CHAPTER X

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.