Variant Collation and CBGM
Also known as: Coherence-Based Genealogical Method, CBGM, Apparatus Criticus Construction, Genealogical Coherence Analysis
Variant collation and the Coherence-Based Genealogical Method (CBGM), developed by Gerd Mink at the Institute for New Testament Textual Research in Münster, address the central obstacle to editing the Greek New Testament: contamination. Because medieval scribes routinely copied from several exemplars at once, the New Testament tradition is too intermixed for a classical bifurcating stemma. Mink's solution, set out in his 2004 chapter in Studies in Stemmatology II, shifts the unit of analysis from whole manuscripts to individual variation passages. At each passage the editor decides which reading gave rise to which (a local stemma), and the method then aggregates these local decisions, using the coherence of agreement among witnesses, to infer the global flow of text and the relationships among witnesses. CBGM now underlies the Editio Critica Maior and the modern Nestle-Aland and UBS Greek New Testaments.
Key highlights
- Handles contamination by analyzing variation passages rather than forcing whole manuscripts into a single tree.
- Keeps text-critical judgment central through editor-built local stemmata while making its aggregation explicit and checkable.
- Uses coherence as a diagnostic to flag implausible readings whose supporters scatter incoherently across the tradition.
- Produces a transparent, revisable global picture of textual flow that underlies modern critical editions of the New Testament.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use variant collation with CBGM when editing a tradition too contaminated for a classical stemma, above all the Greek New Testament and other heavily intermixed corpora where scribes drew on multiple exemplars. It is the appropriate method when you have, or can build, a comprehensive collation and when expert editors can make reasoned passage-by-passage judgments about which reading is prior. CBGM is well suited to producing a defensible reconstructed initial text together with a transparent account of textual flow among witnesses. It is less appropriate for cleanly tree-like traditions, where classical stemmatics is simpler and sufficient, and for small or sparsely attested traditions where there are too few variation passages to compute meaningful coherence. It also presupposes substantial editorial expertise, since the local stemmata, not the algorithm, carry the text-critical judgments.
Strengths & limitations
- Handles contamination by analyzing variation passages rather than forcing whole manuscripts into a single tree.
- Keeps text-critical judgment central through editor-built local stemmata while making its aggregation explicit and checkable.
- Uses coherence as a diagnostic to flag implausible readings whose supporters scatter incoherently across the tradition.
- Produces a transparent, revisable global picture of textual flow that underlies modern critical editions of the New Testament.
- Reconstructs flow among witnesses, not the physical copying history, so it does not yield a manuscript stemma in the classical sense.
- Depends heavily on the editor's local stemmata, so its results inherit the subjectivity and possible errors of those decisions.
- Is iterative and labor-intensive, requiring repeated revision of local stemmata as global coherence is checked.
- Needs a large, comprehensive collation and many variation passages, limiting its use to richly attested traditions.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How does CBGM differ from classical stemmatics?
Classical stemmatics tries to reconstruct which physical manuscript was copied from which, building a tree from shared errors, and it fails when scribes mixed several exemplars. CBGM instead analyzes individual variation passages, has the editor decide locally which reading is prior, and aggregates those decisions into a directed graph of textual flow among witnesses. It does not claim to recover the physical copying chain and can represent the multiple sources that contamination implies, which is exactly why Mink developed it for the Greek New Testament.
What is the difference between pre-genealogical and genealogical coherence?
Pre-genealogical coherence is simply how similar two witnesses are, measured as the proportion of passages where they agree; it is symmetric and carries no direction. Genealogical coherence adds direction by drawing on the local stemmata: it counts, across the passages where two witnesses differ, how often each carries the prior reading, yielding a net flow from one to the other. Pre-genealogical coherence diagnoses closeness and flags incoherent support, while genealogical coherence supplies the oriented ancestor relationships that build the global stemma.
Does CBGM replace the editor's judgment?
No. The decisive text-critical judgments live in the local stemmata, where the editor decides at each passage which reading gave rise to the others on grounds of transcriptional and intrinsic probability. CBGM's contribution is to propagate those judgments coherently across the whole tradition and to expose where they conflict, prompting revision. It is a disciplined, iterative framework for applying and testing editorial expertise, not an algorithm that determines the text by itself, which is why competent application of CBGM still demands a trained textual critic.
Sources
- 1.Mink, G. (2004). Problems of a highly contaminated tradition: the New Testament. Stemmata of variants as a source of a genealogy for witnesses. In P. van Reenen, A. den Hollander, & M. van Mulken (Eds.), Studies in Stemmatology II (pp. 13-85). Amsterdam: John Benjamins.ISBN 9789027232229
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Variant Collation and CBGM. ScholarGate. https://scholargate.app/religious-studies/variant-collation-analysis