If Q was originally an Aramaic document, used by Matthew and Luke in Greek translations going back to different copies of the Aramaic original, it is fair to assume that these two translations would have had different histories. Q would always be growing, by the aid of oral tradition; and if Q was written before Mark, there was ample time, say twenty-five years at least, before it was used by Matthew and Luke, for the two recensions, circulating in different communities and perhaps originally shaped to suit the needs of different readers, to acquire many dissimilar features. Not only would the same saying in many instances become changed to meet the varying need, or to adapt itself to what was considered a better tradition, but many things would be included in either recension which were not included in the other. Matthew will thus have had a recension of Q which we may designate by the sign QMt, and Luke one which we may call QLk.
The following pages represent an attempt to determine the content of Q, as that is represented in both Matthew and Luke.[92] Of the sections of Matthew and Luke examined, some are marked QMt, some QLk, and some merely Q. By this it is not meant that Matthew and Luke each had a document Q, and besides this a document QMt or QLk, and that they took now from one and now from the other. But where the wording of Matthew and Luke is identical, or so closely similar that the variations can be easily explained as changes made by Matthew or Luke, the material is assigned simply to Q. But where the variations are too great, much greater for example than any changes that have been made by Matthew and Luke or by either one of them where they are taking their logian material from Mark, the material is assigned to QMt and QLk. Reasons for the assignment to QMt or to QLk instead of to simple Q are given in each case seeming to require them. The sum of all passages assigned to any form of Q will constitute the total content of Q, so far as it is contained in both Matthew and Luke. This total content will be somewhat larger than the content that could be assigned to Q without the hypothesis of QMt and QLk, since by this hypothesis many sections will be sufficiently alike to be assigned to Q (QMt and QLk) which otherwise would have to be ascribed to different sources.[93]