The basis of all texts of Shakespeare must be that of the earliest Edition of the collected plays, the Folio of 1623, which, for more easy reference, we have designated F1[2]. This we have mainly adopted, unless there exists an earlier edition in quarto, as is the case in more than one half of the thirty-six plays. When the first Folio is corrupt, we have allowed some authority to the emendations of F2 above subsequent conjecture, and secondarily to F3 and F4; but a reference to our notes will show that the authority even of F2 in correcting is very small. Where we have Quartos of authority, their variations from F1 have been generally accepted, except where they are manifest errors, and where the text of the entire passage seems to be of an inferior recension to that of the Folio. To show that the later Folios only corrected the first by conjecture, we may instance two lines in _Midsummer Night's Dream_: Give me your neif, Mounsieur Mustard Seed. IV. 1. 'Neif,' which is spelt 'niefe' in Qq F1, becomes 'newfe' in F2, 'newse' and 'news' in F3 F4. And finds his trusty Thisby's mantle slain. V. 1. F1 omits 'trusty.' F2 makes up the line by inserting 'gentle.' Where the Folios are all obviously wrong, and the Quartos also fail us, we have introduced into the text several conjectural emendations; especially we have often had recourse to Theobald's ingenuity. But it must be confessed that a study of errors detracts very much from the apparent certainty of conjectures, the causelessness of the blunders warning us off the hope of restoring, by general principles or by discovery of causes of error. For example: in the _Midsummer Night's Dream_, I. 1, Or else it stood upon the choice of merit, the reading of the Folios, is certainly wrong; but if we compare the true reading preserved in the Quartos, 'the choice of friends,' we can perceive no way to account for the change of 'friends' to 'merit,' by which we might have retraced the error from 'merit' to 'friends.' Nothing like the 'ductus literarum,' or attraction of the eye to a neighbouring word, can be alleged here. Hence though we have admitted conjectures sometimes, we have not done so as often as perhaps will be expected. For, in the first place, we admit none because we think it better rhythm or grammar or sense, unless we feel sure that the reading of the Folio is altogether impossible. In the second place, the conjecture must appear to us to be the only probable one. If the defect can be made good in more ways than one equally plausible, or, at least, equally possible, we have registered but not adopted these improvements, and the reader is intended to make his own selection out of the notes.