124, _note_ 3), γ᾽, δ᾽, ε᾽ ϛ᾽, ζ᾽, η᾽, ι᾽, ιε᾽ taken from King Henry II’s Library; the rest (i.e. θ᾽, ια᾽, ιβ᾽, ιγ᾽, ιδ᾽, ιϛ᾽) are those ἂ αὐτοὶ πανταχόθεν συνηθροίσαμεν, or, as the Latin runs, “quae undique corrogare licuit:” these, of course, were not necessarily his own, one at least (ιγ᾽, Act. 9, Paul. 11) we are sure was not. Although Robert Stephen professed to have collated the whole sixteen for his two previous editions, and that too ὡς οἷόν τε ἦν ἐπιμελέστατα, this part of his work is now known to be due to his son Henry [1528-98], who in 1546 was only eighteen years old (Wetstein, N. T., Proleg., vol. i. pp. 143-4). The degree of accuracy attained in this collation may be estimated from the single instance of the Complutensian, a book printed in very clear type, widely circulated, and highly valued by Stephen himself. Deducting mere _errata_, itacisms, and such like, it differs from his third edition in more than 2,300 places, of which (including cases where π. or πάντες stands for _all_ his copies) it is cited correctly 554 times (viz. 164 in the Gospels, ninety-four in St. Paul, seventy-six in the Acts and Catholic Epistles, 220 in the Apocalypse), and falsely no less than fifty-six times, again including errors from a too general use of πάντες(176). I would not say with some that these authorities stand in the margin more for parade than use, yet the text is perpetually at variance with the majority of them, and in 119 places with them all(177). If we trust ourselves once more to the guidance of Mill (Proleg. § 1228), the folio of 1550 departs from its smaller predecessors of 1546, 1549, in 284 readings(178), chiefly to adopt the text of Erasmus’ fifth edition, though even now the Complutensian is occasionally preferred (e.g. εὐλογήσας Matt. xxvi. 26), most often in the Apocalypse, and that with very good reason. Of his other fifteen authorities, ια᾽ (= Act. 8) and ιϛ᾽ (= Apoc. 3) have never been identified, but were among the six in private hands: β᾽ certainly is Cod. D or Bezae; the learned have tried, and on the whole successfully, to recognize the remainder, especially those in the Royal (or Imperial, or National) Library at Paris. In that great collection Le Long has satisfied us that γ᾽ is probably Evan. 4; δ᾽ is certainly Evan. 5; ε᾽ Evan. 6; ϛ᾽ Evan. 7; η᾽ Evan. L; ζ᾽ he rightly believed to be Evan. 8 (above, p. 191, note); ι᾽ appears to be Act. 7. Of those in the possession of individuals in Stephen’s time, Bp. Marsh (who in his “Letters to Mr. Archdeacon Travis,” 1795, was led to examine this subject very carefully) has proved that ιγ᾽ is Act. 9; Wetstein thought θ᾽ was Evan. 38 (which however see); Scholz seems to approve of Wetstein’s conjecture which Griesbach doubted (N. T., Proleg., Sect. 1. p. xxxviii), that ιβ᾽ is Evan. 9: Griesbach rightly considers ιδ᾽ to be Evan. 120; ιε᾽ was seen by Le Long to be Act. 10: these last four are now in the Royal Library. It has proved the more difficult to settle them, as Robert