At first sight, the total of fourteen hundred corrections to the Greek seems to indicate extensive changes to the text of Codex Bezae. But in fact over 320 of these are orthographic, whilst 380 are simply the removal of nonsense, and others are slight. In Part III we have devoted our main attention to only about 380 corrections. It is true that, for example, eighty-two definite articles are added to the text. But against the 11,837 articles apparently to be found in the Gospels and Acts of Nestle-Aland26, this is a tiny number. And, given the remarkably distinctive character of the text, one wonders whether the correctors have made much impact in Acts, or any at all in the Gospels. We can test this by comparing the number of divergencies in text between the two columns in the fifth and the ninth centuries. The first hand shows, as we will discover in Part IV, a total of 860 differences between the two. The net increase as a result of corrections is by 142, to 1,002. This is not a startling change. The details (tables 34–6) show us that, apart from corrections by B, the move is slow.

It is clear that only G treats the two columns as a single tradition. Otherwise, there is a steady erosion of the peculiarities of the text, which is accompanied by a widening gap between the columns. The consequence of centuries of correction was the addition, not the diminution, of differences between the columns.