Dorabella & D'Agapeyeff Ciphers - Solutions
D'Agapeyeff Cipher - Solution
Finale
I had assumed up to this point that the phrase April'39 was a Transposition Key - hence, keeping tabs on which columns were being shifted and to where (in the matrix). Armed with what must surely be the correct Key, I tried various attacks on column 13 (or rather - column 2, since I was now adopting the Chinese fill method described by D'Agapeyeff). No joy whatsoever. Column 12 (or 3) also yielded nothing.
I sat back to take stock of things. With another 182 cipher groups still to be solved (91 characters of plain text), probably representing some 17 or 18 words and hence 17 or 18 Porta half-alphabets to recover, the workload seemed daunting. Had I really done all this work just to recover one column? Also, the Porta recovery didn't have any Indicator letter (unlike the examples in the book). Also, the Transposition seems to be rule-based, rather than Keyword-driven - in which case, April '39 has to be the underlying Plain, demanding further Plain from the remaining 182 cipher groups. Would d'Agapeyeff really have set such a stinker in what is after all an introductory book on cryptography? Surely not!
I thus set about finding an alternative explanation.
Solution
The weak point (IMHO) is the 39 recovered in the last cell of the LONDON template. It fits perfectly (in terms of 1939 reduced to '39 and with the Porta recoveries and methodology). And yet ............. And what if? Hmm ............
Thinking back to the Polybius Square, the Keyword was DAGAPEYEFF - which reduced to DAGPEYF after duplicate letters were removed (working from Left to Right). Could a similar process have been applied to the LONDON template, with the last cell being the result of something reducing to '39'? [Since there is already duplication present on the lower cells (the value 93 occurring twice), this argument would need to apply only to the top row values.]
Looking at the 39, the 3 is a 2nd-digit (from [12345]) and the 9 is a 1st-digit (from [67890]). We have already recovered A(65) and L(82) on the top row. It follows that, if duplication rules have been applied, the only value which could precede 3 would be a 6 or an 8 (63=letter F or 83=M). Likewise, the only value which could follow 9 would be a 2 or a 5 (92= O or 95=S). The 95=S can be ruled out as FS or MS would seem an unlikely start to a word. But FO or MO looks promising. Further, any additional letters would need to fall within the same limitations. Click to see.