one character of a set of characters was accepted, other characters in the set were included. It was impossible to cover more than a small part of any one subject, and the list of characters is illustrative rather than comprehensive, but it should be well balanced as far as it goes.
The choice of characters was checked by a closed circuit through the dictionaries 26 27 28 29 30 31 32 33 34 35. Starting with an English to kanji dictionary, the kanji for a selected English word was found, then continuing with the character dictionary, the rōmaji of the given kanji was found, and ending with a rōmaji to English dictionary, the kanji and English for the given rōmaji were found. Thus the final English word could be checked against the initial English word.
In the character dictionaries each character is followed first by the on pronunciation, second by the kun pronunciation, with English translations wherever possible, and finally by a table of compounds wherein the character appears. Although many of the individual characters no longer are used alone and appear only as components of compounds, they still are given archaic English translations, which would unbalance an abridged list of morphemes. Furthermore, certain grammatical morphemes do not occur in the character dictionaries because they have only phonetic renderings. It appears that the best way to illustrate the use of digitalized characters is by a dictionary listing analogous to Sanseido's33 . Each entry in the listing is punched on a separate punch card in the order rōmaji-kanji-kana-English. The deck of cards may be sorted, abridged, or augmented easily. Its present status is illustrated in Appendix E.
Each character in Nelson's dictionary32 is assigned its own number, whereas the characters in other dictionaries are located by page number. Inasmuch as the numbering in Nelson's dictionary provides a natural and definite identification, it has been adopted for the numbering of digitalized characters. It is easy to recover the character
18
