Introduction: Research and Methodology
ond, the alliance treaty must contain language that would qual-
ify it as a defense pact, a neutrality or non-aggression pact, or an
entente; and third, the effective dates of alliance have to be iden-
tifiable. Each of these criteria is discussed in turn.
System membership
The original Correlates of War state system membership criteria
were first developed in Russett, Singer, and Small (1968) and
were later expanded in Singer and Small (1972) and in Small and
Singer (1982). The criteria include states that (1) have a popula-
tion greater than 500,000, and (2) are “sufficiently unencum-
bered by legal, military, economic, or political constraints to
exercise a fair degree of sovereignty and independence” (Singer
and Small 1972, 20). The second criterion was operationalized
prior to 1920 as recognition by Britain and France, and after
1920 as formal recognition by any two major states or member-
ship in the League of Nations or United Nations.^6 The latest ver-
sion of these data is available at http://www.correlatesofwar.org/.
These volumes draw upon an original data set of system
membership for the period between 1648 and 1815. Corre-
sponding to the Correlates of War data set for the post-1815
period, this data set is divided according to whether a system
member is either a major or minor state in the system. Each of
these membership statuses is discussed in turn.
Because Correlates of War definitions of system membership
exist only for the post-1815 periods, this data set required a
functional equivalent that would be compatible with the
traditional definitions of major and minor status. Levy (1983)
provides an important beginning as he relies on careful histori-
cal research to identify the composition of what he terms the
“modern great power system.” A great power, according to Levy
(1983, 16), is “a state that plays a major role in international
politics with respect to security-related issues.” Great powers
possess the military might to impose their will on lesser states
and the ambition to have their own interests commingle with
“systemic interests” (that is, the international balance of power).
These states are also quite active in the international system,
and this activity engenders a certain respect by other great pow-
ers in addition to formal recognition by most other states. As
Levy (1983, 18) notes, the Treaty of Westphalia “named France
and Sweden as the guarantors of the peace settlement” for good
reason. These two states possessed all of the above characteris-
tics, making them the great powers of that era.
Levy’s definition of great powers is used to indicate major
status for the period between 1648 to 1815. This is a valid sub-
stitute measure because of the similarity between Levy’s list of
great powers and the Correlates of War list of major states for
the post-1815 period. These two data sets differ only with respect
to major status entry dates for four states,^7 but they never differ
regarding the identification of which states are considered
major.^8
To meet the Correlates of War definition of minor status a
state must have “all the traditional earmarks of nationhood” in
addition to meeting two specific criteria: “(a) a population
exceeding half a million and (b) de facto diplomatic recognition
from the two nations that come closest to being the interna-
tional community’s legitimizers during the period under inves-
tigation....”(Singer and Small 1968, 3). Singer and Small argue
that Britain and France were those two states for much of the
post-1815 time period. As noted, however, the Treaty of West-
phalia left Sweden and France as guarantors of the peace in- Therefore, Sweden was added to the list of Britain and
France as legitimizing states for the 1648 to 1721 time period.
After 1721, when Sweden lost its major status, Britain and
France remained as legitimizing states.^9
The population threshold for several candidate states was
checked using various demographic histories (especially Babus-
cio and Dunn 1984; Cook and Paxton 1981).^10 A lower popula-
tion threshold was initially considered appropriate because of
the lower world and European populations of this era, but this
idea was dismissed as a revised population threshold for two
reasons. First, no scaled population threshold was used by
Singer and Small (1968), even though their data set spanned
more than 150 years. Therefore, compatibility issues dictated an
extension of the half-million threshold to 1648. Second, a more
practical justification existed in the fact that few ambiguous
cases present themselves; most treaties during this time period
(the set of cases of which alliances are a subset) had major states
as parties. Few treaties involved only minor powers (six out of
lvi- See Gleditsch and Ward (1999) for a discussion and critique of these
criteria; however, use of the Gleditsch and Ward coding rules would
not greatly alter the composition of the alliance data set. For example,
during the years 1816–1944, only five treaties involving ten states had
durations that overlapped the Gleditsch and Ward (1999) data but not
the Correlates of War system membership data.
7. Levy codes the United States as a great power in 1899, but the Corre-
lates of War Project gives it major status in 1898. Likewise, Levy’s defi-
nition differs by one year compared with Correlates of War for China
and Italy (1949 instead of 1950, and 1861 instead of 1860). The only
major difference between these two data sets is the coding of Japan.
Levy states that Japan achieves great power status in 1905; the Corre-
lates of War project identifies Japan as a major state in 1895.
8. Conversations with Jack Levy and historian Paul Schroeder led me to
change the major status exit date of Spain from 1808 to 1743. The new
date is the last year of Spain’s inconclusive war against England and is
thought to be a better marker of Spain’s exit as a great power (Dupuy
and Dupuy 1977, 661).
9. Diplomatic recognition was determined through an examination of
the historical record. States having extensive contact (for example, ex-
change of diplomats or dignitaries or treaty agreements) with one of
the legitimizers were considered to be formally recognized.
10. “Candidate” states were defined as those states that were believed to
have the possibility of crossing any of the three inclusion criteria.
Among the “states” considered were: Hesse-Cassel, Ireland, Scotland,
Romania, Greece, and Bulgaria. Hesse-Cassel never crossed the popu-
lation threshold. The other states were never fully independent ac-
cording to the Singer and Small coding rules.