ERRORS IN ENCIPHERING AND TRANSMISSION
One of the most difficult tasks before the cipher expert, is the
correction of errors which creep into cipher texts in the process of
enciphering and transmission by telegraph or radio.
In some cipher methods a mistake in enciphering one letter, or the
omission of one letter, will so mix up the deciphering process that
only one familiar with such errors can apply the necessary corrections.
The transmission of cipher text over the telegraph or by radio is
a slow process, and many fairly good operators cannot receive such
matter satisfactorily, because they listen for words and guess at
letters at times. The spaced letters in American Morse are the cause
of so many errors in code transmission that the War Department Code
does not employ any groups using them. In fact, this code is limited
to the letters
A B D E F G I K M N S T U X
so that there may be a minimum of such confusion.
In cipher work it is necessary, under ordinary circumstances, to
use any or all of the letters of the alphabet. To assist operators
in keeping the text straight, it is customary to divide cipher text
into groups of four, five, six or ten letters, and usually groups of
five letters are used. The receiving operator may then expect five
letters per group, and if he receives more or less he is sure that
either he or the sending operator has made an error. This division
into groups of a constant number of letters eliminates word forms and,
in the mind of the non-expert, increases the difficulty of solving the
cipher. But the increase in difficulty is more apparent than real;
particularly, as a cipher examiner habitually finds himself dealing
with ciphers without word forms, and the occurrence of a cipher with
word forms usually means that he has an easy one to handle.
Messages are occasionally encountered which consist partly of plain
text and partly of cipher. The cipher part may or may not retain its
word forms, but, when this method is used, it is clearly impossible
to have a fixed number of letters in each cipher group if the word
forms are not used. It is almost impossible to prevent errors of
transmission in such messages, and it often requires considerable
skill and labor to correct them.
For those unfamiliar with the telegraph alphabets, they are given
below. Messages sent by commercial or military telegraphs or buzzer
lines will be transmitted with the American Morse alphabet. Those sent
by radio, visual signalling or submarine cable will be transmitted
by Continental Morse, known also as the International Code. Messages
may be transmitted by both alphabets in course of transmission. For
example, a cablegram from the Philippines to Nome, Alaska, will be
transmitted by Continental Morse (commercial cable) from Manila to San
Francisco, by American Morse (commercial land line) from San Francisco
to Seattle, by Continental Morse (military cable) from Seattle to
Valdez, by American Morse (military land line) from Valdez to Nulato
and by Continental Morse (military radio) from Nulato to Nome.
Prior to February, 1914, the Mexican government telegraph lines used
an alphabet differing slightly from the American and Continental
Morse. However, at that time, the Continental Morse alphabet was
prescribed for use on these lines and it is believed that the use
of the old alphabet has entirely ceased on Mexican lines. However,
skilled American operators would have no difficulty in picking up
this alphabet if it were found to be in use.
Radio communication is, by International Convention, invariably in
Continental Morse.
Telegraph Alphabets
Character American Morse Continental Morse
or International Code
A . - . -
B - . . . - . . .
C . . . - . - .
D - . . - . .
E . .
F . - . . . - .
G - - . - - .
H . . . . . . . .
I . . . .
J - . - . . - - -
K - . - - . -
L -- . - . .
M - - - -
N - . - .
O . . - - -
P . . . . . . - - .
Q . . - . - - . -
R . . . . - .
S . . . . . .
T - -
U . . - . . -
V . . . - . . . -
W . - - . - -
X . - . . - . . -
Y . . . . - . - -
Z . . . . - - . .
1 . - - . . - - - -
2 . . - . . . . - - -
3 . . . - . . . . - -
4 . . . . - . . . . -
5 - - - . . . . .
6 . . . . . . - . . . .
7 - - . . - - . . .
8 - . . . . - - - . .
9 - . . - - - - - .
0 ---- - - - - -
Period . . - - . . . - . - . -
Question Mark - . . - . . . - - . .
Comma . - . - - - . . - -
The following example will show some of the errors that creep into
messages prepared with the cipher disk and transmitted by radio:
Message
Radio Douglas de El Paso, 2 H 71, twenty-fifth, 9:00 a.m.,
Govt. To C.O., Sixth Brigade, Douglas, Arizona:
JPRZI RDJSG XTRMJ USFPC RECLA BCPCB OAXPK QEQKF PPZAE BKUTT JHWEU
AHPZE EZOLT HKXPH KIHAV DRODN IAPZC LVUMP KFUBV VTVNV EFVZV TLVQS
BKAHQ NVKVF MGJTH OWBGN WWEPO LJKFP HEXKW CPDLZ JWSQC JVKIG HTJHT
EGAHA GDXXK BSPPK DIAVZ VQONC HOVDA VZQKW FNVON RPVGH CUFPV SFPIE
TOZOD WGYFE AWNJY KOEDW UMELD NOBUH MUPQL GYOPP ODBAB UFUUC AEOJW
RDIPK WMOKV OMICW CKPIH LUMSY YOSBG WOPHV PKOMO PHGER
Smith.
The key word is ATCHISON, the cipher disk being used and the setting
changed for every letter of the message. The letter X indicates a
period where it is evidently not a letter of a word.
Deciphering the message with this key and method we have:
RELIA BLEIN FORGF TIONF ROMCA SASGR ANDES RECEI LEDHE RETHA TAMOU
NTEDD ETACH MSKTL EFTTH ERELA STNIG HTTOE SCORT SHYMP ENTOF ARMSA
NDAMM UNITI ONTOB ESMUG GLEDA CRJXS BORDE RNEXT FRIDA YNIGH TATAP
OINTT WELVE MIENX FKOSB
Beyond this point the message, if we continue the deciphering process,
is unintelligible. The sense fails at the first P of the cipher group
BSPPK. We have translated B as M with disk A to N and S as I with
disk A to A. The last words that make sense are A POINT TWELVE MI;
clearly the rest of the last word is LES and this is represented by
PPK. Putting P=L then A=A and putting P=E then A=T. In other words,
the encipherer forgot to change his disk setting, A to A, after
enciphering I into S and enciphered L into P with the same setting,
A to A. Continuing the deciphering on this basis, we have:
LES EASTO FDOUG LAS.T HISIS INYOU RDIST RICT. WILLY OUTAK ENECE
SSARV STEPS TOPRE VENTT HISSH IPMEN TFROM GOING SKZRX LEADE ROFSM
UGGLE RSSAL DTOBE JUANH ERNAN DEZOF NACO.
The minor errors underlined above are not difficult to correct except
the sixth word in the eighth line. They will be taken up however for
analysis of cause of error.
Line 1, GF should be MA. Putting the latter into cipher we find the
letters of the cipher should have been GO instead of MJ. This is
clearly a telegrapher's error, --. --- becoming -- .---
Line 2, L should be V. The corresponding cipher letter should be F
instead of P. This is an error of the encipherer in copying.
Line 2, SK should be EN. The corresponding cipher letters should be YU
instead of KX. Another telegrapher's error, -.-- ..- becoming -.- -..-
Line 3, Y should be I. The corresponding cipher letter should be L
instead of V. Another error in copying by the encipherer.
Line 4, JX should be OS. The corresponding cipher letters should be
FK instead of KF; an error on the part of the encipherer in copying.
Line 7, V should be Y. A mistake in copying.
Line 8, SKZRX. If we take X as a period, then this line might be OVER,
the R being correct and SKZ being in question. The corresponding
cipher letters are AEO and if we encipher OVE we get ETJ. Here again
we have a telegrapher's error, . - .--- becoming .- . ---
Line 9, L should be I. The corresponding cipher letter should be K
instead of H; an error in copying by the encipherer.
The errors by the encipherer above noted are fairly common ones. These
and similar errors are usually found when a cipher message, prepared
as a rough draft by the encipherer, is copied by a clerk and a careful
check of the copy is not made. The letters mistaken depend, of course,
on the encipherer's hand writing or printing. Other errors, besides
those noted, are the confusion of C, G, and Q; I, and J; B and R, etc.
The error by the encipherer, in not changing his disk setting for
one letter and thus throwing out the whole process of deciphering,
would not have occurred had he put the message into eight columns or a
multiple thereof and enciphered each column with one disk setting. This
latter method is also very much faster.
Telegraphers' errors in cipher transmission are common and often
very confusing. Note should be taken as to whether Continental or
American Morse was used for transmission. An analysis along the lines
indicated will usually develop the error and correction. If not,
a repetition should be demanded, calling attention, if possible,
to the particular groups that are not clear.
The deciphered and corrected message is:
"Reliable information from Casas Grandes received here that a
mounted detachment left there last night to escort shipment of
arms and ammunition to be smuggled across border next Friday
night, at a point twelve miles east of Douglas. This is in your
district. Will you take necessary steps to prevent this shipment
going over? Leader of smugglers said to be Juan Hernandez of Naco."
Another remarkable example of errors in transmission by American Morse
is the following: A message, partly in cipher and partly in plain text,
contained the cipher words
GA GTXIEIT EIDISXQ
This, deciphered as far as possible by the alphabet determined by
analysis of the rest of the cipher, read
SU SME_Y_M Y_O_GES
It was finally decided that the context required a single word like
SUSPENDIO or SUSPENDIOLES for this cipher group. An examination along
this line showed that the cipher words should have been
received G A G L X C U R D P X G
and were received G A G T X IE IT EI D IS X Q
and that there were five errors in transmission in these three cipher
groups alone.
NOTES
[1] Occurrence rare, usually in proper names.
[2] The method used is not the most satisfactory one for several
reasons and a better method is that of writing the message in multiples
of the key and enciphering the columns as already described.