Assured Results

Textual Criticism has a problem: It doesn't know what is and is not true. There are no assured results. In the sciences, there are some things so thoroughly verified that you don't have to re-re-reconfirm the results. (The obvious examples are from physics: The first two laws of thermodynamics -- the law of conservation of energy and the law of entropy -- have been so thoroughly verified that there is no need to further test them. At least until some strong counter-evidence shows up.)

Why should textual critics care? Because assured results are so useful! What we often see, in textual criticism, is that results which are not assured are treated as assured. Very frequently, textual critics act like medieval natural philosopher appealing to Aristotle: "It's in Hort (or Streeter, or Lake, or Metzger); it must be true!"

The lack of assured results can also lead to skeptical attacks. Just as one group of people may affirm results which have not been verified, another may deny results which have been more than sufficiently verified. (A modern-day equivalent is vaccine skepticism.)

So the question becomes, can we declare any results in textual criticism to be assured? (Note that, for something to be assured, it must be experimentally verified. Universal agreement is not sufficient. It must be supported by evidence.)

The answer is, Yes, but it's a short list. The following are the items I am aware of:

A manuscript cannot be older than its exemplar. A logically necessary conclusion, but an important one. If a manuscript A is dated after manuscript B, then A cannot be the ancestor of B (although it might be descended from a manuscript which is an ancestor of B, and it may be closer to that ancestor than B). This has significant implications for the Coherence Based Genealogical Method, which allows textual "flows" to go from a newer manuscript to an older. Of course, advocates of the CBGM will say that a flow is not an actual relationship....
The Majority Text Exists. That is, there is a textual grouping of manuscripts, quite closely related in terms of readings, to which the majority of manuscripts belongs. Note that all we can say about it is that it is the majority. We cannot call it Byzantine or Syrian and consider that an assured result.
The final verification of this claim came only quite recently, with the Munster "Thousand Readings" project. If one studies the results of this project, which examines hundreds of readings in almost all the manuscripts known at the time (there are a few manuscripts omitted), we find that almost all passages do have a clear majority reading. This alone would not make the existence of a Majority Text certain. (If most readings had only 60% support, and the 60% shifted, there would be no majority text.) But the fact that most readings see one variant supported by 80% or more of the witnesses is significant. So is the fact that the 80+% includes most of the same manuscripts over all these variants. It's only a relatively small group which deviate more than a handful of times.
Note that this does not tell us the nature of the Majority Text. Whether it is good or bad, long or short, edited or inedited is another question altogether.
Textual Groupings exist. This is a very vague statement as such, but the point is that we have examples of all sorts of textual groupings: parent and child (D^p and D^abs), siblings (many of the K^x Cl 74 manuscripts copied by Theodore of Hagiopetros), families (the Lake Group; it appears that the Ferrar Group and Family 2138 are superfamilies), text-types (the Byzantine text). We do not have a clear definition of any of these groupings, and we do not know how many levels of kinship there may be (a typical proposal contains about four: Family, Clan, Sub-Text-Type, Text-Type -- but this is a proposal based on logic, not observation). Nonetheless, we can safely assume that manuscripts can be grouped, and try to group them; we do not have to assume that all manuscripts exist in isolation. This may sound trivial; it is not. It is one of the crucial points of textual criticism. Until it is certain, most of the tools provided by classical textual criticism do not apply.
Mixed Manuscripts exist. This is proved by a handful of manuscripts: D and D^abs (the latter a mixed manuscript derived from the former) and 424^c. In addition, manuscripts like 1881 can hardly be explained by any means other than a Byzantine/1739 mixture.
Like the preceding, this may seem like a trivial point, but the existence of mixture is a vital part of the theories, e.g., regarding the "Cæsarean" text. It is good to be sure that such manuscripts exist.
Note that this does not prove that such manuscripts are common, or that any particular manuscript is mixed. This must be proved on a case-by-case basis.
Assimilation of Parallels occurs. Every manuscript tested shows this phenomenon: Occasional adjustment of passages to match their parallels in other gospels. It appears that all have at least occasional singular assimilations. This demonstrates that the phenomenon takes place.
Note that this does not prove that any particular parallel reading is an assimilation. While it is surely more common for manuscripts to produce harmonized rather than disharmonized readings, scribes do make errors of the other sort.

There is another side to this: Any result which is not assured is just that: Not assured. It may be true, it may be likely, but it is not certain. As new evidence accumulates, these non-assured results need to be re-examined.

The following shows some non-assured results which have been treated as assured:

The Byzantine text is late and derivative. Almost universally believed. But proved? No. (See the article on Byzantine Priority.) Even if one believes the evidence absolutely conclusive at present, what happens if we find a second century Byzantine manuscript or Father?
Most canons of criticism. We take a very high proportion of these on faith, in some cases (e.g. "prefer the shortest reading") rather in the face of the evidence. It's not easy to see what we can do about this -- canons of criticism are more nearly postulates than the result of study; in the absence of autographs, they cannot be proved. But that's precisely the point: they cannot be proved.
That text-types other than the Byzantine exist. The existence of the Alexandrian text is almost assured -- but its boundaries are not assured. Are 𝔓⁴⁶ and B and 1739 Alexandrian in Paul? Yes, say some, scholars; no, say Zuntz and Carlson (and I think they're right). Until the boundaries of the type are established, it's not all that useful.
The cases for the "Western" and "Cæsarean" texts are still less certain. There is certainly a D-F-G text of Paul. But is this the same as the text of Codex Bezae in the Gospels and Acts? Is Codex Bezae a representative member of whatever type it does belong to? The answers, to this point, are largely assumptions; there is no proof. The evidence, if anything, says that Bezae is edited (the obvious evidence being the use of Matthew's genealogy of Jesus in Luke); great care must be used when trying to prove anything from Bezae.
The doubts about the "Cæsarean" text are so well-known that we will not document them here.
The dates of most manuscripts. We tend to treat manuscript colophons as a guarantee of dates, and paleography as nearly certain as well. But colophons can be faked; Colwell, for instance, documented the errors in the colophon of 1505. For undated manuscripts, the situation is worse, because our only evidence is based on inferences based on the dated colophons we have. And even then, the results can be inconsistent. It is not uncommon to see two scholars examine a manuscript independently and offer dates two centuries apart. And that's for minuscules, where dated samples are common! Take a manuscript like B. Everyone dates it to the fourth century. Why? Based on documents with similar writing styles, which we believe to be contemporary, and which we date based primarily on their contents. In other words, we're making multiple assumptions here: First, we're dating other writings based on their contents. Second, we're assuming that the date of B corresponds to the dates of those documents. This is a chancy assumption -- those other documents are mostly secular, and generally official. Can it be assured that those scribes were trained in the same way as the scribes of Christian manuscripts? It's quite possible that Christian scribes would adopt an archaic style.
Chances are that our paleographic results are generally correct. But they are not assured. One cannot treat them as a guarantee of anything.