they came respectively under different sets of influences, and each in the course of time lost more or less of its original purity. With certain limited exceptions already noticed, the concordance of Β and א marks that residual portion of the text of their primitive archetype in which neither of the two ancestries had at any point adopted or originated a wrong reading. Where their readings differ, at least one of the ancestries must have departed from the archetypal text. The possibility that both have gone astray in different ways must remain open, for it would be only natural that there should be an occasional coincidence of place between corruptions admitted into the one line of transmission and corruptions admitted into the other; and as a matter of fact there are a few passages where it is difficult to think that either Β or א has preserved the reading of the common original. But these coincidences are likely to be only exceptional; and all that has been observed up to this point respecting the character of our two MSS justifies a strong initial presumption in each particular case that the text of their archetype is preserved in one or other of them.
328. It follows that any subsingular, or even singular, reading of either Β or א may owe the limitation of its attestation to either of two totally different sets of antecedents. A subsingular reading of Β (or א) may be, first, equivalent to a subsingular reading of אB combined, which has lost part of its attestation by the accidental defection of א (or B); it may be, secondly, an early corruption limited in range of acceptance. Both explanations being in all cases possible, the antecedent probabilities differ widely according as the one or the other MS is in question. The ancestry of Β posterior to the common archetype was probably a chain of very few
