Manuscript Rahlfs 803 (4QLXXNum), dated to around the turn of the era, is, so to speak, an example of what is disputed: if Patrick Skehan, John William Wevers, Udo Quast and Nicholas Petersen are correct in their characterisation, it is an example of a textual witness to an early text recensionally reworked on the basis of the Hebrew.^[i] Eugene Ulrich, on the other hand, sees Ra 803 as a witness to the Old Greek, which he characterises as follows:^[ii] “There are seventeen variants in 4QLXXNum, thirteen of which are unique, only four finding support in other Greek manuscripts.”^[iii] This remarkable evidence notwithstanding, Ulrich argues that in Num 3:40 Ra 803 witnesses to the Old Greek and in his opinion the two editors of the Book of Numbers in the Göttingen Septuagint, Wevers and Quast, wrongly chose a recensional reading (although uniformly attested) as original.^[iv] But the main criterion for Wevers and Quast to have rejected the reading of Ra 803 to Num 3:40 is that they gave preference to the translator's normal way of rendering the Hebrew. Emanuel Tov has put his finger on the problem here and stated that the tradition has become standardised over time, thus the less uniform readings and deviations from the MT tradition tend to be more original. At the same time, however, he emphasised that the abandonment of this editorial principle and the preference for individual readings based on fragmented papyri would have led to “an almost chaotic situation in the reconstruction procedure”.^[v] In this respect, in the case of Ra 803, the question of whether it is an early recensional text or not is perhaps not to be decided at the moment, but Wevers’ text-critical decision can be justified in any case.

What the readings show, in my opinion, is the diversity of transmission at an early stage of textual history. The early daughter versions paint the same picture: they show that in the early phase of the history of transmission, a pluriformity of textual forms prevailed, which only later became stratified. The statement that we are dealing with several strands, stages or layers of editing, that the text was subject to constant change in its transmission and development, is therefore certainly capable of consensus, but sometimes not sufficiently substantiated. This development took place in different directions: Copies of a biblical text could be revised in different places and at different times according to the same scheme. This explains, for instance, why there was not one καίγε-revision, but why we can observe different καίγε-revisions, not only of different biblical books, but also within one and the same book.