^{page 1 note 1} In his Callimachea, vol. i. (Leipzig, Teubner, 1870)Google Scholar.

^{page 2 note 1} His collation of K is complete only for these and VI.

^{page 2 note 2} Called e in this paper.

^{page 2 note 3} Praef. XXXII., XXXV.–XXXVI.

^{page 3 note 1} See the Praefatio of his Weidmann text (Berlin, third edition, 1907)Google Scholar. It cannot be fairly said thai A ‘unus eminet’ (p. 11)in the x family, for C is nearly his equal; and Wilamowitz’ concession that a fourth apograph of the archetype may have existed, with GHI as its descendants (p. 13), is not borne out. His dependence on Schneider has caused several errors in his slight apparatus criticus.

^{page 3 note 2} Nigra, Costantino, in Rivista di filologia e d'istruzione classica (Turin), XX. (1892), pp. 194–232, 414–455, 516–543Google Scholar; XXI. (1893), pp. 51–96.

^{page 3 note 3} In J.H.S. XV. (1895), pp. 136–183, 251–313Google Scholar; XVII. (1897), pp. 45–62, 241–267.

^{page 4 note 1} So Omont, H., Inventaire sommairc des manuscrits grecs de la Bibliothèqve Nationale (Paris, 1886–1898)Google Scholar. Schneider gives it to the early sixteenth (Pratf. XXXIV.), on B. Haase's authority.

^{page 4 note 2} In Catalogus Codicum Graecorum Bibliothecae Ambrosianae, Martini, A. and Bassi, D. (Milan, 1906)Google Scholar.

^{page 4 note 3} By Delisle, L. in Bibl. Nationale, Catalogue des Manuscrits des Fonds Libri et Barrois (1888), p. 125Google Scholar.

^{page 4 note 4} In the British Museum Catalogue.

^{page 4 note 5} But the authenticity of this date is open to dispute.

^{page 4 note 6} Probably in or after 1496 (see below).

^{page 4 note 7} Praef. XXXV.

^{page 4 note 8} Ibid.

^{page 5 note 1} Angeli Politiani miscellaneorum centuriae primae ad Laurentium Medicem (Florence, Antonio Miscomini, 1489)Google Scholar.

^{page 5 note 2} Nigra uses V ² for the second edition (Brescia, 1496) and V ³ for the third (Aldine, 1498).

^{page 5 note 3} See Iriarte, , Regiae Bibliothecae Matritensis Codices Graeci MSS. (Madrid, 1769), pp. 499 sqqGoogle Scholar.

^{page 5 note 4} Wilamowitz (see op. cit. 1907, Praef. p. 14) calls it Taur. in his ap. crit. It is treated at length by Nigra, , op. cit., XX., pp. 225–232Google Scholar.

^{page 5 note 5} See Schneider Praef. XL., who refers to Miller, , Catal. des MSS. grecs de la bibl. de l'Escurial (Paris, 1848), p. 337Google Scholar. The Librarian of the Escurial kindly told me of its fate.

^{page 5 note 6} Schneider, Praef. XL.; cf. Montfaucon, , Bibl. bibl., p. 490Google Scholar.

^{page 5 note 7} See Montfaucon, , op. cit., p. 530bGoogle Scholar; and Schneider, Praef. XL., where for Herodoti librum de malignitate Homeri read Herodoti de uita Homeri.

^{page 5 note 8} Praef. XL.

^{page 5 note 9} Ibid.

^{page 5 note 10} Praef. XXXIX.

^{page 5 note 11} In the library of the Collegio di Spagna, No. 130; see Allen, T. W., Notes on Greek Manuscripts in Italian Libraries (1890), pp. 25–26Google Scholar.

^{page 5 note 12} By a ‘primary codex’ is here meant one which is not a copy of any other codex known to be extant.

^{page 6 note 1} See Schneider, Praef. XXIX.; Ludwich, A., Eudociae Augustae Procli Lycii Claudiani carminum Graecorum reliquiae (Teubner text, 1897), pp. 123–124Google Scholar. For Proclus it is L.

^{page 6 note 2} Orpheus and Proclus are not distinguished (cf. B, C, etc.), the end of the latter being followed by τλος τν ργοναυτικν το ỏρϕως.

^{page 6 note 3} I.e., two marginalia (the usual scholium at 1. 14; and at σπείρωσε 1. 33, ήλɩʂε παρ τ σπεɩρб⍵ τò είλσσ⍵ öθεѵ σπείρɩμα [?] τ σπαρτι +?), and five interlinear notes (over έοɩ τί κεѵ, 1. I, ђγоṳѵ í εọίκεѵ αλλọ over ѵδη 1. 33,: óѵọμα ѵύμɸης over τ Χεŭμα 1. 37, τ ѵ ρκαδíα: over εúτε, 1. 42, ỏτε: over τọѵτκί, 1. 44, τóτε).

^{page 6 note 4} C is ascribed to Johannes Rhosus, as is e where the hand is somewhat different; but the writing agrees in some notable characteristics, and the diversities are comprehensible when one remembers that Rhosus' activity as a scribe extended over fifty years (1447–1497); see Gardthausen, V., Griechische Palaeographie (Teubner, Leipzig, 1879) pp. 326–327Google Scholar.

^{page 6 note 5} See Schneider, Praef. XXIX.; Ludwich, A., op. cit., p. 123Google Scholar. For Proclus it is J.

^{page 7 note 1} See Nigra, , op. cit., XX.(1892), pp. 200–201Google Scholar; Schneider, Praef. XXIX.–XXX.; Ludwich, , op. cit., p. 122Google Scholar. For Proclus it is A.

^{page 7 note 2} Op. cit., XX. (1892), p. 200Google Scholar.

^{page 7 note 3} Ibid. p. 201.

^{page 7 note 4} See Codices Urbinates Graeci Bibliothecae Vaticanae (Rome, 1895), pp. 280–281Google Scholar.

^{page 7 note 5} Urbinas 146 consists of Apollonius' Argonautica, in the same hand as the latter part of 145 (Codices Urbinates… p. 281).

^{page 7 note 6} Ibid. p. 280.

^{page 8 note 1} Did he add Nicander from C ?

^{page 8 note 2} As, inter alia, their agreement at the larger lacunae shows.

^{page 8 note 3} Omissions of words are reserved for separate consideration.

^{page 9 note 1} In C, the parent of B, III. 182 ends a page, and in turning over the scribe (of C) omits 183 owing to homoeographon with 184.

^{page 9 note 2} K has this sign (∴) to attract attention, also at the end of the lacuna with which the x group close V. 17.

^{page 9 note 3} At IV. 147 (ἂṪεѵτ' K: γεѵτ' A by correction from γ-) A' alteration, though undoubted, is not very clear, and probably Rhosus did not discern it.

^{page 10 note 1} Add III. 173, where A has οίκήσοѵσα, K οІκήσασα, which C originally had but corrected to -οѵσα (B also -οѵσα). Presumably Rhosus conjectured -ασα in writing K and C, but repented of it in the latter case.

^{page 11 note 1} On this hypothesis C's correction of κεχρημѵ⍵ѵ (which B has) to -οѵ (I. 12) must be later than the copying of B from C. So for I. 10, where B has παρѵασíη, C παρρασíη by correction from παρѵ-. At V. 58 C's ταρᾶѵ (so AKeπ) shows no accent over the first α, but -ѵ has been added after erasure; B alone has ρραѵ, which is probably Rhosus' own conjecture.

^{page 11 note 2} Throughout this paper ‘major variations’ (or ‘mistakes’) refers to permutation of letters or syllables; omission or insertion of words or syllables; mis-division and non-division; and other points of importance. ‘Minor’ cases are those concerning accent, breathing, addition or omission of v ephelk. or of iota subscr.

^{page 11 note 3} Also, at I. 20 C's μλ' ἒüδροζ is nearer the true reading than B's μλα ἕüδροζ. At IV. 114, C rightly π' μεῖο B, π' μοῖο, with some others.

^{page 11 note 4} At I. 90 B's blurred reading is probably the correct ѵκλασαζ (-ασαζ cett.). At IV. 23 his πύργοσίѵ (-οίσί cett.) is due to later addition of -ѵ, perhaps by a second hand, such as added α in his α in his α χαίτδ, V. 13. At IV. 44 his αύ δείѵῖο is not so good as σύ of ACK, and no more metrical (see below, p. 14). At IV. 271 he gives ποσεῖ ωѵί δοѵτί, his first reading being influenced by κροѵτί just before.

^{page 12 note 1} A, alone among all the MSS., except II at VI. 108, sometimes has ‶in place of ‵ (μѵ, IV. 55, 185, 196; V. 93; VI. 65. δἒ, V. 23. μ, VI. 17). That this error is probably not original in A is suggested by VI. 108 (τἃѵ στíα Aπ: ταѵòѵ στíα SQ). Perhaps an original αi״, with the second ‵misread as (or corrected to) εѵ is responsible for aαιѵ of x (αι' cett.). at IV. 184.

^{page 12 note 2} C is twice guilty of sins of omission, fusing III. 130–131 (… στχѵѵ εύ δ' πμα) through parablepsy, and leaving out III. 183 owing to homoeographon with 184. A's text is normal in both places.

^{page 12 note 3} Also, at I. I, AK have ἢγοѵѵ (or ἒοί κεѵ ?) ἂλλο suprascript to οι τι κεѵ: BC the plain text.

^{page 13 note 1} Perhaps we ought to add at least some of the passages where BCK together seem to show emendations (cited above, p. 10).