CNT = Bruce Metzger, The Canon Of The New Testament (New York: Oxford University Press, 1997)
FGO = Martin Hengel, The Four Gospels And The One Gospel Of Jesus Christ (Harrisburg, Pennsylvania: Trinity Press International, 2000)
INT = D.A. Carson and Douglas Moo, An Introduction To The New Testament (Grand Rapids, Michigan: Zondervan, 2005)
TCD = Lee McDonald and James Sanders, edd., The Canon Debate (Peabody, Massachusetts: Hendrickson Publishers, 2002)
The large majority of the earliest Christians whose writings are extant haven't left us a record of every book they considered scripture. We can identify a portion of their canon, but not its entirety. Martin Hengel observes that "of the second-century Christian writings known to us by title, around 85% have been lost. The real loss must be substantially higher." (FGO, 55) And we wouldn't expect the early Christians to discuss the full extent of their canon in the vast majority of the contexts in which they wrote.
In a recent post, I argued that Athanasius' twenty-seven-book New Testament was held by some of his contemporaries and by some before his time. Origen is the earliest source I'm aware of who advocates the twenty-seven-book canon in the records extant to us. Though Athanasius' Festal Letter 39 from the middle of the fourth century is often cited as the first reference to the twenty-seven-book canon, the evidence suggests that it was advocated at least more than a century earlier. Thus, the plausibility is increased that the twenty-seven-book canon was among the New Testament canons held by the many ante-Nicene sources who haven't left us a record of the full extent of their canon of scripture. Since we only have a partial record of the canon of most of the earliest sources, how close does the portion of their canon we know about come to the twenty-seven-book collection?
Even if a source is known to have held some other canon, it's important to know how different it was. It would be one thing if the earliest Christians had something like a five-book or eight-book collection of scripture. It would be something else if they had a twenty- or twenty-five-book canon, which would fall short of our twenty-seven-book canon by a much smaller margin. When critics of Christianity, or of Evangelicalism in particular, tell us that some Christians held a different canon, it's important to know how different those other canons were. The closer they come to the twenty-seven-book canon, the less significant the objection that's being raised.
Harry Gamble writes:
"It is recognized by all that (1) by the end of the second century the four gospels, the letters of Paul, and 1 Peter and 1 John had acquired very broad use and high authority in almost all regions of early Christianity...it is widely recognized today that Paul's letters were consistently known and used throughout the second century, and by the late second century had become fully and universally established as apostolic scriptures....both the Gospels and the Pauline Letters were shaped into firm collections during the second century" (TCD, 271, 286-287)
Those nineteen documents are nearly identical to the twenty Eusebius refers to as undisputed in his Church History (3:25), which he composed in the late third and early fourth centuries. He writes that his assessment of the documents is derived from "the tradition of the church" (3:25), and he appeals to how the documents were received in earlier generations, so he isn't just describing the state of the canon in his day.
In the research of Franz Stuhlhofer, we have further corroboration of this near agreement between the scholarly consensus Gamble refers to and Eusebius:
"First, drawing on the work of Stuhlhofer, Barton counts the number of times the New Testament (and other) books are actually cited by the Fathers in proportion to each book's length. He discovers there are three clear groups: those New Testament books that are quoted frequently (viz., the four gospels and the major Pauline letters), those quoted less frequently (the rest of the New Testament), and books that are scarcely quoted at all (viz., those that were excluded from the canon). In other words, there is a sharp demarcation in actual frequency of usage between the New Testament books and all other claimants: actual usage was establishing the canon." (INT, 733-734)
Whether we go with Gamble's nineteen books or Eusebius' twenty books, we have the large majority of the New Testament, more than two-thirds of it. But there are problems with how both Gamble and Eusebius arrive at their numbers. Both suggest that the number could be higher.
Eusebius changed his view of Revelation over the years (INT, n. 20 on 734), and he described the book and its acceptance as scripture by the church in different ways in different places in his writings. He'll acknowledge that Revelation could be classified as one of the widely accepted books of the New Testament, but will go on shortly afterward to comment that it could be placed in a lesser category as well (Church History, 3:25). As D.A. Carson and Douglas Moo note, Revelation seems to have been "almost universally recognized as Scripture in the second century" (INT, n. 20 on 734). Its more disputed status later shouldn't prevent us from including it among the widely accepted books in earlier generations. Gamble notes the early widespread acceptance of Revelation and acknowledges that the disputed status of the book that people often refer to arose later in church history (TCD, 289). If we add Revelation to the lists of Gamble and Eusebius, we have twenty and twenty-one books, respectively, that were widely received early on.
Gamble excludes Acts, but not because the evidence is contrary to Eusebius' inclusion of Acts among the undisputed books. Rather, he believes that "The early history of Acts is largely obscure and needs further investigation." (TCD, 288) But Eusebius had access to a lot of sources no longer extant. The relatively few sources who comment on the canonicity of Acts in the earliest generations, such as Irenaeus, are supportive of what Eusebius reports, and, as Gamble notes (TCD, 288), Acts is a companion piece to the gospel of Luke. It seems unlikely that Luke would be so widely accepted without a similar acceptance of Acts. I see no reason to not include Acts among the books widely accepted as canonical early on.
The history of Hebrews is similar to that of Revelation. It was widely accepted in both the West and East early on, but became more controversial later. The later disputed status of the book, referred to by Eusebius and others, shouldn't prevent us from including it among the books widely accepted as canonical early on. Gamble acknowledges some of the sources who accepted the book in early generations (TCD, 289), but neglects to mention others, gives no evidence of any significant early opposition to the document, and seems to read later opposition to the book among some in the West into the earlier generations, without any justification that I can perceive from anything Gamble says. Hebrews ought to be included among the books widely accepted in the ante-Nicene era.
That leaves us with five others: James, 2 Peter, 2 John, 3 John, and Jude. They all seem to have been less widely accepted early on, but I'm not aware of any reason to conclude that any of them were rejected by the majority in the earliest centuries. Rather, their widespread acceptance in the post-Nicene era, accompanied by some attestation earlier, suggests that they were accepted by a majority even in the ante-Nicene era, though a smaller majority than the other twenty-two books had.
Gamble argues that many people in the early centuries of the church included some books that aren't part of the twenty-seven-book canon we have today. He cites The Shepherd Of Hermas as his primary example, though he also mentions others:
"More widely popular than either of these [First Clement and The Epistle Of Barnabas], however, was the Shepherd of Hermas, which was fully acknowledged as scripture by Irenaeus (Haer. 4.20.2), Clement of Alexandria (Strom. 1.17.29, 2.1.9, 12), and Tertullian (Or. 16). Its strong representation among early Christian papyri discovered in Egypt probably reflects its popularity....the esteem and use attaching to them [documents like First Clement, The Epistle Of Barnabas, The Shepherd Of Hermas, and The Didache] was appreciably earlier, more continuous, and more widespread than to many of the writings that were finally accepted in the canon, including Hebrews, 2 Peter, James, and 2 and 3 John....If the scope of the canon had been determined in the second century, it seems likely that they would have found a place in it." (TCD, 289-290)
Notice, first of all, that Gamble only cites three patristic references to the scriptural status of The Shepherd Of Hermas, which is his primary example. Yet, he doesn't include books like Hebrews and Revelation among those widely accepted early on, even though both had much more early attestation and less opposition. And he even includes Hebrews among the documents that allegedly weren't as well attested as The Shepherd Of Hermas.
Second, notice Gamble's failure to interact with the early opposition to the canonicity of the documents he cites, even though he refers to their "more continuous" acceptance. As Everett Ferguson points out later in the same book in which Gamble wrote those comments (TCD, n. 60 on 308), Tertullian tells us that The Shepherd Of Hermas was excluded from the canon of the Christians of his day (On Modesty, 10). Tertullian's view of the document changed. Gamble cites Tertullian's earlier positive assessment without commenting on his later negative assessment, an assessment Tertullian tells us was shared by the Christians of that era in general. Tertullian's comments are corroborated by Origen, who tells us that The Shepherd Of Hermas wasn't generally accepted by the churches (CNT, 188). The dating of the Muratorian Canon is disputed, but most scholars place it in the second century, and it, too, rejects The Shepherd Of Hermas (CNT, 307). Apparently, the book was accepted only by a minority. I'm not aware of any early source who gives as negative an assessment of any of the books in our twenty-seven-book canon as Tertullian gives us for The Shepherd Of Hermas. Where does an early source tell us that Hebrews, 2 Peter, or Revelation, for example, has been as widely rejected as Tertullian claims The Shepherd Of Hermas has been? I don't know of any such source. Tertullian goes on, in the same document, to refer to Hebrews as more widely accepted among the churches than The Shepherd Of Hermas (On Modesty, 20). Gamble's analysis of The Shepherd Of Hermas is far off the mark, and its his primary example.
Similar observations can be made about the other documents Gamble cites. Clement of Alexandria would often speak highly of non-canonical documents in one place, but then qualify those comments elsewhere. I've written about that tendency in Clement in another article. To cite an example involving one of the documents Gamble mentions, Clement referred to The Epistle Of Barnabas as inspired and wrote a commentary on it, yet in other passages he criticized the document (CNT, 134 and n. 43 on 134). Though Jerome thought highly of The Epistle Of Barnabas, he refers to it as "reckoned among the apocrypha" (Lives Of Illustrious Men, 6), apparently a reference to a general rejection of the document. Etc. Gamble gives a positive assessment of these documents that he doesn't sufficiently support, and he doesn't interact with the evidence against his assessment.
It seems that twenty-two of the twenty-seven New Testament documents were widely accepted during the ante-Nicene era, five were accepted by a smaller majority, and documents like The Shepherd Of Hermas and The Epistle Of Barnabas were accepted only by a minority. Bruce Metzger discusses which books were accepted by the writers of that era in the documents that are extant, and he concludes that there are twenty-two in Irenaeus (CNT, 155), twenty-three in Tertullian (160), twenty-two and sometimes more in Clement of Alexandria (134-135), twenty-two in Hippolytus (150), etc. The numbers could be higher. Men like Irenaeus and Tertullian don't tell us the full extent of their canon. As Metzger explains, the absence of a reference to a document like 2 or 3 John in Tertullian, for example, doesn't have much significance, since it would be easy for somebody to not cite such a short and unoriginal document, yet consider it scripture (160).
Lee McDonald lists the New Testaments of seventeen fourth-century sources (TCD, 592-595). Almost all of them contain twenty or more of the twenty-seven books.
Most of the New Testament canons that differ from the twenty-seven-book canon don't differ by much. Books like 2 and 3 John significantly increase the percentage of books that were disputed, but they don't make much of a difference in the textual length or theological content of the canon.
And the high number of books in so many of the early canons suggests that it's plausible that the twenty-seven-book canon was held by some Christians even earlier than Origen. If our incomplete knowledge of the canon of men like Clement of Alexandria and Tertullian leads us to the conclusion that they accepted twenty-two or more of the twenty-seven books, then our twenty-seven-book canon isn't too far off, and it may have already been present.