Ἰουδαίαν in Acts 2:9: Reverse Engineering Textual Emendations

Abstract Building on a plethora of conjectured emendations for IOϒΔAIAN, this article approaches the issue once again to test the viability of a quantitative tool and to establish the possibility of palaeographical confusion of IOϒΔAIAN with an alternative topo- or demonym. The article starts with an experiment using Greek topo- and demonyms derived from contemporary Christian, Jewish, and Classical sources to establish a palaeographical confusion score. Next the likeliness of the “closest” alternatives in the geographical arrangement of Acts 2:9–11 is explored. The article ends with an evaluation of the possible implications for the text in Acts 2:9 as well as a critical appraisal of the method for conjectural criticism.


Introduction
In the first part of this study, "Ἰουδαίαν in Acts 2:9: a diachronic overview of its conjectured emendations", we demonstrated that several textual emendations have been conjectured to solve the alleged interpretive problems with IOϒΔAIAN in Acts 2:9. In the history of interpretation, solutions have been explored in three directions: (1) IOϒΔAIAN has been interpreted as an adjective instead of a noun, (2) IOϒΔAIAN has been regarded as a corruption or later interpolation in the text, and (3) several toponyms have been proposed as a solution to emend the text. None of these suggestions have been generally accepted.
The first aim of this article is to contribute to the discussion by testing the hypothesis that the text might have been corrupted during transmission due to palaeographical confusion of Greek characters, i.e. the text might originally have contained a different Greek word.¹ To test this hypothesis, we need to identify a Greek toponym that is palaeographically, historically, and geographically suitable to replace IOϒΔAIAN in Acts 2:9. Therefore, we use a reverse engineering approach: instead of starting from the text itself, trying to find a fitting explanation of the interpretive problem concerning IOϒΔAIAN, we start with a list of toponyms and research which toponyms are likely alternatives based on the criterion of palaeographical confusion.² The results of this test will be further analysed to see whether we can solve the interpretive problem in Acts 2:9 or whether we can exclude palaeographical confusion as a possible explanation for the alleged corruption of the text.
The second aim of this article is to evaluate whether our method is helpful for identifying possible cases of palaeographical confusion in problematic texts.
The article has been organised as follows: Section 2 is concerned with the method and data used for this study. Section 3 presents the findings after the application of the method for Acts 2. Section 4 evaluates these results by considering the implications of the experiment for Acts 2 and provides a critical appraisal of whether our experiment recommends a wider application of the method for other cases. Section 5 finalises the article with a conclusion and outlook for further research.

Method, algorithm, and data
To gauge the probability of confusion between a toponym and the supposedly corrupted IOϒΔAIAN, we research how easy it might have been that during textual transmission a scribe misreads one character for another character and that this resulted in the current reading.
We propose the following method (illustrated in Figure 1): we use a computer algorithm and three data sets comprising toponyms occurring in early Christian literature, Greek Jewish Scriptures, and Classic texts to identify possible palaeographic alternatives. Furthermore, we test the appropriateness of the most likely results in their wider historical and geographical contexts. Louw   2 Reverse engineering is the terminology employed in technology-related disciplines. The Cambridge Dictionary defines it as "the process of studying another company's product to see how it is made, sometimes in order to be able to copy it". In this article, we use this concept in a more loose way: we start with the hypothesis about the textual transmission process and a data set of semantical candidates, and subsequently prune the number of likely toponyms by adding additional criteria.

Confusion distances
The confusion distance has been proposed by Van Altena et al. as a metric to calculate the probability of palaeographical confusion between two words.³ The underlying assumption is that the text could have been corrupted at a very early stage in the transmission process. When this corruption was unintentional, it might have been caused by a misreading of the Vorlage, possibly due to the confusion of Greek majuscule characters. In their study, Van Altena et al. propose that the probability of confusion of Greek majuscule characters is not equally likely and that combinations of characters might be confused for a single character and vice versa. They therefore propose a confusion table that contains character pairs and a digit representing the assumed ease of confusion. Our study uses this algorithm and the accompanying confusion table to establish confusion distances of toponyms.

Data
The second ingredient for our study is a collection of place names (i.e. toponyms) with geographic locations, also known as an onomasticon or gazetteer.⁴ For the purpose of our study, this data set should contain Greek toponyms that correspond in grammatical function with IOϒΔAIAN, are supposed to be familiar to the original author of Acts, and fit the geographical arrangement in Acts 2:9-11.⁵ To the best of our knowledge, a gazetteer that fits these requirements does not exist. We therefore retrieved toponyms from textual resources, inflected them to the proper grammatical case (i.e. accusative or nominative in case of uninflected nouns), and assigned geographical coordinates to each toponym.

Retrieval of toponyms
We used three types of literary resources. For the New Testament, we created a data set from a semantic dictionary.⁶ This data set contains 259 location-related pronouns, including toponyms, demonyms, and derivatives such as adjectives, which occur in the 27 books of the New Testament. Furthermore, we argue that it is reasonable to construct a list of toponyms, which appear in the Septuagint (LXX), and take them to reflect contemporary toponyms that might have been known to an author contributing to the tradition of early Christian literature in the time of writing the book of Acts.⁷ For this purpose, we retrieved  3 van Altena et al., "Spatial Analysis of New Testament Textual Emendations Utilizing Confusion Distances." 4 In a modern sense, a gazetteer consists of "three core elements: toponyms (and their history), spatial location (in various representations, such as points, lines, and polygons), and classification (e.g., types and categories of places)", cf. Yuan, "Mapping Text," 109-23. 5 van Altena et al., "Spatial Analysis of New Testament Textual Emendations Utilizing Confusion Distances," 48-9. 6 Louw and Nida, Greek-English Lexicon of the New Testament. We selected all entries in the range 93.389-93.615 (Places) and exported them to a spreadsheet. 7 Van der Meer identifies five problems to temper expectations in a reconstruction of LXX topography: "[1] most of the geography and topography of the world in which the Septuagint was made […] is not reflected in the Greek translations of Hebrew Scripture, [2] that the corpus for the study is a heterogeneous collection, [3] that several Greek renderings of Hebrew toponyms rest on misinterpretations or [4] deliberate actualizations, and finally [5] that several Greek toponyms were misunderstood in the course of textual transmission", cf. van der Meer, "The Natural and Geographical Context of the Septuagint: Some Preliminary Observations," 393. However, since the aim of this article is not to reconstruct an actual topography of the Lxx but to construct a list of toponyms that the author of Acts might have been familiar with, these caveats will not affect our main conclusion. toponyms from the Septuagint.⁸ The third group pertains to toponyms occurring in classical authors. We used the website Topostext⁹ and the Greek-English dictionary by Montanari et al.¹⁰ as the two major sources to compile this list of toponyms. Especially the retrieval of topo-and demonyms from a dictionary appeared to be a tedious endeavour.¹¹ Eventually, we made the resulting lists of toponyms in singular nominative form available online.¹²

Inflection of toponyms
Subsequently, we had to inflect these lists of toponyms, since the data retrieved from dictionaries and the online databases are in nominative singular form. Instead of looking up almost 10,000 individual toponyms ( Table 1), we created a computer script¹³ that inflected the toponyms to the different cases in singular and plural forms.¹⁴ The results of these processes will be evaluated using qualitative criteria (see Section 4).

Geographical parsing
The position of each toponym on the skin of the earth has to be expressed in x,y-coordinates to enable geographical analysis. This is achieved by geographical parsing, which could be defined as "the task of identifying and resolving toponyms to their geographical coordinates".¹⁵ Due to their nature, it is not equally straightforward to acquire coordinates by geoparsing for the three data sets that we introduced in the previous section.  8 The search of the Septuagint was performed with the aid of the computer program Logos Bible Software version 8.6.0.0052. See https://www.logos.com/. We used a two-step approach. First, we created a concordance for the whole Septuagint that we filtered on Biblical Entities and subsequently on Places. Next, we used the retrieved place names to query the Septuagint and retrieve the Greek words and the biblical references. The Septuagint text is based on: Swete, The Old Testament in Greek. We also used the tagged text that is provided by Tan et al., The Lexham Greek-English Interlinear Septuagint. 9 Kiesling and Aikaterini Laskaridis Foundation, "ToposText," Gazetteer, ToposText Web Version 3.0. 10 Montanari, The Brill Dictionary of Ancient Greek. 11 Montanari's dictionary is lacking an arrangement according to semantic field. Neither is there any explicit indication (e.g. the abbreviation n. pr. loc.-nomen proprium loci) that the particular lemma is dealing with a toponym. As such, the dictionary conforms to the general practice to which Louw-Nida and Swanson appear to be an exception. To tackle this obstacle, we retrieved toponyms from Montanari doing multiple searches for the terms cape, city, country, district, ethnic, island, mountain, pass, people, place, port, province, region, settlement, tribe, urban, and village. 12 van Altena, "New Testament Toponyms;" van Altena, "Greek Old Testament Toponyms;" van Altena, "Greek Toponyms Collected from Classical Literature." 13 van Altena, Greek Noun Inflection Script. 14 Our inflection script worked well for regular nouns, but we needed a solution for exceptions and nouns, which theoretically can be inflected in multiple ways if its gender is unknown (e.g. nouns ending on -ος could either be inflected to the accusative singular ending -ος or -ον be dependent on the gender of the noun). Instead of figuring out the correct inflection for each individual case, we implemented a solution that produces all possible variant inflections. This means that the script inflects these nouns in multiple ways. Though we expect only one of these inflections will be correct for most of the cases, we can use this inflected list of toponyms as an intermediate input for subsequent processing. New Testament toponyms are easily tied to the Tyndale Individualised Proper Names with all References (TIPNR).¹⁶ Toponyms from the Septuagint can be linked to a toponym in TIPNR using their Hebrew equivalents but need to be checked since some Greek toponyms seem to have been actualised during translation.¹⁷ Therefore, the LXX toponyms have also been linked to their equivalents in the Pleiades project¹⁸ as far as possible. Toponyms from classical resources had already been given a location in Topostext.¹⁹ For the geoparsing of the additional toponyms we retrieved from the Montanari dictionary, we used the Pleiades project.

Results
The confusion distance algorithm was used to estimate the probability of palaeographical confusion with IOϒΔAIAN for each toponym in our list. From the results shown in Figures 2 and 3 and Table 2, it is apparent that very few toponyms are palaeographically close to IOϒΔAIAN.
Only 6% of the toponyms has a confusion distance smaller than 4 and even 0.9% a confusion distance smaller than 3. Table 3 presents the toponyms that are palaeographically closest to IOϒΔAIAN, which means they are the most likely to be confused.
Taken on its own, the confusion distance might however give a distorted picture since a low confusion distance does not automatically imply a small number of operations to change one word into another. For example, IEPA ΠϒΛH has a small confusion distance with IOϒΔAIAN, but this metric conceals that the transformation needed to change IEPA ΠϒΛH into IOϒΔAIAN requires seven operations (neglecting the space). To account for this bias, we here introduce a second metric, the probability index, ProbIdx, which accounts for bias in the confusion distance, confdist, by multiplying it with the required number of operations, ops: The results show a small shift to the right in the frequency diagram of the operations compared to the distribution in the histogram of the confusion distance. This can be easily explained, since the confusion distance is the result of a weighted score of the individual operations.
For the remainder of this study, we have selected the first 49 toponyms that have the closest probability index with IOϒΔAIAN.²⁰ This is of course an arbitrary selection based on the assumption that a maximum of four unintentional confusions might still be conceivable.


16 The TIPNR is a collection of every proper or geographical name in the Bible, which are linked to the original languages. That is, Hebrew for names occurring in the Old Testament, and Greek for those found in the New Testament. The data set does not contain alternative readings, neither references to Greek names found in the Septuagint nor toponyms occurring in the Aramaic parts of the Old Testament. Geographical names have been enriched with a geolocation, which was derived from the OpenBible data set, Smith, "Bible Geocoding -Bible Maps in Google Earth and Google Maps." and Instone-Brewer, "Tyndale Individualised Proper Names with All References." 17 See also note 7. 18 The Pleiades data set is a community-built gazetteer of ancient places, covering the Greek and Roman worlds extensively, and currently broadening its scope to Ancient Near Eastern, Byzantine, Celtic, and Early Medieval geography. The data set is maintained by the Ancient World Mapping Center and the Institute for the Study of the Ancient World and is available through services to individual human researches as well as for consummation by computational humanities research, cf. Bagnall et al., "Pleiades: A Gazetteer of Past Places." 19 ToposText is an online database that contains location references relevant to Greek history and mythology. More than 5,000 references to Greek toponyms were collected from ancient texts covering a time span starting in the Neolithic period up through the second century CE, cf. Kiesling and Aikaterini Laskaridis Foundation, "ToposText." The 56 entries in Table 3 can be divided into 32 places (settlements, towns, or cities) 16 regions, 8 miscellaneous features (hydrography, building structures, or landmarks), and one false positive. In the remainder of this article, these entries have been reduced to 48 unique toponyms by eliminating the false positive ΙΕΡΕΙΑΝ and redundant toponyms. Redundancy occurs in the cases of ΕΟΡΔΑΙΑ, ΕΟΡΔΑΙΑN, and ΕΟΡΔΙΑN; ΙΟϒΔΑΙΑ, ΙΟϒΔΑΙΟΝ, ΙΟϒΔΑΙΟΣ, and ΙΟϒΔΑΝ; ΙΣΤΙΑΙΑ and ΙΣΤΙΑΙΑΝ; ΡΟϒΔΙΑ and ΡΟϒΔΙΑΝ.

Discussion
This section discusses the semantical, geographical, and historical likeliness of the palaeographically fitting toponyms and evaluates the applicability of the method.

Candidature for emendation
The results in the previous section show that only for a few instances it is reasonable to argue for palaeographical confusion. To advance the establishment of criteria for the evaluation whether any of the   Table 3 might be a plausible substitute for IOϒΔAIAN in Acts 2:9, we first discuss considerations about the variety of attestation, the validity of temporal attestation, and the probability of Jewish communities in a region.

toponyms listed in
Widespread attestation in Classical, Jewish, and/or Christian literary sources does not help: evidence for the familiarity of a toponym can be used to advocate the likeliness as well as the improbability of a specific toponym simultaneously. The ambiguity is that while a wider attestation in literary sources might increase the probability that the author of Acts might have used the toponym, simultaneously such an attestation might reduce the chance that a scribe would have replaced a toponym for a better known candidate.
Likewise, the time span for which a toponym is attested does not contribute much to resolve the issue. Though it is tempting to advocate that a particular toponym might have been anachronistic to the author of Acts since we only have attestation from later sources, this argumentum ex silentio can never be conclusive since we cannot prove that the author was unfamiliar with the toponym. We simply do not know.
On a similar vein, the observation that the catalogue in Acts 2 designates Jews from several nations and regions (Acts 2:5, 11) does not help to further exclude toponyms as possible candidates, since the  presence or absence of Jewish settlements in a specific region cannot be demonstrated beyond doubt from the partial historical evidence, nor do rhetorical statements like Philo's "not only are the main lands full of Jewish colonies but also the most highly esteemed of the islands" (Leg. Ad Gaium 282) contribute to settle the issue.²¹ The statement is too generic to establish evidence for the Jewish population in a specific location. More is to be expected from exclusion according to semantic subtype, fitting in the geographic progression, and contextual appropriateness within Acts 2:9. The results from Table 3 will be discussed using these three criteria.²²

Exclusion according to semantic subtype
The majority of palaeographically probable toponyms can be discarded as likely replacements for IOϒΔAIAN when their semantical subtype is considered in relation to the structure of the catalogue. The function of the catalogue is to clarify the identity of the audience in Acts 2:8. The catalogue consists of four units ordered in a chiastic structure.²³ The first and last units consist of groups of demonyms, while the second and third groups start with an active participle, respectively, οἱ κατοικοῦντες (residents) and οἱ ἐπιδημοῦντες (sojourners). IOϒΔAIAN appears in the second group as an object of κατοικοῦντες. This constrains the possibility of toponyms by excluding the hydrography-related toponyms Λουδιας, Iορδάνης, and Ἰσάρας, and names pertaining to building structures, i.e. Ἱερά πύλη, Προπύλαια, Eὐπαλία, Θυαμία, or a landmark like Λευκάτας.
Besides pairing κατοικοῦντες with regions (Acts 2:9, 13), Luke combines it with city names (Jerusalem in Acts 2:14; 13:27, and Lydda in Acts 9:35). The author of Revelation uses the nomen in a construction to refer to "those who dwell on the earth" (Rev 11:10; 13:8; 17:2, 8). However, although a city name would technically fit the second unit of the catalogue, this would be a surprising deviation of its pattern: in parallel to the other regional toponyms one would also expect a region name at the spot of IOϒΔAIAN. Heuristic support for this observation is found in the historical discussion of proposed emendations, which all describe regions.²⁴ Therefore, the palaeographical-related settlements …Ἄστραια, Γραῖα, Ἐλαία, Oἰχαλία, and Ὑδραμία; the towns Xετταία, Σουάτρα, Tριταία, Γεραρα, and Ἰμίχαρα; and the cities Bούδεια, Γάδαρα, Γόρδιον, Ἑστίαια, Eὔταια, Ἰουλίς, Ἱστίαια, Ἰστρία, Λῐλαια, Mονδαία, Nουαρία, Oὐγαυά, Oὐάγα, Oὐαρία, Oὐελία, Oὐλία, Ποτείδαια, Ῥουδία, Στάδια, and Tουία are not found to be fitting alternatives to emend IOϒΔAIAN in Acts 2:9. This means that 38 of the palaeographically likely unique toponyms can be safely discarded as potential emendations of IOϒΔAIAN on the basis of their semantic subtype.

Fitting in the geographic progression
The remaining 11 toponyms pertain to regions that all are palaeographically close to IOϒΔAIAN. For these regions, the question needs to be addressed whether they do fit in the geographical progression of Acts 2:9-11, somewhere between Mesopotamia and Cappadocia.
To visualise the potential search area, we used ArcGIS Pro software to create a multiple ring buffer from the centre point of the straight line between the centre points of Mesopotamia and Cappadocia Unfortunately, none of the palaeographically related regions form geographically fitting alternatives for IOϒΔAIAN. Ῥιπαῖα refers to a mythical range of mountains that were thought to form the northern edge of the world. The region was thought to be uninhabited, and already in classical times authors doubted its existence.
The other alternatives ( Figure 5) break the geographical progression and have additional issues: Ἰστρία could be either a region or people on the north-eastern Adriatic coast as well as a city at the Black Sea. The region Πισιδία in Asia Minor is attested both in the book of Acts and in classical works. Ἰταλία roughly corresponds to modern Italy and is very well attested in Classical, Jewish, and Christian texts. It is also hard to conceive in which way this toponym could fit the geographical progression and how the overlap with Rome (Acts 2:10) should be explained.
Additionally, the African region Nομαδία, the Macedonian landscape Ἐορδαία or Ἐορδία, the Aegean islands Ἱεράκια and Ῥοδία, and the island group Πεταλία do not fit the geographical arrangement better than IOϒΔAIAN.

Contextual appropriateness of Ἰδουμαία
The remaining toponym ΙΔΟϒΜΑΙΑ has already been proposed as an alternative to IOϒΔAIAN by several authors.²⁵ The designation was first applied to the country of Edom and later also to the southern part of Judah. This region had an important strategic position for controlling trade routes.  25 Barthius, Adversariorum commentariorum libri LX quibus ex universa antiquitatis serie, omnis generis, ad vicies octies centum, auctorum, plus centum quinquaginta millibus, loci, tam gentilium quam christianorum, theologorum, iureconsultorum, medicorum, philosophorum, philologorum, oratorum, rhetorum etc. obscuri, dubii, maculati, illustrantur, constituuntur, emendantur, cum rituum, morum, legum, sanctionum, sacrorum, ceremoniarum, pacis bellique artium, formularum, locutionum denique, In favour of ΙΔΟϒMΑΙΑΝ is that it is palaeographically close to IOϒΔAIAN.²⁶ Only two operations are required to transform it to IOϒΔAIAN and a variant reading on Mark 3:7 transposes the two words.²⁷ Furthermore, the toponym was valid and common during the time of writing, and the region had Jews among its population from various ethnic backgrounds (Mark 3:8; JW 2.43).
If ΙΔΟϒMΑΙΑΝ is assumed as the original reading, additional support is still needed to show how the current reading IOϒΔAIAN could have come into existence. One might speculate that a copyist wanted to remove a reference to ΙΔΟϒMΑΙΑ especially because of the turbulent relationship between ΙΔΟϒMΑΙΑ and IOϒΔAIA: during Nebuchadnezzar's besiege of Jerusalem the Edomites scorned and mocked the endangered Jews. After Jerusalem was defeated in 587 BCE, its Jewish population was deported into Babylonian captivity and their homes and possessions were confiscated by Edomites (Am. 10-14, cf. Ps. 137:7). Decades later when the Jews returned from being exiled, they found the homes of their families occupied by Edomites. A few centuries later, after the Arabs had invaded the territory, the Jewish high priest John Hyrcanus forced the Idumeans to follow the Jewish law and thus had them circumcised (Jos. Ant 13.9.1). Later on, the tyrant king Herod the Great was a born Idumean.²⁸ A mutual hostility between these two peoples is not difficult to perceive. This is however pure conjecture and it does not explain why such a well-known topographic name like ΙΔΟϒMΑΙΑ could have been confused with ΙΟϒΔΑΙΑ, which introduces even more difficulties. Furthermore, ΙΔΟϒMΑΙΑ does not provide a better solution to the odd geographic progression²⁹ (which however could also be argued in favour of its originality).

Summary of findings on alternatives
Though it is tempting to draw firm conclusions about the likely candidates for IOϒΔAIAN in Acts 2:9, there is no significant evidence to warrant any of them.
What can be concluded is that an unintentional scribal error based on palaeographical confusion of IOϒΔAIAN with one of the toponyms we retrieved from the New Testament, Greek Old Testament, and Classical literature seems very unlikely. We simply did not find a fitting candidate that was palaeographically close enough to IOϒΔAIAN and might have been altered unintentionally in only a few operations and that-on top of this-is also meeting the historical, geographical, and contextual requirements to fit the catalogue in Acts 2:9-11.
For the moment, this leaves the problem unsolved. However, the possibility remains that a new artefact might be discovered that contains a different spelling of an existing or even a completely unknown toponym that fits the geographical progression, corresponds to the expected semantic subtype (i.e. being a region), and also suits the wider historical and literary context. Such a discovery would provide an ideal toponym to emend IOϒΔAIAN in Acts 2:9.
Until such a discovery is done, it might be better to refrain from speculation about unintentional palaeographical confusion and to accept that the catalogue of nations in Acts 2 is a peculiar list in multiple facets. The mention of IOϒΔAIAN is just one of these peculiarities.

Critical appraisal of method
Our assessment of the method elaborated in this research is however much more positive. Given the right conditions (see below), this method provides new insights that otherwise are difficult to obtain or would be cumbersome since they require an enormous amount of manual work. Using a computer algorithm to gauge whether palaeographical confusion of words might be a solution to an interpretive problem seems promising for several cases.
A first case can be found in the gospel narratives (see Matt 3:4 and its parallel text Mark 1:6), where the character of John the Baptist is introduced. His portrayal as someone who is wearing a camel-skin garment and eating locusts has been interpreted as a reference to Old Testament prophets. However, some have doubted that a man could have actually been eating locusts and therefore conjectured what John could have been eating instead. At this point, the method presented in this research might be helpful to identify a Greek noun belonging to the semantic field of food to test whether alternatives could be detected that might be explained by palaeographical confusion. Similar to this study, these results should be analysed along qualitative conditions to decide between possible and plausible alternatives.
A second case where the method might be helpful is the quest for a fitting verb in Hebr. 11:37. While most of the manuscripts here have ἐπειράσθησαν ("they were tempted"), this is strange within the enumeration of rather specific violent deaths. To solve this puzzle 17 divergent emendations have been proposed. It might be helpful to test which verbs are likely to be confused palaeographically with  29 So Michelsen, in Submission to Prijsvraag G 94: een verhandeling over de toepassing van de conjecturaal-kritiek op den tekst van de schriften des Nieuwen Testaments, II-13-II-14.
ἐπειράσθησαν. This requires an inflection of the verbs to the identical grammatical-morphological form and a semantic discussion of the found alternatives.
A third case where our method can add value is in the interpretation of 2 Pet. 3:10. Here the verb εὑρεθήσεται ("shall be found") is unintelligible and several other verbs with a diversity of meanings have been conjectured.³⁰ Our algorithm might contribute to identify verbs that are palaeographically close to εὑρεθήσεται and would make sense in the sentence.
Each of these cases could greatly benefit from a grammatical-morphological data set as well as from a semantic dictionary that should be both available in digital format.
The scope of the method could also be extended to different types of confusion, for instance, phonetical. This would require a second confusion table expressing the ease of mispronunciation, mishearing (or mis-vocalisation) of a text. Furthermore, scenarios assuming Hebrew, Aramaic, or Syriac sources to Greek texts can be scrutinised using our algorithm and a tailored confusion table. The same exercise could for instance be performed, supposing a Hebrew origin for the list of nations.

Conclusion
In this study, we dealt with the alleged corruption of IOϒΔAIAN in Acts 2:9. As we summarised in part 1 of this article series, a plethora of suggestions have been proposed to solve the problem in the text and these suggestions were argued from differing perspectives: a background in contemporary textual sources, the intrinsic logic of the narrative, the fittingness in the geographical arrangement, and the probability of palaeographical confusion.
To offer a contribution to this discussion, this study set out to investigate the fittingness of alternative toponyms in the text assuming palaeographical confusion. It therefore used a confusion table, which enabled the researcher to quantify the likeliness by which a scribe would confuse particular letter combinations, and an algorithm designed to simulate operations during confusion. The results of this experiment showed there are only a few candidates that could be regarded as real alternatives. Due to the ambiguity of additional qualitative historical, geographical, linguistic, and contextual arguments, we did not produce conclusive arguments in favour of a particular conjecture. Since the possibilities for palaeographical confusion appear to be very limited, the solution to the textual problem concerning IOϒΔAIAN in Acts 2: 9 needs to be sought in a different direction. On the one hand, our research did not exclude the possibility of a very early intentional scribal alteration of an unknown toponym into the transmitted reading. But due to the lack of any manuscript evidence, an attempt to resolve the issue in this direction easily becomes mere speculation. On the other hand, our research did neither provide conclusive reasoning that demonstrated the originality of the transmitted reading IOϒΔAIAN. Several theories have been proposed by respected scholars to advocate the authenticity of IOϒΔAIAN,³¹ but neither of these has yet been convincing. Therefore, either direction needs further investigation to see whether it could solve the issue.
The second aim of this study was to scrutinise whether reverse engineering palaeographical confusion has potential as a method. The experiment confirmed the suitability of the algorithm to test the probability of palaeographical confusion and proposed three cases for further experimentation. To be successful in establishing probable conjectures for these cases, there is a need for a dictionary data set arranged by semantic fields.