71 



selves. He was furnished with materials for doing this by the 

 kindness of Captain Larcom and Mr. Petrie, the former of 

 whom placed at his disposal all the drawings of Ogham inscrip- 

 tions collected by the draughtsmen employed on the Ordnance 

 Survey of Ireland ; whilst the latter furnished him with nu- 

 merous and accurate tracings of inscriptions taken from his 

 own sketch-books. And here arose a question as to the best 

 mode of employing these materials. The common methods 

 of deciphering, which assume that the writing to be deciphered 

 is divided into words, were at once found to be inapplicable 

 to the Ogham character, the inscriptions in which are written 

 continuously. In seeking to frame a method applicable in 

 this and similar cases, Mr. Graves conceived the one which 

 he then proceeded to describe. 



This method rests upon the following principle : that in 

 any given language, or group of cognate languages, there is a 

 preference for particular sounds, and particular sequences of 

 sounds. 



In order to determine what are the favourite sounds or 

 sequences in a language, we must analyze considerable por- 

 tions of it in such a way as to exhibit its tendencies to repeat 

 and combine the several letters of its alphabet. This end is 

 arrived at by the construction of a table, which shows how 

 often, on an average, each letter is followed by each of the 

 remaining ones, in a passage of some determined length ; as, 

 for instance, a passage consisting of ten thousand letters. 

 With such a table at hand, it is not difficult to assign their 

 proper powers to the secret characters or ciphers in which a 

 document in that language is written. We have merely to 

 tabulate the sequences of the ciphers ; and, by comparing their 

 tendencies to repetition and combination with those of the 

 known letters, we readily arrive at a knowledge of their re- 

 spective powers. It is here assumed that the document to be 

 deciphered is of a reasonable length. This condition is indis- 

 pensable, inasmuch as the distribution of the letters in a pas- 



