - only 10 wheels and they don't seem to match any of the M-94 wheels.

Mystery Device

This device uses similar construction, but can not be used like the M-94 because there's no way you can compose a message since instead of each wheel containing a full alphabet each drum has 3 or 4 columns of letters where each of those columns does not contain a complete alphabet. The drums are numbered 1 to 7.
There are 26 columns (black letters). The red letters are associated with how the device is keyed.

In all the photos the drums appear to be in the same order: 3C, 5C, 4C, 3C, 3C, 4C, 4C (were C means columns).

The drums are:

Red No.	No. of Coul
1	3
2	5
3	4
4	3
5	3
6	4
7	4
8a & 8b	1 & 1

Photos

Fig 1
Drum order

Left 1	2	3	4	5	6	7	Right 8
(3C)	2	(4C)	(3C)	(3C)	(4C)	(4C)	a&b
QVE EPM MDA FMN P??	TL[]NA RCISL 1AELi NiENE	LCPT ASLU PQVG XEiT	MSE PMC EBM UJG	RTC SEN AiE ONU	EAUA ELUE EOUE CROA	OAHE UAL[] SAXS SLD1	O

Fig 2

1	2	3	4	5	6	7
(3C)	(5C)	(4C)	(3C)	(3C)	(4C)	(4C)
AFT D[]P ENH UDK ENL CHQ MPC QOB OD?	DM[]EF NOAEi iTOND BVPFV []LDCY iLDPE UOGPV EOAPV ERSKT	4ETE TEET RILA LCPT ASLU POUG XEIT OOTE	NSR CAF MSE PMC EBM UJG LDU OUD	R[]T SRO Q9N 88E 8RE SA1 RiL SEX	1EEA EFLU CEOS EAUA ELUE EOUE CROA ENUE	RLTA FTWS OAME UAL[] SAX6 SLD1 CLOA

Fig 3

1	2	3	4	5	6	7
(3C)	(5C)	(4C)	(3C)	(3C)	(4C)	(4C)
iPF iET Eii AFY D[]P EMH UKD ENL GHG	EREFA TiTAO OLPOZ ERLOA NASiL SRLUi DMNEP NOAEi iTOND BVPFV	UOLE PMRL AEiT OiVE HSEE DASU ELQN EEi8 LNNU	QUD ODL UVi PRA TU7 R6N 5EE TRN SAR ETN	TNR JFE YEE MEW E{}A VEC DES EEC 7CC	BARA RiOT iOUi OTNR HGRi XZ{}K iSNT 3RNi UiST	B9NC 8SiP UDPE TEEE JTEA ECSE DEFA RDMH EiEE WAOF

Fig 4

1	2	3	4	5	6	7
(3C)	(5C)	(4C)	(3C)	(3C)	(4C)	(4C)
Eii AFY D[]P EMH UDK ENL CHG MPC GGB	OLFOZ PRLOA NASiL SRLUi DM{}EP NOAEi iTOND BVPFV []LDCY iLDPE	AEit OiVC HSEE OASU ELQN EEiS LNNU ETUO LUOR iLMO	UVi PRA TU7 R6N 5EE TRN SAR ETN EAE NSU	YEE MEW E[]A VEC DES EEC TCE ESN CRV DLM	iOUi OTNR HGRi XZ[]K iSNT ARNi UiST iSAT DMPR DFMU	TLEE JTEA ECSE LEEA ROMH EiSE WAOE EEAL ANU[]

Bibliography

Aegean Park Press - C32 - Computer Simulation of Calssical Substution Cypher Systems, page 51
                              - C-4    - Kullback, Solmon, Statistical Methods in Cryptography - not M-94 but for cryptanalysis
                              - C-8 - Advanced Military Cryptography, William F. Friedman (U.S. Army Technical Manual 11-485)
                              - C-24 - Riverbank Publications, Volume II, Several Machine Ciphers and Methods for their Solution, Publication 20, Riverbank Laboratories, 1918
Amercian Cryptogram Association (ACA) Jly-Aug 82 Vol XLVII #6 Has 25 test messages on page 4 & 5
Amercian Cryptogram Association (ACA) Nov-Dec 82 Vol XLVII #8 has solutions for the test messages & a description of how they were solved
Amercian Cryptogram Association (ACA) Nov - Dec 1994 Vol LX, No 6, Bazeries Cylinder, page 12 Computer Column with program
                                                                there is a typo in program line 1080 (Z appears twice in the line, which is wrong?)
The Code Breakers, David Kahn, Pages: 192-194=-Jefferson, 268+2 pages=M-94 photo, 325= Bazeries, 385=Friedman, 584=Luzon
Cryptologia January 1997, Vol. 1, Number 1 The Cryptology of Multiplex Systems by Greg Mellen and Lloyd Greenwood, - the first two issues feature the M-94
Cryptologia April     1997, Vol. 1, Number 2 The Cryptology of Multiplex Systems. Part 2: Simulation and Cryptanalysis by Greg Mellen and Lloyd Greenwood. Consideres the cases: Known Alphabets & Known Crib, Unknown Alphabets & Known Crib, and Unknown Alphabets & Unknown Crib. The heart of the cryptanalysis depends on finding which disks are possible for the known ct = pt based on the offset or gap between the ct and pt. For a computer version the authors recommend (1) a table with the disk definations (2) a table with the disks ordered based on the key and (3) a table with the disks ordered by the key AND rotated based on the plain text.
They used the log weight method described by Abraham Sinkov in Elementary Cryptanalysis: A Mathematical Approach. I prefer using the sum of the squares of the difference between the normalized frequency count and a count for a language like English to recognize plain text.

But the more practical case of Known Alphabets & Unknown Crib is not covered. I think this is possible by the use of a computer, i.e. a generalized solution.
Cryptologia Vol 5, No 4, The Genesis of the Jefferson/Bazeris Cyper Device, page 193

Historical Survey Of Strip Cipher Systems. This is available from NARA; NSA Historical Collections 190/37/7/1, NR 3525 CBRK24 12957A 19450000.
SRH-366 History of Army Strip Cipher. This is available from NARA; RG 0457: NSA/CSS Finding Aid A1, 9020 US Navy Records Relating to Cryptology 1918-1950 Stack 190 Begin Loc 36/12/04 Location 1-19
U.S. Patent Office 2,395,863 (380/56) (1939 patent by William Friedman for a strip system, M-138)

Scribd: SRH-366 US Strip Ciphers -

Patents

2022 added this paragraph after finding this patent. Have not made a search.

875070 Apparatus for correspondence in cipher, Carl Haas, Heinrich Studt, 1907-12-31, 380/58 -

The Wiki page for Robert Patterson has information about the cipher Patterson included in a letter to Jefferson dated December 19, 1801. No one decoded the message until 2007. The method used has a slight similarity to the pencil and paper code used in the Spy Nickel case.

Value

There have been a couple of M-94 cypher devices sold on eBay prior to 2002 and they went for about $7,000.
World War II Army Signal Corps Cipher Device - eBay Item # 737471549 Nov-24-02 US $6,115.00
Signal Corps M-94 Code Wheel - eBay Item # 738184625 Nov-26-02 $Auction ended by seller
US Army M-94 Cipher Divice - eBay Item # 747564079 Dec-16-02 18:44:20 PST US$ Auction ended by seller
A Pair of US Army M-94 Cipher Devices eBay Item # 749853381 Jan-05-03 19:30:00 PST US $7,600.00 (reserve not yet met)

[an error occurred while processing this directive] page created 17 Sep. 2000.

M-94 Cipher Wheels

Background

Variants

Colonial Williamsburg Cypher Wheel Secret Decoder

Mystery Device

Photos

M-209

Bibliography

Patents

Robert Patterson Cipher to Thomas Jefferson

Related

Links

Value