We believe that the resulting cryptogram could prove puzzling to any cryptanalyst who has met the cipher for the first time. It is true that he has, in the original square, a large number of intact minor units, provided he can restore them.
|
Figure 48 Cipher Requiring Little Analysis Plaintext: THIS WOULD SEEM TO BE... Cryptogram: T H W S E S I O E.... |
But these units are very tiny, and the several of them which stand on any one row are not continuous among themselves; thus his vowel-distribution, while approximately normal, would probably not satisfy his expectations. If, however, having failed to find a more satisfactory block arrangement, he attempts to match columns (it being remembered that he is accustomed to reading in all sorts of directions in order to discover plaintext fragments), he will most certainly discover the trigrams and trace their route. Afterward, however, having met and analyzed the cipher, it would probably occur to him to look for exactly this complexity whenever he discovers that he is dealing with a square whose key-length is divisible by 3. We have mentioned before the assistance which may be had from the mere knowledge that a certain method exists.
Complete-unit ciphers, of course, may be troublesome, but their complexities are necessarily confined to one small area. An irregular cipher, on the other hand, usually involves the entire text, and its complexities may be real. In this class we occasionally find ciphers in which a single cryptogram is impossible to break; and we find others in which the eventual solution of one cryptogram will not instantaneously provide the key to another enciphered exactly like it. Such ciphers are well worth analyzing, for surely, somewhere, they have their weaknesses; and most certainly any two cryptograms enciphered exactly alike should be decipherable with the same key.
The cipher shown in Fig. 49 is of the double-columnar type known in this country as the “United States Army” double transposition, and has, in fact, been authorized for use, under suitable conditions, in the military service of more than one country. As may be seen from the figure, this cipher is, in all respects, the columnar transposition of the [preceding chapter] accomplished twice in succession on the same text (decipherment being, as usual, the reversal of the encipherment process). The same rules apply here, as in the single columnar transposition, to the use of nulls, and to the advisability of avoiding the key-length which is a multiple of 5. It goes without saying that the block should never, under any circumstances, be allowed to work out as a square; this, in substance, would be the block unit of the Nihilist cipher. While the figure shows a primary cryptogram, taken off from the upper block, this, in practice, is never actually done. The columns of the upper block are always transferred direct to the rows of the lower one, and only the columns of the lower block are taken off as an actual cryptogram. In preparing this cryptogram, both blocks should be laid out at the same time; otherwise, there is danger that the operator may apply the first transposition and forget the second,
| Figure 49 The "United States Army" Double Transposition 1st Encipherment P A R A D I S E 6 1 7 2 3 5 8 4 R E G R E T C H Primary CryptogramA N G E I N S Y S T E M S X X X (Not usuallyX taken off) 2d Encipherment ENT, REM, EIS, HYX, TNX, RASX, GGE, CSX. P A R A D I S E 6 1 7 2 3 5 8 4 E N T R E M E I S H Y X T N X R A S X G G E C S Final Cryptogram: X N H S R X, G E T G I, R S M N E, E S A X T, Y X E X C. |
thus sending out a simple columnar transposition which carries the key to all of his other cryptograms. This cipher, as may be seen, has its points. Yet it will have been noticed that its use for military purposes was not authorized without restrictions.
The special hazards of military correspondence have already been mentioned: the huge volume of interceptable cryptograms; the ever-present knowledge as to probable subject-matter and more-than-probable words, including numbers and dates; the personal habits of individual operators; above all, the fact that much of the enciphering is necessarily done by operators who are not, in the first place, trained for their work, and who, very often, must perform this work rapidly under conditions which are far from conducive to clear thinking. These, however, are chiefly the hazards of the firing line. Back of the lines, where hazards are reduced, there may be a chance that a cryptogram will not be intercepted at all. It becomes possible, for many purposes, to make use of a cipher in which a single cryptogram, though probably read in the end, will resist the decryptor for the necessary length of time, several hours or several days. The double columnar transposition can be very resistant, especially when the key is long and the columns short, and can be made even more complicated by carrying it through still a third block, perhaps using a different key with each new block.
Why, then, would it not be possible to use such a cipher for general communication? To this, there are two answers. For transposition cipher, taken as a whole, has two very serious drawbacks.
First, a transposition, in order to be a good one, must be a transposition of the whole text, and not a series of short individual transpositions. Thus, it becomes possible that an error, either in the encipherment or in the transmission, will not be confined to one small area, but will garble the whole message. In this way, we have not only the delay during which the legitimate decipherer is attempting to decrypt his own message, but, should he fail, the danger which lies in having it repeated. The decryptor who has been provided with both the correct and the incorrect version of a same cryptogram, is often able to figure out both the system and the key.