HE art of secret writing has, in its long and eventful history, masqueraded under many different names. Some of these are steganography, steganology, cryptology, cryptography, and cipher writing.
At the present time cipher, which in ordinary usage is synonymous with cryptogram, has come more especially to mean any particular system of secret writing, as the Gronsfeld cipher, or the Blair cipher, rather than some specimen of writing, or cryptogram, in that system.
But cryptography, too, derived from the Greek words kruptos — hidden, covered, secret—and graphia—writing—has survived in this struggle for existence, and may now be styled as the accepted scientific title.
Following are the names by which you can shake hands with this most interesting study in several other languages. Cryptography and cipher will be seen to crop up in most of them, in some form or other.
| English: | cryptography; cipher. |
|---|---|
| Esperanto: | kriptografio; secreta scribado. |
| French: | cryptographic; chiffre. |
| German: | kryptographie; chiffrirkunsi; chiffrcschrift; geheimschrijt; geheimschreibekunst. |
| Italian: | criptografia; crittografia; caracteres secretos; cifra. |
| Portuguese: | cryptographia. |
| Russian: | tainopisanie; shifre. |
| Spanish: | criptografia; cifra. |
But the name under which this obscure department of human activities has been known is not all that has been changing. Cryptography itself has evolved from an art dealing with relatively simple methods of secret writing into a science of a complexity probably undreamed of by the old cryptographers.
Cryptanalysis is that branch of the science which treats of the resolution of ciphers without their keys. Several of the more elementary of these methods have already been described in Solving Cipher Secrets.
Two of these articles, published in FLYNN'S for February 21, and August 15, contained alphabetical frequency tables, one of the uses of which is to assist in the solution of the simple substitution cipher.
It may be remembered from the above that E is the most frequently used letter in the English alphabet. Consequently, a tabulation of the characters of a simple substitution cipher will ordinarily reveal at once just what cipher character is used for E for the reason that it generally predominates over all others by a considerable margin.
Unfortunately E is about the only letter that can be determined by this method. And in many messages, the shorter ones in particular, even E has an uncanny knack of not being the most frequently used letter. Quite often it comes out second or third in frequency, or even still further down the scale, being overtopped by T, O, or some other letter or letters of high frequency.
From this it is obvious that some further method is needed in the solution of such ciphers unless mere guess work is to be depended upon.
To overcome this difficulty many methods of determining letters, and of differentiating between them, have been devised.
Various of these methods will be treated from time to time in Solving Cipher Secrets. That described herewith may be used advantageously with a simple substitution cipher having normal divisions between words.
And it depends on the fact that the various letters of the alphabet, considered from their total frequencies as compared with their frequencies as initials and finals of words, bear certain well defined and easily recognized characteristic relationships, which will enable the decipherer to distinguish between them to a greater extent than would be possible by a consideration of their total frequencies alone.
A study of the graphic table of total, initial, and final frequencies herewith will make this clear.
This table, which is almost self-explanatory, has been constructed to conform with the alphabetic frequencies enumerated in the next column:
| TABLE OF ALPHABETIC FREQUENCIES | |||
| Total | Initial | Final | |
| A....................... | 78.1 | 17.4 | 0.9 |
| B....................... | 12.8 | 9.2 | 0.2 |
| C....................... | 29.3 | 10.0 | 0.3 |
| D....................... | 41.1 | 7.0 | 20.7 |
| E....................... | 130.5 | 4.6 | 42.0 |
| F....................... | 28.8 | 8.2 | 8.6 |
| G....................... | 13.9 | 2.7 | 5.6 |
| H....................... | 58.5 | 13.1 | 3.9 |
| I....................... | 67.7 | 13.0 | 0.0 |
| J....................... | 2.3 | 1.1 | 0.0 |
| K....................... | 4.2 | 1.2 | 2.2 |
| L....................... | 36.0 | 3.4 | 6.3 |
| M....................... | 26.2 | 7.0 | 2.8 |
| N....................... | 72.8 | 4.0 | 20.1 |
| O....................... | 82.1 | 14.7 | 10.3 |
| P....................... | 21.5 | 8.4 | 1.5 |
| Q....................... | 1.4 | 0.5 | 0.0 |
| R....................... | 66.4 | 5.3 | 13.1 |
| S....................... | 64.6 | 13.4 | 25.4 |
| T....................... | 90.2 | 36.1 | 24.6 |
| U....................... | 27.7 | 3.0 | 0.2 |
| V....................... | 10.0 | 1.3 | 0.0 |
| W....................... | 14.9 | 10.8 | 2.3 |
| X....................... | 3.0 | 0.0 | 0.3 |
| Y....................... | 15.1 | 4.4 | 8.7 |
| Z....................... | 0.9 | 0.2 | 0.0 |
| ———— | ———— | ———— | |
| ....................... | 1000.0 | 200.0 | 200.0 |
In the graphic table the unbroken line shows the total frequency per 1,000 letters of each letter of the alphabet, based on a count of 5,000 letters straight English text,
including initials and finals. For working convenience the letters are arranged in the order of their descending frequencies.
The dash line, and the dot line show the frequencies per 200 of each letter of the alphabet, based respectively on counts of 5000 initials only and 5,000 finals only, in straight English text. Words of but a single letter are, of course, excluded from this count.
Since English words have an average length of 5 letters—see FLYNN'S for August 15, the frequency per 200 of any letter as an initial or final is practically equivalent to its frequency per 1,000 in a total count of all letters, including initials and finals.
Thus, in a message of 1,000 letters—or 200 words—E should average a total occurrence, including its use as initials and finals, of 130.5 times.
As an initial it would occur on an average of 4.6 times per 200 in a count of initials only; or the same number of times as an initial in the total count of 1,000.
And as a final it would occur on an average of 42.0 times per 200 in a count of finals only, or about the same number of times as a final in the total count of 1,000.
Now we have arrived at the main event, namely, the method of applying the tables toward the identification of letters in the solution of ciphers.
To illustrate, suppose you have a cipher of whose characters you have prepared a frequency table. It might be that you would be unable to decide which of the most used characters was E. And it would be hardly possible from their total frequencies alone, to positively determine the identities of any of the other characters.
But if, now, you proceed to make frequency tables of the initials and finals, important differences will at once be noted.
For example, as the tables herewith would lead you to expect, E will ordinarily occur more frequently as a final than as an initial, while T will occur more often as an initial than as a final. This one difference should identify these two characters at once.
Further, O will occur about equally as initial or final; A often as an initial, but seldom as a final; and so on. The relative total frequencies, as compared with initial and final frequencies, also play a prominent part in these determinations.
Considerable variations in cryptograms from this, or any other table, must, of course, be expected, and allowed for. The longer the cipher, the more nearly will the counts made from it correspond with those of the table.
Favorable results should be had with messages of fifty words—two hundred and fifty letters—or more. But even in shorter ones the tables are of definite aid. Of course, it is possible to solve the simple substitution cipher with normal divisions between words by the method described in FLYNN'S for May 16. But the point is here that we are getting away from guess work.
By this method the decipherer should be able to determine the values of a number of the cipher characters by their mathematical relations in total, initial, and final counts, before any attempt is made to decipher the message itself.
To find only E and T in this way is to know the values of nearly one-fourth of all the characters in the cipher. And to determine E, T, O, and A, would mean that nearly forty per cent of the letters had been identified. The rest should then be easily discovered by context.
This method of determining letters by their occurrence as initials or finals is not at all new. It is foreseen even in the thirteen rules of Sicco Simonetta, written in 1474 A.D., the first of which takes account in a general way of the terminations of Latin words.
However, as presented here, this idea is not limited merely to the identification of letters as above described.
To illustrate, you will find appended two ciphers. In one of these the normal word divisions have been retained throughout. While in the other arbitrary divisions have been used.
Consequently, if both of these be tested by the above method, only one will react normally. Similarly, a message in some other cipher, say, the Gronsfeld normally spaced, would not come out in agreement with the tables.
Again, the relative frequencies of letters in different languages betray certain characteristic differences in their total, initial, and final counts, which should assist the decipherer in discovering if a cipher of the type being discussed is in English or some other language.
Equipped with tables similar to the above for other languages, the decipherer can go a long way identifying beforehand the language used in a given specimen. Besides there are numerous other tests that assist in this of which you will hear later on. So, in order to derive the greatest benefit from the following two ciphers, the fans should confine themselves as far as possible to the principle explained in this article.
Try first to find which of the two ciphers is normally spaced. Then decide on tentative values for as many letters of this one as you can before trying to decipher it yourself.
Finally, after you have read the cipher, compare the true values with your predetermined ones, and see how many you had correct.
We would be glad to hear what good fortune attends your first experiment with this method.
Let us know how many, and what letters you were able to predetermine correctly.
And also of any special method used in assigning their values that would be of interest to the fans.
Which of the two following ciphers can you solve by this method, and how will you solve the other one?
Look in the next Solving Cipher Secrets for an explanation of the mystery. In the meantime— Here's your material:
CIPHER No. 1.
WK L VFUBSWR JUDPLVZU LWW HQ L QDF LS KHU XVHGIR UV HFUHW PHPR UDQGDD QGF RUUHVSR QGHQF HW ZRWKRX VDQGBHD UV DJREBMX OLXVFDHVDU, WK HU RPDQJHQ- HUDO, VWDWHVPDQ, DQGZUL WHU. HDFK OHWW HUR IW KHPHVVD JHL VUHS UHVHQWHGLQF LS KHUEB WKHIRX UWK OHWW HULQD GYDQ- FHR ILWLQD OS KDEHWL FDOR UGHU, GEH LQJ XVHG IRUD, HIRUE, DQGVRR QWKUR XJKWK HDOSKD EHW, FIL QDOO BEH LQJ XVHG IRUC. DPRUJW KHV HYHUDOP HWKRG VWKDWF DQE HXV HGW RUHVR OYHDF LS KHU RIWK L VWBSHPDBE HPH QWLRQHGW KDWXVHGZ LW KWKHDXJX VWXVF LS KHU GHVFULE HGLQI QBQQV IRUI HEUXDU BW ZHQW BILUVW.
CIPHER No. 2.
VK KSZ ZEGTIVKTPO PQ V NPOKSJ KTNZ KSZ JZV SVY JBOL JP MPD KSVK KSZIZ IZNVTOZY WZKDZZO NZ VOY KSZ XPOKTOZOK WBK V JNVMM JKIZVN. KSTJ T XIPJJZY, VOY DSZO T SVY GIPXZZYZY JPNZ YTJKVOXZ QIPN KSZ JVZ, T JVDV RPPY DVF WZQPIZ NZ JPNZKSTOR KSVK IZJZN- WMZY V RIZVK QTIZ, WBK VJ T YIZD QZVIZI T YTJXPCZIZY NF ZHPI, QPI DSVK T SVY KVLZO QPI V QTIZ DVJ V XVJKMZ PQ IZY XPGGZI, DSTXS KSZ WZVNJ PQ KSZ JBO NYVZ KP VGGZVI VK V YTJKVOXZ MTLZ QMVNZJ.
The first of the two ciphers in Solving Cipher Secrets for August 15 was a modification of the dot writing explained in that article, first printed in FLYNN'S for July 25.
This cipher made use of a triformed alphabet, the three different shapes of types acting as substitutes for the three different positions of the dots, as represented by the figures 1,2, and 3, in accordance with the following plan:
Each dot over the line, or (1), was represented by a Roman type; each dot upon the line, or (2), by an Italic type; and, each dot under the line, or (3), by a bold face type.
Substituting these values for the types, this cryptogram can be deciphered in exactly the same manner as the original dot writing, already explained in the "above-mentioned issue.
Here is a small sample of it; enough to show how it is done:
| CIPHER: | The bearer of this message is—etc. |
|---|---|
| SUBSTITUTING: | 111 111323 13 2322 1 133321 23—etc. |
| REGROUPING: | 1111 1132 3132 3221 1333 2123—etc. |
| DECIPHERING: | B E A R E R —etc. |
The apparent, or external meaning of this cipher is thus quite different from the internal meaning, which follows:
BEARER IS A COWARDLY SPY. SHOOT HIM AT SUNRISE!
Imagine how the poor fellow must have felt. The message he carried was really his death warrant. He expected to be treated like a gentleman and a soldier. What he got was a dose of lead early the next morning.
In illustrating this triformed alphabet cipher of Blair's, three easily distinguishable forms of type have intentionally been used.
In practice, whether in printed or written character, these differences could be made so small that, as Blair himself put it, a correspondence could be conducted " without any suspicion of a cipher being present." Thi s same is, of course, also true of Bacon's biliteral cipher, described in FLYNN'S for April 25.
The second of the two Blair ciphers in last Solving Cipher Secrets is the most complicated of all his examples.
If you failed to note that the three left-hand columns of Blair's original Alphabet and Key used the (.), or space, and each letter of the alphabet once only, you probably failed also to decipher the cipher. For this arrangement, or, in other words, this key within a key, is the secret of Blair's literal cipher.
To solve the cipher, first number the nine rows of these three columns thus:
| 1 | b | c | s | 4 | k | d | y | 7 | w | h | i | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2 | g | m | t | 5 | q | f | a | 8 | x | n | o | ||
| 3 | j | p | u | 6 | v | l | e | 9 | z | r | . |
Then you must substitute for each letter of the cipher the figure that appears to the left of the row of that letter in the above portion of the key.
The cipher then becomes a series of figures that can be resolved by the same method used for Cipher No. 2 in the July 25 issue, as explained in FLYNN'S for August 15.
A short fragment is sufficient for illustration:
| CIPHER: | i i j i j . r j m s v u j l y a | —etc. |
|---|---|---|
| SUBSTITUTING: | 7 7 3 7 3 9 9 3 2 1 6 3 3 6 4 5 | —etc. |
| GROUPING: | 77 37 39 93 21 63 36 45 | —etc. |
| DECIPHERING: | T H E . G E N E | —etc. |
The message in full is: "THE GENERALITY OF CIPHERS ARE COMPLEX AND DIFFICULT TO WRITE IN PROPORTION TO THEIR INTRICACY."
As a short cut in deciphering the above, the first letter of each pair may be said directly to indicate the row, and the figure of the second letter the column, of the intended letter in Blair's key.
A careful consideration of this system will show that there are just nine substitutes possible for each letter of the Alphabet and Key. This variation of Blair's cipher consists then in the use of a multisubstitutional alphabet of (9x81=) 729 pairs of characters.
" GIVE US MORE CIPHERS! " say the fans.
So the cipher editor, who is nothing if not obliging, has gathered together this choice assortment of home grown nuts for the fans to wreak their vengeance on.
Look them over!
No finer array of nuts ever graced the boughs of any nut tree on earth.
Each particular nut is guaranteed by its producer to be first class. And the editor, who has critically inspected them all, absolutely backs up this guarantee without any mental reservation whatsoever.
So if you have any strength left after cracking the first two nuts in this article, don't put the old nut cracker away until you have tried it on these:
CIPHER No. 3.
This one is from Delbert M. Skelly, Newark, Ohio, who says it might be called a humorous cipher, although there is sense in its nonsense. Mr. Skelly says he got the idea from the cipher in Arthur P. Hankin's story, "The Sea-going Elephant," in Argosy-Allstory for May 2. He has entitled his cryptogram:
A TRAGEDY
The interest in the rifle shots that brought down the parachute on the One Stick left us with a payless day. That is, we of the Circle-A. The tadpole in the parachute saw more interest in the star seen by the six-gone-wrong than the rifle shots. The six-gone-wrong of the Circle-A and the tadpole all saw the star. The right-face moon and the six-gone-wrong twice fell on his head and body. The Circle-A went to the One Stick to get what we need. Although the rifle shots hit the six-gone-wrong from the Circle-A, the twins were not hurt. The star seen by the Circle-A and caused by the rifle shots didn't get us what we need. The left-face moon seeing the One Stick and the six-gone-wrong together, swallowed the sinker and hook.
CIPHER No. 4.
Says W. B. Lava, of Chicago, Illinois: " I am offering here a cryptogram which I challenge anybody to decipher without the key. It is certainly indecipherable. This cipher was done on a typewriter, but if the fans can solve other typewriter ciphers, they certainly cannot solve this one."
5& )&( 4''%?6 )&( ;#? 9&71$ 4''%9 ;2)34&,2#!
CIPHER No. 5.
Without any hint as to its type, this excellent cipher, submitted by M. Walker, Akron, Ohio, should give you a run for your money:
EIHNE EORSN EOAHF RTTNI SOSPI OTUIT OEYHP SKKLO IDTLE NONOH OLWWY AGSTI ETKES OVOUL TOYDI WNTOA HKIOQ OURDT
CIPHER No. 6.
A real gem of cipher wisdom is buried in this cryptogram from C. W. T. Weldon, New York City.
36-66-78-76-44-39-49-33-59-49-32-56-78-76- 57-58-78-45-75-66-46-26-68-59-36-39-77-54- 76-26-36-48-87-48-57-59-69-54-87-39-23-64- 59-68-47-76-77-35-88-30-34-44-86-38-28-66- 69-44-59-58.
CIPHER No. 7.
Here is a cipher, submitted by Rev. James Veale, D. D., South Ozone Park, Long Island, the key to which is based on the principle of the ordinary clock dial. The message might be a very important engagement. Would you be able to keep it?
2-5-1-G-G-8-5-C-5-A-A-F-L-11-H-1-A-9-1- F-G-1-G-9-B-A-A-5-J-L-B-E-10-3-9-G-L-B- A-G-8-H-E-F-4-1-L-A-5-K-G-1-G-G-J-5-11- H-5-A-B-B-A.
That the ciphers in the more recent articles have been more difficult of solution is shown by the fact that fewer correct answers are being received to all the ciphers in each article.
Thus, while many succeeded in solving the No. 1 Gronsfeld, in the June 6 issue, only the following few also submitted the correct answer to the Gronsfeld No. 2 in time for their names to appear in this issue:
Frank Spalding, Wrangell, Alaska; Francis A. Gauntt, Chicago, Illinois; C. W. T. Weldon, New York, N. Y.; C. M. Eddy, Jr, Providence, Rhode Island; E W. Harlan, Chicago, Illinois; W. Walker, Akron, Ohio; F. D. Jackson, Denver, Colorado; James Olden, Medicine Hat, Alberta, Canada.
And the No. 6 Nihilist cipher, in June 27 Solving Cipher Secrets, stopped all of our correspondents up to date except the following, who succeeded in mastering all six:
M. E. Toevs, Bureau of Identification, Police Department, Detroit, Michigan; Francis A. Gauntt, Chicago, Illinois; Jame s Olden, Medicine Hat, Alberta, Canada; L. B. Pennock, Philadelphia, Pennsylvania; Mrs. S. J. E. SoUey, Rockport, Massachusetts.