Dorabella Transposition

the case for a second layer Dorabella's letter frequencies fit English at the 30th percentile. Its n-gram order fails at the 0th. Transposition is the one operation that does exactly that: it preserves frequencies perfectly and destroys order completely. So a rearrangement layered on the substitution is one of only two hypotheses that fits both halves of the evidence.

and the reason it is dangerous At 87 characters, a search free to pick both a substitution and a rearrangement will manufacture English-looking output by chance. That is not a solution, it is a coincidence with good PR.

So this bench applies rules you specify from a catalogue of real historical schemes. It does not hunt for one. A transposition theory has to arrive with an independent reason for its particular rule — something Elgar might plausibly have done — not a rule reverse-fitted to make the output read nicely.

the symbol stream The Dorabella symbols, or anything you paste. Transposition acts on whole symbols, so nothing is created or destroyed — the same 87 glyphs come out in a different order, and every frequency statistic is unchanged by construction.

the only honest way to judge it Anneal a substitution over the rearranged symbols and compare against real English solved at the same effort. Because the rearrangement is fixed by you, this is a fair single test — not a search that gets to pick its own answer. Watch the percentile, not the prose: a genuine solution sits near the middle of the English band.

a bounded comparison, not a search Runs the fixed catalogue once each, at one effort, and ranks them. This is legitimate because the catalogue is small and chosen for historical reasons rather than fitted to the data — but the same caution applies: the best of a dozen arbitrary rearrangements will beat the worst by chance alone. Read it as a screen, and treat a single standout as a lead to investigate, never as a result.