CHAPTER THREE. The Development of Places of Refuge in the Bible

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 


EACH LEGAL corpus of the Pentateuch makes reference to refuges for a

killer. These sources differ on certain fundamental characteristics of such

refuges, raising the question of whether these differences reflect historical

development in the system of asylum.

Julius Wellhausen, whose work represents the capstone of nineteenthcentury

critical scholarship, argues that there were major changes in the

adjudication of homicide.1 In the earliest period, he proposes, sanctuaries

were places of asylum: A fugitive would enter a sanctuary and take hold

of the horns of the altar, obtaining sanctuary from an avenger in hot pursuit,

according to the evidence of Exod 21:13–14 and 1 Kgs 2:28. Then,

during the Deuteronomic reform in the seventh century b.c.e., in order to

prevent the loss of the institution of asylum when local altars were abolished,

Josiah appointed special cities of refuge (Deut 19:1–13). Wellhausen

concludes that the Priestly legislator affirmed this arrangement and specified

six cities of refuge, three on each side of the Jordan (Num 35:9–34;

Josh 20:1–9).2 Since four of these cities were cultic sites known from other

1Julius Wellhausen, Prolegomena to the History of Israel (preface by W. Robertson Smith;

foreword by Douglas A. Knight; 1885; reprint, Atlanta: Scholars Press, 1994), 33, 162, 375.

2Wellhausen, Prolegomena, 162, asserts that Deut 4:41–43, which specifies three cities in Transjordan,

could not be original. He does recognize that Joshua 20 contained a Priestly kernel

biblical texts and the cities of refuge were also Levitical cities, it seemed obvious

to Wellhausen that the special status of the cities of refuge was linked

to the presence of an ancient altar. He recognized, to be sure, that not every

sacred site was a city of refuge. Some that became infamous as sites of pagan

worship, such as Bethel, Dan, Gilgal, and Beersheba, were intentionally


Those who have followed in Wellhausen’s wake have argued for minor

nuances in his scheme of development. The date of the establishment of the

cities of refuge has shifted back and forth. Max Lo¨hr pushes the date back

to the reign of either David or Solomon and argues that the early monarchy

sought to limit blood revenge by means of the cities of refuge.3 Roland

de Vaux is of the opinion that the cities of refuge dated back to the reign

of Solomon: He believes that although the texts that refer to them may be

Deuteronomic, one, Josh 20:1–9, rests on older traditions because the cities

it lists were under Israelite control only during Solomon’s reign.4 Milgrom

agrees with a Solomonic date and argues that the cities of refuge were established

by Solomon, who innovated a whole host of cultic procedures,

including a type of altar that precluded altar asylum.5

Others have stood firm on a late-seventh-century Deuteronomic date.

N. M. Nicolsky holds that the cities of refuge did not exist before the Deuteronomic

revolution of the latter seventh century.6 Henry McKeating argues

that the cities of refuge were established in the seventh century when the

authorities were still trying to regulate, not oust, clan-based justice.7

All these arguments are based on the assumption of institutional change

from altar asylum to cities of refuge, whether during the first century of the

monarchy or the Deuteronomic revolution. It is worthwhile, therefore, to

assess the evidence for radical institutional change.

First, we will examine the argument that in the earliest stage of legal

development in ancient Israel, a sacrificial altar served as a refuge for a

slayer: A killer would flee to an altar and remain safe as long as he remained

there. This view has been based on texts recounting the flight of Adonijah,

one of David’s sons, and the flight of Joab, David’s army commander, to the

altar in the Tent-shrine during Solomon’s reign (1 Kgs 1:50–53; 2:28–34)

and on the statute on homicide in the Covenant Code (Exod 21:12–14).

surrounded by Deuteronomic retouching and explains that later additions to the Pentateuch

imitated Deuteronomic language (Prolegomena, 375). His refusal to recognize these additions

as products of Deuteronomy and thereby to redate Deuteronomy after P is, no doubt, due to

his belief in the late date of P. In his defense, it must be noted that passages in Deuteronomic

style are found in texts, such as Jeremiah, that are certainly not part of Deuteronomy.

3Max Lo¨hr, Das Asylwesen im Alten Testament (Halle: Max Niemeyer, 1930), 35.

4De Vaux, Ancient Israel, 162–163.

5Milgrom, Numbers, 506–507.

6N. M. Nicolsky, “Das Asylrecht in Israel,” ZAW 48 (1930), 156–157.

7McKeating, “Development of the Law of Homicide,” 64–66.

This interpretation overlooks the fact that Adonijah is not portrayed as

ever having killed another person. He seeks sanctuary because he is a political

offender. He attempted to claim the throne during David’s reign, and as a

direct result of this act, Solomon was proclaimed by David as the rightful

heir. Adonijah, now fearing for his life, flees to the altar and holds on to its

“horns.” Solomon promises that no harm will come to Adonijah as long as

he behaves. (Of course, Adonijah cannot resist and asks for one of David’s

concubines, a request that questions Solomon’s legitimacy, another political

misdeed, and he is finally put to death.) When Adonijah flees to a sanctuary,

he is seeking refuge from his political predicament, not from committing a


Joab’s flight has similarly been misinterpreted. Although his act has been

interpreted as seeking sanctuary from the penalty for homicide, the text

reflects a different motive for his flight. Joab had sided with Adonijah,

Solomon’s rival for the throne (1 Kgs 2:28), and when he hears the news

that Solomon has executed Adonijah and has dismissed Abiathar, a supporter

of Adonijah, from his post as priest, Joab realizes that it is political

payback and takes refuge at the Tent-shrine. It does him little good: He is

taken from the sacred precincts and killed. Joab’s seeking sanctuary at an

altar was politically motivated. It was not based on evading criminal culpability

for a killing. Altar asylum protects from political intrigue, not from

the punishment for homicide.8

Seeking respite from political enemies is reflected in a curious episode

in Nehemiah’s memoirs.9 During the Persian period, a prophet advises

Nehemiah to protect himself from those sent to kill him in the night by

taking shelter in the Temple (Neh 6:10–13). Nehemiah refuses, objecting

that a person in his position should not show cowardice by taking

8Ake Viberg, Symbols of Law: A Contextual Analysis of Legal Symbolic Acts in the Old Testament

(Coniectanea Biblical Old Testament Series 34; Stockholm: Almqvist & Wiksell, 1992),

122ff., recognizes that Adonijah is not seeking asylum for homicide and that his action has

no apparent relation to asylum for homicide. Viberg also realizes that Joab’s case was problematic

for two reasons: 1) Why would Joab have sought asylum since the type of homicide

he committed would have allowed him to be removed from the altar? 2) If Joab committed

culpable homicide, why would there have been a problem with removing him from the altar?

Viberg solves this problem by arguing that Joab was appealing for some form of general asylum

whereas Solomon was applying Exod 21:12–14, albeit incorrectly, since Joab was struck down

at the altar without any indication that he was removed. Of course, the answer may be simpler:

Joab thought that seeking asylum at the altar would or might protect him, and Solomon either

thought it did not or simply killed Joab anyway. The narrator’s framing of Joab’s offense as the

shedding of the blood of war in peacetime makes Joab’s actions appear illegal. However, taking

vengeance on others for what they did in war is considered acceptable, according to Judg 8:


9The use of the Temple as a sanctuary from political machinations is attested for the First

Temple period as well. Joash is kept hidden from his mother Athaliah within the Temple after

she has killed everyone else of royal stock (2 Kgs 11:2–3).

flight.10 As far as he is concerned, sanctuary asylum still operated in

his day.

The theory that altar asylum was abrogated is also based on the reference

to homicide in the Covenant Code, and it must be stated at the outset that the

Covenant Code’s statute is enigmatic. The evidence for the changeover from

altar asylum to cities of refuge is found in the early dating of the Covenant

Code and its reference to taking refuge at an altar. This evidence is in fact


Scholarly consensus identifies “the Covenant Code,” Exod 21:1–23:19,

as an independent collection of laws incorporated in a larger literary unit,

the Book of the Covenant. Although nineteenth-century scholarship was

divided on whether the Covenant Code belonged to the Pentateuchal sources

J11 or E,12 there was a wide consensus during the past century that the

Covenant Code was independent of J and E13 and was a product of the

tribal period. Since the Covenant Code does not contain any references to

the monarchy and reflects the circumstances of small landholders, the period

of the judges when the Israelites were settled on patrimonial estates in their

tribal territories seemed appropriate.14 The tribal period was considered

10Nehemiah does not object that entering the Temple precincts or the inner sanctuary would

make him liable to be killed (Milgrom, Numbers, 507). Nehemiah does enter the Temple

precincts later in the narrative in order to remove the furnishings of Tobiah’s room (Neh 13:8).

Those who were not cultic personnel were barred from the area of the altar in both the First and

Second Temple. When concerns arose about the apparent misappropriation of money brought

into the First Temple, the solution was to set aside a box for the collection on the right side

of the altar (2 Kgs 12:5–12). Since the laity did not have access to the area of the altar, the

Temple guards, who were priests, were made responsible for taking the money from the laity

and depositing it in the collection box. 2 Chr 26:16–20 explains Uzziah’s leprosy as a result of

his encroachment upon the prerogative of the priests to offer incense upon the altar. However,

according to 1 Kings 2, Adonijah and Joab do enter the area of the altar and take hold of it,

contradicting this prohibition. The fact that the area of the altar was forbidden to the laity in

the First and Second Temples may not hold true for pre-Temple shrines, like the Tent-shrine in

pre-Temple Solomonic Jerusalem.

11Wellhausen, Prolegomena to the History of Israel, 12–13, 392, considers it to be an expression

of a legislative element of the Jahwist, a document of the main part of the Assyrian period,

ca. 800 b.c.e.

12Adolf Ju¨ licher, who argues for the relationship of the Covenant Code to E, is referred to

in Bruno B¨ antsch, Das Bundesbuch, Ex. XX 22XXIII 33 (Halle: Max Niemeyer, 1892),


13B¨antsch was the first to argue for the independence of the Covenant Code from the documentary

sources (Das Bundesbuch, 73). He holds that the individual !yfp`m were known to J at

the end of the ninth century, while the collection of them was known to E in the middle of the

eighth century (Das Bundesbuch, 122). He also holds the opinion that the !yrbd were a product

of the prophetic movement of the mid-eighth century (Das Bundesbuch, 121).

14Cf. Paul, Studies in the Book of the Covenant, 44; Childs, The Book of Exodus, 456.

Ludger Schwienhorst-Scho¨nberger, Das Bundesbuch (BZAW 111; Berlin: Walter de Gruyter,

1990), 268, distinguishes between the casuistic law of the eleventh–tenth centuries, later put

together as a book in the ninth–eighth centuries, and then placed as part of divine law in the

to be the least socially differentiated period in Israelite development. Legal

thought and institutions were at their least developed as well. According to

this opinion, the dating of the Covenant Code to the tribal period would

be the launching point for extrapolating earlier stages in the adjudication of


The consensus on the date of the Covenant Code has been undermined

in recent years. One line of attack on a tribal period for its dating has focused

on recognizing reformist elements within the Code. Arguments for an

eighth-century date utilize three proofs highlighting a reformist tendency:

1) The amalgam of literary forms that makes up the entire complex of the

Covenant Code suggests the dissolution or deliberate combination of formerly

separate legal traditions.15 2) The presence of laws protecting the

poor and regulating slavery for debt is evidence for a date well into the

monarchy because there would be no need for them during the poor economic

conditions of the prestate period.16 Slavery gained significance only

during the later monarchic era. 3) The Covenant Code in its attention to

the alien presupposes widespread population shifts, which would fit well

with the serious refugee problem dealt to Judah after the fall of the northern


However, the proofs for an eighth-century date are flawed. While it is true

that the laws in the Covenant Code appear to have barely been reworked

from their original formulation because texts of varying literary form had

been placed side by side without any attempt to make thematic or linguistic

unity, it is, however, possible to offer many different points in Israelite history

when the combination of formerly separate legal traditions could have occurred.

Other considerations are also equivocal. First, laws addressing social

issues appear in all the legal corpora of the Bible and would be applicable

in almost any period of Israelite history. Second, the social conditions that

would incur debt slavery were prevalent during most of the period of the

eighth–seventh centuries. Henri Cazelles, “L’Auteur du Code de l’Alliance,” RB 52 (1945),

188, argues for a slightly earlier date, holding that Moses is the direct source.

15Frank Cru¨ semann, The Torah: Theology and Social History of Old Testament Law (trans.

Allan W. Mahnke; Edinburgh: T & T Clark, 1996), 165–169; Rainer Albertz, A History of

Israelite Religion in the Old Testament Period (trans. John Bowden; 1992; reprint, Louisville,

Kentucky: Westminster/John Knox, 1994), 1.183–184.

16One problem with this claim is the references to “slave” in biblical literature depicting

the premonarchic period. Cru¨ semann argues that the term db[ in Joshua, Judges, and 1 and

2 Samuel was most often used as a polite designation for a subordinate speaking to a superior

and, therefore, the use of the term for “slave” must date from the later centuries of the monarchy,

not the tribal period nor the beginning of the monarchy (The Torah, 152). The real conflict

during the early monarchy, according to Cru¨ semann, was between the king and the people, not

between free and slave, according to the evidence of 1 Sam 8:16–17. However, there were slaves

even during the premonarchic period, and the word used for them was db[ – there is no other

term in Hebrew.

17Albertz, A History of Israelite Religion, 1.183, 336, n. 166.

monarchy as reflected in both historical texts (e.g., 2 Kgs 4:1–7) and in the

social criticism of the prophets.18 Third, poor economic conditions were not

the only circumstances under which a person might be sold into slavery. A

thief who could not pay the penalty for his theft would become a slave: The

thief’s illegal act and his inability to pay, presumably because he was poor,

were the circumstances that caused him to be sold into slavery. Fourth, the

absence of a king could be taken as surprising if the Deuteronomic laws,

more firmly situated in the monarchic period, were replete with references

to the royal establishment, but Deuteronomy makes only a few references to

the institution of kingship (Deut 17:14–20). Fifth, the agrarian nature of the

society reflected in the Covenant Code cannot serve as proof of an early date

because Israelite society persisted in remaining agriculturally based in patrimonial

estates throughout the First Temple period.19 Lastly, the Covenant

Code does not exhibit archaic language, a characteristic expected of early

texts and the only sure proof of an early date.20 In general, it is difficult to

determine a specific date for the Covenant Code. It appears, then, that its

date cannot be chronologically set within any specific time during the First

Temple period.

The second issue is whether the refuge mentioned in the Covenant Code

is an altar. In the statute, there are two references to places involved in the

adjudication of homicide, one of which appears to be a safe haven, the other

a place from which a killer can be taken. Exodus 21:13–14 reads, “If [the

killer] did not do it by design, but God caused it to meet his hand, I will assign

you a place to which he can flee. When a man schemes against another and

kills him treacherously, you shall take him from my very altar to be put to

death.” The referent of “a place to which he can flee” in verse 13 is the

crux of the matter. Does it refer to the altar mentioned in the following

verse? Some have argued that the implication of taking away a killer even

from God’s altar is that the place of asylum in Exod 21:13 is identical to

the altar mentioned in Exod 21:14.21 However, if the statute in the first

verse was referring to the “altar,” why did it not simply state “altar”?22 The

implication, thus, is that !Oqt and jAoe.y were distinct. If so, two possibilities

arise: 1) !Oqt andjAoe.y were two completely different places; 2) one was part

of the other, the jAoe.y being part of the !Oqt.

18Ibid., 1.183–184.

19See Chapter Two on the social structure of ancient Israel.

20The Covenant Code does contain the phrase, htn[w htsk hra`, which appears to be a legal

phrase of long usage. See Shalom M. Paul, “Exod. 21:10: A Threefold Maintenance Clause,”

JNES 28 (1969), 48–53.

21E.g., A. Graeme Auld, “The Cities of Refuge in Israelite Tradition,” JSOT 10 (1978), 135.

22Cf. Moshe Greenberg, “The Biblical Conception of Asylum,” in Studies in the Bible and Jewish

Thought (JPS Scholar of Distinction Series; Philadelphia: The Jewish Publication Society, 1995),

43; Alexander Rof´e, “The History of the Cities of Refuge in Biblical Law,” in Studies in Bible

(ed. Sarah Japhet; ScrHier 31; Jerusalem: Magnes Press, 1986), 205.

The word jbzm has one meaning, “altar,” but the semantic range of the

word !wqm is more complicated. It is the common word for “place.” Sometimes,

words with a general meaning have a technical meaning in specific

genres of literature.23 To see whether a specialized meaning exists for !wqm

in legal terminology, we need to examine the usage of the word in a legal

text that contains the word enough times to indicate a specific reference.

The other statutes in the Covenant Code are of no use here since !wqm is

found only in the passage in question. The word does appear twice in the

entire complex of the Book of the Covenant, but unfortunately, it has a

specific reference in one place and a general reference in the other. So in

Exod 20:24, it refers to a sacred place. But in Exod 23:20, it refers to the

land of Israel. These references do not allow us to extrapolate a technical

meaning for !wqm. If we turn to another biblical legal corpus of greater

length, like Deuteronomy, !wqm is used to refer to cultic sites for both the

central sanctuary chosen by God (Deut 12:5, 11, 14, 18, 21, 26; 14:23,

24, 25; 15:20; 16:2, 6, 7, 11, 15, 16; 17:8, 10; 18:6; 26:2; 31:11 – these

are all described as either . . . rjby r`a !wqmh or . . . rjby r`a !wqmb) and non-

Israelite sanctuaries (singular 12:3, 13; plural 12:2).24 This word, then, is

a technical term for “sanctuary,” at least as can be identified in this legal


More importantly, this specialized meaning of “sacred site” is not limited

to legal terminology. The word !wqm is replaced with `dqm in later biblical

Hebrew texts: `dqm substitutes for the !wqm of Pss 96:6, 8 in 1 Chr 16:27, 29.25

There are several texts in Genesis that presuppose !wqm to refer to a sacred

site (12:6; 28:11).26 In 2 Sam 7:10, the future sanctuary in Jerusalem is called

!wqm.27 In Jer 7:12, 14, the word is used to refer to the Shiloh sanctuary and

to the Jerusalem Temple, respectively. This is further confirmed by the use of

mqm to denote a sanctuary in a number of Punic inscriptions,28 and in the

existence of jars with a dedicatory inscription to lmqm “[dedicated] to the

23Gary A. Anderson, Sacrifices and Offerings in Ancient Israel: Studies in their Social and

Political Importance (HSM 41; Atlanta: Scholars Press, 1987), 31–33. For example, the Hebrew

word hjnm generally refers to “a gift,” but in the Priestly law, it signifies a particular type of

sacrifice, a grain offering.

24Milgrom, Numbers, 506.

25A. Gelston, “A Note on II Samuel 7:10,” ZAW 84 (1972), 92–94. Gelston also notes that 4

Q Florilegium equates !wqm of 2 Sam 7:10 with the Jerusalem Temple. Japhet, Chronicles, 317–

318, suggests that the change in the verse was made to avoid the implication that the Temple

was standing in David’s time.

26A. Cowley, “The Meaning of !wqm in Hebrew,” Journal of Theological Studies 17 (1916),


27David Vanderhooft, “Dwelling Beneath the Sacred Place: A Proposal for Reading 2 Samuel

7:10,” JBL 118 (1999), 625–633.

28KAI 119.7 and 173.5. Cf. J. Hoftijzer and K. Jongeling, Dictionary of the North-West Semitic

Inscriptions (Leiden: Brill, 1995), 2.679–680. The word !qm appears in two Hebrew inscriptions

from the Hellenistic period referring to a synagogue, CIJ 973, 974.

shrine/sanctuary” in Philistine sites close geographically and linguistically to

ancient Israel.29

At the same time that this word has a specialized meaning, it is used for

“place” in general. The Deuteronomic corpus uses it to indicate a sacred

site, as well as to refer to a place in general without any special status

(Deut 1:31 [in the land of Israel], 33 [in the wilderness]; 9:7 [the final stop

in the wilderness]; 11:5 [the final stop in the wilderness]; 11:24 [territory

in general]; 23:17 [a chance place to which a slave may flee]; 29:6 [in

wilderness]).30 In fact, throughout the Bible, !wqm tends to be the general

word for “a place,” not one with specific status. It can, though not necessarily,

operate as a technical word for “a sanctuary.”

Furthermore, the word !wqm also refers to “town.” This can be proved

from its use in two passages: It is in parallel with ry[ in Deut 21:19, and

wmqm r[` is used synonymously with ry[ in Ruth 4:10.31

To what, then, does !wqm refer in Exod 21:13? Despite the evidence we

have adduced, the term remains ambiguous, except in one regard. It seems

safe to conclude that !wqm does not refer to the altar mentioned in Exod

21:14.32 The import of Exod 21:14 is that intentional, premeditated homicide

is so heinous that the one who commits such a transgression could even

be arrested at an altar, generally an area with restrictions against interlopers

and encroachers who have no ritual business there.33

We cannot use Exod 21:12–14 as evidence for altar asylum for killers,

the first part of the theory on the development of asylum.We can turn to the

texts from Deuteronomy to see whether the second part of the theory, the

abrogation of altar asylum, holds true. Determining what is innovative and

what is assumed as existing legal practice can be based on an explicit statement

of what passed for contemporary practice and what should be the norm

in the future. For example, the unification of worship in Deut 12:8–14 is presented

as a comparison of contemporary versus future practice. We do not

29Seymour Gitin, “Seventh Century B.C.E. Cultic Elements at Ekron,” in Biblical Archaeology

Today, 1990 (ed. A. Biran and J. Aviram; Jerusalem: Israel Exploration Society, 1993),


30Two references are ambiguous. Deut 12:13 may refer either to a place in general or to a

Canaanite altar. Deut 26:9 may refer to the land or to the central sanctuary.

31Levine, Numbers 2136, 567.

32Schwienhorst-Scho¨nberger, in Das Bundesbuch, 40–41, argues that two terms, !wqm and jbzm,

are used for the same object because God does not erect an altar but appoints a place, where

human beings erect the altar. However, this still leaves the question open as to why an altar

is not mentioned in Exod 21:13. (Schwienhorst-Scho¨nberger’s theory does not even apply to

Exod 20:22–26.)

33Cf. the execution of the unauthorized encroacher in Num 1:51; 3:10, 38; 18:7 by sanctuary

guards and in Exod 28:43; 30:20–21; Lev 10:6, 9; Num 4:15, 19–20 presumably by divine

means. Targum Pseudo-Jonathan and Targum Neofiti on Exod 21:14 assume that the killer

to be taken from the altar was a priest, presumably because only a priest would have been

authorized to be at an altar.

have an explicit comparison of present and future practice in the description

of the cities of refuge in Deuteronomy 19.

Such a comparison would be the ideal type of evidence to indicate innovation,

but in its absence, could other elements serve to indicate innovation?

A motive clause offering a rationale for the statute is present – “The blood

of the innocent shall not be shed in the land which the Lord your God is

giving to you, imputing bloodguilt upon you. . . .You shall not have pity on

[the intentional killer], but shall make expiation of the blood of the innocent,

and it will be well with you” (Deut 19:10, 13) – but it cannot be construed

as a justification for a new legal process for two reasons: 1) It does not explicitly

contrast a practice to be abrogated with one to be put into effect;

in comparison, the statute prescribing the centralization of worship in Deut

12:8–14 makes a clear-cut contrast between past (incorrect) worship and

(correct) worship in the future. 2) The entire Deuteronomic corpus is replete

with motive clauses, and it is a characteristic of Deuteronomy’s general

rhetorical style.

A comparison with another, presumably earlier, statute could serve as

evidence. For example, the laws of the slave in Exod 21:2–6 and in Deut

15:12–18 contain many parallels in language as well as contradictions in

content. These parallels and contradictions between the passages highlight

the changes made in the law. First, much of the language of Exod 21:1–

6 appears in Deut 15:12–18, including the rare usage of the term yrb[, its

only appearance in Deuteronomy. Second, the release of the slave after six

years applies solely to men, according to Exod 21:2, but in Deuteronomy

the provision is extended to women as well. The manner in which this transformation

is expressed is, in fact, evidence for the priority of the Exodus

passage. While Exodus has separate laws for a male slave (21:2–6) and for

a female slave (21:7–11), the Deuteronomy passage in its opening clause,

verse 12, stipulates that the law is to apply to both male and female slaves

in an wa formula, hyrb[h wa yrb[h Ayja, “If your brother, a male Hebrew or

female Hebrew.” The secondary nature of the /a formula is suggested by

the gratuitous repetition in verse 17b of the stipulation that the law is to

be applied to both male and female.34 Third, Deut 15:13–14 dutifully takes

up the next topic in the Exodus statute, the question of monetary payment.

The slave leaves the master’s charge without payment, according to Exod

21:2, but the Deuteronomy passages reverse the issue from a payment paid

by the slave to a payment paid to the slave and requires that the slave be

34Michael Fishbane adds an additional reason, arguing that the use of the masculine form in

the succeeding verses (vv. 12b–17a), where the interpolator failed to change the grammatical

formulation of the statute, indicates an interpolation (Biblical Interpretation in Ancient Israel

[Oxford: Oxford University Press, 1985], 171, 211, n. 99). In the same manner, Fishbane holds,

the use of the masculine form subsequently in v. 18 suggests again that v. 17b is also an interpolation.

However, it would have been unnecessary for the writer to change the grammatical

form since the masculine is used elsewhere to apply to both male and female.

given provisions. All this suggests that Deut 15:12–18 has reworked Exod


In light of this type of analysis, can such parallels and contradictions

between Deut 19:1–13 and Exod 21:12–14 be found? First, the statutes

on homicide lack the parallels in language that would indicate literary

dependence. For example, the place of refuge is described differently: in

Exod 21:13, hm` swny r`a !wqm, versus jxr lk hm` swnl . . . !yr[ `l`, and so on,

in Deut 19:2, 3, 4. Second, the formal structures of the statutes are different.

Exod 21:12–14 begins with a general prohibition of killing, followed

by provisions on the procedure to be followed in accidental and intentional

homicide. Deut 19:1–13, by contrast, begins with the command to establish

places of refuge, with information on which acts of homicide allow a

killer to gain entrance to the refuge, followed by the motive for establishing

refuges and concluding with the acts of homicide for which a killer is expelled

from a city of refuge to be executed. Third, the content of the laws

is different. The distinctions, for example, between categories of homicide

are drawn differently. Exod 21:14 defines an intentional killer as one who

willfully attacks another in treachey, whereas Deut 19:11 distinguishes one

who hates another and lies in wait for him as an intentional killer. Therefore,

no evidence exists for the dependence of Deut 19:1–13 on Exod 21:12–14.

It appears, then, that Deut 19:1–13 assumes that the cities of refuge were an

institution of long standing, not an innovation.35

Deut 19:1–13 and its parallel tradition, Num 35:9–34, do not present

the cities of refuge as something new or as a replacement. At the same time,

it must be noted that the cities of refuge are nowhere mentioned in any of the

texts that purport to tell about the early monarchy. An innovation can be seen

as a discontinuity with the past, which is how previous scholarship has seen a

change from altar asylum to cities of refuge, or as having continuity with the

past. There are no texts that depict cities of refuge as a radical discontinuity.

The cities of refuge are presented as having continuity with past practice.

The ambiguity of Exod 21:13 cannot be resolved completely. It refers to

the refuge for an accidental homicide as !wqm, a word that can have, as we

have seen, the technical meaning of “sacred place” or “town.” Exod 21:14

explicitly contrasts this with the declaration that the intentional killer may

even be taken from the altar, the most sacred part of a sacred place. The

background of the statutes in Exod 21:13–14 can either be that of a sanctuary

used as a refuge or that of a city of refuge. Furthermore, because there are

no texts that depict the cities of refuge as a radical innovation, we cannot

determine whether cities of refuge were always part of the Israelite legal

system as reflected in the Bible or whether they were a development from

sanctuary refuge.

35Levinson argues that centralization, a Deuteronomic innovation, profoundly affected the

judicial system (Deuteronomy and the Hermeneutics of Legal Innovation, 98–143).

There are other biblical texts that refer to refuge at a sacred place.36

The Psalms make reference to YHWH as being a refuge and a high tower

(Pss 59:17, 18; 144:2), protection in the shadow of YHWH’s wings (Pss

17:8, 57:2, 61:5), dwelling in the tent of YHWH (Pss 15:1, 61:5), and hiding

in the cover of his tent (Ps 27:5). While it is possible to understand these

references as metaphorical, many interpreters have taken them as describing

an actual situation, most likely because of the examples of taking refuge

at an altar. Lo¨hr understands these as making direct reference to altar asylum.

37 Other scholars have connected psalms to the cities of refuge. B. Dinur

identifies one psalm, Psalm 27, as the ritual of admission allowing manslayers

into a city of refuge.38 L. Delekat even goes so far as to argue that the

city of refuge was the Sitz im Leben of many Psalms: Psalms were written

by and for those accused of homicide.39 However, this identification of the

speaker of the Psalms as a fugitive killer is incorrect. While it is true that

the Psalms present the enemies of the speaker as wanting to kill him and

that the Psalms refer to YHWH as protector and YHWH’s dwelling as a

place of refuge, there is no indication that the refugee has been accused of

homicide. The psalms do show that a sanctuary was a refuge from a variety

of enemies.

The arguments that scholars have used to prove that altar asylum for

killers was abrogated and replaced by the cities of refuge has proven faulty.

The documentation for seeking refuge at an altar shows that the fugitives

were political offenders, not killers. The statute in the Covenant Code is obscure

and does not clearly refer to altar asylum. The passages in Deuteronomy

on the cities of refuge do not present them as an innovation. The evidence

for radical historical change is weak.

Historical change has been based on inserting P and D into a scheme of

historical development, but the evidence for prioritizing one over the other

is weak. Num 35:9–34, P’s set of laws on homicide, is an amalgam of the

Priestly traditions P and H. These sources consist of several strata laid out

side by side and result from literary activity over several centuries. Although

the P source has been dated to the Second Temple period, this has been based

on a scheme of cultic and social development.40 However, the most reliable

36One inscription, Arad letter 18, contains an elusive reference to a person who is staying in

the House of the Lord. The description of the problem is far too vague to be used as evidence

for asylum for homicide.

37Lo¨hr, Das Asylwesen im Alten Testament, 209.

38B. Dinur, “The Cultic Aspect of the Cities of Refuge and the Ceremony of Gaining Sanctuary

in Them” [Hebrew], EI 3 (1954), 144–146.

39L. Delekat, Asylie und Schutzorakel am Zionheiligtum: Eine Untersuchung zu den privaten

Feindpsalmen (Leiden: Brill, 1967), 11–39.

40Cf. Wellhausen, Prolegomena to the History of Israel, 34–51; Karl Heinrich Graf, “Die

sogenannte Grundschrift des Pentateuchs,” Archiv fu r wissenschaftliche Erforschung des Alten

Testaments 1 (1869), 466–477.

determination has been made on a linguistic basis. Certain Priestly terms

were not used during the postexilic period, others were replaced by synonyms

in the postexilic period, and others experienced a change in meaning in the

Second Temple period that contradicted their preexilic meaning.41 According

to linguistic evidence, the P source dates from the First Temple period. H,

which incorporated P, can be dated by linguistic criteria as well to the exilic

or early Persian period.42

The third main text on homicide is Deut 19:1–13. Deuteronomy as a

whole is a striated book, and although scholars have been able to divide

Deuteronomy into strata, they have had difficulty linking them to particular

time periods. Deuteronomy has correctly been linked to the cultic reform of

Josiah in the late seventh century b.c.e.43 However, the differences between

Josiah’s reform and the book of Deuteronomy, as we now have it,44 force

us to admit that while the book certainly is related to the Josianic reform,

it is unclear which literary accretions of Deuteronomy are directly tied to it.

C. Steuernagel and J. G. Staerk separated out the layers of the text on the

basis of the Israelites being addressed in the singular or in the plural,45 but

the divided text cannot be dated more precisely. The structure of Deuteronomy

as a whole has many affinities to the Vassal Treaties of Esarhaddon

(672 b.c.e.). Deuteronomy has a distinctive literary style, and from the seventh

century onward, the historiographic and prophetic texts in the Bible

exhibit many of this style’s features. Deuteronomy is definitely linked to the

seventh century, but whether any particular stratum of Deuteronomy predates

or postdates the seventh century is sheer guesswork.

The main Deuteronomic text on the adjudication of homicide, Deut 19:1–

13, can be divided into layers,46 but these strata can be assigned a date only

on a relative basis. If verses 1 and 9 are Deuteronomic, then verses 5 and

41Cf. Avi Hurvitz, A Linguistic Study of the Relationship Between the Priestly Source and the

Book of Ezekiel (CahRB 20; Rome: Pontifical Biblical Institute, 1982); Hurvitz, “Dating the

Priestly Source in Light of the Historical Study of Biblical Hebrew a Century AfterWellhausen,”

ZAW100 (1988), 88–99; Jacob Milgrom, Studies in Levitic Terminology I (Berkeley: University

of California Press, 1970), 8–16, 60–87.

42There are some indications that H may include additions from the exilic and early Persian

periods. Cf. Israel Knohl, The Sanctuary of Silence (Minneapolis: Fortress, 1995), 201–204;

Jacob Milgrom, Leviticus 116 (AB; New York: Doubleday, 1991), 27. It should be noted that

the priority of P or H relative to one another has no effect on my analysis.

43W. M. L. deWette, Dissertatio critica exegetica qua Deuteronomium a prioribus Pentateuchi

libris diversum, alius cujusdam recentioris actoris opus esse monstratur (Halle, 1805).

44For example, in contrast to 2 Kgs 23:20, there is no indication that the priests of the bamot

must be killed. Deut 18:6 and 2 Kgs 23:9 directly contradict each other. The title of the lawbook

in Josiah’s reform, tyrbh rps, does not refer to Deuteronomy but to the book referred to in

Exod 24:7.

45C. Steuernagel, Der Rahmen des Deuteronomium (2d edition; 1894; reprint, Halle: Max

Niemeyer, 1923); W. Staerk, Das Deuteronomium (Leipzig: J. C. Hinrichs, 1894).

46A. D. H. Mayes, Deuteronomy (New Century Bible Commentary; Grand Rapids, Michigan:

Eerdmans, 1979), 283–285, 297.

8 can be considered post-Deuteronomic, because they refer to Israel as a

whole rather than to a specific community waiting to enter the land of Israel

and because they use a late expression to refer to the priests, ywl ynb !ynhkh.

There also may be pre-Deuteronomic law in verses 4–5, 11–12, assuming

that casuistic formulation indicates earlier material, and post-Deuteronomic

supplements in verses 8–10, a contradiction to 4:41–43.

The biblical laws on homicide, Num 35:9–34 and Deut 19:1–13, as

well as Exod 21:12–14, are mutually independent sources. They do not

have a common literary origin, and they stem from diverse historical and

ideological/theological settings. Although previous studies of biblical law

have assumed that each of the legal parts of the Pentateuch can be securely

dated to different periods and have devised schemes of historical development

in biblical law based on that dating, the legal portions of the Pentateuch

cannot be dated with such precision and, in fact, all date from

some time in the First Temple period with no clear evidence for historical


The legal sources in the Pentateuch, P and D, that differ on the characteristics

of the places of refuge do so because they conceptualize the sanctuaries

within their ideological/theological program. They disagree on the number,

the sacred status, the rationale, and the existence of a technical term for the

cities of refuge.

The statute in Num 35:9–34 calls for the establishment of six cities as

refuges for the slayer from !dh lag, denoting them by the technical term

47The narrative sources are also difficult to date. Two examples, the killing of Joab (1 Kgs 2:5–

6, 28–34) and the execution of those who assassinated Joash (2 Kgs 14:5–6), can illustrate the

quandary. The text of 1 Kgs 2:5–6, 28–34 forms part of the Succession Narrative (2 Sam 9–20;

2 Kgs 1–2), a product of an author in Solomon’s court who collected a number of independent

stories and wove them together in order to justify the new king’s accession to the throne. (Cf.

Leonhard Rost, The Succession to the Throne of David [trans. by Michael D. Rutter and David

M. Gunn; intro. by Edward Ball; 1926; reprint, Sheffield: Almond Press, 1982]). The rest of

Samuel is assigned to the earlier reign of David. However, can the Succession Narrative so

easily be separated from the rest of Samuel? David’s admonition to Solomon to kill Joab, 1 Kgs

2:5–6, a part of the Succession Narrative, depends directly on the murders of Amasa and Abner

in 2 Sam 2:18–23 and 3:28–30, which are not identified as part of the Succession Narrative.

More critically, the dating of these texts to the reigns of David and Solomon is problematic.

It has simply been assumed that the narratives are contemporary with the events described,

but there are a number of details that would indicate a later date. (Cf. David M. Gunn, The

Story of King David [JSOTSup 6; Sheffield: JSOT Press, 1978], 32ff.) Indeed, the reigns of

David and Solomon may have been an alluring setting to writers for centuries. Other narrative

texts are also problematic. For example, the account of the execution of those who assassinated

Joash, in 2 Kgs 14:5–6, contains a quotation of Deut 24:16. This could be used as a means of

fixing the date of 2 Kgs 14:5–6. Unfortunately, I can argue for mutually contradictory dating

with virtually the same reasoning. Does the quotation of Deut 24:16 by 2 Kgs 14:5–6 signify

that 2 Kgs 14:5–6 must have been composed after the promulgation of Deuteronomy?

Or does this show that Deut 24:16 circulated independently before the public release of


flqm, “refuge; confines.”48 This term is not found in the other legal sections

of the Pentateuch, and its appearance here reflects a characteristically Priestly

concern for technical and ritual terminology.

Three cities were to be established on each side of the Jordan River, and

all six were to be appointed simultaneously after the crossing of the Jordan

River, reflecting P’s emphasis that the Land of Canaan was to be conquered in

one fell swoop and that nothing was to be established there until the conquest

was complete. Canaan was not to be distributed to the tribes piecemeal, but

rather was apportioned at a single official ceremony. The cities of refuge were

part of a scheme of forty-eight cities assigned to the Levites. The question

of whether the Levitical cities ever existed in reality has been the subject of

heated debate.49 Yet little doubt has been cast on the existence of the cities

of refuge.

According to Num 35, six cities of refuge were to be established, and the

number six develops from the Priestly law’s endeavor to schematize Israel’s

antiquity.50 God is revealed in stages and, therefore, the relationship between

God and human beings is described in a series of four covenants (implicitly

with Adam, Gen 1:28–2:4a; Noah, Gen 9:1–17; Abraham, Gen 17; Israelites

in the wilderness, Exod 24:1–8).51 The Israelites are divided into twelve tribes

ruled by twelve chieftains. The tribes encamp in four groups of three around

four standards on the four sides of the Tabernacle, the wilderness sanctuary.

On the march, two standards precede the Tabernacle and two follow.Within

the Tabernacle, four families serve. The priests and the Levites are given fortyeight

cities, of which six are cities of refuge, three on each side of the Jordan.

All these numbers are based upon the numeral 12, its multiple 48, and its

divisors 2,3,4,6. The precise number of cities of refuge is therefore generated

by the Priestly law’s theological numerology.

The Priestly law links the cities assigned to the Levites and the places

of refuge. Many scholars have, therefore, extrapolated that the reason for

assigning a certain city as a place of refuge was due to its status as a sacred city

and/or to the existence of an altar, presuming that the institution of asylum

48The choice of this term for the cities of refuge reflects the theology/criminology informing

the significance of the cities of refuge for the accidental killer. See the analysis of the accidental

killer’s confinement in Chapter Four.

49E.g., W. F. Albright, “The List of Levitic Cities,” in Louis Ginzberg Jubilee Volume (New

York: American Academy for Jewish Research, 1945), 1. 49–73; Benjamin Mazar, “The Cities

of the Priests and the Levites,” Congress Volume 1960 (SVT 7; Leiden: Brill, 1960), 193–204;

Menahem Haran, “Studies in the Account of the Levitical Cities,” JBL 80 (1961), 45–54, 156–

165 (also Temples and Temple-Service in Ancient Israel [Winona Lake, Indiana: Eisenbrauns,

1995], 122–131); J. P. Ross, “The ‘Cities of the Levites’ in Joshua XXI and I Chronicles VI,”

Ph.D. diss., University of Edinburgh, 1973; John R. Spencer, “The Levitical Cities: A Study of

the Role and Function of the Levites in the History of Israel,” Ph.D. diss., University of Chicago,


50Rof´e, “The History of the Cities of Refuge in Biblical Law,” 225.

51Wellhausen, Prolegomena, 338–342.

developed historically from altar asylum to sanctuary in a sacred city. This

begs the question as to why the status of city of refuge was not extended to

all the Levitic cities: If they were sacred or possessed an altar, why did they

not qualify as a city of refuge? Furthermore, a distinction must be drawn

between a city for the Levites to dwell in and the location of the sanctuaries

in which they performed cultic functions; there was no direct link between a

Levitic city and a sacred place. A Levite might live in one place but officiate

as a Levite in a different location.52 Abiathar, for example, owns a field in

Anathoth, but officiates in Jerusalem and Nob (1 Kgs 2:26).

There is another aspect to the cities of the Levites that is critical to the

functioning of a refuge. They were to be distributed evenly throughout the

Land of Israel and Transjordan (Num 35:8). In order to provide equal and

easy access for a slayer, the cities of refuge needed to be distributed evenly

throughout the territory: If each tribe had its own place of refuge, slayers

from tribes with smaller territory would have easier access than those

from tribes with greater territory. This is reflected in the specific language

of the command, “You shall make accessible to yourselves cities of refuge”

(Num 35:11). It is not the sacredness of a Levitic city that determines its

status as a refuge but rather its geographic distribution.

Understanding that the cities of refuge were selected for reasons of geography

clarifies a difficulty with associating the right of asylum with a function

of cult sites, with an altar and sacred space. It is assumed that while one who

is in the presence of God is under his protection, those who are not supposed

to be in a sacred space are executed. If the Priestly statute avers that the slaying

pollutes (Num 35:33–34), would a slayer be permitted to encroach upon

a sacred place? Indeed, the point of the Levitic cities is that they are manned

by personnel not limited by tribal geography and who had access to nonsacred

aspects of sacred items. The Priestly law stipulates that the priesthood

was stratified into priests, strictly defined, and Levites, service personnel

limited to assigned tasks of the transport, maintenance, and handling of

cultic items (Num 3–4; 8:5–22). The Levites were the corps of subordinate

servitors, relegated to nonsacral functions of sacred sites and rites.53 They

assisted the priests (Num 18:2,4) and performed acts that do not pertain to

the altar (Num 16:9). The functions of the Levites are outside cultic sanctity,

according to the Priestly traditions (Ezek 44:11; 46:24).54 Being divorced

from the sacred and being geographically distributed, the Levites are therefore

the appropriate personnel to oversee the cities of refuge. A killer was to

be kept away from all that was sacred because his offense polluted the land.

52Menahem Haran, Temples and Temple-Service in Ancient Israel (Winona Lake, Indiana:

Eisembrauns, 1995), 119–120.

53The distinction between priests and Levites may be a distinction only in Numbers. Cf. Levine,

Numbers 120, 65, 81, 104–105.

54Haran, Temples and Temple-Service in Ancient Israel, 61.

Just as the theological and social program of Numbers informs the

statutes establishing the cities of refuge in that book, so too does the social

and theological program of Deuteronomy shape its statutes on homicide.

Deuteronomy manifests anxiety over the possibility that !dh lag might commit

an accidental homicide because he could kill any slayer with impunity

outside the city of refuge – “Whoever came with his fellow into the forest

to cut wood: as his hand swings the ax to cut down the tree, the ax-head

falls off the handle and hits the other so that he dies – that man shall flee

to one of these cities and live, lest the blood avenger pursuing him in his

hot anger, overtakes him and slays him because the distance is too great, yet

he was not liable to the death penalty because he was not hostile to him in

the past” (Deut 19:5–6). Deuteronomy is concerned with the slaying of the

accidental killer and the effect it would have upon the Israelite people. In

contrast to the Priestly law, where the slaying of the accidental killer does

not incur culpability at all, the Deuteronomic statute evinces the belief that

the killing of a fugitive who has not yet reached a refuge does. In his case,

the fugitive’s status as an accidental or intentional killer is not yet clear: He

may potentially be an accidental killer, whose death is unwarranted. In an

ironic transformation, the same term for culpability for the victim, yqn !d,

“innocent blood” (Deut 19:13), is used to refer to culpability for the killer

slain before he reaches the city of refuge (Deut 19:10).

If the blood avenger manages to overtake the fugitive and kill him, the

Israelite people as a whole are responsible, according to Deut 19:10. The

Priestly law, by contrast, avers that the land, not the people, will be polluted

by the presence of the unpunished slayer, not the death of the accidental

slayer. The Priestly law is concerned with the purity and the pollution of

space, the Deuteronomic with that of the Israelite people. The Priestly law

is concerned with the pervading presence of God in the midst of Israel,

while Deuteronomy focuses on the conduct and fate of the Israelite people.55

Indeed, Deuteronomy is completely unconcerned with the immanence of

God: for Deuteronomy, the Temple in Jerusalem is not the dwelling place of

God; it is the place where God causes his name to dwell. The Priestly law

is concerned with the polluting effects of a slaying, whereas D is concerned

with the social aspects of the law.

Like the statutes in Numbers, Deuteronomy’s places of refuge are divorced

from any link with the sacred or the priesthood, but this is informed by

the Deuteronomic trend toward secularization and the separation of sacred

and secular. This tendency is part of the larger program of Deuteronomy.56

Warfare, for example, is stripped of its sacred ritual in Deuteronomy. There

55Moshe Weinfeld, Deuteronomy 111 (AB; New York: Doubleday, 1991), 25.

56Weinfeld, Deuteronomy and the Deuteronomic School, 233–243. The term secularization

may be too strong a term for this aspect of Deuteronomy’s program, as Weinfeld himself

notes (“On ‘Demythologization and Secularization’ in Deuteronomy,” IEJ 23 [1973],

is no mention of the sounding of the priestly horns or of the plunder that

is to be dedicated to the sanctuary.57 The use of the ark and the holy vessels

is missing in Deuteronomic warfare. Even the function of the priests is

secularized. In P, the priests are to sound their horns so that the warriors

are remembered by YHWH (Num 10:9). In D, by contrast, the priest addresses

the people to inspire their courage (Deut 20:1–4). This secularization

is carried through in the Deuteronomic tradition of the cities of refuge. The

Levites, who are considered the Priestly class in Deuteronomy, are not connected

with the cities of refuge. The high priest is not a factor in the stay of

the accidental homicide in a city of refuge.

Like Numbers, the Deuteronomic refuges are established on the basis of

geographic considerations: The country is to be divided up into three parts

with a city of refuge established in each area so as to enable a slayer to seek

refuge efficiently.

Unlike the refuges in Numbers, the tally of the cities of refuge in Deuteronomy

is complex because, as the present text of Deuteronomy reads, the calculation

is linked to a multistaged conquest, in contrast to the comprehensive

conquest in Numbers. Three stages are indicated in the establishment of

the cities of refuge. In Deut 4:41–43, Moses designates three cities after the

territory east of the Jordan is conquered.58 In Deut 19:1–7, the Israelites

are commanded to designate three cities after the conquest of the Land of

Israel. In Deut 19:8–9, three more cities are to be added after additional

conquests. If we were to read the chapters of Deuteronomy in succession,

the total number of the cities of refuge is nine. However, in Deut 19:1–7,

the command to set up three cities after the conquest of the Land of Israel is

presented as a new injunction without any reference to the establishment of

earlier cities in the territory east of the Jordan River.59 Indeed, in Deut 19:7–

9, the text reads: “When the LORD your God enlarges your territory . . . you

shall add three more cities to these three.” If the law was to include the

three cities mentioned in Deut 4:41–43 as well as the three established by

Deut 19:1–7, it would have stated “these six.” It clearly appears that the

author of Deuteronomy 19 did not know of Moses’ action in Deut 4:41–43.

In fact, Deut 4:41–43 is placed between Moses’ lengthy orations of Deut

1:1–4:40 and 4:44–26:19, an appropriate place for an insertion.

Deut 19:8–9, the command to establish three more cities after additional

conquests in the land of Israel, is apparently parenthetical to the command

in Deut 19:1–7.60 Deut 19:10 does not refer back to verse 9 but to verse 7

230). Cf. Milgrom, “The Alleged ‘Demythologization and Secularization’ in Deuteronomy,”

IEJ 23 (1973), 156–161.

57Compare Num 10:9; 31:6, 50–54; Judg 7:19–20; 2 Sam 8:11; 11:11.

58This section and Josh 20:8 are almost identical, but it is difficult to say which has priority.

Cf. Auld, “Cities of Refuge in Israelite Tradition,” 138.

59Rof´e, “History of the Cities of Refuge,” 222.

60Ibid., 222–224.

because it deals with “the land that the LORD God is allotting to you,” not

to the enlarged territory. Furthermore, the motive cited by verse 10, “Thus

the blood of the innocent will not be shed,” is linked to verse 6, “yet he

was not guilty of a capital crime.” Deut 19:8–9 is a secondary layer whose

purpose is to adapt the law of the cities of refuge to Num 35:9–34, which

stipulates the establishment of six cities. The author of Deut 19:8–9 attempts

to reconcile the two laws. However, another attempt was made to reconcile

the sources, Deut 4:41–43. These texts in Deut 4:41–43 and 19:7–8 are not

to be seen as connected with historical reality but as examples of intrabiblical

exegesis reconciling contradictions in inherited legal literature. In fact, the

original command in Deuteronomy was to establish only three cities.61

Both Numbers and Deuteronomy outline a specific procedure to adjudicate

whether the killer committed intentional or accidental homicide. According

to Numbers, once the fugitive has reached the city of refuge, the trial

is conducted before the hd[, “assembly.”62 The hd[ and the leaders of the

hd[ play an important role in the book of Numbers. The term hd[ refers to

the entirety of the Israelites,63 who witness public ceremonies.64 The term

can refer more specifically to the assembly of adult Israelite males.65 The

chiefs of the hd[ hold executive powers.66 They take the initiative in dealing

with certain problems and represent the entirety of the Israelites in situations

where the presence of all the Israelites would be impossible.67 When

it appears in biblical literature portraying a later period, it seems to be a

61Rof´e argues that Jerusalem was also to be included as a place of refuge even though the

statute in Deuteronomy does not allude to Jerusalem (“History of the Cities of Refuge,”

215, 224). Rof´e understands “(geographic name) $ra lwbg” in Deut 19:3 as Israelite territory

outside of Jerusalem and Benjamin which does not include Jerusalem and Benjamin. Therefore,

the threefold division was established in addition to the territorial unit of Jerusalem and

Benjamin. There are two difficulties with this. First, if Jerusalem were to be included as a

city of refuge, why did the text not stipulate a fourfold division of the country? Second, a

study of the phrase “(geographic name) lwbg” indicates that when it refers to dividing territory,

it includes the entire territory of a country within the actual line serving as a boundary:

There is no indication that the capital city is excluded. (Cf. the dividing of the country in

Num 34 and Josh 15. Also, an entire country: Gen 10:19; Exod 7:27; 10:4; Num 20:16, 17.

The phrase “(geographic name) lwbg lk”: Exod 10:14, 19; 13:7; Judg 19:29; 1 Sam 11:3, 7;

27:1; 1 Kgs 1:3; 2 Kgs 10:32; 14:25; 1 Chr 21:12.)

62The term hd[ can be used as proof for the First Temple date of the Priestly tradition. Cf. Avi

Hurvitz, “The Use of the Priestly Term ‘hd[’ in Biblical Literature,” [Hebrew] Tarbiz 40 (1970):


63Exod 12:3, 6, 47; 16:1, 2, 9, 10, 19; 17:1; 35:1; Lev 4:13; 10:6; 16:5; 19:2; Num 1:53; 3:7;

8:9, 20; 10:2; 13:26; 14:1, 2, 3, 5, 7, 10, 25, 27; 15:24, 25, 26, 33, 35, 36; 16:3; 17:6, 7, 10,

11; 19:9; 20:1, 2, 8, 11, 22, 27, 29; 25:6, 7; 27:2, 3, 14, 16, 17, 19, 20, 21, 22; 31:12, 16, 43;

Josh 9:19, 21; 18:1; 22:12, 16, 17, 18, 20; Ps 74:2; Jer 6:18; 30:20.

64Lev 8:3, 4, 5; 9:5; Num 10:3; 1 Kgs 8:5 [1 Chr 5:6].

65Exod 35:4, 10; Num 1:2, 18; 26:2.

66Num 1:16.

67Exod 16:22; 35:31; Lev 4:15; Num 4:34; 16:2; 31:13, 26, 27; 32:2, 4; Josh 9:15, 18, 27;


pan-Israelite assembly. In Judg 20:1 and 21:10, 16, it has both political and

judicial aspects. The hd[ arbitrates between an individual and a tribe, declares

war on a particular tribe, and accepts terms of peace. Its political role

is otherwise vague. None of the judges is portrayed as consulting with the

hd[. There is only a single reference to it in the history of the monarchy: 1

Kgs 12:20 refers to hd[ as the body that crowns Jeroboam I.

It would seem very unwieldy to convene all the Israelites to judge a case

of homicide, as Numbers 35 prescribes. The wilderness setting of the book

of Numbers makes the judicial function of the hd[ appear to be an archetype

for local communities and sanctuaries. Just as the law of slaughter in Lev

17, as an example, applies to local sanctuaries, not to a central sanctuary,

so too does the term hd[ apply to a small local court, not a central assembly.

Numbers 35 is the only reference to the judicial function of the hd[. The

references elsewhere to the role of the hd[ in the punishment of a violator

of the Sabbath and of a blasphemer (Num 15:33, 35, 36; Lev 24:13, 16)

are misleading because the hd[ in these cases does not exercise any role in

the adjudication. Rather, the hd[ in Num 15 and Lev 24 is the entirety of

the Israelites from whose midst the transgressor is extirpated. The term hd[

appears to signify one meaning in almost all of the Hebrew Bible and another

in Numbers 35. It appears to be a judicial body outside of the city of refuge

because of the stipulation that the hd[ will return the slayer to the city of

refuge upon deciding that the death was inadvertent.

In Deuteronomy, the obligation for adjudicating the case devolves upon

the killer’s home city. According to Deut 19:12, if intentional homicide took

place, the elders of the killer’s city,68 not a pan-Israelite body like the hd[,

should take the killer from the city of refuge and have the blood avenger

execute him. The accused must be judged in his own city in absentia because

1) the text describes the action of the elders as the implementation of a

judgment already made, and 2) the accused is in one of the cities of refuge to

which he fled in fear of the blood avenger after committing the killing. Who

makes this judgment? It is likely that these elders are the ones, because 1)

they are explicitly mentioned as extraditing the intentional killer, and 2) the

68Deut 19:12 refers to wry[ ynqz, “the elders of his city.” Although one might argue that this

refers to the elders of the victim’s city, it is much more likely that the elders of the killer’s city

were involved. (Cf. Driver, Deuteronomy, 233; von Rad, Deuteronomy, 127; Rof´e, “History

of the Cities of Refuge,” 228.) The elders of the offender’s city are more inclined to keep one

of their own safe and hold a fair trial than the victim’s city with the victim’s kin thirsting

for revenge. If they were partial to the slayer because he was one of their own, bloodguilt

would fall upon them. (Cf. Hanokh Reviv, The Institution of the Elders in Ancient Israel

[Hebrew], [Text and Studies; Jerusalem: Magnes, 1983], 66.) Another Deuteronomic stipulation,

Deut 21:1–9, contains a ceremony designed to address the guilt of the city nearest

to the place where a human corpse is found. If we may call upon evidence of the role of

elders from another legal action, Deut 25:5–10 stipulates that in the case of levirate marriage,

the responsibility for resolving the dispute devolves upon the elders of the offender’s


elders do exercise judicial functions in general.69 If the fugitive is condemned

as having committed homicide intentionally and with prior malice, the elders

send for him and deliver him to the blood avenger.

As we have seen, Numbers 35 and Deuteronomy 19 reflect the theological

and social programs of the Priestly literature and of Deuteronomy. Their

conceptualization of the adjudication of a slaying coincides and diverges

because their distinctive theological and social programs shape the process


A new element in procedure, a hearing of admission to gain entrance into

the city of refuge, is found in another tradition about the cities of refuge,

Josh 20:1–9:

1 The Lord said to Joshua, 2 “Speak to the Israelites, saying, ‘Assign

the cities of refuge, about which I commanded you through Moses,

3 to which a slayer who strikes down a person by mistake unintentionally

may flee; they shall be a refuge for you from the blood avenger.

4He shall flee to one of these cities, present himself at the entrance to

the city gate, and plead his case before the elders of that city. They shall

admit him into the city and give him a place to live among them. 5 If

the blood avenger should pursue him, they shall not hand the slayer

over to him, for he struck his neighbor unintentionally and had not

been his enemy before. 6He shall live in that city until he stands before

the assembly for trial, [and remain there] until the death of the

high priest who is in office at that time. Then the slayer may return

to his town and his home from where he fled.’” 7 They sanctified

Kedesh in the Galilee, in the hill country of Naphtali, Shechem in the

hill country of Ephraim, and Kiryat Arba, that is, Hebron, in the hill

country of Judah. 8On the other side of the Jordan, eastward, they

assigned Bezer in the wilderness, in the steppe of the tribe of Reuben,

Ramot in Gilead, of the tribe of Gad, and Golan in the Bashan, of

the tribe of Menasseh. 9 These are the designated cities for all the

Israelites and the alien who dwells among them, so that anyone who

69Cf. Deut 21:18–21 (although there is no explicit mention of a trial; if the child’s parents did

have the right to condemn him without need of official judgment, why would the parents be

required to present the situation before the elders?); 22:13–21; 25:5–10; 1 Kgs 21:8–13; Ruth

4:1–12. However, Rof´e suggests that if the families of the victim and the killer agree that the

killing was unintentional, there is no need for the elders to be involved in any capacity (“The

History of the Cities of Refuge,” 229). Only if they agree on the culpability of the killer, then

the elders must exercise their executive function in extraditing the fugitive. If the families of

the victim and the killer do not agree, according to Rof´e, the determination is made by the

consensus of the local community. This solution of Rof´e’s seems unwieldy. A dispute cannot

be resolved by an amorphous body deciding on the basis of rumors and hearsay. If a formal

presentation is required for other offenses that are less serious than a charge of murder, it would

be unlikely that a slaying would be adjudicated in a less organized manner. Therefore, a formal

trial in absentia before the elders was warranted.

kills a person by mistake may flee there and not die by the hand of

the blood avenger before he has stood trial before the assembly.

In order to understand the origin of this innovation, the relationship of this

tradition to the other two must be analyzed.

The presence of elements from both Numbers 35 and Deuteronomy 19

in Joshua 20 is apparent:70

Josh 20:2 hm dOE !eyo3 y#iAAr3 flaŽU yiAt0 !el Wn!, “Assign the cities

of refuge” – The term flay yi, “cities of refuge,” appears only in Numbers

35 (vv. 11, 14, 25, 26, 27, 28, 32). The verb used for assigning cities here

and in Num 35:13–14 is n-t-n.

Josh 20:3 .!U l5y flaym !el WyW¨ t!EAyp< hPP•– vE} hŠr U\/r hO§ sWnl,

“to which a slayer who strikes down a person by mistake unintentionally

may flee; they shall be a refuge for you from the blood avenger.” –

The entire verse is an almost complete parallel to Num 35:11b–12a, which

reads l5y flaym !yiV !el WyW¨ hPP•– vE} hŠr \\o hO§ sn¨ “to which a slayer

who strikes down a person unintentionally may flee; the cities shall be as a

refuge from the avenger.” The double characterization of this type of murder,

t!EAyp< hPP•– “by mistake unintentionally,” is a conflation of the criteria of

Numbers and Deuteronomy. Num 35:11b denotes this category of killing by

the term hgg`b, “by mistake,” while Deut 19:4 uses t[d ylbb, “unintentionally.”

Josh 20:4a hk· 5W !yiVx t\.Al0 sn¨, “He shall flee to one of these cities.” –

This clause is similar to Deut 19:5b, yk hk· 5W !yiV t\.Al0 sWnŸ aWh, “that man

shall flee to one of these cities and live.”

Josh 20:5aα wye`. !U l5 #D


ia yi¨, “If the blood avenger should pursue

him” – This is paralleled in Deut 19:6, \\oW yi`. !U l5 #D


iaA@X, “lest the

blood avenger pursue him.”

Josh 20:5b !/vms l/m!y /l aWh aqcA4l¨ WhiAt0 hE[ t!DAypij yH, “for he

struck his neighbor unintentionally and had not been his enemy before.” –

This is paralleled in Deut 19:4b, l/m!y /l aqc ·

4l aWh¨ t!EAypi WhiAt0 hD r3

!Rms, “whoever slays his fellow without intent and was not hostile to him in

the past.”

Josh 20:6aβ fYym hEW yqOp /du d!, “until he stands before the assembly

for trial.” – A trial before the assembly is stipulated in Num 35:24.

Josh 20:6γ lOdfiU @ZIU tOm d!, “until the death of the priest” – The release

date of the accidental homicide is the same as in Num 35:28.

Josh 20:6b .!&y sn r3 ryWAl0 OtyoeAl0¨ OryAl0 a;W jA\OrW bWvŸ z1; “Then the

killer may return to his town and his home from where he fled.” – Num

70There may be another parallel, if a textual emendation is warranted. In v. 7, the cities west of

the Jordan are set aside (wvdqyw), and I would argue that this is an error for wrqyw, the same verb

used in Num 35:11. The error was caused by the proximity of the place name Kadesh (vFa) in

the same verse.

35:28 stipulates that the killer may return to his patrimonial estate, hl`m.

Deut 19:12 mentions that the killer departed from “his town,” /ry.

Joshua 20, however, does contain one verse that constitutes a departure

from Numbers 35 and Deuteronomy 19. Josh 20:4 mandates that before the

fugitive is permitted to enter the city of refuge, a hearing must take place

in order to determine whether he is eligible for admission to the city at all.

The accidental killer is to be stopped at the gate of the city of refuge. He

can only gain admittance after he presents his case to the elders of the city

of refuge that the slaying was accidental. This appears to be separate from

his trial, which must still take place before the assembly (Josh 20:6). The

hearing is apparently a way to prevent intentional slayers from entering the

city of refuge at all.

This new element, the procedure of admission in Josh 20:4, has a relationship

to certain elements in Deuteronomy. In Deut 19:12, the elders

of the slayer’s city play a role in determining his guilt and, if he is found

culpable, deliver him to the blood avenger. City elders resolve disputes in

Deut 22:13–21 and 25:5–10. In the same vein, Josh 20:4, in introducing a

procedure of admission, assigns it to the elders, albeit of the city of refuge,

not the elders of the killer’s city, as in Deut 19:12. This is appropriate since

the elders of the city of refuge are protecting it from the presence of intentional

killers. In addition, the elaboration in Josh 20:5–6 that the accidental

slayer will not be delivered to the blood avenger and will live in the city of

refuge, b`yw . . . wrgsyAalw, uses language reminiscent of Deut 23:16–17, which

mandates that an escaped slave will not be delivered over (rygst) to his master

and will live among (b`y Am[) the Israelites.71

How can the presence of a new procedure in Deuteronomic language that

does not appear in Deuteronomy be explained? The simplest explanation is

that Joshua 20 is a Deuteronomic reworking of a Priestly kernel.72 In fact,

one manuscript of the LXX, Codex Vaticanus, contains none of the Deuteronomic

additions, including Josh 20:4, and therefore provides evidence for the

independent existence of a Priestly pericope. The new procedure in Josh 20:4

was worded in Deuteronomic style, although its content differs significantly.

While the other versions, Numbers 35 and Deuteronomy 19, eventually restrict

asylum to accidental killers, Joshua 20 limits initial entrance into

the city of refuge only to accidental killers. The probability of a Deuteronomic

reworking of Priestly material is further heightened by the fact that

where Joshua 20 has parallels to both Numbers 35 and Deuteronomy 19,

it is invariably closer to, or identical with, Deuteronomy 19. This makes

sense since both are part of the Deuteronomic literature. A Deuteronomic

author, later than both Numbers 35 and Deuteronomy 19, designed this new

71Rof´e, “Joshua 20: Historico-Literary Criticism Illustrated,” in Empirical Models for Biblical

Criticism (ed. Jeffrey H. Tigay; Philadelphia: University of Pennsylvania Press, 1985), 137–138.

72Ibid., 141–143.

procedure to allay any anxiety over the presence of intentional killers in a

city of refuge by preventing them from gaining entrance in the first place.

The intentional slayer has no way of escaping the blood avenger even for a

limited time in a city of refuge until he is convicted and handed over to the

avenger to be killed.

The legal sources of the Pentateuch P and D differ as a direct result of

their distinctive ideological and theological programs. The statute in Joshua

is an attempt to reconcile these differences. I have shown that the cities of

refuge were not a Deuteronomic innovation nor were they an innovation

of the early monarchy. However, there may have been a development from

sanctuary asylum to the cities of refuge at a glacial speed, starting with the

ability of others besides killers to seek sanctuary at an altar and the indication

in the Psalms that the Sanctuary was a place of refuge from danger. The

development of homicide in the Hebrew Bible follows along the lines of a

steady-state theory, with the recognition that even a steady-state universe

experiences change from time to time.