this observation does not in any case give the preference to one reading over another. It must also be remarked, that those instances are few indeed, where addition, subtraction, or muta tion leave the sense precisely the same. If I perceive no diffe rence, it is, perchance, perceived by another : if I see it not to-day, I may have seen it yesterday, or I may see it to-morrow. If there be no difference as to doctrine, there may be as to elegance, simplicity, emphasis, connection, or some kind of parallelism.
9. A single manuscript does not establish a various reading, because it argues merely the carelessness of the transcriber, especially in the case of omission ; provided only that the received reading is according to the analogy of the faith, — otherwise, Canon 22 comes into play.
In Canons 9—12, and 40—43 (compare his Prolegomena n. 108), our author treats of the number of manuscripts. But, in the first place, the antiquity and diverse origin1 of MSS. is
1 " Bengel," says Tregelles, " clearly observed the difference existing in MSS. and versions, so that he saw that in a general manner they belonged to two different families. The one embraces the most ancient documents whether MSS. or versions, the other comprises -the greater part of those that are more recent. It was thus that a ground-plan of a division into Alexandrian and Byzantine families was laid down : these were termed by him, African and Asiatic."
Bengel thus expresses himself in his App. Grit. Ed. II., pp. 425, 426, —
"1. Codices, Versiones, et patres in duas discedunt familias, Asiaticam, et Africanam.
" 2. Ex Africana est cod. Al. pane solus ; (quia codices Africani fere deleti sunt), at quamlibet multis par : cum versione ^Eth. Copt. Lat. Ex Asia- tica ceteri fere testes. Latinae version! subordinantur cod. Grsecolatini et Latinizantes.
"3. Lectio familiar Africana; semper antiqua est, sed tamen non semper genuina : prsesertim ubi aberratio in proclivi erat.
" 4. Codices Asiatici, quamvis multi, exiguum sa3pe pondus habent : nulla pnesertim antiqua versione stipati.
" 5. Africana lectio ssepius excessum Asiaticum redarguit : Asiatica lectio interdum medetur hiatui Africano.
" 6. Consensus plurium vel certe prsecipuorum testium ex utraque familia magnum est genuinse lectionis criterium.
" 7. Prasclarum esset adjumentum, si duo testes, insignis codex Graecus, et insignis aliqua versio, sumerentur : quorum consensio primum, deinde dis- crepantia non ipsam quidem ubique decisionem daret, sed tamen iter ad