19 See Chapter Two on the social structure of ancient Israel.
20 The Covenant Code does contain the phrase, htn[w htsk hra`, which appears to be a legal
phrase of long usage. See Shalom M. Paul, ‚ÄúExod. 21:10: A Threefold Maintenance Clause,‚ÄĚ
JNES 28 (1969), 48‚Ä“53.
21 E.g., A. Graeme Auld, ‚ÄúThe Cities of Refuge in Israelite Tradition,‚ÄĚ JSOT 10 (1978), 135.
22 Cf. Moshe Greenberg, ‚ÄúThe Biblical Conception of Asylum,‚ÄĚ in Studies in the Bible and Jewish
Thought (JPS Scholar of Distinction Series; Philadelphia: The Jewish Publication Society, 1995),
43; Alexander Rof¬ī , ‚ÄúThe History of the Cities of Refuge in Biblical Law,‚ÄĚ in Studies in Bible
(ed. Sarah Japhet; ScrHier 31; Jerusalem: Magnes Press, 1986), 205.
DEVELOPMENT OF PLACES OF REFUGE IN THE BIBLE
The word jbzm has one meaning, ‚Äúaltar,‚ÄĚ but the semantic range of the
word !wqm is more complicated. It is the common word for ‚Äúplace.‚ÄĚ Some-
times, words with a general meaning have a technical meaning in speciÔ¬Āc
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 speciÔ¬Āc 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
speciÔ¬Āc 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 techni-
cal 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 ‚Ä“ 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 ‚Äúsanctuary,‚ÄĚ at least as can be identiÔ¬Āed in this legal
More importantly, this specialized meaning of ‚Äúsacred site‚ÄĚ 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 conÔ¬Ārmed 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 ‚Äú[dedicated] to the
23 Gary A. Anderson, SacriÔ¬Āces and Offerings in Ancient Israel: Studies in their Social and
Political Importance (HSM 41; Atlanta: Scholars Press, 1987), 31‚Ä“33. For example, the Hebrew
word hjnm generally refers to ‚Äúa gift,‚ÄĚ but in the Priestly law, it signiÔ¬Āes a particular type of
sacriÔ¬Āce, a grain offering.
24 Milgrom, Numbers, 506.
25 A. Gelston, ‚ÄúA Note on II Samuel 7:10,‚ÄĚ ZAW 84 (1972), 92‚Ä“94. Gelston also notes that 4
Q Florilegium equates !wqm of 2 Sam 7:10 with the Jerusalem Temple. Japhet, Chronicles, 317‚Ä“
318, suggests that the change in the verse was made to avoid the implication that the Temple
was standing in David‚Ä™s time.
26 A. Cowley, ‚ÄúThe Meaning of !wqm in Hebrew,‚ÄĚ Journal of Theological Studies 17 (1916),
27 David Vanderhooft, ‚ÄúDwelling Beneath the Sacred Place: A Proposal for Reading 2 Samuel
7:10,‚ÄĚ JBL 118 (1999), 625‚Ä“633.
28 KAI 119.7 and 173.5. Cf. J. Hoftijzer and K. Jongeling, Dictionary of the North-West Semitic
Inscriptions (Leiden: Brill, 1995), 2.679‚Ä“680. The word !qm appears in two Hebrew inscriptions
from the Hellenistic period referring to a synagogue, CIJ 973, 974.
78 HOMICIDE IN THE BIBLICAL WORLD
shrine/sanctuary‚ÄĚ in Philistine sites close geographically and linguistically to
At the same time that this word has a specialized meaning, it is used for
‚Äúplace‚ÄĚ 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 Ô¬Ānal stop
in the wilderness]; 11:5 [the Ô¬Ānal stop in the wilderness]; 11:24 [territory
in general]; 23:17 [a chance place to which a slave may Ô¬‚ee]; 29:6 [in
wilderness]).30 In fact, throughout the Bible, !wqm tends to be the general
word for ‚Äúa place,‚ÄĚ not one with speciÔ¬Āc status. It can, though not necessar-
ily, operate as a technical word for ‚Äúa sanctuary.‚ÄĚ
Furthermore, the word !wqm also refers to ‚Äútown.‚ÄĚ 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 homi-
cide 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‚Ä“14 as evidence for altar asylum for killers,
the Ô¬Ārst 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 state-
ment of what passed for contemporary practice and what should be the norm
in the future. For example, the uniÔ¬Ācation of worship in Deut 12:8‚Ä“14 is pre-
sented as a comparison of contemporary versus future practice. We do not
29 Seymour Gitin, ‚ÄúSeventh Century B.C.E. Cultic Elements at Ekron,‚ÄĚ in Biblical Archaeol-
ogy Today, 1990 (ed. A. Biran and J. Aviram; Jerusalem: Israel Exploration Society, 1993),
30 Two 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.
31 Levine, Numbers 21‚Ä“36, 567.
32 Schwienhorst-Schonberger, in Das Bundesbuch, 40‚Ä“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-Schonberger‚Ä™s theory does not even apply to
33 Cf. 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‚Ä“21; Lev 10:6, 9; Num 4:15, 19‚Ä“20 presumably by divine
means. Targum Pseudo-Jonathan and Targum NeoÔ¬Āti 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.
DEVELOPMENT OF PLACES OF REFUGE IN THE BIBLE
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 inno-
vation, but in its absence, could other elements serve to indicate innovation?
A motive clause offering a rationale for the statute is present ‚Ä“ ‚ÄúThe 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‚ÄĚ (Deut 19:10, 13) ‚Ä“ but it cannot be construed
as a justiÔ¬Ācation for a new legal process for two reasons: 1) It does not ex-
plicitly 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‚Ä“14 makes a clear-cut contrast between past (incorrect) worship and
(correct) worship in the future. 2) The entire Deuteronomic corpus is re-
plete with motive clauses, and it is a characteristic of Deuteronomy‚Ä™s general
A comparison with another, presumably earlier, statute could serve as
evidence. For example, the laws of the slave in Exod 21:2‚Ä“6 and in Deut
15:12‚Ä“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‚Ä“
6 appears in Deut 15:12‚Ä“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 trans-
formation is expressed is, in fact, evidence for the priority of the Exodus
passage. While Exodus has separate laws for a male slave (21:2‚Ä“6) and for
a female slave (21:7‚Ä“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, ‚ÄúIf your brother, a male Hebrew or
female Hebrew.‚ÄĚ 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‚Ä“14 dutifully takes
up the next topic in the Exodus statute, the question of monetary payment.
The slave leaves the master‚Ä™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
34 Michael Fishbane adds an additional reason, arguing that the use of the masculine form in
the succeeding verses (vv. 12b‚Ä“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 inter-
polation. 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.
80 HOMICIDE IN THE BIBLICAL WORLD
given provisions. All this suggests that Deut 15:12‚Ä“18 has reworked Exod
In light of this type of analysis, can such parallels and contradictions
between Deut 19:1‚Ä“13 and Exod 21:12‚Ä“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 differ-
ent. Exod 21:12‚Ä“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‚Ä“13, by contrast, begins with the command to estab-
lish 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 ex-
pelled 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 deÔ¬Ānes 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‚Ä“13 on Exod 21:12‚Ä“14.
It appears, then, that Deut 19:1‚Ä“13 assumes that the cities of refuge were an
institution of long standing, not an innovation.35
Deut 19:1‚Ä“13 and its parallel tradition, Num 35:9‚Ä“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 ‚Äúsacred place‚ÄĚ or ‚Äútown.‚ÄĚ 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‚Ä“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 reÔ¬‚ected in the Bible or whether they were a development from
35 Levinson argues that centralization, a Deuteronomic innovation, profoundly affected the
judicial system (Deuteronomy and the Hermeneutics of Legal Innovation, 98‚Ä“143).
DEVELOPMENT OF PLACES OF REFUGE IN THE BIBLE
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‚Ä™s wings (Pss
17:8, 57:2, 61:5), dwelling in the tent of YHWH (Pss 15:1, 61:5), and hid-
ing 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. Lohr understands these as making direct reference to altar asy-
lum. Other scholars have connected psalms to the cities of refuge. B. Dinur
identiÔ¬Āes one psalm, Psalm 27, as the ritual of admission allowing manslay-
ers 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 identiÔ¬Ācation 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‚Ä™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
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 ob-
scure 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‚Ä“34, P‚Ä™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
36 One 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.
37 Lohr, Das Asylwesen im Alten Testament, 209.
38 B. Dinur, ‚ÄúThe Cultic Aspect of the Cities of Refuge and the Ceremony of Gaining Sanctuary
in Them‚ÄĚ [Hebrew], EI 3 (1954), 144‚Ä“146.
39 L. Delekat, Asylie und Schutzorakel am Zionheiligtum: Eine Untersuchung zu den privaten
Feindpsalmen (Leiden: Brill, 1967), 11‚Ä“39.
40 Cf. Wellhausen, Prolegomena to the History of Israel, 34‚Ä“51; Karl Heinrich Graf, ‚ÄúDie