larly in B. Hence Β has to be left virtually out of account as an authority against unclassical forms with ι, and א against unclassical forms with ει; while in the converse cases the value of their evidence remains unimpaired, or rather is enhanced, allowance being made for the possible contingency of irregular permutations here and there. Till the unsifted mass of orthographical peculiarities of a MS has been cleared from the large irrelevant element thus contributed by what are probably mere itacisms, no true estimate can be formed of its proper orthographical character. When this rectification has been made, it becomes clear that the unclassical forms and spellings abound most in the MSS having the most ancient text, and that their occurrence in cursives is almost entirely limited to cursives in which relics of a specially ancient text are independently known to exist.
400. To accept however every ancient spelling differing from the late spellings would be as rash as to accept every Western reading because it is very ancient. Curiously enough, but quite naturally, the Western documents are rich in forms and spellings not found in other documents, and some few are also confined to documents in which the Alexandrian text is very prominent. Here again Β holds a neutral place, having many spellings in common with each class of text. We have as a rule taken only such unclassical spellings as had the support of both classes, or of either alone with B. Even where Β stands alone, we have usually followed it for the text, unless forbidden by some tolerably strong internal or analogical reason to the contrary. But in many cases there is no room for hesitation about the reading, all the best uncials being concordant.
401. The irregularity of the extant orthographical evidence is so great that it would have often been unsatisfactory to decide on the form to be given to a word in any one place without previous comparison of the evidence in all or nearly all places where the same or similar words occur. Most orthographical variations have been carefully tabulated, and the readings decided on consecutively as they stood in the tables, not as they occur scattered among substantive readings. Many of the particulars required were not to be found in the published apparatus critici: but the labour involved in collecting them has not been fruitless. Examination of the columnar tables of attestation, by bringing to light approximate uniformi-