^{page177 note 1} I have used the letters employed by the editors for such MSS. as they have treated. C and M, which they have not quoted, I have supplied. It may be said inpassing that Hirschig seems to use V to describe both V and X.

^{page178 note 1} Reason is given below for suspecting that β did not contain the passage in which these two difficulties occur, in which case they give no support to the theory here put forward. But the other evidence is, I think, sufficient.

^{page178 note 2} E.g. Teubner, p. 4. 18, тς νεώς CP: тς тε νεώς BMV. P. 119. 17, т δ νν PMV:omitted by BC.

^{page178 note 3} P. 61 verso, the last words being εί παρασ κενάΖεσθαι πρδς μάχην (Teubner, p. 258. 13), which are immediately followed by ὃтΨ δεήσει тρόπΨ κтλ. (Teubner, p. 92. 17).

^{page178 note 4} P. 200 verso. The last words are παтρψανγνω … (Teubner, p. 186. 1).

^{page179 note 1} E.g. Teubner, p. 191. 27, μικρῷ MV, p. 221. 2, ξήтει BP: Ζήтει MV, p. 246. 19, γρ 'тδσπης και тούтου προυνόησε BP:omitted byMV.

^{page179 note 2} Whether it is right or wrong to assume that β was defective, the main question of the existence and authority of the two families is not affected. The only difference is that if β was defective, P has no more, and probably less, value than M and V in the passage in question; whereas if β was a complete MS., then P might be judged to retain some of its readings when it is in opposition to MV.

^{page179 note 3} Basileae ex Officina Hervagiana, February, 1534; edited by Vincentius Obsopoeus [cf. H. Simonsfeld, Einige hunst- und literaturgeschichtliche Funde, pp. 539–547(Sitzungsberichti der philosphilol. und der histor. Classe der kgl. bayer. Ahademic der Wissenschaften, 1902, Heft IV.].

^{page180 note 1} Scriptores Erotici Graeci, Vol. II. 1792–4Google Scholar.

^{page180 note 2} Charitonis Chaereas et Callirhoe. Text and notes by , J. P. D'Orville. Amsterdam, 1750Google Scholar.

^{page180 note 3} Erotici Scriptores. Didot. Paris, 1856Google Scholar.

^{page180 note 4} Heliodori, , Aethiopica. Teubner. Leipzig, 1855Google Scholar.

^{page180 note 5} ‘Ηλιοδώρου Ạίθιοπικά, ἃ … μεтά σημειώσεων ξδωκε’ … , Ο. Δ. κοράηςParis, 1804Google Scholar.

^{page180 note 6} For examples see above.

^{page180 note 7} E.g. Teubner, p. 3. 12, тρίтον ζωσтρος CP: тρίтον ζωσтρα MVB. P. 119. 17, т δ νûν MVP:omitted by BC.

^{page180 note 8} E.g. Teubner, p. 25. 21, παρελσθαι παρανομήσανтος C: σαρελέσθαι omitted by BP: σαρανομήσανтος omitted by MV (though V has it added in the margin by a later hand.

^{page181 note 1} P. 179, note 1.

^{page181 note 2} Since B and P, when they disagree with C, only rarely agree with one another, it would perhaps be better to assume that their ancestors broke off from β independently instead of postulating, as I have done in the tree, a common parent for them (δ). However, there are cases such as that quoted on p. 180, note 8. The very fact that BP rarely agree against C when it is available is strong evidence in support of the contention that an agreement of BP after C's failure gives a reading which C and so β would have givenhad they been extant.