The Coptic text is one tractate in Nag Hammadi Codex II, part of the jar of thirteen codices found near Nag Hammadi in Upper Egypt in 1945; the manuscript is of about the middle of the fourth century, and its Coptic translates a Greek original. Three Greek fragments had already been found at Oxyrhynchus: P.Oxy. 654, 1 and 655, recovered between 1897 and 1904 and published promptly by Grenfell and Hunt as unidentified sayings of Jesus. They are dated roughly between 130 and 250. Only after 1945 could they be matched to the Coptic and recognised as three separate copies of the same work, which is itself evidence that the book circulated widely in Egypt.
The dating question is the substance of the dispute, and it is not a matter of scholars hedging. One position holds that the collection, or a core of it, was assembled in the first century and draws on sayings traditions independent of Matthew, Mark and Luke, and that its shorter and less allegorised parables are sometimes closer to what was said than the canonical versions. Helmut Koester, Stephen Patterson, John Dominic Crossan and, for an early stratum with later additions, April DeConick argue in that direction. The opposing position holds that the parallels show knowledge of the finished gospels, including their distinctive editorial wording, and dates composition to the second century. Simon Gathercole, Mark Goodacre and Nicholas Perrin argue this, Perrin on the further ground that the sayings order reflects catchword links in Syriac and so depends on Tatian's second-century gospel harmony.
The evidence genuinely cuts both ways, because a collection can preserve old material and still have been compiled late, and the two questions are routinely conflated. What can be said without controversy is that the Greek fragments put copies in circulation in Egypt by the late second century, that the three copies differ from each other and from the Coptic, and that the text was not fixed.