date of the readings taken together, and the consequent presumption in favour of any one of them.
298. When however we go on, secondly, to compare the identical readings of אB with the readings of א unsupported by Β and of Β unsupported by א, the first alternative obtains so much positive corroboration that the second and third may be safely dismissed. For the present purpose we must neglect the numerous readings in which א or Β forms part of a large group, and attend to those readings only in which they stand respectively in opposition to all or almost all other Greek MSS, but with some other support: with the places where they stand absolutely alone we are not for the present concerned. It is then seen that a large proportion of the small groups containing one or other of the two MSS contain also other documents (versions or quotations) attesting a high antiquity of text. Many of the readings of Β having this accessory attestation are doubtless wrong, and, as we shall see presently, a much greater number of the readings of א: what we are now concerned with however is not genuineness but antiquity. Each of the two MSS is proved by these readings to be at least in part derived from an original preserving an extremely ancient text, for the most part not represented by our other extant MSS: and these two texts are by the nature of the case different from each other.
299. The distinct existence of these two independent texts is further illustrated by places where they emerge into view simultaneously; that is, in a certain number of those ternary or yet more composite variations in which the readings of א and of Β are different from each other, but are closely connected together in opposition to the reading or readings of the great bulk of docu-
