Secondly, languages which is he said to have spoken well, but as to which the evidence of sufficient trial is not so complete.

Thirdly, languages which he spoke freely, but less perfectly.

Fourthly, languages in which he could merely express himself and initiate a conversation. I shall add:—

Fifthly, certain other languages which he had studied from books, but does not appear to have spoken.

And lastly, dialects of the principal languages. This order, of course, precludes all idea of a scientific classification[568] of the languages according to families.

I.—Languages frequently tested, and spoken with rare excellence.[569]

II.—Stated to have been spoken fluently, but hardly sufficiently tested.