Scribes omit words more often than add them.

All recent NT textual research is moving one direction: with regard to the earliest Egyptian manuscripts of Matthew a recent scholar has stated: "What is customary with the old papyri, and also remarkable, is that omissions are more than twice as likely to be found than additions" (Kyong Shick Min, Die früheste Überlieferung des Matthäusevangeliums [Arbeiten zur neutestamentlichen Textforschung 34; Berlin: De Gruyter, 2005], 97). Other scholars like Peter M. Head, James P. Royse, Maurice Robinson, etc., say the same thing.

What I think is truly remarkable with regard to the Byzantine Textform is that the consensus of the later Byzantine manuscripts is not any longer than the consensus of the earliest Byzantine manuscripts copied one thousand years earlier. The consensus of the common text has remained remarkably consistent throughout history. The early Egyptian texts are far too inconsistent and their one hallmark feature, again and again, seems to be a tendency to omit rather than add. The majority of Byzantine manuscripts, on the other hand, demonstrate a tendency neither to add nor to omit, but rather simply to leave the text alone.

Jonathan C. Borland