I've just finished reading the following fascinating article:
August den Hollander and Ulrich Schmid, 'The Gospel of Barnabas, the Diatessaron and Method', Vigiliae Christianae 61 (2007) 1-20.
The article argues cogently against the need to postulate that there was an Old Latin (pre-Vulgate) version of the Diatessaron using the following arguments:
'a) The OL Diatessaron hypothesis raises more questions than it answers;
b) The OL Diatessaron hypothesis is based on unsafe textual analysis;
c) The OL Diatessaron hypothesis is based on anachronistic assumptions and anachronistic use of source material' (p. 5)
I particularly like the following quotation:
'But this is exactly what the Old Latin Diatessaron hypothesis wants us to believe: The Old Latin Diatessaron tradition, despite its powerful presence in the 13th-15th centuries—as evidenced by the many vernacular representatives—is gone without a trace in its original language. From a purely historical perspective this is very difficult to believe.' (p. 6)
Since the beginning of the 20th century research aiming at reconstructing Tatian’s lost Gospel harmony Diatessaron utilizes a growing number of late 13-15 c. texts extant invarious Western vernaculars for this purpose. As the most recent example Jan Joosten introduced the so-called Gospel of Barnabas, a composition perhaps as late as the 16th or17th century as a potential source for readings of the Diatessaron (2nd c.). With special emphasis on methodological issues, this essay offers a detailed critique of Joosten’s analysis as well as a general critique of that type of research as carried out by other scholars in the past.