Blotting Out the Name

Scribal Methods of Erasing the Tetragrammaton in Medieval Hebrew Bible Manuscripts, Part 1

In: Textus
Nehemia Gordon Bar-Ilan University Israel

Search for other papers by Nehemia Gordon in
Current site
Google Scholar
Open Access


Early rabbinic interpretation of Deut 12:4 prohibited erasure of the Tetragrammaton, which required Jewish scribes to employ creative methods to resolve extraneous instances of the divine name. This may be foreshadowed in the writing of divine appellations in Paleo-Hebrew in the Dead Sea Scrolls. Medieval Jewish scribes employed diverse methods to deal with errors involving the Tetragrammaton. In codices this involved marking God’s name with dots, rectangles (also used in liturgical scrolls), lines, and supralinear circelli. Some scribes indicated the Tetragrammaton’s erasure by leaving it unpointed and recording a correction (usually Adonai) in the margin, without any additional notation. A special procedure involved a nonstandard usage of the Qere notation. All of these methods were performed in accordance with rabbinic strictures. Part 2 of this study will consider exceptions to the rule and the special case of liturgical Torah scrolls.


Early rabbinic interpretation of Deut 12:4 prohibited erasure of the Tetragrammaton, which required Jewish scribes to employ creative methods to resolve extraneous instances of the divine name. This may be foreshadowed in the writing of divine appellations in Paleo-Hebrew in the Dead Sea Scrolls. Medieval Jewish scribes employed diverse methods to deal with errors involving the Tetragrammaton. In codices this involved marking God’s name with dots, rectangles (also used in liturgical scrolls), lines, and supralinear circelli. Some scribes indicated the Tetragrammaton’s erasure by leaving it unpointed and recording a correction (usually Adonai) in the margin, without any additional notation. A special procedure involved a nonstandard usage of the Qere notation. All of these methods were performed in accordance with rabbinic strictures. Part 2 of this study will consider exceptions to the rule and the special case of liturgical Torah scrolls.

This study1 will examine methods used to erase the Tetragrammaton in Hebrew Bible manuscripts. Rabbinic strictures prevented scribes from physically erasing the divine name, requiring them to employ notations to mark it as extraneous without defacing it or to remove it from the writing surface (animal skin or paper) without damaging it.2 Some of these scribal methods were already used in the Second Temple period (as will be seen from parallels in the Dead Sea Scrolls) and have continued in use until modern times.

1 Halakic Prohibition against Erasing the Tetragrammaton

Deuteronomy 12:3 contains an injunction to destroy places of worship dedicated to the names of Canaanite deities. Verse 4 then warns the Israelites not to worship God in the same manner as the Canaanites did. Sifre Deut reimagines this warning as a proscription against erasing even a single letter of God’s name:

Rabbi Ishmael says, “From where ⟨do we know⟩ that he who erases a single letter of the name violates a prohibition? As it says, ‘And you shall destroy their name … you shall not do so to the Lord your God’ ” (Deut 12:3–4).3

Rabbi Ishmael (early second century) seems to be seeking a biblical justification for a preexisting cultural taboo against erasing the divine name.4 To be sure, metaphorically erasing a name had the sense of a curse in biblical times, as we see in the Psalms: “in the generation following, let their name be blotted out.”5 The biblical curse uses the root ‮מח״ה‬‎, which literally means to “wash off” (or “wipe off”) the ink with liquid, possibly using a sponge (see § 5.12, below).6 Rabbi Ishmael uses the verb ‮מח״ק‬‎, which can mean both to wash off and to scrape off (abrade), the latter also referred to more specifically by the verb ‮גר״ר‬‎ or ‮גר״ד‬‎ (see § 5.11, below).7 Rabbi Ishmael’s midrashic interpretation extends the prohibition against cursing God by erasing his name into the realm of correcting scribal errors.8

According to the minor tractate Sefer Torah (ca. third century), the prohibition against erasing single letters of the Tetragrammaton goes into effect only once the first two letters have been written.9 This prohibition does not extend to ink that drips on the letters of the Tetragrammaton, “for ⟨the scribe⟩ never had any intention ⟨concerning the dripped ink⟩ other than to correct.”10 Intention is a key part of the prohibition, as can be gleaned from the minor tractate Soferim (ca. mid-eighth century), which incorporates large sections from the earlier Sefer Torah:

⟨The scribe who⟩ was supposed to write the ⟨divine⟩ name, ⟨mistakenly⟩ wrote “Judah” with intention, but had not yet placed the dalet; he erases it and writes the ⟨divine⟩ name. Rabbi Judah says, “He traces over it with the pen and sanctifies it …”11

Writing the Tetragrammaton was thus understood as a sacred act. Even though the first three letters of “Judah” are identical to the beginning of the Tetragrammaton, the scribe had to have the specific intent to record the divine name for these letters to be considered sacred. This even applies to a biblical context where the divine name was supposed to have been written, but the scribe formed the requisite intent to write “Judah.”12 On the other hand, writing those same letters in the wrong place with the intention of recording the Tetragrammaton, resulted in a sacred word that it was forbidden to physically erase in whole or in part.13

Jonathan P. Siegel argued that the Tetragrammaton and other divine appellations (e.g., ‮אדוני‬‎, ‮אלוהינו‬‎, ‮צבאות‬‎) were written in Paleo-Hebrew script in some of the Dead Sea Scrolls, specifically to ensure that they would not be erased.14 He observed that thirty errors in 11QPsa were physically erased using abrasion, that is, scratching the ink off the writing surface with a sharp instrument. In contrast, the two errors involving the Tetragrammaton were marked with dots to indicate that they were superfluous (see § 5.1.2, below). According to Siegel, this special treatment reflects an early adoption of the prohibition against erasing the divine name. However, as Siegel himself pointed out, there is one instance where the Tetragrammaton was physically erased through abrasion in 1QIsaa (see § 5.11.1, below).

2 Methods of Erasure in the Ancient World

Scribes in the ancient world employed a variety of methods to erase texts written in ink. Egyptian scribes as early as the Old Kingdom included a sponge in their standard set of equipment for washing off ink. This was a method suited for papyrus, not animal skins or medieval paper.15 Removal of ink from animal skins through abrasion with a sharp instrument, found extensively in the Dead Sea Scrolls, may have also been one of the functions of the scribe’s razor mentioned in Jer 36:23 (see § 5.11.1, below).16 Another method employed in the ancient world, especially with large blocks of text, was excision, that is, cutting out the text and replacing it with a patch containing the correction (see § 5.9, below).17

Scribes and readers did not always resort to physically removing the ink. Greek scribes marked erroneous words with dots over, or next to, a word or letters, an invention credited to Aristarchus (third century BCE). Erasure dots and strokes are also found extensively in the Dead Sea Scrolls. These erasure symbols are placed above, below, and adjacent to words and letters, and even sometimes completely surrounding a word (see § 5.1, below).18

Another method found in the Greek world and the Dead Sea Scrolls is to mark the beginning and end of a word with parenthesis marks.19 A variation of this method connects the parenthesis marks with horizontal lines, resulting in the word being surrounding by a rectangle or ellipse (see § 5.3, below).20 Egyptian scribes in the New Kingdom marked text with a strikethrough, that is, by passing a horizontal or oblique line through the text (see § 5.7, below). This method was also employed in Aramaic papyri from Elephantine, in Ptolemaic Egypt, and in the Dead Sea Scrolls.21

3 Methods of Erasure in Early Rabbinic Literature

The prohibition against erasing a single letter of the Tetragrammaton produced a practical dilemma for Jewish scribes when errors occurred. The minor tractate Soferim distinguishes between physically “erasing” (‮מחק‬‎) the Tetragrammaton and “invalidating” (‮מעכב‬‎) it.22 In contrast to the prohibition against physically erasing the Tetragrammaton already cited, Soferim provides a permissible way to deal with an error caused by dittography:

He who writes two divine names, he maintains the first one and invalidates (‮מעכב‬‎) the second.23

The verb ‮מעכב‬‎ may have the broad sense of invalidating the Tetragrammaton by marking it in various ways to indicate that it does not belong in the text, although this is by no means certain. Rabbi Chaim Hirschensohn (1857–1935) understood the term as a linguistic variant of ‮מעקף‬‎, meaning to surround the erroneous word with an ellipse or parenthesis marks (see § 5.3, below).24 In contrast, Rabbi Samuel Halevi Segal (1625–1681) interpreted ‮מעכב‬‎ as a reference to excision; this involved cutting out the Tetragrammaton, thereby leaving a hole in the writing surface that could be replaced with a patch containing the correction (see § 5.9, below).25

Excision was already practiced in talmudic times, although it was referred to by the verb ‮קד״ר‬‎. The Babylonian Talmud contains a discussion about saving gilyonin (apparently a disparaging epithet for εὐαγγέλιον) and other Jewish Christian books (‮ספרי מינין‬‎) from a fire on the Sabbath.26 The Sages might have been content to let them burn had it not been for the divine appellations (‮אזכרות‬‎) that they evidently contained.27 In this context, Rabbi Jose the Galilean (early second century) declared, “I would excise (‮קודר‬‎) the divine appellations (‮האזכרות‬‎) in them on a weekday, put them in a genizah, and burn the rest.”28 Similarly, a talmudic tradition accuses King Ahaziah of Judah of excising (‮קודר‬‎) the divine appellations (‮אזכרות‬‎) in the Torah and replacing them with patches containing idolatrous names.29 Although these and similar talmudic sources do not describe excision to fix scribal errors, they reflect an awareness of this method and specifically in the context of dealing with the Tetragrammaton.

The meaning of ‮מעכב‬‎ in the tractate of Soferim is complicated by the matter of textual variants. The lost Babylonian version of Sop. 5:1, cited in the eleventh- or twelfth-century work Hilkhot Sefer Torah, replaces ‮מעכב‬‎ with ‮מנקד/מנקיד‬‎, i.e., to mark with dots (see § 5.1, below).30 Furthermore, this section of Soferim incorporates the earlier minor tractate Sefer Torah, which instead of ‮מעכב‬‎ uses the (hiphil) verb ‮מחטיב‬‎. In one of his responsa, Maimonides apparently read the piel form ‮מחטב‬‎ in Soferim itself.31

Like the meaning of ‮מעכב‬‎, that of ‮מחטב/מחטיב‬‎ is unclear.32 Haran understood it as referring to abrasion (see § 5.11, below).33 Ben Yehuda’s dictionary interpreted it as referring to a strikethrough (see § 5.7, below).34 Peretz suggested it referred to skiving, that is, exfoliating the layer of skin containing the error, also referred to by the verb ‮קל״ף‬‎, literally, “peeling” (see § 5.8, below).35

As the above interpretations and variants illustrate, Soferim, with its parallel in Sefer Torah, might generally describe invalidating the Tetragrammaton with some notation, or more specifically, skiving, excision, abrasion, surrounding it with a rectangle, or marking it with a strikethrough or dots. The aforementioned variants may reflect different methods used in different periods or regions to remove the Tetragrammaton or mark it as extraneous; the diverse interpretations of the various verbs certainly do.

One further method to resolve an error involving the Tetragrammaton, mentioned in early rabbinic literature, is to remove the entire sheet or membrane (‮יריעה‬‎) containing the mistake (see § 5.10, below).36 All of these methods were employed by Jewish scribes in different periods to invalidate or remove erroneous instances of the Tetragrammaton. In addition, medieval Jewish scribes marked the Tetragrammaton as extraneous in other ways, such as with a supralinear circellus or a nonstandard usage of the qere notation. Some even corrected the Tetragrammaton in the margin without any notation. In some manuscripts, the method of dealing with the Tetragrammaton does not differ from the overall scribal approach to errors although, in most, the sacred nature of the Tetragrammaton required a special solution.37

Of the above methods, abrasion, excision, marking an error with dots, surrounding the error with an enclosure, and strikethroughs are carry-overs from the ancient world. Removing an entire sheet due to an error involving a divine name may be a Jewish innovation of the Tannaitic period. Pending further investigation, it can tentatively be suggested that skiving is a Jewish innovation of the gaonic period.38 Leaving a word unpointed to indicate its erasure, as well as adapting notations from the masorah parva (supralinear circellus and text-correcting qere) to resolve scribal errors are presumably internal medieval Jewish developments.

4 Date and Origin of Manuscripts Considered in this Study

The fields of Hebrew paleography and codicology identify Jewish manuscripts as originating in five primary geocultural regions based on their script and the physical characteristics of the codex.

  1. The Mizraḥi codicological and paleographical type comes from Egypt, the Levant, eastern Asia Minor, Iraq, Persia, and even as far away as Central Asia. A subtype of the Mizraḥi type developed into a distinctive script in Yemen beginning around the thirteenth century.39

  2. The Sefardic type may have originated in northwest Africa and spread to the Iberian Peninsula, Sicily, and southern France (Provence; Languedoc).40

  3. The Ashkenazic type comes from central Europe beginning in the twelfth century and includes northern France, England, and later Poland.41

  4. The Italian type seems to have originated in Puglia in southern Italy in the ninth century and later spread to the rest of the Italian peninsula.42

  5. The Byzantine type is characteristic of Greece, western Asia Minor, the lands around the Black Sea, and the Balkans.43

The manuscripts considered in this study come from the major geocultural regions that produced Hebrew manuscripts. In addition, some important examples come from Torah scrolls produced by the isolated Jewish community in Kaifeng, China. The distinctive Chinese Jewish calligraphic style developed out of the Mizraḥi Jewish script used in Persia.44 In some circumstances, manuscripts are identified and dated based on the Hebrew language tradition reflected in their vocalization. In this study, this scenario pertains to those from the Cairo Genizah with Babylonian pointing, which could have been produced anywhere from Persia to Yemen and even Egypt.45 All dates should be considered approximate, except when based on a colophon or radiocarbon dating.

The “medieval” manuscripts considered at the core of this study span the centuries from the ninth century until the advent of printing in the fifteenth century. Evidence is also considered from later periods, because Jewish scribes continued medieval scribal traditions long past the end of the European Middle Ages, especially in the case of liturgical scrolls. Some of the scribal practices discussed here (e.g., dots) have early parallels in the Dead Sea Scrolls and some (e.g., skiving) continue until modern times. With a few exceptions noted below, the various methods are not distinctive enough to reveal the date or origin of a given manuscript, since they were used by scribes in different geocultural regions and periods.

A significant question when discussing scribal corrections and erasures is whether they were performed by the original scribe or by a second hand. In most instances it has not been possible to determine this.46 Nevertheless, in some cases, different colors of ink, different shapes of letters, or different usage of symbols reveal the work of a second (or third) scribe. In other instances, the placement of a correction immediately after the error without breaking the flow of the text, or before vowel points were inserted, reveals the work of the original scribe.47 In the case of codices, the scribe who inserted the vowel points and accents may have differed from the scribe who wrote the letters, and hence this second hand participated in the original writing of the manuscript.

5 Methods of Erasing the Tetragrammaton

5.1 Erasure Dots

5.1.1 Erasure Dots in Early Rabbinic Literature and the Masorah

Perhaps the oldest and most enduring method of invalidating words without physically erasing them is to mark them with scribal erasure dots. The MT contains fifteen places where words are permanently marked with extraordinary points, originally indicating their erasure.48 ʾAbot de Rabbi Nathan imagines that Ezra the Scribe inserted these dots to mark doubtful words that might not belong in the text:

Why are all these letters marked with dots? Ezra said thus: “If Elijah comes and says, ‘Why did you write ⟨these letters⟩?’ I will say to him, ‘I already put dots on them.’ If he says to ⟨me⟩, ‘You wrote it beautifully,’ I will then remove the dots over them.”49

These instances include fourteen places where the dots are written above the letters and one in which they are written both above and below.50 None of these extraordinary points marks the Tetragrammaton.

5.1.2 Continuity of Use from Antiquity through the Middle Ages

An early parallel to the medieval use of erasure dots can already be found in the Dead Sea Scrolls, which contain numerous words, including the Tetragrammaton, that are marked with erasure dots. For example, 11QPsa XXI, 2 (Ps 138:1) contains an extraneous instance of the Tetragrammaton written in Paleo-Hebrew and marked with erasure dots (see Fig. 1).51 Other errors in this scroll were physically erased using abrasion, sometimes after being initially marked with dots.52 As already mentioned (see § 1, above), Jonathan P. Siegel argued that the scribe deliberately avoided using abrasion due to the sacredness of the divine name.

Figure 1

11QPsa XXI, 2 (Ps 138:1)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

image b-314640; courtesy of the leon levy dead sea scrolls digital library; israel antiquities authority, photo: shai halevi

The use of erasure dots in 11QPsa is remarkably similar to an example in the sixteenth-century Yemenite manuscript BL Or. 1473 in Jer 14:8 (fol. 25r, see Fig. 2).53 Whereas 11QPsa meticulously recorded a single dot above and a single dot below each letter, the Yemenite scribe of BL Or. 1473 was less precise, using five dots (or dashes) above the Tetragrammaton and six below it.54 The error may have been caused by the phrase ‮מקוה ישראל‬‎, which appears both here and in Jer 17:13, but with the Tetragrammaton in the latter. The effect of invalidating the Tetragrammaton, i.e., marking it as a mistake that should not be read, is nevertheless clearly identifiable in both texts, separated by 1,500 years and their differing geocultural regions. Like 11QPsa, BL Or. 1473 normally erased words through abrasion, but made do with erasure dots to avoid physical erasure of the divine name.55

Figure 2

BL Or. 1473, Jer 14:8 (fol. 25r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

© british library board

5.1.3 Placement of the Dots

The placement of the erasure dots varied without changing their meaning, perhaps based on the preference of the scribe or in some instances the idiosyncrasies of a given scribal tradition. For example, 1QIsaa III, 25 (Isa 3:18, see Fig. 3), dated 201–93 BCE, placed one dot below each letter of the Tetragrammaton as well as one dot before the yod and another after the final he. The divine appellation Adonai is written above the line as a correction. The previous line (III, 24 [Isa 3:17]) contains the opposite correction, having Adonai marked with erasure dots and the Tetragrammaton written above the line.56 In these instances, the placement of the dots below the letters may have been a function of the need to insert the supralinear corrections. In a third example, 1QIsaa III, 20 (Isa 3:15), Adonai is written as a supralinear correction, but the Tetragrammaton is not marked with erasure dots. In this instance, the correction was presumably intended to complement the Tetragrammaton rather than replace it; ‮אדני יהוה‬‎ is also found in the MT (cf. § 5.6.2, below).57

Figure 3

1QIsaa III, 20 Isa 3:15 (top); III, 24 Isa 3:17 (middle); III, 25 Isa 3:18 (bottom)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © israel museum, jerusalem by ardon bar hama

In 1QIsaa III, 24 (Isa 3:17), the scribe who changed Adonai to the Tetragrammaton may not have been correcting an “error” as much as he was bringing the text in line with a particular textual tradition. RNL Evr. II B 9 (Mizraḥi, ca. 1000) has the Tetragrammaton in Isa 3:17 (fol. 3v) with the masorah parva note “⟨according to the tradition of⟩ Neharde⟨ʿa⟩ it is written Adonai, and it is a matter of dispute” (see Fig. 4).58 Thus the two readings in 1QIsaa—the original and the correction—may reflect two different textual traditions, which later were also held by the great rabbinic schools of Nehardeʿa and Sura, respectively.59 Many erasures considered below may be a matter of one textual tradition favored by the original scribe and another favored by the corrector, or one favored by the scribe who wrote the letters and the other by the scribe who wrote the vowel points.

Figure 4

RNL Evr. II B 9, Isa 3:17 (fol. 3v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

Concerning the placement of the dots, the aforementioned Hilkhot Sefer Torah (eleventh or twelfth century) explains, “The dots that they spoke about ⟨in Soferim⟩ are above ⟨the word⟩, not below.”60 This rule was usually, although not always, followed in medieval manuscripts. Hence, a ca. tenth-century manuscript in Mizraḥi square script, RNL Evr. II B 119, contains a dot above each of the four letters of the Tetragrammaton in Isa 21:6 (fol. 9v), with Adonai written as a correction above the line (see Fig. 5). The scribe of this manuscript physically erased other words besides the Tetragrammaton, sometimes also using a strikethrough, but limited himself to erasure dots when it came to the divine name.61

Figure 5

RNL Evr. II B 119, Isa 21:6 (fol. 9v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.1.4 Number of Erasure Dots

The supralinear erasure dots do not always correspond to the individual letters of the Tetragrammaton. RNL Evr. II B 63, a tenth- or eleventh-century Mizraḥi manuscript, marks a dittography of the Tetragrammaton in 2 Kgs 19:21 (fol. 148v, see Fig. 6) with a series of six short lines above the letters; the Tetragrammaton was left unpointed. The same page of the manuscript contains the pointed Tetragrammaton marked with eight supralinear dots in 2 Kgs 19:23 and Adonai as a correction in the margin (see Fig. 7).62 Leaving a word unpointed was in itself a way to indicate its erasure.63 However, in 2 Kgs 19:23 and in many of the examples considered below, the error was evidently not discovered until after the Tetragrammaton was already pointed. Most errors in this manuscript were erased through abrasion or marked with a strikethrough.64 In contrast, errors involving the Tetragrammaton were generally marked with supralinear dots or short lines.65

Figure 6

RNL Evr. II B 63, 2 Kgs 19:21 (fol. 148v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg
Figure 7

RNL Evr. II B 63, 2 Kgs 19:23 (fol. 148v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

A remarkable scribal error involving the Tetragrammaton appears in a ca. ninth-century Cairo Genizah fragment with early Old Babylonian pointing, RNL Antonin B 812, containing the alternating Hebrew text of Deuteronomy and the targum.66 In the targum of Deut 17:19 (fol. 1r, see Fig. 8), the scribe accidentally wrote two words, the second of which was the Tetragrammaton, instead of the Aramaic ‮חיווהי‬‎ (“his life”). The scribe then apparently abraded most of the first word and marked the Tetragrammaton with twelve supralinear apostrophe-like dashes.67 The scribe must have noticed his error as he was writing, because immediately after the error he corrected himself on the same line.

Figure 8

RNL Antonin B 812, Deut 17:19 (fol. 1r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.1.5 “Mid-letter” Erasure Dots

Sometimes the erasure dots above the Tetragrammaton were combined with mid-letter dots, i.e., dots that are about vertically centered inside or adjacent to the letters. For example, RNL Evr. II B 10, written in Mizraḥi script in the tenth century, has a dittography of the phrase ‮אשר יהוה‬‎ in Deut 4:1 (fol. 180v; see Fig. 9). The second instance of the Tetragrammaton was marked with mid-letter dots as well as with a second dot above each letter. In contrast, the bottom portion of the adjacent preposition ‮אשר‬‎ was physically erased through abrasion, the top portion of the letters being left to avoid creating a closed parashah in the middle of the verse.68

Figure 9

RNL Evr. II B 10, Deut 4:1 (fol. 180v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.1.6 Mid-letter Circelli

A noteworthy variation of invalidating the Tetragrammaton consists of mid-letter circelli, combining the form of the circellus from the masoretic note with the concept of erasure dots. This practice appears in a Mizraḥi manuscript, RNL Evr. II B 52, from the second half of the eleventh century. For example, in Gen 20:4 (see Fig. 10) each of the four letters of the unpointed Tetragrammaton is marked with a mid-letter circellus, with the pointed correction Adonai in the left margin.69

Figure 10

RNL Evr. II B 52, Gen 20:4 (fol. 15r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.1.7 Surrounding Error with Dots

Sometimes the scribe appears to surround the Tetragrammaton with erasure dots, as in the ca. twelfth-century Ox. Heb. e 72/4 written in Mizraḥi square script. In Gen 18:27 (fol. 8r; see Fig. 11), the scribe repurposed the shewa under the yod and the holem over the first he as erasure dots.70 The qamets under the vav, consisting of a horizontal line and a dot, also contributes to encircling the Tetragrammaton with no fewer than thirteen erasure dots. Adonai is written with vowel points (and possibly a faded zaqef qaṭon) as a correction in the left margin of the previous line.71

Figure 11

Ox. Heb. e 72/4, Gen 18:27 (fol. 8r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

courtesy of the bodleian libraries, the university of oxford

5.2 Lines and Bars

5.2.1 Scope and Usage

Another method of invalidating the Tetragrammaton was to mark it with a solid continuous line above, below, or above and below. This method of erasure is found beginning in the tenth century and as late as the nineteenth century in virtually every geocultural region where Hebrew manuscripts were produced.72 For example, RNL Evr. I Bibl. 85, dated to the tenth or eleventh century and written in Mizraḥi square script, contains an erroneous instance of the Tetragrammaton in Deut 7:12 (fol. 129r, see Fig. 12). The Tetragrammaton was left unpointed and was invalidated with a supralinear line. The error was caused by the familiar idiom “as ‮יהוה‬‎ swore” being written instead of “as he swore” in this verse.73 Erasures of other words in this manuscript were done using abrasion, but abrasion was avoided when it came to the Tetragrammaton.74

Figure 12

RNL Evr. I Bibl. 85, Deut 7:12 (fol. 129r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.2.2 Sefardic Erasure “Bar”

Manuscripts written in the Sefardic script in the fourteenth and fifteenth centuries have lines with circles at either end forming a distinctive “bar.” For example, the fifteenth-century Ox. Pococke 348, written in Sefardic square script, invalidates the Tetragrammaton in Isa 6:8 (fol. 3v) with a supralinear bar and the correction Adonai in the right margin (see Fig. 13). The same error and correction appear six lines lower on the page in Isa 6:11. In both instances, the Tetragrammaton was already pointed before the error was detected and Adonai is also pointed in the margin.

Figure 13

Ox. Pococke 348, Isa 6:8, 11 (fol. 3v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

courtesy of the bodleian libraries, the university of oxford

5.2.3 Position of Lines and Bars

The position of the line or bar seems not to have had any significance, and some manuscripts contain variety in this matter. The aforementioned Ox. Pococke 348 has the Tetragrammaton marked with bars both above and below, with Adonai in the left margin, in Isa 11:11 (fol. 5v, see Fig. 14).75 In this case, the error was caught before the Tetragrammaton was pointed. Most corrections in this manuscript were done using abrasion, but the sacredness of the Tetragrammaton required a different approach.76

Figure 14

Ox. Pococke 348, Isa 11:11 (fol. 5v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

courtesy of the bodleian libraries, the university of oxford

5.2.4 Stylized Semi-cursive Sefardic Bar

A stylized variation of the Sefardic erasure bar appears in the fourteenth-century BL Or. 2415 written in the Sefardic semi-cursive script. In Num 14:17 (fol. 70v), the scribe invalidated the Tetragrammaton with a supralinear bar before inserting the vowel points, and then immediately wrote the correction Adonai after it on the same line (see Fig. 15).77 Erasures of other words in this manuscript were accomplished using abrasion.78

Figure 15

BL Or. 2415, Num 14:17 (fol. 70v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

© british library board

5.3 Rectangles around the Tetragrammaton

5.3.1 Rectangles in Codices

Another common way to invalidate an erroneous instance of the Tetragrammaton was to surround it with ink in the form of an ellipse or rectangle. This method is a continuation of the inverted nuns in the MT and has a parallel in the parenthesis marks in the Dead Sea Scrolls, which when connected form an enclosure.79 Surrounding erroneous instances of the Tetragrammaton with such an enclosure was performed in different periods and geocultural regions. For example, the fourteenth- or fifteenth-century JTS Lutzki 442, written in Sefardic script, has a rectangle made of four bars around the Tetragrammaton in Ps 33:11 (fol. 3r, see Fig. 16). The rest of the line was erased and rewritten, but the erroneous Tetragrammaton had to be left intact due to rabbinic strictures. The same manuscript has bars above and below an erroneous instance of the Tetragrammaton written instead of Adonai in Ps 38:10 (fol. 1v, see Fig. 17). The three words following the Tetragrammaton were physically erased and rewritten with spacing that was narrower than the surrounding text, in order to squeeze Adonai onto the same line. Physically erasing the Tetragrammaton would have been much simpler but was forbidden.

Figure 16

JTS Lutzki 442, Ps 33:11 (fol. 3r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

courtesy of the library of the jewish theological seminary
Figure 17

JTS Lutzki 442, Ps 38:10 (fol. 1v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

courtesy of the library of the jewish theological seminary

5.3.2 Rectangles in Torah Scrolls

Peretz found the rectangle, in combination with leaving the word unpointed, as the only method used to invalidate the Tetragrammaton in pointed Ashkenazic codices.80 It turns out this practice was also applied to unpointed Ashkenazic Torah scrolls. For example, the fourteenth-century Vat. ebr. 2 contains two instances of this practice in Exod 34:9 (see Fig. 18). The scribe may have originally written ‮בעיניך יהוה ילך נא יהוה‬‎ with the Tetragrammaton instead of Adonai. Realizing the error, the first scribe, or another scribe, invalidated the two erroneous instances of the Tetragrammaton with rectangles and worked around them. The corrector probably abraded the words ‮ילך נא‬‎ and wrote ‮אדני ילך‬‎ over the erasure. What appears in the final version is: ‮בעיניך {יהוה} אדני ילך {יהוה} נא אדני‬‎.81

Figure 18

Vat. ebr. 2, Exod 34:9 (sheet 32)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

© 2020 biblioteca apostolica vaticana

5.4 Circelli

Another method for invalidating an erroneous instance of the Tetragrammaton was to mark it with a circellus or small circle. This method was undoubtedly an outgrowth of the masorah parva, which uses circelli to mark words of special interest, referring the reader to masoretic notes in the margin.

The ca. 1300 Parma, de Rossi 668, written in Italian script, contains the unpointed Tetragrammaton at Gen 18:27 (fol. 14r, see Fig. 19), marked with one circellus above and another below. Adonai is written in the right margin with pointing and its own circellus. This marginal correction has the opposite pointing scheme of ketiv-qere in this manuscript, which records the pointing in the ketiv and leaves the qere unpointed.82


Figure 19

Parma, de Rossi 668, Gen 18:27 (fol. 14r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

complesso monumentale della pilotta, biblioteca palatina, ms. parm. 1849, by concession of ministry of national heritage and culture

This scribe did not always add a second circellus below the line. In Exod 15:17 (fol. 63v, see Fig. 20), he marked the unpointed Tetragrammaton with a single supralinear circellus, with the pointed Adonai and a circellus in the margin.83 The scribe of de Rossi 668 usually marked erroneous words with mid-letter dots in each of the letters (in addition to a circellus and marginal correction),84 so it is possible that he considered marking the Tetragrammaton with mid-letter dots to be disrespectful.85


Figure 20

Parma, de Rossi 668, Exod 15:17 (fol. 63v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

complesso monumentale della pilotta, biblioteca palatina, ms. parm. 1849, by concession of ministry of national heritage and culture

5.5 Text-Correcting Qere

5.5.1 Qere Used to Correct Errors

Another variation using the circellus involves a nonstandard usage of the qere notation. The standard qere marks the way words were meant to be read, or in some cases, interpreted.86 As Michael V. Fox explains, “The qerayin were not corrections or words to be incorporated in the body of the text by the next copyist, as is the case with interlinear and marginal variants in the Qumran scrolls …”87 In contrast, “text-correcting qere” actually intends for the ketiv to be replaced in the body of the text the next time the manuscript is copied.

Examples of text-correcting qere can be found in RNL Evr. I Bibl. 102, containing the five scrolls and hafṭarot, apparently written in Ashkenazic square script, possibly in France, at the end of the thirteenth or beginning of the fourteenth century.88 In Lam 3:31 (fol. 15v, see Fig. 21), the scribe who added the vowel points and accents, marked an erroneous instance of the Tetragrammaton with a supralinear circellus, which is itself not unusual. He then wrote ‮אדני ק̇‬‎, “it is read Adonai” (without vowel points) in the margin.89 The Tetragrammaton has been read as Adonai in the Hebrew Bible since at least talmudic times. However, it is not accompanied by a marginal qere note.90 This makes the note “it is read Adonai” in RNL Evr. I Bibl. 102 redundant. The Leningrad Codex has Adonai rather than the Tetragrammaton in Lam 3:31, so this marginal note was meant to correct a scribal error (or stamp out a textual variant). Hence, this note is not really about pronunciation and must mean something like: “it should be read as if it were written Adonai.” The note deals with how the reader was meant to visually assess the Tetragrammaton and how the next copyist was meant to reproduce the body of the text, not about how one was meant to read it aloud, since in any event that would be Adonai.

Figure 21

RNL Evr. I Bibl. 102, Lam 3:31 (fol. 15v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.5.2 Qere Used Alongside Strikethrough

It is worth noting that this manuscript contains numerous instances of text-correcting qere, unrelated to the Tetragrammaton. The word being corrected is generally marked not only with a circellus, but also with a strikethrough.91 For example, the manuscript originally had the erroneous reading ‮כעדר הקצבות‬‎ in Cant 6:6 (fol. 5v). The word ‮הקצבות‬‎ is marked with a circellus and crossed out, with the vocalized correction ‮הרחלים ק̇‬‎ in the margin (see Fig. 22).

Figure 22

RNL Evr. I Bibl. 102, Cant 6:6 (fol. 5v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.5.3 Qere to Correct Scribal Errors

When the word in the body of the text was not crossed out, it is difficult to distinguish a text-correcting qere from a standard qere. For example, Isa 42:17 (fol. 26r) has ‮יסגו‬‎ in the body of the text marked with a circellus, but not crossed out, with ‮נסגו ק̇‬‎ in the margin (see Fig. 23). Presumably this is a correction of an error, but it could also be an alternative ketiv-qere not found in the Leningrad Codex or Aleppo Codex.92 In contrast, Adonai qere in Lam 3:31 could not be an alternative ketiv-qere because, with or without the marginal note, the Tetragrammaton was read as Adonai. The scribe refrained from crossing out the Tetragrammaton when employing the text-correcting qere notation due to the sacredness of the divine name.93

Figure 23

RNL Evr. I Bibl. 102, Isa 42:17 (fol. 26r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

Obad 1:1 (fol. 32r, see Fig. 24) contains a complex text-correcting qere related to the Tetragrammaton. The base text has an error of metathesis with the phrase ‮יהוה אדני‬‎ instead of ‮אדני יהוה‬‎ found in the Leningrad Codex. The scribe marked the unpointed Tetragrammaton with a tailed circle and wrote in the margin ‮אֲדֹנַי ק̇‬‎, “it is read Adonai.” As in Lam 3:31, the meaning must be “it should be read as if it were written Adonai.” However, in this case the scribe included vowel points in the marginal note, but the wrong vowel points. Adonai in the margin is written with a patakh, identifying it as the nonsacred “my lords.”94

To make matters worse, the word Adonai in the base text (the first word on fol. 32v, see Fig. 25) is unpointed and marked with a very faint circle indicating its having been erased, but with no notation to indicate that it should be replaced with the Tetragrammaton. A note such as ‮אלהים ק̇‬‎ or ‮יהוה ק̇‬‎ would have fixed the error. As it stands now, a reader following the instructions of this manuscript would read (aloud) ʾădōnay instead of ʾădōnāy ĕlōhîm.95

Figure 24

RNL Evr. I Bibl. 102, Obad 1:1 (fol. 32r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg
Figure 25

RNL Evr. I Bibl. 102, Obad 1:1 (fol. 32v), contrast enhanced

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.5.4 Text-Correcting Ketiv

A variation of this method can be found in RNL Evr. II B 9 written in Mizraḥi square script, dated to ca. 1000. In Isa 7:14 (fol. 5v; see Fig. 26), the scribe mistakenly wrote the Tetragrammaton instead of Adonai. To remedy the situation, he marked the Tetragrammaton with a circellus and wrote ‮אדני כת̇‬‎, “⟨it should be⟩ written Adonai” in the margin.96 Most corrections in this manuscript were carried out using abrasion.97

Figure 26

RNL Evr. II B 9, Isa 7:14 (fol. 5v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

photograph © national library of russia, st. petersburg

5.6 Unpointed Tetragrammaton without Notation and Marginal Correction

5.6.1 Adonai instead of the Tetragrammaton

An erasure method that could be used only in a pointed codex is to leave the Tetragrammaton unpointed without additional notation and, where necessary, to insert a correction in the margin. Kolodni found this method of erasure in fourteen Italian manuscripts.98 This method produces ambiguity when employed in manuscripts that do not point every word. For example, a ca. fifteenth-century Yemenite manuscript, BL Or. 2366, contains a correction of the Tetragrammaton in Gen 18:27 (fol. 11v) without any specific notation; the divine name was left unpointed, and Adonai was added above the line (see Fig. 27). Other than the lack of vowel points, there is nothing to indicate that the Tetragrammaton has been invalidated.99 Most erasures in this manuscript were accomplished through abrasion. However, BL Or. 2366 occasionally leaves the Tetragrammaton unpointed without implying an erasure (e.g., Exod 19:9 [fol. 93v]). Thus, this correction is somewhat ambiguous and could theoretically be read as ‮אדני יהוה‬‎ (read aloud as ʾădōnāy ĕlōhîm).

Figure 27

BL Or. 2366, Gen 18:27 (fol. 11v)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

© british library board

5.6.2 Adonai in addition to the Tetragrammaton

A nearly identical correction can be found in the ca. tenth-century Cairo Genizah fragment with late Old Babylonian pointing, JTS Lutzki 507 in Isa 7:7 (fol. 4r, see Fig. 28).100 However, this older manuscript contains only partial pointing, which means that many words were left unpointed, without the lack of pointing indicating their erasure.101 In this case, the correction apparently aims at ‮אדני יהוה‬‎ (read aloud as ʾădōnāy ĕlōhîm), found in other masoretic manuscripts and reflected in the targum on the following line. Although nearly identical to the previous example, this is not an erasure at all.102

Figure 28

JTS Lutzki 507, Isa 7:7 (fol. 4r)

Citation: Textus 29, 1 (2020) ; 10.1163/2589255X-02901008

courtesy of the library of the jewish theological seminary

Part 1 of this study has dealt with the standard methods of erasure performed (primarily) in codices, in accordance with rabbinic halakah. Part 2 of this study will consider some exceptions to the rule in which rabbinic strictures were violated by erasing the divine name with strikethroughs and abrasion. Special techniques employed in Torah scrolls will also be considered, namely, skiving, excision, and removing sheets. Washing off, which was not performed in the corpus examined, will also be briefly discussed.


This article consists of a chapter of my doctoral thesis conducted at the Bible Department at Bar-Ilan University, carried out under the supervision of Prof. Yosef Ofer. An early version of this article was presented at the SBL/EABS Conference in Helsinki in 2018. I am indebted to Prof. Jordan Penkower for his invaluable insights and guidance, as well as for various examples that he called to my attention. I want to thank David Marcus, Viktor Golinets, Mordechai Veintrob, Estara Arrant, Jen Taylor Friedman, Dawn Erickson-Irion, and James Walker for their input. I especially want to thank Troy Bryant for his tireless efforts. The following abbreviations are used for manuscripts:


Paris, Alliance Israélite Universelle


London, British Library


Cambridge, University Library


New York, Jewish Theological Seminary


Washington, D.C., Museum of the Bible


St. Petersburg, Russian National Library


Oxford, Bodleian Libraries


Vatican, Biblioteca Apostolica


Wolfenbüttel, Herzog August Bibliothek.


On the different writing surfaces (and confusion surrounding the terminology), see Malachi Beit-Arié, Hebrew Codicology: Historical and Comparative Typology of Medieval Hebrew Codices Based on the Documentation of the Extant Dated Manuscripts until 1540 Using a Quantitative Approach, preprint internet version 0.9 (April 2018), 159–200 (Heb.).


Sifre Deut., Reʾeh § 61; t. Mak. 5:9; Sep. Torah 5:9; Sop. 5:9.


A related taboo can be found in the Second Temple period Megillat Taʿanit, which records: “On the third of Tishri, the ‮אדכרתא‬‎ was abolished in contracts.” The scholion elaborates, “for the ⟨Seleucid⟩ Greek kingdom decreed that the name of Heaven must not be pronounced with their mouth, so when the Hasmoneans prevailed and defeated them, they decreed that the name of Heaven should be mentioned even in contracts. They used to write thus: ‘In the year such and such of John, High Priest to the Most-high God (‮כהן גדול לאל עליון‬‎).’ When the sages heard about this, they said, ‘Tomorrow this ⟨person⟩ will pay his debt, and the contract will be tossed into the rubbish.’ So ⟨the Sages⟩ invalidated them, and they made it a day of celebration” (b. Roš. Haš. 18b; for other versions, see Vered Noam, Megillat Taʿanit: Versions, Interpretations, History [Jerusalem: Yad Ben-Zvi, 2003], 235 [Heb.]). Heinrich Graetz, Geschichte der Juden, Vol. 3.2 (Leipzig: Leiner, 1888), 572, suggested concerning the formula ‮כהן גדול לאל עליון‬‎ (adapted from Gen 14:18, 22), “oder vielleicht gar mit dem Tetragrammaton ‮לה׳‬‎, wofür ‮אדכרתא‬‎ spricht”; cf. Marcus Jastrow, A Dictionary of the Targumim, Talmud Babli, Talmud Yerushalmi and Midrashic Literature, 2 vols. (London: Luzac, 1903), 1.16, “‮אדכרה‬‎”; Noam, Megillat Taʿanit, 235–238.


Ps 109:13 [KJV]; cf. Deut 9:14; 25:6; 29:19; 2 Kgs 14:27; Ps 9:6. For a parallel in the ancient world, see Robert Kriech Ritner, The Mechanics of Ancient Egyptian Magical Practice (Chicago: Oriental Institute, 2008), 137.


Num 5:23; 2 Kgs 21:13; Prov 30:20; see Menahem Haran, “Book Scrolls in Israel in Pre-Exilic Times,” JJS 33 (1982): 161–173 (esp. 168–169); June Ashton, Scribal Habits in the Ancient Near East: c. 3000 BCE to the Emergence of the Codex (Sydney: Mandelbaum, 2008), 156–158.


Jastrow, Dictionary 2.763, “‮מחק‬‎”; Haran, “Book-Scrolls,” 170 n. 20; Emanuel Tov, Scribal Practices and Approaches Reflected in the Texts Found in the Judean Desert (Leiden: Brill, 2004), 201 n. 246, 230; cf. Maimonides, Commentary on m. Kelim 13:2. In Biblical Hebrew, the verb ‮מח״ק‬‎ appears in Judg 5:26 in the sense of crushing, as evidenced by the linguistic variant ‮מח״ץ‬‎ later in the verse, and perhaps this evolved into the sense of scraping, cf. Shelomo Morag, Lectures on the Phonology of Hebrew, Phonetics and Phonology and the Morphology of Biblical Hebrew (Jerusalem: Magnes, 2009), 32 (Heb.). ‮גר״ד‬‎ already appears in the sense of scraping in Job 2:8, while a related meaning for ‮גר״ר‬‎ may already be anticipated in 1 Kgs 7:9.


Exceptions to the prohibition are four instances of the Tetragrammaton in Judges (17:2–3, 13; 18:6) that Sep. Torah 4:5 deemed to be profane (contra Sop. 4:12).


“He who writes … YH from ⟨the⟩ four letters, they may not be erased” (Sep. Torah 4:2). This seems to indicate that if the scribe wrote only the yod of the Tetragrammaton before realizing his error, it would be permissible to physically erase it. The parallel passages in y. Meg. 1:9, 71d and b. Šebu. 35b are discussed in Jonathan P. Siegel, “The Employment of Palaeo-Hebrew Characters for the Divine Names at Qumran in the Light of Tannaitic Sources,” HUCA 42 (1971): 159–172 (esp. 166–169); idem, “The Scribes of Qumran” (PhD diss., Brandeis University, 1972), 36–41. Concerning the date of the minor tractate Sefer Torah, see Hermann Leberecht Strack and Günter Stemberger, Introduction to the Talmud and Midrash, trans. and ed. Markus Bockmuehl (Minneapolis: Fortress, 1996), 232: “the basic form probably dates from the third century, albeit subsequently revised.”


Sep. Torah 5:9; cf. Joseph Peretz, “Ways of Correcting the Biblical Text in Ashkenazic Manuscripts from the Middle Ages,” Talele Orot 15 (2009): 19–34 (esp. 29 n. 44 [Heb.]).


Sop. 5:4; according to Strack and Stemberger, Introduction to the Talmud, 228, “in its present form ⟨Soferim⟩ cannot be dated prior to the middle of the eighth century, even if earlier forms must be assumed.”


A remarkable example of the importance of intent when writing divine appellations is evidenced by a ca. 1400 Ashkenazic Torah scroll, MOTB SCR.1181. In Gen 24:12, the scribe was supposed to write the nonsacred ‮אדני‬‎ (ʾădōnî) but must have formed the requisite intent to write the graphically identical ‮אדני‬‎ (ʾădōnāy). To remedy this (assuming this reconstruction is correct), he excised the wrongly intended divine ‮אדני‬‎ (ʾădōnāy), replacing it with a blank patch, and wrote the correctly intended nonsacred ‮אדני‬‎ (ʾădōnî) above the line; cf. Solomon Ganzfried, Qeset Hasofer: Part 1, 2nd ed. (Uzhhorod: Jäger, 1871), 18a (§ 10.15), 22b; 23b (§ 11.15; 12.3).


Ganzfried, Qeset Hasofer, 18b–21a (§ 11.1–7). The prohibition against erasing the Tetragrammaton was not limited to scriptural texts. Sep. Torah 5:12 and Sop. 5:14 prohibit a person who wrote the Tetragrammaton in impermanent ink on his flesh from washing it off.


Siegel, “The Employment of Palaeo-Hebrew,” 162, 169; idem, “The Scribes of Qumran,” 32, 41–42; cf. Tov, Scribal Practices, 245.


Ashton, Scribal Habits, 157; cf. David Diringer, The Hand-Produced Book (Watford: Hutchinson’s, 1953), 550; Eric G. Turner, Greek Manuscripts of the Ancient World (Princeton: University Press, 1971), 8. For a possible example involving animal skin in 4QLevc 4, 1 (4:23), see Tov, Scribal Practices, 230.


So Frederic G. Kenyon, Books and Readers in Ancient Greece and Rome, 2nd ed. (Oxford: Clarendon, 1951), 45; contra Haran, “Book-Scrolls,” 170; see also Ashton, Scribal Habits, 157–158; Tov, Scribal Practices, 229–230. The medieval scribe’s razor appears in an illumination in the Ashkenazic maḥzor Ox. Laud. Or. 321 (ca. 1270–1280), fol. 43v in the context of erasing the name of Amalek (Parashat Zakhor). For a parallel involving abrasion in Samaritan manuscripts, see Alan D. Crown, Samaritan Scribes and Manuscripts (Tübingen: Mohr Siebeck, 2001), 72.


Ashton, Scribal Habits, 158; Haran, “Book-Scrolls,” 163; for a parallel in Samaritan manuscripts, see Crown, Samaritan Scribes, 72–73.


Ashton, Scribal Habits, 159, 175–176; Tov, Scribal Practices, 188–198; Kenyon, Books and Readers, 73; Beit-Arié, Hebrew Codicology, 375; Turner, Greek Manuscripts, 18. For a parallel involving erasure dots in Samaritan manuscripts, see Crown, Samaritan Scribes, 71.


This procedure is known in Greek as περιγράφειν; the marks are known as σίγμα and ἀντίσιγμα (LSJ, 162, “ἀντίσιγμα”; 1371, “περιγράφω”; Turner, Greek Manuscripts, 18; Tov, Scribal Practices, 201). This may be the purpose of the inverted nuns in the MT at Num 10:35–36.


Tov, Scribal Practices, 201–203. See Ashton, Scribal Habits, 162 for a possible parallel in Egyptian practice.


Ashton, Scribal Habits, 159–160; Tov, Scribal Practices, 198–201. This procedure is known in Greek as διαγράφειν (LSJ, 392, “διαγράφω”; Turner, Greek Manuscripts, 18).


Cf. Peretz, “Ways of Correcting,” 31–32; Jastrow, Dictionary, 2.1077, “‮עכב‬‎” translates the verb as “to invalidate,” the interpretation tentatively adopted in this study. The root ‮עכ״ב‬‎ often has the meaning of “to delay” (ibid.), a sense that may allude to ʾAbot R. Nat. A:34; B:37 where erasure dots are said to postpone the resolution of a doubtful word until the arrival of Elijah (see § 5.1.1, below). Regardless of the precise meaning of ‮עכ״ב‬‎, it must somehow differ from physical erasure (‮מח״ק‬‎) that involves defacing the letters of the Tetragrammaton, since the latter is explicitly forbidden in both Sep. Torah 5:9 and Sop. 5:9.


Sop. 5:1. In this context, “names” refers both to God’s personal name, i.e., the Tetragrammaton (including its abbreviated form ‮יה‬‎), and other divine appellations, e.g., ‮אל, אלהיך, אלהים, אלהיכם, אהיה אשר אהיה, אדני, שדי‬‎ (and according to some opinions ‮צבאות‬‎); see Sep. Torah 4:1–2; Sop. 4:1–2; b. Šebu. 35a; cf. m. Soṭah 7:6; Siegel, “The Employment of Palaeo-Hebrew,” 166–170.


Chaim Hirschensohn, Malki Ba-Kodesh (St. Louis: Moinester, 1919), 2.247: “This is the meaning of the tractate of Soferim ⟨when it says⟩ ‘he maintains the first one and surrounds (‮ומעקף‬‎) the second,’ meaning, he surrounds it with a circle or with two parenthesis marks to show that it is extraneous, as we are accustomed to doing in talmudic texts.”


Samuel Halevi Segal, Naḥalat Shivʿah: Part 2 (Fürth: Shneur, 1692), 23a–b (responsum 37). The early rabbis were divided on the issue of writing a correction on a patch (see Sep. Torah 2:13; Sop. 2:17; Tov, Scribal Practices, 123–124). Some later sources prescribe doing so as the recommended course of action (see § 5.8–9, below). However, Ganzfried, Qeset Hasofer, 22a (§ 11.13) ruled it forbidden to write divine appellations on a patch for fear that it might eventually fall off.


The censor removed the following from the standard edition of b. Šabb. 116a: “Rabbi Meir used to call it, ʾāwen gīllāyôn (wicked blank page) and Rabbi Joḥanan used to call it, ʿāwôn gīllāyôn (iniquitous blank page)” (Vat. ebr. 108, fol. 57r; Vat. ebr. 487/9, fol. 82v; Munich, Bayerische Staatsbibliothek, Cod. hebr. 95, fol. 27v); cf. Jastrow, Dictionary, 1.248–249, “‮גליון‬‎.”


According to J. Fürst, “Askara Oder Schem Hammephorasch, Das Ausdrücklich Ausgesprochene Tetragrammaton,” ZDMG 36 (1882): 410–416 ‮אזכרה‬‎ originally referred to the Tetragrammaton alone, but by the third century CE sometimes took on the meaning of any divine appellation, including ‮אלהים‬‎. Even so, David ben Solomon Ibn Abi Zimra (ca. 1479–1573), Shu”t Haradba”z (Warsaw: Iuzefov, 1882), 1.33 (responsum 102) could still maintain that “no ⟨other divine appellation⟩ is called ‮אזכרה‬‎ other than the Tetragrammaton (‮שם ההויה‬‎).”


B. Šabb. 116a; cf. t. Šabb. 13:5.


B. Sanh. 102b; cf. Moʿed Qaṭ. 26a; Sanh. 103b.


Only the first two chapters of the Babylonian version survive. See Michael Higger, Seven Minor Treatises (New York: Bloch, 1930), x–xiv (Heb.); Strack and Stemberger, Introduction to the Talmud, 228. The citations of Sop. 5:1 survive in JTS R2063 (ENA 222), 7–8, published in Elkan Nathan Adler, Ginze Miṣrayim: Hilkhot Sefer Torah (Oxford: Hart, 1897), 13–48; idem, “An Eleventh Century Introduction to the Hebrew Bible,” JQR 9 (1897): 669–716. The first citation reads: “He who writes a divine appellation (‮אזכרה‬‎) twice, he puts dots (‮מנקד‬‎) on the first one” (ibid., 695 [JTS R2063, 7]). The second citation reads: “He wrote the divine appellation (‮האזכרה‬‎) twice, he puts dots (‮מנקיד‬‎) on the first one and maintains the second” (ibid., 697 [JTS R2063, 8]). On the possible identity of the author of this work, see Lipa Ginat, Rabbi Joseph Rosh Hasseder and His Manuscript of the Toshba and Halacha (from the Geniza) (Petach Tikva: Ginat, 2004), 35–37 (Heb.) and the literature cited there.


Maimonides’s responsum citing Sop. 5:3 (or possibly Sep. Torah 5:3) is in turn cited by Joseph Caro, Bet Yosef, Yoreh Deʿah 276 in the name of the fifteenth-century Rabbi Joseph Iskandarani: ‮וכן הורה הרמב״ם בתשובה על יריעה שנתקלקל בה השם שגונזין אותה יריעה או מחטב את השם כדאמר ר׳ יהודה מחטבו וכותב יהודה אחר עכ״ל‬‎. Maimonides’s responsum is also cited, nearly verbatim, in the name of the same Iskandarani, by Ibn Abi Zimra, Shu”t Haradba”z, 2.2 (responsum 596).


Cf. Academy of the Hebrew Language, Maʾagarim: The Historical Dictionary of the Hebrew Language, “‮חטב ‭2‬‬‎,” (Heb.); Jastrow, Dictionary, 1.447, “‮חטב‬‎ I.” The only examples of the meaning “erase” cited by the dictionaries are from the passages under discussion in Sefer Torah and Soferim. Rabbi Benjamin Aaron Slonik, Masʾat Binyamin (Kraków: Meisels, 1633), 97b, remarks on Sop. 5:1, “ ‘and he invalidates (‮ומעכב‬‎) the second.’ Some read ‮מחטב‬‎, and both of them mean excision (‮קדירה‬‎).” The variant readings of ‮חט״ב‬‎ in Sop. 5:1 (bis), 3, 6 are also found in Elijah Gaon of Vilna (1720–1797), Masekhet Avot … with Minor Tractates (Shkloŭ: Aryeh ben Menaḥem, 1804), 53b.


Haran, “Book Scrolls,” 170 n. 20.


Eliezer Ben Yehuda, Moses Hirsh, Naphtali Herz Tur-Sinai, Complete Dictionary of Ancient and Modern Hebrew, 17 vols. (New York: Yoseloff, 1960), 3.1506–1507, “‮החטיב‬‎” (Heb.).


Peretz, “Ways of Correcting,” 31–32 n. 50. This is also apparently the interpretation of the verb assumed by Rabbi Chaim Benveniste (1603–1673), Sheyare Kenesset Hagedolah (Istanbul: Ashkenazi, 1729), 117a (Yoreh Deʿah 276 n. 25) who wrote, “the meaning of ‮מחטבו‬‎ in ⟨the⟩ context ⟨of Sef. Torah 5:3⟩ is to raise it from off the gevil, as the scribes were accustomed to doing.” Skiving can leave gashes in the skin that resemble the gashes made by an adze hewing wood—the literal meaning of ‮חט״ב‬‎—which supports Peretz’s explanation. Cf. Yaakov Shmuel Spiegel, Chapters in the History of the Jewish Book: Scholars and their Annotations, 2nd ed. (Ramat Gan: Bar-Ilan University, 2005), 70–71 n. 132, 76 n. 159 (Heb.); Tov, Scribal Practices, 201 n. 246. Cf. also the variants listed in Michael Higger, Masekhet Soferim (New York: Deve Rabanan, 1937), 152–155; idem, Seven Minor Treatises, 33.


Sep. Torah 5:5; Sop. 5:5; b. Menaḥ. 30b.


Prior treatments of the subject include Siegel, “The Employment of Palaeo-Hebrew,” 159–172; idem, “The Scribes of Qumran,” 29–41; Tov, Scribal Practices, 198, 201 n. 246, 218–221, 245; Ashton, Scribal Habits, 173–176; Peretz, “Ways of Correcting,” 29–32; Orlit Kolodni, “The History of Transmission and Methods of Corrections of the Consonantal Text of the Torah in Medieval Italian Codices and Tiqqunei Soferim,” Italia 22 (2013): 33–44 (Heb.).


The earliest usage of the verb ‮קלף‬‎ to describe removal of the Tetragrammaton may be Lam. Rab., Petiḥta 24 (ed. Buber, 24); cf. Jastrow, Dictionary 2.1381, “‮קלף‬‎.” However, the context there is planing a layer of wood(?) from a weapon inscribed with the Tetragrammaton, not skiving a layer of skin from a scroll or codex.


Beit-Arié, Hebrew Codicology, 61; idem, Specimens of Medieval Hebrew Scripts, Volume 1: Oriental and Yemenite Scripts (Jerusalem: Israel Academy of Sciences and Humanities, 1987), iii–vii; Edna Engel, “Styles of Hebrew Script in the Tenth–Eleventh Century in Light of the Dated and Datable Genizah Documents,” Teuda 15 (1999): 365–398 (Heb.).


Beit-Arié, Hebrew Codicology, 61–64; Malachi Beit-Arié and Edna Engel, Specimens of Medieval Hebrew Scripts, Volume 2: Sefardic Script (Jerusalem: Israel Academy of Sciences and Humanities, 2002), v–xvii.


Edna Engel and Malachi Beit-Arié, Specimens of Medieval Hebrew Scripts, Volume 3: Ashkenazic Script (Jerusalem: Israel Academy of Sciences and Humanities, 2017), vii–xliv; Edna Engel, “Between France and Germany: Gothic Characteristics in Ashkenazi Script” in Manuscrits hébreux et arabes: Mélanges en l’ honneur de Colette Sirat, ed. Nicholas De Lange and Judith Olszowy-Schlanger (Turnhout: Brepols, 2014), 197–219.


Beit-Arié, Hebrew Codicology, 64–65.


Beit-Arié, Hebrew Codicology, 65. It should be noted that manuscripts were often produced by Jews who migrated from one region to another, bringing their script with them (ibid., 48–50). Furthermore, codicological characteristics can sometimes be found in several geocultural regions (ibid., 52).


Salomo A. Birnbaum, The Hebrew Scripts, 2 vols. (Leiden: Brill, 1971; London: Palaeographia, 1954), 1.243; 2.215–216.


Israel Yeivin, The Hebrew Language Tradition as Reflected in Babylonian Vocalization (Jerusalem: Academy of the Hebrew Language, 1985), 26–27 (Heb.).


On the methodological problems, see Beit-Arié, Hebrew Codicology, 374.


In this study, the term vowel points is used to refer to the Hebrew pointing system (‮ניקוד‬‎), even though it includes elements that are not vowels (e.g., silent shewa) and matres lectiones are included in letters.


Emanuel Tov, Textual Criticism of the Hebrew Bible, 3rd ed. (Minneapolis: Fortress, 2012), 51–52, 202–204; idem, Textual Criticism of the Hebrew Bible, Qumran, Septuagint (Leiden: Brill, 2015), 156–157. On this method of erasure in nonbiblical medieval manuscripts, see Beit-Arié, Hebrew Codicology, 375.


ʾAbot R. Nat. A:33; B:37; see further, Saul Lieberman, Hellenism in Jewish Palestine (New York: Jewish Theological Seminary, 1962), 43–46.


A different use of supralinear dots was employed by Babylonian Masoretes to indicate disputes on such matters as open and closed parashot, word division, ketiv-qere, full and deficient orthography, and vav copulative; see Yosef Ofer, The Babylonian Masora of the Pentateuch: Its Principles and Methods (Jerusalem: Academy of the Hebrew Language, 2001), 240–253 (Heb.).


Each letter of the Tetragrammaton is also marked above and below with erasure dots in 11QPsa XVI, 7 (Ps 145:1); 4QIsad 6–10, 10 (Isa 49:4). The first three letters of the Tetragrammaton were mistakenly written and then marked with erasure dots above and below each letter in 4QIsad 6–10, 7 (Isa 49:1). See further Tov, Scribal Practices, 193; Al Wolters, “The Tetragrammaton in the Psalms Scrolls,” Textus 18 (1995): 87–99; Patrick W. Skehan and Eugene Ulrich, “58. 4QIsad,” in Eugene Ulrich et al., Qumran Cave 4.X: The Prophets, DJD XV (Oxford: Clarendon, 1997), 77.


Siegel, “The Employment of Palaeo-Hebrew,” 162; James A. Sanders, The Psalms Scroll of Qumrân Cave 11 (11QPsa), DJD IV (Oxford: Clarendon, 1965), 13–14; Florentino García Martínez et al., Qumran Cave 11.II: (11Q2–18, 11Q20–31), DJD XXIII (Oxford: Clarendon, 1998), 32, 35.


BHS notes here “nonn Mss 𝔊 + ‮יהוה‬‎” suggesting this was not an error but rather a variant textual tradition (see further Benjamin Kennicott, Vetus Testamentum Hebraicum cum Variis Lectionibus, 2 vols. [Oxford: Clarendon, 1776–1780], 2.110). This may be a discrepancy between the textual traditions of the scribe who wrote the letters and the Masorete who added the vowel points and accents; cf. Joseph Peretz, “Unpointed Biblical Passages in Masoretic Codices of the Middle Ages,” Studies in Bible and Exegesis 7 (2015): 183–209, (esp. 189–192 [Heb.]).


The scribes of the Dead Sea Scrolls varied in their use of erasure dots. For example, one dot could mark a single letter or an entire word, and a word could be marked with a large number of dots unrelated to its number of letters. For a detailed treatment of the subject, see Tov, Scribal Practices, 188–198.


In another instance (fol. 146v [Ezek 35:13]), BL Or. 1473 invalidated the unpointed Tetragrammaton along with a second word with continuous horizontal lines above and below (see § 5.2, below). Examples of abrasion of other words include Jer 7:15 (fol. 14r); Jer 32:4 (fol. 54v).


Cf. Tov, Scribal Practices, 188, 194, 215; Shemaryahu Talmon, “Aspects of the Textual Transmission of the Bible in Light of Qumran Manuscripts,” Textus 4 (1964): 95–132 (esp. 118).


Cf. Patrick W. Skehan, “The Divine Name at Qumran, in the Masada Scroll, and in the Septuagint,” BIOSCS 13 (1980): 14–44 (esp. 40–41 n. 14).


Folio 3v. Cf. the masoretic list of differences between Westerners (Land of Israel) and Easterners (Babylonia) in the Leningrad Codex, fol. 468v for Lam 5:21. Similarly, ‮לאדני‬‎ in Dan 9:9 has a masorah parva note in the Leningrad Codex (fol. 445r), “for the Easterners, written ‮ליהוה‬‎.” See also Israel Yeivin, Introduction to the Tiberian Masorah, trans. and ed. Ernest John Revel (Atlanta: Scholars Press, 1980), 99.


Whether these two traditions developed independently in both the Dead Sea Scrolls and Babylonia or derived from a common source is beyond the scope of this study.


JTS R2063, 7–8; Adler, “An Eleventh Century Introduction,” 695, 697.


Corrections in RNL Evr. II B 119 may have been made by multiple hands that used abrasion (e.g., Jer 34:5), strikethrough (e.g., Jer 52:32), and abrasion combined with strikethrough (e.g., Jer 35:5).


BHS: “nonn Mss ‮יהוה‬‎” (see further Kennicott, Vetus Testamentum, 1.674). The thirteenth- or fourteenth-century JTS Lutzki 49 in Mizraḥi script contains the Tetragrammaton in Exod 4:10 (fol. 8v) with the correction Adonai written above the initial yod and five supralinear dots over the final three letters. Lower on the same page in Exod 4:13, Adonai is written as a correction in the right margin, with as many as nine(?) supralinear dots spread out over the four letters of the Tetragrammaton. Lewis-Gibson, Bible 2.52 (Mizraḥi, eleventh–thirteenth century) has ten dots written over an erroneous instance of the Tetragrammaton in Lev 3:16 (fol. 1v), but no dot over the inseparable preposition lamed, cf. Siegel, “The Employment of Palaeo-Hebrew,” 167–169.


That is, both the vowel points and accents were omitted. See Peretz, “Unpointed Biblical Passages,” 188, 205.


However, lengthy erasures were sometimes marked with supralinear dots and short lines, e.g., 1 Kgs 7:19; 12:9; erasures accomplished through abrasion include Judg 18:10; 1 Kgs 17:17; erasures marked with strikethroughs include 2 Sam 6:3; 22:44.


The Tetragrammaton is also marked for erasure with dots in 1 Kgs 3:10 and with short lines in Isa 30:12. There are two exceptions in which the Tetragrammaton was defaced with a strikethrough: Isa 6:1; 7:14 (see § 5.7, below). This treatment of the Tetragrammaton appears to have been the work of a different scribe, as also evidenced by the different handwriting of the supralinear correction ‮אדני‬‎ in each case; this scribe also abraded an error, replacing it with ‮אברהם‬‎ in Isa 41:8. According to Edna Engel of the Hebrew Palaeography Project (pers. comm.), the script of this other scribe “could be Mizraḥi, written at about the same time as the manuscript.”


Paul Kahle, “Die hebräischen Bibelhandschriften aus Babylonien,” ZAW 46 (1928): 113–137 (esp. 121), identified this as one of several fragments belonging to a manuscript he designated Ea 12. Yeivin, The Hebrew Language Tradition, 27, 104 categorizes this manuscript as early Old Babylonian, which he dates to the first half of the ninth century (ibid., 27). However, Yeivin warns that some manuscripts could date later than their language tradition suggests (ibid., 91) or reflect different regional dialects of Babylonian Hebrew pronunciation (ibid., 26). Despite this possibility, Yeivin maintains the early date of his Old Babylonian type (ibid., 27, 91). According to Edna Engel (pers. comm.): “Mizraḥi pre-square script, ca. ninth cent. (prior to the tenth century), Jerusalem? Ramle? Perhaps a Babylonian immigrant in the land of Israel?”


The first word was abraded and the empty space filled in with two x-like symbols. The erroneous phrase may have been ‮חי יהוה‬‎, a familiar idiom occurring forty-four times in the Bible that is graphically similar to ‮חיווהי‬‎. Another fragment of manuscript Ea 12, CUL T-S B4.20, contains a dittography of Num 35:31 (fol. 1v); the second instance of the verse is marked with numerous supralinear dots. However, short errors in this manuscript were generally corrected using abrasion. For example, ‮את‬‎ was abraded and corrected to ‮עם‬‎ in Deut 9:24 (Ox. Heb. d. 26/2, fol. 5v). In contrast, the Tetragrammaton is only marked with supralinear dots, as in Deut 15:2 (Ox. Heb. d. 26/2, fol. 9v). The supralinear dots over ‮בכל‬‎ in Deut 28:52 may be part of a dittography of several words due to homoioarchton involving ‮בכל ארצך‬‎ (CUL T-S AS 62.415, fol. 1r attaches to the upper left portion of T-S B4.21, fol. 2r [4a]). For a list of fragments belonging to this manuscript, see Yeivin, The Hebrew Language Tradition, 104.


Other instances where the Tetragrammaton is marked with erasure dots include Exod 13:14 (fol. 65r); Lev 27:30 (fol. 130v); Num 30:2 (fol. 167r). Numerous other erasures in this manuscript were carried out using abrasion, sometimes leaving the tops of the erased letters to avoid creating an open or closed parashah. For example, Exod 21:24; Lev 15:5; 20:1; 27:19; Num 7:11; 14:16; Deut 2:22; 8:19. Num 2:12 (fol. 132v) contains a strikethrough on the word ‮צורי‬‎, perhaps the work of a second hand.


BHS: “nonn Mss ‮יהוה‬‎” (see further Kennicott, Vetus Testamentum, 1.32). The mid-letter circellus is reminiscent of the procedure used for extraneous instances of vav and yod described in Beit-Arié, Hebrew Codicology, 375. Circelli inside the letters were also used to invalidate the Tetragrammaton in RNL Evr. II B 52 in Exod 39:26 (fol. 96r), with a circellus in the yod, first he, and final he, but not in the vav. The same verse contains an entire line erased through abrasion, with the correction written as three and a half lines over the erasure in smaller script. Marking a word with mid-letter circelli was not limited to the Tetragrammaton. The word ‮ביום‬‎ in Lev 7:18 (fol. 105r) was invalidated with circelli inside the letters. In Num 22:36 (fol. 166r), ‮מואב‬‎ was initially invalidated with circelli inside the letters and ‮ארנן‬‎ as a supralinear correction, but then a double strikethrough was added to ‮מואב‬‎ for no apparent reason. The words marked for erasure are unpointed in the above examples. In Num 10:6 (fol. 149r), the words ‮תימנה תרועה‬‎ were erased through abrasion and replaced with ten words in order to fill in a lacuna caused by homoioteleuton. Mid-letter circelli were used to mark the Tetragrammaton for erasure in RNL Evr. II B 4 (Mizraḥi square script, fourteenth century), Jer 26:13. However, this manuscript contains decorative “hollow” vowel points and accents, and hence these circelli are actually equivalent to dots in other manuscripts.


At first glance, there may appear to be only a single dot under the yod, but a closer examination and comparison to other instances of shewa under yod in this manuscript reveals a shewa in this instance as well. The Tetragrammaton was also surrounded with dots in RNL Evr. II B 748 (Mizraḥi, eleventh–thirteenth century), Ezek 28:22; CUL T-S B4.28 (Late Mixed Babylonian pointing), Deut 6:10 (fol. 1v).


BHS: “ℭ3 mlt Mss ‮יהוה‬‎” (see further Kennicott, Vetus Testamentum, 1.29). Throughout Ox. Heb. e. 72/4, the Tetragrammaton is vocalized with a holem over the first he; cf. Pavlos D. Vasileiadis and Nehemia Gordon, “Transmission of the Tetragrammaton in Judeo-Greek and Christian Sources,” Accademia: Revue de la Société Marsile Ficin 18 (forthcoming). There should also have been a zaqef qaṭon above the vav, but perhaps the scribe realized the error before inserting it. The phrase ‮יהוה אלהים‬‎ is also surrounded with erasure dots in Gen 15:8 (fol. 6r) and replaced in the margin with ‮אדני יהוה‬‎. In contrast, ‮לבא‬‎ was apparently corrected from the plene ‮לבוא‬‎ using abrasion in Gen 15:12 (fol. 6r). The ca. tenth-century JTS Lutzki 589, fol. 10v in Mizraḥi script, has erasure dots above the letters of the Tetragrammaton and to the left of the final he in Lam 1:14. Other examples of the Tetragrammaton marked with erasure dots or short lines of varying placements include BL Harley 5498 (Sefardic, fourteenth century), Exod 8:22; CUL T-S A15.10 (Mizraḥi square, ca. 1000), Ezek 2:4 (fol. 1r); Ezek 11:16 (fol. 2r); CUL T-S A16.22 (Mizraḥi, tenth or eleventh century), Joel 2:20 (fol. 1r); CUL T-S A22.176 (Mizraḥi, eleventh–thirteenth century), Gen 6:13 (fol. 1r); CUL T-S A32.117 (Mizraḥi, twelfth or thirteenth century), Ps 99:5 (fol. 2r); CUL T-S A34.3 (Mizraḥi, eleventh or twelfth century), Job 1:12 (fol. 1v); CUL T-S A35.56 (Mizraḥi, twelfth or thirteenth century), Lam 3:58 (fol. 1v); CUL T-S NS7.14 (Mizraḥi, eleventh–thirteenth century), Exod 21:6 (fol. 1v); RNL Antonin B 746 (Mizraḥi, tenth or eleventh century), Num 31:42 (fol. 2r); RNL Evr. II B 426 (Mizraḥi, probably eleventh or twelfth century), Gen 19:29 (fol. 1v); RNL Evr. II B 644 (Mizraḥi, eleventh–thirteenth century), Deut 14:23 (fol. 1v); RNL Evr. II B 777 (Mizraḥi, eleventh–thirteenth century), Ps 89:18 (fol. 2v); RNL Evr. II B 789 (Mizraḥi, probably twelfth or thirteenth century), Ps 55:10 (fol. 3v); RNL Evr. II B 1141 (Mizraḥi square, ca. tenth century), Jer 9:23; RNL Evr. II B 1150 (Mizraḥi, probably eleventh century), Jer 7:20; RNL Evr. II B 1254 (Mizraḥi, tenth or eleventh century), 2 Sam 7:2; RNL Evr. II B 1344 (Mizraḥi, tenth or eleventh century), Isa 9:16; RNL Evr. II B 1365 (Mizraḥi square, ca. tenth century), Ezek 33:17, 20; RNL Evr. II B 1375 (Mizraḥi, maybe eleventh or twelfth century), 2 Sam 7:18; RNL Evr. II B 1384 (Mizraḥi, tenth or eleventh century), 2 Sam 7:18; RNL Evr. II B 1418 (Mizraḥi, probably eleventh century), Isa 49:13; RNL Evr. II C 114 (Mizraḥi square [Karaite?] script, ca. fifteenth or sixteenth century), Ps 104:27; RNL Evr. II C 124 (Mizraḥi, twelfth century?), Ps 62:13; RNL Evr. II C 217 (Mizraḥi, thirteenth century?), Judg 13:16; RNL Evr. II C 408 (Mizraḥi, probably twelfth or thirteenth century), Isa 66:6; RNL Evr. II C 613 (Mizraḥi, probably twelfth–fourteenth century), Num 14:17.


On this method of erasure in nonbiblical manuscripts, see Beit-Arié, Hebrew Codicology, 375.


This sort of error is mentioned in the Jerusalem Talmud as ‮השגרת לשון‬‎ (y. Ber. 2:4 4d) or ‮אשגרת לשון‬‎ (y. Meg. 2:5 73c; Ter. 1:1), literally “fluency of tongue,” in the sense of a “familiar idiom” (cf. Jacob Neusner and Tzvee Zahavy, The Jerusalem Talmud: A Translation and Commentary [Accordance ed., version 2.2], ad loc.; Jastrow, Dictionary, 1.126, “‮השגרה, אשגרה‬‎”: “current phraseology”). The context is an error in m. Meg. 2:4, according to which the “deaf-mute, fool, and minor” are not qualified to read publicly from the Scroll of Esther on Purim. Rabbi Ḥisda argues that the deaf-mute (who cannot read aloud) was included only because it is part of a familiar idiom that normally lists all three categories of people together. Jonathan Vroom, “The Role of Memory in Vorlage-based Transmission: Evidence from Erasures and Corrections,” Textus 27 (2018): 258–273 calls this type of mistake a “memory-cued error.”


For example, in Lev 9:15 (fol. 29r) and Lev 12:4 (fol. 32v), the scribe erased the bottom portion of the extraneous words to avoid creating an open parashah. In Lev 11:5–7 (fol. 30v), four lines were erased through abrasion and replaced with five lines in a smaller script.


The same variation can be found in RNL Antonin B 252, a fourteenth- or fifteenth-century Sefardic manuscript. It has erroneous instances of the Tetragrammaton in 1 Chr 28:6 (fol. 6v) and 2 Chr 5:14 (fol. 11v), marked with a supralinear bar in the first instance and bars both above and below in the second. The error in the latter verse was influenced by the parallel in 1 Kgs 8:11, which employs the Tetragrammaton instead of “God.” The scribe apparently caught his mistake before adding the vowel points and wrote the correction ‮האלהים‬‎ immediately after the erased Tetragrammaton (at the end of the verse).


Examples of abrasion include Isa 8:13 (fol. 4r); Isa 13:16 (fol. 6r); Isa 20:6 (fol. 8r). A notable exception is the use of a strikethrough in Isa 24:8 (fol. 9v). Interestingly, the divine appellation ‮צבאות‬‎ was marked with a supralinear bar in Isa 19:19 (fol. 8r) rather than erased with abrasion, in accordance with the majority opinion in Sop. 4:1; cf. Siegel, “The Employment of Palaeo-Hebrew,” 163–164, 170.


BHS: “mlt Mss Edd ‮יהוה‬‎” (see further Kennicott, Vetus Testamentum, 1.302).


Erasures in BL Or. 2415 using abrasion include Gen 35:14 (fol. 5r); Gen 42:28 (fol. 9v). In Exod 26:25 (fol. 30v), the bottom portion of the letters were removed through abrasion, leaving the top portion behind to avoid the appearance of an open parashah. Other manuscripts with lines or bars above the Tetragrammaton include Sassoon 369 (Ashkenazic, twelfth or thirteenth century), Lam 3:31; BL Or. 2375 (Yemen, 1460–1483), Ps 94:23 (fol. 60v); Lewis-Gibson, Bible 2.45 (Mizraḥi, thirteenth century), Lev 3:16 (fol. 1v); CUL T-S A32.189 (Yemenite?, ca. fifteenth century), Ps 64:2 (fol. 1r); BL Or. 2201 (Sefardic square script, early fourteenth century), Dan 9:4, 9 (fol. 324v); BL Add. 9399 (Ashkenazic square script, thirteenth century), Dan 9:3, 9 (fol. 188v); BL Add. 15251 (Sefardic square script written in Italy, 1498), Dan 9:9 (fol. 414r); JTS Lutzki 229 (Yemen, thirteenth century), 1 Sam 1:11 (fol. 9r); RNL Antonin B 125 (Sefardic square script, ca. sixteenth century), Gen 15:8 (fol. 1r); RNL Evr. II B 734 (Byzantine, 1307), Isa 38:14; RNL Evr. II B 807 (Mizraḥi, thirteenth century or later), Ps 15:1; RNL Evr. II B 1366 (Mizraḥi, tenth or eleventh century), Neh 9:6; RNL Evr. II B 1373 (Mizraḥi, probably eleventh century), Ps 30:9; RNL Evr. II B 1385 (Mizraḥi, possibly eleventh century or later), Job 28:28; RNL Evr. II B 1455 (Sefardic, fourteenth or fifteenth century), 1 Sam 10:22; RNL Evr. II C 443 (Mizraḥi, twelfth–fourteenth century?), Exod 12:14; Zürich, Jeselsohn 5 (Sefardic script, 1477), Gen 4:5 (fol. 11v), cf. Jordan Penkower, Masorah and Text Criticism in the Early Modern Mediterranean (Jerusalem: Magnes, 2014), 283–284. RNL Evr. II B 555 (Mizraḥi, twelfth or thirteenth century), Lev 16:7–8 (fols. 3v–4r) contains a lengthy error—caused by homoioteleuton—marked with supralinear lines, dashes, and dots; the Tetragrammaton is marked with three supralinear dots and an additional supralinear line. JTS Lutzki 420 (Ashkenazic, thirteenth century), Dan 9:16 (fol. 1r) marks the error with a tailed circle above the Tetragrammaton, pointing to Adonai in the margin (cf. Beit-Arié, Hebrew Codicology, 375). BL Add 27167 (Sefardic square script, ca. 1450–1475), Lam 2:1 (fol. 421v) has the erasure line below the Tetragrammaton. Manuscripts with lines or bars both above and below the Tetragrammaton include Ox. Marshall Or. 51 (Ashkenazic, ca. 1251–1300), Num 4:46 (fol. 1r); JTS Lutzki 80 (Sefardic, fifteenth century), Gen 6:22 (fol. 8r); JTS Lutzki 200 (Yemenite, 1847), Gen 3:13 (fol. 6v); JTS Lutzki 420 (Ashkenazic, thirteenth century), Dan 9:15 (fol. 1r); BL Add. 9399 (Ashkenazic square script, thirteenth century), Dan 9:4 (fol. 188v); RNL Evr. II B 1385 (Mizraḥi, possibly eleventh century or later), Ps 89:50. AIU X.4 (Mizraḥi, eleventh or twelfth century), Deut 11:12 (fol. 1r) combines dots above each letter of ‮יהוה אלהיך‬‎ with lines in a different color ink above and below the phrase to resolve a lengthy dittography caused by homoioteleuton. The other extra words—presumably ‮דרש אתה תמיד עיני‬‎—were erased through vigorous abrasion, which left holes in the page. RNL Evr. II B 624 (Mizraḥi, probably eleventh or twelfth century), Deut 18:3 combines mid-letter dots with lines above and below the Tetragrammaton.


Tov, Scribal Practices, 201–203. The Jewish practice of invalidating the divine name with an enclosure rather than physically erasing it was so well known in the Middle Ages that it was even mentioned by the thirteenth-century English Franciscan theologian William de La Mare. See Judith Olszowy-Schlanger, Les manuscrits hébreux dans l’ Angleterre médievale: étude historique et paléographique (Paris: Peeters, 2003), 58–59.


Peretz, “Ways of Correcting,” 29–30.


My thanks to the staff of the Biblioteca Apostolica Vaticana for giving me the opportunity to examine the scroll. Vat. ebr. 2 has several other instances of erasure rectangles, such as Num 14:17 (sheet 52) around the Tetragrammaton, and Deut 9:26 (sheet 65) around ‮אלהים‬‎. Num 29:13 (sheet 58) has a clumsy double rectangle around ‮ליהוה‬‎, with the inner rectangle crossing the bottom of the first he and the vav. The complexity of these sorts of corrections can be seen in Exod 18:10 (sheet 26), which has a rectangle around the phrase ‮יהוה האלהים‬‎ but no mark on a second erroneous instance of the Tetragrammaton, resulting in the impossible phrase ‮מתחת יד יהוה מצרים‬‎. Other manuscripts that employed ellipses or rectangles to invalidate the Tetragrammaton include Vat. Ross 1169 part C (Ashkenazic, late thirteenth century), Ezek 21:8 (fol. 12r); BL Harley 5709 (Ashkenazic, ca. 1300–1339), Num 14:17 (fol. 168r); Ox. Canonici Or. 87 (produced in Rimini, Italy with Ashkenazic script, 1374–1378), Judg 6:15 (fol. 130r); BL Or. 2367 (Yemen, fifteenth century), Gen 4:8 (fol. 30v); Meir of Padua Torah, Institute of Microfilmed Hebrew Manuscripts F 76338 (Italian square script, 1545), Deut 27:3; Wolf. Cod. Guelf. 3 (Ashkenazic, thirteenth or fourteenth century), Exod 16:6; Wolf. Cod. Guelf. 148 (Ashkenazic, Torah scroll, fourteenth century), Deut 3:24; 4:24; Lisbon, National Library, MS 72 (Sefardic script, 1300) 1 Kgs 22:53 (fol. 199r); Vienna, Austrian National Library, Cod. Hebr. 11 (Ashkenazic square script, 1302), Gen 2:21 (fol. 4v); RNL Evr. II B 725 (Sefardic, fourteenth or fifteenth century), 1 Kgs 12:15; RNL Evr. II B 820 (probably North African, thirteenth, maybe fourteenth or fifteenth century?), Ps 48:6; RNL Evr. II B 917 (Mizraḥi, eleventh–thirteenth century), Exod 40:35; RNL Evr. II B 942 (Mizraḥi, eleventh–thirteenth century), Judg 13:21 (fol. 2r); RNL Evr. II B 966 (Mizraḥi, eleventh or twelfth century), Gen 20:4; RNL Evr. II B 983 (Sefardic square script, ca. seventeenth century), Josh 4:10 (fol. 1v); RNL Evr. II B 1391 (Mizraḥi, tenth or eleventh century), 2 Chr 5:1; RNL Evr. II B 1485 (Mizraḥi, tenth or eleventh century), Ezek 24:24; JTS Lutzki 195 (Yemenite, nineteenth century), Gen 1:4 (fol. 2r). In JTS Lutzki 48 (Mizraḥi square script, twelfth century), Lev 9:21 (fol. 5v) and RNL Evr. II B 9 (Mizraḥi square script, ca. 1000), Ezek 21:14, mid-letter dots were placed inside each letter of the Tetragrammaton, which was also surrounded by an ellipse, partially formed using the outer lines of the letters. JTS Lutzki 515.3 (Mizraḥi square, thirteenth century) contains both a rectangle around the Tetragrammaton and a circellus above it (cf. § 5.4, below) in Gen 21:12 (fol. 5v) with Elohim as a correction in the right margin.


An example of a ketiv-qere can be found in Gen 8:17 (fol. 7r).


BHS: “ℭ mlt Mss ⅏ ‮יהוה‬‎” (see further Kennicott, Vetus Testamentum, 1.138). There are many corrections throughout the manuscript marked with a single supralinear circellus, indicating missing words that are recorded with full pointing in the margin, e.g., Gen 4:20–21 (fol. 4r). Other methods were also used to indicate missing words, perhaps by different scribes, such as an inverted supralinear segol, e.g., Gen 16:3 (fol. 12r) and a tailed circle in the margin, e.g., Gen 42:11 (fol. 39v).


For example, Gen 17:5 (fol. 12v); Gen 35:6 (fol. 32r); Gen 47:8 (fol. 45r).


This manuscript has been identified as containing so-called Palestino-Tiberian vocalization; see Nehemya Allony, “Which is ‘Our Pointing’ in ‘Machzor Vitry?’ ” in Studies in Medieval Philology and Literature, 6 vols. (Jerusalem: Ben Zvi, 1986–1992), 2.529–539 (esp. 534 n. 40 [Heb.]); cf. Shai Heijmans, “Vocalization, Palestino-Tiberian,” Encyclopedia of Hebrew Language and Linguistics, ed. Geoffrey Khan (Leiden: Brill, 2013), 3.967–971.


For theories about the origin of the qere phenomenon and its characteristics, see Yosef Ofer, “Ketiv and Qere: The Phenomenon, Its Notation, and Its Reflection in Early Rabbinic Literature,” Leshonenu 70 (2008): 55–73; 71 (2009): 255–279 (Heb.); idem, The Masora on Scripture and Its Methods (Berlin: de Gruyter, 2019), 85–107.


Michael V. Fox, “The Qeré in the Context of the Masorah Parva,” in Le-Maʿan Ziony: Essays in Honor of Ziony Zevit, ed. Frederick E. Greenspahn and Gary A. Rendsburg (Eugene, OR: Cascade, 2016), 173. Similarly, Emanuel Tov, Hebrew Bible, Greek, Bible, and Qumran (Tübingen: Mohr Siebeck, 2008), 202, “We are probably misled by the manuscripts of MT and modern editions, both of which represent the qere as marginal corrections or footnotes. But the Masoretes had no such intention …”


On the pointing of this manuscript, see Allony, “Which is ‘Our Pointing,’ ” 534 n. 49.


Or possibly “read it as Adonai!” as an imperative. Qere has been understood as a passive participle following Emil Kautzsch, Grammatik des Biblisch-Aramäischen (Leipzig: Vogel, 1884), 81 n. 1 who cites Samuel David Luzzato, Elementi grammaticali del caldeo biblico e del dialetto talmudico babilonese (Padua: Bianchi, 1865), 38. However, the English translator remarked in idem, Grammar of the Biblical Chaldaic Language and the Talmud Babli. Idioms, trans. J.S. Goldammer (New York: Wiley, 1876), 37, “But why not ‮קְרִי‬‎ as the imperative? corresponding to ‘lege! legatur! read!’ ” Cf. also the note by the German translator in idem, Grammatik der biblisch-chaldäischen Sprache und des Idioms des Thalmud Babli, trans. Marcus Salomon Krüger (Wrocław: Schletter’sche Buchhandlung, 1873), 32.


See b. Pesaḥ. 50a; GKC § 17c; Joüon § 16 f. Most instances of the Tetragrammaton in the manuscript (e.g., Lam 3:18, 22, 24–26 [fol. 15r]) are not marked with a qere notation.


However, Lam 3:35 (fol. 15v) contains a correction without the qere notation. The scribe’s eye must have jumped, causing him to copy ‮לעות אדם‬‎ from the following verse. The two erroneous words were crossed out, a circellus was written over the following word ‮גבר‬‎, and then ‮להטות משפט‬‎ was written in the margin, without the qere notation.


An example of alternative ketiv-qere traditions is Prov 31:27, which in some manuscripts (Aleppo Codex; CUL Add. 1753; BL 9879; CUL T-S A41.1; AS2.205; RNL Evr. II B 1520) is ketiv ‮הילכות‬‎ and qere ‮הליכות‬‎ but in others (Leningrad Codex; Sassoon 1053) ‮הליכות‬‎ is in the body of the text without a separate qere (cf. Fox, “The Qeré,” 172).


The strikethrough was not always used, e.g., in 1 Kgs 1:20 (fol. 30r), where the base text has ‮ועתה‬‎ fully pointed and marked with a supralinear circellus and no strikethrough. The margin has ‮ואתה ק̇‬‎ (BHS remarks on this word “mlt Mss 𝔗Ms𝔘 ‮ועתה‬‎”; see further Kennicott, Vetus Testamentum, 1.604).


Cf. Baḥya ben Asher ibn Ḥalawa’s comment on Gen 18:3. However, the scribe employed Palestino-Tiberian vocalization and apparently made no distinction in pronunciation between patakh and qamets. See Allony, “Which is ‘Our Pointing,’ ” 536–537.


A text-correcting qere was also used in Ox. Pococke 348 (Sefardic square, fifteenth century), Ezek 18:14 (fol. 65v), where the scribe initially abraded the final letter in the word ‮ויראה‬‎, then unsuccessfully attempted to restore the erased letter, leading the scribe to record ‮ויראה ק̇‬‎ in the margin. A similar phenomenon may have happened with the word ‮והוא‬‎ in the Leningrad Codex, 2 Kgs 14:21 (fol. 212r). Another variation of text-correcting qere can be found in the ca. tenth-century CUL T-S A5.1 written in Mizraḥi square script in Deut 3:21 (fol. 1r), where the scribe accidentally left out the Tetragrammaton. To resolve the error, ‮יעשה‬‎ was marked with a circellus and ‮יהוה ק‬‎ (with no dot over the qoph) “⟨it should be⟩ read ‮יהוה‬‎” inscribed in the margin; the marginal nun also appears above the Tetragrammaton. Rather than the Tetragrammaton replacing the word marked with the circellus as in a standard qere, it is to be read in addition as ‮יעשה יהוה‬‎. On the marginal nun, see Elvira Martín-Contreras, “The Marginal Nun in the Masora of the Cairo Codex of the Prophets: Use and Function,” VT 65 (2015): 81–90.


The same method was employed in Isa 6:1 (fol. 4v), 8 (fol. 5r); 8:7 (fol. 6r). RNL Evr. II B 983 (Sefardic square script, ca. seventeenth century) contains a similar correction in Josh 3:16 (fol. 1r) where the scribe accidentally wrote the qere ‮מאדם‬‎ in the body of the text. He resolved this by writing in the margin ‮באדם כת̇‬‎ “⟨it should be⟩ written ‮באדם‬‎.”


Examples of abrasion can be found in Ezek 47:22; Jer 22:22, 28. The scribe occasionally also used strikethroughs, as in Zeph 3:6. Another method used to mark the Tetragrammaton for erasure in this manuscript is described above in n. 81.


Kolodni, “The History of Transmission,” 41; idem, “The Torah in Italian Codices and Tiqqunei Soferim in the Middle Ages” (PhD diss., Bar-Ilan University, 2008), 137–138. Two of the manuscripts had a rectangle drawn around the correction Adonai, but there was no notation on the Tetragrammaton itself. Kolodni does not give the dates for these manuscripts, but her corpus concerns texts from the thirteenth through sixteenth centuries, with the exception of one from ca. 1100 and another from ca. 1550–1650 (ibid., 23–47). For a similar method used in the Dead Sea Scrolls, but not involving vowel points, see Tov, Scribal Practices, 228.


BHS: “ℭ3 mlt Mss ‮יהוה‬‎” (see further Kennicott, Vetus Testamentum, 1.29).


ENA 2342; Yeivin, The Hebrew Language Tradition, 27, 128; Kahle, “Die hebräischen Bibelhandschriften,” 126 (Eb 16); cf. n. 66, above. No other corrections are visible in this manuscript; JTS Lutzki 507, fol. 7 belongs to a different manuscript.


Cf. Peretz, “Unpointed Biblical Passages,” 184.


The targum on the following lines contains the phrase ‮ייי ﭏהים‬‎—with an aleph lamed ligature—read aloud as ʾădōnāy ĕlōhîm. According to Judith Olszowy-Schlanger (pers. comm.), this ligature is characteristic of southwestern Mizraḥi script.

Content Metrics

All Time Past Year Past 30 Days
Abstract Views 379 0 0
Full Text Views 1207 571 14
PDF Views & Downloads 1794 695 19