picking out all the words beginning with B; then a third time for
those with C, and so on with D, E, and the rest, till he had
transcribed the whole, and his copy was no longer in the fortuitous
disorder of the original, but in what we call _first-letter_ order.

A still later scribe making a copy of this vocabulary, or possibly
combining two or three lists already in first-letter order, carried
the alphabetical arrangement one stage further; instead of
transcribing the A-words as they stood, he went through them, picking
out first those that began with Aa-, then those in Ab-, then those in
Ac-, and so on, to Az. Then he did the same with the B-words, picking
out first all in Ba-, then Be-, Bi-, Bl-, Bo-, Br-, Bu-, By-; and so
exhausting the B-words. Thus, at length, in this second recension, the
Vocabulary stood, not yet completely alphabetical, but alphabetized as
far as the second letter of each word.

All these stages can actually be seen in four of the most ancient
glossaries of English origin that have come down to us, known
respectively, from the libraries to which they now belong, as the
Leiden, the Epinal, the Erfurt, and the Corpus (the last at Corpus
Christi College, Cambridge). The Leiden Glossary represents the
earliest stage of such a work, being really, in the main, a collection
of smaller glossaries, or rather sets of glosses, each set entered
under the name of the treatise from which it was extracted, the words
in each being left in the order in which they happened to come in the
treatise or work, without any further arrangement, alphabetical or
other. It appears also to incorporate in a final section some small
earlier vocabularies or lists of names of animals and other classes of
things. In order to discover whether any particular word occurs in
this glossary, the whole work from beginning to end must be looked