In his Four Gospels, Burnett Streeter articulates his view of the sources of Luke and proto-Luke as follows:
The hypothesis I propose in no way conflicts with the generally accepted view that Matthew and Luke are ultimately dependent not only on Mark but on Q—meaning by Q a single written source. Most, if not all, of the agreements of Matthew and Luke, where Mark is absent, are, I think, to be referred to Q; but I desire to interpolate a stage between Q and the editor of the Third Gospel. I conceive that what this editor had before him was, not Q in its original form—which, I hold, included hardly any narrative and no account of the Passion—but Q+L, that is, Q embodied in a larger document, a kind of “Gospel” in fact, which I will call Proto-Luke. This Proto-Luke would have been slightly longer than Mark, and about one-third of its total contents consisted of materials derived from Q (Streeter 208).
Click here for a diagram of Streeter’s understanding of the synoptics’ sources. For an online version of the Four Gospels with sectional pagination, see κατα~Π (1924 ed.).
Herder thought that Mark most exactly reproduced UrevOr. Matthew reproduced it with expansions, and Luke, aware of these expansions, “wished to create ‘an actual historical account’ after a wholly Hellenistic pattern.” Herder also hypothesized that “[s]ome forty years later John . . . wrote an ‘echo of the earlier Gospels at a higher pitch’ which undertook to set forth Jesus as the Savior of the world. . . .”