Cryptography is a science of deduction and controlled experiment; hypotheses are formed, tested and often discarded. But the residue which passes the test grows and grows until finally there comes a point when the experimenter feels solid ground beneath his feet: his hypotheses cohere, and fragments of sense emerge from their camouflage. The code ‘breaks’. Perhaps this is best defined as the point when the likely leads appear faster than they can be followed up. It is like the initiation of a chain-reaction in atomic physics; once the critical threshold is passed, the reaction propagates itself. Only in the simplest experiments or codes does it complete itself with explosive violence. In the more difficult cases there is much work still to be done, and the small areas of sense, though sure proof of the break, remain for a while isolated; only gradually does the picture become filled out.

In June 1952 Ventris felt that the Linear B script had broken. Admittedly the tentative Greek words suggested in Work Note 20 were too few to carry conviction; in particular they implied an unlikely set of spelling conventions. But as he transcribed more and more texts, so the Greek words began to emerge in greater numbers; new signs could now be identified by recognizing a word in which one sign only was a blank, and this value could then be tested elsewhere. The spelling rules received confirmation, and the pattern of the decipherment became clear.