The Staicu lexicon in relation to lexicons belonging to the Berynda family: orthography and structure

The first manuscripts with Slavonic-Romanian dictionaries date from the 17 th century and six of these have been preserved to the present day almost intact. They are all based on Pamvo Berynda’s lexicon (1627), to which several word lists were added in some of the manuscripts. These manuscripts have mostly been studied from a lexical point of view. Our study aims at describing the way in which the lexicons were elaborated, at observing the significant orthographic variations and also at becoming an additional tool for establishing the network of filiations from which the Romanian lexicons resulted. Our observations focus on the entries for the first three letters in the above-mentioned lexicons. As all these lexicons were available in elec-tronic format, we identified the propagation of several errors in the Romanian antigraphs, emphasized various aspects related to spelling and compared the additional lists. The data indicates that Romanian manuscripts preserve the orthography of Berynda’s Lexicon and the additional lists to a different degree. These variations suggest the existence of at least one protograph between the source and the Romanian lexicons. Applied to the entire content of the manuscripts, the investigation of these variations could contribute to the identification of the manuscript that is the closest to the initial version.


Introduction
The "Berynda family" designates the Slavonian-Ruthenian lexicon of Pamvo Berynda [= ber], printed for the first time in 1627 in Kiev, as well as a group of Slavonic-Romanian lexicons from the 17 th century which were based on the Ruthenian edition. The following works are included: The Staicu Lexicon (Romanian manuscript 312 bar, mentioned in our study as Lex.Sta.), the lexicon in Romanian manuscript 1348 bar written by Mihai Logofătul (= Lex.1348), the lexicon in Romanian manuscript 3473 bar copied by a certain Mihaiu (= Lex.3473), the lexicon of Petersburg (ОР Q.XVI.5 bnr, mentioned in our study as Lex.Pet.), the lexicon of Mardarie in the Romanian manuscript 450 bar (= Lex.Mard.) and the lexicon of Moscow (F.188, Op.1. nr.1383, rgada, mentioned in our study as Lex.Mosc.). The Romanian manuscripts have benefitted of inconsistent attention throughout the years, mainly because of the fact that two of them are located in Russia. Although the list is far from thoroughgoing, we should mention that Lex.Mard. was edited by Crețu (1900) in a volume that still proves useful, whereas Lex.Mosc. was partly studied by Ciobanu (1914). Lex.Sta. and its author were studied by Strungaru (1973) and Gînsac & Ungureanu (2019). Overall perspectives were provided by Strungaru (1966), Левичкин & Сухачев (2015) and Gînsac & Ungureanu (2018). Since the above-mentioned works focused primarily on the lexical component of the dictionaries, we considered that the filiation should also be approached from the less exploited perspective of orthography. Several manuscripts have become accessible through digitization processes, more information is available about the orthographic norms of the time, and fonts such as those created within the Ponomar project 1 provide us with an opportunity to explain differences that used to be overlooked. Our secondary aim was to catalogue those divergences that cannot be explained solely by the degree of freedom that scholars took when copying from one manuscript to another.
Before the issue of the complete comparative edition we intend to elaborate within the eRomLex project (see scriptadacoromanica.ro), this study used the material for the first three letters: А, Б and В. All the Slavonic words were transcribed taking into account the whole range of orthographic details whenever fidelity was deemed necessary. The translations from the Romanian manuscripts are rendered in Italics.

Structure
The Staicu Lexicon, which can be found on pages 41 v -216 v of the Romanian manuscript 312 bar, represents a collaborative endeavour of several people. It contains around 6,000 proper names, Slavonic common nouns and a smaller percentage of Greek and Hebrew nouns listed according to the Slavonic alphabet. Towards the end, it includes two additional wordlists, considerably shorter than the basic lexicon. These are followed by a series of thematic lexical lists, such as the translation of animal names from Lev, 11. The basic lexicon is an adaptation of the ber, with a few notable differences. Compared to the Ruthenian edition, the proper names are listed immediately after the common ones, not separately. The second notable difference is the fact that after the material in ber, each letter contains at least one page of content that is independent from the Ruthenian lexicon. These additional entries were definitely written by the same hand and most of them are Slavonic, Russian and Ruthenian words. Further on we shall use the convention Staicu 1 for the definitions with an equivalent in ber, whereas the ones copied from an unidentified source shall be considered written by Staicu 2 , without overlooking the fact that this is actually the same person. The content taken from ber by Staicu 1 varies, yet for certain letters it is retrieved almost completely. The words written by Staicu 2 are arranged almost alphabetically, meaning that they appear in groups of 3-5 words listed in alphabetical order, between which some words that do not seem to be in the right place were added. This apparently chaotic organization could be explained in relation to other types of early lexicons. It is possible that our lexicons completed the lexical lists in Berynda with entries taken from one or several Slavonic proto-dictionaries. In order to justify this hypothesis, a variety of information can be considered. This hypothesis is supported by information on the structure of medieval lexicons, which were not always arranged according to the expectations one may have today with regard to a dictionary (Ковтун, 1989, p. 23). A common method was to organize the entries in chapters where words were grouped according to their initial and the second vowel they contained (Карпов, 1877, p. 137); in other cases, the definitions were separated according to thematic categories: months of the year, body parts, proper names, plants, etc. The first principle could explain why in Lex.Sta. we come across successive entries that seem to be organized without any particular logic, such as Амфілоіхїй, Алѵпїй, Акіндѵнъ or Артемъ, Аверкїй, Ареѳа. The Staicu 2 section for В starts with the following sequence: бы́ті, веззлобїе, бденїе, бдит, бок, брод, блистанїе, бдю, бѣсъ, безщадїй, безпечалїе, бос, бꙋбен, бѣлка, бервно, бꙋдꙋ, бꙋдеши, бꙋди, бꙋдет, бых. The first term is a paradigm supplement of the будущїй entry, the last one in ber. If we accept that блистанїе is a paradigm supplement for блѣскъ and reconstruct the historical forms of some words (бѫбьнъ, бьрьвьно), we obtain two almost regular consecutive series: е, е, и, о, о, о, ѣ, (ю), ѣ / е, е, о, ѫ, ѣ, ѣ, ꙋ, ꙋ, ꙋ, ꙋ, ы. This approach can only partially explain the order in Staicu 2 and the other Romanian manuscripts, respectively; however, if this hypothesis proves to be true, it indicates that the entries in Staicu 2 were copied from at least one early Slavonic lexicon. The definitions copied by Staicu 2 do not seem to be grouped thematically, yet some of them appear where one would expect to find them in a thematic dictionary. For instance, at the end of Б, we find the succession бонифантий, борис, бориѳъ. The proper names categories are also placed at the end of the entries for one letter in other Slavonic dictionaries (Карпов, 1877, p. 136; for the classification of Russian lexicons, see Ковтун, 1989, p. 9). A similar pseudo-alphabetical order can be found in manuscript no. 1649 of the Pogodin collection belonging to the Russian National Library, a small lexicon dating from mid-16 th century 2 . We therefore believe that it is possible for the translation from Staicu 2 to have been done after a manuscript which was similar in form and content to the ones that appeared in the Russian area starting with the 15 th century. To provide such an example we shall refer to the entry бориѳъ (translated o iarbă ascuțită la miros [a sharp-smelling type of grass] in Lex.Mosc., Lex.1348, and o iarbă iute la miros [a pungent-smelling type of grass] in Lex.Sta.), relating it to Sergey's lexicon, a manuscript copied around the year 1650 at the Russian Solovetsky Monastery 3 and described by Левичкин & Сухачев (2015, p. 440) as the most comprehensive Russian medieval lexicon. Sergey's lexicon has the structure described above and section Бо includes the entry Бориѳ, зелье о҆стрыя сласᲅи, к мыᲅью̀способно, ꙗ҆́коже ꙋ ҆ наⷭ҇мы́ло. и҆ вчищению силꙋ и҆ мѣетъ, ꙗ҆́коже силиᲅра. І҆ ереⷨ , гⷧ҇в҃ , which translates: "Borith. plantă iute la gust și utilizabilă la spălat, cum e la noi săpunul, și putere de curățare are precum silitra. Ierem 2" [Borith. A sharp-tasting plant that can be used for washing, as we use soap, which cleans as well as salpetre. Jerem 2]. We can thus infer that two of our lexicons selected the first meaning of острый, namely 'ascuțit' [sharp], whereas Staicu opted for the other meaning, which is more appropriate 'iute' [pungent], either because he had the Slavonic text in front of him or because he understood the source of the error. Бориѳ and a few other adjacent entries are missing from Lex.1348, yet the latter still contains around 30 definitions that are not recorded in other manuscripts. This entry cannot be found in either Lex.Mard. or Lex.Pet.
Obviously, the source of the definitions copied by Staicu 2 cannot possibly be Sergey's lexicon, yet the fact that most entries for Б can also be found, although in a different order, in the Russian manuscript, is still relevant. It is possible that they were part of a common body of definitions that migrated from one Slavonic lexicon to the other, in some instances becoming simpler, as in the Romanian manuscripts and in others gaining an encyclopaedic character, as in the Solovetsky dictionary. These additional lists from Staicu do not necessarily originate from the same source. In Lex.Mard. the letter Б ends with the entries бодець, боденець and бодеж. In his lexicon, Staicu 2 draws the line after the last entry in the succession бодець, боденець, боище, then continues to copy entries that can also be found in the other Romanian lexicons. We cannot be certain that Staicu 2 used two different lists, but the line can as well signal the end of a list that was related to Lex.Mard.

The scribes of the Staicu lexicon
Besides the text of the scribe, the lexicon was modified by several readers. Some of these interventions consisted in definitions or observations inserted sporadically, while others produced consistent changes regarding the content of the dictionary. It is difficult to assess how many people contributed to the content of the manuscript, as the handwriting style can be modified by several factors: the writing instrument, the type of intervention -side note, end note or restoration of the manuscript, the language in which it is written.

The Lupașcu hand
The first two pages of letter A, between Азъ and Агрипнїѧ, were written by the same hand. According to a note on page 41 v from May, 1, 1758, the manuscript was owned by a certain priest, Lupașcu. Although the orthography is much neater, one can recognize in it the hand that will add numerous entries to the lexicon. Page 41 v was left half blank and subsequently another hand added some etymological notes. Definitions 1-9, up to Алчба, can also be found in ber. The others, up to Агрипнїѧ, were most probably copied from another manuscript, with little connection to ber. This conclusion is supported by: the approximate alphabetical order, the extremely tenuous selection, if it were to be explained through ber, differences in the translation of the same terms as compared to Staicu -Апⷭлъ:  Lupașcu also left some lexicographic attempts in the form of small groups of words on the pages that precede the lexicon. Many of these have a correspondent in ber, yet some are obviously Ruthenian, without any relation with the Slavonic or Southern-Slavonic languages: кꙋлѧ: glonț [bullet] (31 r ), реестр꙽ или саѧ: rînduială, -ire [order]. The former is attested with its military meaning, according to gsbm (s.v. куля), as early as the 16 th century, while the latter is attested in Russian in 1665 (Vasmer, s.v. реестр). Both examples were borrowed in Russian from Polish, either directly or through a Ruthenian intermediary. mda 2 attests glonț in Costin's Chronicle, but it would be interesting to find out if any occurrences can be detected prior to 1670.
Lupașcu also noted an entry at the end of letter A and several entries at the end of B, of which some are worth mentioning: The capital В with protruding serif, ᲅ with three legs and a serif at the upper end of the left leg, the б, the triangle-shaped ф, the ꙟ, the slightly bent ꙋ, the ц, the к with sinuous strokes, the overwritten р with its leg bent at approximately 45 degrees from the horizontal, all these letters must have been written by the same hand or at least by people with extremely similar writing styles. Some words can be found on the entire Slavonic areal, such as вѣно or вравиѩ. The former exists both in pre-modern Russian (sdi, s.v. вѣно) and in Bulgarian, its use being currently dialectal and obsolete. The latter is a Greek borrowing of a Southern-Slavonic ecclesiastic background, from βραβεῖον, which meant 'prize' (accompanied by a palm-leaves garland, cf. palmares) in Slavonic and pre-modern Russian and then evolved towards 'palm-tree branch' or any convenient substitute for the celebration of Palm Sunday (srs18, s.v. вравие). Other marginal glosses are related exclusively to the North-Eastern Slavic area. In the phrase Воскиⷨ леⷮ пеⷭ: va chelălăi cîinele [the dog will yelp], the verb is an Ukrainian regional version (sum, s.v. скімлити), where the vowels often close at [i]. It could be a Biblical gloss extracted from Ex, 11, 7.

The апфо hand
At the end of letter A we find two entries, one noted by Lupașcu-Архивах: писмо хранилищах: cămări de scrisori [letter rooms]-and one by an unknown author-апфо: acum încă [now yet]-, an entry which is missing from ber and which notes a difficult term from the second book of Kings. In the Hebrew text (bhs, 2Rg, 2, 14) we find the combination 'ap hū, which sept leaves untranslated, rendering it by αφφω. This phrasing was also challenging for the Slavic translators. Some Slavic texts render it as апфо -bis 6 , for instance, or аффо, ost and pan avoid it completely, the Moscow Bible (a re-editing from 1663 of ost) omits it, yet notes афо̀on the side. sept.frank transcribes it as απφὼ, whereas vulg translates it by "etiam nunc", meaning "right now". This апфо hand occasionally notes at the end of a letter definitions that present pleophony and at times complete the definitions provided by Staicu [брашно: bucate, mîncare [food] (Staicu), făină [flour] (апфо)].

Hands of uncertain origin
Other additions that seem to have a Ruthenian origin are билиница and врꙋтка. The first example is actually a rewriting of a Staicu entry, placed one line above. The entry changes the ѣ into и and slightly rephrases the translation. The intervention does not enrich the lexicon, yet the closing ѣ indicates a speaker who was familiar with the Slavic languages spoken north of Moldova: rusn. білый, ucr. бíлий, ceh. bílý. However, south of the Danube, the isogloss of ѣ divided the speakers into two large groups: eastern speakers, who pronounce an Iotated a (bg. бял), and western speakers, who pronounce the sound as e 7 . In the second case, next to the Lex.Sta. translation: țeavie sau duda, someone added ᲂуⷬ лою или рꙋра. Over рьци there seems to be another level, with the letter y. As rură is placed by dlrlc in the northern dialects (Moldova or Bukovina) and the etymon is a Ruthenian one, entered via Polish from the German Rohr = 'țeavă' [pipe] (wsjp, s.v. rura), we can assume that the person who wrote that gloss was a scholar from the Northern part of the country. The lexicons Lex.Mosc., Lex.Pet., Lex.3473 and Lex.1348 translate it identically by țeavie de aramă [brass pipe], with no additional explanation, whereas Lex.Mard. erroneously notes îndemînă [handy], probably under the influence of the previous entry врꙋченїе.
On the first page for letter B, someone wrote the note бащанъ, pe bulgărie zice tată [in Bulgarian it is called father]. It is not clear whether the author is a Bulgarian speaker or, on the contrary, a Russian or Ruthenian speaker who notes a form he is unfamiliar with. The second hypothesis could be supported by the modern Ukrainian баштан 'harbuzărie' [watermelon farm]. The Romanian manuscripts provide many examples of homophone pairs, both in the sections translated after ber, and in the other sections. It is also possible that the person who noted бащанъ wanted to oppose the Bulgarian word to a Ruthenian one he was already familiar with.
In Lex.Pet., at the end of letter В there are three entries that cannot be found in any other dictionary, one of them being веприни: agrișu [gooseberry bush]. Most probably, the scribe was already familiar with the term веприна 'carne de mistreț' [wild boar meat] or 'femela porcului mistreț' [female wild boar], common to several Slavonic languages (ucr., sb. etc. cf. вепрь: porcu, gliganu in Lex.Pet.) and wanted to add a regionalism that was specific to South-Western Ukraine, an adjacent area of the Northern Romanian areal (sum, s.v. 2. веприна).
The most plausible scenario is the one according to which the first pages from the original Staicu lexicon had been lost or were not copied at all, and Lupașcu compensated these gaps subsequently, selectively extracting several definitions from at least two sources and completing the dictionary with a series 6 ᲂубо гдѐ еⷭ҇б҃ г ілиїнь, апфо, pag. 261 r9. 7 Cf. mc. бел, sb-cr. bèo, bije ̯ l, be ̯ l, but dialectally also бѝ`о.
of definitions either at the beginning or at the end of a letter. His additions are generally placed at the beginning, right after the Staicu version, but in some cases, as it happens at the end of the letter Г, entries were added between two of his interventions. In this scenario, Lupașcu is also the one who eliminates two of Staicu's definitions -Азок and Алектор, as these had already been noted and translated in his reconstructed version. Finally, on manuscript 312, at least two other owners left their graphic imprint in a later epoch. One of them, Jean de Talmatzy 8 , has rare interventions with equivalents in Latin and French. On page 50 r , the entry браздна: holdă [fields] is surrounded by a frame containing translations into Greek and Latin, French (champ laboure), Russian (земля обработаная) and Romanian (holdă, pămînt lucrat), to which another writer added a second frame containing the translations holdă sau dealniță.

The Staicu hand
On page 42 r the manuscript continues with the entry Адонай written by an author we identified as Staicu. Our choice indicates that we accept the demonstration of Strungaru (1966), who attributed the paternity of this writing to the scribe from Tîrgoviște. The lexicon owes over 95% of its content to Staicu. After he finished processing the lexical material in ber, Staicu 2 adds 66 entries in which we can distinguish around five groups arranged in an approximately alphabetical order. The Staicu 2 additions are not found in Lex.Pet., Lex.1348 or Lex.Mosc. (in Lex.3473 the letter A was not preserved, whereas in Lex.Mard. it is abruptly interrupted). With only two exceptions, А҆ гаѳоⷩ҇and Анаѯиѡⷭ҇9, all the entries added by Staicu 2 are also found in ber. Of these 64 entries, only five are new, the rest being duplicates of previously entries, some of them repeated even twice. It is almost as if we are looking at the notebook of a student who tries to learn a list of names, translating them repeatedly until he acquires them properly.

Orthography
Generally, the orthography of the Slavonic content in Romanian manuscripts differs from the one in ber. Even if we were to accept the hypothesis of a Romanian protograph which would be extremely faithful to ber, the successive antigraphs gradually took distance from it by applying the Slavonic orthography of Romanian redaction. In Lex.Sta., unlike ber, the combination breathing mark -acute accent ( ҆) is consequently placed on the initial vowel, consonants in adjacent positions coda -onset at the borderline between syllables are relatively consequently separated by Staicu with a payerok ( ꙽ ) or even an yer (Амьма)-as it also happens in ber, owever, only sporadically-, in some instances the consonants are doubled: Lex.Sta. Ам꙽ малиⷦ҇vs ber Амаликъ, and in some cases the Greek letters are phonetically adapted: a. ѵ by и, в or ф: Авѯентїи vs Аѵѯентїи, Афтоноⷨ for Аѵтоноⷨ , which in two other cases Staicu had, however, transcribed with ѵ. As Staicu 1 writes Аѵтоном right after Авѯентїи, we can assume that the scribe was either accustomed to write в instead of ѵ, or to pronounce it as such. b. ѳ by т: Антиногеⷩ҇vs Аѳиноґен꙽ . c. х by ф: А҆ мфилохїй vs Ааⷨ филофїе. d. б by п: А҆ неⷨ бодиⷭть vs Анемподістъ. All the three occurrences in Lex.Sta. appear with б. e. The group гг written нг. Агаѳаггелъ is noted as in Greek only in Lex.Pet., all the other manuscripts we analyzed operated the replacement. f. The groups liquid-yer (рь/лъ/рь/ль) are often noted as we believe they used to be pronounced at the time, namely with the vocalization of the yer before the liquid sound, depending on the etymon: ăr, er, or, ăl, el, ol etc.
In any case, substitutions as the ones emphasized above indicate the possibility that the lexicon was copied after dictation.
ber occasionally uses the spelling with vocalization when a word has a vernacular correspondent (Волна = 'lînă' [wool]), but generally preserves the yer in words belonging to standard Slavonic (Влъненїе = 'vălurire' [waving], Влъ́хвъ = 'vrăjitor' [wizard]). Moreover, it preserves the traditional spelling even in cases where the pronunciation with metathesis and vocalization is obvious; for instance, between the entries вериѓа and весь the following six terms are printed: врᲆста, врᲆтепъ, врътограⷣ , врᲆтоградаръ, врᲆховный and врᲆшба. A significant detail is the fact that almost all these terms have an oversized/ tall version of the back yer. ber does not signal the vocalization using a tall back yer as a rule, yet this tendency can be observed, especially in the combination liquid sound-yer and it could reflect the orthography of the sources from which Pamvo Berynda extracted his definitions. With a single exception, our lexicons do not seem to use a tall yer, although they occasionally seem to distinguish between a vocalized yer and a final yer or between back yer and front yer, especially where the distinction is exemplified with minimal pairs in ber (Lex. ). This succession of six definitions occurs in all instances exactly as in ber between вериѓа and весь. The entries are transcribed in Lex.Pet., Lex.1348, Lex.3473 and Lex.Mosc. also with a yer, either as in ber, or with metathesis (Lex.Mosc., Lex.3473: въртепъ; Lex.3473: върховный), or by vocalizing the yer either selectively (Lex.1348, Lex.Pet.: верховный; Lex.Pet.: вер꙽ тепъ) or in all instances (Lex.Sta., except for the first entryвръшба).
The exception mentioned above is Lex.Mard. Whereas in all the other sources, as in all manuscripts, it is difficult to make a comprehensive inventory of the set of characters, in Lex.Mard. it is obvious that the scribe was familiar with the difference between tall and normal back yers, on the one hand, and the front yers, on the other hand. The tall back yer is also used in the group of words we analysed above from where only the entry врᲆшба̀is omitted, although it is not exclusively used in a strong position (cf. вᲆсаⷣ никь) and occasionally it is conveyed in an orthographic dimension that differs from ber (cf. въ́нь: afară [outside], вᲆ́нь: într-însul [in it/him]).
The Romanian manuscripts incorporate the graphic signs overwritten in ber in different degrees. An interesting case is that of the kamora (), which marked for a long time in Slavonic the stress or the palatalization. In the first half of the 17 th century, Smotrițki grammars propose its introduction in the antistoechon, namely its transformation into a symbol that would emphasize morphological homophony. The new norm was rapidly adopted by the texts printed in Kiev (Kusmaul', 2017). For instance, Berynda uses it to differentiate between Вдовъ ('văduv' [widower], masc. nom. sg.) de Вдѡ̑въ (dat. pl.). The Romanian texts treat this pair in various ways. First of all, Lex.Mosc., Lex.Sta. and Lex.3473 have an extra entry -Вдѡва (fem. sg.). Lex.Pet. and Lex.1348 preserve Вдѡва, but eliminate the entry Вдѡ̑въ, whereas Lex.Mard. preserves the version in ber, including the diacritics. Of all the entries, only Вдѡ̑ва in Lex.Mosc. can be described with no doubt as having an inverted breve 10 ; on Вдѡ҆ въ, the previous entry in the same manuscript, it rather looks like a smooth breathing mark. The author of Lex.1348 also seems to use a smooth breathing mark, Lex.Sta. and Lex.1348 use the acute accent, whereas Lex.Pet. adopts a slightly different orthographic convention, placing the acute accent on о́and a double acute accent or a double grave accent on ѡ̏. We cannot be sure that the Romanian authors had a good understanding of the use of the inverted breve in ber, although attempts were made, at least at the beginning, to transcribe it; in the instances they did not understand it, they chose to transcribe it according to their own norms.
Unlike in the case of ber, some proper names in Staicu are rendered with a suffix: Агаѳаггелъ vs Агаѳангелѡⷭ҇, Акиндѵнъ vs Акиндиноⷩ, Анеќⷮ vs Анектоⷩ.
Proper names are often adapted to Romanian in Lex.Sta., the endings in їа, ꙗ being replaced by ие and їѧ. This observation is valid almost with no exception for the last groups in Staicu 2 : Алонїй vs Алонїе, and so on. Other Romanian lexicons are more conservative from this viewpoint.
Another error that could be explained in relation to ber is the spelling of Авелѡ, where, at one point in the history of the text, a Romanian scribe mistook the front yer for o, and obviously the next step was to render it by ѡ. All the manuscripts contain sections in which they do not follow exactly the order in ber. These cases are not necessarily frequent and could have been caused by negligence.
A series of hilarious errors in Lex.Pet. shed light on the work process of medieval scribes. Each Slavonic entry starts with a cinnabar, yet in some instances the initial is missing and in others it is placed before a Romanian word at the beginning of a line: А҆ фродіта: дꙋ ⷨнезїѡѧ Амилей: лꙋчафърꙋ ⷧȏ r з꙽ дѧй: чела че аⷹ ꙟблаⷮ къларе In the first example, the entry Afrodita had the explanation dumnezioaia milei, Luceafărul [the goddess of mercy, Morning Star]. The scribe added by mistake an A with cinnabar to the next line, turning one entry into two. In the second example, the letter Ꙗ was obviously omitted. The scribe first wrote the entries without initial, then added the initial in cinnabar, thus making mistakes as the ones mentioned above. Strangely enough, he inserted a colon after milei, as if luceafărul would have actually explained the term Amilei, so it is actually possible that the error was not his, but it was instead perpetuated from another manuscript.

Wording
The entry А(в)дий, translated as cărunteațe [hoariness] in Lex.Pet. and Lex.Mosc., illustrate the way in which the first Romanian translators related to the Ruthenian text. In ber, Авдій is not a separate entry, but part of the entry for Авдѣй, the chief cook of King Ahab. Berynda specifies that Авдій is a prophet celebrated on the 19 th of November and provides his description in Подлѣнник 11 : "cărunt, cu barbă nu prea mare, rotundă, părul vîlvoi, haina de culoarea ocrului, verzuie la poale, ținînd în mînă un sul" [greyhaired, with a round beard that was not so long, ruffled hair, with an ochre coat with green edges, holding a roll]. As the proper name is placed at the beginning of the line, the Romanian translator selected the first word of the quotation -Сѣдъ and translated it, not necessarily accurately, by "cărunteațe". As the original version was no longer available, the other lexicons assumed this modification without any changes; Lex.Pet. even omitted the в, as it is probably following a manuscript similar to Lex.Mosc., where в is overwritten in a way that pushes it towards the previous entry. This manner of translating at most one of the Ruthenian synonyms, namely the one the Romanian translator is most familiar with, can be observed in several instances in the Romanian lexicons. Another error can be observed at the entry Аꙁмодеосъ, where the Romanian manuscripts mistake the Ruthenian каꙁитель 'cel care strică' [the one who breaks] for казнитель 'cel care chinuie' [the one who tortures] and consequently mistranslate it: muncitor, căznitor [harasser, one who submits to ordeals]. This strategy alternates with one in which the Ruthenian part is ignored, a (sometimes literal) translation being provided for the Slavonic entry.
In the case of Асиръ, Lex.Sta., Lex.Pet., Lex.Mosc. and Lex.1348 (with the mention that the last manuscript in the series transcribes it Арись) combine the two entries in ber about which the Ruthenian scholar tells us that they are written differently in Hebrew, one with samekh and the other with shin.
There are often cases when the Romanian manuscripts contain additional words when compared to ber. Most of these words are extended paradigms of terms included in ber, various verbal forms, adjectives derived from nouns or nouns derived from adjectives. The Ruthenian printing contains grammatical notes related to aspects such as the voice of the verb, gender, etymological origin, orthographic explanations, and bibliographic references. Some of these notes are also conveyed in the Romanian texts. There are instances in which the Romanian author inserts such notes on his own initiative, without taking them from ber; for instance, in the case of бл҃ го: bine, bun [well, good]. сред., the addition сред (= slav. 'neutral') has no correspondent in ber.

Case study
Let us consider a succession of 12 definitions printed in ber on column 355 12 : The table emphasizes a series of aspects. First of all, Lex.Sta. is certainly a copy of at least the second generation of Romanian lexicons. The author's tendency is to reduce the number of definitions, while increasing the number of explanations. Staicu 1 omits the entry Асᲅролоґъ, but copies by mistake the corresponding explanation at the entry Асᲅролоґіа. Right after this he repeats the mistake, translating А҆ сіѓкриⷮ by crescut [grown] from the following definition and correcting it by crossing it with a line. The second obvious aspect is the fact that Lex.Mard. is the result of a different redaction. The third aspect beyond doubt is that Lex.Mosc. cannot possibly be a copy after any of the manuscripts we analysed. The possibility of a filiation from Lex.1348 is significantly diminished by examples provided elsewhere.
Before an in-depth analysis of the entire lexicographic material is conducted, I shall only mention that the scheme proposed by Gînsac & Ungureanu (2018) seem to reflect the reality better than the filiations previously suggested in literature. These two researchers from Iași have advanced the theory of the existence of at least one intermediary between ber and the group Lex.Pet., Lex.Sta. and Lex.3473. The data presented above suggests the profile of a manuscript that is very close to Lex.Mosc.. Besides the fact that Lex.Mosc. has some entries that are missing from the other manuscripts and the translations of the other lexicographic witnesses are convergent with the ones in Lex.Mosc., the spelling of the analysed entries is identical to the one in ber, whose letter variants are also often imitated. Unlike Lex.Sta., Lex.Pet. and Lex.1348, Lex.Mosc. does not use breathing marks on the initial; it does not replace ᲄ with ᲅ or ѵ with и/і and does not insert a payerok between neighbouring consonants, as Lex.Sta. and Lex.1348 (and occasionally Lex.Pet.), neither does it overwrite the final letter, or replace а by ѧ as Lex.1348. Furthermore, it seems to be the only manuscript that occasionally reproduces ґ with an upwards serif from the Ruthenian printing 14 , see Асѵґкріⷮ , and also Архісѵнагѡ́ґ, where both versions occur, the difference between them being obvious.
However, two of the details contained in the table oppose the identification of the assumed intermediary as Lex.Mosc. The first detail is related to the correction crescut in Lex.Sta. In Lex.Mosc. the entry Асѵґкріⷮ closes the page, whereas Аᲄᲅалъ is at the top of the next page, so Staicu could not possibly overlook it and go to the next definition. Nevertheless, I have to admit that one can as well imagine an explanation according to which the turning of the page causes the person copying the text to make a mistake. Moreover, it is not clear whether Аᲄᲅалиа in ber was transcribed with и or with н by the Lex.Mosc. copyist. I am almost sure that he opted for the wrong version. Fortunately, things are as clear as possible in the case of the other three manuscripts. Lex.Pet. and Lex.1348 erroneously transcribe with н and this aspect is confirmed by the presence of the payerok after л. In this case, the entry in Lex.Sta., accurately and univocally transcribed with ї, is revealed as important, as it motivates us to search for another protograph. Nonetheless, we can exclude neither the possibility of a correction by chance, nor that of an informed correction. Apart from his negligence, which can be frustrating at times, Staicu seems to have a good command of the Slavonic language and proves to be a competent scholar.
We should also mention the case of the proper name Ахиафскъ, transcribed erroneously by all the Romanian manuscripts we analysed: А҆ хиафиⷦ: fărmecător [charming] (Lex.Sta.), А҆хиафїкъ: fărămecătoriul (Lex.Pet.), Ахиафкъ: fărmecătoriul (Lex.1348) and Ахиафькъ: fărmecătoriul (Lex.Mosc.). This confusion is actually generated by Berynda himself, who wanted to enrich the entry Achsaph: Veneficus 14 We have to specify that this type of Ge was used in the Slavonic books printed in the Polish-Lithuanian State Union to indicate g in words of Greek origin and was not part of the orthographic norms of the Slavonic spoken in the Romanian Principalities. Consequently, its occurrence on the Romanian territory can only indicate the contact with a Ruthenian original. For further details, see Andreev et al. (2015, p. 11). In ber this character has the height of a capital letter, its height being actually the criterion according to which we identified it in Lex.Mosc. Unfortunately, the font used in our article does not reflect this difference.
[etc] from steph 15 with its equivalent in the Slavonic canon. In the Slavonic versions of Ios, 1, 11 we find къ цр҃ ю а҆ хиафъскꙋ (ost) or къ царю а҆ хифскꙋ (bis 152) 16 , meaning "on the king of Achshaph", but Berynda interprets the adjectival suffix used for geographical localization -ск as part of the name (cf. sept.frank: αχιαφ), and the Romanian copyists are thus challenged by a group of consonants they are unable to decipher. With regard to the aspects discussed above, the order of the events is relatively simple. Lex.Mosc. sees in ber a ь instead of c, Lex.1348 overwrites the yer, whereas Lex.Pet. and Lex.Sta. vocalize it at ї/и.

Conclusions
The Berynda lexicon represented a remarkable intellectual achievement and, in the actual case of the Romanian manuscript bar 312, it was the source of a text used for over a century by several users who improved it permanently. The significant number of Romanian lexicons issued in a relatively short period of time illustrates the notable impact it produced. They could be the result of a synthesis between the Ruthenian version and a Slavonic dictionary of small dimensions. The structure of this hypothetical dictionary differs from the one of the current dictionaries and, as many other similar proto-dictionaries belonging to the Slavonic cultural space, was probably the result of a collection of glosses. We are actually facing a double challenge: on the one hand, to describe this hypothetical proto-dictionary as accurately as possible and on the other hand to retrace the filiations network of the Romanian dictionaries belonging to the Berynda family. With regard to the first challenge, we should conduct a more thorough investigation of the Slavonic manuscripts to be found in Romania or abroad, hoping that we have the chance of discovering a manuscript that would explain as well as possible the lexical lists included in the Romanian manuscripts. The second challenge makes us wonder not only how related our lexicons are to each other, but also how close they are to Berynda's edition. Besides the translation of the terms, other approaches are also worth exploring, which refer to the texts layout, the structure and organization of the lists, the interpretation, correction and perpetuation of errors. Apart from Lex.Mard., which has a slightly different writing and is also the oldest, Lex.Sta. is estimated by most researchers to be from an earlier date than all the other lexicons. However, the other manuscripts contain numerous elements that indicate an older edition and are closer to ber in many respects. Previous studies place Lex.Mosc. between the 17 th and the 18 th century, yet its orthography, inventory of terms and the layout of lists composing it indicate an edition that was very close to the first Romanian translation. Although we should remain cautious and keep the number of hypothetical manuscripts as low as possible, the significant variations between the lists, translations and errors renders the hypothesis of only one protograph quite unsatisfactory for the attempt to provide a valid explanation regarding the filiations of the 17 th century Romanian manuscripts.