Gerd Mink
Guide to “Genealogical Queries”
(Version 2.0)
Preface
Version 2.0 of the “Genealogical Queries” facility is designed to be used in connection with the second edition of the Editio Critica Maior (ECM) of the Catholic Letters. It differs from the first edition, because, in several passages, word and variant addresses have been altered.
“Genealogical Queries” provides five web-based modules that permit investigation of genealogical relationships inherent in the New Testament textual tradition.
The data brought together in the Editio Critica Maior (ECM) has been analysed and structured by reference to the Coherence-Based Genealogical Method (CBGM). The CBGM was developed while work on the ECM was in progress and additional studies had to be carried out during the editorial work using the rich and complex material.
Only after the last fascicle of ECM IV was published, was a comprehensive set of data available for the entire corpus. This data led to a revision of the previous reconstruction of the ECM text and thus to the second edition of the Catholic Letters. The revision resulted in 12 textual changes as compared with the first edition. Moreover, a split guiding line appears with 43 passages in the second edition where the editors could not decide in favour of one of the variants in question.
Consequently, the set of genealogical data was changed. In this data the assumed priority and posteriority of variants and the resulting genealogical relationship between witnesses have been recorded. On this data version 2 of “Genealogical Queries” is based. The changes in the data basis affect the results produced by the modules of “Genealogical Queries”. Hence the output of versions 1 and 2 is different.
The reduction from 3046 to 3043 variant passages as a result of the apparatus revision is of minimal importance. The new textual decisions are more significant. The data set on which the first version of “Genealogical Queries” was based comprised 125 passages where the decision about the initial text was left open. In the second version there are only 43 such passages. The new decisions led to changes in the genealogical data for all witnesses supporting variants that are now assessed differently.
One important feature of the CBGM is its potential to improve results through an iterative procedure. First, local stemmata of the variants are constructed taking into account pregenealogical coherence (see below). As a whole, the local stemmata make for a first overview of genealogical relationships between the manuscript texts that attest the genealogically arranged variants. Based on this newly derived external criterion, the local stemmata are put to the test and revised.
The next step is a revision of the local stemmata of variants in light of the newly established genealogical relationship between witnesses in the full corpus.This revision has been completed as far as the reconstruction of the initial text and its relationship to the variants is concerned. A systematic revision of the local stemmata of passages where the reconstruction of the text did not require revision is still pending. That work may again lead to changes in the genealogical data and have impact on the local stemmata.
Hence, what we present by “Genealogical Queries” is still work in progress.
The softeware’s layout, functionality and user guidance will continue to be enhanced. Even now the facility can be employed to examine the data, assess the stability of relationships, and so on. Due to continuing revision of the local stemmata, results obtained with the facility today may differ from those obtained tomorrow. Nevertheless, it remains a useful analytical tool. As one of many current uses, the facility can be employed to examine the degree of stability of relations, thus helping to establish whether the ongoing revision process is likely to change ancestor-descendant relationships in particular cases.
Important: When results achieved by the CBGM are used, attention has to be paid to the data source on which they are based. This applies to “Genealogical Queries” itself as well as to literature.
0. Database and Premises
The ECM apparatus has 3043 variant passages in the Catholic Letters. For every instance a local stemma of variants has been created. Each stemma represents a hypothesis about which variant originates from which, thus declaring variants as prior (source variants) or posterior in their relations with one another. (In cases where a hypothesis appeared to be too unreliable, we refrained from naming a specific source variant.)
The source variant is not posterior to any other variant of the same passage. It represents the starting point for the entire textual history at this place, and is thus considered to be part of the initial text of the tradition. Generally, the initial text is consistent with the base text of the ECM. However, there are a number of instances where we have abstained from re-constructing the initial text by means of the CBGM, and such cases have been treated as if they were lacunae in the initial text. A hypothetical witness “A” was assigned to the hypothetical initial text.
The local stemmata of the variants are based on the
usual text critical methods. Additionally, at the first compilation of these
stemmata, the pre-genealogical coherence was examined. This type of coherence
is solely based on the agreement of witnesses and is characterised by the
statement:
Witness x agrees with witness y in
n out of all cases (or p % of cases).
If the pre-genealogical coherence between the attestations of two variants is very low, a genealogical relation between those two variants can be regarded as highly unlikely.
The local stemmata of the variants also establish a
genealogical relation between the witnesses. This results from the fact that
when two witnesses are compared across all variant passages, one is associated
with a source variant more often than the other (unless the respective counts
are equal). A genealogical coherence emerges, consistent with two statements:
In n out of all cases (or p
% of cases), witness x reads the
source variant of the variant associated with witness y.
In n out of all cases (or p
% of cases), witness y reads the
source variant of the variant of witness x.
This results in one of the following three statements being true:
Witness x reads the respective source variant in more cases than does witness y.
Witness y reads more cases of the
respective source variant than does witness x.
Witness x and witness y read the
respective source variant with equal
frequency.
If the first of the three statements applies, witness x is a potential ancestor of witness y. If the second statement applies then witness y is a potential ancestor of witness x. If the third statement applies, there is no genealogical direction between the two witnesses.
Important! We consider as a witness the state of the text (without corrections) that we find in a continuous text manuscript, not the manuscript itself. As the text of a manuscript can be much older than the manuscript itself, a witness that is preserved in a late manuscript (e.g. 10th century) can be a potential ancestor of a witness preserved in an early manuscript (e.g. 5th century).
The stemmatic coherence, which is the definite simplest hypothesis about the most elementary genealogical connection between all witnesses, has not yet been implemented in the “Genealogical Queries” facility.
1. “Potential Ancestors and Descendants” Module
a. Purpose of the Module
This module analyses the genealogical coherence between witnesses and can thus differentiate between potential ancestors and descendants. The relation between ancestor and descendant, and thus the genealogical coherence, is established by the proportion of variants in a witness that are prior (source variants) or posterior compared to the corresponding proportions of the compared witnesses. The degree of agreement determines the strength of the genealogical coherence.
The genealogical coherence between ancestor and descendant is determined by which witness contains more prior variants, and the strength of coherence corresponds to the level of agreement between ancestor and descendant.
The “Potential Ancestors and Descendants” module allows a sound judgment to be formed about the text-historical environment of a witness. It also permits a good guess to be made concerning which witnesses have the potential to be ancestors in a stemma.
b. The Data Entry Form
In general, Gregory-Aland manuscript numbers are used to identify witnesses in all data entry forms. Normally, only continuous text manuscripts are processed; however, the option to include fragmentary texts exists. (There is no option to include lectionaries.) The option to include fragments will normally be waived because (1) due to their small extent they are not very significant and therefore only rarely play a role in the genealogical analysis, and (2) they undermine clarity. When included, minor fragments often occupy the beginning or end of the list of results. If they cover passages that barely vary then they tend to exhibit 100% agreement with nearly all witnesses. On the other hand, they tend towards the opposite extreme in passages where there is a lot of variation. (As an example, take witness 307 and choose the ‘Include Minor Fragments’ option.)
In addition to the potential ancestors, the potential descendants can be displayed as well. This is of interest if one wants to see the entire close environment of a witness and especially if one assumes an unusual witness to be at a dead end of the tradition.
The data source can also be selected. Choosing a single Letter is especially interesting if the textual character of a witness changes within the Catholic Letters so that we can assume it has different exemplars for the respective Letters. 2Jn, 3Jn and Jd must be treated with caution because the corresponding databases are small. A step of approximately 1% in the proportion of agreement is caused by a difference of only 1 variation unit in 2Jn and 3Jn, 2 variation units in Jd, but 7 in Jas, and 30 in the entire Catholic Letters.
c. The Result
The header displays three percentages referring to the witness in question which help to put the agreement values in the table below into perspective: the share of variants pertaining to the majority text (MT)*, the percentage of the average agreement (AA) and the median agreement (MA) with all witnesses included.
With the default options, the result presents the potential ancestors (W2 = witness 2) of the selected witness (W1 = witness 1) sorted in descending order according to the degree of agreement with W1. (Try witness 81, for example.) Levels of agreement are found as percentages under the heading PERC1 (= percentage 1) or as absolute numbers under the heading EQ (= equal). These values are given with reference to the total number of variant passages that can be found in both W1 and W2. This total appears under the heading PASS (= passages). The highest possible number is 3043, if a comparison includes the entire Catholic Letters corpus and both compared witnesses contain all variant passages of the corpus.
With all genealogical evaluations, the assumed priority or posteriority of variants plays a decisive role. Where variants are called prior or posterior in the following, these statements refer to assumptions established in the local stemmata of variants.
The column W1<W2 shows the number of cases where W2 has a variant prior to the one in W1, meaning a variant from which the variant in W1 has presumably developed. The column W1>W2 shows the number of cases where W1 contains the prior variant. A higher value in W1<W2 compared to W1>W2 characterises W2 as potential ancestor. The difference between both values also makes a statement about the stability of the relation of W1 and W2 (the so-called “particular textual flow”). To give an example, the relation between 81 and 35 can be inverted if the genealogy of their variants is changed at only one instance. Major differences between these values, however, indicate less agreement accordingly, and thus more distant kinship.
If the values in W1<W2 and in W1>W2 are equal (as is the case for 1448), there is no genealogical direction between the witnesses. In this case, a “-” appears in column D (= direction). These undirected genealogical relationships must be taken into account, especially in cases of a close relationship between W1 and W2. In such instances, too, the situation would change, if the genealogy of variants were assessed differently at only one passage. The result would be a directed genealogical relationship.
In column NR (= number), the potential ancestors are ranked according to their degree of agreement. If ancestors have the same degree of agreement they receive the same ranking number. Column NR contains a “0” if (1) an undirected genealogical relationship exists (e.g. 1448 in the table referring to 81), or (2) if a witness is not included in Jas, 1Pt, 2Pt, 1Jh and at least one of the minor Catholic Letters found under the option “Data Source: All Catholic Letters”. Since a genealogical direction may nevertheless exist between the witnesses, a “>” will appear in column D in the second case. This occurs only rarely. An example can be found among the potential ancestors of 33 (048 under W2).
A “0” is also placed in column NR with the small fragments that are displayed, if the option “Include Minor Fragments” has been chosen. At any rate, the “0” means that the pertinent witnesses under W2 are not included as potential ancestors in subsequent genealogical evaluations.
A high rank in column NR improves the odds that a potential ancestor is also a stemmatic ancestor. Since a stemma should explain the state of a text based on as small a number of contamination sources as possible, even a highly ranked (which means very similar) witness can be superfluous when constructing a stemma. This is because there may be one or more other highly ranked witnesses as contamination sources that explain all variants of the descendant without needing the one that has a higher rank. According to Ockham’s razor, a stemma that requires fewer witnesses is preferable to one with more.
In column UNCL (= unclear) one can find the number of variant passages where no decision has been made whether W1 or W2 contains the prior variant.
Column NOREL (= no relation) displays the number of variant passages where W1 and W2 read different variants which, however, have no direct relation to each other (e.g. if W1 reads variant b and W2 variant c, but the prior variant for both is variant a).
If the option “Show Descendants” is selected, potential descendants are also displayed. The list conforms in structure to the one for potential ancestors except in this case the values in column W1<W2 are less than those in column W1>W2.
d. How to interpret the result?
(Example witness 468, default options)
The hypothetical initial text (witness “A”) occupies the highest rank among the potential ancestors of witness 468. If “A” is near the top of such a list, witness W1 is of special interest. In such cases W1 may reasonably be presumed to have conserved the oldest text at some instances without the intermediation of any other surviving witness. However, the level of agreement also has to be taken into account. Since nearly all textual links are missing in the oldest textual tradition, a value of more than 91% agreement with “A” is good. This value is surpassed only by 18 witnesses in relationship with “A”.
The following parameters may help to assess which agreement values can be regarded as relatively high or low. “A” agrees with the majority text (MT) at 89.63%. The average agreement (AA) of all witnesses with “A” is 89.10%. The median (MA) (above and below which the agreement values of respectively one half of the witnesses are positioned) is 89.27%. Tip: A list of all relations with “A” in descending order is displayed, if you enter “A” in this module and choose the option “Show Descendants”.
For comparison: The lowest value of agreement currently is 77.805% (with 1241 and 1838), the highest 99.056% (with 614 and 2412). The average agreement value overall is 87.495%, the median 87.312%. Note that these figures are based on a genealogical evaluation referring to the witnesses included among which the witnesses of the Byzantine text are represented by a selection. The average value would be considerably higher, if all preserved witnesses would have been taken into account.
468 agrees with the majority text (MT) at 94.30%. The average agreement (AA) of all witnesses with 468 is 90.54%. The median (MA) (above and below which the agreement values of respectively one half of the witnesses are positioned) is 91.02%.
The special quality of 468 is demonstrated already by the small number of witnesses in its list of potential ancestors. At 87 %, the agreements with 03, 1739, 04 (all of them first-rank witnesses) are still good for older levels of the tradition. They can all be considered as contamination sources taking into account that the intermediate witnesses, by which the connection to the earliest layers of transmission could be established, are obviously lost. Close relatives of 468 from later levels of the tradition survive, however, including ten witnesses whose agreement values are clearly higher than the MT value of 468 but still do not rank among its potential ancestors.
If the option “Show Descendants” is chosen, a large number of close relatives of 468 are displayed featuring high agreement values. With these, again, the differences between the figures under W1<W2 and W1>W2 have to be considered. If the difference is minor, a potential descendant could become a potential ancestor by changing a few local stemmata.
An examination of as many such tables as possible will make it very clear how the different values of agreement and the characteristic way these values drop within the tables should be understood.
(Example witness 1739, default options)
Only two potential ancestors are shown. These have also proven to be stemmatic ancestors. That is, all instances where 1739 is extant either (1) concur with A or 03, or (2) have, at least indirectly, developed from variants of A or 03. Agreement with “A” is particularly high (being exceeded only by 03).
(Example witness 81, Option “Show Descendants”)
This list is longer than that of 468, but it should be noted that the range of percentages is basically rather narrow (with a difference of only 1.2 percentage points between rank numbers 2 and 13, which means that the difference between ranking numbers in this range is of minor significance). “A” is in first place at a good agreement value which clearly exceeds even the one of 1739. This witness shares the second rank with 35. The variants conforming with “A” could be largely identical to the agreements with 1739 since 1739 is also very similar to “A”. 1739 and 35, the latter of which is a witness of the Byzantine text, two very different potential contamination sources come into consideration. Further down the list quite a few representatives of the Byzantine text follow, but finally 04 and 03 appear with only slightly lower agreement values. The values under W1<W2 and W1>W2 do not differ much with 35 and other “Byzantines”, getting more significant only with 468. Again, assessing some relationships between variants differently could make potential ancestors descendants. Considering the given situation contamination from different text historical directions is possible.
There are no close relatives among the potential descendants. Those coming relatively close feature differences between W1<W2 and W1>W2 high enough to exclude that one of them could become an ancestor by changing some assessments of variants.
(Example witness 1735, Option “Show Descendants”)
1735, too, is a witness without very close relatives, as is shown by the list of both, potential ancestors and descendants alike. 02 and “A” appear at ranks 1 and 2, 81 at rank 5. These three are very different already. The remaining witnesses point into yet another direction. Core witnesses of the Byzantine text are among them, and others follow at about the same distance from 1735. Differences between W1<W2 and W1>W2 are large enough to make it less probable than in the previous example that these potential ancestors could become descendants by changing a few local stemmata.
(Example witness 01, Option “Show Descendants”)
“A” shows the greatest agreement. The values sink below 85% rather quickly, at which point there is an almost equal distance to an extremely great number of witnesses whose levels of agreement are in the range 84-85% and which occupy diverse positions in textual history.
This pattern is typical for a witness with frequent individual variants (variants that it alone has, or that it coincidentally shares with others).
The list of potential descendants suggests that the text of 01 occupies a dead end in textual history. (There is an extremely great distance to its closest relatives.)
(Example witness 1661, Option “Show Descendants”)
There exists a very close potential ancestor, 996. Then, from 90% downwards, follows a sequence of Byzantine witnesses. 996 and 1661 are apparently special cases that have separated from the Byzantine text, differing from it at 300 places.
In special cases such as these, the few potential descendants have such great distances from them that they could never be considered as descendants in a stemma. For example, 1661 is 97th in the list of potential ancestors for 378.
2. “Comparison of Witnesses” Module
a. Purpose of the Module
This module allows comparison of the relationship of witnesses from writing to writing in order to detect any changes of Vorlagen or, rather, changes of contamination sources. If there are any conspicuous changes of the agreement values or the values under W1<W2 and W1>W2 for part of the corpus, the matter can be pursued by taking a look at the potential ancestors of the witnesses based on the data of single letters.
b. The Data Entry Form
The witnesses to be compared can be entered here. Principle uses of this module include comparison of a witness with the hypothetical initial text (witness “A”) and comparison of witnesses that are relatively closely related.
c. The Result
The result places the compared witnesses under W1 and W2, and the genealogical direction of the relationship under DIR (= direction). (The DIR column has “-->” if W1>W2 has a higher value than W1<W2, or “<--” for the opposite case.) Figures in a row refer to the letter given under the heading WRIT (= writing); in the last row (CL = Catholic Letters), values for the entire corpus are shown. Column NR contains the rank in the list of potential ancestors of the descendant witness, based on the data source indicated in the WRIT column. All other columns have the same meaning as in the “Potential Ancestors and Descendants” module.
If a letter is missing under WRIT, at least one of the compared witnesses is lacking there. The data for 2Jn, 3Jn, and Jd should be used with caution on account of the very limited textual base. By selecting “View Differences”, the diverging variants of the compared witnesses can be displayed letter by letter. The table displays the compared witnesses under W1 and W2 and the respective variants under VAR1 and VAR2. The entry under DIR indicates the genealogical direction between the variants. “U” (= unclear) appears in this column, if no decision was made regarding the pair of variants in question. “N” (= no relation) appears, if no direct relation between the variants was established.
d. How to interpret the result?
(Example witness “A” and 1739)
There is a high level of agreement for each data source, with an extremely high percentage in 1Jn. The values referring to 3Jn and Jd are clearly lower. Due to the small amount of data for 3Jn, this is probably not significant, because each difference in this letter deminishes the agreement value by a percentage point.
(Example witness “A” and 03)
In this case, the level of agreement is yet higher and more uniform.
(Example witness 18 and 35)
These two important representatives of the Byzantine text are very similar. The difference of the values under W1<W2 and W1>W2 is either very insignificant or zero. It is not significant that the genealogical direction in Jas differs from the general trend; the differences are so slight as to indicate a weakly developed trend, if any.
(Example witness “A” and 665)
Even though there is a sensationally high level of agreement for 2Pt, we have to bear in mind that about half of this letter is missing in 665. The highest ranked potential ancestor of 665 in 2Pt is 93. If one compares this witness with 665 then extremely high agreement values are obtained, both in general, and especially for 2Pt. One could now compare “A” and 93 at 2Pt, the text of which is nearly complete in 93. The level of agreement in 2Pt is not as extreme for “A” and 93 as for “A” and 665, but it is still remarkably high.
(Example witness “A” and 044)
The agreement values are very inconsistent and show an extreme drop at 2Pt. Even there, however, “A” is a potential ancestor with a rank of 8 (rank 1 being obtained by 623); for the entire Catholic Letters “A” ranks at position 1. (Compare the values under NR.) 044 has a considerable number of peculiarities in 2Pt not shared with any other witness. This fits with the pattern of the agreement values, and seems to indicate that 044 occupies a dead end in one textual tradition.
3. “Coherence in Attestations” Module
a. Purpose of the Module
This module allows for the creation of so-called textual flow diagrams within attestations and also displays possible connections to attestations of other variants at the same variant passage.
Excursus: Textual flow is the consequence of a genealogical decline between the witnesses. The textual flow contains variants that constitute agreement or disagreement between witnesses.
We differentiate different kinds of textual flow. The general textual flow commonly proceeds from earlier to later forms of text. The position of any witness within the general textual flow can be determined by the genealogical coherence of its closest relatives (compare the “Potential Ancestors and Descendants” Module). The particular textual flow is the one that can be established between ancestor and descendant respectively. It contains variants adopted by the descendant and those taken by the descendant as starting points of change. The global textual flow prevails within a global stemma of the witnesses and presumes stemmatic coherence, i.e. the definite hypothesis about which preferably small combination of ancestors entirely explains the text of a descendant. The local textual flow is the one taking place within a certain attestation or within the attestations of all variants at a certain text passage. The local textual flow can be based upon genealogical or stemmatic coherence.
This module looks at local textual flow based on genealogical coherencies.
b. The Data Entry Form
The variant address should be entered based on the common address system of the ECM (chapter, verse, and word number of beginning and end of the variant passage, letter address of the variant; with 2Jn, 3Jn and Jd chapter number “1” has to be entered).
Normally, the entire Catholic Letters should be selected as the database. The genealogical coherencies for the relations of the witnesses which have been collected by analysis of the entire corpus will then form the base.
The single writing should be selected as base if you find incoherencies when the “Catholic Letters” option is selected that give rise to the suspicion that witnesses within the corpus have changed Vorlage.
If the option “Catholic Letters” is selected, those witnesses that are not extant in Jas, 1Pt, 2Pt, 1Jn and at least one of the smaller letters will be considered as fragments. Too much fragmentation can render a genealogical assessment impossible.
It is possible to conduct an evaluation of the connectivity of a variant. Connectivity is the capacity of a variant to connect ancestors and descendants genealogically. It depends on two parameters: (1) the degree of relationship of the witnesses involved, that is on their general level of agreement, and (2) on the character of the variant. In other words, for closely related witnesses even the agreement of insignificant details supports the relationship — the common variants are connective; with a lower degree of general agreement one would rather consider multiple coincidental emergence of a variant if its character advises us to do so. However, in cases where coincidental multiple emergence is unlikely, the variant in question is connective, even for witnesses that have a lesser degree of relationship.
Depending on the assessment of connectivity, the potential ancestors are analysed as long as the appropriate ranking number is not exceeded. The default value of the ranking number is 10. For run-of-the-mill variants where a high fluctuation of the witnesses is assumed, one chooses the option “Low” or makes use of the possibility of a user-defined lower value. A higher value can be selected if the variant — due to its character — is so connective that it should also join less closely related witnesses. The option “Absolute” enforces coherence of all witnesses of the same variant. This option offers a good way to see how remote the witnesses of an attestation are since the ranking number from the list of potential ancestors is displayed (see below).
It is possible to select the variant that constitutes the initial text of the tradition. An entry is mandatory, because the program does not permit a blank here. In the default setting, the program assumes that this is variant a, regardless whether or not a decision about the initial text was made in the local stemma.
c. The Result
The result is a graph displaying the local textual flow based on genealogical coherencies. The arrows point respectively from the (potential) ancestor to the (potential) descendant within the attestation. If one looked up a list of potential ancestors for a descendant in the graph shown here in the module “Potential Ancestors and Descendants”, one would find a ranking number in column NR for the ancestor from which the descendant in this graph is derived. If this ranking number is 1 it will not be noted for the descendant in the graph since this case dominates. If the ranking number is greater than 1, it will be displayed together with the descendant after a slash. The level of connectivity selected in the data entry form determines up to which ranking number the potential ancestors will be charted.
If no potential ancestor can be found for a descendant within the same attestation then the attestations of other variants at this place will be searched. The locations (variants) will be displayed outside the frame that encircles the current attestation, in each case together with the corresponding potential ancestor. Next, the ranking number the potential ancestor obtains in the ancestor list of the descendant will be displayed, along with the identification of the ancestor.
Thus, one can obtain an overview as to whether a variant can be deduced from the current variant at all.
The contents of the graph can also be displayed as a table (“Show Table”). This is useful if the course of the arrows is too confusing. In the chart, the witnesses will be arranged in generations (G1-G10). G0 is the generation of witnesses located with the variant from which the current one is deducible. The identification of this variant is to be found under VARID. N0-N10 contain the ranking numbers. The reading of such a chart will be explained in the next example.
There is also a “Show Local Stemma” option that can be used to display a local stemma of the variants at the selected passage. Such local variant stemmata have been created for all variant passages in order to calculate the genealogical coherence of the witnesses in a writing or in the entire corpus based on the variant relations. In the module “Coherence in Attestations”, those genealogical coherencies of witnesses are projected back to the variants of a variant passage. In doing so, it may happen that relations presumed in the local stemma of the variants are not compatible with the result obtained in this module (e.g. if the local stemma deduces one variant from another one for which there is no relation through genealogically coherent witnesses according to the result of this module). Inconsistencies like this one are important indications that the local stemma should probably be revised. The local stemmata are based on genealogical data as collected for the single writings. Textual flow diagrams, however, can now be based on the data for the entire corpus. Cf. the preface and the note at the end of this guide. In some local stemmata, a question mark may occur in place of a source variant. Moreover, variant identifiers can be split (e.g. b into b1 and b2) or merged (e.g. b and c to b!). Cf. the explanations on the module “Local Stemmata”.
d. How to interpret the result?
(Example Jas 2,3/50-56b, default options)
In the frame the entire attestation of the variant is visible, divided into three branches (one consists of 999 only). In view of the selected connectivity this result indicates that the variant may have evolved three times. The sources are displayed outside the frame. In each case, only variant a is quoted as a source. The closest related potential ancestor for 642 would be 424. Within the attestation of variant b no potential ancestor of 642 with a ranking number between 1 and 10 can be found. This applies to 1739 as well. Its most closely related potential ancestor is “A”, the reconstructed initial text (variant a).
For most of the witnesses within the attestation of variant b, a potential ancestor with the ranking number 1 (which will not be displayed) can be found. However, 1739 has the ranking number 2 in reference to the descendant 025 apparent in the itemisation 025/2. 025 has the ranking number 8 in reference to the descendant 1448. That the υπο of variant a has been repeatedly and independently altered into επι (variant b) is plausible, as it facilitates reading.
If, however, the option “Low” is chosen for the connectivity, 025 will no longer be accepted as an ancestor of 1448. 1448 receives its own branch and is derived from 35, its next most closely related potential ancestor in variant a.
If one wants to find out how distantly the witnesses that developed independently from variant a are related, the “Absolute” option has to be selected. It is then indicated that according to the option only one connection to variant a via 1739 remains. However, please note the path from 025 to 999 via 1448 and 642 with its constantly weakening genealogical coherencies. In the list of potential ancestors of 999 (accessible by selecting “Include Minor Fragments”), 642 appears at ranking number 39!
I return once more to the “Average” option for the connectivity and open the chart connected to the graph (“Show Table”) then compare the two. In the chart you can find the witnesses displayed outside the frame (from which witnesses inside the frame are derived) under G0. The variant they read is indicated under VARID. The descendants of witnesses under the heading G0 can be found under G1, e.g. 642 as descendant of 424. 424 has the ranking number 1 as potential ancestor, apparent under N0. In column G2, the descendants of 642 are found in the line of 642 and below — until a new witness appears in column G1, in this case 999. Under the heading N1, one can find the ranking number that the ancestor (under G1) possesses in the ancestor list of a descendant (under G2). All these details can be obtained with the graph, too.
However, there is important additional information in the table. It is found in the columns PEV0, PEV1 etc (PEV = percentage of equal variants). E.g. the percentage in the first line of the column PEV0, 93.371, specifies the agreement of 424 (G0) and 642 (G1). It is the same percentage we find when we look for the potential ancestors of 642 in the “Potential Ancestors” module (column PERC1). The ranking numbers may indicate that there is a chance that a potential ancestor may become a stemmatic ancestor in an optimal substemma. However, the percentages of agreement must additionally be consulted in order to assess the possibility of merely coincidental agreements.
The following example shows how different the relation of ranking number and percentage of agreement may be. For witness 025 (column G2) 1739 is shown as the most closely related potential ancestor within this attestation. 1739 has the ranking number 2, found in column N1. In column PEV1 the agreement with 025 is given: 88.769 %. Witness 025 itself is a potential ancestor of 1448 (cf column G3). According to column PEV2, the percentage of agreement is 88.772, and the ranking number (N2) is 8. We see that different ranking numbers have similar percentages of agreement. For a more precise assessment one should use the “Potential Ancestors” module, where you can see that the closer genealogical environment of 025 — as far as ancestors are concerned — is virtually lost. Only 2298 is shown above 1739. However, there is no genealogical direction from 2298 to 025 or vice versa. Traces of the closer genealogical environment can only be found with less affiliated witnesses. The case of 1448 is quite different. According to the table of potential ancestors, there is a dense genealogical environment in the range of 92 and 91%. Compared with these values, the agreement with 025 is clearly lower.
Returning to “Coherence in Attestations” for Jas 2:3/50-56b I choose the option “Low” in order to find out what happens if I declare variant b to be the initial text (Initial Reading b). The graph will transform decisively due to the fact that the hypothetical witness “A” now constitutes its source. Theoretically, it is possible that variant b is the initial text. However, part of the attestation (642, 999 and 1448) would have developed the text from variant a and both, 642 and 1448, passed it on to descendants.
If under the same option the diagram for variant a is displayed, the entire attestation of this variant will be attached via 03 to “A” at variant b. This is theoretically possible as well. However, it is very unusual that the witnesses ranging in the highest area, close to 03, are never seen to have received the variant from ancestors with ranking number 1 (cf. the figures after the slash). This is because most of them have “A” as their most closely related potential ancestor, which is shown, if a is selected at Initial Reading.
If we rather assume — since επι has definitely developed from υπο for part of the witnesses (if variant b is the initial text) — a trend from υπο to επι, the question of whether variant a (υπο) has incidentally developed from variant b (επι) arises. If a is entered at Initial Reading, one obtains a diagram featuring a perfect coherence. Nearly all descendants are tied to potential ancestors with rank number 1.
One could get the idea that the shortest variant c presents the initial text. If c is entered under Initial Reading and the diagram for c displayed, the hypothetical initial text witness “A” is indeed selected as potential ancestor with rank number 1 for the witness 044. However, there is more to this than the ranking number.
If one opens the program again in a new browser window and takes a look at the potential ancestors of 044 in the module “Potential Ancestors and Descendants”, one will discover that this witness has no close relations. The agreement with “A” is the closest, but at 88.671%, it falls short of what would be expected of a good witness. (The best witness at variant a is 03 with 95.995% agreement; at reading b it is 1739 with 93.003% agreement; the best Byzantine witness is 468 with 91.341% agreement.)
If one selects variant a as Initial Reading and takes a look at the diagram, it appears likely that variant c emerged from variant a or b, although the wording of b fits more closely.
At variants f and g, which share the pleonastic ποδων and only differ through υπο … σου (f) and επι … μου (g), one could wonder whether they are connected with each other and whether there is a connection between υπο at variant a and επι at variant b. The diagram for variant f offers a and b as source variants and one will decide to accept a, both because of the same preposition and maybe because of the potential ancestor with the better ranking number. The diagram for variant g offers, under the option “Low”, a and f as ancestor variants for the connectivity of which only f could be accepted due to the wording; however, a connection to the επι of variant b does not emerge. This will only arise under option “Average”, and then to the potential ancestor 218 with the ranking number 9. “Show Table” displays a low agreement value with 33 at g for 02 at f. However, the agreement value of 218 at b is even considerably lower. 33 has actually only one closely related potential ancestor, namely 2344 under a. If we assume that επι has repeatedly developed from υπο anyway we can forego this connection.
At variant d (παρα), which theoretically could have developed from a or b, the textual flow diagram shows that only a development from variant a is possible. If one should presume to consider variant d to be the initial text (Initial Reading = d), the diagram demonstrates that we would not have a surviving witness for this initial text since 398 cannot be connected with “A”. Only if connectivity is set to “Absolute”, a connection between “A” and 398 appears. It agrees with “A” at 89.595%, which is low and does not permit far-reaching conclusions, the more so because 398 has many relatives at the range of 91%. Internal criteria would have to be overwhelming to justify a textual decision in favour of variant d here.
4. “Coherence at Variant Passages” Module
a. Purpose of the Module
This module should assist the user to develop an overview of all the relations between the variants of a variant passage.
The result can be regarded as a summary of all results that one would receive if the module “Coherence in Attestations” was applied to all variants of a variant passage.
b. The Data Entry Form
This form is derived from the one used in the “Coherence in Attestations” module, where most of the same options are selectable (see above). Additionally, we find the option group “Display Mode”.
The “Qualified: All Interrelations” option only displays relations that are possible on the basis of the respective strongest genealogical coherence between witnesses of different variants. On the other hand, the “Qualified: Interrelations of the first order only” option causes a limitation of the result: relations between variants will only be displayed if — in reference to the descendant — the potential ancestor in an ancestor list has ranking number 1.
c. Result and Interpretation
(Example Jas 2,3/50-56b, default option)
The textual flows that emanate from potential ancestors with ranking number 1 are displayed in blue, the others in green. The numbers shown with the green relations indicate the ranking number of the potential ancestor.
5. “Local Stemmata” Module
This module documents the local stemmata which are graphical representations of the presumed genealogy of variants. Local stemmata are only included for variants assigned to the first hands of Greek continuous text manuscripts.
In order to display a local stemma, a variant address that conforms to the ECM system has to be entered into the input form using numbers for chapter, verse, and beginning and ending word(s) of the variant passage; “1” has to be entered as the chapter number for 2Jn, 3Jn and Jd.
As a rule, variant a appears at the top of the local stemma, being the text of the reconstructed initial text “A”. A question mark is placed at the top of the local stemma in passages where we abstained from a reconstruction of the initial text for the purposes of genealogical analyses (e.g. Jas 4:14/8). A question mark was also used whenever the source of a variant could not be determined (e.g. 1Pt 2:21/16-22). In such cases, for which we have here a particularly impressive example, “?” does not represent one common source variant. Instead, the respective source variant, to which the arrows point is uncertain. In the present example the source variant is uncertain in seven instances.
Note: The local stemmata are based on genealogical relationships between witnesses that have been discerned from analyses restricted to the Catholic Letters as individual writings. Data for the entire corpus were not at our disposal until after the last installment was published. The local stemmata are currently being revised on the basis of the now comprehensive data (cf. the preface of this guide).
Note: A local stemma of variants may contain two instances of the same variant (e.g. Jas 1:12/31). For example, a variant b may appear as b1 and b2. In such cases, a variant has been divided logically into two due to profound genealogical divergences so as to make it possible to relate different parts of the attestation to witnesses of different variants. The opposite has happened if, for example, variant b! appears in the local stemma (e.g. Jas 3:16/8-10). The exclamation mark signifies that variant b has been combined with other variants to form one logical variant and, accordingly, the attestations to form one attestation. This takes place if, for example, variants depend on such changes as ει/η or ο/ω (in the case of many indicative/subjunctive variants), and it is not possible to draw conclusions about the exemplar being used by a scribe.
* The variants of the majority text were determined by a selection of nearly pure representatives of the Byzantine text (cf. ECM IV.2, Supplementary Material, 2.2 Codices Byzantini). Where these representatives point to a split Byzantine attestation, it is not possible to determine the majority reading with certainty. Such passages were not included for determining the MT value in the “Potential Ancestors and Descendants” module.